Citation
Healthy aging

Material Information

Title:
Healthy aging factors that contribute to positive perceived health in an older population
Creator:
Bryant, Lucinda Lynne Bruner
Publication Date:
Language:
English
Physical Description:
x, 221 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Aging -- Case studies -- Colorado ( lcsh )
Health behavior -- Age factors -- Case studies -- Colorado ( lcsh )
Older people -- Attitudes -- Colorado ( lcsh )
Aging ( fast )
Health behavior -- Age factors ( fast )
Older people -- Attitudes ( fast )
Colorado ( fast )
Genre:
Case studies. ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )
Case studies ( fast )

Notes

Bibliography:
Includes bibliographical references (leaves 200-221).
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by Lucinda Lynne Bruner Bryant.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
41462236 ( OCLC )
ocm41462236
Classification:
LD1190.L566 1998d .B79 ( lcc )

Downloads

This item has the following downloads:


Full Text
HEALTHY AGING: FACTORS THAT CONTRIBUTE TO POSITIVE
PERCEIVED HEALTH IN AN OLDER POPULATION
by
Lucinda Lynne Bruner Bryant
B.A., Brown University, 1962
M.S. Health Administrating University of Colorado at Denver, 1993
M.B.A., University of Colorado at Denver, 1993
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Health and Behavioral Sciences
1998


1998 by Lucinda Lynne Bruner Bryant
All rights reserved.


This thesis for the Doctor of Philosophy
degree by
Lucinda Lynne Bruner Bryant
has been approved
by
Arne Beck
hWgnrteg/ 12., 1^3%
Date


Bryant, Lucinda Lynne Bruner (PhJD., Health and Behavioral Sciences)
Healthy Aging: Factors That Contribute to Positive Perceived Health in an Older
Population
Thesis directed by Professor Craig R. Janes
ABSTRACT
The purpose of this study was to identify factors contributing to healthy aging,
represented by positive perceived health status, among community-dwelling older
people with chronic conditions and a history of high utilization of health care
services. The research incorporated quantitative and qualitative methods, with
regression modeling and grounded theory-type analysis of interviews.
Quantitative data came from the first year of a prospective randomized trial of
a managed care organizations outpatient group visit program. Nearly 700 subjects
(mean age 73) completed questionnaires concerning variables thought to be important
to older peoples health. Administrative records provided data concerning utilization
of health services. A. multiple linear regression model of perceived health status on
these variables: 1) identified factors significantly associated with positive perceived
health and 2) provided predicted values of perceived health for each study subject.
Comparisons between predicted and observed values of perceived health identified
individuals whose reported status most differed from predicted values. Semi-
structured interviews with a random sample of 22 of the discrepant individuals probed
for information about their interpretations of health and well-being and the relative
importance of contributing factors, with emphasis on characteristics that might
account for their differential health status assessments.
xv


1
A model of healthy aging emerged from the research. The qualitative phase of
the research found that these older people equated positive health status with
meaningful going and doing. Four categories of factors contributed to that positive
outcome: having something worthwhile and desirable to do, possessing the required
abilities, obtaining the necessary resources, and having the will or positive attitude to
go and do. The quantitative portion of the analysis, which explained almost 40
percent of the variation in perceived health status, supported these conclusions. It
found mobility, physical performance, and worsening chronic conditions
(fundamental components of ability) and depression (which is related to attitude) the
most important of the available variables.
Understanding how older people define healthy aging and identifying its most
important components provide insights into possible interventions, from both medical
and broader community sources. Future studies of health and aging might well
benefit from inclusion of the factors identified in this study.
This abstract accurately represents the content of the candidates thesis. I recommend
its publication.
Signed _
Craig R. Janes
v


DEDICATION
I dedicate this dissertation with gratitude
to Peter Bryant, who secured the roots and freed the wings to make it possible,
to Edward and Katherine Bryant, who supported and encouraged from near
and far,
to Louise Graham Bryant and Lynwood Bryant, whose graceful aging
provided the finest model,
and to the memory of Laura Lynne Bruner Boyle, whose experiences first
made me aware of the issues of healthy aging.


ACKNOWLEDGEMENTS
This research was supported in part through grants from the Robert Wood
Johnson Foundation Chronic Care Initiative to the Kaiser Foundation Health Plan of
Colorado and from the Graduate Research Opportunities Program of the University of
Colorado at Denver.
I also wish to acknowledge Dr. Craig Janes, who opened my eyes to the
multiple aspects of health; Dr. Ame Beck, who offered access to the wonderful people
whose stories lie within, provided support, and periodically got me unstuck when data
and words overwhelmed; Dr. Diane Fairclough, who with good humor and much
patience allowed me to learn how to deal with the data; Dr. Kitty Corbett, who
supported and guided the journey through research design and analysis; Dr. Jean
Kutner, who read every page, listened to every concern, validated coding, and shared
her experience with qualitative research and gerontology; friends and colleagues in
the Health and Behavioral Science program, who enriched the learning; and, last but
most certainly not least, the research subjects who so graciously shared their stories of
health and aging.


CONTENTS
CHAPTER
1. INTRODUCTION...................................................1
Healthy Aging................................................1
The Importance of Healthy Aging..............................2
Specific Aims................................................4
Overview of Research Methods.................................5
Study Setting................................................7
Description of the Study Population..........................8
Structure of the Dissertation................................8
2. HEALTH........................................................10
Historical Perspectives.....................................11
Philosophical Models of Inquiry.............................12
Definitions of Health.......................................16
Concepts of Health.......................................17
Quality of Life..........................................19
Models of Health.........................................22
Operational Definition of Health.........................27
Possible Determinants of Health.............................28
Sociodemographic Factors.................................29
Physical and Functional Status...........................29
Social and Temporal Comparisons of Status................30
Challenges...............................................31
Personal Resources and Responses.........................34


Social Support Resources....................................38
Health Care Delivery........................................42
Societal and Environmental Challenges Specific to Aging.....42
Risks and Predictors of (Unhealthy Aging: Previous Studies.....46
Mortality...................................................46
Institutionalization........................................47
Function....................................................48
Perceived Health...............................................50
Summary........................................................55
3. QUANTITATIVE ANALYSIS............................................57
Methods........................................................57
Quantitative Study Design...................................57
Study Population and Data Collection........................58
Measures....................................................59
Analysis....................................................61
Results........................................................62
Study Attrition.............................................62
Missing Data................................................64
Characteristics of the Study Sample.........................66
Regression Models...........................................72
Discussion.....................................................79
Limitations....................................................83
Summary........................................................84
4. QUALITATIVE ANALYSIS.............................................85
Methods........................................................86
Sample Selection............................................86
Interview Instrument....................................... 87


Structure of the Interviews................................88
Analytic Methods...........................................88
Interviewees..................................................91
Results.......................................................93
Ratings of Health Status...................................93
Well-Being.................................................95
Emerging Themes............................................98
Differences Between Under-raters and Over-raters..........117
Summary of Qualitative Findings..............................119
Going and Doing...........................................119
Challenges and Resources..................................120
Comparative Case Studies..................................123
Limitations..................................................125
5. A MODEL OF HEALTHY AGING.......................................127
Review of Results............................................128
The Model....................................................129
Components of the Model...................................131
Interactive Processes.....................................134
Evaluating the Model.........................................136
Conclusion...................................................138
APPENDICES........................................................140
A. Research Instruments.....................................140
B. Literature Review Tables.................................150
C. Data Description Tables..................................194
BIBLIOGRAPHY......................................................200
x


CHAPTER 1
INTRODUCTION
Healthy Aging
MacArthur Foundation Research Network researchers, in their studies on
successful aging, defined healthy aging as the absence of any deficits in Activities of
Daily Living (ADLs) and the presence of no more than one physical performance
disability (Berkman et al., 1993). That definition seems too strict. Aging is a
process, not a static entity observed at a single point in time, and it is often associated
with decreases in physiological, cognitive, and functional abilities. The process
varies among individuals, indicating that it is influenced to some extent by
environmental, behavioral, or genetic conditions, which themselves are highly
variable (Berkman, 1988, p. 39). Healthy aging does not mean the absence of
decline but the best possible health status for the individual and his or her best
adaptation to the aging process. In addition to ability levels, it depends on care and on
coping with, reacting to, and managing the aging process. Sharing this view of
healthy aging, or successful aging in their terms, Thomas and Chambers (1989)
described it as first, self-evaluated life satisfaction, relating to the adaptive tasks of
aging, and, second, an external evaluation of the success with which the individual
has handled the developmental tasks of coming to terms with the problems of bodily
decline and eventual death (p. 185). Curb et al. (1990) called it effective aging,
deriving from the adaptation and rehabilitation that permits the maintenance of
relatively high levels of functioning in many older people, despite physiologic
declines, increased numbers of risk factors, and the presence of diagnosed disease.
1


The Importance of Healthy Aging
Older Americans, aged 65 and older, represent an increasingly larger
percentage of the population. As Table 1.1 shows, since 1900 the percentage of those Table 1.1. . Actual U.S. Population, 1900-1996
Year Total population Aged 65 and older Aged 65+ as percent of total
1900 762 3.1 4.1%
aged 65 and older in the 1920 106.0 4.9 4.6%
United States has more than 1940 132.2 9.0 6.8%
1960 1793 16.7 9.3%
tripled, to almost 13 percent 1980 226.5 25.7 11.3%
of the population; the number 1996 265.2 33.9 12.8%
has increased more than Note. Population figures in millions. Adapted from U.S.
Administration on Aging, Profile of older Americans: 1997.
tenfold (U.S. Administration Available: http://pr.aoa.gov/aoa/stats/profile (30 July 1998).
on Aging, 1998). The U.S. Bureau of the Census (1998) predicted they will represent
20 percent of the population by 2050. As Table 1.2 indicates, the number of the
oldest-old, those 85 and older, is growing at the greatest rate. Now 3.7 million, their
numbers are projected to increase nearly fivefold by 2050.
Table 12. Projected U.S. Population, 1996-2050 (population in millions)
year total population aged 65 and older aged 65+ as percent of total aged 85 and older aged 85+ as percent of total
1996 2652 33.9 12.8% 3.7 1.4%
2000 274.6 34.7 12.6% 43 1.6%
2010 297.7 39.4 13.2% 5.7 1.9%
2020 322.7 532 16.5% 6.5 2.0%
2030 346.9 69.4 20.0% 8.4 2.4%
2040 370.0 75 2 203% 13.6 3.7%
2050 393.9 78.8 20.0% 182 4.6%
Note. Population in millions. Adapted from U.S. Census Bureau, Resident population of the United
States: Middle series projections, 1996-2050. Available: http://www.census.gov/population/
projections/nation/nas (28 July 1998).
2


People in general are living longer. Compared to a child bom in 1900. who
had an average life expectancy of 47 years, a child bom in 1996 could expect to live
76 years. Most of this increase has come from reduced death rates for children and
young adults, not from increases in the maximum possible age (U.S. Administration
on Aging, 1998).
An ongoing debate questions whether we are reaching the limits of increased
life expectancy, with related concerns that increased life spans may be burdened with
longer end-of-life periods of disability and dysfunction. Fries (1980) reported that,
while life expectancy has increased, the maximum life span has not. His studies
suggested that longer years of life do not necessarily mean longer periods of
disability, but with great variability among individuals. Present approaches to social
interaction, promotion of health, and personal autonomy may postpone many of the
phenomena usually associated with aging. The rectangularization of the survival
curve may be followed by the rectangularization of the morbidity curve and by
compression of morbidity (p. 135). Manton and Soldo (1985) took a less optimistic
view, focusing on the reduction of mortality at advanced ages that will lead to a rapid
increase in the population aged 85 and older, the population group with the highest
per capita service needs. They too, like Fries, cited evidence that functional
impairment among the elderly is not a natural consequence of aging and must be
evaluated on an individual basis but warned that whatever interventions may be
introduced, the demographic aging of the population will cause large increases in the
number of disabled elderly (p. 53). Fozard, Metter, and Brant (1990) cautioned that
the aging of the population may mean that in contrast to the past, when the oldest old
was a very select group who had the grit for survival, now a larger part of the
population is surviving to these ages, and may reflect a different kind of aging with
less of a survivorship issue (p. PI 17).
3


Whether or not the onset of morbidity has been delayed, the healthiness of
older people obviously affects the individuals themselves. As Curb et al. (1990) so
clearly stated,
It is important that scientists and clinicians adopt a unified concept of
aging that allows for prevention, treatment, and compensation or
rehabilitation with the ultimate goal of developing health-care practice
and policies that will maximize the quality of life for the largest
number of older people, (p. 828)
Healthiness, or its absence, in this rapidly increasing population also has
serious implications for demands on health care and other social resources. Although
the average annual increase in national health expenditures decreased from nearly 13
percent in 1980 to less than five percent in 1995 and 1996, the total amounted to more
than a trillion dollars in 1996. Medicare provided $203 billion of that total, one
measure of the cost of providing health care for those aged 65 and older. In 1995,
non-institutionalized older people accounted for 38 percent of hospital stays and 48
percent of days of care in hospitals, and older people averaged more than twice as
many contacts with doctors as those under 65 (11 vs. five contacts). This older
segment of the population accounted for 36 percent of personal health care
expenditures in 1987, for a total of $162 billion and an average of $5,360 per person
compared to $1,290 for younger people (U.S. Health Care Financing Administration,
1998). These numbers underscore the need to understand the factors that contribute
to healthier aging, in order to support better lives for older individuals with the most
appropriate and prudent allocation of resources.
Specific Aims
In order to explore healthy aging and identify factors that contribute to it, this
project analyzed both health status questionnaire and utilization data originally
collected during a prospective randomized trial of a group model health maintenance
4


organizations (HMOs) outpatient group visit program and, subsequently, results of
interviews with a subset of that population. The study pursued the following specific
aims:
Specific Aim 1:
Specific Aim 2:
Specific Aim 3:
Specific Aim 4:
Construct a statistical model to describe and predict
positive perceived health from a variety of demographic,
clinical, functional, social support, and utilization variables
previously associated primarily with negative outcomes
related to aging.
Identify discrepant cases, those individuals whose reported
perceived health differed substantially from values
predicted by the model.
Ascertain and describe the characteristics of those
individuals whose perceived health ratings differed from
those predicted by the model, with the goal of identifying
additional factors that support successful, healthy aging and
retard declining health.
Describe a model of healthy aging for this older population.
Overview of Research Methods
It is possible that no single methodological approach or simple set of
assumptions will provide definitive health predictors. We emphasize
the need for a series of imaginative studies in a variety of populations,
done in diverse ways so that the more robust predictors will emerge
[to] create a realistic background for determining the most efficient
methods for reducing the burden of morbidity and mortality in the
rapidly increasing elderly population. (Benfante, Reed, & Brody, 1985,
p. 394)
To achieve the aims of the study, research employed both quantitative and
qualitative methods. Construction of a multiple linear regression model had two
5


purposes. First, it identified factors significantly associated with perceived health,
from the parent randomized trials variables and data. Second, comparisons between
values of perceived health predicted by the model and those actually reported by the
studys subjects identified those whose reports most diverged from their predicted
values, both positively and negatively. A randomly selected sample of the discrepant
cases participated in semi-structured interviews. The interviews explored their
definitions of well-being, reasons for the assessments they made of their health status,
reactions to factors suggested by the literature as contributors to health, and their ideas
about why people might under- or over-rate their health. Grounded theory-type
qualitative analysis of the interviews generated additional insights into characteristics
of and factors associated with healthy aging. The final step of the research combined
the results of the analyses to construct a model of healthy aging and the factors that
contribute to it
The following diagram illustrates the relationships among the parent study, the
quantitative and qualitative portions of this study, and the final model.
6


Study Setting
Quantitative data and the pool of study subjects for this study came from the
randomised trial of Kaiser Permanente of Colorados (Kaisers) innovative program
to provide primary health care to older members in a group setting. The Cooperative
Health Care Clinic (CHCC) program combined clinical care, health education, and
social interaction. It targeted the health care needs of an older, community-dwelling
subpopulation with a history of chronic disease and greater-than-average utilization of
provider services. Physicians could choose this type of care for their older patients
but were not obligated to do so.
Groups of about 20 members each met monthly for approximately two hours.
They received basic health care during the meetings, along with educational
presentations concerning relevant issues; presenters included the groups physicians,
psychologists, health educators, dietitians, physical therapists, and many others.
According to Kaisers guide, the CHCC program focused on the elderly populations
complex health problems, their coping ability, the assistance they need to maintain
independence, the need for coordination among resources, and the desire to promote
longer life with decreased illness and disability (Cooperative Health Care Clinics,
1993).
To evaluate the impact of the CHCC model of care, Kaiser conducted a
randomized trial of nearly 800 members, half of whom had the opportunity to receive
care in the CHCC setting and half who received usual care. It is that parent study that
provided quantitative data and study subjects for this study.
7


Description of the Study Population
The mean age of the study subjects was 73 at the beginning of the parent
study. All lived in the
communitythis was not a frail
population in need of
institutional care. On average,
each reported 2.7 chronic
conditions; 60 percent indicated
they had arthritis or rheumatism,
49 percent hypertension, and 20
percent trouble hearing. In
many ways they mirrored the
Table 13. Study Population Compared to the U. S.
Population: Selected Characteristics
Characteristic CHCC Population U.S. Population
Gender 62 % female 59 % female
Race 89% White 85% White
Live Alone 29% 30%
Live with Spouse 62% 60%
Married 63% 56%
Male 78% 76%
Female 51% 43%
Note. U.S. data from U.S. Administration on Aging,
Profile of Older Americans: 1997. Available:
http://pr.aoa.gov/aoa/stats/profile (30 July 1998).
noninstitutionaiized population of older people in the United States. Table 1.3
compares characteristics of this sample with the U.S. Department of Health and
Human Services 1997 profile of noninstitutionaiized older Americans (U.S.
Administration on Aging, 1998). Chapter 3 and Appendix C contain more complete
descriptions of the study population.
Structure of the Dissertation
This chapter of the dissertation introduced the issues; stated the specific aims
of the research; described the research methods, study setting, and population; and
now outlines the structure of the following chapters. Chapter 2 reviews perspectives
on health, previously described definitions and models of health, factors hypothesized
in the literature to contribute to health, and previously published studies of risks and
8


I
I
I
predictors of (unhealthy aging. It also explains the choice of perceived health as the
outcome of interest
Chapters 3 and 4 provide more complete details of methods for the
quantitative and qualitative analyses, respectively, as well as results and discussions
of results specific to the analyses. Chapter 5 brings together the results of both types
of analysis to produce a model of healthy aging for this older population.
I
i
t
i
9


CHAPTER 2
HEALTH
Exploration into factors that contribute to healthy aging logically begins with
attempts to construct an operational definition of health or healthy. As R.G. Evans
and Stoddart (1990) cautioned, definitions and determinants of health are separate
although related enterprises: Whatever the level of definition of health being
employed, however, it is important to distinguish this from the question of [its]
determinants (p. 1348). Susser (1974) established the causal progression, stating
that the way health is defined is a necessary antecedent of the way health is
measured (p. 539). He explained the important and complicating dependence of
definitions on the value systems of those who do the defining. Definitions contain
ethical components that rest on those value systems, which are consequences of
economic and class interests, political and social power, and culture created by
societies historical evolution. Tesh (1988), although referring to theories of disease
(and consequent preferences for different prevention policies) rather than definitions
of health, relevantly identified the importance of understanding these hidden
arguments that underlie definitions and theories.
Wallace (1994), with special reference to the health of older people, also
emphasized the need to begin with definition as the basis for the assessment of
factors. He, like Susser (1974), asserted that values, ethics, and personal preferences
affect definitions of health and may differ among health professionals, sick patients,
and otherwise healthy nonprofessionals, as might personal and family disease
experiences and levels of general health education. He also referred to structural
environmental factors:
10


Health definitions must also be considered in their social, political, and
economic contexts. Health state perceptions may be highly
conditioned by the presence or absence of war or famine, political or
social unrest, or threats from the natural or man-made environment
Health perceptions may also vary according to access to special
helping programs, or adherence to a particular ideology, either political
or religious. (p. 449)
Health then, for any age, is a complex construct whose definition depends on
historical, social-political-economic, and philosophical perspectives as well as clinical
ones.
Historical Perspectives
Historically the level of knowledge of biological systems has affected
concepts of health. The Greeks, who defined biological processes in terms of bodily
humors, believed in health as a state of equilibrium and understood that health
depends on ones heritage, economic situation, physical surroundings, and
poiitical/social environment (Hippocrates, n.d.). The uncertainties and fears of the
Dark Ages displaced these rational perceptions of health, replacing them with beliefs
that man had no control over threats to health, which were considered to be Divine
punishment.
The view of health (and nature in general) as Divine plan faded (at least in
Western cultures) with the rise of Cartesian mechanistic views of the body as a
machine. As Dubos (1959) wrote, the myths of Hygeia and Asclepius symbolize ...
two different points of view.... For the worshipers of Hygeia, health is the natural
order of things, a positive attribute to which men are entitled if they govern their lives
wisely.... More skeptical or wiser in the ways of the world, the followers of
Asclepius believe [in treating] disease, to restore health by correcting any
imperfection caused by the accidents of birth or of life (p. 131). By the early
11


nineteenth century, hospital-based physicians examined disease in specific organs, not
the health of the person as a whole. The discovery of the germ theory of disease,
emphasis on laboratory science, and the Flexner reports scientific orientation to
medicine reinforced the health-as-absence-of-disease perspective (Jones & Moon,
1987). From the mid-nineteenth until at least the mid-twentieth century, in the United
States, advances in biomedicine and the position of professional sovereignty held by
physicians focused attention on clinical intervention to cure or mediate specific
medical conditions (Starr, 1982). With market changes in recent years, vesting
greater authority in payers and insurers and less in providers of care, health
increasingly has been viewed as commodity and health care as a business.
Philosophical Models of Inquiry
Philosophical models of inquiry provide tools for exploring the multiple levels
of factors that constitute health. Models range from a positivist belief in a single,
knowable, fact-based reality driven by immutable natural laws and mechanisms
(Guba, 1990, p. 20); to an interpretive-constructivist vision of reality as multiple
mental constructions whose form and content depend on the social and experiential
base of the persons who define them; and further to a critical view that ideology and
prevailing political power determine, often falsely, what people see as reality. In
health terms, a strictly positivist, biomedical definition restricts evaluation to clinical
indicators of changes in disease states related to medical inputs. The interpretive-
constructivist view includes social and cultural factors related to wellness. A
political-economic (critical) construct of health declares that structural, population-
based variables such as income differentials among social groups, access to political
power, and environmental degradation determine health-related quality of life.
Kleinman (1988) clarified the distinctions among these constructs, but from a
perspective of negative rather than positive health states. He described disease as an
12


I
entity circumscribed by biological structure or functioning; illness as the innately
human experience of symptoms and suffering... how the sick person and the
members of the family or wider social network perceive, live with, and respond to
symptoms and disability (p. 3); and sickness as the broader, population-level
understanding of a disorder in its generic sense across a population in relation to
macrosocial (economic, political, institutional) forces ... seeing it as a reflection of
political oppression, economic deprivation, and other social sources of human
misery (p. 6).
Frankenberg (1988) summarized the temporal and social relationships among
these levels of challenges to health. He identified disease as a condition without
temporal boundaries, a disturbance of body functions and performance seen in
biological terms of the kind that physicians are primarily trained to detect, diagnose,
and treat (p. 16). Disease made personal by the sufferers present-tense experience
becomes illness, and sickness provides the social and cultural framework... to
analyze the social consequences of illness and disease (p. 17).
These constructs of challenges to health lead to definitions of health as the
absence of something. The empiricist or positivist biomedical model focuses on
disease resulting from the detrimental effects of a clinically-defined agent on the host,
whose genetic character determines its susceptibility. Health in this context means
absence of disease, or as Doll (1992) defined it, a state distinguished by the absence
of disease or of physical or mental defect, that is, the absence of conditions that
detract from functional capacity whose incidence can be measured objectively (p.
933).
In the broader context of absence of illness, health becomes an interpretive-
constructivist concept of wellness. This construct has substantial subjective
componentsindividual, social, and cultural. Dubos (1959) stated that health (and
disease) cannot be defined just in terms of biomedical attributes but must be measured
I
i
i
i
13
I


I
by the ability of the individual to function in a manner acceptable to himself and to
the group of which he is a part (p. 261). He continued further,
Men naturally desire health and happiness. For some of them,
however, perhaps for all, these words have implications that transcend
ordinary biological concepts. The kind of health that men desire most
is not necessarily a state in which they experience physical vigor and a
sense of well-being, not even one giving them a long life. It is,
instead, the condition best suited to reach goals that each individual
formulates for himself. Usually these goals bear no relation to
biological necessity; at times, indeed, they are antithetic to biological
usefulness. More often than not the pursuit of health and happiness is
guided by urges which are social rather than biological; urges which
are so peculiar to men as to be meaningless for other living things
because they are of no importance for the survival of the individual or
of the species, (pp. 278-279)
Jones and Moon (1987), combining interpretive and critical viewpoints, explained
this view of health as a social and moral judgment that varies according to the
societys norms, expectations, and culturally shared rules of interpretation.
A strictly critical perspective views health as the absence of sickness, which
incorporates those societal and political conditions that create harmful environments.
Environment includes social-psychological, political-economic, and historical-cultural
influences as well as physical conditions. Ulich (1976), from this perspective,
explained that health, after all, is simply an everyday word that is used to designate
the intensity with which individuals cope with their internal states and their
environmental conditions (p. 7). In the same vein, Gil (1993) defined health as not
only the absence of specific diseases but also the ability to develop and function
according to ones innate physical, intellectual, emotional, and social potentials
without blockage or waste from the environment, in order to satisfy ones intrinsic
needs and enrich society. He felt that if societys established ways thwarted the
individuals fulfillment, then constructive developmental energy would become
destructive energy resulting in physical, emotional, intellectual and social sickness.
14


Labonte (as cited in Eisen, 1994), noted that the prerequisites to health are no longer
simply disease prevention, or proper lifestyles, but include peace, shelter, education,
food, income, a stable ecosystem, social justice and equity (p. 237). To this list Link
and Phelan (1996) added knowledge, money, power, prestige, and social connections,
all culturally constituted by the community.
As Liang (1986) pointed out, although these points of view treat health as a
residual category defined primarily by absence of disease, illness, or sickness, health
incorporates more than that. Dubos (1959) explained further that
solving problems of disease is not the same thing as creating health
and happiness. This task demands a kind of wisdom and vision which
transcends specialized knowledge of remedies and treatments and
which apprehends in all their complexities and subtleties the relation
between living things and their total environment Health and
happiness are the expression of the manner in which the individual
responds and adapts to the challenges that he meets in everyday life.
And these challenges are not only those arising from the external
world, physical and social, since the most compelling factors of the
environment, those most commonly involved in the causation of
disease, are the goals that the individual sets for himself, often without
regard to biological necessity, (p. 26)
Jylha (1994) explained the difference between health and disease this way:
constructs of health and disease differ in a profound, qualitative sense: definitions of
disease are formal while definitions of health are social and contextual (p. 985).
Lewis (1953) suggested that the dualist language of health and disease is
unavoidable and the fictions, health and disease, serve a useful intellectual purpose,
though we know they refer merely to uplands and lowlands in a continuously graded
and terraced country (p. 110). The point of exploring factors related to positive
health, rather than focusing on negative risks that challenge it, is to view the range
from the healthy end rather than the more usual obverse. Antonovsky (1987) called
this focus salutogenesis and framed his inquiries into contributing factors in this way:
15


A salutogenic orientation (which focuses on the origins of health)
poses a radically different question: why are people located toward the
positive end of the health ease/dis-ease continuum, or why do they
move toward this end, whatever their location at any given time? (p.
xii)
He characterized his work with the metaphor of a river as the stream of life, observing
that many people choose to jump into the river while refusing to leam to swim, or in
terms of health, while not adopting beneficial health behaviors. He then asked,
Wherever one is in the streamwhose nature is determined by historical, social-
cultural, and physical environmental conditionswhat shapes ones ability to swim
well? (p. 90).
Definitions of Health
The challenge is to construct a definition of health as a positive attribute or
condition, a definition that incorporates the multiple perspectives outlined above and
specifies factors that might be amenable to intervention. It is not an easy task. An
operational definition of health has become increasingly difficult as the emphasis of
medical and health care has shifted from decrease in mortality and increase in
longevity to improvement in the health-related quality of life (Bergner & Rothman,
1987, p. 191). Such an inclusive definition of health must contain more than the
absence of specific biomedical conditions. For example, the World Health
Organization defined health as complete physical, mental, and social well-being, not
merely the absence of disease or infirmity (Breslow, 1990, p. 9). Most conceptual
constructs of health would include some aspect of physiological or biological status,
mental state, physical and social functioning, and health behaviors and attitudes but
probably different views on the parts of the construct that contribute most to overall
health (Bergner & Rothman, p. 192).
16


Concepts of Health
Many have tried to capture the concept in words. Liang (1986) focused on
physical health and suggested three major approaches, which encompass the positivist
and interpretive-constructivist perspectives. A medical model or physical definition
speaks of health as a residual category defined by absence of disease (p. 248). In a
functional model or social definition, health is a state of optimum capacity for the
performance of ones roles and tasks for which the person has been socialized, and
physical health is equated with conformity to norms of physical and mental capacity
for adequate participation in social activities (p. 249). (References to the ability to
fulfill social roles appear repeatedly in attempts to define health.) A psychological
model incorporates a subjective evaluation of health as the individuals perception
and evaluation of his or her overall physical health... which represents a summary
statement concerning the ways in which various aspects of health, subjective as well
as objective, are combined within ones perceptual framework (p.249).
Murray, Dunn, and Tamopolsky (1982) provided similar psychological
definitions but with a focus on the psychoanalytical. First, a traditional medical-
psychiatric definition views healthy people as the residue when those with manifest
pathology have been identified. A psychoanalytic view defines health as the
achievement of optimal integration of the individual. Finally, a sociopsychological
definition again includes the optimum capacity to perform effectively the roles and
tasks for which the individual has been socialized, i.e., to meet expectations. Murray
et al. considered health to be interchangeable with normality, which provides the
standard against which to judge the abnormal. It is not a new concept Centuries ago
Chaucer counseled, Ocupye the meene by stydefast strengthes, for al that ever is
undir the meene or elles al that overpasseth the meene despiseth welefulness
(Boethius, iv, in prose 7).
I
I
i
i
i
17
i


Murray et al. (1982) noted the difficulty of establishing such a standard. Since
no objective criteria of health have been established, they wrote, health-normality
must be placed in the context of human variability, with reference to the individuals
material and social environment For example, an athletes perception of disability
from impaired mobility would differ from that of an older, more sedentary person.
J.G. Evans (1984) cautioned that although the traditional conception of aging
presumes normal and pathological'disease processes, there is an absence of any
theoretical or practical means of defining normal aging [and] certainly no grounds
for assigning... that what occurs normally, i.e. commonly, in Western societies is
necessarily normal, i.e. healthy or optimal (p. 355).
Focusing on health as ones level of functioning, Patrick, Bush, and Chen
(1973) considered two components: function at a point in time, evaluated in terms of
social preferences for various functional levels, and the expected transition to other,
more or less favorable, levels in the future. They defined optimum function, i.e.
health, as the ability to conform to societys standards of physical and mental well-
being and to perform the activities usual for the individuals age and social role. In
contrast to Liang (1986) cited above, these authors emphasized performance, not
capacity. Susser (1974) too believed that health has multiple dimensions that include
individual-level organic (disease, impairment) and functional (illness, disability)
components as well as social levels of social dysfunction, sickness, and handicap.
Bergner (1985) added health potential to the definition and identified five
dimensions of health: genetic or inherited basic physiological structure; biochemical,
physiologic, or anatomic condition (disease state, disability or handicap state);
functional condition (social-role, physical, and cognitive performance); mental
condition (mood or feeling state, affective state); and health potential (longevity,
functional potential, disease and disability, disadvantage, prognosis). She further
listed factors that affect health status, which include societal issues such as
18


environmental quality, housing, crowding, and sanitation; the availability of and
access to health care; social and familial issues including personal health attitudes and
behavior, resources, and the physical condition of those close to individual; and
personal issues related to personal health care, social network, coping skills, and
available resources. Feinstein (1993) discriminated between two types of personal
characteristics that may cause differences in health status: non-resource-dependent
behavioral characteristics including psychological, genetic, and cultural factors and
resource-dependent characteristics such as wealth and material ownership.
One study actually asked people aged 60 and over What do you think most
people around your age mean when they say they are in good health? (Strain, 1993,
p. 343). From a selection of responses, over 40 percent chose the ability to perform
usual activities, about one third selected a definition of good health as a general
feeling of well-being, and fewer than 20 percent chose the absence of symptoms.
Quality of Life
According to Wallace (1994), conventional interpretations of health status
have a positivist flavor, with their focus on clinical processes (e.g., signs and
symptoms, diseases, mortality), alterations in anatomy and physiology, and individual
functional attributes. He suggested that definitions should also embrace broader
measures of the quality of life, incorporating a range of clinical, social, and
psychological health constructs. Additionally, he continued, they might include
prenatal and birth events, features of human development and maturation, an
assessment of culturally conditioned psychological and behavioral factors (e.g.,
abstract cognitive and social abilities, performance in typical environmental settings,
response to challenges), and hygienic and preventive behaviors. He added an
additional level, environmental factorsphysical, biological, and socialas well as
potential services and resources in the community and the medical care system.
19


I
Edlund and Tancredi (1985) enumerated five possible definitions of quality of
life, pointing out what they believed to be the ideological underpinnings of each (their
hidden arguments, in Teshs, 1988, terms). First, quality of life as fulfillment of
personal goals, based on telic theory, mirrors the American cultural emphasis on self-
actualization. A second definition, the ability to lead a normal life, reflects the
political popularity of normalcy in times of great change. The ability to lead a
socially useful life, their third definition of quality of life, depends on the size of ones
sociological frame and might be restricted to political and economic concerns with
employment or expanded to include social and personal roles. The fourth definition,
that of the rational man, assumes the ability to generate objective, expert criteria
against which to gauge individual status. Dangers exist when the experts underlying
assumptions are obscured. Finally, the individualistic view defines quality of life as
whatever an individual personally declares it to be. Consonant with democratic
choice valued by American culture, it also brings risks of unchecked individualism,
with values that may run counter to those of the general community.
Diener (1984) reviewed psychological theories that attempt to explain what
contributes to well-being (or happiness, which he equated with subjective well-being).
Telic, or endpoint theories, assert that people gain happiness when they achieve their
goals. Maslows (1970) work on motivation exemplifies this category. He suggested
levels of needs that, when fulfilled, contribute to health-as-well-being. Dependent on
the environment but focused more on the individual, his hierarchy begins with
biological-physiological-material needs for the basic resources for existence and
progresses to the needs for safety and security, love and belongingness, esteem based
on ones own productivity and creativity, and self-actualization or self-fulfillment.
Diener cited a related set of theories that also explain well-being in terms of goal
achievement, but only to the extent that success has overcome an existing deficit or
pain. This point of view suggests that permanent fulfillment of all needs precludes
i
I
!
20
(


complete satisfaction or well-being. Activity theories maintain that well-being
depends on the activity involved in reaching a goal rather than attainment of the goal
itself. Diener distinguished between bottom-up and top-down theories. Bottom-up
theories maintain that well-being or happiness is an accumulation of small pleasures,
tallied by the individual. Top-down theorists believe that global personality features
influence an individuals reaction. Cognitive association, the ability to associate
current emotional situations with cumulative past experience, may help explain the
development of essentially happy or unhappy temperaments, effectively linking
bottom-up and top-down perspectives. Judgment theories postulate that comparisons
between actual conditions and not-necessarily-conscious standards determine well-
being. Any deviation from normal will therefore change an individuals state of well-
being.
Farquhar (1995a) also reviewed the literature on quality of life. As in the
descriptions of theories above, she determined that global definitions usually
incorporate judgmental or cognitive experiences of (dis)satisfaction and affective
experiences of (un)happiness. Quality of life includes both conditions of life and the
experience of life. She found references to the possession of adequate resources to
satisfy needs, participation in activities that lead to self-actualization, and satisfactory
comparisons between self and others. Farquhars review contributed a useful listing
of specific components that contribute to quality of life. They include general health
and functional status, level of activity, comfort, mental state and longevity,
socioeconomic status, subjective evaluations of life satisfaction and self esteem, and,
especially for older people, concepts of privacy, freedom, respect for the individual,
freedom of choice, emotional wellbeing and maintenance of dignity (p. 504).
Expanding the definition of health to incorporate quality of life complicates
the exercise. By its nature, quality of life is a subjective assessment of well-being
rather than a more objective enumeration of clinical, functional, or even social
21


attributes. The primary difficulties, however, are of measurement but not of
definition; they do not diminish the benefits of the broader perspective that enriches
our understanding of the construct of health. One additional warning may be useful.
Robertson and Minkler (1994) cautioned that viewing health broadly, as with the
World Health Organizations definition of health that refers to the extent to which an
individual or group is able, on the one hand, to realise aspirations and satisfy needs;
and on the other hand, to change or cope with the environment (p. 298), may lead us
to minimize the individuals everyday reality of pain, discomfort and difficulty
Models of Health
The previous subsections of this chapter explored concepts of health and well-
being. Some theorists and researchers have taken another approach to defining
health, by constructing models.
Before investigating these models of health, one should note Kasls (1983)
cautionary point: the development of broad principles of human functioning can
generate beautiful, complex constructs that presume a multi-factorial whole when
they are actually composed of quite discrete heterogeneous elements. For example,
social support comprises a number of components: breadth and size of network;
amount of support given, received, and perceived; and type of support (e.g.,
instrumental, emotional, informational). Echoing Robertson and Minklers (1994)
concern about too broad definitions of health as well-being, Kasl warned that
combining components into a theoretical whole may mask important individual
effects and interactions. These thoughts suggest that useful models of health should
incorporate factors and interactions with the following characteristics:
they are specific and conceptually clear;
they are related to individuals and their well-being;
they can be measured and assessed; and
22


they are amenable to intervention, at least distally.
Theorists and researchers have constructed a variety of models to describe the
relationships between and among the various components of health. Barsky, Cleary,
and Klerman (1992) noted that most patients experience their health as a global
experience and a level of function, as an overall state of well being (p. 1147), in a
model that incorporates a number of factors. The factors include medical status;
functional impairment; psychological dysfunction and emotional distress; social
factors, role demands, and stressful life events (e.g., social isolation, adverse life
events, unemployment, dissatisfaction with ones life circumstances); and age.
Boult, Kane, Louis, Boult, and McCaffrey (1994) diagrammed the
relationships between chronic conditions and disability. They adapted World Health
Organization and U.S. Academy of Sciences Institute of Medicine models,
suggesting a sequential relationshipfrom chronic medical conditions to impairment
of abilities, to functional limitation, and finally to disabilitybut also observing that
disability ... may be caused, not only by physical or mental limitation, but also by
cultural expectations, environmental obstacles, or lack of motivation and training (p.
M28).
Blaum, Liang, and Liu (1994) also hypothesized an interaction among chronic
diseases, disability, and self-rated health status, controlled for exogenous factors of
age, gender, race, education, and social isolation. They proposed that these three
components combine to create physical health status and that all three separately and
together affect utilization of health services. They referred to the Anderson Health
Behavior Model of Utilization that, although focused on utilization, offers insights
into the multiple factors that may affect an individuals sense of well-being. The
factors include predisposing characteristics (e.g., health beliefs and attitudes),
enabling characteristics (e.g., insurance and socioeconomic status), and need
characteristics (e.g., disability, health status, disease; Anderson & Bartkus, 1973).
23


I
Wolinsky, Callahan, Fitzgerald, and Johnson (1992) explained these constructs more
fully: predisposing refers to the propensity for using services, dependent on
sociocultural characteristics (e.g., demographics, social structure, health beliefs);
enabling encompasses the (economic) means for obtaining services; and need refers
to the perception of illness or its possibility, involving both the individuals
assessment of the amount of illness and professionally evaluated need.
The preceding authors focused on disability rather than on health, but they
suggested factors that belong in any model of health: physical or medical condition,
functional ability, demographics, social influences, health beliefs, perceived well-
being, and socioeconomic status. R.G. Evans and Stoddart (1990) proposed a model
of health that includes and expands upon these factors, based on their observed need
for a framework to describe the complex causal patterns among components of health
and also between health and health care. They suggested important features of such a
framework, which they asserted should
accommodate distinctions among disease, as defined and treated by the
health care system, health and function, as perceived and experienced
by individuals, and well-being, a still broader concept to which health
is an important, but not the only, contributor [and include]
consideration of both behavioural and biological responses to social
and physical environments, (p. 1362)
Building on the positive and negative biomedical and economic interactions between
health care and disease (and so still, at least initially, focused on disease rather than
health), they incorporated the following components, in subsequent stages:
lifestyle (in terms of individual behavior), environment, and human
biology as contributors to disease;
an expansion of lifestyle and environment to acknowledge the impact of
social and physical environment on lifestyle, or rather on individual
response, which in turn evolves from behavior and biology (genetic
endowment);
24


a feedback loop from individual response to health and function, to well-
being, and back to individual response; and
the introduction of prosperity as a measure of economic trade-offs and
their impact on well-being and the individuals environment
Figure 2.1 shows a graphical representation of the final stage of their model,
adapted from their diagrams. They cautioned that such a representation
oversimplifies the concepts: The entities which form the components of our
framework are themselves categories, with a rich internal structure (p. 1349). They
explained that each box represents a complex concept that in general cannot be
adequately represented by a single homogeneous variable and added that it may be
interactions between factors from different categories that are critical to an
individuals or a populations health.
Figure 2.1. Representation of R.G. Evans and Stoddarts Model of Health
Wilson and Cleary (1995) constructed a more sequential conceptual model of
health, or taxonomy of patient outcomes, that categorizes measures according to the
underlying health concepts they represent and suggests causal relationships among
25


Characteristics
of the
Individual
Sodaiand Social and
Economic . Psychological
Supports Supports
X T X"
nwivtwkHfic Nonmedical
of the Factors
Environment
Figure 2.2. Wilson and Clearys Multifactorial Model of Health
Reprinted, by permission, from Journal of the American Medical Association, 1995,273(1):
60. 1995, American Medical Association
them. Figure 2.2 shows the relationships they described. Their model contains five
main categories: biological and physiological factors, symptoms, functioning, general
health perceptions, and overall quality of life. It also acknowledges the importance of
individuals characteristics, including personality and values, and characteristics of
the environment including nonmedical factors.
Moore, Van Arsdale, Glittenberg, and Aldrich (1980) proposed a model of the
human ecosystem to explain the multiple interacting factors that affect health. Their
focus was on disease and biological capacity, but the model has broader applicability.
It consists of two subsystems, the first concerning the macrolevel environment
(structural environmental characteristics) in which the individual lives and the second
the individual. Health in this model means equilibrium, which individuals achieve by
coping with and repairing the disequilibrium, or sickness, that has occurred when
26


their adaptive capacities have been threatened and exceeded. The dynamic, context-
specific processes that produce sickness and restore health evoke behavioral
responses that, insofar as they express and reinforce an individuals attitude toward
self and the world,... very definitely influence the individuals adaptive powers (p.
24), which then may affect future coping and repair responses as well as the
macrolevel environment
Operational Definition of Health
Based on the concepts and models described above, an operational definition
of health or healthiness (which must precede explorations into determinants of
healthy aging) should include, in Farquhars (1995a) felicitous phrase, both
conditions and experiences of life that include the following:
sociodemographics;
biological, physiological, or anatomic status, including genetic background
and the absence of disease or disorder;
physical functional abilities; level of activity;
cognitive abilities;
affective state; absence of depression;
personality traits, e.g., optimistic or pessimistic attitude;
ability to fulfill ones social role and to receive respect and dignity;
health behaviors and attitudes;
satisfaction and happiness;
achievement of goals; self-actualization; worthwhile activity; self-esteem;
favorable comparisons with context-dependent normality or standards;
social environment; social and familial resources and challenges; and
physical, economic, and political environment, including access to health
care.
I
27
I


Categories of factors alone do not provide a definition. That requires an
understanding of the relationships among them. The models proposed by Boult et al.
(1994) and Blaum et al. (1994) included the first (physical and functional) categories.
Blaum et al. added beliefs about health and interactions with perceived health. R.G.
Evans and Stoddarts (1990) model introduced the complexity of relationships
between the individual and the environment and the interactions between those
relationships and disease. Wilson and Cleary (1995) expanded that model,
introducing characteristics of the individual (personality, motivation, and values) and
environmental supports. They also offered a simpler model, suggesting a sequential
progression of health status outcomes from strictly biomedical concerns to the
broadest construct of quality of life. The ecological model from Moore et al. (1980)
placed the individuals equilibrium-disequilibrium processes in an environmental
context and included important interactive relationships among that process, the
environment, and the individuals behavioral responses.
None of the models described here gives a complete definition of health, but
taken together they provide an understanding that health is a dynamic, context-
specific process involving individual and environmental physical, social, and
psychological characteristics, challenges, and resources.
Possible Determinants of Health
The preceding attempt to define health established a framework for
exploration into specific factors that contribute to health and healthy aging. The
following subsections present a variety of such factors, drawn from the literature. It is
not intended to be all-inclusive or restrictive but to give structure to the exploration.
28


Sociodemographic Factors
Agegetting olderis the most obvious factor that might affect health in
older people. Other possible relevant sociodemographic characteristics include
gender, race or ethnicity, and socioeconomic status (for which education and income
often serve as proxy measures). Marital status and living arrangements too may
reflect socioeconomic status and may also provide information about social support.
A later section in this chapter reviews previous studies and provides more complete
references to these factors.
Physical and Functional Status
Physical status refers primarily to biomedical symptoms and signs that
indicate the presence or absence of acute illness, chronic disease, and any other
underlying physiological conditions. Although not often cited in studies of the risks
associated with aging (again, a later section provides more complete information),
pain too may be an important factor.
Functional status has a number of componentsphysical capability, the ability
to care for oneself, and cognitive and affective status. Physical capability includes
performance (e.g., the ability to climb stairs) and mobility. Both basic (e.g., bathing,
eating) and instrumental (e.g., cooking, taking medication) skills determine how
independently one can live. Dementia and depression, as well as other cognitive and
affective disabilities, have substantial impacts on functional status.
Quantitative assessment of specific functional limitations may not, however,
give a complete picture of effects on well-being, or more global perceptions of health.
As Johnson and Wolinsky (1993) observed, these distinctions become apparent when
identically disabled individuals demonstrate differential adaptation when confronted
with the functional demands of mobility, personal hygiene, and housekeeping (p.
29


109). As one way of assessing ability, their study asked respondents to compare their
activity levels with those of their peers. Berg, Hallauer, and Berk (1976) found that
asking people to place values on various abilities and activities provided important
information about perceived quality of life. They asked respondents to select from a
list the functions they valued most or would feel worst about not having. Examples
included abilities to do the following:
walk, see, and read
have friends, maintain contact with family and friends
love and be loved
think clearly, use mental abilities
not be depressed
make choices (in day-to-day activities)
participate in particular activities
control body functions
be free of pain
On average, respondents most valued the abilities to use mental abilities, see, think
clearly, love and be loved, make decisions, live at home, walk, maintain contact with
family and friends, and talk.
Social and Temporal Comparisons of Status
As Patrick et al. (1973) pointed out, perceptions of health status involve
expected transitions as well as levels at a point in time. Perceived status may then be
comparative rather than absolute.
Social Comparisons and Normalcy. Mechanic (1978) raised the issue that
people evaluate their health status, in part at least, by comparing it with normal, a
construct whose definition depends on the social group. They may define normalcy in
positivist terms with the ideal as a point of reference, in statistical terms as an
30


observation of what others do, or with a social-reaction approach, expressing a social
reaction to difference and social identity. Suls, Marco, and Tobin (1991) discovered
that older people who mentioned others worse off than themselves tended to rate then-
own health more positively. They proposed that the elderly compare themselves
with a cognitively constructed stereotypical standard of the frail elderly rather than
with a specific other and because few actually fit this stereotype, most elderly feel
they are doing well (p. 1125).
Temporal Comparisons. Suls et al.s (1991) study also found that those who
mentioned thinking about past or anticipated health gave more negative health ratings.
Kind and Dolan (1995) explored the influence of past and/or current illness
experience on the evaluation of health status. They found no difference between past
experience of ill health and of better health, but they discovered that those in currently
poor health (and valuing their health as low) selected a more limited range of values
than those in better health and less often chose better status categories.
Challenges
All sorts of physiological, psychological, social, and environmental factors can
challenge health. The literature suggests that the nature of the issues, the way
individuals respond to them, and the resources they have available to deal with them
all affect the extent to which people feel healthy.
Physical. Cognitive, and Functional Factors. Disease, dysfunction, disability,
and painthe signs and symptoms of physiological disorder and functional deficit
threaten well-being. Schulz and Williamson (1993) pointed out that the nature of the
onset of illness, the perceived prognosis, and the visibility of the disability may also
affect perceptions of health. Depression and other forms of mental illness both
challenge health and make it more difficult to respond effectively. Culture and
socialization within a culture affect responses to these challenges. Zborowski (1952)
31


found that Jewish and Italian patients in a New York hospital responded to pain
emotionally, old Americans appeared stoical, and Irish patients tended to deny the
pain. Kleinmans (1988) illness narratives told similar stories about different cultural
responses to psychiatric disorder.
Childhood. Social background, or social history, may help to determine how
people perceive and evaluate their health and may create threats to well-being or
provide resources to support it. Mechanic (1972) proposed that the responses given to
illness and symptoms during childhood affect later interpretations. Barker (1994)
elaborated that people tend to retain the understanding of illness and treatment that
were prevalent when they were young, because beliefs, practices, and behaviors of
individuals around health and illness depend on the groups fundamental values and
worldview (p. 10). Older people, when their values conflict with medical values
because of the nature of their more prevalent chronic illness patterns, may suffer when
they desire palliation and support more than clinical medicine.
Worries. Health Troubles, and Control Over Health. Murray et al. (1982)
found that preoccupation with, and expressed concern about health were consistently
related to a poor self-rating [of health] (p. 376). Rakowski and Cyran (1990) also
mentioned worry due to health, as did Wolinsky and Johnson (1991). Fillenbaum
(1979) suggested the influence of health troubles that get in the way of what people
want to do. The feeling of having control over ones health, now or in the future, may
provide a helpful response to worries about health (Mechanic, 1972; Rakowski &
Cyran; Wolinsky & Johnson).
Isolation. Isolation leads to loneliness, loss of personal integrity, and the loss
of connection with other social resources. It can be social or emotional. Quantity of
daily contacts, marital status, living arrangement, and number of companions and
confidants determine social isolation. Emotional isolation concerns the quality of
relationshipsit is possible to be lonely in a crowd. Chappell and Badger (1989)
32


found that emotional isolation is more important to well-being than social and that the
lack of a confidant or companion is more important to well-being than unmarried
status or living alone.
Environment. As discussed earlier, critical theorists identified social-
psychological, political-economic, and historical-cultural influences as components of
the environment that affect health. Specifically, they named peace, shelter,
education, food, income, a stable ecosystem, social justice and equity (Labonte, as
cited in Eisen, 1994, p.237) as well as knowledge, money, power, prestige, and social
connections (Link and Phelan 1996).
Awareness and Focus. The tendency to make comparisons indicates an
awareness of self or condition, which may have positive or negative effects. Others
have used the term appraisal to mean much the same thing (Quayhagen &
Quayhagen, 1996). Mechanic (1978) considered social comparison to be one aspect
of adaptation. All of them dependent on the social system, the other aspects include
the search for meaning, attribution of stress or change to a specific cause, level of
dependence, and power. In a later work he explained:
The fact that people cope much of the time without awareness is a
central point in understanding personal and social adaptation. To
become aware, to become self-conscious, is an indication of more than
a routine problem, a greater challenge, a break in the flow of normal
activity___But it is not psychologically economical to worry about
what one cannot predict or control, and individuals maintain a sense of
invulnerability by inattention to potential threat (Mechanic, 1986, p. 5)
Mechanic believed that introspection or awareness, which is conditioned by
sociocultural factors and childhood socialization, increases prevalence of symptoms
and negative self-evaluations, causes more distress and greater upset from stressful
life events, generates increased health care utilization, and appears to exaggerate the
experience of distress and illness.
I
33


Personal Resources and Responses
Adaptation. Cobb (1976) explained the difference between adaptation and
coping. Adaptation refers to individuals abilities to change themselves to improve
the person-environment fit Coping, on the other hand, depends on managing the
environment rather than oneself.
Mechanic (1978) described social adaptations to illness as a coping process:
Illness, illness behavior, and reactions to the ill are aspects of an
adaptive social process in which participants are often actively striving
to meet their social roles and responsibilities, to control their
environment, and to make their everyday circumstances less uncertain
and, therefore, more tolerable and predictable, (p. 1)
He continued further that health and disease are not entities but processes of
adaptation to lifes changing demands and the changing meanings people give to life.
Adaptive resources include material resources, appropriate skills, adequate defenses,
social supports, and sustained motivation.
From a critical perspective, Dlich (1976) stated that in part at least, the health
of a population depends on the way in which political actions condition the milieu and
create those circumstances that favor self-reliance, autonomy, and dignity for all,
particularly the weaker, and he continued, in consequence, health levels will be at
their optimum when the environment brings out autonomous personal, responsible
coping ability (p 7). Fries (1980) observed that the older person requires
opportunity for expression and experience and autonomy and accomplishment [rather
than] support and care and feeding and empathy (p. 135).
Attitude. Borawski, Kinney, and Kahana (1996), exploring congruence
between health appraisal and so-called objective measures of health status, identified
four attitudinal categories. Good-health realists and poor-health realists made
appraisals that were congruent with observed clinical characteristics. Health
pessimists rated their status lower than indicated by clinical observations; health
34


optimists rated theirs higher. When asked why they rated their health as they did, a
substantial number of respondents (ranging from 12 percent of the poor-health realists
to 30 percent of the health optimists) referred to positive-negative attitudes,
optimistic-pessimistic outlooks, and lifestyle behaviors.
Control. Explanatory Style, and Self-EfScacv. People with an internal locus
of control perceive events as contingent upon their own behavior or personal
characteristics. According to Rodin, Timko, and Harris (1985), they feel that they
have the freedom to choose opportunities or at least have an awareness of them,
enabling them to select their preferred goals and means (personal competence). They
accept responsibility for outcomes that come from their own efforts (action-outcome
contingency). Rakowski and Cyran (1990) asserted that perceived internal locus of
control and self-efficacy, as components of perceived health, provide the stimulus to
act or not act on matters related to health. Those with an external locus of control, on
the other hand, perceive events as resulting from chance, the control of powerful
others, or unpredictable fate. Their attribution of ills to non-modifiable sources leads
to poorer outcomes and less impetus to make changes.
The level of perceived internal control may decrease with age. Wolinsky and
Stump (1996) found a strong negative linear relationship between age and the sense of
control. Rodin (1986) suggested that aging increases the risk of losing control
because of the increased number and intensity of significant negative experiences
(e.g., loss of spouse, retirement, fear of institutionalization), deterioration of
functional and physical health, and learned dependence with respect to health
professionals.
One way of relinquishing control may be to lower ones aspiration levels,
helping older individuals [to] come to terms with various aches and pains better than
younger individuals (p. 270), according to Tomstam (1975). He further pointed out
that such a response is, of course, a mercy, but it also involves an ethical dilemma
35


