an.

1

.
M

._§fi

$5

I. hitb
. humur‘dwuca: an.
650'... .461. £th
.a‘. . .

I. (a; .
V1.53. a
I.

 

Ix. :IFNIU-a-‘l .

This is to certify that the
dissertation entitled

PATTERNS AND PREDICTORS OF COMPLEMENTARY
AND ALTERNATIVE THERAPY USE IN A CANCER
POPULATION:

A SECONDARY ANALYSIS OF THE 2002 NATIONAL
HEALTH INTERVIEW SURVEY

presented by

JUDITH M. FOULADBAKHSH

has been accepted towards fulfillment
of the requirements for the

PhD degree in Nursing

 

 

%ajor Professor’s Signatu‘r'e

October 20, 2006

 

Date

MSU is an Affirmative 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
; K 8
Eli‘RZQSW MR 0. 7 ZUIU
091309

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p1/ClRC/Date0uemddop1

 

 

 

PATTERNS AND PREDICTORS OF COMPLEMENTARY AND ALTERNATIVE
THERAPY USE IN A CANCER POPULATION: A SECONDARY ANALYSIS
OF THE 2002 NATIONAL HEALTH INTERVIEW SURVEY
By

Judith M. Fouladbakhsh

A DISSERTATION

Submitted to
Michigan State University
in partial fiilfillment of the requirements
for the degree of

DOCTOR OF PPHLOSOPHY
College of Nursing

2006

ABSTRACT
PATTERNS AND PREDICTORS OF COMPLEMENTARY AND ALTERNATIVE

THERAPY USE IN A CANCER POPULATION: A SECONDARY ANALYSIS
OF THE 2002 NATIONAL HEALTH INTERVIEW SURVEY

By

Judith M. Fouladbakhsh

This secondary analysis of data from the 2002 National Health Interview Survey
has determined factors that predict the use of complementary and alternative medicine
(CAM) therapies in the United States cancer population, and patterns of use of CAM
providers, practices, and products. The CAM Healthcare ModeL a modification of
Andersen’s Behavioral Model for Health Services Use, guided the study. Predisposition
to use CAM provider services, practices, and products, factors that enabled or impeded
use, and need for care factors were examined with respect to CAM use for treatment
and/or for health promotion purposes. Multivariate analysis has identified characteristics
that distinguish CAM users from non-users in the US. cancer population thereby
extending what has been reported in the research literature. Analysis using STATA 9.2
software has allowed for determination of significant predictors of overall CAM use and
use of specific categories of CAM in the estimated U.S. cancer population of more than
14.3 million individuals who have been diagnosed with cancer. The patterns of CAM use
have been compared among recent and long-term cancer survivors. The empirical
findings confirm that overall CAM use in the cancer population was more prevalent
among female, middle-aged, white, and well-educated people. Data further reveal that

women were specifically more likely to use CAM practices, but not more likely to use

providers or products than men. Those with private insurance were more likely to use
only CAM providers. Higher income, contact with nurse practitioners, physician
assistants, therapists, and mental health professionals, and presence of symptoms and co-

existing co-morbidity were strong predictors of CAM use in the cancer population.

Copyright by
Judith M. Fouladbakhsh
2006

ACKNOWLEDGMENTS

The work and scholarship reflected in this dissertation were made possible by the
combined energy, effort, and love of those who surrounded me during my doctoral
journey. It is with the most heartfelt gratitude that I thank my family for their
never ending support, their willingness to join forces, and their always-present sense of
humor. I sincerely thank my wonderful husband Hersel, and my children Joe, Eric,
David, and Lisa, and of course, little Julian, for always reminding me what the journey
was really all about. I further extend a very sincere note of thanks to my wonderful
friends who patiently listened, offered advice, and provided support to finish this work.

My gratitude is also extended to Dr. Manfred Stommel, a most brilliant visionary
and mentor, for his guidance, ability to listen patiently, expertise at creating exciting new
challenges, and his willingness to provide learning beyond my highest expectations.
Thank you to an advisor that was a gift on this journey. A sincere note of thanks is also
extended to doctoral program committee members, Dr. Barbara A. Given, Dr. Linda
Spence, Dr. Harry Perlstadt and Dr. Margaret Holmes-Rovner for their expertise,
encouragement, and teamwork. In addition, a word of appreciation for Dr. Audrey Gift
whose faith in my ability to achieve this honor was invaluable. This research was
possible with the collaborative effort of all the faculty members who have made the
doctoral program at the College of Nursing a reality. Funding for this research was
provided by the Oncology Nursing Society and Foundation, Blue Cross and Blue Shield
Foundation of Michigan and Sigma Theta Tau International Honor Society, Lambda

Chapter.

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES

CHAPTER 1
INTRODUCTION, SIGNIFICANCE AND CONCEPTUAL FRAMEWORK
Introduction and Significance
Trends
Importance for Study
Definitions and Categorization of CAM
NCCAM Categorization
Alternative Categorization
Purpose for CAM Use
Interface: CAM and Conventional Healthcare
Conceptual Framework
Overview of the Behavioral Model for Health Services Utilization
The CAM Healthcare Model: Rationale for Model Revisions
Defining CAM in a Health Services — Health Practice Framework
The CAM Healthcare Model: Model Constructs
Aggregate Level Determinants
Individual Determinants
Using the CAM Healthcare Model with the NHIS 2002 Dataset
Purpose of the Study

CHAPTER 2
LITERATURE REVIEW
CAM Healthcare Model for NHIS Secondary Analysis
Predisposing Variables
Enabling Variables
Need Variables
CAM Healthcare Model for NHIS Secondary Analysis: Cancer Specific
Predisposing Variables
Enabling Variables
Need Variables
Research Questions & Hypotheses
Predisposing Variables
Enabling Variables
Need Variables

Vi

viii

ix

NNu—au—ay—ni—In—s

23
23
23
26
29
35
35
36
36
38
38
39
40

CHAPTER 3
METHODOLOGY
Description ofNHIS 2002
Data Collection Procedures
Sample
Instruments
Predisposing and Enabling Variables
Need Variables
CAM Use —Outcomes
Study Variables
Dependent Variables
Independent Variables
Procedure
Data Analysis

CHAPTER 4
DATA ANALYSIS & RESULTS
Analytical Procedures
Model Variables - Descriptive Statistics
Predisposing Variables
Enabling Variables
Need Variables
CAM Use & Purpose -— Descriptive Statistics
Outcome Variable — CAM Use
Outcome Variable - CAM Purpose
Predictors of CAM Use
CAM Use—Overall
CAM Use—Specific Categories
Predictors of Purpose of CAM Use
Summary: Predictors & Purpose of CAM Use

CHAPTER 5
DISCUSSION & CONCLUSIONS
Cancer Population
Predictors of CAM Use and CAM Purpose
Predisposing Variables
Enabling Variables
Need Variables
Use of the CAM Healthcare Model
Conclusions
Limitations

REFERENCES

vii

42
42
42
43
44
45
45
46
46
46
48
49
50

52
52
54
54
56
58
61
61
62
62
63
66
70
72

74
74
77
77
82
86
89
91
94

146

LIST OF TABLES

Table 1. List of CAM Therapies in the 2002 NHIS 96
Table 2. NCCAM Categorization of CAM Therapies 97
Table 3. Alternative Categorization of CAM Providers,

Practices, and Products 98
Table 4. Independent Variables in the CAM Healthcare Model 100
Table 5. Comparison of the Non-cancer Population and the Cancer Population 103
Table 6. Comparison of Recent and Long-term Survivors in

the Cancer Population 110
Table 7. Prevalence of CAM Use in the Cancer and Non-cancer Population 118
Table 8. Frequency of Use of Specific Types of CAM in the Cancer Population 120
Table 9. Purpose of CAM Use in the Cancer & Non-cancer Population 125

Table 10. Logistic Regression Model: Predictors of CAM Use
in the Cancer Population 126

Table 11. Multinomial Logistic Regression Model: Predictors of Use of

Specific Categories of CAM in the Cancer Population 129
Table 12. Multinomial Logistic Regression Model: Predictors of

Purpose for CAM Use in the Cancer Population 132
Table 13. Summary of Predictors of Specific CAM Use Categories , 135

Table 14. Summary of Predictors of Purpose for CAM Use 136

viii

Figure 1.
Figure 2.
Figure 3.

Figure 4.

Figure 5.

Figure 6.

LIST OF FIGURES

Dual Nature of CAM
CAM Healthcare Model
CAM Healthcare Model for NHIS Secondary Analysis

CAM Healthcare Model for NHIS Secondary Analysis:
Cancer Specific

Graph Relating Probability of Overall CAM Use to Age

Graphs Relating Probability of Use of Different Categories
of CAM to Age

ix

137

138

140

141

142

144

CHAPTER 1

INTRODUCTION, SIGNIFICANCE AND CONCEPTUAL FRAMEWORK

Introduction

Complementary and alternative medicine (CAM) therapies are widely used across
the United States (US) by healthy individuals and by those experiencing illness. Data
indicate that those with cancer and/or chronic disease often seek CAM as a form of
complementary treatment to promote health, treat illness and manage related symptoms
and functional limitations. It is anticipated that CAM use will continue, thus reflecting
changes in health care behavior with the potential to effect use of conventional health
services (Astin, 1998; Barnes, Powell-Griner, McFann & Nahin et al., 2004; Eisenberg,
Davis, Ettner, Appel, Wilkey, van Rompay, et al., 1998; Ernst & Cassileth, 1998;
Richardson & Straus, 2002; Wolsko et al., 2003).

Trends

Trend data from population studies have revealed a continuing increase in CAM
use over the past decade, with visits to CAM providers surpassing number of visits to
conventional primary care providers in the US. Uses of CAM health products, such as
herbs and vitamins, have increased by 380% and 130%, respectively. Increases have also
been noted in the use of massage, self-help groups, folk remedies, energy healing and
homeopathy (Eisenberg et al., 1998; Barnes et al., 2004). In contrast, however, other
survey data have revealed lower percentages of use than the seminal work of Eisenberg
and colleagues (Druss & Rosenheck, 1999). These inconsistencies may be attributed to

the operational definition of CAM, and the varying list of therapies included for study.

Whereas some studies examine a very extensive array of provider services, product use
and self-care practices, other studies may solely focus on use of CAM providers. In
addition, findings may vary due to different sampling procedures and research
methodology, for example, use of face-to-face interviews versus telephone surveys, and
differences in socioeconomic status of the original populations studied (Bausell, Lee &
Berman, 2001; Druss & Rosenheck, 1999). Hence, one must acknowledge that
limitations in comparing prevalence and patterns of CAM use exist across studies.

Data from the National Health Interview Survey (NHIS, 2002) also support
widespread CAM use, revealing that 75% of the US population used CAM during their
lifetime, 62% using within the preceding year (Barnes et al., 2004). In the NIHS dataset,
CAM use is determined by responses to items asking about 22 different CAM provider
services, practices and products, including use of prayer for health reasons (Table 1).
Frequently used CAM therapies included personal prayer (43%), prayer by others (24%),
natural products (19%), deep breathing exercises (12%), participation in a prayer group
(10%), meditation (8 %), yoga (5%), massage (5 %), and diet-based therapies (4%).
When prayer for health reasons was excluded in the NHIS analysis the percent of the
population using CAM within the preceding year decreased from 62% to 36%. Tindle,
Davis, Phillips and Eisenberg (2005) have reported that the prevalence of CAM use
between 1997 and 2002 has remained consistent. Cost for CAM use, most often requiring
out-of-pocket expenditures, has been estimated to range from $27 to $47 billion per year
(Barnes et al., 2004; Eisenberg et al., 1998).

Data from smaller scale cross-sectional studies also reveal widespread use of

CAM across urban, suburban and rural communities. Data revealed that 76% of

community residents used CAM therapies, products, and practices for the self-treatment
of pain, often without informing their primary care providers. CAM therapies used to
manage pain included yoga, massage, relaxation, prayer, meditation and chiropractic
treatment (Vallerand, Fouladbakhsh & Templin, 2003). Studies examining different
arrays of CAM therapies and practices among diverse groups also support increasing
prevalence of CAM use worldwide. Varying estimates range from 38% in Belgium, 47%
in the United Kingdom, 48% in Australia, 70% in Canada to 75% in France (Lewith,
Broomfield, & Prescott et al., 2002; World Health Organization, 2004).
Importance for Study

Given the apparently widespread use and rising popularity of CAM therapies, it is
important to more fully understand who uses CAM and for what purposes. In order to do
so, it is imperative to clearly identify what is included in the operational definition of
CAM. Only then will we be able to sort the predictors of use of CAM provider services,
products and practices to determine similarities and differences among diverse
populations of users. This information will then be available for nurses, physicians, and
other conventional healthcare providers caring for individuals who are CAM users.
Ultimately, this will increase our understanding of how CAM interfaces with the use of
conventional healthcare services, and its’ influence on healthcare outcomes.

Definitions and Categorization of CAM

NC CAM Categorization

The National Center for Complementary and Alternative Medicine (NCCAM)
within the National Institutes of Health defines CAM as a group of diverse medical and

health care systems, practices, and products that are not presently considered to be a part

of conventional medicine. A complementary therapy is defined as a selected therapeutic
method, product, or treatment by a practitioner used in combination with conventional
mainstream medicine as a health service for patients. Alternative therapy is defined as a
selected therapeutic method, product or treatment by a practitioner used in place of
conventional medical therapy (NCCAM, 2004). Distinguishing between a
complementary therapy and alternative therapy requires examination of the context of
use, and whether the CAM user is seeking services or products for health reasons outside
of the conventional health service sector. For example, herbs used in place of
pharmaceutical products would be considered alternative, whereas herbs used in
combination with pharmaceutical products would be considered complementary. The
task of determining whether individuals use nonconventional therapies instead of, or in
addition to, conventional biomedical treatments is not possible in the NHIS, hence the
term CAM is used throughout this dissertation. The NCCAM categorization of CAM
includes four areas and one overarching domain that include the following:
1. Mind-body therapies: behavioral, social, psychological and spiritual
approaches to health, e.g., yoga, Tai Chi, meditation, hypnosis
2. Biological-based therapies: natural and biologically-based products, practices
and interventions, e.g., herbs, supplements, diet therapy
3. Manipulative and body-based systems: systems based on manipulation and/or
movement of the body, e.g., massage, Feldenkrais
4. Energy therapies: systems that use subtle energy fields in and around the
body to promote healing, e.g. Healing Touch, Therapeutic Touch,

acupuncture, Reiki

5. Alternative medical systems: this overarching domain, that may include
therapies within the other four areas, is defined as complete systems of theory
and practice developed outside of a western, conventional biomedical
approach to health and illness, e.g., homeopathy, naturopathy, traditional
Chinese medicine (Table 2)

The NCCAM categorization is one means of dividing up the vast array of
therapies and practices inherent within the definition of CAM. It does not, however,
allow for the distinction of therapies based upon use or non-use of a provider that delivers
services. In addition, it must be noted that the current categorization does not fully
acknowledge the overlap that occurs among the categories. For example, yoga and Tai
Chi, listed as mind-body therapies, are also based on the principle of energy flow, i.e.,
chi, through the body to promote healing. Hence, limiting these CAM practices to one
category is not fully explanatory. Since this dissertation views CAM within a health
service-health practice framework, an alternative categorization will be used.

Alternative Categorization

Whereas many CAM therapies require services of practitioners, others are readily
available for self-directed use by consumers for health related purposes. Thus, for this
dissertation an alternative categorization of CAM is utilized (Table 3). CAM is
conceptualized from a health service-health practice perspective, and it includes the
following categories:

1. CAM health services (individual and group) provided by CAM practitioners,

e. g., acupuncture, massage, naturopathic, and chiropractic treatment

2. CAM products and resources such as herbs, supplements, essential oils, self-
help manuals, books, and other instructional materials
3. CAM health practices (individual and communal), such as meditation, yoga
practice, breath work, and use of music

These categories are not considered mutually exclusive (Figure 1) as CAM practices and
products are often used and recommended by CAM providers. This alternative
categorization will allow for more specific identification of the category of CAM that is
used for illness treatment as compared to health promotion, among those with and
without illness.
Purpose for CAM Use

CAM, an area of great complexity, is viewed as a form of health care that
includes services and therapies provided by CAM practitioners and an extensive array of
products and practices that can be used independently for self-care. Although not very
different from self-care in conventional medicine, what differs may be the purpose, focus,
and resources available for self-care in CAM. The author postulates that this difference
may be related to the philosophical premise of CAM that places greater responsibility for
healing on the individual. This premise differs from conventional health care where the
burden of responsibility rests with the physician. Thus, CAM in essence may encourage
self-care and health promotion for those who are healthy. CAM is also viewed as
meeting a need for symptom management that exists among those with chronic illness
where no cure through conventional medicine is possible. Hence, it is imperative to
understand benefits and risks of long-term use, impact on cost of care, and health

outcomes of CAM use.

Interface: CAM and Conventional Healthcare

The pattern of increasing CAM use by those with or without illness calls attention
to the need for information about the interface of CAM use with conventional health
services utilization. The potential effect of CAM use on the delay, disruption, or
enhancement of conventional health service use for illness prevention, health promotion,
diagnosis, and treatment remains largely unexplored. CAM therapies, when used
concurrently with conventional health care services must be evaluated for (a) potential
positive and negative interactions with conventional biomedical and pharmaceutical
treatment, (b) levels of effectiveness, and (c) influence of CAM use on concurrent use of
conventional health services. In addition, the converse should also be examined: How
does the lack of access to and cost of conventional health services influence the use of
CAM? Emerging results indicate significant differences in use of CAM therapies
between those reporting and those not reporting difficulty obtaining medical care because
of cost (Pagan & Pauly, 2005).

Changes in policy and research funding priorities at the local and national level
further validate increasing CAM prevalence and potential impact on conventional health
care delivery. Continuing demand for the development of guidelines for CAM use in
diverse populations is evident. Thus, as an increasing number of CAM users present for
conventional health services, and as healthcare systems throughout the nation explore the
benefits and risks of integrating CAM therapies as a health service option, it becomes
imperative to understand the following:

1. Characteristics of users

2. Patterns of use (visits to CAM practitioners, and self-directed use of CAM

health practices and products)

3. Factors that predict who will use this complementary form of healthcare

across diverse populations (what increases likelihood of CAM use)

4. Purpose of CAM use (health promotion, illness/symptom treatment)

In summary, the increasing prevalence of CAM has created a need to identify
predictors, patterns, and purposes of use among diverse populations. The aim of this
dissertation was to identify characteristics of CAM users, as well as factors associated
with use in the US. cancer population. Information from this study will increase our
understanding of CAM use, and promote early recognition of CAM users. Data may also
serve to inform the integration process as CAM becomes more available in conventional
healthcare settings. Further, detailed understanding of patterns of use can illuminate
areas for CAM effectiveness research. Overall, identification of characteristics, predictors
and patterns of use is viewed as vital to promoting quality health care and maximizing
positive health outcomes among those who choose to use CAM (Fouladbakhsh, Stommel,

Given & Given, 2005; Jordan & Delunas, 2001; Off & Lynch, 2002).

Conceptual Framework

Overview of the Behavioral Model for Health Services Utilization

The Behavioral Model for Health Services Utilization is a theoretical fi'amework
with a consistent and longstanding ability to guide research that examines factors related
to utilization of conventional health services and resources (Aday & Andersen, 1974;

Aday & Awe, 1997; Andersen, 1995; Andersen & Newman, 1973; Andersen, 1968). The

Behavioral Model uses the individual as the unit of analysis, with societal determinants
(technology and norms), and the health services system (resources and organization) as
aggregate determinants of an individual’s health-care-seeking behavior (Awe & Aday,
1997; Andersen & Newman, 1974).

The Behavioral Model proposes that the use of health services is a function of
individual determinants identified as: (a) predisposing variables - an individual’s
propensity to use services; (b) enabling variables - the means an individual has available
for the use of health services, and (c) need variables - the individual’s need for care
(Andersen & Newman, 1974; Andersen, 1995). The outcome of the utilization of
conventional healthcare services and resources is quality of life (Andersen, 1995).
Predisposing variables include demographic characteristics such as gender, age, and
marital status, social structure attributes identified as education, race, ethnicity and
employment, and health beliefs as indicated by individual values and attitudes about
health services, good health, physician services, and health insurance. The primary focus
of the Behavioral Model, in general, and the predisposing variables, in particular, is on
factors that influence the demand for services, about which information can usually be
gathered through survey methods.

Enabling variables identified in the model are those conditions or factors that
allow (enable) or impede use of health services. Included are resources specific to
individuals and families that may potentially influence conventional health Service use
such as income, health insurance, employment, and regular source of care. Also viewed
as potential predictors of health service use are community attributes, such as physician

and hospital bed ratios, and region of the country. In addition, place of residence, e.g.,

urban or rural, influences proximity to sources of conventional health services, and may
be considered as a factor affecting utilization.

The theoretical construct of need variables in the Behavioral Model includes both
evaluated and perceived need. Evaluated need refers to objective measurable indicators
of health status, such as actual diagnostic reports, symptom severity measurement, and
treatments received. Perceived need refers to the individual’s perception of health status
and illness state, as measured by perception of health scales scores, and other perceived
health status indicators. Need for care has also been measured through: (a) presence of
illness (symptoms, limitations, number of days disabled, etc.), (b) individual’s responses
to illness (going to a physician, clinic visits, etc.), and (c) measures taken to prevent
illness and maintain health (physician exams, etc.) (Aday & Awe, 1997).

The Behavioral Model was developed to measure the effect of individual
determinants on the utilization of conventional health services by individuals and
families. Health services are defined as those services provided by physicians, nurses,
and other health care practitioners at conventional care settings, such as physician offices,
hospitals, outpatient clinics, and emergency departments, for preventative, curative, and
restorative health care. Model revisions, over time, have also allowed for the study of
access to health care services and utilization patterns, among general and vulnerable
populations.

The CAM Healthcare Model: Rationale for Model Revisions

The CAM Healthcare Model has been developed by the author as an extension or

modification of the Behavioral Model of Health Services Use for research on CAM use.

The CAM Healthcare Model is Shown in its entirety in Figure 2, and is discussed from a

10

theoretical perspective. Recognizing that not all the model variables were available in the
NHIS 2002 dataset, the full model was subsequently modified as reflected in Figures 3
and 4. For this dissertation research, the model shown in Figure 4, which is specific to
cancer, was used. This progression from the full model to a more limited version reflects
the process of model development used for this research. Although the entire CAM
Healthcare Model (Figure 2) was not used, it is anticipated that it will guide future CAM
research endeavors.

Studies focused on CAM are increasingly noted in the literature with widespread
variations in findings related to patterns and predictors of use. A conceptual framework
is needed at this time to guide CAM research so that more consistency and comparability
exists across studies, thereby strengthening findings. The CAM Healthcare Model was
employed to increase understanding of CAM utilization to promote health, prevent
disease, treat illness, and manage symptoms in a cancer population (Figure 4). This
model conceptualizes CAM within a health service utilization framework, hence the
applicability of the Behavioral Model constructs. The Behavioral Model however is
modified to include all aspects of CAM use (Table 3). Primarily, the CAM Healthcare
Model differs from the original Behavioral Model at the level of health services use. The
CAM Model includes utilization of health services AND/OR health practices. As
depicted in Figure 1, the use of resources and products is noted to overlap with both
health services and health practices.

The inclusion of CAM health practices in the revised model reflects emphasis on
a self-directed (self-care) component of health care that is viewed as an internal

individual resource for health. Self-care behavior is also a vital component within

11

conventional health care, as noted by the vast array of pharmaceutical products and
medical supplies available to consumers. Conventional self-care behavior, however, is
most often based on previous contact with and recommendations from conventional
healthcare providers, for example, doctors who guide diabetic care and pharmacists who
are available in community-based retail sites for over-the-counter sales. Inherent within
the CAM Healthcare model is a higher level of personal involvement in the self-directed
component of CAM health practices. Decisions to use CAM are often in opposition to
established healthcare norms, and often used without conventional provider awareness
(American Society of Radiation Oncology, 2005; Giveon, Liberman, Klang & Kahan,
2003; Vallerand, Fouladbakhsh & Templin, 2003). CAM health practices generally
require intense self-directed information seeking from a myriad of sources, and often
include an expanded sense of personal involvement in health and the healing process.
Frequently, emphasis is placed on the promotion of health to maintain an existing state of
wellness. This is notably different when contrasted with the predominant treatment-
focused, “let’s promote a return to wellness” approach of conventional medicine. Thus,
the Behavioral Model is modified in the “use of health services” construct to include both
of the following: use of CAM health services (provider required or provider directed) and
use of CAM health practices (provider not required/self-directed). Direct and indirect
relationships among the model variables will be examined to identify factors predicting
use of CAM provider services, products and health practices in the study sample.
Defining CAM in a Health Service-Health Practice Framework

Health service is defined as the assessment, treatment and/or referral for treatment

of medical/health conditions provided by physicians, nurses and other healthcare

12

practitioners. Standards for conventional health services are mandated, as are
certification and/or licensing of providers with established mechanisms of control.
Within the CAM world, variation still exists in defining what CAM actually is, and who
the providers are. Licensing and credentialing exists in some areas, but may be lacking in
others, as are standards and guidelines for practice. Differences exist in what may be
designated as a CAM therapy or practice. It is evident in the research literature that a
widely varying selection of therapies, services, products, and practices are studied under
the label of CAM. Disagreement occurs over, e.g., the inclusion of prayer as a CAM
practice, with inclusion in some studies, but not others. Thus, one might ask whether
prayer, something that is inherent within religions across the world, is indeed a
complementary or alternative health therapy or practice. How is prayer, albeit personal
or not, defined? What is termed CAM also differs now that conventional healthcare
systems have started to integrate nonconventional therapies into their health service
options. Is massage no longer a CAM therapy now that it is offered to hospitalized
patients? These are just a few of the questions raised in the literature about defining
CAM in research and practice. Another health service perspective considers dollars spent
on CAM health care. CAM health services and products are “purchasable” fi'om
providers, whereas CAM practices are not, although they may involve the “purchase” of
instruction and/or guidance through service providers and/or the purchase of products.
Thus one may conclude that given the variation that exists within CAM, defining CAM
within a health services context is challenging.

