../......« {p a.
x 1!; .32
I...

 

5'..r'-.

i

u? .
n. .s|

.I ‘35‘

an)“.

a 3
ﬁx.“ a)

i a... wig... A ,
. 1. :ipwmﬂu: a... ..

i. La:
$1.“..uﬂllthﬂx
(93%;. a.
. a f‘
s. .. i
2.3.3....” mm
. 9!.

.5. :J .
”0.3%.

n: Jr.
car: v.
k. » .f 4

L. r

 

om...
g
3;.
1&8” 1.!
’
....
ya”...

I...“

; E14...»

.. .3... Run»:

4‘

POWER. ‘Wslibtﬂlt
$397.32. 1:33...

. (7.... x: . a

. guy. 4... mun...“

.pnur

:le I09:

, t Vl.‘

~ 39‘ tvl..ﬂ\ 5, ‘1 t

.2- .. . 1

-2- . I
shszs... 70.2
‘5, 1’:

1. n ..
l )n‘
IKI. .u..r.
\
{i

 

uni

\. I: .

 

.3. mun-lg...
.
3., pit}.
. fr .5
1:33.. Lu. .
i. 5.4. L)... :1 c.
,ﬂ...v..r;.:.a:. 5.8.
..r nwruvﬂvas an
'I'IJ!’ , . v. A .
. i .o . . é.
. v a .
New... :rhﬂumW..n.ﬂ§.m9NRnt _
jhv i
10.4.. I .0 III. a" 'II
41 I; 44- mount—H44

.A

This is to certify that the ;
dissertation entitled ’

THE ROLE OF CHRONIC ILLNESS, PERCEIVED CONTROL.
SOCIAL SUPPORT, AND HEALTHCARE DISCRIMINATION
IN MENTAL WELLBEING AND ALTERNATIVE
HEALTHCARE

presented by

KIMBERLY RENEE BARBER

has been accepted towards fulﬁllment
of the requirements for the

-—--.--a-c--e-A- ~ '~ -

PhD. degree in Sociology

 

// __7 ’7 7 ,7
x" '/ \ /
; rat/zgi’mVC/éz, M w

/;Wlajor‘Profes§or's SignatiE
/ 2/1/;“ / L? 5/
/ / I
Date

MSU is an Afﬁnnatr‘ve Action/Equal Opportunity Institution

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8/01 c:/C|RC/DateDue.p65-p.15

THE ROLE OF CHRONIC ILLNESS, PERCEIVED CONTROL, SOCIAL SUPPORT,
AND HEALTHCARE DISCRIMINATION IN
MENTAL WELLBEING AND ALTERNATIVE HEALTHCARE
By

Kimberly Renee Barber

A DISSERTATION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements
For the degree of

DOCTOR OF PHILOSOPHY
Department of Sociology

2004

ABSTRACT
THE ROLE OF CHRONIC ILLNESS, PERCEIVED CONTROL, SOCIAL SUPPORT,
AND HEALTHCARE DISCRIMINATION IN
MENTAL WELLBEING AND ALTERNATIVE HEALTHCARE
By

Kimberly Renee Barber

Background: Because it is a chronic process that includes the proliferation of multiple
stressors, a chronic illness can lead to signiﬁcant psychological distress. Within the
framework of the stress proliferation model, a chronic illness is a signiﬁcant predictor of
such distress. Coping with the distress through greater personal agency and utilization of
alternative therapies is proposed to mediate the effect on distress. Discrimination in
seeking healthcare, as a moderating factor, was hypothesized to add signiﬁcantly to the
prediction of the distress.

Methodology: The 1996 National Survey of Midlife Development in the United States
(MIDUS) was utilized for this secondary analysis. There were 4242 respondents, aged 25
— 74 years, surveyed by phone on physical health, midlife development, and social
responsibility. Of these, 3680 responded to a mail survey on aspects of health lifestyles
and psychological health. The 728 on current treatment for depression were excluded
leaving 2934 respondents for the current analysis. Abbreviated versions of a depression,
anxiety, and perceived control scales were applied. Chronic illness and alternative
therapy use were measured by index lists. Discrimination was measured by health
seeking events per lifetime. Multiple regression analysis was conducted on models for

each dependent variable. Family status, social support, personal agency, and

socioeconomic status were controlled in the analysis. The ﬁnal linear regression model
constructed estimated the signiﬁcant contribution of each independent variable to illness
status in predicting stress and psychological distress.

Results: Chronic illness was reported by 2095 (71%) respondents. Chronic illness was a
signiﬁcant predictor of depression (p< .05). The effect was mediated by perceived control
(p< .01 ). Discrimination measured by perceived interference in life was signiﬁcant (p<
.05). Chronic illness was a signiﬁcant predictor of alternative therapy use (p< .01).
Female gender and high educational attainment were signiﬁcant independent predictors
ofuse (p< .05 & p< .01).

Conclusions: The current analysis supports the premise that chronic illness is a stressor
that is associated with the emergence of other stressors, such as distress from
discrimination. Chronic illness is also a strong predictor of health seeking behavior and
may reﬂect greater unmet healthcare need among those experiencing ongoing, long-terrn

SITCSSCS.

ACKNOWLEDGMENTS

This dissertation, as well as a clear understanding of sociologic concepts, would not have
been accomplished without the time and effort put forth by Professor Cliﬁ‘ord Broman.
His insight and assistance have been invaluable throughout the entire doctoral process
and I thank him dearly for his support, guidance, and patience. His combination of
academic intellect, personal integrity, and quiet leadership was an inspiration to me. I am
deeply grateful for his mentoring, as well as, his friendship. I would also like to
acknowledge the dedication and extra effort put forth by my committee members; Harry
Perlstadt, Stan Kaplowitz, Steve Gold, and Margaret Holmes-Rovner. Their guidance has
contributed greatly to my growth as an academic professor and research investigator and

I will always be grateful to them for it.

iv

TABLE OF CONTENTS

CHAPTERIBACKGROUND
TheStressProcess'I‘heory......
The StressProliferationModel.........................................
Chromcrllness
Discrimination
AlternativeMedicine..........
Dissertationlntent......................................................
DatasetDescription................................................
Snengths&Limitationsofl\41DUS..........................
DescriptionofMeasures....... .
DependentVarrables
IndependentVariables...................................................
PlanofAnalysis.....

CHAPTER 3. RESULTS...
Univariate Analysis... ..
Mental Health Outcome Analysis
Health Behavior Outcome Analysis

viii

1X

11

13

.21

..23

.27

27

.32

...36

.36

39

.44

47

......48
.....51
....52
ChroniclllnessSubgroupAnalysis.............

66

CHAPTER4:DISCUSSION..............
Limitations..................
Conclusions............

APPENDICES...........

Appendix A: Table i'é'iiis'ikii'viiiiiiiéélf.’ Z fifflf.’ III if
Appendix B: Scales and Indexes...
Appendix C: CorrelationMatrix Tabels...

vi

.75
.81

83

...88
.88

93
.97

100

LIST OF TABLES

Table l —- MIDUS Sample Disposition Prior to Refusal Conversion... ..

Table 2- Comparison of MIDUS Sample and National Population
CharacteristicProportions...

Table 3 - Comparison of Respondents Returning and Not Returning

the MailedQuestionnaire.............
Table4-DistributionofDepressionResponses........
Table5-DistributionofAlternativeTherapiesUsed......................
Table6-DistributionofPerceivedControl......
Table7—DistributiononuaIityofSupport...

Table 8—Distribution ofDiscrimination...... ..

Table 9- Prevalence of Reported Chronic Conditions In Final
Sample(n=2934)... .

Table 10 — Comparison of Demographic Characteristics Among

Those withand Those without Chronic Illness..................... ..

Table 11— Univariate Analysis of Proportions with the Outcome by
Illness Group Status...

Table 12 — Univariate Analysis of mean Values for Each independent

Variable by Illness Group Status... ..
Table 13 - Regression of Chronic Illness and Other Predictors on Outcomes...
Table 14 —Regression Analysis ofPredictors on Depression...
Table 15 — Depression Rates According to Control Variable Subgroups... ..
Table 16 — Depression Rates According to Independent Variable Subgroups...
Table 17 -- Depression Rates among Illness and Discrimination Subgroups...

Table l8—Regression Analysis for Predictors on Anxiety... ..

vii

.28

...30

.32

.37

..39

.42

.43

...49

.50

...51

.52

.53

.54

.55

.58

60

....62

Table 19 — Regression Analysis for Predictors on Alternative Therapy Use ............. 63

Table 20 — Alternative Therapy Rates of Use for Illness and
Discrimination Status ............................................................................. 64

Table 21 — Regression Analysis of Outcomes According to Selected
Serious Chronic Conditions ...................................................................... 67

Table 22 — Regression Analysis of Outcomes According to Perceived

Physical Health ..................................................................................... 67
Table 23 — Regression Analysis for Depression on Predictors According

to Perceived Physical Health ..................................................................... 69
Table 24 — Regression Analysis for Anxiety on Predictors According to

Perceived Physical Health ........................................................................ 70
Table 25 — Regression Analysis for CAM on Predictors According to

Perceived Physical Health ........................................................................ 71
Table 26 — Zero-order Correlations for Mental Health Outcomes .......................... 72

Table 27 — Zero-order Correlations for Complimentary Alternative Medicine Use ...... 73
Table 28 - List of Variables ....................................................................... 88
Table 29 — Pearson’s Correlation Matrix for Depression .................................... 99

Table 30 — Pearson’s Correlation Matrix for Complimentary Alternative Medicine. . ...100

Table 31 - Regression Analysis of Predictors on Depression including
Interaction Terms ................................................................................. 101

Table 32 - Regression Analysis for Predictors on Anxiety including
Interaction Terms ................................................................................. 102

Table 33 - Regression Analyses for Predictors on CAM including
Interaction Terms ................................................................................. 103

viii

LIST OF FIGURES

Figure1—TheStressProcessModel............................. . .. . .. ...
FigureZ—Primary and Secondary Stressors: PathwaystoDepression.......
Figure3—RespondentSampleFlowchart...............................................
Figure4—InteractionBetween IllnessandAge..........
Figure 5 —- Interaction Between Perceived Control and Support... ..

Figure 6 — Interaction Between Chronic Illness and Discrimination... ..

Figure 7— Alternative Therapy Use According to Illness Status and
Demographics... . .

Figure 8-— Alternative Therapy Use According to Illness Status and
Discrimination...

ix

..48

57

..59

.61

....64

...66

INTRODUCTION

This proposal examined contemporary issues in social-psychological distress
related to chronic illness. Chronic illness (both disease and disability) is prevalent in
American families today with incidence rates continuing to rise for many conditions.
Some chronic illnesses (e.g., diabetes and asthma) have become signiﬁcant public health
concerns (Mannino et al 2002; Eisenmann 2003; Mascie-Taylor and Karim 2003). Such
ongoing, long term conditions must be managed within the context of social status while
negotiating a healthcare system that is geared toward acute illness and accidents. Those
with chronic illness must not only cope with this larger context but also must cope with
both their role in society (i.e., spouse, parent, employee) and that of long term patient.
The latter role may emerge unexpectedly or gradually but none-the-less become a new
long-term commitment, and thus deserves the label of what Leonard Pearlin refers to as
an ‘unexpected career’ (1994). Various types of role-related strains emerge for a person
when managing a chronic illness that can be conceptualized as chronic stressors.
Furthermore, the relationship of these multiple stressors ﬁt within the framework of the
stress proliferation model as demonstrated from multiple secondary stressors initiated by
the primary stressor of a chronic illness.

This analysis sought to show that chronic illness contributes a unique strain to
persons within the context of other psychosocial factors as conceptualized by the model
of the Stress Proliferation Process. The key dependent variables for this analysis were
depression, anxiety and utilization of alternative therapies. The main independent

variables included the presence of a chronic illness, discrimination in seeking medical

care, and perceived control. Key variables to be controlled in the analysis were socio-
economic status, marital status, quality of spousal support, and prior use of health care
services.

Distress, as measured by depression and anxiety indexes, is the major dependent
variable for this analysis. Chronic stressors have the power to disrupt activities,
relationships, and life and once established may have an adverse impact on mental health
and well-being (Pearlin, et al. 1990). Long-term illness presents a person with role
strains, relationship conﬂicts, and violations of normative life-course expectations as they
deal with each serious exacerbation of symptoms and/or the long-term morbidity of the
condition. The persistent nature of these chronic stressors places the ill person at
considerable risk of such negative mental health outcomes as increased distress, anxiety,
and depression.

The main independent variable for this analysis was the presence of a chronic
illness. Unlike urgent accidents and acute illnesses, a chronic illness is ongoing and long
term, often emerging insidiously and persisting through growth phases. The expansion of
emerging, persistent stressors over time has the potential to give rise to an accumulation
of stressors and thus a proliferation of stress. Chronically ill persons can ﬁnd themselves
directing more and more of their energies toward illness-oriented needs rather than health
maintenance. Illness demands are added followed by additional concerns, worries, and
feelings of guilt.

A second key independent variable in this analysis was that of discrimination-
both the experience of discrimination in general and in seeking medical care speciﬁcally.

The experience of discrimination is directly related to poor mental health outcomes

because of the stressful environment that is created (Harrell 2000; King 1996). In regards
to the chronically ill'person, the experience of discrimination may be a mediating factor
which increases the likelihood of chronic illness or of severe disease from environmental
hazards and barriers to quality medical care (Weinick et al 2000). However, for persons
seeking medical care for a chronic illness, the barriers presented by discrimination during
the illness trajectory create additional strains and frustrations that may signiﬁcantly
multiply the impact on mental strain. The effect of discrimination when seeking medical
care was expected to moderate the association between managing a chronic illness and
distress. For some acute illnesses, the inability to obtain desired care can be ignored when
faced with barriers due to discrimination, although it remains a stressful experience. Yet
for chronically ill persons, the inability to achieve desired health outcomes may be
ongoing as well, as encounters with the healthcare system multiply.

Perceived control (i.e., having personal resources such as high self esteem and a
sense of control in life) was a key independent variable that may mediate the effect of
chronic illness on distress. 111 persons with the personal resources to overcome barriers in
access to care, to persist in challenging difﬁculties presented by others, and to seek out
and utilize alternatives in order to manage their illness over time are likely to experience
less distress than those without such personal resources. Perceived control was expected
to mediate the effect of chronic illness on distress. Persons with high perceived control
were expected to experience less daily stress and to utilize more alternative methods in
managing their illness and therefore be at lower risk of mental distress and/or depression.

Because socioeconomic status (SES) is associated with all three independent

variables and with the dependent variable, it was controlled for in the statistical analysis.

For the same reasons, marital status, the quality of support, presence of children, and
prior use of health care services was analytically controlled

We proposed that the presence of chronic illness is a signiﬁcant predictor of
social-psychological distress, discrimination signiﬁcantly increases the effect, and that
personal agency mediates the effect beyond that of socio-economic status and
demographic characteristic factors. The tendency within the initial domain of chronic
illness for new stressors, such as discrimination when seeking care, to present themselves
and for preexisting stressors to be exacerbated by barriers to care, demonstrates the Stress
Proliferation concept. This concept was tested to show whether a chronic illness is indeed
a signiﬁcant predictor of distress and if discrimination increases the effect signiﬁcantly,

and whether personal agency tempers the effect by decreasing the risk.

CHAPTER ONE

BACKGROUND

The Stress Process Theory

Early stress models quantiﬁed stressors as signiﬁcant life experiences that are
relatively time-limited and occur as events across various domains of life experience
(Holmes and Rahe I967). The occurrence of a major event (such as the death of a spouse)
or of multiple events (such as divorce and being ﬁred from a job) signaled an increased
risk of stress-related illness. The development of the Holmes-Rahe Social Readjustment
Scale meant that such risks could be quantiﬁed according to critical values that signify
varying levels of risk (Holmes and Rahe 1967). Established deﬁnitions of stress as a
process emphasize the importance of viewing stress as a dynamic process rather than a
single, acute event (Pearlin 1989;Lester et al., 1994; Wheaton 1996). Stress may originate
from multiple, ongoing sources that are not as immediately recognizable as an acute, life
event yet nonetheless can have a profound negative effect on health. These chronic, more
continuous sources of stress (called stressors) are more inherently insidious in onset and
ongoing than discrete life events, but may result in high levels of stress or lead to
maladaptive behaviors (Wheaton 1994). Blair Wheaton has extended the traditional stress
model to more fully represent the concept of the stress process as a continuum. He
contends that stress can result as much from too little change as from too much sudden
change, as when an individual is exposed to continuously difficult or demanding
environments that do not change. Wheaton conceptualizes these stressors as 1)

developing as continuing problematic conditions in our social environment and roles, and

2) typically having a longer time course than life events from onset to resolution. Chronic
stressors are either deﬁned by the nature of daily role enactments or are so regular in the
enactment of daily roles that they behave as if they are continuous for the individual.

Brown and Harris (1978) refer to chronic stressors as life difﬁculties. They
distinguish chronic stressors as difﬁculties in ongoing adversities that people face. They
provide such examples as a spouse’s drinking problem or a partner’s chronic health
problem. Brown and Harris contend that such ongoing difficulties have importance
independent of life events in predicting negative health outcomes such as depression.

In reﬂecting on stress process concepts, Leonard Pearlin (1999) notes that the
stress process occurs within a broader social context. A key assumption of the model is
that diverse factors that are interrelated converge on people’s well-being. Among these
converging factors are the social statuses of individuals, the contexts that envelope their
daily lives, their exposure to stressors, the resources upon which they are able to call
upon, and the stress manifested in their psychological and physical functioning. A second
assumption of the stress process perspective is that people engaged in the ordinary
pursuits of life can be exposed to multiple and highly complex stressors depending upon
the context within which they live. A person’s standing in the social order, economic
class, gender, race, and ethnicity have the potential to pervade the structure of their daily
lives and the experiences that ﬂow from it. Thus, status placement of people can create a
universe of stressors that set the stage for life difﬁculties, strains, and chronic stressors
(Figure 1).

An implication of this “web of interconnections” among multiple factors is that a

change in one can effect changes in the others and set in motion chains of effect (Pearlin

L1 1999). Stressful events, whether life events or chronic stressors, rarely occur in

isolation — rather events may initiate, precipitate and then inﬂuence each other in a

process that unfolds, accumulates, and feeds back on previous stressors (as seen in Figure

l).

 

 

Social and Economic Statuses

 

 

 

 

 

Neighborhood Ambient
Stressors

 

 

 

 

 

Primary
Stressors

 

 

 

 

 

 

 

 

 

 

 

 

Secondary . Mental health
T Stressors T. outw‘ms
Moderating Resources

 

 

FIGURE 1: The Stress Process Model (Pearlin, L1 1999).

That the structure of an experience in one setting (i.e., social isolation or social

aspiration disconnect) can structure action in another setting (i.e., suicide or anomie) has

been noted empirically throughout history (Thomas, WI 1928, Durkheim E 1951,

Merton, RK 1968, Mechanic 1978, Pearlin L1 1989). Durkheim (1951) showed suicide

not as a rash act of deviant behavior but as a consequence of people’s attachments to

others. Thus, the values attached to social status and the conditions surrounding them

contribute directly to well-being and stressful outcomes (Pearlin, L1 1999). For example,
not only does divorce have implications for an individual’s role as spouse, but it also has
social implications through means of guilt, shame, and status decline.

