An Assessment of ethnic differences in drug use decision making among African American and White American hypertensive p...

MISSING IMAGE

Material Information

Title:
An Assessment of ethnic differences in drug use decision making among African American and White American hypertensive patients implications for hypertension management
Alternate title:
Implications for hypertension management
Physical Description:
viii, 203 leaves : ill. ; 29 cm.
Language:
English
Creator:
Brown, Carolyn Marie, 1966-
Publication Date:

Subjects

Subjects / Keywords:
Research   ( mesh )
Patient Compliance   ( mesh )
Hypertension -- drug therapy   ( mesh )
Decision Making   ( mesh )
Health Knowledge, Attitudes, Practice   ( mesh )
Cross-Cultural Comparison   ( mesh )
Health Behavior   ( mesh )
African Americans   ( mesh )
European Continental Ancestry Group   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1994.
Bibliography:
Bibliography: leaves 195-202.
Statement of Responsibility:
by Carolyn Marie Brown.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 80382708
ocm80382708
System ID:
AA00011218:00001

Full Text













AN ASSESSMENT OF ETHNIC DIFFERENCES IN DRUG USE DECISION
MAKING AMONG AFRICAN AMERICAN AND WHITE AMERICAN HYPERTENSIVE
PATIENTS: IMPLICATIONS FOR HYPERTENSION MANAGEMENT












By

CAROLYN MARIE BROWN












A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1994














This manuscript is dedicated to my parents, Herbert and

Beverly Brown, who have provided me with unyielding support

over the years. My mother served as an exemplar of strength

under the toughest circumstances. My father has always

backed me in my educational decisions even when he had no

idea how they were to be financed. I thank my two brothers,

Jerome and Dexter, for serving as my connections to reality

outside of graduate school and always keeping me in check.

I would like to thank my grandmothers, aunts, uncles and

cousins for their support as well as their acceptance of my

collect calls. Finally, thanks go out to all of my

wonderful friends and to my colleagues at Xavier University

of Louisiana for their unwavering support and confidence in

me.














ACKNOWLEDGEMENTS


My sincere gratitude goes to Dr. Richard Segal, my

advisor, for his unyielding support, guidance and stamina

throughout my graduate studies. I would also like to thank:

the following committee members for their encouragement,

support and interest in my research project: Dr. Donna

Berardo, Dr. Carole Kimberlin, and Dr. John Lynch.

I would like to express my appreciation to the staff cf

Pharmacy Health Care Administration for their support and

patience over the years. My thanks go to Dennis, Ned and

Dan Nissen for their computer support; the postdoctoral

fellows and graduate students for being great colleagues;

the PHCA faculty for helping me grow; Drs. Ried, Gums,

Rutledge and Ron Stewart for their support and service on rn

expert panel; Drs. Coward and Duncan for their interest in

my project; the pretest sites and Ms. Copeland for their

invaluable assistance with patient recruitment; the American

Foundations for Pharmaceutical Education (AFPE), the Center

on Rural Health and Aging, the Florida Rural Health Research

Center, Bristol-Meyers Squibb, and Marion Merrell-Dow for

providing financial support.

Finally, I would like to extend my genuine love and

appreciation to my parents, family, and friends.















TABLE OF CONTENTS

page

ACKNOWLEDGEMENTS................. ..................... 111

ABSTRACT .............................. ...... ........ vi

CHAPTERS

1 INTRODUCTION.................... ............... 1

The Need for the Study....................... 1
Problem Statement.............................. 7
Purpose and Significance............. ......... 8
Research Questions............................. 9

2 REVIEW OF LITERATURE ........... ......... ...... 11

Drug Use Behavior Research................... 11
Temporal Orientation Research................ 31
Summary......................................... 37

3 RATIONALE AND THEORETICAL FRAMEWORK........... 39

4 METHODOLOGY.................................... 54

Sample Selection............................... 54
Data Collection Procedures.................... 56
Nonrespondent Bias.................... ........ 59
Study Variables....................... .......... 60
Instrument Development and Validation......... 67
Data Analysis.................................. 80
Limitations .................................... 87

5 RESULTS....................................... 89

Sample Description....... ...................... 89
Evaluation of Nonrespondent Bias.............. 91
Evaluation of Final Instrument................ 91
Health Perceptions and Hypertension Temporal
Orientation .... ............. ... ........... 98
Patient Self-Reported Drug Use Behavior........ 100
Ethnic Differences in Health Perceptions and
Hypertension Temporal Orientation .......... 102









page

Hypertension Temporal Orientation and Health
Perceptions..... ................. ...... ...... 106
Health Perceptions and Drug Use Behavior....... 117

6 DISCUSSION, CONCLUSIONS,
AND RECOMMENDATIONS ......................... 129

The Health Belief Model (HBM) and Drug Use
Behavior............ ........................ 129
Ethnicity, Hypertension Temporal Orientation
and Health Perceptions............. ........ 130
Effects of Health Perceptions :n Compliance
with Rx and Use of HR....................... 138
Study Limitations ............................. 142
Conclusions................................... 144
Implications.................................. 144

APPENDICES

A INTERITEM AND CORRECTED ITEM T] TOTAL
CORRELATIONS OF PRETEST SCALES............. 151

B INTERITEM AND CORRECTED ITEM TO TOTAL
CORRELATIONS OF FINAL SCALES................ 154

C PRETEST SURVEY INSTRUMENT..................... 157

D TEXT OF SCREENER AND SURVEY INSTRUMENT OF
FINAL STUDY ................................. 176

E DESCRIPTION OF HOME REMEDIES A-;D THEIR USES... 194

REFERENCES........................................... 195

BIOGRAPHICAL SKETCH ........ ..... ................... 203















Abstract of Dissertation Presented to the Graduate School
of the University of Fl:rida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

AN ASSESSMENT OF ETHI;.C DIFFERENCES IN DRUG USE DECISION
MAKING AMONG AFRICAN AMEICAN AND WHITE AMERICAN HYPERTENSIVE
PATIENTS: IMPLICAT::NS FOR HYPERTENSION MANAGEMENT

By
Car:lyn Marie Brown

:acember 1994

Chairman: Professor Richard Segal
Major Department: Pharmacy Health Care Administration

The purpose of this study was to explain and to predict

drug use behavior among African American and White American

hypertensive patients wi-hin the context of health

perceptions and their subsequent effect on drug use

practices. Using the Health Belief Model (HBM) as the

theoretical framework, this study examined both the

associations of cultural variables (ethnicity and

hypertension temporal or-entation) with health perceptions

and the effects of these culturally-related health

perceptions on compliance with prescription medication (Rx)

and use of home remedies (HR). Data were collected via

telephone interviews with 300 individuals who had been

medically diagnosed as having hypertension and who had not

experienced any major complications of uncontrolled

hypertension.








The study's findings revealed that, after controlling

for selected covariates, African Americans perceived that

the costs of Rx were more burdensome compared to White

Americans. Ethnic group differences, however, were more

apparent when the health perceptions were evaluated within

the cultural context of hypertension temporal orientation.

Factor analysis results indicated that hypertension

temporal orientation can be categorized into three groups:

(1) nonexperiential domain which represented potential

consequences of hypertension; (2) experiential disease

domain which involved the day-to-day dealing with

hypertension; and (3) experiential treatment domain which

involved the daily management of hypertension. African

Americans were rare present oriented than White Americans

concerning their daily management of hypertension. A more

future orientation regarding hypertension management was

associated with higher perceptions of severity,

susceptibility, benefits of Rx, and costs of HR and with

lower perceptions of costs of Rx and benefits of HR.