Therefore, the conceptual model developed for this dissertation defines CAM

with a dual nature that consists of health services and health practices, respectively

13

provider-directed CAM and self-directed CAM (Figure 1). Viewing the self-directed
health practices as an internal resource of the individual for health and healing allows for
use of the Behavioral Model of Health Services Utilization (Figure 2) that is modified for
this dissertation (Andersen, 1995). Although internally driven, self-directed CAM often
involves external resources, such as the purchase of vitamins, herbs, and other products,
and/or information seeking. External resources and CAM practices are also
recommended and used by providers in the delivery of CAM health services as depicted
in the overlapping area of products and resources in Figure 1.

Whether patients are using CAM as a health service, i.e., seeking services from a
CAM provider, or using CAM, as a health practice/self-care activity requires distinction
for several reasons. First and foremost is the issue of what constitutes CAM. Often
viewed as natural therapies, the complexity of CAM is fi'equently underestimated. Thus,
the potential for harmful interactions, as well as positive effects, may not be considered
fully. Provider- directed CAM use allows for supervised health care, whereas self-
directed use does not. Using a CAM provider may promote use of the client’s medical
health history information and coordination of health services with conventional health
care providers. Ultimately, however, whether provider-directed or self-directed, CAM
use is viewed as having the potential to affect the provision, cost and quality of
conventional health services. This may result from (a) positive therapeutic effects of
CAM that decrease need for conventional health services, (b) delay in seeking
conventional treatment as one tries CAM first, and (c) potential negative interactions and
adverse events resulting fi'om concurrent and unrecognized use of CAM products and

practices with conventional treatments. In addition, the CAM trend can be viewed as a

14

healthcare movement that is altering the relationships among the public and the medical
community, in essence challenging the authority of medical expertise. In this sense, even
those CAM practices and products not viewed as a health service in the traditional sense
because of the absence of a provider, are presenting a challenge to the conventional
health service sector.

The CAM Healthcare Model: Model Constructs

Aggregate Level Determinants

Societal determinants in relation to CAM health care behavior include increased
acceptance of CAM use and availability throughout the US. The acceptability of certain
CAM therapies and practices is evidence of an emerging aggregate level norm, as noted
by, but not limited to: (a) Media coverage, e.g., advertisements, news articles, television
commercials, (b) CAM therapy availability at health care facilities, (c) course offerings in
schools of medicine, nursing, pharmacy, and health sciences, and ((1) programs in
integrative medicine. Technological advances, most notably the Internet, allow for rapid
access to CAM providers, products and health practice information for self-directed use.
Integration of CAM into conventional health-care delivery systems, and CAM
reimbursement by health insurance companies are also powerful aggregate level
determinants that influence CAM utilization.

Individual Determinants

The CAM Healthcare Model, although retaining all of the individual
determinants, fiirther extends and adds to the original Behavioral Model by including
variables and potential empirical indicators that are considered to be specifically related

to CAM use of provider services and self-directed health practices (Figure 2).

15

Predisposing variables.

The fill] CAM Healthcare Model (Figure 2) includes six categories of
predisposing variables as noted below. Variables and potential indicators that
specifically refer to CAM are included in the model and are noted with an asterisk.

1. Demographic characteristics (indicators: gender, age and marital status)

2. Social structure attributes (indicators: education, race, ethnicity, employment,
community lifestyle,* and cultural practices*). Culture of origin and culture
of community living are both viewed as having the potential to influence
individual or family predisposition to use CAM.

3. Health beliefs (indicators: values and attitudes about good health,
responsibility for health care,* and CAM practitioners, health practices, and
products“)

4. Risk perception“ (indicators: illness perception" and CAM perception“).
Illness perception includes perceived risk and perceived severity of illness and
perceived healthcare options. CAM perception includes perceived safety,
perceived efficacy, and perceived acceptability of CAM health services,
practices and products.

5. Personal Knowledge“ (indicators: level of CAM knowledge“ and verbalized
need for CAM information")
6. Personality characteristics“ (indicators: measures of self-efficacy", risk-
taking ability", perception of control“, and self-care propensity)

(Figure 2).

16

Enabling variables.

The full CAM Healthcare Model (Figure 2) retains the same enabling variables
identified in the Behavioral model of resources and community attributes. Specific
potential empirical indicators relevant to CAM health services, products and health
practices are defined for the enabling variables as follows:

1. Resources (personal and other) (potential indicators: income, health insurance,

CAM health coverage, employment, regular source of conventional
healthcare, and regular source of CAM health care“)

2. Community attributes (potential indicators: geographic location and
availability/access to CAM resources“). Specifically, measurable indicators
for geographic location include region of the US, and community of living
defined as rural, urban and suburban neighborhoods. Availability and access
refers to presence of CAM within a community, for example, CAM provider
services offered within conventional healthcare systems, CAM practices
offered at community centers, and CAM product availability in retail stores.
Access to other CAM users is also viewed as important, as word of mouth
recommendations and referrals (referral network) may prompt and facilitate
CAM use (Figure 2).

Need variables.

Need variables in the CAM Healthcare Model are defined under the middle range

level construct of “illness experience” (Figure 2). Potential evaluated need indicators that
can be examined for association with CAM use include morbidity diagnosis, such as

cancer and chronic illness. Potential evaluated need indicators can be further delineated

l7

among those with cancer by examining cancer site, stage, symptoms and treatment as
related to CAM use. Evaluated need indicators to measure relationship between chronic
illness and CAM use include diagnosis, symptoms and treatment. Perceived need
indicators include reported perception of health status, and perceived need for CAM“.
Determining variations in CAM use by specific illness states is an anticipated future use
of the CAM Healthcare Model.
CAM use — outcomes of care

The model construct of health service use in the Behavioral model is modified in
the CAM Healthcare Model to include self-directed CAM use, as well as provider-
directed health services. Thus, the theoretical level construct for the CAM Model is
changed to CAM Health Service - Health Practice Use (Figure 2). Potential empirical
indicators in the CAM Healthcare Model for this construct include visits to CAM
providers, * use of CAM products and resources, "‘ and use of CAM health practices. *
The outcome variable of the Behavioral model, quality of life, although not included in
this study, remains the same in the comprehensive CAM model (Figure 2), as this is also
the desired result of CAM health care interventions. Improved quality of life for an
illness experience of cancer and/or chronic disease is reflected in health promotion and
illness treatment. Whether health is promoted and illness treated can be measured
through the following potential empirical indicators of improved quality of life:
decreased symptom, increased sense of well-being, decreased filnctional limitations,
increased satisfaction, diagnostic verification of condition improvement and increased
perception of control over health as noted in the model (Figure 2). Thus, conceptualizing

CAM within a health service - health practice perspective, the revised model (Figure 2):

18

1. Uses the major constructs within the Behavioral Model as factors influencing
utilization of CAM

2. Adds potential empirical indicators specific to CAM (highlighted above)

3. Modifies the Behavioral model so self-directed CAM health practice and
product use is included as well as provider-directed CAM use under the
model construct of use of services and health practices

It must be noted that the CAM Healthcare Model is an attempt to identify in a
consistent manner those factors that are related to the use of CAM services, products, and
practices. The new model incorporates all of the factors from the original behavioral
model to examine their effect on CAM utilization in a systematic way. In addition, the
CAM Healthcare Model extends the original model by adding those empirical indicators
that are proposed as potential factors influencing CAM use (noted above by asterisk“).
These have been added based on the CAM literature and clinical expertise of the author.
The self-directed aspect of care that is inherent within CAM is emphasized in the new
model. Further, categorizing CAM as use of provider services, practices, and products
will allow for examination of differences in predictors of use in these categories. Thus,
the extended model should help identify more specific information about CAM use in the
selected populations.

In sum, extensive research related to the use of conventional health services using
the Behavioral Model has been documented in the literature. Use of complementary and
alternative medicine as a health service has received less attention. Studies documenting
predictors of CAM use are emerging in the research literature, however theory predicting

the use of CAM health services and practices is lacking. Development of a conceptual

19

model to explain and predict CAM use is a needed innovation at this time of change in
health care service utilization. The CAM Healthcare Model, by identifying measurable
empirical indicators associated with use of CAM, may serve to guide CAM research and
clinical practice. The desire to use the Behavioral Model is motivated by the apparent
strength of its structure, and applicability to modification for a variant form of health
care. The CAM Healthcare Model aims to provide a uniform categorization system that
will promote comparison across studies on CAM. In addition, the model aims to increase
understanding about the relationships among factors associated with CAM use. To
reiterate, Figure 3 represents a fiirther delineation of the CAM Healthcare Model for use
with the NIHS 2002 dataset. The model further becomes more specific in Figure 4 for
use with the cancer population within the 2002 NHIS dataset. The three model figures
show progressive steps that make the model usable for specific purposes.
Using the CAM Healthcare Model with the NHIS 2002 Dataset

The CAM Healthcare Model was used in this research to identify potential
predictors of CAM use in the US. cancer population using the NHIS 2002 dataset. The
estimated cancer population included recent cancer survivors (individuals with a cancer
diagnosis within the year preceding the NHIS interview) and long-term survivors
(individuals with a cancer diagnosis more than one year before the interview).
Preliminary review of the 2002 NHIS dataset revealed that not all empirical indicators
identified in the CAM Healthcare Model were available for analysis. Hence, this
dissertation research examined variables in the model in relationship to CAM use in the
cancer population (Figure 4). The variables used in the analysis are listed in Table 4, and

identified as follows: predisposing variable — gender, age, marital status, education and

20

race; enabling variable — income, insurance status, provider connection (contact with a
conventional healthcare provider); need variables — recent cancer diagnosis, primary
cancer site, pain, fatigue, depression, co-existing co-morbid conditions, and reported
health status. The outcome of the study was CAM use for treatment and health
promotion purposes, rather than quality of life as depicted in the full CAM Healthcare
Model.

Purpose of the Study

This dissertation research has extended what is known about CAM use in the
US. cancer population. CAM use has been examined by aggregating use of all CAM
therapies, and by regrouping CAM use within the alternative categorization proposed.
Thus, patterns of use of provider services, use of products and use of practices have been
examined separately. This research aimed to:

1. Identify attributes and characteristics that distinguish CAM users from non-
users in a cancer population.

2. Compare and contrast patterns of CAM use among recent cancer survivors
(cancer diagnosis during preceding 12 months), and among long-term
survivors (cancer diagnosis more than one year ago).

3. Examine how the type of cancer, time since diagnosis, symptoms experienced,
presence of co-morbid conditions and contact with conventional healthcare
provider affect patterns of CAM use.

4. Identify the "delivery mechanism" for CAM use in the cancer population.

Delivery mechanism refers to the use of CAM health services/therapies that

21

require a service provider as contrasted with CAM practices that are self-
directed.
. Explore purpose of CAM use in a cancer population, defined as use for

treatment and/or health promotion.

22

CHAPTER 2

LITERATURE REVIEW

The review of the literature presents data on use of CAM health services,
practices, and products in the general population and a cancer population. Information is
presented according to the individual determinants of CAM health service - health
practice use as identified in the CAM Healthcare Model and discussed in the Introduction
of this paper. The organization of the literature review is guided by two models: the
CAM Healthcare Model for NHIS Secondary Analysis, and the CAM Healthcare Model
for NHIS Secondary Analysis that is specific to cancer (Figure 3 and 4). The dependent
variables for this research were: use of CAM in general, use of CAM provider services,
use of CAM health practices, and use of CAM products in the US. cancer population.
Predisposing, enabling, and need factors were the independent variables that were
analyzed for ability to predict use of each type of CAM. Empirical indicators for each

study variable are noted in Figure 3 and Figure 4.

CAM Healthcare Model for NHIS Secondary Analysis

Predisposing Variables
Data reveal variations in the relationship of predisposing variables and CAM use
in population studies and smaller scale research. National population studies consistently
reveal that CAM use is more common among women than men (Bausell et al., 2001;
Eisenberg et al., 1998; Lee, Charn, Chew & Ng, 2004; Wolsko, Eisenberg, Davis, Ettner
& Phillips, 2002). Sparber and Wootton (2001), however, note that the difference in

CAM use among women is often marginally significant and in proportion to the gender

23

of those seeking any type of healthcare treatment. Odds of use of CAM provider services
are also noted to be greater among women in some studies (Wolsko et al., 2002). It must
be stressed, however, that the definition of what constitutes CAM often varies across
studies. Being a CAM user in one study may differ from “being a user” in another.
Hence, it is imperative to examine the specific therapies, provider services, products,
and/or practices included in the study definition of CAM use. Higher rates of use are also
noted among middle-aged individuals (Bausell et al., 2001; Eisenberg et al., 1998; Lee,
Charn, Chew & Ng, 2004). More specifically, Eisenberg’s data indicate higher use
among those 35-49 years of age, questionably middle-aged by today’s standards. Higher
prevalence of use was noted in adults aged 30-49 years of age by Bausell and associates
(2001)

The literature points to a predictive effect of social structure attributes such as
education, with higher level of educational attainment associated with increased
likelihood of CAM use in western societies (Astin, 1998; Barnes et al., 2004; Bausell et
al., 2001; Egede, Ye, Zheng & Silverstein, 2002; Eisenberg et al., 1998; Mackenzie,
Taylor, Bloom, Hufford & Johnson, 2003). This, however, is notably different among
eastern societies possibly related to the commitment of the less educated to cultural and
traditional health practices and behaviors inherent in CAM (Lee et al., 2004). This
cultural aspect is relevant given the changing ethnicity of the US. population and the
increasing number of immigrants fi'om the east.

Race has been shown in some national surveys to be associated with CAM use,
with higher use among Caucasians (Eisenberg et al., 1993, 1998). It must be noted,

however, that Eisenberg’s study populations were predominantly Caucasian. In contrast

24

to Eisenberg’s findings, the NHIS of 2002 found that Afiican-American adults were more
likely than Caucasian or Asian adults to use CAM. This racial difference in CAM usage
was noted when megavitamin therapy and use of prayer for health reasons were included
in the definition of CAM. NHIS data provide the first indication that black adults and
Asian adults are substantial users of CAM, with use noted at 71% and 62% respectively
in these subpopulations (Barnes et al., 2004). Secondary analysis of the Medical
Expenditure Survey of 1996 revealed that unlike Hispanics and African Americans,
Caucasians were also more likely to visit CAM practitioners, independent of the effect of
educational attainment (Bausell et al., 2001). Mackenzie, Taylor, Bloom, Hufford and
Johnson (2003) found no difference in CAM use among white, African-American black,
Latino, Asian, and Native American populations in the US. It has been suggested that
CAM use may vary by race and ethnicity when one explores individual therapies and
practices, especially those that are prevalent within a designated culture. For example,
when compared with Caucasians data reveals that: (a) Afiican-Americans, Asian
Americans and Latinos were more likely to use herbs, (b) Asian-Americans were more
likely to use acupuncture and less likely to use home remedies, and (c) Asian Americans
and Latinos were less likely to use chiropractic. The study authors conclude that CAM
use is “prevalent among all ethnic groups, and as such, is probably an important
component of delivering culturally competent care” (Mackenzie et al., 2003, p. 55).
Enabling Variables

Personal, family, and community resource availability and use of CAM have been
explored in the literature to varying degrees. Geographic region has been significantly

associated with CAM use further elucidating the role of regional trends and cultural

25

values on health practices (Bausell et al., 2001; Eisenberg et al., 1998; Lee et al., 2004;
MacLannan, Wilson & Taylor, 1996; Mujaharine et al., 2000). Bausell et al., (2001) note
the highest odds of CAM use were associated with living in the western region of the
US. Insurance coverage for CAM therapies and products, viewed as a resource and thus
a potential enabling factor related to CAM use, varies across regions of the US. Data
reveal that insurance coverage for CAM is related to increased use of CAM provider
services. Wolsko et al., (2002) noted that having fisll insurance coverage for CAM
providers increased the odds of fiequent use, defined as eight visits or more per year, by a
factor of five. Odds of those with partial coverage using provider services were more
than 3 times greater than those with no insurance coverage. Increased demand for CAM
services is expected to follow changes in insurance plans providing a wider scope of
CAM coverage.

CAM therapies also differ fiom conventional health services at the level of out-of-
pocket costs. Whereas most conventional health care therapies are paid for by private
and public insurance across the country, third-party payers infrequently cover CAM
therapies. Increasing reimbursement for CAM health services has been occurring across
the US. during the past decade, a change that coincides with decreasing reimbursement
for conventional health care. Survey data reveal current insurance coverage is limited to
chiropractic treatment, massage therapy, acupuncture and naturopathic medicine, with
differences noted among health plans, policies and practitioner requirements (Cleary-
Guida, Okvat, Oz & Ting, 2001; Lafferty et al., 2004). A billing code system for CAM
therapies (ABC Codes) has been developed and is currently being tested by CAM

providers (altemativelink.com, 2005) in anticipation of increased insurance

26

reimbursement. Movement toward third-party payment is related to establishment of: (a)
evidence of effectiveness, (b) assurance of safety, and (c) cost-benefit ratio.

Some studies on CAM have examined the relationship of income level and use,
since CAM often requires out-of-pocket expenditures. Cherniack, Senzel and Pan (2001)
found no correlation between income and use of CAM among elderly individuals in an
urban population. Income was also not a significant predictor of CAM use among
individuals with cancer (Boon, Westlake, Stewart, Gray, Fleshner et al., 2003;
Fouladbakhsh, Stommel, Given & Given, 2005).

By contrast, one disabling supply factor that affected use of CAM may be an
individual’s rural residence. The literature indicates rural areas often have less developed
health service infrastructures. Thus, distances to source of care, which affect the use of
conventional health care services, should also be of consequence for access to CAM
providers, products, and therapies. In fact, in a study of patients with pain (N=595), place
of residence was significantly related to use of CAM, with highest use in suburban
communities (82%) (Vallerand, Fouladbakhsh & Templin, 2003). CAM use in urban and
rural communities was 77% and 58%, respectively. The CAM therapies most frequently
used were herbal products and supplements, with highest use among suburban
participants and lowest use among rural residents. NHIS data reveal almost 63% of
urban residents and 60% of rural residents use CAM (Barnes et a1, 2004) when prayer for
healing is included. In part, such variations go back to the definition of CAM, which
includes not only personal services and practices such as self-prayer, but also products
that can be mailed, or otherwise shipped to users in remote areas. In fact, some

researchers have argued that distance from health care providers, as experienced by those

27

living in rural areas, increases the potential for self-care and self-treatment (Bartlome,
Bartlome, & Bradham, 1992). This may provide an impetus to seek out CAM, which is a
mode of treatment that relies more heavily on self-care. Thus, the impact of rural-urban
residency on CAM use remains uncertain, involving numerous factors such as
availability, access, and affordability of conventional as well as CAM services, and
individual tendency to engage in self-care behaviors.

Provider connection, that is having a relationship with a healthcare provider,
whether conventional or CAM, is viewed as an enabling factor with the potential to
predict CAM use. Seeing a conventional provider may enable or impede CAM use. Use
patterns may be influenced by physician and/or nurse (a) attitudes and knowledge about
CAM efficacy and safety, and (b) subsequent ability and willingness to refer to CAM
providers, and to recommend products and/or practices. In addition, having a
relationship or connection with a CAM provider is viewed as an enabling factor, often
serving as a source of CAM referrals for treatment or health promotion. Further, the
literature indicates that concurrent use of conventional medical doctors and CAM
practitioners has increased significantly over time. CAM use for chronic illness often
serves as an adjunct to physician visits, with few CAM users seeking solely alternative
care. Concurrent use of both types of providers has been noted to vary by age, gender,
chronic conditions, long-term disability, hospital days, psychosocial distress, and spiritual
values. Interestingly, some studies reveal that more men than women visited both CAM
and conventional providers for health services. Individuals under the age of 65 with one
or more chronic medical conditions were also significantly more likely to see both types

of providers. Thus, data reveal a pattern of concurrent use of CAM and conventional

28

health services for common chronic illnesses and related symptoms (Muhajarine,
Neudorf & Martin, 2000).
Need Variables

Among individuals with cancer, current estimates of use vary widely from 7% to

 

64% (Ernst & Cassileth, 1998), with 30% of women and 28% of men either continuing or
beginning to use CAM therapies after becoming ill with cancer (Salmenpera, 2002).
National survey data reveal that having cancer increased the odds of seeing a CAM
provider by a factor of three (Wolsko, Eisenberg, Davis, Ettner & Phillips, 2002). CAM
in the cancer population has been widely documented in the literature examining an array
of different complementary therapies (Alferi, Antoni, Ironson, Kilbourn & Carver, 2001;
Jordan & Delunas, 2001; Patterson, et a1, 2002; Salmenpera, 2002). However, the
specific measures of CAM use, the particular services and practices included, and the
populations examined, vary widely fiom study to study, making comparisons of results
somewhat hazardous.

The frequent use of CAM for chronic illness and health-related problems is
supported throughout the literature (Astin, 1999; Barnes et al., 2002; Bausell et al., 2001;
Eisenberg et al., 1993, 1998; Wolsko et al., 2002). Informed consumers for symptom
management often seek CAM health care for long-term chronic illness. Population data
reveal that 34% of study respondents reported medical conditions, with the highest
condition specific use of CAM for neck and back problems (Eisenberg et al., 1998).
Wolsko et al., (2002) found back and neck problems to be significant predictors of use of

CAM provider services. CAM therapies most often used for chronic medical conditions

29

included chiropractic treatment, relaxation techniques, and massage therapy (Eisenberg et
aL,1998)

Population data from the 1996 Medical Expenditure Panel Survey (MEPS)
identified three broad categories of the most fi'equently occurring chronic illnesses and
health problems for which CAM was used. These categories include the following:

(a) musculoskeletal (e.g., arthritis, back, and joint disorders), (b) mental (e.g., anxiety-
somatoform-dissociative, affective, malaise-fatigue, stress related disorders), and (c)
metabolic (e. g., diabetes and endocrine disorders). Individuals who experienced health
problems within these three categories were more than three times more likely to use
CAM. Likelihood of use increased with co-morbidity, that is, having health problems in
more than one category (Bausell et al., 2001). Data further reveal an extensive range of
chronic illnesses among CAM users. The authors concluded that the highest users of
CAM therapies provided by practitioners are individuals with co-morbid, non-life
threatening health problems. Egede et a1, (2002) also found that individuals with
diabetes, cancer, and hypertension were more likely to use CAM as compared with
individuals with no chronic illness.

The literature assures us that the course of chronic illness is often complex and
unpredictable; individuals experience a vast array of symptoms, often of long-term
duration and with great personal, emotional, and psychological cost. A wide range of
CAM practitioners, therapies, and practices are often used for relief of symptoms I
associated with chronic illness (Astin, 1998; Egede et al., 2002; Eisenberg et al., 1998;
Ernst, 1998; Vallerand & Fouladbakhle 2003, 2004; Willison & Andrews, 2004;

Wolsko, Eisenberg, Davis, Kessler & Phillips, 2003). Faced with the lack of a cure from

30

conventional medicine, those with chronic illness may experience a sense of helplessness,
thereby seeking other avenues of care for symptom management to improve quality of
life. In addition, CAM is viewed as a component of self-care management, suggesting
the need for personal responsibility for health while experiencing chronic illness (Thorne,
Paterson, Russell & Schultz, 2002). Further, the goal of self-healing through
enhancement of psychological and physical well being often contributes to the decision to
use CAM by those with chronic illness (Ritvo, Irvine, Katz et al., 1999).

With increasing age, the prevalence of chronic co-morbid conditions also
increases, resulting in additional symptom burden. It is estimated that 80% of older
adults have at least one chronic condition and up to 50% have two conditions that cause
pain, disability, and fiinctional limitations (Gerberding, 2006). Millar (1997) found that
elders reporting three of more chronic conditions (26%) were more likely to consult with
CAM providers than those with only one chronic condition (9%).

Symptom Experience

Pain, which affects millions on a daily basis, whether attributed to chronic illness,
or cancer and its treatment, proliferates throughout modern society despite widespread
medical, political, and legal efforts directed at management. Data reveal that the under-
treatment of pain is a major public health problem within the US. Research findings
indicate that more than 60% of the general population has experienced pain for more than
five years; for 40% this pain is constant. Although more than 90% have seen a
conventional medical doctor, pain persists, and quality of life is affected. Most state their
daily life is altered by pain, and that prescription analgesic treatment is less than very

effective (Lazarus & Neumann, 2001 ). It is further noted that conventional healthcare

31

treatments for pain focus almost solely on medication, a method acknowledged to be
potentially very effective for pain control. This conventional approach however, often
fails to acknowledge the (a) inherent side effects that disrupt quality of life while abating
pain, (b) cost, both financial and psychosocial, and (c) individual belief systems and
values about pain and its treatment.

CAM treatment for cancer pain, fatigue, anxiety, and depression is currently

 

receiving increased attention from the National Cancer Institute (N C1), the Office of
Cancer Complementary and Alternative Medicine (OCCAM) and NCCAM with
numerous clinical trials in progress (http:l/clinicaltrials.gov). A summary of data
includes the following findings on CAM and cancer-related symptoms. Auricular
acupuncture has been noted to decrease cancer pain intensity by 36% among patients
receiving analgesics (Alimi, Rubino, Pichard-Leandri, Fermand-Brule, Dubreuil-Lemaire
et al., 2003). Clinical trial data also found yoga to be effective in reducing sleep
disturbances, improving overall sleep quality and duration, however not affecting fatigue,
anxiety, or depression (Cohen, Warneke, Fouladi, Rodriguez & Chaoul-Reich, 2004).
Healing Touch and massage were more effective than standard care among cancer
patients undergoing chemotherapy in reducing pain, mood disturbances, and fatigue
(Post-White, Kinney, Savik, Gau, Wilcox et al., 2003). Massage provided for cancer
outpatients resulted in a 50% reduction in pain, fatigue, stress, anxiety, nausea, and
depression that lasted over a two day time period (Cassileth & Vickers, 2004).

Prevalent chronic illnesses, such as rheumatologic conditions affect a significant
percent of the US. population, resulting in chronic pain and functional impairment (Rao,

Mihaliak, Kroenke, Bradley, Tierney & Weinberger, 1999). Studies of individuals with

32

arthritis, and other rheumatologic conditions indicate 40-90% have used CAM therapies
(Arcury etal., 1996; Anderson et al., 2000; Wilson et al., 1999; Rao, et a1, 1999). Severe
pain is also noted as a significant predictor of regular CAM use in this population, with
frequent use associated with illness duration (Rao et al., 1999). The most commonly
used CAM therapies and practices in studies of individuals with rheumatologic
conditions include: prayer, relaxation, positive thinking, exercise, massage, and hot-tub
use (Wootten & Sparber, 2001). These authors suggest that these frequently used
therapies and practices indicate “a yearning for less stress, more relaxation, and a greater
emphasis on care in health care” (Wootten & Sparber, 2001, p.720).