Within this web of interconnectedness various types of role strains emerge as
chronic stressors (Pearlin, 1983). They include being in a role whose demands exceed
capacities, being a captive of the role, being in conﬂict with others in the role set and
trying to reconcile the simultaneous demands of dual roles. The issue of role strain from
an unexpected or emergent role is speciﬁcally pertinent to the topic of chronic illness.
Pearlin refers to these as care-giving roles that become long-term commitments and thus
deserving of the label of “unexpected careers” (Pearlin and Aneshensel, 1994). Emergent
illness roles surface within a context of pre-existing multiple roles and the
accommodation of the new role demands restructuring of existing roles. Such
restructuring may bring, depending on available resources, restriction of activities,
conﬂicts with others, and irreconcilable strains from competing demands (Pearlin 1999;
Pearlin, et a1. 1990). Activities directly associated with long-term care giving can also
become chronic stressors and can themselves lead to stressors in other domains of life.

This multiple loading of stress is referred to as stress proliferation (Pearlin, et a1. 1997).

The Stress Proliferation Model

The Stress Process model provides a conceptual framework that accounts for
multiple, ongoing sources of chronic stress. However, sources of stress are not always
independent. Not only do stressors occur within a complex process, but any particular
stressor within the web can lead sequentially to other stressors (as is demonstrated in the

mid section of the model in Figure l). Pearlin (1989) developed a conceptual framework

referred to as the stress proliferation model. Stress proliferation occurs when an initial
primary event causes stress that leads to the emergence of other secondary stressors.
Stress proliferation is critically important to the stress process model due to the creation

of secondary stressors. Rarely are people exposed to only one severe stressor (see Figure

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2).
PRIMARY STRESSORS
SECONDARY STRESSORS
OUTCOMES
ILLNESS
DEMANDS
IDENTITY
STRAIN
ROLE
CONFLICTS DEPRESSION
+ , RESTRICTION IN
1 SOCIAL/LEISURE
ACTIVITIES
ROLE
H CAPTIVITY

 

 

 

 

FIGURE 2: Primary and Secondary Stressors: Pathways to Depression
(Pearlin et a1. 1997).

An initial stressful event, such as divorce, oﬁen leads to additional event-related
stressors, such as ﬁnancial losses and single parenting (Pearlin, 1999). The initial stressor
is referred to as primary and those that follow as secondary (Pearlin, 1989). The terms do
not imply importance, but are intended to distinguish stressors according to the temporal

order in which they can be observed. Underlying the conceptual model is the presumption

that stressors appear sequentially, rather than simultaneously, as the stress process
unfolds (Pearlin 1999)

The dynamic aspect of the stress proliferation process underscores the multiple
number and types of stressors to which people can be exposed. The process is described
by Pearlin as a changing conﬁguration of stressors in which people become enmeshed
(Pearlin, et al., 1981). The various stressors surface and recede, as well as combining
with each other over time. For example, a person may experience stress from a spouse’s
illness as an initial acute and major event which leads to other ongoing stressors that can
precipitate additional acute and major events, particularly when the illness is severe. The
stress proliferation model is suggested to explain some of the differences among
individuals in outcomes when they experience similar stressors. Beyond the effect of
resources to moderate negative outcomes in mental health, differences in outcomes
among those exposed to similar stressors remain (Pearlin, 1999). Secondary stressors
then help to explain the differences in this relationship. Individuals in similar
circumstances may be very different in the presence and effect of secondary stressors.

According to Pearlin (1999), the major mechanism giving rise to stress
proliferation in the form of secondary stressors resides in the basic feature of social and
economic life. Speciﬁcally, he points out that people are the incumbents of multiple roles
located both within formal institutions, such as the family, and informal domains, such as
friendships. There exists a structural separation in people’s multiple roles which helps
them shiﬁ ground as they move between their roles and interact with others in the role
set. However, the structure of the experience in one setting comes to structure action in

another setting. Secondary stressors then arise from the consequent setting where the

10

obligations of one role make the enactment of another role difﬁcult. A pertinent example
in the literature is observed among women in the workplace who ﬁnd it difﬁcult to satisfy
the requirements of their jobs and the demands associated with the care of young children
(Lennon and Rosenﬁeld, 1992; Menaghan, 1991). In this case, secondary stressors
surface when the obligations of the parental role make the obligations of the occupational
role difficult to meet. It holds true even more so when the illness role is added to the
parental/worker role. The “priority claims of one role, then, can render the enactment of
another role problematic” (Pearlin, 1999).

However, as Aneshensel cautions, the appearance and intensity of secondary
stressors may also be inﬂuenced by one’s standing in systems of inequality (1999). Thus,
the conditions contributing to the creation or exacerbation of a primary stressor (low SES,
minority status, or comorbidity) may also contribute to the emergence and intensity of
secondary stressors (i.e., severity of illness).

Chronic Illness

The conceptual framework developed by Pearlin (1989) referred to as stress
proliferation was employed here to explain the effect of chronic illness and its impact on
mental health The model is particularly pertinent because of the ongoing, chronic, and
cumulative effect that long-term illness can have on the psychological well-being of a
person. Stress proliferation occurs when an initial primary event, a diagnosis of a chronic
illness for example, causes stress that leads to the emergence of other secondary stressors,
such as role strain. The illness does not occur in isolation, it initiates and then inﬂuences
other events and stressors and, along with the social environment, conditions action that

is taken.

11

The effect of an ongoing illness on mental health can be considered a type of
chronic stressor. The illness role can itself cause strain on the individual’s psychological
health. Various types of role strains emerge that have been identiﬁed by Pearlin (1983) as
chronic stressors. They include being in a role whose demands exceed personal
capacities, being a captive to an illness role, roles conﬂicting with other schedules , and
trying to reconcile the simultaneous demands of the dual roles of such as maintaining the
roles of spouse, employee, and health advocate.

The issue of an unexpected or emergent role is speciﬁcally pertinent to the topic
of chronic illness. Illnesses are rarely known in advance and therefore by deﬁnition not
planned for. Even for conditions that have a gradual onset, few people are prepared for
the changes, the demands, and the novel responsibilities that are created by the illness.
Various degrees of restructuring, depending on available resources, are necessary to
accommodate an illness that is ongoing. 111 persons become restricted in their usual
activities. Conﬂicts with other activities occur as when schedules need to accommodate
emergent and urgent medical visits. Competing demands strain mental resources as when
work obligations and personal duties collide. Unlike acute illness in which demands,
conﬂicts, and restrictions have a ﬁnite time frame, chronic illness can be endless and
therefore stretch the capacity of parents to cope without negative psychological
outcomes.

For these reasons and more, the presence of a chronic illness involves multiple
stressors that unfold from the initial diagnosis (primary stressor) to illness management
and coping (successive and secondary stressors). Renegotiating multiple roles (e.g., as

parent, spouse, and/or employee), is the major mechanism through which secondary

12

stressors emerge from the primary stressor. For example, a person who works fulltime
and deals with an ongoing illness must reschedule time off work to accommodate the
multiple medical visits or must cope with the mental strain of missing work to do so.

In addition, the social status of an ill person inﬂuences the likelihood of presence
of the primary stressor (e.g., developing severe disease) and the consequent secondary
stressors (problems obtaining care to manage the disease). Other roles such as lone
parenting is associated, through lower socioeconomic status, with greater severity of
illness and is associated with a lack of resources to manage the illness, as well as, to
manage the distress (Benzeval 1998, Montgomery, et al 1996). Once established, these
consequent problems become added independent, albeit related, cumulative stressors in
the life of the person with a chronic illness.

The stress proliferation model illustrates how chronic stressors can be additive in
their effect on mental health outcomes. However, some stressors in the chronic stress
process can modify the effect by exacerbating an existing problem with the appearance of
a stressful exposure that modiﬁes the impact of life stresses on psychological adjustment.
Discrimination, through unfair treatment and feelings of exclusion, can condition the
exposure, experience, and/or response of chronic stress for a Moular group and thereby
place them at greater risk of poor mental health outcomes.

Discrimination

A recent comprehensive literature review of representative ﬁndings on racial and
ethnic disparities in medical diagnosis and treatment describes the current state of health
status among minorities in historical terms and highlights the ethical implications of

disparities in health (Geiger 2003). Geiger points out that at no time in the history of the

13

United States has the health status of minority populations equaled, or even
approximated, that of white Americans (2001). Although the health of all Americans has
improved dramatically in the last several decades, excess morbidity and decreased life
expectancy for minorities has stubbornly persisted. Most recently, a signiﬁcant disparity
in life expectancy at birth has been shown among black American populations even for a
chronic disease with inheremly low rates of mortality (Barber and Johnson 2004). Geiger
emphasizes this persistence of inequality in his review by illustrating the ﬁndings of
Williams and Rucker (2000) that in 1995, the overall African-American mortality rate
was 60 percent higher than that of whites — “precisely what it had been in 1950”.

Research continues to show poorer outcomes for racial minorities compared to
non minorities across various diseases including coronary artery disease (Hemingway et
al., 2001), diabetes (Rucker-Whitaker et al., 2003; McBean et al., 2003), hypertension
(Jackson et al., 1996), cancer (Elston-Lafata et al., 2001; Dominitz et al., 2002;
McMahon et al., 1999), asthma (Bosco, Gerstrnan, and Tomita 1993), and mental distress
(Pak et al., 1991) and depression (Comas-Diaz and Greene 1994). The disparity has been
observed in both adult and pediatric populations. In the case of asthma, a study of black
and white Medicaid insured children in Detroit found that black children were much
more likely than white children to receive inadequate therapy (i.e., obsolete medications
that do not follow current guidelines) and were less likely to receive steroids or
adrenergic inhalers (Bosco, Gerstrnan, and Tomita 1993; Joseph et al., 1998), despite
higher rates of health care visits and higher rates of hospitalization.

Emperical studies have variously attributed racial disparities in health to such

causes as socioeconomic status, environmental hazards, inferior housing, poor nutrition,

14

lifestyle and behavioral choices, and cultural beliefs. However, even when these factors
are held constant disparities in health persist. Two other explanations have come to the
forefront of health disparity research, healthcare access and discrimination as an
independent effect.

Lack of minority access to health care has been shown to occur through multiple
mechanisms from residential segregation and geographic restrictions on quality
healthcare to rationing of medical services through the availability of comprehensive
health insurance (Blendon et all., 1989; Weinick, Zuvekas, and Cohen, 2000). Current
research has shown that racial minorities receive different treatments than non minorities
for similar conditions, such as heart disease, even after controlling for access or barriers
to care related to insurance, socioeconomic status, location, and regular physician
utilization (Conrad et al., 2003; Suresh et al., 2002). The existence of such a situation can
itself be considered a form of discrimination. Thus, racial or ethnic discrimination may be
an important contributor to health disparities, not merely through the historic and
persistent disadvantages it creates for minorities in the social structure but also
speciﬁcally through individual and institutional level health provider bias (Geiger 2003).
Geiger (2003) emphasizes that , whatever the causes, the experience of minorities within
the health care system differs from that of comparable whites across a broad range of
disease categories. The problem of racial and ethnic disparities in diagnosis and treatment
particularly, had become up until most recently, an important subset of the issue of
achieving equity in health status (Department of Health and Human Services, 2000).

Large-scale studies have examined racial differences in the adequacy, intensity

and quality of diagnostic and therapeutic procedures. One of the largest reviewed more

15

than 1.7 million hospital discharge charts to examine use of procedures in 77 disease
categories among more than 500 acute care hospitals (Harris, Andrews, and Elixhauser
1997). Black Americans were signiﬁcantly less likely than whites to receive therapeutic
interventions for half of the 77 disease categories even after controlling for patient age,
severity of disease, health insurance coverage, and hospital type. A study comparing
experiences of Hispanic with non-Hispanic patients in California, Florida, and New York
found that Hispanics were less likely to undergo major surgical procedures in 38 percent
of 63 disease categories (Andrews and Elixhauser 2000). Similar disparities have been
observed in more common and simple procedures such as laboratory testing and drug
therapies. Ayanian et al., (1999) examined in-hospital patients with congestive heart
failure and pneumonia. After adjustment for socioeconomic status, health coverage, and
type of hospital, black Medicare patients were signiﬁcantly less likely than whites to
receive adequate laboratory and diagnostic tests and/or drugs such as diuretics and
antibiotics.

General and orthopedic studies present similar ﬁndings consistent with the
disparities in diagnosis and therapy. Blacks hospitalized in Maryland had lower rates of
surgeries and the incidence among blacks was particularly low for more elective surgeries
(Gittelsohn, Halpem, and Sanchez 1991). Even more disconcerting is that the ﬁndings of
a large retrospective cohort study of amputation and leg-sparing surgery for peripheral
vascular disease among Medicare beneﬁciaries. This study comparing African Americans
with white Americans with and without diabetes, found that African Americans were
signiﬁcantly less likely to receive lower-extremity arterial revascularization (Guadagnoli

et al., 1995). African Americans are 32 percent less likely than whites to receive

16

laparoscopic surgery (Arozullah et al., 1999), to receive total hip replacement (Wilson,
May,and Kelly 1994) or total knee replacement (Baron et al., 1996). Even in intensive
care units (ICU) where critical care is provided round the clock, disparities exist between
racial and ethnic patient populations. In a national sample if ICUs in the United States,
African American patients received signiﬁcantly less monitoring, fewer tests, and less
life-support treatments than whites in the ﬁrst 24 hours, even after adjusting for age, type
and severity of disease, and hospital characteristics (Williams et al., 1995).

Findings that likely reﬂect inadequate primary and preventive care was
highlighted by a study in which Aﬁican American Medicare beneﬁciaries were more
likely to undergo bilateral orchiectomy for prostate cancer (late stage surgical
intervention) and more likely to undergo lower limb amputation (late stage surgery)
(Gomick, Eggers, and Reilly 1996). Using claims data for Medicare beneﬁciatires in 10
states and the District of Columbia, a subset was studied that matched beneﬁciaries on
the basis of zipcode to neutralize the effects of black-white differences in provider access
and regional practice patterns. Despite adequate health coverage, black patients’
utilization was substantially weighted toward lower-cost procedures (Lee et al., 1997).
The authors concluded that providers appeared to be giving “less intensive care to
otherwise similar black Medicare beneﬁciaries”.

That these disparities can in many cases be explained by discrimination and bias
in diagnosis and treatment, and that such differences do contribute to the excess burdens
of morbidity, disability, impaired quality of life, and premature mortality may be an
indication of problems of race and ethnicity discrimination in the larger society (Geiger

2001 ). The role of individual or institutional bias in creating racial and ethnic disparities

17

in care is difficult to quantify empirically. However, inﬂuences on decision making by
clinicians has been identiﬁed since 1957 (Geiger, HJ). Recent studies have linked
provider perceptions at every medical level, from medical student to physician resident to
experience practitioner, to decisions as varied as judgments of patients’ quality of life
(Rathore et al., 2000), physician-patient commtmication (Waitzkin 1985; Cooper-Patrick
et al., 1999), and the management of pain (Weisse et al., 2001). Geiger points out that
neither the health care system as a whole nor individual providers are fully insulated from
attitudes toward race, ethnicity, and social class that are prevalent in the larger society
(2001)

Racial stereotyping has been shown to contribute to health and wellbeing
disparities. African Americans are more likely than whites to be diagnosed as psychotic
but less likely to receive antipsychotic medications (Abreu 1999). They are more likely to
be hospitalized involuntarily, to be regarded as potentially violent, and to be placed in
restraints in both the inpatient and the outpatient settings (Benson 1983; Rosenﬁeld 1984;
Sleath, Svarstad, and Roter 1998; Kales et al., 2000; DelBellow et al., 2001).
Furthermore, provider and institutional bias have been shown to be signiﬁcant
contributors of disparities in health care. A number of empirical studies have supported
this conclusion in which provider views are assessed or decision-making of physicians
are blinded to patient race and ethnicity. Professional responses to white and non-white
patient subjects enrolled in experimental, controlled studies have been found to differ
signiﬁcantly in diagnosis, prognosis and therapeutic recommendations (Geiger 2003).

Unequal treatment as a result of racial/ethnicity or gender stereotyping, racial

bias, or discrimination has been repeatedly demonstrated in empirical research. Despite

18

the capacity of medicine to diagnose, treat and cure disease, minorities have not beneﬁted
fully or equitably from advances in medical science. A chronic lack of beneﬁt due to
discrimination can itselfbe a source of stress. The lesser likelihood of receiving
medication for pain (Todd, Samaroo, and Hoffman 1993) or of receiving adequate
analgesic for cancer treatment (Cleeland et al., 1997) in situations that normally require
pain amelioration adds to the burden of stress from the disease itself. A greater incidence
of failed patient-provider communication and lesser satisfaction with care received
reported by minorities suggests greater stress in relation to the disparate care than for
their white counterparts. Discrimination in the pursuit and acquisition of health and
medical care has been conceptualized as one of several types of racism-related stresses
(Harrell, SP. 2000). Harrell (2000) puts forth at least six types of stressors related to
racism of which daily hassles (racism microstressors) and chronic strain (chronic-
contextual) are two types that are illustrative of unfair treatment related to health care

access.

Unfair treatment stemming from discrimination thus is an important class of
chronic stressors (Kessler, Mickelson, & Williams 1999). It has long been established
that exposure to discriminatory behavior is an important feature of life for socially
disadvantaged groups, including racial minorities (Essed, 1991; Thompson, 1996),
women (Gardner, 1995 ; Krieger, 1990), and the poor (Turner & Lloyd, 1995). Empirical
research has been less consistent in showing that differential exposure is a cause of
mental distress, in part due to confusions in measurement with differential vulnerability
(Kessler et al., 1999). It has previously been suggested that differential exposure to stress

plays an important role in explaining the higher prevalence of psychological distress

l9

among lower status persons (Thoits, 1983). Yet, research has been slow to incorporate
measures of discrimination along with major life events which has led to incorrect
conclusions that minorities, women, and others exposed to high levels of discrimination
are failing to cope when in reality they are being exposed to more secondary stresses
(Kessler, Mickelson, and Williams 1999).

Perceived discrimination is one of the most important secondary stressors associated
with major stressor events (Williams, et a1. 1997). It has been signiﬁcantly associated
with job loss and exposure to violence (Thompson, 1996). A positive association has
been observed in an analysis evaluating the effects of discrimination, utilizing a unique
inventory measure of perceived discrimination, on psychological outcomes (Kessler,
Mickelson, and Williams 1999). Investigators utilized the MIDUS national survey (Brim
et al., 1996) which assessed perceived discrimination of any type rather than only on the
basis of gender, race, or social class. Perceived discrimination and lifetime exposure to
discrimination were measured, as well as chronic daily discrimination. Authors
constructed a continuous scale of frequency of perceived daily discrimination. Lifetime
perceived discrimination signiﬁcantly predicted nonspeciﬁc distress (p< .001) and major
depression (p < .001). Frequency of daily discrimination was found to be statistically
signiﬁcant in predicting major depression and generalized anxiety disorder (p < .05). No
evidence was found of a cumulative effect in predicting outcomes. In other words, the
presence of any perceived discrimination was just as signiﬁcant of a stressor in predicting
depression or anxiety as was the report of multiple types of perceived discrimination.

Discrimination then, the experience of and perceived, may be one form of stress

proliferation whereby the expansion of a primary stressor, such as illness, is reﬂected in

20

the addition of secondary stressors such as discrimination in receipt of medical services.
Discrimination proliferates and compounds the experience of stress directly through
unfair treatment and indirectly through the failure to receive a needed or wanted
treatment. In both, discrimination ﬁts the chronic stressor deﬁned by Brown and Harris
(Brown & Harris, 197 8) and conceptualized by Wheaton (1999) as life difﬁculties that
develop as continuing problematic conditions in our social environment and roles with a
longer time course than that for life events.

