Citation
Primary care providers' approaches to mental health care for socially and economically disadvantaged adults with chronic disease

Material Information

Title:
Primary care providers' approaches to mental health care for socially and economically disadvantaged adults with chronic disease
Creator:
Darr, Carol Ann
Publication Date:
Language:
English
Physical Description:
xi, 228 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Chronically ill -- Mental health services ( lcsh )
People with social disabilities -- Mental health services ( lcsh )
Primary health care ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 215-228).
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by Carol Ann Darr.

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:
66460890 ( OCLC )
ocm66460890
Classification:
LD1193.L566 2005d D37 ( lcc )

Full Text
PRIMARY CARE PROVIDERS APPROACHES TO MENTAL HEALTH CARE
FOR SOCIALLY AND ECONOMICALLY DISADVANTAGED ADULTS
WITH CHRONIC DISEASE
by
Carol Ann Darr
B.A., Colorado State University, 1983
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
for the requirements for the degree of
Doctor of Philosophy
Health and Behavioral Sciences
2005


This thesis for the Doctor of Philosophy
degree by
Carol Ann Darr
has been approved
by
Tillman Farley
Date


Darr, Carol Ann (Ph.D., Health and Behavioral Sciences)
Primary Care Providers Approaches to Mental Health Care
For Socially and Economically Disadvantaged Adults with Chronic Disease
Thesis directed by Assistant Professor John A. Brett
ABSTRACT
This studys purpose was to describe primary care providers (PCPs) approaches
to mental health care for socially and economically disadvantaged adults with chronic
disease who receive their care in community health centers (CHC), and to identify
factors that influence those approaches.
Because mental illness increases symptom burden, functional impairment, and
medical costs, and impedes self care in people with chronic disease, management of
mental disorders constitutes an important component of their care. Challenges in
providing integrated physical and mental health care are magnified for PCPs whose
patients are from culturally diverse backgrounds, live in poverty, and tend to have
poorer physical and mental health than the general population. This study sought to
provide better insight into current mental health care practices in these settings as a
means to help identify opportunities for improvement.
This qualitative study was conducted in seven clinics representing three CHC
systems in the Denver, Colorado metropolitan area. Interviews with 48 PCPs explored
recognition and responses to patient psychosocial issues in the context of chronic
disease management. Observation of clinic activity and interviews with 12 non-PCP
clinic staff providing psychosocial services in the three systems provided additional
insight into influences on the process of care.
A systematic analysis of PCPs descriptions of their thoughts and actions in
specific patient situations revealed the practice of reluctant psychiatry a PCPs
approach to mental health care when the patients needs are exceptionally complex and
m


the PCP is unable to refer the patient for specialty care. As specialty resources for low-
income individuals with serious mental illness have become less and less accessible to
patients and their PCPs due to cuts in public mental health spending, PCPs face the
dilemma of providing care that is outside their scope of expertise and/or struggling to
manage chronic disease in patients who are otherwise unable to engage in self care.
Reluctant psychiatry exacts a toll on the individuals forced to practice it, the
patients whose care is compromised for lack of appropriate specialty care, and the
society that absorbs the cost of supporting increasing numbers of individuals whose
physical and mental health are poorly controlled.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
!
i
IV


DEDICATION
This dissertation is dedicated to Danny Darr, with whom I have shared a fantastic
journey for most of my life, and who has unselfishly supported me in realizing this and
all my dreams, and to my beautiful and accomplished daughters, Kristin and Keryn,
who inspire me to be all that I can be.
I dedicate this as well to the many, fine health care professionals working in community
health centers, who care for those who are most in need. I am humbled by the
magnitude of their service to others and inspired to use my newly acquired knowledge
and skills in a way that makes a difference.


ACKNOWLEDGEMENTS
I wish to acknowledge my academic and clinical advisors, the agencies that provided
financial support for this research, and the community health centers that participated in
this study.
Dr. John Brett patiently and wisely guided me in every aspect of this endeavor, and his
unswerving belief in the value of this research and in my ability to succeed sustained me
throughout this long process. He is everything that one could wish for in a committee
chair and will remain a trusted advisor. Dr. Kitty Corbett held me to the highest
standards of scholarship and provided research training opportunities that uniquely
contributed to my education. Dr. Tillman Farley and Dr. Shale Wong kept me anchored
in the realities of clinical practice and provided invaluable insights into the intricacies of
patient care. Dr. Farley also took time from his many clinical, administrative and
academic responsibilities to act as mentor for my pre-doctoral training.
Others who unselfishly contributed to this learning experience include Dr. Lucinda
Bryant and Dr. Sheana Bull, whose wise counsel and expertise guided me at many steps
along the way. Dr. Stephen Koester enthusiastically supported my work, especially in its
formative stages. Finally, my growth as a scholar was enriched by my fellow students,
with whom I shared this experience and from whom I have learned so much. I am
grateful for all these contributions to my education and to my development as a
researcher.
I also wish to acknowledge funding support for this research by the National Institutes
of Health under the Ruth L. Kirschstein National Research Service Award #F31
MH070978 from the National Institute of Mental Health, and by the Agency for
Healthcare Research and Quality under Health Services Dissertation Research Grant
#R36 HS014153.
Finally, this project would not have happened without the participation of the
community health centers in which I conducted my research, and I am grateful to
Carolyn Shepherd, M.D. of Clinica Campesina Family Health Services, Thomas
MacKenzie, M.D. of Denver Health, and Tillman Farley, M.D. of Salud Family Health
Centers for welcoming me into their clinics. I also wish to acknowledge the many
clinicians and other professionals who took time from their important work to
participate in this study. In particular, I wish to acknowledge Brian Boley, PA-C; Jennie
Bravo, PA-C; Don Gutstadt, PA-C; Rebecca Hanratty, M.D.; Mary Karner, NP; Terri
Richardson, M.D.; Jeanne Rozwadowski, M.D.; and David Wu, M.D, for generously
facilitating and promoting my work within their climes.


CONTENTS
Figures ..................................................................x
Tables ...................................................................xi
CHAPTER
1. INTRODUCTION..........................................................1
Study Purpose and Rationale........................................2
Overview of Research Methods.......................................3
Study Setting and Participants.................................3
Terminology........................................................4
Usage and Conventions..........................................7
Structure of the Dissertation......................................8
2. BACKGROUND............................................................9
Primary Care and Integrated Care..................................14
Chronic Disease and Mental Health Care for Disadvantaged Populations.... 16
Non-recognition of Mental Disorders in Primary Care Settings:
Issues, Evidence, and Factors.....................................20
Theoretical and Philosophical Assumptions.........................30
Constructivism................................................30
"Practice Ecosystem": The Context of Care.....................32
The Social Context of Mental Health and Mental Health Care....35
Competing Views of Mental Health and Illness..................38
Clinical Explanatory Models...................................39
Clinical Decision Making......................................41
Researcher's Presuppositions..................................43
3. METHODS..............................................................46
Research Design...................................................47
Sampling Strategy, Study Sites, and Participants..............48
Research Phases...............................................51
Design for Trustworthiness....................................52
Data Collection Techniques........................................54
Qualitative Analysis..............................................58
Verification......................................................67
Limitations.......................................................68
vii


4. PCPS APPROACHES TO MENTAL HEALTH CARE FOR SOCIALLY
AND ECONOMICALLY DISADVANTAGED ADULTS WITH
CHRONIC DISEASE.....................................................70
Characterizing PCPs' Approaches to Mental Health Care...........71
Describing the Process of Recognition........................74
Describing Responses to Mental Health Needs..................86
Types of Approaches..........................................93
Influences on PCPs' Approaches to Mental Health Care............95
What Influences Recognition?.................................95
What Influences Response?...................................103
Why One Approach versus Another?...............................117
Where Does It Lead?............................................121
Qualitative Outcomes........................................121
Fears, Doubts, Misgivings...................................124
Coping......................................................127
Insights Into PCPs' Approaches to Mental Health Care in CHCs...129
5. THE PRACTICE OF RELUCTANT PSYCHIATRY IN COLORADO
COMMUNITY HEALTH CENTERS...........................................131
The Practice Ecosystem for CHCs in Colorado....................131
Colorados Mental Health Care System and the Policies that Created and
Sustain It.....................................................133
Implications to Community Health Centers....................138
Practice Setting Resources for Mental Health Care..............145
Local Communities: Patients Served and Community Resources.....150
Characteristics and Mental Health Needs of CHC Patients.....150
Community Resources Available to PCPs and Their Patients....156
PCPs: The Mental Health Care Comfort Zone and Expanding
the Scope of Practice.......................................157
Challenges PCPs Face in Providing Mental Health Care in
Colorado's CHCs................................................163
Recognition Challenges......................................164
Challenges in Managing Mental Health........................175
6. DISCUSSION.........................................................184
What These Findings Contribute to the Current Base of Knowledge.185
How PCP Characteristics Influence Provision
of Mental Health Care.......................................186
How Characteristics of the Practice Setting
Influence Mental Health Care................................190
Expanding the View: The Practice Ecosystem
and Integrated Care.........................................192
viii


Unexpected Findings.......................................195
Implications of This Research to Theory and Methodology......199
A Blending of Theoretical Perspectives.............199
Assessing Methodology.....................................200
Conclusions..................................................201
APPENDIXES
A. Glossary of Terms...............................................204
B. Interview Guides................................................212
REFERENCES.............................................................215


FIGURES
Figure
2- 1 Ecosystem of Primary Care Practice......................................34
3- 1 Research Design.........................................................48
3-2 Data Collection Techniques..............................................55
3-3 Approach to Analyzing Study Data........................................59
3- 4 Analytic Tool for Characterizing PCPs Approaches to Mental Health Care.63
4- 1 Analytic Model for Characterizing PCPs Approaches to Mental Health Care....74
4-2 Behaviors Descriptive of PCPs Approaches to Mental Health Care.........94
4- 3 Zones of Operation Used to Describe PCPs Approaches to Mental
Health Care...........................................................117
5- 1 The Ecosystem of Primary Care Practice.................................132
x


TABLES
Table
2- 1 Factors Influencing Recognition of Mental Disorders in Primary Care............28
3- 1 Characteristics of PCPs........................................................50
3-2 Participation by Clinic System.................................................51
3- 3 Situation Summaries............................................................64
4- 1 Summary of Cues Signaling Potential Psychosocial Issues.......................80


CHAPTER 1
INTRODUCTION
The 1999 U.S. Surgeon Generals report on mental health confirmed that
mental disorders are real, disabling and treatable health conditions and underscored the
importance and urgency of treating and preventing mental disorders and of promoting
mental health in our society (U.S. Department of Health and Human Services, 1999).
Despite this imperative, a large proportion of mental disorders continue to be
unrecognized and/or untreated in the U.S. population (Hirschfeld et al., 1997; Kessler
et al., 2005). Only one-third to one-half of patients with mental disorders are correcdy
diagnosed in primary care settings (Callahan et al., 1997; deGruy, 1996), the sole source
of care for most people with mental health problems or disorders (Borowsky et al.,
2000; Callahan et al., 1997; Docherty, 1997; Nutting, Rost, Smith, Werner, & Elliot,
2000; Wang, Berflund, & Kessler, 2000), and most patients with mental disorders do
not receive treatment. Recognition and management of mental disorders takes on added
importance in the context of physical illness, especially chronic disease (Black, 1999;
Chapman, Perry, & Strine, 2005; Davidson & Melzer-Brody, 1999; Garfield, 2002).
Growing evidence shows that mental illness increases symptom burden, functional
impairment and medical costs, and impedes self-management efforts in people with
chronic disease (Chapman et al., 2005; Katon & Ciechanowski, 2002; Rubin & Peyrot,
2001). Much remains to be understood, however, about how to optimally manage the
co-occurrence of physical and mental health issues in primary care settings (deGruy,
1996; Von Korff, Katon, Uniitzer, Wells, & Wagner, 2001).
The co-existence of physical and psychological signs and symptoms often
confounds recognition of mental disorders (Craven, Cohen, Campbell, Williams, &
Kates, 1997; Docherty, 1997; Goldman, Nielsen, & Champion, 1999; Jacobson, 1996;
Rost et al., 2000), and both recognition and associated diagnostic decisions are further
1


complicated by sociocultural, economic and environmental factors unique to people
from culturally diverse backgrounds and to those living in poverty (U.S. Department of
Health and Human Services, 2001). Providing integrated physical and mental health care
is thus especially challenging for primary care providers caring for individuals who are
socially and economically disadvantaged and who tend to have poorer physical and
mental health than the general population (Bruce, Takeuchi, & Leaf, 1991; Cameron &
Mauksch, 2002; Everson, Maty, Lynch, & Kaplan, 2002; Fiscella, 2002; Geronimus,
2000; Norquist & Wells, 1991; Waitzkin, 2000).
Study Purpose and Rationale
Evidence of high rates of non-recognition of mental disorders in primary care
settings and that socially and economically disadvantaged patients are at greatest risk for
non-recognition and inappropriate treatment (Borowsky et al., 2000) prompted the
question that framed this study: Why do mental disorders among socially and economically
disadvantaged adults often go unrecognised or undiagnosed in primary care settings? To begin to
address this question, the study focused on primary care providers (PCPs) recognition,
interpretation and responses to patient psychosocial issues in the context of chronic
disease management.
The studys goal was to better understand PCPs approaches to providing
mental health care for adult patients with chronic disease who receive their care in
community health centers (CHCs), a key source of primary care for low-income and
culturally diverse populations (Lewin & Altman, 2000). The specific aims for
accomplishing the study goal were to:
Aim 1 Describe how CHC providers frame the psychosocial concerns of and structure
the provision of mental health care for patients with chronic disease
1.1 Explore CHC providers attitudes and beliefs about and approaches to
psychosocial aspects of chronic disease management
2


1.2 Identify criteria CHC providers use to distinguish treatable mental
disorders from other psychosocial issues affecting patients with chronic
disease
1.3 Discover the cues that prompt recognition and diagnosis of mental
disorders in chronically ill, socially and economically disadvantaged
patients
Aim 2 Identify factors that influence PCPs beliefs, attitudes and clinical decisions in
providing mental health care to patients with chronic disease in CHCs
Overview of Research Methods
This qualitative study used a combination of different types of interviews and
observation in CHC clime settings to address the study aims. Semi-structured interviews
explored PCPs beliefs about mental health care and about their roles as providers of
that care, as well as decisions and actions they take in situations in which a patients
psychosocial issues and physical health issues both demand attention. Observation of
day-to-day clinic operations and interaction and interviews with other clinic staff
provided insights into forces within the practice environment that influence provision
of mental health care. Study data used to produce findings consisted of text comprising
transcriptions from audio recordings and/or notes from 72 interviews, augmented by
field notes from 198 hours of observation in clinic settings. A combination of analytic
techniques facilitated systematic examination and interpretation of these data.
Verification consisted of reviews of the findings with academic advisors, colleagues, and
most importantly PCPs themselves, to ensure accuracy and neutrality of interpretations
and conclusions.
Study Setting and Participants
Forty-eight family practice and internal medicine primary care providers,
including physicians (MDs), nurse practitioners (NPs), and physicians assistants (PAs)
3


whose patient panels included adults with chronic disease, and 12 non-PCP clinic staff
who provided mental health-related services, participated in this study. Seven
community health center clinics representing three clinic systems that provide primary
care services to low-income and uninsured people in the Denver, Colorado
metropolitan area served as study sites. Participating clinics were located in urban,
suburban and rural communities within and near the city. With the intent of identifying
patterns across settings and among individuals, a variety of settings and locations were
purposively identified and included to ensure that the data captured as wide a range as
possible of contextual influences on clinical decision making, and to gain access to as
broad a range of PCPs as possible.
Terminology
Acknowledging the complexity of terms used to frame and describe this study, I
have adopted definitions developed specifically for communicating to a broad audience
about mental health and illness, mental health care, and chronic disease management to
ensure clarity and consistency in terminology. These include reports from the U.S.
Surgeon General (U.S. Department of Health and Human Services, 1999, 2001), health
communications from the Centers for Disease Prevention and Control (CDC), and a
report on the status of mental health care in Colorado (TriWest Group, 2003). Citations
in this section and footnotes in a glossary of terms (Appendix A) identify these and
other sources of the definitions of terms that are central to this study.
The U.S. Surgeon Generals report on mental health views mental health and
mental illness as points on a continuum rather than as opposites (U.S. Department
of Health and Human Services, 1999, p. 4). While acknowledging that mental health
is not easy to quantify or interpret uniformly across cultures, the report defines it as the
successful performance of mental function, resulting in productive activities, fulfilling
relationships with other people, and the ability to adapt to change and to cope with
adversity (U.S. Department of Health and Human Services, 1999, p. 4). The U.S.
4