Significant predictors of compliance with Rx were age, costs

of Rx and benefits of HR. Severity, benefits and costs of

HR, ethnicity, and a poverty by education interaction

significantly explained use of HR.

The results of this study supported a number of

principal conclusions: (1) cultural variables influence

individual perceptions of disease and evaluations of








treatment alternatives; (2) compliance with Rx is affected

by demographic characteristics and by beliefs about costs of

Rx and benefits of HR; and (3) use of HR is a function of

demographic characteristics, perceptions of disease

severity, and evaluations of benefits and costs of HR.


viii














CHAPTER 1
INTRODUCTION


The Need for the Study


Problem of Hypertension


An estimated 60 million persons in the United States

have hypertension. Of these, only about 32 million are

aware of having this disease (The American Heart

Association, 1990). The findings of a study conducted by

the South Carolina Department of Health in 1987 revealed

that 28% of persons with hypertension were unaware of their

hypertension, 12% were aware but not treated, 28% were being

treated but not controlled, and 32% were being treated and

controlled (Gorlin, 1991). Clark (1991) also cited evidence

indicating that more than half of the hypertensives are

either untreated or inadequately controlled. Similarly,

findings from a national report indicated that of the

estimated 60 million hypertensives, possibly one-third are

receiving medical treatment, and only six million are

appropriately managed (The American Heart Association,

1990). Thus, it is reasonable to conclude that somewhere

between 10% to 32% of all people with hypertension are

adequately controlled.










Hypertension has been an enduring problem in the

African American community. Blacks have both a higher

prevalence of hypertension and a higher age-specific mean

blood pressure than that of whites (Klag et al., 1991).

Most studies indicate that statistically significant blood

pressure differences between the races occur sometime after

age 17 (Hildreth and Saunders, 1991). The frequency with

which hypertension occurs in the African American community

ranks as one of the highest in the world. The prevalence of

Stage I (mild) and Stage II (moderate) hypertension is more

that twice as high in black adults as in white adults, and

the incidence of Stage III (severe) hypertension is five to

seven times higher in blacks than in whites (Hildreth and

Saunders, 1991). Moreover, since blacks develop more severe

hypertension earlier and remain untreated or uncontrolled

for longer periods of time, more severe consequences of

longstanding hypertension result (Kannel, 1974). Millions

of African-Americans die each year as a result of strokes,

heart attacks, and other cardiovascular-related illnesses

that are directly linked to uncontrolled hypertension.

Among hypertensives in the United States, blacks are 10 to

18 times more likely to experience kidney failure, and three

to five times more likely to develop chronic heart failure

than whites (Fackelmann, 1991).

Although the treatment of hypertension has generally

improved over the last two decades, the rate of improvement










among African Americans is slower compared to that of White

Americans (Hildreth and Saunders, 1991). The reasons for

this discrepancy are generally unclear. Nevertheless, among

those diseases that plague the black community, hypertension

("the silent killer") ranks as one of the most prevalent.


Compliance with Treatment Regimens


As described earlier, more than half of all the people

with hypertension are either untreated or treated without

sufficient blood pressure control. One of the major reasons

cited for inadequate control of high blood pressure is the

failure to adhere to treatment regimens (Gorlin, 1991). In

general, noncompliance rates for many medications have been

found to be in the 25% to 50% range (Fletcher, 1989).

Based upon a review of literature before 1984,

Christensen (1985) cited a number of factors that have been

found to be associated with compliance. These factors

include the perceived seriousness of the disease, the

perceived efficacy of the treatment, the complexity of the

prescribed regimen, the duration of therapy, adverse effects

of the prescribed regimen, the amount of medical supervision

the patient receives, and patient satisfaction with the

physician visit. An additional factor, not included in his

review, is locus of control which has also been found to be

related to compliance (Rotter, 1954; Balsmeyer, 1984). Some

evidence suggests that sociodemographic variables (e.g.,










socioeconomic status and race) are related to compliance

(Kirscht and Rosenstock, 1977; Nelson et al., 1978).

However, a later review of compliance literature failed to

provide any evidence indicating that general compliance

rates remarkably differed across sociodemographic groups

(Fletcher, 1989).

Noncompliance rates among hypertensive patients vary

greatly, with an approximate range of 30% to 70% (Sackett

and Snow, 1979; Clark, 1991). For example, Klein (1988)

found that approximately 60% of hypertensive patients took

their medications as prescribed. In a national survey of

patients who stopped taking their antihypertensive

medications, approximately half (46%) reported that they

stopped because they thought that they were cured, and 25%

indicated they stopped because they thought their doctor had

told them to stop (Gallup and Cutugno, 1986). Surprisingly,

only 11% of them reported stopping because of side effects

and 6% stopped because they could not pay for the

medication. These findings are especially important since

they highlight the role of patient perceptions of illness in

drug use decisions, as evidenced in the nearly 70% who

stopped taking their medications because of subjective

reasons rather than because of drug-related or financial

reasons.

Hypertension and its management provide an interesting

topic of study for three reasons: (1) hypertension has both










folk-based and scientifically-based interpretations among

various groups; (2) hypertension is an asymptomatic

condition; and (3) the value of managing hypertension is

based upon reducing the probability of future negative

consequences. In the first case, patients with folk-based

interpretations of high blood pressure may be more inclined

to use alternative treatments than those who hold scientific

views. In the second case, the asymptomaticity of

hypertension is generally problematic in all patients in

that they find it difficult to engage in illness behavior

(e.g., taking medication) when they do not feel ill.

Moreover, in the short run, many prescribed antihypertensive

regimens make patients feel worse than when the condition

goes untreated. In the third case, because the management

of hypertension is based on future events, patients must be

motivated to take their medications such that the future

benefits of therapy are recognized. Essentially, adherence

to prescribed treatment regimens poses a special medical

care dilemma; all three factors just mentioned figure into

patients' decisions about and evaluations of treatment

options.


Culture, Health Beliefs, and Health Behaviors


Of the many factors shaping health beliefs and

behaviors, culture, as indicated by ethnicity, has been

shown to be of particular relevance (Harwood, 1981). This










assertion was supported by a review of sociological,

anthropological, and psych-ological health-related research

conducted by Landrine an: Klonoff (1992) showing that health

beliefs of White Americans and of ethnic-minorities in this

country are culturally shaped and organized. Accordingly,

the health beliefs of Whtre Americans differ substantially

from those of African-Arericans, and there also is

considerable variation in health beliefs within each of

these groups (Landrine and Klonoff, 1992; Harwood, 1981).

Through life experiences, members of an ethnic group

often acquire certain perceptions and beliefs about health

and illness that are consistent with the customs and values

of their particular cult-re. Moreover, these culturally-

relevant health beliefs and perceptions guide individuals'

health actions and behaviors. Health beliefs and behavior

encompass a wide variety of knowledge and activities ranging

from estimates of the seriousness of various illnesses and

one's susceptibility to :hem to adherence to drug therapy

regimens (Becker and Mairan, 1975). An inherent assumption

behind the concept of "compliance" is that the patient has

no prior beliefs and attitudes about illness that may affect

whether or not medical regimens are carried out. Clearly,

if the indicated treatment is not congruent with the

patient's model of illness, then it is likely that the

patient will not "comply." Insight as to how ethnicity, as

manifested in culture, influences health beliefs and










perceptions would prove fruitful in advancing our

understanding cf drug use decision making.


Problem Statement


As indicated by the preceding discussion, patient

compliance has seen an enduring problem in medical care,

especially among patients with chronic diseases (e.g.,

hypertension). In the area of hypertension, adherence to

therapy has been found to be lower in black patients than in

white patients, especially in black males (Gorlin, 1991).