Pain and the need for pain relief are also very common experiences among the
increasing aging population, most of whom endure chronic illnesses (N ajm, Reinsch,
Hoehler & Tobis, 2003; Vallerand & F ouladbakhsh, 2004). Data indicate older adults’
reasons for using CAM often include pain relief, improved quality of life, maintenance of
health and fitness, stress relief, and prevention (Williamson, Fletcher & Dawson, 2003).
Commonly used CAM therapies include chiropractic treatment, herbal medicine, dietary
supplements, massage, and acupuncture (Najm et al., 2003; Williamson et al., 2003).
Variations in use of CAM therapies have also been noted among the ethnic elderly. Pain
was noted to be a higher indicator of CAM use among Asians, whereas gastrointestinal
problems and diabetes prompted use among Hispanics. Stress, fatigue, and
cardiovascular problems were associated with CAM use among white non-Hispanics
(Najm et al., 2003). Estimates of population grth indicate an ongoing shift in the
increasing proportion of elders in the US, among them the ethnic elderly. The vast

majority of elders will develop chronic illness with accompanying symptoms, oo-

33

morbidity and fiinctional limitations, all impacting quality of life. The sheer number of
these older adults is expected to influence health care services. Thus, their health
behavior, cultural traditions, and subsequent treatment choices require further
illumination.

Population studies also indicate that back and neck pain are prevalent health
conditions that result in considerable morbidity, disability, functional limitations, lost
revenue, and psychological distress. It is estimated that almost 33% of all visits to CAM
providers were for treatment of back and neck pain (Eisenberg et al., 1998; Wolsko et al.,
2003), with more than 50% of individuals with this type of pain seeking CAM treatment.
Frequently used CAM therapies include chiropractic treatment, massage, and relaxation,
all rated as very helpful by CAM users. Effectiveness of massage as a CAM therapy for
chronic pain, back pain, and musculoskeletal disorders is widely documented in the
literature (Cherkin, Eisenberg, Sherman, Barlow, William et al., 2001; Walach, Guthlin
& Konig, 2003). Clinical trials identify long-term benefits of massage lasting up to one
year with subsequent reductions in conventional health care utilization for this prevalent
chronic condition. In addition, use of prescription pain medication and costs of
outpatient care were significantly lower for chronic back pain patients who received
massage therapy. Study results suggest that the cost of CAM therapy use may be offset
by the reductions in subsequent conventional healthcare costs (Cherkin et al., 2001).
Systematic reviews of acupuncture as a CAM treatment for chronic low back pain
conclude that acupuncture, when used as a complement to other conventional therapies,

relieves pain, and improves functioning (Furlan, van Tulder, Cherkin, Tsukayama, Lao &

34

et al., 2005). It is noted that complementary use is often more effective than conventional

therapies alone.

CAM Healthcare Model for NHIS: Cancer Specific

Predisposing Variables

In terms of predictors of CAM use, most studies have examined predisposing
variables. In a study of breast, colon, and prostate cancer patients (Patterson et al., 2002),
the reported odds of using a CAM therapy were more than 2.5 times greater among the
female participants (p< 0.05). The odds of a female patient seeing a CAM provider were
even greater, exceeding those of a male by a factor of 5.5. Spiegel et al., (2002) also
found a significantly higher number of female oncology patients used CAM than male
patients (33% & 20.5% respectively). This pattern of more frequent CAM use by women
is consistent with study results from the National Health Interview Survey (Barnes et al.,
2004). Predictors of CAM use by cancer patients also include younger age and higher
education (Alferi et al., 2001; Burstein, 1999; Lee et al., 2000; Richardson et al., 2000).
Variations in use by ethnicity have also been documented in cancer and general
populations (Alferi et al., 2001; Barnes et al., 2004; Lee et al., 2000) with higher use
among Afi'ican-American women. While not all study results support the notion of CAM
use rates varying by age (Shen et al., 2002), education, and ethnicity, there appears to be
strong support that women are more likely to be CAM users (Fouladbakhsh et al., 2005;

Lengacher et al., 2002; Sparber et al., 2001; Speigel et al., 2002).

35

Enabling Variables

The impact of enabling variables on CAM use has been reported in the literature.
Lee et al. (2000) found higher income to be a significant predictor of CAM use by
women of varying ethnicity. Other data, however, have not supported a relationship
between the enabling variables of income and family support identified in the Andersen
model, and CAM use by cancer patients (Shen et al., 2002). Having a caregiver, also
viewed as a potentially enabling factor, was unrelated to use of CAM among community
dwelling cancer patients (Fouladbakhsh et al., 2005). While current enrollment in health
insurance is a potent predictor of the use of conventional health services, it is less clear to
what extent insurance coverage is relevant to services that are generally not covered.

Need Variables

Among the need variables identified for this study, cancer site is often reported in
the literature, with most frequent CAM use by breast cancer patients (Boon, Stewart &
Kennard, 2000; Morris, Johnson, Homer & Walts, 2000; Richardson, Sanders, Palmer,
Greisinger & Singletary, 2000). Results concerning the impact of cancer staging on
CAM use are mixed, with some studies reporting greater use among late stage cancer
patients (Lee, 2000; Shen et al., 2002), and others showing no association with use of
alternative therapies (Patterson, 2002). Fouladbakhsh et al., (2005) found patients with
early stage cancer were more likely to use CAM. Cancer treatment has been reported in
association with CAM use, with more than half (56%) of patients using at least one
complementary therapy during conventional cancer treatment (Alferi et al., 2001).
Dissatisfaction with medical treatment, pain, emotional distress, concern about cancer,

and expectation of recurrence were unrelated to use. Data reveal patients treated with

36

surgery and chemotherapy were more likely to use CAM (Fouladbakhsh et al., 2005;
Richardson, 2002). Research indicates cancer patients use CAM to enhance benefits from
conventional cancer treatment and to improve general well being with use significantly
associated with receiving multiple cancer treatments (Alferi et al., 2001; Patterson et al.,
2002; VandeCreek, Rodgers, & Lester, 1999). Jacobson, Workman and Kronenberg
(2000) note that breast cancer patients are increasingly seeking CAM, on their own as
well as through conventional healthcare providers, to improve chances of survival,
decrease risk of cancer recurrence, and relieve cancer and treatment related symptoms.
CAM therapies used for treatment side effects of cancer, and symptom management
include acupuncture for chemotherapy associated nausea and vomiting, massage therapy
for post-mastectomy lymphedema and pain, and mind/body therapies to reduce stress and
anxiety related to illness state and treatment measures. lData reveal that most cancer
patients experience a vast array of symptoms related to illness and treatment strategies
(Fouladbakhsh et al., 2005). Frequent symptoms include fatigue, anxiety and pain, with
number of symptoms (three or greater) being a significant predictor of CAM use
(Fouladbakhsh et al., 2005). Data indicate that fatigue is often the most debilitating
cancer related symptom, with more frequent use of healthcare services, including CAM
use, among those who experience this symptom (Ashbury, Findlay, Reynolds &
McKerracher, 1998).

In sum, the current state of the existing literature on CAM use provides clues and
partial evidence, but the complex patterns of use of nonconventional therapies, either as
alternatives to or complements of conventional medical approaches require further

illumination. This study offers a further exploration of how predisposing, enabling and

37

need variables affect CAM use for health promotion and illness treatment among those
with illness and related symptoms. For this secondary analysis, CAM use is defined as
including 21 of the 22 CAM health services, products, and practices included in the NHIS
dataset (Table 1). Prayer for health related purposes is not included in the dissertation
analysis. CAM use patterns will be examined in three different ways as follows: (a)
CAM use in general, i.e., having used any one of the 21 identified CAM therapies, (b)
CAM use according to the alternative categorization presented in the Introduction
section, that is, use of CAM health services, use of CAM products and/or use of CAM
health practices, and (c) CAM use of services, products, and practices for the purpose of
health promotion and/or illness and symptom treatment. Thus, this dissertation aims to
address the following research questions and hypotheses for the cancer population

surveyed in the 2002 NHIS.

Research Questions and Hypotheses

Predisposing Variables
§§n_de_r
Question 1: Is there a gender difference in the use of CAM health services, CAM
health practices, and CAM products for illness/symptom treatment and
health promotion?
Hypothesis 1: Overall, women are more likely to use CAM health services, CAM health
practices and CAM products for treatment and health promotion than men.

Age, race & marital status

Question 2: Are age, race, and marital status related to use of CAM across all three

38

categories of use?

Hypotheses 2a: Whites have a greater likelihood of using CAM provider services, CAM
health practices and CAM products than Afiican Americans, Hispanics, or
Asians.

Hypotheses 2b: Age is predictive of use of CAM across all three categories of CAM use,

with increased likelihood of use among those 35 to 55 years of age.

Enabling Variables

m

Question 3: What is the relationship of income to use of CAM health services, CAM
practices and CAM products

Hypotheses 3a: Higher income is associated with increased likelihood of use of CAM
therapies illness/ symptom treatment and health promotion that must be
purchased (that is, health services and products) for

Hypotheses 3b: There is no association between income level and use of CAM health

practices that do not require a purchase for their performance.

Education

Question 4: Is level of education attainment related to use of CAM health services,
CAM practices and CAM products?

Hypothesis 4: There is a positive association between level of education completed and

use of CAM across all three categories (provider services, practices and products).

39

Insurance

Question 5:

Hypothesis 5:

What is the relationship between conventional health insurance coverage
and CAM use?

Individuals without health insurance coverage are more likely to use CAM
health services, practices and products for both health promotion and

illness/ symptom treatment.

Concurrent use of conventional providers and services

Question 6:

Hypothesis 6:

Cancer site

Question 7:

Symptoms

Question 8:

How do patterns of CAM use correlate with use patterns of conventional
health care?

Individuals who have visited conventional medical providers during the
preceding year are more likely to use CAM for illness/symptom treatment

and/or health promotion.

Need Variables

Is cancer site associated with likelihood of use of CAM health services,

practices, or products?

Are reported symptoms (pain, fatigue, anxiety, and depression) associated
with likelihood of use of CAM health services, practices and products?
Is there a symptom threshold (number of reported symptoms) associated

with likelihood to use CAM?

Hypotheses 8a: Pain, fatigue, anxiety, and depression are significant predictors of use of

CAM health services, practices and products.

40

Hypotheses 8b: There is a positive association between number of reported symptoms
and likelihood of CAM use.

Co-existing co-morbid conditions

Question 9: Do patterns of CAM use, that is use of services, practices, and products,
differ among individuals depending on the number and types of co-morbid
conditions?

Hypothesis 9: Having a co-existing co-morbid condition increases likelihood of use of
CAM health services, practices and products.

Time since cancer diagnosis

Question 10: Is time since cancer diagnosis, defined as recent diagnosis (within past
year) and survivor diagnosis (more than one year ago), associated with

differences in CAM use patterns and reasons for use?

41

CHAPTER 3

METHODOLOGY

Description of NHIS 2002

This dissertation used data from the National Health Interview Survey (NHIS,
2002) for a secondary analysis focused on patterns and predictors of CAM use in a cancer
population in the US. The NHIS is a multi-purpose health survey conducted on a regular
basis by the National Center for Health Statistic (N CHS), Centers for Disease Control
and Prevention (CDC), to provide information on the health of the adult civilian,
noninstitutionalized, household population in the US. The 2002 cross-sectional NHIS
contains the Alternative Medicine/Complementary and Alternative Medicine (ALT)
supplement, asking adult respondents (18+) about their use and experience with 22 types
of CAM therapies, products, and practices (Table 1). Data fiom this supplement were
merged with the regular NHIS sample adult file, as well as parts of the Family-level and
Person-level files, to access additional information on the respondents' health status,
history of health conditions, access and utilization of conventional health care,
sociodemographic information and information on income and health insurance coverage

(NHIS Survey Description, 2003).

Data Collection Procedures

The NHIS methodology has been well described in the literature (Botman et al.,
2000). Specifically, the 2002 NHIS employs a multistage probability cluster sampling

design, representative of the NHIS target universe, which was defined as "all dwelling

42

units in the US. that contain members of the civilian noninstutionalized population.”
(NHIS Survey Description, 2003). In the first stage, 339 primary sampling units (PSUS)
were selected from approximately 1900 area sampling units representing counties, groups
of adjacent counties or metropolitan areas covering the 50 states and the District of
Columbia. The selection included all of the most populous PSUs in the US. and
stratified probability samples (by state, area poverty level and population size) of the less
populous ones. In a second step, PSUs were partitioned into substrata (up to 21) based on
concentrations of Black and Hispanic populations. In a third step, clusters of dwelling
units form the secondary sampling units (SSUs) selected from each substratum. Finally,
within each SSU, all Black and Hispanic households were selected for interviews, while
other households were sampled at differing rates within the substrata. Census
interviewers who were trained and directed by survey supervisors within the US. Census
Bureau Regional Offices across the US. conducted household interviews. A computer-
assisted personal interviewing (CAPI) system was used allowing interviewers to read
items fiom and enter responses directly into the computer during interviews. Total
household response rate, that is total number of responding households divided by the
total number of eligible households, was reported at 89.6%. Household was defined as
two more related persons living together in the same housing unit, although in some

cases, unrelated persons living together fit these criteria.

Sample

The sample for the secondary analysis included information provided by 31,044

adult respondents to the NHIS household interviews of 2002. Of these respondents, 2262

43

reported a diagnosis of cancer by a physician, 461 of whom received a cancer diagnosis
within the preceding twelve months. NHIS interview questions asked about the diagnosis
of thirty different types of cancer; responses indicating multiple types and sites of cancer
were allowed. The current analysis was focused on different subgroups of the 2002 NHIS
survey:
1. CAM users diagnosed with cancer:
a. Recent survivors: diagnosed within the past 12 months
b. Long-term survivors: diagnosed more than one year before the NHIS
interview
2. Users of CAM provider services
3. Users of CAM practices

4. Users of CAM products

Instruments

Instruments used in the NHIS are discussed according to predisposing, enabling,
need variables, and the outcome categories identified in the model that guided this
dissertation research.

Predisposing and Enabling Variables

Extensive demographic data, information on social structure attributes, resources,
and community attributes was available in the NHIS, 2002. This included age, gender,
marital status, education, race, ethnicity, income, insurance coverage, and information
about conventional healthcare providers. This data were used to determine the

relationship of these variables to CAM use in the estimated total US. cancer population.

44

 

The Socio-demographic Section (FSD) of the NHIS Person-level file contained items that
were used to determine educational attainment of survey respondents.

The Health Care Access and Utilization Section (FAU) of the Person-level file in
the NHIS contained information about health care and utilization services. Data on visits
to conventional doctors’ offices, as well as use of other conventional health services,
were available in this section for analysis. The Health Insurance Section (FHI) of the
Person-level file contained a full range of items addressing health insurance. The item in
this section of relevance to this dissertation was type of health care coverage, which
included: Medicare, Medicaid, military/V A, CHAMPUS/TRICARE/CHAMP-VA, State
sponsored health plan, Indian Health Service, other government programs, private
insurance, and single service plans.

Need Variables

Information was available on participant health status in the Health Status and
Limitation of Activity Section (FHS) of the Person-level file; additional information was
also included in the Sample Adult file and the Alternative/Complementary and
Alternative Medicine (ALT) supplement. Information on health conditions including
cancer and a wide range of chronic illnesses and reported symptoms was available for
analysis of need predictors in the model. Cancer was further delineated by site, and the
chronic illness list followed the [CD-9 codes. In addition, NHIS participants who
indicated use of a CAM therapy, practice, or practitioner were asked to identify three
specific health conditions for their use of CAM. This data were available in the ALT
section. An NHIS item asking respondents to rate their health status was used to measure

perception of health.

45

 

 

 

CAM Use -0utcomes

Responses to a comprehensive set of items related to the use of complementary
and alternative therapies, products, practices and practitioners were available in the NHIS
2002 dataset. Participants were asked whether they used any of the following CAM
therapies, products, or practices: acupuncture, avurvedic treatment, biofeedback,
chelation, chiropractic treatment, special diets, energy healing, folk medicine, herbs,
homeopathy, hypnosis, massage, naturopathy, vitamins, prayer and healing rituals,
progressive relaxation, deep breathing, guided imagery, meditation, yoga, Tai Chi, or
Qigong (Table 1). Subcategories of two of these items were also included. Use of diet
asked participants about six different diets, and use of prayer had four separate categories
to which participants could respond (Table 1). All respondents were asked to identify up
to three conditions for CAM treatment, and whether CAM was found to be helpfiil.
Responses to 21 of the CAM use item questions were used for this study. Prayer was not

included in this analysis.

Study Variables

Dependent Variables
The dependent (outcome) variables used for this analysis were:
1. Overall use of CAM with 2 categories: users vs. non-users
2. Type of CAM used with 5 categories: providers only, products only, practices
only, combined types of CAM, no use at all
3. Purpose of CAM use with 4 categories: CAM used for treatment only, CAM

used for health promotion only, CAM used for both treatment and health

46

promotion, no CAM use.

Dependent Variable I

The first dependent (outcome) variable for this analysis was a simple dichotomy:
use or nonuse of any of the identified CAM providers, practices or products listed in the
ALT section of the NHIS. A positive response to at least one of the 21 CAM USE items
in the ALT section of the NHIS determined a respondent’s classification as a “user of
CAM,” whereas a negative response to all of these items resulted in the classification of
the respondent as a “nonuser.” Use of prayer for health reasons in the original NHIS list
of 22 items was n_ot included.
Dependent Variable 2

The second dependent variable focused specifically on the use of specific
categories (types) of CAM used according to the model presented in the Introduction
section and the recategorization presented in Table 3. This was possible with the NHIS
dataset as 10 items in the ALT section asked respondents if they saw a provider for
specific CAM treatments. These included “saw a provider for” the following treatments:
acupuncture, ayurvedic, biofeedback, chelation, chiropractic, energy healing, folk
medicine, hypnosis, massage, and naturopathic. These items are coded 1=yes and 2=no
for CAM use of provider services. The remaining 11 items on CAM use in the ALT
section that were used for this analysis were further subdivided according to the model
and are presented in Table 3. Use of CAM practices included responses to eight items
asking about use of any of the following: diets, yoga, tai chi, qigong, meditation, guided
imagery, progressive relaxation, and deep-breathing exercises (1=yes, 2=no). Use of

CAM products included responses to the remaining three items in the ALT section of the

47

NHIS asking about use of herbs, vitamins (high dose) and/or homeopathic treatment
(1=yes, 2=no).
Dependent Variable 3

The third dependent variable, purpose of overall use of CAM, was also analyzed.
Responses to each of the 21 CAM items, indicating if CAM use was “to treat a specific
condition” (coded as 1=yes, 2=no) were analyzed. For the responses indicating CAM
was LLQILISLCI “to treat a specific condition,” the purpose of CAM use was considered to
be “for health promotion.” Thus, a four-category variable was used in this analysis to
examine CAM usage patterns: (a) use only for treatment, (b) use M for health
promotion, (c) mixed usage (some CAM therapies used for treatment, some used for
health promotion), and (d) no CAM use.

Independent Variables

The independent (predictor) variables used in the analyses are discussed according to the
theoretical constructs in the model.
Predisposing and Enabling Variables

Variable indicators included in the analysis are: gender (male=1, female=2), age
(18-29, 30—44, 45-59, 60-75, > 75), education (high school or less, some college/associate
degree, bachelor/master’s degree, doctorate), marital status (married=1, widowed=2,
divorced=3, separated/single=4), and race/ethnicity (white=1, black/African American=2,
Indian American =3, Asian-Indian=9, Chinese=10, Filipino=11). Empirical indicators for
the enabling predictor variables are: insurance (yes= l, no=0), income (under $20,000 =

1, $20, 000 and over =2), and healthcare provider contact (yes =1, no=0).

48

Need Variables

Independent variables in the need category of the model were measured with the
following empirical indicators: cancer diagnosis ever (yes=1, no=2), recent cancer
diagnosis (yes=1 , no=0), cancer survivor (yes=], no=0), pain (yes=1,no=0), fatigue
(yes=1, no=0), depression (yes=l, no=0), reported health status (excellent=1, very
good=2, good=3, fair=4, poor=5), co-existing co-morbidity (yes=1, no=0), and number of
chronic co-morbid conditions. Chronic illness was defined as having one or more of 32

specified non-cancer co-morbidities in the past 12 months as listed below:

Hypertension Poor circulation Stroke
Coronary heart disease Diabetic retinopathy Angina

Heart condition Irritable bowel Asthma
Emphesema High cholesterol ‘ Glaucoma
Congestive heart failure Bowel problems Ulcers
Thyroid problems Urinary problems Diabetes

Skin problems Allergies (hay fever) Sinus problems
Allergies (food/odor) Allergies (medications) Arthritis
Kidney problems Chronic bronchitis Liver problems
Gynecological problems Prostate problems Heart attack
Cataracts Macular degeneration

Information on chronic conditions was available in the 2002 NHIS dataset. Respondents
who reported that they had one or more of the following conditions were identified as
having a chronic health problem/ illness. For more information on the independent

variables refer to Table 4.

Procedure

This secondary data analysis made use of the public use files available for

downloading on the NCHS website. The data are available in ASCII format with

49

separate command files already prepared by NCHS to convert them into STATA data
files. Protection of human rights of study participants was assured through the Michigan
State University Committee on Research Involving Human Subjects (U CRIHS).
Approval was granted on Nov. 10, 2005, with exempt status. NHIS data has been
de-identified to protect subject rights and confidentiality of responses. The sampling
design of the 2002 NHIS allowed for representation of women and minorities in the

dataset, with over-sampling to promote minority representation.
Data Analysis

The complex sampling design of the NHIS required special methods of variance
estimation in the analysis, since, with many multistage designs, weighted parameter
estimates are weighted functions of several random quantities. Currently, three
alternative methods of variance estimation are used under these circumstances:
linearization, the jackknife and balanced repeated replications (Levy & Lemeshow, 1999;
Korn & Graubard, 1999). The sampling design and appropriate weighting information
for the 2002 NHIS is contained in three variables (stratum, psu, wtfa) of the public
release files, which can be used for correct parameter estimations.

All statistical analyses were carried out using the STATA 9.2 software (Special
Edition). The “svy” commands of STATA contain a comprehensive set of estimation
procedures ranging from mean, proportion and ratio estimation to linear, logistic, I
multinomial logit, and ordered logit regression models. All procedures are implemented
for variance estimation involving linearization, the jackknife and balanced repeated

replications. Furthermore, the “subpopulation” estimation command allowed for the

50

analysis of subsetted data, taking full advantage of the complete sampling design
information in the data. Binary and multinomial logit regressions were the primary
statistical models employed in the analysis, focusing on between- subject differences in
CAM use.

Steps in the analysis process included initial identification of subpopulations to be
studied. Descriptive analyses were conducted using cross tabulation and Pearson Chi
square computation for the (a) cancer population as compared to the non-cancer
population, and (b) recent cancer survivors as compared to long-term survivors. Binary
and multinomial logit regression analyses proceeded in several stages. Using the CAM
Healthcare Model for the NHIS Secondary Analysis: Cancer Specific (Appendix B.
Figure 4), the independent variables were entered into the logistic regression model to
identify ability to predict overall CAM use. Multinomial logit regression analyses were
also used contrasting treatment users to non-users, health promotion users to non-users,
and combined treatment/health promotion as compared with non-users. Significance
level of p< 0.05 was used to determine significant predictors of CAM use in the identified
subpopulation. Adjusted odds ratios for significant predictors reflect the estimated

likelihood of use among the population groups.

51

CHAPTER 4

DATA ANALYSIS AND RESULTS

Analytical Procedures

Data from the National Health Interview Survey (NHIS, 2002) were used as the
source for this secondary analysis. The Alternative Medicine/Complementary and
Alternative Medicine (ALT) supplement of the 2002 NHIS, asking 31,044 adult
respondents (l 8yrs and older) about their use and experience with 22 types of CAM
therapies, products and practices was merged with the regular NHIS sample adult file.
Parts of the Family-level and Personal-level files were also merged to access additional
information on the respondents' health status, limitations in daily activity, history of
health conditions, access and utilization of conventional health care, socio-demographic
information and information on income and assets. For this dissertation, the analysis
focused on the subset of the sample that had reported a cancer diagnosis within the past
year (recent cancer survivor) or at some point in their lifetime (long-term survivor) (N =
2262). This sub-sample represented an estimated 14,304,621 individuals in the US. who
had been diagnosed with cancer prior to the interview. That number equaled almost 7%
of the total estimated adult (age 18 and over) US. population of more than 205 million in
2002.

The complex sampling design of the NHIS required special methods of variance
estimation in the analysis, since, with many multistage designs, weighted parameter
estimates are weighted functions of several random quantities. The sampling design and

appropriate weighting information for the 2002 NHIS is contained in three variables

52

(stratum, psu, wtfa) of the public release files, which were used for correct parameter
estimations. All statistical analyses were carried out using the STATA 9.2 software
(Special Edition). The 'svy' commands of STATA, which contain a comprehensive set of
estimation procedures, were used for calculation of descriptive summary statistics
(frequencies, means, proportions, and ratio estimation) and adjusted odds ratios using
logistic regression and multinomial logistic regression to determine significant predictors
of CAM use in the cancer population (Agresti, 1990; Hosmer & Lemeshow, 2000). All
procedures were implemented for variance estimation using linearization (Levy &
Lemeshow, 1999). Two-way tables using the “svy” commands were created using
Pearson Chi Square analysis and F values to determine significant differences among
subpopulations. The “subpopulation” estimation command allows for the analysis of
subsetted data, taking full advantage of the complete sampling design information in the
dataset. Binary and multinomial logit regressions were the primary statistical models
employed in the analysis (Hosmer & Lemeshow, 2000), focusing on between-subject
differences in CAM use. A stepwise procedure was followed in which potential predictor
variables were excluded from the model if their p-value exceeded 0.10.

Descriptive information comparing the estimated cancer population and the non-
cancer population in the US. is presented in Table 5, with the appropriate confidence
intervals for the population estimates. For some variables the estimated totals are
somewhat smaller than the overall estimates for the cancer and non-cancer populations,
which is due to missing information as noted in the table. Descriptive information on
recent cancer survivors, as juxtaposed to the long-term cancer survivors, is presented in

Table 6. F-values and significance levels are also provided indicating significant

53

 

 

differences between the subpopulations compared in both tables. The discussion of

results will be presented in the following order:

1.