Surgeon General uses the term mental illness to refer collectively to all diagnosable
mental disorders (U.S. Department of Health and Human Services, 1999, p. 5), and
mental disorder to refer to a health condition that is characterized by alterations in
thinking, mood, or behavior (or some combination thereof) associated with distress
and/or impaired functioning (U.S. Department of Health and Human Services, 1999,
p. 5). For the purposes of this study, the term mental illness is used in reference to an
individuals health status to suggest the possibility that a persons condition may be
affected by more than one mental disorder, and mental disorder'll used to refer to labeled
mental health conditions (e.g., depression) that meet diagnostic criteria such as those
developed and published by the American Psychiatric Association in the Diagnostic and
Statistical Manual for Mental Disorder, Fourth Editions (DSM-IV). Finally, because of this
studys focus on providers interpretation of signs and symptoms, the term mental health
issues is used to refer to signs or expressions of a patients mental distress that the
provider detects but around which diagnostic uncertainty may still remain. Such signs
and symptoms, for example, those seen with bereavement, may be debilitating if left
unattended (U.S. Department of Health and Human Services, 1999, p. 5) but may not
meet the intensity or duration criteria for any mental disorder. I thus use this term to
refer to indications of potential mental disorder(s) in absence of a diagnosis.
This studys formulation also assumed that a distinction exists between mental
disorders and psychosocial issues which the U.S. Surgeon General considers to derive
from sources that are either psychological (e.g., stressful life events, affect, personality,
and gender) or social (e.g., parents; socioeconomic status; racial, cultural or religious
background; interpersonal relationships), or both (U.S. Department of Health and
Human Services, 1999, p. 55). While this research sought to discover inductively how
CHC providers define and apply the terms psychosocial issues and mental
disorders and how they differentiate between the two (Aim 1.2), it became clear as the
study progressed that focused analysis of these particular data would not be necessary to
illuminate how these PCPs frame the psychosocial concerns of and structure the
5


provision of mental health care to patients with chronic disease (Aim 1). I have thus
retained the working definition of the terms used to communicate the studys purpose
and design, which highlighted the challenges PCPs face in determining the threshold
for functional impairment (U.S. Department of Health and Human Services, 1999, p.
49) that guides mental health diagnosis and treatment. While the distinction in usage
between mental health issues and psychosocial issues in this report is nuanced, the
term mental health issues is used to emphasize the existence of detectable signs of mental
distress and the term psychosocial issues is used to emphasize sources of potential mental
distress that may cause the provider to suspect that mental health may be impacted,
whether the patient directly expresses distress or not.
Certain references to health status also deserve clarification. The U.S. Surgeon
General distinguishes disease from disorder, where disease is generally reserved
for conditions with known pathology (detectable physical change) and disorder
refers to clusters of symptoms and signs associated with distress and disability (i.e.,
impairment of functioning), yet whose pathology and etiology are unknown (U.S.
Department of Health and Human Services, 1999, p. 44). Following accepted
convention (U.S. Department of Health and Human Services, 1999), this study uses the
term disorder to refer to mental health conditions and the term disease is reserved to refer
to somatic, or general medical conditions, while acknowledging that in fact physical and
mental health are inseparable (deGruy, 1996, U.S. Department of Health and Human
Services, 1999). This study uses the term chronic illness as a more encompassing term to
refer to both physical and mental health conditions that are prolonged, do not resolve
spontaneously, and are rarely cured completely (Centers for Disease Control and
Prevention/Washington, 2003) and chronic disease to refer specifically to those conditions
in which alterations in non-mental functions predominate (U.S. Department of
Health and Human Services, 1999, p. 6).
This study uses the descriptor socially and economically disadvantaged to refer to
groups affected by issues of both poverty and limited political power and social
6


resources, as well as unequal access to opportunities, social rewards, and social status
(U.S. Department of Health and Human Services, 2001, p. 5). The latter component of
this definition mirrors the use of the term minority in the U.S. Surgeon Generals
report, which adds that it is not used to connote inferiority or to indicate small
demographic size (U.S. Department of Health and Human Services, 2001, p. 5) of a
group. In adopting a unique descriptor for the population of concern for this study, I
wish to avoid assigning characteristics such as poverty and powerlessness to all
individuals in such categories but to instead refer specifically to the issues that render
individuals in these groups vulnerable.
Finally, the literature often categorizes groups according to race and
ethnicity, and usage often conflates these with issues of poverty and/or social status.
Acknowledging that these terms represent not biologically and genetically distinct, but
rather socially-constructed categories (U.S. Department of Health and Human Services,
2001), I use the terms race and ethnicity as well as the term minority where necessary to
maintain consistency with references in the literature, particularly relative to health
disparities. Also, to report study findings and statistics, it is often necessary to use U.S.
Census categories, such as Hispanic to delineate among groups. Because the
participating clinic systems use the term Hispanic in characterizing the ethnicity of
their patients, I use it in this report to refer to that patient group.
Usage and Conventions
Because this study focused on the unique perspectives of individual PCPs, I
have attempted to provide as much insight into this topic as possible in the words of the
PCPs themselves. The number that appears in brackets following a quotation is the
coded participant identifier. I also use this convention with key words or phrases
embedded within the discussion to indicate that the words are the PCPs, not mine.
To protect the identities of individuals and their clinic affiliations, I have
omitted words or phrases within quotations that might reveal this information. These
7


changes are denoted either by a double slash (//) indicating that text has been removed,
or the inclusion of a substitution, shown in brackets. For example, [CMHC] may appear
in place of the name of a specific community mental health center, whose location
could be associated with a particular clinic located in the same service area. Similarly,
within the quotations generic references such as s/he and PCP mask gender and
specialty or occupation, respectively. Because quotations are excerpted, in some cases I
insert a word or phrase to clarify what is being said. These also appear in brackets
within the quotation. Additionally, a double slash (//) within quotes indicate instances
where I have deleted words or phrases for the sake of brevity.
Structure of the Dissertation
Following this introduction, Chapter 2 discusses the background supporting this
studys focus and aims, including a summary of the literature containing both empirical
findings and discourse on the major concerns that underlie the provision of mental
health care in primary care and health care for socially and economically disadvantaged
individuals. It also outlines the philosophical and theoretical assumptions guiding the
studys methodology and interpretation of results. Chapter 3 details the research design
and methods used for data collection, the analysis used to produce the findings, and the
steps taken to verify accuracy and neutrality of findings. Study findings are discussed in
Chapters 4 and 5, with Chapter 4 focusing on the description of PCPs approaches to
mental health care and Chapter 5 situating these phenomena within the broader context
of care to further elucidate those findings. Chapter 6 includes discussion of the findings
implications of this work, both as a contribution
as the basis for future research.
relative to the literature, and identifies
to the current body of knowledge and
8


CHAPTER 2
BACKGROUND
An estimated 28 to 30% of the adult population in the United States is affected
in a given year by mental disorders, as defined by prevailing diagnostic criteria 1 (Kessler,
Chiu, Dernier, & Walters, 2005; U.S. Department of Health and Human Services, 1999).
Six percent of adults in the United States experience significant functional impairment
from these disorders, and another 10% are at least moderately impaired in their ability
to function (Kessler et al., 2005). Although estimation of prevalence of mental illness is
hampered by lack of clear thresholds for measurement (Kessler & Zhao, 1999; Switzer,
Dew, & Bromet, 1999; U.S. Department of Health and Human Services, 1999), these
figures indicate the potential burden of these conditions on individuals and society in
the United States. The costs in individual suffering and economic burden of mental
disorders are tremendous. In a massive 1990 study, researchers from Harvard School of
Public Health and the World Health Organization estimated that mental illnesses
account for more than 11% of disease burden worldwide (measured by premature
mortality 1 2 and disability as well as morbidity), and for 15% of disease burden from all
causes in the United States and other countries with established market economies
(Murray & Lopez, 1996). Depression alone is estimated to cost $83.1 billion annually in
the United States, considering the cost of medical care, mortality-related suicide, and
workplace costs associated with depression (Greenberg et al., 2003), making it one of
1 Reported estimates vary depending upon which disorders are included. The Surgeon Generals report
on mental health (U.S. Department of Health and Human Services 1999), for example, treats addictive
disorders separately from other mental disorders, whereas recent epidemiological analyses by Kessler et
al. (2005) include substance abuse disorders but have excluded schizophrenia and other non-affective
psychoses because these are not included in some data sets used for their analysis. In both cases, because
of the high rate of co-occurrence of these conditions with other mental disorders, estimates identify most
individuals with mental disorders.
2 Premature mortality is defined as death that occurs before reaching the age of highest life expectancy of
any world population (Murray & Lopez 1996).
9


the 10 most costly illnesses in the United States (Hirschfeld et al., 1997). Quality-of-life
issues, if similarly quantified, would substantially increase this estimate (Hirschfeld et al.,
1997).
Chronic illnesses, both physical and mental, rank among the leading causes of
disability and mortality in the world (Murray & Lopez, 1996), and both types of
conditions co-occur in the same individual more often than chance would predict
(deGruy, 1996; Eaton, 2002; Katon & Ciechanowski, 2002; Muehrer, 2002; Rubin &
Peyrot, 2001; U.S. Department of Health and Human Services, 1999). For example,
several studies have found a significantly higher prevalence of depression or depressive
symptoms among diabetics than non-diabetics (Black, 2002; Delamater et al., 2001;
Katon & Ciechanowski, 2002; Talbot & Nouwen, 2000), and similar comorbidity exists
in patients with other chronic diseases (Chapman, Perry, & Strine, 2005; deGruy, 1996;
Garfield, 2002; Goldman, Nielsen, & Champion, 1999). The reciprocal effects of
chronic disease and depression are well-substantiated, with evidence that depression is
associated with poor glycemic control and accelerated rates of coronary heart disease in
diabetic patients (Chapman et al., 2005; Lustman & Clouse, 2002) and that diabetes and
heart disease may increase risk for depression (Chapman et al., 2005; Muehrer, 2002).
Concomitant depression has been shown to substantially increase the health burden of
diabetes (Black, 1999) and arthritis (Chapman et al., 2005). A review of studies of
comorbidity of depression and several chronic illnesses found evidence of increased
use of medical resources and costs, amplification of physical symptoms, additive
functional impairment, decreased ability to adhere to medications... [and necessary
lifestyle changes]..., and increased mortality (Katon & Ciechanowski, 2002, p. 859).
Greenberg and associates point out that estimates of the economic burden of
depression, already astronomical, do not consider the incremental health care costs for
managing other health conditions, either physical or psychiatric, that co-occur with
depression and whose management are complicated by it (Greenberg et al., 2003).
10


The interaction between physical and mental health, particularly in the context
of chronic illness, thus has been the focus of increasing attention by federal agencies
and research institutions concerned with addressing public health problems in the
United States (Abeles, 2002; Chapman et al., 2005; Garfield, 2002; Office of Behavioral
and Social Sciences Research, 2001; U.S. Department of Health and Human Services,
1999). One emphasis is to improve management of co-existing physical and mental
illness in primary care settings (Donaldson, Yordy, Lohr, & Vanselow, 1996), where
most patients with these conditions receive their care (deGruy, 1996; Klinkman, 1997).
This understanding is critical to reducing the burden of chronic illness at both the
individual and societal levels, as well as to fulfilling the expectation for delivery of
integrated care in primary care settings, especially with increasing recognition of primary
cares importance in the de facto mental health network (deGruy, 1996; Volk, Nease,
& Cass, 1997). Each year, between 10 and 20% of the general population seeks mental
health care in a primary care setting (deGruy, 1996), and for more than 50% of patients
with mental illness, primary care providers are their sole health care contacts (Robinson,
Geske, Prest, & Barnacle, 2005). As much as 40% of primary care patients have
diagnosable mental disorders associated with significant functional impairment
(deGruy, 1996; Muehrer, 2002). One concern is that while evidence exists that mental
disorders are treatable (U.S. Department of Health and Human Services, 1999) and that
treatment for these disorders can improve the course and outcomes of some chronic
physical conditions (Lustman & Clouse, 2002), a large proportion, especially mood
disorders such as depression, remain unrecognized or misdiagnosed in these settings
(Hirschfeld et al., 1997). Only about half of those who receive treatment meet
diagnostic criteria for a mental disorder (Kessler et al., 2005; Wang, Lane, Pincus, Wells,
& Kessler, 2005), further indicating the need for improvement in mental health
assessment in these settings.
Improving recognition and treatment of mental disorders in primary care
settings is particularly important for improving health care for vulnerable populations
11


(Fiscella, 2002). The U.S. Surgeon Generals report on mental health drew attention to
the disproportionate disability burden from unmet mental health needs borne by racial
and ethnic minority groups, which prompted further examination of disparities in
mental health care for these groups (U.S. Department of Health and Human Services,
2001). The most current analyses of psychiatric epidemiologic data concluded that while
most people with mental disorders are untreated or poorly treated (Kessler et al., 2005),
unmet needs for treatment are greatest in traditionally underserved groups, (Kessler
et al., 2005, p. 629) among them, racial and ethnic minorities and people with low
incomes and/or who lack insurance (Wang et al., 2005).
The consequences of mental illness are magnified for socially and economically
disadvantaged groups, who also have higher rates and severity of physical illness
(Smedley, Stith, & Nelson, 2002). These groups are more likely to lack access to needed
health and social services and to realize more profound economic and social
consequences from disabling conditions (U.S. Department of Health and Human
Services, 2001). While evidence does not associate ethnicity with increased risk for
mental disorders, it does indicate that members of ethnic minority groups, regardless of
socioeconomic status, tend to have disorders that are more persistent (Breslau,
Kendler, Su, Gaxiola-Aguilar, & Kessler, 2005). In addition, racial and ethnic minorities
are less likely than other populations to seek mental health care from specialists
(Borowsky et al., 2000; U.S. Department of Health and Human Services, 2001) and are
more likely to access care in primary care settings (deGruy, 1996; U.S. Department of
Health and Human Services, 2001), and both poorer health and higher rates of non-
recognition and/or under-treatment for mental illness are associated with these groups
(Borowsky et al., 2000; Snowden, 2003; U.S. Department of Health and Human
Services, 2001; Wang et al., 2005; Young, Klap, Sherbourne, & Wells, 2001;).
The inherent complexities and constraints primary care providers face in
recognizing and treating mental disorders are compounded for socially and
economically disadvantaged groups. Characteristics associated with ethnic and cultural
12


diversity, including differences in treatment-seeking behaviors, symptom interpretation
and presentation, patient mistrust based on historical influences of racism and
discrimination, language barriers, and level of acculturation may increase the risk of
non-recognition and inadequate treatment by clinicians (U.S. Department of Health and
Human Services, 2001). One hypothesis suggests that clinician bias and stereotyping,
subconsciously invoked in response to heightened diagnostic uncertainty from factors
such as these, also play a role in disparate recognition and treatment, particularly in
racial and ethnic minorities (Smedley et al., 2002; Snowden, 2003; U.S. Department of
Health and Human Services, 2001; Van Ryn & Fu, 2003). Some argue that issues such
as patient-provider class differences and influences of the culture of medicine (Good,
James, Good, & Becker, 2002) also impact clinical decision making (U.S. Department of
Health and Human Services, 2001; Waitzkin, 2000).
While disparities in mental health care clearly have been linked with race and
ethnicity, the reasons behind these disparities are not fully understood (Breslau et al.,
2005; U.S. Department of Health and Human Services, 2001). Areas identified by the
U.S. Surgeon General as priorities for further scrutiny include cultural factors related to
patients, providers, and service delivery; and larger social, economic, and political
factors such as racism and discrimination, stigma, poverty, and structure and financing
of mental heath services (U.S. Department of Health and Human Services, 2001). As
suggested in recommendations and priorities from the National Institutes of Health for
research on mental health care disparities (Office for Special Populations, 2001; Office
of Behavioral and Social Sciences Research, 2001), this study explored the intersection
of social, cultural, and environmental influences in the clinical encounter, as reflected in
providers judgments and consequent courses of action in response to patients
psychosocial issues.
The following background sections discuss in more detail what is known relative
to the study question. The terminology used in describing the background for this study
follows definitions provided by the Institute of Medicine, U.S. Surgeon General, and
13


others in recently published reports on primary care (Donaldson et al., 1996), mental
health (Kessler, 2004; Kessler et al., 2005; TriWest Group, 2003; U.S. Department of
Health and Human Services, 1999, 2001) and health care disparities ( Smedley et al.,
2002; U.S. Department of Health and Human Services, 2001).
Primary Care and Integrated Care
A critical element in the Institute of Medicines definition of primary care is its
responsibility for the majority of personal health care needs, which include physical,
mental, emotional and social concerns.. .that involve the functioning of an individual
(Donaldson et al., 1996, p. 33). This definition places primary care providers in the
pivotal role of [receiving] all problems that patients bring unrestricted by problem or
organ system (Donaldson et al., 1996, p. 33) with the expectation that these providers
will diagnose and manage most health problems and will involve other health care
providers as needed to ensure that patients receive integrated care (Donaldson et al.,
1996). This view presupposes that physical health and mental health are inseparable, a
belief underpinning both the vision for primary care (deGruy, 1996) and the U.S.
Surgeon Generals recommendations on mental health (U.S. Department of Health and
Human Services, 1999; U.S. Department of Health and Human Services, 2001). For
most primary care providers, however, the inclusion of mental, emotional and social
concerns within the realm of primary care presents a formidable expectation and one
for which typical medical training is inadequate preparation and which is not always well
supported by the environments within which these providers practice (deGruy, 1996).
The literature contains evidence that while primary care providers recognize and accept
this mandate, not all are confident in their ability to function as mental health providers
(Cameron & Mauksch, 2002; Craven, Cohen, Campbell, Williams, & Kates, 1997), and
it identifies many factors related to patient characteristics and the practice setting that
may impede the ability of these providers to provide mental health care. Thus,
researchers have begun to focus on the gap between idealized practice and actual care
14


(Robinson, Prest, Susman, Rouse, & Crabtree, 2001 p. 864) for mental health problems
in these settings.
Delivery of mental health care in primary care settings differs in important ways
from that delivered by mental health specialists (Craven et al., 1997; deGruy, 1996;
Klinkman, 1997). Three issues in particular support this studys concern. One is the
undifferentiated nature (Craven et al., 1997, p. 943) of problems that many primary
care providers see, making it difficult to sort out the respective contributions of
physical problems, adverse life events, and psychiatric disorders (Craven et al, 1997, p.
947). Most patients with mental health issues see their primary care provider for
physical complaints rather than seeking treatment for emotional distress (deGruy, 1996;
Klinkman, 1997; Robinson et al., 2001). Nearly half of primary care visits involve
somatization, or clinically significant physical symptoms that cannot be entirely
explained by physical disease (Dickinson et al., 2003, p. 1) which poses a particularly
difficult obstacle to identifying mental health issues.
A second issue and related confounding factor is the comorbidity of mental and
physical illness (Docherty, 1997; Goldman et al., 1999; Rost et al., 2000; Volk et al.,
1997), which occurs in two-thirds of primary care patients with diagnosed mental
disorders (deGruy, 1996; Goldman et al., 1999). The tension and overlap between
physical and mental health problems as experienced in primary care settings has
received attention in primary care studies of competing demands (Nutting, Rost,
Smith, Werner, & Elliot, 2000; Rost et al., 2000) that attempt to explain clinical decision
making given these conditions. A third issue is that the mental health problems that
primary care providers see, particularly mood disorders, often fail to meet diagnostic
criteria such as those in the DSM-IV, the de facto standard classification system
developed primarily for use in specialty care settings (deGruy, 1996; Klinkman, 1997).
Such mixed syndromes and sub-threshold conditions (deGruy, 1996, p 294) also
may benefit from treatment, but evidence to identify which are amenable to treatment is
still lacking (deGruy, 1996).
15


Besides lacking a diagnostic classification system that accommodates their
unique needs, primary care providers also lack guidance for managing mental health
issues within the context of physical illness (deGruy, 1996; Farley, 2000). Care
guidelines and quality of care measures for chronic diseases tend to focus on
management of physiologic disease indicators while failing to mention appropriate
psychosocial care (American Diabetes Association, 2002; Larme & Pugh, 2001). The
American Diabetes Association, for example, does not include guidance for
management of depression within its standards of diabetes care (American Diabetes
Association, 2002), despite both the prevalence of comorbid depression and diabetes
and evidence that optimal outcome relies on comprehensive treatment, giving equal
emphasis to both the medical and mental disorders (Lustman & Clouse, 2002, p. 919).
Researchers and health care providers are increasingly underscoring the importance of
recognizing and treating mental disorders that co-occur with medical conditions
(Chapman et al., 2005; Glasgow et al., 2001; Goldman et al., 1999; Lustman & Clouse,
2002; Muehrer, 2002; Nutting et al., 2002; Rost et al., 2000; Rubin & Peyrot, 2001). An
identified need for primary care providers is for diagnostic systems that are more
congenial to the demands, constraints, and phenomenology (deGruy, 1996, p. 294) of
mental health issues encountered in these settings, including formulations that allow
providers to address the question of whether [mental health] conditions should be
regarded and managed differently when they occur in the presence of physical illness
(deGruy, 1996, p. 295).
Chronic Disease and Mental Health Care
for Disadvantaged Populations
Achieving an integrated care model is especially challenging for primary care
providers caring for individuals who are socially and economically disadvantaged
(Garfield, 2002). People who are poorer also tend to have poorer physical and mental
health than the general population (Bruce, Takeuchi, & Leaf, 1991; Everson, Maty,
16