[concerning the] extent [to which it is] defensible to make aging individuals contented
by means of decreasing their different aspiration levels (p. 270).
Peterson, Seligman, and Vaillant (1988) proposed and have continued to study
the hypothesis that a pessimistic explanatory style, which corresponds substantially
with an internal locus of control, increases the risk of physical illness. By explanatory
style they mean the habitual way in which people explain the bad events that befall
them (p. 23). Helplessness theory, on which they based their hypothesis,
distinguished three dimensions of explanatory style: intemality (e.g., Its me vs.
The road was icy), stability (e.g., Its the way the world is vs. It was a one-time
spill), and globality (e.g., It will affect everything I do vs. It made me late to work
that day). According to Peterson et al., the individual who explains bad events
pessimistically, with stable, global, and internal causes shows more severe
helplessness deficits than a person who explains them with unstable, specific, and
external causes (pp. 23-24). Those deficits include passivity, pessimism, and low
morale, which foreshadow disease and death (Peterson & Seligman, 1987, p. 237).
Perceived self-efficacy may serve as a counter to helplessness. Bandura
(1977) defined self-efficacy as the conviction that one can successfully execute the
behavior required to produce the [desired] outcomes (p. 193). Self-efficacy,
according to a later work, not only affects willingness to initiate coping behavior but
also determines the amount and duration of effort that people will expend in the face
of obstacles or aversive experiences. Perceived self-efficacy alone, however, cannot
prevent feelings of futility and helplessness. People may feel fully competent of their
capabilities but give up trying because they expect their efforts to produce no results
due to the unresponsiveness, negative bias, or punitiveness of the environment
(Bandura, 1982, p. 140).
Coping. Schulz and Williamson (1993), in a study of the impact of physical
frailty on patient and caregiver, constructed a conceptual model that has relevance
36


beyond that relationship. After discussing conditions that may generate patients
stress (physical and functional signs and symptoms), they listed conditioning
variables that might intervene between stressors and outcomes. The variables include
economic resources, personality attributes (optimism, perceived control, and self-
esteem), social support, and coping strategies.
Pearlin (1989) explained copings mechanisms Coping, he wrote, serves to
change the situation from which stressors arise, manage the meaning of the situation
in a manner that reduces its threat, or keep the stress symptoms within manageable
bounds.
Independence. Independence incorporates both autonomy, or the ability to set
and follow ones own rules, and the ability to act J.G. Evans (1984) offered an
example related to mobility: being able to choose when and where to go sustains
autonomy, but full independence is constrained if one needs assistance with actually
moving. He suggested that autonomy may be the more useful global objective for
older people.
Sense of Coherence. Antonovsky (1987) wished to understand what predicts
good health outcomes and what characterizes those who have them. To answer his
questions, he introduced the concept of sense of coherence, which he defined as
follows:
a global orientation that expresses the extent to which one has a
pervasive, enduring though dynamic, feeling of confidence that (1) the
stimuli deriving from ones internal and external environments in the
course of living are structured, predictable, and explicable; (2) the
resources are available to one to meet the demands posed by these
stimuli; and (3) these demands are challenges, worthy of investment
and engagement, (p. 19)
Components of a sense of coherence, according to Antonovsky (1987),
include 1) comprehensibility, or the degree to which stimuli make cognitive sense as
information that is ordered, consistent, structured, and clear, rather than as noise (p.
37


17) and will be predictable or at least orderable, explicable, and understandable; 2)
manageability, or the extent to which one feels s/he has adequate resources to meet
the demands of stimuli, as opposed to feeling burdened and overwhelmed or, at the
other extreme, under-challenged; and 3) meaningfulness, or significance., or the sense
that the challenges one faces are worthy of emotional commitment and investment.
To explore sense of coherence, Antonovsky (1987) proposed looking at good
deviant cases, individuals who were exposed to risk but suffered no ill effects and
who, as outliers, can provide information about their exceptional status. He felt that
those who acquire what he called generalized resistance resourcesincluding
autonomy, competence, money, ego strength, cultural stability, and social supports
can maintain coherence in the face of maladaptive social-relational processes
(Antonovsky, 1979). Motzer and Stewart (1996) added these factors first proposed by
Flanagan (1982) and Burckhardt, Woods, Schultz, and Ziebarth (1989): physical and
material well-being; relations with other people; social, community, and civic
activities; personal development and fulfillment; recreation; and independence.
Social Support Resources
Much has been written about social support and its importance as a resource
for dealing with lifes challenges. House, Umberson, and Landis (1988), in their
extensive review, defined social support as an element of social relationships that
refers to positive aspects such as emotional caring and concern, instrumental
(tangible) aid, and information. Its sources can be informal, from friends and family,
or formal, from institutions and commercial services. Antonucci and Akiyama (1987)
described six beneficial results of support: a confidant relationship, reassurance,
respect, provision of care when ill, an outlet for talk when upset, and a source of
health information.
I
38


M. J. Stewart (1989) outlined five theoretical perspectives concerning why
social support has beneficial effects:
Attribution theory: Individuals formulate attributions (often by assigning
blame) in order to understand, predict, and control the environment.
Fairness mattersrecipients deserve help to the extent that they did not
cause their need, and donors feel obligated to the extent that they are not
responsible for their surplus of resources.
Coping theory: Support affects individuals abilities to undertake ever-
changing cognitive and behavioral efforts to manage demands that exceed
their resources. Sources of social support provide information about
events and stressors; broaden the number of options, resources, strategies,
and referrals; provide norms; inhibit maladaptive responses, and provide
tangible aid and emotional sustenance.
Social-exchange (or equity) theory: Individuals desire to maintain equity
in their exchanges. They attempt to maximize their outcomes, but they
feel distress when in inequitable relationships and will try to restore
equity.
Social-comparison theory (see also above): People tend to evaluate
themselves and to gather information about their behavior, characteristics,
opinions, and abilities by comparing themselves with others. Social
support assists the process by supporting appraisal, affirming, supporting
esteem, and providing feedback.
Loneliness theory (see also isolation as a challenge above): Lack of
desired social support leads to a subjective, unpleasant experience derived
from a perceived deficiency in social relationships or relational provisions.
House et al. (1988) concluded from the studies they reviewed that social
support promotes health by fostering a sense of meaning or coherence, facilitating
health-promoting behaviors, providing motivation and emotional support, and/or
affecting neuroendocrinal responses. Newman, Struyk, Wright, and Rice (1990),
studying the role of social support in determining institutionalization, summarized
five hypotheses about haw social support operates:
Main effects hypothesis: Informal or formal support reduces existing
risks, with an independent or main effect on risk net of background, health
or other relevant factors.
39


Buffering effects hypothesis: Ones support system moderates or buffers
the effects of risk factors by improving coping abilities. Social
relationships affect well-being only in the presence of stress or challenge.
Supplementation hypothesis: Formal support permits informal systems to
function better and longer.
Accommodating environment hypothesis: Situational variables (in his
example, housing characteristics) modify the willingness of the informal
support systems members to provide care.
Mediation hypothesis: Informal and formal support intervene between the
challenge (deteriorating health, in his example) and negative outcomes
(e.g., institutionalization).
Social support comes from the members of ones social network. Pearlin
(1989) explained that while the social network can be regarded as the totality of the
social resources on which one potentially may draw, social support represents the
resources that one actually uses in dealing with life problems (p. 251). House et al.
(1988) suggested three classes of social support variables. The first two are elements
of the social relationship structure and the third concerns function. The first, social
integration, refers to the quantity of relationships, their type and frequency. Social
network structure, the second, refers to the properties that characterize the set of
relationships. The third class of variables, relational content, refers to the functional
nature of relationships, processes that include emotional and instrumental social
support, relational demands and conflicts, and social regulation or control.
S. Cohen (1988), extracting from network theory, listed characteristics
associated with social network structure: size, density (which gives a measure of the
richness and complexity of interrelationships within the network), multiplexity (the
number of available alternatives), reciprocity, durability, intensity, frequency,
dispersion, and homogeneity. B.H. Kaplan, Cassel, and Gore (1977) identified
anchorage (the length of the path to others) and reachability as additional attributes.
Kaplan et al. also described the interactional properties of support: direction
40


(reciprocity), intensity, frequency, and content or meanings. Content/meanings
includes rituals, values and beliefs, shared norms, interpersonal exchanges, a fit
between role and dependency, opportunities for closeness, self support, the
disposition of negative effects, and social status.
Functional types of social support are generally classified as emotional,
tangible, and informational. Emotional support provides the sense of being cared for
and loved, esteemed and valued. Tangible support offers goods and services, while
informational support shares information. All types are part of a network of mutual
obligation that both defines and reflects ones place in society (Cobb, 1976). S.
Cohen (1988) described four types of support: information-based assistance, identity
and self-esteem, social influence, and tangible resource. Information-based support
provides instrumental information, or at least the knowledge that information will be
available if needed. Identity-and-self-esteem support increases feelings of personal
control. Social-influence support emphasizes social norms and normative coping
behaviors. Tangible resources refer to actual services provided or made available.
According to social exchange theory, reciprocity matters (Flaherty &
Richman, 1989). Individuals desire to maintain equity in their exchanges, attempting
to maximize their own outcomes, but they feel distress in inequitable relationships
and will attempt to restore equity (M. J. Stewart, 1989). Reciprocity can be between
two individuals and contemporaneous, between two individuals over a period of time,
or more communal and over time (e.g., person A helps person B, who assists person
B, who at a later time does something that benefits person A). An equitable exchange
balance promotes satisfaction (N. Krause, 1995).
Although social support appears to benefit well-being, not all social support
generates positive effects, and more is not always better. Studies have indicated that
to be beneficial, support must be autonomy-enhancing (Rowe & Kahn, 1987), fair,
and reciprocated (M. J. Stewart, 1989). N. Krause (1995) discovered that averting
41


negative interaction had a greater positive effect on satisfaction that did the amount of
support In an earlier study, he found that support increased feelings of control up to a
threshold level, after which increased support could lead to dependence,
entanglement and decreased feelings of control. Cobb (1976) also cautioned that
goods and services may foster dependency but suggested that informational support
might not.
Health Care Delivery
How health care is deliveredquantity, style, and settingalso may affect
health, as either challenge or resource. A large number of research projects have
studied specialized hospital-based programs, often called geriatric evaluation and
management programs. (For an introduction to this very extensive literature, see the
review by Applegate, Deyo, Kramer, & Meehan, 1991, and the meta-analysis by
Stuck, Siu, Wieland, Adams, & Rubenstein, 1993.) A search of the literature,
however, discovered very little concerning outpatient programs.
Societal and Environmental Challenges Specific to Aging
Bengtson, Burgess, and Parrott (1997) described two theories of social
gerontology that give structure to the following discussion of environment and
socialization. The macro-level political economy theory of aging explores how the
interaction of economic and political forces, which constitutes the environment of
aging, determines allocation of social resources (e.g., variations in treatment, status of
older people). Its emphasis on the effects of social structure and social power may
mask the impact of aging on the individual. Social constructionism, on the other
hand, examines the social construction of age and aging, exploring the influence of
social definitions and structures on individual processes of aging. It takes a micro-
42


level view of individual agency and social behavior, specifically attitudes toward
aging, within the larger structure of society. Bengtson et al. suggested that political
economy of aging can... be linked with social constructionist perspectives to point
to the ways in which structural forces manage and control the social construction of
aging and how old age is experienced (p. S83).
Especially for older people, who are prone to multiple health problems that
include psychological and social as well as physical dimensions, the perception of
well-being depends on more than just clinical and functional status (for example,
Mechanic, 1995; Rickelman, Gallman, & Parra, 1994; Schoenfeld, Malmrose, Blazer,
Gold, & Seeman, 1994). Farquhar (1995b) listed the dimensions of quality of life
most frequently mentioned by older people: family, social contacts, (physical) health,
mobility/ability, material circumstances, activities, happiness, youthfulness, and home
environment Wan (1986) proposed seven determinants of health particularly relevant
to older people: physiological condition, presence of degenerative illness, mental
status, age, sex, functional dependencies (e.g., problems with walking, eating, and
toileting), and strength of social support network. Degenerative chronic illness,
decreased mental functioning, and limitations to activity have more of an impact on
the health of the elderly, compared to younger populations, than a less robust
physiological condition, which many consider to be a natural and therefore less
threatening part of the aging process. (Younger populations likely perceive a greater
impact from physical deficits because they expect physical vigor.) Active life can be
constrained not only by physical or mental limitation, but also by cultural
expectations, environmental obstacles, or lack of motivation and training (Boult et
al., 1994, p. M28). Older people may also perceive that their care is less intense than
that offered younger people because society, and often they themselves, devalue their
lack of productivity and consider the expense of care too great for the expected cure
or relief.
I
43


J. G. Evans (1984) spoke of aggravated aging, the increase in morbidity due
to social/environmental hazards/challenges (e.g., poorer housing, insufficient heating,
diminished access to geriatric services). The inclusion of environment as a relevant
determinant of health suggests its importance but does not describe its breadth or
level of importance (Wan, 1986). Threats to well-being, according to Pearlin (1989),
largely arise from and are influenced by various structural arrangements in which
individuals are imbedded (p. 241). These arrangements correspond to the various
environmentsphysical, political, and socialin the models described earlier.
Pearlin suggested that individual experiences derive from interrelated levels of social
structuresocial stratification, social institutions, and interpersonal relationships
and he specified age as one of the determinants of social strata.
Parsons (1951) pointed out that societys expectations and norms concerning
individuals roles and status have a substantial effect on health. Health in this context
refers to individuals capabilities to perform adequately the roles and tasks for which
they have been socialized, based on their societies norms (Liang, 1986). The role of
significant others (Kasls, 1983, neutral label intended to encompass the sometimes
controversial topics of social networks, social support, and social isolation), belief in
ones ability to perform adequately, and congruence between beliefs and societal
expectations are then important contributors to health, along with clinical
interventions to support and maintain physical and functional capabilities.
Rosow (1974) explored socialization to old age (in our culture), with specific
reference to negative changes in relationships with society and social institutions.
Socialization serves to inculcate both values and behavior in members of a society in
order to encourage conformity with social norms. For older people, Rosow
suggested, our society has not defined social norms or expectations. Aging members
of the society do not have clearly defined roles, rites of passage, positive goals, or
44


markers of successful performance. Old age, then, has become a separate subculture,
surrounded by a different, less supportive environment
Mechanic (1978) concurred, describing the social position of the aging as a
socially caused negative experience, characterized by isolation, lack of social support
health decrements, lack of respectable identity, and loss of work involvement. Social
maladaptation, according to Lewis (1953), occurs when a persons own version of
his social role is not in conformity with societys version: and in so far as each person
has many social roles, it is the dominant role, or those which have precedence in the
daily organization of his activities that are important (p. 116). He cautioned further
that involuntary changes of social status will clearly favour conflicts over a mans
own conception of his social role and that which society fastens upon him (p. 116),
leading to non-conforming behavior that can but may not lead to illness, which he
thought depends to a much greater extent on physiological and psychological
dysfunction than on social adaptation. Manton (1989) warned of a greater impact:
Given the dimensions of the problem [of an increasingly large disabled
elderly population], it is likely that no response will be satisfactory
unless fundamental changes in the sociocultural perception of the
functioning of elderly people, and the provision of family and other
social resources to maintaining that functioning, are developed, (p. 55)
Investigating the substantial variation in longevity and rates of decline among
older people, Berkman (1988) expressed the need to identify some behavioral and
socioenvironmental conditions that appear to influence the way and rate at which
people age and die (p. 39). She suggested that social status among older people may
be determined more by relative standing among peers than by income, education, or
occupation, which may decrease in importance as indicators.
45


Risks and Predictors of fUrilhealthv Aging: Previous Studies
Describing healthy aging and developing a model of health for an older
population require an understanding of the relative importance of the relevant factors,
including those suggested in the preceding subsections. The literature concerning
risks and predictors of (unhealthy aging is dauntingly large. It includes studies of
predictors and correlates of mortality, institutionalization, and maintained or
decreased levels of function. The studies reported in the literature varied widely in
characteristics of the sample (e.g., sample size, frail vs. unimpaired, of all adult ages
vs. elderly only, single gender or not, ethnically diverse or not, with various levels of
income or not, institutionalized or community-dwelling), study design (e.g., cross-
sectional, longitudinal, prospective, retrospective), and analytic methods (e.g.,
bivariate, multivariate, linear regression, hazard analysis, logistic regression,
correlation). Making comparisons among them or generalizations from them even
more difficult, the studies defined constructs of health differently, chose a wide
variety of indicators to measure them, and included (or excluded) different
combinations of them. Most models failed to account for a substantial amount of
observed variation. Some factors did, however, repeatedly show significant
predictive or correlational associations with mortality, institutionalization, function,
and perceived health. The following review identifies those factors while noting
categories of variables excluded from the various studies.1
Mortality
Almost all of the 52 studies of mortality reviewed here adjusted for older age
and male gender. Poorer clinical condition and comorbidities were the next most
1 Citations in the text refer only to a limited number of studies that exemplify the issues and variables.
Appendix B contains extensive tables of the reported studies, displaying more complete information
about research designs and variables.
46


commonly cited predictors or correlates of mortality (Deeg, van Zonneveld, van der
Maas, & Habbema, 1989; Wolinsky, Johnson, & Stump, 1995). Decreased functional
statususually measured as Activities of Daily Living (ADL), Instrumental Activities
of Daily Living (IADL), and/or physical performance (Manton, 1988; Mor, Wilcox,
Rakowski, & Hiris, 1994; Reuben, Siu, & Kimpau, 1992)and cognitive status
(Narain et al., 1988) also appeared often. Other variables predictive of mortality
included obesity (measured as body mass index, BMI; Idler & Kasl, 1991), larger
number of medications (G. Kaplan, Bareli, & Lusky, 1988), and health behaviors such
as smoking, alcohol consumption, and little exercise (GA. Kaplan, Seeman, Cohen,
Knudsen, & Guralnik, 1987). Individuals ratings of their health, or perceived health,
contributed significantly to 22 of the 38 models that included them (Idler & Kasl,
1991; Mossey & Shapiro, 1982; Wolinsky & Johnson, 1992). Informal social support
appeared to protect against mortality: 24 of the 38 studies that included at least one
measure other than marital status found a significant effect (Berkman & Syme, 1979;
Blazer, 1982). Most of the studies included clinical measures of health, but many
failed to include any measure of functioning level, depression or other measure of
psychological status, social support other than marital status, health behaviors, or
socioeconomic status.
Institutionalization
Admission to a nursing home, especially for longer periods of time, generally
signals increased frailty. The 46 published studies reported here found that increased
age (Branch & Jette, 1982; Shapiro & Tate, 1985) and decreased functional and/or
physical abilities (M. A. Cohen, Tell, & Wallack, 1986; Reuben et al., 1992)
predicted or correlated with institutionalization, as did clinical indicators such as
stroke (Teresi et al., 1989). Cognitive impairment (Coughlin, McBride, & Liu, 1990;
Shapiro & Tate, 1988) and other forms of mental impairment (Branch & Jette, 1982)
47


contributed significantly to the risk of nursing home admission. Being White (Greene
& Ondrich, 1990; Liu, Coughlin, & McBride, 1991) and female (Shapiro & Tate,
1988) increased overall risk, but being male correlated with earlier (younger)
admission (Liu, Coughlin, & McBride, 1991). Only 14 of these studiesfewer than
the mortality studiesincluded perceived health as a variable, but nine of them found
it a significant predictor (M. A. Cohen et al., 1986; Steinbach, 1992). Social and
socioeconomic factors appeared to contribute significantly to the risk of
institutional ization. Living alone (Boaz & Muller, 1994; Wolinsky et al., 1992), not
owning ones home (Coughlin et al., 1990; Greene & Ondrich, 1990), lower SES (Liu
et al., 1991), feeling lack of control (Wolinsky et al., 1992), and receiving formal
support such as home care (Boaz & Muller, 1994) predicted institutionalization;
informal support (including having a living spouse) protected against it (M. A. Cohen
et al., 1986; Pearlman & Crown, 1992). Many of the studies failed to include the
following categories of variables: SES, clinical information other than diagnosis,
physical performance and mobility (only 12 studies included any indicator), and
depression (only three studies included this variable; of them, two found it
significant).
Function
Decreased functional abilities also suggest increased frailty. Common
measures of function include ADLs such as toileting, bathing, dressing, and walking,
and IADLs concerning housework, meal preparation, managing money, and eating. In
most of the 29 published studies reviewed here, clinically assessed problems related
to specific conditions (Boult et al., 1994; Idler & Kasl, 1995) and BMI (Pinsky et al.,
1985) contributed significantly to the risk of functional decline. Age also had a strong
association with function (Mor, Wilcox, et al., 1994; Roos & Havens, 1991).
Hoeyman, Feskens, van den Bos, and Kromhout (1997) determined that decreased
I
48


functional status resulted from population aging, not from cohort-specific trends.
Female gender (Roos & Havens, 1991), non-White ethnicity (House et al., 1994), and
lower SES (House et al., 1994) correlated with or predicted decline, or, in the rare
cases of studies of positive outcomes, supported continued physical ability (Guralnik
& Kaplan, 1989; Harris, Kovar, Suzman, Kleinman, & Feldman, 1989). Fewer than
Half of the studies included perceived health, but in seven of the 11 that did, it
correlated significantly and positively with increased function (Idler & Kasl, 1995;
Mor, Wilcox, et al., 1994). A few studies suggested an association between health
behavior and function (Guralnik & Kaplan, 1989); others found relationships between
function and psychological well-being (Berkman et al., 1993) and social role status
(A. L. Stewart et al., 1989). Informal support associated negatively with decline
(Boult et al., 1994); that is, informal support appeared to protect against decline.
Only 14 studies included psychological factors; few mentioned social measures other
than marital status.
Three studies assessed change in functional ability over time. These studies
reported that baseline status, age, perceived health, clinical condition, and/or
psychological status affected changes in functional status (Crimmins & Sito, 1993; G.
A. Kaplan, Strawbridge, Camacho, & Cohen, 1993; Mor, Wilcox, et al., 1994).
In summary, the literature consistently identified age, functional status,
clinical condition, and cognitive status as factors related to the risks of, generally,
unhealthy outcomes. Perceptions of health, when included in the models, often
associated significantly with the outcomes. The studies offered a less clear picture of
the importance of social or psychological factors other than cognitive function. In
great part that inconsistency derived from the paucity of variables included and the
variability in measures used, if the topics were considered at all. Because most
reported studies focused primarily on risks to health, they can only suggest, in
negative form, what may contribute positively to healthy aging.
49


Perceived Health
Having established that many factors may affect health and healthy aging, this
study required a means of identifying important ones. Moriyama (1968) listed the
following characteristics of good measures, which should:
be meaningful and understandable;
be sensitive;
be clear, justifiable and reasonable in assumptions;
be composed of clearly defined components;
consist of independent parts, contribute independently to total variance;
and
use available or obtainable data.
Health care measures are commonly classified as indicators of structure,
process, and outcome (Donabedian, 1988). Structure encompasses attributes of the
setting, components that are relatively easy to measure but not particularly well
correlated with care (Schwartz & Lurie, 1990). Process describes actual activities and
procedures. Service performed well does not necessarily guarantee effectiveness; the
evaluation of process therefore may not provide information about the significance of
inputs. Outcomes represent results (Lohr, 1988). Positive outcomes do not
necessarily improve healthpositive does not always equate with appropriatebut
they do serve as indicators of significance. Brook et al. (1977) asserted that because
outcomes assessment is based directly on measures of health status, it has more
validity than process assessment, which looks only at the quality of the services
delivered.
Outcomes assessment can focus narrowly or broadly. Measures based on the
biomedical model focus on the provision of care rather than on improvements in
health status. Outcomes of this sort include mortality, hospital readmissions, length
of stay, standard clinical measurements (e.g., blood pressure and laboratory test
50


results), the efficacy of a new drug, improvement in (or loss of) mobility after hip
replacement, and remission from cancer after radiation therapy. These have particular
relevance to the treatment of acute illness and conditions, and they serve a useful
purpose in that environment. They also appear to have many technical advantages
(e.g., hard clinical data, well-defined end points, sensitive and specific tests,
statistical validity and reliability), but they address only limited aspects of health.
They have shortcomings even as measures of the effectiveness of specialized
scientific medicine because of the shift from acute to chronic disease, the aging of the
population, the increasing prevalence of diseases with a social component and multi-
factorial etiologies, and a growing criticism of the business of health (Turner, 1992).
They ignore important factors such as the presence or absence of increased or
decreased pain, the ability to return to work, and change in mental state.
At the other end of the health outcomes continuum, global well-being, often
referred to as quality of life, encompasses the broadest concepts of health. It
incorporates everything from genetic composition to political and economic
environmental influences on life. That strength is also its weakness as an outcome
indicatorhealth care intervention cannot directly affect many of its components in
measurable ways.
Perceived health, or self-rated health, is the outcome measure selected for this
study. It combines the strengths of both more clinical and more general indicators
and, of course, shares some of their weaknesses. Like many clinical indicators, it is
reasonably unambiguous and straightforward to measure. Researchers frequently
assess perceived health with a single simple question. Most such questions, including
the following, request responses of excellent/very good/good/fair/poor or a similar
range:
How would you rate your overall health? (S. H. Kaplan, 1987)
How would you rate your health at the present time? (Idler & Kasl, 1995)
51


Compared to others your age, would you say your health is excellent, very
good, good, fair, or poor? (Greiner, Snowdon, & Greiner, 1996)
For your age would you say, in general, your health is excellent, good, fair,
poor, or bad? (Mossey & Shapiro, 1982)
Do you consider yourself a healthy, fairly healthy, sick, or very sick
person? (G. Kaplan et al., 1988)
A richer measure than clinical indicators, ones subjective sense of physical
health and well-being does not stand in a fixed, one-to-one relationship with objective
clinical assessments of ones medical status____In part, this discrepancy reflects the
fact that most patients experience their health as a global experience and a level of
function, as an overall state of well being (Barsky et al., 1992, p. 1147) that
incorporates medical status, functional impairment, psychological dysphoria and
emotional distress, social factors, role demands, stressful life events, and age.
According to Liang (1986), perceived health provides a global evaluation of an
individuals health status, incorporating comprehensive domains of health and serving
as a summary of an individuals perceptions of various objective and subjective
aspects of health. Although global in scope, perceived health as a concept and
measure remains personal enough to avoid the separation between individuals and
their health (p. 299) that Robertson and Minkler (1994) worried would make health a
theoretical objective rather than an everyday reality amenable to intervention.
As the section on previous studies of risks of aging indicated, many
researchers have reported that perceived health, in turn, predicts other, more distal
outcomes such as mortality and institutionalization. Because it predicts other
outcomes so well, Johnson and Wolinsky (1993) pointed out the importance of
finding out how it is actually related to other dimensions of health status (p. 109).
Perceived health may have other attributes as wellRakowski and Cyran (1990)
found that it not only serves as a static indicator but can provide the stimulus to act or
not act on matters related to health. Ware (1976) found self-evaluation a valid
52


measure of health status. It meets Moriyamas (1968) criteria to be meaningful and
understandable, as well as clear, justifiable, and reasonable in assumptions.
For older people, subjectively assessed health may be an especially important
measure because older people are prone to multiple health problems that include
psychological and social aspects as well as physical components (Schoenfeld et ai.,
1994).2 inn and Linn (1980) observed that how the elderly view their own health
may be an extremely useful clinical guide as to their overall health status (p. 311).
Studies have shown that perceived health rated by older individuals correlates with or
is predicted by more objective health indicators (Johnson & Wolinsky, 1993; Wan,
1976), physical exam or physician rating (LaRue, Bank, Jarvik, & Hetland, 1979;
Maddox, 1962), chronic conditions (Jylha, Leskinen, Alanen, Leskinen, & Heikkinen,
1986), functional status (Barsky et al., 1992; Johnson & Wolinsky, 1993), depression
and other psychological characteristics (Blazer & Houpt, 1979; Tissue, 1972), and
SES or economic condition (Markides & Lee, 1990; Wan, 1976). Linn and Linn
(1980) determined that it provides a better marker than age. Older persons (65 years
old and older) tend to perceive their own health significantly more positively than
younger adults (Cockerham, Sharp, & Wilcox, 1983). Older-old persons (75 years
old and older), who report more health-related problems than the younger old, tend to
rate their health even more positively (Ferraro, 1980). These individuals appear to
expect and discount some level of decreasing function, or healthy decline, when
assessing their health.
Hays and Stewart (1990) identified the following as components of self-
reported health in chronic disease patients:
physical health
2 As in the earlier section on risks associated with (un)heaithy aging, citations in the text represent only
a limited number of studies, ones that exemplify the issues and variables. Appendix B contains
extensive tables of the reported studies, displaying more complete information about research designs
and variables.
53


role limitations due to physical health
physical functioning
satisfaction with physical ability
mobility
mental health
depression/behavioral emotional control
positive affect
anxiety
feelings of belonging.
No measure, of course, is perfect, and self-reported health has limitations. As
Mechanic (1978) observed,
It is much easier to develop measures of physical incapacity for
specified population groups than it is to devise reliable and valid
measures of overall health status ... which is so involved with
subjective perceptions, social expectations and role demands, and
value judgments that it is extraordinarily difficult to translate it into
any set of empirical measures. A variety of proxy measures are
frequently used to measure health status, such as subjective appraisals
of ones health and the absence of chronic conditions, but these are
only poor approximations of the concepts investigators really wish to
study. (p. 183)
Self-reported information in particular raises concerns about intended or
unintended subjective bias. No assessment, however, can be completely free of bias,
either in the definition of the indicator or in the process of measurement. As Patton
(1990) wrote, all... data are based on someones definition of what to measure and
how to measure it (p. 480). Cultural and technological issues affect even definitions
of life and death, as we see in debates over abortion and the timing of organ harvest
for transplantation. In addition, perceived health may not be an entirely subjective
measure. Kutner (1987) observed that perceived health status is a concept that
includes both an objective and a subjective component (p. 30).
54


Perceived healths global nature, its reliance on a single self-reported
indicator, and the multiple contributors to an individuals self-appraisal may mean
that it does not meet Moriyamas (1968) requirements for clearly defined components
and sensitivity. Hyland (1993) noted that
perceptions of the self are not unitary. Two distinctions are
particularly relevant to the patients perceptions of health. The first is
that knowledge of events and evaluative appraisals of events are
independent and involve different neurological processes. Thus a [sic]
patients knowledge of their health problems are likely to be
independent of evaluations of how much distress those problems
cause. Second, affective evaluations of positivity tend to be
independent of affective evaluations of negativity. That is, a patient
may be both happy and unhappy with life or happy about some aspects
and unhappy about others, (p. 1021)
As a measure of change, however, perceived health appears to be reasonably
stable while capable of detecting change. Goldstein, Siegel, and Boyer (1984)
reported that, over a one-year period, one-third of a study group changed their ratings
by at least one category of four (poor, fair, good, excellent), in both positive and
negative directions; his study included adults of all ages, not just the elderly. Rodin
and McAvay (1992) found similar results. In their study, they incorporated up to
seven measurements of perceived health over three years. Pairwise correlations
between adjacent measurements ranged from 0.7 to 0.8. Comparisons of the most
distal measurements found that 75 percent did not change, 11 percent improved, and
14 percent declined.
Summary
For the purpose of this research into healthy aging, perceived health will serve
as a proxy for health. The literature described in this chapter identified a variety of
55


possible contributing factors that formed the foundation for the analyses whose
reports follow.
Regression modeling had to be limited to the available data from the parent
study, a selection of variables much like those incorporated into the risks of
(un)healthy aging studies described earlier. These data primarily concern age,
functional and physical performance status, clinical condition, and cognitive status,
with a few measures of psychological or social characteristics and health care
utilization.
Interviews with a subset of the study population offered the opportunity to cast
a wider net Semi-structured questions addressed the following subjects: well-being
in general; valued activities and abilities; social and temporal comparisons of health
status; challenges, including childhood experiences, worries, isolation, and internal
focus; personal resources, including sense of control, attitude, and coherence; support
resources; and health care delivery.
56


CHAPTER 3
QUANTITATIVE ANALYSIS
This chapter describes the methods and results of a quantitative analysis that
had two goals, to identify factors that support positive perceived health and to provide
criteria for selecting discrepant cases for subsequent interviews. The analysis
addressed parts of three of the overall studys specific aims:
to construct a statistical model to describe and predict perceived health
from a variety of demographic, clinical, functional, social support, and
utilization variables previously associated primarily with negative
outcomes related to aging (Specific Aim 1);
to identify discrepant cases, those individuals whose reported perceived
health differed substantially from values predicted by the model (Specific
Aim 2); and
to identify factors that affect perceived health, to contribute to a model of
healthy aging for this older population (Specific Aim 4).
Methods
Quantitative Study Design
Data for this study came from a two-year prospective randomized trial of
Kaisers Cooperative Health Care Clinic (CHCC) program, which delivers outpatient
primary care in a group visit format. The quantitative portion of the study reported
here is a cross-sectional snapshot of all of the parent studys subjects, whether or not
recipients of the CHCC intervention, at the end of the first year of that study. Because
baseline explanatory data were available as well, it was possible to analyze both
baseline and 12-month change components of the variables.
57


Study Population and Data Collection
The parent study drew subjects from among community-dwelling pre-frail
Kaiser members 60 years old and older who had a history of chronic conditions and
high utilization, defined as 12 or more contacts with a provider within the 18 months
prior to the study. These members constituted about 16 percent of Kaisers Medicare-
risk population at the beginning of the study (approximately 46,000 members).
Kaisers research team mailed the Health Screening Form (HSF) to Kaiser
members who met the initial age and utilization selection criteria. (Appendix A
contains a copy of the instrument.) The form included questions related to the study
variables (e.g., perception of health, ability to perform activities of daily living,
presence of chronic conditions) along with participation screening questions (e.g., the
likelihood of changing to another health care provider within the study period or the
presence of a condition or serious illness that would make it impossible to attend
meetings or participate in a group). In addition, the questionnaire asked respondents
if they would be interested in group-based health education and patient care clinics, if
such clinics were started. Physicians familiar with the members then assessed their
functional capability to participate in a group-based clinic. Those who returned the
HSF, met selection criteria, and expressed interest in the group concept were
randomized into two arms of the study, experimental (CHCC intervention) and
control (usual care). Subjects entered the study as physicians began groups, between
March 1995 and June 1996, and numbered nearly 400 cases and 400 controls, or
almost 11 percent of those who met the initial selection criteria and 71 percent of the
randomized sample.
As each group ended the first year of the study, subjects again received the
same HSF instrument CHCC intervention subjects completed the form dining their
12-month meetings if they attended them; all others received the instruments by mail.
58


Study staff aggressively followed up with non-completers to ensure the greatest
possible response.
Measures
Outcome (Dependent) Variable: Perceived Health Status at 12 Months. The
outcome (dependent) variable, health status at 12 months, comprised responses to this
question:
Compared to other persons your age, would you say
your health is:
Excellent Fair
Very Good Poor
Good
Although the perceived health status variable was measured on an ordinal scale, the
size of the sample permitted considering it a continuous variable for the purposes of
linear regression.
Explanatory (Independent) Variables. Explanatory variables included clinical
and functional data from the HSF questionnaire and additional information from
administrative databases. For analysis, variables were grouped into conceptual
categories: sociodemographics, chronic conditions, physical function, cognitive and
psychological function, health behaviors (smoking), informal and formal social
support, and health care services utilization. Sociodemographic information included
age, gender, race/ethnicity, education, and (optionally) household income. The
questionnaire asked about the presence or absence of eight health conditions that an
individual might self-identify (e.g., chronic lung disease, trouble seeing, asthma,
arthritis) and of six other health conditions that would more likely be identified by a
physician (e.g., hypertension, kidney disease, cancer). Physical condition variables
included indicators of Activities of Daily Living (ADL; from Katz, Ford, Moskowitz,
59


Jackson, & Jaffee, 1963), Instrumental Activities of Daily Living (IADL; from Duke
University Center for the Study of Aging and Human Development, 1978), physical
performance and mobility (from Rosow & Breslau, 1966; Nagi, 1976), and urinary
incontinence and bowel control. A depression item and a five-part question about
dementia measured cognitive and psychological function. The questionnaire assessed
informal social support by asking about marital status, current work status (including
volunteer activities), living arrangements, and transportation requirements.
Respondents were also asked about receipt of agency-based formal social support
services. Health care services utilization variables included number of medications,
days of hospitalization, outpatient services, skilled nursing care, patient education,
percent attendance in the CHCC group visit program, and a physician identifier. Data
were collected at baseline and at the end of 12 months. Table C.l in Appendix C
contains a complete listing and description of the variables at baseline. Table C.2
similarly reports 12-month change information.
Recoding Based on Distribution of Responses. The HSF generated ordinal or
categorical responses. For some items, the distribution of the responses suggested
recoding into a smaller number of categories. The following figures give examples.
Sum of APIs at Oassino (Q=no aependenaas)
Sum of lAOLs at Basaine (0=no dependencies)
Figure 3.1. Sum of Activities of Daily
Living Dependencies
Figure 3.2. Sum of Instrumental Activities of
Daily Living Dependencies
60
i


As illustrated in Figures 3.1 and 3.2, very
few respondents reported any ADL or
IADL dependencies. All sums greater than
zero therefore were combined into a single
category any, to contrast with none in the
recoded dichotomous variable. Because of
its distribution (Figure 3.3), the sum of
physical performance indicators was
recoded into three values, combining the
original 0 and l into a single category indicating substantially diminished physical
ability, the original 2 and 3 into a category associated with moderate disability, and
leaving the original 5 (indicating the presence of all abilities) as the third category.
Analysis
Analysis began with descriptive summaries of the data and examination of
correlations between variables. Tables C.l and C.2 in Appendix C present summary
statistics and measures of correlation between the dependent variable and each
independent baseline and 12-month change variable.
Theoretical assumptions described in Chapter 2 suggested that health consists
of many factors that contribute sequentially to an individuals perception of it Based
on those assumptions, variables were grouped into categories (e.g.,
sociodemographics, functional status). Stepwise multiple linear regression modeling
identified the variables within each category that had a significant association with
perceived health at 12 months, the dependent (outcome) variable. The analysis used
the regression functions of SPSS 8.0 for Windows, entering and removing variables
stepwise with default selection criteria probabilities (for F) oip = 0.10 to enter and p
= 0.15 to remove.
Sum of Ptiyucal Performance Indicators at Basefine
Figure 33. Physical performance
61


The model-building process continued by hierarchically forcing entry of each
categorys .significant variables, as identified in the previous stage of the analysis, into
a multiple linear regression model. Sequential, or hierarchical, entry of categories
began with sociodemographics, followed by chronic conditions, functional status,
physical performance and mobility, depression and cognition, informal social support,
formal social support, and health care services utilization. Entering categories in this
fashion permitted examination of the incremental effects of progressively less
individual and more social/behavioral and social/delivery-of-care factors.
Results
Study Attrition
Of the 793 study subjects who completed the HSF questionnaire at baseline,
705 completed the same instrument again approximately 12 months later. Twenty-
Table 3.1. 12-Month HSF: Comparisons Between Respondents and Non-Respondents
Responded to HSF at 12 months?
No Yes
Variable (mean at baseline)__________________________(N= 88)_______(Y=70S)________pa
Age 74.09 73.46 0.47
Gender (0 = male, 1 = female) 0.57 0.62 0.38
Sum of chronic conditions IDd by patient 2.06 1.54 BIKE
Sum of chronic conditions IDd by physician 1.38 1.10
Sum of ADLs4 0.24 0.14 024
Sum of IADLsc 1.20 0.54 OtCKB
Sum of physical performance indicators 2.41 3.07 0:000
Depression (0 = no, 1 = yes) 0.30 0.18 0:01
Live alone (0 = no, 1 = yes) 0.34 0.30 0.39
Number of daily medications 6.12 4.76 0:003
Health status at baseline 2.77 3.16 6.000
Note. Shading indicates statistically significant differences, /K0.05.
a p based on Students t-test for continuous data, Mann-Whitney U test for dichotomous data.
b ADL = dependencies in Activities of Daily Living.
c IADL = dependencies in Instrumental Activities of Daily Living.
62


four of the 88 non-respondents died during the year, 13 terminated their Kaiser
memberships (four others who terminated membership did complete the 12-month
HSF), and nine switched providers (37 others who switched did complete the 12-
month HSF). Excluding those who died, 91.7 percent of those who could have
completed the 12-month HSF (705 of769) did so. The 88 non-respondents not
surprisingly were on average in poorer condition at baseline than the 705 respondents,
as Table 3.1 shows.
Thirteen of the 705 participants who completed the HSF at 12 months failed to
rate their health status, the outcome variable for this analysis. Data in Table 3.2
compare mean values of selected variables between those who did and did not
respond to the health status question at 12 months. Those with missing data were
older, more likely to be male, less functionally impaired, and more likely to live
alone, but only the differences in IADL dependencies are statistically significant
(p<0.05). As with those who did not complete any of the HSF at 12 months, these
Table 3.2. 12-Month HSF: Comparisons Between Those Who Did and Did Not Report
Health Status
Variable (mean at baseline) 12-month health status missing? No (JV= 692) Yes (N= 13) P
Age 73.4 77.1 0.08
Gender (0 = male, 1 = female) .62 .54 0.56
Sum of chronic conditions IDd by patient 1.55 123 0.32
Sum of chronic conditions EDd by physician 1.11 .85 024
Sum of ADLs4 .14 .00 0.46
Sum of IADLsc .54 .15 0:00
Sum of physical performance indicators 3.07 323 0.62
Depression (0= no, 1 = yes) .18 .23 0.63
Live alone (0 = no, 1 = yes) 29 .54 0.05
Number of daily medications 4.76 4.25 0.66
Health status 3.15 3.42 0.29
Note. Shading indicates statistically significant differences, p<0.05.
a p based on Students t-test for continuous data, Mann-Whitney U test for dichotomous data.
4 ADL = dependencies in Activities of Daily Living.
f IADL = dependencies in Instrumental Activities of Daily Living.
63


I
individuals data have been excluded from the analysis, leaving a study sample of 692
and a response rate of 90.0 percent (692 of the 769 who survived).
Missing Data
There were several baseline questionnaire items that respondents left blank
more often than other items. Thirty-two respondents failed to answer the depression
question, 30 did not provide the number of daily medications, and 195 left blank the
optional income item. To assess whether other significant differences existed
between respondents and non-respondents to these items, data reported in Table 3.3
compare a subset of the variables between the respondents and non-respondents.
Those who did not respond to the depression and income questions reported
significantly lower health status at baseline but no other signigicant differences.
Based on 35 baseline and 12-month variablesJ, 391 (56.5 percent) study
members provided complete data, 227 (32.8 percent) failed to respond to one item,
and 69 (10.0 percent) left two to five items unanswered. Only five (0.7 percent) failed
to respond to more than five items; four of these missed a complete two-sided page of
the HSF questionnaire.
Because there were so few significant differences between those who did and
did not respond to some of the questions, it seemed appropriate to substitute imputed
values for the missing explanatory variable data in order not to exclude cases
unnecessarily and to ensure the most complete possible set of predicted values on
which to base outlier selection. There seemed to be no obvious pattern to missing I
Health status at baseline and at 12 months, baseline age, gender, marital status, race/ethnicity,
education, employment status, income, baseline and 12-month sums of chronic conditions, worsening
of chronic conditions, bowel and urinary incontinence, sums of Activities of Daily Living and
Instrumental Activities of Daily Living dependencies, sums of physical performance indicators,
mobility, depression, type of housing, sums of needed social support, and transportation requirements.
I
64
i


data either by variable or by case; therefore mean values were imputed from existing
data and substituted for missing values of the independent (explanatory) variables.
Table 3 3. Comparisons of Selected Baseline Variables, Cases with Data Present vs. Cases
with Data Missing
# of daily
Depression__________medications_____________Income
Missing? Missing? Missing?
Variable (mean at baseline) No #=660 Yes #=32 P No #=662 Yes #=30 Pa No #=497 Yes #=195 Pa
Age 73.4 74.5 0.43 73.4 753 0.16 73.4 73.6 0.81
Gender .62 .63 0.94 .61 .74 0.19 .60 .66 0.14
Sum of chronic 1.55 139 0.47 1.55 1.33 033 1.57 1.48 0.38
conditions IDd by patient Sum of chronic 1.11 1.00 0.57 1.11 .85 0.18 1.11 1.07 0.60
conditions IDd by physician Sum of ADLs* .14 .12 0.84 .14 .07 0.61 .13 .16 0.60
Sum of IADLsc .52 .97 038 .53 .77 031 .55 .51 0.75
Sum of physical 3.09 2.66 0.05 3.06 3.32 0.24 3.08 3.05 0.81
performance indicators Depression (0 = no, 1 NA NA .18 .19 0.78 .17 30 0.31
= yes) Live alone (0 = no, 1 = .30 38 0.89 39 .39 0.36 .30 .28 0.47
yes) Number of daily 4.77 430 0.54 NA NA 4.82 4.58 0.50
medications Health status at 3.17 2.89 3.15 332 0.70 3.21 3.02 0.01
baseline Health status at 12 3.08 2.91 "034~ 3.07 3.03 0.84 3.11 2.97 0 12
months_________________________________________________________________________
Note. Shading indicates statistically significant differences, p<0.05.
a p based on Students t-test for continuous data, Mann-Whitney U test for dichotomous data.
b ADL = dependencies in Activities of Daily Living.
c IADL = dependencies in Instrumental Activities of Daily Living.
65


Characteristics of the Study Samnle
Perceived Health. Mean perceived health status at baseline was slightly
greater than good (3.15 on a scale from 1 for poor to 5 for excellent). The mean
decreased only slightly during the year (to 3.07), but at 12 months fewer people
reported very good and more reported fair. Tables 3.4 and 3.5 and Figures 3.4 and 3.5
show the distributions of the ratings at baseline and 12 months.
Table 3.4. Distribution of Baseline
Perceived Health Status Ratings
Status Frequency Percent
Poor 25 3.6 %
Fair 128 18.5
Good 287 41.5
Very good 207 29.9
Excellent 37 53
Total valid 684 98.8 %
Missing 8 13
Table 3.5. Distribution of 12-Month Perceived Health Status Ratings
Status Frequency Percent
Poor 34 4.9 %
Fair 165 23.8
Good 266 38.4
Very good 173 25.0
Excellent 54 7.8
Total 692 100.0 %
Parenvad HaaMt Sana at Bamne
Figure 3.4. Distribution of health status at
baseline
Paraned HaaWi Statu* at 12 Months
Figure 3.5. Distribution of health status at
12 months
66


The nature of the perceived health outcome variable introduced concerns
about floor and ceiling limits. Individuals who reported poor or excellent health
status at baseline could indicate change in only one direction at 12 months. The
numbers affected, however, were very small. Of the 25 who reported poor health at
baseline, 14 indicated improved status at 12 months, leaving only 11 who might have
reported poorer health if such a category had been available. Similarly, of the 37
respondents who reported excellent health at baseline, 12 indicated poorer status at 12
months; that is, a maximum of 25 might have selected a better status at 12 months if
given the opportunity. Given these small numbers, the analysis did not exclude
individuals who reported poor or excellent health status at baseline.
Baseline Explanatory Variables. Table 3.6 summarizes some of the baseline
characteristics of the study sample, with information about the response rate for each
item. Table C. 1 in Appendix C, which more completely describes the variables,
presents responses to the baseline HSF, attendance, and administrative data; variables
that have been recoded or aggregated from those data; and correlations between the
baseline variables and one-year perceived health. Table C.2 provides similar
information for changes in variables over the 12-month period.
The mean baseline age was 73.4 years; 45 percent of the individuals were
between 65 and 75 years old. Approximately 62 percent were female; more than 90
percent identified their race/ethnicity as White/Caucasian (it is likely that some of the
4.6 percent who identified as Native American misunderstood the classification); 50
percent had at least a high school education; and, for those who responded to the item,
mean income lay between $15,000 and $25,000.
The HSF distinguished between chronic conditions that the individual would
be aware of (e.g., trouble seeing or hearing, asthma, chronic lung disease, arthritis)
and other conditions more likely to be identified by the physician (e.g., hypertension,
heart attack, kidney disease, cancer). Five percent of the study members reported no
67


Table 3.6. Frequencies (Prevalence) of Selected Baseline Characteristics (N = 692)
Cateeorv Variable % with condition present % with data missing % with Cateeorv condition Variable present % with data missing
Perceived health Phvsical performance
Poor 3.6% 13% Sum of abilities 0.4%
Fair 18.5 0 6.4%
Good 41.5 1 4.6
Very good 29.9 2 13.9
Excellent 5.3 3 26.0
Demographics 4 48.7
Age 0.6 Mobility 0.9
<65 14.8% Need aid 22.1
65-75 45.4 No limits 77.0
>75 39.1 Transportation 0.6
Gender (female) 61.8 0.0 Drive self 81.8
Chronic conditions Others drive 17.6
Sum IDd by patient 0.6 Dont go out 0.0
0 17.8 Depression, dementia
1 35.0 Feel depressed 17.1 4.6
2 28.3 Sum of dementia 0.4
3 12.6 indicators
4 43 0 78.8
5 12 >1 20.8
6 03
Sum IDd by physician 0.6 Informal social support
0 28.3 Married 62.4 0.4
1 43.4 Full- or part-time 22.0 0.3
2 19.4 Employment
3 5.8 Living arrangement 0.0
4 2.5 Alone 29 2
5 0.0 With spouse 61.8
6 0.1 With child(ren) 7.1
Fimctional status With relative(s) 1.9
Sum of ADLs 1.0 W/ non-relative(s) 23
0 92.1 With pet(s) 15.6
>1 6.9 Housing 0.3
Sum of IADLs 0.0 Own home 97.0
0 79.6 Formal social support
>1 20.4 Sum of supports 0.0
0 94.4
>1 5.6
68


chronic conditions at baseline; 26 percent reported four or more of the 14 listed
conditions. As Table 3.7 shows, the most commonly cited conditions included
arthritis or rheumatism (60 percent), hypertension (49 percent), and deafness or
trouble hearing (29 percent). Between
10 and 20 percent of the respondents
reported cancer, heart attack or
myocardial infarction, chronic lung
disease, asthma, angina, blindness or
trouble seeing, and/or diabetes. Only a
few individuals indicated that they had
experienced a stroke, congestive heart
failure, kidney disease, or an ulcer or
gastrointestinal bleeding. Nine percent
felt that at least one of these conditions
was getting worse.
These community-dwelling older people, for the most part, reported functional
independence, as measured by dependencies in Activities and Instrumental Activities
of Daily Living (ADLs and IADLs), physical capabilities, and mobility. They
reported almost no ADL dependencies.4 For three of the eight IADLsgrocery
shopping, chores, and transportationnearly 12 percent indicated the need for
assistance.5 It is interesting to note that these are the three IADLs most dependent on
physical ability and mobility. Few study respondents (10 percent) had difficulty
walking up and down stairs or going out to a movie or other such activity, but more of
them needed help (from a person or special equipment, e.g., a walker) doing heavy
Table 3.7. Prevalence of Chronic Conditions
Condition Prevalence
Arthritis/rheumatism 60 %
Hypertension 49
Deafness/trouble hearing 29
Cancer 18
Heart attack 17
Chronic lung disease 15
Asthma 14
Angina 13
Blindness/trouble seeing 13
Diabetes 13
Stroke 8
Congestive heart failure 7
Kidney disease 4
Ulcer or intestinal bleeding 4
4 ADLs included in the HSF: need help toileting, bathing, dressing, eating, and getting in and out of
bed.
5 IADLs included in the HSF: need help with meal preparation, grocery shopping, household chores,
money management, laundry, taking medications, transportation, and using the telephone.
69


work (nearly 50 percent) or walking a half mile (almost 30 percent). They remained
quite mobile; all of them could get around inside and outside the house with at most
aid from a cane or wheelchair (and that for only a few). Seventy-two percent drove
their own cars.
Approximately 18 percent of the 660 individuals who responded to an item
about depression reported often feeling sad or depressed. Seventeen percent of the
study sample reported increasing problems with forgetting dates and names, but few
indicated any problems with the other indicators of dementia.
These individuals had access to sources of informal social support, to the
extent that the HSF measured that type of support. Sixty-three percent were married,
22 percent worked full- or part-time, and 97 percent lived in their own homes. They
reported little contact with (and presumably little need for) formal support services.
Administrative data confirmed generally high utilization of primary care (82
percent had at least one visit other than CHCC), hospital services (24 percent had at
least one admission), and emergency care (26 percent had at least one visit). Some
concerns exist about the accuracy of these administrative datawhen different
systems supposedly contained related longitudinal data, they did not always agree.
There is no reason to believe, however, that any particular bias existed between
individuals or at a particular point in time. Even allowing for possibly inflated
numbers, this group of older people continued to consume a large number of health
care services.
Approximately 22 percent of the intervention group members (78 of the 356
who survived and did not terminate their Kaiser memberships) attended no CHCC
meetings; five percent (19) attended only one. There were few significant HSF
differences between those who attended and those who did not. Non-attenders were
more likely to die during the 12-month period (p 0.05, based on the Mann-Whitney
U test), to be depressed (p < 0.01), and to report fewer medications ip = 0.01).
70


Correlations Between Baseline Explanatory Variables Aand 12-Month
Perceived Health. Baseline perceived health status, as one would expect, correlated
positively with 12-month status
(Kendall = 0.581). The only
explanatory variables that
correlated even moderately (|zi|
> 0.200) with 12-month
perceived health status were
sums of chronic conditions
identified by the patient, ADLs,
IADLs, physical performance
indicators, mobility, need for Figure 3.6. Change in Perceived Health Status
transportation, and number of medications.
Changes Over 12 Months. More than half of the study members reported no
change in health status over the 12 month period (see Figure 3.6). Most who reported
any change indicated transitions between adjacent categories; 147 (21.2 percent)
reported a decline of one category, 129 (18.6 percent) a one-category increase. Fewer
than five percent reported greater changes. Table 3.8 provides details.
Table 3.8. Changes in Health Status over 12 Months
Health status_________________Health status at baseline (N = 684)
at 12 months Very Excel-
(N= 692) Poor Fair Good good lent Missing Total
Poor ;TI 16 5 1 1 34 ( 4.9%)
Fair 11 75 59 15 5 165 ( 23.8%)
Good 2 35 160 64 *> 2 266 ( 38.4%)
Very good 60 104 8 1 173 ( 25.0%)
Excellent 1 2 3 23 :25 54 ( 7.8%)
Total 25 128 287 207 37 8 692 (100.0%)
3.6% 18.5% 41.5% 29.9% 5.3% 12%
Note. Shading indicates no change.
71


There were no dramatic changes in numbers of chronic conditions or any of
the measures of functional and physical status during the year. The changes that were
reported occurred in the direction one would anticipate: increased dependencies in
IADLs, decreased physical performance and mobility, greater assistance needed with
transportation, and an increase in the number of medications.
Regression Models
Table 3.9 presents results of the stepwise linear regression selection of
variables within each conceptual predictor category, showing only those that
contributed significantly (p < 0.10) to the variation in 12-month perceived health.
The stepwise regression procedure confirmed the relationships suggested by
correlation comparisons. Categories that individually explained the greatest amount
of variation in 12-month perceived health status included baseline perceived health
status (41 percent); the number of chronic conditions, change in the number of
reported conditions over the 12-month period, and reported worsening of any of the
conditions (20 percent); sums of ADL/IADL dependencies and changes between
baseline and 12 months (16 percent); stun of physical performance indicators,
mobility, and changes in these variables over the 12-month period (26 percent); and
utilization of some health care services (15 percent). Demographics, depression and
dementia, informal social support, and formal social support added little explanatory
power. Neither membership in the parent studys experimental group nor percentage
attendance at group visit meetings showed any significant association with perceived
health.
72


Table 3.9. Perceived Health Status at 12 Months: Multiple Linear
Regression, Stepwise Within Categories
Category Variable Step Adjusted R2 Beta
Baseline health status 1 .414 .709
Sociodemoeraohics Education 1 050 .152
Income 2 .053 .060
Chronic conditions Sum, identified by patient at baseline 1 .095 -26 9
Sum, identified by patient, change 2 .132 -.171
Sum, identified by physician at baseline 3 .152 -.164
Sum, identified by physician, change 6 .202 -.152
Condhion(s) getting worse at baseline 5 .195 -.688
Condition(s) getting worse, change 4 .168 -.605
Functional status Sum of ADLs at baseline 3 .138 -.836
Sum of ADLs, change 4 .159 -.565
Sum of IADLs at baseline 1 .088 -.713
Sum of IADL, change 2 .128 -.374
Phvsical Derformance Physical performance at baseline 4 261 .195
Physical performance, change 3 .240 .188
Mobility at baseline 1 .132 .642
Mobility, change 2 231 .490
TransDortation Transportation at baseline 1 .050 -.737
Transportation, change 2 .085 -.548
Degression, dementia Depression at baseline 1 .044 -.851
Depression, change 2 .076 -.497
(Dementia not significant) Informal social simvort Employment at baseline 1 .026 .526
Employment, change 2 .035 .288
Live with spouse 3 .039 .336
Formal social suDDort Formal social support at baseline 2 .030 -.704
Formal social support, change 1 .013 -.699
Utilization Number of daily medications at baseline 2 .081 -.073
Number of daily medications, change 3 .108 -.054
73


TableS^P^CConL^
Category Variable Step Adjusted R2 Beta
Durable medical equipment/supplies 1 .047 -397
Home health care 4 .130 -306
Patient education visits 6 .145 -315
Skilled nursing facility visits 8 .151 -367
Emergency visits 5 .141 -323
# of days of hospitalization 7 .149 -.004
Note. Adjusted R2 is cumulative by category; non-standardized betas.
Table 3.10 contains the results of the hierarchical linear regression analysis of
perceived health status, beginning with the sociodemographics category and adding an
additional category at each subsequent stage of the modeling process. The table
shows unstandardized beta coefficients for those variables that retained statistical
significance at each stage of the modeling process. It also contains cumulative and
incremental coefficients of multiple determination (R2), to measure the amount of
variation in perceived health that each model explains. Incremental changes in R2
values are presented in two ways. The first describes how much the model improved
when a new category was added in the hierarchical order suggested by theoretical
concerns (Type I sum of squares). The second measures the incremental
improvement if the category were added last to the model, with all other categories
already in the model (Type in sum of squares).
The unstandardized beta coefficient quantifies the difference in perceived
health status that results from a unit change in a variable. Perceived health is scaled
in discrete steps from poor (1) to excellent (5). Although reported status can take on
only integer values, predicted values range along the scale continuously. In this report
of results of the modeling, effects will be described as increments (or decrements) of
steps along the perceived health status scale. For example, the effect on perceived
health status of a variable with a positive beta coefficient of .030 will be reported as
an improvement of 0.03 step. For continuous variables (e.g., age) each increase of
74


Table 3.10. (Cont.)
Category
Variable
Model Model Model Model Model Model
1 2 3 4 5 6
Model Model Model
7 8 9
Ul
Informal social support
Employment at baseline
Employment, change
Live with spouse at baseline
Live with spouse, change
Formal social support
Formal support at baseline
Formal support, change
Utilization and Intervention
Number of daily medications at baseline
Number of daily medications, change
DME/supplies, any vs. none
Home health care, any vs. none
Emergency visits, any vs. none
Non-CHCC education visits, any vs. none
# of days of hospitalization
SNF episodes, any vs. none
Percent of CHCC meetings attended
Constant
Adjusted R2
Category added
Cumulative R}
Incremental R\ hierarchical
Incremental R1, if category added last
.177*
NS
.221J
NS
.180*
NS
.22 If
NS
NS
NS
NS
NS
.21 If
NS
NS
NS
NS
NS
NS
NS
NS
NS
-.194f
2.223 3.098 3.168 1.966 2.065 2.072 2.041 0.003 NS NS 2.210 -.004* 2.235
Socio Chron Funct Phys Depr Inform Formal Util
.053 .233 .301 .351 .363 .375 .376 .382 .377
.180 .068 .050 .012 .012 .001 .006 -.005
.014 .041 .000 .042 .009 .010 .001 .006
Note. Unstandardized betas; change refers to change over 12-month period; J p <, 0.001; t p £ 0.01; p <. 0.05; NS p> 0.05.