Descriptive statistics for the independent model variables grouped as
predisposing, enabling, and need with comparisons across subpopulations
(Tables 5 and 6),

Descriptive statistics for the dependent variables of overall CAM use, CAM

types, and CAM purpose (Tables 7 to 9),

. Logistic regression analysis of the overall CAM use variable (Table 10);

Multinomial logit analysis of specific categories of CAM use
(Tables 11 and 13), and

Multinomial logit analysis of reasons for CAM use (Tables 12 and 14).

Model Variables - Descriptive Statistics

Predisposing Variables

The estimated U.S. cancer population in 2002 included a higher percentage of

women (56.9%) than men (43.1%). As the 95% confidence intervals (CI) indicate, these

estimates have a margin of error of plus or minus 2.3% (Table 5). This difference by

gender was also evident in the recent survivor and the long-term survivor groups, which

represent more than 2.9 and 11.3 million individuals respectively (Table 6). Although the

two survivor groups did not differ significantly by gender (p < 0.08), the gap between

males and females is larger by more than ten percentage points in the long-term survivor

subpopulation (Table 6). This reflects increased longevity among women and higher

long-term survival rates for breast cancer (Health, US, 2005). In contrast, the

54

population estimates for gender in the (younger) non-cancer population are more
balanced than in the cancer population with 48.3% males and 51.2% females in the US.
adult population. Overall, the proportion of cancer survivors who are women is higher
than in the non-cancer population (p < 0.01) (Table 5).

As mentioned, the age distribution within the total cancer population differs
substantially (p < 0.00) from the non-cancer population, such that the cancer population
is much older. The mean age for the cancer population is 62, whereas the mean age in
the non-cancer population is 43.9. Because the cancer population is substantially older,
many of the other demographic differences noted in the data, for example race and co-
morbidity and the aforementioned gender, are influenced by this age effect. Therefore,
adjustments for age are included in later results when indicated. The level of educational
attainment in both the cancer and non-cancer populations was very similar. Of the cancer
population, 18% had less than a high school education, a third had completed high
school, and half had attended college and/or attained various degrees.

Racial composition, however, was noticeably different (p < 0.01), with
substantially more non-Hispanic whites in the cancer population than in the non-cancer
population (91% as compared to 72%) (Table 5). The only minority that is as often
found among cancer survivors as among the non-cancer population is Native Americans.
By contrast, the percentages of African Americans and Hispanics in the cancer
population are lower than in the non-cancer population. Cross tabulation of age and race
reveal higher population estimates of both African Americans and Hispanics in the
youngest age category (18-34) and lower estimates in the older age categories (65 and

older). Thus the difference in racial composition between the cancer and non-cancer

55

populations is, in part, explained by the large difference in the age distribution, since
overall cancer incidence rates increase with age (N CI, SEER Public Use Data, 2003).
With the adjustment for age, the comparison of the two populations reveals smaller
variation in racial composition.

Marital status was significantly different in the cancer and non-cancer populations
(p < 0.00). The higher proportion of widowed individuals in the cancer population again
reflects, to a substantial degree, but not completely, the difference in the age distributions
of the two populations (Table 5). No significant differences were noted in education,
race or marital status between the recent survivor and the long-term survivor
subpopulations (Table 6).

Enabling Variables

More than 94% of the total cancer population had health insurance defined as
private insurance, Medicare, Medicaid, and/or other government insurance, such as
military insurance (VA, CHAMPS), Indian Health Service (IHS) or other public, state-
level insurance (Table 5). This did not differ significantly among the recent and long-
term survivor subpopulations (p < 0.15) (Table 6), but did differ from the adult non-
cancer population, in which 83.3% were estimated with insurance, while cancer survivors
more often had multiple types of insurance (p < 0.00) (Table 5). More individuals in the
cancer population had Medicare, Medicaid, and other public government insurance,
reflecting the older age distribution and corresponding lower income level of this
population (Table 5).

A significant difference (p < 0.01) was evident when examining household

income defined as above or below $20,000. (While this measure is a crude indicator of

56

income, it is based on a substantially larger number of responses compared to the more
detailed income questions asked in the 2002 NHIS.) Although the majority of individuals
in both the cancer and non-cancer populations were in the $20,000 or above category, a
higher percent of those with cancer had incomes below $20,000 (21%). This again
reflects the substantially higher proportion of elderly in the cancer population, many of
whom are retired and on fixed incomes. In addition, the ratios of household incomes to
the 2002 poverty threshold show that cancer survivors’ median income was between 2.5
and 3 times the poverty level compared to between 3 and 3.5 among the non-cancer
population (Table 5). However, the lower household income among cancer survivors
only pertained to the long-term survivors (Table 6), again confirming the importance of
the age distributions in the population group comparisons. Overall, although one out of
four had incomes larger than five times the poverty level across the recent survivors and
the long-term survivors, almost 10% had family incomes at or below the poverty level,
which is in line with overall Census Bureau estimates of poverty. It should be noted that
the ratios of family income to poverty thresholds were adjusted for family size, which
ranged fiom one to eight people per household.

Contact with conventional healthcare providers was analyzed as an enabling
variable in the CAM Healthcare model to determine relationships between contact with
physicians, nurses, and other healthcare providers and CAM use. Physician contact,
defined as “saw or talked with a doctor who was a medical specialist, general doctor, or
OB GYN doctor in the past year,” was reported by more than 93% of the cancer
population. This is significantly higher than the 75% who reported this type of contact in

the non-cancer population (p <0.01). Contact with other healthcare providers, including

57

nurse practitioners, physician assistants, midwives, therapists (occupational/physical
therapy) and mental health professionals, was also significantly higher (p < 0.01 and
p < 0.03) in the cancer population (30% and 7%) versus the non-cancer population (18%
and 6%) (Table 5). Likewise, visits to the emergency department were significantly
higher in the cancer population (p< 0.01) (Table 5), with highest proportion of visits
among recent survivors (Table 6). Contact with all conventional providers was highest in
the recent survivor group, as would be expected following cancer diagnosis. Although
not significantly different, noted is the decreased frequency of conventional provider
contacts across all categories in the long-term survivor population (Table 6). Thus, it is
evident that those with a cancer diagnosis have a higher frequency of contact with
conventional healthcare providers, in particular physicians, and that contact is highest
among recent cancer survivors (Table 5 and 6).
Need Variables

Based on the 2002 NHIS, primary cancer diagnoses were distributed as follows in
the total population of cancer survivors: breast (14%), melanoma (9.5%), colon (6.2%),
lymphoma (5.8%), prostate (2.9%), and lung (2.5%) (the breakdown by long-term and
recent survivors is shown in Table 6). This prevalence distribution only partially reflects
the incidence rates of the most common cancers identified by the National Cancer
Institute and the American Cancer Society (2006). There were a higher number of long-
term breast cancer survivors (14.9%), and a lower number of long-term survivors with
melanoma, lung and prostate cancer when compared with the recent survivor
subpopulation (Table 6). These differences in the prevalence of various cancer types

among the long-term survivors is related to several factors, including: (a) longer lifespan

58

for women in the US, (b) the high five-year survival rate for breast cancer (75.9%), (c)
the greater incidence of lung cancer among males (73.3 per 100,000) as compared with
women (47.7 per 100,000), and (d) the corresponding lower survival rate for those with
lung cancer (11.6%) (American Cancer Society, 2006).

Co-existing, non-cancer co-morbidity was examined as a separate dichotomous
variable, that is, “had one or more co-morbid conditions,” and as a continuous variable
for number of conditions reported. A significantly higher percent (89.2%) in the cancer
population reported having one or more co-morbid condition as compared with 57.8% in
the non-cancer population (p < 0.00) (Table 5). When adjusted for age, the difference in
probability of having a co-existing co-morbid condition decreases in comparison of the
cancer and non-cancer population (83.1% and 62.5%, respectively). Analysis using co-
morbidity as a continuous variable revealed a significant difference between the cancer
and non-cancer populations, with a mean number of co-existing co-morbid conditions of
3.5 and 1.5 respectively (p < 0.00) (Table 5). Range of the number of conditions reported
was similar in the two populations (Table 5). No significant differences were found
between the recent and long-term cancer survivors when co-morbidity was analyzed as
“having versus not having” a condition (Table 6). Mean number of co-morbid conditions
also did not differ among the recent (mean 3.4) and long-term survivors (mean 3.5)
(Table 5).

Respondents were asked about different types of pain that they might have
experienced over a one-month, three-month, and/or twelve-month period. During the
interview they could report up to nine types of pain based on pain location and

recurring/nonrecurring status. Pain was reported by 70.5% of the total cancer population

59

(Table 5) with approximately 21% of recent and long-term survivors reporting only one
kind of pain (Table 6). Types of reported pain were: joint pain (50%), back pain (36%),
leg pain (39%), neck pain (21%), and severe headaches (16%) The non-cancer population
differed significantly with a smaller proportion reporting any kind of pain (54%)

(p < 0.000). However, while reports of recurring pain were lower in the non-cancer
population (17% as compared with 32.9%), the proportions experiencing nonrecurring
pain (37%) were similar in the cancer and non-cancer populations (Table 5). In addition,
pain was reported by a higher percent of long-term cancer survivors (Table 6). After
adjusting for age, in order to take into account that musculoskeletal pain, particularly
recurring pain because of arthritis, all increase with age, differences in reported pain
between the cancer and non-cancer populations decreased, but did not disappear: The
estimated probability of having pain in the cancer population, when adjusted for age,
reduces from 70% to 66% and increases from 54% to 55% in the non-cancer population.
Thus, even if the age distribution were the same between the two populations, pain would
remain more prevalent in the cancer population.

Other symptoms reported in the cancer population were insomnia (28.5%),
depression (24.7%), and fatigue (16.1%), with 43% reporting one or more symptoms
(Table 5). All symptom reports were highest among those in the recent survivor group,
with a significant difference in number of symptoms experienced (p < 0.01) (Table 6).
Frequencies of all reported symptoms in the non-cancer population were significantly
lower (p < 0.01) (Table 5). When adjusted for age, the difference in symptom fi‘equency
between the two populations decreases. However, fatigue, insomnia, and depression

remain more prevalent among those who had cancer. Of the total cancer population, 72%

60

rated their health status as good, very good, or excellent, and 28% as fair or poor.
Reported health status differed significantly in the non-cancer population with more
(88.9%) reporting good to excellent health (p < 0.00) (Table 5). Reported health status
was also significantly different among survivors, with lower ratings among recent

survivors (p < 0.02) (Table 6).

CAM Use and Purpose - Descriptive Statistics

Outcome Variable — CAM Use

There were an estimated 5.6 million users of at least one CAM provider, practice,
or product in the US. cancer population during 2002, amounting to more than 39% of the
14.3 million cancer survivors in that year. CAM use in the cancer population was
slightly higher than in the non-cancer population with greatest difference in use of CAM
combinations (p < 0.01) (Table 7). Divided into mutually exclusive categories, CAM use
in the cancer population was reported as follows: (a) CAM providers only 5.2%, (b)
practices only 8.6%, (c) products only 10.3%, and (d) CAM combinations 15.1% (Table
7). The latter amounts to the use of CAM combinations by more than 2.2 million
individuals who have been diagnosed with cancer at some point in their lifetime. Use of
CAM, however, is not significantly different in the recent and long-term survivor
subpopulations (p < 0.62) (Table 7).

Table 8 shows proportions among cancer survivors who used individual types of
CAM provider services, practices, and products in the cancer population. CAM provider
use ranged fiom a low of less than 1% for ayurvedic treatment to a high of 8.2% for

chiropractic care. More than 5% of the cancer population received massage therapy,

61

which is an estimated 780,000 people during the year preceding the NHIS 2002. Of the
CAM practices reported, deep breathing exercises, meditation, and guided imagery were
the used most frequently. Herbs were used more frequently for healing than vitamins and
homeopathic products (Table 8), with use reported by one out of five individuals in the
cancer population. Different types of prayer and healing rituals were also among the
most highly reported practices. Although frequency of prayer use is included in the table,
it was not a part of this CAM use study.
Outcome Variable — CAM Purpose

The purpose of CAM use was significantly different in the cancer and non-cancer
populations (p < 0.01). Whereas, CAM was most often used for treatment purposes only
in the cancer population (18%), CAM in the non-cancer population was used most
frequently for health promotion only (14.8%). There was no statistically significant
difference in purpose of CAM use between recent survivors and long-term cancer

survivors (p < 0.63) (Table 9).

Predictors of CAM Use

Variation in CAM use was examined in two distinct ways. At the conception of
this research, “CAM use” was defined as “use of at least one CAM provider, practice, o_r
product” listed in the NHIS 2002. This binary categorical outcome variable (“used any
CAM” versus “did not use any CAM”) was used in the logistic regression model to
determine predictors of CAM use (Table 10). After examination of this binary outcome,
the analysis focused on the use of different categories of CAM, defined as (a) use of

providers only, (b) use of practices only, (c) use of products only, and ((1) use of

62

providers, practices, and products in combination (combined CAM). A multinomial logit
regression model was employed for the second analysis providing information on use of
the four mutually exclusive categories of CAM (Table 11). This allowed for a fine-
tuning of the operational variable, CAM use, and provided a clear-cut delineation of use
of different categories of CAM. The independent/predictor variables remained the same
in both logit models. The logistic regression model results for overall CAM use (Table
10) are presented first, followed by the multinomial model results for use of specific
CAM categories (Table 11).
CAM Use - Overall

Among cancer survivors, gender and age were strong predictors of CAM use.
Women were significantly more likely to use CAM than men, with the odds of use higher
by a factor of 1.4 (p < 0.01). When the relationship of age to CAM use was initially
examined using multiple age categories, it showed a pattern of higher use with greater
age up until the middle-ages, followed by a downward trend for older people. Thus, to
represent this nonlinear relationship between age and CAM use efficiently, a polynomial
transformation of the continuous age variable was tested, with the quadratic polynomial
providing the best fit, i.e., the final regression model contained both the linear and
squared age term as predictors. As the results in Table 10 Show, both the linear and
squared age terms were highly significant (p < 0.01; OR 1.06 & p < 0.01; OR 0.9993
respectively), supporting the notion that age is not linearly related to CAM use in the
cancer population. When testing at which age CAM use was at a maximum (Figure 5),
one finds highest the adjusted odds (and probabilities) of use at age 47. Those who are

younger and older have lower probabilities of use, with a steadily declining increase fiom

63

age 18 to 47, followed by a steadily rising decrease from age 48 on (Figure 5). As
Figure 5 shows, the probabilities of CAM use at age 20 and 75, for example, were
Similar, with use only 63% as high when compared with the 47 year olds.

To determine the relationship of education and CAM use, the original NHIS 2002
education variable was modified based on post-hoc testing. Several categories of the
education variable did not differ significantly and were combined. By contrast,
preliminary analysis also revealed a pattern of differing CAM use among those with
professional doctorates (MD, JD, etc.) and academic doctorates (PhD), so these were kept
as separate categories. Thus, levels of education in the model included: (a) high school
completion or less, (b) attended college/ received an Associates, Bachelors or Master’s
degree, (c) professional doctorate, and (d) academic doctorate. Results show that having
a college education/college degree did significantly predict CAM use (p < 0.01), with
odds of use 1.9 times higher than in the reference group of people with high school
education or less. While having an academic doctorate raised the odds of using CAM by
a factor of 3.7 (p < 0.01), those with professional doctorates did not have higher odds of
CAM usage than the reference group.

In order to examine race in the logistic model, some modifications were made to
the race variable in the NHIS 2002. Because of the small numbers in the Native
American, Chinese, Filipino, and Asian Indian categories in the cancer population, these
racial groups were combined into the “other/mixed race” category for the logistic I
regression analysis. Thus, identification of differences in CAM use between Asians and
whites was not possible in this study. Race was recoded as (a) non-Hispanic-whites, (b)

Afi'ican Americans, (c) Hispanics, and (d) other/mixed race, and was a significant

64

predictor of CAM use in the model. Data reveal that Afiican Americans in the cancer
population were less likely to use CAM (p < 0.01; OR 0.40) than non-Hispanic whites.
No significant differences, however, were found between Hispanics and whites in the
cancer population (Table 10).

CAM use also differed significantly by income level. Those with incomes below
$20,000 were less likely to use CAM, with odds of use only 60% as high as odds in the
higher income group (p < 0.01; OR 0.64). Adjusted household income, although
included as a descriptive statistic, was not included in the logistic or the multinomial logit
model because of the high number of respondents with missing information (Table 10).
Having various types of health insurance coverage was unrelated to CAM use.

Provider contact was included in the model to examine relationship of contact
with various, diverse conventional providers and CAM use. Having contact with nurse
practitioners, physician assistants, professional midwives, and therapists in the year
preceding the NHIS 2002 interview did significantly predict overall use of CAM
(p < 0.01) in the cancer population. Odds of use by those who saw or spoke with this
type of nonphysician healthcare provider were 170% higher than odds of use by those
who did not have contact. Those who had contact with mental health professionals had
even higher odds of using CAM (p < 0.01; OR 2.53). In contrast, contact with physicians
was not predictive of CAM use (p < 0.95) (Table 10).

Primary cancer diagnosis, a measure of evaluated need, was entered into the
analysis as a multi-category independent variable that included six of the most prevalent
cancer diagnoses in the NHIS 2002 dataset: breast, colon, lung, prostate, lymphoma, and

melanoma. All primary diagnoses were compared with the reference category defined as

65

“other site” cancer. None of the primary cancer diagnoses were predictive of overall
CAM use in the logistic regression model (Table 10). The p-value for prostate cancer,
however, approached significance (p < 0.058). The corresponding odds ratio suggests the
possibility of decreased likelihood of CAM use among those with this primary cancer
diagnosis (OR 0.54) (Table 10).

Reported symptoms, a measure of perceived need in the conceptual model, were
examined to determine ability to predict CAM use. Pain as a symptom was examined
separately from insomnia, fatigue, and depression because of its apparent strong
predictive ability in preliminary logistic regression analyses. In the final logistic model,
pain was a significant predictor of overall CAM use (p < 0.01). The odds of using at least
one CAM provider, practice or product among those with pain in the cancer population
were almost 1.8 times greater than the odds of use by those with no report of pain
(Table 10). Health status, another measure of perceived need, was also a significant
predictor of using at least one CAM provider, practice or product. Those who reported
good to excellent health had increased likelihood of use (p < 0.02; OR 1.34) (Table 10).

The following variables were not predictive of overall CAM use in the logistic
regression model: marital status, insurance, number of emergency room visits, number of
reported symptoms, co-morbidity, and recent diagnosis (Table 10).

CAM Us — Specific Categories

A multinomial logit model was employed to show predictors of specific
categories of CAM (provider use only, practice use only, product use only, combined
use) (Table 11). This more detailed model revealed that women in the cancer population

were significantly more likely than men to use CAM practices only (p < 0.00), and the

66

combined use of providers, practices, and products (p < 0.00), but did not differ fi'om
men in their probability of using a CAM provider or product. The odds of using only
CAM practices were more than 2.2 times greater and odds of combined CAM use were
1.8 times greater among the women (Table 11). An adjusted Wald statistic (F = 0.00)
supports a significant gender effect for CAM use in the cancer population.

Age showed a similar pattern as a predictor of the specific CAM use categories.
Age significantly predicted use of CAM practices (p < 0.00; OR 1.13) and the combined
use of providers, practices, and products (p < 0.00; OR 1.14), but had no predictive effect
on use of providers only or products only (Table 11). As shown in Figure 6 (Appendix
B), the probability of using CAM practices only and combined CAM both increased with
increasing age up to a point, at which point a downward trend in probability of use
occurred. Using elementary calculus, we estimate that for CAM practices only, the
highest probability of use is at age 51, and for the use of combined CAM, the highest
probability occurs at age 45 (Figure 6). The probability of using a CAM provider only or
a CAM product only does not change with age. Images in this dissertation for Figure 6,
Appendix B, are presented in color to distinguish the categories of CAM use. Both the
linear and quadratic age coefficients for these categories of use were nonsignificant
(Table 11).

Having a college education and/or college degree significantly predicted use of
CAM practices only (p < 0.01; OR 2.1), CAM products only (p < 0.01; OR 1.8) and the
combined use of providers, practices, and products (p < 0.01; 2.1). Those with a college
education had odds close to 200% as high of using these types of CAM when compared

with the reference group (high school completion or less). In addition, odds of combined

67

use of CAM providers, practices and/or products were almost 8.5 times greater among
those with academic doctorates when compared with the lowest educated group

(p < 0.00). Having a professional doctorate did not predict use of any category of CAM
(Table 11).

As with overall CAM use, African Americans were the only minority that differed
in their use of the specific categories of CAM fiom non-Hispanic whites. African
Americans in the cancer population were less likely to use CAM products (p < 0.03), and
the combined use of providers, practices and products (p < 0.00). Compared to whites in
the cancer population, the odds of using a CAM product were more than two times
smaller (0.45) among African Americans; the odds of using all of the CAM categories
were even 8.5 times smaller (0.14) among Afiican Americans. None of the other
racial/ethnic groups differed significantly in their use of specific CAM categories. A
decreased likelihood of use of CAM products only was also predicted by marital status,
with widowed respondents less likely to use than married respondents (p < 0.04; OR
0.61) (Table 11).

Patterns of CAM use also varied by income level. The odds of using CAM
practices were almost 2 times greater, and the odds of combined use of providers,
practices and/or products were 3 times greater among those with incomes of $20,000 or
more. Interestingly, having private health insurance was a significant predictor of CAM
provider use, raising the odds by a factor of 1.9 (p < 0.04). Having contact with
nonphysician providers (i.e., nurse practitioners, physician assistants, professional

midwives, and therapists) significantly predicted use of all categories of CAM.

68

Particularly noteworthy is the fact that contact with this group of nurses and other
healthcare providers raised the odds of using a CAM provider by a factor of 2.2

(p < 0.01). The use of CAM practices (p < 0.01; OR 3.1), products (p < 0.02; OR 2.2)
and the combined use of providers, practices and products (p < 0.01; OR 2.7) were all
significantly predicted by contact with mental health professionals. Contact with this
group of healthcare providers did not, however, predict use of CAM providers (p < 0.42).
Contact with a physician over the preceding twelve months and visits to the emergency
department were not predictive of any of the specific categories of CAM use (Table 11).
When examined in the multivariate model, only the primary diagnosis of prostate cancer
was a significant predictor of the use of some CAM category. Specifically, those in the
cancer population with this type of cancer were less likely to use CAM practices (p <
0.50; OR 0.22), with odds of use only 22% as high when compared with “other site”
cancer survivors (Table 11).

Reported pain, entered into the multinomial logit model as an independent
variable, was a significant predictor of CAM provider use only (p <0.01), and combined
use of providers, practices and/or products (p < 0.01). Among those who reported pain in
the cancer population, odds of using CAM providers only were 2.9 times greater than
among those who did not report pain. Odds of using combined CAM were more than 2.5
times greater (Table 11). Number of symptoms, including insomnia, depression and
fatigue, was predictive of use of CAM practices only, with a 29% increase in odds of use
with each additional symptom a person reported (p < 0.02). The number of symptoms
had no predictive ability for use of any other category of CAM (Table 11). Non-cancer

co-morbidity predicted increased likelihood of the combined use of providers, practices

69

and/or products (p < 0.01; OR 1.09). The odds of using a combination of CAM
approaches increased by 9% for each additional non-cancer co-morbid condition that a
cancer survivor reported in the NHIS 2002 interview.

The following variables were not predictive of any specific category of CAM use
in the multinomial logit model: number of emergency room visits, reported health status,

and recent cancer diagnosis.

Predictors of Purpose of CAM Use

A multinomial logit model was used to determine predisposing, enabling and need
factors that predicted the use of CAM for (a) treatment purposes only, (b) health
promotion purposes only, and (c) treatment and health promotion purposes combined.
Gender was a significant predictor (p < 0.01) of using CAM for treatment and health
promotion combined. The odds of a woman using CAM for these purposes were 2 times
greater than among men in the cancer population (Table 12). Age was also predictive of
using CAM for combined purposes (p < 0.01; 0.9984) and for health promotion reasons
only (p < 0.02; 0.9991). Those in the cancer population with a college education were
significantly more likely to use CAM for all three defined purposes when compared with
those having a high school education or less. The odds were highest for using CAM for
treatment and health promotion combined (p < 0.01; OR 2.6). In addition, among those
with an academic doctorate, odds of use for both treatment and health promotion were
more than 10 times greater (p < 0.01) (Table 12).

Among African Americans in the cancer population, CAM use was significantly

less likely for treatment only (p < 0.01; OR 0.29) and for treatment and health promotion

70

combined (p < 0.04; OR 0.28). The odds of a white cancer survivor using CAM for these
purposes were more than 3 times greater (1/0.28; Table 12). Income also predicted use of
CAM for health promotion only and for treatment/health promotion combined, with
decreased odds in the under $20,000 group (p < 0.01; OR 0.33 & p < 0.05; OR 0.65
respectively). Those with higher incomes were more likely to choose CAM to promote
health and to treat illness/symptoms. Contact with nurse practitioners, physician
assistants, midwives, and therapists predicted CAM use for treatment only
(p < 0.01) and for treatment/health promotion combined (p < 0.01). Odds of use for these
purposes were close to 2 times greater among those who had contact with these
conventional healthcare providers (Table 12). Odds of using CAM for all purposes were
greatest among those who had contact with a mental health professional (Table 12).
Primary diagnosis, pain, co-morbidity and recent cancer diagnosis were also
significant predictors of the purpose of CAM use. Those with prostate cancer were
significantly less likely to use CAM for treatment only (p < 0.01; OR 0.26) and for health
promotion only (p < 0.05; OR 0.31) in comparison to those with “other site” cancer.
Among those reporting pain in the cancer population, likelihood of use for treatment and
health promotion combined increased by a factor of almost 3 (p < 0.00), and by a factor
of 2.2 for treatment only (p < 0.00). Co-morbidity predicted use for treatment only
(p < 0.03) and for treatment/health promotion combined (p < 0.02), with a 6% and a 9%
increase in odds respectively, for each additional co-morbid non-cancer condition
reported (Table 12). Those reporting good to excellent health status were more likely to
use CAM for health promotion only (p < 0.01; OR 1.96) when compared with those

reporting poor to fair health status.