Lynch, & Kaplan, 2002; Geronimus, 2000; Link & Phelan, 1995; Mauksch et al., 2001;
Norquist & Wells, 1991; U.S. Department of Health and Human Services, 1999;
Waitzkin, 2000; Yu & Williams, 1999) and evidence has linked poverty with biomedical
and psychosocial comorbidity (Berkman & Kawachi, 2000; Fiscella, 1999). It has been
well established that people in the lowest socioeconomic strata are two to three times
more likely to have a mental disorder and to have higher levels of psychological distress
(U.S. Department of Health and Human Services, 2001). These individuals are more
likely to experience stressful circumstances of daily living, including exposure to more
extreme environmental stressors such as violence and unemployment, and are less likely
to have social or economic resources to buffer these stresses (Fiscella, 1999; U.S.
Department of Health and Human Services, 2001).
People with lower incomes are also more likely to be dependent on drugs or
alcohol3 (Fiscella, 2002; Norquist & Wells, 1991; Rosenbaum, Shin & Darnell, 2004;
Wu, Kouzis, & Schlenger, 2003; Wu & Ringwalt, 2005), which further complicates
management of both physical and mental health conditions. Studies show that people
with either mental disorders or substance use problems are roughly twice as likely to
have both types of issues as people who are affected by neither (Kessler, 2004; National
Health Policy Forum, 1998). Referred to as dual diagnosis, this comorbidity renders
both conditions more severe, more chronic, and more resistant to treatment than either
would be alone (Kessler, 2004). An attribute that is strongly associated with dual
diagnosis is lack of health insurance, which is characteristic of many people with low
incomes, and which also poses substantial barriers to health services that are specialized
for treating either mental health or substance use problems (Wu et al., 2003). Most
people who receive treatment for co-occurring mental and substance abuse disorders
3 Studies show that lack of health insurance increases the risk of dependency on substances. Both ethnic
minorities and those with low incomes are disproportionately represented among the uninsured. Fiscella
(2002) also proposes that higher rates of chemical dependency in low-income populations are related in
part to the detrimental effect of chemical dependency on employment and earnings (Fiscella, 2002, p.
368).
17


obtain it in general medical rather specialty settings (National Health Policy Forum,
1998).
Although poverty affects mental health status and disproportionately affects
racial and ethnic minorities, evidence shows that socioeconomic status alone does not
explain mental health disparities documented for these groups, whose social situations
are also shaped by historical influences of racism, oppression, and discrimination at
societal and institutional as well as individual levels (Breslau et al., 2005; U.S.
Department of Health and Human Services, 2001). While chronic diseases constitute a
tremendous health burden across all populations in the United States, affecting 90
million people and accounting for 70% of all deaths (National Center for Chronic
Disease Prevention and Health Promotion, 2005), minority populations are
disproportionately burdened. For example, on average among adults, African
Americans and Hispanics are two to three times more likely than non-Hispanic whites
to have diabetes (National Institute of Diabetes and Digestive and Kidney Diseases
[NIDDK], 2002). Because chronic disease is a risk factor for mental disorders
(Chapman et al., 2005; deGruy, 1996; Katon & Ciechanowski, 2002; Rubin & Peyrot,
2001; U.S. Department of Health and Human Services, 2001), this also places minority
populations at higher risk for mental illness (U.S. Department of Health and Human
Services, 2001).
The chronic disease and mental illness burden on low-income and minority
populations is thus an important consideration in community health centers the
settings that are the focus of this study two-thirds of whose patients are minorities (on
average, nationwide), and nearly all of whose patients are uninsured or underinsured
(Bailey, Legaspi, Bloom, Campbell, & Regan, 2000). For individuals whose lack of
financial resources denies them access to most of the health care system in the United
States, primary care is available to only a limited number, mainly through the safety
net system of providers, which includes public hospital systems, publicly supported
community health centers, and local health departments (Lewin & Altman, 2000). These
18


providers operate under a variety of structural arrangements, but all grapple to some
degree with more limited resources than their privately funded counterparts (Chin et al.,
2001; Lewin & Altman, 2000). A particularly difficult limitation for many of these
providers is the inability to refer most of their patients to specialists, including mental
health specialists. Lacking this option, the primary care provider is left to do what s/he
can to attend to the patients needs (Goldman et al., 1999). This, coupled with the
poorer baseline physical and mental health and higher prevalence of biomedical and
psychosocial comorbidity among patients (Berkman & Kawachi, 2000; Fiscella, 1999),
describes the dilemma that provided the backdrop for this studys focus on provision of
integrated care in these settings.
A recently published Institute of Medicine study on health care disparities
examined, among other factors, mechanisms in the clinical encounter that might be
operative.. .from the providers side of the exchange: bias (or prejudice) against
minorities; greater clinical uncertainty when interacting with minority patients; and
beliefs (or stereotypes) held by the provider about the behavior or health of minorities
(Smedley et al., 2002, p. 9). The report, echoing discussion in the U.S. Surgeon
Generals report on mental health (U.S. Department of Health and Human Services,
2001), suggests that clinician behavior in turn may elicit responses, such as reluctance to
disclose emotional distress or to follow up on a referral to a specialist, from patients
who are often already mistrustful of the medical establishment. This may then further
contribute to disparate outcomes. A factor contributing to clinical uncertainty in dealing
with these populations is that patients often bring into the encounter psychosocial
issues that further confound recognition and accurate diagnosis of mental disorders
(Goldman et al., 1999).
Providers may perceive that patients preoccupation with psychosocial issues
derails their effort to deal with physical health issues (Fiscella, 1999; Good et al., 2002;
Mechanic, 1992), and it is unclear if or how providers who routinely work with these
populations distinguish psychosocial issues from mental distress that indicates treatable
19


illness (Chew-Graham, Mullin, May, Hedley, & Cole, 2002). With evidence of the
impact of poverty, racism, and discrimination on both mental and physical health
(Bruce et al., 1991; Everson et al., 2002; Fiscella, 1999; Geronimus, 2000; Lynch,
Kaplan, & Shema, 1997; Norquist & Wells, 1991; Waitzkin, 2000), this constitutes an
important gap in our understanding of the psychosocial aspects of primary care for
patients whose life circumstances may, through manifestation in symptoms of mental
health issues or disorders, impede or prevent their active engagement in managing their
illness. As Black (2002) observed about disadvantaged diabetics, Control of diabetes is
a highly demanding endeavor, requiring substantial vigilance, lifestyle change,
medication adherence, and motivation. Those who suffer most from the disease are
often those least prepared to deal with it (Black, 2002, p. 547).
The importance of this studys focus on chronic disease and mental health care
in community health center settings is reflected in the priorities for a multi-year
program undertaken by several federal agencies, known as the Health Disparities
Collaboratives (Bureau of Primary Health Care, 2003). Intended to transform the way
safety net providers engage and treat their patients who may have chronic physical
health conditions as well as mental health problems (U.S. Department of Health and
Human Services, 2001, p. 163), these initiatives have been implemented in several
community health centers nationwide to improve care of diabetes, asthma,
cardiovascular disease, and depression for medically underserved patients.
Non-recognition of Mental Disorders in Primary Care
Settings: Issues. Evidence, and Factors
While the rate of treatment for mental disorders has increased in the last 10
years, still only one-third of adults with diagnosable mental disorders receive mental
health services within a given year in the United States (Kessler et al., 2005). Under-
recognition of these disorders contributes to under-treatment (Volk et al., 1997).
Researchers have approached the issue of detection or recognition of mental
20


disorders in a number of ways. Because the two terms appear to be used
interchangeably in the literature, the term recognition is used for consistency in this
discussion. Recognition may be assessed either direcdy, through operationalizing and
measuring the construct, or indirectly, by comparing treatment rates (prescriptions,
counseling, specialty referral) against patient diagnoses or mental health screening
results. Rates as high as 60% to 70% have been reported for recognition of depression
in primary care (Davidson & Melzer-Brody, 1999) and between 30% to 40% for
accurate diagnosis of depression in these settings (Docherty, 1997). Because of the
broad variation in methods and measures used to determine these rates, more precise
quantification of the issue is difficult, but results from various studies tend to agree that
between 30% and 50% of diagnosable mental disorders go unrecognized in primary
care settings (deGruy, 1996; Goldman et al., 1999). Although these approaches provide
some indication of the size of the non-recognition problem, they typically are unable to
identify instances in which the provider may have recognized mental health problems
but chose not to record this in medical records (Callahan et al., 1997) or not to follow
up on the recognized issues (Robinson et al., 2001; Rost et al., 2000; Nutting et al.,
2002).
Insights into providers approaches to treatment as well as recognition are
important. Based on direct observation of primary care outpatient visits, Robinson, et
al. (2001) described a three-phased process by which physicians responded to
emotional distress (Robinson et al., 2001, p. 864), indicating that the boundary
between recognition and treatment is not necessarily clear. Recognition does not
necessarily predict that treatment will follow or that the treatment offered will be
adequate (Callahan, Dittus, & Tierney, 1996; Davidson & Melzer-Brody, 1999; deGruy,
1996; Melfi, Croghan, & Hanna, 1999). Only about half of patients whose mental
disorders are recognized receive any treatment, and treatment discrepancies (i.e., under-
treatment or treatment for conditions that do not meet diagnostic criteria) occur more
often in primary care than specialty settings (Goldman et al., 1999; Kessler et al., 2005).
21


Some authors cite evidence of under-treatment as an indicator of under-
recognition (Davidson & Melzer-Brody, 1999). Much of this evidence comes from
studies evaluating either presence or absence, or appropriateness (using evidence-
based criteria) of treatment based on patient surveys that include questions to screen for
depression and/or anxiety and ask patients about receipt of mental health care. Such
studies have found appropriate treatment for between 11% and 33% of patients
whose screening indicated probable mental disorders (Davidson & Melzer-Brody, 1999;
Kessler et al., 2005; Wang, Berflund, & Kessler, 2000; Young et al., 2001). Because of
this disconnect, an equal emphasis on ensuring that providers are properly equipped
and supported for providing appropriate treatment is necessary (Goldman et al., 1999).
Several correlates of recognition of mental disorders have been identified in
various studies. Recognition has been found to be higher among patients who are older
(Borowsky et al., 2000; Callahan et al., 1996), female (Borowsky et al., 2000; Callahan et
al., 1997), and have lower perceived health status (Callahan et al., 1997), as well as
among those who are white, not married, and have lower income (Borowsky et al.,
2000). A lower level of education has been associated with both higher (Borowsky et al.,
2000) and lower (Callahan et al., 1997) recognition rates. One recent study in particular
(Borowsky et al., 2000) provided confirming evidence that patient demographic factors
influence non-recognition, further fueling efforts to address mental health care
disparities in racial and ethnic minorities (U.S. Department of Health and Human
Services, 2001). Borowsky, et al. (2000) adopted an intentionally inclusive definition
of detection in a cross-sectional survey-based study of patient and provider
characteristics associated with non-recognition of mental disorders in primary care.
Under this broadened definition, besides diagnosis, report of referral or counseling was
considered indicative of the providers awareness of a patients mental health issues.
Using a nested sampling scheme comprising three samples, each more tightly screened
than the one containing it, and comparing these assessments against responses of
physicians (who were blinded to screening results) on a post-visit survey, these
22


researchers were able to test associations among several variables and indicators of
recognition. Variables tested included patient demographics and self-reported health
and functional status, physician demographics and their proclivity toward providing
counseling for depression (Borowsky, et al., 2000, p. 381), and diagnoses of specific
depressive disorders. This study provided the most convincing and recent evidence to
date that race/ethnicity and gender influence recognition, showing reduced odds of
recognition for men and for African Americans and Hispanics. The analysis found no
significant interaction between race/ethnicity and gender. Another important finding
was that physicians caring for minority patients tended to self-report a lower preference
for counseling for depression.
The nature and severity of the presenting symptoms in primary care also impact
recognition. Borowsky et al. (2000) found an overall recognition rate of less than 50%
for patients whose screening showed the most distinct indicators of diagnosable
depression disorders about twice the rate of recognition for patients whose screening
indicated possible depression. Not surprisingly, physicians in this study were more likely
to recognize major depression than sub-threshold conditions or conditions manifesting
in less severe or fewer symptoms, a finding supported by other research (Callahan et al.,
1997; deGruy, 1996). Recognition also is more likely for patients whose presenting
complaint is emotional rather than physical (Borowsky et al., 2000; deGruy, 1996).
Whereas primary care providers accurately diagnose mental disorders in only about 50%
of patients who present with physical complaints, they are able to do so in more than
90% of patients whose presenting complaint is emotional distress or a psychological
symptom (deGruy, 1996). Competing demands studies have found that physical
problems tend to take priority over mental health issues for time and attention in
primary care visits (Nutting et al., 2000; Rost et al., 2000). Other studies have found an
association between existence of comorbid physical problems and non-recognition of
depression (Docherty, 1997; Goldman et al., 1999). These seem to contradict studies
that have found that recognition and/or diagnosis of mental disorders is more likely to
23


occur in the presence of physical illness (Borowsky et al., 2000) and more often during a
chronic care visit than acute or well-care visits (Callahan et al., 1998).
While these associations provide an intriguing glimpse into the process of
recognition (Callahan et al., 1997, p. 175) of mental disorders in primary care, little has
been done to delve more deeply into the process itself. In particular, litde is known
about the experiences of primary care providers in providing mental health care; the
meanings they attach to various signs, symptoms, or patient characteristics; or their
strategies for sorting through and responding to the psychosocial and physical health
issues their patients talk to them about. As importantly, few of the studies on
recognition generalize to providers caring for socially and economically disadvantaged
populations or for culturally diverse populations. The sample for the study by
Borowsky, et al. (2000), for example, drew from private practice and health maintenance
organization (HMO) settings, and included only English-speaking adults. And while
Young et al. (2001) over-sampled people with low-income and psychiatric disorders to
assess demographic differences in quality of care for depression and anxiety in a
population-based sample, they excluded people with more severe disorders and those
who indicated they had experienced severe life difficulties (e.g., witnessing violence,
being homeless) within the past year. Criteria such as these not only exclude an
important subset of people receiving their care in community health centers, but also
obscure much of the complexity faced by providers in these settings in recognizing
treatable mental disorders. While several studies have used qualitative techniques to
explore providers views of and approaches to chronic disease management for low-
income and uninsured individuals (Hunt & Arar, 2001; Larme & Pugh, 1998; Larme &
Pugh, 2001; Loewe, Schwartzman, Freeman, Quinn, & Zuckerman, 1998; Loewe &
Freeman, 2000), the literature search for this study found only one qualitative study
(reviewed later in this section) that provides specific insights into primary care
providers reported experiences with providing mental health care to socially and
economically disadvantaged patients.
24


Qualitative research that explores providers attitudes and beliefs about or
approaches to mental health care for any population is scarce. A Canadian study used
focus groups of family physicians to obtain descriptions of how these providers
function in their roles as mental health providers (Craven et al., 1997 p. 943). Another
recent study (Robinson et al., 2001) analyzed detailed descriptions of directly observed
clinical encounters to characterize providers mental health care strategies. The resulting
typology of [family practice] physicians responses to patients expressed mental health
needs (Robinson et al., 2001, p. 864) accounted for both physicians philosophy, or
inclination (biomedical versus biopsychosocial4), and skill level (basic or advanced).
Most of the physicians observed in this study exhibited basic skills and a biomedical
inclination in their interactions with patients. Other studies seeking the providers
perspective have included interviews with providers to formulate questionnaires used to
examine perceived barriers to mental health care (Nutting et al., 2002) and/or self-
reported practices (Craven et al., 1997; Williams et al., 1999).
Under-recognition is more likely to occur in primary than specialty care settings
(deGruy, 1996; Goldman et al., 1999) largely due to the need for primary care providers
to attend to mental health issues concurrently with physical issues (deGruy, 1996;
Goldman et al., 1999; Klinkman, 1997). Accurate recognition and appropriate treatment
of mental disorders is complicated for primary care providers by the matrix of
explained and unexplained physical symptoms, as well as acute and chronic medical
illnesses in which signs and symptoms are embedded (deGruy, 1996, p. 286), and the
multi-faceted influences on the process of health care that derive from social,
cultural,and structural dimensions of the context within which that care occurs. Mental
health and illness, and the boundaries defining mental disorders, are in fact blurred by
their definitional and interpretational reliance on value judgments that may vary across
4 The authors considered a biomedical philosophy to be indicated by a concentration on medical aspects of
care, with minimal exploration of the patients psychosocial milieu, and a biopsychosocial orientation to
be indicated in physicians who addressed the patients emotional, physical, social, and sometimes
spiritual well-being (Robinson et a., 2001, p. 867).
25


individuals and cultures (U.S. Department of Health and Human Services, 2001, p. 6)
a key conception framing the U.S. Surgeon Generals supplementary report on mental
health disparities related to race, ethnicity and culture (U.S. Department of Health
and Human Services, 2001). The report stresses the important role played by the
providers judgment in interpreting the patients signs and reported symptoms and in
assessing the degree to which these impair the patients ability to function (U.S.
Department of Health and Human Services, 2001), and it is this subjective boundary
that is the focus of this study.
Snowden (2003) discussed two kinds of bias suggested by studies of mental
health decision making overpathologizing bias and minimi2ation bias.
Overpathologizing bias occurs when providers interpret unfamiliar behavior of
minority individuals as indicating mental illness, whereas minimization bias occurs
when providers ignore genuine manifestations of mental illness, perhaps attributing
these to cultural variations (Snowden, 2003, p. 241). DeGruy (1996) suggested that a
[primary care] providers inattention to a problem represents not negligence or
unwillingness but a rational setting of priorities among a list of competing demands
(deGruy, 1996 p. 298). Whatever the influence, the discretionary judgment of providers
ultimately determines whether and how the medical system responds to a patients
suffering.
Findings from a recent qualitative study by British researchers (Chew-Graham
et al., 2002) underscore the potential impact of providers attitudes and beliefs on the
courses of action taken with patients. In exploring general practitioners (GPs)
approaches to depression management through semi-structured interviews and
comparing those of GPs in socio-economically deprived areas with those serving
more affluent populations (Chew-Graham et al., 2002, p. 632), these researchers found
distinctively different views associated with the different patient populations. Whereas
the GPs working with the more affluent patients considered depression to be a treatable
illness and derived satisfaction from helping those patients, the GPs caring for patients
26


in deprived inner-city areas viewed ongoing management of depression as an an
interactional problem (emphasis in original) (Chew-Graham et al., 2002, p.632), resulting
from the need to respond to despair made manifest in illness (Chew-Graham et al.,
2002, p. 636). The researchers analysis revealed that confronting the intractable
problems of disadvantaged patients was considered burdensome by these GPs.
Attribution of emotional distress to life or situational problems is among
several factors suggested by various studies as contributing to non-recognition,
inaccurate diagnosis, failure to respond to suspected mental health problems, or
treatment that is not guideline-concordant. These factors, detailed in several review
articles and reports (Davidson & Melzer-Brody, 1999; Docherty, 1997; Goldman et al.,
1999; Hirschfeld et al., 1997; U.S. Department of Health and Human Services, 2001),
typically fall into three domains: those associated with patient characteristics, those
deriving from provider characteristics, or those that affect the clinical context in which
recognition occurs. Table 2-1 summarizes these factors, the sheer number of which
underscores the fact that influences on these clinical decisions are multi-faceted and
complex (Callahan et al., 1998; Klinkman, 1997; Rost et al., 2000), and thus more likely
to be illuminated through qualitative study techniques that examine phenomena in
context (Erlandson, Harris, Skipper, & Allen, 1993; Hinds, Chaves, & Cypess, 1992).
Limited information is available on specific cues that prompt recognition and
diagnosis of mental disorders, although studies have shown that recognition is more
likely to occur when a patient discloses emotional distress or brings up emotional issues
in an encounter (Callahan et al., 1998; Goldman et al., 1999; Williams et al., 1999).
Providers also reportedly respond to patients who are notably sad or who appear
depressed (Goldman et al., 1999; Williams et al., 1999). In a questionnaire-based study
to assess primary care physicians approaches to providing depression care, most
respondents indicated that they based their diagnosis on their overall clinical
impression, with some inquiry about specific symptoms (Williams et al., 1999, p. 64).
27


