The problem of non-adherence

Material Information

The problem of non-adherence an ethnographic study of the symbolic meanings associated with human immunodeficiency virus (HIV) infection and treatment with highly active antiretroviral therapy (HAART)
Grodesky, Michael J
Publication Date:
Physical Description:
xvii, 289 leaves : ; 28 cm


Subjects / Keywords:
Patient compliance ( lcsh )
Antiretroviral agents ( lcsh )
HIV-positive gay men -- Health and hygiene ( lcsh )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 276-289).
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by Michael J. Grodesky.

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:
122938776 ( OCLC )
LD1193.L566 2006d G76 ( lcc )

Full Text
B.S., Loretto Heights College, 1986
M.S., University of Colorado, 1994
A dissertation submitted to the Faculty of the University of Colorado in
partial fulfillment of the requirements for the degree of
Doctor of Philosophy
Health and Behavioral Sciences

2006 by Michael J. Grodesky
All Rights Reserved

This thesis for the Doctor of Philosophy degree by
Michael J. Grodesky
has been approved for the
Program in Health & Behavioral Sciences
Steve Koester
Susan Dreisbach
Claire Zilber

Grodesky, Michael (Ph.D., Health and Behavioral Sciences)
The Problem of Non-Adherence: An Ethnographic Study of the Symbolic
Meanings Associated with Human Immunodeficiency Virus (HIV) Infection
and Treatment with Highly Active Antiretroviral Therapy (HAART).
Dissertation Directed by Associate Professor Steve Koester
Background: The purpose of this study was to gain deeper contextual insight
into adherence behavior as it relates to HIV infection and treatment a
population of gay men. The introduction of highly active antiretroviral
therapy (HAART) was a turning point in the biomedical treatment of Human
Immunodeficiency virus (HIV) infection and significantly reduced morbidity
and mortality. For HAART regimens to have a sustained effect, high levels
of adherence (>95%) are needed. However, studies have suggested that non-
adhierence is common and may be present in 50 to 70% of patients. While
numerous social and demographic factors have been identified as being
associated with non-adherence, effective interventions to improve adherence
have not been developed suggesting a fundamental understanding of this
issue is lacking.
Study Objectives: (1) To discover and describe the contextual and symbolic
factors that influence adherence to HAART regimens in patients selected
from an urban clinic environment. (2) To explore and gain understanding
and insight into the symbolic and contextual interactions between the urban
medical clinic setting and the individual. (3) To explore symbolic and
contextual influence of the individual perception of self on adherence

Design and Method: The study design was qualitative, ethnographic.
Twenty-four in-depth, open-ended, semi-structured interviews were
conducted on eighteen participants. Sampling was carried out utilizing a
critical case purposive strategy with key informants identifying participants
who were informant rich. The sampling frame was the University of
Colorado Hospital Infectious Disease Group Practice (IDGP), a large,
hospital-based primary care HIV/AIDS practice.
Data Analysis: Transcribed interview data were coded using Atlas Ti, a
qualitative data analysis computer software package. Nine domains of
understanding were synthesized from the coded data and individual
experiences with the domains were summarized. From the summaries
emerged three behavioral categories or typologies.
Findings: The data suggested that the interaction between the individual
and the medical institution is not threatening to the individuals who, in this
study, were generally in an alliance with the medical system. Rather, the
potentially threatening interaction was between the individual and the
illness (HIV-infection) itself. Nine domains of understanding emerged as
factors that influence the self and the symbolic meaning one may ascribe to
HIV illness.
Three typologies emerged as explanatory models of adherence behavior and
were termed Symbolic Alignment, Symbolic Partial Alignment and Symbolic
Misalignment. Those in the symbolic alignment category tended to be highly

adherent to their treatment regimens while those in the symbolic
misalignment group were least likely to be adherent. Those in the symbolic
partial alignment typology were more or less adherent depending on
circumstances. The results from the study help to add insight into the
complex nature of adherence behaviors, help to integrate concepts from the
current scientific literature, and suggest that those who are interested
helping individuals with HIV-infection better adhere to treatment regimens
must look at a host of structural elements with which individuals must
interact in order to understand and interpret their illness and its attendant
This abstract accurately represents the content of the candidate's thesis. I
recommend its publication.
Steve Koester

To the memory of my parents, John and Jean.
Gifts come in unexpected ways. With love and gratitude I honor you for the
gifts you have given me.
Because I know that time is always time
and place is always and only place
and that what is actual is actual only for one time, and only for one place
I rejoice that things are as they are
and renounce the blessed face.
-T.S. Eliot

A project as complex as a doctoral dissertation requires the support,
insight and assistance of numerous individuals and groups. Some have
provided direct conceptual and technical insight and support, while others
have provided inspiration and encouragement simply through their presence
in my life. All have been integral to the successful completion of this
First, I wish to express my admiration for the men and women who
live with and who have died from HIV and AIDS. Throughout my years of
work with the HIV Community, I have been truly humbled and inspired by
the transformative, soulful and sometimes mysterious journeys of those who
must confront this illness on a daily basis. My life has been changed, not
only because HIV effects all of us, but because of my interactions with those
whom it has effected most severely.
I am most thankful to my friends and colleagues who are the staff of
the University of Colorado Hospital Infectious Disease Group Practice
(IDGP). I am deeply grateful to Steven C. Johnson, M.D., the Medical
Director of the IDGP, who has been one of the significant mentors of my life
and has inspired me both as a clinician and as a man.

I am thankful to all those in the IDGP who provided referrals for the
study especially, Claire Meams, L.C.S. W., who took a lead role in this
process. My decade-long association with the IDGP is a source of pride and I
view that time as the most satisfying part of my career to date.
I am indebted to Patricia Burdick who functioned as a research
assistant on this project and who performed most of the data transcriptions.
Her devotion to the project and her attention to detail provided the basis for
a high-quality research project.
A special word of gratitude to those who were a part of the
Dissertation Committee for this project including Steve Koester, Ph.D., Susan
Dreisbach, Ph.D., Claire Zilber, M.D., Kees Rietmejier, M.D., Ph.D., and Craig
Janes, Ph.D. It was an honor for me to work with individuals of this caliber.
I am also thankful to other faculty and students from the Health and
Behavioral Sciences Program who added insight and support, including
Teresa Sharp, Ph.D., and Bonnie Castillo, Ph.D. A special thanks also the
Abby Fitch, Program Administrative Manager for her continual help and
logistical support and to Christine Pon, the former Program Administrative
Manager whose ongoing support was integral to the project.

I would like to thank Mark Wicks, Ph.D.(c), of the University of
Washington School of Medicine and to Amy Benjamin, Ph.D. (c), of Seattle,
Washington, who provided ongoing insight, enlightenment and
encouragement throughout the past two years and who assisted with
colleague debriefing which added significantly to the quality and internal
validity of the project.
I have been blessed throughout the years to be surrounded by a
group of loving friends who have supported and sustained me in this long
journey through graduate school and through life. My deepest thanks to
Carol Anne Bosco, friend, soul mate and ongoing source of light,
encouragement and inspiration, and to Ron Bucknam, my teacher and
mentor. I also wish to thank Amy Benjamin, Paul Brown, Marla Corwin,
Lydia Frederick, Stephen Heiling, Gerald Higgins, Barbara Klaus, Barry
North, Teresa Sharp, Sen Tally, William Threlfall, Mark Wicks, Jane
Wolbach-Lowes, and to my cousin Kathy Pusatory who has become a sister
to me.
I also continue to be inspired by the memory of those departed who
have, in their own ways, contributed to this project. Andrew Kottenstette,

John Fesenmeyer, Glenn Brown, Luke Davis, Patricia Mijer and Ron
And, finally, I am blessed to have in my life and in my work the gentle
and encouraging touch of James Myron Simandl.

1. INTRODUCTION AND OVERVEW...................................1
The Research Problem............................... 1
Research Questions.................................. 8
Scope of the Research............................... 12
Organization of the Thesis........................ 13
2. REVIEW OF THE LITERATURE................................. 16
Introduction...................................... 16
Biomedical Imperative For Adherence................18
Factors Associated with Non-adherence..............21
Measurements of Adherence..........................28
Theory in the Current Literature ..................30
Coping Strategies...........................35
Locus of Control............................37
Limitations of Previous Work..............................38

3. THEORETICAL CONCEPTS............................................ 40
Symbolic Interactionism ............................... 43
Erving Goffman...........................................50
Arthur Kleinman .........................................54
4. METHOD.......................................................... 58
Overview ............................................... 58
Research Location....................................... 63
IDGP History, Culture and Demographics............66
Sampling and Recruiting.......................... 73
Data Collection......................................... 84
Semi-structured Question Development..............86
Informed Consent and Ethical Considerations.......87
Audio Recording...................................91
Field Notes.......................................92
Pilot Interviews..................................94
Semi-structured Interviews.......................102
Preliminary Data Analysis........................107
Mini Focus Group ................................109
Development of Follow up Interview Questions.....112

Reliability and Validity of the Data................116
5. DATA ANALYSIS AND SYNTHESIS..................................120
Definition of Terms.........................................121
Final Sample Characteristics................................122
Post Fieldwork Data Analysis and Synthesis................. 127
Organization and Management of Data.................127
Approach to Data Analysis...........................128
6. FINDINGS.......................................................139
Introduction to Findings....................................139
Section One: Symbols of Interaction and Threat..............141
Medical Institution as Symbol of Interactive Threat.... 141
HIV Illness as Symbol of Interactive Threat.........146
Summary of Section One..............................152
Section Two: The Typologies.................................153
Introduction to Typologies..........................153
Symbolic Alignment Typology.........................158
Symbolic Misalignment Typology......................178
Symbolic Partial Alignment Typology.................191
Summary of Findings...............................208
7. DISCUSSION.................................................... 210

Methodological Discussion...........................215
Discussion of the Findings..........................226
The Role of the Researcher......................... 236
Theoretical Extensions............................. 239
Conclusions and Implications...................... 256
A. HUMAN SUBJECTS APPROVAL...................................260
B. INFORMED CONSENT DOCUMENTS................................265
C. CONFIDENTIALITY AGREEMENT.................................274

2-1. Mead's Self-Interaction.......................................... 48
4-1. IDGP Sampling Frame Demographics..................................74
4-2. IDGP Entrance.....................................................96
4-3. The IDGP Waiting Room.............................................97
4-4. The Interview Room................................................106
6-1. Symbolic Alignment...............................................155
6-2. Symbolic Misalignment............................................156
6- 3. Symbolic Partial Alignment.......................................157
7- 1. Goffman's Secondary Adjustment...................................249
7-2. Modified Goffman's Secondary Adjustment......................... 250

4-1. Qualitative Sampling Strategy...........................................79
4-2. Basic Inclusion Criteria................................................81
4-3. Revised Initial Semi-Structured Questions..............................101
4- 4. Follow-Up Questions for Second Round Interviews........................114
5- 1. Definition of Terms....................................................122
5-2. Selected Characteristics of the Sample................................ 124
5-3. Final Sampling Distribution............................................125
5-4. Second Round Questioning Sample Distribution........................... 126
5-5. Final Code Titles and their Meanings...................................133
5- 6. Domains of Understanding and Related Codes............................ 136
6- 1. Overview of Symbols of Interaction and Typologies..................... 140