From this ﬁ'amework of differential exposure, discrimination may modify the
effect of a primary stressor through increased exposure to secondary stressors and/or the
proliferation of increased severity of the stressor compared to those who are not
discriminated against. Within the stress process concept, discrimination adds to the stress
continuum by accounting for additional, multiple, and ongoing sources of chronic stress.
Alternative Medicine

Complimentary and alternative medicine (CAM) has become increasingly
common in the United States. It encompasses a wide range of treatment options including
herbal supplements and vitamins, acupressure and acupuncture, meditation, and prayer.
Up to one-third of the American adult population reported in 1993 use of CAM to treat a
problem in the previous year (Eisenberg et al. 1993). The number of visits to CAM
providers in 1998, an estimated 425 million, has surpassed the number of visits made to
primary care physicians (Wagner et a1. 1999). The core values of CAM (e.g.,
individualism, personal responsibility) offer a response to the growing complaints of
conventional medicine, with its lack of emphasis on the individual, on prevention, and on

caring (Goldstein 2000). The increasing popularity of CAM in recent years has further

21

increased the options to patients regarding health care interventions (Pachter et al. 1998).
A major incentive reported by users of CAM is the ease of access to the alternative
medicines (Wagner et al. 1999). Furthermore, individuals with unmet healthcare needs
are signiﬁcantly more likely to use CAM (Sturm and Sherbourne 2001).

Alternative medicine is now a common option for many sectors of society
including the elderly (Smola et al. 2001), children (Pachter et al. 1998), minorities and
non minorities (Chen MS 1999; Mackenzie et al 2003; Pachter et al. 1998). Among rural
residents in Illinois, nearly two thirds of 176 respondents reported use of CAM (Herron
and Glasser 2003). CAM has also become a common treatment for a multitude of
conditions from the common cold (Pachter et al. 1998) to diabetes (Egede et al. 2002)
and disability (Ong et al. 2002). Independent predictors of CAM include chronic pain
(Haetzman et a1 2003), disability (Ong et a1. 2002), diabetes (Egede et al. 2002), older
age , higher educational attainment (Cherniack et al. 2001; Mackenzie et al. 2003) and
female gender (Cherniack et al. 2001; Mackenzie et al. 2003). Among a population of
elderly primary care patients (Cherniack et al. 2001) and among Taiwanese hospitalized
patients (Yang et al 2002) use of CAM was not correlated with income, race, or self-
perceived health. However, among respondents under 65 years of age, having a chronic
medical condition signiﬁcantly increased the likelihood of concurrent use of care
(Muhajarine et al. 2000). Men, those reporting high levels of distress, and those for
whom spiritual values were important were also more likely to use CAM as an adjunct to
conventional medicine.

The relationship between the use of CAM to the presence of chronic illness is

expected to be increased signiﬁcantly by perceived control. Perceived control (i.e., self

22

esteem) is positively associated with behaviors that protect health (Lefcourt and
Davidson-Katz 1991). Empirical evidence shows that individuals with greater perceived
control behave more adaptively when faced with health concerns are more likely to
engage in preventive health behaviors, and to seek appropriate medical care (Lefcourt
and Davidson-Katz 1991). They are more likely to consider alternatives in the face of
barriers, to take action, and to choose adaptive strategies for dealing with stressors
(Cohen and Edwards 1989). Therefore, persons with high perceived control are more
likely to take an active role in their own healing and when encountering barriers to care
are more likely to seek out alternatives in order to gain the care they need (Goldstein

2000).

Dissertation Intent

The aim of this examination was to present survey data on the national prevalence
of chronic illness, the prevalence of discrimination related to seeking medical care, and
their relationship to the mental health and behavioral outcomes of depression and
utilization of alternative healthcare.
Primary Hypotheses:
Chronic illness signiﬁcantly increases the risk of mental distress/depression and
also signiﬁcantly increases the likelihood of reported use of alternative healthcare.
Secondary Hypotheses:
1) The relationship between chronic illness and depression is modiﬁed by the
experience of discrimination in general and by the experience of discrimination in

seeking medical care speciﬁcally. Those with chronic illness who report discrimination

23

will have increased risk of depression compared with those who experienced no
discrimination.

2) The relationship of chronic illness and depression is mediated by perceived
control. Those with chronic illness with high perceived control will have decreased risk
of depression.

3) The relationship between chronic illness and alternative care utilization is
modiﬁed by discrimination. Those chronically ill who report discrimination will have an
increased likelihood of alternative care compared with the non-discriminated.

4) The relationship between chronic illness and alternative care utilization will be
mediated by perceived control. Those with higher perceived control will report greater
alternative care use even among those with chronic illness.

The design of this empirical examination is a quasi-experimental case comparison
study in which cases are survey respondents with a chronic disease and are compared to
respondents who do not have a chronic disease. The analysis attempted to provide
empirical support that the presence of chronic illness is a unique source of stress and that
differential exposure to discrimination in general and in medical care services speciﬁcally
accounts for a greater magnitude of mental strain for those reporting the discrimination.
Speciﬁcally, it is hypothesized that persons with a chronic illness will report more
experiences of 1) daily stress, 2) chronic stress, and 3) depression, and will be more likely
to turn to alternative treatments in their attempt to cope with the ongoing struggle of
dealing with a chronic illness. F urtherrnore, the additional burden of experiencing
discrimination when seeking medical care for their illness may place ill persons at greater

risk of mental distress than their non-ill counterparts.

24

Daily stress, as measured by the report of daily coping demands was compared
between households reporting the presence of a chronic illness and those reporting no
chronic illness. Chronic stress, as measured by the report of life’s demands, sadness,
wony, anxiety and depression was compared between the case and comparison groups.
Alternative therapies, as measured by reported use of herbal supplements, mega vitamins,
and other alternative care was compared between the case and comparison groups.

The primary hypothesis was that distress (depression and anxiety indexes) will be
signiﬁcantly greater among those with illness than among those without an illness beyond
that which is attributed to socioeconomic status, educational attainment, or race.
Furthermore, those experiencing discrimination, and speciﬁcally those experiencing
discrimination in seeking medical care, will have signiﬁcantly more mental distress than
those who do not experience discrimination. We believed the addition of discrimination
as a modifying factor in the relationship between the presence of a chronic illness and
mental distress would increase the association signiﬁcantly. Finally, among chronically
ill (persons, those reporting experiences with discrimination would be signiﬁcantly more
likely to report alternative therapy use than those not experiencing discrimination.

The proposed hypothesis was tested by secondary analysis utilizing a population-
based national survey conducted in 1996. Subjects for the current analysis were
classiﬁed from the original dataset of over 4000 respondents according to family status
and the presence of a chronic illness. Index measures were constructed ﬁom survey
scales and validated with reliability and correlation testing. The dependent variables
(daily stress and depression/anxiety) were hypothesized to predict the presence of a

chronic illness compared to those not chronically ill. Coping, by means of higher

25

perceived control and greater alternative therapy use, was hypothesized to mediate the
effect between illness stressors and depression/anxiety helping to diminish the effect but
not abolish it altogether. Discrimination was examined for its moderating effect on the
chronic illness and the depression/anxiety relationship. The experience of discrimination
was expected to increase the effect on depression and anxiety. However, its effect was
hypothesized to be larger for those with a chronic illness than those without such an

illness.

26

CHAPTER TWO

METHODOLOGY

Dataset Description

The National Survey of Midlife Development in the United States (MIDUS) was
a collaborative, interdisciplinary investigation of patterns, predictors, and consequences
of midlife development in the areas of physical health, psychological well-being, and
social responsibility. Respondents were drawn from a nationally representative random-
digit-dial (RDD) sample of non-institutionalized, English-speaking adults, aged 25-74,
selected from working telephone banks in the United States. The survey was conducted
during 1995 and the database made available electronically by the Inter-university
Consortium for Political and Social Research (ICPSR) as a transportable ﬁle provided as
SPSS and SAS export ﬁles. Variables within the data ﬁles were recoded to ensure
respondent anonymity. The telephone interview of 4242 respondents lasted an average of
thirty minutes and mailed questionnaires took an estimated average of an additional two
hours to complete. The RDD screening was attempted on 20,000 households. Table 1
provides the sample disposition prior to an attempt to convert refusals into usable
interviews by offering additional incentives.

The most common reason for an incomplete screen was due to a non-working
number (44.1% or 5242). The most common reason among complete screens for an
unusable interview was the 44.9% (n=3636) who were ineligible due to age restrictions
(i.e., no one available within the 25 — 74 year age range or the eligible respondent was

terminated to maintain probability sampling).

27

TABLE 1: MIDUS Sample Disposition Prior to Refusal Conversion

 

 

 

Screening Totals (%) Screening Not Compléted 7 j Totals (%)
Completed (n=l 1,900 or 59.5%) (n=20,000)
(n=8100 or 40.1%)
Completed Interview:
Usable 3323
Unusable 2
Total 3325
(41.1)
Appointment not
completed 131 (1.6) No contact 1246 (10.5)
Refused:
Prior to interview 932 Non-household 2743 (23.1)
After interview 78 Non-working number 5242 (44.1)
begun 1008 Total 7985 (67.1%)
Total (12.4)
Total Eligible 4464
(55.1)
Ineligible: Ineligible:
No one 25 -74 988 Language problem 362
years 2473 Circumstantial 154
Rsswndent age 71 Appointment missed 66
tmnated 104 Refused 2087
Language P‘Pblem 3636 Total 2669 (22.4)
Circumstantial (443)
Total

 

 

A total of 5676 people were eligible overall, for an initial telephone response rate of
70.0% (3971/5676).

Sample representation was increased by using a series of weights adjusted by the
MIDUS investigators for differences in the probability of selection and differential non
response. They developed a total of six weights and the product of the weights was used
to create a ﬁnal summary weight. Each telephone number was classiﬁed according to
census area, age group, race, education, and income. These variables were analde to

help predict variation in survey cooperation and probability of response was applied to

28

each type of respondent. Success in obtaining a listing was negatively associated with
neighborhood proportions of Hispanics, positively associated with neighborhood
proportions of young adults (age 0 — 24 years), and negatively associated with any
household in the Northeast section of the United States. Thus, greater population
representation was achieved with this adjustment. In addition, the sample was also
weighted to achieve equal numbers of men and women in each of ﬁve decades of age.
Younger women in small households were much more common than older men
Therefore, the probability of rejecting women in younger age groups (25 -- 34 years) was
much greater than for men in older ages (65 — 74 years). However, a random selection
process was used to reject these more abundant interviews. A consistent selection
probability sampling was always used to guarantee that easy-to-reach and hard-to-reach
respondents had the same probability of being recnrited into the sample. This would not
have been the case if a non-probability quota design had been used. Additional attempts
and incentives were utilized to increase the likelihood of a completed survey among those
with a low probability so as to minimize bias.

This weighted sample is as representative as possible regarding the
demographic variables of gender, race, marital status, and age. Age was collected as a
continuous variable but reported here in groups for comparison to the national dataset.
The weighted sample leans more heavily to males because older men were intentionally
over sampled to account for their differing population proportions and for the tendency of
men to have a lower probability of agreeing to be in surveys than women.

Of the 4464 eligible interviews, an 86.8% conditional response rate was observed

for mail surveys providing a ﬁnal sample of completed, matched telephone and mail

29

surveys of 3682. Post stratiﬁcation revealed that the resulting MIDUS weighted sample
matched much more closely with the national population than the original un-weighted
sample (Table 2).

TABLE 2: Comparison of MIDUS Sample and National Population
Characteristic Proportions

 

 

 

 

 

 

 

 

in" E “PM 7E1}? "I"??? ’
Region:

Northeast 20.6 18.0 18.7

Midwest 23.7 27.4 25. 1

South 34.1 35.2 37.7

West 21.6 19.5 18.4

Gender:
Male 48.3 48.5 43.5
Female 51.7 51.5 56.5
Race:

White 84.8 87.3 84.1
Black 11.2 6.1 10.8
Others 4.0 6.5 5.1

Age:

25-34 27.6 20.8 26.0

35-44 27.0 24.2 27.8

45-54 19.2 24.0 19.1

55-64 13.9 19.9 15.2

65-74 12.2 11.1 11.8

Education:
< 12 15.8 10.0 13.2
12 36.4 29.3 38.3
> 12 47.8 60.8 48.5
Marital Status:
Married 67.5 64.0 68.1
Not married 32.5 36.0 31.9

 

 

 

Participating respondents were queried in the initial telephone interview and
responded to a mail questionnaire. The main dataset contains responses from the survey
of 4,242 respondents. They were asked to provide extensive information on their physical

and mental health throughout their adult lives, and to assess the ways in which their

30

lifestyles, including relationships and work-related demands, contributed to the
conditions experienced. Histories of physical ailments, treatment and/or lifestyle
behaviors, and lifestyle changes were obtained with responses on how these factors
affected the respondent’s physical and mental wellbeing. Additional questions addressed
the respondents’ sense of control over their health, awareness of changes in one’s medical
condition, commitment to healthy behaviors, and experiences with nontraditional
remedies or therapies. Information on work histories, on signiﬁcant others, on physical
and emotional demands, and on how their personal health correlated to their jobs were
also elicited. Respondents were also asked to consider their personal feelings of
accomplishment, desire to learn, and sense of control over their lives. Background
information on each respondents included age, gender, education, religion, marital status,
employment status, age of children, household income, race, ethnicity, sexual orientation,
height, weight, insurance coverage, spouse’s employment, and parental histories.

The data used for this evaluation came from the 86.8% or 3690 respondents who
also returned the mail survey. Because the measure for chronic illness was obtained from
the mailed questionnaire, this analysis was conducted on the 3690 respondents
completing both the telephone and mail survey. The subset of respondents not completing
the mail survey differed signiﬁcantly in some characteristics from those who completed
both. Those who did not return the mailed survey were signiﬁcantly younger (mean 42
years v5.47 years), more likely to be male (56% vs. 50%), less likely to be married (49%

vs. 63%), and less likely to have a high school degree (56% vs. 63%). (Table 3).

31

TABLE 3: Comparison of Respondents Returning and Not
Returning the Mailed Questionnaire

 

 

 

 

 

 

 

Percents Returned Mail Did Not Return Chi-square
(unless otherwise Surveys Survey (p—value)
Ln_oted) (n=3690) (n=552)
Age (mean, sd) 47.0 (13.25) 41.9 (13.32) 8.4 (<.001)
Gender: Male 50.0 56.0 6.8 (.009)
Female 50.0 44.0
Working now: Yes 61.4 39.3 0.4 (.51)
No 38.0 59.6
0.6 1.1
Missing
Marital Status: 63.3 48.6 42.6 (<.001)
Married 36.7 51.4
Not
married
Education: 9.7 (.002)
<=H.S. degree 37.3 44.2
> H.S. degree 62.7 55.8

 

 

Because the group not returning the mail survey differed signiﬁcantly on demographic
parameters, they are likely to differ on other characteristics compared to those who did
return the survey. Therefore, they could signiﬁcantly bias the resulting estimate regarding
chronic illness, distress, and alternative treatment use. The analysis for this study thus
concentrated on the 3690 who participated in both the telephone interview and mail
survey.
Strengths and Limitations of Using MIDUS

The use of the MIDUS Survey to address the question of mental distress among
those with a chronic illness has several strengths. The survey sample is a nationally
representative sample with a broad range of ages among families dealing with an ill adult
and represents a broad range of chronic illnesses or conditions. This dataset is thus

representative of the general population and provides a great deal of generalizability of

32

ﬁndings. The MIDUS survey covers a broad range of chronic illnesses or conditions and
is therefore more representative of all those struggling with a chronic illness. Like other
national datasets, the MIDUS total sample has a small proportion of minorities (5%).
However, the proportion of Aﬁican American minorities in each subgroup is similar
(7.5% versus 5.7%).

Respondent report of chronic illnesses or conditions is a strength of the MIDUS
study. Chronic illness was quantiﬁed by separate items for a broad range of chronic
conditions. The conditions deﬁned as chronic was clearly labeled as persistent illnesses
or conditions that require ongoing medical care. Items included commonly understood
illnesses as asthma, diabetes, and HIV but also included less commonly considered
conditions as arthritis, joint disease, and migraines. A total of 29 illnesses and conditions
are considered and therefore would validly classify anyone with a persistent chronic
condition into the group of those with chronic illness or chronic condition.

The MIDUS Survey also has strengths for addressing our hypothesis because of
the measures included in the survey. The measures for depression and perceived control
have previously been tested and validated in population studies and the scales developed
for this analysis resulted in reliability coefﬁcients greater than 0.80. The use of
depression and perceived control inventories to identify respondents at risk for chronic
and daily stress has been extensively tested in national population studies as a valid and
reliable tool for screening depressive and distressed states (Mirowsky 1994). Although
clinical diagnosis of depressed states is more speciﬁc, inventories can reveal patterns of
correlation that are obscured by diagnostic symptoms (Mirowsky 1994). By utilizing

items from the inventories in the MDUS Survey dataset, patterns of distress among

33

families may be distinguished according to the presence of a chronic illness that might
otherwise be missed using diagnostic symptoms found in other datasets.

Use of the MIDUS Survey to test the hypothesis that personal distress differs
according to the presence of illness has some limitations. Tlmt minority discrimination is
not the focus of the MIDUS survey is one limitation. The hypothesis results could suffer
from a lack of power since a small sample of Aﬁican Americans with chronic illness
(n=163) participated in the study and even fewer reported having no chronic illness
(n=63). This represents only 5.3% of the total sample of 4242 respondents.
Discrimination in obtaining healthcare services is particularly acute for Black minorities
and because minorities represent a small proportion of this survey, reported
discrimination among them represents only 7.5% (n=17) of the 230 respondents
identifying themselves as a black minority. Reported discrimination in any service is,
however, a much larger number. There were 175 or 76% of the 230 Black respondents
reporting day-to-day discrimination in the receipt of any service.

Depression indexes include both psychological and somatic symptoms in
identifying people at risk for depressed states. Using a combined index then can create
problems from somatic symptoms that may imitate depression but actually be indicative
of physical illness. However, questions identifying somatic symptoms were removed
from the ﬁnal depression index of this study. The depression screening questions used for
the scale in this study (presence of chronic sadness, hopelessness, or unworthiness) have
been shown to be accurate in identifying those with chronic stress intense enough to
cause a negative impact on mental health (Mirowsky 1994). The index for perceived

control in MIDUS infers only daily stress from the demands of life in general. Ideally,

34

questions speciﬁc to illness-related demands and dual role demands would have been
measured. However, the general perceived control index does not discount symptoms that
coincide with stressors speciﬁc to dual role and illness that may be causing the stress.

The use of traditional measures for socio-economic status is a liability in this
survey. Household annual income was reported separately for respondent, spouse, or
other income earner in categories of one thousand units up to $20,000. The categories
are then reported only by units of 5000 up to $50,000 after which they become units of
25000 up to $100,000, units of 50000 up to $200,000 and units of 100000 thereafter.
Accurate comparisons are skewed toward the lower income brackets and adding all
income earners of a household obscures calculated means. Therefore, it was necessary to
use respondent income and measures of perceived ﬁnancial situation (i.e., belief that one
has enough money to meet needs and level of difficulty in paying bills) to compare
ﬁnancial SES.

Caution must also be observed in the ability of this database to suggest
associations that are true cause and effects. The design of the MIDUS Study is cross-
sectional and any associations may be confounded by unmeasured factors or by selection
biases. Temporal bias is a particular threat to the validity of any ﬁnding from the current
analysis. For example, those with a tendency toward depression or anxiety may be more
likely to categorize acute illnesses as a chronic condition. However, they are just as likely
to misclassify a chronic condition. For example, asthma and diabetes often go
undiagnosed and untreated for some time among persons who shun the health care
system because their own mental problems may make them more likely to be unaware of

a chronic condition early on. However, the proportion in the study that this would

35

represent is likely small and the many strengths of using MIDUS for this analysis out
weighs these few limitations.