Table 2-1: Factors Influencing Recognition
of Mental Disorders in Primary Care
Factor
Provider characteristics: Attitudes, beliefs, preferences, knowledge, skills Level of comfort with psychosocial issues; interpersonal skills; self efficacy 4,7
Provider philosophy/style 7>11
Knowledge (e.g., insufficient knowledge of diagnostic criteria)3-4-5- 6> 7>8-15 16
Interviewing skills 4
Fear of jeopardizing patients insurance status or stigmatizing patient3
Beliefs about treatment effectiveness '-7
Does not consider the mental disorder a real condition; assumes it will resolve on its own 2
Medical training: biological orientation; lack of emphasis on psychosocial factors or mental health care skills 3
Reluctance or failure to inquire or follow up on cues (spoken or non- verbal) to patients distress 4
Provider-perceived role responsibility for mental health care 4
Attribution of blame to patient 4
Perception that there exists a good reason for patients distress (e.g., physical illness or serious problem in patients life) 4
Bias (race/ethnicity, gender, age); cultural and class differences 13>14
Clinical uncertainty 13
Patient characteristics or behaviors Lack of access to services and/or treatments 6
Reluctance or resistance due to stigma 3
Resistance to diagnosis or treatment 3
Fear, mistrust16
Low awareness; failure to recognize symptoms 3
Underestimate severity; feel they can deal with it themselves 6
Justify feelings as appropriate reaction to life situation 6
Cultural or ethnic beliefs/attitudes about emotional distress 6
Substance abuse 5
Factors affecting the clinical encounter Level of support in practice setting; resources available (e.g., access to mental health specialists) 3'7
Reimbursement limitations or patients ability to pay 3
Competing demands on providers time and attention; limited time to address all patients issues '.3,5,7,8,11,16
Language barriers 14
Lack of continuity of care or personal knowledge of patients 4
Lack of time to deal with emotional issues (real or provider-perceived) 2-4-5> 6, 7, 10, 16
Lack of adequate/relevant tools (screening, diagnostic), including ones tuned for primary care use and use with diverse populations 3
28


Table 2-1: Factors Influencing Recognition
of Mental Disorders in Primary Care (Cont.)
Factor
Confounding factors in patients presentation 3
Somatization1
Patient presents with physical complaints; medicalization of symptoms 3, 4, 5, 7,16,17
Patient fails to disclose distress >12
Comorbid medical illness or mental disorders; sub-threshold conditions 4*9- 10
Note. 1 Callahan et al., 1996 2 Davidson & Melzer-Brody, 1999 3 deGruy, 1996 4 Docherty, 1997 5 Goldman et al., 1999 6 Hirschfeld et al., 1997 7 Klinkman, 1997 8 Muehrer, 2002 9 National Health Policy Forum, 1998 10 Nutting, et al., 2002 11 Robinson et al., 2001 12 Rost et al., 2000 13 Smedley et al., 2002 14 Snowden, 2003 15 Williams et al., 1999 16 U.S. Department of Health and Human Services, 2001 17 Young et al. 2001
What is lacking is evidence deriving from direct inquiry of providers about what
constitutes or supports these impressions and what prompts them to proceed to
diagnose and/or treat a patient for a mental disorder. Because recognition of mental
disorders opens the way for treatment, it is important to better understand reasons for
non-recognition, especially in the context of chronic disease, where the patients mental
health has high potential impact on the burden of the disease and on disease
management, and particularly for patients who are disproportionately affected by both
mental and physical illness.
I
29


Theoretical and Philosophical Assumptions
Although the intent of this exploratory study was to describe PCPs practices
rather than to test specific theoretical constructs, it is important to acknowledge the
assumptions guiding the studys methodology. Strauss and Corbin (1994), in discussing
the injunction against use of restrictive prior theories and theoretical models (Strauss
& Corbin, 1998, p. 277) in the grounded theory approach, also acknowledged the
advantage and potential of researchers being theoretically sensitized. (Strauss &
Corbin, 1998, p 280). Such sensitivity allows the researcher to perceive the subde
nuances and meanings in data and to recognize the connections between concepts
(Strauss & Corbin, 1998. pp. 42-43) as they search for meaning in qualitative data. At
the same time, awareness of taken-for-granted presuppositions (Toombs, 2001, p. 1)
helps the researcher approach the study reflexively. A fundamental assumption reflected
in this studys approach is that all research is inherently value-laden and that the way to
ensure a level of objectivity is to understand assumptions that inform the research and
that may influence analysis and interpretation (Addison, 1999; Dewalt, Dewalt, &
Wayland, 1998; Hammersley & Atkinson, 1995; Van Manen, 2001). The following
sections summarize both the theory-informed assumptions and personal
presuppositions reflexively recognized by the researcher.
Constructivism
Epistemological and ontological assumptions constituting a constructivist
paradigm, as described by Guba (1990), Erlandson, et al. (1993), and Miller and
Crabtree (1999) informed the approach to the study question. Supporting the goal of
better understanding how PCPs frame and structure the process of integrated care, the
approach to this study assumed, consistent with this paradigm, that these constructions
are socially and experientially based (Guba, 1990, p. 27) in clinical encounters, and
that access to this insight would best be obtained directly from the PCPs themselves,
who construct the reality of their clinical practice in the process of interacting with
30


patients. Also instrumental in this process are PCPs taken-for-granted presuppositions,
or predispositions stemming from their past experiences and socialization
(Moustakas, 1995). This study sought to understand what constitutes the reality of
PCPs practices and how the subjectively conditioned truth (Waksler, 2001, p. 74) on
which practices are based influence recognition, interpretation, and responses to patient
psychosocial issues in clinical encounters.
To do so, the research design drew on the strength of the related naturalistic, or
constructivist, approach to inquiry (Erlandson et al., 1993; Miller & Crabtree, 1999) for
addressing the study aims. This approach stresses the importance of context in studying
social phenomena and argues for the need to study these phenomena in the natural
settings in which they occur (Erlandson et al., 1993), with the assumption that behavior
both shapes and is shaped by peoples responses to stimuli from their environment, and
importantly, that it is the persons interpretation of those stimuli that matters
(Hammersley & Atkinson, 1995). A closely associated assumption is that meaning is
expressed both verbally and through behaviors and that accessing those meanings
requires interpretation (Addison, 1999). These assumptions are best supported by a
qualitative approach that relies on both reported and observed action to explore and
understand in depth the constructed realities (Erlandson et al., 1993, p. 21) of the
decision makers on whom the study focuses because these techniques allow for thick
data to be collected that demonstrate their interrelationship within their context
(Erlandson et al., 1993, p. 16).
Van Manen (2001) identified a specific need for a research model suited to
examining issues and concerns that arise in the lifeworld of professional practitioners
(Van Manen, 2001, p. 459), such as clinicians. Asserting that such issues and concerns
find their source in situational contexts where practical action is necessary (Van
Manen, 2001, p. 468) Van Manen is among proponents of naturalistic inquiry who
advocate a context-sensitive research approach (Addison, 1999; Erlandson et al., 1993;
Hinds et al., 1992; Van Manen, 2001). Central to such an approach is the practice of
31


looking beyond individual actions, events, and behaviors to a larger background
context and its relationship to the individual events (Addison, 1999, p. 151).
Practice Ecosystem: The Context of Care
The contexts in which clinical encounters occur contain multiple factors at
multiple levels that condition PCPs beliefs and attitudes as well as constraining their
clinical decisions. A key assumption underlying this research is that the ability to attach
meaning to and to understand a phenomenon dependjs] on .. .knowledge of the
multiple contexts in which that phenomenon occurs (Hinds et al., 1992, p. 32). Hinds
and colleagues proposed that meaning in a clinical encounter, for example will be
different for PCPs and patients because their contexts differ (Hinds et al., 1992). This
parallels the constructivist idea that there exist not one, but multiple social realities
among participants in any interaction. Not only do the predispositions of both the PCP
and patient condition their interactions, but the features of the clinical setting in which
they occur bound the possibilities for therapeutic action as well.
It is now recognized that biological, psychological, and sociocultural factors are
interrelated and interact in the production of physical and mental health and illness
(deGruy, 1996; Engel, 1980; Hahn & Kleinman, 1983; Lock, 1995; U.S. Department of
Health and Human Services, 1999), and that these same factors influence the process
and outcomes of clinical encounters (Clark, Potter, & McKinlay, 1991; Engel, 1980;
Smedley et al., 2002; van Ryn & Fu, 2003). Various models have been constructed to
describe the multiple influences at play in the process of care. The models share in
common the idea that both health and behavior are influenced by multiple dimensions
(physical, social/cultural, interpersonal) of the environment at multiple levels
(individual, group, community, society), and that these dimensions and levels all interact
(Engel, 1980; Hahn & Kleinman, 1983; Sallis & Owen, 1997). Such conceptualizations
provide a framework for better understanding and describing the realities of everyday
32


life and professional practices (Van Manen, 2001 p. 458) that occur within clinical
practice settings.
Recognizing the substantial amount of mental health care that is provided and
the added complexity of identifying and treating mental disorders in primary care
settings, Klinkman (1997) recently adapted the competing demands model, developed
to frame studies on preventive services in primary care (Jaen, Stange, & Nutting, 1994),
to describe factors impacting the process by which primary care providers recognize,
identify, and treat psychosocial problems (Klinkman, 1997, p. 100). Researchers have
used the concept of competing demands to study depression treatment in primary care
(Nutting et al., 2000; Rost et al., 2000), examining the trade-offs that PCPs make when
faced with multiple, often conflicting demands, which may be the result of differing
agendas between patient and PCP, equally compelling physical and psychosocial needs,
or time and resource limitations (Jaen et al., 1994; Nutting et al., 2000; Rost et al., 2000).
The domains used to capture sources of these demands PCP, patient and
practice environment mirror those used in various models to describe forces at play in
clinical encounters (Clark et al., 1991) and to situate the encounter within the practice
setting and/or broader structural context in which it occurs5 (Crabtree, Miller, & Stange,
2001). Klinkman (1997) broadened the conceptualization of practice environment to
include elements beyond the immediate office environment, such as community
mental health resources available to the clinician (Klinkman, 1997, p. 101) and called
this context the practice ecosystem, the term adopted to refer to the context of
interest for this study. Extensive descriptive research to characterize primary care
settings using direct observation resulted in a model of primary care practice (Crabtree
et al., 2001), adapted on Figure 2-1 to reflect the components of the practice ecosystem
that influence decisions and interactions in clinical encounters.
5 These three domains also mirror those used to frame factors associated with non-recognition of mental
disorders, as summarized in Table 2-1.
33


Adapted from:
Klinkman (1997);
Crabtree & Miller (2001)
Figure 2-1: Ecosystem of Primary Care Practice
Recent research on the process of chronic disease management similarly has
recognized the importance of context on delivery of care (Larme & Pugh, 2001) as
reflected in models emphasizing a multifaceted approach that creates an environment
supportive of patients, providers, and settings in implementing guidelines for effective
care and management (Glasgow et al., 2001; Rundall et al., 2002). This study was
designed to examine factors in the practice ecosystem that influence providers
approaches to providing integrated care for chronic diseases (Chin et al., 2001; Larme &
Pugh, 1998), which demand close and ongoing attention to many dimensions of the
patient, including emotional well-being and health behaviors, as well as control of
physiologic disease factors. Particularly important for this study is consideration of
structural factors, such as funding and health system structure and policy, which not
only limit access to needed services for disadvantaged patients, but also uniquely
constrain the providers working with these patients.
34


Awareness of context also was important for this study because of the strong
association between social and economic conditions and health and illness, both
physical and mental. Experts agree that poverty and race/ethnicity are not only
interrelated, but also interact to produce health disparities (Krieger, 2003; U.S.
Department of Health and Human Services, 2001). Others have suggested that the
same holds true for disparities in health care through mechanisms such as stigma, bias,
stereotyping, and discrimination, which may come into play in clinical decision making,
albeit subconsciously (Clark et al., 1991; Smedley et al., 2002; van Ryn, 2002; van Ryn &
Fu, 2003; U.S. Department of Health and Human Services, 2001). To disregard the
weight of broader influences on decision making and focus only on the process of
recognition would vastly oversimplif[y] the complex problem (Rost et al., 2000, p.
153) of recognizing mental disorders in these settings.
The Social Context of Mental Health and Mental Health Care
These broader influences are rooted in centuries-old societal views and are
reflected in conceptualizations of health and illness in Western cultures. Characterized
by a doctrine of mind-body dualism, which sees the mental as a distinct and
separate domain from the physical, the belief system underlying the theory and
practice of medicine in the United States is so pervasive that it affects not only the
illness experiences of individuals, but also the organization and financing of health care
delivery systems, and thus the work of professionals practicing within them. Referred to
as biomedicine because of its focus on measurable biological (somatic) variables
(Engel, 1977, p. 271), the system of medicine founded and dominant in Euro-
American societies (Hahn & Kleinman, 1983, p. 306) places preeminence on
physiological and biological processes. A reflection of the pervasiveness of this view of
health and illness is the generally accepted use of the term medical to refer to physical
health conditions. Similarly, mental illness, which cannot necessarily be fully
explained in terms of derangement of underlying physical mechanisms (Engel, 1977,
35


p. 272), is not granted the same social legitimacy as health conditions that can be
explained and measured within that framework. Asa result, individuals who experience
mental illness often also experience the fear and embarrassment of having a condition
that is not viewed as real and that is stigmatized by others (U.S. Department of
Health and Human Services, 1999).
The strict conceptual separation of the mental and physical domains has
translated to a model of health care delivery consisting of separately operated and
funded systems of care that mirror that separation6. Physical and professional distance
between practitioners in both systems discourages or even prevents collaboration, and
thus impedes provision of integrated care focused on the whole person (deGruy,
1996, p. 287). Not only does this delivery model place undue burden on individuals by
requiring them to navigate multiple systems to receive care, but it also contributes to
issues of stigma by requiring individuals to seek care in a mental health care specialty
setting and risk being seen and possibly labeled by others. Noting the impact of this
health care system structure on primary care providers, who are the first point of
contact for many individuals with mental disorders, deGruy (1996) argued that
operating within a disintegrated system
... consigns the clinicians in each arm to a misconceived and incomplete
clinical reality that produces duplication of effort, undermines
comprehensiveness of care, hamstrings clinicians with incomplete data,
and ensures that the patient cannot be completely understood. (deGruy,
1996, p. 286).
Another important aspect of the social context of care is its cultural
composition. In the preface to the supplement to his report on mental health, the U.S.
Surgeon General underscored the main message of [the] Supplement that culture
count? (emphasis added) (U.S. Department of Health and Human Services, 2001).
6 In reality, the health care delivery system consists of multiple systems that operate separately from one
another. The mental health care delivery system itself can be conceptualized as consisting of several
systems drat all operate under different funding, institutional, or organizational arrangements, as
discussed in Chapter 5. Substance abuse services are similarly delivered through a distinct system that is
separate from both mental and physical health care systems.
36


Because culture is often proposed, as in the U.S. Surgeon Generals reports, as an
important variable impacting clinical interactions and health behaviors, and because of
this studys focus on delivery of health care for culturally diverse populations, the
conceptualization of culture used in framing this study is an important assumption to
make explicit. The U.S. Surgeon Generals reports broadly defined culture as a common
heritage or set of beliefs, norms, and values that are shared and largely learned by a
group of people, but pointed out that the term is often wrongly understood or used to
generalize about values and beliefs of people who are categorized into broad groups
(e.g., Hispanic), given that not all members grouped together in a given category will
share the same culture (U.S. Department of Health and Human Services, 2001, p. 9).
One such broad categorization is the culture of clinicians (U.S. Department of Health
and Human Services, 2001), or more broadly, the culture of medicine (Good et al.,
2002; Hahn & Kleinman, 1983). Hahn and Kleinman (1983) argued that biomedicine is
a sociocultural system that is socially constructed and that guides societal members
(e.g. patients and healers) in their construals and responses to health and illness (Hahn
& Kleinman, 1983). Acknowledging that such a system exists and exerts influence on
the way providers conceive of and interpret a patients complaints or symptoms as well
as on clinical interactions and decisions, the assumption for this study nonetheless was
the degree to which values and meanings commonly attributed to this belief system are
shared among providers will vary and should thus not be assumed.
An associated is that, just as membership in a group does not guarantee that an
individual subscribes to values and beliefs attributed to that group, individual members
may identify with multiple groups and their associated cultures. As such, this study
focused on analysis of individual experiences and meanings, suspending preconceptions
about the source(s) of such meanings, and looking for patterns that transcend
individuals and persist across contexts.
37