Table 3.10. Hierarchical Multiple Linear Regression of 12-Month Perceived Health on Previously -Selected Variables
Category Model Model Model Model Model Model Model Model Model
Variable 1 2 3 4 5 6 7 8 9
Demoeraohics Education .152$ .124$ .110$ .107$ .105$ .087$ .083$ .078$ .111$
Income Chronic conditions NS NS NS NS NS NS NS NS
Sum, identified by patient at baseline -.237$ -.184$ -.126$ -.109$ -.108$ -.106$ -.093$ -.112$
Sum, IDd by patient, change for worse -.166$ -.132$ -.102$ -.095$ -.093$ -.091$ -.098$ -.098$
Sum, identified by physician at baseline -.160$ -.149$ -.133$ -.144$ -.141$ -.136$ -.122$ -.114$
Sum, IDd by physician, change for worse -.138* -.117* NS NS NS NS NS
Condition(s) getting worse at baseline -.720$ -.602$ -.506$ -.449$ -.479$ -.491$ -.464$ -.466$
Condition(s) getting worse, change Functional status -.620$ -.531$ -.452$ -.414$ -.418$ -.415$ -.420$ -.392$
Sum of adls, at baseline -.708$ -.405* -.355* -.343* -.335* NS -.299*
Sum of adls, change for worse -.600$ -.332$ -.328$ -.297* -.278* NS -.269*
Sum of iadls at baseline -.349$ NS NS NS NS NS
Sum of iadls, change for worse Physical performance -.214* NS NS NS NS NS
Sum, physical performance at baseline .099* NS .089* .094* NS .096*
Sum, performance, change for better .123$ .115$ .122$ .123$ .107$ .124$
Mobility at baseline .439$ .472$ .473$ .473$ .451$ .459$
Mobility, change for better .355$ .357$ .342$ .335$ .335$ .348$
Transportation needs at baseline NS NS NS NS NS
Transportation needs, change for worse Depression (and dementia) NS NS NS NS NS
Depression at baseline -.386$ -.389$ -.376$ -.356$ -.404$
Depression, change for worse -.206* -.197* -.192* -.174* -.204*


one unit (e.g., one year of age) changes predicted health status by the beta number of
units. The change is positive or negative depending on the sign of the beta
coefficient; a positive sign denotes an increase in health status, a negative sign a
decrease. For dichotomous and ordinal variables, the presence of the variable (e.g.,
depression) indicates a change of beta units (for each ordinal increase) in predicted
health status compared to health status in the variables absence, if all other variables
remain the same.
The first column of Table 3.10 shows the results of Model 1, when only
sociodemographic variables were included. Education had a fairly small but
statistically significant effect. (Throughout this discussion, significant denotes
statistical significance.) Each increased level of schooling corresponded to about 0.10
step increase in health status.
Model 2 introduced chronic conditions and worsening of those conditions.
All of these factors had significant negative effects. Conditions identifiable by the
individual (e.g., chronic lung disease, deafiiess, arthritis) had a greater effect than
those more likely identified by a physician (e.g., cancer, kidney disease, stroke).
Increased numbers of conditions over time had smaller but significant effects.
Individuals reports that these conditions were worsening, and worsening more over
time, had the greatest negative effect on perceived health status in this model, each by
more than half a step.
Introducing functional status (Model 3) and physical performance and
mobility (Model 4) increased the total amount of variation explained from 23 to 35
percent. Education continued to be significant in these models but with reduced
effect. Similarly, chronic conditions (and their worsening) remained significant but
with a slightly reduced magnitude. The presence of any AJDL dependency (vs. none)
corresponded to a decrease of about 0.70 step in health status (reduced to 0.40 after
the addition of physical performance to the model). An increase (decrease) in the
77


number of ADL dependencies during the 12 months decreased (increased) health
status by another 0.60 step (reduced to 0.33 after the addition of physical performance
to the model). Dependencies in IADLs, many of which require physical ability and
mobility (e.g., grocery shopping, chores, and laundry), had significant effects only in
the absence of the physical performance and mobility measures. Better baseline
mobility increased perceived health by more than 0.40 step, and improvement
(worsening) over the 12-month period increased (decreased) it by about 0.35 step.
Both better baseline physical performance ability and improvement over the 12-month
period had significant but smaller positive effects of about 0.10 step each.
Model 5 shows the results of adding the presence of depression and change in
depression during the 12-month period. Both had significant effects, of, respectively,
nearly 0.40 and about 0.20 step decrease in health status. The effects of previously
entered variables remained virtually unchanged.
Incorporating informal and formal social support variables (Models 6 and 7)
added litde explanatory power. Being employed or volunteering full- or part-time
increased health status by about 0.18 step. Living with ones spouse had a significant
negative effect on health status, representing a 0.22 step decrease. This result, which
contradicts previous studies reports, reflects correlations noted abovenegative
between perceived health and living with ones spouse and positive between
perceived health and living alone.
Model 8 introduced utilization and intervention variables. Adding these
variables, none of which contributed significantly (although hospital days nearly
reached significance,/? = 0.056), decreased the effect of ADLs and physical
performance, which lost statistical significance; they were especially sensitive to the
inclusion of durable medical equipment (DME). The final model, Model 9, included
those variables that remained significant in Model 8. In addition, ADLs and physical
performance were reintroduced in the absence of DME, and the hospitalization
78


variable was added because of its near significance in the previous model. ADLs and
physical performance regained statistical significance in this model, and
hospitalization days showed a small but significant effect. This final model explained
38 percent of the variation in perceived health. The intercept value of2.235 defined
the average perceived health status of individuals who reported the referent values of
all variables; that corresponds to a rating of a little better than fair.
Discussion
The purpose of the quantitative portion of the study was to construct a
statistical model of perceived health status from a set of factors (within theoretically
aggregated categories) commonly associated with negative outcomes of aging. There
were two reasons for doing so. The first goal was to determine which of the factors
appeared to contribute most to positive perceived health status. The second, primarily
a methodological issue, was to provide predicted values of perceived health with
which to compare individuals reports, in order to identify a sample of individuals
whose reported status differed from that predicted by the model. That process will be
more fully described in the next chapter.
The results indicated that many but not all of the factors associated with
negative outcomes do contribute (in the opposite direction) to healthy aging, as
measured by perceived health, and that they explain about 38 percent of the observed
variation in perceived health. Of the sociodemographic variables, only education
remained in the model; perceived health increased by 0.11 step for each increased
level of education. This means, for example, that if all other characteristics were
identical, an individual with some college education would be expected to report
health status 0.11 higher (on the scale of 1 = poor to 5 = excellent) than another who
had only completed high school. So, if the high school graduates expected health
status were good (measured as 3 on a 5-point scale), then the otherwise identical
79


college attendee's would be slightly better (3.11, an improvement of 11 percent of the
step between 3 and 4). Table 3.11 summarizes the expected (predicted) incremental
changes due to each variable in the final model.
The study found that physical health status, as represented by chronic
conditions, detracted substantially from perceived health. Perceived health status
decreased by 0.11 step for each additional reported baseline chronic condition, 0.47
step for baseline worsening of conditions), and 0.40 step if a worsening of
condition(s) was reported at 12 months but not at baseline.
Functional dependencies (ADLs and IADLs) explained some of the variation
in perceived health, but the importance of these variables diminished (and, for IADLs,
disappeared) when combined with measures of physical performance and mobility,
which significantly affected perceptions of health status. Perceived health status
decreased by 0.30 step in the presence of any baseline ADL dependencies and by 0.27
Table 3.11. Incremental Changes in Expected Values of Perceived Health
Factor_________________________________________________________________________Change
Expected perceived health for an individual who reported referent values for 2.235
every variable in the model.......................................... (fair +)
Adjustments:
for each increased level of education................................ +0.111
for each additional baseline chronic condition identified by patient. -0.112
for more (fewer) chronic conditions identified by patient at 12-months... - (+) 0.098
for each additional baseline condition identified by physician....... -0.114
if conditions getting worse at baseline.............................. 0.466
if conditions getting worse at 12 months but not baseline............ 0.392
if any baseline dependencies in Activities of Daily Living (ADL)..... - 0.299
if more ADL dependencies at 12 months................................ 0.269
for each positive baseline physical performance indicator............ + 0.096
for each positive (negative) change in physical performance at 12 months + (-) 0.124
for each better baseline level of mobility........................... + 0.459
for each positive (negative) change in mobility at 12 months......... + (-) 0.348
if depressed at baseline............................................. 0.404
if more (less) depressed at 12 months................................ (+) 0.204
if live with spouse.................................................. 0.194
for each day of hospitalization during 12 months..............................- 0.004
80


step with the report of a greater number of dependencies over the 12-month period; it
increased by 0.09 step for each additional baseline indicator of physical performance,
increased (decreased) by 0.12 step with each change for the better (worse), increased
by 0.46 step for each better baseline level of mobility status, and increased
(decreased) by 0.35 step for improvement in (worsening of) mobility over the 12-
month period.
Depression and change to a depressed state over the 12 months had a
substantial negative impact, independent of other variables and greater than suggested
by previously published studies of more distal outcomes. The presence of depression
decreased perceived health status by 0.40 step, and change to a depressed state
decreased it by 0.20 step.
Among measures of informal social support, only living with ones spouse
contributed to the model, for a decrease in perceived health status of 0.19 step. The
negative effect of living with ones spouse contradicted results from most previous
studies. Marital status did not affect the outcome, either independently or in
combination with the living with... variable. More men than women (78 vs. 52
percent) lived with their spouses, and more women than men were widowed (36 vs.
11 percent), but the individuals gender did not significantly interact with living with
ones spouse to explain variation in perceived health status, nor did gender alone
affect perceived health. It may be that the spouses of the married study subjects
needed substantial care, creating caregiver stress that outweighed previously reported
benefits of living with ones spouse, but quantitative data do not exist to explore this
potential relationship. It may also be the case, as Berkman (1988) suggested, that
marital status may no longer tap a sense of intimacy or social integration for older
people as widowhood becomes a normative experience (p. 59). Among utilization
measures, only hospitalization affected perceived health status, with a minimal
negative effect
81


Some factors that contributed to poor health outcomes in earlier studies of
older people did not affect perceived health in this study population. Although being
male generally predicts mortality, it did not here predict perceived health status.
Neither age nor race/ethnicity had an effect, most likely because of the homogeneity
of this group. To the extent that disability and chronic illness correlate with age, it
may be that the model excluded age not because it had no explanatory power but
because collinearity with other variables masked its effect. There was, however, very
little correlation between age and any of the ability or performance variables.
Although living with ones spouse contributed (negatively) to the model of perceived
health, marital status did not, either independently or in combination with the living
with .. variable. IADL abilities, many of which depend on physical abilities, had no
impact once physical performance and mobility measures were introduced into the
model. Again multicollinearity may explain the absence of effect; IADL
dependencies correlated significantly with the sum of physical performance indicators
(Kendall %& = -0.434). Dementia had no effect, but although more than 20 percent of
respondents reported at least one indicator of dementia, they generally only reported
forgetting dates and names, not other more serious cognitive disabilities. Neither
housing type nor any measures of formal support predicted perceived health, likely
because this community-dwelling population did not yet need environmental
supports.
Of the various health care utilization variables, only the number of days of
hospitalization contributed significantly to the model; each incremental day of
hospitalization decreased perceived health status by 0.004 step. Because of their
HMO membership, these individuals had few if any barriers impeding access to
health care. This homogeneity of access may explain why utilization explained so
little of the variation in their perceptions of health status.
82


The analysis here did not test the sequential causal relationships among
categories or domains of health status suggested by Wilson and Clearys (1995)
model, but it did support the notion that some of these categories have cumulative,
interdependent effects. Physiological factors and symptoms (measured here by
chronic conditions) and broadly defined physical functional status (daily living
abilities, physical performance, and mobility) each contributed similarly to the model,
on the margin (Type III sum of squares). Their overall effects when combined were
interdependent The addition of functional status to a model based on chronic
conditions improved the explanatory power of the model but reduced the impact of
chronic conditions, suggesting an interrelationship between the categories. As in
Wilson and Clearys model, social support at least as measured here, had only a
peripheral effect
Limitations
There are limitations to this quantitative analysis. Almost all data were
obtained from self reports. A comparison of questionnaire utilization responses with
information from administrative databases illustrates the hazards of self-reported data.
For example, of 640 individuals who self-reported no utilization of home health
services at 12 months, administrative utilization records indicated that 69 had
received at least one episode of home health service during that time.
As in the example concerning IADLs and physical performance described
earlier, if variables are highly correlated, then a factor retained in the regression model
may be explaining the same variation in outcome that an excluded factor affects. In
that case, a factor found not to have a significant effect on perceived health in the
regression model may in fact be important to the outcome. Very few variables in this
study, however, had any substantial correlation with any of the others.
83


The homogeneous nature of the population studied here limits the
generalizability of the results. All subjects were community-dwelling Medicare-
eligible older people with similar sociodemographic characteristics (age, gender,
race/ethnicity, education, and income), a history of chronic conditions, and access to
the same health care. Replacing missing explanatory covariates with imputed mean
values also decreased variability. The lack of differences between respondents and
non-respondents, however, diminished the negative impact of that replacement.
Summary
In summary, the quantitative part of the study identified factors that predicted
perceived (self-reported) health, as representative of healthy aging. Some of the
factors reported in earlier studies of negative health outcomes among older people
especially smaller numbers of chronic conditions, better physical performance, greater
mobility, and absence of depressionshowed predictive capability for positive
perceived health and in fact explained 38 percent of the variation observed in this
study population. Still, the amount of variation not explained by the model derived
here62 percentsuggests that other factors also have substantial effects. As
Wilson and Cleary (1995) hypothesized, those factors may include characteristics of
the individual and the environment such as personality, values, and societal supports
beyond those measured quantitatively in this study. The next chapter further
describes interview-based qualitative explorations.
84


I
CHAPTER 4
QUALITATIVE ANALYSIS
Exact sciences give correct answers to certain aspects of life problems,
but very incomplete answers. It is important of course to count and
measure what is countable and measurable, but the most precious
values in human life are aspirations which laboratory experiments
cannot yet reproduce. (Dubos, 1959, p. 279)
I would call attention to an unfortunate byproduct of scientific
methodology that is ignored by the pathogenic orientation. The good
scientist formulates a hypothesis, when there is a basis for doing so,
rigorously submits it to testing, and rejoices when it is supported in
repeated testing_The pathogenicist is content with hypothesis
confirmation; the salutogenicist, without disdaining the importance of
what has been learned, looks at the deviant case. (Antonovsky, 1987,
P-ID
The research described in the previous chapter counted and measured. It
produced a statistical model that identified factors, from those commonly associated
with negative risks of aging, that also supported positive outcomes. From the model
it was possible to compute predicted values of perceived health with which to
compare individuals reports of health status. That comparison supplied a method for
identifying, in Antonovskys (1987) words, deviant cases and allowing further
exploration into what makes the difference between those older people who feel
healthy and those who do not. This chapter describes the methods used to select and
interview deviant, or discordant, cases; explains the qualitative analytic methods used;
provides information about these cases; and describes and discusses the results of
qualitative analysis of the interviews.
!
85


Methods
Sample Selection
The sampling frame for this study was more structured than that often
associated with grounded theory-type work because the regression model offered a
method of obtaining what Kuzel (1992) called maximum variation, to benefit from
points of view as disparate as possible. The interview sample was selected to include
different types of discordance between predicted and reported perceived health status:
extreme under-rating, by people who reported poor or fair status compared
to better predicted values;
moderate under-rating, by people who reported good status compared to
better predicted values;
moderate over-rating, by people who reported good status compared to
poorer predicted values; and
extreme over-rating, by people who reported very good or excellent status
compared to poorer predicted values.
Table 4.1 shows the distribution, by category of reported health status, of
study subjects whose reported 12-month health status differed from values predicted
by the regression model (total N= 199). Boxes identify the four discrepant groups.
For example, there were 61 individuals whose reported fair health status deviated (in
a negative direction) between one and two units from the values predicted by the
model, on a standard deviation scale. There were 86 extreme under-raters (76 after
adjusting for deaths and changes in the providers providing care), 17 moderate under-
raters (15 after adjustments), 11 moderate over-raters, and 85 extreme under-raters
(83 after adjustments). Individuals in each group were ordered randomly and
contacted in that order. Prospective interviewees first received a letter outlining the
project, explaining confidentiality, and requesting permission. Approximately one
86


week later, they were called to solicit participation and to make arrangements for the
interview.
Qualitative
sample sizes are hard to
predict. The
interviewing process
should continue until no
new information or
insights seem to be
forthcoming and the
emerging theory has been
well-tested against
counterexamples, or
Table 4.1. Discrepancy from Predicted Health Status
standardized residual" reported health status
poor fair good very good excel- lent total
-3 to -2 2 8 10
-2 to -1 15 61 17 | 93
-1 to 0 13 74 155 7 249
Oto 1 4 21 83 135 243
1 to 2 1 10 29 38 78
2 to 3 1 2 15 18
>3 l 1
total 34 165 266 173 54 692
Note: Numbers refer to number of cases. Boxes identify groups
discussed in the text.
a Standardized residual = standardized value of reported rating less
predicted value; negative residual indicates under-rating.
disconfirming cases in Kuzels (1992) terminology. Kuzel advised that although the
rules are not hard and fast, experience has shown that 12-20 [data sources] commonly
are needed when looking for disconfirming evidence or trying to achieve maximum
variation (p. 41). I initially planned to speak with 20 to 30 people to reach
saturation, with more from the extreme categories than the other two.
Interview Instrument
The structured interview questions solicited information specifically about a
number of factors derived from the literature discussed in Chapter 2, covering
perceptions of health status and reasons for individuals assessments, well-being,
function (valued abilities, activities, and relationships), social support, control, sense
of coherence, and personal outlook. In addition, the interview offered respondents an
opportunity to speak without structure.
87


Three Kaiser CHCC participants who were not part of the parent study pilot
tested the instrument They seemed to enjoy the process, and did not find the
approximately 45-minute sessions a burden. An expert qualitative researcher sat in
on one of the pilot interviews and reviewed the results of all of them. One major
change in the instrument evolved from the pilot tests. Based on the assumption that
the best way to find out what you want to know is to ask it, subsequent interviewees
were asked directly for their thoughts about factors that cause people to perceive their
health differently than expected. Appendix A contains a copy of the final instrument.
Structure of the Interviews
Participants selected the interview locations most convenient for them.
Generally that was their usual Kaiser clinic, but many preferred to be interviewed at
home. Each interview took approximately 45 minutes, although one lasted an hour
and a half. The instrument provided structure, but conversations tended to expand
beyond its borders. Most people seemed genuinely interested in the topic, and they
appeared to contribute thoughtfully and completely.
Interviews were audiotaped. Undergraduate students hired for the purpose
transcribed the tapes into computer text files, for analysis using ATLAS/ti software.
I reviewed each transcript for accuracy and made corrections as needed.
Analytic Methods
Analysis of the interviews employed what has been termed grounded theory-
type (Strauss & Corbin, 1990) immersion into the material. The grounded theory
method of qualitative analysis has quite specific components, a systematic set of
procedures to develop an inductively derived grounded theory about a phenomenon
88


(p. 24). The analysis described here did not follow all of its precepts religiously, but
the precepts gave structure to the process.
Grounded theory-type analysis begins with open coding, the naming and
categorizing of phenomena through close examination of data (Strauss & Corbin,
1990, p. 62). In this case, structured interview questions suggested an initial set of
codes (e.g., childhood, control, well-being, ability, activity, relationships, social
support, getting older). ATLAS/ti software facilitated the process of attaching
codes to segments of text in transcriptions of the interviews. Additional codes
emerged, representing both refinements of initial codes (e.g., types of medical care as
a subset of childhood, loci as a subset of control) and new concepts (e.g.,
independence, normality, something to do).
Expert colleagues reviewed the code dictionary and independently coded
randomly selected interviews. There were few discrepancies among individuals
codings. The experts recommendations were incorporated into the ongoing, iterative
coding process.
The next step in grounded theory-type analysis, Strauss and Corbins (1990)
axial coding, combines the original codes into categories by connecting them in terms
of conditions that give rise to them, properties that are common to them, strategies
that guide them, and consequences they share. As the interview process progressed,
responses began to suggest an analytic framework, that of the (im)balance between
challenges and resources. ATLAS/ti provides tools for constructing and visually
representing networks that incorporate these relationships, and the analysis here made
use of them. For example, one can illustrate a directional causal relationship between
code-factor A and code-factor B(Atya cause o/B) this way:
89


or show code-factor A as a member of the set represented by code-factor B (A is a B):
or a non-directional relationship (A is associated with B) between them:
Figure 4.1 illustrates the analytic thought processes that directed the following
report of the challenges and resources that interviewees thought important. Expert
associates reviewed the analytic process, and their input greatly aided the analysis
while also providing validation.
The figure is an imperfect representation because peoples perceptions are far
more complex than boxes and one-way relationships. Additionally, many factors can
be challenges or resources depending on degree and on individuals points of view
arrows might then be headed in the wrong direction. The chart is included here as an
illustration, not as a constraint. (The numbers in the brackets refer to, first, the
Figure 4.1. Challenges and resources network
90


Full Text

PAGE 1

. I HEAL TIIY AGING: FACTORS THAT CONfRIBUTE TO POSITIVE PERCEIVED HEAL Til IN AN OLDER POPULATION by Lucinda Lynne Bruner Bryant B.A., Brown University, 1962 M.S. Health University of Colorado at Denver, 1993 M.B.A., University of Colorado at Denver, 1993 A thesis submitted to the University of Colorado at Denver in partial fulfiJiment of the requirements for the degree of Doctor of Philosophy Health and Behavioral Sciences 1998

PAGE 2

1998 by Lucinda Lynne Bruner Bryant All rights reserved. I

PAGE 3

i I I : I I This thesis for the Doctor of Philosophy degree by Lucinda Lynne Bruner Bryant has been approved by Arne Beck Nc'J'ern\9e/ \2., Date

PAGE 4

Bryant, Lucinda Lynne Bruner (Ph.D., Health and Behavioral Sciences) Healthy Aging: Factors That Contribute to Positive Perceived Health in an Older Population : I : 1 Thesis directed by Professor Craig R. Janes I I I ABSTRACT The purpose of this study was to identify factors contributing to healthy aging, represented by positive perceived health status, among community-dwelling older people with chronic conditions and a history of high utilization of health care services. The research incorporated quantitative and qualitative methods, with regression modeling and grounded theory-type analysis of interviews. Quantitative data came from the first year of a prospective randomized trial of a managed care organization's outpatient group visit program. Nearly 700 subjects (mean age 73) completed questionnaires concerning variables thought to be important to older people's health. Administrative records provided data concerning utilization of health services. A.multiple linear regression model of perceived health status on these variables: I) identified factors significantly associated with positive perceived health and 2) provided predicted values of perceived health for each study subject. Comparisons between predicted and observed values of perceived health identified individuals whose reported status most differed from predicted values. Semi structured interviews with a random sample of22 of the discrepant individuals probed for information about their interpretations of health and well-being and the relative importance of contributing factors, with emphasis on characteristics that might account for their differential health status assessments. iv

PAGE 5

. I A model of healthy aging emerged from the research The qualitative phase of the research found that these older people equated positive health status with meaningful going and doing. Four categories of factors contributed to that positive outcome: having something worthwhile and desirable to do, possessing the required abilities, obtaining the necessary resources, and having the will or positive attitude to go and do. The quantitative portion of the analysis, which explained almost 40 percent of the variation in perceived health status, supported these conclusions. It found mobility, physical performance, and worsening chronic conditions (fundamental components of ability) and depression (which is related to attitude) the most important of the available variables. Understanding how older people define healthy aging and identifying its most important components provide insights into possible interventions, from both medical and broader community sources. Future studies of health and aging might well benefit from inclusion of the factors identified in this study. This abstract accurately represents the content of the candidate's thesis I recommend its publication. Signed Craig R. 1 anes v

PAGE 6

DEDICATION I dedicate this dissertation with gratitude to Peter Bryant, who secured the roots and freed the wings to make it possible, to Edward and Katherine Bryant, who supported and encouraged from near and far, to Louise Graham Bryant and Lynwood Bryant, whose graceful aging provided the finest model, and to the memory of Laura Lynne Bruner Boyle, whose experiences first made me aware of the issues of healthy aging.

PAGE 7

This research was supported in part through grants from the Robert Wood Johnson Foundation Chronic Care Initiative to the Kaiser Foundation Health Plan of Colorado and from the Graduate Research Opportunities Program of the University of Colorado at Denver. I also wish to acknowledge Dr. Craig Janes, who opened my eyes to the multiple aspects of health; Dr. Arne who offered access to the wonderful people whose stories lie within, provided support, and periodically got me unstuck when data and words overwhelmed; Dr. Diane Fairclough, who with good humor and much patience allowed me to learn how to deal with the data; Dr. Kitty Corbett. who 1 supported and guided the journey through research design and analysis; Dr. Jean Kutner, who read every page, listened to every validated coding, and shared her experience with qualitative research and gerontology; friends and colleagues in the Health and Behavioral Science who enriched the learning; and, last but : I most certainly not least, the research subjects who so graciously shared their stories of health and aging.

PAGE 8

CONTENTS CHAPTER 1. IN1"RODUCTION ........................................................................................ l Healthy Aging ........................................................................................... 1 The Importance of Healthy Aging ............................................................ 2 Specific Aims ............................................................................................ 4 .I I Overview of Research Methods ................................................................ 5 Study Setting ............................................................................................. 7 : I I I Description of the Study Population ......................................................... 8 Structure of the Dissertation ..................................................................... & 2. IlEAL TH .................................................................................................... 1 0 Historical Perspectives ............................................................................ 11 Philosophical Models of Inquiry ............................................................. 12 Definitions of Health ............................................................................... 16 Concepts of Health ............................................................................ 17 Quality ofLife ................................................................................... 19 Models of Health .............................................................................. 22 Operational Definition of Health ...................................................... 27 Possible Determinants ofHealth ............................................................. 28 Sociodemographic Factors ............................................................... .29 Physical and Functional Status ........................................................ .29 Social and Temporal Comparisons ofStatus .................................... 30 Challenges ......................................................................................... 31 Personal Resources and Responses ................................................... 34 viii

PAGE 9

. I I I I :I Social Support Resources ................................................................. 38 Health Care Delivery ........................................................................ 4 2 Societal and Environmental Challenges Specific to Aging .............. 42 Risks and Predictors of (Un)healthy Aging: Previous Studies ............. .46 Mortality ........................................................................................... 46 Institutionalization ............................................................................ 47 Function ............................................................................................ 48 Perceived Health ..................................................................................... 50 Summary ................................................................................................. 55 3. QUANTITATIVE ANALYSIS ................................................................. 51 Methods ................................................................................................... 57 Quantitative Study Design ................................................................ 57 Study Population and Data Collection ............................................. .58 Measures ........................................................................................... 59 Analysis ............................................................................................. 61 Results ..................................................................................................... 62 Study Attrition .................................................................................. 62 Missing Data ..................................................................................... 64 Characteristics of the Study Sample ................................................. 66 Regression Models ............................................................................ 72 Discussion ............................................................................................... 79 Limitations .............................................................................................. 83 Summary ................................................................................................. 84 4. QUALITATIVE ANALYSIS .................................................................... 85 Methods ................................................................................................... 86 Sample Selection ............................................................................... 86 Interview Instrument .......................................................................... 87 lX

PAGE 10

Structure of the Interviews ................................................................ 88 .Analytic Methods .................................................... ......................... 88 Interviewees ................................ ............. ................... ......................... 91 Results ..................................................................................................... 93 Ratings of Health. Status ................................................................... 93 Well-Being ....................................... ................................................ 95 I I Emerging Themes ............................................................................. 9 8 Differences Between Under-raters and Over-raters ........................ 117 Summary of Qualitative Findings ...... .................................................. 119 Going and Doing ............................................................................. 119 Challenges and Resources ............................................................... 120 Comparative Case Studies ......................................... .. ................. 123 Limitations ............................................................................................ 125 5. A MODEL OF HEALTHY AGING ........................................................ 127 Review of Results ................................................................................. 128 The Mode1 ............................................................................................. 129 Components of the Model.. ................................. ........................... 131 Interactive Processes ....................................................................... 134 Evaluating the Model ........................................................................ ... 136 Conclusion ............................................................................................ 13 8 APPENDICES .................................. ............................................................ 140 A. Research Instruments ...................................................................... 140 B. Literature Review Tables ................. ....................... ...................... 150 C. Data Description Tables .................................................................. 194 BIBLIOGRAPHY ...................................................................................... ... 200 X

PAGE 11

: I CHAPTER I INTRODUCTION Healthv Aging MacArthur Foundation Research Network researchers, in their studies on successful aging, defined healthy aging as the absence of any deficits in Activities of Daily Living (ADLs) and the presence of no more than one physical performance disability (Berkman et al., 1993). That definition seems too strict Aging is a process, not a static entity observed at a single point in time, and it is often associated with decreases in physiological, cognitive, and functional abilities. The process varies among individuals, indicating that it is "influenced to some extent by environmental, behavioral, or genetic conditions, which themselves are highly variable" (Berkman, 1988, p. 39). Healthy aging does not mean the absence of decline but the best possible health status for the individual and his or her best adaptation to the aging process. In addition to ability levels, it depends on care and on coping with, reacting to, and managing the aging process. Sharing this view of healthy aging, or successful aging in their terms, Thomas and Chambers ( 1989) described it as "first, self-evaluated 'life satisfaction,' relating to the adaptive tasks of aging, and, second, an external evaluation of the success with which the individual has handled the developmental tasks of coming to terms with the problems of bodily decline and eventual death" (p 185). Curb et al. (1990) called it e.ffoctive aging, deriving from the adaptation and rehabilitation that permits the maintenance of relatively high levels of functioning in many older people, despite physiologic declines, increased numbers of risk factors, and the presence of diagnosed disease. 1

PAGE 12

The Importance of Healthy Aging Older Americans, aged 65 and older, represent an increasingly larger percentage of the population. As Table 1.1 shows. since 1900 the percentage of those aged 65 and older in the United States has more than tripled, to almost 13 percent of the population; the number has increased more than tenfold (U.S. Administration Table 1.1. Actual U.S. Population, 1900-1996 Total Aged65 and Aged 65+as Year population older percent of total 1900 162 3.1 4.1% 1920 106.0 4.9 4.6% 1940 132.2 9.0 6.8% 1960 1793 16.7 9.3% 1980 226.5 25.7 11.3% 1996 265.2 33.9 12.8% Note. Population figures in millions. Adapted from U.S. Administration on Aging, "Profile of older Americans: 1997 .., Available: http://pr.aoa.gov/aoa/stats/profile (30 July 1998). on Aging, 1998). The U.S. Bureau of the Census (1998) predicted they will represent 20 percent of the population by 2050. As Table 1.2 indicates, the number of the oldest-old, those 85 and older, is growing at the greatest rate. Now 3.7 million, their numbers are projected to increase nearly fivefold by 2050. Table 1.2. Projected U.S. 1996-2050 (population in millions) total aged 65 and aged 65+ as aged 85 and aged 85+as older of total older of total 1996 2652 33.9 12.8% 3.7 1.4% 2000 274.6 34.7 12.6% 43 1.6% 2010 297.7 39.4 13.2% 5.7 1.9% 2020 322 7 532 16.5% 6.5 2.0% 2030 346.9 69.4 20.0% 8.4 2.4% 2040 370 0 752 20.3% 13. 6 3.7% 2050 393.9 78.8 20.0% 18.2 4.6% Note. Population in millions. Adapted from U.S. Census Bureau, "Resident population of the United States: Middle series projections, 1996-2050." Available: http://www.census.gov/populationl projections/nationlnas (28 July 1998). 2

PAGE 13

People in general are living longer. Compared to a child born in 1900, who had an average life expectancy of 47 years, a child born in 1996 could expect to live 76 years. Most of this increase has come from reduced death rates for children and young adults, not from increases in the maximum possible age (U.S. Administration on Aging, 1998). An ongoing debate questions whether we are reaching the limits of increased life expectancy, with related concerns that increased life spans may be burdened with longer end-of-life periods of disability and dysfunction. Fries (1980) reported that, while life expectancy has increased, the maximum life span has not. His studies suggested that longer years of life do not necessarily mean longer periods of disability, but with great variability among individuals. "Present approaches to social interaction, promotion of health, and personal autonomy may postpone many of the phenomena usually associated with aging. The rectangularization of the survival curve may be followed by the rectangularization of the morbidity curve and by compression of morbidity" (p. 135). Manton and Soldo (1985) took a less optimistic view, focusing on the reduction of mortality at advanced ages that will lead to a rapid increase in the population aged 85 and older, the population group with the highest per capita service needs. They too, like Fries, cited evidence that "functional impairment among the elderly is not a natural consequence of aging and must be evaluated on an individual basis" but warned that ''whatever interventions may be introduced, the demographic aging of the population will cause large increases in the I number of disabled elderly" (p. 53). Fozard, Metter, and Brant (1990) cautioned that I the aging of the population may mean that in contrast to the past, when "the oldest old was a very select group who had the grit for survival," now "a larger part of the population is surviving to these ages, and may reflect a different kind of aging with less of a survivorship issue" (p. P 117). 3

PAGE 14

I I I Whether or not the onset of morbidity has been the healthiness of older people obviously affects the individuals themselves. As Curb et al. (1990) so clearly It is important that scientists and clinicians adopt a unified concept of aging that allows for prevention, and compensation or rehabilitation with the ultimate goal of developing health-care practice and policies that will maximize the quality of life for the largest number of older people. (p. 828) Healthiness, or its absence, in this rapidly increasing population also has serious implications for demands on health care and other social resources. Although the average annual increase in national health expenditures decreased from nearly 13 percent in 1980 to less than five percent in 1995 and 1996, the total amounted to more than a trillion dollars in 1996. Medicare provided $203 billion of that total, one measure of the cost of providing health care for those aged 65 and older. In 1995, non-institutionalized older people accounted for 38 percent of hospital stays and 48 percent of days of care in hospitals, and older people averaged more than twice as many contacts with doctors as those under 65 (11 vs. five contacts). This older segment of the population accounted for 36 percent of personal health care expenditures in 1987, for a total of$162 billion and an average of$5,360 per person compared to $1,290 for younger people (U.S. Health Care Financing Administration, 1998). These numbers underscore the need to understand the factors that contribute to healthier aging, in order to support better lives for older individuals with the most appropriate and prudent allocation of resources. Specific Aims In order to explore healthy aging and identify factors that contribute to this project analyzed both health status questionnaire and utilization data originally collected during a prospective randomized trial of a group model health maintenance 4

PAGE 15

I I organization's (HMO's) outpatient group visit program and, subsequently, results of interviews with a subset of that population. The study pursued the following specific aims: Specific Aim 1: Construct a statistical model to describe and predict positive perceived health from a variety of demographic, clinical, functional, social support, and utilization variables previously associated primarily with negative outcomes related to aging. Specific Aim 2: Identify discrepant cases, those individuals whose reported perceived health differed substantially from values predicted by the model. Specific Aim 3: Ascertain and describe the characteristics of those individuals whose perceived health ratings differed from those predicted by the model, with the goal of identifying additional factors that support successful, healthy aging and retard declining health. Specific Aim 4: Describe a model of healthy aging for this older population. Overview of Research Methods It is possible that no single methodological approach or simple set of assumptions will provide definitive health predictors. We emphasize the need for a series of imaginative studies in a variety of populations, done in diverse ways so that the more robust predictors will emerge [to] create a realistic background for determining the most efficient methods for reducing the burden of morbidity and mortality in the rapidly increasing elderly population. (Benfante, Reed, & Brody, 1985, p. 394) To achieve the aims of the study, research employed both quantitative and qualitative methods. Construction of a multiple linear regression model had two 5

PAGE 16

I purposes. Firs4 it identified factors significantly associated with perceived health, from the parent randomized trial's variables and data. Second, comparisons between values of perceived health predicted by the model and those actually reported by the study's subjects identified those whose reports most diverged from their predicted values, both positively and negatively A randomly selected sample of the discrepant cases participated in semi-structured interviews. The interviews explored their definitions of well-being, reasons for the assessments they made of their health status, reactions to factors suggested by the literature as contributors to health, and their ideas about why people might underor over-rate their health. Grounded theory-type qualitative analysis of the interviews generated additional insights into characteristics of and factors associated with healthy aging. The final step of the research combined the results of the analyses to construct a model of healthy aging and the factors that contribute to it. The following diagram illustrates the relationships among the parent study, the quantitative and qualitative portions of this study, and the final model. -----. : parent study : : provided data and study population ; . ][----------------------guantitative analysis gualitative analvsis what factors, of those commonly associated with negative outcomes of .. based on interviews with discrepant ... cases, what other factors contribute to aging, predict positive health healthy aging? perceptions? L model of healthy asring what constitutes healthy aging and what factors contribute to it? 6

PAGE 17

Study Setting Quantitative data and the pool of study subjects for this study came from the randomized trial ofKaiser Permanente of Colorado's (Kaiser's) innovative program to provide primary health care to older members in a group setting. The Cooperative Health Care Clinic (CHCC) program combined clinical care, health and social interaction. It targeted the health care needs of an older, community-dwelling subpopulation with a history of chronic disease and greater-than-average utilization of provider services. Physicians could choose this type of care for their older patients but were not obligated to do so. Groups of about 20 members each met monthly for approximately two hours. They received basic health care during the meetings, along with educational presentations concerning relevant issues; presenters included the groups' physicians, psychologists, health educators, dietitians, physical therapists, and many others. According to Kaiser's guide, the CHCC program focused on the elderly population's complex health problems, their coping ability, the assistance they need to maintain independence, the need for coordination among resources, and the desire to promote longer life with decreased illness and disability ("Cooperative Health Care Clinics," 1993). To evaluate the impact of the CHCC model of care, Kaiser conducted a randomized trial of nearly 800 members, half of whom had the opportunity to receive care in the CHCC setting and half who received usual care. It is that parent study that provided quantitative data and study subjects for this study. 7

PAGE 18

Description of the Study Population The mean age of the study subjects was 73 at the beginning of the parent study. All lived in the community-this was not a frail population in need of institutional care. On average, each reported 2. 7 chronic conditions; 60 percent indicated they had arthritis or rheumatism, 49 percent hypertension, and 20 percent trouble hearing. In many ways they mirrored the Table 13. Study Population Compared to the U.S. Population: Selected Characteristics CHCC u.s. Characteristic PoEulation PoEulation Gender 62 %female 59% female Race 89% White 85% White Live Alone 29% 30% Live with Spouse 62% 60% Married 63% 56% Male 78% 76% Female 51% 43% Note. U.S. data from U.S. Administration on Aging, "Profile of Older Americans: 1997." Available: http://pr.aoagov/aoalstats/profile (30 July 1998). noninstitutionalized population of older people in the United States. Table 1.3 compares characteristics of this sample with the U.S. Department of Health and Human Service's 1997 profile ofnoninstitutionalized older Americans (U.S. Administration on Aging, 1998). Chapter 3 and Appendix C contain more complete descriptions of the study population. Structure of the Dissertation This chapter of the dissertation introduced the issues; stated the specific aims of the research; described the research methods, study setting, and population; and now outlines the structure of the following chapters. Chapter 2 reviews perspectives on health, previously described definitions and models of health. factors hypothesized in the literature to contribute to health, and previously published studies of risks and 8

PAGE 19

predictors of(un)healthy aging. It also explains the choice of perceived health as the outcome of interest. Chapters 3 and 4 provide more complete details of methods for the quantitative and qualitative analyses, respectively, as well as results and discussions of results specific to the analyses. Chapter 5 brings together the results of both types of analysis to produce a model of healthy aging for this older population. 9

PAGE 20

CHAPTER2 HEALTH Exploration into factors that contribute to healthy aging logically begins with attempts to construct an operational definition of health or healthy. As R.G. Evans and Stoddart (1990) cautioned, definitions and detenninants of health are separate although related enterprises: ''Whatever the level of definition ofhealth being employed, however, it is important to distinguish this from the question of[its] determinants" (p. 1348). Susser ( 197 4) established the causal progression, stating that "the way health is defined is a necessary antecedent of the way health is measured" (p. 539). He explained the important and complicating dependence of definitions on the value systems of those who do the defining. Definitions contain ethical components that rest on those value systems, which are consequences of economic and class interests, political and social power, and culture created by societies' historical evolution. Tesh (1988), although referring to theories of disease (and consequent preferences for different prevention policies) rather than definitions of health, relevantly identified the importance of understanding these "hidden arguments" that underlie definitions and theories. Wallace (1994), with special reference to the health of older people, also emphasized the need to begin with definition as the basis for the assessment of factors. He, like Susser (1974), asserted that values, ethics, and personal preferences affect definitions of health and may differ among health professionals, sick patients, and otherwise healthy nonprofessionals, as might personal and family disease experiences and levels of general health education. He also referred to structural environmental factors: 10

PAGE 21

i I I Health definitions must also be considered in their social, political, and economic contexts. Health state perceptions may be highly conditioned by the presence or absence of war or famine, political or social unrest, or threats from the natural or man-made environment. Health perceptions may also vary according to access to special helping programs, or adherence to a particular ideology, either political or religious." (p. 449) Health then, for any age, is a complex construct whose definition depends on historical, social-political-economic, and philosophical perspectives as well as clinical ones. Historical Perspectives Historically the level of knowledge of biological systems has affected concepts of health. The Greeks, who defined biological processes in terms of bodily humors, believed in health as a state of equilibrium and understood that health depends on one's heritage, economic situation, physical surroundings, and political/social environment {Hippocrates, n.d.). The uncertainties and fears of the Dark Ages displaced these rational perceptions of health, replacing them with beliefs that man had no control over threats to health, which were considered to be Divine punishment. The view of health (and nature in general) as Divine plan faded (at least in Western cultures) with the rise of Cartesian mechanistic views of the body as a machine. As Dubos (1959) wrote, "the myths ofHygeia and Asclepius symbolize ... two different points of view .... For the worshipers ofHygeia, health is the natural order of things, a positive attribute to which men are entitled if they govern their lives wisely .... More skeptical or wiser in the ways of the world, the followers of Asclepius believe [in treating] disease, to restore health by correcting any imperfection caused by the accidents of birth or of life" (p. 131 ). By the early 11

PAGE 22

nineteenth century, hospital-based physicians examined disease in specific organs, not the health of the person as a whole. The discovery of the germ theory of disease, emphasis on laboratory science, and the Flexner report's scientific orientation to medicine reinforced the health-as-absence-of-disease perspective (Jones & Moon, 1987). From the mid-nineteenth until at least the mid-twentieth century, in the United States, advances in biomedicine and the position of professional sovereignty held by physicians focused attention on clinical intervention to cure or mediate specific medical conditions (Starr, 1982). With market changes in recent years, vesting greater authority in payers and insurers and less in providers of care, health increasingly has been viewed as commodity and health care as a business. Philosophical Models of Inguiry Philosophical models of inquiry provide tools for exploring the multiple levels of factors that constitute health. Models range from a positivist belief in a single, knowable, fact-based reality "driven by immutable natural laws and mechanisms" (Guba, 1990, p. 20); to an interpretive-constructivist vision of reality as multiple mental constructions whose form and content depend on the social and experiential base of the persons who define them; and further to a critical view that ideology and prevailing political power determine, often falsely, what people see as reality. In : I health terms, a strictly positivist, biomedical definition restricts evaluation to clinical I indicators of changes in disease states related to medical inputs. The interpretive constructivist view includes social and cultural factors related to wellness. A political-economic (critical) construct of health declares that structural, population based variables such as income differentials among social groups, access to political power, and environmental degradation determine health-related quality oflife. Kleinman (1988) clarified the distinctions among these constructs, but from a perspective of negative rather than positive health states. He described disease as an 12

PAGE 23

. I I : I I I : I entity circumscribed by biological structure or functioning; illness as "the innately human experience of symptoms and suffering ... how the sick person and the members of the family or wider social network perceive, live with, and respond to symptoms and disability'' (p 3); and sickness as the broader, population-level "understanding of a disorder in its generic sense across a population in relation to macrosocial (economic, political, institutional) forces ... seeing it as a reflection of political economic and other social sources of human misery" (p. 6). Frankenberg (1988) summarized the temporal and social relationships among these levels of challenges to health. He identified disease as a condition without temporal boundaries, "a disturbance of body functions and performance seen in biological terms of the kind that physicians are primarily trained to detect, diagnose and treat'' (p. 16). Disease made personal by the sufferer's present-tense experience becomes illness, and sickness ''provides the social and cultural framework ... to analyze the social consequences of illness and disease" (p. 1 7). These constructs of challenges to health lead to definitions of health as the absence of something. The empiricist or positivist biomedical model focuses on disease resulting from the detrimental effects of a clinically-defined agent on the host, whose genetic character determines its susceptibility. Health in this context means absence of disease, or as Doll (1992) defined it, "a state distinguished by the absence of disease or of physical or mental defect, that is, the absence of conditions that detract from functional capacity whose incidence can be measured objectively" (p 933). In the broader context of absence of illness, health becomes an interpretive constructivist concept of wellness. This construct has substantial subjective components-individual, social, and cultural. Dubos (1959) stated that health (and disease) cannot be defined just in terms of biomedical attributes but must be measured 13

PAGE 24

I I by "the ability of the individual to function in a manner acceptable to himself and to the group of which he is a part" (p. 261 ). He continued further, Men naturally desire health and happiness. For some of them, however, perhaps for all, these words have implications that transcend ordinary biological concepts. The kind of health that men desire most is not necessarily a state in which they experience physical vigor and a sense of well-being, not even one giving them a long life. It is, instead, the condition best suited to reach goals that each individual formulates for himself. Usually these goals bear no relation to biological necessity; at times, indeed, they are antithetic to biological usefulness. More often than not the pursuit of health and happiness is guided by urges which are social rather than biological; urges which are so peculiar to men as to be meaningless for other living things because they are of no importance for the survival of the individual or of the species. (pp. 278-279) Jones and Moon (1987), combining interpretive and critical viewpoints, explained this view of health as a social and moral judgment that varies according to the society's norms, expectations, and culturally shared rules of interpretation. A strictly critical perspective views health as the absence of sickness, which incorporates those societal and political conditions that create harmful environments. Environment includes social-psychological, political-economic, and historical-cultural influences as well as physical conditions. lllich (1976), from this perspective, explained that '"health,' after all, is simply an everyday word that is used to designate the intensity with which individuals cope with their internal states and their environmental conditions" (p. 7). In the same vein, Gil (1993) defined health as not only the absence of specific diseases but also the ability to develop and function according to one's innate physical, intellectual, emotional, and social potentials without blockage or waste from the environment, in order to satisfy one's intrinsic needs and enrich society. He felt that if society's established ways thwarted the individual's fulfiJJment, then constructive developmental energy would become destructive energy resulting in physical, emotional, intellectual and social sickness. 14

PAGE 25

Labonte (as cited in Eisen, 1994), noted that the ''prerequisites to health are no longer simply disease prevention, or 'proper' lifestyles, but include peace, shelter, education, food, income, a stable ecosystem, social justice and equity'' (p. 23 7). To this list Link and Phelan (1996) added knowledge, money, power, prestige, and social connections, all culturally constituted by the community. As Liang (1986) pointed out, although these points of view treat health as a residual category defined primarily by absence of disease, illness, or sickness, health incorporates more than that. Dubos ( 1959) explained further that solving problems of disease is not the same thing as creating health and happiness. This task demands a kind of wisdom and vision which transcends specialized knowledge of remedies and treatments and which apprehends in all their complexities and subtleties the relation between living things and their total environment. Health and happiness are the expression of the manner in which the individual responds and adapts to the challenges that he meets in everyday life. And these challenges are not only those arising from the external world, physical and social, since the most compelling factors of the environment, those most commonly involved in the causation of disease, are the goals that the individual sets for himself, often without regard to biological necessity. (p. 26) Jylha (1994) explained the difference between health and disease this way: "constructs of health and disease differ in a profound, qualitative sense: definitions of disease are formal while definitions of health are social and contextual" (p. 985). Lewis (1953) suggested that the dualist language of health and disease is unavoidable and "the fictions, health and disease, serve a useful intellectual purpose, though we know they refer merely to uplands and lowlands in a continuously graded and terraced country'' (p. 11 0). The point of exploring factors related to positive health, rather than focusing on negative risks that challenge it, is to view the range from the healthy end rather than the more usual obverse. Antonovsky (1987) called this focus salutogenesis and framed his inquiries into contributing factors in this way: 15

PAGE 26

; I : i I I A salutogenic orientation (which focuses on the origins of health) poses a radically different question: why are people located toward the positive end of the health ease/dis-ease continuum, or why do they move toward this end, whatever their location at any given time? (p. xii) He characterized his work with the metaphor of a river as the stream of life, observing that many people choose to jump into the river while refusing to learn to swim, or in terms of health, while not adopting beneficial health behaviors He then asked, "Wherever one is in the stream-whose nature is determined by historical, social cultural, and physical environmental conditions-what shapes one's ability to swim well?" (p. 90). Definitions of Health The challenge is to construct a definition of health as a positive attribute or condition, a definition that incorporates the multiple perspectives outlined above and specifies factors that might be amenable to intervention. It is not an easy task. An operational definition of health "has become increasingly difficult as the emphasis of medical and health care has shifted from decrease in mortality and increase in longevity to improvement in the health-related quality of life" (Bergner & Rothman, 1987, p. 191). Such an inclusive definition of health must contain more than the absence of specific biomedical conditions. For example, the World Health Organization defined health as "complete physical, mental, and social well-being, not merely the absence of disease or infirmity'' (Breslow, 1990, p. 9). Most conceptual constructs of health would include "some aspect of physiological or biological status, mental state, physical and social functioning, and health behaviors and attitudes" but probably "different views on the parts of the construct that contribute most to overall health" (Bergner & Rothman, p 192). 16

PAGE 27

I I I Concepts of Health Many have tried to capture the concept in words. Liang (1986) focused on physical health and suggested three major approaches, which encompass the positivist and interpretive-constructivist perspectives. A medical model or physical definition speaks of health as "a residual category defined by absence of disease" (p. 248). In a functional model or social definition, health is "a state of optimum capacity for the performance of one's roles and tasks for which the person has been socialized," and ''physical health is equated with conformity to norms of physical and mental capacity for adequate participation in social activities" (p. 249). (References to the ability to fulfill social roles appear repeatedly in attempts to define health.) A psychological model incorporates a subjective evaluation of health as "the individual's perception and evaluation of his or her overall physical health ... which represents a summary statement concerning the ways in which various aspects of health, subjective as well as objective, are combined within one's perceptual framework" (p.249). Murray, Dunn, and Tarnopolsky (1982) provided similar psychological definitions but with a focus on the psychoanalytical. First, a traditional medical psychiatric definition views healthy people as the residue when those with manifest pathology have been identified. A psychoanalytic view defines health as the achievement of optimal integration of the individual. Finally, a sociopsychological definition again includes the optimum capacity to perform effectively the roles and tasks for which the individual has been socialized, i.e., to meet expectations. Murray et al. considered health to be interchangeable with normality, which provides the standard against which to judge the abnormaL It is not a new concept. Centuries ago Chaucer counseled, "Ocupye the meene by stydefast strengthes, for al that ever is undir the meene or elles al that overpasseth the meene despiseth welefulness" (Boethius, iv, in prose 7). 17

PAGE 28

Mmray et al. ( 1982) noted the difficulty of establishing such a standard. Since no objective criteria of health have been established, they wrote, health-normality must be placed in the context of human variability, with reference to the individual's material and social environment. For example, an athlete's perception of disability from impaired mobility would differ from that of an older, more sedentary person. J.G. Evans (1984) cautioned that although the traditional conception of aging presumes normal and pathological/disease processes, there is an .. absence of any theoretical or practical means of defining 'normal' aging [and] certainly no grounds for assuming ... that what occurs 'normally,' i.e. commonly. in Western societies is necessarily 'normal; i.e. healthy or optimal" (p. 355). Focusing on health as one's level of functioning, Patrick, Bush, and Chen (1973) considered two components: function at a point in time, evaluated in terms of social preferences for various functional levels, and the expected transition to other, more or less favorable, levels in the future. They defined optimum function, i.e. health, as the ability to conform to society's standards of physical and mental well being and to perform the activities usual for the individual's age and social role. In contrast to Liang (1986) cited above, these authors emphasized performance, not capacity. Susser (1974) too believed that health has multiple dimensions that include individual-level organic (disease, impairment) and functional (illness, disability) components as well as social levels of social dysfunction, sickness, and handicap. Bergner (1985) added health potential to the definition and identified five dimensions of health: genetic or inherited basic physiological structure; biochemical, physiologic, or anatomic condition (disease state, disability or handicap state); functional condition (social-role, physical, and cognitive performance); mental condition (mood or feeling state, affective state); and health potential (longevity, functional potential, disease and disability, disadvantage, prognosis). She further listed factors that affect health status, which include societal issues such as 18