Description of Measures

Abbreviated versions of the depression, daily stress, and perceived control scales were
applied for this analysis. The scales were modiﬁed to provide the best ﬁt and items with
high response rates and high construct validity were retained within each scale.
Dependent Variables

Depression: This was inferred from the psychological attributes of sadness,
hopelessness, worry and feeling that everything was an effort deﬁned as the presence of
any one of these psychological attributes at some point within the previous 12 months.
Reponses were dichotomous, yes or no, as to whether they were ever present for two
weeks or more during the previous year. The 4-point scale of excessive worry was based
on a relative measure of whether the respondent felt that in the previous 12 months they
worried more, less, about the same, or not at all compared to most people. The depression
index also used the positive psychological attributes such as cheerfulness, happiness,
peacefulness, and satisfaction with life which was then reverse coded. These were
deﬁned as the frequency of whether such feelings were experienced all of the time, most
of the time, some of the time, or not at all during the previous 30 days.

Face validity identiﬁed 38 potential items for a depression index including other
psychological attributes such as loss of interest, thoughts of death, feeling irritable and
somatic attributes of loss of energy, loss of appetite, trouble sleeping, restlessness, and
aches or pains. These other psychological attributes and the somatic attributes were

dropped from the ﬁnal index due to poor reliability with the above attributes. The alpha

36

score for the original 38 items was poor, reaching only 0.2401. Removal of the other
psychological attributes (e.g., loss of interest) and of the somatic attributes resulted in a
Cronbach’s alpha score of 0.5993. The inter-item correlation is only moderate because
the somatic attributes have lower validity in identifying depression since they can be just
as common in physical illnesses and conditions as they are in psychological conditions.
The response items to each positive attribute had to be transposed due to the inverse
wording of the questions compared to depressed questions. Inversing the positive
attributes of happiness, peacefulness, and satisfaction with life and adding them to the
retained psychological attributes resulted in a ﬁnal reliability coefﬁcient of 0.89. The
ﬁnal ll-item index was signiﬁcantly correlated across the items, ranging in correlation
coefﬁcients of 0.25 to 0.65 (p-value <0.0001).

The items in the retained index were averaged over all eleven questions with a 5-
item Likert scale from 1 (least depressed) to 5 (most depressed). Overall frequencies

resulted in 5% indicating most depressed and 18% indicating not depressed (Table 2).

TABLE 4: Distribution of Depression Responses

 

 

 

 

 

 

 

 

 

2

How Depressed imple Willa-mum Non ﬁgoznders Stefanie

(p-value)
Most 1.6% 28.8%
Somewhat 10.2% 66.5%
A little 66.6% 0.7% 101.4
Not much 21.2% 3.0% (<0.001)
Not at all 0.4% 0.7%
Missing 1 1

 

 

 

 

Alternative Therapy Use: This was measured by the summation of a list of

responses for potential uses of alternative therapies (Appendix A). The survey listed 20

37

items identiﬁed to the respondent as an alternative therapy. Items were expressed as
presence of ever-use behaviors using a dichotomous scale (yes/no use) and included such
common therapies as mega-vitamins, herbal supplements, special diets, special exercises,
prayer, and chiropractic visits. They also included less familiar alternatives such as
acupuncture, biofeedback, energy healing, hypnosis, and meditation. Additional
questions included the practice of using sedatives, stimulants, or prescription painkillers
without a prescription or in a dosage beyond that prescribed. These items are valid
constructs for this examination because people who are stressed and who may perceive
the standard health care system as failing them are more likely to go to such lengths as
using these medications when not prescribed speciﬁcally by a physician for their stress.
The reliability coefficient for the 20-item scale was 0.6886. Removal of any one of the
items did not change the coefﬁcient, the Cronbach’s alpha score ranged between 0.67 and
0.69 regardless of items removed. Therefore, all 20 items were retained for the ﬁnal
scale.

The scale of each alternative therapy was dichotomized as to whether the therapy
was ever used or not used by the respondent. The scales were summed and this scale
item ranged from 0 (have never used any one) to 13 (have used these at least once). A
cumulative total of 99% was reached for those reporting the use of 8 or less therapies.
The distribution was skewed toward zero with 44.7% reported never having used an
alternative therapy. However, a majority (54.8%) did report every using one or more

type of alternative care (Table 5).

38

TABLE 5: Distribution of Alternative Therapies Used

 

 

 

 

 

 

 

 

 

 

 

 

 

2

Sample Mail imam Non geriatrders S tafis-tic
Types Used '/o (p-value)
None 44.7% 99.5%
One 23.2% 0.2%
Two 13.5% 0.2% 100.6
Three 7.9% 0% (<0.001)
Four or more 10.5% 0.2%
MissinL 0 562

Independent Variables

Presence of Chronic Illness: The presence of a chronic illness was deﬁned
speciﬁcally for respondents by a list of conditions and illnesses that are persistent and
require ongoing medical care. Thus, chronic conditions are clearly differentiated from
acute self-limiting conditions. The response for each item on the list is dichotomous,
either present or not. The full list is provided in Table 9 (p. 56). Chronic illness was
initially measured in MIDUS from a broad range of conditions, several of which could be
considered less than chronic. Some of the listed conditions may either be self limiting or
symptoms may be so minimal that they do not require encounters with the health care
system. For example, hay fever or varicose veins which are included in the list of chronic
conditions may manifest with minimal symptoms requiring only occasional trips to the
drug store rather than to a healthcare professional. Conditions such as gall bladder
problems or hernia can manifest acutely and result in surgery that quickly corrects the
condition and therefore not require long term encounters with the healthcare system. The
MIDUS survey did not distinguish between current long term suffering and treatment for

the condition in the past year. Measurement in this way may have obscured the effect of

39

chronic illness. Therefore, the regression model was retested using a revised list of only
serious, long term chronic illnesses. The conditions included for this analysis were
sciatica, bone disease, hypertension, bladder problems, asthma, coronary heart disease,
diabetes, migraines, thyroid disease, autoimmune disease, cancer, and stroke. The other,
self-limiting conditions were included into the non-chronically ill subgroup.

However, whether a condition is chronic or self limiting may differ across
individuals. Some people may suffer for longer periods and with worse symptoms for a
condition than others with the same condition and therefore, one person’s chronic
condition may be another person’s self-limiting condition. Depending on whether the
condition is experienced as debilitating or limiting will determine whether a person
perceives a need for ongoing treatment and thus multiple, ongoing encounters with the
healthcare system. Therefore, a global measure was used to reanalyze the presence of a
chronic condition according to the respondent’s perception of debilitation of physical
health. This perceived health was determined in the MIDUS survey by a Global Health
Questionnaire that queried respondents on a 5-point Likert Scale according to whether, in
general, they would rate their physical health of the previous year as: poor, fair, good,
very good, or excellent.

Perceived Control: This variable refers to self-esteem and labeled perceived
control by the MIDUS survey. Indicators included the psychological attributes of the
belief that certain actions will achieve desired goals. For example, the belief that one can
do anything one desires, that one is responsible for own successes, that misfortunes are
the result of own mistakes, and that one is responsible for ones own failures. The inverse

of perceived control infers the belief of powerlessness as indicated by feeling helpless

4O

dealing with problems, things being beyond own control, multiple things interfering with
own desires, and feelings of little control over problems. These items described
perceptions of self control or powerlessness, of managing the present and controlling the
future, and of changing one’s situation. Items elicited responses from a 7-item Likert
scale on the level of agreement for the extent to which the above items described the
individual from 1 indicating strong agreement to 7 indicating strong disagreement.
Initially, 23 items were identiﬁed by face validity for the perceived control scale. An
initial Cronbach’s alpha score of 0.675 was attained.

Construct validity testing reﬁned the perceived control scale and identiﬁed 19
items of higher reliability. This group of items resulted in an adjusted Cronbach’s alpha
of 0.7899. Items contributing to the initial lower score were those reporting on how
pleased with life one was, how in charge of life one felt, and whether the future depended
mostly on oneself. Six items elicited responses on whether the respondent liked their
personality, was pleased with how things have turned out, and whether they believe they
are good at managing daily life. Because these items tap less into feelings of control than
feeling overwhelmed with demands, they were removed. Removal of these items resulted
in an improved reliability coefficient of 0.865 for a ﬁnal l4-item index. Items retained
were those eliciting level of agreement about perceived powerlessness as in feelings of
helplessness, feelings of powerlessness to make changes, and feelings of a lack of control
over life. The ﬁnal l4-item index was signiﬁcantly correlated across the items, ranging in
correlation coefﬁcients of.0.29 to 0.50 (p-value <0.0001).

The score in the retained index was averaged over all fourteen items. Response

items ranged from 1 (strongly agree that one is powerless) to 7 (strongly disagree). The

41

frequencies resulted in a normal distribution (mean= 4.7, median= 5.0, and mode= 5.0)
with approximately 13% indicating lower control and 22% indicating highest control

(Table 6).

TABLE 6: Distribution of Perceived Control

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I

Sample Maillg-eggggders Non llze-sg‘oznders S tafi; tic
High Control % - — (p-value)
Strongly Agree 0.8% 0%
Somewhat Agree 4.3% ' 0%

Agree 11.3% 0.5%
Neutral 23.1% 0.4% 110
Disagree 32.7% 0.2% (< 0001)
Somewhat Disagree 25.9% 0.4% '
Strorgly Disagree 1.4% 0%
Total 3662 8
Missing 18 (0.5%) 554 (98.6%)

 

Quality of Support indicators included respondent report of perceived level of
spouse or partner caring, appreciation, understanding and reliability. Six measures were
identiﬁed by face validity for a quality of support scale. Respondents were asked to rate
the quality of their relationship with their partner on a scale of l to 5 (1=excellent,
5=poor). Four questions asked respondents to report how much their partner cared for
them, understood them, appreciated them, and how much they could rely on their pmtner
for help. These scales ranged from 1 to 4 (1= A lot, 2= Some, 3= A little, and 4= Not at
all). The sixth question asked respondents to report on whether their partner makes too
many demands using a scale of 1 to 4 (1=oﬁen, 4=never).

Construct validity testing reﬁned the six question scale, with a Crohnbach’s
reliability score of 0.79, to a ﬁve question scale that included the ﬁrst ﬁve items

discussed above. Removal of the question on amount of demands the partner places on

42

the respondent resulted in retaining the ﬁve questions into a single indicator of support

quality which had a ﬁnal Crohnbach’s alpha of 0.90. The items had high correlation

ranging from a correlation coefﬁcient of 0.58 to 0.72. The frequencies in the resulting 4-

item Likert scale were similar between the sub sample and total sample (Table 7).

TABLE 7: Distribution of Quality of Support

 

 

 

 

 

 

 

 

 

 

 

Sample Mail Responders Non Responders X2-

. . . .

Quality of Support /0 N=3680 N=562 Statistic
(p-valne)

A lot (Excellent) 0.4% 0.2%
Some (good) 57.3% 0.9%
A little (Fair) 9.2% 0.2% 115 (<
None (Poor) 2.8% 0% .0001)
Total 2567 7
Missing 1113 (30.2%) 555 (98.8%)

 

 

Discrimination in obtaining medical services was measured by respondent report
of being denied care or provided inferior medical care due to gender, race, or ethnicity
discrimination. A total of 974 respondents in the total sample reported denied or inferior
care one or more times in their life when seeking medical care. There were 533 (12.6%)
of the total sample reporting such discrimination due to race or ethnicity and an
additional 459 (10.8%) reporting denied or inferior care due to gender discrimination. In
addition, the impact of discrimination was measured by two questions that asked
respondents how much discrimination interfered with life and how much harder life was
due to discrimination. Both were based on a 4-point Likert scale (1=A lot, 2= Some, 3== A
little, and 4= Not at all). The two questions were summed and then a correlation
coefﬁcient calculated of 0.985 and a Crohnbach’s alpha score of 0.99. The frequency of

items was similar between the total sample and sub sample (Table 8).

43

TABLE 8: Distribution of Discrimination

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2

Sample Mail 1:de“ Non ge-sspglznders S tﬁistic
Med Discrimination % - _ (p-value)
Denied or Inferior Care:
Ever 2.9% 0.5%
Never 90.0% 1.2% 101
Total 3418 10 (<.0001)
Missigg 262 (7.1%) 552 (98.2%)
General
Discrimination
Interferes very much 2.3% 0%
Interferes Somewhat 4.9% 0.4%
Interferes a little 13.8% 0.4% 110
Interferes not at all 26.5% 0% (.0001)
Total 1749 10
Missing 1931 (52.5%) 552 (98.2)

Plan of Analysis

This examination compared cases (respondents with a chronic illness) and a

comparison group (respondents without chronic illness) across all descriptive and

 

outcomes variables using the Chi-square test for nominal categorical-level variables and
the Wilcoxon two-sarnple test for continuous variables. Variables relevant to the research
question were explored descriptively and included the demographics of the respondent
(i.e., age, race, marital status, and socio—economic status), discrimination (general and
speciﬁc to healthcare access), illness prevalence, perceived control, daily stress,
depression, and use of alternative therapies. Descriptive analysis was also conducted on
variables relevant but not central to the research question and included number of
children, and quality of partner support. These were treated as control variables.

Multiple regression analysis was conducted on models for each dependent

variable of daily stress, depression, and alternative therapy use. These dependent

variables; chronic stress (14 —- 98), depression (11 - 51), and alternative therapy (0 — 19)
have been scored so that lower scores reﬂect greater stress and psychological distress.
Variables regressed for each dependent variable considered central to each model
included cases (chronic illness), lifetime discrimination (1 - 7), level of perceived control,
and socioeconomic status. Variables controlled within each model included prior use of
health care services, number of children in the family, marital status, and quality of the
marriage (i.e., functional support). Predictor variable arrangement differed for each
model. Predictor variables included chronic illness, discrimination, perceived control, and
support. A ﬁnal linear regression model was constructed that estimated the signiﬁcant
contribution of discrimination to the relationship of case status in predicting stress and
psychological distress. The following assumptions were included in the model: the
relationship between each dependent variable and independent factors is linear, the
distributions of the independent variables have equal variances for each value of the
dependent variable, and the values for each explanatory variable are independent across
the dependent variable. The coefﬁcient of determination (R2) was used to describe the
amount of variation explained by the model of chronic illness on each dependent
variable. A multiple linear regression model was developed to predict depression and
anxiety based on the presence of chronic illness (1=has a chronic illness, 0=no chronic
illness); perceived control (1=feel highly powerless to 7=feel highly empowered); and the
quality of support from a signiﬁcant other (1=great deal of positive support to 4=no
support). For alternative therapy, a multiple logistic regression model was used where

reported use of alternative therapy (1=any use, 0=never used) was predicted based on the

45

 

presence of chronic illness, perceived control, and quality of support as above. The model
developed is demonstrated by the formula:
Y= a + le - sz - bX3

Where:

0 Y=dependent variable (depression, anxiety, or CAM)

0 X1 is the presence/absence of chronic illness in the respondent,

0 X2 is the level of personal agency or perceived control score,

0 X3 is the level of support from a signiﬁcant other.
Covariates controlled within the model include gender, race, education, employment,

insurance, income, and having children at home. A stepwise model was employed where:

 

Model # 1: Dependent variable (depression/anxiety or CAM) modeled on chronic illness.
Model #2: Dependent variable on chronic illness and perceived control.

Model #3: Dependent variable on chronic illness, perceived control, and support.

Model #4: Dependent variable on chronic illness, perceived control, quality of support,
and discrimination in medical care.

Model #5: Dependent variable on chronic illness, perceived control, and quality of
support, and discrimination in general.

Model #6: Dependent variable on chronic illness, perceived control, and quality of
support, and both discrimination factors.

Model #7: Interaction term between medical discrimination and chronic illness.

Model #8: Interaction term between perceived discrimination and chronic illness.

 

46

 

Interaction terms were built for discrimination in the receipt of medical services
(Discrl) and perceived discrimination (Discr2) and chronic illness. The product of
perceived discrimination (measured as interferes with life 1= A lot --- 4=Not at all) was
computed with chronic illness (measured as global health 1 = poor ---- 5= excellent). A
second interaction term was created measured as the number of times discriminated
receiving medical care and chronic illness. The scale levels indicating don’t know or
missing (numerical values of 8 or 9) were eliminated. The resulting scaled levels created
for the interaction term numbered from 1=worst up to 20 = Best (excellent health and no
discrimination). The interaction term was entered separately in each regression model in
the seventh and eighth step for each predictor, depression and alternative medicine. The

product terms for each are labeled Discrl * illness and Discr2 * illness.

Summary

Because chronic illness includes the proliferation of multiple stressors, a chronic
illness can lead to signiﬁcant psychological distress. This proposal has suggested that
within the framework of the stress proliferation model, a chronic illness is a signiﬁcant
predictor of distress. Coping through utilization of alternative therapies was proposed to
mediate the effect on distress. Furthermore, the addition of discrimination was
hypothesized to add signiﬁcantly to the prediction. A multivariate regression analysis

examined the contribution of chronic illness, to depression, anxiety, and CAM.

47

CHAPTER THREE

RESULTS

A total of 3680 completed both the telephone interview and the mail survey. Of
these, 728 (19.7%) were being treated for depression or anxiety within in the last nine
months and were eliminated from the ﬁnal analysis because there was no way to

determine the duration or effectiveness of treatment.

Total MIDUS Survey
N=4242

Mail Survey Completed
Returned Mail Survey Not Returned
=3 890 n=552

Depression/Anxiety
Not currentlv under treatment Currentlv un_der tregtrnent
N=2952 n=728

Illness Status
Chronic Illness No chronic illness

n = 2095 n = 839 missing = 18

T g l
l

 

Final n = 2934

FIGURE 3: Respondent Sample Flowchart

48

Chronic illness was reported by 2095 (71.4%) and 839 (28.6%) reported no
chronic illness (Figure 3). Among the respondents reporting a chronic illness, only 21%
had a single condition. Two or more conditions were reported by 56% of ill respondents.
No chronic illness was reported by the remaining 839 or 28.6%. Table 9 presents the
prevalence of each reported chronic illness. The most common were bone disease

(17.7%), hypertension (16.6%), and back problems (16.5%).

Table 9: Prevalence of Reported Chronic Conditions in Final Sample
(n=2934)

 

Cogdition Percent Number
Bone or joint disease 17.7 518
Ulcer or stomach problems 16.8 493
Sciatica or lurnbago 16.5 484
High blood pressure 16.6 486
Hay fever 14.1 414
Urinary or bladder problems 11.1 327
Asthma, bronchitis, or emphysema 10.5 307
Persistent foot trouble 9.9 291
Piles or hemorrhoids 9.4 276
Persistent skin trouble 9.4 277
Persistent trouble with teeth 8.1 237
Migraine headaches 7.3 215
Chronic sleeping problems 7.0 205
Cancer 6.5 190
Gum or mouth trouble 6.0 175
Diabetes or high blood sugar 4.9 145
Constipation 4.0 1 17
Thyroid disease 3.5 104
Other lung problems 2.6 77
Hernia or rupture 2.2 66
Gall bladder trouble 1.7 51
Alcohol or drug problem 1.6 46
Neurological disorders 1.6 47
Heart attack 1.4 42
Autoimmune disorders 1.0 28
Varicose veins requiring treatment 0.9 25
Stroke 0.6 17
AIDS or HIV infection 0.2 6
Tuberculosis 0.2 5

49

The demographic proﬁles for each group (those with chronic illness and those

with no illness) are presented in Table 10.

Table 10: Comparison of Demographic Characteristics among Those
with and those without a Chronic Illness (%)

 

 

Characteristic No Illness Chronic Illness Chi square #
ln=839) (n=2095)
Age: mean (SD) 43.2 (12.61 48.6 (13.5) 9.97 **
Gender: Male 57.3 51.7
Female 42.7 48.3 7.6"
Race: White 83.8 85.1 1.5
Minority 13.7 12.0
Missing 2.5 3.0
Marital Status:
Married 65.8 65.8 0.0
Not married 34.2 34.2
Education:
< H.S. 6.2 9.3
H.S./GED 27.6 28.8 4.4“
Some College 23.1 22.3
College degree 42.9 39.7
Employment:
Work now 70.3 59.0 30.9"
Not working 24.6 34.6
Missing 5.1 6.4
Income: < $10,000 21.2 30.2 33.7 "
$10-$19 14.2 15.1
$20 - $49 31.6 26.4
$50 - $75 21.8 16.4
> $75,000 11.2 11.9
Insurance: Yes 82.6 81.3 0.64
None 17.4 18.7
Any Kids: Yes 45.5 34.7 29.6 **
No 54.5 65.3

 

* p< 0.05, ** p< 0.01, # age signiﬁcance test by Student t-Test.