Competing Views of Mental Health and Illness
An important distinction reflected in sociological theory and research around
mental health and illness concerns the contrary ways of comprehending behaviors
construed within society as indicating mental illness (Aneshensel & Phelan, 1999, p.
5) . Aneshensel and Phelan (1999) distinguished between two fundamental views of the
nature of mental illness, contrasting the view that mental illness is a social construction,
with causes that are social, political, and economic, not medical (Aneshensel and
Phelan, 199, p. 8), from the view that emphasizes natural or physical causes for
mental illness, placing it squarely within the providence of medicine (Aneshensel &
Phelan, 1999, p. 6). This distinction is particularly relevant to better understanding the
approaches of those providing mental health care. If a provider believes that a mental
disorder has a physiological, genetic, or chemical base (Aneshensel & Phelan, 1999, p.
6) , the response is most likely to be to treat it as a disease most often through
means such as medication. Clinical uncertainty is likely to be greater for those
practitioners who are sensitive to the potential impacts of social circumstances and
cultural beliefs and practices on the experience and expression of emotional distress.
While the medical model of mental illness has gained considerable scientific
and lay acceptance in recent years, Aneshensel and Phelan pointed out that [n]o
demonstrable organic pathology has been established for most disorders (Aneshensel
& Phelan, 1999, p. 7), and that medical treatments alleviate symptoms but do not cure
the condition. Further, many question the validity of the medical model based on
culturally-based differences in the ways in which signs and symptoms are interpreted
and dealt with in different social contexts (Agbayani-Siewert, Takeuchi, & Pangan, 1999;
Aneshensel & Phelan, 1999; U.S. Department of Health and Human Services, 2001).
Some assert that, in fact, diagnostic criteria for mental disorders in the DSM-IV are
based on prevailing social norms in the United States, and thus describe culture-bound
38


syndromes7 that do not translate well to the experiences of non-dominant cultural or
ethnic groups in this country. Given the diversity represented in CHC patient
populations, this is a critical consideration for understanding providers assessment of
their patients mental health.
Clinical Explanatory Models
An important reflection of the constructed reality of PCPs is the clinical
explanatory model (Hunt & Arar, 2001; Lock, 1995). Developed and tested by medical
anthropologist Arthur Kleinman and colleagues, the explanatory model concept
describes the way in which patients and healers conceive of and interpret illness,
constituting an important influence on responses to those conditions (Hunt & Arar,
2001; Kleinman, 1988; Lock, 1995). Anthropological studies have shown that in the
practical domain of illness, there is often more than one kind of rationality at work
(Hunt & Mattingly, 1998, p. 269).
The biomedical conceptualization of health and illness has predominated in
Western societies, but many thinkers have challenged its relevance to clinical practice
and have proposed alternative models. Hunt and Mattingly (1998) also argued that
[b]iomedical reasoning may be sufficient to explain the bounded realm of microscopic
events and abstract principles, but other kinds of reasoning are necessary when those
principles are applied to the unbounded universe of the real world of physical,
phenomenological, and social lives (Hunt & Mattingly, 1998, p. 270). The most
important challenge to biomedical dogma from within the medical community came
from George Engel, who in 1977 published his seminal essay that argued that many
factors besides biochemical and physiological processes interact to produce disease and
illness (Engel, 1977) among them, social, cultural, political, economic, and
7 The U.S. Surgeon General defined a culture-bound syndrome as clusters of symptoms much more
common in some cultures than others (U.S. Department of Health and Human Services 2001, p. 11),
and discussed the inherent difficulty of universally applying diagnostic criteria, such as that found in the
DSM-IV, that are based on a specific cultures belief system.
39


environmental (Engel, 1980). Engel proposed the biopsychosocial model as a guiding
framework for more completely understanding and addressing health and illness. The
potent explanatory and therapeutic power (Engel, 1980, p. 82) of this model has been
embraced by many general medical practitioners, and is reflected in the definition of
primary care (deGruy, 2003; Donaldson et al., 1996). It is also particularly relevant to
approaches to mental health care, arguing for attention to mental health care in general
medical settings if not direct delivery of much of that care by providers in those
settings as well as arguing against the structural division that continues to distinguish
physical from mental health care.
Although the explanatory model concept has increased awareness of the gap
between patients and healers interpretations of and responses to illness, critics argue
that, as commonly applied, it has hindered as well as helped the realization of a more
humanistic approach to medicine. Margaret Lock (1995) observed that whereas the
patients explanatory model is seen as idiosyncratic, the clinical explanatory model
used in practice by physicians still strongly based in scientific methods and principles
. .is usually taken as the standard to which patient explanations must be adapted
(Lock, 1995, pp. 108-109). The clinical explanatory model is central to this study, and
consistent with Locks critique, assumptions informing the study included the view that
explanatory models of patients and PCPs are distinct and deserve equal scrutiny and
analysis. The explanatory models of PCPs derive in part from their individual
characteristics, personal socialization, and history, and in part from their professional
socialization through medical education into the culture of medicine (Good et al.,
2002; Lock, 1995; Hahn & Kleinman, 1983). Particularly important to this study was to
understand how the PCPs views of and responses to patients, and approach to care,
may be conditioned by sociocultural influences in their background or experience, or
constrained by economic and political factors affecting the practice setting.
While studies have explored providers explanatory models related to chronic
disease care (Chin et al., 2001; Hunt & Arar, 2001; Loewe & Freeman, 2000), few if any
40


have investigated providers views of dealing with the patients psychosocial issues that
either may be a consequence of the disease or interfere with its management through
their impact on patient self-care behaviors, or both (Black, 2002; Katon &
Ciechanowski, 2002; Muehrer, 2002). In-depth studies of primary care providers
approaches to or practices in delivery of mental health care specifically within the
context of chronic disease care are sparse, despite the strong overlap of physical and
psychosocial issues associated with many of these diseases. This study sought to address
that gap, and reflects a belief in the importance of understanding PCPs explanatory
models of both types of issues, and their treatment in primary care settings, to better
explain those practices.
Clinical Decision Making
Experts agree that recognition and/or diagnosis of illness is inherently
problematic and that it involves judgment beyond mere evaluation of clinical evidence
(deGruy, 1996; Downie, Macnaughton, & Randall, 2000; Engel, 1980; Goldstein, 1979;
Hahn & Kleinman, 1983; Snowden, 2003; U.S. Department of Health and Human
Services, 1999; U.S. Department of Health and Human Services, 2001). Symptoms and
diagnoses of mental disorders in primary care settings occur within a matrix of physical
symptoms, disorders, and diseases; other mental symptoms and disorders; and social
predicaments and stressors (emphasis in original) (deGruy, 1996, p. 287). The
reference to medicine as an art (Downie et al., 2000; Hahn & Kleinman, 1983;
Wigton, 2001) reflects the fact that discretion, or judgment, plays an important role in
clinical decision making (Wigton, 2001). What constitutes the art is the need for the
clinician to integrate many decision factors that are situational and which present as
fallible cues (Hahn & Kleinman, 1983; Wigton, 1988; Wigton, 2001), thus requiring
interpretation (Downie et al., 2000; Wigton, 2001). As a result, given the same set of
clinical indicators, clinicians decisions will vary depending upon the certainty with
which these can be applied (Downie et al., 2000; Wigton, 2001) and on how decisions
41


are framed by the situation and context within which they occur (Clark et al., 1991;
Hunt & Arar, 2001). Clinicians must consider both the clinical evidence and constraints
within the practice ecosystem, as well as factors specific to the individual patient and
their circumstances, and then draw not only on technical knowledge and skill, but also
on humane judgement (Downie, et al., 2000, p. 75) to determine what is most
appropriate for a given patient in a given situation.
A potential provider predisposition worthy of attention in this study was the
extent to which providers view emotional distress or other indicators of mental health
issues as normal response to life events (Chew-Graham et al., 2002; Docherty, 1997;
Goldberg, 1992; Goldman et al., 1999; Mechanic, Angel, & Davies Lorraine, 1991;
Nutting et al., 2002; Snowden, 2003) and the medical diagnosis of such responses as
medicalization of these conditions (Aneshensel & Phelan, 1999; Chew-Graham et al.,
2002; Waitzkin, 2000). Clinicians responses to patients psychosocial issues may be
influenced by the extent to which clinicians view these fundamental causes (Link &
Phelan, 1995) of psychological distress as beyond their control, as Chew-Graham and
colleagues (2002) found in their study of depression care in the United Kingdom. That
study underscored the tension between discourse on medicalization of social problems
(Goldstein, 1979; Kleinman, 1995; Lock, 1995; Waitzkin, 2000) or attribution of the
label mental disorder based on perceived deviant behavior (Aneshensel & Phelan,
1999; Goldstein, 1979; U.S. Department of Health and Human Services, 2001), and the
emphasis on the importance of medical treatment to reduce or eliminate the impact and
burden of mental disorders on individuals and society (U.S. Department of Health and
Human Services, 1999).
In investigating meanings providers attach to their experience as mental health
care providers, this study sought to find indications of their social response (Clausen
& Huffine, 1975; Horowitz, 1999) to the behaviors and complaints of patients whose
socioeconomic conditions and cultural and historical conditioning may obscure or bias
the distinction between what is appropriately or inappropriately labeled a mental
42


disorder. Sociologists have observed that persons in certain social categories and
situations are much more vulnerable to having deviant acts formally labeled [as mental
illness] than are others, whether or not they are more likely to deviate (Clausen &
Huffine, 1975, p. 413). At the same time, patient behaviors deriving from these
sociocultural conditions, such as mistrust, reluctance to disclose symptoms, or variation
in presentation of symptoms, contributes to clinical uncertainty and increases the
subjective interpretation component of clinical decision making (Smedley et al., 2002),
as do health behaviors such as substance abuse, which considerably complicates
recognition of mental disorders that may occur with this condition (Goldman et al.,
1999; Snowden, 2003).
The broad distinction is often drawn between two types of theoretical models
of decision making normative and descriptive that are used across diverse
disciplines (Clark et al., 1991; Garro, 1998). Concurring with the preference of many
medical anthropologists for the descriptive paradigm, central to this study was a view
of human cognition that is both flexible and grounded in social and cultural processes
(Garro, 1998, p. 323) and likewise a rejection of the use of normative models, which
presume a utilitarian rationality behind practical decisions (Garro, 1998; Hunt &
Mattingly, 1998). Descriptive approaches see rationality as more sensitive to contextual
influences (Clark et al., 1991, p. 853) and bounded by mechanisms, such as
heuristics, rules of thumb, or reliance on prior beliefs, that decision makers use to
simplify complex problems (Clark et al., 1991; Garro, 1998; Garro, 1998; Smedley et al.,
2002). This aligns with the constructivist position, with its emphasis on the construction
of realities and the reliance of these constructions on context (Erlandson et al., 1993;
Guba, 1990).
Researchers Presuppositions
In choosing a constructivist approach, a researcher implicitly acknowledges the
role that her values and beliefs play in the conduct of the research and interpretation of
43


data. Reflecting on preconceptions and biases about the study topic and recording these
to be revisited throughout the research process is an integral step in this type of
research (Addison, 1999; Guba, 1990). Initial reflection for this study identified the
following preconceptions.
An Institute of Medicine (IOM) report (Smedley et al., 2002) that was released
during conceptualization of this research, as well as provocative commentary and
evidence presented by authors such as Geiger (1996), McKinlay and Clark (Clark et al.,
1991; McKinlay, Potter, & Fledman, 1996), and Van Ryn (2002), among others, spurred
my curiosity about the extent to which provider bias and stereotyping might influence
clinical decision making and thus contribute to documented disparities in health care
received by racial and ethnic minority individuals. Because the literature lacks studies
that explore with sufficient depth the perspectives of providers delivering care to these
populations, I explicitly structured this study to do so. Anticipating that evidence might
surface in this research to support hypotheses proposed in the literature, and
particularly in the IOM report, I also recognized the need for a balanced approach to
ensure that, besides providers beliefs, attitudes and behaviors, the investigation would
examine external forces in the context that could also impact provision of care.
Another important area of reflection for this study, because of its focus on
providers views of and responses to mental health issues, is the nature of mental illness.
Despite awareness of the sociological debate over whether mental illness is socially
caused and constructed or whether it represents a true medical condition, I recognized
my inclination to accept the medical model, most likely fueled, as for many people, by
the apparent efficacy observed with medications that alter brain chemistry in mitigating
symptoms of mental disorders (the remedy authenticates the disease [Aneshensel &
Phelan, 1999, p. 7]). At the same time, however, I had to acknowledge that my
anthropology background would prevent me from fully accepting a biological
explanation for mental illness. A gnawing question that this study promised to
illuminate was the extent to which social, political, economic, and cultural phenomena
44


are viewed as causal by providers charged with addressing complaints typically associated
with mental illness, and how these views might influence the way they address mental
health issues in patients profoundly impacted by these phenomena. In this sense, I
concluded that lack of a clear personal belief around the nature of mental illness is
optimal for this type of inquiry, and an important area for active dialogue with.. .the
researchers own values, assumptions, interpretations, and understandings (Addison,
1999, p. 151) as well as study participants and the data itself.
Finally, my belief that behavior does, for the most part, reflect underlying beliefs
and attitudes, resulted in the decision to give greater weight, in interpreting the data, to
reported behaviors than to stated beliefs.
45


CHAPTER 3
METHODS
This study sought to better understand PCPs approaches to providing mental
health care for adult patients with chronic disease who receive their care in community
health centers. Its broader intent was to discover reasons why mental disorders among
socially and economically disadvantaged adults often go unrecognized or undiagnosed in
primary care settings. Although the research objectives are supported by the literature,
the absence of detailed explanatory knowledge of the basis for clinical judgments and
therapeutic action related to mental health care in these settings justified use of a
qualitative research strategy (Addison, 1999; Hammersley & Atkinson, 1995), which is
also most suitable for exploring and describing the complex interaction of factors
influencing the process of care. The research strategy capitalized on the use of evidence
drawn from multiple sources in a variety of contexts and of multiple techniques to
obtain a rich data set for inductive analysis.
Study methods drew on the strength of the naturalistic approach to inquiry
(Erlandson, Harris, Skipper, & Allen, 1993; Miller & Crabtree, 1999), which stresses the
importance of studying social phenomena in the settings in which they occur
(Erlandson et al., 1993). Emphasizing a context-sensitive approach (Erlandson et al.,
1993; Hinds, Chaves, & Cypess, 1992; Addison, 1999) that look[s] beyond individual
actions, events, and behaviors to the relationship of context to the individual events
(Addison, 1999, p. 151), naturalistic inquiry is well-suited for studying issues and
concerns that find their source in situational contexts where practical action is
necessary (Van Manen, 2001, p. 468). The study approach was experientially focused to
capture the day-to-day realities of clinical practice. The protocol for this study was
submitted to and received approval from the University of Colorado at Denver
46


(Protocol #957) and the Colorado Multiple Institutional Review Board (Protocol #03-
841) to conduct research with human subjects.
Research Design
The research design incorporated both empirical and reflective methods and
procedures (Van Manen, 2001, p. 461) to examine how CHC providers construct
mental illness in caring for their patients and specifically, the ways in which they
determine when and how to intervene to address their patients mental health issues. It
also was designed to identify influences on the process of care in these complex
situations. Empirical methods included interviews to obtain descriptive, experiential
accounts from providers and observation to better understand the day-to-day
environment in which providers interact with patients and make clinical decisions.
Reflective methods consisted of active dialogue with the research participants;
colleagues and advisors; emerging themes and patterns; and my own values,
assumptions, interpretations, and understandings (Addison, 1999, p. 151) throughout
the process.
Figure 3-1 describes the interaction among the elements of this design
throughout four phases of its execution. Each element is described in more detail in the
discussion that follows.
47


PHASE 1:
Orientation
Brief, recruit PCPs
Plan, schedule
w/key contacts
Conduct baseline
interviews
Pilot core interview
technique
Core interviews
PCPs
Debriefs
CLINIC 1
Observation
Recruitment
Interviews Non- PCPs

Setup Key Contact Wrap-up
CLINIC 2
CLINIC 7
PHASE 2: In-Clinic Data Collection
PHASE 4: Verification
Reviews: Group Reviews:
Key Contacts PCPs
KEY
Process Adjustment
Data
Figure 3-1: Research Design
Sampling Strategy. Study Sites, and Participants
The research design incorporated a purposive sampling strategy (Patton, 2000;
Erlandson et al., 1993) that sought maximum variation (Kuzel, 1999; Erlandson et al.,
1993) in both context and provider characteristics, with the goal of obtaining data that
would be information rich (Erlandson et al., 1993; Kuzel, 1999). Seeking the
broadest range of information and perspectives (Kuzel, 1999, p. 39) possible helped
48


to characterize the variation and identify patterns and themes that persisted in PCPs
approaches despite this variation.
Seven community health center clinics in the Denver, Colorado metropolitan
area served as study sites. These included two clinics in the vertically integrated health
care system that provides care to the citys low-income and uninsured population, and
five climes that serve similar populations in the citys suburban and nearby rural areas -
three in one clinic system and two in another. The three systems represent a range of
structural arrangements (e.g., funding, availability of resources, community support),
whose features constrain or enable PCPs latitude in delivering care and hence influence
clinical judgments. These differences mirror the diversity that exists among the health
care facilities that fall under the rubric of community health center (Lewin & Altman,
2000). Including a variety of settings helped identify the range of contextual influences
on clinical decision making, allowed identification of common patterns that persisted in
diverse settings, and increased the potential range of PCP characteristics represented
among participants. Further variation derived from site-specific organizational culture
and values, as well as from location-specific differences in population needs in the
communities served.
The primary participants were family practice and internal medicine primary
care providers, including physicians (MDs), nurse practitioners (NPs) and physicians
assistants (PAs), whose patient panels included adults with chronic disease. One PCP at
each study site functioned as the key contact, facilitating on-site research activities
and providing access to clinic staff and PCPs, as well as providing insight, such as
identification or interpretation of culture-specific phenomena (Gilchrist & Williams,
1999), not directly accessible through the other planned research activities. To minimize
the impact of the research in the clinic settings, medical directors helped to determine
the way to enlist these individuals that was most suitable to each system. The preference
of one clinic system was for the clinic directors to perform this role and participate in
the initial set of interviews (described later), primarily because these individuals had
49


dedicated administrative time and thus more flexibility to participate. In others, the role
was presented to all the providers as an optional form of voluntary participation. Five
of the seven key contacts were either clinic directors or team leaders in their clinics. All
key contacts, including those with administrative responsibilities, saw patients for a
majority of their time, so their perspectives were anchored in practical clinical
experience.
Forty-five PCPs, including the seven key contacts, participated in one or more
interviews, and another three PCPs participated in verification activities (described later
in this chapter), resulting in participation by 48 PCPs. Table 3-1 describes the
characteristics of PCPs who participated in the study.
Table 3-1: Characteristics of PCPs
Characteristic n (%)
Occupation MD 30 (63)
NP 13 (27)
PA 5(10)
Specialty Family Practice 29 (60)
Internal Medicine 19 (40)
Sex Male 22 (46)
Female 26 (54)
Total 48
Interviews with 12 non-PCP staff, including social workers, psychologists,
counselors, and clinic administration personnel, provided insights to complement the
perspectives and reports of PCPs, bringing the total number of individuals who
participated in the study to 60 (48 PCPs and 12 non-PCPs), as detailed in Table 3-2.
With an overall participation rate among adult care providers of 74%, these
demographics are a good reflection of the characteristics of PCPs working in these
clinics.
50


Table 3-2: Participation by Clinic System
Characteristic Clinic System A Clinic System B Clinic System C
Occupation PCP 18 18 12
Non-PCP 5 5 2
Sex Male 10 6 10
Female 13 17 4
Total 23 23 14
Research Phases
As illustrated on Figure 3-1, during the orientation phase, I visited each clinic to
brief key contacts and interested clinic staff on the purpose of the research, explain
what it involved, and solicit participation. Scheduling
and planning for in-clinic data collection activities
occurred during this time as well. Also during this
phase, I collected baseline data and conducted pilot
interviews to test the core interview guide and process.
Twelve PCPs participated in baseline interviews, and
four key contacts participated in pilot interviews. All three clinic systems were
represented in roughly equal proportions in interview participation during this phase.
Insights from the baseline interviews helped inform and refine activities planned for the
studys second and main data collection phase, as did pilot interviews with key contacts
from each of the three clinic systems.
In the in-clinic data collection phase, I spent an average of 28 hours at each of the
seven clinics observing activity and conducting interviews with both PCPs and non-
PCP staff, resulting in a total of 198 hours of research
within clinic settings, during which I conducted a total
of 55 interviews. Each clinics observation period
spanned two consecutive weeks. I pre-scheduled
interviews with PCPs and, as recommended by the pilot
51


interviews, previewed the interview protocol with them so that they could prepare in
advance. Also while in the climes, I was able to identify other non-PCP staff who played
key roles in delivery of mental health care to patients, and to schedule and conduct
interviews with these individuals. With permission of interviewees I recorded interviews
for later transcription.
Analysis began in tandem with the main data collection phase and continued
throughout the remainder of the project, with initial activity focused on gleaning
insights from the baseline interviews to sensitize myself
to factors and phenomena to further investigate as I
proceeded with subsequent data collection. This phase
also overlapped and interacted with the verification phase
in that feedback from reviews was used to refine and
finalize findings.
Design for Trustworthiness
The research design incorporated several mechanisms aimed at ensuring
trustworthiness of the study throughout. Introduced by Lincoln and Guba, the
concept of trustworthiness describes the basis for assessing naturalistic inquiry
analogous to that for assessing other scientific inquiry that follows the positivist
paradigm (Erlandson et al., 1993). According to this concept, a valid inquiry must
demonstrate its truth value, provide the basis for applying it, and allow for external
judgments to be made about the consistency of its procedures and the neutrality of its
findings or decisions (Erlandson et al., 1993, p. 29).
As recommended by Erlandson, et al. (1993) to enhance credibility, or truth
value, I scheduled observation to allow me to observe normal clinic activity during as
many of each clinics days and hours of operation as possible, thus reducing the chance
that observations would reflect phenomena that were situational. Triangulation of data
sources (PCPs, other clinic staff) and methods of obtaining the data (observation,
52


different types of interviews) also increased credibility. Variation in interview techniques
allowed exploration of themes and patterns in individual PCPs approaches across a
broad range of patient care situations and helped triangulate information by capturing
PCPs perceptions and self-described behaviors both based on abstract reflection and
grounded in descriptions of real patient encounters, as did obtaining reports of
situations that had occurred both very recently and in the past (see data collection
techniques). Applicability, another quality contributing to trustworthiness, (Erlandson et
al., 1993), was also designed into the research through use of a purposive sampling
(Patton, 2000) strategy that maximized both the depth and range of data collected
through achieving the greatest possible variation in context and PCP characteristics.
Checking in frequently with key contacts throughout the on-site data collection
phase, while constituting one element of the reflective dimension of my research
approach, also allowed me to test my ongoing interpretation and associated emergent
themes and patterns with individuals who were familiar with the research but not
immersed in the day-to-day research activities. Besides the dialogue that occurred during
the in-clinic data collection period, upon completion at each clinic, I debriefed with the
key contact to clarify my observations and to discuss the research process. This
technique for assessing trustworthiness is known as member checking (Erlandson et
al., 1993), and was central to my approach in the verification phase of the project (see
discussion on verification).
Finally, thorough and complete record-keeping to maximize accessibility of
project data for audit or review, and maintenance of a project journal to document
logistics and reflexive notes throughout the process further contributes to
trustworthiness by enhancing consistency and neutrality of the research (Erlandson et
al., 1993). Along with notes on observations, I kept a separate journal to record insights
as they emerged, and noted what seemed to be working and not with the research
process, as well as adjustments to the process as I made them.
53