PAGE 29

environmental quality, housing, crowding, and sanitation; the availability of and access to health care; social and familial issues including personal health attitudes and behavior, resources, and the physical condition of those close to individual; and personal issues related to personal health care, social network, coping skills, and available resources. Feinstein (1993) discriminated between two types of personal characteristics that may cause differences in health status: non-resource-dependent behavioral characteristics including psychological, genetic, and cultural factors and resource-dependent characteristics such as wealth and material ownership. One study actually asked people aged 60 and over "What do you think most 1 people around your age mean when they say they are in good health?" 1993, p. 343). From a selection of responses, over 40 percent chose the ability to perform usual activities, about one third selected a definition of good health as a general feeling of well-being, and fewer than 20 percent chose the absence of symptoms. I .I I Quality of Life According to Wallace (1994), conventional interpretations ofhealth status have a positivist flavor, with their focus on clinical processes (e.g., signs and symptoms, diseases, mortality), alterations in anatomy and physiology, and individual functional attributes. He suggested that definitions should also embrace broader measures of the quality of life, incorporating a range of clinical, social, and psychological health constructs. Additionally, he continued, they might include prenatal and birth events, features of human development and maturation, an assessment of culturally conditioned psychological and behavioral factors (e.g., abstract cognitive and social abilities, performance in typical environmental settings, response to challenges), and hygienic and preventive behaviors. He added an additional level, environmental factors-physical, biological, and social-as well as potential services and resources in the community and the medical care system. 19

PAGE 30

Edlund and Tancredi (1985) enumerated five possible definitions of quality of life, pointing out what they believed to be the ideological underpinnings of each (their hidden arguments, in Tesh's, 1988, terms). First, quality of life as fulfillment of personal goals, based on telic theory, mirrors the American cultural emphasis on self actualization. A second definition, the ability to lead a normal life, reflects the political popularity of normalcy in times of great change. The ability to lead a socially useful life, their third definition of quality of life, depends on the size of one's sociological frame and might be restricted to political and economic concerns with employment or expanded to include social and personal roles. The fourth definition, that of the rational man, assumes the ability to generate objective, expert criteria against which to gauge individual status. Dangers exist when the experts' underlying assumptions are obscured. Finally, the individualistic view defines quality of life as whatever an individual personally declares it to be. Consonant with democratic choice valued by American culture, it also brings risks of unchecked individualism, with values that may run counter to those of the general community. Diener (1984) reviewed psychological theories that attempt to explain what contributes to well-being (or happiness, which he equated with subjective well-being). Telic, or endpoint theories, assert that people gain happiness when they achieve their goals. Maslow's (1970) work on motivation exemplifies this category. He suggested levels of needs that, when fulfilled, contribute to health-as-well-being. Dependent on the environment but focused more on the individual, his hierarchy begins with biological-physiological-material needs for the basic resources for existence and progresses to the needs for safety and security, love and belongingness, esteem based on one's own productivity and creativity, and self-actualization or self-fulfillment. I Diener cited a related set of theories that also explain well-being in terms of goal achievement, but only to the extent that success has overcome an existing deficit or pain. This point of view suggests that permanent fuJfiJJment of all needs precludes 20

PAGE 31

, I complete satisfaction or well-being. Activity theories maintain that well-being depends on the activity involved in reaching a goal rather than attainment of the goal itself. Diener distinguished between bottom-up and top-down theories. Bottom-up theories maintain that well-being or happiness is an accumulation of small pleasures, tallied by the individual. Top-down theorists believe that global personality features influence an individual's reaction. Cognitive association, the ability to associate current emotional situations with cumulative past experience, may help explain the development of essentially happy or unhappy temperaments, effectively linking bottom-up and top-down perspectives. Judgment theories postulate that comparisons between actual conditions and not-necessarily-conscious standards determine well being. Any deviation from normal will therefore change an individual's state of wellbeing. Farquhar (1995a) also reviewed the literature on quality of life. As in the I 1 descriptions of theories above, she determined that global definitions usually incorporate judgmental or cognitive experiences of (dis )satisfaction and affective experiences of(un)happiness. Quality of life includes both conditions of life and the experience of life. She found references to the possession of adequate resources to satisfy needs, participation in activities that lead to self-actualization, and satisfactory comparisons between self and others. Farquhar's review contributed a useful listing of specific components that contribute to quality of life. They include general health and functional status, level of activity, comfort, mental state and longevity, socioeconomic status, subjective evaluations of life satisfaction and self esteem, and, especially for older people, "concepts of privacy, freedom, respect for the individual, freedom of choice, emotional wellbeing and maintenance of dignity" (p. 504). Expanding the definition ofhealth to incorporate quality of life complicates the exercise. By its nature, quality of life is a subjective assessment of well-being rather than a more objective enumeration of clinical, functional, or even social 21

PAGE 32

I i attributes. The primary difficulties, however, are of measurement but not of definition; they do not diminish the benefits of the broader perspective that enriches our understanding of the construct of health. One additional warning may be useful. Robertson and Minkler (1994) cautioned that viewing health broadly, as with the World Health Organization's definition ofhealth that refers to ''the extent to which an individual or group is able, on the one hand, to realise aspirations and satisfy needs; and on the other hand, to change or cope with the environment'' (p. 298), may lead us to minimize the individual's everyday reality of pain, discomfort and difficulty. Models of Health The previous subsections of this chapter explored concepts of health and well being. Some theorists and researchers have taken another approach to defining health, by constructing models. Before investigating these models ofhealth, one should note Kasl's (1983) cautionary point: the development of broad principles of human functioning can generate beautiful, complex constructs that presume a multi-factorial whole when they are actually composed of quite discrete heterogeneous elements. For example, social support comprises a number of components: breadth and size of network; amount of support given, received, and perceived; and type of support (e.g., instrumental, emotional, informational). Echoing Robertson and Minkler's (1994) concern about too broad definitions of health as well-being, Kasl warned that combining components into a theoretical whole may mask important individual effects and interactions. These thoughts suggest that useful models of health should incorporate factors and interactions with the following characteristics: they are specific and conceptually clear; they are related to individuals and their well-being; they can be measured and assessed; and 22

PAGE 33

they are amenable to intervention, at least distally. Theorists and researchers have constructed a variety of models to describe the relationships between and among the various components of health. Barsky, Cleary, and Klerman (1992) noted that "most patients experience their health as a global experience and a level of function, as an overall state of well being" (p. 1147), in a model that incorporates a number of factors. The factors include medical status; functional impairment; psychological dysfunction and emotional distress; social factors, role demands, and stressful life events (e.g., social isolation, adverse life events, unemployment, dissatisfaction with one's life circumstances); and age. Boult, Kane, Louis, Boult, and McCaffrey (1994) diagrammed the relationships between chronic conditions and disability. They adapted World Health Organization and U.S. Academy of Science's Institute ofMedicine models, suggesting a sequential relationship--from chronic medical conditions to impairment of abilities, to functional limitation, and finally to disability-but also observing that : I "disability ... may be caused, not only by physical or mental limitation, but also by cultural expectations, environmental obstacles, or lack of motivation and training" (p. M28). Blaum, Liang, and Liu (1994) also hypothesized an interaction among chronic diseases, disability, and self-rated health status, controlled for exogenous factors of age, gender, race, education, and social isolation. They proposed that these three components combine to create physical health status and that all three separately and together affect utilization of health services. They referred to the Anderson Health Behavior Model of Utilization that, although focused on utilization, offers insights into the multiple factors that may affect an individual's sense of well-being. The factors include predisposing characteristics (e.g., health beliefs and attitudes), enabling characteristics (e.g., insurance and socioeconomic status), and need characteristics (e.g., disability, health status, disease; Anderson & Bartkus, 1973). 23

PAGE 34

. I I Wolinsky, Callahan, Fitzgerald, and Johnson (1992) explained these constructs more fully: predisposing refers to the propensity for using services, dependent on sociocultural characteristics (e.g., demographics, social structure, health beliefs); enabling encompasses the (economic) means for obtaining services; and need refers to the perception of illness or its possibility, involving both the individual's assessment of the amount of illness and professionally evaluated need. The preceding authors focused on disability rather than on health, but they suggested factors that belong in any model of health: physical or medical condition, functional ability, demographics, social influences, health beliefs, perceived well being, and socioeconomic status. R.G. Evans and Stoddart (1990) proposed a model of health that includes and expands upon these factors, based on their observed need for a framework to describe the complex causal patterns among components of health and also between health and health care. They suggested important features of such a framework, which they asserted should accommodate distinctions among disease, as defined and treated by the health care health and function, as perceived and experienced by individuals, and well-being, a still broader concept to which health is an important, but not the only, contributor [and include] consideration of both behavioural and biological responses to social and physical environments. (p. 1362) Building on the positive and negative biomedical and economic interactions between health care and disease (and so still, at least initially, focused on disease rather than health), they incorporated the following components, in subsequent stages: lifestyle (in terms of individual behavior), environment, and human biology as contributors to disease; an expansion of lifestyle and environment to acknowledge the impact of social and physical environment on lifestyle, or rather on individual response, which in tum evolves from behavior and biology (genetic endowment); 24

PAGE 35

a feedback loop from individual response to health and function. to well being, and back to individual response; and the introduction of prosperity as a measure of economic trade-offs and their impact on well-being and the individual's environment. Figure 2.1 shows a graphical representation of the final stage of their model, adapted from their diagrams. They cautioned that such a representation oversimplifies the concepts: "The entities which form the components of our framework are themselves categories, with a rich internal structure" (p. 1349). They explained that each box represents a complex concept that in general cannot be adequately represented by a single homogeneous variable and added that it may be interactions between factors from different categories that are critical to an individual's or a population's health. need cost Health& Function l Disease .. f:: Health Care Individual Response Betavior Biology + .. Lor r Well-Being I' Prosperity Social Environment ... Environment .. r Genetic ... Endowment Figure 2.1. Representation ofR.G. Evans and Stoddart's Model ofHealth Wilson and Cleary (1995) constructed a more sequential conceptual model of health, or taxonomy of patient outcomes, that categorizes measures according to the underlying health concepts they represent and suggests causal relationships among 25

PAGE 36

I I : I Figure 2.2. Wilson and Cleary's Multifactorial Model of Health Reprinted. by permission, from Journal of the American Medical Association, 1995, 273( 1 ) : 60 1995, American Medical Association them. Figure 2.2 shows the relationships they described Their model contains five main categories: biological and physiological factors, symptoms, functioning, general health perceptions, and overall quality of life. It also acknowledges the importance of individuals' characteristics, including personality and values, and characteristics of the environment including nonmedical factors. Moore, Van Arsdale, Glittenberg, and Aldrich ( 1980) proposed a model of the human ecosystem to explain the multiple interacting factors that affect health. Their focus was on disease and biological capacity, but the model has broader applicability. It consists of two subsystems, the first concerning the macro level environment (structural environmental characteristics) in which the individual lives and the second the individual. Health in this model means equilibrium, which individuals achieve by coping with and repairing the disequilibrium, or sickness, that has occurred when 26

PAGE 37

. I I I their adaptive capacities have been threatened and exceeded. The dynamic, context specific processes that produce sickness and restore health evoke behavioral responses that, "insofar as they express and reinforce an individual's attitude toward self and the world, ... very definitely influence the individual's adaptive powers" (p. 24), which then may affect future coping and repair responses as well as the macrolevel environment. Operational Definition of Health Based on the concepts and models described above, an operational definition of health or healthiness (which must precede explorations into determinants of healthy aging) should include, in Farquhar's (1995a) felicitous phrase, both conditions and experiences of life that include the following: sociodemographics; biological, physiological, or anatomic status, including genetic background and the absence of disease or disorder; physical functional abilities; level of activity; cognitive abilities; affective state; absence of depression; personality traits, e.g., optimistic or pessimistic attitude; ability to fulfill one's social role and to receive respect and dignity; health behaviors and attitudes; satisfaction and happiness; achievement of goals; self-actualization; worthwhile activity; self-esteem; favorable comparisons with context-dependent normality or standards; social environment; social and familial resources and challenges; and physical, economic, and political environment, including access to health care. 27

PAGE 38

Categories of factors alone do not provide a definition. That requires an understanding of the relationships among them. The models proposed by Boult et aL (1994) and Blaum et al. (1994) included the first (physical and functional) categories. Blaum et al. added beliefs about health and interactions with perceived health. R.G. Evans and Stoddart's (1990) model introduced the complexity of relationships between the individual and the environment and the interactions between those relationships and disease. Wilson and Cleary (1995) expanded that model, introducing characteristics of the individual (personality, motivation, and values) anj environmental supports. They also offered a simpler model, suggesting a sequential progression of health status outcomes from strictly biomedical concerns to the broadest construct of quality of life. The ecological model from Moore et al. ( 1980) placed the individual's equilibrium-disequilibrium processes in an environmental context and included important interactive relationships among that process, the environment, and the individual's behavioral responses. None of the models described here gives a complete definition of health.. but taken together they provide an understanding that health is a dynamic, context! specific process involving individual and environmental physical, social, and psychological characteristics, challenges, and resources. Possible Determinants of Health The preceding attempt to define health established a framework for exploration into specific factors that contribute to health and healthy aging. The following subsections present a variety of such factors, drawn from the literature. It is not intended to be all-inclusive or restrictive but to give structure to the exploration. 28

PAGE 39

Sociodem.ographic Factors Age-getting older-is the most obvious factor that might affect health in older people. Other possible relevant sociodemographic characteristics include gender, race or ethnicity, and socioeconomic status (for which education and income often serve as proxy measures). Marital status and living arrangements too may reflect socioeconomic status and may also provide information about social support. A later section in this chapter reviews previous studies and provides more complete references to these factors. Physical and Functional Status Physical status refers primarily to biomedical symptoms and signs that indicate the presence or absence of acute illness, chronic disease, and any other underlying physiological conditions. Although not often cited in studies of the risks associated with aging (again, a later section provides more complete information), pain too may be an important factor. Functional status has a number of components-physical capability, the ability to care for oneself. and cognitive and affective status. Physical capability includes performance (e.g., the ability to climb stairs) and mobility. Both basic (e.g., bathing, eating) and instrumental (e.g., cooking, taking medication) skills determine how independently one can live. Dementia and depression, as well as other cognitive and affective disabilities, have substantial impacts on functional status. Quantitative assessment of specific functional limitations may not, however, give a complete picture of effects on well-being, or more global perceptions of health. As Johnson and Wolinsky (1993) these distinctions become apparent when identically disabled individuals "demonstrate differential adaptation when confronted with the functional demands of mobility, personal hygiene, and housekeeping" (p. 29

PAGE 40

: I I I 109). As one way of assessing ability, their study asked respondents to compare their activity levels with those of their peers. Berg, Hallauer, and Berk (1976) found that asking people to place values on various abilities and activities provided important information about perceived quality of life. They asked respondents to select from a list the functions they valued most or would feel worst about not having. Examples included abilities to do the following: walk, see, and read have friends, maintain contact with family and friends love and be loved think clearly, use mental abilities not be depressed make choices (in day-to-day activities) participate in particular activities control body functions be free of pain On average, respondents most valued the abilities to use mental abilities, see, think clearly, love and be loved, make decisions, live at home, walk, maintain contact with family and friends, and talk. Social and Temporal Comparisons of Status As Patrick et al. (1973) pointed out, perceptions of health status involve expected transitions as well as levels at a point in time. Perceived status may then be comparative rather than absolute Social Comparisons and Normalcy. Mechanic (1978) raised the issue that people evaluate their health status, in part at least, by comparing it with normal, a construct whose definition depends on the social group. They may define normalcy in positivist terms with the ideal as a point of reference, in statistical terms as an 30

PAGE 41

: I I : I I I observation of what others do, or with a social-reaction expressing a social reaction to difference and social identity. Suls, Marco, and Tobin (1991) discovered that older people who mentioned others worse off than themselves tended to rate their own health more positively. They proposed that "the elderly compare themselves with a cognitively constructed stereotypical standard of the frail elderly rather than with a specific other" and because few "actually 'fit' this stereotype, most elderly feel they are doing well" (p. 1125). Temporal Comparisons Suls et al.'s (1991) study also found that those who mentioned thinking about past or anticipated health gave more negative health ratings. Kind and Dolan (1995) explored the influence of past and/or current illness experience on the evaluation of health status. They found no difference between past experience of ill health and of better but they discovered that those in currently poor health (and valuing their health as low) selected a more limited range of values than those in better health and less often chose better status categories. Challenges All sorts of physiological, psychological, social, and environmental factors can challenge health. The literature suggests that the nature of the issues, the way individuals respond to and the resources they have available to deal with them all affect the extent to which people feel healthy. Physical. Cognitive. and Functional Factors. Disease, dysfunction, disability, and pain-the signs and symptoms of physiological disorder and functional deficitthreaten well-being. Schulz and Williamson (1993) pointed out that the nature of the onset of illness, the perceived prognosis, and the visibility of the disability may also affect perceptions of health. Depression and other forms of mental illness both challenge health and make it more difficult to respond effectively. Culture and socialization within a culture affect responses to these challenges. Zborowski (1952) 31

PAGE 42

found that Jewish and Italian patients in a New York hospital responded to pain emotionally, "old Americans" appeared stoical, and Irish patients tended to deny the pain. Kleinman's (1988) illness narratives told similar stories about different cultural responses to psychiatric disorder. Childhood. Social background, or social history, may help to determine how people perceive and evaluate their health and may create threats to well-being or provide resources to support it. Mechanic (1972) proposed that the responses given to illness and symptoms dwing childhood affect later interpretations. Barker (1994) elaborated that people tend to retain the understanding of illness and treatment that were prevalent when they were young, because ''beliefs, practices, and behaviors of individuals around health and illness depend on the group's fundamental values and worldview" (p. 10). Older people, when their values conflict with medical values because of the nature of their more prevalent chronic illness patterns, may suffer when they desire palliation and support more than clinical medicine. Worries. Health Troubles. and Control Over Health. Murray et al. (1982) found that ''preoccupation with, and expressed concern about health were consistently related to a poor self-rating [of health]" (p. 376). Rakowski and Cyran (1990) also mentioned worry due to health, as did Wolinsky and Johnson (1991). Fillenbaum (1979) suggested the influence ofhealth troubles that get in the way of what people want to do. The feeling of having control over one's health, now or in the future, may provide a helpful response to worries about health (Mechanic, 1972; Rakowski & Cyran; Wolinsky & Johnson). Isolation. Isolation leads to loneliness, loss of personal integrity, and the loss of connection with other social resources. It can be social or emotional. Quantity of daily contacts, marital status, living arrangement, and number of companions and confidants determine social isolation. Emotional isolation concerns the quality of relationships-it is possible to be lonely in a crowd. Chappell and Badger (1989) 32

PAGE 43

found that emotional isolation is more important to well-being than social and that the lack of a confidant or companion is more important to well-being than unmarried status or living alone. Environment. As discussed earlier, critical theorists identified social psychological, political-economic, and historical-cultural influences as components of the environment that affect health. Specifically, they named ''peace, shelter, education, food, income, a stable ecosystem, social justice and equity" (Labonte, as 1 cited in Eisen, 1994, p.237) as well as knowledge, money, power, prestige, and social connections (Link and Phelan 1996). Awareness and Focus. The tendency to make comparisons indicates an awareness of self or condition, which may have positive or negative effects. Others have used the term appraisal to mean much the same thing (Quayhagen & Quayhagen, 1996). Mechanic (1978) considered social comparison to be one aspect of adaptation. All of them dependent on the social system, the other aspects include the search for meaning, attribution of stress or change to a specific cause, level of dependence, and power. In a later work he explained: The fact that people cope much of the time without awareness is a central point in understanding personal and social adaptation. To become aware, to become self-conscious, is an indication of more than a routine problem, a greater challenge, a break in the flow of normal activity .... But it is not psychologically economical to worry about what one cannot predict or control, and individuals maintain a sense of invulnerability by inattention to potential threat. (Mechanic, 1986, p. 5) Mechanic believed that introspection or awareness, which is conditioned by sociocultural factors and childhood socialization, increases prevalence of symptoms and negative self-evaluations, causes more distress and greater upset from stressful life events, generates increased health care utilization, and appears to exaggerate the experience of distress and illness. 33

PAGE 44

Personal Resources and Responses Adaptation. Cobb (1976) explained the difference between adaptation and coping. Adaptation refers to individuals' abilities to change themselves to improve the person-environment fit. Coping, on the other band, depends on managing the environment rather than oneself. Mechanic (1978) described social adaptations to illness as a coping process: lllness, illness behavior, and reactions to the ill are aspects of an adaptive social process in which participants are often actively striving to meet their social roles and responsibilities, to control their environment, and to make their everyday circumstances less uncertain and, therefore, more tolerable and predictable. (p. 1) He continued further that health and disease are not entities but processes of adaptation to life's changing demands and the changing meanings people give to life. Adaptive resources include material resources, appropriate skills, adequate defenses, social supports, and sustained motivation. From a critical perspective, Dlich (1976) stated that "in part at least, the health of a population depends on the way in which political actions condition the milieu and create those circumstances that favor self-reliance, autonomy, and dignity for all, particularly the weaker," and he continued, "in consequence, health levels will be at their optimum when the environment brings out autonomous personal, responsible coping ability'' (p 7). Fries (1980) observed that "the older person requires opportunity for expression and experience and autonomy and accomplishment [rather than) support and care and feeding and empathy" (p. 135). Attitude. Borawski, Kinney, and Kahana (1996), exploring congruence between health appraisal and so-called objective measures of health status, identified four attitudinal categories. Good-health realists and poor-health realists made appraisals that were congruent with observed clinical characteristics. Health pessimists rated their status lower than indicated by clinical observations; health 34

PAGE 45

. i optimists rated theirs higher. When asked why they rated their health as they di
PAGE 46

, I I I [concerning the] extent [to which it is] defensible to make aging individuals contented by means of decreasing their different aspiration levels" (p. 270). Peterson, Seligman, and Vaillant (1988) proposed and have continued to study the hypothesis that a pessimistic explanatory style, which corresponds substantially with an internal locus of control, increases the risk of physical illness. By explanatory style they mean habitual way in which people explain the bad events that befall them" (p. 23). Helplessness theory, on which they based their hypothesis, distinguished three dimensions of explanatory style: internality (e.g., "It's me" vs. "The road was icy''), stability (e.g., "It's the way the world is" vs. "It was a one-time spill''), and globality (e.g., "It will affect everything I do" vs. "It made me late to work that day''). According to Peterson et al., the individual who explains bad events pessimistically, "with stable, global, and internal causes shows more severe helplessness deficits than a person who explains them with unstable, specific, and external causes" (pp. 23-24). Those deficits include passivity, pessimism, and low morale, which "foreshadow disease and death" (Peterson & Seligman, 1987, p. 237). Perceived self-efficacy may serve as a counter to helplessness. Bandura : I (1977) defined self-efficacy as ''the conviction that one can successfully execute the behavior required to produce the [desired] outcomes" (p. 193). Self-efficacy, according to a later work, not only affects willingness to initiate coping behavior but also determines the amount and duration of effort that people will expend in the face of obstacles or aversive experiences. Perceived self-efficacy alone, however, cannot prevent feelings of futility and helplessness. People may feel fully competent of their capabilities but "give up trying because they expect their efforts to produce no results due to the unresponsiveness, negative bias, or punitiveness of the environment" 1982, p. 140). Coping. Schulz and Williamson (1993), in a study of the impact of physical frailty on patient and caregiver, constructed a conceptual model that has relevance 36

PAGE 47

: I beyond that relationship. After discussing conditions that may generate patients stress (physical and functional signs and symptoms), they listed conditioning variables that might intervene between stressors and outcomes. The variables include economic resources, personality attributes (optimism, perceived control, and self esteem), social support, and coping strategies. Pearlin (1989) explained coping's mechanisms. Coping, he wrote, serves to change the situation from which stressors arise, manage the meaning of the situation in a manner that reduces its threat, or keep the stress symptoms within manageable bounds. lndq>endence. Independence incorporates both autonomy, or the ability to set and follow one's own rules, and the ability to act. J.G. Evans (1984) offered an example related to mobility: being able to choose when and where to go sustains autonomy, but full independence is constrained if one needs assistance with actually moving. He suggested that autonomy may be the more useful global objective for older people. Sense of Coherence. Antonovsky (1987) wished to understand what predicts good health outcomes and what characterizes those who have them. To answer his questions, he introduced the concept of sense of coherence, which he defined as follows: a global orientation that expresses the extent to which one has a pervasive, enduring though dynamic, feeling of confidence that (I) the stimuli deriving from one's internal and external environments in the course of living are structured, predictable, and explicable; (2) the resources are available to one to meet the demands posed by these stimuli; and (3) these demands are challenges, worthy of investment and engagement. (p. 19) Components of a sense of coherence, according to Antonovsky (1987), include 1) comprehensibility, or the degree to which stimuli make cognitive sense "as information that is ordered, consistent, structured, and clear, rather than as noise" (p. 37

PAGE 48

17) and will be predictable or at least orderable, explicable, and understandable; 2) manageability, or the extent to which one feels slhe has adequate resources to meet the demands of stimuli, as opposed to feeling burdened and overwhelmed or, at the other extreme, under-challenged; and 3) meaningfulness, or significance, or the sense that the challenges one faces are worthy of emotional commitment and investment. To explore sense of coherence, Antonovsky (1987) proposed looking at good deviant cases, individuals who were exposed to risk but suffered no ill effects and who, as outliers, can provide information about their exceptional status. He felt that those who acquire what he called generalized resistance resources-including autonomy, competence, money, ego strength, cultural stability, and social supports can maintain coherence in the face of maladaptive social-relational processes (Antonovsky, 1979). Motzer and Stewart (1996) added these factors first proposed by Flanagan (1982) and Burckhardt, Woods, Schultz, and Ziebarth (1989): physical and material well-being; relations with other people; social, community, and civic activities; personal development and fulfillment; recreation; and independence. Social Support Resources Much has been written about social support and its importance as a resource for dealing with life's challenges. House, Umberson, and Landis (1988), in their : 1 extensive review, defined social support as an element of social relationships that refers to positive aspects such as emotional caring and concern, instrumental (tangible) aid, and information. Its sources can be informal, from friends and family, or formal, from institutions and commercial services. Antonucci and Akiyama (1987) described six beneficial results of support: a confidant relationship, reassurance, respect, provision of care when ill, an outlet for talk when upset, and a source of health information. 38

PAGE 49

I M. J. Stewart (1989) outlined five theoretical perspectives concerning why social support has beneficial effects: Attribution theory: Individuals formulate attributions (often by assigning blame) in order to understand, predict, and control the environment. Fairness matters-recipients deserve help to the extent that they did not cause their and donors feel obligated to the extent that they are not responsible for their surplus of resources. Coping theory: Support affects individuals' abilities to undertake ever changing cognitive and behavioral efforts to manage demands that exceed their resources. Sources of social support provide information about events and stressors; broaden the number of options, resources, strategies, and referrals; provide norms; inhibit maladaptive responses, and provide tangible aid and emotional sustenance. Social-exchange (or equity) theory: Individuals desire to maintain equity in their exchanges. They attempt to maximize their outcomes, but they feel distress when in inequitable relationships and will try to restore equity. Social-comparison theory (see also above): People tend to evaluate themselves and to gather information about their behavior, characteristics, opinions, and abilities by comparing themselves with others. Social support assists the process by supporting appraisal, affirming, supporting esteem, and providing feedback. Loneliness theory (see also isolation as a challenge above): Lack of desired social support leads to a subjective, unpleasant experience derived from a perceived deficiency in social relationships or relational provisions. House et al. (1988) concluded from the studies they reviewed that social support promotes health by fostering a sense of meaning or coherence, facilitating health-promoting behaviors, providing motivation and emotional support, and/or affecting neuroendocrinal responses. Newman, Struyk, Wright, and Rice (1990), studying the role of social support in determining institutionalization, summarized five hypotheses about how social support operates: Main effects hypothesis: Informal or formal support reduces existing risks, with an independent or main effect on risk net of background, health or other relevant factors. 39

PAGE 50

Buffering effects hypothesis: One's support system moderates or buffers the effects of risk factors by improving coping abilities. Social relationships affect well-being only in the presence of stress or challenge. Supplementation hypothesis: Formal support permits informal systems to function better and longer. Accommodating environment hypothesis: Situational variables (in his example, housing characteristics) modify the willingness of the informal support system's members to provide care. Mediation hypothesis: Informal and formal support intervene between the challenge (deteriorating health, in his example) and negative outcomes (e.g., institutionalization). Social support comes from the members of one's social network. Pearlin (1989) explained that while "the social network can be regarded as the totality of the social resources on which one potentially may draw, social support represents the resources that one actually uses in dealing with life problems" {p. 251 ). House et al. I ( 1988) suggested three classes of social support variables. The first two are elements of the social relationship structure and the third concerns function. The first, social I integration, refers to the quantity of relationships, their type and frequency. Social network structure, the second, refers to the properties that characterize the set of relationships. The third class of variables, relational content, refers to the functional nature of relationships, processes that include emotional and instrumental social support, relational demands and conflicts, and social regulation or control. S. Cohen (1988), extracting from network theory, listed characteristics associated with social network structure: size, density (which gives a measure of the richness and complexity of interrelationships within the network), multiplexity (the number of available alternatives), reciprocity, durability, intensity, frequency, dispersion, and homogeneity. B.H. Kaplan, Cassel, and Gore (1977) identified anchorage (the length of the path to others) and reachability as additional attributes. Kaplan et al. also described the interactional properties of support: direction 40

PAGE 51

I I (reciprocity), intensity, frequency, and content or meanings. Content/meanings includes rituals, values and beliefs, shared norms, interpersonal exchanges, a fit between role and dependency, opportunities for closeness, self support, the disposition of negative effects, and social status. Functional types of social support are generally classified as emotional, tangible, and informational. Emotional support provides the sense of being cared for and loved, esteemed and valued. Tangible support offers goods and services, while informational support shares information. All types are part of a network of mutual obligation that both defines and reflects one's place in society (Cobb, 1976). S. Cohen ( 1988) described four types of support: information-based assistance, identity and self-esteem, social influence, and tangible resource. Information-based support provides instrumental information, or at least the knowledge that information will be available if needed. Identity-and-self-esteem support increases feelings of personal control. Social-influence support emphasizes social norms and normative coping behaviors. Tangible resources refer to actual services provided or made available. According to social exchange theory, reciprocity matters (Flaherty & Richman, 1989). Individuals desire to maintain equity in their exchanges, attempting to maximize their own outcomes, but they feel distress in inequitable relationships and will attempt to restore equity (M. J. Stewart, 1989). Reciprocity CZIIl be between two individuals and contemporaneous, between two individuals over a period of time, or more communal and over time (e.g., person A helps person B, who assists person B, who at a later time does something that benefits person A). An equitable exchange balance promotes satisfaction (N. Krause, 1995). Although social support appears to benefit well-being, not all social support generates positive effects, and more is not always better. Studies have indicated that to be beneficial, support must be autonomy-enhancing (Rowe & Kahn, 1987), fair, and reciprocated (M. J. Stewart, 1989). N. Krause (1995) discovered that averting 41

PAGE 52

I negative interaction had a greater positive effect on satisfaction that did the amount of support. In an earlier study, he found that support increased feelings of control up to a threshold level, after which increased support could lead to dependence, entanglement, and decreased feelings of controL Cobb (1976) also cautioned that goods and services may foster dependency but suggested that informational support might not. Health Care Delivery How health care is delivered-quantity, style, and setting-also may affect health, as either challenge or resource. A large number of research projects have studied specialized hospital-based programs, often called geriatric evaluation and management programs. (For an introduction to this very extensive literature, see the review by Applegate, Deyo, Kramer, & Meehan, 1991, and the meta-analysis by Stuck, Siu, Wieland, Adams, & Rubenstein, 1993.) A search of the literature, however, discovered very little concerning outpatient programs. Societal and Environmental Challenges Specific to Aging Bengtson, Burgess, and Parrott ( 1997) described two theories of social gerontology that give structure to the following discussion of environment and socialization. The macro-level political economy theory of aging explores how the interaction of economic and political forces, which constitutes the environment of aging, determines allocation of social resources (e.g., variations in treatment, status of older people). Its emphasis on the effects of social structure and social power may mask the impact of aging on the individual. Social constructionism, on the other hand, examines the social construction of age and aging, exploring the influence of social definitions and structures on individual processes of aging. It takes a micro42

PAGE 53

I I I level view of individual agency and social behavior, specifically attitudes toward aging, within the larger structure of society. Bengtson et al. suggested that "political economy of aging can ... be linked with social constructionist perspectives to point to the ways in which structural forces manage and control the social construction of aging and how old age is experienced" (p. S83). Especially for older people, who are prone to multiple health problems that include psychological and social as well as physical dimensions, the perception of well-being depends on more than just clinical and functional status (for example, Mechanic, 1995; Rickelman, Gallman, & Parra, 1994; Schoenfeld, Malmrose, Blazer, Gold, & Seeman, 1994). Farquhar (1995b) listed the dimensions of quality of life most frequently mentioned by older people: family, social contacts, (physical) health, mobility/ability, material circumstances, activities, happiness, youthfulness, and home : 1 environment. Wan (1986) proposed seven determinants of health particularly relevant j to older people: physiological condition, presence of degenerative illness, mental status, age, sex, functional dependencies (e.g., problems with walking, eating, and toileting), and strength of social support network. Degenerative chronic illness, decreased mental functioning, and limitations to activity have more of an impact on I the health of the elderly, compared to younger populations, than a less robust physiological condition, which many consider to be a natural and therefore less threatening part of the aging process. (Younger populations likely perceive a greater impact from physical deficits because they expect physical vigor.) Active life can be constrained ''not only by physical or mental limitation, but also by cultural expectations, environmental obstacles, or lack of motivation and training" (Boult et al., 1994, p. M28). Older people may also perceive that their care is less intense than that offered younger people because society, and often they themselves, devalue their lack of productivity and consider the expense of care too great for the expected cure or relief. 43

PAGE 54

:I I I I I :I I J. G. Evans (1984) spoke of aggravated aging, the increase in morbidity due to social/environmental hazards/challenges (e.g . poorer housing, insufficient heating, diminished access to geriatric services). The inclusion of environment as a relevant determinant ofhealth suggests its importance but does not describe its breadth or level of importance (W 1986). 1breats to well-being, according to Pearlin (1989). "largely arise from and are influenced by various structural arrangements in which individuals are imbedded'' (p. 241 ). These arrangements correspond to the various environments-physical, political, and social-in the models described earlier. Pearlin suggested that individual experiences derive from interrelated levels of social structure-social stratification. social institutions, and interpersonal relationshipsand he specified age as one of the determinants of social strata. Parsons ( 1951) pointed out that society's expectations and norms concerning individuals roles and status have a substantial effect on health. Health in this context refers to individuals' capabilities to perform adequately the roles and tasks for which they have been socialized, based on their societies' norms (Liang, 1986). The role of significant others (Kasl's, 1983, neutral label intended to encompass the sometimes controversial topics of social networks, social support, and social isolation). belief in one's ability to perform adequately, and congruence between beliefs and societal expectations are then important contributors to health, along with clinical interventions to support and maintain physical and functional capabilities. Rosow (1974) explored socialization to old age (in our culture), with specific reference to negative changes in relationships with society and social institutions. Socialization serves to inculcate both values and behavior in members of a society in order to encourage conformity with social norms. For older people, Rosow suggested, our society has not defined social norms or expectations. Aging members of the society do not have clearly defined roles, rites of passage, positive goals, or 44

PAGE 55

. I markers of successful performance. Old age, then, has become a separate subcultme, surrounded by a different, less supportive environment. Mechanic ( 1978) concurred, describing the social position of the aging as a socially caused negative experience, characterized by isolation, lack of social support. health decrements, lack of respectable identity, and loss of work involvement. Social maladaptation, according to Lewis (1953), occurs when a person's "own version of his social role is not in conformity with society's version: and in so far as each person has many social roles, it is the dominant role, or those which have precedence in the daily organization ofhis activities that are important'' (p. 116). He cautioned further that "involuntary changes of social status will clearly favour conflicts over a man's own conception of his social role and that which society fastens upon him" (p. 116), leading to non-conforming behavior that can but may not lead to illness, which he thought depends to a much greater extent on physiological and psychological dysfunction than on social adaptation. Manton (1989) warned of a greater impact: Given the dimensions of the problem [of an increasingly large disabled elderly population], it is likely that no response will be satisfactory unless fundamental changes in the sociocultural perception of the functioning of elderly people, and the provision of family and other social resources to maintaining that functioning, are developed. (p. 55) Investigating the substantial variation in longevity and rates of decline among older people, Berkman (1988) expressed the need to "identify some behavioral and socioenvironmental conditions that appear to influence the way and rate at which people age and die" (p. 39). She suggested that social status among older people may be determined more by relative standing among peers than by income, education, or occupation, which may decrease in importance as indicators. 45

PAGE 56

Risks and Predictors of CUn)healthy A!Zing: Previous Studies Describing healthy aging and developing a model of health for an older population require an understanding of the relative importance of the relevant factors, including those suggested in the preceding subsections. The literature concerning risks and predictors of (un)healthy aging is dauntingly large. It includes studies of predictors and correlates of mortality, institutionalization, and maintained or decreased levels of function. The studies reported in the literature varied widely in characteristics of the sample (e.g., sample size, frail vs. unimpaired, of all adult ages vs. elderly only, single gender or not, ethnically diverse or not, with various levels of I income or not, institutionalized or community-dwelling), study design (e.g., cross sectional, longitudinal, prospective, retrospective), and analytic methods (e.g., bivariate, multivariate, linear regression, hazard analysis, logistic regression, I i I I I correlation). Making comparisons among them or generalizations from them even more difficult, the studies defined constructs of health differently, chose a wide variety of indicators to measure them, and included (or excluded) different combinations of them. Most models failed to account for a substantial amount of observed variation. Some factors did, however, repeatedly show significant predictive or correlational associations with mortality, institutionalization, function, and perceived health. The following review identifies those factors while noting categories of variables excluded from the various studies. 1 Mortality Almost all of the 52 studies of mortality reviewed here adjusted for older age and male gender. Poorer clinical condition and comorbidities were the next most 1 Citations in the text refer only to a limited number of studies that exemplify the issues and variables. Appendix B contains extensive tables of the reported studies. displaying more complete information about research designs and variables. 46

PAGE 57

I commonly cited predictors or correlates of mortality (Deeg, van van der & 1989; Wolinsky, Johnson, & Stump, 1995). Decreased functional status-usually measured as Activities of Daily Living (ADL), Instrumental Activities of Daily Living (IADL), and/or physical performance (Manton, 1988; Mor, Wilcox, Rakowski, & Hiris, 1994; Reuben, Siu, & Kimpau, 1992}-and cognitive status (Narain et al., 1988) also appeared often. Other variables predictive of mortality included obesity (measured as body mass BMI; Idler & Kasl, 1991), larger number of medications (G. Kaplan, Barell, & Lusky, 1988), and health behaviors such as smoking, alcohol consumption, and little exercise (GA. Kaplan, Seeman, Cohen, Knudsen, & Guralnik, 1987). Individuals' ratings of their health, or perceived health, contributed significantly to 22 of the 38 models that included them (Idler & Kasl, 1991; Mossey & Shapiro, 1982; Wolinsky & Johnson, 1992). Informal social support appeared to protect against mortality: 24 of the 38 studies that included at least one measure other than marital status found a significant effect (Berkman & Syme, 1979; Blazer, 1982). Most of the studies included clinical measures ofhealth, but many failed to include any measme of functioning level, depression or other measure of psychological status, social support other than marital status, health behaviors, or socioeconomic status. Institutionalization Admission to a nursing home, especially for longer periods of time, generally signals increased frailty. The 46 published studies reported here found that increased age (Branch & Jette, 1982; Shapiro & Tate, 1985) and decreased functional and/or physical abilities (M.A. Cohen, Tell, & 1986; Reuben et al., 1992) predicted or correlated with institutionalization, as did clinical indicators such as stroke (Teresi et al., 1989). Cognitive impairment (Coughlin, McBride, & Liu, 1990; Shapiro & Tate, 1988) and other forms of mental impairment (Branch & Jette, 1982) 47

PAGE 58

contributed significantly to the risk of ntrrsing home admission. Being White (Greene & Ondrich, 1990; Liu, Coughlin, & McBride, 1991) and female (Shapiro & Tate, 1988) increased overall risk, but being male correlated with earlier (younger) admission (Liu, Coughlin, & McBride, 1991). Only 14 of these studies-fewer than the mortality studies-included perceived health as a variable, but nine of them found it a significant predictor (M. A. Cohen et al., 1986; Steinbach, 1992). Social and socioeconomic factors appeared to contribute significantly to the risk of institutionalization. Living alone (Boaz & Muller, 1994; Wolinsky et al., 1992), not owning one's home (Coughlin et al., 1990; Greene & Ondrich, 1990), lower SES (Liu et al., 1991 ), feeling lack of control (Wolinsky et al., 1992), and receiving formal support such as home care (Boaz & Muller, 1994) predicted institutionalization; informal support (including having a living spouse) protected against it (M. A. Cohen : 1 et al., 1986; Pearlman & Crown, 1992). Many of the studies failed to include the following categories of variables: SES, clinical information other than diagnosis, physical performance and mobility (only 12 studies included any indicator), and depression (only three studies included this variable; of them, two found it significant). I I : Function Decreased functional abilities also suggest increased frailty. Common measures of function include ADLs such as toileting, bathing, dressing, and walking, and IADLs concerning housework, meal preparation, managing money, and eating. In most of the 29 published studies reviewed here, clinically assessed problems related to specific conditions (Boult et al., 1994; Idler & Kasl, 1995) and BMI (Pinsky et al., 1985) contributed significantly to the risk of functional decline. Age also had a strong association with function (Mor, Wilcox, et al., 1994; Roos & Havens, 1991 ). Hoeyman, Feskens, van den Bos, and Kromhout (1997) determined that decreased 48

PAGE 59

functional status resulted from population aging, not from cohort-specific trends. Female gender (Roos & Havens, 1991), non-White ethnicity (House et al., 1994), and lower SES (House et al., 1994) correlated with or predicted decline, or, in the rare cases of studies of positive outcomes, supported continued physical ability (Guralnik & Kaplan, 1989; Harris, Kovar, Suzrnan, Kleinman, & Feldman, 1989). Fewer than half of the studies included perceived health, but in seven of the 11 that did, it correlated significantly and positively with increased function (Idler & Kasl, 1995; Mor, et al., 1994). A few studies suggested an association between health behavior and function (Guralnik & Kaplan, 1989); others found relationships between function and psychological well-being (Berkman et al., 1993) and social role status (A. L. Stewart et al., 1989). Informal support associated negatively with decline (Boult et al., 1994); that is, informal support appeared to protect against decline. Only 14 studies included psychological factors; few mentioned social measures other than marital status. Three studies assessed change in functional ability over time. These studies reported that baseline status, age, perceived health, clinical condition, and/or psychological status affected changes in functional status (Crimmins & Sito, 1993; G. A. Kaplan, Strawbridge, Camacho, & Cohen, 1993; Mor, Wilcox, et al., 1994). In summary, the literature consistently identified age, functional status, clinical condition, and cognitive status as factors related to the risks of, generally, unhealthy outcomes. Perceptions of health, when included in the models, often associated significantly with the outcomes. The studies offered a less clear picture of the importance of social or psychological factors other than cognitive function. In great part that inconsistency derived from the paucity of variables included and the variability in measures used, if the topics were considered at all. Because most reported studies focused primarily on risks to health, they can only suggest, in negative form, what may contribute positively to healthy aging. 49

PAGE 60

Perceived Health Having established that many factors may affect health and healthy aging, this study required a means of identifying important ones. Moriyama ( 1968) listed the following characteristics of good measures, which should: be meaningful and understandable; be sensitive; be clear, justifiable and reasonable in assumptions; be composed of clearly defined components; consist of independent parts, contribute independently to total variance; and use available or obtainable data. Health care measures are commonly classified as indicators of structure, process, and outcome (Donabedian, 1988). Structure encompasses attributes of the setting, components that are relatively easy to measure but not particularly well correlated with care (Schwartz & Lurie, 1990). Process describes actual activities and procedures. Service performed well does not necessarily guarantee effectiveness; the evaluation of process therefore may not provide information about the significance of inputs. Outcomes represent results (Lohr, 1988). Positive outcomes do not necessarily improve health--positive does not always equate with appropriate-but they do serve as indicators of significance. Brook et al. (1977) asserted that because outcomes assessment is based directly on measures of health status, it has more validity than process assessment, which looks only at the quality of the services delivered. Outcomes assessment can focus narrowly or broadly. Measures based on the biomedical model focus on the provision of care rather than on improvements in health status. Outcomes of this sort include mortality, hospital readmissions, length of stay, standard clinical measurements (e.g., blood pressure and laboratory test 50

PAGE 61

results), the efficacy of a new drug, improvement in (or loss of) mobility after hip replacement, and remission from cancer after radiation therapy. These have particular relevance to the treatment of acute illness and conditions, and they serve a useful purpose in that environment. They also appear to have many technical advantages (e.g., hard clinical data, well-defined end points, sensitive and specific tests, statistical validity and reliability), but they address only limited aspects of health. They have shortcomings even as measures of the effectiveness of specialized scientific medicine because of the shift from acute to chronic disease, the aging of the population, the increasing prevalence of diseases with a social component and multi factorial etiologies, and a growing criticism of the business of health (Turner, 1992). j I They ignore important factors such as the presence or absence of increased or decreased pain, the ability to return to work, and change in mental state. At the other end of the health outcomes continuum, global well-being, often referred to as quality of life, encompasses the broadest concepts of health. It incorporates everything from genetic composition to political and economic environmental influences on life. That strength is also its weakness as an outcome indicator--health care intervention cannot directly affect many of its components in measurable ways. Perceived health, or self-rated health, is the outcome measure selected for this : I study. It combines the strengths of both more clinical and more general indicators : I and, of course, shares some of their weaknesses. Like many clinical indicators, it is reasonably unambiguous and straightforward to measure. Researchers frequently assess perceived health with a single simple question. Most such questions, including the following, request responses of excellent/very good/good/fair/poor or a similar range: How would you rate your overall health? (S. H. Kaplan, 1987) How would you rate your health at the present time? (Idler & Kasl, 1995) 51

PAGE 62

:I I Compared to others your age, would you say your health is excellent, very good, good, fair, or poor? (Greiner, Snowdon, & Greiner, 1996) For your age would you say, in general, your health is excellent, good, fair, poor, or bad? (Mossey & Shapiro, 1982) Do you consider yourself a healthy, fairly healthy, sick, or very sick person? (G. Kaplan et al., 1988) A richer measure than clinical indicators, "one's subjective sense of physical health and well-being does not stand in a fixed, one-to-one relationship with objective clinical assessments of one's medical status .... In part, this discrepancy reflects the fact that most patients experience their health as a global experience and a level of function, as an overall state of well being" (Barsky et al., 1992, p. 1147) that incorporates medical status, functional impairment, psychological dysphoria and emotional distress, social factors, role demands, stressful life events, and age. According to Liang ( 1986), perceived health provides a global evaluation of an individual's health status, incorporating comprehensive domains of health and serving as a summary of an individual's perceptions of various objective and subjective aspects of health. Although global in scope, perceived health as a concept and measure remains personal enough to avoid the "separation between individuals and their health" (p. 299) that Robertson and Minkler (1994) worried would make health a theoretical objective rather than an everyday reality amenable to intervention. As the section on previous studies of risks of aging indicated, many researchers have reported that perceived health, in turn, predicts other, more distal outcomes such as mortality and institutionalization. Because it predicts other outcomes so well, Johnson and Wolinsky (1993) pointed out the importance of finding out "how it is actually related to other dimensions of health status" (p. 109). Perceived health may have other attributes as well-Rakowski and Cyran (1990) found that it not only serves as a static indicator but can provide the stimulus to act or not act on matters related to health. Ware (1976) found self-evaluation a valid 52

PAGE 63

i i I I measme ofhealth status. It meets Moriyama's (1968) criteria to be meaningful and understandable, as well as clear, justifiable, and reasonable in assumptions. For older people, subjectively assessed health may be an especially important measure because older people are prone to multiple health problems that include psychological and social aspects as well as physical components (Schoenfeld et al., 1994).2 Linn and Linn (1980) observed that ''how the elderly view their own health may be an extremely useful clinical guide as to their overall health status" (p. 311 ). Studies have shown that perceived health rated by older individuals correlates with or is predicted by more "objective" health indicators (Johnson & Wolinsky, 1993; Wan, 1976), physical exam or physician rating (LaRue, Bank, Jarvik, & Hetland, 1979; Maddox, 1962), chronic conditions (Jylha, Leskinen, Alanen, Leskinen, & Heikkinen, 1986), functional status (Barsky et al., 1992; Johnson & Wolinsky, 1993), depression and other psychological characteristics (Blazer & Houpt, 1979; Tissue, 1972), and SES or economic condition (Mar.kides & Lee, 1990; Wan, 1976). Linn and Linn (1980) determined that it provides a better marker than age. Older persons (65 years old and older) tend to perceive their own health significantly more positively than younger adults (Cockerhmn, Sharp, & Wilcox, 1983). Older-old persons (75 years old and older), who report more health-related problems than the younger old, tend to rate their health even more positively (Ferraro, 1980). These individuals appear to expect and discount some level of decreasing function, or healthy decline, when assessing their health. Hays and Stewart (1990) identified the following as components of self reported health in chronic disease patients: physical health 2 As in the earlier section on risks associated with (un)healthy aging. citations in the text represent only a limited number of studies, ones that exemplify the issues and variables. Appendix B contains extensive tables of the reponed studies, displaying more complete information about research designs and variables. 53

PAGE 64

role limitations due to physical health physical functioning satisfaction with physical ability mobility mental health depression/behavioral emotional control positive affect anxiety feelings ofbelonging. No measure, of course, is perfect, and self-reported health has limitations. As Mechanic (1978) observed, It is much easier to develop measures of physical incapacity for specified population groups than it is to devise reliable and valid measures of overall health status ... which is so involved with subjective perceptions, social expectations and role demands, and value judgments that it is extraordinarily difficult to translate it into any set of empirical measures. A variety of proxy measures are frequently used to measure health status, such as subjective appraisals of one's health and the absence of chronic conditions, but these are only poor approximations of the concepts investigators really wish to study." (p. 183) Self-reported information in particular raises concerns about intended or unintended subjective bias. No assessment, however, can be completely free of bias, either in the definition of the indicator or in the process of measurement. As Patton ( 1990) wrote, "all ... data are based on someone 's definition of what to measure and how to measure it'' (p. 480). Cultural and technological issues affect even definitions of life and death, as we see in debates over abortion and the timing of organ harvest for transplantation. In addition, perceived health may not be an entirely subjective measure. Kutner ( 1987) observed that ''perceived health status is a concept that includes both an objective and a subjective component" (p. 30). 54

PAGE 65

, I Perceived health's global nature, its reliance on a single self-reported indicator, and the multiple contributors to an individual's self-appraisal may mean that it does not meet Moriyama's (1968) requirements for clearly defined components and sensitivity. Hyland (1993) noted that perceptions of the self are not unitary. Two distinctions are particularly relevant to the patient's perceptions of health. The first is that knowledge of events and evaluative appraisals of events are independent and involve different nemological processes. Thus a [sic] patients' knowledge of their health problems are likely to be independent of evaluations of how much distress those problems cause. Second, affective evaluations of positivity tend to be independent of affective evaluations of negativity. That is, a patient may be both happy and unhappy with life or happy about some aspects and unhappy about others. (p. I 021) As a measure of change, however, perceived health appears to be reasonably stable while capable of detecting change. Goldstein, Siegel, and Boyer ( 1984) reported that, over a one-year period, one-third of a study group changed their ratings by at least one category of four (poor, fair, good, excellent), in both positive and negative directions; his study included adults of all ages, not just the elderly. Rodin and McAvay (1992) found similar results. In their study, they incorporated up to seven measurements of perceived health over three years. Pairwise correlations between adjacent measurements ranged from 0.7 to 0.8. Comparisons of the most distal measurements found that 75 percent did not change, 11 percent improved, and 14 percent declined. Summary For the pmpose of this research into healthy aging, perceived health will serve as a proxy for health. The literature described in this chapter identified a variety of 55

PAGE 66

possible contnouting factors that formed the foundation for the analyses whose reports follow. Regression modeling had to be limited to the available data from the parent study, a selection of variables much like those incorporated into the risks of (un)healthy aging studies described earlier. These data primarily concern age, functional and physical performance status, clinical condition, and cognitive status, with a few measures of psychological or social characteristics and health care utilization. Interviews with a subset of the study population offered the opportunity to cast a wider net. Semi-structured questions addressed the following subjects: well-being in general; valued activities and abilities; social and temporal comparisons of health status; challenges, including childhood experiences, worries, isolation, and internal focus; personal resources, including sense of control, attitude, and coherence; support resources; and health care delivery. 56

PAGE 67

CHAPTER3 QUANTITATIVE ANALYSIS This chapter describes the methods and results of a quantitative analysis that had two goals, to identify factors that support positive perceived health and to provide criteria for selecting discrepant cases for subsequent interviews. The analysis addressed parts of three of the overall study's specific aims: to construct a statistical model to describe and predict perceived health from a variety of demographic, clinical, functional, social support, and utilization variables previously associated primarily with negative outcomes related to aging (Specific Aim 1); to identify discrepant cases, those individuals whose reported perceived health differed substantially from values predicted by the model (Specific Aim2); and to identify factors that affect perceived health. to contribute to a model of healthy aging for this older population (Specific Aim 4). Methods Quantitative Study Design Data for this study came from a two-year prospective randomized trial of Kaiser s Cooperative Health Care Clinic (CHCC) program. which delivers outpatient primary care in a group visit format. The quantitative portion of the study reported here is a cross-sectional snapshot of all of the parent study's subjects, whether or not recipients of the CHCC intervention, at the end of the first year of that study. Because baseline explanatory data were available as well, it was possible to analyze both baseline and 12-month change components of the variables. 57

PAGE 68

: I :I : I Study Population and Data Collection The parent study drew subjects from among community-dwelling pre-frail Kaiser members 60 years old and older who had a history of chronic conditions and high defined as 12 or more contacts with a provider within the 18 months prior to the study. These members constituted about 16 percent of Kaiser's Medicarerisk population at the beginning of the study (approximately 46,000 members). Kaiser's research team mailed the Health Screening Form (HSF) to Kaiser members who met the initial age and utilization selection criteria. (Appendix A contains a copy of the instrument.) The form included questions related to the study variables (e.g., perception of health, ability to perform activities of daily living, presence of chronic conditions) along with participation screening questions (e.g., the likelihood of changing to another health care provider within the study period or the presence of a condition or serious illness that would make it impossible to attend meetings or participate in a group). In the questionnaire asked respondents if they would be interested in group-based health education and patient care clinics, if such clinics were started. Physicians familiar with the members then assessed their functional capability to participate in a group-based clinic. Those who returned the HSF, met selection criteria, and expressed interest in the group concept were randomized into two arms of the study, experimental (CHCC intervention) and control (usual care). Subjects entered the study as physicians began groups, between March 1995 and June 1996, and numbered nearly 400 cases and 400 controls, or almost 11 percent of those who met the initial selection criteria and 71 percent of the randomized sample. As each group ended the first year of the study, subjects again received the same HSF instrument. CHCC intervention subjects completed the form during their 12-month meetings if they attended them; all others received the instruments by mail. 58

PAGE 69

Study staff aggressively followed up with non-completers to ensure the greatest possible response. Measures Outcome (Dependent) Variable: Perceived Health Status at 12 Months. The outcome (dependent) variable, health status at 12 months, comprised responses to this question: Compared to other persons your age, would you say your health is: 0 Excellent 0 Fair 0 Very Good 0 Poor 0 Good Although the perceived health status variable was measured on an ordinal scale, the size of the sample permitted considering it a continuous variable for the pmposes of linear regression. Explanatory (Independent) Variables. Explanatory variables included clinical and functional data from the HSF questionnaire and additional information from administrative databases. For analysis, variables were grouped into conceptual categories: sociodemographics, chronic conditions, physical function, cognitive and psychological function, health behaviors (smoking), informal and formal social support, and health care services utilization. Sociodemographic information included age, gender, race/ethnicity, education, and (optionally) household income. The questionnaire asked about the presence or absence of eight health conditions that an individual might self-identify (e.g., chronic lung disease, trouble seeing, asthma, arthritis) and of six other health conditions that would more likely be identified by a physician (e.g., hypertension, kidney disease, cancer). Physical condition variables included indicators of Activities of Daily Living (ADL; from Katz, Ford, Moskowitz, 59