50

The two groups were similar in marital status, race/ethnic makeup, and insurance
status. However, the groups did differ signiﬁcantly on several characteristics. The chronic
illness group was composed of more females (48.3% vs. 42.7%) with a slightly higher
group mean age (48.6 yr vs. 43.2 yr). The ill group had fewer currently employed (59.0%
vs. 70.3%), fewer current smokers (38.2% vs. 43.5%), a greater proportion in low income
bracket (30.2% reporting < $20,000 vs. 21 .2%), lower educational attainment (6.2 % vs.
9.3% < HS degree), and fewer with children living at home (34.7% vs. 45.5%). These
factors were statistically controlled in the multivariate analysis.

Univariate Analysis

Univariate analysis was conducted according the dichotomized variable of
chronic illness (presence or absence). Table 11 presents the mean score ﬁndings. The
chronic illness group had signiﬁcantly more respondents meeting criteria for depression

or anxiety and more reporting use of CAM.

Table 11: Univariate Analysis of Proportion with the Outcome by
Illness Group Status.

 

 

Dependent Variable

(Dichotomous: % yes) Chronic Illness No Illness Chi-square
(n=2095) (n=839)

Depression 7.8 4.6 9.2"

Generalized anxiety 1.3 0.8 1.1

CAM 55.2 39.1 62.3”

*p<0.05, "p<0.01

Mean scores for perceived control, quality of support, reported discrimination generally,

and discrimination with medical services speciﬁcally was compared between the chronic

51

illness group and the non-ill group. Table 12 presents the results. The groups differed
only in perceived control with the chronically ill reporting lower control (p<.01). The ill
group also reported more times discriminated in medical care but this was not

signiﬁcantly due to a few reporting a high frequency of discriminations.

Table 12: Univariate Analysis of Mean Values for Each Independent
Variable by Illness Group Status.

 

 

Group Status
Independent Variable
Mean score (SD)
Chronic Illness No Illness t-statistic
(n=2095) (n=839)
Perceived control 4.76 (1.12) 5.06 (1.06) 6.8"
Quality of support 2.17 (0.44) 2.15 (0.48) -1.7
Discrimination
Interferes with life 1.50 (0.76) 1.53 (0.81) -0.63
Discrimination medical
(mean # times in life) 0.13 (2.4) 0.3 (.23) -1.8

 

Mann Whitney-U t-Test for non-parametric data.
* p< 0.05, ** p< 0.01

Mental Health Outcomes Analysis

Unstandardized coefﬁcients and their corresponding p-values are reported. Table
13 presents the regression of the mental health and behavior outcome variables on control
variables. The table demonstrates that chronic illness is a signiﬁcant predictor of both
mental health outcomes and health behaviors. Respondents in the chronically ill group
were signiﬁcantly more likely to report depression and to use alternative medicine than
respondents who did not have a chronic illness (p<.001). Insurance was highly signiﬁcant

for depression (p< .01) but not a predictor of anxiety or CAM use. Respondents without

52

health insurance were more depressed than those with insurance. Age was associated

with depression. Younger respondents were more depressed than their younger

counterparts. Gender was a signiﬁcant predictor of alternative therapies (CAM).

Females trended to be more depressed and anxious, but not signiﬁcantly so, and

signiﬁcantly more likely to use alternative therapies. Education was a predictor of CAM.

Higher education was signiﬁcantly associated with more CAM use (p< .01).

Table 13: Regression of Chronic Illness and Other Predictors on

 

 

Outcomes.

Depression Anxiety CAM
Gender (0=male) .062 .084 .340"
Age (years) -.005"‘ -.002 -.004
Education (1=grade) -.063 -.048 .504”
Employed (0=Yes) -.012 -.018 .007
Income (1=<$10000) .012 -.015 .078
Any kids (0=none home) -.019 -.022 -.103
Insurance (0=no ins) -. 160" -.1 12 .083
Race . (0=White) -.020 -.01 l -.078
Chronic ill (0=not ill) 091* .026 .440”
Constant 3.1 .316 -.297

= .03 .01 .03

n 2674 2674 2674

 

Unstandardized, beta coefﬁcients reported

*p< .05, "p< .01

Table 14 — 16 presents regression results for depression, anxiety, and alternative

therapy use. The variables were added in six steps. The ﬁrst consisted of control variables

only. Chronic illness was considered an independent predictor and added next. The

effects of perceived control, the quality of social support, and the experience of

discrimination were considered to mediate or modify this effect so each successively

added next. The ﬁnal step added all predictor variables to the model. Table 14 presents

the regression results of predictors on depression.

 

Table 14: Regression analysis of Predictors on Depression.

 

 

 

 

 

 

 

 

 

 

Demog. Chronic Control Support Discrl Discr2
Gender 0.064 0.062 0.051 0.036 0.037 0.040
Age -.004* -.005* -.006** -.006" -.006" -.006"""
Education -.067 -.063 -.022 -.020 -.021 -.023
Employmt -.012 -.012 -.007 -.005 -.005 -.005
Income 0.01 1 0.012 0.027 0.026 0.027 0.028
Kids -.019 -.019 -.050 -.069 -.070 -.071
Insurance -.158“ -.157" -.139""" -.137" -.l33"‘ -.131‘
Race -.020 -.020 0.006 0.001 -.002 -.030
Chronic ill 0.091 " 0.047 0.059 0.057 0.061
Control -.124" -.110** -.110" -.105"
Support -.144" -.145*"' -.140"
Discr 1 0.004 0.004
Discr 2 0.052'
Constant 3.23 3.17 3.7 3.4 3.4 3.3

0.02 0.05 0.07 0.08 0.10 0.1 1
N 2674 2674 2667 1945 1867 813

 

 

*p<.05, "p<.01 (Unstandardized, beta coefﬁcients reported)

Chronic illness was initially a signiﬁcant predictor of depression. Respondents
reporting a chronic illness were more depressed than those with no chronic illness (p<
.05). The primary hwothesis that chronic illness increased the risk of depression was
srgvported in this model. However, the effect of chronic illness on depression was weak
to begin with and disappeared with the addition of perceived control. increased, the
likelihood of depression decreased (p< .01). Its effect became stronger with the addition
of perceived control. It remained a highly signiﬁcant predictor of depression in the
ﬁnal model (p< .01). The younger subgroup was more depressed than the older group
regardless of illness status (Table 15).

Health insurance coverage was a signiﬁcant, independent predictor of

depression (p < .01). Those with insurance coverage of any kind were less likely to be

54

depressed than those without insurance regardless of age or the other predictors. Its

effect was not altered by the addition of chronic illness.

Table 15: Depression Rates According to Control Variable Subgroups.

 

Group (% Depressed) Chronic Illness No Illness
(n=2095) (n=839)

Young 7.8% 4.6%
Insured 10.2” 5.7
Uninsured 9.3“ 4.6

Old 4.9” 2.2
Insured 4.9 2.7
Uninsured 4.9 00

 

*p<.01 for illness status, ”p<.001

Table 16 presents depression rates according to the independent variables.
Personal agency as measured by perceived control was a signiﬁcant independent
predictor of depression (p< .01). Low control was associated with higher depression
regardless of illness status. The addition of perceived control to the model analysis
eliminated the signiﬁcant effect seen for chronic illness. Perceived control modiﬁed the
magnitude and signiﬁcance of chronic illness such that chronic illness was no longer a
predictor of depression in the presence of control. Correlation analysis revealed a
statistically signiﬁcant, negative association between chronic illness and perceived
control (r= -.18, p<.01). Low perceived control was associated with the presence of
chronic illness. Therefore, control was shown to mediate the effect between chronic
illness and depression scores. The secondary hypothesis that personal agency mediates

the eﬂect of chronic illness on depression was supported by this analysis.

55

Social support was also a highly signiﬁcant predictor of depression. The presence of a
supportive partner was a protective factor in depression. Higher rated support was
signiﬁcantly associated with less depression (p< .01). Respondents with a very
supportive partner were less likely to be depressed and this effect remained highly
signiﬁcant in the ﬁnal model.

The protective effect of insm'ance coverage diminished somewhat with the
addition of support but remained signiﬁcant (p< .05). As the quality of the support
increased, the importance of insurance as protection from depression diminished
Support, entered the model and remained throughout the analysis, a highly signiﬁcant
predictor of depression (p< .01). Greater support equated with less depression.

Protection from depression may be more conditional on personal agency (as
measured by perceived control) and the presence of a supportive partner rather than on
age per say. Age demonstrated a signiﬁcant negative correlation with perceived control
(r= -.104, p< .01). Older respondents were more likely to have lower control. In
addition, those with low control were more likely to be depressed than those with high
control (r= -.237, p< .001).

The presence of a supportive partner was associated independently with the
absence of depression and so having a supportive partner protects against depression
(p< .001). Age did not correlate with the quality of support (r== -.023, [F .255).
However, perceived control did correlate signiﬁcantly with support (r= .201, p<.001).
Higher perceived control was associated with greater social support. The combination
of higher control and a supportive partner was signiﬁcantly associated with a lower

likelihood of depression even among those with chronic illness.

56

Sample size was diminished by more than half in the ﬁnal analysis, dropping from 2674
cases to only 813 in the ﬁnal model. In general, such a signiﬁcant drop in number of
cases could cause a failure to reach statistical signiﬁcance due to a lack of power. The
introduction of support measures led to a drop in sample size of 729 cases and the
introduction of perceived discrimination led to an additional drop of 1 132 cases.
However, the drop in power is not a major factor in these analyses because the
signiﬁcance level of each independent variable remained consistent throughout each
step of the regression model. This lack of impact on estimates was similar for each

dependent variable of depression, anxiety, and CAM use.

 

% Depressed

18
16 +
14 ~ I
12 w +Chronlc Ill Insured
10 . +No Ill Insured

8 . ' I Chronic Ill No lnsur
a . *No 111 No insurance

4 ,

2 -

0 r r
Young Older

 

 

 

 

 

 

 

 

FIGURE 4: Interaction Between Illness and Age.
Although there was an initial main effect of illness on depression, it disappeared
with the addition of perceived control. Chronic illness may be associated with low
control. Correlation analysis revealed a statistically signiﬁcant, negative association

between chronic illness and perceived control (r= -.18, p<.01). Low control was

57

associated with higher depression regardless of illness status. The addition of perceived
control to the model analysis eliminated the signiﬁcant effect seen for chronic illness

(Table 14). Therefore, control was shown to mediate the effect between chronic illness

and depression scores.

Table 16: Depression Rates According to Independent Variable

 

 

Subgroups.
Group (% Depressed) Chronic Illness No Illness
(n=20951 ln=839J
Overall 7.8" 4.6
High Control 5.7" 3.6
High Support 4.2" 2.8
Low Support 11.5 00
Low Control 11.7 7.1
High Support 9.8 5.8
Low Support 11.3 10.0
*p<.01, "p<.001

Social support was also a highly signiﬁcant predictor of depression. The presence
of a supportive partner was a protective factor in depression. Respondents with a very

supportive partner were less likely to be depressed and this effect remained highly

signiﬁcant in the ﬁnal model (Table 16).

58

% Depressed
ts

 

 

 

 

 

 

 

16 a
1‘ +Chron1c illligh
' control
‘2 ‘ Noilll'lgh control
10 -l
8 - +Chronlo lllLow
8 ~ control
4 , -O-Noill Low control
2 q
0

 

LOW Support “9h 399110"

FIGURE 5: Interaction of Illness and Support on Depression.

The experience of discrimination related to medical care services was not at all
associated with depression. However, less than 2% of the sample reported having
experienced discrimination speciﬁc to medical services one or two times in their lifetime
and thus has very little variation in measurement for the analysis. Therefore, the effect
may be present but not visible in this model. Its addition to the model did strengthen the
effect of age on depression. In the presence of discrimination, older age was even more
protective against depression.

Discrimination as measured by perceived interference in one’s life was a
signiﬁcant, independent predictor of depression (p= .05). The more that discrimination
was reported to interfere with life, the more likely depression was present regardless of
other predictors. Respondents older than 50 years who report discrimination were no
more likely to be depressed than those never experiencing discrimination (5.7% versus
5.6%). The younger group, 50 years or less, who have experienced discrimination report

the most depression (10.2%). Even when controlling for, insurance, perceived control and

59

support discrimination remained associated with higher depression scores. Discrimination
had a main effect on depression in the presence of illness. Depression rates were
signiﬁcantly higher for those reporting discrimination than for those reporting no
discrimination among both the chronically ill and non ill groups (Figure 6). The presence
of chronic illness magniﬁed the difference (Table 17 ). Depression rates were highest for
those with a chronic illness and who reported experiencing discrimination (p< .05). The
secondary hypothesis that discrimination moderates the risk of depression by chronic

illness was supported by the measure of perceived discrimination.

TABLE 17: Depression Rates among Illness and Discrimination Subgroups.

 

 

Group (% Depressed) Chronic illness No illness
(n=2095) 1n=8391
7.8** 4.6
Discrimination (n=1286) 10.0 5.4“
No Discrimination (n=35) 8.0 00 "

 

*p< .01 for chronic illness status, “p<.001, " n=10.

Figure 6 presents the rates of depression according to illness status and subgroups
of those who reported any interferences or effect of discrimination on their life compared
to those reporting no interference. The number of those reporting the times discriminated

according to medical care services was too small to report by subgroups.

6O

% Depressed

1s
1s ‘
14 «-

12 ‘ —e—Chronlc Knees
1o - \ n=2095

3 ‘ +1110 illness III-839
5 .1
4 _
2 , n=10

o .
Uscrlmlnated Against Not Dseriminated

 

 

 

 

 

 

 

 

FIGURE 6: Relationship of Chronic Illness, Depression and Discrimination.

The ﬁnal model was signiﬁcant for two control variables and three of the
independent predictors, explaining 11% of the total variation in depression scores. This
model demonstrated that chronic illness is a signiﬁcant predictor of depression mediated
by perceived control and social support and moderated by discrimination.

The model was tested for generalized anxiety to determine whether these same
factors would predict anxiety. Table 18 presents the regression of generalized anxiety and
its predictor variables. Independent variables in the model failed to show an association
with generalized anxiety and together explained little of the variation in scores (1%).
Chronic illness was not a predictor of anxiety. Perceived control was a signiﬁcant, but not

independent, predictor of anxiety (p< .01).

61

Table 18: Regression Analysis for Predictors on Anxiety.

 

 

Demog. Chronic Control Support Discrml Discrm2
Gender .084 .084 .076 .083 .083 .087
Age -.002 -.002 -.003 -.003 -.003 -.003
Education -.049 -.048 -.017 -.018 -.018 -.020
Employed -.018 -.018 -.014 -.015 -.015 -.015
Income -.016 -.015 -.004 -.004 -.004 -.002
Any kids -.022 -.022 -.046 -.036 -.036 -.037
Insurance -.1 12 -.1 12 -.093 -.094 -.094 -.091
Race -.011 -.011 .008 .011 .011 -.030
Chronic ill .026 -.007 -.013 -.013 -.007
Perceived -.093 " -. 100" -. 100“ -.092
Control -.071 -.071 -.07 8
Social Support .000 .000
Med .077“
Discrimination
Gen .810
Discrimination .04

.331 .316 .736 .910 .909 813
Constant .01 .01 .02 .03 .03
R2 = 2674 2674 2667 1945 1867
N

 

*p< .05, "p< .01 (Unstandardized, beta coefﬁcients reported)

However the addition of perceived discrimination eliminated the effect of
perceived control. In the ﬁnal model, only discrimination was signiﬁcantly associated
with anxiety (p<.05). Respondents reporting higher interference due to discrimination
reported more anxiety than those with none due to discrimination. The ﬁnal model
explained little total variation in anxiety scores (R2 = 4%).

The lack of an association may have more to do with speciﬁcity of measurement.
Anxiety measurement may be less speciﬁc than depression and if the chronic group
consists of illnesses that are more acute tlmn long lasting, any effect would be hidden.
This is explored in a later section that differentiates the chronic illness group as only

those conditions that are clearly persistent and long term compared to the healthy group.

62

Health Behavior Outcome Analysis

Table 19 presents the regression of predictors on alternative therapy use.

Table 19: Regression Analysis for Predictors on Alternative Therapy

 

 

Use.
Demog. Chronic Control Supmrrt Discrml Discrm2
Gender .349" .340“ .344“ .353" .358“ .362"
Age -.001 -.004 -.003 -.003 -.003 -.003
Education .488” .504" .490" .489” .482" .479"
Employed .006 .007 .006 .005 .006 .007
Income .069 .078 .073 .074 .077 .080
Any kids -. 105 -.103 -.092 -.081 -.088 -.089
Insurance .073 .083 .074 .072 .072 .075
Race -.080 -.078 -.086 -.083 -.098 -.144
Chronic ill .440” .455“ .447" .440" .446"
Perceived .042 .034 .031 .040
Control -.089 -.086 -.095
Social Support .019 .019
Med .086
Discrimination
Gen -.362
Discrimination .05
-.032 -.297 -.489 -.271 -.250 813
Constant .03 .03 .03 .04 .04
R’ = 2674 2674 2667 1945 1867
N

 

*p< .05, "p< .01 (Unstandardized, beta coefﬁcients reported)

Gender was a signiﬁcant, independent predictor of alternative therapy use (p<
.05). Females were signiﬁcantly more likely to report use of CAM and to report using
more types of CAM than males. This remained a signiﬁcant effect throughout the model.
Education was also a signiﬁcant, independent predictor of CAM use (Table 19). Those
with higher education were more likely to use CAM (Figure 6). The effect remained

strong and consistent throughout the inclusion of additional variables (p< .01).

63

TABLE 20: Alternative Therapy Rates of Use for Illness and Discrimination
Status.

 

 

 

Group % use CAM N
Chronic illness 55.2 2095
No illness 39.1 839
Females 57.2 1370
Males 44.9 1564
5_ HS education 43.3 1081
> HS education 54.9 1851
Discrimination 58.2 1286
No Discrimination 48.6 35
Chronic illness No illness
55.2 39.1
Female 59.4 45.0
3 HS education 53.0 35.3
> HS education 65 .8 54.7
Male 48.2 3 1.8
3 HS education 41.4 28.5
> HS education 55.0 35.0
* p<.01, "p<.001
%Using CAM
100
go .

 

so . -e—Chronlc Ill
Males

7o «

60 . +Chronlc 111

so . f- Females

40 _ / +Healthy Males

Healthy Females

 

 

 

 

 

o .
<- HS education > H3 education

 

FIGURE 7: Alternative Therapy Use According to Illness Status and Demographics.

An interaction occurred between illness status, gender and education (Figure 7).
Among the lower educated, use of CAM was ordered progressively from ill females, to ill
males, to healthy females, and ﬁnally healthy males. However, among the higher
educated, healthy females were not less likely to use CAM than ill males. Among the
higher educated, the tendency for females to use CAM is so great that they are as likely to
use alternative care as ill males.

Chronic illness was a signiﬁcant, independent predictor of CAM. Those managing
with a chronic illness were signiﬁcantly more likely to use CAM (p< .01). The effect
remained highly signiﬁcant throughout the analysis modeling. The primary hypothesis
regarding CAM was supported by these findings.