"Adherence is a complex dynamic behavior influenced by characteristics of the patient, treatment
regimen, disease, patient-provider relationship and clinic setting" (Ickovics, 2002).
"I's and we's become reduced to systems of it's. And men and women are seen as objects of
information, never subjects in communication." (Foucault, 1972).
The Research Problem
Traditionally, addressing patient compliance to prescribed treatment
regimens has been an uncomfortable, and at times, politically-charged issue.
So much so that in recent years the term compliance has generally been
replaced by adherence in an attempt to signify a fundamental understanding
that the act of adhering to a course of treatment isor should bea set of
circumstances mutually agreed upon by medical provider and patient. As
Ickovics (2002) so eloquently points out, adherence behaviors are
complicated and dynamic interactions between individuals and systems.
The scientific literature is now laden with research that seeks to
understand, at least on some level, why certain persons adhere to a

prescribed course of treatment while others do not, even when the treatment
may be potentially life-saving. The issue of adherence reveals, in many
disease processes, an interesting dichotomy between advancing biomedical
technology and the human behavioral conduit through which such
technology must be conveyed. In no setting is this dichotomy more manifest
than in that of infection with the Human Immunodeficiency Virus (HTV) and
associated Acquired Immunodeficiency Syndrome (AIDS) and the attendant
treatment strategies related to this infection.
HTV is known as a retrovirus because it translates its viral ribonucleic
acid (RNA) into deoxyribonucleic acid (DNA) inside human cells through
the utilization of the viral enzyme reverse transcriptase. In more typical
biological circumstances, it is DNA that is translated into RNA. When

inserted into the nucleus of cells, DNA directs the activities of cellular
function. HIV has an affinity for human immune cells known as CD4
lymphocytes, a type of white blood cell that is a part of the cell-mediated
component of the human immune system. HIV binds to the CD4
lymphocytes and transforms these cells into virtual virus-producing
machines (Streicher, Reitz & Gallo, 2000). The loss of CD4 lymphocytes that
occurs over time eventually leaves individuals with severely depleted

immune systems which are susceptible to various life-threatening
opportunistic infections and malignancies (Vittinghoff, Scheer, O'Malley,
Colfax, Holmberg, & Buchbinder, 1999).
The current treatment technology, which has been highly successful,
involves suppression of the HIV RNA, or viral load, to extremely low levels
so that large-scale damage to the CD4 lymphocytes can no longer occur. This
low level of viral suppression is known as "undetectable" because the
presence of the viral RNA in plasma is so low that it cannot be detected by
typical laboratory assays. This level of viral suppression is accomplished by
using combinations of drugs known as antiretrovirals (Carpenter et al. 2000).
Selected combinations of these antiretroviral agents including
nucleoside reverse transcriptase inhibitors (NRTIs), non-nucleoside reverse
transcriptase inhibitors (NNRTIs) and protease inhibitors (Pis) collectively
known as highly active antiretroviral therapy (HAART) or "the cocktail"
were introduced in the mid-1990s and replaced the previous usage of single-
agent antiretroviral drugs. The introduction of HAART marked the turning
point in the biomedical treatment of HTV/AIDS that significantly reduced
morbidity and mortality (Vittinghoff, et al., 1999).

For HAART regimens to effect a sustained reduction in HTV RNA
levels, patients must maintain very high (greater than 95%) levels of
adherence to dosing requirements for indefinite periods of time. If adherence
is not maintained and HIV RNA is not suppressed to levels below detection,
the remaining virus is subject to the development of mutations that make it
resistant to HAART regimens. Left unchecked, this resistant virus will
become the dominant genotype in the plasma and future viral suppression
will become more difficult and in some cases impossible. Especially in the
early HAART regimens, patients were often faced with high pill burden
(numerous pills, taken at different times) and with differing behavioral
requirements such as dietary restrictions, and debilitating drug side effects
(Altice, Mostashari, & Friedland, 2001).
In addition to the logistical and clinical complexity of HAART
treatment, individuals with HIV-infection have frequently included those
who are among the most stigmatized members of the culture including
injection drug users (IDU) and men who have sex with men (MSM). As the
AIDS epidemic has evolved, the social marginalization of the infected
individuals has further increased. This increased marginalization has been
characterized by complex life circumstances such as poverty, and the myriad

of issues that accompany it, as well as poly-substance addiction and chronic
psychiatric co-morbidity (Altice, et al., 2001).
Researchers and clinicians have been faced with discovering the
intricate and interrelated issues that represent barriers to the successful
administration and maintenance of these life-saving medications.
Understanding non-adherence has required inquiry into the
multidimensional issues related to environmental and sociocultural-
structural circumstances, as well as individual human (agency) phenomena.
The complexity of these phenomena as they relate to human behavior
renders a single etiology for non-adherence unlikely. Moreover, as is the
case with other complex human phenomena, the use of empirical research
methodologies and their attendant quantitative design strategies as a means
of inquiry has significantly limited the types of research questions that may
be asked as well as the content, quality and presentation of the data that may
be collected. As Foucault (1972) has pointed out, devoid of the context in
which they occur, human experiences can easily be reduced to empirical
systems that objectify individuals.
One must also regard much of the existing scientific literature that
addresses the issue of non-adherence as portraying an epistemological

perspective that is based predominately on a Western cultural frame of
reference (See Chapter 2, Review of Literature). This perspective has been
reflected in the use the methodologies of Western science to study the
problem of non-adherence, which, by definition, has kept the scope of
inquiry within the margins of the tradition through which it is being
Clearly, factors derived from the now abundant mass of empiric data
support the implication of factors such as race and ethnicity, age, level of
self-efficacy, socioeconomics, medication side effects and gender differences
in the etiologies of non-adherence to treatment of HIV-infection and other
illnesses. Yet these variables alone have been unable to provide a
comprehensive understanding of the deeper contextual patterns that are
undoubtedly involved, patterns that cannot be discerned through the
empirical tools that have been utilized.
Even in instances where research has been informed by constructs
from behavioral science theories, the reflexive and contextual richness of the
findings frequently fall prey to the limitations of the quantitative nature of
the study design. For example, some studies investigating non-adherence
etiologies have concluded that higher levels of individual perceived self-

efficacy (a construct from Social Cognitive Theory that reflects individual
belief in ability to bring about specific outcomes) are positively correlated
with increased levels of adherence. Yet, the understanding of this correlation,
pursued through purely quantitative approaches, has been abbreviated and
divorced from the deeper contextual meanings, symbols and relationships
that may constitute the basis for the individual experience of self-efficacy.
Perhaps the most striking evidence to date of this reductionist
approach to the understanding of non-adherence is seen in the small
numbers of research-derived interventional strategies that have been
presented for investigation. Moreover, there are virtually no meaningful
outcome data with respect to successful widespread adherence intervention
Clearly, the current empirically derived factors in the scientific
literature that have been proffered as etiologies of non-adherence are neither
irrelevant nor invalid. I would argue, however, that a qualitative foundation
that would inform the meaning of these factors has not been adequately
developed. The present research has sought to add to this foundation. In
addition, while ethnographic work customarily does not commence from a
well-defined theoretical basis, I believe there are several theoretical models

within the behavioral sciences that are compelling to the issue of adherence
and that these positions are well suited to qualitative exploration. (See
Chapter 3). For this reason, I have chosen to amend the traditional
ethnographic format and have proceeded from the start with certain
theoretical concepts in place.
I have been aware of the potential risks and benefits of initiating
inquiry with apriori theoretical concepts in an ethnographic project such as
this. Moreover, as I have previously stated, the research questions
themselves were inspired by theoretical conceptions. One might say that the
underlying theory was, indeed, the guiding passion of this work.
Nonetheless, I have taken care throughout the project to insure that it is the
body of data itself that speaks and not a prejudgment from theoretical
Research Questions
The purpose of this research has been to explore, describe and gain
deeper insight into the nature of the symbolic and contextual meanings and
experiences associated with adherence (and non-adherence) to HAART
regimens and to relate this insight to certain theoretical positions. In the

development of the research questions, and throughout the research, certain
terms have been used that relate to some of the fundamental concepts of the
inquiry. The term symbolic has been used to refer to the personal, social and
cultural meanings through which individuals may represent, or experience
representation of their illnesses. This definition of symbolic is distinct from
though similar to that of Symbolic Interactionism, a social sciences theory on
which this inquiry is based and which will be discussed in Chapter 3.
The term context has been used to refer to the individual, social, cultural and
institutional milieus in which experience takes place. Context, one might say,
is the fabric on which experience is played out. It is the dimension that
clarifies and brings to life the "why" and "how" elements of the "what" in
research. It is the soul of the ethnographic approach.
Creating an unambiguous operational definition for the important
concept of self in this study has been particularly challenging. Because of its
inherent vagueness, the characterization and perception of the self in
research has often been established through the use of surrogate definitions
such as self-efficacy, self-esteem or self-identity. Yet, for the present research, it
was important to let the emerging data itself clarify the term in its most
relevant context. For the purpose of simply adding clarity to the research

questions, the term self has been defined as an individual's consciousness of
his or her own being, the state of having a distinct identity.
The research questions include:
1. What is the symbolic and contextual experience of gay men
with HIV infection taking HA ART regimens and how does this
experience influence adherence to treatment?
2. What is the nature of the contextual experience associated with
the interaction between the individual and the medical
institution in the urban HIV/ AIDS setting, and how might this
experience influence adherence behaviors?
3. What is the experience of individual perception of the self and
how does this perception influence and inform the issue of
Each of the research questions was developed to address the issue of
adherence in a substantive way, from a particular perspective based on
certain theoretical assumptions from Symbolic Interactionism, the work of
Erving Goffman and Arthur Kleinman. These assumptions are discussed in
Chapter 3.
The research questions are interrelated in that the first was intended
to garner general knowledge regarding the adherence issue for the purpose

of discovery and enlightenment, and to begin to illicit the rich context that I
have espoused as being critical to the research. This included the
participants' life circumstances, histories and experiences relative to HTV-
infection, as well as their individual approaches to caring for and treating
their illness.
The second research question brings in the reflexive nature of
interaction between structure and agency and was intended not only to build
on the data derived from question one, but also to incorporate an exploration
of the proposition that adherence behaviors should not be viewed merely as
the result of actions by individual patients, but rather in the structural field
in which they may subsist. Specifically this involved an exploration of the
interaction of the individual with the medical institution from the
individual's perspective.
Finally, the third research question was intended to concentrate on the
role of the self in adherence behaviors. Data that were collected during the
initial in-depth interviews were preliminarily analyzed to assist in focusing
and framing subsequent interview questions for the further examination of
this complex component that is implied in the literature as being important,
yet was previously addressed in only the most one-dimensional manner.