PAGE 70

I I : I ; I & Jaffee, 1963), Instrumental Activities ofDaily Living (IADL; from Duke University Center for the Study of Aging and Human Development, 1978), physical performance and mobility (from Rosow & Breslau, 1966; Nagi, 1976), and urinary incontinence and bowel control. A depression item and a five-part question about dementia measured cognitive and psychological function. The questionnaire assessed informal social support by asking about marital status, current work status (including volunteer activities), living arrangements, and transportation requirements. Respondents were also asked about receipt of agency-based formal social support services. Health care services utilization variables included number of medications, days of hospitalization, outpatient services, skilled nursing care, patient percent attendance in the CHCC group visit program, and a physician identifier. Data were collected at baseline and at the end of 12 months. Table C. I in Appendix C contains a complete listing and description of the variables at baseline. Table C.2 similarly reports 12-month change information. Recoding Based on Distribution of Responses. The HSF generated ordinal or categorical responses. For some items, the distribution of the responses suggested recoding into a smaller number of categories. The following figures give examples. Figure 3.1. Sum of Activities ofDaily Living Dependencies -5GO lGO -iiCX) t 0 0 2 5 Figure 3.2. Sum of Instrumental Activities of Daily Living Dependencies 60

PAGE 71

I As illustrated in Figures 3.1 and 3 .2, very few respondents reported any ADL or IADL dependencies. All sums greater than zero therefore were combined into a single category any, to contrast with none in the recoded dichotomous variable. Because of its distribution (Figure 3.3), the sum of physical performance indicators was recoded into three values, combining the Figure 33. Physical performance original 0 and I into a single category indicating substantially diminished physical ability, the original 2 and 3 into a category associated with moderate disability, and leaving the original5 (indicating the presence of all abilities) as the third category. Analvsis Analysis began with descriptive summaries of the data and examination of correlations between variables. Tables C.l and C2 in Appendix C present summary statistics and measures of correlation between the dependent variable and each independent baseline and 12-month change variable. Theoretical assumptions described in Chapter 2 suggested that health consists of many factors that contribute sequentially to an individual's perception of it. Based on those assumptions, variables were grouped into categories (e.g., sociodemographics, functional status). Stepwise multiple linear regression modeling identified the variables within each category that had a significant association with perceived health at 12 months, the dependent (outcome) variable. The analysis used the regression functions ofSPSS 8.0 for Windows, entering and removing variables stepwise with default selection criteria probabilities (for F) of p = 0.10 to enter and p = 0.15 to remove. 61

PAGE 72

The model-building process continued by hierarchically forcing entry of each category's significant variables, as identified in the previous stage of the analysis, into a multiple linear regression model. Sequential, or hierarchical, entry of categories began with sociodemographics, followed by chronic conditions, functional status, physical performance and mobility, depression and cognition, informal social support, formal social support, and health care services utilization. Entering categories in this fashion permitted examination of the incremental effects of progressively less individual and more social/behavioral and social/delivery-of-care factors. Results Study Attrition Of the 793 study subjects who completed the HSF questionnaire at baseline, 705 completed the same instrument again approximately 12 months later. Twenty-Table 3.1. 12-Month HSF: Comparisons Between Respondents and Non-Respondents Responded to HSF at 12 months? Variable (mean at baseline) Age Gender (0 = male, I = female) Sum of chronic conditions ID'd by patient Sum of chronic conditions ID'd by physician SumofADLl Sum Sum of physical performance indicators Depression (0 =no, I =yes) Live alone (0 =no, I =yes) Number of daily medications Health status at baseline No 1res (N = 88) (N = 705) 74.09 73.46 0.57 0.62 2.06 I.54 1.38 1.10 024 0.14 I.20 0.54 2.41 3.07 0.30 O.I8 0.34 0.30 6.I2 4.76 2.77 3.16 Note. Shading indicates statistically significant differences, p
PAGE 73

four of the 88 non-respondents died during the year, 13 terminated their Kaiser memberships (four others who terminated membership did complete the 12-month : 1 HSF), and nine switched providers (37 others who switched did complete the 12-month HSF). Excluding those who died, 91.7 percent of those who could have completed the 12-month HSF (705 of769) did so. The 88 non-respondents not surprisingly were on average in poorer condition at baseline than the 705 respondents, as Table 3.1 shows. Thirteen of the 705 participants who completed the HSF at 12 months failed to rate their health status, the outcome variable for this analysis. Data in Table 3.2 compare mean values of selected variables between those who did and did not respond to the health status question at 12 months. Those with missing data were older, more likely to be male, less functionally impaired, and more likely to live alone, but only the differences in IADL dependencies are statistically significant {p<0.05). As with those who did not complete any of the HSF at 12 months, these Table 3.2. 12-Month HSF: Comparisons Between Those Who Did and Did Not Repon Health Status Variable (mean at baseline) Age Gender (0 =male, 1 =female) Sum of chronic conditions ID'd by patient Sum of chronic conditions ID'd by physician SumofADLsb Sum of IADLsc Sum of physical performance indicators Depression (0= no, 1 = yes) Live alone (0 = no, 1 = yes) Number of daily medications Health status 12-month health status missing? No (N= 692) Yes (N= 13) pa 73.4 77.1 0.08 .62 .54 0.56 1.55 1.23 032 1.11 .85 034 .14 .00 0.46 .54 .15 Qj)() 3.07 3.23 0.62 .18 .23 0.63 .29 .54 0.05 4. 76 4.25 0.66 3.15 3.42 0.29 Note. Shading indicates statistically significant differences, p
PAGE 74

. I individuals' data have been excluded from the analysis, leaving a study sample of 692 and a response rate of90.0 percent (692 of the 769 who survived). Missing Data There were several baseline questionnaire items that respondents left blank more often than other items. Thirty-two respondents failed to answer the depression question, 30 did not provide the number of daily medications, and 195 left blank the optional income item. To assess whether other significant differences existed between respondents and non-respondents to these items, data reported in Table 3.3 compare a subset of the variables between the respondents and non-respondents. Those who did not respond to the depression and income questions reported significantly lower health status at baseline but no other signigicant differences. Based on 35 baseline and 12-month variables3 391 (56.5 percent) study members provided complete data, 227 (32.8 percent) failed to respond to one item, and 69 (10.0 percent) left two to five items unanswered. Only five (0.7 percent) failed to respond to more than five items; four of these missed a complete two-sided page of the HSF questionnaire Because there were so few significant differences between those who did and did not respond to some of the questions, it seemed appropriate to substitute imputed values for the missing explanatory variable data in order not to exclude cases unnecessarily and to ensure the most complete possible set of predicted values on which to base outlier selection. There seemed to be no obvious pattern to missing 3 Health status at baseline and at 12 months. baseline age, gender, marital status, race/ethnicity, education, employment status, income, baseline and 12-month sums of chronic conditions, worsening of chronic conditions, bowel and urinary incontinence, sums of Activities of Daily Living and Instrumental Activities ofDaily Living dependencies, sums of physical performance indicators, mobility, depression, type of housing. sums of needed social support, and transportation requirements. 64

PAGE 75

: I I data either by variable or by case; therefore mean values were imputed from existing data and substituted for missing values of the independent (explanatory) variables. Table3.3. Comparisons of Selected Baseline Variables, Cases with Data Present vs. Cases with Data Missing #of daily DeEression medications Income Miss. ? mg. Missing? Missing? Variable No Yes No Yes No Yes (mean at baseline) N=660 N=32 ll N--662 N=30 ll N=497 N=l95 2.a Age 73.4 74.5 0.43 73.4 75.3 0.16 73.4 73.6 0.81 Gender .62 .63 0.94 .61 .74 0.19 .60 .66 0.14 Sum of chronic 1.55 1.39 0.47 1.55 1.33 0.33 1.57 1.48 0.38 conditions ID'd by patient Sum of chronic 1.11 1.00 0.57 1.11 .85 0.18 1.11 1.07 0.60 conditions ID'd by physician SumofADLsb .14 .12 0.84 .I4 .07 0.61 .13 .16 0.60 Sum ofiADLsc .52 .97 0.28 .53 .77 O.JI .55 .51 0.75 Sum of physical 3.09 2.66 0.05 3.06 3.32 0.24 3.08 3.05 0.8I performance indicators Depression (0 = no, I NA NA .I8 .I9 0.78 .17 20 0.31 =yes) Live alone (0 = no, I = .30 28 0.89 29 .39 0.36 .30 .28 0.47 yes) Number of daily 4.77 4.30 0.54 NA NA 4.82 4.58 0.50 medications Health status at 3.17 2.89 3.15 322 0.70 3.21 3.02 o.or baseline - --.--... . --Health status at 12 3.08 2.91 0.34 3.07 3.03 0.84 3.1I 2.97 0.12 months Note Shading indicates statistically significant differences, p<0.05. a p based on Student's t-test for continuous data, Mann-Whitney U test for dichotomous data. b ADL =dependencies in Activities of Daily Living. c IADL =dependencies in Instrumental Activities of Daily Living. 65

PAGE 76

Characteristics of the Study Sample Perceived Health. Mean perceived health status at baseline was slightly greater than good (3.15 on a scale from 1 for poor to 5 for excellent). The mean decreased only slightly during the year (to 3.07), but at 12 months fewer people reported very good and more reported/air. Tables 3.4 and 3.5 and Figures 3.4 and 3.5 show the distributions of the ratings at baseline and 12 months. Table 3.4. Distribution of Baseline Status Poor Fair Good Very good Excellent Total valid Missing Perceived Health Status Ratings Frequency 25 128 287 207 37 684 8 Percent 3.6% 18.5 41.5 29.9 53 98.8% 12 Table 3.5. Distribution of 12-Month Perceived Health Status Ratings Status Poor Fair Good Very good Excellent Total Frequency 34 165 266 173 54 692 Percent 4.9% 23.8 38.4 25.0 7.8 100.0% 66 Figure 3.4. Distribution ofbealth status at baseline 100 Figure 3.5. Distribution ofhealth status at 12 months

PAGE 77

The nature of the perceived health outcome variable introduced concerns about floor and ceiling limits. Individuals who reported poor or excellent health status at baseline could indicate change in only one direction at 12 months. The I : 1 numbers affected, however, were very small. Of the 25 who reported poor health at baseline, 14 indicated improved status at 12 months, leaving only II who might have reported poorer health if such a category had been available. Similarly, of the 37 respondents who reported excellent health at baseline, 12 indicated poorer status at 12 months; that is, a maximum of25 might have selected a better status at 12 months if given the opportunity. Given these small numbers, the analysis did not exclude individuals who reported poor or excellent health status at baseline. Baseline Explanatory Variables. Table 3.6 summarizes some of the baseline characteristics of the study sample, with information about the response rate for each item. Table C.I in Appendix C, which more completely describes the variables, presents responses to the baseline HSF, attendance, and administrative data; variables that have been recoded or aggregated from those data; and correlations between the baseline variables and one-year perceived health. Table C.2 provides similar information for changes in variables over the I2-month period. The mean baseline age was 73.4 years; 45 percent of the individuals were between 65 and 75 years old. Approximately 62 percent were female; more than 90 percent identified their race/ethnicity as White/Caucasian (it is likely that some of the 4.6 percent who identified as Native American misunderstood the classification); 50 percent had at least a high school education; and, for those who responded to the item, mean income lay between $15,000 and $25,000. The HSF distinguished between chronic conditions that the individual would be aware of (e.g., trouble seeing or hearing, asthma, chronic lung disease, arthritis) and other conditions more likely to be identified by the physician (e.g., hypertension, heart kidney disease, cancer). Five percent of the study members reported no 67

PAGE 78

Table 3.6. Frequencies (Prevalence) of Selected Baseline Characteristics (N= 692) %with %with %with %with Category condition data Category condition data Variable eresent missing Variable eresent missing Perceived health Poor 3.6% 12% Sum of abilities 0.4% Fair 18.5 0 6.4% Good 41.5 1 4.6 Very good 29.9 2 13.9 Excellent 5.3 3 26.0 Demo'l!:!!I!.hics 4 48.7 Age 0.6 Mobility 0.9 <65 14.8% Need aid 22.1 65-75 45.4 No limits 77.0 >75 39.1 Transportation 0.6 Gender (female) 61.8 0.0 Drive self 81.8 I Chronic conditions Others drive 17.6 I Sum ID'd by patient 0.6 Don't go out 0.0 I 0 17.8 De'{l!:.ession, dementia 1 35.0 Feel depressed 17.1 4.6 2 28.3 Sum of dementia 0.4 3 12.6 indicators 4 4.3 0 78.8 5 12 20.8 6 0.3 Sum ID'd by physician 0.6 ln(grmal social SUJ!ll.Ort 0 28.3 Married 62.4 0.4 1 43.4 Full-or part-time 22.0 0.3 2 19.4 Employment 3 5.8 Living arrangement 0.0 4 2.5 Alone 292 5 0.0 With spouse 61.8 6 0.1 With child(ren) 7.1 Fwzctional status With relative( s) 1.9 SumofADLs 1.0 W/ non-relative(s) 2.3 0 92.1 With pet(s) 15.6 6.9 Housing 0.3 SumofiADLs 0.0 Own home 97.0 0 79.6 F annal social SUJ!Il.Ort 20.4 Sum of supports 0.0 0 94.4 >1 5.6 68

PAGE 79

I 1 I I I I chronic conditions at baseline; 26 percent reported four or more of the 14 listed conditions. As Table 3.7 shows, the most commonly cited conditions included arthritis or rheumatism (60 percent), hypertension (49 percent), and deafuess or trouble hearing (29 percent). Between 10 and 20 percent of the respondents reported cancer, heart attack or myocardial infarction. chronic lung disease, asthma, angina. blindness or trouble seeing, and/or diabetes. Only a few individuals indicated that they had experienced a stroke, congestive heart failure, kidney disease, or an ulcer or gastrointestinal bleeding. Nine percent felt that at least one of these conditions was getting worse. Table 3.7. Prevalence of Chronic Conditions Condition Arthritis/rheumatism Hypertension Deafuess/trouble hearing Cancer Heart attack Chronic lung disease Asthma Angina Blindness/trouble seeing Diabetes Stroke Congestive heart failure Kidney disease Ulcer or intestinal bleeding Prevalence 60% 49 29 18 17 15 14 13 13 13 8 7 4 4 These community-dwelling older people, for the most part, reported functional independence, as measured by dependencies in Activities and Instrumental Activities of Daily Living (ADLs and IADLs), physical capabilities, and mobility. They reported almost no ADL dependencies.4 For three of the eight IADLs-grocery shopping, chores, and transportation-nearly 12 percent indicated the need for assistance.s It is interesting to note that these are the three IADLs most dependent on physical ability and mobility. Few study respondents (10 percent) had difficulty walking up and down stairs or going out to a movie or other such activity, but more of them needed help (from a person or special equipment, e.g., a walker) doing heavy 4 ADLs included in the HSF: need help toileting. bathing, dressing, eating. and getting in and out of bed. s IADLs included in the HSF: need help with meal preparation. grocery shopping. household chores, money management, laundry, taking medications, transportation, and using the telephone. 69

PAGE 80

work (nearly 50 percent) or walking a half mile (almost 30 percent). They remained quite mobile; all of them could get around inside and outside the house with at most aid from a cane or wheelchair (and that for only a few). Seventy-two percent drove their own cars. Approximately 18 percent of the 660 individuals who responded to an item about depression reported often feeling sad or depressed. Seventeen percent of the study sample reported increasing problems with forgetting dates and names, but few indicated any problems with the other indicators of dementia. These individuals had access to sources of informal social support, to the extent that the HSF measmed that type of support. Sixty-three percent were married, 22 percent worked full-or part-time, and 97 percent lived in their own homes. They reported little contact with (and presumably little need for) formal support services. Administrative data confirmed generally high utilization of primary care (82 percent had at least one visit other than CHCC), hospital services (24 percent had at least one admission), and emergency care (26 percent had at least one visit). Some concerns exist about the accuracy of these administrative data--when different systems supposedly contained related longitudinal data, they did not always agree. There is no reason to believe, however, that any particular bias existed between individuals or at a particular point in time. Even allowing for possibly inflated numbers, this group of older people continued to consume a large number of health care services. Approximately 22 percent of the intervention group members (78 of the 3 56 who survived and did not terminate their Kaiser memberships) attended no CHCC meetings; five percent (19) attended only one. There were few significant HSF differences between those who attended and those who did not. Non-attenders were more likely to die during the 12-month period (p = 0.05, based on the Mann-Whitney U test), to be depressed (p < 0.01), and to reportfewer medications (p = 0.01). 70

PAGE 81

Correlations Between Baseline Explanatory Variables Aand 12-Month Perceived Health. Baseline perceived health status, as one would expect, correlated positively with 12-month status (Kendall Tb = 0.581). The only explanatory variables that correlated even moderately Cl Tbl > 0.200) with 12-month perceived health status were sums of chronic conditions '"' g40+------u :s =-J: 30 +-----= .. 20+----identified by the ADLs, IADLs, physical performance -4 -1 0 2 3 4 Change in Health StaniS indicators, mobility, need for Figure 3.6. Change in Perceived Health Status transportation, and number of medications. Changes Over 12 Months. More than half of the study members reported no change in health status over the 12 month period (see Figure 3.6). Most who reported any change indicated transitions between adjacent categories; 147 (21.2 percent) reported a decline of one category, 129 (18.6 percent) a one-category increase. Fewer than five percent reported greater changes. Table 3.8 provides details. Table 3.8. Changes in Health Status over 12 Months Health status Health status at baseline = 684l at 12 months Very Excel-(N=692) Poor Fair Good good lent Missing Total Poor i ,<.:-Jl:: 16 5 .. --,-. 1 1 34( 4.9%) Fair 11 -':"75 59 15 5 165 ( 23.8%) ---Good 2 35 64 ... 2 266 ( 38.4%) ..) Very good 60 ___ xif4 __ 8 1 173 ( 25.0%) Excellent 1 2 3 23 54( 7.8%) Total 25 128 287 207 37 8 692 (100.0%) 3.6% 18.5% 41.5% 29.94'/o 5.3% 1.2% Note. Shading indicates no change. 71

PAGE 82

, I : I I There were no dramatic changes in numbers of chronic conditions or any of the measures of functional and physical status during the year. The changes that were reported occurred in the direction one would anticipate: increased dependencies in IADLs, decreased physical performance and mobility, greater assistance needed with transportation, and an increase in the number of medications. Regression Models Table 3.9 presents results of the stepwise linear regression selection of variables within each conceptual predictor category, showing only those that contributed significantly (p S 0.10) to the variation in 12-month perceived health. The stepwise regression procedure confirmed the relationships suggested by correlation comparisons. Categories that individually explained the greatest amount of variation in 12-month perceived health status included baseline perceived health status ( 41 percent); the number of chronic conditions, change in the number of reported conditions over the 12-month period, and reported worsening of any of the conditions (20 percent); sums of ADUIADL dependencies and changes between baseline and 12 months (16 percent); sum of physical performance indicators, mobility, and changes in these variables over the 12-month period (26 percent); and utilization of some health care services (15 percent). Demographics, depression and dementia, informal social support, and formal social support added little explanatory power. Neither membership in the parent study's experimental group nor percentage attendance at group visit meetings showed any significant association with perceived health. 72

PAGE 83

Table 3.9. Perceived Health Status at 12 Months: Multiple Linear Stepwise Within Categories Category Variable Step Adjustedk Beta Baseline health status l .414 .709 Sociodemographics Education 1 050 .152 Income 2 .053 .060 Chronic conditions SWil, identified by patient at baseline l .095 -269 SWil, identified by patient, change 2 .132 -.171 SWil, identified by physician at baseline 3 .152 -.164 SWil, identified by physician, change 6 202 -.152 : I Condition( s) getting worse at baseline 5 .195 -.688 Condition(s) getting worse, change 4 .168 -.605 Functional status Sum of ADLs at baseline 3 .138 -.836 Sum of ADLs, change 4 .159 -.565 Sum ofiADLs at baseline 1 .088 -.713 ; I Sum ofiADL, change 2 .128 -.374 Physical per(ormance Physical performance at baseline 4 261 .195 Physical performance, change 3 .240 .188 Mobility at baseline l .132 .642 Mobility, change 2 .231 .490 Transportation I 1 .050 -.737 ; I Transportation, change 2 .085 -.548 Depression. dementia Depression at baseline 1 .044 -.851 Depression, change 2 .076 -.497 (Dementia not significant) Infgrmal social S7!J!POrt Employment at baseline 1 .026 .526 Employment, change 2 .035 .288 Live with spouse 3 .039 .336 Formal social support Formal social support at baseline 2 .030 -.704 Formal social support, change l .013 -.699 Utilization Number of daily medications at baseline 2 .081 -.073 Number of daily medications, change 3 .108 -.054 73

PAGE 84

I Table 3.9. (Cont.) Category Variable Durable medical equipment/supplies Home health care Patient education visits Skilled nursing facility visits Emergency visits #of da ofho italization Step I 4 6 8 5 7 Adjustedk .047 .130 .145 .151 .141 .149 Beta -397 -306 -315 -367 -223 -.004 is cumulative by category; non-standardized betas. Table 3.10 contains the results of the hierarchical linear regression analysis of perceived health status, beginning with the sociodemographics category and adding an additional category at each subsequent stage of the modeling process. The table shows unstandardized beta coefficients for those variables that retained statistical significance at each stage of the modeling process. It also contains cumulative and incremental coefficients of multiple determination (K), to measure the amount of variation in perceived health that each model explains. Incremental changes in R2 values are presented in two ways. The first describes how much the model improved when a new category was added in the hierarchical order suggested by theoretical concerns (Type I sum of squares). The second measures the incremental improvement if the category were added last to the model, with all other categories already in the model (Type m sum of squares). The unstandarclized beta coefficient quantifies the difference in perceived health status that results from a unit change in a variable. Perceived health is scaled in discrete steps from poor (1) to excellent (5). Although reported status can take on only integer values, predicted values range along the scale continuously. In this report of results of the modeling, effects will be described as increments (or decrements) of steps along the perceived health status scale. For example, the effect on perceived health status of a variable with a positive beta coefficient of .030 will be reported as an improvement of 0.03 step. For continuous variables (e.g., age) each increase of 74

PAGE 85

Table 3.10. (Cont.) Categorv Model Model Model Model Model Model Model Model Model Variable I 2 3 4 5 6 7 8 9 Informal social support Employment at baseline .177* .180* NS Employment, change NS NS NS Live with spouse at baseline -.221 t -.22lt -.211 t -.194t Live with spouse, change NS NS NS Formal social support Formal support at baseline NS NS Formal support, change NS NS Utilization and Intervention Number of daily medications at baseline NS --..! Number of daily medications, change NS VI DME/supplies, any vs. none NS Home health care, any vs. none NS Emergency visits, any vs. none NS Non-CHCC education visits, any vs. none NS # of days of hospitalization 0.003 -.004* SNF episodes, any vs. none NS Percent of CHCC meetings attended NS Constant 2.223 3.098 3.168 1.966 2.065 2.072 2.041 2.210 2.235 Adjusted R1 Category added Socio Chron Funct Phys Depr Inform Formal Util Cumulative R1 .053 .233 .301 .351 .363 .375 .376 .382 .377 Incremental R1 hierarchical .180 .068 .050 .012 .012 .001 .006 -.005 Incremental R1 if category added last .014 .041 .000 .042 .009 .010 .001 .006 Note. Unstandardized betas; change refers to change over 12-month period; t p s 0.00 I; t p s 0.0 I; p s 0.05; NS p > 0.05.

PAGE 86

. ---------------------------Table 3.10. Hierarchical Multiple Linear Regression of 12-Month Perceived Health on Previously -Selected Variables Category Model Model Model Model Model Model Model Model Model Variable l 2 3 4 5 6 7 8 9 Demographics Education .152t .124t .I JOt .107t .lOSt .087t .083t .078t .Illt Income NS NS NS NS NS NS NS NS Chronic conditions Sum, identified by patient at baseline -.237t -.184t -.126t -.109: -.108t -.106t -.093t -.112: Sum, ID'd by patient, change for worse -.l66t -.132t -.l02t -.095t -.093t -.09lt -.098t -.098t Sum, identified by physician at baseline -.160t -.149t -.133t -.l44t -.141 t -.136t -.122t -.114t Sum,ID'd by physician, change for worse -.138* -.117* NS NS NS NS NS Condition(s) getting worse at baseline -.720t -.602t -.506: -.449: -.419: -.491 t -.464t -.466t Condition(s) getting worse, change -.620t -.531t -.452t -.414t -.41St -.41St -.420t -.392t Functional status -....) Sum of adls, at baseline -.708t -.405* -.355* -.343* -.335* NS -.299* 0\ Sum of adls, change for worse -.600t -.332t -.328t -.297* -.278* NS -.269* Sum of iadls at baseline -.349t NS NS NS NS NS Sum of iadls, change for worse -.214* NS NS NS NS NS Phvsica/ performance Sum, physical performance at baseline .099* NS .089* .094* NS .096* Sum, performance, change for better .123+ .11 5t .122+ 123t .I07t .124t Mobility at baseline .439t .472t .473t .473t .45 It .459t Mobility, change for better .355t .357t .342t .335t .335t .348t Transportation needs at baseline NS NS NS NS NS Transportation needs, change for worse NS NS NS NS NS Depression (and demelllial Depression at baseline -.386t -.389t -.376t -.356t -.404t Depression, change for worse -.206* -.197* -.192* -.174"' -.204*

PAGE 87

I I one tmit (e.g., one year of age) changes predicted health status by the beta number of tmits. The change is positive or negative depending on the sign of the beta coefficient; a positive sign denotes an increase in health status, a negative sign a decrease. For dichotomous and ordinal variables, the presence of the variable (e.g., depression) indicates a change of beta units (for each ordinal increase) in predicted health status compared to health status in the variable's absence, if all other variables remain the same. The first column ofTable 3.10 shows the results of Modell, when only sociodemographic variables were included. Education had a fairly small but statistically significant effect. (Throughout this discussion, significant denotes statistical significance.) Each increased level of schooling corresponded to about 0.10 step increase in health status. Model 2 introduced chronic conditions and worsening of those conditions. All of these factors had significant negative effects. Conditions identifiable by the individual (e.g., chronic lung disease, deafuess, arthritis) had a greater effect than those more likely identified by a physician (e.g., cancer, kidney disease, stroke). Increased numbers of conditions over time had smaller but significant effects. Individuals' reports that these conditions were worsening, and worsening more over time, had the greatest negative effect on perceived health status in this model, each by more than half a step. Introducing functional status (Model3) and physical performance and mobility (Model4) increased the total amount of variation explained from 23 to 35 percent. Education continued to be significant in these models but with reduced effect. Similarly, chronic conditions (and their worsening) remained significant but with a slightly reduced magnitude. The presence of any ADL dependency (vs. none) corresponded to a decrease of about 0. 70 step in health status (reduced to 0.40 after the addition of physical performance to the model). An increase (decrease) in the 77

PAGE 88

. I I number of ADL dependencies during the 12 months decreased (increased) health status by another 0.60 step (reduced to 0.33 after the addition of physical performance to the model). Dependencies in many of which require physical ability and mobility grocery shopping, chores, and laundry), had significant effects only in the absence of the physical performance and mobility measures. Better baseline mobility increased perceived health by more than 0.40 step, and improvement (worsening) over the 12-month period increased (decreased) it by about 0.35 step Both better baseline physical performance ability and improvement over the 12-month period had significant but smaller positive effects of about 0.10 step each. Model 5 shows the results of adding the presence of depression and change in depression during the 12-month period. Both had significant of. respectively, nearly 0.40 and about 0.20 step decrease in health status. The effects of previously entered variables remained virtually unchanged. Incorporating informal and formal social support variables (Models 6 and 7) added little explanatory power. Being employed or volunteering full-or part-time increased health status by about 0.18 step. Living with one's spouse had a significant negative effect on health representing a 0.22 step decrease. This result, which contradicts previous studies' reports, reflects correlations noted above--negative between perceived health and living with one's spouse and positive between perceived health and living alone. Model 8 introduced utilization and intervention variables. Adding these I variables, none of which contributed significantly (although hospital days nearly reached significance, p = 0.056), decreased the effect of ADLs and physical performance, which lost statistical significance; they were especially sensitive to the inclusion of durable medical equipment (D:ME). The final model, Model 9, included those variables that remained significant in Model 8. In ADLs and physical performance were reintroduced in the absence ofDME, and the hospitalization 78

PAGE 89

: I : I : I I I I variable was added because of its near significance in the previous modeL ADLs and physical performance regained statistical significance in this model, and hospitalization days showed a small but significant effect. This final model explained 38 percent of the variation in perceived health. The intercept value of2.235 defined the average perceived health status of individuals who reported the referent values of all variables; that corresponds to a rating of a little better than fair. Discussion The purpose of the quantitative portion of the study was to construct a statistical model of perceived health status from a set of factors (within theoretically aggregated categories) commonly associated with negative outcomes of aging. There were two reasons for doing so. The first goal was to determine which of the factors appeared to contribute most to positive perceived health status. The primarily a methodological issue, was to provide predicted values of perceived health with which to compare individuals' reports, in order to identify a sample of individuals whose reported status differed from that predicted by the model. That process will be more fully described in the next chapter. The results indicated that many but not all of the factors associated with negative outcomes do contribute (in the opposite direction) to healthy aging, as measured by perceived health, and that they explain about 38 percent of the observed variation in perceived health. Of the sociodemographic variables, only education remained in the model; perceived health increased by 0.11 step for each increased level of education. This means, for example, that if all other characteristics were identical, an individual with some college education would be expected to report health status 0.11 higher (on the scale of 1 = poor to 5 = excellent) than another who had only completed high school. So, if the high school graduate's expected health status were good (measured as 3 on a 5-point scale), then the otherwise identical 79

PAGE 90

; I I I college attendee's would be slightly better (3.11, an improvement of 11 percent of the step between 3 and 4). Table 3.11 summarizes the expected (predicted) incremental changes due to each variable in the final model. The study found that physical health status, as represented by chronic conditions, detracted substantially from perceived health. Perceived health status decreased by 0.11 step for each additional reported baseline chronic condition. 0.47 step for baseline worsening of condition(s). and 0.40 step if a worsening of condition(s) was reported at 12 months but not at baseline. Functional dependencies (ADLs and IADLs) explained some of the variation in perceived health. but the importance of these variables diminished (and. for IADLs, disappeared) when combined with measmes of physical performance and mobility, which significantly affected perceptions of health status. Perceived health status decreased by 0.30 step in the presence of any baseline ADL dependencies and by 0.27 Table 3.11. Incremental Changes in Expected Values ofPerceived Health Factor Expected perceived health for an individual who reported referent values for every variable in the model .................................................................... ... Adjustments: for each increased level of education ........................................................ for each additional baseline chronic condition identified by patient ........ for more (fewer) chronic conditions identified by patient at 12-months .. for each additional baseline condition identified by physician ................. if conditions getting worse at baseline ..................................................... .. if conditions getting worse at 12 months but not baseline ........................ if any baseline dependencies in Activities of Daily Living (ADL ) ........... if more ADL dependencies at 12 months .................................................. for each positive baseline physical performance indicator ....................... for each positive (negative) change in physical performance at 12 months for each better baseline level of mobility ................................................. .. for each positive (negative) change in mobility at 12 months ................... if depressed at baseline .............................................................................. if more (less) depressed at 12 months ....................................................... if live with spouse ...................................................................................... for each day of hospitalization during 12 months .................................... .. 80 Change 2.235 (fair+) +0.111 -0.112 (+) 0.098 -0.114 -0.466 -0.392 -0299 -0269 + 0.096 + (-) 0.124 + 0.459 + (-) 0.348 -0.404 (+) 0.204 -0.194 -0.004

PAGE 91

step with the report of a greater number of dependencies over the 12-month period; it increased by 0.09 step for each additional baseline indicator of physical performance, increased (decreased) by 0.12 step with each change for the better (worse), increased by 0.46 step for each better baseline level of mobility status, and increased (decreased) by 0 .35 step for improvement in (worsening of) mobility over the 12: 1 month period. Depression and change to a depressed state over the 12 months had a substantial negative impact, independent of other variables and greater than suggested : I by previously published studies of more distal outcomes. The presence of depression : I decreased perceived health status by 0.40 step, and change to a depressed state I decreased it by 0.20 step. Among measures of informal social support, only living with one's spouse contributed to the model, for a decrease in perceived health status of0.19 step. The negative effect of living with one's spouse contradicted results from most previous studies. Marital status did not affect the outcome, either independently or in combination with the living with. . variable. More men than women (78 vs. 52 percent) lived with their spouses, and more women than men were widowed (36 vs. 11 percent), but the individual's gender did not significantly interact with living with one's spouse to explain variation in perceived health status, nor did gender alone affect perceived health. It may be that the spouses of the married study subjects needed substantial care, creating caregiver stress that outweighed previously reported benefits of living with one's spouse, but quantitative data do not exist to explore this potential relationship. It may also be the case, as Berkman (1988) suggested, that '"marital status may no longer tap a sense of intimacy or social integration for older people as widowhood becomes a normative experience" (p. 59). Among utilization measures, only hospitalization affected perceived health status, with a minimal negative effect. 81

PAGE 92

Some factors that contn"buted to poor health outcomes in earlier studies of older people did not affect perceived health in this study population. Although being male generally predicts mortality, it did not here predict perceived health status. Neither age nor racelethnicity bad an effect, most likely because of the homogeneity of this group. To the extent that disability and chronic illness correlate with age, it may be that the model excluded age not because it bad no explanatory power but because collinearity with other variables masked its effect. There was, however, very little correlation between age and any of the ability or performance variables. Although living with one's spouse contributed (negatively) to the model of perceived health, marital status did not, either independently or in combination with the living witk . variable. IADL abilities, many of which depend on physical abilities, had no impact once physical performance and mobility measures were introduced into the model. Again multicollinearity may explain the absence of effect; IADL dependencies correlated significantly with the sum of physical performance indicators (Kendall Tb = -0.434). Dementia bad no effect, but although more than 20 percent of respondents reported at least one indicator of dementia, they generally only reported forgetting dates and names, not other more serious cognitive disabilities. Neither housing type nor any measures of formal support predicted perceived health, likely because this community-dwelling population did not yet need environmental supports. Of the various health care utilization variables, only the number of days of hospitalization contributed significantly to the model; each incremental day of hospitalization decreased perceived health status by 0.004 step. Because of their HMO membership, these individuals bad few if any barriers impeding access to health care. This homogeneity of access may explain why utilization explained so little of the variation in their perceptions of health status. 82

PAGE 93

II I I I ! The analysis here did not test the sequential causal relationships among categories or domains of health status suggested by Wtlson and Cleary's (1995) model, but it did support the notion that some of these categories have cumulative, interdependent effects. Physiological factors and symptoms (measured here by chronic conditions) and broadly defined physical functional status (daily living abilities, physical performance, and mobility) each contributed similarly to the model, on the margin (Type ill sum of squares). Their overall effects when combined were interdependent. The addition of functional status to a model based on chronic conditions improved the explanatory power of the model but reduced the impact of chronic conditions, suggesting an interrelationship between the categories. As in Wilson and Cleary's model, social support, at least as measured here, had only a peripheral effect. Limitations There are limitations to this quantitative analysis. Almost all data were obtained from self reports. A comparison of questionnaire utilization responses with information from administrative databases illustrates the hazards of self-reported data. For example, of 640 individuals who self-reported no utilization of home health services at 12 months, administrative utilization records indicated that 69 had received at least one episode of home health service during that time. As in the example concerning IADLs and physical performance described earlier, if variables are highly correlated, then a factor retained in the regression model may be explaining the same variation in outcome that an excluded factor affects. In that case, a factor found not to have a significant effect on perceived health in the regression model may in fact be important to the outcome. Very few variables in this study, however, had any substantial correlation with any of the others. 83

PAGE 94

, I : I : I ., The homogeneous nature of the population studied here limits the generalizability of the results. All subjects were community-dwelling Medicare eligible older people with similar sociodemographic characteristics (age, gender, racelethnicity, education. and income), a history of chronic conditions, and access to the same health care. Replacing missing explanatory covariates with imputed mean values also decreased variability. The lack of differences between respondents and non-respondents, however, diminished the negative impact of that replacement. Summary In summary, the quantitative part of the study identified factors that predicted perceived (self-reported) health. as representative ofhealthy aging. Some of the factors reported in earlier studies of negative health outcomes among older peopleespecially smaller numbers of chronic conditions, better physical performance, greater mobility, and absence of depression-showed predictive capability for positive perceived health and in fact explained 38 percent of the variation observed in this study population. Still, the amount of variation not explained by the model derived here--62 percent-suggests that other factors also have substantial effects. As Wilson and Cleary (1995) hypothesized, those factors may include characteristics of the individual and the environment such as personality, values, and societal supports beyond those measured quantitatively in this study. The next chapter further describes interview-based qualitative explorations. 84

PAGE 95

. I I I CHAPTER4 QUALITATIVE ANALYSIS Exact sciences give correct answers to certain aspects of life problems, but very incomplete answers. It is important of course to count and measure what is countable and measurable, but the most precious values in human life are aspirations which laboratory experiments cannot yet reproduce. (Dubos, 1959, p. 279) I would call attention to an unfortunate byproduct of scientific methodology that is ignored by the pathogenic orientation. The good scientist formulates a hypothesis, when there is a basis for doing so, rigorously submits it to testing, and rejoices when it is supported in repeated testing. . The pathogenicist is content with hypothesis confirmation; the salutogenicist, without disdaining the importance of what has been learned, looks at the deviant case. (Antonovsky 1987, p .11) The research described in the previous chapter counted and measured. It produced a statistical model that identified factors, from those commonly associated with negative risks of aging, that also supported positive outcomes. From the model it was possible to compute predicted values of perceived health with which to compare individuals' reports of health status. That comparison supplied a method for identifying, in Antonovsky' s ( 1987) words, deviant cases and allowing further exploration into what makes the difference between those older people who feel healthy and those who do not. This chapter describes the methods used to select and interview deviant, or discordant, cases; explains the qualitative analytic methods used; provides information about these cases; and describes and discusses the results of qualitative analysis of the interviews. 85

PAGE 96

Methods Sample Selection The sampling frame for this study was more structured than that often associated with grounded theory-type work because the regression model offered a method of obtaining what Kuzel (1992) called maximum variation, to benefit from points of view as disparate as possible. The interview sample was selected to include different types of discordance between predicted and reported perceived health status: extreme under-rating, by people who reported poor or fair status compared to better predicted values; moderate under-rating, by people who reported good status compared to better predicted values; moderate over-rating, by people who reported good status compared to poorer predicted values; and extreme over-rating, by people who reported very good or excellent status compared to poorer predicted values. Table 4.1 shows the distribution, by category of reported health status, of study subjects whose reported 12-month health status differed from values predicted by the regression model (total N= 199). Boxes identify the four discrepant groups. For example, there were 61 individuals whose reported fair health status deviated (in a negative direction) between one and two units from the values predicted by the model, on a standard deviation scale. There were 86 extreme under-raters (76 after adjusting for deaths and changes in the providers providing care), 17 moderate under raters ( 15 after adjustments), 11 moderate over-raters, and 85 extreme under-raters (83 after adjustments). Individuals in each group were ordered randomly and contacted in that order. Prospective interviewees first received a letter outlining the project, explaining confidentiality, and requesting permission. Approximately one 86

PAGE 97

:I I week later, they were called to solicit participation and to make arrangements for the interview. Table4.1. Discrepancy from Predicted Health Status Qualitative reEorted health status sample sizes are hard to standardized very excelresiduala poor fair good good lent total predict. The -3 to -2 I 2 8 I 10 interviewing process -2 to -1 15 61 17 I 93 should continue until no -1 to 0 13 74 155 7 249 Oto 1 4 21 83 135 243 new information or 1 to 2 10 29 38 78 2 to 3 1 2 15 18 insights seem to be >3 1 1 forthcoming and the total 34 165 266 173 54 692 emerging theory has been Note: Numbers refer to number of cases. Boxes identify groups discussed in the text. well-tested against a Standardized residual = standardized value of reported rating less predicted value; negative residual indicates under-rating. counterexamples, or disconfirming cases in Kuzel's (1992) terminology. Kuzel advised that "although the rules are not hard and fast, experience has shown that 12-20 (data sources] commonly are needed when looking for d.isconfirming evidence or trying to achieve maximum variation" (p. 41 ). I initially planned to speak with 20 to 30 people to reach saturation, with more from the extreme categories than the other two. Interview Instrument The structured interview questions solicited information specifically about a number of factors derived from the literature discussed in Chapter 2, covering perceptions of health status and reasons for individuals' assessments, well-being, function (valued abilities, activities, and relationships), social support, control, sense of coherence, and personal outlook. In addition, the interview offered respondents an opportunity to speak without structure. 87

PAGE 98

Three Kaiser CHCC participants who were not part of the parent study pilot tested the instrument. They seemed to enjoy the process, and did not find the approximately 45-minute sessions a burden. An expert qualitative researcher sat in on one of the pilot interviews and reviewed the results of all of them. One major change in the instrument evolved from the pilot tests. Based on the assumption that the best way to find out what you want to know is to ask i4 subsequent interviewees were asked directly for their thoughts about factors that cause people to perceive their health differently than expected. Appendix A contains a copy of the final instrument. Structure of the Interviews Participants selected the interview locations most convenient for them. Generally that was their usual Kaiser clinic, but many preferred to be interviewed at home. Each interview took approximately 45 minutes, although one lasted an hour and a half. The instrument provided structure, but conversations tended to expand beyond its borders. Most people seemed genuinely interested in the topic, and they appeared to contribute thoughtfully and completely. Interviews were audiotaped. Undergraduate students hired for the purpose transcribed the tapes into computer text files, for analysis using A TLAS/ti software. I reviewed each transcript for accuracy and made corrections as needed. 1 Analytic Methods i I :I Analysis of the interviews employed what has been termed grounded theory type (Strauss & Corbin, 1990) immersion into the material. The grounded theory method of qualitative analysis has quite specific components, "a systematic set of procedures to develop an inductively derived grounded theory about a phenomenon" 88

PAGE 99

(p. 24). The analysis described here did not follow all of its precepts religiously, but the precepts gave structure to the process. Grounded theory-type analysis begins with open coding, "the naming and categorizing of phenomena through close examination of data" (Strauss & Corbin, 1990, p. 62). In this case, structured interview questions suggested an initial set of codes (e.g., childhood, control, well-being, ability, activity, relationships, social support, getting older). A TLAS/ti software facilitated the process of attaching codes to segments of text in transcriptions of the interviews. Additional codes emerged, representing both refinements of initial codes (e.g., types of medical care as a subset of childhood, loci as a subset of control) and new concepts (e.g., independence, normality, something to do). Expert colleagues reviewed the code dictionary and independently coded randomly selected interviews. There were few discrepancies among individuals' codings. The experts' recommendations were incorporated into the ongoing, iterative coding process. The next step in grounded theory-type analysis, Strauss and Corbin's (1990) axial coding, combines the original codes into categories by connecting them in terms of conditions that give rise to them, properties that are common to them, strategies that guide them, and consequences they share. As the interview process progressed, responses began to suggest an analytic framework, that of the (im)balance between challenges and resources. ATLAS/ti provides tools for constructing and visually representing networks that incorporate these relationships, and the analysis here made use of them. For example, one can illustrate a directional causal relationship between code-factor A and code-factor B (A is a cause ofB) this way: 89

PAGE 100

or show code-factor A as a member of the set represented by code-factor B (A is a B): or a non-directional relationship (A is associated with B) between them: Figure 4.1 illustrates the analytic thought processes that directed the following report of the challenges and resources that interviewees thought important. Expert associates reviewed the analytic process, and their input greatly aided the analysis while also providing validation. The figure is an imperfect representation because people's perceptions are far more complex than boxes and one-way relationships. Additionally, many factors can be challenges or resources depending on degree and on individuals' points of viewarrows might then be headed in the wrong direction. The chart is included here as an illustration, not as a constraint. (The numbers in the brackets refer to, first, the Figure 4.1. Challenges and resources network 90

PAGE 101

.I number of quotations associated with the particular code and, second, the number of other codes with which the particular code has been associated in this and other networks.) The final major step in grounded theory-type analysis is identification of the central component of the emerging theory and the relationships of the other factors with it. Strauss and Corbin (1990) called this process selective coding. The central component, or phenomenon, together with the relationships form a theory of that phenomenon, grounded in the data. The analysis described in this chapter identified major components of a theory of healthy aging. The next, final chapter will combine them with the results of the quantitative analysis to complete the theory-building process. Interviewees Not everyone contacted agreed to be interviewed. Eight individuals refused, five because they Table42. Number of Interview Contacts, by Discordance Group were not interested, Completed Call-back Total two because of Discordance interview Refused a contacted illness, and one Extreme under 6 2 2 10 Moderate under 5 0 0 5 because she had Moderate over 5 3 0 8 Extreme over 6 3 1 10 moved out of the All grOUES 22 8 3 33 area. An additional Note. a Refused = not interested, too ill, moved. two asked to be called back later but were then unavailable, and the interviewer missed connections with another. Table 4.2 provides details. Differences existed between those interviewed and those who refused. By chance associated with random ordering of the sample, members of the parent study's experimental group were over-represented among those contacted. In addition, a greater proportion of the experimental group members, when contacted, agreed to be 91

PAGE 102

, I I interviewed, as Table 4.3 illustrates. Based on HSF data (but not statistically significant, from the Mann-Whitney U test, in part at least because of the small numbers), interviewees were slightly younger than those who refused (p =0.611 ), reported fewer chronic conditions (p = 0.076 and 0.807, chronic conditions identified by patient and Table 43. Interview Contacts, by Parent Study's Group physician, respectively), Parent study's group Control Experimental Total Completed interview 5 (45.5%) 17 (773%) 22 (66.7%) Refused or call-back delay 6(54.5%) 5 (22.7%) 11 (333%) Total 11 22 33 had greater mobility (p = 0.534) and physical ability (p = 0.069), were less depressed (p = 0.251), and reported better health status at baseline (p = 0.486). The discordance groups (all members, not just those contacted for interviews) also differed, as Table 4.4 shows. In every physical and functional category except baseline perceived health (ADLs, IADLs, number of chronic conditions, physical performance, mobility), a greater proportion of the moderate under-raters reported better baseline status than any other group, even the extreme over-raters. More of the moderate over-raters, on the other hand, reported worse status than the others. Table 4.4. Health Status Differences Among Discrepant Groups, Mean Values Discrepant Extreme Moderate Moderate Extreme Baseline variable under-raters under-raters over-raters over-raters E.t: ADLs, any vs. nonea 0.08 0.00 036 0.02 0.000 IADLs, any vs. none6 0.24 0.06 0.64 0.16 0.001 Sum of chronic conditions (identified by patient) 1.74 0.82 2.45 1.42 0.003 Sum of chronic conditions (identified by physician) 1.24 0.47 1.45 1.06 0.011 Sum physical performance indicators 3.01 3.47 227 3.33 0.004 Mobility, least to most 1.74 2.00 1.09 1.79 0.000 Health status 2.61 3.18 2.73 3.98 0.000 Note. a ADLs =dependencies in Activities of Daily Living. 6 IADLs =dependencies in Instrmnental Activities ofDaily Living. c p based on Kruskal-Wallis test. 92

PAGE 103

I Results The following discussion of the results of the interviews has four subsections. The first explores interviewees' descriptions of their health status, the outcome variable for this study, and their reasons for their assessments. The second subsection describes what interviewees had to say about well-being in general, to provide a foundation for understanding and interpreting their subsequent observations about their own health. The third subsection identifies emerging themes, and the final subsection focuses on differences among the four discordant groups. Ratings of Health Status The interview asked respondents to describe their health status and probed for reasons for the assessments. The availability of 12-month HSF ratings, collected as long as two years prior to the interviews, permitted comparisons of ratings over time. Most ratings remained stable. One of the extreme under-raters, however, reported dramatically improved status; pain-relieving surgery in the interim may explain the change. A woman in the middle of an acute attack of shingles not surprisingly rated her current status substantially worse than before, as did a man whose physical condition had indeed deteriorated. As Jylba (1994) discovered from a similar exploration of people's expressed reasons for their health ratings, "health" is context-bound in two senses: first, in their talk people always construct their health status out of individual (albeit culturally shared) elements from their individual life situation(s); and second, the interview situation is typically one in which the interviewee is tempted to take up many different, even contradictory health related issues. In this process, the context of health and the context of discussing health are closely interwoven. (p. 988) 93

PAGE 104

. I The ability to go and do and comparisons with others dominated explanations for ratings, but there were many others as well. One woman said, "I think my health is good, because I can get up and get going in the morning." Another concurred, "I most of the time feel well, I can do all the things I want to do." Still another, who rated her health as seven on a self-constructed scale from one to ten, explained, "I'm able to get around even though I'm not able to do a lot of things." Comparisons were both social and temporal in nature. In general, both extreme under-and over-raters, when asked to compare their health status with that of others their age, felt that they had better health than their contemporaries. For example, an under-rater whose reported status improved over time from fair to good, had been to an informational gathering at a local pharmacy the morning of the interview, and she observed that "at that meeting this morning [I] found out all those medicines everybody takes, so I think I'm in pretty good shape." Her report supports Suls et al. 's (1991) contention that older people compare themselves with a cognitively constructed standard-in this case the number of needed medicationsthat they feel they exceed. This population's experience only partially confirmed another of Suls et al. 's observations, that older people who mentioned others worse off tended to rate their own health more positively. Three people, all extreme under raters, specifically mentioned others worse off, but ten others who were not extreme under-raters rated their health favorably compared to that of their age peers. Comparative ratings are by definition relative and, as Jylha (1994) noted, sometimes contradictory. One woman initially reported that "for my age I would say quite good," but then when thinking of specific other people, added "of course there are some people that do better than I do." Social comparisons require a cohort of peers for comparison. The oldest of those interviewed, an extreme over-rater in her early nineties, responded to the question about social comparisons with the 94

PAGE 105

observation that "I don't know because I don't have many friends my age, ... and I don't have any comparisons to make" but added, "I would say I've been very well." Suls et aL (1991) also found that those who mentioned past or anticipated health were more likely to report poorer health status. Two of the three extreme under-raters who made temporal comparisons did make negative comparisons, one quite explicitly saying, "I feel like I'm going downhill right now." Two of the over raters, however, also made temporal comparisons, one speaking wistfu.lly of how she and her husband "used to dance at [an amusement park]" and the other reveling in much improved physical health. Several under-raters explained their health status ratings in terms of ability to manage or cope with health problems, with an emphasis on the problems and related declines and losses. As one said, "I'm struggling to live a clean life and eat the right foods and do the right things ... I certainly could enjoy life better if my health was better." Feeling well also had an effect, and some felt that luck played a role. "I feel very blessed for my health," said one, who continued, "if you don't have health you don't have anything else to make up for no matter what you have in the way of possessions, material wealth, dollars in the bank." People also attributed their ratings to their satisfaction, or lack of satisfaction, with life. Two extreme under-raters referred to dull or routine lives as contributors, but most interviewees spoke positively, as did this woman: "I've just got so many things in my life that are worthwhile that make me happy." Well-Being Notwithstanding a concern that well-being may encompass more than health, people's definitions of well-being provided a foundation, or perhaps more accurately a vocabulary, for understanding and interpreting their subsequent observations about their own health. Interviewees identified the following characteristics of well-being: 95

PAGE 106

: I physical condition, security, the abilities to do things and be with people, and personal internal characteristics. In general, under-raters spoke more about physical aspects of well-being while over-raters had a more global view, but respondents seldom limited their definitions to any single characteristic. Two of the extreme under-raters referred to physical attributes of well-being. One man defined well-being this way: ''you're able-bodied, you don't have any health problems and do your own work and [handle] problems you have to deal with in your routine." Another also included proper management of problems, while additionally noting the importance of one's attitude. Manageability, as defined by Antonovsky (1987), incorporates the sense of security that one has adequate resources to meet challenges. Again, it was under-raters who incorporated this element into their definitions. One man spoke specifically of security, both financial security and the sense that "mentally everything [is] going OK." Another also mentioned a person's financial situation as a component of well-being, along with physical health, a balanced social life, and a happy marriage. A woman who enjoyed travel with her husband was the only woman who alluded to financial security; she mentioned being able to live in the style you like." About a third of the interviewees referred to ability and social relationships as components of well-being. They defined well-being as ''feeling well enough to do everything you wanted to do and being happy to get up in the morning" or being "happy to get up every day, to be able to get up every day and do what you have to do," especially if without pain. Others agreed that it is being "able to get up and do as much as you want to do when you want to," and one identified '"the fact that you feel well and want to do things and be with people." A woman who referred a number of times to losses in abilities added a more subdued note, speaking of "being able to do some of the things that you used to." Others spoke of the importance of independence. One woman, especially challenged by an immediate acute illness that 96

PAGE 107

made her at least temporarily more dependent than she liked, mentioned care of yourself and being on your feet and doing your own thing," while a woman in her nineties emphasized that she was "able to help myself in all ways.'' A newly widowed woman focused on family and having family around, but, she said, ''it [well being] used to be my work because I had something to do and keep my mind working." Several people included internal personal characteristics such as "'a good mental perception of yourself and your environment and people and relationships around you." Two women described well-being as a sense of contentment and peace, and one added "a sense of being aware that your time is short to learn a lot of things you want to learn." One man emphasized "enjoying life" even when one "can be where it's a pretty rough life." Another referred to those who "can live within their limitations [and] enjoy what they have [by continuing] to aspire but not put themselves in a situation where they are setting themselves up for disappointment and frustration." Interviewees identified a number of things that contribute to or detract from well-being. Many mentioned characteristics of physical health and behavior such as good health or feeling ill, pain, worries about others' health, days when you can't do ., as much, energy (a contributor), rushing around (a detractor), and anger. Other, positive influences included family, friends, happy marriage, and a balanced social life; church and faith; something to do and being active in a variety of things; what others think of you and emotional support; and health care. Respondents believed that family problems, financial constraints, and current external issues concerning government, the news, and pollution detracted from well-being. 97

PAGE 108

i I I Emerging Themes The analysis of people's specific responses concerning factors explicitly explored (e.g., activities and abilities, social support, sense of control) provided information about those factors' importance. As hoped, other issues and factors emerged from grounded theory-type exploration of these interesting individuals' rich and rewarding stories. Getting Older. Many but not all of these older people found getting older itself a challenge. As one woman put it aptly, '4It's not the golden years, it's definitely rusty They expressed concerns about physical decline-for example, one commented that 44I think you just kind of fall apart"-but not everyone thought aging was the problem. One man observed that if one had led as healthy a life as possible, then '4I don't believe that growing old is a major problem ... just the aging process itself does not have to be a major problem." Some found declining ability frustrating ('4J cannot do what I did ten years ago, and I get very angry ... I get disgusted with myself'), and others seemed more accepting ("We're all gonna get older, we're all going to have our aches and pains, and we're all gonna go;" "You can't do a lot of the things that you did or you do, you can't do them as quickly, but that's about it, there's nothing that I can't do that I really want to do"). Several people mentioned benefits of getting older. One woman spoke of compensations: "You're not as frantic as you used to be; you kind of know you can't conquer the world, and you don't really want to." Another admitted that '4sometimes I use my age as a good excuse if I don't want to do [something]." For some, getting older brought a larger and closer circle of multiple generations of family and friends. One of the most physically challenged of the interviewees, the only one almost completely housebound, spoke quite emotionally of the joys of attention from friends and family. For others aging had meant losing those 98