However, the addition of perceived control or social support did not alter the
effect. There was a positive relationship between control and CAM. Respondents with
higher control used more CAM but this was a non signiﬁcant effect. There was a negative
relationship between support and CAM. Respondents with little social support were more
likely to use CAM but the relationship was not signiﬁcant. There was a weak association
between discrimination and use of alternative medicines that was non-signiﬁcant in the
regression model (Table 21 and Figure 7). This may be due to its interaction with chronic
illness (Figure 7). There was no difference in CAM use among those not experiencing
discrimination in life. However, among respondents reporting discrimination, those with
illness were more likely to have used CAM than those who were healthy. The ﬁnal model
explained 5% of the total variation in CAM use frequency. The secondary hypothesis

regarding CAM and discrimination was supported by this model.

65

% Using CAM

 

100
90 ..
80 ..
70 -

so . \ +cnronlc Illness

50 q A 4 n-2095

‘0 - +No Illness nil-839

so ~

20 -

10 -
o .

Discriminant: No Discrimination (n= 35)

 

 

 

 

 

 

 

FIGURE 8: Alternative Therapy Use According to Illness Status and Discrimination.

Chronic Illness Subgroup Analysis

This regrouping eliminated 700 respondents from the chronic list who were
shifted to the comparison group. The resulting subgroup of serious chronic illnesses
included 1562 (53.2%) with one or more chronic illnesses and 1372 (46.8%) without
(Total = 2934). The regression analyses for the subgroup of serious chronic illnesses did
not alter the ﬁndings for depression outcome for all but the effect of chronic illness
(Table 21). Results were nearly identical to the original analysis. Other cut-off points for
grouping respondents according to listed conditions considered chronic or not resulted in

similar ﬁndings.

66

Table 21: Regression Analysis of Outcomes According to Selected

Serious Chronic Conditions.

 

 

 

Mr Depression Anxietv CAM
Gender (0=male) .061 .083 .311‘
Age (years) -.005"‘ -.002 -.004
Education (0= 5 HS) -.022 -.047 .308"
Employed (0=fulltime) -.012 -.015 -.004
Income (1= < $10,000) .012 -.007 .042
Any kids (0=none home) -.017 -.025 -.079
Insurance (0=no ins) -.155‘"" -.092 .013
Race (0=White) -.021 -.017 -.083
Chronic ill (0=no illness) .071 .057 .420”
Constant 3.16 .353 -.037
R2 = .03 .01 .04

n 2674 2674 2674

 

*p< .05, ”p< .0] (Unstandardized, beta coefﬁcients reported)

Chronic illness as measured as Global physical health was a signiﬁcant predictor

for all three outcomes in this initial analysis (Table 22).

Table 22: Regression Analysis of Outcomes According to Global

 

 

Physical Health.

.F_actor Depression Anxietv CAM
Gender (0=male) .087“ . 100 .363”
Age (years) -.005* -.002 -.004
Education (0= 3 HS) -.044 -.030 .520"
Employment (0=fulltime) -.019 -.020 .009
Income (1= < $10,000) .015 -.009 .089
Any kids (0=none home) .021 -.001 -.064
Insurance (0=no ins) -. 165" -. 109 .083
Race (0=White) -.016 -.010 -.103
Global health (1=poor) -.075** -.068* -.176"
Constant 3.23 .423 .400
R2 = .04 .02 .03

n 2934 2934 2934

 

’p< .05, "p< .01 (Unstandardized, beta coefficients reported)

67

Poor perceived health was signiﬁcantly associated with higher levels of
depression, of anxiety, and of more frequent use of complementary alternative medicines
(each p< .001). Depression was regressed on control variables and the independent
factors in the six-block model and reanalyzed.

Table 23 presents results for depression. Gender was initially a signiﬁcant but not
independent predictor of depression. Its effect disappeared with the introduction of social
support (Block 4). Females were more depressed than males until the quality of support
available to them was controlled for. Females were than no more likely to be depressed if
a supportive partner was available. Insurance remained throughout the six-model analysis
a highly signiﬁcant, independent predictor of depression. (p< .01). Respondents with
insurance of any type were less depressed than those without insurance at the time of the
survey. Perceived health was a signiﬁcant predictor of depression. Respondents with
better perceived physical health were less depressed than those with poor perceived
health (p < .01).

The magnitude of the effect weakened somewhat with the introduction of control
but remained a signiﬁcant predictor of depression at p< .05). Control was a highly
signiﬁcant, independent predictor of depression (p< .01). Social support was also a
highly signiﬁcant, independent predictor of depression (p< .01). As the reported quality

of support increased the likelihood of depression decreased.

68

Table 23: Regression Analysis for Depression on Predictors according
to Global Physical Health.

 

 

Easter Demog. Chrong’ Qontrol Suppgrt Discrml Discrm2
Gender .077 .087" .081* .068 .069 .071
Ages -.001 -001 -.001 -.003 -.003 -.003
Education -.060 -.044 -.007 -.005 -.006 -.008
Employment -.019 -.019 -.018 -.017 -.017 -.016
Income .006 .015 .024 .022 .023 .024
Any kids .014 .021 .005 -.011 -.012 -.013
Insurance -.174"“" -.165" -.149"”" -.148** -.148"”" -.146"
Race -.012 -.016 .012 .009 .005 -.021
Global health -.075""" -.048"' -.043"' -.043* -.042"'
Perceived Control -.109“ -.098” -.098” -.094”
Social Support .135" .135" .130"
Med Discrimination .004 .004
Gen Discrimination .049“
Constant 3.04 3.23 3.55 3.21 3.21 3.15

R’ = .02 .04 .08 .09 .09 .10

n 2934 2674 2667 1945 1867 813

 

*p< .05, "p< .01 (Unstandardized, beta coefﬁcients reported)

Discrimination speciﬁc to the receipt of medical services was not signiﬁcant.
However, discrimination reported to interfere with life was signiﬁcantly associated with
more depression (p< .05). The ﬁnal model demonstrated that the lack of insurance, low
control, and poor support are signiﬁcant predictors of depression. In addition, poor
physical health and the experience of discrimination are independent predictors of
depression. This analysis supported the primary hypothesis chronic illness increases the
risk of depression.

Anxiety was regressed on control variables and the independent factors in six

blocks and reanalyzed. Table 24 presents results for generalized anxiety. None of the

69

control variables were associated with anxiety. Perceived health was initially a signiﬁcant

predictor of anxiety but was not independent of personal agency.

Table 24: Regression Analyses for Anxiety on Predictors according to
Perceived Physical Health.

 

 

[actor Dempg. Chronic Control Support Discrml Discrm2
Gender .090 .100 .095 . 103 .102 . 106
Age -.001 -.001 -.002 -.001 -.003 -.003
Education -.045 -.030 -.004 -.006 -.005 -.008
Employment -.021 -.020 -.020 -.020 -.020 -.020
Income -.017 -.009 -.002 -.001 -.001 .001
Any kids -.007 -.001 -.013 -.004 -.004 -.005
Insurance -.1 17 -.109 -.098 -.098 -.098 -.096
Race -.007 -.010 .010 .012 .012 -.027
Global health -.068"' -.048 -.050 -.050 -.048
Perceived Control -.078** -.084” -.084” -.078"""
Social Support .076 .076 .084
Med Discrimination .000 .000
Gen Discrimination .074“
Constant .247 .423 .648 842 .841 .747
R2 = .01 .02 .03 .03 .03 .04

n 2934 2674 2667 1945 1867 813

 

*p< .05, "p< .01 (Unstandardized, beta coefﬁcients reported)

The addition of perceived control eliminated the effect seen with the global
measure of health. Perceived control remained throughout the analysis a signiﬁcant,
independent predictor of generalized anxiety (p< .01). Discrimination reported to
interfere with life was an independent, signiﬁcant predictor of anxiety (p< .05). Anxiety
was more common in respondents who reported greater interference in life due to

discrimination

7O

Use of complementary alternative medicine was regressed on control variables
and the independent factors in six blocks and reanalyzed. Table 25 presents results for

CAM.

Table 25: Regression Analyses for CAM on Predictors according to
Global Physical Health.

 

 

Eager Dempg, Chronic Contrpl Support Discrml Discrm2
Gender .339‘ .363" .367“ .382" .387" .390”
Age -.002 -.002 -.001 -.001 -.001 -.002
Education .481" .520“ .500" .497" .490“ .487"
Employment .008 .009 .008 .007 .009 .009
Income .068 .089 .084 .086 .090 .092
Any kids -.079 -.064 -.055 -.037 -.044 -.045
Insurance .062 .083 .074 .073 .072 .074
Race -.096 -.103 -.119 -.114 -.131 -.l66
Global health -.176""" -.191""" -.196" -.196"“" -.193"”"
Perceived Control .060 .046 .044 .050
Social Support .153 .149 .156
Med Discrimination .021 .022
Gen Discrimination .066
Constant -.054 .400 .223 .616 .627 .543

R2 = .02 .03 .03 .03 .04 .04

n 2934 2674 2667 1945 1867 813

 

‘p< .05, “p< .01 (Unstandardized, beta coefﬁcients reported)

Gender was a signiﬁcant, independent predictor of CAM use. Females
were more likely to report use of CAM than males (p< .05). This effect was enhanced
further with the addition of perceived physical health. Perceived health was a signiﬁcant,
independent predictor of CAM (p< .01). Respondents with poorer health were more
likely to use CAM. Education was also a signiﬁcant, independent predictor of CAM use
(p< .01). Higher education was associated with greater use of CAM. An interaction
occurred between illness status, gender and education (Figure 6). Among the lower

educated, use of CAM was ordered progressively from ill females, to ill males, to healthy

71

females, and ﬁnally healthy males. However, among the higher educated, healthy females
were not less likely to use CAM than ill males. Among the higher educated, the tendency
for females to use CAM is so great that they are as or more likely to use it than are ill
males. Discrimination was not associated with utilization of CAM. The ﬁnal model
demonstrated that higher educated females with poorer physical health were the most

likely to utilize CAM (p< .01).

Table 26: Zero Order Correlations for Mental Health Outcome.

 

 

 

 

Dependent Variable: Depression Chronic Illness
Indpppndent Variables r-Coefﬁcient r-Coefﬁcient

Gender .056 .027
Age -.104" -.092**
Race .001 -.021
Education -.054 .196"
Employment -.045 -.074"'
Income -.005 . 166"
Insurance -. 104” .090"
Children home .029 .049
Chronic Illness -.133""" 1.00
Control -.237” .255”
Support -.181"”" .081"
Discriminationl .014 .007
DiscriminationZ .109" -.094"""

 

* p< .05, .. p<.01

In the zero-order correlation analysis for depression, age, insurance, illness,
control, support, and perceived discrimination were signiﬁcantly associated with
depression (Table 26 & Table 28). Younger age, less insurance, chronic illness, low

control, less support, and more discrimination were signiﬁcantly associated with higher

72

depression scores. In the zero-order correlation analysis for chronic illness, older age, less
education, employment, lower income, less insurance, lower control, less support, and
more discrimination were signiﬁcantly associated with having a chronic illness. The
similar patterns in these correlations indicate that depression itself can manifest as a
chronic insult on health that is very much associated with physical chronic illness.
Because chronic illness is so highly correlated with lower socio-economic status, its
small contribution to the prediction of depression may be accounted for by its correlation

with the previous predictors and the correlation between depression and chronic illness.

Table 27: Zero Order Correlations for Complimentary Alternative
Medicine Use.

 

 

Dependent Variable: CAM Use Chronic Illness
Indemndent Variables Correlation Correlation
Coefﬁcient Coefﬁcient
Gender .075* .027
Age -.013 -.092"
Race -.031 -.021
Education .158" .196 *"
Employment -.005 -.074*
Income .038 . 166"
Insurance .038 .090"
Children home -.023 .049
Chronic Illness -.047 1.00
Control .034 .255"
Support .034 .081“
Discrimination] .041 .007
DiscriminationZ .007 -.094"‘*
" p< .05, ” p<.01

In the zero-order correlation analysis for CAM, gender and education were
signiﬁcantly associated with CAM use (Table 27). Chronic illness approached, but did

not reach, a signiﬁcant association (p=.08). Females, who are highly educated, were most

73

likely to use CAM. However, highly educated females are also more likely to be in better
health, have higher perceived control, and greater support. The zero-order correlation
analysis for chronic illness demonstrated signiﬁcant correlations with older age, less
education, unemployment, less income, less insurance, lower control, less support and
more discrimination. Although the correlation between gender and education is highly
signiﬁcant for CAM use, chronic illness remained a signiﬁcant predictor in the multiple
regression, thus indicating it as a signiﬁcant predictor of CAM use.

The interaction term Discr1*illness was not statistically signiﬁcant for depression
(p=.92). The interaction term for Discr2*illness was not statistically signiﬁcant for
depression (p=.16). The negative direction of the association between the interaction
terms and depression suggests that the effect of illness on depression is lower for those
who experience more discrimination. Although this was not signiﬁcant statistically, the
amount of explained variation increased from 0.11 to 0.14 with the ﬁnal interaction term
included (Table 31). No signiﬁcant effect was observed for the model on anxiety (Table
32). The interaction terms were not statistically signiﬁcant (p> .90 for both Discr1*illness

and Discr2*illness). The ﬁnal amount of variation explained remained at .04.

There was no statistically signiﬁcant effect of the interaction term on the predictor
of alternative medicine use (p> .90 for both Discr1*illness and Discr2*illness).The

amount of variation explained by the model increased slightly from .04 to .046 (Table

25).

74

CHAPTER FOUR

DISCUSSION

The present study examined the relationship between chronic illness and
alternative medicine utilization and psychological distress. Two primary hypotheses
based on six possible models for the relationships with key independent variables were
examined within the domains of illness and discrimination. Findings indicated that the
presence of chronic illness is associated with mental distress and depression. The effect
of illness on mental health is not independent of personal agency and social support.
Furthermore, the experience of discrimination perceived to interfere with life is
associated with greater depression rates, particularly among those managing a chronic
illness use and provides support of a moderating affect of discrimination on mental
wellbeing. In addition, the current analysis demonstrated a highly signiﬁcant main effect
of chronic illness for complementary alternative medicine (CAM).

Hypothesis #1 : There is a significant positive relationship between mental
distress/depression and the presence of a chronic illness. Findings from the current
analysis support this hypothesis. Chronic illness was an independent predictor of
depression and anxiety even in the presence of control variables for age and insurance.
Respondents with a chronic illness had higher rates of depression than those reporting no
illness regardless of age or insurance status. The presence of a chronic illness imposes
chronic stressors unique to the daily stresses of other role strains. An illness that is
ongoing requires adjustment to a different lifestyle and to new health-related behaviors.

There are stressors involved with the illness, as well as strains related to the new lifestyle

75

(Brown and Harris 1978). In addition, the pressure from having to conform to different
rules and regulations given a desire to comply with improved outcomes is difﬁcult. As
demonstrated by Pearlin, et al., these demands cause initial and proliferated stress that
can lead to mental strain (1997). Such chronic stressors have been shown to uniquely
impact negatively on mental health when measured separately from the effect of daily
hassles (Serido, et a1 2004). Serido et al., provided support for chronic home stressors
functioning as a moderating factor on the relationship between daily hassles and
psychological distress (2004). They also utilized the MIDUS study and demonstrated that
chronic stressors themselves increase psychological distress and depression. Because
chronic illness is a chronic stress as conceptualized by Wheaton (1984), their ﬁndings
support the current analysis that chronic illness increases mental distress and/or
depression.

Interestingly, it was the respondents 50 years of age or younger who had
signiﬁcantly higher rates of depression independent of health insurance coverage. The
protective effect of age held for those with chronic illness and for those who were
healthy. The worst off by far were younger respondents (_<_ 50 years of age) who were
managing a chronic illness with no health insurance coverage. Their rates of depression
were over 14% (Figure 4). Other current evidence in the psychological literature has
challenged the previous assumption that older people are more depressed due to their age
(Gottlieb SS, et al. 2004; Schieman S, et al. 2002).

The effect of chronic illness on depression rates was strongly mediated by self
efﬁcacy, as measured in this analysis by perceived control, and was moderated by social

support. As previous research has established and the current analysis supports, low

76

perceived control was associated with higher depression rates even among healthy
respondents. Rapley and Fruin (1999) summarized the juxtaposition of general and
regimen-speciﬁc self efficacy in chronic illness. For those with a chronic illness, changes
in lifestyle are not only difficult but are necessary. These major changes may involve role
adjustments, learning new behaviors, and modifying one’s lifestyle. The ease with which
changes occur depends on the person’s self efﬁcacy beliefs and outcome expectations.
People carryout multiple regimen- and task-speciﬁc behavior in order to manage the
chronic illness. The ability to comply with the new regimen and tasks is directly
dependent on self efﬁcacy. A belief in oneself to tackle new barriers and accomplish new
goals mediates the strain that an illness presents to ones psyche. The greater the self
efﬁcacy, the more likely one will perform behaviors consistent with therapeutic
compliance and the more likely they will achieve positive outcomes, either in managing
the illness or preventing future comorbidities. The current analysis demonstrated that
depression outcomes related to chronic illness are highly mediated but not eliminated by
self efﬁcacy. Chronically ill respondents, even those high in self efﬁcacy, still had higher
depression rates than the 4% seen in the general population (Table 16).

An interaction in illness-related depression rates occurred between social support
and perceived control in the current analysis. Social support did not impact depression
rates for the chronically ill group with low control (Figure 4). However, for those
reporting high perceived control who were managing a chronic illness, a depression rate
of 12% dropped to less than 4% in the presence of support. This suggests an interaction
between support and self efﬁcacy and illness. The ease with which lifestyle adjustments

occur depends on the person’s efﬁcacy but also on their outcome expectations (Rapley

77

and Fruin 1999). 111 respondents with low self efﬁcacy may be unable to harness the
beneﬁt that is available from a highly supportive spouse due to low expectations of the
outcome whereas those with high self efﬁcacy have the conﬁdence to draw upon the
resources provided by a supportive partner.

The group with the highest depression rates were those with a chronic illness who
reported a non-supportive partner and this effect held for both those with low perceived
control (11% depressed) and with high perceived control (12% depressed).

The secondary hypothesis #1 : The relationship between mental
distress/depression to the presence of chronic illness is significantly magnified by the
experience of discrimination in general and by discrimination in receipt of medical
services speciﬁcally. Findings support this hypothesis overall. Depression rates were
signiﬁcantly higher for those reporting that the experience of discrimination interfered in
their life. Those in whom discrimination was perceived to interfere had higher rates of
depression (7%) than those in whom discrimination did not interfere (4%). The true size
of the effect of discrimination regarding illness was difﬁcult to conclude because the
group experiencing no discrimination who where also healthy was extremely small, only
10 respondents. However, depression rates were higher for the chronically ill regardless
of discrimination status and furthermore, among those with a chronic illness the presence
of discrimination increased depression rates (8% versus 10%) (Figure 5).

The ﬁnal model demonstrated a non-signiﬁcant effect of chronic illness on mental
health mediated by perceived control and social support that is clearly magniﬁed by the

experience of discrimination in life.

78

The ﬁndings of the ﬁnal regression model for the outcome of mental wellbeing
indicated no statistically signiﬁcant effect for an interaction of discrimination and chronic
illness on depression. Although the stress process theory supports a role for the
interaction of discrimination and chronic illness through increased encounters with the
healthcare system and increased opportunities for discrimination, the current analysis
shows no clear evidence of this. No effect modiﬁcation was observed with the
interactions of discrimination and chronic illness for anxiety. It is possible that it takes a
signiﬁcant amount of discrimination, even for those dealing with a chronic illness, to be
pushed to the limit regarding stress from such discrimination and thus it would not show
up as mental strain until it reached a severe intensity.