Scope of the Research
This study was conceived and implemented as a qualitative analysis
of the research questions with the goal of gaining substantive insight into the
depth of the complex human issue of non-adherence. The research is based
on a constructivist worldview and is supported in this philosophy by the
design, guiding theory, and analysis.
The study was conducted within the University of Colorado Hospital
Infectious Disease Group Practice (IDGP), in Denver, Colorado, utilizing a
sample of 18 men and a total of 24 in-depth interviews. While the clinic
setting offered considerable population diversity, only gay men were
selected as participants. This was done to maximize the likelihood of
recognizing the emergence of common themes in the data and to minimize
the likelihood that such themes would be so diverse as to have little
interrelatedness (See Chapter 4 for complete explanation of the sampling
One must, of course, understand that while the utilization of this
approach has added to the depth of knowledge, particularly with respect to
applications of theory, the findings are relative only to the individuals
studied. As in all qualitative research, this inquiry did not seek to establish

correlations or causal relationships, nor was there intent to apply the
findings in a manner that would be generalizable to a population. As
legendary astronomer Carl Sagan often said, "Extraordinary claims require
extraordinary evidence" (Sagan, 1980).
The results from this research do not proffer extraordinary claims.
Rather, the scope of this work was to enhance a foundational insight on
which to base a larger conceptual understanding. It is entirely possible that
future researchboth qualitative and quantitativemay be undertaken to
more fully develop and explore the findings from this work.
Organization of the Thesis
Following this brief chapter of introduction and overview, there is a
comprehensive review of the literature. This review includes a complete
description of the biomedical imperative for adherence in HTV treatment, as
well as a rigorous overview of current research findings on the topic. The
review of the literature not only clarifies the currentalbeit limited
understanding of the issue of adherence, but also serves as a means on which
I base my argument for the need for additional qualitative evaluation of the
topic. In the literature review, I have also included a review of the

previously utilized theoretical approaches that have further framed and
clarified my own theoretical positions and approaches.
Following the literature review is an in-depth look at the theoretical
basis through which this research was developed, including a discussion of
Symbolic Interactionism. Included is a summary of the work of George
Herbert Mead who is sometimes referred to at the progenitor of Symbolic
Interactionism. This is followed by a discussion of the work of Irving
Goffman whose ideas on the self provided some of the basic assumptions of
the research. While the findings in this study did not support Goffman's
predictions as I had initially interpreted them, his ideas remained an
essential source of understanding and illumination in the project. Finally, I
have included some of the theoretical concepts that have been put forth by
Arthur Kleinman, whose work has added clarification and depth to the
A detailed look at the qualitative methodology is then presented
including sampling strategies, question-development, interview processes
and data analyses. This chapter also includes a discussion of some of the
components that comprise the culture of the "field" in which the research
took place. Several photographs are included that were taken from a larger

group collection. These photographs are considered part of the data and
were coded along with the in depth interviews and field notes.
Chapters 5 and 6 are a comprehensive presentation of the findings
from the data analyses, discussion of the findings and application to the
underlying theoretical concepts. ,

During the inception of this research in late 1999, a comprehensive
Med Line literature search revealed less than 30 citations dealing directly
with the topic of HIV infection and adherence. By Mid 2005, a repeat search
using the same key words with a time limit of five years yielded well over
1000 citations. Clearly as clinicians and researchers, as well as governmental
sponsors such as the National Institutes of Health, have become more
attentive to the profound impact of non-adherence on treatment success, the
magnitude of research on the topic has increased dramatically.
The literature review in its present form has been derived from
published research ranging from 2006 back as far as 1989, though the major
focus is on the time frame between 1998 and 2006, which captures the
HAART era of antiretroviral therapy relative to the adherence topic. In the
initial review of 1175 abstracts, those studies dealing with adherence as it
relates to specific antiretroviral drug regimensusually a part of a drug

efficacy clinical trialwere eliminated as were most studies dealing with
populations of individuals outside the United States.
With respect to the state-of-the-science based on mostly quantitative
research, there is a significant amount of recurrence in the literature of
certain well-defined factors correlated with non-adherence. These factors
include psychiatric illness and substance abuse, minority and gender issues
and medication side effects. Also included are concepts such as level of
individual self-efficacy, locus of control and coping strategies. Hence, studies
selected for this review are representative of and best depict those findings
that have been replicated numerous times.
Qualitative research is seen much less frequently in the current
literature and the studies selected for this review are those that contain
relevance to the research questions of this study, particularly with respect to
the concepts of symbol, self and individual, and institutional interaction.
The literature review has been organized into several subsections
including a look at the biomedical necessity for high level adherence,
followed by a thorough examination of quantitative and qualitative findings.
To give strength and contrast to the literature review, some citations have
been included that deal with adherence issues outside of HIV/ AIDS in areas

such as diabetes mellitus, rheumatoid arthritis and multiple sclerosis.
Representative studies that explore the biometrics of adherence level
measurement have also been included as well as a review of theoretical
positions that have been utilized in the body of knowledge as it currently
Finally, there is a critical summation of the current state-of-the-science
with respect to adherence research that forms a rationale for the present
research approach.
The Biomedical Imperative for Adherence
In mid-1996, clinicians began observing the successful suppression of
plasma HIV RNA levels with HAART regimens in individuals infected with
HIV. This pharmacologically induced viral suppression frequently resulted
in an increase in CD4 lymphocyte counts and recovery of cellular immunity.
This event had profound implications for those with HIV-infection and not
only decreased the incidence of AIDS-associated morbidity and mortality,
but also effectively transformed HTV-infection and AIDS from a terminal
illness into a chronic, manageable illness (Detels, Munoz, McFarlane,

Kingsley, Margolik, Giorgi, Schrager, &Phair, 1998; Echeverria,
Jonnalagadda, Hopkins, & Rosenbloom, 1999; Vittinghoff, et al. 1999).
However, because of the tendency for HIV to develop genotypic
resistance to specific antiretroviral agents, high-level adherence to prescribed
therapy has been of extreme importance. For example, genotypic resistance
patterns such as the presence of the M184V codon mutation on the reverse
transcriptase gene which confers full resistance to the drug lamivudine, and
many similar mutations, are highly correlated with individual non-
adherence (Descamps, Flandre, Calvez, Peytavin, Meiffredy, Collin,
Delaugerre, Robert-Delmas, Brasin, Aboulker, Raffi, & Brun-Vezinet, 2000).
Early experience with initial HAART regimens suggested that individuals
must achieve and maintain greater than 95 per cent adherence levels in order
to prevent the evolution of viral resistance and hence the maintenance of full
HIV RNA suppression of less than 50 copies per milliliter of blood (Bartlett,
Chesney, & Farthing, 2000; Patterson, Swindells, & Mohr, 2000).
Non-adherence to HAART regimens has been strongly associated
with treatment failure (Carpenter, Cooper, Fischl, Gatell, Gazzard, Hammer,
Hirsch, Jacobsen, Katzenstein, Montaner, Richman, Saag, Schechter,
Schooley, Thompson, Vella, Yeni, & Volberding, 2000). Data suggest,

however, that non-adherence is common and may be present in 50% to 70%
of patients (Chesney, Ivkovics, Hecht, Sikipa, & Rabkin, 1999; Chesney,
Ickovics, Chambers, Gifford, Neidig, Zwickl, & Wu, 2000; Nieuwkerk,
Sprangers, Burger, Hoetelmans, Hugen, Danner, van Der Ende, Schneider,
Schrey, Meenhorst, Sprenger, Kauffmann, Jambroes, Chesney, de Wolf, &
Lang, 2001).
The relationship between adherence levels and the emergence of viral
resistance has traditionally been viewed as a linear process similar to a bell
curve with viral resistance occurring less frequently both at very low level
adherencebecause of the absence of selective drug pressureand at high
level adherence because of full viral suppression (Friedland& Williams,
1999). As HAART regimens have become more advanced in terms of
increased potency, decreased pill burden and improved side effect profiles,
the understanding of genotypic viral resistance has also become more
sophisticated. For example, Bangsberg, Porco, Kagay, Charlebois, Deeks,
Guzman, Clark & Moss, (2004), recently conducted mathematical modeling
of protease inhibitor resistance patterns which suggested that in a treatment
experienced population, maximum drug resistance occurs at the 87 per cent

adherence level (very close to the 70 to 80 per cent adherence ranges
commonly observed in clinical trials).
This research suggests that the highest risk for the development of
drug resistance occurs at the most commonly observed levels of adherence
and underscores the biomedical imperative for fastidious adherence if
treatment with HAART regimens is to be successful both in the short and
long term (Bangsberg et al. 2004).
Factors Associated with Non-Adherence
Factors associated with non-adherence across numerous chronic
disease states, including HIV infection, have included compilations of
attributes and circumstances that have been frequently reproduced in the
scientific literature. One of the most comprehensive studies of non-
adherence in HIV infection has been the Healthy Living Project sponsored by
the National Institutes of Mental Health (NIMH). With a well-powered
sample (n=2735), this theory-based cross sectional study reinforced previous
research suggesting that non-adherence is correlated with factors such as
substance abuse, being part of an ethnic or racial minority, homelessness and

poverty and feelings of discouragement about taking chronic medications
(Johnson, Catz, Remien, Rotheram-Borus, Morin, Charlebois, Gore-Felton,
Goldsten, Wolfe, Lightfoot & Chesney (2003).
This same NIMH study also found that certain attributes may
actually be protective against non-adherence behaviors, These included the
presence of high levels of adherence self-efficacy (the personal belief that one
has the ability to adhere to the prescribed regimen), being able to manage
side-effects and individual belief that non-adherence will make the virus
stronger (Johnson et al. 2003).
The presence of adverse medication side effects has been correlated
with non-adherence (Ickovics, Cameron, Zackin, Bassett, Chesney, Johnson
& Kuritzkes, 2002; Miller, 1997), as has the tendency for patients to forget to
take the medications, particularly if regimens are too inconvenient and pill
burdens are too high (Gifford, Bormann, Shively, Wright, Richman &
Bozzette 2001; Singh, Berman, Swindells, Justis, Mohr, Squier & Wagener,
1999). Other studies have suggested that additional convenience-related
issues including dietary restrictions and high dosing frequency are also
related to non-adherence behaviors (Chesney, 2000; Miller, 1997).

Socioeconomic and demographic correlates of non-adherence have
been identified in large, multi-center inquiries as being younger age (specific
age cut-off undetermined), female gender, non-Caucasian ethnicity, low and
very low income levels, and less than high school education level (Johnson,
et. al, 2003; Kleeberger, Phair, Strathdee, Detels, Kingsley & Jacobson, 2001).
These constituents have been frequently identified in smaller studies as well
(Catz, McClure, Jones & Brantley, 1999; Gordillo, delAmo, Soriano &
Gonzalez-Lahoz, 1999; Laine, Newschaffer, Zhang, Cosier, Hauck & Turner,
2000; Magnus, Schmidt, Kirkhart, Schieffelin, Fuchs, Brown & Kissinger
Similar socioeconomic and demographic findings have been seen in
studies of non-adherence related to cancer screening (Jennings-Dozier &
Lawrence, 2000; Myers, Hyslop, Wolf, Burgh, Kunkel, Oyesanmi & Chodak,
2000) and hypertension management (Rose, Kim, Dennison & Hill, 2000).
While the treatment options for individuals with HIV-infection have
increased, the social and cultural marginalization of those infected has made
this treatment more difficult to administer (Altice et. al, 2001). Complicating
this marginalization are psychiatric and substance addiction co-morbidities
which are common in HIV infection and have been correlated with non-

adherence to HAART regimens (Angelino & Treisman, 2001). Major
depressive disorders among persons with HIV-infection may be as high as 50
per cent, bipolar disorder as high as 20 per cent and personality disorders
may be in the 20 to 25 per cent range (Angelino & Treisman, 2001).
Addiction co-morbidity, highly correlated with non-adherence, ranges
from 25 to 95 per cent depending on the population studied (Adler-Cohen &
Jacobson, 2000). Substance addiction of all types is implicated in non-
adherence in numerous studies that have examined multiple factors
(Johnson & Catz, 2003; Kleeberger, Buechner, Palella, Detels, Riddler,
Godfrey & Jacobson, 2004; Palmer, Salcedo, Miller, Winiarski & Amo, 2003;
Hinkin, Hardy, Mason, Castellon, Durvasula, Lam & Stefaniak, 2004; Samet,
Horton, Meli, Freedberg & Palpepu, 2004; Tucker, Burnam, Sherboume,
Kung & Gifford, 2003).
Attempts to equalize social power differentials between medical
providers and patients have been associated with improved adherence,
though outcome data in this area are limited (Buchmann, 1997; Frank,
Kupfer & Siegel, 1995; Rodin & Janis, 1979). One study reported that a
decrease in patient perception of "powerful others" (through an intervention
that promoted equality between patients and providers) was correlated with