PAGE 109

; I connections: "My best friends are all gone ... and I feel that is one of the worst parts of getting old and surviving; you have to bear the pain of seeing them go." The extreme under-raters appeared to accept or be resigned to limitations due to aging, saying things like "I wouldn't even try to [do something she'd done earlier]," ''I know everybody's got to die sometime," and ''I have reconciled myself with the fact there are some limitations I have to live with now." Over-raters, in comparison, expressed determination to continue to do as much as possible. As one woman observed, "Well, you can either just sit down and do nothing, or you can keep the few things that you like to do and work towards that. When I get up in the morning I usually got an idea of something I'd like to do." A woman in her mid eighties who exercises religiously and continues to create works of art, may have explained the difference: "[Some other] people have an idea that as you get older you're going to be weaker and you're going to deteriorate, so you might as well sit and deteriorate." Burden of Chronic illness and Pain. Information about chronic illness came .I both from the HSF instrument, which included check-off lists of fourteen chronic medical conditions and from the interviews. It is a failing of both the HSF instrument and the structured interview questionnaire that neither asked about pain. I These older people most often reported (in the HSF) or mentioned (in 1 I interviews) arthritis and other musculoskeletal problems, followed by heart conditions, hypertension, and lung disease. Differences existed among the groups, but they are difficult to interpret. Extreme under-raters and moderate over-raters reported the greatest illness burden, moderate under-raters the smallest, with extreme over-raters in the middle. Four of the six extreme over-raters reported cancer--only two others of those interviewed reported any cancer-which suggests that cancer may affect people's sense of health less dramatically than other chronic illnesses. 99

PAGE 110

i I I i I l I Although neither the HSF nor the interview questionnaire asked specifically about it, absence of pain appeared to be a very important contributor to positive perceived health. None of the extreme over-raters mentioned although several reported arthritis and cancer, which can be painful. Almost all of those who specifically complained of pain were extreme under-raters. There are at least two possible explanations. Because the HSF instrument did not assess pain, no data concerning pain were available for the predictive model. It is quite possible that including information about pain would have improved the model and produced lower predicted health status ratings for these individuals. If that were the case, predicted ratings would have more nearly matched their reports. These individuals may therefore have been misclassified as extreme under-raters. On the other hand, they may have experienced no greater pain but have had fewer resources to deal with it. Others also reported painful conditions but did not focus on their pain. One moderate over-rater, a woman with two artificial hips and two lame knees (as well as serious heart disease), spoke of being "happy to get up every day ... and do what you have to do," adding "of course if you can do all that without pain that's good too." Valued Activities and Abilities. A series of interview questions asked respondents to name the activities and abilities that they valued most, or did or would feel worst about not having. Interviewees most often cited mobility or being able to go and do, vision, and mental function as their most valued abilities. Only one person cited hearing; several indicated that loss of hearing would be more tolerable than loss of sight. One person with ongoing stomach problems wished for the ability to eat well. Respondents most valued activities that seem related to the abilities they mentioned--social activities, travel, reading, and housework. Associated with going and doing, individuals mentioned going fishing (as a greatly missed activity), 100

PAGE 111

, I swimming, riding a motorcycle, and observing nature. Additionally, many also mentioned creative activities: dancing, music, creating art, and needlework. Responses to these open-ended requests quite closely matched the ones Berg et al. (1976) elicited when they offered their study members a list from which to choose. Table 4.5lists categories of responses from the twenty-two interviewees, the number who mentioned each category, and Berg et al. 's corresponding most-valued activities and abilities. The sense ofloss of Table 4.5. Valued Activities and Abilities Interviewee category Walk, get around Go and do Drive Travel Vision Read Mental abilities Give advice Social activities Sports/athletics Play bridge Do for others Care for self Housework Shop Other #of responses 11 7 7 7 11 7 10 I 6 3 1 1 3 5 1 15 Bergetal. (1976) category Walk See Use mental abilities Think clearly Make decisions Love and be loved Have contact with family, friends Talk Live at home activities was in some cases substantial. One woman explained her concerns as her husband became more ill: I have quit all my activities. I used to go to [a group] and I quit that, and I used to go to all the theaters and all the plays, and I've quit that. First of all I quit it because I was a little bit afraid of bringing home flu or something and either have my husband or I get sick. I wanted to protect us. Another spoke of how she used to be very athletic and now feels "kind of cheated that I can't do the things as comfortably as I used to." : I A separate question asked about most valued relationships. Almost everyone spoke of friends and family. Three women talked about groups of :friends they had 101

PAGE 112

had for many years. One described a group who have gotten together for dinner once a month for more than forty years: I really value those relationships because those people know you they know your kids, they know your family problems. Those are the ones I value. I have wonderful neighbors, but they're not near as close even though they live next door. They're not as close in your heart. : I Most respondents, even those who had experienced some difficult times with their children, spoke lovingly of family. As one man explained, "I am a very strong believer in close relationships with the family members. We have a beautiful family, and I treasure that, and I value that." Although none of the interview questions asked directly about it, independence emerged as an important contributor to better health among this older population. Three-fourths of the respondents-and all but one of the over-raters made at least some reference to it. They wanted not to lose the abilities to get around, 1 to cope with things, and to be able to care for themselves. A woman with acute disabling discomfort said, "I just want to be able to take care of myself, do my own house, and play cards and stuff like that." One of the younger women noted from her I I I experience with her mother and mother-in-law in their nineties that getting older "makes you less independent" and asserted her own desire not to have help. Many respondents spoke with pride or pleasure of their independence, saying, for example, ''we can take care of ourselves" and "I can do anything I want to." One woman expressed a common concern, "I don't want to go to a nursing home, and I don't want to be a burden." Being dependent on others ''would I guess embarrass me," said one man, who added "I just hope I grow old gracefully." Acknowledging that others didn't mind giving the help he needed, an oxygen-dependent respondent nonetheless regretted his dependence: "I'm kind of getting tired of having to be so dependent on somebody else now ... I hate to impose on [my wife] for everything I have practically." On the other hand, another man who had no such support system 102

PAGE 113

worried about "who's going to take care of me" ifhe lost his ability to care for himself. At the most extreme, in terms of coping with such loss of ability, the father of a physician stated that "I'd rather my son gave me some kind of a pill and put me to sleep and I said good-bye to everybody and that's it." Childhood. Most but not all of these older people indicated that childhood experiences had a lasting impact. Half of the six negative recollections but only one of nine positive ones came from the extreme under-raters. One woman turned to her church and religious faith in childhood to counter a dismal situation that included threats of being sent to an orphanage. She continued substantial church-based service and refused to let the burden of physical discomfort and inconvenience lead to self pity. Two others, with better current physical status, also spoke of difficult childhoods. One, for whom ability to remain physically active defined health, required many years to overcome fears related to his mother's disability: [Her medications] didn't help a great deal and so I lived with a tremendous amount of anxiety about that personally because I was worried about her. I was worried about myself: was I going to also develop this ... and I was blessed and fortunate that I didn't have to contend with that. The other, a woman in her early eighties who attributed her satisfying and perceived healthy life at least in part to self-reliance, explained that ''my immediate family was not a really close family, and that may be part of ... the way I feel ... because I knew : I I had to take care of myself." The family's failing (in her opinion) had encouraged her to develop an attribute valued in the social context of American life. The availability of health care during childhood varied by location (farm vs. town), income, and family values. Comments such as ''we lived way out in the country, we never went to the doctor" and ''you never went to the doctor unless you were nearly dead, it cost money to go to a doctor'' expressed one end of the range, but another person remembered that they "put me to bed, called the doctor-at that time 103

PAGE 114

doctors came to you." Sometimes it was the whole community that provided care. One man described life as the child of a poor minister in a small town: "the whole community served as a support group to some great extent; they made a lot of illness, and they thought that prayer was a major factor in the healing of any type of ailment." A woman who fell seriously ill as a young adult remembered that "the whole town rallied about that." and when her sister struggled with tuberculosis, ''you felt as though everyone in the whole city, it was a small town, was your friend." Health care delivered at home could be pretty rough. One person described ruefully but with a laugh the treatment for croup: '"my dad'd give me horseradish and put in my mouth, or kerosene with sugar in it ... now what kind of stuff, quite a remedy, but I survived it." Another recalled getting mustard plasters and, when asked about the face he made when he spoke of it. allowed that "I don't like mustard for some reason, maybe that's why." A woman whose family never went to the doctor very much, and who now devours all the health information she can find, told of the response to stepping on a nail. "I screamed, and my brother came out and pulled the nail out and poured iodine in. Then when my dad got home from work, I had to soak it in hot Epsom salt water and that was his amount of medicine." Not all treatment hurt so much, and many reported gentle, loving care. One woman remembered her mother's medicinal arsenal, which included Carter's Little Liver Pills, Vicks VapoRub, Vaseline, and mercurochrome. Another reminisced about her hard-working and loving mother, who raised eight children on a farm: If I was sick, she'd make me come and lay out on the couch, and she'd be working in, like, the dining area. She'd either have her sewing machine out or she had her ironing board out or she'd have something. You were always there, you weren't in a bedroom with the door shut. The whole family was there, right there to help take care of you. Social Roles and Social Challenges. As noted earlier, most respondents, when asked to compare their health status with that of others their age, felt that they had 104

PAGE 115

better health than their contemporaries. That comparison, social in nature. provided an example of how people used social norms as a standard against which to measure their own social roles and status. Parsons (1951) and Rosow (1974). among others, emphasized the importance to health of being able to conform to society's expectations and norms. A number of the interviewees referred quite explicitly to normalcy or the perceived standard for older people. They said the following (italics added): I think most of us [would really hate to lose the ability to drive]. I think we all worry [about being dependent]. I think this [thinking too much if one doesn't keep busy] is kind of a natural tendency. My husband and I make a lot of compromises ... I think that's normal or should be. You think about how [decisions will] affect your family, and I think that's true with any normal person. I think I'm moderate or about average. [Concerning family problems], we hear that so much with difforent families. [Getting information] on you're getting older and you can expect that you're going to do these things and here's what to do; I think that would be important. One of the respondents confirmed Mechanic's (1978) concern for the negative social impact of loss of work involvement. The interviewee said, "I finally gave it up and decided just to not struggle with the fact that I am retired, and I should accept that and leave it at that and not keep searching for something that's going to substitute for my employment in the past years." Two of the extreme under-raters expressed disappointment that retirement had not lived up to their expectations, one because physical problems cut short the world travel he had so anticipated and the other 105

PAGE 116

: I I because he felt he didn't have the financial resources to do the things he'd have liked to do. Interviewees identified a number of other social challenges. Pain and disability have social ramifications, as one woman discovered when she had to quit her job because of pain. Surgery ultimately resolved the physical problem but could not restore her employment and associated sense of worth: "[what contributes to well-being] used to be my work, because I had something to do." Then, because she feared bringing colds and such home to her increasingly ill husband, the same woman also quit most of her other activities. Having a spouse did not necessarily guarantee a positive perception of health, especially if the spouse had poor health and required caretaking or if the respondent felt overly dependent on the spouse for care. For those whose spouses had died after long illness, the length of widowhood appeared to matter. Three of the five longer term widows (there were no widowers in this sample) were members of the extreme over-rater group. One of the others had to raise children while coping with her husband's thirty-two years of debilitating illness; she felt that the burden had meant she lost track of other family. Several mentioned worries about the health of family I members, especially spouses .. I Cooperative marriages, on the other hand, appeared to support positive perceptions. Six of the nine who spoke warmly of such relationships were over raters. One explained, with laughter, how she and her husband manage: "We make two separate grocery lists, one for him and one for me, and two baskets, and we put them together, and I get to go sit on a chair and he checks them out, and we go home." "And of course we do have each other," another responded to a question about support; "I depend a lot on [my husband], he does most of the leg work in our twosome." 106

PAGE 117

'I The interview instrmnent did not include questions about sexual activity, but two respondents, one male and one female, expressed regret that their spouses no longer were interested. A third spoke of the pleasme of a "very desirable spouse" and continued sexual satisfaction. The former were extreme under-raters and the latter an extreme over-rater. Social Smwort. Several interview questions addressed social supportavailable, wanted or needed, and reciprocated. Responses supJ}orted all five of the theories cited by Stewart (1989) concerning why social support has an effect: attribution, social-exchange, social-comparison, loneliness, and coping. One man attributed the dullness of his life to his lack of financial resources, blaming the government for the lack of Social Security income for his wife and for insufficient Medicare support for medications. Many respondents mentioned reciprocal support and the desire to keep exchanges equitable. "We have a prayer list at church, and we pray for each other," said one woman. Another told of her daughter-in-law who shoveled the snow, "and I asked her how I could repay her-' Just buy lots of Girl Scout cookies,' she said-so I bought twenty-five dollars' worth." A much older woman reflected on the occasional need to allow inequity: "I don't like the idea of asking [a friend] to do anything for me, but on the other hand there are probably some times when we need help, we need to accept that from others." Social comparison, discussed earlier with specific reference to health status appraisal, requires information about others as a basis for the comparison. Interviewees looked to "family and your close friends if you need some opinions or advice" and relished "being able to answer questions." illustrating the relationship between loneliness and social support, a man long divorced and without close family described fear and loneliness and his resulting need for support: 107

PAGE 118

II I Maybe this is common, maybe it's a nightmare. You know, I live in an apartment. If I die, I might be laying there for months, nobody to check on. The mail would build up and so forth. People would call, but they would think, "Oh. he's out." The friend he had depended on was now quite ill himself, and another friend "kind of like an older brother" had recently died. He mourned, "that [death] was unexpectedly, and it didn't make me mad, but, you know, who can I talk to?" Coping is a larger issue that involves all the varieties of social supportemotional and instrumental. Social support, at its best and most effective, provides enough of the right resources for people to use to meet challenges. When asked about available social support resources (whether needed or not), almost everyone referred to family and friends, for all kinds of assistance, and quite a few mentioned health care or health information. On the other hand, seven of the eleven over-raters said they neither wanted nor needed any additional support. Most of them did in fact subsequently name some desired resources, but their first reactions were that they needed and wanted nothing. None of the under-raters responded similarly. Examples of available social support included: "I get good support from my family." "I have emotional support from my husband and family ... I have a niece who is a muse ... so she and I look things up." "I think your family and your close friends [give support] if you need some opinions or advice." sure I could call the church and get anything I want." family, they're all fantastic, believe me, they come and check on Mom and Dad and take us places." From a woman whose husband's health had recently deteriorated, leaving her weary and worried: "By her [a nurse practitioner who took the time to offer some emotional support] talking, being interested in, asking about me and how I'm doing-everybody says well how's [my husband] doing and she talked a little bit ... and when we were through talking, I could just feel a weight lifted." 108

PAGE 119

: I Although they appeared to have quite a few available sources of social support, many respondents described additional support they needed or wanted, such as the following: "rdjust want friendship, some compassion, understanding, empathy." "I suppose I need, everyone needs, someone to lean on ... because you can't be up all the time." "'r d sme like to find a handyman." "I've got to get somebody out here to clean house because it really needs it." "It's pretty hard to get me to the doctor." Support received was not always desired. "I get so unhappy with my grandchildren," said one with laughter, "they ease me down and help me up, and when I walk into a restaurant, they come take my arm as though I can't walk and things like that, but then I let them do it." Another felt similarly: "My relatives that look after me a little bit, sometimes I could do without some of that." There were some other differences among the groups. Moderate under-raters reported fewer available or desired instrumental resources, with more desire for emotional support. Over-raters indicated they had more instrumental supports available than the other groups, although as noted above, they also said they needed and wanted none. Over-raters, in fact, identified a larger number of both available and desired resources overall. Extreme under-raters identified about as many available resources as the over-raters; they expressed fewer emotional needs and wants than the other groups. Relationship with Provider. A positive relationship with one 's physician provided another resource. There were clear differences among the groups in their descriptions of their relationships with providers. Most of the extreme under-raters limited their participation in the relationship to supplying information and complying with the provider's authority, as opposed to true collaboration. As one said, ''I try to 109

PAGE 120

. I follow directions and take my prescriptions and everything according to order." Moderate under-raters thought they had responsibilities but mainly at the level of discussing options. Moderate over-raters felt more involved and more like active members of a team. One related a story of the source of the Permanente part of Kaiser's name and the importance of teamwork: In California, there are two rivers that come together, and the river that they run into is named Permanente. The doctor and the patient represent the other two. They come together, they talk, converse with one another, and the result of their coming together is the making of good health. The Permanente makes a beautiful valley that grows everything, and so a better life, because its basics come together .... I think that makes so much sense for the doctor and the patient to be able to converse with one another. Extreme over-raters emphasized the patient's responsibilities. One declared, "Oh yes, I, yes yes yes" contribute to decisions about health. Another explained that ''the patient has to be somewhat responsible rather than. just have a doctor make all the decisions; you have to make the choices." Many interviewees commented on the importance of communication and mutual respect in their relationships with providers. They didn't always agree, however, on the appropriate amount of information they wanted. One woman expressed "reluctance to really confide in the doctor because I have been made to feel a number of times like why are you asking about that, and I think I would like to see more doctors open and explain to people." Another, however, was relieved that "they never mentioned cancer'' although she felt pretty sure that was the problem. Several mentioned the benefits of a caring manner. One said, ''I think [my provider] is interested in helping me," and another compared his "satisfaction in knowing that those people are there, and they're interested in you" to ''other times [when] you get a sense of impersonal detached connections at best." 110

PAGE 121

I I I Having to change providers interrupted communication. One person explained, "You get a good doctor ... and first thing you know he's gone somewhere else and you have to start over.n Another worried that changing doctors "is going to be hard, and I have no idea who we will go to." An extreme over-rater reacted to her doctor's retirement and subsequent fragmented care: "One day we got kicked around them all, and I said, finally I said, 'I'm getting tired of this, I'd like to settle down to at least one or two doctors instead of being kicked from here to there,' so they did." Some of the women reported perceived gender and/or age bias from their providers, another type of social role challenge. "The doctors do not listen to women the way they do to men," said one, "because when a woman has a headache they (just] expect her to, but when a man says he's got a headache, they give him all the tests they can to see what's wrong." Another allowed that "I don't know whether it's age or sex; however, I think my husband gets much different answers sometimes." She also wished doctors would "explain to people who are getting older the way they do when you are younger ... in your thirties you're told how your body is reacting, and then as you get in your forties you're told about menopause and you get in your fifties and it's there; you get into your sixties and nobody says anything about what's happening to you." One of the older interviewees reported that "the first time I suggested [surgery] to [my doctor] he said my veins were too small and I was too old." She acknowledged that "they're kind of getting away from doing expensive surgery on older people, and I understand that" but did finally insist on the surgery, which tremendously improved her health and life. Locus of Control and Sense of Self. A set of interview questions probed for information about the extent to which these older people felt they had control over their health and their lives. Kaiser encourages its members to take an active role in their health care, especially within the group-based format but also in general, and it may be that people who wish active involvement self select Kaiser because of that Ill

PAGE 122

reputation, so it is perhaps not surprising that most felt that they had control over their health. One woman explained, ''I control my health by what I do, how I live;" another said similarly, "Control over my health means what I do with it." People did not always identify their locus of control unequivocally as internal or external. One man who indicated that he could control his health behaviors also commented that "just out of the clear blue sky [a bad health incident] can happen to anybody, I guess." Another man who said he had little internal control did in fact very actively conduct research on types of supplementary oxygen and then convinced his physician to supply the type he found to be best. One man explained the importance of internal control, the comfort that comes from feeling that things don't just happen to you: "I am concerned about things that I can't control because I feel comfortable with what I can control; I just feel like that anything I can control myself I'm all right with it." Over-raters were slightly more likely than under-raters to indicate an internal locus of control over life. They said things like "I think I feel like I have control over the future," and "I think I pretty much control what I do.'' Extreme over-raters tended to be even more emphatic: "Well, I think I'm totally in charge of my life," said one, and another agreed, "Well, I think I'm truly in control." Again, though, several made contradictory statements, such as ''basically I feel like I can do what I need to," but then "sometimes I don't feel like I'm in control at all." Related to locus of control (internal vs. external) is an individual's sense of self--self-control, self-esteem, and self-confidence. One woman stated, "I just believe in myself so strongly, that anything I can control is not a worry and not a problem." Another woman attributed her lack of self-esteem to comparisons with an older sister and not having "a good mental perception of yourself and your environment and people and relationships around you." Both of these women, with such dissimilar perceptions, were under-raters. A male over-rater with difficult 112

PAGE 123

I I :I childhood experiences talked about having to overcome a lack of self-assurance: "I think I grew up with a real lack of, you know, confidence, for various reasons in my life [which he then listed]; anyway, it didn't give me a feeling of that I had any abilities." Another over-rater, a woman, may have explained how he recovered from his experiences. She said,"' don't know, but I'm sure when you've had a lot of problems and you've worked your way through them, you're more apt to feel that it will work out." Just as an internal locus of control seemed to support better perceived health, so did an external focus. One of the extreme under-raters could speak only ofhimself and what he did or didn't do. In contrast, a woman who acknowledged tendencies to feel down occasionally, spoke of how visiting sick and shut-in church members ''makes me feel very good after I do something like that." Another said, think if you can look at somebody else and see where you might be able to help them, that helps you." A house-bound over-rater asserted that "only two things are really worthwhile ... what you eat and what you can do for somebody else." An awareness of self or condition had both positive and negative impacts. In some cases, learning more about health behaviors and their own conditions increased the sense of control and provided tools for coping with health problems. Several people mentioned specific classes (and CHCC group presentations, for those who participated in CHCC) that had given them useful information. On the other hand, as Mechanic (1986) cautioned, awareness did sometimes breed worry and anxiety. One woman described that feeling: Sometimes if I get some vague little physical feelings, you know, one can, I think, as they get older, can imagine, can let the imagination run away and say, "Oh my gosh, it could be this, it could be that." Another worried that sister was about nineteen years older than I was, and everything that happened to her in her life about twenty years later it would happen to 113

PAGE 124

i I me." The following exchange concerning the absence of awareness may better explain the possible negative aspects of its presence. When asked if it had been frightening to be out of control with her then untreated bipolar mental illness, the respondent answered, "Nope and I'll tell you why-because you're not aware of it." Several people descn"bed ways of dealing with worries. One woman, whose father became a zealous nutritionist in response to illness, reacted, "'I have met people that are so overly concerned that I thought r d rather be unconcerned and try to make my way." Another's father offered a different example: My father always said, ''Don't worry about anything, because if you worry and it happens, you've worried for nothing, and if you worry about it and it doesn't happen, you've worried for nothing," and I think I've kind of lived by that. Attitude. Almost from the first interview, attitude emerged as an important, perhaps the most important, contributor to positive perceived health. When asked to explain the difference between people who think of themselves as healthy, even though they may have serious physical problems, and people who think of themselves as less healthy, although they may have fewer problems, respondents repeatedly spoke of positive attitude: "It's attitude, see, instead of looking out and seeing the problem, look out and see something else, something nice.'" "They can still keep smiling and have a positive outlook." "A negative attitude is very destructive, and a positive attitude is always in the direction of a better way of some kind." "Getting older is no fun, that's for sure, but I think a lot of it is in outlook." "I think it's a person's mental attitude about life." "They don't dwell and look inward and feel sorry for themselves .... I think if you can look at somebody else and see where you might be able to help them., that helps you rather than being 'poor me."' 114

PAGE 125

''Some people are comageous, and they somehow overcome the obstacles; ... other people may have minor things and complain like it's the end of the world. That's the way people, you know, evolve in their personalities and their make-up." "I really do think just being positive and not sitting around and groan and moan." "I think it's the nature of the person." "I truly think that your mental attitude has more to do with it than anything., or its reverse, a poor me attitude: "People make mountains out of molehills." "I've often looked at people and wondered, 'Why are you feeling so sorry for yourself? Why do you complain all the time?' And then the sky can be as blue as blue can be, and somebody'd bitch about it. ... Every one of these people want to blame their childhood, ... and they want to blame other people .... I really do get angry with 'poor me.'" "I think it goes back to their mental attitude ... if they have this negative attitude toward themselves and toward others, then I think it all kind of closes in." "It was just like 'feed me, water me, whatever,' and they weren't going to try and help themselves." "I know one lady here in the building who I would say is more or less a hypochondriac. and I'm so sorry for her. I mean I wish she could get the strength to overcome these things." "Some people's dispositions are maybe they're born with a more negative outlook on things." Nineteen of the 22 interviewees mentioned attitude, and many were able to suggest factors that they thought contributed to it. Categories of reasons corresponded closely with factors already discussed above: physical issues; childhood experiences; locus of control, focus, and sense of self; and faith. One woman suggested that I have friends who are just tired." Another 115

PAGE 126

. I I I : I I : I could compare her own attitude before and after surgery that greatly improved her physical status; her outlook changed from near-suicidal despondency to satisfaction. About half of the respondents felt that upbringing played a role. One spoke of a "person's earlier life and the ability or inability to accept responsibility for one's actions." Another remembered his father's trust and belief in him ''that built up my own attitude [and] has lived with me all my life." A third explained the connection between childhood and attitude: I think that certainly one's upbringing can be a factor, in what you're trained and taught and what your environment, what you're exposed to. Those experiences form your emotional responses to adversity and for the positive things life provides for most of us. And then you're in command as to whether you're going to feel sorry for yourself and think the world is just the most awful place One man suggested that life experience, including but not limited to childhood, made the difference: [Attitude comes from] how they were bought up and the number of friends and family supports they had. There's some people who do not have the family backing, or their parents have passed away, or anyway they don't the closeness of a family in their later years or do not have a close circle of friends and they feel like nobody loves them .... They, you lose your vigor or lose your outlook on life, and it's sad, really sad. Several people spoke of control or willpower. One said, "If you've got something to do and you make yourself do it, you kinda forget your problems." Others spoke of those who ''have a lot of willpower and control over their life" and ''the ability or inability to accept responsibility for one's own actions." They thought people developed negative attitudes because ''they don't try to help theirselves to alleviate their problems," "they're just too lazy to get up and get on with it," "[they] just sit down and give up, which is what I'm sure a lot of people do," and ''they give up so easily." They also attributed a negative attitude to an inward focus. One 116

PAGE 127

woman suggested that "they don't want to give of themselves, and they don't want to help." Another mentioned some people's need for attention. A number of the interviewees thought faith supported a positive attitude. One made a direct connection, saying, "Both of us have always thought positive-we go to church and we believe in God." Another explained, ''I think [the reason for my positive attitude] is my faith, because I just am so thankful for what the good Lord has given me." Differences Between Under-raters and Over-raters People with various types of discordant reported and predicted health status ratings were selected for interviews to explore the possibility that some factors better distinguish between more and less positive perceived health. As already mentioned I throughout the preceding sections, there were in fact differences among the groups. Over-raters tended to be more assertive and active, even feisty, while under-raters were more cautious and accepting or resigned. As an example, extreme under-raters 1 generally limited their participation in the patient-provider relationship to providing I information and complying with directions, while moderate under-raters accepted a bit more responsibility. Moderate over-raters felt like active members of the team, and extreme over-raters placed emphasis on their roles and responsibilities. Similar differences arose in descriptions of aging. Extreme under-raters accepted limitations I to aging and were resigned to them; over-raters expressed determination to continue to do as much as possible. All but one of the over-raters mentioned the importance of independence. Only three of the six extreme under-raters did so. Over-raters, especially extreme over-raters, were more likely than others to indicate an internal locus of control, to feel in charge of their lives. Under-raters more often defined well-being in terms of physical condition, while over-raters more often took a more global view of well-being that incorporated 117

PAGE 128

doing things and being with people. Extreme tmder-raters and moderate over-raters j I reported greater illness burdens than the other two groups, although extreme tmder raters reported more pain. They reacted differently to that challenge. Two extreme : I : I : I I : I I I I I i under-raters referred to dull or routine lives; moderate over-raters were more likely to just sit down and do nothing." Although they had greater numbers of resources available than under-raters and even mentioned a few others that would be desirable, many over-raters reported that they wanted and needed no social supports. Extreme under-raters indicated they had almost as many available resources as the over-raters, but they expressed fewer emotional needs and wants. These responses matched the great desire of over-raters to remain independent and the tendency of extreme under-raters just to accept. Moderate under-raters reported fewer desired or available instrumental social supports and had more desire for emotional support. Moderate under-raters seemed to have some innately different characteristics that perhaps explained their lower-than-predicted ratings. They generally offered cautious, conservative, thoughtful responses, and they seemed to process the questions more cognitively than emotionally. This response by one of them provides an example: "I've never really thought about that, I can't answer that in a second, I have to think about that." One woman alluded to tendencies toward depression, and another's bipolar mental disorder was controlled by medication. Their conservative ratings of their health status appeared to be indicative more of their personalities than of any specific challenges or resources. Of the factors described by the interviewees, then, the ones that most differentiated between those with optimistic health status ratings and those with more pessimistic ratings seemed to be primarily attitudinal-independence, assertiveness, the determination to continue to be active, and the desire to take charge. 118

PAGE 129

Summary of Qualitative Findings As described earlier, interviewees were asked several different types of questions: how they rated their health status and why, how they defined the more global construct of well-being, what was important in their own lives about a number of factors suggested by the literature (e.g., valued abilities and activities, sense of 1 control and coherence), and what they thought were the reasons that different people perceived apparently similar health states differently. :I : I I I I I I Summarized below, responses to all types of questions seemed to fall into similar categories. The summary also notes the distinction between factors elicited directly and those that emerged from people's stories and explanations. The framework of challenges and resources helped to structure the analysis and continues to be useful here. One frequently repeated item, though, seemed more central and more important than the others. Going and doing deserves its own subsection. Going and Doing Over and over again, respondents spoke of going and doing something meaningful. At first that seemed like just one of the resources people bad (or didn't have) that supported perceived health. Increasingly, however, it emerged as the outcome they were talking about rather than a factor related to it. They said the following: "I like this class I'm taking [to maintain nursing credentials] ... it keeps your mind alive, it keeps you interested, it keeps you going now." "I tried to retire at 68 but developed aches and depression. You have to have something you're expected to do." "If you've got something to do and you make yourself do it, you kind forget your problems." 119

PAGE 130

I I I I "They don't have enough to keep them occupied is why they're always negative." "Another thing too, as long as you can get out and do things and feel good about yourself, that helps this aging." "If I didn't have something to wake up for, I think I might be a basket case." ''You strive until the goal is made then look for one still unattained." well you can either just sit down and do nothing or you can keep the few things that you like to do and work towards that. When I get up in the morning, I usually got an idea of something r d like to do." ''But we do try to get out a little bit every day, and I think that's important-get up and get doing, do what you can." "You have to keep going, you can't just sit down and give up." "We go constantly; we don't stay home." Going and doing didn't have to involve physical activity. One man, quite limited physically, expressed that "I enjoy getting to the root of things, finding out why it works; it's just a pleasure." Another reported, "I have tried, you even not being in school. to maintain a certain mental stimulation in my life by reading or participating in things that give me some mental activity." In this population, being healthy meant going and doing. The final chapter, combining results of the quantitative and qualitative analyses to produce a model of healthy aging, will pursue this further. Challenges and Resources Interviewees identified pain and some combination of getting older and dealing with an increasing burden of chronic illness as major challenges to their perceived health. Uncontrolled pain-or perhaps focus on discomfort-was associated with lowered health status ratings. Some people thought aging and illness were inevitably intertwined ("I think you just kind of fall apart."). For others, aging 120

PAGE 131

and illness were independent factors (''Just the aging process itself does not have to be a major problem."). Getting older, for some, had benefits-a larger network of friends and multiple generations of family, fewer obligations, and less pressure. Feeling healthy depended on mobility, vision, and menta/function to suppon valued activities such as participating in social activities, reading, travel, caring for self and home, and creative activities. From individuals' identification of specific valued abilities and activities, independence or autonomy emerged as a most important contributor to health. Three-quarters of the respondents made some reference to it. Interviewees spoke of social issues both as challenges and as resources. Childhood experiences had at least indirect effects on current perceptions of health for some of them. Very difficult childhood situations appeared to foster self-reliance even if it took some years to achieve. In general, though, happier recollections were associated with more favorable health status assessments. Except for several comments about perceived age (and gender) bias from providers, most of these people did not speak directly about the social role of older : 1 people in our society. They did however refer repeatedly to normalcy or perceived I standards for older people, and a number of them made social comparisons to explain their ratings of their health status. One man made specific reference to the difficulty of adjusting to retirement. Similarly, a woman who had lost her job because of I I problems with pain associated that loss with a decrease in well-being, or at least the need to find alternate sources. Other social challenges emerged from the interviews. Long-term caretaking of an ill spouse or, from the other direction, dependence on one 's spouse due to illness created a substantial physical and/or emotional burden Cooperative marriages, on the other hand, supported independence and the ability to continue life 121

PAGE 132

in its accustomed manner. Lack of spousal interest in sexual activity challenged perceptions of health. For almost everyone, social support provided substantial resources to deal with challenges. All respondents valued relationships with family and friends. Several mentioned very long-term connections with friends. One mail, however, did not have close family and had lost valued relationships with two confidants because of death or illness; his sense of isolation had a negative impact on his perceived health. Respondents identified a wide range of available and desired social supports, both emotional and instrumental (including information as well as tangible services). Several also referred to problems with the wrong amount or type of support, either because help was in a way insulting or because it couldn't be reciprocated. Relationships with health care providers varied. More collaboration between provider and patient, a cooperative team relationship, appeared to support better perceived health status. Respondents valued a caring manner, clear communication, and mutual respect. I lllich (1976) spoke of the importance of .. autonomous personal, responsible I I coping abilitf' (p. 6) to health, and these older people agreed. Both in response to specific questions and more generally, they spoke about personal characteristics, responses, and resources. In general they thought they had control over their health and their lives. A good sense of self contributed to perceived health, as did the ability to focus outward toward others. Awareness sometimes generated worry and decreased perceived health, but that was not always the case. For some, learning more about health behaviors and their own conditions increased the sense of control and provided tools for coping with health problems. Attitude emerged as possibly the most important factor of all. Respondents felt that a positive attitude supported perceived health, and they suggested that a poor 122

PAGE 133

me attitude contributed substantially to a person's negative outlook on health. They attributed attitude to many of the factors already described: physical issues, childhood experiences, personal characteristics, and faith. The factors that this group of older people identified as important to their perceptions of their own health in particular and to differences in how people in general rate their health, can be categorized as follows: determination of something worthwhile to do: social role and normalcy, valued activities; abilities and challenges: mobility, vision, and mental function; independence; burdens of pain, chronic illness, and/or getting older itself; life experiences; taking care of or depending on one's spouse; resources: family and friends, cooperative marriage, relationship with one's provider, faith; and personal attributes: attitude, sense of self and awareness of self, internal locus of control, external focus, assertiveness, the determination to continue to be active, the desire to take charge. Comparative Case Studies The following case studies illustrate some of the characteristics and differences summarized above. The individuals are in some instances composites, and in all cases personal identifying information has been changed to protect confidentiality. Ralph A. and Lucy B. had each faced substantial and painful physical challenges, hers perhaps a bit more extensive. Disability and discomfort put limits on what they could do. At least, one assumes both suffered discomfort. Ralph spoke a great deal about discomfort; Lucy, who had bone disease and breathing problems, said little about it and tended to joke about it when she did. Both talked about happy, supportive, cooperative marriages, but Lucy was concerned about her husband's deteriorating condition. Ralph reacted more passively to his situation. As he said, "I 123

PAGE 134

I accept the things that I can't control ... I have reconciled myself with the fact that there are some limitations I have to live with now," but he seemed to regret what was lost. Lucy, on the other hand, substituted new activities for ones she could no longer do. She was determined to keep going, and her response to difficulties was, "If things get real frustrating, I just cuss and then pray." Although the model predicted good health status for Ralph, he rated it only fair. For Lucy it was the reverse; the model predicted fair status while she said her health was good. Rebecca C. and Anna D. were both extreme under-raters whose perceived ratings actually may have been more accurate assessments than the predicted values. Rebecca had all sorts of ailments that had required surgery and caused pain. She felt that she was on a downward path. Anna too had experienced successive problems, including surgery with uncomfortable results. She expected her disease to return, but considered the surgical results an extension of her life and an opportunity to see her children grow up. Rebecca felt she had little control over health or life in general-'"! live day to day whatever is possible that day''-and attributed her attitude to her mother's pessimistic view of the world. Anna, whose childhood was also quite difficult, had great religious faith and declared it important not to feel sorry for herself but to "get out and do things and feel good about yourself." Although their health ratings and conditions were similar, they appeared quite different, Rebecca subdued and Anna upbeat and busy. Martha E. had a very complex medical history. Jack F., on the other hand, had few physical problems. Martha emphasized the cooperation with her husband that made it possible to stay independent, and she spoke of available family support and affection. Jack also was married and family oriented. He took pleasure from serving as househusband to support his wife's creative efforts; being a househusband after years of his own employment did however mark a change in social role. Although aging had not yet much affected him physically, Jack seemed quite aware of aging-124

PAGE 135

I I I .I related changes. He talked about caution and moderation, and he made repeated comparisons with normalcy and average. Martha, in contrast, emphasized coping: "You just go and there's some things you can't do; you just have to say no and find something else to do and be happy about it. There's no point in being upset, it's not going to help a bit." The predictive model generated a fair rating for Martha. She however perceived her health as good. For Jack, the model predicted his health as very good; he rated it only good. Limitations Possible limitations to the qualitative portion of this study include interviewee selection, history and maturation, and questions about the rigor of qualitative research. The interview sampling frame depended on a regression model that provided health status values to compare with study subjects' reports. If important factors were missing from that model, then the model may have inaccurately predicted values Reports that appeared discordant with predicted values might in fact have matched predicted values from a more complete model. The absence of any data concerning pain may have decreased the predictive value of the model that defined the interview sampling frame for this study. A number of people chose not to be interviewed. Those who refused reported worse baseline HSF characteristics, but none of the differences were statistically significant. A substantial length of time passed between collection of the quantitative data that determined the sampling frame and the interviews. That delay introduced concerns about the effects of history and maturation. In almost all cases, however, respondents made the same assessment of their health status during the interview that they reported on the 12-month HSF. Erlandson, Harris, Skipper, and Allen (1993) provided a framework for determining the quality and rigor, or trustworthiness, of qualitative research. 125

PAGE 136

I : I 0 Trustworthy work must be credible, transferable, dependable, and confirmable. To meet the standard of credibility, this study incorporated extended engagement with the interviewees, comparison with quantitative data, and periodic debriefing by expens in this type of analysis and in gerontology. Purposive sampling and thick description, according to Erlandson et al., support transferability by providing enough information for other observers to judge the applicability of the results to other contexts. This study's sample selection procedures emphasized discordant cases in order to include divergent as well as typical data, and interviewees were encouraged to elaborate on their responses in order to generate thick description. Dependability means that another person might reasonably draw the same conclusions from the data. All of this study's interview tapes and transcripts, as well as the interviewer's notes, remain accessible, providing a dependability audit trail. Confirmability concerns the degree to which the results of the research are products of the inquiry, not of the researcher's biases. As with dependability, the best guarantee lies in a well-documented audit trail; all A TLAS/ti coding, memos, and network records from this project remain available. 126

PAGE 137

CHAPTERS A MODEL OF HEALTHY AGING I pray for work to do and strength to do it. (Words of wisdom from beloved Nanna., Myrtle Belyea Graham) Chapters 1 and 2 explained the background and justification for exploring factors that contribute to positive perceptions of health by older people. A review of the literature identified factors that studies have associated with negative outcomes and other factors less rigorously tested that may contribute positively, and it surveyed some existing models of health. Chapter 3 described a regression model that identified which of a set of primarily of physical condition and function, contributed significantly to perceived health (Specific Aim 1 of the study). That model permitted computation of predicted values of perceived health to compare with individuals' own assessments (Specific Aim 2). Interviews with a subset of the study : I population, randomly selected from those whose reported assessments were discordant with values predicted by the regression model, provided additional information. Interviewees confirmed the importance of some factors suggested by the literature and introduced others. Chapter 4 described the methods and results of the qualitative analysis of the interviews (Specific Aim 3). The multiple methods of analysis generated several different types of information: self-reported questionnaire data; numerical representations of the size and significance of variables derived from the questionnaire data; peoples' talk specifically about themselves; and peoples' talk about aging in in everyone. 127

PAGE 138

. I This chapter will synthesize the results from the previous chapters to construct a model of healthy aging for this older population (Specific Aim 4). Review of Results Based on the regression model, the following variables had significant effects on perceived health (see Table 3.1 0 for complete details). The variables are listed here in descending order by the size of the effect: chronic conditions getting worse; each better baseline level of mobility, change in mobility at 12 months; depression at baseline, change in depression at 12 months; any baseline dependencies in Activities of Daily Living (ADL), change in ADL dependencies at 12 months; living with spouse; each baseline chronic condition identified by physician; each baseline chronic condition identified by patient, change in number of chronic conditions identified by patient at 12 months; each increased level of education; each positive baseline physical performance indicator, change in physical performance at 12 months; and each day of hospitalization during the 12 months. The feeling that chronic conditions were getting worse (negative effect), mobility (positive), and depression (negative) had the largest effects, of nearly half a step each in health status rankings (i.e., a half step increase or decrease from one ranking to the next higher or lower ranking, of possible rankings from poor to excellent). As expected because ADLs represent very basic living skills, the presence of ADL dependencies had a significant negative impact on perceived health. Most of this population, however, had few ADL dependencies; only seven percent of the study members reported any. Variables with the smallest effects-education, having fewer 128

PAGE 139

chronic conditions at the end of the year, and being able to perform any one of the physical performance tasks-added about ten percent of a step, and each day of hospitalization decreased perceived health status by a small amount. Several factors found in previous studies of poor outcomes did not contribute significantly to this model. Likely due to the homogeneity of this population, neither age nor racelethnicity affected perceived health status. Introducing mobility and physical performance indicators into the model erased the impact ofiADL : 1 dependencies, many of which depend on physical abilities. Additionally, contradicting results of most previous studies, living with one's spouse had a negative effect. Hays, Schoenfeld, and Blazer (1996), who found being married inversely related to perceived health, tried to find an explanation and suggested that "the presence of a spouse could increase a respondent's awareness ofhealth problems" (p. 192). Interviewees in this study spoke of the strains of caring for ill spouses and feelings of dependence by those receiving such care. Those burdens may better explain the unexpected negative impact of living with a spouse. Interviewees contributed insights into the meaning of health and factors that affect it. To them, ratings of perceived health measured the extent to which they had something meaningful to do and were able to do it. They identified these categories of factors necessary for healthy aging: the determination of something worthwhile to do, abilities, resources, and personal attributes The Model A definition, and then a model, of healthy aging-aging in a healthy manneremerged from the research described in the previous chapters. For this group of older people healthy aging meant going and doing-older people who do something meaningful feel healthier than those who do not. As one of the interviewees said, it means ''to get up and do as much as you want to when you want to." In a strictly 129

PAGE 140

: I I biomedical positivist framework, this definition sounds backward. It says that being able to get around and do things equates with as opposed to the biomedical view that being able to go and do results from health defined as the absence of disease and functional deficits. For these older people, however, going meant more than mobility and doing more than physical function. They spoke of independence, resolving problems, and contributing advice as well as of more physically active pursuits. Going and doing involves not only having the abilities and resources to meet challenges but also having a reason to want to do so. As Mechanic (1978) and Fries (1980) explained and as Moore et al.'s (1980) ecological model described, health is an adaptive process that depends as much on personal accomplishments and the changing meanings people give to life as it does on "support and care and feeding and empathy" (Fries, p. 135). The model suggested by the interviewees' responses supplemented with the regression results comprises a network of elements necessary for healthy aging and detrimental if absent. Figme 5.1 illustrates the model. The illustration may mistakenly suggest that its elements can always be clearly distinguished from each other, but its intent is to emphasize the following components and their interrelationships: having something worthwhile and desirable to do; possessing the required abilities; obtaining the necessary resources; and having the will to go and do 130

PAGE 141

; I I : I I I I : I I I HEAL THY AGING: A MODEL SOMETffiNG ,:J!ODQ. adaptation .... ..... .. s,;pplement supplement Figure 5.1. A Model ofHealthy Aging Components of the Model ATTITUDE support SOCIAL RESOURCES Something Worthwhile and Desirable to Do. To contribute positively to health, the things that people do need to include activities that matter to them. The respondents confirmed what Dubos (1959) realized many years ago. He observed that the goals people set for themselves-and the manner in which the individuals respond and adapt to them-have as much to do with health and happiness as disease and other challenges from the external world. Interviewees identified a variety of activities that they valued. The quantity of valued activities that a person named did not necessarily associate with a relatively : j more positive assessment of health status (that is, better reported status than the regression model predicted), but the two people who specifically spoke of dull or boring lives substantially under-rated their health status. To some degree, the value interviewees placed on activities related to their social roles-parent, friend, retiree, homemaker, person with disabilities. They made 131

PAGE 142

i I I 'I occasional references to normalcy, comparisons that suggested an attempt to define what they ought to be doing. Both internal and external expectations played roles. One interviewee spoke of the importance ofhaving something she was expected to do. Another commented that she'd "be a basket case" if she didn't have plans for meaningful work each day. To be beneficial and not negatively frustrating, expectations needed to be realistic. Interviewees with more positive relative assessments of health status bad substituted new activities for ones they could no longer do. Depression may affect the ability to identify a meaningful activity. One interviewee compared an earlier depressed period with her current happy and positive status. lllness both contributed to her depression and decreased her abilities, but she also observed that the depression made it difficult to think that any activity was worth doing even if within her abilities. Abilities. The extent to which people have sufficient abilities affects how satisfactorily they can accomplish valued activities. Interviewees spoke of mobility, vision, and mental function as particularly important. They also emphasized the abilities required to maintain independence-driving or otherwise getting around and housekeeping, for example. They felt that both chronic illness and getting older itself presented the greatest challenges to their abilities. Most of the factors that remained significant in the regression model concerned abilities, and they are the ones most amenable to medical intervention. This does not mean, however, that abilities are exclusively medical issues. Interviewees with the poorest relative health status ratings often spoke of pain, which generally can be medically controlled. In these cases, however, expressions of pain may also have indicated other, perhaps more attitudinal challenges to ability. Other challenges mentioned by interviewees included dependence on others and caregiving 132

PAGE 143

obligations that interfered, currently or in the past, with abilities to participate in valued activities. Resources. External resources support going and doing. The study population reported using almost none of the formal support services listed in the HSF instrument, such as home-delivered meals or home health care. They dicL however, I : I identify a variety of informal support services from family and friends. They felt they had and needed both instrumental assistance-transportation and help with chores, for example-and emotional support in the form of :friendship, empathy, and understanding. Religious faith was a substantial source of support for some. Almost all of those who spoke about the resources necessary to remain independent referred to cooperative marriage, with a sharing of responsibilities and abilities. Access to social support seemed to be only part of its effectiveness. Among respondents, too little support bred loneliness and lower relative perceived health, while too much caused less harm but generated bad feelings and challenged self esteem. As with Goldilocks, the amount needed to be just right. People spoke of health care as a resource. Some referred to health information from family and friends, but most discussion, in response to specific interview questions, concerned their relationships with health care providers. Respondents valued a caring manner, good listening skills, and respect. Just having health care did not much affect relative perceptions of health; it was the nature of the relationship that mattered. Those who most felt that they were active partners with their providers rated their health status relatively better than those who merely complied with medical advice. Attitude. Personal characteristics affect people's choices and actions Either by direct attribution or through their own stories, interviewees identified the following characteristics as important: attitude, one's sense of self and awareness of self, an internal locus of control, and a focus on others. When other factors had been 133

PAGE 144

. I accounted for-by the regression model and then from talk about specific other possible contributions to health-these were the characteristics that most discriminated between optimistic and pessimistic evaluations of health status related to aging. Trying to identify the sources of these attributes, respondents mentioned childhood experiences, learned problem-solving, and personality type. They spoke of upbringing and family experiences as important contributors to a person's responses to life's later challenges. Many suggested that personality type-an optimistic or pessimistic orientation-underlay the difference between positive and poor me attitudes. Whatever the sources, people who expressed positive attitudes, assertiveness, the determination to continue to be active, and the desire to take charge perceived their health status relatively more favorably than those who did not. Interactive Processes Like sequential models, such as Wilson and Cleary's ( 1995) taxonomy of outcomes, this model contains biological, functional, social, and psychological factors--attributes of the individual and the environment. This model, in contrast to sequential ones, views health not as a static condition but as an ongoing interactive rearranging and balancing of its components to achieve the goal of meaningful activity. It has more in common with Moore et al.'s (1980) ecological equilibrium but incorporates people's values and personality attributes in addition to the biological and cultural adaptive and coping resources discussed by Moore et al. The four components of the model-something worthwhile and desirable to do, the ability to accomplish the activity, the resources to support the activity, and sufficient will or positive attitude-all directly contribute to the desired outcome, healthy aging. They also interact, supporting and supplementing each other and contributing to or benefiting from adaptation to change and challenge. 134

PAGE 145

Ability denotes the capacity to act but does not guarantee performance. Without a goal (something meaningful to do) or the will or desire to act, ability remains only a necessary but not a sufficient condition for health. In healthy older people, at least in this population, an adaptive feedback interaction exists between ability and something to do. When an ability declined, these people adapted their valued activities or substituted an alternative. Reading replaced needlework when arthritis interfered, for example, as walking did bicycling when eye problems occurred. Conversely, having something to do occasionally affected ability. One woman dedicated hours to working puzzles to regain ability lost with a stroke. Social resources, formal and informal, have direct impacts on going and doing but perhaps even more indirect impact through the other components of the model. Social support, in the form of transportation or help with housework, for example, may supplement decreased ability sufficiently for people to maintain independence and continue with valued activities. Cultural expectations (Boult et al., 1994 ), which derive from the social environment, affect attitude and can, in mainstream American culture, complicate efforts to define a meaningful activity. Family and community resources, on the other hand, can support those efforts by offering emotional support toward feeling worthwhile and instrumental assistance in locating and participating in desired activities. Attitude affects the ways in which a person does or does not choose a worthwhile activity, use abilities, and take advantage of support resources. In tum, having something to do, feeling competent, and benefiting :from available supports can bolster attitude, supplementing and supporting it. In summary, the components of the model directly support the going and doing that is healthy aging and also affect each other, supporting, supplementing and interacting in adaptive feedback relationships. 135

PAGE 146

Evaluating the Model A well-constructed theory, according to Strauss and Corbin (1990), meets four criteria for judging its applicability to a phenomenon: It fits with the data. It is comprehensible and sensible. It has a comprehensive enough base and a conceptually broad enough interpretation to allow abstraction to more general populations or situations. It offers possibilities for action concerning the phenomenon. Similarly, as stated in Chapter 2, a useful model of health should incorporate factors : j and interactions that are specific and conceptually clear, are related to individuals and their well-being, can be measured and assessed, and are amenable to intervention. Although one can never remove a researcher's influence--in qualitative or quantitative research-the model proposed here emerged from the data, not from any preconceived template. It is a conceptually simple model, built of comprehensible and easily understood components that contribute directly to the desired outcome, healthy aging: older people with something meaningful to do and a combination of abilities, support, and the will to do it feel healthier than those who do not. The outcome is an important one for individuals as well as the larger society, and the relevant factors relate directly to their well-being. The study population was a single fairly homogeneous group of community dwelling older people. Details specific to these individuals-sadness about not going fishing, for example, or caregiving obligations-may not translate to other populations and settings, but the concepts seem comprehensive enough to apply beyond just these study subjects. Different populations-older people with different degrees of frailty or in different age groups--will value specific activities differently, but the importance of having something to do and doing it should remain. 136

PAGE 147

The final criterion for judging the relevance of this model concerns its practicality, that is, the degree to which its components can be assessed and acted and the usefulness of the components for improving or maintaining health in an older population. Some of the factors in the model have been measmed with validated indicators, some have not. but most certainly could be. Many scales exist for physical and functional abilities, physical condition. social support. and personality The determination of what constitutes meaningful and desirable activity can only be made by the individual. Simply asking people to compare the activities that they think valuable with those they actually do may be the best measure of their levels of satisfaction. Factors that comprise the model's four components are amenable to intervention. Medicine offers some supports. Even if not curable, depression. pain, j chronic illness, and disease processes respond to medical treatment. Mechanical aids 1 permit greater mobility for those with physical disabilities. I The community has responsibility for other resources. Assisted living arrangements and readily available public for example, supplement reduced abilities and permit continued independence. Senior centers can be a source of valued activities, and adult day care centers provide support both for caregivers and those who need care. Services such as Lifeline provide a check on status that makes it possible for older people to live alone safely. The issue of social role derives from society's perceptions of older people. Designing effective interventions here presents a more difficult task. Perhaps the aging of the population. with the sheer numbers of the Baby Boom generation. will hasten the political process of education about the contributions older people can and do make. Community-level programs exist that help both to define a positive role and to create opportunities for worthwhile activity. Examples include coordinated 137

PAGE 148

I I volunteer activities and foster grandparent programs. As several of the interviewees illustrated, church-based service and activities offer other such opportunities. Personal characteristics may be the least amenable to intervention. Self esteem, locus of control, and personality evolve over a lifetime. On the other hand, Lawton's (1983) research showed that "positive affect is directly associated with social interaction and other types of active time use that may be in short supply for older people for reasons other than of personal preference" (p. 71 ). He continued, "low income, poor health, a lack of accessible age-relevant opportunities, a shrinking of the available social network, and environment barriers such as poor transportation and neighborhood crime may all conspire to limiting such time uses" (p. 71). These negative influences can be "directly changed by attention to such matters as activity programming, [efforts that] will be even more successful when the person is the origin of the activity, which means providing her with the wherewithal to create or choose her own activities" (p. 71). Intervening against negative affect (as opposed to reinforcing positive affect), Lawton suggested, requires focus on "physical health and on long-term benefits to the self, such as counseling, opportunities for, and practice in origin behavior'' (p. 71 ). Possibilities include treatment for depression, monitoring of medications, and attention to nutrition. Showing people that there are worthwhile contributions they can make may help them adopt a more external focus and increase the will to get involved. Conclusion The components of this model are not new. As reviewed in Chapter 2, representatives of various disciplines--medical sociology, psychology, biomedicine, and medical anthropology, to name a few-have talked about health and aging in terms of physical and mental conditions and abilities, psychological attributes, social support, or valued activities. What this model contributes is a new perspective on 138

PAGE 149

: I t I I I I i healthy aging as a multidisciplinary process to achieve the desired outcome, going and doing something meaningful. That process entails balancing internal and external, individual and environmental, challenges and resources that include determining something worthwhile to do, maintaining the abilities to do it, receiving appropriate social resources to support it, and generating the will to continue. To be considered a useful tool, the model must stimulate the development of interventions, unconstrained by the boundaries of specific disciplines, to help older people identify and sustain valued activities. By reframing healthy aging in older people's own terms and emphasizing their perceptions of health, this model encourages the support of desired goals and outcomes rather than only attacks on deficits and challenges. The preceding section suggested some examples of interventions that are not limited to traditional gerontological-medical tools, but these suggestions are only first, small steps. Effective support of healthy aging will require integrated, multidisciplinary efforts to design and implement interventions that act on all of the components of the model. 139

PAGE 150

Instruments APPENDIX A RESEARCH INSTRUMENTS A. I Health Questionnaire (Kaiser Permanente Colorado Region) A.2 Semi-Structured Interview Questions 140

PAGE 151

:I I I I Confidential HSF Kaiser Permanente Colorado Region Health Questionnaire Directions: Complete this form by fdling in the appropriate information in the spaces of boxes provided. Unless otherwise indicated, select the one answer that fits you best and marie an (x) in the box Kaiser Permanente ID II Regular Physician:-------Date of Birth 1. Name: Last First Middle Initial 2. Address: Number Street Apartment# City frown State Zip Code 3. Area Code Home Area Code Office 4. Sex: male 0 female 0 S. In case of emergency, who may we contact? Name Telephone 141 Relationship Day Evening