There is an abundance of literature that examines patient

compliance with treatment regimens with particular emphasis

on the identification of factors associated with compliance

behavior. Although these factors have been crucial to our

ability to predict compliance behavior (Ried and

Christensen, 1938), a much deeper understanding is needed of

the decision process of patients within the context of

health and illness perceptions and its subsequent effect on

health and medical care practices.

A major problem in compliance research is its failure

to acknowledge the multidimensionality of drug use. Failure

to "comply" with prescribed medical regimens does not mean

that the patient goes untreated. Despite the evidence that

culture has been shown to influence the beliefs and

perceptions of members of different ethnic groups, little is

known about possible ethnic differences in the decision-










making associated with drug use behavior. Moreover, it is

not clear how these differential illness and treatment

perceptions are related to the use of formal and informal

treatment alternatives. No research has explicitly assessed

ethnic differences in drug use decision making. It is

essential that we evaluate and integrate these cultural and

ethnic differences into treatment plans so that treatment

programs can have increased potential for success.


Purpose and Significance


The overall goal of this project is to assess drug use

behavior among African American and White American

hypertensive patients. Specifically, this project seeks to

examine the relationships among ethnicity, poverty status,

termoral orientation, and health perceptions and their

individual and combined relationships with drug use

behavior. The specific variables of interest are based upon

the six dimensions of the Health Belief Model (HBM) and will

be discussed in a succeeding section.

Since hypertension affects such a large proportion of

the United States population, especially in the African

American community, it is very important to determine how

ethnic and cultural influences are related to health beliefs

and perceptions that ultimately guide both formal and

informal drug use practices. In addition, a study is needed

to assess this aspect of drug use which is, for the most








9

part, unexamined in the literature. This understanding will

hopefully enable health care practitioners to facilitate

more appropriate drug use among hypertensive patients.

Moreover, health care providers would be better equipped to

plan strategies for the management of hypertension that have

increased potential for effectiveness and that are

culturally-relevant to members of both the African American

and White American populations.


Research Questions


In meeting the goals of this project, the following

research questions will be addressed:

1) What is the relationship of ethnicity, poverty

status, and temporal orientation to health

perceptions of hypertensive patients?

2) What is the relationship of these factors

individually with drug use behavior?

3) What is the relationship between health

perceptions and hypertensive patients' drug use

behavior?

4) Do the two ethnic groups significantly differ with

respect to drug use behavior? Is this difference

accounted for by intergroup differences in poverty

status, temporal orientation and health

perceptions?








10
5) Does the conceptual model adequately explain drug

use behavior among African American and White

American hypertensives?














CHAPTER 2
REVIEW OF LITERATURE


This literature review is divided into two sections:

(1) drug use behavior research, and (2) temporal orientation

research.


Drug Use Behavior Research


Patient drug use behaviors have been the focus of much

research over the years. Most studies investigating drug

usage have been referred to as compliance research. As

implied by the term "compliance," many of these studies have

suggested that patients are at blame for not following the

prescribers' drug orders. In recent years, however,

compliance research has moved away from placing blame on the

patient, acknowledging the patient as an active participant

in drug use decisions. In essence, what is seen as

noncompliance from a practitioner's standpoint may be deemed

as "rational decision making" by the patient (Donovan and

Blake, 1992). Due to the confusion and negative connotation

associated with the term "compliance," the term "drug use

behavior" is used in this research effort. Nevertheless, in

remaining consistent with the most customary term in the










literature, compliance is used in selected parts of this

chapter.


Overview of Compliance with Medication Regimens


In their reviews of empirical studies of patient

compliance, Christensen (1985) and Fletcher (1989)

determined that noncompliance with medication regimens is

between 25% and 50%. Noncompliance rates are even higher

among patients with chronic diseases, such as hypertension,

requiring long-term or lifetime medication therapy. On

average, a little more than 50% of hypertensive patients

take their medication as prescribed (Sackett and Snow,

1979), although compliance rates as low as 35% to 40% have

been found among hypertensives (Dirks and Kinsman, 1982;

Sackett and Snow, 1979). For example, Nelson and his

colleagues (1978) found that of the 121 patients being

treated for hypertension in their study, almost 60% of the

patients had missed one or more doses during the 28 days

prior to the data collection.


Factors Associated with Compliance


Patient compliance studies have been reviewed by

Haynes, Taylor, and Sackett (1979), DiMatteo and DiNicola

(1982), Christensen (1985), and later by Fletcher (1989).

In these reviews, the authors identified many factors

associated with compliance behavior. These factors can be










grouped into three categories: (1) psychological factors,

(2) socio-environmental factors, and (3) patient-provider

relationship factors.

Psychological factors. Many researchers and

practitioners have postulated that a person's beliefs and

attitudes predispose them to either compliant or

noncompliant behaviors. As a result, investigators have

employed a number of psychologically-based approaches to

explain drug use behavior among patients. Some evidence

indicates that noncompliance can be the result of deliberate

attempts by patients to exert control over their illness

(Conrad, 1985). Similarly, researchers have found that the

health locus of control of individuals is associated with

compliance behavior (Johnson and Beardsley, 1978; Schlenk

and Hart, 1984). Other research indicates compliant

behavior is related to the individual's belief about the

threat of the disease or its sequelae and the perceived

benefits versus costs of treatment (Ried and Christensen,

1988; Nelson et al., 1978; Becker and Maiman, 1975). Still

other researchers have found that lay beliefs, particularly

when inconsistent with western medical thought, will lead to

noncompliant behavior (Donovan, Blake and Fleming, 1989).

Socio-environmental factors. A second approach to

understanding compliance involves the social and

environmental influences in peoples' everyday lives. This

perspective recognizes that drug use does not occur in a










vacuum, but that it is one of a myriad of influences in

one's daily life. The process of carrying out medication

regimens requires thought, time, money and other resources

that impact patient behav\ rs (DiMatteo and DiNicola, 1982).

Graveley and Oseasohn (19-1) found that the complexity of

the regimen, the duration of treatment, adverse effects of

the regimen, and financial costs of therapy are all related

to compliance behavior. ::her research indicated that

social support is also instrumental in managing chronic

illness (e.g., hypertension) (Strogatz and James, 1986;

Schlenk and Hart, 1984). Yet other studies have indicated

the importance of social influences (e.g., friends and

family) and culture in predicting medication taking

behaviors (Ried and Chris:ensen, 1988; Becker and Maiman,

1980).

Provider-patient relationship factors. The third

approach to studying compliance is concerned with the

interaction of the patient and the health care provider.

Numerous studies have suggested that the quality of the

physician-patient encounter was partly responsible for

noncompliance. For example, patients' dissatisfaction with

their visits to physicians, physicians' failure to

effectively communicate drug- and disease-related

information, and lack of care or genuine concern on the part

of the provider (Garrity, 1981; Christensen, 1985; Kaplan et

al., 1989) have all been found to be associated with










patients' resistance to following medication regimens.

Garrity (1981) found that noncompliance can be partially

explained by the amount of medical supervision a patient

receives once drug therapy has been initiated.

The large number of factors associated with compliance

behavior is a clear indication of the complexity of drug use

practices. However, DiMatteo and DiNicola (1982) argued

that there is value in this diversity in that it

helps to counteract the tendency of medical professionals to

conceptualize noncompliance as a unitary phenomenon "

(p. 19). Since the decision to comply with medical regimens

ultimately lies with the patient within the context of

her/his beliefs and values, drug use behavior is necessarily

" embedded in an intricate web of social and

psychological factors" (DiMatteo and DiNicola, 1982, p. 19).