decreases in cynical hostility by patients toward the medical environment.
This study hypothesized that the decreases in cynical hostility would be
associated with increases in medication adherence (Christensen, Wiebe &
Lawton, 1997). The establishment and development of favorable physician-
patient relationships has been linked to increased adherence in numerous
chronic diseases. Quantitativeand to a far lesser extent qualitativestudies
have suggested that when patients perceive a supportive, interactive
relationship with their provider, the level of adherence is higher (Lo, 1999;
Miller, 1997; Mostashari, Riley, Selwyn & Altice; Meyers et. al, 2000; Rose, et.
al, 2000; Williams, Rodin, Ryan, Grolnick & Deci, 1998). This individual
perception of external support often extends into family and social realms
where higher levels of adherence are associated with higher levels of
perceived family and social support (Catz, et. al, 1999; Gordillo, et. al, 1999;
Lo, 1999; Magnus, et. al, 2001). These studies suggest an important element
may exist with respect to the individual interaction with the medical
institution as a social institution. Research question 2 of this study attempted
to address this phenomenon. In addition, much of the theoretical
perspective presented in association with this study related to the structure-

agency interaction that is implicit, but not extensively explored in the
Limited research involving additional aspects related to the individual
has addressed a deeper, more complex and more challenging component of
the adherence issue. For example, factors such as individual self-perception
and self-identity or sense of selfhood are relatively abstract and have been
difficult to operationalize and quantify. In some studies the "self" has been
operationalized as self-efficacy, which Bandura has defined as individual
belief in ability to bring about a desired outcome. Higher levels of self-
efficacy have been correlated with higher levels of adherence (Brus, Van de
Laar, Taal, Rasker & Wiegman, 1999; Kneckt, Syrjala, Laukkanen &
Knuuttila, 1999; Mohr, Boudewyn, Likosky, Levine & Goodkin, 2001).
Self-efficacy has been associated with self-esteem and self-identity.
Higher levels of self-esteem have been correlated with higher degrees of self-
efficacy and higher levels of adherence (Fraser, Hadjimichael & Vollmer,
2001; Kneckt, Keinanen-Kiukaanniemi, Knuuttila & Syrjala, 2001). This has
suggested that self-esteem and self-identity may be expressed through self-
efficacy behaviors and that measures of self-efficacy may also be indirect
measures of self-esteem and self-identity. Further, self-efficacy has been

related to questions regarding individual perception of social power and the
ways in which patients interact with the medical system. Patients with
greater knowledge regarding their disease and a belief that the disease can
be managed effectively are more likely to be adherent to treatment (DeGeest,
Moons, Dobbels, Martin & Vanhaecke, 2001; Miller, 1997; Myers, et. al, 2000).
By contrast, belief systems that are based in hopelessness have been
correlated with poor adherence (Catz, et. al, 1999), as have irrational belief
systems such as conspiracy theories about the origins of disease
(Christensen, Moran & Wiebe, 1999). Both of these health belief systems have
frequently been associated with self-efficacy (DeGeest, et. al, 2001; Gifford,
et. al, 2000).
The formation of personal health belief systems has been associated
with individual appraisal of threat, thus the perception of less severe disease
has been linked with lower levels of adherence in various chronic illnesses.
10'24 Perceived equality of social power, higher self-efficacy and self-esteem
have also been related to autonomous self-regulatory behaviors where
individuals utilize self-efficacy for self-advantageousor pro-agency
behaviors, rather than agency-detrimental behaviors (Bandura, Caprara,
Barbaranelli, Pastorelli & Regelia, 2001; Williams, et al, 1998). Trust in the

medical care system and trust in societal institutions in general are positively
correlated with increased levels of HAART adherence, though these specific
variables have not been clearly linked to self-efficacy (Altice, et. al, 2001).
Self-efficacy has been correlated with other variables concerning health
beliefs such as locus of control (identification with the notion that one has
control of one's future as opposed to external entities) and coping style. For
the latter, higher levels of self-efficacy are related to problem-oriented coping
styles and such coping patterns have been predictors of higher adherence
levels across chronic disease processes (Avants, Warburton, & Margolin,
2001; Goldbeck, Baving & Kohne, 2000; Singh, et. al, 1999). In contrast, the
perception of external locus of control is associated with non-adherence
(Christensen, et. al, 1997; Goldbeck, et. al, 2000; Murphy, Foreman, Simpson,
Molloy, & Molloy, 1999; Tollotson, & Smith, 1996).
Measurements of Adherence
The difficulty of implementing adequate measurements for adherence
is obvious throughout the literature. There is no "gold standard" of
measurement. Though generally regarded as less than ideal, one of the most

common measurements has been patient self-report. Studies involving
multiple disease processes have correlated patient self-report with other
mechanisms of measurement such as pill count or plasma drug levels
(DeGeest, et. al, 2001; Gifford, et. al, 2000; Gordillo, et. al, 1999; Wagner,
Justice, Chesney, Sinclair, Weissman & Rodriguez-Barradas, 2001). Biological
markers such as Hemoglobin A1C values in diabetes mellitus or plasma HIV
RNA levels in HIV-infection have added some objectivity to the assessment
of adherence (Christensen, et. al, 1999; Kneckt, et. al, 1999; Kneckt, et. al,
2001). These measures, however, are subject to significant individual
variability and in the case of HIV-infection detectable plasma HIV RNA
levels could result from causes other than non-adherence (Bangsberg, Hecht,
Clague, Charlebois, Ciccarone, Chesney & Moss, 2001; Gifford, et. al, 2000;
Lucas, Cheever, Chaisson & Moore, 2001; Patterson, et. al, 2000; Rabkin,
Ferrando, van Gorp, Rieppi, McElhiney & Sewell, 2000; Wagner, et. al, 2001).
The measurement of serum drug levels has demonstrated efficacy in
evaluating adherence levels, though again, these levels are subject to
individual clinical variation and to factors other than non-adherence, which
may produce variation (Nieuwkerk et. al, 2001). The use of patient pharmacy
refill records has been widely used in research as an indirect measurement of

adherence and in some cases has correlated with other concurrent measures
such as self report (Avants, et. al, 2001; Laine, et. al, 2000; Singh, et. al, 1999).
In the clinical trial setting, electronic medication event measuring devices
have been employed to correlate number of container openings with
scheduled doses (DeGeest, et. al, 2001). Since no clear "gold standard" for
adherence measurement has been identified, methodologies that have
employed more than one assessment modality have been considered more
rigorous (DeGeest, et. al, 2001; Gifford, et. al, 2000; Gordillo, et. al, 1999; Lee,
Greene, Douglan, Grim, Kirk, Kusek, Milligan, Smith & Whelton, 1996;
Williams, et. al, 1998). This study defines adherence based on both patient
self-report and documented HIV RNA levels.
Theory in the Current Literature
A review of the adherence literature suggests that the use of guiding
theory is generally uncommon. A preponderance of the research involving
this topic has been approached through biomedical methodologies that have
typically not employed theory. Moreover, in reviewing this literature one
finds a clear distinction between those studies that have been philosophically

and methodologically guided by theoretical underpinnings, and those that
have simply garnered theoretical constructs from other research and
employed them as independent variables.
Theories pertaining to health behaviors such as the Transtheoretical
Model and its constructs that have been merged from the Stages of Change,
Decisional Balance and Processes of Change models have been identified as
being the guiding foundation of several studies reviewed (Buckworth &
Wallace, 2002; Willey, Redding, Stafford, Garfield, Geletko, Flanigan,
Melbourne, Mitty & Caro, 2000).
The Theory of Reasoned Action with its emphasis on the subjective
norm, that is, the belief about whether an associated referent approves or
disapproves of a certain behavior, has been utilized in two of the studies
reviewed (Syrjala, Niskanen & Knuuttila, 2002; Smith & Biddle, 1999).
Constructs from the Theory of Reasoned Action have been expanded in the
Theory of Planned Behavior, which adds the construct of perceived behavior
control (a concept somewhat similar to self-efficacy) along with the
subjective norm. Planned Behavior is presented as the guiding theoretical
basis in three adherence studies reviewed (Elliott, Armitage & Baughan,
2003; Murgraff, McDermott & Walsh, 2001; Smith & Biddle, 1999).

One study reviewed with respect to adherence issues utilized
Prospect Theory. Originally developed by Tversky and Hahneman, this
theory refers to the "framing effects" of individual perception. The premise
involves the assumption that individuals will likely view situations of certain
gain as more valuable than those situations of possible gain. Conversely,
individuals will view possible loss as more favorable than certain loss (Finney
& Iannotti, 2002). The well-researched Health Belief Model that supports
individual appraisal of perceived risk versus benefit has been utilized to
some degree in the adherence literature (Christensen, et. al, 1999; Tanner,
Craig, Bartolucci, Allon, Fox, Geiger & Wilson, 1998).
Research geared toward interpersonal approaches to health behavior
primarily includes the use of Social Cognitive Theory. In the early 1960s
Albert Bandura proposed a framework designed to broaden the
understanding of Social Learning Theory that had been previously
developed by Miller and Dollard nearly two decades before (Bandura, 1962;
Baranowski, Perry & Parcel, 1997). Social Learning Theory had emphasized

the significance of the roles played by vicarious, symbolic and self-regulatory
processes in human behavior (Bandura, 1977). Theoretical propositions had
been developed around research findings that linked behavioral response
associations to individual experience (Bandura, 1962; Bandura, 1986).
Bandura had further suggested that the learning response involved a
reciprocal interaction between individual agency and social structure, and
introduced the concept of the self into this interactive process (Bandura, 1962;
Bandura, 1986).
By 1995 Bandura had stipulated that self efficacy was the major
theoretical construct underlying the core components of thought, motivation
and action which he suggested operate in reciprocal concert with outside
influences (Bandura, 1962; Bandura, 1986; Bandura, 1995). Self-efficacy was
described as individual belief in one's capability to organize and execute
one's own course of action required to produce given attainments
(Baranowski et. al, 1997). In comparative studies Bandura showed that self-
efficacy beliefs were more predictive of behavior change than self-esteem
beliefs and that self-esteem perception is less predictive of behavior change
when individual discernment of self-efficacy is factored out (Bandura, 1995;
Baranowski et. al, 1997).