Data Collection Techniques
A combination of different interview techniques and observation in clinic
settings produced data for addressing the studys aims. Interview techniques included two
types of semi-structured interviews lasting 45 to 60 minutes and informal 10- to 15-
minute debriefs with PCPs, and 30- to 45-minute semi-structured interviews with
non-PCP staff. The semi-structured format was chosen to minirrme restrictions on
responses while ensuring a level of comparability across respondents through the use of
a guide that provides consistency in questions that are posed and the order in which
they are asked (Bernard, 1994). Appendix B contains interview guides and Figure 3-2
illustrates how the various data collection techniques fit within the research design. The
first type of PCP interview, conducted in the project initiation phase to provide baseline
understanding of the phenomenon central to the study aims, explored PCPs views of
and perceptions about mental health care in the context of chronic disease management,
their self-described approaches to this care, and factors they identified that impact their
ability to provide this care. The second type the core technique was used in the in-
clinic data collection phase, and consisted of semi-structured interviews that explored
how PCPs respond to competing psychosocial and biomedical demands in complex
chronic care visits by asking for detailed descriptions of specific situations.
Based on the principle that a persons past behavior in a given situation will
predict their future behavior in similar situations (Anderson & Wilson, 1997), the core
technique derived from the critical incident technique (CIT), which originally was
developed by a military psychologist to identify flight crew behaviors in World War II
(Kemppainen, 2000; Chell, 1998; Anderson & Wilson, 1997). CIT has since been
applied for a variety of purposes, mostly in business and industry for job analysis and
performance assessment (Anderson & Wilson, 1997; Chell, 1998). More recendy it has
been applied in health services research to studies of health care processes and service
quality (Kemppainen, 2000).
54


Core interview pilots '
PHASE 1:
Orientation
Baseline
interviews
Core interviews
PCPs
Debriefs
CLINIC 1
Observation
Recruitment
Interviews Non- PCPs

Setup Key Contact Wrap-up
CLINIC 2
CLINIC 7
PHASE 2: In-Clinic Data Collection

Figure 3-2: Data Collection Techniques
CIT and its variations are used to analyze specific behaviors in defined situations toward
evaluating or predicting analogous behaviors in similar situations. The technique
involves asking the respondent to identify an incident that meets criteria relevant to the
topic of investigation a key one being its criticality in determining a particular outcome
and then to recount specific information about the incident (Kemppainen, 2000).
Although the accounts are retrospective and may be in the distant past, recall is
facilitated by the focus on an event that the respondent identifies as memorable and
important (Chell, 1998). Analysis of data gathered through this technique reveals key
factors and processes that are problematic as well as problem-solving.
For this study, I adapted a variation of CIT called the behavioral event
interview (Chell, 1998) to elicit PCPs descriptions of their experiences in providing
55


integrated care. An important adjustment to this approach based on the baseline and
pilot interviews was to explore situations rather than incidents with participants, which
acknowledged the strong indication in these early interviews that it most often takes a
series of visits for the PCP and patient to conclude that mental health issues exist and to
initiate steps to address them. Each participant identified and reflexively recounted two
or more memorable situations in which a patient exhibited evidence of psychosocial
issues during care for chronic disease at least one that resulted in diagnosis/treatment
or referral for a mental disorder, and one that did not. The two types of scenarios were
designed to allow exploration of the uncertainties surrounding recognition of mental
disorders as well as decisions influencing courses of action. An advantage to this
technique is that it is context-rich (Erlandson et al., 1993) and particularly suited to
delving into the PCPs perspective in that the context is developed entirely from the
subjects perspective (Chell, 1998, p. 55). Using a semi-structured format allowed me to
probe into the respondents thoughts, feelings, and rationale for their actions and their
assessment of how these impacted the outcome. Providing the scenarios to PCPs in
advance of the interview, as suggested by the pilot interview experiences, helped them
better identify relevant, specific situations to discuss. Responses illuminated how PCPs
frame and structure the process of care in complex encounters, how they interpret and
respond to psychosocial issues in these situations, and how they craft and rationalize
their actions and decisions. PCPs accounts also disclosed factors that influence their
clinical decisions.
The third type of interview administered to PCPs took the form of a brief ad
hoc discussion I referred to as debriefs, which were designed to capture reflections on
the days encounters while details were still fresh. The need to engage in these
discussions on the spot reduced my opportunity to capture them in audio recordings,
so while several such discussions occurred, only nine were recorded and transcribed,
with the rest documented in observation notes.
56


Drawing on the approach used by Loewe, et al. (1998) to investigate the
attitudes and practices of clinicians providing diabetes care, the combination of
interview techniques chosen for this study was designed to explore providers
perceptions and self-described behaviors in more than one way. Loewe and colleagues
used semi-structured interviews with providers in contexts removed from clinical
processes and direct observation of interactions between providers in precepting
sessions (Loewe, Schwartzman, Freeman, Quinn, & Zuckerman, 1998; Freeman &
Loewe, 2000). This variation also allowed these researchers to explore how providers
framed their discourse about patients differently with the researchers than with other
providers (Loewe et al., 1998).
A second type of data collection consisted of in-clinic observation and interaction
with clime staff, both through informal discussion and in a more structured manner
through interviews. Fundamental to ethnographic field work, participant observation
is a relatively unstructured way to collect data in naturalistic settings by which the
researcher observes and interacts with the people being studied in context of their day-
to-day activities (Dewalt, Dewalt, & Wayland, 1998). This form of data collection can
vary in both degree of structure (Miller & Crabtree, 1999) and degree of
participation (Miller & Crabtree, 1999; Dewalt et al., 1998; Creswell, 1997). The form
this method took for this study was unstructured background observation (Miller &
Crabtree, 1999, p. 18) in that I primarily observed and occasionally interacted with PCPs
and clime staff (Dewalt et al., 1998), and recorded observations in notes rather than
using structured instruments, such as checklists or rating scales (Miller & Crabtree,
1999).
In-clinic observation provided data on the practice ecosystem (Klinkman,
1997), a term adopted for this study to refer to the environment in which care is
provided and to emphasize the multiple factors at multiple levels that influence clinical
practice. I did not directly observe clinical encounters, but focused instead on activity
and interactions occurring among PCPs and other staff outside the exam room to
57


follow through on patient care. Precepting interactions with residents and medical
students in some of the clinics presented an unanticipated opportunity to hear how
MDs instructed less experienced physicians on this facet of care. Besides contributing to
my understanding of the environment within which the process of care occurs,
assessing the practice ecosystem helped identify barriers to and enablers for provision
of mental health care in these settings, as well as organizational factors that influence
PCPs approaches to patient care. Observations helped confirm or augment PCP
reports obtained in interviews. Anchored in the natural daily flow of activity,
observation helped capture meaning that is expressed in action and practices as well
as verbalized by PCPs (Addison, 1999, p. 150).
Spending time in each clinic provided two unanticipated benefits. First, my
presence in the clinic generated curiosity and interest among PCPs who had not
expressed an interest in participating in the study during the project initiation briefings,
so overall participation increased as a result. Similar interest among other clinic staff
resulted in their sharing unsolicited insights, which I also had not expected. For the
most part, I was able to obtain the clarification or additional information I needed from
non-PCP staff without the need to pre-schedule time with them. Second, more
opportunities arose to interview other clinic staff members who address patients
psychosocial needs because of my access to these individuals during those periods.
Qualitative Analysis
Study data, consisting of text (field notes and transcribed interviews from audio
recordings), were interpreted using a combination of techniques for systematically
examining such data. The primary analytic technique most resembled the process
termed immersion/crystallization by Crabtree and Miller (1999) and described by
Borkan (1999), which consists of multiple systematic, in-depth readings of transcripts
and notes, each focusing on different aspects or details of the content (immersion), and
58


successive refinement of the interpretation of those data (crystallization). Figure 3-3
illustrates the analytic process used to derive study results.
Immersion
Verify transcripts
Template coding
Detailed coding
Comprehensive
reading
u
rystallization
J
Unifying theme
Analytic tools
1
Preliminary
findings
immersion
Summarize
situations
Selected
depth coding
Search for confirming/
disconfirming evidence
Figure 3-3: Approach to Analyzing Study Data
The core interview technique called for sampling to the point of redundancy
(Kemppainen, 2000). Thus, the goal was to conduct these interviews with as many
PCPs caring for adults with chronic illness in each of the study sites as needed to
adequately search for disconfirming cases (Kuzel, 1999; Patton, 2000) and verify
consistency of findings. General guidelines suggested that 20 to 25 interviews would
support this studys methodology (Kuzel, 1999). This studys data set included
transcripts of 36 core and 12 baseline interviews with PCPs, as well as transcripts and
59


notes from debriefs, and transcripts and notes from another 24 interviews conducted
with key contacts and non-PCP staff.
Using a process that was much more iterative or recursive than linear
(Borkan, 1999, p. 183), I examined the data several times and in multiple ways,
alternating between horizontal passes through the data reading transcripts from
beginning to end and vertical passes focusing on specific information within each
transcript (Borkan, 1999). Although in reality, crystallization occurs throughout, Figure
3-3 shows a discrete crystallization step that facilitated the transition from a more
broad, exploratory analytic process that produced a conceptual model to a more
concrete and focused analysis based on that model.
As prescribed by this technique, crystallization began before the first data were
collected, as I recorded a priori biases [184] that might influence my research and
interpretation of the data (Borkan, 1999). These are summarized in Chapter 2 with
other background information that informed this study. With these in mind, and also as
described for this technique, I specifically set out in the project initiation phase to
identify initial patterns and themes in the baseline interviews that would inform my
subsequent work.
I made the first horizontal pass through the data in the process of verifying the
accuracy of each transcript against the recorded data, which provided the opportunity to
begin to detect broad themes and patterns. Also during this process, I sanitized each
transcript, removing text that could reveal the identity of the speaker, the clinic system
or clinic location s/he was affiliated with, or any other individuals discussed or
mentioned in the interview. I also assigned coded identifiers at this point to obscure the
identities of participants. Because transcription occurred as data collection continued,
the process of verifying transcripts continually updated and further crystallized insights
about these broad themes and patterns, but the pass through the baseline interviews
provided the initial insight I sought.
60


After completing the in-clinic data collection activities, I began to immerse
myself in the data in a different way by using the qualitative data analysis software,
Atlas/ti which aids the process of assigning and managing codes assigned to text
segments to perform the vertical passes through the data. Given the number of
interviews in the data set, and for efficiency, I divided the interviews into two groups
that were matched on key characteristics (i.e., clinic systems represented, PCP gender,
occupation, and specialty), such that each groups profile matched that of the overall
groups as closely as possible. I used the first group for coding and reserved the second
as the basis for searching for confirming and disconfirming evidence.
The software allowed me to chunk the data into manageable pieces and to
identify and compare related text segments across all interviews for example, all
passages discussing barriers to recognition. This permitted more in-depth examination
of subsets of similarly coded text segments in subsequent steps in the analysis to identify
themes and patterns related to specific phenomena. I created a template (Crabtree &
Miller, 1999) for decomposing the interviews into segments that answered specific
interview questions and/or mapped to study aims. This template consisted of several,
relatively broad a priori codes based on types of information I had specifically set out to
obtain. For example, the code CUES was used to identify all passages in interviews in
which the participant discussed cues that prompt recognition and diagnosis of mental
disorders (Aim 1.3). After coding interviews with the template codes, I re-examined
these data to identify conceptual themes and patterns in the text, this time setting aside
the template categories and working with the text as a whole. This produced a second
set of codes I referred to as thematic codes, and I was able to begin to appreciate the
strength of various concepts in the data through the number of text segments that
exemplified a given concept. A more important measure of this strength was the
number interviews with different participants that contained any particular concept, and
the settings represented in those interviews.
61


Following this immersion into the data and noting insights from this
exploration, I then returned to a higher level view to make another comprehensive,
horizontal pass through the entire set of interviews. Through these various passes, a
central organizing theme (Addison, 1999, p. 155) emerged, which helped focus
further interpretation. This initial engagement with the data also suggested a scheme for
further organizing the data for systematic examination, which guided the next steps in
the process. Specifically, it became clear that PCPs approaches to mental health care
were highly adapted to constraints that were situation-specific, and that the most
meaningful characterization of those approaches would be achieved through careful
examination of situations rather than attempting to categorize or typify the general
approaches of individuals. The focus of the analysis then shifted to the situations
described by PCPs in the interviews.
The data suggested that examining those situations relative to two aspects of
mental health care recognition and response would be most helpful in characterizing
PCPs practices and behaviors, as well as discovering factors that influenced them, and
that a PCPs behavior in any given situation could be described by points along
continuums that described the range of potential behaviors related to these two aspects.
In addition, because specific behaviors around recognition of mental health issues did
not seem to associate with particular responses to those issues, examining each
separately, then mapping behaviors to the intersection of the two, offered the best
scheme for categorizing the approach used in any particular situation. Chapter 4
describes these continuums and approaches defined by their intersections in greater
detail.
Figure 3-4 illustrates the tool used to characterize the approaches PCPs
described in each of the situations they recounted in the interviews.
62


05
o
2
O
s
CD
O
O
LU
CC
(0
cc
to
0.
Low-
No Risk
RESPONSE
High
Risk
Figure 3-4: Analytic Tool for Characterizing PCPs
Approaches to Mental Health Care
Referring to this categorization scheme, I summarized the 83 situations
described in the various interviews to allow further analysis and comparison, using a
spreadsheet template to consistendy document and display specific details about the
situations and the associated PCP behaviors. Because not all interviews yielded
sufficiendy detailed situation descriptions, the ones analyzed were drawn from 28 of the
core interviews and 5 debriefs representing the experiences of 29 PCPs. Table 3-3
contains some situation summaries chosen to illustrate this data organization scheme.
By including fields, such as clinic system, I was able to more easily compare various
aspects of the situations across individuals and settings, and to ensure that patterns
persisted across settings.
63


Table 3-3: Situation Summaries
p#, Situation# Clinic System "Zone" Medical status Factors, Circumstances Mental health diagnosis PCP's Thoughts, Comments
PI, 1 1 3 > 8 -> 5 Patient complains of long- term sinus infection, GERD, back pain multiple complaints Transferred from another clinic location; previously seen in CMHC and they "fired him" for non- engagement in Tx program; patient is uninsured "horrible" anxiety disorder (previously diagnosed, but PCP didn't have that info at first) Patients come in for a medical opinion, so that's where PCP always starts. Believes patient needs CBT but thinks maybe patient has not been given the same priority at CMI-IC as Medicaid patients might.
PI 8,1 1 3 As yet undiagnosed; complains of difficult walking, shortness of breath; has presented to ER w/symptoms; patient initially thought he had MS 45 y/o male living w/parents who are "frustrated at continuing to support him." Patient disclosed psychiatric history. Patient is uninsured, but can get tests for clinic visit co-pay. Specialist referral not an option. PCP suspects that "psychiatric issues could be the major thing going on with him" probably depression but doesn't suspect psychosis Senses patient's resistance to MH Dx; also senses patient may have "some investment" in his illness though says patient may be somatizing and not malingering. PCP suspected Mil basis for complaints "fairly early on" but "didn't want to just dismiss that there might be something real that was going on." Believes that somatization should be "diagnosis by exclusion"
PI 8, 6 1 V 00 Presented w/back pain Patient had MH Dx and "needed the resources to be on the medication that he needed" Major depression; recendy diagnosed with bipolar (not by PCP) PCP was very vocal about system- and society-wide issues that create these situations
31,3 2 8 Diabetes (uncontrolled); uncircumscribed penis w/infection in pain for 2 months waiting for Tx Male in 30's; uninsured and unable to pay co-pay for one-time consult w/psychiatrist. Patient is ".. .constantly hounded by auditory hallucinations of a negative manner and used marijuana to make them go away. Also "very hypersexual" and unable to have sex because of pain from the infection. Decompensated paranoid schizophrenia ".. .his schizophrenia was so.. .teetering on the edge, and his pain was so bad, that I didn't even have time to attend to his diabetes, really..." "But I don't know what more.. .with my level of training, I could have done." This situation was "frustrating" because patient wanted Tx for mental disorder and couldn't get it.


Table 3-3: Situation Summaries (Cont.)
P20, 1
3
2 > 9
Prior to incident, seen only
for acute care; has
hypertension and
dislipodemia "and probably
someday will soon be a
diabetic."
Female, 20 y/o; married
w/children; undocumented.
Had been seen psychologist
for issues w/marriage.
Psychologist hadn't taken
any action follow-up was
planned
Psychosis
Though PCP is cautious about
placing people on psych hold, the
need here was clear cut. Patient had
plan for suicide and may have taken
others with her. Didn't have a psych
note on patient may not have had
chart when she showed up. Thankful
that patient is compliant on psych
meds, but "this woman truly does
need to see a psychiatrist...somebody
truly who understands psychosis"
P#, Situation# Recognition Response Outcome Comments
PI, 1 Concentrated on medical
"... he does this classic thing that always alerts me to those sort of things and they switch...they switch every few sentences to another topic. Seems to be a concentration sort of issue." Also, patient has complaints that nothing seems to remedy, has multiple complaints, and is a frequent flier. issues first did extensive work-ups and tests to exclude organic causes for complaints. Based on cue (pattern of patient's complaints), talked to previous PCP and MU specialists (upon learning of MH Dx). Tried all anxiety meds finally Rx for benzo, and looking for STUCK: Patient won't accept MH diagnosis and is high utilizer PCP continues to instruct patient that he needs to see a psychiatrist. PCP is trying to keep patient "in some sort of control" via benzos to reduce patient's utilization (clinic, ER) PCP observes that patient "needs to see somebody, and we've left the door open to him..All PCP can do is continue to urge patient to see counselor
community resources.