Primary hypothesis #2: There is a signiﬁcant positive relationship between
chronic illness and the utilization of complementary alternative medicine (CAM). This
hypothesis was also supported by the analytic ﬁndings. Chronic illness was a signiﬁcant
predictor of CAM use. Respondents who were managing with a chronic illness were
signiﬁcantly more likely to report alternative methods of care. This effect was
independent of, but not entirely unrelated to, two other factors that were associated with
CAM use. Gender and educational attainment were also signiﬁcant predictors of CAM.

The willingness to try an alternative means for obtaining healthcare was
distinctly a female tendency. The gender effect was independent of education. Women
are more likely then men to report CAM use. Even in health, women were more likely
than men to use CAM. This effect supports what we know for health behaviors and
medical care in general. Women are signiﬁcantly more likely to utilize primary care

services (Bertakis KD, et al. 2000), to make outpatient visits more often (Bertakis KD, et

79

(Bertakis KD, et al. 2000; Burns MJ, et al. 2001), and to be prescribed multiple classes of
medications (Roe CM, et a1. 2002) than men, who tend to avoid or forestall healthcare
issues.

The use of CAM was strongly inﬂuenced by educational attainment. Respondents
with some college or a college degree were much more likely to utilize CAM than those
with a high school degree or less. This is supported by research on CAM conducted in
many other populations. Among veterans (Smola S, et al. 2001), among the elderly
(Muhajarine N, et al. 2000), and among patients hospitalized for surgery (Cherniack EP,
et al. 2001) the higher educated are more likely to use alternative methods of care. This
may be due to a greater knowledge of alternatives that are available, a larger social
network that provides opportunities to alternatives, or due to more resources attained by
the highly educated. All three of these factors may be operating for the higher educated.
However, these factors do not explain the gender effect of CAM use. Women in society
do not necessarily have more resources, a higher education, and greater knowledge of
health than men. In the current analysis, even healthy females with low educational
attainment were more likely to use CAM than males (Figure 6).

The three signiﬁcant predictors (gender, education, illness) were not entirely
independent of each other in their effect on CAM. Women, in general, are more likely to
suffer chronic illness and consequently there would be more women seeking care.
Although women in general have higher rates of chronic illness, the gender effect on
CAM use was independent of illness status in the current analysis. Since women in
general are more likely to interact with the health care profession it is not surprising that

they are more likely to seek and use CAM.

80

The effect of illness on CAM use was modiﬁed by education. Among the lower educated
group, ill .males were more likely to use CAM. Yet, among the higher educated, healthy
females were just as likely to use CAM as ill males. Illness was a stronger predictor of
CAM use than gender among those with lower education. This interaction may be
highlighting the relationship of resources in health seeking behavior. For those with few
resources, illness necessitates seeking and use alternatives even among males who prefer
to avoid encounters with the health care system. Those who have higher education also
have access to more resources and more importantly access to extra resources for
prevention. The gender effect is magniﬁed in this group because females are the most
likely to visit a health care professional and with resources, are more likely to seek out
healthcare services to prevent illness. There was no interaction between discrimination
and CAM use (Table 33). The results do show that discrimination in medical services
does increase depression, anxiety and CAM —even though there is no interaction effect.
The primary hypotheses and secondary hypotheses #2 and #4 were signiﬁcant and

consistent. However, hypotheses #1 and #3 were not supported for an interaction.

Limitations

Some subgroups suffered from small sample sizes, although the overall sample of
4000 respondents was large. A greater concern is the small amount of variation in each
dependent variable that the models were able to explain. The low values for the
coefﬁcients of variation in each model (10% and 4%) are an indication of much
measurement error or random variation or both. Measurement error likely played a role

in the analysis that measured chronic illness by inventory. The model’s ability to explain

81

the variation in depression more than doubled when chronic illness was measured using
global perceived health. However, even though it was strengthened, the ﬁnal model
explained only 10% of the variation in depression rates. This may indicate more random
variation and thus make it less likely that the ﬁndings would be duplicated. Although
low, coefﬁcients of variation are often disappointedly small in epidemiologic studies
(Friedman GD 2004). The consistent statistical signiﬁcance of the regression coefﬁcients
for the tested predictive factors makes a chance ﬁnding less likely.

The error rate was high regarding CAM use and this threatens the validity of the
ﬁndings that chronic illness is a predictor of CAM. However, chronic illness was
clearly a statistically signiﬁcant predictor of CAM use even in the presence of error. In
addition, whether measured by inventory or perceived health, it signiﬁcantly predicted
alternative medicine use. This may indicate that the analysis suffered less ﬁ'om
measurement error than from random error as evidenced by the large and statistically
signiﬁcant regression coefﬁcients. Therefore, the ﬁnding that chronic illness can be a
signiﬁcant predictor of CAM use is important.

This analysis does have limited generalizability because those not responding to
the mailed survey were signiﬁcantly different than survey respondents on several
demographic characteristics. These results may not apply to younger, single males
without a high school degree. Previous research has shown that respondents who
complete surveys differ on reported physical and mental health with higher scores for
each (Powers JR, et al. 2003). Among those included in the analysis, generalizability
may be limited regarding minority status. Only 13% of the 2934 respondents were a

minority and the impact of chronic illness on mental health may be under estimated for

82

them. Future analyses will need to be conducted on much larger nationally representative
samples in order to validate these ﬁndings.
Conclusions

Previous studies have established that chronic stress and its related stressors
impacts negatively on mental health (Brown & Harris 1978, Wheaton 1994).
Furthermore, chronic stress occurs not in isolation but in relation to a ‘web of
interconnectedness’ among multiple factors both acute and chronic (Pearlin 1999). The
broader social context of work and family environment and the roles established for
individuals can effect whether an illness precipitates or exacerbates mental wellbeing
(Wheaton 1994, Peralin 1999). For illness, not only do stressors occur within a complex
process, but any particular stressor within the web can lead sequentially to other stressors.
The presence of illness can lead to the emergence and the proliferation of other chronic
stressors (Peralin 1989) and the buildup of stressors manifests as anxiety and depression.

As important as the broader social context is in coping with an illness, other
situatioml factors contribute to the context within which one experiences stressors
(Wheaton 1996, Seredo et. al., 2004). Most recently, research on stress and
psychological health has demonstrated that chronic stressors may have a unique effect on
distress (Seredo et. al., 2004). The current analysis supports this thesis that chronic
stressors present an ongoing threat to the individual, the ever-present potential to erupt in
various ways that impact on daily life. Chronic illness is a primary chronic stress that not
only can lead to poor mental health but causes the emergence of other stressors as 7
chronic, such as discrimination when seeking ongoing health care. This study

demonstrated that chronic illness increases the likelihood of depression even when

83

controlling for other characteristics such as age, gender, and insurance. However, as
expected, the presence of personal agency and positive support mediated its effect. The
weak effect of chronic illness on depression may be due in large part to the impact of
secondary and proliferating stressors introduced by the ongoing illness that were not
accurately reﬂected by the survey’s measures. The use of a global measure of physical
health increased the association but it remained somewhat weak because of other factors
that may have had greater precision in their measurement. The presence of illness was no
longer signiﬁcant when perceived control and social support were included and these
variables may have had greater accuracy in their measurement because of previously
validated instruments that were used in MIDUS.

While there are associations between chronic stressors and daily hassles, they may
each have a unique effect on psychological distress at the aggregate and daily levels. .
Seredo, et. al., suggest that daily hassles, as a vast array of minor disruptions, can force
the individual to act acutely and thereby increase psychological distress distinct from the
chronic stress (2004). The current study supports their ﬁndings. Discrimination was a
distinct predictor of depression and anxiety independent of the effect from chronic
illness. Our study was unable to show whether this was indeed due to discrimination
speciﬁc to the receipt of health care services because of the small numbers involved.
However, it did demonstrate that perceived discrimination in general can lead to
increased levels of depression, especially for those with a chronic illness.

Health seeking behavior can also have an impact on mental wellbeing and can
account for some disparities in mental health outcomes. Health behaviors occur in the

broader social context within which the chronic stressor of illness emerges and access is a

84

key initiator to seeking care. Previous research has demonstrated that women seek health
care more often than men and individuals with unmet healthcare needs are more likely to
seek and use CAM (Sturm and Sherboume 2001). This may explain the current ﬁndings
that education modiﬁed the gender effect of illness on CAM use. Lower education
results in fewer resources and greater unmet need for health care. Under these
circumstances, an illness may more likely prompt even males to seek alternative care.
Healthcare seeking behavior may also be impacted by the experience of discrimination.
Discrimination interacted with illness in the current analysis suggesting greater unmet
need for healthcare services among those with a chronic illness.

Even though the current study lacked power in some analyses and may have
suffered from low speciﬁcity in some measures, suggested trends add to what is already
known in the mental health research. Chronic illness is a chronic stressor that can
increase depression. The regression analysis suggests a causal relationship because the
addition of illness was signiﬁcant even after controlling for known factors that increase
the risk of depression and when measured by global health, the effect remained
signiﬁcant after adding all other independent variables. Furthermore, the experience of
discrimination adds to the distress as either a unique chronic stressor or a factor
increasing daily hassles. Chronic illness is also an independent predictor of healthcare
seeking behavior and may reﬂect the greater unmet healthcare need among those
experiencing ongoing, long-term stresses.

Given these ﬁndings, an accurate measure of chronic illness that validly reﬂects
the chronic nature of living with a chronic illness needs to be developed and tested.

Future research may then be able to isolate, not only the effect of chronic illness on

85

mental health outcomes but discern the effects according to the proliferation of secondary
stressors on depression and anxiety. Such a measure would also assist future research in
identifying the effect of discrimination from ongoing encounters with the healthcare

system regarding the management of chronic illness.

86

APPENDIX A

LIST OF VARIABLES

87

 

 

 

 

 

 

 

 

 

 

 

TABLE 28: List Of Variables
Dem ra hics:
Number of respondents #
Households with children under 18 #
years of age
Gender M/F
Age Years
Race: Caucasian
Minority:
Black / African American
Native American /Eskimo
Asian or Paciﬁc Islander
Multiracial
Other
Marital status: Married
Not married:
Separated
Divorced
Widowed
Never married
Partnership: Currently living with someone in a steady,
marriage like relationship:
Y/ N
Children have: Number
Education: Up to High School
High School Graduate
Beyond High School:
1 — 2 yrs college
3 or more college
4 yr degree
Master degree
Advanced dgggee
Employment: Working now/self-employed
Not working:
Looking for work
unemployed
Laid off /leave temporary
Retired
Homemaker
Disabled mananently

 

 

 

 

 

88

 

 

 

 

 

 

 

 

 

 

Respondent Income: Annual: $00 - $9,999
$10,000-$19,999
(also Spouse / other adult income) $20,000-29,999
$30,000-39,999
$40,000-49,999
$50,000-$74,999
$75,000-$99,999
$100,000 and above
Perceived ﬁnancial status: 1. More money than need
Would you say you have more money, just 2, Just enough money
m or not enough to meet your family’s 3. Not enough money
8
Perceived ﬁnancial status: 1. Very difﬁcult
How difficult is it for you and your family to 2. Somewhat difﬁcult
Pay your monthly bills? 3. Not very difﬁcult
4. Not at all difﬁcult
Health Insurance: Private insurance w/ insurer
Private insurance w/ employer
Medicare
Medicaid
Champus/ other government insurance
Perceived physical health: 1. Poor
2. Fair
(also perceived mental health) 3. Good
4. Very Good
5. Excellent
Do you smoke cigarettes regularly Yes / No
new?
Usual Care: Yes / No
Do you have one particular place where you
usually get medical care?
Usual Care: Yes / No
Do you have one particular doctor who you
usually see?
Alternative Therapy Utilization: Acupuncture
Have you used any in the past 12 months to Biofeedback
treat, prevent, illness or to maintain health? Chiropractic
Energy healing
Exercise therapy
Herbal therapy
High dose mega vitamins
Homeopathy
Hypnosis
Imagery techniques
Message therapy
Prayer or spiritual practice
Meditation techniques

 

 

89

 

 

 

 

 

 

Special diets Other
Scales and Indexes
Depression Index:
Ever a time in last 12 months felt sad, blue, or Yes / NO / On medication
depressed for two weeks or more in a row?
In past 12 months do you worry more, less, or More / Less / About same
same as most people?
During past 30 days how much did you feel: All the time
3° 3“" ““1““3 °°“ld Most of the time
cheer you up? S f th -
Hopeless? (Prue 0 6 time
That everything is an thtIC Of the trme
effort? None of the time
All the time
During past 30 days how much did you feel: MOSt 0f the “.1116
Cheerful? Some of the trme
In good spirits? Little of the time
Extremely happy? None of the time
Calm and peaceful?
Satisﬁed?
Full of life?
Perceived Control Index:
Level of agreement with:
Strongly Agree
The demands of everyday life often get me Somewhat Agree
down. A Littl
In many ways I feel disappointed about my D??? :n e
achievements in life. , n 0w ,
Maintaining close relationships has been Dlsagree a thﬂe
difﬁcult and frustrating for me. Somewhat Disagree
I live life one day at a time and don’t really Strongly Disagree

think about the future.

I sometimes feel as if I’ve done all there is to
do in life.

I gave up trying to make big changes in my
life.

There is little I can do to change the
important things in my life.

I often feel helpless in dealing with the
problems of life.

Other people determine most of what I can
and cannot do.

What happens in my life is often beyond my
control.

 

 

90

 

 

 

There are many things that interfere with
what I want to do.

I have little control over the things that
happen to me.

There is really no way I can solve the
problems I have.

I sometimes feel I am being pushed around in
life.

 

 

 

 

 

Daily Stress/Hassles Index:
How often experienced each of the
following in the past year:
All the time
You have too many demands made on you. Most of the time
You have a lot of interruptions at home. Some times
Responsibilities at home reduce the errort can Rarely
devote to job. Never
Personal or family worries distract you when
at work.
Activities at home prevent you from getting
the sleep you need to do your job.
Stress at home makes you irritable at work.
Quality of Support Index:
Excellent
Would you describe your relationship Very good
as: Good
Fair
Poor
A lot
How much does your partner care about you? Some
Howfmgch does he/she understand the way A little
0U .
I-Iow :uch does he/she appreciate you? N“ at all
How rrnlch can you rely on him/her for help
with a serious problem?
Discrimination Index:
(due to race, gender, age, disability,
other) A lot
Some
To what extent does discrimination: A little
interfere with your life? Rarely
Not at all

 

 

91

 

 

 

How many times in life have you Number
been discriminated against by being
denied or provided inferior medical
care?

 

 

92

 

APPENDIX B

SCALES AND INDEX MEASURES

93

Depression Index

1. During the past 12 months was there ever a time when you felt sad, blue, or depressed for
two weeks or more in a row? Yes / No

2. People differ a lot in how much they worry. Considering how things have been going in
your life over the past 12 months, do you worry:

More, Less, About the same, or Not at all

3. During the past 30 days, how much of the time did you feel so sad that nothing could
cheer you up?

All the time, most, some, a little, or None of the time

4. During the past 30 days, how much of the time did you fell hopeless?
All the time, most, some, a little, or None of the time

5. During the past 30 days, how much of the time did you feel that everything was an effort?
All the time, most, some, a little, or None of the time

6. During the past 30 days, how much of the time did you feel satisﬁed?
All the time, most, some, a little, or None of the time

7. During the past 30 days, how much of the time did you feel in good spirits?
All the time, most, some, a little, or None of the time

8. During the past 30 days, how much of the time did you feel extremely happy?
All the time, most, some, a little, or None of the time

9. During the past 30 days, how much of the time did you feel calm and peaceful?
All the time, most, some, a little, or None of the time

10. During the past 30 days, how much of the time did you feel satisﬁed?
All the time, most, some, a little, or None of the time

11. During the past 30 days, how much of the time did you feel full of life?

All the time, most, some, a little, or None of the time

94

Perceived Control Index

All item responses on a 7-item Likert Scale:

Agree Strongly / Somewhat / A Little / DK / A Little / Somewhat / Disagree Strongly

1.

2.

8.

9.

The demands of everyday life often get me down.

In many ways I feel disappointed about my achievements in life.

Maintaining close relationships has been difﬁcult and frustrating for me.

I live life one day at a time and don’t really think about the future.

I sometimes feel as if I’ve done all there is to do in life.

I gave up trying to make big improvements or changes in my life a long time ago.
There is little I can do to change the important things in my life.

I often feel helpless in dealing with the problems of life.

Other people determine most of what I can and cannot do.

10. What happens inymy life is often beyond my control.

11. There are many things that interfere with what I want to do.

12. I have little control over the things that happen to me.

13. There is really no way I can solve the problems I have.

14. I sometimes feel I am being pushed around in my life.

95

Alternative Therapies
Each item response was dichotomous (Yes or No) as to ever using it.

1. Have you ever used the following items to treat a physical health problem, to treat
an emotional or personal problem, to maintain or enlmnce your wellness, or to
prevent the onset of illness?

a. Acupuncture

b. Biofeedback

c. Chiropractic

(1. Energy healing

e. Exercise or

f. Herbal therapy

g. High doses of vitamins
h. Homeopathy

i. Hypnosis

j. Imagery techniques

k. Message therapy 1. Prayer or other spiritual practice
m. Relaxation or meditation techniques 11. Special diets
o. Spiritual healing by others p. Any other non-traditional remedy

2. Did you ever use the following items on your own without a prescription or in
doses beyond what was prescribed?
a. Sedatives, either barbiturates or sleeping pills
b. Tranquillizers or nerve pills

c. Amphetamines or other stimulants

96

APPENDD( C

Correlation Matrix Tables

97

Svd: wove.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Q85
5. :85
9.2.- 1.2.- teen—am
*2: .30.. ssvmr 35:00
No.- ......oo.. 8. 1.2.- v.35:
vow we. .13.- 8. mo. mEM
tea.- anN. mor mo. ﬁo. 1.2.- .9:
0; atom.- nem. vor mor 1mm; 5.- 2:85
3.2.- 1.3.. o; no: Loy ......mm. ......9. wow 9cm
vor tLN. ......wm. 8.. o; Low mo: vor 8.. .eoznm
No; we. 8.. vor ...8.- ...wor o; 1.3.- vor So. comm
1.2.- 1.8.- 1.3.- 3. mo: 3mm. 8.. 2.3.- og 1.3.- 1.3.- om<
32.- Loy mo. mo. 8. ......mm.- ......om. 3.- 3.- 1.3.- o._ co. becoO
3:. 1.3.- ...nvmr 1.2.- mo. 1.3.- 5.- 3.. 3.. co. 1.3.- oo. A: mini
_
N520 tandem 8:80 32:: mug .m5 .885 «ram 46m coax ow< €50 me

 

 

.=e_eeo.aon ..8 5.5a: .5329.an 9:83; "3 93am-

98

HQVQS. .mo.VQ*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.985

3. :85

......oﬁ .- vmo. toanzm

:2 .- {b.- 30. 1550

nor ......oor mo. 53.- mace—z

3.- 3. 1.3.- mo. mmor mug

IN.- IN. nor mo. 5. wmo. .mE

o; ......m.- ......m. 3.- 8.. ......mm.- wmo. 0885

1.2.- 1.00- o; 3.- For mm. .19. 5.- 9cm

mmor ....Lm. ......wm. nor o; two: *8; 3.- 1.2. .835

8.- v3. 53.- wmor 48.- two: o; 8.- 3.- mo: coax

$3.- 13.- 1.3.- Eo. mo: 1mm. 8.. *wor o; 3.- S.- ow<

13.- *8: Re. mmo. :o. 3mm: ......om. 3.- 3.- ti.- o; two. beﬁo

8. 3o. 3o. 53.- mmor wmo. wmo. 8.- IS. 8.. S.- awo. o; 320
NEED tandem 1:80 32:: meg .mE .Eooﬁ dam 46m 08% om< .680 $20

 

 

 

 

 

 

 

 

 

.2: 25:32 grain“: ESEEQEEeU ..eu its: zeta—0530 9.89.55 ”on 93:.