Bandura argued that in understanding and interpreting the role of
intentionality in human agency one needs to distinguish between the unique
action itself and the effects of the action on the surrounding environment.
Thus, mediated by level of self-efficacy, individuals may contribute to their
own pain and suffering, or resolution of these afflictions through intentional
behaviors (Bandura, 1962; Bandura, 1986; Bandura, 1995). Bandura
correlated high levels of self-efficacy with what is termed prosocial behavior,
that is, behaviors which are agency supportive, and low levels of self-efficacy
with moral disengagement behavior, that is, behaviors which are agency
detrimental through intention (Bandura, 1995; Bandura et. al, 2001). Moral
disengagement may result in behaviors that are transgressive, or anti-
institutional (Bandura et. al, 2001).
In addition to self-efficacy, other important constructs of Social
Cognitive Theory include environment, reinforcements and situation, which
form the basis for the dynamic interaction of the person, the behavior and the
environment known as reciprocal determinism (Bandura, 1962; Bandura, 1977;
Baranowski, et. al, 1997). Most of the research reviewed with respect to
adherence essentially ignores reciprocal determinism, and hence the
structure-agency interaction. In addition, numerous studies of adherence

that have utilized the construct of self-efficacy have not necessarily
employed this construct within the context of its association to Social
Cognitive Theory (Bandura, 2001; Brus, et. al, 1999; Buchmann, 1997;
Chesney, et. al, 2000; Fraser, et. al, 2001; Kneckt, et. al, 1999; Kneckt, et. al,
2001; Mohr, et. al, 2001; Tanner, et. al, 1998; Smith, Rublein, Marcus, Brock &
Chesney, 2003; Senecal, Nouwen & White, 2000).
While it was tempting to include Social Cognitive Theory as a guiding
basis in the present proposal, the concept of self that is explicit in the work of
Mead and Goffman is fundamentally different than the self-efficacy
construct of Social Cognitive Theory. Whereas Bandura may view self-
efficacy as a surrogate marker for other conceptions of the self such as self-
esteem or self-identity, Mead and Goffman would likely view self-efficacy as
simply an individual component that makes up a larger conception of the
Coping Strategies
Several theory-based conceptual models have been developed around
the phenomenon of coping with illness. These models emanate from various

disciplines and range from elaborate, well-researched theoretical frameworks
to simple constructs. One of the better developed works has been the Corbin
and Strauss Illness Trajectory Model (Corbin & Strauss, 1991;Corbin, 1998).
In this model, the concept of coping is placed within the context of a series of
phases beginning with the onset of a particular illness process and extending
in a linear fashion toward a variety of outcomes ranging from health stability
to involvement with the dying process.
Elements of the Lazarus Schema of Coping and Adaptation and the
Schlotffeldt Health Seeking and Coping Paradigm have been described in the
coping literature. These models view human coping as a transactional
process. Interactive variables such as situation, personal factors, resources
and socio-demographic components are interrelated with individual health
goals and cognitive appraisal (Lazarus, Averill & Opton, 1974; Nyamathi,
1988). The cognitive appraisal component of the model involves the
constructs of emotion-focused versus problem-focused coping strategies that
are frequently variables of interest in the coping literature. Interestingly, the
models from the coping literature generally ignore influences from cultural
or other structural influences. Also interesting is that while models from the
coping literature have been frequently included in health-illness research, the

coping literature and the adherence literature have not intersected in a
significant way.
Locus of Control
Variables such as coping strategies and locus of control are present in
some adherence research but have not typically been presented within the
context of a theoretical framework (Avants, et. al, 2001; Murphy, et. al, 1999;
Singh, et. al, 1999; Tollotson & Smith, 1996).
From a community and group intervention perspective, only one
study reviewed has been supported by theoretical propositions. Remin and
colleagues utilized Social Action Theory in a qualitative study assessing
barriers to adherence in diverse populations. This theory focuses on the
nature of communities and the social systems that exist within them (Remin,
Hirky, Johnson, Weinhardt & Whitter, 2003).
Finally, three studies reviewed profess grounded theory, a qualitative
methodology as well as a technique for analysis, as their guiding basis
(Power, Tate, McGill & Taylor, 2003; Wilson, Hutchison & Holzemer, 2002;
Schilder, Kennedy, Goldstone, Ogden, Hogg & Oshaughnessy, 2001). The

connotation of grounded theory as a function of data analysis may lead to
the generation of new theoretical concepts that are grounded in the data
obtained during the course of the study. This proposed research utilizes
grounded theory as a data analysis technique, in this case primarily geared
toward description rather than theory generation.
Limitations of Previous Work
Previous research involving the health issue of adherence is limited
both in its depth and in its breadth. The presence of qualitative methods in
the scientific literature is relatively uncommon suggesting that preliminary
work aimed at identification of relevant concepts has not been adequately
carried out. This lack of groundwork has led to a limited understanding of
the issues that may actually be involved, as well as the context in which they
may exist. Structural factors outside the individual such as cultural
influences have been largely omitted or ignored. Moreover, the reliance on
quantitative approaches supported by positivist theoretical perspectives has
significantly limited the types of research questions that can be asked about
adherence behaviors and how such questions might be answered.

The rather large body of quantitative work that has been done with
respect to the issue of adherence has been limited by methodological and
validity problems which characterize this field of research. There has been
difficulty in operationalizing the constructs of adherence (the outcome
variable), and the numerous independent variables that have been tested.
Moreover, previous research has limited applicability to diverse populations
and small sample sizes have often minimized the generalizability of the
Though this trend may be changing, considerable research in this field
has not been informed or guided by creative use of multi-disciplinary
theoryparticularly the creative use of theory that supports and validates
the reflexive nature of human interaction. For example, a large portion of the
scientific literature places responsibility for adherence exclusively on the
patient while the interaction between the patient, the medical institution and
the host of other reciprocal constituents including meaning and symbol have
been largely ignored. The use of symbolic interactionism as a philosophical
basis that supported a constructivist methodology was chosen for this study
as a means of investigating the validity the reflexive nature of adherence.

"The practice of reserving something of oneself from the clutch of an institution is very visible in
mental hospitals and prisons but can be found in more benign and less totalistic institutions too."
(Goffman, 1961).
The quest for the development of common principle has historically
been the keystone of most scientific and humanistic domains of knowledge
(Grudin, 1990). In the behavioral sciences the construction of theoretical
concepts to guide inquiry has undoubtedly led to new insight and to a
beneficial integration of complex phenomena, though at times the use of
theory has been limited by positivist methodologies, deductive analyses and
subsequent one-dimensional findings (Craib, 1984; Ritzer, 1975).
As demonstrated in the review of the literature, in the investigation of
a human issue as complex and multidimensional as adherence, research
questions posed from a positivist epistemology typically yield quantitative
findings that can be statistically analyzed and interpreted. Conversely,

proceeding from a philosophical position that is constructivist in its tenants
and contextual in its methodological approach allows researchers to pose
questions that capture the depth and meaning that could ultimately result in
a deeper fundamental understanding of the phenomenon. In the interest of
furthering this depth of understanding that has been identified as lacking in
the current literature and to be consistent with the research questions posed
in this study, I have elected not only to pursue a method that is constructivist
in its philosophical basis, but also to approach the inquiry with a developed
body of underlying theory.
There is disagreement among scholars regarding the incorporation of
theoretical concepts into ethnographic research. Certainly the use of
underlying theory carries with it certain risks such as the potential for
findings that are, however unintentional, formed to fit theoretical concepts of
which the researcher may have significant emotional or academic
investment. Others, however, have suggested that qualitative research is
rarely a purely inductive process and that it often includes the use of
deductive elements such as common sense expectations, certain stereotypes
and theoretical concepts (Hammersley & Atkinson, 2000).

In this ethnography, I have chosen to incorporate theoretical concepts
for two main reasons. First, it has been desirable to place the work on an
epistemological foundation that not only supports a reflexive, constructivist
worldview, but one that also philosophically supportsif not mandates
the use of a qualitative methodology. The second reason for focusing the
research on previously developed theoretical ideas is based on the
understanding that there exists a developed body of knowledge within the
behavioral sciences that may be highly applicable to the current research
questions, but has been almost completely unutilized in previous research.
In the investigation of the topic of adherence the use of theory, I would
argue, offers the potential for enhancement of expanded knowledge and
increased understanding.
In approaching the issue of adherence from an ethnographic
standpoint then, I have looked to a guiding paradigm that supports such a
position from both epistemological and methodological perspectives. Since,
in the research questions, I have proposed that meaning, symbol, self,
interaction and context are critical elements that are lacking in the current
scientific research involving adherence, I have looked to the paradigm of

Symbolic Interactionism as an overall guiding theory, and with this the work
of George Herbert Mead.
In addition, Erving Goffman's work on the self and its interaction
with structural elements of society and culture has been both inspirational
and enlightening. Goffman's work combined with Symbolic Interactionism
allowed this study to explore another critical element that is missing from
the adherence literature as well as from the illness-coping literature. That is
the influence of structural elements such as culture on the behavior o the
Finally, I have included the theoretical work of contemporary
psychiatrist Arthur Kleinman whose work has included the experiential
perspective of chronic illness and the relationship of this illness to social
structuresparticularly the institution of medicine.
Symbolic Interactionism
To understand Symbolic Interactionism as a paradigm and as a
philosophical worldview, it is helpful to position this theoretical work in
relation to other classic paradigms as they relate to positivist and
constructivist visions. Although a clear understanding of the association

between Symbolic Interactionism and the constructivist position within the
philosophy of human science spectrum is essential, Symbolic Interactionism
must first be viewed from its primary roots in the social sciences.
Social theories have sometimes been presented as belonging to one of
three major paradigms that are based on the principles of numerous
theoretical frameworks on a spectrum ranging from positivist to
constructivist thinking. The first of these is the Social Facts Paradigm that is
based in the ideals of the structural-functionalism in a social system. This
paradigm supports the epistemological constituent of positivism in that it
primarily relates to those social structures, institutions and processes that are
accessible to inquiry through location and quantification (Craib, 1984; Ritzer,
1975). Structural Functionalism, particularly as it has been delineated
through the works of Parsons and Marx engages the debate between
structure and agency with a decided emphasis on the macro interaction
between institutional structures (Mouzelis, 1995). Durkheim, for example,
suggested that social facts should be related to and clarified by other social
facts and that these so-called macro-facts are interrelated to other macro-facts
(Mouzelis, 1995).

The second major sociological paradigm is that of Social Behavior. The
theorists in this group view the elements of social behavior as the most
relevant subject matter within the social sciences. Positions in this paradigm
held by theorists such as Skinner or Homans relate to such classic lines of
research as conditioned response and behavioral psychology. The research
in this paradigm is quantitative and the methods used for inquiry are
generally based on experimental designs (Ritzer, 1975).
By contrast, the Social Definitions Paradigm is a constructivist
perspective that is concerned philosophically and theoretically with the
manner in which individuals define and discover meaning in social
situations rather than the structures and institutions themselves (Ritzer,
1975). The symbolic interactionist view of George Herbert Mead
acknowledges the critical nature and immediacy of individual meaning and
context. In contrast to empirical methodologies of the social facts paradigm,
the theorists from the social definitions perspective have used a wide range
of more contextual, exploratory methodologies to address complex human
phenomena where meanings can be extracted, developed and interpreted.
The tenets that have collectively become known as the symbolic
interactionist perspective are rooted in the work of Mead and his

contemporaries and include the core constructs of individual and collective
acts, symbols, meanings and interpretations (Miller, 1973; Musolf, 2003).
Human beings are seen as carrying out actions toward things or events based
on the meanings these elements have and the context in which they occur.
The meanings may arise out of the social interactions one has with others
and are processed and modified through an interpretive process used by the
individual person in dealing with what is encountered (Miller, 1973). The
emphasis on meaning and interpretation distinguishes the nature of
symbolic interactionism from the more positivist viewpoints and provides a
unique perspective from which to proceed with inquiry that is primarily
interested in illuminating the subjective face of interaction between
individuals and between individuals and structure (Musolf, 2003).
Mead, an American philosopher and social theorist known for his
association with the Chicago School of Pragmatism, was sometimes referred
to as the father of symbolic interactionism. His work emphasized the
importance of individual experience and interaction with the self, as well as
with structural influences, as the primary process for development of the self
(Mead, 1982). Mead's writings focused intensely on the nature and function
of the self and rejected the prevailing mind-body dissociation of Cartesian