Table 3-3: Situation Summaries(Cont.)
PI 8, 1 Can't find "medical etiology" for patient's complaints, but hasn't yet ruled out every possibility. PCP aware of social stressors (living w/parents who aren't happy about it). Patient's presentation is "not really part of a pattern that I recognize." PCP has suspected MH issues from early on. PCP has done exhaustive workups and says s/he hasn't "given up on finding something organic" is considering testing for some rare conditions. Unless patient presents with acute complaints, has "left it up to him to follow up." Put patient on TCA for pain and also suspected depression. STUCK: Plans to rule out rare conditions, then try to talk to patient about depression "but he's very resistant to that" PCP says s/he often prescribes antidepressants for chronic pain, knowing it's often accompanied by and exacerbated by depression. Seems to use this to hedge his/her bets when s/he's unsure about depression and doesn't want to confront patient
PI 8, 6 N /A patient carried Dx Handed the bipolar meds situation off to the in-clinic psychosocial worker, who directed the patient to the CMHC. When patient reports to PCP that it will be a 2-year wait to get into CMI IC, PCP decides to start the patient on meds for bipolar. Status TBD Not clear how patient received the Dx but apparently did not have a prescription the PCP could carry forward
31,3 N/A patient carried Dx; PCP was aware of physical condition that caused patient's condition to decompensate While patient was in pain (until infection was resolved) "I was just scrambling...to try to keep him together." PCP "[added] antipsycho tics to his regimen" which worked "until the pain got so bad that the voices would come back" LOST/TICKING TIME BOMB: Finally found a way to get patient into psychiatrist, but unable to reach patient, who w/wife had fled the state due to criminal charges. "I dont know what happened to him."
P20, 1 A friend brought patient in because she was exhibiting very bizarre behavior. Patient shared her belief that she was appointed to sacrifice herself to 'save' the people of the community. Called police to have patient committed. Now manages patients antipsychotic meds that were prescribed by hospital (clinic was able to obtain what was prescribed). TICKING TIME BOMB: Patient is doing well on meds. Patients chronic conditions arc "pretty poorly controlled" but PCP doesn't want to "push her on that" for fear of'losing her. PCP knows that eventually the family will move (migrant work) and she may not have access to meds any longer. This situation is "a huge concern" to PCP, who recognizes that with mental illness uncontrolled, patient was suicidal and possibly homicidal.


As this example indicates, I also noted movement from one mode of operation
represented by numbered squares in the figure and labeled zone in the table to
another within the same situation. These instances helped me assess how changing
circumstances impacted PCPs decisions and behaviors. As constructed, this tool also
provides the basis for future analysis that may examine differences related to PCP
characteristics such as experience level, gender, or occupation. Besides summarizing
pertinent details about each situation, I represented in PCPs own words as much of
the detail as possible.
Also during this second major period of immersion, I returned to coding to
gain further clarity around two specific aspects of PCPs approaches: the cues they use
in recognizing mental health issues, and the challenges they encounter in both
recognizing and responding to the mental health needs of their patients. I coded all of
the interviews for these details to allow me to confirm or disconfirm the evidence I
had found in the first group of interviews, and in the process, refined my
understanding of these specific phenomena. Summarizing the situations across the
entire set of interview data also provided the opportunity to test my initial
interpretations. Finally, comparison between the phenomena suggested by the
situational accounts and the more general descriptions provided in other interviews
provided another opportunity to test for confirming or disconfirming evidence in the
interviews and to triangulate findings (Erlandson et al., 1993; Borkan, 1999).
From these analysis steps, I produced a summary of preliminary findings in
presentation format to review with colleagues, advisors, and PCPs from all study sites
to verify accuracy and neutrality of my interpretations and conclusions.
Verification
In the verification phase, I tested viability of my findings (Erlandson et al.,
1993) in individual and group reviews, which produced feedback from 20 individuals,
including 13 PCPs from participating clinics, 3 of whom had not participated in the
67


interviews. Reviewers included academic advisors, colleagues, key contacts, and PCPs.
I first reviewed findings with advisors and colleagues to refine and clarify the
presentation for review with PCPs. I then reviewed findings with key contacts in
individual meetings, and with other PCPs in two group reviews, each conducted at
study sites within different clinic systems and each including PCPs who did not
participate in the interviews as well as those who did. The reactions of these PCPs
affirmed broader applicability of the findings. While the findings were enthusiastically
received and endorsed by most reviewers, comments and discussion in the reviews
helped further crystallize the findings. I was thus able to use this feedback to fine tune
as well as to ensure the credibility of the study findings.
Limitations
Although PCPs from the study settings endorsed the findings as accurately
representing the reality of their practices, strict applicability of this studys findings is
limited to these specific settings. Purposeful selection of study sites and high individual
participation rates argue for logical extension to similar clinic settings in the
metropolitan area operating within the same state-wide practice ecosystem. Findings
are less applicable, however, outside the state context that establishes funding levels
and health care system structure. Similarly, while this study included high-need areas in
the state, it did not include the high-need area in which the fewest services are
delivered, so findings do not reflect the full impact of mental health care resource
constraints in the state.
The recruitment strategy for individual PCPs created potential bias toward
those who felt more strongly about this aspect of care, and did not necessarily produce
a participant pool representative of all PCPs working in community health centers
state-wide. High participation rates contributed to the intended variation in gender,
experience level, occupation, and specialty, however, and indicated generally high
interest in the issues being investigated, thus reducing the threat of this selection bias.
68


Finally, because PCPs were asked to recount memorable situations, many
described extreme cases, which are not necessarily indicative of typical mental health
care activity in these settings. To temper interpretation, I explicidy asked each PCP
how typical the situation was of their practice and to describe how each differed from
the norm.
69


CHAPTER 4
PCPS APPROACHES TO MENTAL HEALTH CARE
FOR SOCIALLY AND ECONOMICALLY
DISADVANTAGED ADULTS
WITH CHRONIC DISEASE
Primary care providers approaches to mental health care appear to come in
nearly as many varieties as there are PCPs and as many more as there are situations that
each individual encounters in his or her practice. This uniqueness may derive in part
from the need to address maladies that involve the psyche, for which, in contrast to
physical ailments, theres no hard and fast rules//theres no labs that you can draw
[P2], The adage that medicine is an art applies especially to these non-organic
aspects of patient care those that draw heavily on the providers natural inclinations
and interactional skills, which are based in backgrounds, experiences, knowledge,
attitudes, and beliefs that are unique to each individual. Further contributing to this
variation are differences among the environments in which PCPs function day to day,
which determine the options available for therapeutic action.
Acknowledging this variation, this studys goal was to discover patterns in
thoughts and behaviors reported by a variety of providers working in CHCs that might
help illuminate their approaches while highlighting factors that seem to influence their
choices and actions. Insights into characteristics of these approaches came indirectly
from PCPs descriptions of specific situations they had experienced and more directly
from their general descriptions of how they conceive of, recognize, diagnose, and treat
mental disorders in their patients. This chapter describes the range of approaches
reflected in the data using an analytic scheme that the data suggested would best frame
those descriptions.
70


As noted in the discussion of methods in the previous chapter, the data
demonstrated that, rather than adhering to a single, preferred style, most of these
PCPs adapt their approach to best address the situation at hand. Consequently, the
descriptions in the following sections do not attempt to categorize individual PCPs
according to a style or type of approach, but rather to identify situation-specific
influences on clinical decisions and actions taken in providing mental health care in
these settings. While approaches are categorized for the purposes of this discussion,
these categories are used to describe behaviors reported by PCPs in specific situations
and to represent the range of behaviors reported by PCPs in this study.
The following sections introduce and explain the analytic model; provide
detail about PCPs reported thoughts and behaviors in recognizing and responding to
the mental health needs of their patients; discuss influences on the provision of
mental health care in these settings; identify conditions that prompt PCPs to adopt a
particular approach in a given situation; and discuss consequences PCPs must
consider in choosing a course of action.
Characterizing PCPs Approaches to Mental Health Care
Analysis of the interviews suggested that examining two dimensions in
particular recognition and response would be particularly helpful in characterizing
the approaches, and would provide a basis for comparing and contrasting the ways in
which different individuals behaved in similar situations or the same individual behaved
in different situations. These dimensions also promised to be useful in identifying what
might account for variability. In isolating and examining recognition separately from
response as well as in combination, this scheme also supports the studys focus on
recognition of mental disorders through the framing question: Why do mental
disorders among socially and economically disadvantaged adults often go unrecognized
or undiagnosed in primary care settings?
71


The wide variability in approaches reflected in these data suggested that PCP
behavior would fall along a continuum in each dimension. In the recognition dimension,
stories indicated that PCPs operate along a passive-to-active continuum, from not
noticing cues or ignoring suspicions that psychosocial issues need to be addressed, to
proactively seeking information on the patients social circumstances, state of mind,
concerns, and so forth, without being prompted to do so by the patients behavior,
complaints, or symptoms. PCPs descriptions of their responses to mental health needs
suggested a continuum reflecting their inclinations around risk-taking, which ranged
from risk-avoidance to risk-taking. Specifically, PCPs seemed to vary in their willingness
to try to address issues they felt were beyond the scope of their expertise.
PCPs stories also suggested that actions are not always consistent with their
stated beliefs and attitudes, and that behaviors in one dimension are not always
intuitively consistent with those in the other. For example, PCPs who reported
proactively seeking information to help assess their patients psychosocial situations in
order to integrate the whole person [P45] perspective into the health care they
provided did not always actively deal with issues that surfaced. Some would instead
respond by referring the patient to a counselor or psychosocial worker1, preferring to
have someone else address those issues so that the PCP could concentrate on the
patients medical 1 2 care. In these cases, while the biomedical focus seemed to belie a
biopsychosocial orientation to patient care, PCPs use of these services most often
represented a pragmatic choice given time limitations. Similarly, some PCPs recounted
situations in which they did not routinely seek information on their patients
1 The term psychosocial worker used in this discussion to refer to professionals who focus on the
psychosocial needs of patients. This category may include social workers, case managers, health educators,
and counselors. The staffs of many of the study climes included these types of professionals, and climes
encourage PCPs to use these services in the interest of efficiency.
2 Even study informants who expressed the belief that mental disorders have a physical basis tended to
use the term medical to refer specifically to physical health conditions or care as distinguished from mental
health conditions or care, as with popular usage.
72


psychosocial concerns, but when mental health issues surfaced, they actively engaged in
addressing them.
Representing a combination of the recognition and response dimensions in an
analytic tool allowed for comparison among the approaches described by the PCPs for
various situations they were facing and for further examination of circumstances of
those situations. The tool consists of a three-by-three grid that forms zones from the
intersections of each of three descriptive divisions of the two continuums, with each
zone describing a particular approach category.
Because in many cases PCPs reported behavior seemed to fall somewhere
between the two extremes on either dimension, a third category of behavior describing
a more moderated mode of operation helped more accurately characterize providers
approaches in terms of recognition and response. The middle-ground category for
recognition describes behavior reported by most study informants for most situations
they encountered that is, recognizing the need to address psychosocial issues by
means of cues. So while they may not actively seek information about non-organic
aspects of their patients well-being, for the most part, PCPs do not necessarily avoid
delving into these issues, either. Instead, they respond to some indication that
psychosocial issues may need to be addressed. In terms of responses, while the extreme
behaviors consisted of either avoiding direct involvement in managing mental health or
attempting to address mental health issues that were clearly outside the scope of their
expertise, most PCPs acknowledged and accepted their responsibility to diagnose and
treat the common mental disorders they had been trained to recognize and manage,
and this constituted the middle ground of the response continuum.
Figure 4-1 shows the nine zones defined by the grid.
73


Actively
Seek
s
o
o
P:
$
e>
o
o
Uj
cc
Respond
to Cues
Dont
Ask
Q_ Refer Within Scope of Expertise Outside Scope of Expertise
Low-
No Risk * -------- RESPONSE
High
Risk
Figure 4-1: Analytic Model for Characterizing
PCPs Approaches to Mental Health Care
Describing the Process of Recognition
The Ends of the Continuum. The active end of the recognition continuum
describes the behavior of PCPs who Actively Seek information from patients concerning
their sense of well-being, social circumstances, and health beliefs and behaviors. This
includes PCPs reports of routinely and early in their relationship with the patient,
inquiring beyond the details of specific complaints to learn about the context within
which the patient has developed or is experiencing them and assessing the patients
general sense of well-being.
.. Tm very cognizant of those issues. And so, when I go in to see a
patient, from the time I walk in the room, Pm thinking, you know, how
74


are they managing at home? How are they what is their financial
situation? How are they coping in general? Do they look depressed? Do
they look anxious? [P34]
Because many patients are reluctant to share personal and/or emotional details
with someone unfamiliar to them, these PCPs also described techniques to create
openings for patients to talk. Many reported routinely posing questions as simple and
open-ended as How are you feeling?.. .Whats bothering you the most today? [P32],
or Are you enjoying life?.. .Hows life? Is it stressful or not stressful? [P2]. If their
questions dont draw a response on the first visit, PCPs who are more proactive toward
recognition will persist over subsequent visits to open the door [PI2] for these
discussions with patients whether the patient sends signals or not, knowing that it often
takes time for the patient to feel safe and comfortable in talking about psychosocial
concerns.
Also fitting this recognition category were reports by PCPs that while they may
not necessarily look for all indications of their patients mental health status, they will
routinely screen their patients for specific issues, such as domestic violence or substance
abuse both prevalent phenomena in the population and often associated with mental
health issues.
But as far as, how often do I see women with mental health problems? I
would say, you know, 30 to 50% of the women I see. And thats a lot
because I ask I always ask. I always screen for DV//because domestic
violence is so prevalent in our population and so.. .missed by so many
health care providers. And its an easy way for us to get in and get these
women help. [P2]
Some said they believe, as a matter of personal philosophy or as a consequence of their
training, that a whole person [P45] perspective is essential to their ability to
adequately address their patients health needs, and that they seek this information so
they can integrate it into the health care they provide.
Others whose behavior mapped to this end of the continuum, however,
reported doing so as a consequence of having developed, over the course of caring for
this population for some time, a high index of suspicion [P21] that psychosocial
75


issues exist that may affect the patients health or their ability to effectively address
health issues. Many observed that if they faced life circumstances as difficult as most of
their patients, they would be depressed, too [P9, P14]. Similarly, the high incidence of
health risk behaviors and high-risk environmental factors affecting their patient
populations further compels these PCPs to proactively seek to understand all the
potential influences on their patients health or barriers to being able to adequately
address health issues in these patients. Some who maintained a high index of suspicion
about psychosocial issues said they would often cut to the chase [P4, P35] and test
their suspicions with patients.
So a lot of times Ill just sort of cut to the chase Do you feel
depressed? And then, often theyll break down in tears.. .and then you
know you sort of struck the right chord, so that you can go about it, as
opposed to trying to figure out, well, whats all this other stuff there?
[P4]
But while the tendency of some was to immediately explore those issues with
their patients, others acknowledged the same difficult life circumstances and risk factors
but did not pursue these discussions if they could help it, lacking confidence in their
ability to respond in some meaningful way. This behavior instead mapped to the passive
end of the continuum, which is characterized by a nearly singular focus on physical
complaints and no explicit inquiry into psychosocial concerns. Others whose reports
reflected this passive approach tended to do so for other reasons as well.
When operating in Dont Ask mode, PCPs may fail to notice or may even
actively ignore indications of mental health issues. Personal philosophy and/or training
also influence actions consistent with this behavior.
My training is in evaluating, examining, looking at lab and imaging
studies, and then deciding what to do from there, rather than kind of
interacting with someone and automatically attributing their issues to
their mental condition. So that is my standard kind of framework, is to
exclude, I guess.. .organic disease is the best way I can describe it
before I attribute things to mental. [P31]
76


With this belief and many described this explicidy the PCP approaches each
situation from a biomedical perspective first, and then proceeds to explore mental health
issues only when an organic [P3, P4, P5, P31] basis for a patients symptoms or
complaints has been ruled out. Recognition, in this sense, is effectively only postponed
until physical health issues are addressed to the PCPs satisfaction, and resolution of
those issues or the failure to find a physical basis for the patients symptoms becomes
the cue that prompts the PCP to delve into potential psychosocial causes.
While exploring mental health issues with patients is certainly not an activity
most PCPs expressed a preference to engage in, circumstances may sometimes justify
not pursuing these issues at a particular point in time, if at all. For example, when a
patient carries a mental health diagnosis [P50] and discloses this to his/her PCP or
when a patient comes to the PCP specifically seeking mental health care, the PCP has a
basis for determining an appropriate response and there is no real need to take steps to
discover if such issues exist. Postponing discussions about mental health concerns is
also often justified by the relative severity of physical health issues that may claim higher
priority for the PCPs time and attention until they are brought under control.
Middle Ground: Responds to Cues. By their report, many PCPs tend to operate
more reactively than proactively in recognizing mental health issues with their patients.
Trained primarily to recognize signs and symptoms of physical disease, most consider
their primary role to be provider of medical care, with mental health care being
ancillary to their real job. And because most of their patients present with physical
health concerns or symptoms, mental health issues typically are not the first thing these
PCPs tend to evaluate with their patients.
Part of the problem, I think, is that were so focused on organic kind
of the organic pathway. A patient comes in with headaches, or back
pain, or whatever physical complaint, and I think were taught in medical
school to really kind of go down the organic pathway and look into
various organic etiologies, either through tests, procedures, or pure
therapy give somebody a stomach acid pill for stomach pain and
77


then really considering the psychosocial factors and their influences at
the later stages, even after we have explored really rare organic diseases.
[P5]
Instead, they will most often follow up, with varying degrees of aggressiveness, on cues
that indicate that psychosocial issues exist and are affecting their patients health or well-
being.
While most recognition occurs through this mechanism, the study data suggest
that PCPs may arrive at this mode of operation differently and at different times in the
course of working with a patient. The more direct path to recognition via cues occurs
when PCPs encounter behaviors or symptoms they associate with psychosocial issues or
mental disorders, which provide the best evidence for considering this possibility. By
contrast, others arrive by process of elimination that is, when all potential organic
causes for the patients complaints have been ruled out. In the latter case, the lack of
objective [PI] findings to support or substantiate the patients symptoms is the cue
that prompts the PCP to further investigate the possibility of psychosocial issues.
Noticing a pattern of behavior or patient complaints over time creates another indirect
path to recognition.
We are sometimes only after the fact putting together this patients
number and frequency of somatic complaints where we dont find very
much.. .should have had me thinking about three visits ago, is there
depression underneath this? And so I write a little note to myself to
more specifically address depression underneath this the next time they
come in telling me about a new pain in their lower part of their left
scapula. And I think thats probably more typical just in terms of the
number of patients that we have. [P8]
Once a PCP notes a cue, s/he may or may not follow up immediately on that
insight, if at all. While some may proceed immediately to exploring their suspicions with
the patient, others may choose to wait to explore them further. Delay may occur for
several reasons. Some PCPs said that as long as uncertainty still exists as to whether the
patients symptoms represent a physical ailment, they are reluctant to prematurely or
incorrectly attribute a real [P6, P21, P43] physical condition to a mental state.
78