71

The following variables in the multinomial model were not predictive of the
purpose of CAM use: marital status, insurance, number of emergency room visits,

number of reported symptoms, and recent cancer diagnosis.

Summary: Predictors and Purpose of CAM Use

Table 13 shows the significant predisposing, enabling, and need predictors of the
specific categories of CAM use. Although women in the cancer population were
significantly more likely to use CAM, there is a distinct pattern of use identified in the
multinomial logit model. The data do support the hypothesis that women are more likely
to be CAM users; however, this hypothesis must be qualified, as analyses clearly show
that women are only more likely to use CAM practices and combined CAM. The
relationship of age and overall CAM use is best depicted using the quadratic polynomial
transformation, which shows a curvilinear pattern of overall CAM use associated with
age (Figure 5). The relationship of age and CAM use however, differs for the specific
categories CAM. Whereas CAM practice use and combined CAM use change with age,
use of CAM providers and CAM products do not. The age of highest probability of use
also differed for the different categories of CAM (Figure 6). The data thus lend strong
support to the hypothesis that age is related to CAM use.

A positive association between level of education and CAM use is evident, but
not uniform. Having a college education or an academic doctorate were both strong
predictors of most categories of CAM use. The contrast between academic and
professional doctorates, however, calls into question the influence of other factors on

CAM use, such as professional socialization, values, and role expectations.

72

Results also support the hypothesis suggesting a relationship between income and
use of CAM. Data clearly reveal an association between income level and use of both
purchasable and nonpurchasable CAM. Being uninsured was not related to CAM use,
thereby lending no support in the data for the hypothesis that the uninsured were more
likely to use CAM. In contrast, use of CAM providers only was more likely by the
privately insured in the cancer population. This may be explained by increasing health
insurance coverage for certain CAM therapies, for example chiropractic, biofeedback,
and massage therapy. Contact with conventional healthcare providers was also strongly
predictive of CAM use; however, the type of provider was an important factor. Pain, as a
reported symptom, was very strongly related to CAM use, especially combined use of
providers, practices, and/or products. Thus, data support the hypotheses that pain,
number of symptoms, and co-existing co-morbidity are associated with CAM use in the
cancer population.

The hypothesis that women are more likely to use CAM for both treatment and
health promotion was supported by the data. Age, education, race, income, and provider
contact are also significantly associated with purposes for CAM use, with the exception
of contact with physicians. Contact with mental health professionals was the only
predictor of CAM use for all three purposes. Reported pain and co-morbidity predicted
use for both treatment only and for treatment/health promotion combined. Health status
was the only variable in the model that predicted use of CAM for health promotion
purposes only in the cancer population. Most of the model variables predicted use of

CAM for combined purposes (Table 14).

73

CHAPTER 5
DISCUSSION AND CONCLUSIONS

This research has identified patterns and predictors of CAM use in the US. cancer
population. The CAM Healthcare Model, an extension of the Behavioral Model for
Health Services’ Use, has provided the conceptual framework to guide this research
study. Evidence from the 2002 NHIS has supported the notion that many of the
predisposing, enabling, and need factors identified in the model were related to the use of
CAM in the population of recent and long-term cancer survivors. Furthermore, while the
use of the CAM Healthcare Model as a framework has led to the identification of hitherto
unexplored predictor variables of CAM use, the decision to widen the definition of CAM
into specific categories based on CAM types and purposes, has added considerable
refinement to our understanding of CAM use. The empirical results of the multivariate
analyses presented here confirm, and often go beyond, what has been reported in the
literature on CAM use. The following discussion is organized around the major
constructs of the CAM Healthcare Model. This includes highlights of the comparison of
the cancer and non-cancer populations in the US, and implications of the findings
concerning predictors, patterns and purposes of CAM use. This study has identified
characteristics of CAM users in the cancer population, and has increased our

understanding of how and why they used CAM.

Cancer Population

The prevalence of overall CAM use in the cancer population (39.2% used CAM)

was higher than previously reported by Sahnenpera (2002), and falls within the wide

74

range of estimated use reported by Ernst & Cassileth (1998). It is important to note the
wide variation that exists among studies reported in the literature on CAM use among
those with cancer, and in general, populations. The variation in the CAM therapies that
have been examined influences interpretation of results across studies. The current study
has also identified use of the different categories of CAM identified in the CAM
Healthcare Model. Highest use was reported for the combined use of providers,
practices, and/or products among cancer survivors, followed by product use, practice use,
and lastly, use of CAM providers only. Overall CAM use was significantly lower in the
non-cancer population (34.9% as compared with 39.2%) (p < 0.01), with the greatest
difference noted in use of combined CAM. The use of CAM practices only, products
only, and providers only, were similar in both populations. CAM use, both overall and
the use of the specific categories, did not differ among the recent and long-term cancer
survivors (p< 0.63) (Table 7).

The cancer population differed significantly from the non-cancer population in
gender, age, race, and marital status. The cancer population included more women,
especially in the long-term survivor group. This probably reflects the longer lifespan of
women in the US, and the higher prevalence and survival rate for those diagnosed with
breast cancer, a disease that primarily affects women. Breast cancer was also one of the
most prevalent cancers in the NHIS 2002 subset of respondents used for analysis of the
estimated total cancer population in the US. The age distribution and mean age of the
cancer population indicated that this population was substantially older, necessitating
adjustments for age in the analysis for many of the study variables. Age was adjusted in

the analysis using the population age distribution from the 2000 US. Census. Once

75

adjusted for age, differences in the gender and race composition of the two populations
decreased, but did not disappear completely. The cancer population still had a higher
number of women and non-Hispanic whites. It is our belief that this reflects the greater
number of whites and women that reach the 6th and higher decades of life in the US.
Factors that negatively influence longevity among African Americans, such as higher
morbidity and mortality because of chronic illness, violence, and lack of access to
healthcare, have been widely reported in the literature (Gillum, Mussolino & Madans,
1997; Schneider, Staggers, Alexander, Sheppard, Rainforth et al., 1995; Williams, 1999;
Zoratti, Havstad, Rodriguez, Robens-Paradise, Lafata et al., 1998). This has contributed
to (a) lower numbers of African Americans who will reach an age when cancer is likely
to occur, and (b) decreased cancer survival rates related to high co-existing co-morbidity.
The difference in the number of Hispanic respondents in the cancer population may
represent the lower numbers of Hispanic elders in the US. as compared with young
Hispanic individuals, possibly related to increased immigration among the young. The
differences in the age distribution of the two populations also influenced findings on
marital status. It is not surprising that there were more widowed, and fewer single
individuals in the cancer population given the higher number of elderly.

Recent and long-term cancer survivors had lower incomes, were more insured,
and had more contact with health care providers than those in the non-cancer population.
Both recent and long-term cancer survivors had a higher mean number of co-existing
co-morbid conditions, and higher fiequency of reported pain and other symptoms. This
is not surprising since many had recent cancer, and that the cancer population is older

and, thus, subject to additional health problems and symptoms associated with aging.

76

This presents the picture of a population that has more health needs requiring more
conventional healthcare services, and possibly a greater need for CAM health services
and practices. The finding that a higher percentage of the cancer population used CAM

therapies, when compared with the non-cancer population, may reflect this greater need.

Predictors of CAM Use and CAM purpose

Predisposing Variables
All of the predisposing variables included in the conceptual model significantly

predicted overall use of CAM and/or the use of specific categories of CAM. These

 

empirical findings confirm that overall CAM use in the cancer population was more
prevalent among female, middle-aged, white, and well-educated people, as has been
previously reported for the general population in the research literature (Bausell et al.,
2001; Eisenberg et al., 1998; Lee et al., 2004; Wolsko, Eisenberg, Davis, Ettner &
Phillips, 2002). In addition to confirming these predictive relationships between the
model variables and overall CAM use, we have gone one step further by providing a
more complete picture of the patterns and purposes of use. Breaking down overall CAM
use, based on types and purposes, it could be shown that women were more likely to use
only CAM practices, or a combination of providers, practices, and/or products, for both
treatment and health promotion purposes. On the other hand, this greater female
emphasis on CAM practices does not extend to CAM providers or products. Thus, the
greater likelihood of use of combined CAM therapies among women alone accounts for
the generally reported higher odds of using all CAM approaches. This supports findings

by Sparber (2000) that women with cancer were more likely to use a wide variety of

77

CAM therapies. The likelihood of using practices such as meditation, guided imagery,
and deep breathing for relaxation may reflect the tendency of women to be more actively
involved in self-care than men are, as has been suggested in the healthcare literature. It
may also reflect a belief that these types of practices are important for health and
symptom management. Further study can ensue with the NHIS 2002 dataset to determine
if having symptoms, for example, is a better predictor of CAM use among women than
among men. Further research is also needed to address additional questions on gender
and CAM use behavior, such as: (a) do women chose CAM practices because of their
health beliefs or self-care tendencies, (b) do women assume more self-responsibility for
their healthcare and/or (c) are they more confident about their healthcare choices.

The curvilinear relationship of age and CAM use in the cancer population was an
interesting and important finding that extends what has been presented in the CAM
research literature. Like gender, age specifically predicted the use of CAM practices
only, and the combined use of providers, practices and/or products. It is interesting to
note that odds of use by younger individuals (20-30 yr olds) mirror the odds of use by
older individuals (70 +). Peak use occurred in the hypothesized age range with highest
probability of use occurring fiom ages 45—51 for the different CAM categories (Figures 5
& 6). This confirms previous findings by Bausell, et al. (2001), Eisenberg, et al. (1998)
and Lee et al. (2004) who found higher rates of use among middle-aged individuals. Our
findings, however, are much more precise regarding the age effect on CAM use. In
contrast to the broad age categories mentioned in the literature, we have identified the
specific age of highest probability of use for the different categories of CAM that were

studied. The relationship between age and purpose of CAM use is similar, indicating

78

higher probability of use for treatment and health promotion during middle-age. There
remains, however, a fundamental ambiguity in interpreting the “age effect” in this case.
In cross-sectional surveys, age effects are completely confounded with cohort effects.
That means we cannot know from the data whether the age cohort of 45-50 is historically
uniquely predisposed to CAM use, or whether their greater CAM use is somehow
intrinsically connected to age itself. It is worth emphasizing though that, at least in a
crude way, income and health status were controlled for in the multivariate analysis.
Interestingly, the lower likelihood of use for health promotion in the younger age range
coincides with a pre-illness time in life when health promotion (and illness prevention)
should be a prime concern. Most of the time, since they do not experience many adverse
health problems, young people are unconcerned about their health and simply do not go
to any provider, whether conventional or CAM. In that respect, the use of CAM perfectly
mirrors the use of conventional health services. Similarly, the decreased likelihood of
use for treatment and health promotion in the later years coincides with increased
prevalence of chronic illness and symptoms that need treatment.

Having attended college and/or completing a college degree was one of the
strongest predictors of CAM use, with a variety of significant patterns of CAM use
identified. Interestingly, this level of education was not predictive of the use of providers
only, suggesting that after income and insurance status are controlled for, higher
education may primarily affect self-directed activities, such as CAM practices, and
seeking out CAM products. However, there is one exception to this rule: the difference
between those with academic and professional doctorates. One may speculate that many

of the professional doctorates included physicians and other conventional health care

79

providers with doctorates. These groups may be least inclined to see merit in CAM
therapies, but at this point, this finding raises more interesting questions for firrther study
than it suggests answers. The greater likelihood of CAM use by academic doctorates,
and use for both treatment and health promotion, is consistent with the overall trend of a
positive association of education and greater CAM use. Overall, the findings on
education confirm that likelihood of CAM use increases with higher education among
cancer survivors just as in the general population, as has been suggested in the literature
(Ashikaga, Boomer, O’Brien, & Nelson, 2002; Astin, 1998; Barnes et al., 2004; Bausell
et al., 2001; Egede, Ye, Zheng & Silverstein, 2002; Eisenberg et al., 1998; Mackenzie,
Taylor, Bloom, Hufford & Johnson, 2003).

Regarding race, our data show that Afiican Americans in the cancer population
were less likely to use CAM than whites, supporting earlier findings by Eisenberg et al.,
(1993, 1998). These findings, however, differ from Barnes et al., (2004) who found
higher CAM use among African Americans, when compared to whites using the same
data, the NHIS 2002. We suggest that the difference in these findings on race and overall
CAM use in the NHIS 2002 dataset are primarily because of the exclusion of the prayer
variables as a CAM therapy in our study. Barnes et al., (2004) also reported that the
percentage of the general US. population that used CAM decreased from 62% to 36%
when prayer was excluded in the analysis, and prayer, as a way of dealing with health
issues, is more prevalent among African-Americans. The proportion of Afiican
Americans in the cancer population that used prayer was significantly higher than among
non-African Americans (82% as compared to 69.8%) (p < 0.0035). Furthermore, Barnes’

estimates were not adjusted for age, income, and all the other variables in the current

80

multivariate model, making any comparison to the current findings hazardous. The
finding of a greater likelihood of whites using CAM providers, in the context of
combined CAM use, is supported by previous findings that whites were more likely to
visit CAM practitioners (Bausell et al., 2001). There were no significant differences in
CAM use by Hispanics when compared with whites. African-American race did
significantly predict decreased likelihood of CAM use for treatment only, and use for
both treatment and health promotion by the African Americans in the cancer population.
Widowed individuals in the cancer population were significantly less likely to use
only CAM products when compared with those who were married, despite the fact that
more than three-fourths (80.4%) of the widowed were women. Here we might conclude
that this pattern emerged because married couples, as compared with widowed
individuals, have more financial resources, just by virtue of single versus combined
household incomes, and the effect of age on income level. It is true that the multivariate
analysis included an income variable, but the binary division of income into less than or
more than $20,000 is a rather crude measure and conceals substantial remaining variation
in income. CAM products cost money, and are usually not reimbursable through
insurance. Although no precise theoretical reasons exist for considering marital status as
a predisposing predictor of CAM use, it was included in this study for several reasons.
Marital status has been examined in relationship to CAM use with interesting findings
(Fouladbakhsh et al., 2005). Further, in research examining use of conventional I
healthcare services, marital status is often interpreted as a proxy for social support.
Viewed as such, married individuals would have access to more financial resources

(income etc.) and interpersonal social support that may influence healthcare decisions and

81

use of healthcare services. If this proves to be valid with CAM use behavior, marital
status may more likely be an enabling factor, rather than a predisposing factor, and
should be considered in this light. Further study is thus indicated to sort the influence of
marital status, social support, and financial resources on CAM use in this population, and
how these variables should be categorized in the CAM Healthcare Model.
Enabling Variables

Provider contact was highest among recent cancer survivors as would be expected
in the first year following diagnosis. It is safe to conclude that CAM use in the cancer
population was complementary, rather than alternative, given the high percentage that
had contact with a physician, nurse practitioner, physician assistant, therapists, or mental
health professional over a one-year period. When these contacts were examined
independently, we find that more than 94% of the cancer population had seen or spoken
with a conventional healthcare provider.

In examining factors that might enable or impede use of CAM provider services,
CAM practices, and CAM products, the analysis focused on the predictive ability of
income, insurance, and connection with a conventional healthcare provider. Those in the
cancer population with private insurance, the most widely held insurance, were more
likely to use CAM providers, with use for treatment purposes only. A high percentage of
those who saw CAM providers had private insurance as follows: acupuncture (91.7%),
chiropractic (83.8%), naturopathic (76.7%) and massage (79%). Therefore, use of CAM
providers by those with private insurance may be attributed to: (a) health insurance
coverage for certain CAM provider services such as chiropractic, acupuncture. and

massage, which are generally offered for treatment purposes, and/or (b) more available

82

financial resources for the purchase of non-reimbursable CAM provider services because
the cost of conventional healthcare was covered by insurance (this is what an economist
would call an “income effect”). This finding supports previous research that reported
greater odds of using provider services among those with even partial insurance coverage
for some CAM health services, when compared with those with no insurance (Wolsko et
al., 2002). The literature reveals that insurance coverage for CAM provider services is
increasing across the US, but is often limited to chiropractic treatment, massage therapy,
acupuncture, and naturopathic medicine, Cleary-Guida, Okvat, Oz & Ting, 2001;
Lafferty et al., 2004). More study is needed to identify the relationship between personal
financial resources, cost of CAM, and factors that influence the decision to use CAM,
particularly as some CAM services have become a more integral part of “conventional”
health care.

The findings in the literature regarding the relationship of income and CAM use
have been contradictory. This may be because of varying operational definitions of
CAM, and the different populations that have been examined. It may also be because of
the complexity of a measure such as income. Income has generally been defined as the
amount of money coming into a household. Of great importance is the composition and
size of the household, i.e., the number of family members who are supported by this
income. Our measure of adjusted household income took this into consideration.
Although not included in the logistic and multinomial models, it has still informed us that
the cancer population has more individuals with lower income levels, most likely related

to the age distribution. It is not, however, a complete indicator of financial resources, as

83

it does not reflect savings and other sources of available money that could be Spent on
CAM health care.

In our study, higher income was a significant predictor of CAM use in the cancer
population, confirming findings in the literature that have shown a positive relationship
between CAM and income level (Kao & Devine, 2000; Lee, Lin, Wrensch, Adler &
Eisenberg, 2000; Tindle, Davis, Phillips, & Eisenberg, 2005). Interestingly, in the
current study, those with higher incomes were more likely to use CAM practices only,
which sometimes involve no financial costs, as well as combined CAM approaches, for
both treatment and health promotion. Many CAM practices, although largely cost-flee
when self-directed, except for the time commitment, are also available through private
and community-based classes with a fee requirement. Hence, higher household income
would provide additional financial resources for guided learning of CAM practices (class
instruction, learning materials, such as CDs, book etc.), and for out-of-pocket costs of
other CAM services and products. There are likely to be other reasons for the association
between the income variable and CAM use. The binary income variable, with a cut-off at
$20,000 essentially divides the population into “poor” and “not-poor” groups. Yet,
poverty status is associated with a host of barriers to access (Bodenheimer & Grumbach,
2004), not least among them is simply access to information about CAM services and
practices. It is clear, however, that more research needs to be conducted to understand
more completely the relationship between available financial resources and use of CAM
therapies that require purchase.

Contact with a conventional healthcare provider was another enabling factor that

significantly predicted overall use of CAM, and use of the specific CAM categories. The

84

conventional healthcare provider categories were created based on available data in the
NHIS 2002, and served to represent physician providers, nonphysician providers (nurse
practitioners, physician assistants, physical, and occupational therapists), and mental
health professionals (physician and nonphysician). Although contact was highest with
physicians, this was not predictive of any CAM use in the cancer population. Thus, there
appears to be neither a substitution effect, e.g., cancer patients seeking out CAM instead
of conventional care, nor a re-enforcement/complementary effect, for example, cancer
patients who go to conventional physicians being more likely to use CAM. However,
any conclusion needs to be tempered by the realization that more than 90% of the cancer
survivors had contact with a physician in the previous year, which means that there is
insufficient variation to establish a relationship between physician contacts and CAM
use. By contrast, contact with nurse practitioners, physician assistants, or physical and
occupational therapists, which occurred at a lower rate of 29.5%, was a strong predictor
of a_ll categories of CAM use, providing evidence for the complementary use of CAM
and conventional healthcare services among cancer patients. This type of contact also
predicted use of CAM for treatment purposes only, and for both treatment and health
promotion combined. These providers primarily care for those who are ill and/or
experiencing symptoms, such as pain, hence it makes sense that contact would be mostly
predictive of CAM use for treatment purposes. It also suggests that CAM therapies
provide an additional source of care for symptoms that may not respond to conventional
treatment. Further study is indicated to understand more fully the relationship of contact

with nurses and CAM use. This should include advanced practice nurses, such as nurse

85

 

 

practitioners, as well as other nurses, who provide care in hospitals, clinics, and
community-based settings.

Although contact with a mental health professional was a strong predictor of
CAM use, especially use of CAM practices, it did not predict use of providers only.
Since many CAM practices focus on promoting relaxation and reducing stress, the need
for this is understandable among those with mental health concerns. Practices, such as
deep breathing, meditation, and guided imagery, have been widely used in the mental
health field by nurses, psychotherapists, and psychiatrists for symptom management. Use
of CAM products included herbs, vitamins, and homeopathic preparations, many of
which are used for the management of physical and psychological symptoms such as
pain, anxiety, depression, insomnia, and fatigue, which were prevalent in the cancer
population.

Need Variables

This research has confirmed that many need variables identified in the conceptual
model were related to the use of CAM provider services, CAM practices, and the use of a
combined CAM approaches by those in the cancer population. Perceived and evaluated
need, as measured by reported health status, pain, number of other symptoms, and
number of co-existing co-morbid conditions, was predictive of CAM use in the cancer
population. Having a primary diagnosis of prostate cancer was also a significant
evaluated need factor predictive of use.

Our findings concerning the relationship of good to excellent health status and use
of CAM for health promotion purposes only, have interesting implications. The need for

health promotion among those who are ill has been well documented. This finding.

86

however, suggests that those CAM therapies that have a strong health promotion
component were not readily used by those with lower health status ratings in the cancer
population. A possible factor for lower likelihood of use for health promotion among
those with poorer health status is the self-care aspect of CAM, which requires active
involvement of the individual. Those in fair to poor health may simply lack the energy
and focus to become involved in self-care activities that are not directed at treatment and
relief of symptoms. Clearly, at this point, however, such interpretations are speculative,
and firrther study is indicated to clarify why CAM therapies were chosen for health
promotion purposes by those who reported better health, and if this was prompted by the
beliefs about use.

Overall, the probability of using CAM services, products, or engaging in CAM
practices does not appear to be affected by the specific primary cancer diagnosis. The
only significant difference involved a decreased likelihood of use of CAM practices for
treatment and health promotion by those with prostate cancer. This finding may be the
result of (a) a possible gender difference in the use of self-directed, self-care health
activities, and (b) the extensive information in the media regarding a CAM treatment for
prostate cancer, specifically a product (PC-specs) that was banned by the FDA around the
time of the 2002 NHIS. Hearing about a contaminated, nonconventional cancer treatment
that caused serious, unexpected effects may have prompted hesitancy to use anything
outside of the conventional healthcare realm among those with prostate cancer. Only
data from a different year could Show if the current finding is the result of a temporary

“history” effect (Shadish, Cook, & Campbell, 2002)

87

That the presence of co-existing co-morbid conditions predicted CAM use in the
cancer population is not surprising and confirms previous findings in the CAM research
literature (Bausell et al., 2001; Egede et al., 2002; Eisenberg et al., 1998; Wolsko, 2002).
CAM therapies have frequently been used for the treatment of symptoms related to
chronic illness. Indeed, the current findings support the notion that, the more co-morbid
conditions cancer patients have, the more likely they are to use CAM for treatment
purposes primarily. The finding that co-morbidity was a significant predictor of CAM
use as treatment provides valuable information for nurses and other healthcare providers
caring for recent and long-term survivors. It points out that cancer survivors with co-
existing co-morbidity are seeking relief from the burden of chronic illness by including
CAM in their healthcare regimen. This is fiirther supported by the significance of pain as
a predictor of the use. Having pain was one of only three variables in the CAM
Healthcare Model that predicted use of CAM provider services only. This reflects the
use of chiropractors, massage therapists, acupuncturists, and other CAM providers for the
very common problems of back, neck, joint, and other types of pain. More than 56% of
those with low-back pain and 36.7 % with neck pain had visited a chiropractor (p < 0.00).
Almost 88% of those who saw an acupuncturist (p < 0.05), and 82% of those who
received massage (p < 0.05) in the cancer population reported pain.

The relationship of pain, chronic illness, and CAM use has been reported in the
research literature and is confirmed by our findings (Eisenberg et al., 1998; Wolsko et al.,
2003). The strong ability of pain to predict use of a combination of CAM approaches for
both treatment and health promotion further suggests that those who have pain are

seeking a variety of care options to address this widespread problem. Our data also Show

88

 

 

that pain significantly predicts use of CAM as a treatment option only. The number of
other symptoms reported also predicted the use of CAM, but only the use of practices. It
is clear that as the number of symptoms increases, the burden experienced by the
individual intensifies. This burden may very likely prompt use of self-directed
approaches to promote relaxation, ease stress and manage symptoms. Further study,
however, is necessary to sort the relationship of other symptoms experienced (depression,

insomnia, and fatigue) and use of specific CAM practices.

Use of the CAM Healthcare Model

The CAM Healthcare Model, an extension of the Anderson Behavioral Model for
Health Services’ Use, provided a useful framework for enumerating and showing the
potential predictive relationships between the predisposing, enabling, and need factors,
and the outcome variable defined as CAM use. Using this model to guide the secondary
analysis of the 2002 NHIS dataset allowed us to determine the significance of these
relationships. Further revision of the operational definition of our outcome variable to
include specific categories of CAM allowed for a more specific delineation of use in the
cancer population. This multicategory outcome variable was based on the alternative
categorization of CAM presented in the Introduction section. One of the more surprising
aspects of employing the CAM Healthcare Model is that it appears to apply to CAM
practices, and products, as well as to CAM services offered by providers, even though the
original Anderson Behavioral Model was specifically devised to predict use of
conventional health services. Perhaps, the pursuit of self-care practices is not all that

different from seeking out health care services: both require time, resourcefulness

89

(“enabling factors”), an urge to make the effort (“needs”), and a predisposition to use
them.

It is in the measurement of the predisposing factors that the current analysis has
its biggest shortcomings. Age, education, and race are only rough proxies for the actual
attitudes underlying the predisposition to use CAM. Only actual measures of these
attitudes will fully clarify the role that predisposing factors actually play in seeking out
and using CAM.

It is also important to note that some factors included in this study as
predisposing, based upon previous research and the original Behavioral Model, may also
serve as enabling factors when examined in a different context. For example, education,
viewed in the literature as a factor that predisposes one to use healthcare services, may
also be considered an enabling factor for CAM use. This is noted by the difference in use
among academic and professional doctorates in the current study. Perhaps it is
professional socialization, collegial relationships, and/or the work environment
(conventional healthcare) created through the education process that influences
healthcare choices, rather than solely acquired knowledge. The distinctions for
classifying variables in the model categories are not absolute or clear-cut. Further study
is needed to illuminate how the factors identified in the CAM Healthcare Model
influence CAM use, with subsequent recategorization in the model as indicated.