dualism. He emphasized that there is only a functional distinction between
what might be considered mental or symbolic processes and those actions
that are simply bodily behavior (Mead, 1964; Miller, 1973).
Mead's theoretical perspective valued the processes of experience and
asserts that the temporal dimension (context) of understanding cannot be
excluded from experience. The real human experience, he suggests, is not
timeless but consists of acts and events that form patterns of human
interaction and that are subject to contextual interpretation. Mead places
these acts and events in the framework of time and history out of which may
emerge new and, at times, unpredictable events (Mead, 1982; Musolf, 2003).
In addition to the general philosophical application of symbolic
interactionism to the present research, I have been particularly interested in
Mead's ideas on the nature of the individual self and self-interaction. Unlike
the behaviorists in the Social Behavior Paradigm, Mead's view of the self is a
much more complex entity than simply a reservoir that passively receives
and reacts to stimuli that come from structural influences such as values,
norms, roles and status. Moreover, he proposed that, while structural
components such as values, norms, roles and status may influence the
development and function of the self, they are not in themselves

deterministic. Rather, the self interacts with structural influences, in an
interpretive way that ultimately concludes in behavior. He also pointed out
that these interactive processes are also reciprocal in that they may move in
both directions (Mead, 1934). This interaction between structure and the self
is depicted in Figure 2-1.
*Adapted from Comments on Parsons as a Symbolic Interactionist (Blummer, 1975).
In symbolic interactionism, Mead pointed to two specific phases that
represent the experience of the self. He referred to the "I" phase that
represents the manner in which individuals respond to the influences of
others. This is the "acting" component where the self is the subject, and is in

contrast to the "me" phase that involves a group of attitudes that have
originated from others that an individual may assume. The "me" phase
represents the perspective on the self that one may learn from others. The
"me" phase is the "acted upon" component where the self is the object
(Mead, 1934).
One of the central tenants of symbolic interactionism involves Mead's
concept of the generalized other. In its most basic sense, the generalized other
refers to the organized influences of structure on agency (for better or worse,
the influences generated by the dominant institutions) that the self
internalizes and which become the basis on which the mature self is built.
One must be a member of a community, Mead says, to be a developed self
(Mead, 1934).
In essence then, symbolic interactionism is a perspective that has
contributed to the understanding of human nature and human experience, as
it exists in the reflexive relationship between social structure and human
agency (Mead, 1964; Miller, 1973; Musolf, 2003).

Erving Goffman
American social psychologist Erving Goffman (1922-1982), like many
others, was influenced by symbolic interactionism and by Mead's work,
particularly with respect to the nature and meaning of the self. Goffman's
work in the social sciences embodied the core constructs associated with
symbolic interactionism including acts, contextual meanings and
interpretations. Three of Goffman's most important works have contributed
to the development of the present research. These works include The
Presentation of the Self in Everyday Life (Goffman, 1959), Asylums, Essays on the
Social Situations of Mental Patients and Other Inmates (Goffman, 1961), and
Stigma: Notes on the Management of Spoiled Identity (Goffman, 1963). All three
of these works deal with self-interaction and the interaction between
structure and agency.
In his classic work The Presentation of the Self in Everyday Life,
Goffman pointed out that when an individual appears before others he could
have many reasons for trying to control the impression that is received as a
result of the presentation. Goffman suggested that it is individual experience
that is the joint, interactive influence of one individual on the immediate
presence of another (Goffman, 1959). In this work, which significantly

informed his subsequent writings, Goffman illustrated the action of common
individuals by placing them on a theatrical stage and portraying as acting
out their daily roles in this setting. This work collectively became known as
Social Exchange Theory and included the major construct of impression
management (Goffman, 1959). Impression management refers to the means
by which and individual may influence or control the impressions that
others may have on that individual.
In 1963 Goffman presented these rather abstract concepts involving
the presentation of self into more fully developed propositions in his
ethnographic portrayal of prisoners and mental patients entitled Asylums:
Essays on the Social Situations of Mental Patients and Other Inmates. In this
important work, Goffman framed contextual and experiential interaction in
the concept of the total institution, that is, the extreme structural setting where
care or confinement is given to those who cannot care for themselves, or who
would be a danger to those outside the institution (Goffman, 1961).
Characteristics of total institutions include the complete obstruction of social
interaction with the outside world symbolized by the physical barriers
present within the institution itself such as high fences, locked doors, and
electronic surveillance (Goffman, 1961).

As an explanatory model for the processes and meanings of what
Goffman viewed as compensatory behaviors of the inmates within the total
institution setting, he further proposed a means through which individuals
regulate their experiences within the extreme environment. He referred to
these compensatory behaviors as primary and secondary adjustments. He
described primary adjustments as those in which an individual willingly
contributes expected activities to an organization under required conditions,
in essence adjusting to the institution. Goffman suggested that in our society
institutionalized standards of well-being are supported with the motivation
from incentives and joint values by the prompting from designated penalties.
He postulated that the self plays an integral role in whether or not one
engages in primary adjustment behaviors (Goffman, 1961). Thus, in
Goffman's view, when individuals have a primary adjustment to an
institutional setting, those with a strong sense of self (hence not threatened or
intimidated by the total institution environment itself) can rationally assess
the personal advantage of cooperation with institutional requirementsif in
fact such and advantage existsand exhibit behaviors that are consistent
with the expectations of the institution and there is no conflict between
inmates and those in authority.

Goffman also described the secondary adjustment to an institution. In
this case, a member of an organization may perceive institutional
cooperation as a threat to the self and may employ unauthorized (non-
compliant, non-adherent) means to thwart the institution's assumptions as to
what is or is not in the best interest of that individual. In Goffman's
submission model, both primary and secondary adjustments are explicitly
linked to the self (Goffman, 1959; 1961; 1963; 1971). Goffman suggested that
primary adjustments might take place when the effects of the total institution
do not threaten loss of the individual self. Conversely, secondary adjustment
behaviors may emerge when there is intimidation to the self by the
overwhelming pressure of institutional structures and forces. In Goffman's
view, the unauthorized behavior may constitute a process of reaffirming the
self that has been threatened by the contextual interaction in the extreme
setting (Goffman, 1961).
The practice of reserving something of oneself from the clutch of an institution is
very visible in mental hospitals and prisons but can be found in more benign and
less totalistic institutions, too. I want to argue that this recalcitrance is not an
incidental mechanism of defense but rather an essential constituent of the self...
Our sense of being a person can come from being drawn into a wider social unit; our
sense of selfhood can rise through the little ways we resist the pull. Our status is
backed by the solid buildings of the world, while our sense of personal identity
often resides in the cracks (Goffman, 1961).

Goffman's work is particularly appealing with respect to the present
research in that primary and secondary adjustments are interesting concepts
involving the interaction between structure and agency and the implications
for the relationship of the self in such interactions.
Arthur Kleinman
Contemporary psychiatrist Arthur Kleinman pointed out the
powerful concentration of experience that is created by the presence of
serious illness. His work The Illness Narratives: Suffering, Healing and the
Human Condition, is a study in the processes through which individuals
discover the meaning and symbol associated with the illness experience and
how these meanings interact with the complex social structure that is
modern medicine (Kleinman, 1968). Kleinman addressed the issue of chronic
illness from an experiential (self) perspective and the relationship of such
experience to social structures. For example, he pointed out that one of the
unintended results of the transformation of the modern medical institution is
the continual deflection of the attention of the practitioner away from the
individual experience of illness (Kleinman, 1988).

In The Illness Narratives Kleinman presented explanatory models to
explore the fundamental relationship between the subjective illness
experience of patients and the objective experience of health practitioners,
both of which are influenced and informed by social and cultural meanings.
Often such relationships are in conflict both in terms of content and meaning.
Kleinman pointed out that such conflictsor conflicting explanatory
modelsare representative of the "cultural flow of life experience." Patient
explanatory models are often filled with symbol and relevant meanings that,
when validated by practitioners offer not only important insight into illness
experiences, but also offer validation to the self in a way that is not unlike the
primary and secondary adjustments suggested by Goffman (Kleinman,
Kleinman proposed that the purpose of medicine is both "control of
disease processes and care for the illness experience"(Kleinman, 1988). In
Kleinman's view, the relationship between agency and structure in the
setting of chronic severe illness transcends the need to simply control
symptoms. Care for the difficulties encountered because of the presence of
the illness is the primary concern. He suggested that the social and cultural
pressures that influence the medical institution are rooted in the manner in

which medical education is conducted; in the health care system itself which
gives validity only to the structures of biomedicine such as hospitals and
clinics, and to medical research which has not traditionally been informed by
the medical social sciences and humanities (Kleinman, 1988).
As a guiding philosophical and methodological paradigm, symbolic
interactionism, with its emphasis on acts, meanings and contextual
interpretations was well suited to this research. By definition, this
perspective supports an in-depth exploration of phenomena. This is
apparent not only in the pursuit of contextual meanings that are immediately
relevant to the individual, but also in the focus on the relationship between
the individual and the sociocultural structures in which the individual may
interact. Symbolic interactionism validates and provides an imperative for
evaluating the patterns of human interaction that are subject to interpretation
within a contextual perspective. This approach supports all three of the
research questions.
Coffman's work provides an interesting explanatory model for
adherence issues in that he has created the specific and complex constructs of

primary and secondary adjustments as they apply to the concept of the total
institution. These constructs address the presentation of the self and its
interaction with the medical clinic as a societal institution. Whether these
constructs can or should be applied to the health issue of adherence is
explored in this study. Goffman's work relates to all three of the research
questions in this study; however questions two and three, with their
emphasis on exploration of the nature of the self within the context of the
medical setting and the illness were specifically targeted.
The work of Kleinman is particularly relevant in the approach to the
development of questioning in the semi-structured interviews of the study.
Kleinman's concept of conflicting explanatory models suggests that while the
medical practitioner (underwritten by the medical community) may have
one idea as to how care should proceed in a given set of circumstances,
individual patients may have an entirely different concept. The processes
involved in this potentially discordant association call into play the core
concepts of symbolic interactionism including acts, meanings and
interpretations, and the idea of institutional oppression depicted in
Goffman's primary and secondary adjustment ideas.

The purpose of this study has been to gain deeper, contextual insight
and understanding into the health behavioral issue of adherence and non-
adherence in a sample of gay men with HIV-infection who have been
prescribed highly active antiretroviral therapy (HAART) regimens. The
qualitative methodology was selected based on two fundamental research
goals. First, findings from the literature review have indicated an overall
limited understanding of this issue, and while it has been broadly
investigated from an exterior perspective, a deeper understanding is clearly
lacking. Quantitative inquiries, based in a positivist epistemology, have
helped to establish correlations between adherence behaviors and factors
such as sociodemographic attributes, medication side effects, pill burden and
provider-patient relationships in certain populations with HIV-infection.
Moreover, these findings have been frequently replicated in the literature,
enhancing their validity. Nevertheless, the significance of these findings is
limited by their lack of depth and contextual meaning. For acquisition of this

needed insight into such a complex human behavioral issue as adherence,
quantitative findings are limited.
A qualitative approach, rooted in a constructivist epistemology offers
the potential for exploration of the depth and context that seem to be absent
in the current literature. This contextual depth includes the relationships,
meanings, symbols and experiences of the individuals who must take
chronic medications and the context in which these circumstances occur. The
understanding of these deeper issues may be critical to future successful
adherence-promoting interventions.
The second reason for choosing a qualitative design for this research
was based on my own interest in exploring the application of certain
theoretical propositions to the research questions. While quantitative
designs have been increasingly informed by theoretical positions from the
behavioral sciences, I believe that the complex theoretical concepts I was
interested in exploring such as meanings, symbols, reflexive interactions and
the relationships of these factors with the experience of self, could only be
implemented through a qualitative approach. To that extent the heart of this
research has been the exploration and application of theoretical ideas to the
experiential elements of the participants.