I think I sort of feel compelled.. .1 worry about blowing people off
because of their psyche. And so I usually need to get to a point of
feeling like, okay, Im at a point where I have enough data that Im okay
to.. .Im not really blowing them off Im deciding thats not the
diagnosis. [P6]
Besides needing personal reassurance that they have not missed important
physical health issues, PCPs also talked about the need to reassure patients that they
take physical complaints seriously, describing this as a means to engender trust. PCPs
may postpone a discussion around mental health until they feel they have a sufficiendy
trusting relationship with the patient or as long as they perceive that the patient will not
be open to that possibility. Unlike with actively seeking behavior, however, rather
than proactively and persistendy creating openings for discussion, a PCP who is
responding to cues may harbor suspicions for several visits but hesitate to broach the
topic with a patient, anticipating that the patient may be resistant to the suggestion
that their issues are more psychiatric [PI 8] perhaps based on an earlier experience
with the patient. In these cases, the PCP relies on what s/he perceives or assumes about
the patients current frame of mind rather than explicidy testing it. This behavior was
often described by PCPs recounting situations centering on somatic complaints and
syndromes involving chronic pain, where uncertainty is high and the patient insists the
issues are physical, not mental. While it is admittedly a delicate line to walk,
postponement can unduly prolong the recognition process.
Depending on their inclination toward one end of the recognition continuum or
the other, PCPs also vary in how readily or not they pick up on cues that psychosocial
or mental health issues are at play. Whereas some may not actively seek evidence of
psychosocial issues in interacting with their patients, they will maintain a relatively high
sensitivity and notice cues as they emerge. By contrast, many study informants reported
situations in which recognition occurred only after they realized a pattern that had
emerged over a series of visits. And while all PCPs are likely to miss cues at one time or
another, many are also likely to actively ignore cues under certain circumstances.
Table 4-1 summarizes cues to recognition mentioned by PCPs in this study.
79


Table 4-1: Summary of Cues Signaling Potential Psychosocial Issues
Cue Type Definition Examples
Obvious indicators The patient presents with mental health concerns or exhibits behaviors that clearly indicate the need for the PCP to attend to mental health Volunteers having suicidal or homicidal thoughts, describes hearing voices Behavior is erratic [P14, P44] or bizarre [P2] Cries or shows other clear signs of distress in an encounter PCP notices a change in appearance or behavior in a patient s/he is familiar with Patients family or friends report concern about behaviors
Complaints, conditions The patient presents with certain types of complaints or comorbid diagnoses that the PCP associates with depression or other mental disorders Multiple and/or vague complaints (e.g., ache all over [P9]), often difficult to objectify, [P31] and intractable or unresponsive to treatment Complaints often labeled somatic or psychosomatic e.g., chronic pain (headaches, low back pain), chronic fatigue, musculoskeletal problems Diagnoses often comorbid with depression e.g., irritable bowel syndrome; fibromyalgia; diabetes and other chronic diseases; substance abuse Specific symptoms of depression or anxiety, such as anhedonia, insomnia, unexpected weight loss or gain, memory or concentration problems, palpitations
Patient characteristics The PCP is aware of stressful circumstances or issues in a patients history that suggest or indicate their vulnerability to mental health issues OR the PCPs experience has identified certain patient types or groups as being more likely to have such issues Patient has described stressors, such as loss of job, marital problems, financial worries, or a destitute existence [P7] or PCP is aware of issues such as social isolation or family dysfunction Patient discloses history of physical abuse or other trauma, family history of mental illness or substance abuse Aspects of patients situation e.g., inability to keep a job or maintain relationships, unusual dependence on or isolation from family or friends suggests some level of dysfunction PCP sees indication that patient is unable to cope well with stress or to take control, or seems disorganized [PI, P9, P42, P43]


Table 4-1: Summary of Cues Signaling Potential Psychosocial Issues (Cont.)
Cue Type Definition Examples
PCP routinely encounters distress or certain mental disorders in certain groups or patient types (e.g, depression or dementia in seniors)
Intuition The PCP gets a strong feeling difficult for most to characterise that triggers them to probe further * PCP feels the emotion [PI5] when s/he walks into the exam room PCP reads potential issues from the patients affect, or body language
Health behaviors The patient exhibits certain tendencies in help-seeking, self-care, or adherence that are extreme or seem counterproductive Poor self-care patient does not take steps necessary to control or mitigate chronic disease progression; does not seem to care about their disease [P37] Patient seems to worry excessively about relatively minor health issues Non-adherence: patient does not take medications necessary for disease management, refuses to take meds at all, or goes on and off [PI] medications Frequent flier [P45] patient presents frequently for care, often for relatively minor complaints or for medical complaint of the month. [PI6] Patient tends to over-utilize emergency room and urgent care facilities and/or calls PCP or clinic multiple times each day or week. Doctor shopping [P24] patient presents to different providers and/or switches providers frequently, sometimes because of dissatisfaction with care, or may be seeking drugs
No biomedical explanation Things dont add up [P4] or dont really fit into anything [P43] from a biomedical perspective PCP sees no evidence in test results or work-ups that explain why patient is experiencing symptoms or the severity of symptoms s/he reports Complaints seem inconsistent with patients overall health state or age Patients symptoms dont describe any disease that PCP is aware of Disease processes flare up or remain poorly controlled despite PCPs attempt to manage them Patient complains of side effects from any medication prescribed or claims they are ineffective


Cues to Recognition. Because the literature contains little information on
specific cues that prompt recognition and diagnosis of mental disorders, the interviews
conducted for this study explicidy asked about what will cause PCPs to focus on and/or
pursue discussion of psychosocial issues with their patients. Responses to this inquiry
revealed six broad categories of cues. Table 4-1 lists these in order (most to least) by the
number of individuals who mentioned them in their responses to direct inquiry or who
highlighted them in the situations they described. Taken together, these cues revealed
broader insights about the process of recognition.
Insight #1: Some Cues Are More Obvious than Others. While some cues are
very subtle, others are screaming that this is //the primary thing that needs to get
taken care of [P40]. For example, several PCPs reported having a feeling [PI5] more
so than any concrete indication that a patient had underlying mental health issues, and
proceeding on that intuition to test their suspicions. At the other extreme, a patients
obviously bizarre [P2] behavior in the context of a psychotic break, or a mention of
suicidal or homicidal thoughts, prompted immediate and definitive intervention by the
PCP, either direcdy or through referral for acute care. Beyond the most obvious ones,
some cues scream louder at individual PCPs than others. PCPs who highlighted these
said that, based on their experiences with patients, certain cues nearly always are a red
flag [P7, P43] for them.
If they have discrete medical complaints, then Im not going to address
this.. .with the exception of and youre going to think this sounds
ridiculous, but middle-aged Hispanic women who come in
complaining of bizarre physical complaints like, The left side of my
body is numb, or I have this weird tingling in my fingers. That is
depression untilproven otherwise, (emphasis added) //[F]or some reason I
think in that ethnic group there is a tendency to somaticize [sic] more
theyre more acceptable, you know, the medical complaint, than a
psychiatric complaint. So when I hear a Hispanic female complain of
something that//sort of sounds like a weird, maybe non-physiologic
complaint, I immediately ask, How are things at home? Are you sad?
82


thats a cue and I kick right into that, and 90% of the time.. .theyre
crying in about two minutes. [P44]
Insight #2: Many Cues Derive From the Biomedical Lens Through Which
PCPs and Patients View Health Issues. Difficulties addressing physical health issues or
the nature of patients physical complaints often make the PCP suspect involvement of
psychosocial issues. For example, several study informants talked about the challenge of
trying to resolve complaints typically associated with somatization, primarily due to the
intractability and elusiveness of those complaints to causal explanation. In situations in
which things dont add up [P4] biomedically for example, in which reported
symptoms do not describe any diagnostic entity the PCP is aware of PCPs reported
they are likely to suspect a psychological basis for the complaints. Illustrating this, PCPs
described patients who have multiple, vague complaints such as I ache all over my
bones ache [P9] or My whole body hurts [P4] that are defying diagnosis [P21]. A
diagnosis of chronic disease or other conditions such as chronic pain, fibromyalgia, and
irritable bowel syndrome, which are known to co-exist with depression in particular,
also cue many providers to the existence of potential mental health issues.
Also indicating a biomedical orientation, poor self-care was frequendy
mentioned by study informants as raising concern again capturing the PCPs attention
because of its interference with managing physical health conditions. This may have
been in part due to this studys focus on mental health care in the context of chronic
disease management. However, it is clear that patients health beliefs and behaviors
often create barriers to controlling the physiologic course of chronic conditions and
that mental disorders can often manifest in poor self-care (Katon & Ciechanowski,
2002). Typically revealing itself through a patients pattern of non-adherence, erratic
follow-up [P44], failure to keep appointments, or apparent lack of concern for their
health, this type of cue reveals itself over time.
While notable to most PCPs, poor self-care does not always trigger the PCP to
explore mental health issues, however, because such behavior also can result from very
83


real barriers associated with socioeconomic disadvantage that CHC patients often
confront.
I have to admit thats not the first thing I look at when they come in for
treatment of diabetes or high blood pressure. If Im not going
somewhere with it, then Ill start probably to investigate more the
physical aspects. Can they access.. .do they have the money to access the
meds? Or is it there support at home as far as their diet lets say if
theyre diabetic. Is that a problem? Then we need to have the family
members come in. And probably the last thing would be well, maybe its
the depression they dont care about their disease. [P37]
Similarly, because non-adherence and other behaviors indicative of a patients non-
engagement in managing their chronic disease are so prevalent and expected in this
population, PCPs may not readily identify them as a symptomatic of a mental disorder.
One PCP reflected on one such situation:
And I just thought she was my typical.. .non-compliant patient, which is
very common.. .you know, a lot offatalismo life happens.. .Just, you
know, I dont feel bad now. So//but now that I really look back on
it, shes probably just been depressed enough where she just hasnt cared
nor had the energy to manage her chronic illnesses. [P20]
In this case, however, and as other PCPs reported, the cue that ultimately revealed this
patients depression was not the pattern of poor self-care, but rather emotional behavior
resulting from a crisis in the patients life that finally caught the PCPs attention and
spurred the PCP to ask, and the patient to admit to, specific symptoms. It was only
upon reflection, then, that the PCP realized that the patients non-engagement in
managing her chronic disease was likely associated with depression.
A contrasting health behavior also may signal potential mental health
underpinnings to PCPs. Patients who made frequent visits to their PCPs and exhibited
what seemed an inordinate concern for seemingly minor health issues, came in with
different complaints each time, or paid frequent visits to the emergency room or urgent
care climes were dubbed frequent fliers, [P45] and PCPs were likely to suspect
existence of a mental disorder. One PCP noted:
84


... theres sort of a certain natural relationship between the thickness of
someones medical chart and their age. You know, someone whos 80,
you might expect to have a very thick medical chart, and someone whos
25, youd expect to have a pretty thin medical chart. Well, anytime that I
walk up to the door and// I see a thick chart//and I pull it out to look
at the age, and I see that its someone thats 25, 30, 35, maybe even
40.. .1 almost equate that with psychiatric disease. [PI6]
By several reports, these were among the most difficult patient situations to manage,
because these patients were often firmly convinced that their symptoms were indicative
of real [P6, P21, P43] medical problems and the PCP could neither find a biomedical
explanation for them, nor seem to make headway in exploring potential mental health
issues with them.
Insight #3: Cues Also May Derive from PCPs Beliefs about the Patient and/or
Knowledge of the Patients Life Circumstances. While providers in any primary care
environment encounter patients with stressful life situations, PCPs working in
community health centers routinely see patients whose fives are exceptionally difficult
and seem to justify substantial mental distress. Most study informants noted the high
prevalence of mental disorders, especially depression, in their patient populations, and
with social and economic hardships figuring so prominently in the fives and health of
the patients they see, reported a relatively high sensitivity to the effects of extreme fife
stressors on patients health and their ability to manage it. But while PCPs may, in some
situations, anticipate such issues and inquire about them proactively, in others, they may
note such information as it emerges in conversation, and integrate this knowledge with
other cues during the process of recognition. In other words, such information
becomes one of the puzzle pieces that leads to recognition.
.. .Im just all of a sudden suspicious of one thing over another, because
Ive been talking to them, and sometimes it only happens as Ive got my
hand on the door, and they go, Oh by the way... Well, a lot of times
its one totally new visit that they want in this visit, you know a totally
unrelated problem. But sometimes its the last fitde piece in the puzzle
85


i
that you needed, where theyve got all these unrelated symptoms and
that one thing ties it together, and you go, Aha! [P39]
PCPs may be similarly tuned to other patient characteristics besides life
circumstances. As found in other studies on recognition, PCPs reported anticipating
mental health issues more in women than men, more in the elderly than in younger
patients, and more in their patients who are very ill especially those with chronic
disease (Borowsky et al., 2000; Callahan et al., 1997). Additionally, however, individual
PCPs form expectations based on their experiences with patients, and may associate
such characteristics as ethnicity, education level, life stage (for example, menopause or
pregnancy), level of acculturation, or marital status with higher likelihood of mental
health problems.
Insight #4: Delayed Recognition is Often Due to the Ambiguity of Cues. Of
the cues reported by PCPs, only the ones categorized in Table 4-1 as obvious
indicators are unambiguous enough to justify an immediate response by the PCP in
most cases. Enough uncertainty surrounds the other types of cues to cause potential
delay in recognition. A built-in delay exists, for example, with cues related to health
behaviors, and this is even more obviously the case where the cue is that no
biomedical explanation can be found for the patients complaints. In absence of any
more definitive indicator, it will take time for the PCP to determine that a pattern exists
with these types of cues. Also, while many different cues may arouse the PCPs
suspicion, the ability of the PCP to address mental health issues will be delayed as long
as the patient denies the existence of emotional distress or otherwise resists the PCPs
attempt to explore those issues.
Describing Responses to Mental Health Needs
Once a PCP suspects or recognizes mental health issues in a patient, s/he must
then determine and implement an appropriate response. The interview data suggest that

i
1
86


PCPs responses may be categorized according to the willingness or unwillingness of
PCPs to step outside their comfort zones or to assume risks associated with practicing
outside the scope of their expertise. Setting aside some amount of reluctance around
addressing mental health issues that might be expected of anyone other than a trained
mental health worker, and granting that most PCPs practice relatively comfortably in
the realm supported by their training, the practices of particular interest that these PCPs
described are the ones associated with the extremes of the risk-taking continuum
those that reflect risk aversion, and those that reflect bolder risk taking. The discussion
that follows again references the model shown on Figure 4-1.
The Ends of the Continuum. At one end of the continuum, PCPs described
low- to no-risk behavior in response to mental health issues that were both inside and
outside the realm of their expertise, indicating that their comfort in providing mental
health care is not related only to their training. When PCPs operate in the Avoid, Defer,
Kefer response zone, they are unwilling to diagnose, treat, or adjust psychotropic
medications for patients whom they suspect or know have mental disorders other than
the ones they have confidence in their ability to manage. In some cases, PCPs lacked
confidence in assessing and managing even the common disorders covered in their
training, although these individuals seemed to be less typical than not among this
studys informants. While most were comfortable in recognizing and treating
depression, however, several study informants described behavior in at least some
situations that was consistent with this response zone.
The low-to-no-risk behavior is characterized by the PCPs focus on medical
or physical care and reliance on others whenever possible to handle mental health. One
form the behavior takes is to refer the patient to another person for mental health needs.
Referral may be to an external mental health provider (e.g., to a community mental
health center) when that option exists, or the PCP may simply advise the patient,
insured or not, that s/he should seek specialty care. Referral is, of course, the
87


arrangement that best supports the primary care model, around which PCPs skills and
expectations are developed and the process of care is structured. In this model, the PCP
has the pivotal role of [receiving] all problems that patients bring unrestricted by
problem or organ system with the expectation that these providers will diagnose and
manage most of those problems and will involve other health care providers as needed
to ensure that patients receive integrated care (Donaldson, Yordy, Lohr, & Vanselow,
1996, p. 33). From this standpoint, referral is not only a perfectly appropriate response,
it is the assumed ideal. What is important, however, is that referral is not a viable option
for addressing many of the mental health needs PCPs encounter in CHCs, so PCPs
working in these settings often must craft an alternative response.
Referral also may occur within the practice setting itself. Where resources
permit, PCPs may respond to suspected or recognized mental health needs by handing
off the patient to a clinic staff member, such as a social worker or case manager, to
manage psychosocial issues. In the extreme, a PCP may call in someone else on staff to
talk to a patient whose mental or emotional stability is questionable or to screen a
patient for depression or anxiety. PCPs who reported doing so explained that time
limitations often prohibit them from providing this care themselves, while others
reasoned that their lack of expertise justified using that of others whenever possible.
I dont have any [screening questionnaires] on hand. Im sure I can find
some if I was.. .really wanting to know and I couldnt get [the patient] to
see [clinic psychosocial worker (CPW)]. Thats kind of what I do -
because if I dont know and I want to know, then I send them to
[CPW]. And they always go or they dont come back. Or I send them
to [CPW] and then they dont go and then they dont come back and see
me about it either in which case then I kind of think, well if its really a
problem youd come back and say, you know, Actually I really dont
want to see [CPW] and I really want the meds or I really want to get
rid of this problem. And then Id say, Fine, okay.. .1 dont know why
you dont want to talk to [CPW] or whatever, but Im sure we can talk
about something else.. .or we can further explore in this issue. [PI7]
Nurse practitioners (NPs) and physicians assistants (PAs) have the additional option to
ask an MD colleague to handle a patient whose medical and/or mental health issues are
88


more complex. Clearly, however, if no such back-up exists, the PCP faces the choice of
taking on the tasks or leaving a patients needs unaddressed.
Also falling within the low-to-no-risk zone is hands-off behavior around a
patients mental health issues that are managed elsewhere. Some PCPs reported a more
active role than others in monitoring the patients mental health status and maintaining
contact with the patients mental health specialist. This analysis considered the more
low-involvement stance to fall within the slvoid, Defer, Refer response category, and more
active involvement as evidence that the PCP retained a level of responsibility for broad
management of the patients health including mental health consistent with the
primary care integration role. While some seemed to choose one stance over the other,
in many instances, the PCPs role appeared to be more dictated by the availability or
willingness of mental health specialists to participate in collaborative care.
When the PCP is left without referral options of any kind and determines for
whatever reason that s/he is unwilling to attempt to address serious mental health
issues, the response may be to avoid going there [P45] with the patient at all, instead
staying within the familiar realm of evaluating, examining, looking at lab and imaging
studies, and then deciding what to do from there [P31]. Or, as discussed relative to
recognition, oftentimes PCPs will start out in this mode and defer dealing with suspected
mental health issues by remaining there throughout the process of recognition, and then
shift their focus when they have collected sufficient evidence to be convinced that the
need is clear and/or they have not missed an organic basis for the patients complaints.
Deferring also may allow the PCP to avoid delving into mental health issues for as long
as possible.
Operating at the high-risk end of the continuum are PCPs who, when they feel
they have no other choice, will venture outside the scope of their direct expertise and
step in to manage more serious mental disorders that typically are handled by mental
health specialists. PCPs providing this level of care will do their best to make a diagnosis
89