Peoples' beliefs about and evaluations of health and medical

care practices affect their choices among alternative

actions. Accordingly, most models of health behavior and

subsequent studies of patient drug use practices have

generally included patient beliefs and perceptions as a key

element in health and medical care research (Cummings,

Becker, and Maile, 1980).


Health Beliefs and Health Behaviors

Health belief model

One of the most commonly cited models of health

research is the Health Belief Model (HBM). The HBM was










originally developed to explain preventive health behavior

(Rosenstock, 1966). However, it has since been broadened to

explain compliance with health recommendations and

medication regimens (Becker and Maiman, 1975). The central

tenet of the HBM is its focus on subjective beliefs in the

context of individual behavior. The HBM is rooted in the

value expectancy tradition and thus serves as a framework to

explain patients' motivations to engage in health behaviors

(Becker, 1974). A key assumption of the HBM is that

motivation is a necessary prerequisite for action and that

individual perceptions are selectively governed by motives

(Becker, 1974).

There are six primary elements of the HBM which include

(1) perceived susceptibility to the disease or its sequelae,

(2) perceived seriousness of the disease, (3) perceived

benefits of treatment alternatives, (4) perceived barriers

to (or costs of) treatment alternatives, (5) cues to action,

and (6) modifying factors. In the context of individual

health behavior, the HBM posits that the probability that an

individual will engage in a recommended health action

depends upon: (1) one's perception of the threat of the

disease, which is determined by the perceived seriousness of

the disease and the perceived susceptibility to the disease

or its consequences; (2) one's weighing of the perceived

benefits of action as compared to the perceived barriers to

action; and (3) cues to action to "trigger" the appropriate










response. These cues could be internal (e.g., symptoms) or

external (e.g., mass media campaigns). The "motivating

force" that results from the four perception variables is

believed to be affected by modifying factors such as

demographic and sociopsychological variables. In essence,

the HBM examines health behavior in terms of an individual's

psychological readiness to take action in combination with

the degree to which the individual believes the recommended

action will reduce the perceived threat (Becker, 1974).

Empirical studies of HBM in relation to drug use behaviors

In 1974, an entire issue of Health Education Monographs

was dedicated to "The Health Belief Model and Personal

Health Behavior" (Becker, 1974). Findings from HBM research

were reviewed in this monograph and the author cited

considerable support for the HBM in explaining health

actions. Since then, the HBM continues to be a leading

conceptual framework for understanding and predicting health

and medical care behaviors (Rosenstock, 1990). Of all the

basic elements of the HBM, the dimensions that have received

the most empirical testing include perceived susceptibility,

perceived severity, perceived benefits, and perceived

barriers (Janz and Becker, 1984; Harrison, Mullen, and

Green, 1992). The other elements, namely action cues and

modifying factors, have received the least attention in the

empirical literature using the HBM. Although the action

cues may be instrumental in facilitating the "appropriate"








18

behavior, this category has seldom been included in studies

that used the HBM as a theoretical framework. A large

number of studies, performed in many types of settings with

varying research designs, have used the HBM to explain

prevention, illness, and sick role behaviors. Since this

research effort has a primary emphasis on sick role

behaviors, two primary review articles summarizing HBM

research will be presented with particular emphasis on

findings involving sick role behaviors.

Janz and Becker (1984) summarized HBM research over a

10-year period (1974-1984) and also reviewed selected

findings prior to 1974 in order to assess the HBM's overall

performance. In their review of HBM studies published

between 1974 and 1984, the authors employed the following

inclusion criteria: (1) each study had to include at least

one behavioral outcome; (2) only studies that examined the

four primary HBM perception variables; and (3) only studies

involving medical conditions and adults, thus excluding

dental studies and studies with children. On the basis of

these criteria, 29 HBM-related studies published between

1974 and 1984 were included in their review. In addition,

17 studies prior to 1974 were also included. In total, 46

studies were summarized: 18 prospective and 28

retrospective. The 46 studies covered different types of

health behaviors: preventive health behaviors (24), sick

role behaviors (19), and clinic utilization (3).










Significance ratios (SR) were calculated by dividing the

number of positive, statistically significant findings cf a

HBM dimension by the total number of studies reporting

significance levels for that dimension. In general, the

authors found considerable support for the HBM with

prospective results being generally as strong as

retrospective ones (Janz and Becker, 1984).

In the 19 studies examining sick role behaviors, which

included compliance with hypertension regimens, Janz and

Becker (1984) found the perceived barriers dimension

(SR=92%) of the HBM to be the most powerful predictor of

compliance behavior, followed by the perceived severity

(SR=88%) and the perceived benefits (SR=80%) dimensions.

The perceived susceptibility category (SR=77%) was found to

be the weakest predictor of compliance, although this may be

due to the difficulty of operationalizing vulnerability when

individuals already have the disease (Janz and Becker,

1984). Until recently, Janz and Becker's (1984) review

represented the only formal attempt to offer conclusions

about the overall predictive validity of the HBM. However,

in questioning the conclusions drawn by Janz and Becker

(1984), a second examination of HBM variables was carried

out by Harrison and his colleagues (1992).

Harrison et al. (1992) conducted a meta-analysis of the

relationships between the four HBM dimensions and health

behavior. They applied similar inclusion criteria to those










employed by Janz and Becker (1984) with one notable

exception: the studies had to include some measure of

reliability. Only 16 studies met the criteria for

inclusion, most of these studies were not includedd in the

Janz and Becker (1984) review. Mean effec- sizes were

calculated across all studies and for subcazegories of

studies which were grouped by study design (Prospective and

Retrospective) and by type of dependent variable (Screening

Behaviors, Risk Reduction Behaviors, and Adherence to

Medical Regimens) resulting in 24 mean effect sizes. Of

these 24 mean effect sizes, 22 were found zo be positive and

statistically significant. Significant differences were

found between prospective and retrospective designs. For

example, for retrospective designs, perceived benefits and

perceived costs had larger effect sizes and perceived

severity had smaller effect sizes. However, homogeneity was

rejected in 15 of the 22 effect sizes, indicating that the

same construct was not being measured across studies.

Overall, the mean effect sizes of four HBM dimensions

ranged from 0.01 to 0.30. Accordingly, the authors

concluded that any one of the HBM dimensions could not

explain more than 10% of the variance in health behavior.

In regard to adherence to medical regimens (including

antihypertensive regimens), the mean effect sizes for each

dimension ranged from 0.10 to 0.21. Thus, any one dimension








21

of the HBM could explain no -ore than 5% of the variance in

compliance with medical regirens.

The two reviews are see-ingly at opposite ends of a

continuum in that Janz and Eszker (1984) present a more

favorable view of the HBM cz-ared to Harrison et al.

(1992). Nevertheless, there is so-e agreement between the

two. Both reviews show that HBM variables are significantly

related to health behavior. However, the strength of these

relationships across different study designs and health

behaviors represents the primary point of departure between

the two reviews.

Harrison et al. (1992) arguedd that the use of

significance ratios by Janz and Becker (1984) led to biased

evaluations of the studies since significance ratios are

affected by factors such as effect size, sample size, and

homogeneity of variances across studies. These differences

in methodologies used by Harrison et al. (1992) and Janz and

Becker (1984) seemingly led co very different conclusions.

In addition, the lack of honcgeneity found among 15 of the

22 significant mean effect sizes led Harrison et al. (1992)

to question the validity of .anz and Becker's (1984) summary

of HBM studies. One possible source of the problem is that

Janz and Becker (1984) inclined HBM studies regardless of

whether reliability was assessed or not.