In reflecting on the range of possible qualitative approaches that
would be amenable to addressing the research questions, three main
methodologies were considered. First considered was the use of
phenomenology. As is the case with other non-quantitative approaches to
inquiry, phenomenology occupies a position not only as a respected
methodological technique, but also as an epistemological world view
(Patton, 2002; Bernard, 2002). Phenomenologists have rejected what they
believe to be a one-dimensional approach to knowledge that had been touted
by the positivists as being the definitive gateway to truth. While the
positivists sought to explain causal relationships of observable features,
phenomenologists, by contrast, searched for intense descriptions of the lived
experience (Husserl, 1989; Reeder, 1987; Cohen, 1987). It has been suggested
that a well done ethnographic work contains within it a phenomenology,
that is, a rich description of a component of a culture (Bernard, 2002;
Hammersley & Atkinson, 2002).
Grounded Theory is another qualitative approach that expresses both a
methodology in itself and also a set of systematic procedures used in
qualitative data analysis. Researches who are interested in building surface
theoretical associations from their own data have often used grounded

theory as a means of extracting certain core variables that may then be
developed into theoretical constructs (Glaser & Strauss, 1967). While tenets
from grounded theory were utilized in the analysis of the qualitative data in
this research (see section on data analysis) the methodology in itself did not
allow for the discovery of crucial reflexive cultural underpinnings that I felt
were necessary for the present research to have substantive meaning.
By contrast, the use of the ethnographic method, in its typical form,
allows the researcher (ethnographer) to be involved in the experience of the
participants within a culture. This involvement may consist of observing,
asking questions and examining elements of culture such as relationships,
learning, meaning and symbols. While in the present research I did not, per
se, participate in the daily lives of the individuals under study in the
traditional sense of a cultural anthropologist, I did appreciate from the
inception of the research, the deep cultural implications involved in the
research questions. I felt that the clinical setting in which the research took
place did, in fact, constitute a culture. And while the primary means of data
collection involved open-ended, semi-structured (and later even less
structured) interviews, the use of observation, primarily reflected in the field

notes, also proved to be an important component in the acquisition of the
Therefore, it was because of the reflexive, cultural-based nature of the
research questions that the ethnographic method was chosen for this
research. However, the reader should look for elements of phenomenology
and grounded theory in the body of this research which certainly played a
significant role in shaping the final work.
For purposes of clarification, it may be helpful to briefly discuss the
meaning of the term variable as it is used in the context of this research
design. In quantitative research, this term has typically been used to denote
observational entities on which measurements may be obtained (Glass &
Hopkins, 1996). In the qualitative venue, however, the utilization and
meaning of factors that may constitute variables is dissimilar to the
quantitative designation. For example, certain factors were identified in the
literature, which, while not positioned for measurement or statistical
application in this research, have definitely helped to shape and inform
critical decisions regarding the final design, particularly the sampling
strategy. Some of these factors, with respect to their association with non-
adherence, included individual sociodemographics, race and ethnicity, and

psychiatric and substance addiction co-morbidities. My references to these
factors as variables in the descriptions that follow are not meant to imply
that they themselves are under study. They are simply components that
must be considered based on findings in the review of the literature.
Research Location
In choosing the proper venue in which to carry out this ethnographic
research, I considered various locations that were available within the
Denver metropolitan area. There were three main criteria for selection. First,
it was necessary to have access to a large and diverse population of patients
with HIV/AIDS who were getting HAART regimens. I was seeking a setting
where patients would get multiple services in one place (medical, mental
health, pharmacy, laboratory) and have interactions with multiple staff
members, creating a kind of clinic culture. The second consideration was
logistical access to the research setting including reasonable proximity within
the Denver metropolitan area and hours of operation that could be
incorporated into a research schedule. Finally, it was critical to select a
research venue where I could gain entree. The privilege of gaining entree

into the field is a trademark of ethnographic research and includes logistical
elements such as acquiring permission from proper authorities to enter and
work within the designated setting. More than simple logistics, however,
acquiring entree involves finessing at least a beginning level of acceptance
and trust by those who are in the field, in this case staff, clinicians and
patients. These people needed to feel comfortable with me, and I with them.
Populations with diverse sociodemographics were available through
several agencies in the Denver area including the University Hospital
Infectious Disease Group Practice (IDGP) the Denver Health Medical Center
(DHMC) and the Veterans' Administration Medical Center (VAMC). While I
also had potential access to participants from various community private
practice settings, I eliminated these locations because they did not appear to
reflect the diversity in population I was seeking to reach, and moreover, did
not portray the cultural milieu I wished to engage.
Logistical considerations were of minimal significance since all three
of the potential locations were within relative close proximity and operated
on similar daily schedules. I gave the most weight for consideration of
location selection to the issue of gaining entree. This decision proved to be
complicated and was the subject of extensive discussion with colleagues,

research mentors and fellow research students. Part of the complication
involved my employment as a staff clinician at the IDGP facility. Some
argued that because of my affiliation with the University, I should choose
either the DHMC or VAMC facilities as a means of minimizing the
introduction of bias into the research sample. For example, because of a pre-
existing relationship with the researcher, participants might fear provider
retribution or feel they had to answer questions in certain affirming ways
that could bias the data. In the DHMC or VAMC settings I would be
generally unknown to potential participants, significantly decreasing the risk
of biased data.
I also considered the value of including participants from two or more
of the settings available which would not only limit potential bias, but would
provide a component of data comparability that could strengthen the
findings if such findings proved to be similar throughout the locations
studied. However, in my consideration of these options, I was quite aware
of the intimate nature of the subject matter I wished to address. This
intimate nature made the acquisition of entree into the research setting of
critical significance. Moreover, I believed that choosing the IDGP setting
would yield not only a high level of acceptance from both staff and patients

but also a significant apriori understanding of the culture that would benefit
the research. I felt that I would be a "familiar face" and that this familiarity
would yield a level of access and cooperation that could not be achieved in
the other settings. I was reasonably confident that this potential for
achieving the richest possible data in the IDGP setting outweighed the risk of
potential bias associated with participant knowledge of my previous
presence in the setting. I also believed I could incorporate elements in the
study design that would limit the introduction of sample bias (see section on
sampling strategy).
IDGP History, Culture and Demographics
The IDGP is a large hospital-based practice that is affiliated with the
University of Colorado Health Sciences Center in Denver, Colorado. While
the clinic layout itself is similar to that of any outpatient private practice, it is
set in the backdrop of not only a large urban hospital, but also a large
academic medical facility comprised of approximately 25 separate buildings
located on a campus in the heart of downtown Denver.
As it exists today, the University of Colorado Hospital building is the
focal point of the University of Colorado Health Sciences Center. The

campus comprises several city blocks bordered on the north and west by the
Hale and Congress Park neighborhoods respectively. At the western boarder
is Colorado Boulevard, a very busy six-lane thoroughfare, and to the north is
Eleventh Avenue. The eastern border extends to Clermont Street and to the
south is Eighth Avenue.
To understand certain current cultural and political influences in the
University Hospital and the University of Colorado Health Sciences Center
as the two entities co-exist as an academic medical facility today, and to
understand the impact of these influences on the IDGP and its patients, it is
helpful to look back to the original mission of the hospital. In 1958 the
University of Colorado Regents developed a master plan to serve the
residents of Colorado (and nearby states), which included the construction of
a 405-bed teaching hospital to replace the outdated Colorado General
Hospital. The plan further called for significant expansion of facilities for
research and teaching in the School of Medicine and the addition of a School
of Dentistry. This portion of the plan was completed in 1971 and included
the University Hospital building and a "research bridge" across Ninth
Avenue connecting to the School of Medicine.

The expansion and development of Colorado General Hospital into
the University Hospital as a part of the University of Colorado Medical
Center (later renamed the Health Sciences Center) had profound
implications for the mission statement of the institution. The relationship
between large academic medical centers, their affiliated hospitals and the
communities in which they operated had a long cultural and fiscal history
across the United States (Starr, 1982). The expansion of the institution was
conceived by the Regents in the last half of the 1950s, which was a time still
significantly defined by the confidence in post World War II American
By the time the new hospital was actually built and functioning in the
early 1970s, this climate had changed significantly in part due to events such
as economic recession and increased unemployment. There was a new and
deepening concern for the availability of health care, and doctors and nurses
were said to be in short supply. It was projected that in the period between
1959 and 1975, the numbers of doctors graduating from medical schools each
year would have to be increased from 7,400 to 11,000 (Starr, 1982). This
transformation, which enhanced the role of the academic medical center,
reinforced the long held unofficial agreement between communities and

teaching hospitals, which allowed for recruitment of patients into teaching
hospitals and clinics in exchange for the safety net these facilities provided to
the medically indigent. University Hospital and facilities like it were viewed
as friends of the public not only for their emphasis on cutting edge
technology, but because they had become the cornerstones of public access to
health care for the medically indigent. Indeed, serving the medically
indigent was a significant part of the University Hospital's mission
Over the past two decades, however, University Hospital, like many
health institutions affiliated with large academic medical centers, has
struggled to maintain financial solvency. In part, this financial concern has
led to the development of a corporate ethos and as Starr (1982) points out,
the cultural (and managerial) movement has been from one of health care
planning to one of health care marketing. The emphasis on providing high-
technology (and high profit) services to insured individuals has eroded the
indigent care mission statement. This erosion has lead to decreased access
primarily by the medically indigentto both primary and specialty care.
As further evidence of this transformation to a health care marketing
mentality, the University of Colorado Health Sciences Center is currently in

the mid-stages of an immense relocation from the campus at Ninth Avenue
in central Denver to a larger space that has been acquired from the former
Fitzsimmons Army Medical Center east of Denver in the suburb of Aurora.
The new space has given a corporal presence to the new ideology of
marketing and has caused a significant rift between the faculty of the
University of Colorado and the Chief Executive Officer and associates of the
University Hospital. Fearing a shift from research support to high-end
medical marketing, many of the University's principal medical researchers
who are in some cases also the heads of specialty divisionshave departed,
and with them has gone substantial research funding.
The medical providers in the IDGP are physicians and nurse
practitioners who are part of the School of Medicine faculty at the University
of Colorado Health Sciences Center, Division of Infectious Diseases. Of the
over thirteen providers, a core group of approximately six have full-time
clinical appointments and see clinic patients daily. The others combine
research responsibilities with part time clinical practice. There is a full staff of
nurses, social workers, medical assistants and patient care coordinators who,
along with the providers, are the primary individuals with whom patients