99

nomHmEEtomE 0030800 mm whine
8338 3302: Mo 3502 E 02358.:86 E 285+
€3.80»: 8:22.008 Son 6363050330 Sven... Hove...

 

 

 

 

00w mg m 3 $3 33 Son 30m 30m 2
3 .0 :0 :0 000 w0.0 50.0 m0.0 «0.0 -~
v.m m.m mam Wm Em 5m Cd mad E5300
0 5.- 985038me
gum—wee. T mo-mHBA- 305:... :85
33.0 33.0 ...Nm00 N 5me
30.0 30.0 30.0 . 30.0 0 $me
......§.- test ...-o:- ......§.. ...-3:- tandem
.2005.- ..zamoﬁ- .3..er 1.0:.- ...*0:.- .130- 30:00
000.0 $0.0 $0.0 $0.0 $0.0 53.0 $000 E 068.0
03.- 0m0.- 0m0.- «00.- 30.0 000.0 0N0.- 0N0.- comm
...Nmﬁr ..;er .Lmﬁ- ...mmﬁ- .32.- ...aer act-2.- ..;me 8:8sz
050.- :0.- Ko.- 03.- 000.- 0m0.- a _0.- 20.- meg
R00 80.0 300 R00 200 50.0 20.0 0 3.0 @885
m00.- m00.- m00.- moor m00.- 500.- NS.- NS.- ESoEEm
30.- 30.- mmor 30.- 0N0.- NNO.- $0.. $0.- :058300
......000.- ......000.- 3000.- ......coor .1000.- .Iboor *moor *voor om<
wmo. 3.0 03.0 30.0 08.0 30.0 N000 30.0 $050
$05: $05:
$85 ... Coma ~8me :85 toadsm 6.980 088:0 .moEoQ

 

 

 

 

 

 

 

 

 

 

0288000 no E32005 mo mambmcaa‘ coﬁmoawom ”mm 2an

100

50058086 0030200 0: Q0050
0003000 08:00:: 00 “5:000: :0 5005:0806 00 2005+
80:000.: 0500000000 800 630000053030 5.v:...... .00.V0...

 

 

 

 

 

 

 

 

 

 

 

 

000 0 ﬂ 0 0 ﬂ 0 500 H 0v0~ 500N V50N VO0N Z
V0 .00 5.0 00.0 00.0 00.0 N0.0 5.0 -~
50. 005. , 55. 30. N00. 0V0. 0Nv. 5VN. 0:80:00
50.- 0005:...N00EQ
50-m0V5. T 50-0-:0 M . T mm05018£ﬂ
...050.0 350.0 .3500 N $me
000.0 000.0 000.0 000.0 H 8me
V000 N00.0 .000.0 050.0 050.0 00009.0
......050.- ......050.- ......050.- ......V00.- ......V00.- 050.- 50:00
0v0.- 0.00.- 00.0.- 000.- 000.0 000- ...000.- E 2:030
0N0.- 0N0.- 5N0.- N5.0 N 5.0 05.0 05.- 500.- 000m
000.- 000.- 000.- 000.- 000.- 000.- 007 50 0. 00:00:05
000.- 000.- 000.- 000.- woof 0 5.- 50.- 500.- 02M
50.0 50.0 50.0 50.0 50.- 50.0 000.- 55.- 0:805
0N0.- 50.- 0N0.- 0N0.- 0N0.- 0N0.- 0N0.- _N0.- ago—08m
500.- 000.- 000.- 000.- 000.- 50.- 000.- 000.- :000000m
......000.- ......N00.- ......000.- 000.- 50.- N00.- 50.- 50.- 00<
000.0 000.0 020 N0 0 .0 00 0 .0 000.0 00—0 000.0 80:00
0005: 000:5
N505 1:0me N505 :0me tommsm 00::00 2:030 00800

 

50092 :0 00000605 :00 06.3050 :ommm0hw0m ”N0 030-5

 

101

5005:0006 00.00000 00 N085».
0003000 30:00:: 00 000000: :0 5003:0086 00 2005+
0000.800: 00:0000-«000 S00 030000050300 5.V0...... .00.V0...

 

 

 

 

 

 

 

 

 

 

 

 

 

000 000 000 5000 $0.— 500N V50N VO0N Z
00. V0. v0. 00. mo. mo. 00. N0. NM
~Nw. 005. 04%. 5N0. 000. MNN. 00?. V00.- “55:00
N00.- 00005:...N52Q
HO-meNQ- ﬁolmﬁwmé- +mmu£_m*:ome
*0500 000.0 000.0 N .5me
000.0 ~N00 NN00 ~N00 0 Sme
#000 0000 0000 0.2.0 mm~0 tomasm
**050.- 0v00 0000 $0.0 0v00 000.0 00.5000
000.- al.000- **MOH.- ......000r 33.000.- 1.000.- ......05~.- :0 060.30
wNor 00.- 000.- 000.- v:.- 0:.- 00~r 000.- 000M
000.- 050.0 V500 N500 M500 V500 0000 N000 025.500:
000.- $0.. $0.- vvo... 500.- 000,- v00.- 050.- 00g
#000 #000 N000 000.0 0000 #000 000.0 000.0 08005
0N0.- 000.0 000.0 000.0 500.0 000.0 000.0 000.0 “EOE—hm
500.- ......H0v0 ......5wv0 ......00V0 **50.v.0 ......0000 **0Nm.0 ......va0 55:005me
......moor N00.- N00.- 000.- door 000.- N00.- N00.- 0w<
0000 ......0wm0 ......00M0 ......5wm0 **Nwm.0 ......5000 ......m0m0 *0MM. 50:00
wmoc=m wmoqﬁ:
*NHOme * whomwn— Nhomma Mhommo togmﬂm ﬁchu—COU QMCOEU .MOEOQ

 

.320 :0 00000605 :00 0003050 :2000000m 00 050.5

102

REFERENCES

Aneshensel, C. S. (1999). 'Outcomes of the stress process' in Howitz, A.V. and Scheid,
T. L. (eds.), A Handbook for the Study of Mental Health: Social Contexts,
Theories, and Systems (First Edition), New York, Cambridge University Press.

Aneshensel, C.S. (1992). Social stress: theory and research. Annual Review of Sociology,
18:15-38. .

Aneshensel, C.S., Pearlin, L.I., Mnllan, T., Zarit, S.H., and Whitlatch, C.J. (1995).
Proﬁles in Caregiving: The Unexpected Career. San Diego, CA: Academic Press.

Aneshensel, C. S. and Phelan, Jo C. (1999). Handbook of the Sociology of Mental
Health, New York, NY, Kluwar Academic/Plenum Publishers.

Bertakis K.D., Azari R., Helms L.J., Callahan E.J., and Robbins J.A. (2000). Gender
differences in the utilization of health care services. Journal Family Practice 49(2): 147-
152.

Brim, O.G., Baltes, P.B., Bumpass, L.L. et al., (1996). 'National Survey of Midlife
Development in the United States (MIDUS), 1995-1996.‘ (2nd Edition), Harvard
Medical School, Department of Health Care Policy.

Brown, G. and Harris, T. (1978). The Social Origins of Depression: A Study of
Psychiatric Disorders in Women (First Edition), New York, The Free Press.

Burns M.J., Cain V.A., and Husaini B.A. (2001). Depression, service utilization, and
treatment costs among Medicare elderly: gender differences. Home Health Care
Services Quarterly 19(3):35-44.

Chen MS Jr. (1999). Informal care and the empowerment of minority communities:
comparisons between the USA and the UK. Ethnic Health 4(3): 139-151.

Cherniack EP, Senzel RS, and Pan CK. (2001). Correlates of use of alternative
medicine by the elderly in an urban population. J Altern Complement Med 7(3):
277-280.

Cohen S and Edwards JR. (1989). Personality characteristics as moderators of the
relationship between stress and disorder. In Advances in the Invesg'ggtion of
Psycological Stress. RWJ Neufeld (Ed.). New York: Wiley, pp. 235-283.

Conrad, M.F., Shepard, A.D., Pandurangi, K., Parikshak, M., Nypaver, T.J., Reddy,

d.J., and Cho, J.S. (2003). Outcome of carotid endarterectomy in African
Americans: is race a factor? Journal of Vascular Surgery, 38(1):]29-137.

103

Dominitz, J.A., Maynard, C., Billingsley, K.G., and Boyko, E.J. (2002). Medical
Care, 40(1 Suppl):l 14-126.

Durkheim, E. (1951). Suicide. Glencoe, IL, Free Press.

Egede LE, Ye X, Zheng D, and Silverstein MD. (2002). The prevalence and pattern of
complementary and alternative medicine use in individuals with diabetes.
Diabetes Care 25(2): 324-329.

Eisenberg DM, Kessler RC, Foster C, Norlock FE, Calkins D, and Delbanco TL.
(1993). Unconventional medicine in the United States: prevalence, costs, and
patterns of use. NEJM 328: 246-252.

Eisenmann, J .C. (2003). Secular trends in variables associated with the metabolic
syndrome of North American children and adolescents: a review and synthesis.
American Journal of Human Biology, 15(6): 786-94.

Elston, L.J., Cole, J.C., Ben-Menachem, T., and Morlock, RJ. (2001).
Sociodemographic differences in the receipt of colorectal cancer surveillance care
following treatment with curative intent. Medical Care, 39(4):361-372.

Ensel W.M. and Lin, N. (1991). The life stress paradigm and psychological distress.
Journal of Health and Social Behavior, 33: 321-341.

Essed, P. (1991). Understanding Everyday Racism, Newbury Park, CA, Sage.

Friedman G.D. (2004). Primer of Epidemiology (5m Ed.), New York, NY, The McGraw-
Hill Companies, Inc.

Gardner, CB. (1995). Passing by: Gender and Public Harassment, Berkely, CA,
University of California Press.

Geiger, H.J. (2003). 'Racial and ethnic disparities in diagnosis and treatment: A review
of the evidence and a consideration of causes'. In, Smedley, B.D., Stith, A.Y., and
Nelson, A.R. (eds.), Unequal Treatment: Confronting Racial and Ethnic
Disparities in Health Care (CD Rom), pp. 417-440, Washington, DC, the
National Academies Press.

Goldstein MS. (2000). The growing acceptance of complementary and alternative
medicine. In H_andbool< of Medical Sociology. 5 ed. Chloe Bird, Peter Conrad,
and Allen Fremont (Eds.). New Jersey: Prentice Hall, pp. 284-297.

Gottlieb S.S., Hkatta M., Friedmann E., Einbinder L., Katzen S., Baker B.,
Marshall J ., Minshall 8., Robinson 8., Fisher M.L., Potenza M., Sigler B.,

104

Baldwin C., and Thomas S.A. (2004). The inﬂuence of age, gender, and race on
the prevalence of depression in heart failure patients. J Am College Cardiology 43
(9): 1 542-1549. '

Grembowsky D, Patrick D, Diehr P, Durham M, Beresford S, Kay E, and Hecht J.
(1993). Self-efficacy and health behavior among older adults. J Health and Social
Behavior 34289-104.

Haetzman M, Elliott AM, Smith BH, Hannaford P, and Chambers WA. (2003).
Chronic pain and the use of conventional and alternative therapy. Family Practice
20(2): 147-154.

Harrell, S. P. (2000). A multidimensional conceptualization of racism-related stress:
implications for the well-being of people of color. American Journal of
Orthopsychiatry, 70(1): 42-57.

Herron M and Glasser M. (2003). Use of and attitudes toward complementary and
alternative medicine among family practice patients in small rural Illinois
communities. J Rural Health 19(3): 279-284.

Holmes, T. and Rahe, . (1967). The social readjustment rating scale. Journal of
Psychosomatic Research, 1 1 :213-218.

Kessler, R.C., Mickelson, KD. and Williams, DR. (1999). 'The prevalence,
distribution, and mental health correlates of perceived discrimination in the
United States', Journal of Health and Social Behavior, 40, 208-230.

King, Gary. (1996). Institutional racism and the medical/health complex: a conceptual
analysis. Ethnicity & Disease, 6, 30-46.

Krieger, Nancy. (1990). 'Racial and gender discrimination: risk factors for high blood
pressure?‘ Social Science and Medicine, 30, 1273-1281.

Lefcourt HM and Davidson-Katz K. (1991). Locus of control and health. In Handbook
of Social and Clinical Psychology: The Health Perspective. CR Snyder and RR
Forsyth (Eds). New York: Pergamon, pp. 246-266.

Lennon, M.C. and Rosenfield, S. (1992). Gender and mental health: interactions of job
and family. Journal of Health and Social Behavior, 33, 316-327.

Lester, N., Nebel, L. and Baum, A. (1994). 'Psychophysiological and Behavioral
Measurement of Stress: Applications to Mental Health.‘ in Avison, W.R. and
Gotlib, I.H. (eds.), Stress and Mental Health: Contemporary Issues and Prospects
for the Future (First Edition), New York, Plenum Press.

105

Levin JS. (1994). Religion and health: is there an association, is it valid, is it causal?
Social Science and Medicine 28: 1475-1482.

Mackenzie ER, Taylor L, Bloom BS, Hufford DJ, and Johnson JC. (2003). Ethnic
minority use of complementary and alternative medicine (CAM): a national
probability survey of CAM utilizers. Altern Ther Health Med 9(4): 50-56.

Mannino, D.M., Homa, D.M., Akinbami, L.J., Morrman, J.E., Gwynn, C., and
Redd, SC. (2002). Surveillance for asthma--United States, 1980-1999. MM WR
Surveillance, 29(51):1-13.

Mascie-Taylor, CG. and Karim E. (2003). The burden of chronic disease. Science,
302(5652): 1921-22.

McBean, A.M., Huang, Z., Virnig, B.A., Lurie, N., and Musgrave, D. (2003). Racial
variation in the control of diabetes among elderly medicare managed care
beneﬁciaries. Diabetes Care, 26(12):3250—3256.

McMahon, L.F., Wolfe, R.A., Huang, 8., Tedeschi, P., Manning, W., and Edlund,
M.J. (1999). Racial and gender variation in use of diagnostic colonic procedures
in the Michigan Medicare population. Medical Care, 37(7):712-717.

Menaghan, E.G. (1991). Work experiences and family interaction processes: the long
reach of the job. Annual Review of Sociology, 17, 419-444.

Merton, R. K. (1968). 'Social Structure and Anomie.‘ In R.K. Merton (ed.), Social
Theory and Social Structure (3rd ed.), pp. 85-124. New York, Free Press.

Mirowsky, J. (1994), 'The advantages of indexes over diagnoses in scientiﬁc
assessment', in Avison, W.R. and Gotlib, I.H. (eds), Stress and Mental Health:
Contemporary Issues and Prospects for the Future (1st Edition), New York,
Plenum Press.

Muhajarine N, Neudorf C, and Martin K. (2000). Concurrent consultations with
physicians and providers of alternative care: results from a population-based
survey. Can J Public Health 91(6): 449-453.

Ong CK, Peterson S, Bodeker GC, and Stewart-Brown S. (2002). Health status of
people using complementary and alternative medical practitioner services in four
English counties. Am J Public Health 92(10): 1653-1656.

Pachter LM, Sumber T, Fontan A, Sneed M, and Bernstein BA. (1998): Home-based
therapies for the common cold among European American and ethnic minority

families: the interface between altemative/complementary and folk medicine.
Arch Pediatr Adolesc Med 152(11): 1083-1088.

106

Pearlin, L. I. (1983). 'Role Strains and Personal Stress', in Kaplan, H. (ed.), Psychosoical
Stress: Trends in Theory and Research, New York, Academic Press.

Pearlin, LL (1989). 'The sociological study of stress', Journal of Health and Social
Behavior, 30, 241-256.

Pearlin, LL (1999). 'The Stress Concept Revisited', in Aneshensel, CA. and Phelan, J .C.
(eds), Handbook of the Sociology of Mental Health, New York, Plenum

Publishers.

Pearlin, LJ. and Aneshensel, C.S. (1994). Caregiving: the unexpected career. Social
Justice Research, 72373-90.

Pearlin, L.I., Aneshensel, C.S., and LeBlanc, A.J. (1997). The forms and mechanisms
of stress proliferation: the case of AIDS caregivers. Journal of Health and Social

Behavior, 38(Sept.):223-36.

Pearlin, LI. and Aneshensel, C. S. (1994). 'Caregiving: the unexpected career'. Social
Justice Research, 7, 373-390.

Pearlin, L.I., Mullan, J.T., Semple, S.J., and Skaff, M.M. (1990). Caregiving and the
stress process: an overview of concepts and measures. The Gerontologist, 30:583-

94.

Pearlin, L.I., Lieberman, M.A., Menaghan, E.G., and Mullan, J.T. (1981). 'The stress
process'. Journal of Health and Social Behavior, 22, 337-356.

Powers J.R., Young A.F., Russell A., and Pachana NA. (2003). Implications of non-
response of older women to a short form of the Center for Epidemiologic Studies
Depression Scale. Int J Aging and Human Development 57(1): 37-54.

Rapley, P. and Fruin DJ. (1999). Self-efﬁcacy in chronic illness: the juxtaposition of
general and regimen-speciﬁc efﬁcacy. International Journal of Nursing Practice,

5:209-215.

Roe C.M., McNamara A.M., and Motheral BR. (2002). Gender and age related
prescription drug use patterns. Annals Pharmacotherapy 36(1):30-39.

Rucker-Whitaker, C., Feinglass, J., and Pearce, W.H. (2003). Explaining racial
variation in lower extremity amputation: a 5-year retrospective claims data and
medical record review at an urban teaching hospital. Archives of Surgery,

138(12):1347-1351.

Schieman S., Van Gundy K., and Taylor J. (2002). The relationship between age and
depressive symptomsL a test of competing explanatory and suppression
inﬂuences. J Aging Health 14(2):260-285.

107

Serido J., Almeida D.M. and Wethington E. (2004). Chronic stressors and daily
hassles: unique and interactive relationships with psychological distress. Journal

of Health and Social Behavior, 45(March):l7-33.

Smola S, Justice AC, Wagner J, Rabeneck L, Weissman S, Rodriguez-Barradas M.
(2001). VACS 3 Project Team: Veterans aging cohort three-site study: overview

and description. J Clinical Epidemiology 54(12): 861-876.

Sturm R and Sherboume CD. (2001). Are barriers to mental health and substance
abuse care still rising? J Behavioral Health Services Research, 28(1):81-88.

Thompson, V.L. (1996). 'Perceived experiences of racism as stressful life events.‘
Community Mental Health, 32, 223-233.

Turner, R.J. and Lloyd, D.A. (1995). 'Lifetime traumas and mental health: the
signiﬁcance of cumulative adversity', Journal of Health and Social Behavior, 36,

360-376.

Wagner PJ, Jester D, LeClair B, Taylor AT, Woodward L and Lambert J. (1999).
Taking the edge off: why patients choose St. J ohn's Wort. J Family Practice,

48(8):615-619.

Weinick, Robin M., Zuvekas, S.H., and Cohen J.W. (2000). Racial and ethnic
differences in access to and use of health care services, 1977 to 1996. Medical

Care Research and Review, 57(81): 36-54.

Wheaton, Blair. (1994). 'Sampling the Stress Universe', in Avison, W.R. and Gotlib, I.H.
(eds), Stress and Mental Health: Contemporary Issues and Prospects for the
Future (First Edition), New York, Plenum Press.

Wheaton, Blair. (1996). The domains and boundaries of stress concepts. In HR
Kaplan (Ed.), Psychosocial stress: perspectives on structure, theory, life-course, and

methods (pp. 29-70). New York: Academic Press.

108

 

lijl'llljjlﬂﬂljllm1