These related issues of effect size and homogeneity are

evident in the criticisms that have been leveled against the










HBM since its inception. Although they are necessarily

related, each issue is taken separately for the sake of

discussion. In the first case, the HBM has been criticized

for its lack of ability to explain significant amounts of

variance in health behavior, suggesting that additions to

the model may be necessary. Some research has demonstrated

the value of adding other variables such as self-efficacy

and social norms to the HBM, thus improving its predictive

power (Ried and Christensen, 1988; Bandura, 1977). However,

in defense of the HBM, Janz and Becker (1984, p. 44) argued

that "it is important to remember that the HBM is a

psychosocial model; as such, it is limited to accounting for

as much of the variance in individual's health-related

behaviors as can be explained by their attitudes and

beliefs." The second issue of homogeneity of variance

across studies has also been a criticism of the model. Lack

of homogeneity cf variances may be a result of the lack of

standardized instruments of HBM variables. As a result,

researchers have developed their own approaches to

operationalizing HBM variables, resulting in problems with

the interpretation and comparison of findings across studies

(Janz and Becker, 1984; Harrison, Mullen, and Green, 1992).


Culture and Druo Use Behavior

Culture and health beliefs

Although there is no universally accepted definition of

culture (Goodencugh, 1981), for the purposes of this










research, culture is defined as the learned ways in which

individuals were taught to perceive and react to their world

(Henderson and Primeaux, 1981). Culture is not necessarily

an inherited biological attribute, but it is a set of

beliefs or standards that reflect the ways in which people

have learned to adapt and survive in their contextual world.

Accordingly, culture is the result of both natural and

environmental influences (Tosi, Rizzo and Carroll, 1990).

These influences create the emergence of a fundamental set

of values and beliefs regarding what kinds of behavior are

appropriate in given contexts (Goodenough, 1981). In

essence, culture constitutes a system of standards for

behavior relevant to a particular situation. An assumption

underlying this research effort is that African Americans

and White Americans differ culturally due to differences in

life chances and experiences (Willie, 1979; Willie, 1980;

Reiners, 1984).

Goodenough (1981) proposed that an individual's overall

culture is made up of different "subcultures." These

components of culture or "subcultures" are assumed to be

abstracted from higher level individual principles and

standards that dictate which system of behaviors are

appropriate for a given situation (Goodenough, 1981). These

subcultures become more or less relevant depending upon

which level a person is functioning at given a particular

situation. Thus, individuals can be thought of as being










"multicultural" since they can choose which standards for

behavior are appropriate given the situation in which they

are functioning. This idea is exemplified in Goodenough's

(1981) concept of "operating culture."

Operating culture has been described as an individual

selecting the culture in which (s)he wishes to operate for

any given situation. Goodenough (1981, p. 99) points out

that "this is frequently the case among educated Americans

of foreign parentage or among people who have gained

acceptance in a higher social class than that in which they

were raised as children." For example, the beliefs and

norms for behavior associated with an individual's personal

life (i.e., personal subculture) may not necessarily be the

same as those associated with her/his professional life

(i.e., professional subculture). Hence, an individual in

this country could "profess and practice the ethics of

Christianity, of laissez-faire business enterprise, and of

power politics, each in its own segregated context, without

feeling that" (s)he is "being inconsistent" (Goodenough,

1981, p. 107). In essence, a person, given her/his

repertoire of behavioral standards, can selectively choose

which cultural principles are appropriate given the

situation and the people (s)he is dealing with (Graham,

1981).

One aspect of culture relates to individuals' views

about health and illness. In this research effort, beliefs








25

and values associated with health matters are an element of

one's personal subculture and an individual's personal

subculture is believed to be largely influenced by her/his

ethnic background. These culturally-related beliefs and

values guide subsequent behaviors in health and medical care

situations.

In a review of sociological, anthropological, and

psychological health-related research, Landrine and Klonoff

(1992, p. 267) reported that "the health beliefs of White

Americans, of ethnic-minority groups in the United States,

and of cultures around the world are culturally constituted

and situated." As such, the health beliefs of many White

Americans and that of many African Americans likely differ,

and there are variations in health beliefs and behavior

within each of these groups. The intent of this research is

not to imply that all African Americans or all White

Americans within each group hold the same views about

disease and illness. Rather, the intent is to show enough

homogeneity, as indicated by culture, within each of these

groups to warrant empirical investigation in order to

increase understanding of these effects (Landrine and

Klonoff, 1992).

In general, White Americans view illness as an

internal, circumstantial abnormality that is caused by some

natural causal agents such as genes, viruses, bacteria, and

stress (Landrine and Klonoff, 1992). In contrast, many










ethnic-cultural minority groups, including African

Americans, view illness as a "a long term, fluid, and

continuous manifestation of long term changing relationships

and dysfunctions in the family, the community or nature as

well as in the relationship between the individual and any

one of these" (Landrine and Klonoff, 1992, p. 268). These

culturally situated health beliefs imply causal agents that

are based in external, supernatural forces. Such views are

not necessarily or totally conducive to drug therapy when

the illness may not be viewed as entirely within them

(Landrine and Klonoff, 1992). As a result, many ethnic-

cultural minority groups have their own remedies for illness

and their own faith healers and root doctors who are

believed to better fulfill their needs. Often times,

African Americans choose alternative medicine because the

practices of folk healers and the usage of remedies are more

consistent with their model of illness and its causes (Snow,

1981). Bailey (1987) argued that the health belief system

and subsequent health care seeking behaviors of many African

Americans was largely derived from historical experience

whereby they were systematically excluded from mainstream

medical services on an explicitly racial basis. It is

believed that this exclusion played a pivotal role in the

employment of alternative mechanisms for obtaining medical

care by many African Americans. Moreover, since religion

was seen as the major means of coping, many approaches to










healing had a spiritual or religious basis (Henderson and

Primeaux, 1981). This spiritualism serves as a vital

component of folk medicine.

In an ethnographic study of folk beliefs among low-

income African Americans in the southwest, Snow (1981) found

that their belief system could be categorized into natural

and unnatural events. This dichotomy can best be understood

in terms of the extent to which the harmony of nature is

disrupted. Natural events involved "the world as God made

it and as He intended it to be" (Snow, 1981, p. 82).

Natural events are characterized by illnesses derived from

natural sources (e.g., impurities from air, food, and water)

and also from God's punishment, because it is believed to be

His will. Natural illnesses include various theories about

the conditions of the blood (e.g., hot-cold, thin-thick,

high-low). Unnatural events involved, at best, an

interruption of God's plan; and, at worst, they represented

the work of the devil. Unnatural events included those

having to do with the supernatural (e.g., a mojo, roots,

hexes, spells). In those cases where the cause of illness

is attributed to divine punishment or unnatural sources, it

is believed that the physician is useless in that "you

cannot fight God or the Devil with drugs" (Snow, 1981, p.

84).

In an illustrative example of folk beliefs and

compliance among 54 southern African American hypertensive










females, Heurtin-Roberts and Reisin (1990) found that one

group believed in biomedical hypertension and the other

group believed in folk-based interpretations of high blood

pressure. Folk-based illnesses consisted of two diseases,

"high blood" and "high-pertension." "High blood" was

described as a disease characterized by the thickness,

richness, and hotness of the blood such that the blood rose

slowly and remained there for prolonged periods. "High

blood" was considered treatable through dietary means such

as less spicy foods and the ingestion of folk remedies

(e.g., vinegar). These treatments were believed to cool and

thin the blood. On the other hand, "high-pertension" was

conceived as a blood disorder caused by "the nerves." Thus,

"high-pertension" was an episodic, unstable disease caused

by emotional excitement and stress. The treatment most

amenable for "high-pertension" was one's control over

her/his emotions and the alleviation of stress. In relating

these health beliefs to compliance (measured by pill

counts), the authors found that of the group who believed

they had either of the two folk illnesses, 63% complied

poorly (pill use < 60%) with antihypertensive treatment

compared with 27% of those who believed in the biomedical

hypertension (Heurtin-Roberts and Reisin, 1990). In a

related study about folk beliefs, Bailey (1987) found that

the most influential factor affecting health care seeking

behavior among a sample of 203 African Americans was self-








29

care. Self-care regimens in this study ranged from "living

life according to God's standards" to consuming "a pinch of

garlic" after meals. In this study, individuals sought

formal medical care services only after traditional remedies

had been tried and/or folk healers had been consulted. Some

participants still maintained their self-care practices even

after seeking formal medical help (Bailey, 1987).