To an extent, the IDGP and its patients have been somewhat exempt
from the access-limiting political and fiscal ramifications described. This
exemption has been the result of two main factors. First, despite the fact that
infectious disease is a sub-specialty of internal medicine, the nature of caring
for those infected with HIV lends itself to a primary care approach to patient
management. Early in the history of the practice, the IDGP Medical Director
pointed to the value of a primary care philosophy, not only from a quality of
care perspective, but from a fiscal accountability standpoint as well. He
fostered a climate of caring for all of the patients' needsHIV-related or
otherwiseand of having IDGP attending physicians manage patients
during in-hospital stays which inevitably led to more efficient, less lengthy
inpatient stays.
This approach, coupled with the arrival of HAART regimens,
drastically reduced the enormous expense of in-patient hospital days that
had previously been associated with caring for HTV-infected patients across
the country. Data depicting these savings gained favor with the University
Hospital Administration whose increasing corporate ethos was more
enamored by marketing high-end medicine to insured patients than by
providing primary care to HIV-infected individuals. This administrative

favor at times translated into improved staffing and increasingly better
facilities for patients.
The second reason for the ability of the IDGP to function well within
the limitations of deteriorating access to care involved the yearly receipt of
funds from the Ryan White CARE Act that provides large-scale federal
funding each year to primary care clinics, mental health facilities and
community service organizations in large cities across the United States that
provide services to those with HIV-infection. The Ryan White Funding
insured that no patients with HIV-infection were turned away, not only from
HIV specialty care, but also from the primary medical care with which it is
In the IDGP, patients receive care through one of three major payer
mechanisms: the state-funded program known as the Colorado Indigent
Care Program (CICP), Medicare/Medicaid Plans and through private
insurance organizations. Funds from the Ryan White CARE Act provide a
portion of staff salary support as well as funding for the AIDS Drug
Assistance Program (ADAP) that funds access to most of the more than
twenty licensed antiretroviral agents to patients seen under the CICP

Sampling and Recruiting
The University Hospital IDGP setting provided a richly
heterogeneous sampling frame from which to draw participants. This clinic
treats approximately 1400 established patients with HTV-infection, of which
approximately sixty-five per cent meet the Centers for Disease Control
(CDC) criteria for a diagnosis of AIDS. Approximately 25 percent of the clinic
patients have documented addiction and psychiatric disorder co-morbidities.
The patient demographics include 85 percent males, 68 percent Caucasians,
14 percent Hispanics, and 11 percent African Americans. Approximately 55
percent of the patients identify as being gay and approximately 15 percent
have documented current or past injection drug use (IDU). These
demographics are similar to State of Colorado statistics with respect to
HIV/ AIDS sociodemographic prevalence.
Approximately 35 percent of patients are medically indigent and are
seen through the Colorado Indigent Care Program (CICP) or the state
Medicaid Program (Figure 4-1).


1 T 1 l if**" 1 '1 1

|=7I t=n
'I . ' 1 . ] ; 1 * i y
(0 o o Q.
Q < (0 Q; C0 S (0 o 3 5 a <0 O
*AIDS=CDC-specific AIDS diagnosis; ADD/Psych=psychiatric or addiction co-morbidities; Hisp=Hispanic; AA=African
American; MSM=men who have sex with men; lVDU=injection drug users; Indigent=patients without private insurance
In order to approach this highly diverse population for sampling in
this ethnographic design, a critical case strategy was employed. Critical cases
are those participants who, because of their experience, circumstances or
other pertinent attributes, may make dramatic contributions to the research.
These participants were considered to be key informants because they were
felt, because of their experience, to possess special knowledge and were

willing to share this knowledge (Gilchrist, 1992). Patton (2002) has pointed
out that the logic of using such a strategy in the qualitative setting lies in the
ability to select participants who are informant-rich and who will "illuminate"
the research questions. Hence, for this study such a sampling strategy was
employed to support the acquisition of the depth of data that has been
identified as lacking in the current scientific literature.
While the clinic-sampling frame was clearly rich in multiplicity, such
heterogeneity was also problematic for a qualitative investigation where the
sample size was to be based on acquisition of depth rather than breadth.
Choosing a sample that included representative from all of the clinic's
sociodemographic areas would have led to the in-depth exploration of
individual experiences and circumstances that differed so widely from each
other that the emergence of meaningful common themes would be unlikely
(Patton, 2002).
Thus, applying a sampling strategy to this setting required a carefully
planned and balanced approach in order to attain the degree of homogeneity
needed for the emergence of common themes, while at the same time
drawing upon components of the diversity of the sampling frame. In
addition to the question of which individuals to sample as key informants,

was the decision of how large the sample size should be to achieve
appropriate levels of validity as that term is understood in the qualitative
Identifying the appropriate sample size is somewhat of an ambiguous
consideration in qualitative research. Texts generally agree that depth is
more important than breadth, and that sampling is likely adequate when the
researcher reaches a point where there is a repetition in the responses known
as data saturation (Patton, 2002; Hammersley & Atkinson, 2002; Bernard,
2002). Moreover, Patton (2002) has suggested that in qualitative designs,
validity is more a function of the richness of the informants and the depth of
the data than of the quantity of the sample size.
Nevertheless, it was necessary to approximate a sampling size goal at
least for the initial data collection to begin. Trost (1986) had proposed a basis
on which one might estimate adequate qualitative sample sizes based on the
number of variables identified as being associated with a particular research
question (see previous discussion regarding the use and meaning of
variables in qualitative research). In this strategy, variables can be
dichotomized such as adherence versus non-adherence, or trichotomized as
in die case of low, medium and high socioeconomic levels. This approach

creates various categories that result in the sampling of at least one
participant for each category. While this strategy may be useful in
approaching a diverse sampling frame, the presence of numerous
variablesparticularly those that are trichotomizedcan lead to a potential
sample size that is quite unmanageable in a qualitative study. To decrease
this effect, Trost suggests restricting the number of variables included,
particularly those that are trichotomized.
After careful consideration of the variables identified in the review of
the literature with respect to certain factors thought to be associated with
non-adherence, and of the diversity present in the sampling frame, I elected
to use Trost's strategy as a guide to recruitment would be helpful in shaping
the sample in a way that would maximize the homogeneity/heterogeneity
balance I was hoping to achieve. I did not, however, require that the sample
of participants fit precisely into the categories generated through the Trost
technique nor did I insist that a specific, pre-determined number of
participants be reached. Rather, I mapped the potential variables based on
the review of the literature, applied them to Trost's matrix and used this
matrix as guide in my meetings with those who were designated to refer

I attempted to rigorously restrict the variables to be included in the
sample selection while retaining potentially important ones. In approaching
this strategy, it became clear that restricting the sample to certain sub-groups
within the sampling frame could eliminate several of the variables
previously identified. For example, sampling only medically indigent
patients (defined as those who were seen in the clinic under the CICP or
Medicaid Programs) removed the socioeconomic variable. Sampling only
gay men allowed for recruitment of a sub-population of particular research
interest to me, and effectively eliminated the gender and sexual orientation
factors. The variables addressing psychiatric and substance addiction co-
morbidities and race/ ethnicity were substantive to the research questions
and, therefore, were incorporated in the sampling plan to the extent possible.
Prior to beginning the research, I believed that it would be important
to include in the sample a balance of those who had a history of strict
adherence to HAART regimens and those who did not. However, at the
completion of the pilot interviews and as subsequent interviews progressed,
it became clear that categorizing participants in this way would not
necessarily be helpful (or even relevant) to the acquisition of meaningful
information and could actually create an unnecessary distraction from

accessing information-rich participants. Whether or not individuals were
adherent to their HAART regimens was less important than the information
they provided about the subject in general. Moreover, it was more important
that, because of the stigmatization and social and cultural
disenfranchisement, the sub-group of gay men possess both specific and
general insight into the complex issues of the research questions, yet
maintained the desired degree of homogeneity. The final sample guide
based on three variables discussed in the current adherence literature is
summarized in Table 4-1.
Table 4-1. Qualitative Sampling Strategy Based on Three Variables*
Psychiatric Diagnosis Y N
Substance Abuse Y N Y N
Race / Ethnicity** A C H A C H A C H A C H
* Adapted from Trost, J.E. (1986) Qualitative Sociology. 9(1)
**A=African American, C=Caucasian, H=Hispan
The application of Trost's strategy suggested a minimum sample size of
12 participants if one participant from each category was to be sampled. As
previously noted, the study utilized a critical case strategy in the sampling of

key informant participants with consideration of, though not strict adherence
to, the Trost matrix. Critical cases were considered those who were
identified by the clinic staff (physicians, nurse practitioners, nurses and
social workers) as being those who, because of their experiences and their
ability to relate these experiences, were likely to be capable of making
dramatic or particularly important contributions to the research questions.
Prior to the initiation of the research, a formal training was held with
clinic staff to review the nature of the research, the research questions and
the types of individuals who would be needed. Written materials associated
with these issues were provided as well as a schema describing how the flow
of referrals from the clinic staff to the researcher should be carried out. Basic
inclusion criteria for participation included that the referrals must be males
ages 18 to 65 who have had a relationship with the IDGP as an established
patient and who had self-identified as being gay. Obviously all participants
had to be HIV-antibody positive and had to have a HAART regimen
prescribed by their primary provider. In an attempt to minimize
socioeconomic variability, only those participants who were seen in the clinic
through the state CICP or Medicaid Programs were included.

Finally, in order to avoid the introduction of participant response bias,
since I, myself, was an IDGP clinic provider, it was explicitly stipulated that
referrals would not include any patients with whom I had a past or present
clinical relationship. Given the large number of patients in the clinic-
sampling frame, this limitation was not difficult to enforce. Table 4-2
summarizes the inclusion criteria for the study.
Table 4-2. Basic Inclusion Criteria
Males ages 18-65 who identify as being gay
HIV-Antibody positive
HAART Regimen prescribed by medical provider
Medically Indigent (CICP* or Medicaid payer source)
No clinical relationship with researcher
Willing and able to give informed consent
*Colorado Indigent Care Program
During the training meeting, I discussed in detail with the staff what
elements should constitute a critical case. For example, I explained that
participants should be those individuals who, in the opinion of the clinic
staff, could address the subjects in the research questions including the
issues of context, clinic interaction and sense of self in particularly dramatic

or important ways. This ability could be due to obvious experience or
particular ability to articulate experiences or circumstances. I stressed to the
staff that I was not looking for participants who had unusual or extreme
circumstances or perspectives, but those with typical experience who were
especially capable of expressing, relating or portraying that experience. The
participants, I suggested, should be the voice for the other patients in the
This training with the clinic staff was pleasant, supportive and light
spirited and I felt they were significantly invested in the research and that
they clearly understood the importance of the subject matter. I recognized
also that, having defined my terms to the extent I have described here, it
would be necessary to trust the experience and intuition of the clinic staff in
referring those participants who would represent critical cases and who
could, in fact, act as key informants.
Following the meeting, four from the clinical staff were identified as
being likely to make most of the critical case referrals. I continued to work
closely with these providers, reviewing and refining the critical case
sampling strategy. Indeed, twelve of the initial eighteen participants came

from these providers. I continually monitored the referrals received with the
Trost sampling matrix to gauge the developing heterogeneity of the sample.
When the clinic providers identified a potential participant, they
briefly discussed the nature of the research with that person. If the potential
participant was interested in further information, he was asked to sign a
Health Insurance Portability and Accountability (HIPAA) Form A. This
signed form gave me permission to contact the potential participant, and
arrange a time to meet. In all, 36 individuals were referred from six
providers, 75% coming from the core four providers. Those who agreed to
participate were paid twenty dollars for, each interview.
The providers offered the referrals, but were kept blinded to
knowledge of whether or not their referrals actually participated. This was
done to further assure the privacy of the actual participants and to minimize
any concern the participants may have had about negative repercussions
from their providers.
During the period of recruitment, a total of 36 referrals were
made. Of these, 5 indicated, after further discussion on initial contact, that
they would not be interested in participating. Four of the referrals did not
meet the inclusion criteria and were not considered. Nine did meet criteria