PAGE 152

! I :I 6. Comp:m:d to other persons your age:. would you say your hc-.dth is: (Check one bmt.) C Excellent c Very Good c Good c Fair c Poor 7. Do you now have any of the following health conditions? (Check yes or no for condition.) Cl Cl a. Congestive bean failure Cl Cl b. Chronic lung disease (including bronchitis or emphysema) Cl Cl c. Blindness or trouble seeing even when wearing glasses Cl [J d. Deafness or trouble hearing Cl [J e. Sugar diabetes (diabetes mellitus) [J Cl f. Asthma [J [J g. Ulcer or gastrointestinal bleeding (not counting hemorrhoids) [J [J h. Arthritis or rheumatism 8. Has a doctor told you that you had any of the following conditions? (Check yes or no for gdl condition.) [J [J a. Hypertension or high blood pressure [J [J b. Angina [J 0 c. Hean attack or myocardial infarction [J [J d. Stroke [J 0 e. Kidney disease [J 0 f. Cancer (not including skin cancer) 9. Are any of these conditions getting worse? C YES C NO If yes, which ones?------------------------10. During the past few months, have you had increasing problems with: (Check all that apply.) C a. Forgetting to keep an important appointment? C b. Getting lost traveling from one, usually familiar location to another? C c. Capability in paying bills? C d. Seeing things that are not there? c e. Forgetting the date or the names of familiar friends? 142

PAGE 153

, I I I I I I 11. How many different medications do you take each day? __ # (Include prescription and non prescription drugs). 12. Do you currently smoke cigarettes? 0 YES o NO 13. Which of the following activities are you still healthy enough to do by yourself; that is, without the help of another person or special equipment, such as a cane. {Check all that apply.} o a. Can you walk up and down stairs without help? 0 b. Can you go out to a movie, meeting, to church or synagogue or to visit friends without help? o c. Can you do heavy worlc around the bouse, lilce shoveling snow or washing walls, without help? 0 d. Can you walk half a mile {about eight ordinary city blocks) without help? 14. Which of the following statements fits you best in terms of health? {Read all statements before answering -check only one box.) 0 a. Must stay in bed all or most of the time 0 b. Must stay in the house all or most of the time 0 c. Need the help of another person in getting around inside or outside the bouse 0 d. Need the help of some special aid, such as a cane or wheelchair, in getting around inside or outside the bouse o e. Do not need the belp of another person or a special aid, but have trouble getting around freely 0 f. Not limited in any of these ways 15. Because of a disability or health problem, do you need or receive help from another person for any of the following activities? (Check "never" if no help is needed.) I receive or need hl!lp of another person: MOST OF SOME OF NEVER THE TIME THE TIME a. Using the toilet in bathroom 0 0 0 b. Bathing. including sponge baths 0 0 0 c. Dressing 0 0 0 d. Eating 0 0 0 e. Getting in/out of bed or chairs 0 0 0 16. Do you often feel sad or depressed? 0 YES 0 NO 143

PAGE 154

17. During the past 6 months, bave you lost your urine or been wet on at least three days? o YES 0 NO 18. During the past 6 months, bave you bad accidental loss of your bowels? D YES D NO 19. Because of a disability or health problem do you receive or need belp from another person for the following activities? (Check all that apply.) Ca. Preparing meals o b. Shopping for groceries, etc. 0 c. Doing routine household chores D d. Managing money D e. Doing laundry D f. Taking medication 0 g. Getting to places, out of walking distance o h. Using the telephone 20. If you receive help with any of the above activities, wbo are your helpers? (Give the name and relationship, e.g. daughter, spouse, etc.) Name Relationship Phone Number Name Relationship Phone Number 21. Are you currently receiving any of the following services from an agency? (Check all that apply.) o a. VISiting nurse o b. Therapist (physical, occupational, speech) 0 c. Homemaker/home health aide 0 d. Social Worker o e. Adult day health center 0 f. Assistance with transpOrtation 0 g. Home delivered meals 22. Do you spend more than four months a year out of the state or traveling? D YES D NO 23. Are you planning to move out of the metro area, or the state? D YES D NO 24. Have you terminated or are you considering terminating your Kaiser Permanence membership? 0 YES D NO 144

PAGE 155

25. Which of the following Statements best describe your transportation requirements? (Check just one box.) 0 a. I drive my own car. 0 b. I have a spouse, friend, or family member who drives me anytime I need. 0 c. I have a spouse, friend, or family member who drives me when they can. 0 d. I take a taxi. o e. I take the bus. 0 f. I take the Amb-O-Cab. 0 g. I am home-bound or live in a nursing home and don't go ouL 26. During the last two months, have you changed your primary care physician? o YES o NO 27. Are you considering changing your primary care physician? o YES o NO 28. What is your current living arrangement? (Check each box that applies.) 0 a. Live alone o b. With spouse 0 c. With child(ren) 0 d. With other relative(s) 0 e. With non relative(s) o f. With pet(s) 29. What type of housing do you live in? (Check one box.) 0 a. I live in my own residence (house, condominium, apartment, mobile home) 0 b. I live in the residence of a friend or relative o c. I live in a group home, foster care or adult home o d. I live in assisted living arrangement 0 e. I live in a nursing home 30. What is your current marital swus? 0 a. Married o b. Widowed 0 c. Divorced 31. What is your race or ethnic group? 0 a. Asian/Pacific Islander o b. Black/African-American o c. Hispanic 145 o d. Separated 0 e. Never Married 0 d. Native American 0 e. White/Caucasian 0 f. Other

PAGE 156

32. How many years of regular school have you completed? o a. Grades 0-8 o b. Grades 9-11 0 c. High school graduate o d. Some college 0 e. College graduate 0 f. Post college worlc 33. Do you now work (including volunteer work): o e. Homemaker 0 a. Full time 0 b. Part time o c. Retired 0 f. Temporary medical leave 0 g. Permanently disabled o d. Unemployed 34. Following is a question regarding annual income. Please be aware that responding to this question is in no way associated with Kaiser Permanence benefits, services, rates, etc. You may choose to answer this question and it is strictly voluntary on your pan. Below is a list if annual income groups. Which group comes the closest to your total household income (defined as income before taxes). This includes income of each person in the household, including Social Security pensions, rent from property, dividends, interest, earned income, help from relatives and any other income. (Check one box..) Less than $3,000 per year 0 $3,000 $5,999 0 $6,000-$9,999 0 $10,000 $14,999 0 $15,000 $24,999 S25,000 $34,999 $35,000 and more 0 0 0 35. Have you completed any of the following? (Check all boxes that apply.) 0 Advance Directives 0 Living Will (Colorado Declaration as to Medical or Surgical Treatment) 0 Durable Power of Attorney for Health Care If necessary a. Who may we contact? Name Responsible Party Phone Number 146

PAGE 157

:I i I 36 Is then: anything dsc you would liL;r: us tu know Jbuut you? 37 Did you receive help filling out lhis form? 0 YES 0 NO /fyu a Please give lhe name, relationship, and phone number of your helper (whom we may conract if we need additional information ) Name Relationship Phone Number Kaiser Permanente is considering starting regular group health education and patient care clinics for our adult members. The groups and their personal physician and nurse would meet oace a month for bealtb maintenance checks, health education, and discussion of both medical and non-medical issues. Would you be interested in attending these group clinics, sbould they be started? CYES CNO 0 INTERESTED, but aeed more information 147

PAGE 158

SEMI-STRUCTURED INTERVIEW QUESTIONS Health and Well-Being What words would you use to describe your health now? Why? I How would you define well-being? What or who contributes to your sense of well-being (or lack of it)? What or who detracts or takes away from your sense of well-being? When you were growing up, how did your family and/or community react to illness? In what ways does getting older affect health, yours and others'? How does your health compare to that of others your age? What, if anything, worries you about your health? How easily do you become ill or develop other health problems? Function What abilities do you v3lue most--or, from a more negative point of view, which ones would you/do you feel worst about not having? (e.g., ability to walk, see, read, ... ; think clearly, use mental abilities; not be depressed; control body functions; be free of pain) What activities do you value most--or, from a more negative point of view, which ones would you/do you feel worst about not having? (e.g., participate in particular activities; make choices) What relationships do you value most--or, from a more negative point of view, which ones would you/do you feel worst about not having? (e.g., have friends, maintain contacts with family and friends; love and be loved) In what way( s) do health troubles get in the way of your activities? 148

PAGE 159

Social Support Please describe the support or help (e.g., health information; tangible help; emotional support from family, friends, community, groups, church, ... ) that you: 1. have available from others 2. feel you need from others (whether available or not) 3. want from others 4. give to others :I 1 Control, Coherence. Optimism/Pessimism I i I Please tell me about your ability to predict what your life will bring. Describe your plans for the future. Tell me about how the health of others close to you affects your life and life plans. Please tell me about your sense of control over: 1. your life in general 2. your health 3. the things you need to and want to do now and in the future Some people who have a number of medical problems have a very positive outlook while some people with few or no medical problems have a negative outlook What are your thoughts about why that is? Additional prompts: Health Care What other things do you think contribute to how people balance their physical condition and their outlook? Why is that, what makes the difference? Please describe your relationship with your physician (and others who provide care). Conclusion : I Overall, how do you feel about your life now? What have I not asked that you think I should know? 149

PAGE 160

. I I I l I APPENDIXB LITERATURE REVIEW TABLES Table B.l Characteristics of Studies of Risks Associated with Aging B.2 Risks of Mortality 8.3 Risks oflnstitutionalization 8.4 Risks of Functional Decline 8.5 Characteristics of Studies ofPredictors ofPerceived Health B.6 Predictors of Perceived Health 150

PAGE 161

Table B. I. Characteristics ofStudies ofRisks Associated with Aging Type of study & method of Dependent SamE lea variable6 Abramson, Gofin, & Israel Longitudinal ( 5 years) Mortality Peritz ( 1982) N= 387 (all male; 75 Retrospective died) Discriminant function Age60+ analysis Arling & McAuley N= 2,259 (219 impaired) Cross-sectional NH admission (1983) Age60+ Descriptive Bauer ( 1996) N=2,923 Longitudinal (3 years) NH admission Meanage68 Retrospective Proportional hazard analysis Beckett et al. ( 1996) EPESE ( 4 sites) Longitudinal ( 5 years) Change in N= 17,646 Retrospective physical Age 65+ Markov & random effects function models, repeated measures Berkman & Syme Alameda Longitudinal (9 years) Mortality (1979) N=6,928 Retrospective Age 30-69 x2 age-adjusted relative rates Berkman et al. MacArthur, EPESE Cross-sectional Function (1993) N= 1,354 (1,192 high, 80 x2 comparison of means middle, 82low function) Age 70-79 Berkman, Leo-N= 194 Longitudinal ( 6 months) Mortality after Summers, & Age 65+ Prospective cohort hospitalization Horwitz ( 1992) Logistic regression Blazer (1982) N=331 Longitudinal (30 months) Mortality :I Age65+ Retrospective Binary linear regression 151

PAGE 162

Table B. I. (Cont.) Type of study & method of Dependent SamE lea variableb Boaz & Muller Natl. LTC Survey Longitudinal (2 years) Permanent vs. (1994) N = 4,832 (292 L T; 227 Retrospective transitory NH ST; 945 died. no NH; Multinomial logistic stay alive, no NH) regression Age65+ Borawski, Kinney, N = 885 (79 died) Longitudinal (3 years) Mortality & Kahana ( 1996) Age 73+, mean 80 Retrospective Logistic regression Kane, Louis, LSOA Cross-sectional Mortality & McCaffrey N = 5,210 Dichotomous & Function (1994) Age70+ trichotomous logistic regression Branch & Jette N = 825 (8% NH) Longitudinal ( 6 years) NH admission (1982) Age 65+, mean >75 Retrospective Logistic regression Branch & Ku ( 1989) MA Health Care Panel Longitudinal(l25,6, 10 Mortality Study years) NH admission N= 1,625 Retrospective Function Age 65+ Transition probabilities (ANOVA) Brody, Poulshock, & N= 186 (140 SNF, 46 Cross-sectional NO admission Masciocchi (1978) home health) Comparisons Age means 78 & 73 Campbell, Diep, New Zealand Longitudinal (3-5 years) Mortality Reinken, & McCosh N= 535 (227 died) Retrospective (1985) Age65+ Cox regression Chipperfield ( 1993) Aging in Manitoba Study Longitudinal (12 years) Mortality N=4,303 Retrospective Age 65+ Logistic regression 152

PAGE 163

Table B. I. (Cont.) Type of srudy & method of Dependent srudx SamE lea anaixsis variable6 Cohen. C.A .. et al. N = 196 caregiver/careLongitudinal ( 18 months) NH admission (1993) receiver dyads Retrospective Age=64n7 MANOVA M.A., Tell, N=4,400 Longitudinal (1 year) NH admission & Wallack ( 1986) Age mean 74 Retrospective Logistic regression Coughlin. McBride, Natl LTC Channeling Longitudinal (1 year) NH admission & Liu (1990) N=3,940 Retrospective Age 65+, mean 80 Multinomial logit Crimmins & Sito LSOA Longitudinal (2 years) Change in (1993) N= 3,169 Retrospective function Age70+ Logistic regression Davis, R.B., et al. N= 2,169 Longitudinal ( 5 years) Mortality in (1995) Age adult Retrospective hospital Logistic regression Deeg, van Netherlands Longitudinal (28 years) Mortality Zonneveld, van der N = 2,645 (2,617 died) Retrospective I Maas, & Habbema Age 65-99 Logistic regression I (1989) ; I Dolinsky& U.S. 1980 Census Cross-sectional NH admission : I Rosenwaike (1988) N=495,199 Logistic regression Age 75+ : I Ford, Roy, Haug, N= 1,221 (632 died) Longitudinal(9years) NH admission Folmar, & Jones Age 65+ Retrospective (1991) Logistic regression Freedman, Berkman, EPESE Longitudinal (3 years) NH admission Rappo, & Ostfeld N= 2,812 (354 NH) Retrospective (1994) Age 65+ Logistic regression 153

PAGE 164

Table B.L (Cont.) Type of study & method of Dependent lea variableb Glazebrook. Canadian Study ofHealth Cross-sectional NH admission Rockwood, Stolee, and Aging Logistic regression Fisk. & Gray (1994) N= 641 (108 NH. 533 community controls) Age65+ Grosclaude, Rural France Longitudinal ( 4 years) Mortality Bocquet, Pous, & N= 645 (Ill died) Retrospective Albarede (1990) Age60+ Cox regression Greenberg & Ginn N= 266 (139 NH. 127 Cross-sectional NH admission (1979) home health) Retrospective or home health Age60+ Linear regression care ; I Greene & Ondrich Natl LTC Channeling Longitudinal ( 12 months) NH admission (1990) N= 3,332 Retrospective &exit Age 65+, mean 80 Discrete-time hazard function Greiner, Snowdon, Nun Study Longitudinal (2 years) Mortality & Greiner (1996) N= 629 (all female) Retrospective Function Age 75+ Linear & logistic regression Guralnik & Kaplan Alameda Longitudinal ( 19 years) Function (1989) N = 841 (345 died) Retrospective Age 65-89 (survivors) Dichotomous & I trichotomous logistic regression Hanley, Alecxih, Natl LTC Survey Longitudinal (2 years) NH admission Wiener, & Kennell N = sample of disabled, Retrospective (1990) population N = 4 8 million Logistic regression Age 65+ Hanson, Isacsson, Sweden Longitudinal (5 years) Mortality Janzon, & Lindell N= 500 (all male) Retrospective (1989) Age 68+ Proportional hazard analysis 154

PAGE 165

Table B. I. (Cont.) Type of study & method of Dependent Study Sample a analysis variablei1 Harris, Kovar, LSOA Longitudinal (2 years) Function Snzman, Kleinman, N= 1,791 (338 died) Retrospective (physical & Feldman ( 1989) Age80+ Logistic regression ability) Hirdes & Forbes Ontario LSOA Mortality (1992) N = I, 702 (all male: good Retrospective or excellent perceived Logistic regression health) Age45 I Ho (1991) China Longitudinal (2 years) Mortality N= 1,054 (89 died) Retrospective Age 70+, mean 77 Logistic regression : I Hodkinson & ExtonUnited Kingdom Longitudinal ( 5 years) Mortality Smith (1976) N = 852 (33% died) Retrospective Age65+ Linear regression Hoeymans, Feskens, N= 721 (269-340 at Longitudinal ( 5 years) Function van den Bos, & follow-up) Retrospective Kromhout ( 1997) Age70+ Bivariate proportions House et al. (1994) N= 3,617 (2,867 Longitudinal (2-3 years) Function survivors) Retrospective Age25+ Logistic regression House, Robbins, & N= 2,754 Longitudinal (9-12 years) Mortality Metzner (1982) Age 35-69 Retrospective Logistic regression Idler & Angel NHANES-1 Longitudinal ( 12 years) Mortality (1990) N=6,440 Retrospective Age25-74 Proportional hazard analysis 155

PAGE 166

Table B. I. (Cont.) Type of study & method of Dependent Study Sam}! lea variableb Idler & Kasl (1991) EPESE (Yale) Longitudinal ( 4 years) Mortality N=2,812 Retrospective Age 65+, mean 75 Logistic regression, proportional hazard analysis Idler & Kasl ( 1995) EPESE (Yale) Longitudinal ( 1-6 years) Function N= 2,812 (1,424-1,455 at Retrospective I 6 years) Linear & logistic Age 65-99, mean 75 regression Idler, Kasl, & EPESE Longitudinal ( 4 years) Mortality Lemke (1990) N= 6,485 (2,812 CT, Retrospective 3,673 IA) Logistic regression Age65+ Incalzi et al. (1992) N = 178 (34 died) Longitudinal (l year) Mortality Age 70-95, mean 76 Retrospective Logistic regression Jagger & Clarke England Longitudinal ( 5 years) Mortality (1988) N= 1,124 community Retrospective (388 died) & 79 Cox proportional hazard institution ( 63 died) analysis Age75+ Kane, Matthias, & N=23,557 Cross-sectional NH admission Sampson ( 1983) Age65+ Comparisons Kaplan, GA., et al. Sweden Longitudinal (71 months) Mortality (1994) N = 2,682 ( 167 died) Retrospective Age42-60 Proportional hazard analysis i l Kaplan, GA., Alameda Longitudinal ( 17 years) Mortality Seeman, Cohen, N=4,174 (1,219 died) Retrospective Knudsen, & Age38+ Proportional hazard Guralnik (1987) analysis 156

PAGE 167

Table B I. (Cont.) I Type of study & method of Dependent studx SamE lea anaixsis variableb Kaplan, G.A., Alameda Longitudinal ( 6 years) Change in Strawbridge, N=356 Retrospective function Camacho, & Cohen Age65+ Linear regression (1993) Keil et al. (1989) Charleston Heart Study Longitudinal (25 years) Function N= 1,022 Retrospective (physical Age 60+ at 25 years, mean Linear regression disability) : I 64-70 I Kraus et al. (1976) Canada Cross-sectional NH admission I N= 193 NH & 141 Bivariate comparisons community Age 65+, mean 80 & 74 Lammi et al (1989) Finland Longitudinal (15-25 years) Function N= 716 (all male) Retrospective (physical Age65-84 Linear regression disability) LaRue, Bank, N = 69 surviving twins Longitudinal (>5 years) Mortality Jarvik, & Hetland Age median 84 Retrospective (1979) Descriptions Liu, Coughlin, & Natl LTC Survey Longitudinal (2 years) NH admission McBride ( 1991) N= 5,795 Retrospective Age65+ Hazard analysis Manton (1988) Natl LTC Survey Longitudinal (2 years) Mortality N=26,911 Retrospective NH admission Age65+ Transition probabilities Markides & Lee N=254 Longitudinal (8 years) Function (1990) Age68+ Retrospective Linear regression 157

PAGE 168

Table B. I. (Cont.) Type of study & method of Dependent Study Sam2le0 anaix:sis variableb Markides & Pappas N = 338 (70% MexicanLongitudinal ( 4 years) Mortality (1982) American; 59 known Retrospective (survival) died) Discriminant analysis Age60+ McAuley& N= 1,160 acute care & Cross-sectional NH admission Prohaska ( 1981) 645 community Discriminant analysis I Age60+ McCoy & Edwards 1973 Survey of LowLongitudinal (1 year) NH admission (1981) Income Aged & Disabled Retrospective N = 4,884 welfare Logistic regression recipients Age65+ McFall & Miller Natl LTC Survey Longitudinal (2 years) NH admission (1992) N= 940 (127 NH) Retrospective Age65+ Logistic regression I Mor, Murphy, et al. LSOA Longitudinal (2 years) Function (1989) N= 1,745 (852 Retrospective functionally intact) Logistic regression Age 70-74 I Mor, Wilcox, et al. LSOA Longitudinal ( 6 years) Change in (1994) N=7,521 Retrospective function Age 70-99 Transition table; logistic I regression ., Mossey & Shapiro Manitoba Longitudinal ( 6 years) Mortality (1982) N= 3,128 (877 died) Retrospective Age65+ Logistic & log-linear regression Murtaugh, Kemper, Natl LTC Survey Longitudinal (2 years) NH admission & Spillman (1990) N=2,710 Retrospective Age65+ Comparisons 158

PAGE 169

Table 8.1. (Cont.) Type of study & method of Dependent Same lea variableb Narain etal (1988) Veterans Administration Longitudinal ( 6 months) Mortality N = 396 hospitalized Retrospective NH admission (290/o died) Linear & logistic after Age 70+, median 74 regression hospitalization Newman, Natl LTC Survey Longitudinal (2 years) NH admission Wright, & Rice N=5,580 Retrospective (1990) Age65+ Logistic regression Newsom & Schulz Cardiovascular Health Cross-sectional Function (1996) Study Structural equation model N=4,134 Age 65+, mean 73 Nocks, Leamer, N = 284 intervention Longitudinal ( 18 months) NH admission I Blackman, & Brown (community care) & 340 Prospective (1986) control Logistic regression Age 18+, mean 74 Orth-Gomer & Sweden Longitudinal ( 6 years) Mortality Johnson ( 1987) N= 17,433 (841 died) Retrospective Age29-74 Logistic regression Palmore (1976) LSOA Longitudinal (20 years) NH admission N= 207 who died <1977 Retrospective Age66%>59 Linear regression Palmore, Nowlin, & N = 297 Longitudinal ( 1 0 years) Function Wang(1985) Age 72+ at follow-up, Retrospective mean 81 Linear regression Parker, Thorslund, Sweden Longitudinal ( 4 years) Mortality & Nordstrom (1992) N=421 Retrospective Age 75+ Logistic regression Parkerson et al. N=249 Cross-sectional Function (1989) Age 18-49 Linear regression 159

PAGE 170

Table B.l. (Cont.) Type of study & method of Dependent Study lea analysis variableb Pearlman & Crown Natl LTC Survey Longitudinal (2 years) NH admission (1992) N= 5,273 Retrospective Age65+ Logistic regression pfeiffer (1970) N= 37long-lived matched Longitudinal (10 years) Mortality with 37 shorter-lived Retrospective (longevity) Age mean 66-68 Linear regression Pijls, Feskens, & Netherlands Longitudinal(25years) Mortality Kromhout (1993) N= 783 (all male; 48% Retrospective "healthy," 12% least Cox proportional hazard healthy; 181 died) analysis Age40-59 I Pinsky et al. ( 1985) Framingham Longitudinal (27 years) Function N=2,021 Retrospective :I Age 56-88 at follow-up Logistic regression Pinsky, Leaverton, Framingham Longitudinal (21 years) Function & Stokes ( 1987) N= 1,474 Retrospective j Age 35-68 Logistic regression I I Rakowski, Mor, & LSOA Longitudinal (2 years) Mortality Hiris (1991) N= 1,252 Retrospective Age70+ Logistic regression Reuben, Siu, & N= 149 Longitudinal (2 years) Mortality Kimpau (1992) Age 64-94 Retrospective NH admission Logistic regression Rockwood, Stolee, Canadian Study of Health Cross-sectional NH admission & McDowell ( 1996) & Aging Logistic regression (proxy for N= 1,258 NH & 9,113 frailty) community Age65+ Roos & Havens Manitoba; LSOA Longitudinal ( 12 years) Function (1991) N=2,943 Retrospective (successful Age 77-96 Logistic regression aging) 160

PAGE 171

Table B. I. (Cont.) Type of study & method of Dependent Study Sam2Iea is variableb Sager et al. ( 1996) N=827 Longitudinal (3 months) Change in Age70+ Retrospective function after Logistic regression hospitalization Schoenbach. N=2,059 Longitudinal ( 13 years) Mortality Kaplan, & Age 22-87 Retrospective Kleinbaum (1986) Hazard analysis Schoenfeld, MacArthur; EPESE Longitudinal (3 years) Mortality Malmrose, Blazer, N= 1,192 Retrospective Gold, & Seeman Age 70-79 Logistic regression (1994) Seeman et al. ( 1994) MacArthur Longitudinal (3 years) Change in N= 1,015 Retrospective function Age 70-79, mean 74 Logistic regression (physical performance) Seeman et al. ( 1995) MacArthur Longitudinal (2.5 years) Change in N= 1,189 Retrospective function Age 70-79 Linear & logistic (physical ; I regression performance) I Seeman, Kaplan, Alameda Longitudinal ( 17 years) Mortality I Knudsen. Cohen, & N= 4,174 (1.219 died) Retrospective I I Guralnik (1987) Age 38-94 Cox proportional hazard analysis Shapiro & Tate Manitoba LSOA Longitudinal (2 & 7 years) NH admission (1985) N = 3,383 (395 NH) Retrospective (short-& long-Age65+ Logistic regression term) Smyer ( 1980) N = 66 matched NH vs. Cross-sectional NH admission community Discriminant function Age 62+, mean 75-76 Sorensen (1988) Denmark Longitudinal (3 & 5 years) Mortality N=585 Retrospective Age 75, 80, 85 Logistic regression 161

PAGE 172

Table B. I. (Cont.) Type of study & method of Dependent Same lea anal:x_sis variableb Speare, Avery, & LSOA Longitudinal (2 years) NH admission Lawton (1991) N= 5,151 Retrospective Age70+ Logistic regression Steinbach (1992) LSOA Longitudinal (2 years) Mortality N= 5,151 (??died) Retrospective NH admission Age70+ Logistic regression Stewart et al. (1989) MOS Cross-sectional Function N=9,385 Linear regression Age 18+, mean 46 Strawbridge, Cohen, Alameda Longitudinal ( 6 years) Function Shema, & Kaplan N=356 Retrospective (successful (1996) Age65+ Logistic regression aging) I Sugisawa, Liang. & Japan Longitudinal (3 years) Mortality Liu (1994) N=2,200 Retrospective I Age60+ Cox hazard analysis : I Teresi et al. (1989) Israel Cross-sectional; NH admission I N = 269 ( 113 informal Linear regression caregivers, 156 formal) I Age65+ I I Tsuji, Whalen, & N= 334 (84 NIL 143 Longitudinal NH admission : I Finucane ( 1995) died) Retrospective I Age 24-105, mean 76 Cox proportional hazard analysis Verbrugge, Reoma, N= 165 Longitudinal (1-2 years) Function after & Gruber-Baldini Age 55+ Retrospective hospitalization (1994) Linear regression Vicente, Wiley, & Alameda Longitudinal NH admission Carrington ( 1979) N= 455 who died Retrospective from death Age65+ backto 1965 Discriminant analysis 162

PAGE 173

, I Table B.l. (Coot.) Type of study & method of Dependent studx analxsis variableb Vogt. Mullooly, Kaiser Northwest Longitudinal ( 15 years) Mortality Ernst, Pope, & N= 2,573 (502 died) Retrospective Hollis ( 1992) Age 18+ Cox hazard analysis Wachtel, Derby, & N = 50 discharged to NH. Longitudinal NH admission Fulton (1984) SO to home Retrospective after Age65+ Discriminant function hospitalization Wan & Weissert N= 1,119 Longitudinal (3 years) NH admission (1981) Age65+ Retrospective Linear regression : I Weinberger et al. N = 155 public housing Longitudinal ( 1 year) NH admission (1986) tenants Retrospective : I Age 45+, mean 71 Logistic regression :I Welin et al. (1985) N= 989 (151 died) Longitudinal (9 years) Mortality Age 50,60 Prospective Logistic regression Wingard. Williams-N= 1,504 Longitudinal ( 14 years) NH admission Jones, McPhillips, Age40+ Retrospective Kaplan, & BarrettDescriptions Connor (1990) I Winograd et al. N= 401 (all male) Longitudinal ( 1 year) Mortality (1991) Age65+ Prospective cohort NH admission Proportional hazard analysis, logistic regression Wolinsky& LSOA Longitudinal ( 4 years) Mortality Johnson (1992) N=4,503 Retrospective Age70+ Logistic regression, proportional hazard analysis Wolinsky, Callahan, LSOA Longitudinal ( 4 years) Mortality Fitzgerald. & N= 3,646 (549 NH) Retrospective NH admission Johnson ( 1992) Age70+ Logistic regression 163

PAGE 174

I I I I :I I I Table 8.1. (Cont.) Study Samplea Wolinsky, Johnson, LSOA & Stump (1995) N= 7,627 (2,867 died) Age 55+ Young, Forbes, & Canada; survey of Hirdes (1994) disabled N=?? Age6S+? Zuckerman, Kasl, & N = 398 ( 47 died) Ostfeld (1984) Age 62+ Type of study & method of analysis Longitudinal (8 years) Retrospective Proportional hazard analysis Cross-sectional Logistic regression Longitudinal (2 years) Retrospective Logistic regression Dependent variableb Mortality NH admission Mortality Note. a Alameda= Alameda Human Population Laboratory, EPESE =Established Populations for Epidemiological Study of the Elderly, Framingham= Framingham Heart Study, LSOA = Longitudinal Study on Aging, MacArthur= MacArthur Foundation Research Network on Successful Aging, MOS = Medical Outcomes Study, Natl LTC Channeling= National Long Term Care Channeling Demonstration Project, Natl LTC Survey =National Long Term Care Surveys 1982 & 1984, NHANES-1 =National Health & Nutrition Examination Survey Epidemiologic Follow-Up Study; bNH =nursing home. 164

PAGE 175

Table B.2. Risks of Mortality Category of variable Study perc ADL-pfm-psy-own/ fit statistict age sex dem SES heallh s1111 dx clin h beh IADL mob funcl cogn depr psych soc mslal alone ncl soc ulil univariate, bivariate anal:'ises Berkman & Syme ./ ./ ./ ./ ./ ( 1979) LaRue, Bank, ./c ./c Jarvik, & Hetland (1979) Manton ( 1988) ./ -Linear regression, discriminant function analyses 0'\ U\ Abramson, Gofin, ./ ./ ./ ./ ./ ./ &, Peritz ( 1982) (R1 = 0.21) Blazer ( 1982) ./ ./ ./ ./ ./ ./ ./ Branch & Ku ./ ./ ./ ./ ./ ./ ./ (1989) Campbell, Diep, ./ ./ ./ ./ ./ ./ Reinken, & McCosh ( 1985) (R == 0.228)

PAGE 176

Table B.2. (Cont.) Category of variable Study perc ADL-pfm-psy-own/ fit statlstict age sex dem SES health stat dx clin h beh IADL mob funct cogn depr psych soc mstat alone net soc util Hodkinson & -/ .J'C y"C y"C Exton-Smith ( 1976) (R = 0.44 M, 0.50 F) Markides & ./ ./ ./ ./ -/ -/ Pappas ( 1982) Pfeiffer (I 970) tl tl ./ ./ ./ ./ ./ ./ (R2 = 0.58M, 0.38F) -Logistic regression, proportional hazard analyses 0\ 0\ Boult, Kane, -/ ./ ./ ./ Louis, Boult & McCaffrey ( 1994) Pijls, Feskens, & -/ tl ./ -/ ./ Kromhout ( 1993) Berkman, Leo./ ./ .; .; -/ .; .; .; Summers, & Horwitz (I 992) Borawski, Kinney, .; .; -/ .; & Kahana ( 1996) (G1 = 112.762 df23)

PAGE 177

Table 8.2. (Cont.) Category of variable Study perc ADL pfm-psy-own/ fit statistict age sex dem SES health stat dx clin h bch IADL mob funct cogn depr psych soc mstat alone net soc utll Chipperfield .;' .;' .;' .;'C .;'C .;'C (1993) Davis, R.B., et al. .;' .;' (1995) Deeg, van .;' .;' Zonneveld, van derMaas, & Habbema ( 1989) ..... (R2 = 0.202) 0\ .;' ./ .;' ./ ./ -....J Grand, ./ ./ ./ Grose laude, Bocquet, Pous, & Albarede ( 1990) ('l = 161 df8 p<.OOI, R2 = 0.126) Greiner, Snowdon, cl ./ ./ & Greiner ( 1996) Hanson, lsacsson, d ./ .;' ./ Janzon, & Lindell (1989)

PAGE 178

----------------------------------------------------------------_ _ __ _ -_-_ _-_ -. ---------. --------------------Table 8.2. (Cont.) Category of variable Study perc ADL-pfm-psy-own/ (model fit statistict age sex dem SES health stat dx clin h beh IADL mob 1\mct cogn depr psych soc mstat alone net soc utll Hirdes & Forbes t7 ./ ./ ./ ./ ( 1992) (R = 0.209) Ho (1991) ./ ./ ./ ./ ./ ./ ./ House, Robbins, ./ ./ ./ ./ ./ ./ & Metzner ( 1982) Idler & Angel ./ ./" ./ e ./ (1990) ..... (01 "'186.6M 74.2F) 0\ 00 Idler & Kasl ./ ./ ./ ./ ./ ./ ( 1991) (01;:: 206.97 14 df) Idler, Kasl, & ./ ./ ./ ./ ./ Lemke (1990) (01"' 153.8814df to 232.97 IOdf) lncalzi et at. ./ ./ ./ (1992) Jagger & Clarke ./ ./ ./ ./ ./ ./ ./ ( 1988) Kaplan, G., Barell, ./ ./ ./ ./ ./ ./ ./ & Lusky ( 1988)

PAGE 179

Table 8.2. (Cont.) Cateso!:X of variable Study perc ADL-pfmpsyown/ (model fit statistict nge sex dcm SES health stat dx clin h bch IADL mob funct cugn depr psych sue mstat alone net soc utll Kaplan, G.A., & "' "' "' "' "' "' Camacho ( 1983) Kaplan, G.A., "' "' Seeman, Cohen, Knudsen, & Guralnik ( 1987) Kaplan, G.A., et "' "' "' "' "' al. (1994) Mor, Wilcox, "' "' "' "' "' -Rakowski, & Hiris 0'1 \0 (1994) Mossey & Shapiro ./ "' "' "' "' "' "' "' (1982) Narain et al. "' "' "' "' (1988) Orth-Gomer & "' "' Johnson ( 1987) Parker, Thorslund, ./'' "' "', "', & Nordstrom (1992) Rakowski, Mor, & ./ "' "' ./ ./ Hiris (1991)

PAGE 180

w - -- - -- --- --------Table 8.2. (Cont.) Categm:x of variable Study perc ADL pfm psy own/ (model fit statistict nge sex dem SES health stat dx clin h bch IADL mob funct cogn dcpr psych soc mstnt alone net soc util Reuben, Siu, & ./ ./ Kimpau ( 1992) (x2= 11.93 to 16.95) Schoen bach, ./ .I .I .I ./ Kaplan, Fredman, & Kleinbaum (1986) Schoenfeld, ./ ./ ./ .I .I Malmrose, Blazer, .... Gold, & Seeman ....:I (1994) 0 Seeman, Kaplan, .!'' .I ./' Knudsen, Cohen, & Guralnik ( 1987) Sorensen ( 1988) ./ ./ .I ./ ./ .I .I Steinbach ( 1992) ./ ./ .I ./ ./ (F= 34.12 df9) Sugisawa, Liang, ./ .I .I ./ .I & Liu (1994) (LL = -513.684) Vogt, Mullooly, .I ./ Ernst, Pope, & Hollis ( 1992)

PAGE 181

Table B.2. (Cont.) Category of variable Study perc ADLpfm psyown! (model tit statistic)" age sex dem SES health stat dx clln h beh IAIJL mob funct cogn depr psych soc mslal 1llone nel soc ulil Welin et at. (1985) ./ " ./ ./ ./ Winograd et al. d ./ (1991) Wolinsky, ./ ./ ./ ./ ./ e Callahan, Fitzgerald, & Johnson ( 1992) (G = 515 df 483) Wolinsky & ./ ./ c ./ c ./ ./ ./ :::; Johnson ( 1992) ..(G1 = 136 df25) Wolinsky, ./ ./ ./ ./ ./ ./ ./ Johnson, & Stump ( 1995) (X2 = 2067 df47 p = 0 001) Zuckerman, Kasl, ./ ./ ./ ./ ./ & Ostfeld ( 1984)

PAGE 182

--...1 N Table Il2. (Cont.) Catego!1 of variable11 Study perc ADL-pfm-psy-own/ fit statistic}" age scK dcm SES health stat dK clin h bch IADL mob funct cogn depr psych soc mstnt nlonc net soc utll # studies including 49 41 23 36 38 4 39 34 27 30 21 16 10 9 12 18 32 21 22 32 27 variable # studies finding 29 20 3 15 22 2 25 20 I 0 22 13 6 4 0 2 5 6 5 8 16 I 0 significant relationship Note. a model fit statistics (not always given): R2 =coefficient of multiple determination, G = -2LL =(-2)(log likelihood)= likelihood ratio test, df = degrees of freedom; h age = age in years, sex = male/female gender, dem = other demographics, SES = income/education, perc health = self-reported health status, stat= other measures of health status, dx = # of diagnoses/specific diagnosis/chronic conditions, clin =other clinical measures (e.g., blood pressure), h beh"" health behaviors (e.g., smoking), ADL-IADL =(Instrumental) Activities of Daily Living, pfm-mob = physical performance & mobility, funct =other measures of physical function, cogn =cognition, depr =depression, psych= other measures of mental health, psy-soc =psychosocial (e.g., satisfaction, efficacy), mstat = marital status, own/alone = own home/live alone, net = social network/informal social support, soc = other measures of social support, util = health care utilization; c significant for some but not all subpopulations; '1 matched or single sex ./=independent variable with significant relationship with outcome, p < 0.05; =non-significant independent variable (or no significance reported).

PAGE 183

Table 8.3. Risks of Institutionalization Category of variable Study perc ADLpfm-psyown/ (model fit statistic)" age sex dem SES health stat dx clin h beh IADL mob funct cogn depr psych soc mstat alone net soc utll Descriptive. univariate. bivariate analyses Arling& McAuley (1983) Brody, Poulshock, ./ & Masciocchi (1978) Kane, Matthias, & ./ ./ ./ ./ Sampson (1983) ....,J w Kraus et at. ( 1976) ./ ./ ./ ./ ./ ./ ./ ./ ./ Murtaugh, ./ ./ Kemper, & Spillman ( 1990) Wingard, ./ ./ ./ ./ ./ Williams-Jones, McPhillips, Kaplan, & SarrettConnor ( J 990) Linear regression. discriminant function analyses Branch & Ku ./ ./ ./ ./ ./ ./ (1989)

PAGE 184

------------------Table B .3. (Cont.) Category of variable Study perc ADL-pfm-psy-own/ (model fit statistic)" age sex dem SES health stat dx clin h beh IADL mob funct cogn depr psych soc mstat alone net soc utll Cohen, C .A., et at. ./ ./ ./ (1993) Greenburg & Ginn ./ ./ ./ ./ ./ ./ (1979) (R1 = 0 68) Manton ( 1988) ./ McAuley & ./ ./ ./ ./ Prohaska ( 1981) ,_. Palmore(1976) ./ ./ -...l .c=a. Smyer ( 1980) '1 ./ ./ ./ Teresi et at. ./ ./ ./ ./ (1989) (R1 = 0.45, 0.32) Vicente, Wiley, & ./ ./ Carrington ( 1979) Wachtel, Derby, & ./ ./ ./ ./ ./ Fulton ( 1984) Wan & Weissert ./ ./ ./ ./ ./ ( 1981) (R1 = 0.46)

PAGE 185

. -----------------Table BJ. (Cont.) Category of variable Study perc ADJ.-pfm-psy-own/ (model fit statistic)" age sex dcm SES hcallh sial dx clin h beh IADL mob funcl cogn dcpr psych soc mslal alone nel soc ulil Logistic regression. proportional hazard analyses Bauer (1996) -1' (-2LL = 10607, x2 = 124.97 p < 0.0001) Boaz & Muller (1994) Branch & Jette -1' (1982)
PAGE 186

Table B.3. (Cont.) Catego!l of variable Study perc ADLpfmpsyown/ fit statistict' ogc sex dcm SES henlth slot dx clin h bch IADL mob funct cogn depr psych soc mstot alone net soc util Ford, Roy, Haug, -/ -/ -/ ./ Folmar, & Jones ( 1991) Freedman, ,;c ,;c e Berkman, Rapp, & Ostfeld ( 1994) Glazebrook, ./ ./ ./ ./ ./ ./ ./ ./ Rockwood, Stolee, Fisk, & Gray ..... (1994) ....,J (-2LL = 671.677 0\ dfl5 p
PAGE 187

Table B.3. (Cont.) Category of variable Study perc ADL-pfm-psy-own/ (model fit statistic)a age sex dem SES health stat dx clln h beh IADL mob funct cogn dcpr psych soc mstllt nlonc net soc utll McFall & Miller -1' -ol' -1' -1' (1992) (X2 = 94.4 df II) Mor, Wilcox, et -1' -ol' -1' at. (1994) Narain et al. '1 -1' -1' -1' -1' ( 1988) Newman, Struyk, -1' -1' -ol' -ol' -1' -ol' -1' -1' Wright, & Rice j 1587.3) Nocks, Learner, -1' -1' -1' Blackman, & Brown ( 1986)
PAGE 188

-Table 8.3. (Cont.) Category of variable Study perc ADL-pfm-psy-own/ (model fit statistict oge sex dem SES hculth stot dx clin h beh IADL mob funct cogn depr psych soc mstol alone net soc utll Rockwood, Stolee, v' v' v' v' v' v' & McDowell (1996) (X.2 = 2602,9 df24, p < 0.0001) Shapiro & Tate v' v' v' v' v' v' v' v' v' v' (1985) (x.2 = 218.0, 482.4) Speare, Avery, & v' v' v' v' -1 Lawton ( 199 I ) Steinbach (I 992) v' v' v' -1 v' (F= 27,1 df8) Tsuji, Whalen, & v' v' v' Finucane ( 1995) Young, Forbes, & v' ./ ./ ./ ./ ./ Hirdes ( 1994 ) Weinberger et al. ./ ./ ./ (1986) Winograd et al. ./ (1991)

PAGE 189

--....l \0 Table 8.3. (Cont.) Category of variable Study perc ADL pfm psy ownl (model fit statistic}" age sex dem SES health stat dx clln h beh IADL mob funct cogn depr psych soc mstat alone net soc utll Wolinsky, -/ -/ -/ -/ ./ -/ ./ Callahan, Fitzgerald, & Johnson ( 1992) (X.2 = 4739 df 4872 p < 0.001) #studies including 43 42 33 27 14 2 25 7 2 38 12 15 27 3 14 9 35 29 27 31 28 variable # studies finding 25 8 13 7 9 0 I I 3 0 28 6 8 17 2 8 5 13 17 6 18 13 significant relationship Note. a model fit statistics (not always given): R2 =coefficient of multiple determination, G = -2LL =(-2)(1og likelihood)= likelihood ratio test, df = degrees of freedom; h age = age in years, sex = male/female gender, dem = other demographics, SES = income/education, perc health = self-reported health status, stat = other measures of health status, dx = II of diagnoses/specific diagnosis/chronic conditions, clin =other clinical measures (e.g., blood pressure), h beh =health behaviors (e.g., smoking), ADL-IADL =(Instrumental) Activities of Daily Living, pfm-mob =physical performance & mobility, funct =other measures of physical function, cogn =cognition, depr =depression, psych= other measures of mental health, psy-soc =psychosocial (e.g., satisfaction, efficacy), mstat =marital status, own/alone= own home/live alone, net= social network/informal social support, soc =other measures of social support, util =health care utilization; c significant for some but not all subpopulations; 11 matched or single sex. -/=independent variable with significant relationship with outcome, p < 0.05; =non-significant independent variable (or no significance reported).

PAGE 190

Table B.4. Risks of Functional Decline Category of variable Study perc ADL-pfin-psy-own/ fit statistic)" age sex dem SES health stat dx clin h beh IADL mob funct cogn depr psych soc mstal alone net soc utll univariate, bivarite Berkman et al. ,, ,, (1993) Hoeymans, Feskens, van den Bos, & Kromhout (1997) Linear regression, discriminant function -Branch & Ku 00 0 (1989) Kaplan, G.A., Strawbridge, Camacho, & Cohen (I 993) Keil et al. (I 989) .. (R1 = 0.14-0.26) Lammi et al. ,, (1989) (R1 = 0.07) Mnrkides & Lee (1990)

PAGE 191

Table B.4. (Cont.) Category of variable Study perc ADL-pfm-psy-own/ (model fit statistict age sex dem SES health stat dx clin h beh IADL mob funct cogn depr psych soc mstat alone net soc utll Palmore, Nowlin, ./ ./ ./ ./ & Wang (1985) (adj. R1 = 0.37) Parkerson et al. ./ ./ ./ (1989) (R1 = 0.11 0.41) Stewart et al. ./ ( 1989) (R1 = 0.12-0.24) Verbrugge, ./ ./ ........ Reoma, & Gruber00 ........ Baldini ( 1994) Logistic regression, RrORortiona1 hazard Beckett et al. ./ ./ (1996) Boult, Kane, ./ ./ ./ ./ Louis, Boult, & McCaffrey ( 1994) Crimmins & Sito ./ ./ ./ ./ ./ ./ ./ ( 1993)

PAGE 192

Table 8.4. (Cont.) Category of variable Study perc ADL-pfm-psy-own! (model fit statistic}" ngc sex dcm SES henlth stnl dx clin h beh IADL mob funct cogn depr psych soc mstat alone net soc utll Greiner, Snowdon, J -1' & Greiner ( 1996) (R1 = 0.40) Guralnik & -1' -1' -1' -1' Kaplan ( 1989) Harris, Kovar, -1' Suzman, Kleinman, & Feldman ( 1989) ,..... House et al. (1994) Idler & Kasl -1' -1' -1' -1' -1' -1' -1' -1' ( 1995) (R1 = 0.34) Mor, Murphy, et -1' al. (1989) Mor, Wilcox, et -1' -1' al. ( 1994) Pinsky et al. -1' ( 1985) Pinsky, Leavet1on, -1' -1' & Stokes ( 1987)

PAGE 193

Table 8.4. (Cont.) Category of variable Study perc AOLpfmpsyown/ fit statistict age sex dem SES health stat dx clin h beh IADL mob funct cogn depr psych soc mstat alone net soc utll Roos & Havens (1991) Sager et al. ( 1996) Seeman et al. (1994) Seeman et al. ./ ./ (1995) (MLR R1 = 0.47) Strawbridge, ./ ./ Cohen, Shema, & 00 VJ Kaplan (1996) Structural eguation models, Rath Newsom & Schulz (1996)

PAGE 194

-----------. . Table 8.4. (Cont.) --------Category of variable perc ADL-pfmpsyown/ Study (model fit statistict nge sex dem SES henlth stat dx clln h beh IADL mob funct cogn depr psych soc mstnt ulone net soc utll # studies including variable 28 27 22 23 II 3 23 17 12 10 5 7 6 4 6 8 17 8 9 14 12 # studies finding 19 8 4 IS 7 0 16 14 9 6 3 3 3 3 I 4 3 2 4 6 3 significant relationship Note. c model fit statistics (not always given): R1 =coefficient of multiple determination, G = -2LL =( -2)(log likelihood)= likelihood ratio test, df = degrees of freedom; h age = age in years, sex = male/female gender, dem = other demographics, SES = income/education, perc health = self-reported health status, stat = other measures of health s t atus, dx = # of diagn oses/specific diagnosis/chronic conditions, clin =other clinical measures (e.g., blood pressure), h beh =hea lth behaviors (e .g., smoking), ADL-IADL =(Instrumental) Activities of Daily Living, pfm-mob = physical performance & mobility, funct = oth er measures o f physical function, cogn =cognition, depr = depression, psych =other measures of mental health, psy-soc = psychosocial (e.g satisfaction, efficacy), mstat = marital status, own/alone= own home/live alone, net = social network/informal social support, soc =other measures of social support, util =health care utilization; c significant for some but not all subpopulations; d matched or single sex ./ = independent variable with significant relationship with outcome, p < 0 .05; =non-significant independent variable (or no significance reported).

PAGE 195

Table B.S. Characteristics of Studies ofPredictors ofPerceived Health Type of study & method of Dependent Study Samplea analysis variable I Barsky, Cleary, & N= 188 (88 Cross-sectional Perceived Klerman (1992) hypochondriacs, 100 not) Linear regression health I Age mean 58 : I Blaum, Liang, & Liu N= 11,497 Cross-sectional Perceived : I (1994) Age65+ Linear regression, path health : I analysis Blazer & Houpt Durham, NC Cross-sectional Perceived (1979) N= 719 unimpaired of Comparison health 977 Age65+ Fylkesnes & Forde Norway Cross-sectional Perceived (1991) N= 18,560 Logistic regression health Age 20-61 Goldstein & N= 1,034 Cross-sectional Perceived Hurwicz ( 1989) Age65+ Linear regression health I : I Goldstein, Siegel, & Los Angeles Health Longitudinal ( 1 year) Change in I Boyer ( 1984) Survey Retrospective perceived I N=903 Correlation health Age 18+ I I Gottlieb & Green National Survey of Cross-sectional Perceived (1984) Personal Health Practices Linear regression health and Consequences I N= 3,025 Age20-64 : I Hays, Schoenfeld, & EPESE Cross-sectional Perceived Blazer (1996) N=4,162 Linear regression health Age65+ 185

PAGE 196

Table 8.5. (Cont.) Type of study & method of Dependent Study Same lea variable Hirdes & Forbes Ontario Longitudinal Longitudinal (10, 16 years) Decline in (1993) Study of Aging Retrospective perceived N= 1,702 (rated good Logistic regression health health at base) Age45 Hoeymans, Feskens, N= 721 Longitudinal ( 5 years) Decline in van den Bos, & Age70+ Retrospective perceived Kromhout ( 1997) Bivariate health Johnson& LSOA Cross-sectional Perceived Wolinsky (1993) N= 5,151 Structural equation health Age70+ I Jylha, Leskinen, N= 183, 188, 176 (all Cross-sectional Perceived Alan en, Leskinen, & male) Logistic regression, health Heikkinen (1986) Age 31-35, 51-55, 71-75 structural equation Levkoff, Cleary, & N=460 Cross-sectional Perceived Wetle ( 1987) Age45+ Linear regression health Liang ( 1986) 1968 National Senior Cross-sectional Self-reported I Citizens' Survey Structural equation physical health I I N = 4 samples @725-764 Age65+ I I Lindgren, Albertina, Sweden Cross-sectional Perceived Svardsudd, & N=706 Linear regression health Tibblin (1994) Age 75+ Linn & Linn (1980) N=286 Cross-sectional Perceived Age 65+, mean 74 MANOVA health Markides & Lee N=254 Longitudinal (8 years) Perceived (1990) Age 68+ Retrospective health Linear regression 186

PAGE 197

Table B.S. (Cont.) Type of study & method of Dependent Study Samplea analysis variable Markides & Martin N= 480 (375 Mexican Cross-sectional Perceived (1979) American) Path analysis health Age60+ Minkler& Alameda Longitudinal (5 years) Change in Langhauser(l988) N=280 Retrospective perceived Age60+ Discriminant analysis health :I Mor-Barak. N= 175 Cross-sectional Perceived I I Scharlach, Birba, & Age 55+, mean 70 Linear regression health Sokolov (1992) Murray, Dunn, & London Cross-sectional Perceived Tarnopolsky (1982) N= 5,904 Logistic regression health Age 16-75+ I I Pilpel, Carmel, & Israel Cross-sectional Perceived Galinsky ( 1988) N=605 Linear regression health Age65+ Rodin & McAvay N=2SI Longitudinal (3 repeated Change in (1992) Age62+ measures of 7 possible) perceived Synthetic cohort health Logistic regression Stewart et al. ( 1989) MOS Cross-sectional Perceived N=9,385 Linear regression health Age 18+, mean 46 Sugisawa, Liang, & Japan Longitudinal (3 years) Perceived Liu (1994) N=2,200 Retrospective health Age60+ Linear regression Tessler & Mechanic N= 989 WI; 1,391 Cross-sectional Perceived (1978) students; 339 prison; 379 Linear regression health Duke Age most
PAGE 198

i I Table B.S. (Cont.) Type of study & method of Dependent Study Sample" analysis variable Tissue ( 1972) N = 256 non-institution Cross-sectional Perceived welfare recipients Correlation health Age median 68 Verbrugge& N= 165 Longitudinal ( 1 -2 years) Perceived Balaban ( 1989) Age 55+ Retrospective health Linear regression, ANOV 9 repeated measures Wan(1976) N= 11,153 Cross-sectional Perceived Age 58-63 Linar regression health Zonderman, Leu, & Baltimore Longitudinal Cross-sectional Perceived : I Costa ( 1986) Study of Aging ANOVA health N=910 I Age 17-95, mean 50 -55 I Note. -a Alameda= Alameda Human Population Laboratory, EPESE = Established Populations for Epidemiological Study of the Elderly, LSOA = Longitudinal Study on Aging, MOS =Medical Outcomes Study 188

PAGE 199

Table 8.6. Predictors of Perceived Health Cateso!X of variable Study perc ADL-pfm-psy-own/ (model fit statistict age sex dem SES health stat dx clin h beh IADL mob funcl cogn dcpr psych soc mstnt alone net soc utll DescriRtive, univariate, bivariate Blazer & Houpt -/ -/ -/ -/ (1979) Goldstein, Siegel, -/ -/ -/ -/ & Boyer (1984) Hoeymans, -/ Feskens, van den -Bos, & Kromhout 00 (1997) \0 Tissue ( 1972) -/ -/ -/ Linear regression, discriminant function Barsky, Cleary, & -/ -/ -/ Klerman (1992) (R1 = 0.76) Goldstein & -/ -/ -/ -/ -/ -/ -/ -/ Hurwicz ( 1989) (R1=0.13) Gottlieb & Green -/ -/ -/'' o/c ( 1984) (R1 = 0 18-0.20)

PAGE 200

. .. . ---..... --------------------------Table 8.6. (Cont.) Catesory of variable Study perc ADL-pfm-psy-own/ fit statisticf age sex dem SES health sial dx clln h beh IADL mob funcl cogn depr psych soc mslat alone net soc util Hays, Schoenfeld, ./ ./ ./ ./ ./ ./ ./ ./ ./ & Blazer ( 1996) (R1 = 0.30) Levkoff, Cleary, ./ ./ ./ ./ & Wetle (1987) (K = 0 .17) Lindgren, ./ o/ Svardsudd, & Tibblin (1994) -(R1 = 0.27) \0 Linn & Linn 0 ./ ./ o/ (1980) Markides & Lee o/ ./ o/ (1990) Minkler& ./ ./ ./ Langhauser ( 1988) Mor-Barak, ./ ./ ./ Scharlach, Birba, & Sokolov ( 1992) (R1 = 0.28) Pilpel, Carmel, & o/ ./ ./ ./ Galinsky ( 1988) (R1 = 0 .32 0.39)

PAGE 201

Table 8.6. (Cont.) Category of variable Study perc ADL-pfm-psy-own/ (model fit statistlct oge sex dem SES health slot dx clin h beh IADL mob funct cogn dcpr psych soc mstat alone net soc utll Stewart et al. y" (1989) (R1:: 0.29) Sugisawa, Liang, y" y" y" y" y" y" y" y" & Liu (1994) (R1 = 0 29) Tessler & y" y" y" y" y" y" y" Mechanic ( 1978) (R1 = 0.05 0.28) -Verbrugge & y" y" y" \0 -Balaban ( 1989) Wan (1976) y" y" y" o/ o/ (R' = 0.44) Zonderman, Leu, .f o/ o/ & Costa ( 1986) Logistic regression, RfORortional hazard Fylkesnes & Forde .f o/ o/ o/ o/ y" o/ o/ o/ .f o/ ( 1991) (R1 = 0.21-0.24)

PAGE 202

Table 8.6. (Cont.) Category of variable Study perc ADL pfm psy own/ (model fit statistic)" age sex dcm SES health stnt dx clin h beh IADL mob funct cogn depr psych soc mstat alone net soc util Hirdes & Forbes ./ ./ ./ ./ ./ (1993) (-2LL = 39.2 175.8, R = 0.201 0.242) Markides & ./ ./ ./ ./ ./ ./ Martin ( 1979) Murray, Dunn, & ./ ./ ./ ./ ./ ./ ./ ./ ,_ Tarnopolsky IS (1982) (G2 = 644 df 568) Rodin & McAvay ./ .tc ./ ./ e e ./ (1992) (-2LL =-120.24) Structural equation model. path analysis Blaum, Liang, & ./ ./ Liu (1994) (R2 = 0.31) Johnson & ./ ./ ./ ./ ./ Wolinsky ( 1993)
PAGE 203

Table B.6. (Cont.) Category of variable Study perc ADL-pfmpsyown/ (model fit statistict age sex dem SES health stat dx clin h beh IAOL moh funct cogn depr psych soc mstat alone net soc utll Jylhli, Leskinen, ./ '' ./ ./ ./ c Alanen, Leskinen, & Heikkinen ( 1986) (R1= 0.45 0.51) Liang ( 1986) ./ ./ ./ ./ #studies including 28 21 16 22 1 2 25 17 9 14 11 10 2 9 8 13 14 3 10 14 16 variable .... # studies finding 15 3 5 13 1 0 20 8 7 9 8 3 0 7 6 8 4 1 6 5 6 \0 'fi w sJgnt Jcant relationship Note. a model fit statistics (not always given): R2 =coefficient of multiple determination, G = -2LL =(-2)(1og likelihood)= likelihood ratio test, df =degrees of freedom; 6 age= age in years, sex= male/female gender, dem =other demographics, SES = income/education, perc health = self-reported health status, stat = other measures of health status, dx = # of diagnoses/specific diagnosis/chronic conditions, clin =other clinical measures (e.g., blood pressure), h beh =health behaviors (e.g., smoking), ADL-IADL =(Instrumental) Activities of Daily Living, pfm-mob =physical performance & mobility, funct =other measures of physical function, cogn =cognition, depr =depression, psych= other measures of mental health, psy-soc =psychosocial (e.g., satisfaction, efficacy), mstat =marital status, own/alone= own home/live alone, net= social network/informal social supp011, soc = other measures of social support, uti I = health care utilization; c significant for some but not all subpopulations; "matched or single sex ./=independent variable with significant relationship with outcome,p < 0.05; ""non-significant independent variable (or no significance reported).