Role of social class and ethnicity

Numerous reports have described the disparities between

African Americans and White Americans in the United States

with respect to income, occupation, and education, three of

the major factors utilized for assessing socioeconomic

status (Department of Health and Human Services, 1985;

Reimers, 1984). On average, blacks have less education than

whites. Those with equivalent education have access to

fewer job opportunities than whites. Moreover, African

Americans with equivalent occupations are likely to be paid

less than White Americans (Reimers, 1984). Clearly, these

major socioeconomic differentials that persist between

African Americans and White Americans affect their

respective life experiences and access to and utilization of

medical care services (Reimers, 1984). Beliefs are formed

on the basis of learning through life experiences and are

thus fundamentally related to socioeconomic status. As

such, many African Americans necessarily approach health and

illness in different ways from the dominant culture (Bailey,










1987). Snow (1981, p. 93) argues that "since many avenues

to advancement are largely closed for African Americans,

they fall back on the cultural system where religion .

allows them to deal with a hostile world." -.is may

partially explains why the beliefs in folk medicine seem

more pronounced in lower-class African Americans coupled

with the recognition that folk belief studies typically are

carried out among selected groups of lower-cl'ss urban and

rural populations.

DiMatteo and DiNicola (1982) found that lower-class

people value health less than the middle and .pper classes.

These authors speculated that this finding may be the result

of the lower classes having many competing basic needs

(e.g., food and shelter) that are already satisfied among

the middle and upper classes. However, Goering and Coe

(1970), in challenging this victim centered explanation,

found among lower-class blacks and whites tha: they desired

medical care but were limited by situational constraints.

In addition, the authors found significant differences

between black and white subjects in their utilization of

physician services. Blacks were significantly less likely

than whites to have seen a physician during the year before

the study.

Although the preceding study was conducted over 20

years ago, similar evidence is found today whereby access to

professional medical services for African Americans still








31

lingers behind that of White Americans (Department of Health

and Human Services, 1985). Snow (19=1, p. 79) warns us,

however, that "it would be fallacious to assume that a lack

of professional health care neans nc health care at all."

It is no wonder that "a viable folk belief system is part of

American black culture" (Sno-;, 1981, p. 79).


Temporal Orientation -esearch


Common Models of Time Perspective


Some researchers have examined -he role of time

orientation in peoples' choices and actions (Hendriks and

Zeltzer, 1986; Bergadaa, 1990). Three well-known time

models are: (1) the linear-separable, (2) the circular-

traditional, and (3) the procedural-traditional models

(Graham, 1981). The linear-separable model is characterized

by the perception of the linear nature of time, i.e.,

distinct past, present, and future coponents. This model,

consistent with the Anglo perception of time, incorporates a

heavy future orientation and how time is spent is based on

future desirabilities. Individuals who perceive time in

this way see past time as gone and t-at it never comes back

(Graham, 1981).

The second model, circular-traditional, portrays time

as a circular system. This model, consistent with the time

perspective predominant in many Latin countries, embodies a

heavy present orientation since time is conceived as a










circular system whereby the same events are repeated

according to some cyclical order (Graham, 1981). In this

view, time "comes bac- around" and there is generally no

reason to believe th= future will be any different from the

past (Graham, 1981). In the United States, this perception

of time sees most consistent with individuals who are poor

and less educated (Harwood, 1981; Graham, 1981).

Finally, the prc:edural-traditional time model is

characterized by a procedure-driven orientation as opposed

to a time-driven one (Graham, 1981). In this model, time,

in a mathematical sense, is virtually irrelevant to

activities undertaker. This perception is typical of Native

Americans or American Indians. Things are done when the

"time is right" having little or nothing to do with time per

se. This model incorporates a profound present orientation.


The Role of Time in A-erican Society


The time orientation of individuals is one of the most

powerful influences cn how we think and act both as a

society and as individuals (Graham, 1981). Time perception

is a part of culture, influencing our perception of the

world and our ensuing behaviors (Graham, 1981). The way

time is perceived has been a distinguishing characteristic

among cultures the world over. In western industrial

societies, like the United States, much of what we do is

heavily embedded in a future orientation. For example, many










American corporations, such as insurance companies, became

prosperous only after American society developed a strong

sense of future (Gonzalez and Zimbardo, 1985). Graham

(1981) argued that the United States, as a society, exhibits

a time perspective that is consistent with a linear-

separable model of time (i.e., a futuristic society). This

emphasis on future orientation was referred to as the

"public culture" of the US (See Graham, 1981, p. 107).

However, since the United States is made up of several

different cultural groups, the traditions guiding their

"private cultures" may not necessarily agree with those

associated with the dominant culture. Goodenough (1981, p.

107) points out that "other systems of standards associated

with a specific subgroup may be used by its members

regularly when they deal only with one another." This

assertion makes sense intuitively in that an individual's

professional endeavors and achievements are largely

dependent on the "rules of society." Therefore, one's

professional subculture would likely conform to the

standards associated with the overall society (i.e., public

culture) because this is one domain where much public

interaction occurs.


Time Perspective and Operating Culture


Given the meanings of culture and subculture presented

earlier, individuals are, in a sense, "multicultural" in










that they mentally house more than one group of standards

for behavior (Goodenough, 1981). That is, "a person shares

many different cultures with many groups" (Graham, 1981, p.

338) and the particular standards that are deemed applicable

are dependent upon the group and the activities involved.

Consequently, an individual can choose which culture or set

of standards is relevant to a given situation, i.e., the

choice of an operating culture (Goodenough, 1981). Since

temporal orientation is a part of culture, individuals

should be capable of moving from one temporal model to

another depending upon the people involved and the

particular circumstance (Graham, 1981). Thus, a person can

operate under a linear-separable model (future orientation)

in professional situations, but operate under one of the

other two models (both present orientations) in personal

circumstances. For example, the recognition of this ability

is useful in explaining why progressive professionals

(future oriented) continue to smoke cigarettes (present

oriented) despite probable long-term consequences.


Temporal Orientation. Health Beliefs and Health Behaviors


Graham's (1981) and Goodenough's (1981) research

suggests that individuals have the capability to operate

under a different set of beliefs and time perceptions

depending upon the particular context. Likewise, the health

component of one's personal subculture is comprised of








35

health-related beliefs and temporal orientations that guide

subsequent health actions. Temporal orientation is

multidimensional and individuals can move between these

dimensions given a particular situation.

The existing research on general temporal orientation

suggests that individuals who are future oriented will be

more motivated to act in ways to achieve desired future

positions. For example, in a naturalistic inquiry of the

temporal system of French consumers, Bergadaa' (1990) found

that future oriented respondents held an "attitude of

action" in that they actively sought opportunities for self-

improvement. On the other hand, present oriented

individuals held an "attitude of reaction" whereby they

reacted to a situation if or when it occurred. No research

could be identified that explicitly examined individual time

perception specific to health and illness and its

relationship with health beliefs and illness behavior.