In sum, the extended model provided a conceptual framework that was
theoretically sound, easy to implement, and amenable to changes as familiarity with the
dataset increased. The model identified measurable empirical indicators associated with

CAM use in the cancer population. This allowed us to study CAM on a health service —

90

 

 

health practice continuum, thereby reflecting the nature of CAM use defined in the
Introduction section. Examining the distinct categories of CAM has provided valuable
information on both the self-directed and the provider-directed components of CAM
health care. Thus, we conclude that the extended model was very usefiil in guiding this
research. Continued use of this model for future CAM research may promote increased
comparability across studies, and allow for confirmation of findings about characteristics

of CAM users, and the patterns and predictors of use.

Conclusions

This dissertation research has provided a wealth of information about CAM use in
the US. cancer population that confirms and extends what has been presented in the
research literature. Our findings are strengthened by the methodology of the NHIS 2002,
which employed a multistage cluster sampling design, and represents the civilian,
noninstitutionalized adult US. population. Using STATA 9.2 for analysis of the NHIS
2002 dataset allowed us to estimate the total surviving adult U.S. cancer population of
more than 14 million individuals, and to compare this population group to the more than
190 million adult individuals who did not have cancer. Analysis using logistic regression
and multinomial logit models has provided conclusive findings on significant predictors
of overall CAM use, the use of specific categories of CAM, and the purpose for use in the
cancer population.

These findings provide important information for nurses, physicians, and other
conventional and CAM healthcare providers who care for oncology patients and cancer

survivors. Characteristics of users have been clearly identified, which will assist with the

91

 

 

assessment process as cancer survivors navigate the conventional healthcare system for
care. This information will serve to enhance health outcomes by promoting early
recognition of users and their patterns of use. This recognition can assist nurses and other
healthcare providers maximize outcomes of healthcare by (a) preventing potential
negative interactions and adverse events that are possible between CAM use and
conventional treatment, (b) incorporating those CAM therapies that benefit patients into
the overall nursing plan of care, and (c) allowing for integration, and hence, availability
of CAM, in the conventional healthcare arena. Overall, this will allow for a more
comprehensive system of healthcare that addresses the needs of those who have survived
cancer. As it grows in popularity, CAM may change the boundaries of the healthcare
arena as we know it today.

Nurses who are directly involved in the care of oncology patients and cancer
survivors, whether hospitalized or in the community, should comprehensively assess
CAM use by their patients. In the assessment process, nurses should recognize that
patients might not reveal their use of CAM. Data have shown that many cancer patients
are reluctant to discuss use of CAM therapies and practices with healthcare providers.
Hence, nurses should establish open communication with patients and their families, and
use assessment tools that include a CAM component or are specifically focused on CAM
use (Fouladbakhsh, 2005). Nurses should identify potential benefits, risks of use, and
how CAM has been incorporated into the patient’s program of healthcare.

CAM therapies and practices that provide assistance with the management of
symptoms, whether related to the disease process or the treatment regimen, should be

considered for inclusion in the nursing care plan when possible. As adjuncts to

92

conventional care, certain CAM therapies and practices can also be provided and/or
guided by nurses who are knowledgeable and/or credentialed to do so. Nurses are also
in a prime position to provide education about CAM. Nurses can independently
incorporate teaching about CAM practices, such as guided imagery, meditation, and other
relaxation techniques, which may be useful for self-care during chemotherapy, radiation,
and before and after cancer surgery. Increased involvement in self-care for symptom
management may increase a patient’s sense of control during the illness experience and
treatment process, ultimately affecting comfort, quality of life, and overall satisfaction
with healthcare. Nurses should also be attuned to those patients with co-existing
co-morbid conditions and related symptoms such as pain, depression, insomnia, and
fatigue, recognizing their greater likelihood to use CAM. Recognizing that women are
more likely to use CAM practices, nurses should allow for continued, appropriate use
while CAM users are in their care. They should also include male cancer patients in their
teaching about beneficial CAM practices, noting their lower likelihood of use. Nurses
are also in an ideal position to provide resource and referral information for those who
use CAM.

In summary, nurses are often the healthcare providers most consistently and most
intensely involved with cancer survivors. Therefore, it is important that nurses (a)
recognize who uses CAM, (b) understand patterns and purposes of use, (c) assess risks
and benefits of use, (d) prevent negative interactions and adverse events through
recognition and documentation of use, (e) educate patients about benefits, and (f)
ultimately, serve as a link and resource for their patients who concurrently use CAM and

conventional health care services, practices and products. Nurses are in the ideal position

93

 

 

 

to provide comprehensive care and guidance about CAM to cancer survivors, thereby
promoting positive health outcomes.

This research has also raised further questions about CAM use in the cancer
population. The CAM Healthcare Model includes empirical indicators for the
predisposing, enabling, and need factors that we were unable to measure because of
limitations imposed by the use of secondary data. Information on health beliefs, values,
cultural lifestyle factors, community norms, personality factors, social support, and more
were not available in the 2002 NHIS data. Clearly, a crucial factor in understanding
CAM use is the motivation that prompts one to select a CAM provider, practice, and/or
product to address health needs. This study has highlighted that in the US. cancer
population, the use of CAM appeared to be complementary in nature. More research
specific to cancer is needed to sort out clearly why individuals chose a specific CAM
therapy, what factors led this decision, and whether this was used concurrently with
conventional treatment, and with provider awareness and direction. Understanding why
individuals make the healthcare choices they do, is of course, complex. It is only with
further study that we can illuminate additional factors that prompt a cancer survivor to
choose CAM.

Limitations

This study was limited to information available in the 2002 NHIS dataset.
Because of lack of information, we were unable to explore the relationship between
factors that might potentially predispose an individual to use CAM, such as health
beliefs, cultural background, values, and attitudes. Beliefs about health and personal

responsibility for healthcare are very important in understanding the health services one

94

seeks and uses. Without this information, one cannot understand fully the underlying
motivations that influence CAM use behavior. No variables were available in the NHIS
2002 dataset to explore the link between beliefs and attitudes and CAM use in the cancer
population. Thus, further study is warranted to examine these predisposing factors in
relationship to CAM use.

Specific information about need factors such cancer stage, cancer treatment and
length of survivorship was unavailable in the dataset. Although we could determine the
timing of the cancer diagnosis within a one-year, or more than one-year classification, we
were unable to sort the respondents who were concurrently undergoing conventional
cancer treatment at the time of the NHIS interview. In addition, although respondents
were able to report symptoms, there was no way to identify (a) relationship of reported
symptoms to the cancer experience or treatment, and (b) symptom severity. These need
factors are important in understanding the use of CAM by those with a cancer diagnosis
and whether use is for management of symptoms related to: (a) ongoing cancer
treatment, (b) the cancer diagnosis, (c) the progression of disease, and/or ((1) co-existing
co-morbidities. In addition, although we were able to examine the relationship of CAM
and the enabling factor of income, the categories used were restrictive (limited).
Extended analysis with finer distinctions in income level, and examination of the
relationship between overall financial resources, and cost of CAM in relationship to use

in the cancer population is indicated.

95

 

WEB: wczeo:
30:8 .5 how
.935

miocaxz

 

2.05282

 

 

FEED
§tSowo>
Extra 8: .3232:
ocoN om: EE§>~on
2.52 22:2: :5
$05 mEEa: >
32603 oEummooEo: 2580532
Baum owwmmmz

 

3&8on mat—~35 moon

6528:. V=6“.

 

20:886. 0232on

mazeo: $.25

 

 

 

 

 

 

Domes: 82:0 0385055
538622 coca—25
meow 5 o_oseoeam

£0 B... 2322.3.

«mgr 2:85.334

 

 

 

 

mHEZ Noom 2: E 83955. 320 mo 35 g 2an

96

 

30¢
0:50.550: £050... 3500805. £000... 330: ”03:30“. m
w:_30: 05050.5 0.

 

300: 05 0:003 0:: 5 020m 33:0 0.3% 05 35 0505.3 0380:: 3:05
£83020: .0335: ”043$
>00: 05 080%,?

:0 305005 5:50 5335.35 :0 003: 03 35 050025

003950: 0:3 0333932

 

380:: 3% $5050.53 .30: ”0353mm
52302005
0:3 30:35 0:: 80:00.5 003:-b32wo—05 0:: 3.532

00380:: 0035303205

 

05060::

000:20 3:03:35 a4335053: $580050: ”gag
$05:

0:3 :30: 00 :03053 3060503 3:233:00 5.00003 3 :0
02300 005.000 00:03.5 0:3 D005 :0 0.5009? 0003500

050?? 30305 03:36:?

 

 

0305;: 6230305 .20 3:. dwox ”0355m—

.530:
00 00:30:00: 353% 0:: 3030—0330 .338 .3:0_>3:0m

 

8380:: 30:65::

 

 

 

03:35: 0% :0358000

:3an

ease: 25 .8 803060660 32002 .N 2%:

97

00:30: 3:03:08: 0:

b::00:0:00:_ 003: 0:
:30 5:03:08:

30:30:: :0 0:30 0030:: 0:

 

 

 

 

 

 

 

 

 

 

>_::00:0:00:_ 0033 2<0 A: 003: :35 3:03:08: 2<0 3:300”:

1003:: 00030 330:. 33:05 0:32

0w3\mm3::..:0m 3:080:53 3:03:32 3:23.30

u93003003 333:3 0% :0»3:: u..‘o.::—:3:_> “3:30: 3:05:
33:3 “333—:

w::E::::0 .w:_:0:m= 6:32 3:303 358:: o_r—3:00:82
330503.33

a. 3:0: £5853 :0 03: 33:03:: 33:32 u..0w3mm3§
3E3:3:3:fi0_:03:30:m

.. w:_:30:: :000 30:80:03 .3003

303:0:0x0 w:_:30:m .300 ”33:03:: 3:230:0m *0_00>::%<

u..m:0:0 M32333 3:03:32 300: :_0:..:0m 30333532

2003002 *m:0:00:m 3::0: *0_:03::0:::0

0::mm0flm:0< 3:0 30:03”: u..0:::0::::0<

 

 

 

 

 

E3003 0:30.:3m
30001042: E<0

 

33:038- 80:00::
30:00:: 2.40

 

303:8 :30:
3:0030:: $20

 

Amoom £22 0:: :_ 00020530 30:00:: 0:3 30003:: .30030:: 30:0 :0 :0::3N::0w030 0>::3::0:_< .m 033.:

98

.83 £52 5 cog—2:...

 

3,: ES 38%.x3a

@8302 3:25an 0:

33022.85 com: on

:3 5:05:85

mu: V53 hfiuméhinSER

“5:58.“ .o 88 0233 3

 

 

 

 

 

b23533. com: 320 3 com: :35 8:25an E<U mobs—Voy—
$893035
*flosvoa oESQOoEoI $50622 Eon
329:5 am 382; t<
:ozaflfisfl > mac—$22 amuofimfiofix
2»: 5.58% :5

20:88:: 0238on

505323 wE—mo:

2805 3.88 REED

 

 

 

 

 

wfiocflmn «%sofiom mambo 3033.350 26:0:

W530: Quocm 32¢th 333mm @583 o_SwEwsm

mEBvoE

$33 .563 HEBE 258.2% 8025.5 teen .95an mo 33-.., 95mm: 333%
.150 BE Q80 5 dwo.» moEwawE 368%»:

 

 

 

 

 

brig 28.2%
32842; E<U

 

mEtoEE- £0355
£0355 E<U

 

3228 5.3:
mHoE>o£ E<U

 

.9253 m 23S

99

 

v n 8323 , wasm

 

 

 

100

 

 

 

 

 

 

 

 

N n o_wsm 32836 295.: 3%:
_u @252
v H 35,5850
m n 2.89mi
N u 85¢. =ao£< Eotomoumo 30830: 83.
_ u 853
v u cow £8393
m H occuoi
m n owe—BU Eotowowmu 88260 cosmoswm
_ H 32 8 m:
32.5280 alowwcm d. glow“ ow<
m u o_mEom Eotowoumo x8 .350
_ n 232
fizz—£10:—
58065 .83qu 3b oEata> «Em: oimta> oEmta>

.3on 2: E $2.528 303285 2: 9 @5508 3%: 08 39:28 2: E weigh; COBEBE E02335 2;

382 032:8: 25 2: 5 83%; E2803 .v 2an

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0:350:00 00=U0¢Qm . 0:03 Mm
_ u 00>
c u 02 33.8380
20220.5 5305852
_ u 00>
c u 02 32.8330
05253 $32.20
_ n 00 > Bow—$0000
o u 02 N00220:. 550%:
-8235
_ n 00 > 32.8380
c H 02 01E0>ow05
_ n 00 > .motomeU
o n 02 040820255
_ n 00> 325330
c u 02 0I080€0E0E
_ u 00 > 328330 .I08350E 00:00:05
o u 02
N n 25.on v 30:03:00 6030.: 0:805
_ H 00>0 0% 25de
33.2.0
5:00:05 Rota—cm 0gb 033:; 0E0: 033:; 030wa>

.6083 0 2000

101

 

 

macaw“?

 

 

 

 

 

 

 

 

 

 

 

 

_ H mm; 02.6
c u 02 Eotowouao E325 9:: .5223
N” 82.35680
_ u twice; Eotowofio $823 33% .23:
30:52.00 38:80 bfifiofiéu
2525280 993.550 mEoEExm
_ n mo>
o u 02 Rotomoso Ema—b8 53
n M «€28.02
0 n «Sop—act?—
m H 838.5
v u 530
m n was:
m n 63.5 xv 50ch
_ n 86.850 Bow—0330 wflvEta baEtm
coal
830%.: EotEEm cab 033:3 2.8: 033.2; 233;

.5303 v 23

102

m.mm I m.mm
b.0H I v.0a
m.rm I N.mm
v.0m I 0.NN
m.am I 0.5H
n.>H I m.MH
0.0a I v.r
m.n I H.m
.manmu mcu
N.mm I 0.vm
v.mv I 0.0v
Ho mmm

n.nv
0.Hm
0.0a

0000.0

N.mN
m.vm
m.mH
n.ma
0.0
N.o

5.00
I b.0N 0.0N
I m.mH m.0fi

H m mmmm.mbm "Amm.mmwa~wm.vv m

m.0
m.>
v.NH
0.0H
m.mm
m.mm

mowaaou
wmumwv mm
mm cmnp mqu

coflumosvm
memnIcoflme

+m>
vnlmm
vmlmm
vmlmv
vamm
vMImH

mmm

cfl Umuoc mums: pamuxm mEmm mnu cflmEmu wwumEHumm coflumasaomI

0000.0

*HNQVOMVH

unwoumm

COHpmHsmom uwocmu Hmuoe

Hm

 

umocmoIcoz HmuOB

Nqofi.om "Ammm.fiv m nmmmnucmflmmo
«vmvommamfl amuOB
I m.0m >.Hm mHmEmm
I m.hv m.mv mam:
mmmmwm
mmqdem¢> UZHmOmmHommm
Ho wmm “smoumm
coflpmasmom

 

coflumHsmom umocmo mgu Ucm coflumasmom umocmoIcoz mcu mo COmHHmQEOU .m mHQMB

103

H.0N I

0000.0

mv00.0

u m 00H0.00

v.0 I 0.

m.N I m.

0.0 I v.

v.0 I n.

N.HN I n.

m.0m I 0.

H m mmmw.vm
0.m I m.N

0.H I 0.0
0.NH I 0.0H

0.0 I 0.0
0.NH I v.HH
0.Nw I 0.05

H m omvm.m
N.H I 0.0

"An>.nmmfi.mv.mvm ummeICOHmmo

LON

chuuma
nu“: ocfl>flq
Uwumnmawm
Uwzoufiz
Umuuo>flo
mamcflm
Umfluumz

 

msumum Hmuflumz

”Amm.qmmfi.mm.vvm memn-cmflmmo

HE\Hmnuo

mmmCHno

oacmamflm
cmoflumE< m>fiumz
cmofluwem cmofiuwd
Aamflm cocv muflnz

momm

“Avw.nmmfi.mm.mv m nmmmn-cmflmmo

mumnouooo
UflEmUmod
mumuouooo
HMCOHmmmwoum
“mummz
muoamcomm

muMflUOmmd
mmmaaoo mEom

.An.pcoov m magma

104

ONNLD
0001
LONG

VOL!)
OV‘c—i

0000.0 u 0 0000.HN "AH0.000 .00.000 UmmmnIcoflmmo

 

 

 

0.0 0.0 I v.0 0.0 czochD

H.00 0.00 I 0.00 0.00 oz

0.0 0.v I 0.0 0.0 mm»

ucchuw>oo umcuo I mocmusmcH spammm

0000.0 u 0 0000.0uh00.000.00.avm memQIcmflmmo
0.0 0.0 I v.0 0.0 czocxcs

0.00 0.00 I H.00 0.00 oz

a.» 0.0 I 0.0 0.0 mm»

Uflmoflvmz I wocmusmcH spammm

0000.0 n m 000H.VHO HAH0.H00.N0.H00 wmmnIcmflmmo
0.0 0.0 I 0.0 0.0 czocxca

0.00 0.00 I 0.00 0.00 oz

0.00 n.0a I 0.0a 0.0H mm»

mumoflumz I mocmusmcH Spammm

0NNH.0 u m voma.m “A00.N00 .00.va UmmmnIcmflmmo
0.0 0.0 I 0.0 0.0 czocxca

H.0N 0.00 I 0.00 H.0N oz

>.H> N.Hn I 0.00 0.00 mm»

 

mum>HHm I QUCMHSWCH SUHQwI

 

mmgdem¢> OzHQmflzm

.AU.uCOUV 0 wHQmB

105

0.00 0.00
0.00 v.00
umflamuwzh
0.0 0.0
0.v0 0.00
0.0 0.0
0.00 0.00
0.00 0.00
0.0 0.0
0.00 0.00
0.00 0.00
0.0 0.v

0.00 0.00 I 0.00 0.00 02
0.00 0.00 I 0.00 0.00 mm»

 

.ucmumflmmd c0000m>nm .zm "um00>oum Hmcoflucm>coo nuflz uomucou

0000.0 u 0 0000.000 "0000 .000 ummmnI000000
0.0 0.00 I 0.00 0.00 oz
0.00 0.00 I 0.00 0.00 000

 

02 “umU0>ouQ Hmcoflucw>coo £u03 uomucoo

00000v00 000000000 HMUOB

00.0

I 00.0 00.0 I 00.0 000002

UHOQmmunu >uum>oa ou mEOUC0 >008mw 00 oflumm I 0 mEoocH

0000.0 n 0 0000.0 "A00.0>0.00.va UmmeIcmflme
0.0 0.0 I 0.0 0.0 UmCHmuumomm uoz
0.00 0.00 I 0.00 0.00 muoE no 000.000
0.00 0.00 I 0.00 0.00 000.000 umvcs

mEoocH Uwasouw >H0Emm I H wEoucH

0000.0 u 0 0000.000 H A00.>00 ~00.070 UmmMQICOHmwQ

0.0 0.0 I H.0 0.0 0
0.00 0.00 I 0.0 0.00 0
0.00 v.00 I 0.00 0.00 H
0.0 0.00 I 0.00 0.00 0

 

Ammmxu 00 ucsoov mocmusmcH zuamwm

.Av.ucoov 0 magma

106

0000.0 u 0 0000.000 "000.0000 .00.000 00009-000000

0.00 I 0.00 0.00 0.0 I 0.0 0.0 + 0
0.00 I 0.00 0.00 0.00 I 0.00 0.00 0I0
0.00 I 0.00 0.00 0.00 I 0.00 0.00 0IH
0.00 I 0.0 0.00 0.00 I v.0v 0.00 0

 

00000000000 00 0095220 000002005Iou

 

0000.0 n 0 0000.000 "0000 .000 ummmnI000000
0.00 I 0.0 0.00 0.00 I 0.00 0.00 oz
0.00 I 0.00 0.00 0.00 I 0.00 0.00 00»
A200u00coo U0QuoEIoo umocmoIcoc m mmmv I muHUHQuoEIoo

 

mmqdem¢> Qmmz

 

0000.0 H m 000N.HH "Avh.mmvm.vm.hvh meMQICOHme

00 I 0 ”mocmm 00 I 0 ”00:00

000.0 I 0H0.Q 0.0 "cme 00m.o I 00m.o v.0 “Gmwz
.mm Cu muflmfl>,numUH>ouQIamc00ucm>cou £003 uomucou
0000.0 n 0 0000.0 u A000 .000 00000-000000
0.m0 I v.H0 F.N0 m.v0 I 0.m0 0.m0 OZ
0.0 I H.0 m.b 0.0 I 0.0 H.0 00%

 

Hmcoflmmmwoua £0000: 000cm: ”HwUH>ofiw.2#R§ucw>coo £u03 uomucoo
0000.0 u 0 0000.000 HA000 .va Ummmfllcoflmmo

.00.ucoov 0 00200

107

0000.0 n 0 0000.00 u 000.000.00.000 00000I000000

0.0h I 0.00 0.0n 0.00 I 0.00 0.00 oz
v.00 I 0.00 v.00 0.00 I 0.00 0.00 mm»

c00mmwummo

0000.0 n 0 0000.00 n 00m.000.0m.000 memnIc00m00
>.m0 I 0.00 0.00 0.mm I 0.00 m.m0 oz
0.0m I 0.00 0.00 0.00 I 0.00 0.00 mmw

00cEOmcH

0000.0 n 0 0000.00 u Am0.0m0 .00.000 UmmmpIc00m00
0.00 I 0.00 0.m0 0.00 I 0.00 0.00 02
0.00 I 0.00 0.00 0.0 I 0.0 m.m mm»

0300000

0000.0 H m 0000.MHH ”000.0000 .mm.0vm UmmmnIcmflmmO

 

0.00 I 0.00 0.00 0.00 I 0.00 0.00 000000000 002
0.00 I 0.00 0.00 0.00 I 0.00 0.00 000000000
0.00 I 0.00 0.00 0.00 I 0.00 0.00 0000 oz
chm
0000.0 u 0 0000.00 "000.0000.00.0000 00000I00000o

00 I 0 ”00cmm MN I H ”mmcmm

00.0 I 00.0 0.0 "0002 00.0 I 00.0 0.0 ”0002
AmcofluHUcoo 00 0038:: c0020 I muHUHQHOEIOU

.00.0coov 0 00000

108

0.0a
P.Hm
h.vm
N.bm
>.mH

m.h

H.mH
v.mm
0.00

N.>

m.>H
H.0m
N.mm
v.NH

v.m

P.HH
N.HN
m.vm

0000.0 u m wwwm.

0000.0 H m Naom.0HH "Amm.w©ma

ovm

0.0 m.m I m.m
n.0a 0.0 I 0.>
v.mm w.mm I v.vm
H.0N m.mm I m.Hm
0.va 0.mm I H.Hm
0000.0 H m Nmmm.vw
m.o 0.m I m.m
m.ma 0.0 I H.h
N.mm >.mH I ©.va
0.0m m.vn I 0.mn

Amm.vww .mm.flvm nmmmnIconmo

0.HH

uflmm I Hoom

0.00 ucmaamomeUooo.>IUoow

 

Ammfluoowumu UwcHQEoov msumum nuammm

.om.m0m Ummmnucoflmmo

Hoom
uflmm
Uoow
U000 >uw>
ucmaamoxm

 

msumum cuammm

“Amm.ooofi.mm.mvm nmmeIcmflmmo

0.m
0.0
H.0H
>.mn

m

N
H
0

 

AMHCEOmcfl w wsmflume4cOHmmwummUv HmQESZ Ecumewm

.An.ucoov m magma

109

0.0m
H.vm
0m

0.0m
m.mm
m.NN
m.ma
v.0H
m.h

I m.mv
I m.mN
I m.0H

I v.m~
I 0.HN
I 0.0H
I N.vH
I N.>
m.v

m.mm
0.mm
0.0m
m.mfi
v.0
m.m

b.0m
N.>m
®.HN

mmov.0 u

H.0N
v.Nm
w.mm
0.0a
0.HH
0.HH

m

0 m
0.0H
v.0
H.0H
0.0v 0.mv
h.nm m.mm
N.vH w.wH

m.umummz
m.u0amnomm
m.mumfl00mm¢
womaaoo mEom

wmwaaoo

mm nmumamgoo

mm cmcu mmmq

cofluMUSUm

mmfio.fi "Amm.vvofi.mm.vvm vwmmn-coflmmo

b.0H b.mm
v.mm b.0N
0.vH 0.0H
m.0H m.mH
0.0 v.0
m.m >.n

+mn
vhlmo
voImm
vamv
VVImm
vMI0H

mm<

.manmu mgu ca Umuoc wumnz umwoxm mEmm wcu chEmu mmumEHpmm coflumHsaom+

HU wmm

muo>fl>nsm EumuImcoq

«mmwwvmaa

0.00
0.Nv

UCQUHmm

mm>0.0 u

m «HHH.m
.mmmwmmm

N.nv n.mm
m.Hv m.wv
Ho wmm usmoumm

mno>fl>usm ucmomm

Ammm .va vmmmnIcmflmma
Hmuoe

mHmem
mamz

umncmw

 

mmqmfiHmm> UZHmOmmHQmmm

 

coflpmHsmom uwocmo mnu CH

muo>fl>u5m Euwulmcoq Ucm ucmowm

wo COmflHmQEOU .0 mHQMH

110

hmvm.0 n m mHmN.0 HA00.NH0H .mm.mvm UmmmgIcoflmwQ

H m nmcuumm nuflz ocfl>flq
H OH Uwuuo>fl0
m ma Uwzovflz
v.m Umumumawm
N o mamcflm
0 mm Uwfluumz

 

msumum Hmuflumz
00vm.0 n m mv0N.0 "Amw.0omfi.mv.v0m UmmmnIcmflmmo

m H nmxfle\umguo
o o mmmcfizo
m.m uflcmamflm
m 0 CMUHMmE¢ w>flumz
v v cmoHHwEd CMUHHMQ
v om Auflcmamflm cocv wuflnz

mumm

nmmmnava HmuOH

m.m I 0.H m.m wumnouoov
UHEwUmum

v.v I H.H N.N mumuouuov
Hmcoflmmmwoum

.An.ucoov m manme

11]

NOKO
oovo
ON

Q'mko
oomo
OW

PIDIH
w>~o0
O\

mmr—I
LDHO
ON

@0303
\DNO
OW

Hf‘m
[\NO
ON

Hmmv.0 H m Hmmn.0

.0 I
.00 I
.0

®HKQ
(DNO

.0
.H0
.m

Am0.mmm .mm.