PAGE 204

I I APPENDIXC DATA DESCRIPTION TABLES These tables provide information about coding and ranges, the number of non missing responses, and mean values for baseline variables, as well as bivariate correlations between the baseline explanatory variables and perceived health at 12 months (Kendall tau-b correlations for dichotomous or ordinal variables, Pearson r correlations for continuous variables; all correlations 10.0701 are significant at p < 0.05). For dichotomous variables, the mean provides a measure of the prevalence of the condition or indicator. Table C. I. Description of Baseline Variables Conceptual category Baseline variable Baseline perceived health Sociodemographics Age Gender Race/ethnicity Education Income Chronic conditions Congestive heart failure Chronic lung disease Blindness/trouble seeing Deamessltrouble hearing Diabetes Asthma Coding; range 1 = poor, 2 = fair, 3 = good, 4 = very good. 5 =excellent Range: 58.7 to I 00.4 years 0 = male, 1 = female 1 =Asian/PI, 2 =Black/Aft Amer, 3 = Hispanic, 4 =Native American, 5 = White/Caucasian, 6 = other 1 = grades 0-8, 2 = grades 9-11, 3 = HS grad. 4 = some college, 5 = college grad. 6 = post-college 1 = <$3,000; 2 = $3,000-$5,999; 3 = $6,000-$9,999; 4 = $10,000-$14,999; 5 = $15,000-$24,999; 6 = $25,000-$34,999; 7 =>$35,000 0 =no, 1 =yes 194 N 684 688 692 688 688 497 684 684 683 683 684 685 Corr w/ perc Mean hltha 3.15 .581 7339 .008 .62 024 3.59 .196 5.03 .157 .07 -.172 .15 -.186 .13 -.090 29 -.028 .13 -.140 .14 -.138

PAGE 205

Table C.l. (cont.) Corr : I Concef!.tual Categ,o!J!. w/perc Baseline Variable Codin!i ranse N Mean hltha Ulcer or GI bleeding 684 .04 -.031 Arthritis/rheumatism 688 .60 -.I44 Sum of chronic conditions Range: 0 to 6 (of possible range 0 to 8) 688 1.55 -254 ID'd by patient Hypertension 0 =no. I =yes 686 .49 -.086 Angina 685 .13 -.092 Heart attack or MI 684 .17 102 Stroke 687 .08 -.128 Kidney disease 684 .04 050 Cancer 685 .18 -.029 Sum of chronic conditions Range: 0 to 6 (of possible range 0 to 6) 688 l.ll -.147 ID' d by doctor Worsening condition(s) 0 =no. 1 =yes 677 .09 162 Functional status Activities of Daily Living (ADL) Need help to toilet 0 =never. l =sometimes, 2 =most of the 685 .01 -.081 time Need help to bathe 685 .03 -.132 Need help to dress 685 .03 138 Need help to eat 685 .01 -.071 I Need help in/out of bed 685 .06 -.169 ADL (none vs. any) 0 = no dependencies, 1 = any dependencies 685 .07 -216 Instrumental Activities of Daily Living (IADL) Need help preparing O=no. 1 =yes 692 .06 -.195 meals Need help grocery 692 .12 -241 shopping Need help with chores 692 .11 -227 Need help managing 692 .02 -.088 money Need help with laundry 692 .06 -.180 Need help taking 692 .02 -.130 medications Need help with 692 .12 -209 transportation Need help using 692 .02 099 telephone IADL (none vs. any) 0 = no dependencies, I = any dependencies 692 20 -212 P!.D!..sical f!.gd"prmance Able to walk up/down 0 =no, I =yes 689 .91 .190 stairs 195

PAGE 206

II I I I Table C I (cont.) Corr Conce2.tual categoa w / perc Baseline variable Coding; range N Mean hltha Able to go out to movie, 689 .90 208 etc Able to do heavy work 689 .53 280 Able to walk Y: mile 689 73 286 Smn of physical Range: 0 to 4 (of possible range 0 to 4) 689 3 .07 .315 performance indicators Mobility 0 = bed-riddenneed help of person; 1 = 686 1.78 .339 need help of aid. have trouble; 2 = no limits Transportation 0 = drive own car, 1 = others drive, [not 688 .18 -212 reported at baseline: 2 =don't go out] Dgl,ression and dementia Feel depressed 0 =no, 1 =yes 660 .18 -204 Dementia Forget appointment 0 =no, I =yes 689 .07 -.026 Get lost traveling 689 .04 -.001 Problem paying bills 689 .03 -.024 See things not there 689 .03 -.042 Forget date/names 689 17 .099 Dementia (none vs. any) 0 = no indicators 1 = any indicators 689 21 .086 Health behaviors Smoking 0 =no, 1 =yes 670 .07 -.076 I!![grmal social srmoort Marital status 0 = not now married. I = married 689 .63 .052 Employment status, 0 = not now employed, I = fulllpart time 690 22 145 I Living arrangements Live alone 0 =no, I =yes 692 29 .049 Live with spouse 692 .62 -.060 Live with child(ren) 692 .07 -.010 Live with relative(s) 692 .02 -.047 Live with non-relative 692 .02 .014 Live with pet(s) 692 16 045 Type of housing, I = own residence, 0 = all other 690 .97 001 I Formal suooort Visiting nurse 0 =no, 1 = yes 692 .OI -.053 Visiting therapist 692 01 -.045 Home heal th aide 692 .OI -.076 Social worker 692 .004 -.073 Adult day care center 692 .004 .008 Transportation 692 .03 -.032 Home delivered meals 692 .007 -.102 Support (any vs none) 0 = no indicators, 1 = any indicators 692 .06 -.0 5 9 196

PAGE 207

:I Table C. I (cont.) Corr Conce'{!_tua/ categoo: w/perc Baseline variable Coding; range N Mean hltha Utilization (I 2 monthsl Daily medications baseline range: 0 to 41 662 4.76 -.215 I2-month range: 0 to 22 66I 5.I3 Group Experimental group 0 = control. I = participation 692 50 .013 .I Meetings attended range: 0 to 13 346 5.97 -.Oll (experimental group) I Percent of meetings range: 0 to 100% 346 48.2 .OI2 anended{experanental group) Physician 0 = not this physician. 1 = this physician; I 19 categorical variables Administrative data Primary care office visits 0 = none. I = any; range: Otoi 692 .82 .OOI Non-CHCC patient 0 = none. 1 = any; range: 0 to I 692 .06 -.082 I education services I Hospitalizations 0 = none. I = any; range: Otoi 692 .24 -.liS I Hospital length-of-stay Range: 0 to 251 days 692 3.97 -.I3I I Emergency visits 0 = none. 1 = any; range: 0 to I 692 .26 -.184 Skilled DID'Sing facility 0 = none. I = any; range: 0 to I 692 .03 -.140 episodes Home health episodes 0 = none. 1 = any; range: 0 to 1 692 .14 -.179 Durable medical 0 = none. I = any; range: 0 to I 692 .12 -.I99 eguiementiS!!J:!J:!Iies I Note. a Corr w/ perc hlth = Correlation with perceived health. I I :I 197

PAGE 208

TableC2 Description of Change Indicators Corr Concgo_tual w/ perc Cbanse variable Coding; ranse N Mean hlth0 Perceived health status <0 = poorer status, 0 = no change, >0 = 684 -.08 438 better status; range: -4 to 4 Chronic conditions : I Change in sum of <0 = decrease in number of conditions, 0 = 688 .09 -.058 conditions ID' d by no change, >0 = increase in number of patient conditions range: -5 to 3 (of possible range -8 to 8) Change in sum of <0 = decrease in number of conditions, 0 = 688 .07 068 conditions ID' d by no change, >0 = increase in number of doctor conditions range: -2 to 3 (of possible range -6 to 6) Change in conditions -1 = getting worse at baseline but not at 12 642 .04 088 getting worse months, 0 = no difference, 1 = getting worse at 12 months but not at baseline Function Change in Activities of <0 = decrease in dependencies, 0 = no 685 .03 -.08I Daily Living change, >0 = increase in dependencies; I range: -I to 1 I Change in Instrumental <0 = decrease in dependencies, 0 = no 690 .04 -.066 Activities of Daily Living change, >0 = increase in dependencies; range: -1 to 1 P!!J!.sical2f!!:[grmance Change in physical <0 = decreased performance, 0 = no 689 -23 .I 58 performance indicators change, >0 = improved performance; range: -4 to 4 (of possible range -4 to 4) Change in mobility <0 =decreased mobility, 0 =no change, 679 -.09 I83 >0 =increased mobility; range: -2 to 1 (of possible range -2 to 2) Change in transportation <0 = increased independence, 0 = no 679 .08 120 needed change, >0 = decreased independence; range: -1 to 2 (of possible range -2 to 2) Derz.ression and Dementia Change in depression <0 = less depressed, 0 = no change, >0 = 630 .02 062 more depressed; range : -I to 1 Change in dementia <0 = fewer indicators, 0 = no change, >0 = 689 .02 -.030 indicators more indicators; range: -1 to 1 Health behaviors Change in smoking -1 =smoke at baseline but not at 12 670 -.008 -.007 months, 0 =no change, 1 =smoke at 12 months but not at baseline 198

PAGE 209

Table C.2. (Cont.) Corr Concep_tual categoa w/ perc Change variable Coding; range N Mean hltha J!![grmal social Change in marital status -I =married at baseline but not at I2 686 -.007 .071 months, 0 = no change. I =married at I2 months but not at baseline Change in employment -I= FTIPT at baseline but not at 12 652 -.005 .020 status months. 0 = no change. I = FTIPT at 12 months but not at baseline Change in individual living -I = arrangement at baseline but not at I2 arrangement months. 0 = no change. I = arrangement at I2 months but not at baseline Live alone 689 .OOI .000 Live with spouse 689 -.009 .074 Live with child(ren} 689 -.004 -.034 Live with relative(s} 689 .004 -.022 Live with non-relative(s) 689 .003 -.033 Live with pet( s} 689 -.02 -.015 Change in housing <0 = less independence, 0 = no change. >0 686 -.01 -.014 = greater independence; range: -I to I Change in formal support <0 = fewer needs., 0 = no change, >0 = 690 .02 -.097 more needs; range: -I to 1 Utilization Change in number of daily Range: -36 to IS 637 .39 -.022 medications Note. Change= 12 month value minus baseline value. a Corr w/ perc hlth = Correlation with perceived health. 199

PAGE 210

. I : I BIBUOGRAPHY Abramson, J. H., Go fin. R., & Peritz, E. ( 1982). Risk markers for mortality among elderly men-a community study in Jerusalem. Journal of Chronic Diseases, 35, 565-572. Anderson, J. G., & Bartkus, D. E. (1973). Choice of medical care: A behavioral model of health and illness behavior. Journal of Health and Social Behavior, 14, 348-362. Antonovsky, A. (1979). Health, stress, and coping. San Francisco: Jossey Bass Publishers Antonovsky, A. (I 987). Unraveling the mystery of health: How people manage stress and stay well. San Francisco: Jossey-Bass Publishers. Antonucci, T. C., & Akiyama, H. (1987). Social networks in adult life and a preliminary examination of the convoy model. Journal of Gerontology, 42, 519-527. Applegate, W. B., Deyo, R., Kramer, A., & Meehan, S. (1991). Geriatric evaluation and management: Current status and future research directions. Journal of the American Geriatrics Society, 39(Suppl.), 2S-7S. Arling, G., & McAuley, W. J. (1983). The feasibility of public payments for family caregiving. Gerontologist, 23, 300-306. Bandura, A. (1977). Self-efficacy: Toward a unifying theory ofbehavioral change Psychological Review, 84, 191-215. Bandura, A. (I 982). Self-efficacy mechanism in human agency. American Psychologist, 37, 122-147. Barker, J. C. (1994). Recognizing cultural differences: Health-care providers and elderly patients. Gerontology and Geriatrics Education, 15, 9-21. Barsky, A. J., Cleary, P. D., & Klerman, G. L. (1992). Detenninants of perceived health status of medical outpatients. Social Science and Medicine, 34, 1147-1154. 200

PAGE 211

I I I Bauer, E. J. (1996). Transitions from home to musing home in a capitated long-term care program: The role of individual support systems. HSR: Health Services Research, 3I, 309-326. Beckett, L.A., Brock, D. B., Lemke, J. H., Mendes De Leon, C. F., Guralnik, J. M., FillenbaUill, G. G., Branch, L. G., Wetle, T. T., & Evans, D. A. (1996). Analysis of change in self-reported physical function among older persons in four population studies. American Journal of Epidemiology, I43, 766-778. Benfante, R., Reed, D., & Brody, J. (1985). Biological and social predictors of health in an aging cohort. JoW7Zt1l of Chronic Diseases, 38, 385-395. Bengtson, V. L., Burgess, E. 0., & Parrott, T. M. (1997). Theory, explanation, and a third generation of theoretical development in social gerontology. Journal of Gerontology, 52B, S72-S88. Berg, R. L., Hallauer, D. S., & Berk, S. N. (1976). Neglected aspects of the quality of life. HSR:Health Services Research, II, 391-395. Bergner, M. (1985). Measurement ofhealth status. Medical Care, 23,696704. Bergner, M., & Rothman, M. L. (1987). Health status measures: An overview and guide for selection. Annual Review of Public Health 8, 191-21 0. Berkman, L. F. (1988). The changing and heterogeneous nature of aging and longevity: A social and biomedical perspective. Annual Review of Gerontology and Geriatrics, 8, 37-68. Berkman, L. F., Leo-Summers, L., & Horwitz, R.I. (1992). Emotional support and survival after myocardial infarction: A prospective, population-based study of the elderly. Annals of Internal Medicine, 117, 1003-1009. Berkman, L. F., Seeman, T. E., Albert, M., Blazer, D., Kahn, R., Mohs, R., Finch, C., Schneider, E., Cotman, C., McCleam, G., Nesselroade, J., Featherman, D., Garmezy, N., McKhann, G., Brim, G., Prager, D., & Rowe, J. (1993). High, usual and impaired functioning in community-dwelling older men and women: Findings from the Mac Arthur foundation research network on successful aging. Journal of Clinical Epidemiology, 46, 1129-1140. 201

PAGE 212

I I I I I I ; I Berkman, L. F., & Syme, S. L. (1979). Social networks, host resistance, and mortality: A nine-year follow-up study of Alameda county residents. American Journal of Epidemiology, I09, 186-204. Blaum, C. S., Liang, J., & X. (1994). The relationship of chronic diseases and health status to the health services utilization of older Americans. Journal of the American Geriatrics Society, 42, 1087-1093. Blazer, D. G. (1982). Social support and mortality in an elderly community population. American Journal of Epidemiology, I I 5, 684-694. Blazer, D. G., & Houpt, J. L. (1979). Perception of poor health in the healthy older adult. Journal of the American Geriatrics Society, 27,330-334. Boaz, R. F., & Muller, C. F. (1994). Predicting the risk of"permanent'' nursing home residence: The role of community help as indicated by family helpers and prior living arrangements. HSR:Health Services Research, 29,391-414. Borawski, E. A., Kinney, J. M., & Kahana, E. (1996). The meaning of older adults' health appraisals: Congruence with health status and determinant of mortality Journal ofGerontology, 5IB, S157-S170. Boult, C., Kane, R. L., Louis, T. A., Boult, L, & McCaffrey, D. (1994). Chronic conditions that lead to functional limitation in the elderly. Journal of Gerontology, 49, M28-M36. L G., & Jette, A.M. (1982). A prospective study of long-term care institutionalization among the aged. American Journal of Public Health, 72, 13731379. L G., & Ku, L (1989). Transition probabilities to dependency, institutionalization, and death among the elderly over a decade. Journal of Aging and Health, I, 3 70-408. Breslow, L (1990). A health promotion primer for the 1990s. Health Affairs, 9, 6-21. Brody, S. J., Poulshock, S. W ., & Masciocchi, C. F. (1978). The family caring unit: A major consideration in the long-term support system. Gerontologist, I8, 556-561. 202

PAGE 213

. I I I ; I I I Brook, R H., Davies-Avery, A., Greenfield, S., Harris, L. J., Lelah, T., Solomon, N. E., Ware, J. E., Jr. (1977). Assessing the quality of medical care using outcome methods: An overview of the method. Medical Care, 15(Suppl. 9). Burckhardt, C. S., Woods, S. L., Schultz, A. A., & Ziebarth, D. M. (1989). Quality of life in adults with chronic illness: A psychometric study. Research in Nursing and Health, 12,347-354. Campbell, A. J., Diep, C., Reinken, J., & McCosh, L. (1985). Factors predicting mortality in a total population sample of the elderly. Journal of Epidemiology and Community Health, 39, 337-342. Chappell, N. L., & Badger, M. (1989). Social isolation and well-being. Journal ofGerontology, 44, S169-S176 Chipperfield, J. G. (1993). Incongruence between health perceptions and health problems: Implications for survival among seniors. Journal of Aging and Health, 5, 475-496. Cobb, S. (1976). Social support as a moderator of life stress. Psychosomatic Medicine, 38,300-314. W. C . Sharp, K., & Wilcox, J. A. (1983). Aging and perceived health status. Journal of Gerontology. 38,349-355. Cohen, C. A., Gold, D.P., Schulman, K. I., Wortley, J. T., McDonald, G., & Wargon, M. (1993). Factors determining the decision to institutionalize dementing individuals: A prospective study. Gerontologist, 33, 714-720. Cohen, M.A., Tell, E. J., & Wallack, S. S. (1986). Client-related risk factors of nursing home entry among elderly adults. Journal of Gerontology, 41, 785-792. Cohen, S. (1988). Psychosocial models of the role of social support in the etiology of physical disease. Health Psychology, 7, 269-297. Cooperative health care clinics; a new concept in the delivery of health care; a guide to establish CHCC programs. (1993). Wheat Ridge CO: Kaiser Foundation Health Plan of Colorado. Coughlin, T. A., McBride, T. D., & Liu K. (1990). Determinants of transitory and permanent nursing home admissions. Medical Care, 28, 616-631. 203

PAGE 214

I E. M., & Sito, Y. (1993). Getting better and getting worse: Transitions in functional status among older Americans. Journal of Aging and Health, 5, 3-36. Curb, J.D., J. M., LaCroix, A. Z., Korper, S. P., Deeg, D., Miles, T., & White, L. (1990). Effective aging: Meeting the challenge of growing older. Journal of the American Geriatrics Society, 38,827-828. Davis, R. B., lezzoni, L.l., Phillips, R. S., Reiley, P., Coffman, G. A., & Safran, C. (1995). Predicting in-hospital mortality: The importance of functional status information. Medical Care, 33,906-921. Deeg, D. J. H., van Zonneveld, R. J., van der Maas, P. J., & Habbema J.D. F. (1989). Medical and social predictors of longevity in the elderly: Total predictive value and interdependence. Social Science and Medicine, 29, 1271-1280. Diener, E. (1984). Subjective well-being. Psychology Bulletin, 95, 542-575. Dolinsky, A. L., & Rosenwaike, I. (1988). The role of demographic factors in the institutionalization of the elderly. Research on Aging, 10,235-257. Doll, R. (1992). Health and the environment in the 1990s. American Journal of Public Health, 82,933-941. Donabedian, A. (1988). The quality of care: How can it be assessed? Journal of the American Medical Association, 260, 1743-1748. Dubos, R. (1959). Mirage of health. New York: Harper. Duke University Center for the Study of Aging and Human Development. (1978). Multidimensional functional assessment: The OARS methodology. Durham, NC: Duke University Press. Edlund, M., & Tancredi, R. (1985). Quality of life: An ideological critique. Perspectives in Biology and Medicine, 28,591-607. Eisen, A. (1994). Survey of neighborhood-based, comprehensive community empowerment initiatives. Health Education Quarterly, 21, 235-252. Erlandson, D. A., Harris, E. L., Skipper, B. L., & Allen, S.D. (1993). Doing naturalistic inquiry: A guide to methods. Newbury Park, CA: Sage Publications. 204

PAGE 215

Evans, J. G. (1984). Prevention of age-associated loss of autonomy: Epidemiological approaches. Journal of Chronic Diseases. 37, 353-363. Evans, R. G., & Stoddart, G. L. (1990). Producing health, consuming health care. Social Science and Medicine, 31, 1347-1363. Farquhar, M. (1995a). Definitions of quality of life: A taxonomy. Journal of Advanced Nursing, 22, 502-508. Farquhar, M. (1995b ). Elderly people's definitions of quality of life. Social Science and Medicine, 41, 1439-1446. J. S. (1993). The relationship between socioeconomic status and health: A review of the literature. Milbank Quarterly, 71, 279-322. Ferraro, K. F. (1980). Self-ratings of health among the old and the old-old. Journal of Health and Social Behavior, 21, 377-383. Fillenbaum, G. G. (1979). Social context and self-assessments of health among the elderly. Journal of Health and Social Behavior, 20,45-51. Flaherty, J ., & Richman, J. ( 1989). Gender differences in the perception and utilization of social support: Theoretical perspectives and an empirical test. Social Science and Medicine, 28, 1221-1228. Flanagan, J. C. ( 1982). Measurement of quality of life: Current state of the art. Archives of Physical Medicine and Rehabilitation, 63, 56-59. Ford, A. B., Roy, A. W., Haug, M. R., Folmar, S. 1., & Jones, P. K. (1991). Impaired and disabled elderly in the community. American Journal of Public Health, 81, 1207-1209. Fozard, J. L., Metter, E. J., & Brant, L. J. (1990). Next steps in describing aging and disease in longitudinal studies. Journal of Gerontology, 45, P116-P127. Frankenberg, R. ( 1988). Your time or mine? An anthropological view of the tragic temporal contradictions of biomedical practice. International Journal of Health Services, 18, 11-34. Freedman, V. A., Berkman, L. F., Rapp, S. R., & Ostfeld, A.M. (1994). Family networks: Predictors of nursing home entry. American Journal of Public Health, 84, 843-845. 205

PAGE 216

I I I Fries, J. F. (1980). Aging, natural death, and the compression of morbidity. New England Journal of Medicine, 303, 130-135. Fylkesnes, K., & Forde, 0. H. (1991). The Tromso study: Predictors of self evaluated health-bas society adopted the expanded health concept? Social Science and Medicine, 32, 141-146. Gil, D. G. (1993). Beyond access to medical care: Pursuit of health and prevention of ills. Evaluation and the Health Profossions, /6,251-277. K., Rockwood, K., Stolee, P., J., & Gray, J. M. (1994). A case control study of the risks for institutionalization of elderly people in Nova Scotia. Canadian Journal on Aging, 13, 104-117. Goldstein, M.S., & Hwwicz, M.-L. (1989). Psychosocial distress and perceived health status among elderly users of a health maintenance organization. Journal of Gerontology, 44, P154-P156. Goldstein, M. S., Siegel, J. M., & Boyer, R. (1984). Predicting changes in perceived health status. American Journal of Public Health, 74, 611-614. Gottlieb, N.H., & Green, L. W. (1984). Life events, social life-style, and health: An analysis of the 1979 national survey of personal health practices and consequences. Health Education Quarterly, 11, 91-1 OS. Grand, A., Grosclaude, P., Bocquet, H., Pous, J., & Albarede, J. L. (1990). Disability, psychosocial factors and mortality among the elderly in a rural French population. Journal ofClinical Epidemiology, 43, 773-782. Greenberg, J. N., & Ginn, A. (1979). A multivariate analysis of the predictors oflong-term care placement. Home Health Care Services, 1, 75-99. Greene, V. L., & Ondrich, J.l. (1990). Risk factors for nursing home admissions and exits: A discrete-time hazard function approach. Journal of Gerontology, 45, S250-S258. Greiner, P. A., Snowdon, D. A., & Greiner, L. H. (1996). The relationship of self-rated function and self-rated health to concurrent fuctional ability, functional decline, and mortality: Findings from the nun study. Journal of Gerontology, 51B, S234-S241. 206

PAGE 217

i I I Guba, E. G. (Ed.). (1990). The paradigm dialog. Newbury Park CA: Sage Publications. Guralnik, J. M., & Kaplan, G. A. (1989). Predictors ofhealthy aging: Prospective evidence from the Alameda county study. American Journal of Public Health. 79,703-708. Hanley, R J., Alecxih, L. M. B., Wiener, J. M., & Kennell, D. L. (1990). Predicting elderly nursing home admissions; results from the 1982-1984 national long-term care smvey. Research on Aging, 12, 199-228. Hanson, B.S., lsacsson, S.-0., Janzon_ L., & Lindell, S.-E. (1989). Social network and social support influence mortality in elderly men: The prospective population study of"Men born in 1914." American Journal of Epidemiology. 130, 100-111. Harris T., Kovar, M.G., Snzman, R, Kleinman, J. C., & J. J. (1989). Longitudinal smdy of physical ability in the oldest-old. American Journal of Public Health. 79,698-702. Hays, J. C., Schoenfeld, D. E & Blazer, D. G. (1996). Determinants of poor self-rated health in late life. American Journal of Geriatric Psychiatry, 4, 188-196. Hays, R D., & Stewart, A. L. (1990). The structure of self-reported health in chronic disease patients. Psychological Assessment, 2, 22-30. Hippocrates. (n.d.). Airs, waters and places. Hirdes, J.P., & Forbes, W. F. (1992). The importance of social relationships, socioeconomic status and health practices with respect to mortality among healthy Ontario males. Journal of Clinical Epidemiology, 45, 175-182. Hirdes, J.P., & Forbes, W. F. (1993). Factors associated with the maintenance of good self-rated health. Joumal of Aging and Health. 5, 101-122. Ho, S. C. (1991 ). Health and social predictors of mortality in an elderly Chinese cohort. American Journal of Epidemiology, 133, 907-921. Hodkinson, H. M., & Exton-Smith, A. N. (1976). Factors predicting mortality in the elderly in the community. Age and Ageing, 5, 110-115. 207

PAGE 218

.I I Hoeymans, N., Feskens, E. J. M., van den Bos, G. A. M., & D. (1997). Age, time, and cohort effects on functional status and self-rated health in elderly men. American Journal of Public Health, 87, 1620-1625. House, J. Lepkowski, J. M., Kinney, A.M., Mero, R. P., Kessler, R. C., & Herzog, A. R. (1994). The social stratification of aging and health. Journal of Health and Social Behavior, 35,213-234. House, J. S., Robbins, C., & Metmer, H. L. (1982). The association of social relationships and activities with mortality: Prospective evidence from the Tecumseh community health study. American Journal of Epidemiology, 116, 123-140. House, J. S., Umberson, D., & Landis, K. R. (1988). Structures and processes of social support. Annual Review of Sociology, 14, 293-318. Hyland, M. E. (1993). The validity of health assessments: Resolving some recent differences. Journal of Clinical Epidemiology, 46, 1019-1023. Idler, E. L., & Angel, R. J. (1990). Self-rated health and mortality in the NHANES-1 epidemiologic follow-up study. American Journal of Public Health, 80, 446-452 Idler, E. L., & Kasl, S. (1991). Health perceptions and survival: Do global evaluations ofhealth status really predict mortality? Journal of Gerontology, 46, S55S65. Idler, E. L., & Kasl, S. V. (1995). Self-ratings ofhealth: Do they also predict change in functional ability? Journal of Gerontology, 50B, S344-S353. Idler, E. L., Kasl, S., & Lemke, J. H. (1990). Self-evaluated health and mortality among the elderly in New Haven, Connecticut and Iowa and Washington counties, Iowa, 1982-1986. American Journal of Epidemiology, 131, 91-103. Illich, I. (1976). Medical nemesis: The expropriation of health. New York: Pantheon Books. Incalzi, A. R., Capparella, 0., Gemma, A., Porcedda, P., Raccis, G., Sommella, L., & Carbonin, P. U. (1992). A simple method of recognizing geriatric patients at risk for death and disability. Journal of the American Geriatric Society, 40, 34-38. 208

PAGE 219

' I I ; I I I I : I I I I I I I I Jagger, C., & Clarke, M. (1988). Mortality risks in the elderly: Five-year follow-up of a total population. International Journal of Epidemiology, 17, 111-114. R. J., & Wolinsky, F. D. (1993). The structure ofhealth status among older adults: Disease, disability, functional limitation, and perceived health. Journal of Health and Social Behavior 34, 105-121. Jones, K., & Moon, G. (1987). Health disease and society. London: Routledge. Jylha, M. (1994). Self-rated health revisited: Exploring survey interview episodes with elderly respondents. Social Science and Medicine, 39, 983-990. Jylha, M., Leskinen, E., E., Leskinen, A. L., & E. (1986). Self-rated health and associated factors among men of different ages. Journal of Gerontology, 41, 710-717. Kane, R L., Matthias, R., & S. (1983). The risk of placement in a nursing home after acute hospitalization. Medical Care, 21, 1055-1061. B. H., Cassel, J. C., & Gore, S. (I 977). Social support and health. Medical Care, 15,47-58. G., Barell, V., & Lusky, A (1988). Subjective state of health and survival in elderly adults. Journal of Gerontology, 43, S 114-S 120. G. A., & Camacho, T (1983). Perceived health and mortality: A nine year follow-up of the human population laboratory cohort. American Journal of Epidemiology, 117, 292-304. G. A., Seeman, T. E., Cohen, R. D., L. P., & Guralnik, J. (1987). Mortality among the elderly in the Alameda county study: Behavioral and demographic risk factors American Journal of Public Health, 77, 307-312. G. A., Strawbridge, W. J., Camacho, T., & RD. (1993). Factors associated with change in physical functioning in the elderly: A six-year prospective study. Journal of Aging and Health, 5, 140-153. G. A Wilson, T. W Cohen, R. D., Kauhanen, J., Wu, M., & J. T. (1994). Social functioning and overall mortality: Prospective evidence from the Kuopio ischemic heart disease risk factor study. Epidemiology, 5, 495-500. 209

PAGE 220

: I I I Kaplan, S. H. (1987). Patient reports ofhealth status as predictors of physiological health measures in chronic disease. Journal of Chronic Diseases, 40(Suppl. 1), 27S-35S. Kasl, S. V. (1983). Social and psychological factors affecting the course of disease: An epidemiological perspective. In D. Mechanic (Ed.). Handbook of health. health care, and the health professions (pp. 683-708). New York: The Free Press. Katz, S., Ford, A. B., Moskowitz, R. W., Jackson, B. A., & Jaffee, M. W. (1963). Studies of illness in the aged: The index of ADL: A standardized measure of biological and psychosocial function. Journal of the American Medical Association, 185(12), 914-919. Keil, J. E., Gazes, P. C., Sutherland, S. E., Rust, P. F., Branch, L. G., & Tyroler, H. A. (1989). Predictors of physical disability in elderly blacks and whites of the Charleston heart study. Journal of Clinical Epidemiology, 42, 521-529. Kind, P., & Dolan, P. (1995). The effect of past and present illness experience on the valuations of health status. Medical Care, 33(Suppl. 4), AS255-AS263. Kleinman, A. (1988). The illness narratives: Suffering, healing and the human condition. Basic Books. Kraus, A. S., Spasoff: R. A., Beattie, E. J., Holden, E. W., Lawson, J. S., Rodenburg, M., & Woodcock, G. M. (1976). Elderly applicants to long-term care institutions; I: Their characteristics, health problems and state of mind. Journal of the American Geriatrics Society, 24, 117-125. Krause, N. (1995). Negative interaction and satisfaction with social support among older adults. Journals of Gerontology, 50B, P59-P73. Kutner N. G. (1987). Social ties, social support, and perceived health status among chronically disabled people. Social Science and Medicine, 25, 29-34. Kuzel, A. J. (1992). Sampling in qualitative inquiry. In B. F. Crabtree & W. L. Miller (Eds. ), Doing qualitative research (pp. 31-44). Newbury Park, CA: Sage Publications. Lammi, U.-K., Kivela, S.-L., Nissinen, A., Punsar, S., Puska, P., & Karvonen, M. (1989). Predictors of disability in elderly Finnish men-a longitudinal study. Journal of Clinical Epidemiology. 42, 1215-1225. 210

PAGE 221

, I I : I I LaRue, A., Bank, L., L., & Hetland, M. (1979). Health in old age: How do physicians' ratings and self-ratings compare? Journal of Gerontology, 34,687-691. Lawton, M.P. (1983). The varieties of well-being. Erperimental Aging Research, 9(2), 65-72. Levkoff, S. E., Cleary, P D., & Wetle T. (1987). Differences in the appraisal ofhealth between aged and middle-aged adults. Journal of Gerontology, 42, 114-120. Lewis, A. (1953). Health as a social concept. British Journal of Sociology, 4, 109-124. Liang, J. (1986). Self-reported physical health among aged adults. Journal of Gerontology, 41, 248-260. Lindgren, A.-M., Svardsudd, K., & Tibblin, G. (1994). Factors related to perceived health among elderly people: The Albertina project. Age and Ageing, 23, 328-333. Link, B. G., & Phelan, J. C. (1996). Understanding sociodemographic differences in health-the role of fundamental social causes [Editorial]. American Journal of Public Health, 86,471-473. Linn, B.S., & Linn, M. W. (1980). Objective and self-assessed health in the old and very old. Social Science and Medicine, 14A, 311-315. Liu, K., Coughlin, T., & McBride, T. (1991). Predicting nursing home admissions and length of stay: A duration analysis. Medical Care, 29, 125-141. Lohr, K. N. (1988). Outcomes measurement: Concepts and questions. Inquiry, 25,37-50. Maddox, G. L. (1962). Some correlates of differences in self-assessment of health status among the elderly. Journal of Gerontology, 17, 180-185. Manton, K. G. (1988). A longitudinal study of functional change and mortality in the United States. Journal ofGerontology, 43, S153-S161. Manton, K. G. (1989). Epidemiological, demographic, and social correlates of disability among the elderly. Milbank Quarterly, 67(Suppl. 2), 13-58. 211

PAGE 222

Manton, K. G., & Soldo, B. J. (1985). Dynamics ofhealth changes in the oldest old: New perspectives and evidence. Milbank Quarterly, 63, 206-285. Markides, K. S., & Lee, D. J. (1990). Predictors of well-being and functioning in older Mexican Americans and Anglos: An eight-year follow-up. Journal of Gerontology, 45, S69-S73. Marlcides, K. S., & Martin, H. W. (1979). Predicting self-related health among the aged. Research on Aging, 1, 97-112. Marlcides, K. S., & Pappas, C. (1982). Subjective age, health, and survivorship in old age. Research on Aging, 4, 87-96. Maslow, A. H. (1970). Motivation and personality. New York: Harper and Row. McAuley, W. J., & Prohaska, T. (1981). Professional recommendations for a long-term care placement: A comparison of two groups of institutionally vulnerable elderly. Home Health Care Services Quarterly, 2, 41-57. McCoy, J. L., & Edwards, B. E. (1981). Contextual and sociodemographic antecedents of institutionalization among aged welfare recipients. Medical Care, 19, 907-921. McFall, S., & Miller, B. H. (1992). Caregiver burden and nursing home admission of frail elderly persons. Journal of Gerontology, 47, S73-S79. Mechanic, D. (1972). Social psychologic factors affecting the presentation of bodily complaints. New England Journal of Medicine, 286, 1132-1139. Mechanic, D. (1978). Medical sociology (2nd ed.). New York: The Free Press. Mechanic, D. (1986). The concept of illness behavior: Culture, situation, and personal predispositions. Psychological Medicine, 16, 1-7. Mechanic, D. (1995). Sociological dimensions of illness behavior. Social Science and Medicine, 41, 1207-1216. Minkler, M., & Langhauser, C. (1988). Assessing health differences in an elderly population: A five-year follow up. Journal of the American Geriatrics Society, 36, 113-118. 212

PAGE 223

. I Moore, L. G., VanArsdale, P. W., GUttenberg, J. E., & Aldrich, R. A. (1980). The biocultwal basis of health; expanding views of medical anthropology. Prospect Heights, IL: Waveland Press, Inc. Mor, V ., Murphy, J, Masterson-Allen, S., Willey, C., Razmpour, A., Jackson, M. E., Greer, D,. & Katz, S. (1989). Risk of functional decline among well elders Journal of Clinical Epidemiology, 42, 895-904 Mor, V., Wilcox, V., Rakowski, W., & Hiris, J. (1994). Functional transitions among the elderly: Patterns, predictors, and related hospital use. American Journal of Public Health, 84, 1274-1280. Mor-Barak. M., Scharlach, A. E., Birba, L., & Sokolov, J. (1992). Employment, social networks, and health in the retirement years. International Journal of Aging and Human Development, 35, 145-159. Moriyama, I. M. (1968). Problems in the measurement of health status. In E B. Sheldon & W. E. Moore (Eds.), Indicators of social change (pp. 573-600). New York: Russell Sage Foundation. Mossey, J. M., & Shapiro, E. (1982). Self-rated health: A predictor of mortality among the elderly. American Journal of Public Health, 72, 800-808. Motzer, S. U., & Stewart, B. J. (1996). Sense of coherence as a predictor of quality of life in persons with coronary heart disease surviving cardiac arrest. Research in Nursing and Health, 19,287-298 Mmray, J., Dunn, G., & Tamopolsky, A. (1982). Self-assessment of health: An exploration of the effects of physical and psychological symptoms. Psychological Medicine, 12,371-378. Murtaugh, C. M., Kemper, P., & SpiiJman, B. C. (1990). The risk of nursing home use in later life. Medical Care, 28, 952-962. Nagi, S. Z. (1976). An epidemiology of disability among adults in the United States. Milbank Quarterly, 54, 439-468. Narain, P., Rubenstein, L. Z., Wieland, G. D., Rosbrook, B., Strome, L. S., Pietruszka, F., & Morley, J. E. (1988). Predictors of immediate and 6-month outcomes in hospitalized elderly patients: The importance of functional status. Journal ofthe American Geriatrics Society, 36,775-783 213

PAGE 224

S. J., Struyk, R, Wright, P., & Rice, M. (1990). Overwhelming odds: Caregiving and the risk of institutionalization. Journal of Gerontology, 45, SI73-Sl83. Newsom, J. T., & Schulz. R. (1996). Social support as a mediator in the relation between functional status and quality of life in older adults. Psychology and : 1 Aging. 11, 34-44. Nocks, B. C., Learner, R. M., Blackman, D., & T. E. (I 986). The effects of a community-based long term care project on nursing home utilization. Gerontologist, 26, 150-157. Orth-Gomer, K., & Johnson, J. V. (1987). Social network interaction and mortality. A six year follow-up study of a random sample of the Swedish population. Journal ofChronic Diseases, 40,949-951. Palmore, E. (1976). Total chance of institutionalization among the aged. Gerontologist, 16,504-501. Palmore, E. B., J. B., & Wang, H. S. (1985). Predictors of function among the old-old: A 10-year follow-up. Journal of Gerontology, 40,244-250. Parker, M. G., Thorslund, M., & Nordstrom, M.-L. (1992). Predictors of mortality for the oldest-old: A 4-year follow-up of community-based elderly in Sweden. Archives of Gerontology and Geriatrics, 14, 227-23 7. Parkerson, G. R, Jr., Michener, J. L., Wu, L. R., Finch, J. N., Muhlbaier, L. H., Magruder-Habib, K., Kertesz, J. W., Clapp-Channing, N., Morrow, D. S., Chen, A. L.-T., & Jokerst, E. (1989). Associations among family support, family stress, and personal functional health status. Journal of Clinical Epidemiology, 42, 217-229. Parsons, T. (1951). The social system. New York: Free Press. Patrick, D. L., Bush, J. W., & Chen, M. M. (1973). Toward an operational definition of health. Journal of Health and Social Behavior, 14,6-23. Patton, M. Q. (1990). Qualitative evaluation and research methods (2nd ed.). Newbury Park. CA: Sage Publications. Pearlin, L. I. (1989). The sociological study of stress. Journal of Health and Social Behavior, 30, 241-256. 214

PAGE 225

I i I I Pearlman, D. N., & W. H. (1992). Alternative sources of social support and their impacts on institutional risk.. Gerontologist, 32, 527-535. Peterson, C., & Seligman, M. E. P. (1987). Explanatory style and illness. Jownal of Personality, 55,237-265. Peterson, C., Seligman, M. E. P., & Vaillan4 G. E. (1988). Pessimistic explanatory style is a risk factor for physical illness: A thirty-five-year longitudinal study. Journal of Personality and Social Psychology, 55, 23-27. : I Pfeiffer, E. (1970). Survival in old age: Physical, psychological and social ; 1 correlates of longevity. Journal of the American Geriatrics Society, 18,273-285. I Pijls, L. T. J., Feskens, E. J. M., & Kromhout, D. (1993). Self-rated health, mortality, and chronic diseases in elderly men: The Zutphen study, 1985-1990. American Journal of Epidemiology, I 38, 840-848. Pilpel, D., Carmel, S., & Galinsky, D. (1988). Self-rated health among the elderly; a comparative analysis of health status measures, leisure activities and social contacts in age/sex groups. Comprehensive Gerontolology B, 2, 110-116. Pinsky, J. L., Branch, L. G., Jette, A.M., Haynes, S. G., Feinleib, M., Comoni-Huntley, J. C., & Bailey, K. R. (1985). Framingham disability study: Relationship of disability to cardiovascular risk factors among persons free of diagnosed cardiovascular disease. American Journal of Epidemiology, I 22, 644-656. Pinsky, J. L., Leaverton, P. E., & Stokes, J., ill. (1987). Predictors of good function: The Framingham study. Journal of Chronic Diseases, 40(suppl. 1), 159S167S. Quayhagen, M. P ., & Quayhagen, M. (1996). Discovering life quality in coping with dementia. Western Journal of Nursing Research, 18, 120-135. Rakowski, W., & Cyran, C. D. (1990). Associations among health perceptions and health status within three age groups. Journal of Aging and Health 2, 58-80. Rakowski, W., Mor, V., & Hiris, J. (1991). The association of self-rated health with two-year mortality in a sample of well elderly. Journal of Aging and Health, 3, 527-545. 215

PAGE 226

: I Reuben, D. B., Siu, A. L., & Kimpau, S. (1992). The predictive validity of self-report and performance-based measures of function and health. Journal of Gerontology, 47, M106-M110. Rickelm.an, B. L., Gallman, L., & Parra, H. (1994). Attachment and quality of life in older, community-residing men. Nursing Research. 43,68-72. Robertson, A., & Minkler, M. (1994). New health promotion movement: A critical examination. Health Education Quarterly, 21, 295-312. Rockwood, K., Stolee, P ., & McDowell, I. (1996). Factors associated with institutionalization of older people in Canada Testing a multifactorial definition of frailty. Journal of the American Geriatrics Society, 44,578-582. Rodin, I. ( 1986). Aging and health: Effects of the sense of control. Science, 233, 1271-1276. Rodin, J., & McAvay, G. (1992). Determinants of change in perceived health in a longitudinal study of older adults. Journal of Gerontology, 46, P373-P384. Rodin, J., Timko, C., & Harris, S. (1985). The construct of control: Biological and psychosocial correlates. Annual Review of Gerontology and Geriatrics, 5, 3-55. Roos, N., & Havens, B. (1991). Predictors of successful aging: A twelve-year study of Manitoba elderly. American Journal of Public Health. 81, 63-68. Rosow, I. (1974). Socialization to old age. Berkeley, CA: University of California Press. Rosow, 1., & Breslau, N. (1966). Guttman health scale for the aged. Journal of Gerontology, 21, 556-559. Rowe, J. W., & Kahn, R. L. (1987). Human aging: Usual and successful. Science, 237, 143-149. Sager, M.A., Rudberg, M.A., Jaluddin, M., Franke, T., Inouye, S. K., Landefeld, C.S., Siebens, H., & Winograd, C. H. (1996). Hospital admission risk profile (HARP): Identifying older patients at risk for functional decline following acute medical illness and hospitalization. Journal of the American Geriatrics Society, 44, 251-257. 216

PAGE 227

i I I I I :I Schoenbach, V. J., Kaplan, B. H., Fredman, L., & Kleinbaum, D. G. (1986). Social ties and mortality in Evans county, Georgia American Journal of Epidemiology, 123,577-591. Schoenfeld, D. E., Malmrose, L. C., Blazer, D. G., Gold, D. T., & Seeman. T. E. (1994). Self-rated health and mortality in the high-functioning elderly: A closer look at healthy individuals: MacArthm field study of successful aging. Journal of Gerontology, 49, M109-M115. Schulz, R, & G. M. (1993). Psychosocial and behavioral dimensions of physical frailty. Journal of Gerontology, 48(special issue), 39-43. Schwartz, J. S., & Lmie, N. (1990). Assessment of medical outcomes: New opportunities for achieving a long sought-after objective. International Journal of Technology Assessment in Health Care, 6, 333-339. Seeman, T. E., Berkman, L. F., Charpentier, P., Blazer, D., Albert, M., & Tinetti, M. (1995). Behavioral and psychosocial predictors of physical performance: MacArthur studies of successful aging. Journal of Gerontology, 50A, M177-M183. Seeman, T. E., Charpentier, P., L., Tinetti, M. E., Guralnik, J. M., Albert, M., Blazer, D., & Rowe, J. W. (1994). Predicting changes in physical performance in a high fimctioning elderly cohort: MacArthm studies of successful aging. Jow-nal of Gerontology, 49, M97-M108. Seeman, T. E., G. A., L., R, & J. (1987). Social network ties and mortality among the elderly in the Alameda county study. American Journal of Epidemiology, 126,714-723 Shapiro, E., & Tate, R B. (1985). Predictors oflong term care facility use among the elderly. Canadian Journal on Aging, 4, 11-18. Shapiro, E., & Tate, R B. (1988). Who is really at risk of institutionalization? Gerontologist, 28, 237-245. Smyer, M. (1980). The differential usage of services by impaired elderly. Journal of Gerontology, 35,249-255. K. H. (1988). State of health and its association with death among old people at three-years follow-up. Danish Medical Bulletin, 35, 597-600. 217

PAGE 228

I I Speare, A., Avery, R., & Lawton, L. (1991). Disability, residential mobility, and changes in living arrangements. Journal of Gerontology, 46, Sl33-Sl42. Starr, P. (1982). The social transformation of American medicine. New York: Basic Books. U. (1992). Social networks, institutionalization, and mortality among elderly people in the United States. Journal of Gerontology, 47, S183-S190. Stewart, A. L., S., Hays, R. D., Wells, K., Rogers, W. H., Berry, S. D., McGlynn, E. A., & Ware, J. E., Jr. (1989). Functional status and well-being of patients with chronic conditions: Results from the medical outcomes study. Journal of the American Medical Association, 262,907-913. Stewart, M. J. ( 1989). Social support: Diverse theoretical perspectives. Social Science and Medicine, 28, 1275-1282. Strain, L. A. (1993). Good health-what does it mean in later life? Journal of Aging and Health, 5, 338-364 Strauss, A., & Corbin, 1. (1990). Basics of qualitative research: Grounded theory procedures and techniques. Newbury Park, CA: Sage Publications. Strawbridge, W. J., Cohen, R. D., S. J., & Kaplan, G. A. (1996). Successful aging: Predictors and associated activities. American Journal of Epidemiology, 144, 135-141. Stuck, A. E., Siu, A. L., Wieland, D., Adams, J., & Rubenstein, L. Z. (1993). Comprehensive geriatric assessment: A meta-analysis of controlled trials. Lancet, 342, 1032-1036. Sugisawa, H., Liang, J., & Liu, X. (1994). Social networks, social support, and mortality among older people in Japan. Journal of Gerontology. 49, S3-Sl3. Suls, J., Marco, C. A., & Tobin, S. (1991). The role of temporal comparison, social comparison, and direct appraisal in the elderly's self-evaluations of health. Journal of Applied Social Psychology, 21, 1125-1144. Susser, M. (1974). Ethical components in the definition of health. International Journal of Health Services, 4, 539-548. 218

PAGE 229

i I J., Holmes, Bergman. King, Y., & Bentur, N. (1989). Factors relating to institutional risk among elderly members oflsraeli kibbutzim. Gerontologist, 203-208. S. N. (1988). Hidden arguments: Political ideology and disease prevention policy. New Brunswick, NJ: Rutgers University Press. & Mechanic, D. (1978). Psychological distress and perceived health status. Journal of Health and Social Behavior, 19,254-262. Thomas, L. B., & Chambers, K. 0. (1989). "Successful aging" among elderly men in England and India: A phenomenological comparison In L. E. Thomas (Ed.), Research on adulthood and aging: The human science approach (pp. 183-203). Albany, NY: State University ofNew York Press. Tissue, T (1972). Another look at self-rated health among the elderly. Journal ofGerontology, 27,91-94. Tomstam, L. (1975). Health and self-perception: A systems theoretical approach. Gerontologist, 15, 264-270. Tsuji, I., Whalen, S., & Finucane, T. E. (1995). Predictors of nursing home placement in community-based long-term care. Journal of the American Geriatrics Society, 43, 761-766. Turner, B.S. (1992). Regulating bodies: Essays in medical sociology. London: Routledge. : 1 U.S. Administration on Aging. (1998, July 30). Profile of older Americans: 1997 [On-line]. Available: http//pr.aoadhhs.gov/aoa/stats/profile U.S Bureau of the Census. (1998, June 25). Resident population of the United States: Estimates, by age and sex [On-line]. Available: http://www.census gov/populationlestimates/nationlintfile2-1.txt U.S. Health Care Financing Administration. (1998, September 9). National health expenditures aggregate amounts and average annual percent change, by type of expenditure: Selected calendar years 1960-96 [On-line]. Available: http://www .hcfa.gov/stats/nhe-oact/tables/t1 O.htm Verbrugge, L. M., & Balaban, D. J. (1989). Patterns of change in disability and well-being. Medical Care, 27, S128-S147. 219

PAGE 230

Verbrugge, L. M., Reoma, J. M., & Gruber-Baldini, A. L. (1994). Short-term dynamics of disability and well-being. Journal of Health and Social Behavior. 35, 97117. Vicente, L., Wiley, J. A., & Carrington, R. A. (1979). The risk of institutionalization before death. Gerontologist, 19, 361-367. Vogt, T. M., Mullooly, J.P., Ernst, D., Pope, C. R., & Hollis, J. F. (1992). Social networks as predictors of ischemic heart disease, cancer, stroke, and 1 hypertension: Incidence, survival, and mortality. Journal of Clinical Epidemiology. 45, 659-666. Wachtel, T. J., Derby, C., & Fulton, J.P. (1984). Predicting the outcome of hospitalization for elderly persons: Home versus nursing home. Southern Medical Journal, 77, 1283-1285. Wallace, R. B. (1994). Assessing the health of individuals and populations in surveys of the elderly: Some concepts and approaches. Gerontologist, 34,449-453. Wan, T. T. H. (1976). Predicting self-assessed health status: A multivariate approach. HSR:Health Services Research. 11, 464-4 77. Wan, T. T H. (1986). Evaluation research in long-term care: Scope and methodology. Research on Aging, 8, 559-585. Wan, T. T. H., & Weissert, W. G. (1981). Social support networks, patient status, and institutionalization. Research on Aging, 3, 240-256. Ware, J. E., Jr. (1976, winter). Scales for measuring general health perceptions. HSR:Hea/th Services Research, 396-413. Weinberger, M., Darnell, J. C., Tierney, W. M., Martz, B. L., Hiner, S. L, Barker, J., & Neill, P. J. (1986). Self-rated health as a predictor of hospital admission and nursing home placement in elderly public housing tenants. American Journal of Public Health, 76, 457-459. Welin, L., Tibblin, G., Svardsudd, K., Tibblin, B., Ander-Peciva, S., Larsson, B., & Wilhelmsen, L. (1985). Prospective study of social influences on mortality: The study of men born in 1913 and 1923. Lancet, 915-918. 220

PAGE 231

Wilson, LB., & Cleary, P. D. (1995). Linking clinical variables with health related quality of life: A conceptual model of patient outcomes. Journal of the American Medical Association, 273,59-65. Wingard, D. L., Williams-Jones, D., McPhillips, J., Kaplan, R. M., & Sarrett Connor, E. (1990). Nursing home utilization in adults: A prospective population based study. Journal of Aging and Health. 2, 179-193. Winograd, C. H., Gerety, M. B., Chung, M., Goldstein, M. K., Dominguez, F., Jr., & Vallone, R. (1991 ). Screening for frailty: Criteria and predictors of outcomes. Journal of the American Geriatrics Society. 3 9, 778-784. Wolinsky, F. D., Callahan, C. M., Fitzgerald, J. F., & Johnson, R. J. (1992). The risk of musing home placement and subsequent death among older adults. Journal ofGerontology. 47, S173-S182. Wolinsky, F. D., & Johnson, R. J. (1991). The use ofhealth services by older adults. Journal of Gerontology. 46, S345-S357. Wolinsky, F. D., & Johnson, R. J. (1992). Perceived health status and mortality among older men and women. Journal of Gerontology. 47, S304-S312. Wolinsky, F. D., Johnson, R. L., & Stump, T. E. (1995). The risk of mortality among older adults over an eight-year period. Gerontologist, 35, 150-161. Wolinsky, F. D., & Stump, T. E. (1996). Age and the sense of control among older adults. Journal of Gerontology, SIB, S217-S220. Young, J. E., Forbes, W. F., & Hirdes, J.P. (1994). The association of disability with long-term care institutionalization of the elderly. Canadian Journal on Aging, 13, 15-29. Zborowski, M. (1952). Cultural components in response to pain. Journal of Social Issues, 8, 16-31. Zonderman, A. B., Leu, V. L., & Costa, P. T., Jr. (1986). Effects of age, hypertension history, and neuroticism on health perceptions. Experimental Gerontology. 21, 449-458. Zuckerman, D. M., Kasl, S. V & Ostfeld, A.M. (1984). Psychosocial predictors of mortality among the elderly poor: The role of religion, well-being, and social contacts. American Journal of Epidemiology, 119, 410-423. 221