However, some research has been conducted involving time

orientation and the use of preventative services. As

expected, it is generally found that future oriented people

are more likely to engage in health promoting behaviors than

those who are less future oriented (Harwood, 1981).

Although the explanation of preventative health practices is

important, it is just as important to understand how the

different dimensions of temporal orientation relate to

disease management as opposed to disease prevention,










particularly among individuals who have chronic diseases

like hypertension.


Social Class and Ethnic Variations in Time Perspective


Some evidence suggests that temporal orientation is

related to social class (Leshan, 1952). For example,

Bergadaa (1990) noted that people who have suffered poverty

tend to orient themselves in the present. Moreover, other

sociodemographic factors such as education seem to influence

time orientation (Leshan, 1952). Thus, lower classes and

less educated individuals are more past and present oriented

than other groups (Leshan, 1952; Bergadaa, 1990).

Past experiences influence the temporal orientation of

individuals (Bergadaa, 1990). Social class is a part of

life experiences and thus a part of culture. Bailey (1987)

argued that many cultural attributes found among African

Americans such as the significance of spiritualism and an

emphasis on present orientation are generally in conflict

with mainstream values such as individualism and a future

orientation. As discussed earlier, because of the direct

correlation of ethnicity and socioeconomic status in this

country, it is possible that those things that are

attributed to class differences may very well be explained

by cultural factors.










Summary


Drug use behavior is a complex and multifaceted

phenomenon. Noncompliance can occur with anyone at anytime.

Moreover, even though the ultimate behavior (i.e.,

compliance or noncompliance) manifested by various ethnic

groups is similar, the underlying reasons for the behavior

may be different.

One troubling aspect with compliance research concerns

how compliance is defined. Often, compliance is viewed as a

unidimensional view of drug use practices, resulting in

patients being described as compliant or not compliant. The

general label of "compliance" fails to acknowledge the

multidimensionality of drug use behavior, resulting in the

loss of vital patient information. Compliance studies have

not investigated the role of alternative medicines in

patients' treatment decisions. It is conceivable that

patients who are labelled as noncompliant by the medical

establishment may actually be managing their conditions in

ways most consistent with their models of illness. For

example, a person may not be "compliant" with her/his

prescribed regimen but be perfectly "compliant" with some

alternative regimen. In fact, many ethnic minority patients

make simultaneous use of both the formal and informal

medical care systems (Scott, 1981).

Temporal orientation, as a component of individual

belief systems, influences people's preferences for










particular courses of action (Graham, 1981). Whether a

person is past, present, or future oriented has been shown

to be related to how (s)he thinks, feels, and behaves

(Gonzalez and Zimbardo, 1985). In relation to drug use

practices, temporal orientation can systematically alter an

individual's perceptions of the costs and benefits

associated with alternative options. Essentially, time

perception plays a key role in an individual's evaluation of

a treatment's value which ultimately leads to a preferred

course of action.

As was shown in the preceding sections of this review,

individual beliefs and perceptions, influenced by culture,

can and do serve as primary determinants of drug use

practices. What is needed is an understanding of how

patient variables (e.g., ethnicity, poverty status, and

temporal orientation) are related to patient beliefs and

perceptions about health and illness and how these belief

systems are manifested in actual drug use practices among

hypertensive individuals.














CHAPTER 3
RATIONALE AND THEORETICAL FRAMEWORK


Numerous cognitive models of health behavior have been

proposed to explain individuals' health actions in response

to medical recommendations [See Cummings, Becker and Maile

(1980) and DiMatteo and DiNicola (1982) for reviews]. A

common element among all of these conceptual frameworks is

the recognition that individual health beliefs and

perceptions play principal roles in compliance behaviors.

Of these models, one of the most frequently cited and

researched is the Health Belief Model (HBM) (Rosenstock,

1966; Becker, 1974; Becker and Maiman, 1975; Janz and

Becker, 1984; Harrison, Mullen and Green, 1992). The

theoretical framework for this study is the HBM. The

rationale for selecting the HBM for this research is

twofold. First, the HBM has received empirical support in

compliance research (including compliance with medication

regimens) using both prospective and retrospective designs

(Janz and Becker, 1984; Harrison, Mullen, and Green, 1992).

Second, the HBM was chosen because it incorporates

individual subjective assessments of health situations and

their relationships with health behaviors and actions. The










health behavior to be examined in this research effort is

drug use behavior to combat hypertension.

Drug use behavior represents an individual's treatment

responses to a diagnosis of hypertension or high blood

pressure. This research effort recognizes that a treatment

response could involve the use of prescribed

antihypertensive medications, as well as the use of

alternative or home remedies. The importance of

representing drug use behavior in this fashion is that it

models the use of two different systems, formal and

informal. The use of these systems are supported by Scott's

(1981) research among ethnic minority groups where she

reported that many participants in her study made concurrent

use of both the western medical and traditional or

alternative systems. Another treatment behavior to a

diagnosis of hypertension could be when a person chooses not

to use any medication at all, even under circumstances in

which a physician prescribed an antihypertensive medication.

For individuals who use prescribed antihypertensive

medication, the pertinent health behavior in this study is

the extent to which a person takes the medication as

medically prescribed, that is, their degree of "compliance."

In the case of home remedies, whether or not people use home

remedies is the health action of interest. Individuals

exhibiting these health behaviors may be classified as: (1)

compliers or noncompliers with prescribed antihypertensive










medication (Rx); or (2) users or nonusers of home remedies

(HR). These two types of treatment behaviors represent how

the dependent variable in this study--drug use behavior--is

operationalized.

As discussed in Chapter 2, peoples' perceptions and

judgments of health and medical care situations influence

their choices among alternative actions. Beliefs about

illness conditions often influence drug use practices

(Heurtin-Roberts and Reisin, 1990; Snow, 1981).

Understanding how different illness and treatment beliefs

relate to different aspects of drug use behavior may be

important in devising strategies to improve patient outcomes

in culturally diverse populations. The HBM provides a

conceptual framework in which relationships between health

and treatment perceptions and drug use behavior can be

studied in culturally diverse populations.

The Health Belief Model, shown in Figure 1, consists of

six dimensions: perceived susceptibility, perceived

severity, perceived benefits, perceived barriers, cues to

action and modifying factors. The most notable dimensions

of the HBM, in terms of theoretical propositions and

empirical investigations, are the four perception variables

(Janz and Becker, 1984; Harrison, Mullen, and Green, 1992).

Relationships among these perception variables are

hypothesized to represent a decision process that

individuals may go through when choosing among alternative

























(I,





001
L f r i|









-' I -'--; ;-L----"- a
ix




/-l, *.: 1 :1 i
R2 W .r "
~c ti



>-1 I( l m
> 0 i > < .; */;m I
o^ ^ *^" ^a
M: U ss"t 1O
Q < Is '; T -










health actions. The model proposes that individual beliefs

about severity and susceptibility of a disease and its

consequences are associated with engaging in treatment

action. Subjective assessments of disease threat are

assumed to provide individuals with the motivating force to

take action. Once an individual feels substantially

threatened by a disease and its sequelae, (s)he must decide

among alternative actions. According to the HBM, it is at

this point that individuals perform a type of cost/benefit

analysis such that alternatives are subjectively evaluated

in terms of their benefits and costs (or barriers). This

cost/benefit analysis then results in a preferred course of

action.

This research, based on cognitive theory, recognizes

that when an individual is faced with a choice among

alternative behaviors, (s)he will likely choose the

alternative that is considered most beneficial relative to

the other alternatives in the choice set (Jaccard, 1981;

Lynch, 1984). Each treatment alternative is evaluated in

terms of the benefits and costs associated with all of the

available options. The particular sets of perceived

benefits and costs pertaining to each behavioral alternative

are believed to jointly influence an individual's preferred

course of action, i.e, a person's choice among alternative

behaviors. Figure 2 models the hypothesized relationships




















b 4




















0
I "
/








45

between the four HBM perception variables and the two levels

of drug use behavior.