.0
.v0
.m

(IDr-‘lx—i

va nmmmnlcmflmmo

czocxco
OZ
mww

 

mocmusmcfl ucchuw>om nguo I mocmusmcH guammm

mnfiw.0 H m NmHv.0

0.0 I 0.0
m.vm I v.00
m.0a I 0.0

anvv.0 u m 00v>.0

0.0 I 0.0

m.mm I 0.mv

0.00 I n.0v
hmmv.0 n m HHN0.0

0.0 I 0.0

m.Hm I v.mm

0.00 I 0.00

Amm.mvm .Nw.

vflmoflumz I

AH0.0vm .Nm.

mumoflvmz I

Amm.nmm .wm.

mpm>flum I

avm UmmmnIconmo

CBOCxCD
OZ
mww

 

mUCMHSmcH zuammm
va nmmmpucmflmmo
czocxca

OZ
mww

 

wucmusmcH nuammm
va ummmnucoflmmo
czocxco

OZ
mmw

 

@UCMHSWCH QUHMQI

 

mmqmfiHm¢> UZHAmdzm

Av.ucoov o mapms

112

momfi.o I m movm.m u Ammm .va nmmmpIconmo

m.m I 0.0 N.n m.n I H.m m.v oz
0.vm I m.Hm 0.Nm 0.00 I m.mm H.mm mmw

zwwImo .02 Hmumcwm .umHHmHoQO HmoHUmE
02 "umUH>oum HmcoHucm>coo Lqu uumucoo

 

mfimnmmm homQOHm mmovovofi Hmuoe

mm.m I om.m "cmflumz mv.m I mm.m "cmflnmz
UHonmmu£H >uum>om ou mEoocH >HHEmm mo oHumm I N mEoocH
mmvv.o I m Hfiom.o I Amm.fiqw .mm.fivm vmmmpIcaflmwa
H.@ I w.m h.o m.HH I m.m H.m UmCHmuumomm uoz
H.vh I o.mm m.H> m.mn I N.mm o.mp muoe Ho ooo.omm
m.mm I v.mfi q.Hm H.mm I o.mH m.mH ooo.omm “macs
mEoocH Umasouw >HHEmm I H mEoucH
mHmH.o u m mvmn.fi I Amo.mmm .mm.mvm nmmmnIcmHmmo
H.m I w.o v.H m.H I H.o v.0 mma>u mmung
v.0v I v.mm m.>m m.ov I m.mm m.ov mma>u oze
0.5m I N.Nm m.vm n.mm I m.mv m.vm mmxu mco
v.n I H.m N.m 0.0 I 0.N m.v mucmusmcH oz

 

Amwa>u mo ucsoov mocmmsmcH LuHmmm

.Au.ucoov o mHnme

113

mmm.0

0000.0

0H I H “mmcmm 0H
mmv.0 mm.0 cmwz

>H0.0 I mmm.0

0 0m mmuHm umcuo
n HH mEocmHmz
N 0 mEonQE>H
H.v mumumoum
m 0 mega
n m coHoo
m OH ummmum

 

mHmocmmHo umocmu mumEHum

 

mmqmflHm¢> Ommz

m 00mm.m HA0N.0mmN .N0.>vm UmmmnIconwQ

I H ”wocmm
$0.0 "cmwz

 

.mm on muHmH> "umUH>o~QIHmcoHucm>coo zqu pomucoo

vam.0 u m
n.Hm H.mm m.m0 I 0.nm
n.m 0.0 N.NH I H.m

umHQmuwQB .ucmumHmmm CMHon>cm .zm

mmvm.H n Ammm .va UmmmnlconmQ
m.Hm oz
>.0 mmw

 

"HmUH>ouQ HmcoHucm>coo nqu uomucoo

oHoN.H I Ammm .va ommmnIcmHmma
m.mo oz
m.Hm mm»

 

”HmUH>oum HmcoHucm>coo nqu uomucoo

.An.ucoov m mHQME

114

mm00.0 u m va0.v u AmN.HHm .m0.m0m memQIconwo

0.Hv I m.mm 0.0m N.mm I v.mm >.mm mcHHusomu uoz
o.mm I o.Hm v.mm m.mm I m.mm o.Hm cflmm mcfluusomm
m.om I >.mm H.mm o.o¢ I m.om H.mm cflma oz
chm ocHocmHquxm Umuuoamn I mmNu chm

mm I H "mmcmm ON I H ”wmcmm

mo.m I mm.m m.m "cmwz no.m I 0H.m v.m "cmwz

 

AmCOHuHUcoo UHQuoEIoo mo quESC cmmzv %uHUHQHoE IOU

 

vmmm.o I m monm.o IAww.QOOH .mm.mvm ummmnIconma

Nmmovmfifi macammm mvopmmvfi Hmuoe

m.mm I m.mH o.Hm v.vm I m.ofi m.om wuoe no @
m.nm I m.mm m.qm m.Hv I m.mm o.pm mIm
v.mm I v.0m m.mm N.mm I m.mm p.mm NIH
m.mH I p.m m.HH m.mH I m.© o.m o
Amcofluflucoo Mo “magszv I NufluflnuoEIou

mmom.o I m mmmm.H I Ammm .va nmmmnIconmo

mmwwqmfifi @HOHmmm mvopmNVH Hmuoe

N.mfi I ~.m m.HH m.NH I m.m o.m oz
m.om I m.wm >.mm m.mm I F.Hm o.Hm mmw

 

ACOHuHUcoo UHQuoEIoo HmocmoIcoc mmmv NuHUHQHoEIOU

.Au.ucoov w manme

115

LDON'Q‘OFLDNI—IOO

Nc—IHI—I

[\NNKOV‘HLDFKOI—i

mHmo.o I o ommm.o I Hom.ooo .Ho.Hvz oomooIoonoo

m.m> 0h I m.nw m.mn oz
m.vm ®.Hm I v.HN N.mm mmw

conmwumwQ

NmHv.o I m mmwm.o I on.Hoo .mm.Hvz oomooIoonoo

NH H.vo I m.mo m.mo oz
o.mm o.mm I p.mm o.om mo»
MHCEOWCH

ovoH.o I o Homm.H I Amm.Hmm .om.Hvz oomooIconoo

m.om o.mm I o.oo H.Hm oz
H.mH o.mm I m.VH o.mH mo»
oomHooz

vooH.o I o mmov.H Izoo.nmnm .mm.mvz oooooIconoo

v.0 N.N I m.o m.o m
H.H o.m I m.o m.o m
m.m m.v I m.o o.m H
o.m o.o I m.H m.m o
m.o o.> I H.m m.v m
m.m o.m I o.« o.o v
N.NH m.nH I o.oH o.mH m
o.oH m.mH I m.m m.mH N
m.HN m.mm I H.oH m.om H
H.mm N.ov I 5.0m m.mm o
UmocmHumaxm CHmQ mo mma>u mo ucsoo I pcsou chm

.AU.ucoov 0 mHQmB

116

v.0

0.0N
v.0m
m.nm
0.0H

v.0

P.0H
m.0m
v.mm
m.NH

H.0

0.0H
0.Nm
0.0N
m.vH

mmm0.0 u m

m.vH I
m.nN I
0.0m I
0.0m I
n.mH I

0000.0 H m

N.mH
0.0H
H.0N
0.00

Awsmemm w chEOmcH

VONO

mom0.hH

.0
.m
.m
.0

HvH0.m

A00.vmw

.mm.Hvz oomooIconoo

HOOQIHHmm
ucmHHmome0000.>I0000

 

HmmHuoomumo UmcHnEoov m mapmum nuHmmm

H
v

m.0H
N.mm
0.0m
m.mm
0.NH

AmH.>00H

0.0

0.NH
N.mm
m.mm

Hoom
HHmm
0000
0000 >um>
ucwHHmoxm

 

msumum auHmmm

.om.mvz oomooIconoo

OHNM

 

.conmmuQmUV HmQEsz Eoumfiwm

.A0.ucoov m mHQMB

117

hmvnmmmm NwhmwmmH vvwmmowH mommoom

0v.NH Iv.HH0 Am.0H I v.00 00.0 I m.nv AH.mI v.vv

m.HH 0.0 v.0 n.v

:0H0mcH0600 >Hco >Hco >Hco

240 0050000 00H0000m 000H>00m
mmoomwmw

Am.mm I H.vm0 m.vm
0050000 00 00H0000Q .000H>00Q 2&0 0 0000

000 0mm 0 0cmo0mmv

mmmomHm HmmmnVH anmmmH momwvw

0m.oH I m.mHv 0o.HH I H.mv 00H I m.>0 Av.m I N.vv
H.mH m.0H 0.0 N.m
00H000HQEO0 >Hco >Hco >Hco
2¢0 0050000 00H0000m 000H>00m

mHmonm

Hm.Hv I nmvm.mm

0000000 00 00H0000Q .000H>00Q 240 m 0000

Hopoe

NNvomnva H0009
00.00 I v.vmv H.mm

003 240 OZ

 

ZOHHQHDmOm MMUZ¢UIZOZ

Houoe

000m000 H0009
0mm I 0.0mvm.00

003 240 OZ

 

AHU wmm 0 00000000 ZOHBQHDmOm mmuz<o

 

00H00Hsaom 000000Icoz 000 000000 000 CH mmD 2&0 00 0000H0>00m .m mHnme

118

HON0.0

m >vv0.0 HAN».Nva.0m.mvm 00mmnIC0Hm0D

mC0H00HDQOQQSm E000I0C0H w 0C0000 CH 009 2&0 00 C00H000800

m000.0

z omom.¢H Izmmm .va oomooIconoo

0COHO0Hsaoa 000C00IC0C w 000C00 CH 00: 2&0 00 C00H0MQEO0

 

v.mH
VH

C0H00CHQE00
2&0

m.0H
HH

>HC0
0050000

m.0 0.0 m.mm E000I0C00
m.m H.v m.mm 0cmomm
>HC0 >HC0
00H0000m 000H>00m 0m: 2&0

 

A>HCO 0C000000 mmO>H>mDm EmmBIDZOH w Bzmomm

.00.0coov w oHnoe

119

0.0m m.mm
H.H v.0
m.Hm 0.mm
5.0 v.0
0.00 0.00
H.0 0.0
0.00 0.00
m.0 m.0
0.00 0.H
0.H 0.0
0.00 0.00
N.H 0.H
0C0000m 0C0000m

00>H>0sm E000I0COH 00>H>0sm 0C000m

«.00 I N.0m
0.H I 0.0
0.mm I m.00
0.0 I 0.0
0.H I 0.00
0.0 I 0.0
0.H I 0.00
H.0 I 0.0
0.H I 0.00
0.0 I 0.0
m.0m I m.0m
n.H I 0.0
H0 000

000COE

0>H030 00mm CH00H3 000I

0.00
m.0

0C0000m
C0H00Hsaom 000Cm0 H0009

oz
00%

HxH0m\Bmm

OZ
00%

 

0H0000m00HL0

OZ
00%

C0H00H0L0

OZ
00%

 

0oonooo0on

OZ
00%

000>09%&

OZ
00%

 

00:00C5Q00&

 

IC0H00H5000 000C00 000 CH 2&0 00

000%8 0HOH0000 00 000 00 %0C000000 .0 0Hnme

120

0.00 OZ
H.N 00%

 

00H0 C0H0000m0> 0 0000

m.00 OZ
0.H 00%

 

0C0E00000 0Hn00Q00EOQ 0000

 

 

m.mn oz
0.0m 00%
00000 0000

>.mm oz
m.o 00%
%0000000002

m.vm oz
0.0 00%
0000002

v.mm oz
0.0 00%
0H0000N0

0.00 oz
m.0 00%

 

ocHoHooe 0Ho0

.00.0coo0 0 0H000

121

m.vm
0.0

KO
[\
m

H
H

 

 

 

 

 

0.00 oz
m.v 00»
000%

0.0m oz
0.H 00%
0CHE00H> 0000

H.mm oz
m.o 00%
0000 meow 000 0000
0.0m oz
m.o 00%
00H0 zmHC0o 0:0 0000
m.mm oz
m.o 00%
00H0 cH0H000m 000 0000
0.00 oz
H.H 00»
00H0 m.cH00< 0:0 0000
H.mm oz
m.o 00»

 

00H0 0H00H00000E 0 0000

.00.0coov

m mHnoe

122

0.Hv OZ
m.0m 00%

 

00H00c C30 000 00%000

0.00 oz
v.mH 00%

 

000H000x0 0CH£00000 @000 0000

0.00 oz
m.v 00%

 

COH00x0H00 0>H0000000m 0000

pm oz
m 00%

 

%m0mWEH 000HsmI0000

0.00 OZ
N.m 00%

 

coHumuHer 000:

0.mm oz
0.0 000
moomHm
m.mm oz
H.H 000
H00 000

.zo.0ooov m mHnoe

123

anmvHN

000000

0.00 oz
m.v 00%

 

00H00n C30 000 H000H0 0CHH000
N900H9N H0009

H.0h OZ
0 mm 00%

 

@0000 00%00Q 0 CH 0000QH0H000m

0.m0 OZ
0.0 00%

 

L0H00L 000% 000 002000 000000

.00.0coov 0 0H000

124

0000.0 n 0 movm.o "Amm.moo0smm.mv0 U0000I00000Q
00000000000000 00>0>000 8000:0000 0 000000 00 000 240 00 0000000 00 0000000800

oooo.o u 0 mmw0.m0 n Avm.mmm.mm.mv0 00000u00000o
00000000000 000000|000 0 000000 000 00 000 200 00 0000000 00 0000000800

 

 

m.m m.m0 m.>0 2009:0200
0.0 m.m0 m.m0 920000
000008000 000000\000800009 >000 000008000 00000: >000 000800009
000 240 00 0000000
>000 0000000»1
0000>0>0Dm 2009:0200 0 920000
V9vomm0m0 00009
000.000m.00 0m.m0\m.v0a 0m.m0.m.m00
m.n m.v0 n.m0
000008000 000000\000800009 >000 000008000 000000 0000 000800009
000 240 00 0000000
2009400000 mmoz¢oszoz
0N©vomv0 00009
0vm.m.mm.>_ 0v.v0.m.00_ 0m.m0~m.m0_
m.m >.N0 m0
000008000 000000\000800009 >000 000008000 00000: >000 000800009

000 200 00 0000000

 

ZOH 9.009000 mMUZ¢U

0000000000 000000|002 000 000000 000 00 000 200 00 0000000 .0 00009

125

00x08\00000
00000000
.0080 0000000
<00003

0000

 

000000000
00800000

000000000
000000000000
0000000
m0000 00 00
000000000

0000000 000
00000 00 000
000

000800
m.0002
000000

 

mm0m¢000> UZHmO0mHQmm0

000.v 000.0 00.0 nv.0
000.0 000.0 00.0 00.0
000.0 000.0 00.0 00.0
000.0 00v.0 00.0 00.0
000.0 000.0 v0.0 00.0
000.0 m0v.0 00.0 00.0
000.0 000.0 00.0 0000.0
v00.0 000.0 00.0 00.0
000.0 9v0.0 00.0 00.0
00>00000 0000000000 000 0000> 0 00

 

0000000 00 00000000

~000>000 0000>000 240 000 00000 00 0000

00000000>
00000000000

 

0000000000 000000 000 00 000 200 00 0000000000

“00002 0000000000 00000000 .00 00009

126

vmm.m

vv0.m

000.0

mnm.o

mvv.0

mmn.0

000.0

mom.0

000.0
omm.0

000.0

Nvm.0

N00.o

N0v.o

000.0

000.0

000.0

000.0

mo0.o
000.0

00.0

00.0

00.0

00.0

00.0

00.0

00.0

00.0

mm.m

00.0

00.0

00.0

00.0

00.0

00.0

00.0

00x
002

000000 000002
00»
moz

00000 0 00002
00w
002

000000000

 

0000000 0000>000

ooo.om0v

m00>0 0 ooo~oN0
080000

000

moz
0008000000

00»

moz
00000002

00w

002
00000002

00%

moz
000>000
Mmmmmmmmm

 

mm0m0000> OZquflzm

0030003

000000
m.0000002
000000 0000002

 

.00.0coov 00 00000

127

Amom.mmvm mmam n mnemonsm cfl mcoflum>ummno mo .02
mu coflmmo hmfiom n mcoflum>uwmno mo .02
moflumflumum vaoz

oooo.o n m A noum mn.m
mmm
Hmmaaova u mNHm coflumasqoansm UmumEflumm

 

mmv.H mmm.o «v.0 HH.H mm»
«02

mflmocmmflo ucmowm
Hmn.H ovo.H No.0 vm.H Hmoxwunoow
muflmwnuoom

mzumum spammm

mmo.H vmm.o mo.o vo.H muflnflnnofiou

HNN.H mam.o mv.o wo.a mEoumawm umcuo
me.m mbm.a oo.o mn.a mm»
moz

cflmm
mmo.a mmv.o mH.o mn.o mEocmHmz
omv.H mmm.o n>.o mm.o mEO£QE>A
HNQ.H mmm.o wmo.o vm.o mumumoum
mmm.H va.o m>.o mo.H ocsq
vom.H nom.o Hm.o no.0 coHoo
mmm.H nmm.o mm.o om.H ummwum
swuflm Hmnuo

mflmocmMfiQINHmEHum

mmqm<Hm<> Qmmz

vmo.H mvm.o mw.o Ho.H muflmfl> mm

>uoompmo mocmummwu

m

 

 

 

 

 

 

.Au.ucoov OH mflnme

128

ov.o mm.H ov.o Nw.H vm.o mH.H mm.o pH.H waflE\um£uO

 

wm.o or.o Hm.o NV.H $0.0 mm.o mH.o vm.o UHcmamflm
00.0 Vfi.o mo.o mv.o vm.o Hp.o om.o hm.o “mad cmofluwm
wmufizz
wumm
00.0 mv.m «m.o mo.fi mfi.o mH.m nm.o mm.H mumuouoov
UHEOUGOQ
Hm.o mm.o mv.o mv.o Hm.o mm.m vm.o mo.H mpmuouoon
HMCOHmmmMOHm
00.0 NH.N 00.0 mm.H 00.0 mH.m OH.o mv.a um nmocm>u¢
mmmma no mm
coflumosbm
00.0 mmmm.o nn.o oo.H 00.0 nmmm.o mm.o nmmm.o mumsvm mom
00.0 vH.H Nn.o mm.o 00.0 mH.H mm.o mo.H mumw> CH mom
mm<
00.0 mm.fi n>.o mo.H 00.0 Hm.m mm.o mm.o mHQEmm
mwamz
umvcmo

 

mmqdem<> UZHmOmmHQmmm

 

msHm>nm mo msam>um mo msam>um mo msam>nm mo
>Hco >Hco >Hco mwanmflum>
mmD 240 UmcflQEou mm: nonvoum mmD wofiuomum mm: umUH>oum ucmvchmUcH

Amm: Emu oz "msoum comHHmQEouv mmD 2&0 mo mmfluoomumo UHMfloQO

 

coflumasmom Hwocmu mcu Cfl 240 mo mmfiuomwumo
Uflwflummm mo me mo muouoflbwum "HmUoz coflmmwummm oflumflmoq AMHEocHuasz .HH magma

129

00.0

00.0

00.0

00.0

00.0

00.0

ma.0

wm.0

mn.m

N0.H

No.0

mm.0

v0.0

H0.H

00.0

m0.0

No.0

v0.0

m0.0

00.0

0v.0

v0.0

0H.0

00.0

0H.N

mv.fi

v0.0

00.H

vm.fi

v0.0

00.0

mm.0

00.0

v0.0

H0.0

No.0

vm.0

0N.0

00.0

Hm.0

00.m

mv.a

0H.H

00.0

0v.H

mv.H

00.H

00.0

Nv.0

00.0

v0.0

mm.0

HH.0

om.0

00.0

v0.0

H0.H mmw
moz

spammn Hmucwz
vm.m mm»
902

uwnuo w mmusz
mm.H mm»
moz

cmfloflm>cm

pomucoo quH>oum

 

v0.0 000.0va

auw>o w 000.0Nm
mEoocH

0m.0 mmw
«02

ucwficuw>ow
00.H mm»
moz

Uflmoflvwz
OH.H mm»
moz

mumoflvmz
vm.a mm»
moz

mum>flum

mucmusmcH

 

mmqdem¢> wZHAmmzm

mn.o umzonflz
mm.o maocflm
mvmfiunmz

 

magnum Hmuflumz

.An.ucoov HH manmg

130

 

0000.0 n 0A noum mm; H 336$; oamnanewnsmfi2232038902

0mm n mu caflmmo nmflom u mcoflum>ummno mo .02

Hmmaaovfi n wNHm cofiumHsQoansm UmumEHumm mUHumHumum Hmvoz

>u00wumo mocmummmu m
«0.0 00.0 vm.0 HN.H mH.0 mv.a 00.0 H0.0 mmw
qoz

mflmocmmflo ucmomm
m0.0 NN.H «H.0 0m.H HH.0 mm.H Hm.0 wv.H Hmoxmuvoow
auflmwnmoom

msumum nuamwm

H0.0 00.H mv.0 m0.H Hm.0 v0.H mv.0 00.0 xuflvfinnoEoo

00.0 v0.H mv.0 No.0 No.0 mm.fi H0.0 m0.H mEouQE>m Hwnuo
00.0 nm.m mH.0 mm.H nm.0 0N.H 00.0 om.m mww
moz

cflmm
m0.0 00.H HH.0 00.0 00.0 00.0 0m.0 v0.0 mEocmamz
00.0 00.0 00.0 00.0 00.0 0N.H 00.0 H0.H mEocae>0
00.0 Hm.0 0H.0 mm.0 m0.0 NN.0 NH.0 0H.0 mumumoum
00.0 0v.H 0m.0 0v.0 m0.0 mm.H 00.0 mv.H 0:54
00.0 00.H 0m.0 00.0 Hm.0 mv.fi 00.0 m0.H COHOU
m0.0 NH.H 00.0 0H.H 0N.0 Hm.H 0N.0 0m.H ummmum
amuflm uwzuo

mflmocmmfla mumeflum

mmqmmHmd> ommz

mm.0 «0.0 00.0 0.H mv.0 00.0 00.0 m0.H muflmfl> mm

 

 

 

 

 

 

.Ac‘pcoov Ha manme

131

 

 

 

0m.0 vm.m v0.0 00.0 0H.0 ov.m UmxflE\umnuO
00.0 0N.H HH.0 00.0 0N.0 00.0 00cmamflm
0H.0 00.0 00.0 0N.0 00.0 0N.0 umEd cmoflumm
wumm
0H.0 ov.m va.0 mm.m 00.0 0N.0H mumuouooo
UHEwUmod
Hv.0 00.H 00.0 Hm.0 00.0 00.0 mumuouoov
Hmcoflmmmwoum
00.0 00.N 00.0 hm.H 00.0 vo.m mowaaou
m.mmma no mm
cofiumosnm
No.0 H000.0 vm.0 0000.0 00.0 0000.0 Umumsvm 00¢
No.0 00.H 00.0 m0.H 00.0 0H.H mumm> cfl mod
mmd
00.0 0m.H NH.0 0N.H 00.0 N0.N meEmm
mmamz
umvcmw
mmqmflHmd> oszOmmHQmmm
moam>um mo msam>|m mo moam>um mo
Amwmucv Aoovncv Annfiucv
>aco >Hco cofluoEouQ spammm mmanmflum>
cofluoeoum spammm ucwEumeB Ucm ucwEummuB uchchchH

Amms Emu oz "Quoum COmflquEouv

2&0 mo mmomusm

 

Hmocmu mg» CH mmD 2&0 mo mmoausm

coflumHsmom
”amvoz coflmmmummm UHumflmoq HmflEocHuasz .NH manmh

132

00.0

0N.0

00.0

00.0

H0.0

00.0

00.0

H0.0

00.N

vm.a

00.H

mm.0

00.H

hm.H

00.0

00.H

00.0

00.0

N0.0

0H.0

00.0

N0.0

00.0

00.0

mv.N

N0.H

00.H

00.0

00.0

00.0

N0.0

H0.H

«0.0

00.0

00.0

00.0

mm.0

mm.0

HH.0

N>.0

m0.H

00.N

H0.0

00.0

00.0

hm.H

00.0

H0.0

mmw
moz

nuamwn Hmucwz
mm»
002

Hmcuo w mmusz

mm»
.02
amaoflm>nm
uUMpcoo umvfl>oum

 

000.0va

mum>o w 000.0Nw
mEoocH

mmw

nOZ
ucmEQHm>ow

mm»

002
00000002

mm»

.02
wu000umz

mm»

002
mum>Hum
MMdmmwmdM

 

mmdm<Hmd> UZHdezm

Umzonflz
waacflm
wuwfluumz

mapmum Hmuflumz

 

.Av.ucoov NH magma

133

 

Hmmfiaovfi u wmflm coflumasaoansm UmumEflumm 00am n msoumnsm CH mcoflum>uwmno mo .02

 

 

 

 

 

 

 

0000.0 u m A comm 0m.00H u Avvm.0000 00H00 n mcoflum>nwmno mo .02

000 u mu 000me moflumflumum Hmuoz

>u00wumo mocmummmu a
HN.0 0N.H 00.0 VH.H 00.0 00.0 mmw
.02

mflmoc0MHQ ucmomm
00.0 00.H m0.0 00.H mm.0 00.H Hwoxwunoow
fluflmwnuoom

msumum nuammz

00.0 00.0 00.0 00.H No.0 00.H NuflbflnuoEoo
00.0 00.H 0H.0 ma.H 00.0 00.0 mw»
.02

mEOuQEhm umsuo
0N.0 mm.H 00.0 0N.m 00.0 m0.m mmw
.02

cflmm
mm.0 00.0 vfi.0 00.0 00.0 mm.H mEocmez
00.0 00.H 00.0 00.0 H0.0 00.0 mEO£QE>q
00.0 Hm.0 00.0 0m.0 mm.0 «0.H wumumonm
00.0 Hm.0 0H.0 mm.H H0.0 00.0 mung
00.0 00.0 00.0 vm.H 0H.0 Hv.0 coHoo
00.0 mv.fi 00.0 0H.H 00.0 00.0 ummmum
wmuflm umguo

mflmocmmflo mumEflum

mmqmflHmm> Qmmz

00.0 00.0 00.0 H0.H 00.0 00.H muflmfl> mm

.An.ucouv NH manme

I34

Houoflbwua ucmoHWHcmflm I x

 

mflmocmmfiv ucmowm

 

msumum zuamwm

 

muflnflnuoEou

 

 

mEoummNm

 

cflmm

 

mflmommmfiv umucmo

 

nmmz

 

 

XX

XX
><><

muflmfl> mm
nuamwm Hmucwz
Bm\HO ~<m ~m2

DE
uomucou uwufl>oum

 

wocmusmcH

 

mEoocH

 

mcflanmcm

 

 

 

><><

msumum Hmuflnmz
womm

 

coaumosvm

 

m0<

 

><><><><

XX

umvcwo

 

 

 

 

 

 

0cflmommflkum

 

 

amcHnEOU

>Hco xaco
mposuoum mwofluomnm

mmfluomwumo mmD 2&0 oawfluwmm

>Hco
muwnfl>oum

manaum>

wo muOuoHUme mo >HmEE5m .mH wHQMB

135

MOuoflUqu ucmoHWHcmflm I x

 

mflmocmMflU unmowm

 

msumum nuamwm

 

aufivflnuoe-ou

 

mEoummNm

 

><

aflmm

 

X

mfimomwdflv “wocmo

 

mez

 

 

X x

>C><

mufimfl> mm
guHmwm Hmucmz
em\eo .mm .02

02
HUMUCOU HOUfl>OHm

 

wocmuflmfiH

 

mEoocH

 

mcflanmcm

 

 

mflumum HMUflHME

 

><

wumm

 

><

>4

coflumosvm

 

><

mom

 

><><><><

uwvcww

 

 

 

 

 

mafimomwfikum

 

 

COHuoEoum cuamwm
w ucwEumwuB

>Hco
cofluoaoum suamwx

>Hco
ucmfiumwne

mHQMHum>

mmD 2&0 How wmomusm wo muOuUHUmum wo >umEEsm .va wanme

136

moomfivflum gun—Ugh
88000.0_om----:::--

033.60 8 02082:.
............. wouoobvéowgoi

 

 

 

 

 

25% 2302 025 .0 2:0;

137

25 2 2.060% _.

.bzfifiaoooa £2555 £5.08

07.0:me - xv E25350 375
0.3250 28 $3550 .52..

53850 005::—

 

 

 

_.._2<U .8 .55: 03585; $50: 2<U cloning
0380 5:5: “528.50 0.200558 92.3.8
Bum: hméwuxmk a. 8888: $7.0 88.0.8 0. 5:5:
$5320 E<U 50350 a. 0520;
.3285 25 gm
20:0an 2558:. 9 00583§=fi=0><
5:05; 60 2<U 05208.3 _. 052580 3.52:0
$5058.: 8.0 E<U ~55: hmBzBam 0.5108,:— bEsEfioU
xv 0055: 3:020 m3 mo 5&2 Evie—08m
0.053005 2.4.0 :82 3255553—
_ob:oo “528.500 53.5 258585 05:833.
5:80QO .0 3553.00.58. 05—58:. 0.2082 0202080 23250 Ecom
.5030 E08800 a 0.03:an 2<U 0:500:53 553.80 5255
0:59:53 .0 508m 0380 8:83:— 0380 :3th
0:25:02: 9 hmbhViuhSEQ 85 5:50:— ow<
08908.03, $523.5 E<U Em: RESEM Eofixoifim .5950 0.808.005
H.013. 30.3 a 829.2onan Eutfifim

cotaccefi 5:5:

2200850 5:5:
0.522. 5:5:
8:555 £38800 3:535 2:53.50 .5500 35m
400 e um: 2<U occur—53m 0055: 058:83— 82030250 mmzcxfim:
550 50: 0.855 0550“— 08500 35a
00 0580820 T 83.50 5:5: T 200 .8 0552 T 0532.0 T 0:580:55 35.053550

 

A503 moomtom 5:53 How 5002 goggom .«o :ozmomEOSC 5002 285—053 320 .m 550E

138

 

25 2 2.02am ..

£50 28.5.50
was—S x0: £55555£5m

 

053355825 d. 55:50.50

.50 055: a. .554
00.102305 25

 

.5580 N 2:0:

139

0505855

0:55: 550 320
“15505555 55.0 200

.0055055 320
055.555.05.055.
.00550555 320

05055505535530

.0050355 320
05 500

0.5000 200

0585530 I 500
553.50 02553

03500 0:550 05 55005550
5320 550 0555 0535550

@5085

555

550050

0855800

00055050
0055—: 5055.30
05550500 55:50
055: 05555550
0552

 

55550593,— 0055:

055550
T 5550 550 055 Z

550555555 50350
00 .«5 550055
555..
5505:

550552 53050550

05550 5555§m£
58555—

mDQSOmOM

0505505 b35550

05555050

055550
T 053550

50505585
6585555 55055500
55355.50 3550

 

20% 5552

50<

50550
0535505850

 

05520.55 55355.50 #5550
05355505850

055555”—
T 05005000055

555055.055
3.5.2.350

35.0
50:55.33

3550
3505555000

 

2222. 550850 052 8.. 0502 952055: 25 m 2005

140

“2505588 0:55: 8m $20
£88555 8m 320

38505.8 320

0585$5$55
3550558 2.50

02.555.50-550358.

“55550358 3050
,8 5mD

5mD 320

5558580 .. 5m:
55_E5m 0:553

585% £255: .8 550555."?—

0555 053555532

3520555
0588.55 50830
58:82:
55258850
£55
55355.“
388503
5% 555550
83355 555555
8555255555
$553520 85:50
055: 058§5>m
0552

 

555505962 355:

m55555m
T 5550 8.2 0552

5555

55.85

5586535
85855 80355
$5555 5555.555
5:855
555.3832
5585555 bmggou

$585532

3585.2
T wE—ngm

55253855
85855555 805553—
58835 ~585m

 

355 5:52

5w<

550550
5535558850

 

55:88.55 55355.55 ~585m
55255588522

5883
T w5555£05£

5.885.505
NSEQEM

N580
555255.32

N580
352585§

 

255053555 85553 2.255% 85550 ”$53555. >805555m mi 2 8m #5052 55555—55: $20 .5 58%”—

141

5mg :85 - 5585 55m 25058 5235.3. 5528— E 5535.55.» :5 8.. 0565.203

 

 

 

 

 

5.12 5555 59. 355
5525> 055“. | _0 $55
3.55.» cc .5225 .55550 .5 55¢.
55 55 50 on
_ _ p _ .
I 7.
d
m
I . m
25
M.
I V
I .9

 

552058855 5:88:35 5555558 5555.35 55.55:. 525.5 3505:. 555. 55:55:05 x55 ”.5. 05.512

.9355 .5855 .0 55< 3 mm: .25 .0 E5585

5w< 55 55: E<U =5.5>O ,8 8.235595 wag—5m 055.0 .m 5.525

142

 

.55_5> 5.580558 5 5>55 5.5 .5>55m55 m. 5 555.5 6 v wean—cod u 5w<~©§~© ”5 53.53.50 055555 555.

.o .1. 33033390 - 55:58.5 H 5w<©\>©

“25.55 5. .5555 5. 855.088 5.0 .5 5>.55 5.0 .8 555—5 55. 555.3 5.55 5. 5
.5m 05 5w< 5. 85555. .005 53.53.50 655 5.0 5.555 53 5.55.085. 5 .5 m. > 55.53 .55 55_5> 53. 5.0 £5.55 5. .508 55

2.5.5 n8205.85.55 55055505 £50 555555.»
.9285 .250 =5 a. 555m 5. 55 22.3 .53 w + 55555855585 - 355528555 + .5528 u 3.53:. n >

”5502555558 05805.5 5. 05..5>555 5.5 500552 5.0 555% 5.0 5. .55 A5355? 555055550 55 555 25.0
.8 500555. 055 5.5585555 5500.55.55. w5_m5v 5555555 w50$5=55 5.0 55 05555 5 555.0 59. - 553 $20 555. ”582

55,585 m 2:5...

143

5w55 .555 I 55.55 55m £5055. 52555.3. .mw5_ 35.55255. 5:. 5. 5535..5> =5 .5. 05.55.85.

 

m_IZ NOON E9. 5.50
555 525 52558 I .8595 .25

55555.5 520 III 50325 5.4.0

 

 

 

.5225 .8550 .5 59‘

 

 

 

8 8 8 55
_ _ _ _
/ IO
/ k d
l\ 5
q
.8
[IN
MI”.
S
[.75
he

 

5550.95.58 55.5.5655 5555556 5555.555 558:. 55.5.5 _5._.5E 555. 55555505 .85 55. 05.55.05.

.550o.n_ .5 55:55.5 50355 $20 9:53 .6 55_.___5555.n_ 05.2095

5w< 5. E<U ..5 55553.50 85.5555 .5 55 D ..5 b23555... w5_.5_5m 5555.0 .0 5.5m...

144

.mv 5w5 .5 5.5555 b53555... .555w... 55. 5.5035... 055 5.5505... 555.65... .5 555 055.5585

55. .5. . . m 5w5 .5 5. 555 .5 b53555... .8535 5... a.555 55555... 2.5.0 .5. .55. 5.55.55 53 55.55.55 55585.5 $555 5.5m...
...555_...=w.5-555 5.5 5.55.5558 5w5 555.055..

055 .555: 55. 5.55. 5w5 5.05 5w5555 H02 5550 5.55 .5505... E<U 5 .5 5.55 .5035... 2.55 5 9555 .5 b55555... 55% ”5.52

.5558. 5 2:55

145

REFERENCES

Aday, L., & Andersen, R. (1974). A framework for the study of access to medical care.
Health Services Research, 9, 208-220.

Aday, L., & Awe, W. C. (1997). Health Service Utilization Models. In D. S. Gochman
(Ed.), Handbook of Health Behavior Research, Vol I, New York: Plenum Press.
153-172.

Agesti, A. (1990). Categorical Data Analysis. NY: John Wiley & Sons.

Alferi, S. M., Antoni, M. H., Ironson, G., Kilbourn, K M., & Carver, C. S. (2001).
Factors predicting the use of complementary therapies in a multi-ethnic sample of
early-stage breast cancer patients. JAMWA, 56 (3), 120-123.

Alimi, D., Rubino, C., Pichard-Leandri, E., F ermand-Brule, S., Dubreuil-Lemaire, M.,
& Hill, C. (2003). Analgesic effect of auricular acupuncture for cancer pain: A
randomized, blinded, controlled trial. Journal of Clinical Oncology, 21, 4120-26.

American Cancer Society. (2006). More communication needed on complementary and
alternative medicine. Retrieved July 20, 2006, fi'om http:// www.cancer.org.

American Society of Radiation Oncology. (2005). Cancer patients do not report use of
complementary and alternative treatments. Complementary and Alternative
Medicine News. Retrieved September 15, 2006, from

http://www.patient.cancerconsultants.com/cam cancer news.aspx? id=35250.

Andersen, R. M. (1968). Behavioral models of families’ use of health services.
Research, Series No. 15. Center for Administration Studies. Chicago: University
of Chicago.

Andersen, R. M. (1995). Revisiting the behavioral model and access to medical care:
Does it matter? Journal of Health and Social Behavior, 36, 1-10.

Andersen, R. M., & Newman, J. F. (1973). Societal and individual determinants of

medical care utilization in the United States. Milbank Memorial Fund
Quarterly, 51(1), 95-124.

Arcury, T. A., Quandt, S. A., Bell. R. A., & Vitolins, M. Z. (2002). Complementary &
alternative medicine use among rural older adults. Complementary Health
Practice Review, 7(3), 167-86.

Ashbury, F. D., Findlay., H., Reynolds, 3., & McKerracher, K. (1998). A Canadian

survey of cancer patients’ experiences: Are their needs being met? Journal of
Pain & Symptom Management, 16(5), 298-306.

146

Ashikaga, T., Bosompra, K., O’Brien, P., & Nelson, L. (2002). Use of complimentary
and alternative medicine by breast cancer patients: prevalence, patterns and
communication. Supportive Care in Cancer, 10(7) 542-548.

Astin, J. A. (1998). Why patients use alternative medicine. Journal of the American
Medical Association, 279(19), 1548-1553.

Barnes, P. M., Powell-Griner, E., McFann, K., & Nahin, R. L. (2004, May 27.).
Complementary and alternative medicine use among adults: United States, 2002.
Advance Data, 343.

Bartlome, J.A., Bartlome, P., & Bradham, DD. (1992). Self-care and illness response
behaviors in a frontier area. Journal of Rural Health, 8, 4—12.

Bausell, R. B., Lee, W., & Berman, B. M. (2001). Demographic and health-related
correlates of visits to complementary & alternative providers. Medical Care,
39(2), 190-196.

Boon, H., Westlake, K., Stewart, M., Gray, R., Fleshner, N., Gavin, A., Brown, J. B.
& Goel, V. (2003). Use of complementary/altemative medicine by men
diagnosed with prostate cancer: prevalence and characteristics. Urology, 62(5),
849- 853.

Botman, S. L., Moore, T. F., Moriarty, C. L., & Parsons, V .L. (2000). Design and
estimation for the national health interview survey, 1995-2004. National Center
for Health Statistics, Vital Health Statistics, Series 2, No. 130, 1-32.

Burstein, H. J., Gelber, S., Guadagnoli, E., & Weeks, J. C. (1999). Use of alternative
medicine by women with early-stage breast cancer. New England Journal of
Medicine, 340, 1733-1939.

Cassileth, B. R., & Vickers, A. J. (2004). Massage therapy for symptom control:
Outcome study at a major cancer center. Journal of Pain & Symptom
Management, 28, 244-9.

Chemiack, E. P., Senzel, R. S., & Pan, C. X. (2001). Correlates of use of alternative
medicine by the elderly in an urban population. Journal of A lternative and
Complementary Medicine, 7(3), 277-280.

Cherkin, D. C., Eisenberg, D., Sherman, K., Barlow, W., Kaptchuk, T. J., Street, J., et
al. (2001). Randomized trial comparing traditional Chinese medical acupuncture,

therapeutic massage & self-care education for chronic low back pain. Archives of
Internal Medicine, 161(8), 1081-88.

147

Fo

Fou.

Furla

GlHUn:

Cleary-Guida, M. B., Okvat, H. A., Oz, M. C., & Ting, W. (2001). A regional survey of
health insurance coverage for complementary & alternative medicine: Current

status & future ramifications. Journal of Alternative & Complementary Medicine,
7(3), 269-73.

Cohen, L., Warnecke, C., Fouladi, R. T., Rodriguez, M. A., & Chaoul-Reich, A.
(2004). Psychological adjustment & sleep quality in a randomized trial of the

effects of a Tibetan yoga intervention in patients with lymphoma. Cancer, 100,
2253-2260.

Division of Health Interview Statistics, NCHS. (2003). NHIS Survey Description.
Hyattsville, MD: Centers for Disease Control & Prevention, USDHHS.

Druss, B. G., & Rosenheck, R. A. (1999). Association between use of unconventional
therapies and conventional medical services. JAMA, 282(7), 651-656.

Egede, L., Ye, X., Zheng, D., & Silverstein, M. D. (2002). The prevalence and pattern
of Complementary and alternative medicine use in individuals with diabetes.
Diabetes Care, 25(2), 324-329.

Eisenberg, D. M., Davis, R. B., Ettner, S. L., Appel, S., Wilkey, S., Van Rompay,
M.I., et al. (1998). Trends in alternative medicine use in the United States,

1990-1997: Results of a follow-up national survey. Journal of the American
Medical Association, 280, 1569- 1675.

Ernst, E., & Cassileth, B. R. (1998). The prevalence of complementary/alternative
medicine in cancer: A systematic review. Cancer, 83, 777-782.

Fouladbakhsh, J. M. (2005). Complementary Therapies: Tools to Guide & Assess Use
in Clinical Practice. Proceedings of the American Society of Pain Management
in Nursing National Conference, Albuquerque, NM.

Fouladbakhsh, J.M., Stommel, M., Given, B. A., & Given, C. (2005). Predictors of use
of complementary and alternative therapies among cancer patients. Oncology
Nursing Forum, 32 1115-1122.

Furlan, A. D., van Tulder, M., Cherkin, D., Tsukayama, H., Lao, L., Koes, B. W., et
al. (2005). Acupuncture & ‘dry-needling’ for low back pain: An updated

systematic review within the fi‘amework of the Cochrane Collaboration. Spine, 30
(8), 944-63.

Gillum, R. F., Mussolino, M. E., & Madans, J. H. (1997). Incidence and survival in

African American women and men: The NHANES 1 Epidemiologic F ollow-up
Study. Coronary Heart Disease, 127(2), 111-118.

148

Giveon, S. M., Liberman, N., Klang, S., & Kahan, E. (2003). A survey of primary care
physicians’ perceptions of their patients’ use of complementary medicine.
Complementary Therapies in Medicine, 4, 254-260.

Grumbach, K., & Bodenheimer, T. (2004). Can health care teams improve primary
care practice? JAMA, 291, 1246-51.

Health, US. (2005). Health Trend Data. Retrieved July 20, 2006, from
http://www.cdc.gov/nchs/hus.

Hosmer, D. W., & Lemeshow, S. (2000). Applied Logistic Regression. NY: John
Wiley & Sons.

Jacobson, J. S., Workman, S. B., & Kronenberg, F. (2000). Research on
complementary/alternative medicine for patients with breast cancer: A review
of the biomedical literature. Journal of Clinical Oncology, 18, 668—683.

Jordan, M., & Delunas, L. R. (2001). Quality of life & patterns of nontraditional
therapy use by patients with cancer. Oncology Nursing Forum, 28(7), 1107-13.

Kelner, M., & Wellman, B. (1997). Health care and consumer choice: Medical &
alternative therapies. Social Science & Medicine, 45(2), 203-212.

Kessler, R. C., Davis, R. B., Foster, D. F ., Van Rompay, M., Walters, E. E., Wilkey,
S., et al. (2001). Long-term trends in the use of complementary and alternative
medical therapies in the United States. Annals of Internal Medicine, 135(4), 262-
268.

Lafl'erty, W. 13., Bellas, A., Baden, A. C., Tyree, P. T., Standish, L. J., & Patterson, R.
(2004). The use of complementary and alternative medical providers by insured
cancer patients in Washington state. Cancer, 100(7), 1522-30.

Lazarus, H., & Neumann, C. J., (2001). Assessing undertreatment of pain: the patients’
Perspectives. Journal of Pharmaceutical Care in Pain & Symptom Control, 9(4),
5-34.

Lee, M. M., Lin, S. S., Wrensch, M. R., Adler, S. R., & Eisenberg, D. (2000).
Alternative therapies used by women with breast cancer. Journal of the
National Cancer Institute, 92, 42-47.

Lengacher, C. A., Bennett, M. P., Berarducci, A., & Cox, C. E. (2003). Design and
testing of the use of a complementary and alternative therapies survey in women

with breast cancer. Oncology Nursing Forum, 30(5), 811-821.

Levy, P. S., & Lemeshow, S. (1999). Sampling of populations: Methods and
Applications. New York, NY: Wiley & Sons.

149

Lewith G. T., Broomfield J., & Prescott, P. (2002). Complementary cancer care in

Southampton: A survey of staff and patients. Complementary Therapies in
Medicine, 10(2), 100-6.

MacKenzie, E. R., Taylor, L., Bloom, B. S., Hufford, D. J., & Johnson, J. C. (2003).
Ethnic minority use of complementary and alternative medicine (CAM): A

national probability survey of CAM utilizers. Alternative Therapies, 9(4),
50-6.

Morris, K. T., Johnson, N., Homer, L., & Walts, D. (2000). Comparison of
complementary therapy use between cancer patients and patients with other
primary tumor sites. American Journal of Surgery, 1 79, 407—41 1.

Muhajarine, N., Neudorf, C., & Martin, K. (2000). Concurrent consultations with
physicians and providers of alternative care: Results fiom a population-based
study. Canadian Journal of Public Health, 91(6), 449-453,

Najm, W., Reinsch, s., Hoehler, F., & Tobis, J. (2003). Use of complementary &
alternative medicine among ethnic elderly. Alternative Therapies in Health &
Medicine, 9(3), 50-57.

National Cancer Institute, SEER Public Use Data. (2003). Available at www.
seer.cancer.gov/publicdata, accessed 7/20/06.

NIH National Center for Complementary and Alternative Medicine: Classification of
complementary and alternative medical practices. NCCAM Publication # D156,
2002. Retrieved June 12, 2004, http:// www.nccam.nih.gov.

Pagan, J. A., & Pauly, M. V. (2005). Access to conventional medical care & use of
complementary & alternative medicine. Health Aflairs, 24 (1), 255-62.

Patterson, R. E., Neuhouser, M. L., Hedderson, M. M., Schwartz, S. M., Standish, L.
J ., Bowen, D. J ., et al. (2002). Types of alternative medicine used by patients
with breast, colon or prostate cancer: Predictors, motives and costs. Journal of
Alternative & Complementary Medicine, 8(4), 477-488.

Post-White, J., Kinney, M. E., Savik, K. S., Gau, J. B., Wilcox, C., & Lerner, I. (2003).
Therapeutic massage and Healing Touch improve symptoms in cancer.
Integrative Cancer Therapies, 2, 332-44.

Rao, J., Mihaliak, K, Kroenke, K, Bradley, J ., Tierney, W., & Weinberger, M. (1999).

Use of complementary therapies for arthritis among patients of rheumatologists.
Annals of Internal Medicine, 131(6), 409-16.

150

Richardson, M. A., & Straus, S. E. (2002). Complementary and alternative medicine:
Opportunities and challenges for cancer management and research. Seminars in
Oncology, 29(6), 531-45.

Richardson, M. A., Sanders, T., Palmer, J .L., Greisinger, A., & Singletary, S. E.
(2000). Complementary/alternative medicine use in a comprehensive cancer

center and implications for oncology. Journal of Clinical Oncology, 18(13),
2505-2514.

Ritvo, P., Irvine, J., Katz, J., Matthew, A., Sacamano, J., & Shaw, B. F. (1999). The
Patient’s motivation in seeking complementary therapies. Patient Education &
Counseling, 38, 161-165.

Salmenpera, L. (2002). The use of complementary therapies among breast and prostate
cancer patients in Finland. European Journal of Cancer Care, 11, 44-50.

Schneider, R. H., Staggers, F., Alexander, C. N., Sheppard, W., Rainforth, M.,
Kondwani, K., et al. (1995). A randomized controlled trial of stress reduction for
hypertension in older African Americans. Hypertension, 26, 820-827.

Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and Quasi-
experimental designs for generalized causal inferences. Boston: Houghton
Mifflin.

Shen, J., Andersen, R. M., Albert, P. S., Wenger, N., Glaspy, J., Cole, M., et al.
(2002). Use of complementary/alternative therapies by women with advanced-
stage breast cancer. BMC Complementary and Alternative Medicine, 2(8), 1-7.

Shumay, D.M. (2002). Determinants of the degree of complementary and alternative
medicine use among patients with cancer. Journal of A lternative &
Complementary Medicine, 8(5), 661-671.

Sparber, A., Bauer, L., Curt, G., Eisenberg, D., Levin, T., Parks, 8., et a1 (2000). Use
of complementary medicine by adult patients participating in clinical trials.
Oncology Nursing Forum, 27(4), 623-30.

Sparber, A., & Wootton, J. C. (2001). Surveys of complementary & alternative
medicine: Part II. Use of alternative & complementary cancer therapies. The
Journal of Alternative & Complementary Medicine, 7(3), 281-287.

Spiegel, W., Zidek, T., Vutuc, C., Maier, M., Isak, K., & Micksche, M. (2003).

Complementary therapies in cancer patients: prevalence and patients’ motives.
Middle European Journal of Medicine, 115(19-20), 705-709.

151

Thome, S., Paterson, B., Russell, C., & Schultz, A. (2002). Complementary medicine
in chronic illness as informed self-care decision making. International Journal
of Nursing Studies, 39, 671-83.

Tindle, H. A., Davis, R. B., Phillips, R. S., & Eisenberg, D. M. (2005). Trends in use of
complementary and alternative medicine by US. adults: 1997-2002. Alternative
Therapies in Health & Medicine, 1, 42-49.

Vallerand, A. H., Fouladbakhsh, J. M., & Templin, T. (2004). Self-treatment of pain in
a rural community. Journal of Rural Health, 20(2), 166-172.

Vallerand, A. H., Fouladbakhsh, J. M., & Templin, T. (2003). Use of complementary
and alternative therapies in urban, suburban and rural communities. American
Journal of Public Health, 93(6), 923-925.

VandeCreek, L., Rodgers, E., & Lester, J. (1999). Use of alternative therapies among

breast cancer outpatients compared with the general population. Alternative
Therapies, 5(1), 71-76.

Walach, H., Guthlin, C., & Konig, M. (2003). Eflicacy of massage therapy in chronic
pain: A pragmatic randomized trial. Journal of A lternative & Complementary
Medicine, 9(6), 837-46.

Williams, D. R. (1999). Race, socioeconomic status & health: The added effects of

racism and discrimination. Annals of the New York Academy of Sciences, 896,
173-188.

Williamson, A. T., Fletcher, P. C., & Dawson, K. A. (2003). Complementary and

alternative medicine: Use in an older population. Journal of Gerontological
Nursing, 29, 5, 20-28.

Willison, K. D., & Andrews, G. J. (2004). Complementary medicine and older people:

Past research and future directions. Complementary Therapies in Nursing &
Midwifery, 10, 80-91.

Wolsko, P. M., Eisenberg, D., Davis, R. B., & Phillips, R. S. (2003). Patterns &
perceptions of care for treatment of back & neck pain: Results of a national
survey. Spine, 28(3), 292-8.

Wolsko, P.M., Eisenberg, D., Davis, R. B., Ettner, S.L., & Phillips, RS. (2002).
Insurance coverage, medical conditions, and visits to alternative medicine

providers. Archives of Internal Medicine, 162, 281-87.

World Health Organization. (2005). Alternative Medicine. Retrieved January 2, 2005,
from http://www. who.int/medicines/library/trm/trm-strat-engpdfi

152

Wootton, J. C., & Sparber, A. (2001). Surveys of complementary and alternative
medicine: Part I. General trends and demographic groups. The Journal of
Alternative & Complementary Medicine, 7(2), 195-208.

Zoratti, E. M., Havstad, S., Rodriguez, J ., Robens-Paradise, Y., Lafat, J. E., &
McCarthy, B. (1998). Health service use by Afiican Americans and Caucasians

with asthma in a managed care setting. American Journal of Respiratory Critical
Care Medicine, 158, 371-377.

153

NST

will; 11111131111113le