HI: The more susceptible individuals perceive

themselves to be to consequences of hypertension

and perceive that hyperternson is a serious

condition, the greater the_r likelihood of being

compliant with prescribed -atihypertensive

medication and the greater their likelihood of

being users of hone remedies.

H2a: Perceived benefits of prescribed antihypertensive

medication will increase t-. probability of being

compliant with prescribed antihypertensive

medication and perceived coets of antihypertensive

medication will decrease t-h probability of being

compliant with prescribed a-tihypertensive

medication.

H2b: Perceived benefits of home remedies will increase

the probability of using h:-e remedies and

perceived costs of home re-edies will decrease the

probability of using home remedies.

H3a: Perceived benefits of prescribed antihypertensive

medication will decrease t-h probability of using

home remedies and perceived costs of prescribed

antihypertensive medication will increase the

probability of using home remedies.








46

H3b: Perceived benefits of home remedies will decrease

the probability of being compliant with prescribed

antihypertensive medication and perceived costs of

home remedies .-11 increase the probability of

being complianr with prescribed antihypertensive

medication.

The two remaining components of the HBM, cues to action

and modifying factors, are not conceptualized as core

elements of the decision process itself, but as catalysts

and shapers of the decision process. Accordingly, the HBM

proposes that some type cf internal or external cues to

action are needed to activate this decision process.

Modifying factors such as demographic and sociopsychological

variables are proposed tc relate to health actions primarily

through their influences on individual perceptions (Becker,

1974). Cues to action ani modifying factors have seldom

been included or explicitly examined in HBM studies (Janz

and Becker, 1984). As such, we lack a clear understanding

of how cues to action and modifying factors relate to health

beliefs and health behaviors. The role of cues to action

will not be assessed in -his research. However, the

research hypotheses in the remainder of this chapter focus

on the associations between selected modifying factors and

health perceptions and drug use behavior within the HBM

framework.










The model, in Figure 2, assesses the relationships

among the ethnicity of two selected groups, their poverty

status, temporal orientation, health-illness and treatment

perceptions, a-d drug use behavior. This theoretical model

proposes that ethnic group membership, poverty status, and

temporal orientation are related to health-illness and

treatment perceptions; and that these health-illness and

treatment perceptions are related to drug use behavior. A

description of these theoretical constructs and their

postulated relationships are discussed below.


Ethnicity and Socioeconomic Status


The race/class debate has been one of the most abiding

controversies to be found in the sociological literature

(Rex and Mason, 1986). Because ethnicity and socioeconomic

status are so closely related in this society, it has been

extremely difficult for researchers to separate the two

(Harwood, 1981). Although the intention is not to settle

this debate here, this research studies the unique influence

of ethnic group membership after accounting for

socioeconomic status. A study of this contribution is

supported by research that has shown that ethnic

differences, with attendant cultural differences, persist in

divergent health and illness beliefs and ensuing behaviors

irrespective of social class and education (Landrine and

Klonoff, 1992; Harwood, 1981; Berkanovic and Reeder, 1973;










Suchman, 1964). For example, Suchman (1964) and Greenblum

(1974) have found that ethnic differences in health

orientations such as knowledge about disease and skepticism

about medical care persist across class boundaries. It is

not suggested that all members of an ethnic group are

homogenous with respect to health beliefs and behaviors.

There is considerable variance in the extent to which

members of an ethnic group uphold the beliefs and behaviors

that are characteristic of their particular ethnic group.

However, it is not clear how health beliefs and drug use

behaviors are related to only ethnic and cultural factors,

only socioeconomic factors, or some combination of the

three. Nevertheless, after reviewing multidisciplinary,

health-related literature among various ethnic groups,

Harwood (1981) and Landrine and Klonoff (1992) suggested

that although ethnic differences diminish among members of

upper socioeconomic classes, they do not disappear

altogether. Based upon their research, the following

hypotheses were formulated:

H4a: Controlling for poverty status, the severity of

hypertension will be rated higher among African

Americans than among White Americans.

H4b: Controlling for poverty status, African Americans

will see themselves as less vulnerable to

hypertension consequences as compared to

susceptibility beliefs among White Americans.








49

H4c: Controlling for poverty status, African Americans

will rate the benefits of home remedies and the

costs of prescription medication higher than that

of White Americans.

H4d: Controlling for poverty status, African Americans

will rate the costs of home remedies and the

benefits of prescription medication lower than

that of White Americans.


Temporal Orientation


Temporal orientation is conceptualized as a linear,

multidimensional construct with distinct past, present, and

future components. Temporal orientation, as a part of

culture, is both learned and context-specific and thus

individuals are capable of movement between these three

dimensions given the particular situation. The model,

studied in this research, seeks to examine the relationship

between temporal orientation and health and treatment

perceptions among hypertensive individuals.

It would be valuable to understand the relationship

between temporal orientation and health perceptions

especially among people with hypertension. Because so much

of the value of treating hypertension is focused on some

future health state, it seems likely that individuals

holding different time perspectives concerning hypertension

may very well exhibit notably different perceptions










regarding hypertension and its management. Existing

evidence pertaining to general (not health-specific)

temporal orientation (Leshan, 1952; Bergadaa', 1990; Bailey,

1987; Harwood, 1981) suggests that future oriented

individuals place much more stock in abstract future events

than individuals who are oriented in the present.

Therefore, it is likely that future oriented people would

identify with the benefits of treatment versus costs over

the long run. On the other hand, the costs of treatment may

become magnified among present oriented individuals because

they would likely fail to fully appreciate the benefits

associated with some abstract future. Moreover, because

present oriented people tend to rely on pragmatic life

experiences rather than events in an obscure future, it is

also likely that they may not have thought about their

vulnerability to repercussions of uncontrolled blood

pressure. Based upon these conceptions, the following

hypotheses were formulated:

H5: Compared to people with more future orientations,

individuals who are more present oriented on each

domain of hypertension will see themselves as less

susceptible to hypertension consequences.

H6a: More future oriented individuals on each domain of

hypertension will give greater weight to the

temporally distant benefits of either prescribed

antihypertensive medication or home remedies as









51
compared to the presently salient costs of either

treatment modality.

H6b: More present oriented individuals on each domain

of hypertension will give greater weight to the

presently salient costs of either prescribed

antihypertensive medication or home remedies as

compared to the temporally distant benefits of

either treatment modality.

Hypertension, as a part of one's personal subculture,

is believed to be largely influenced by one's ethnic

background and experiences. Therefore, temporal orientation

related to hypertension may differ among African Americans

and White Americans. Moreover, because temporal orientation

is learned through experiences, African Americans and White

Americans may not be temporally different when it comes to

those things they have not yet experienced. While it is

unknown whether temporal orientation differs among African

Americans and White Americans, there is evidence that

nonspecific temporal orientation differs among levels of

socioeconomic status (Leshan, 1952). This study found that

lower classes tend to orient themselves in the present. The

extent to which these socioeconomic differentials are

related to ethnic differences is unknown. However, in the

United States, the popular concept of "C.P. Time" is a

consistent indication of the cultural difference of the

Black-White concept of time (Houston, 1990). "C.P. Time" is










an acronym for "colored peoples' time" and it is

characterized by an expectation that African Americans will

be "late" in relation to some prescheduled time.

Several authors have supported the belief that there

are commonalities (e.g., time perception) that unite African

Americans which have origins in African culture [See

Anderson (1989) for a review]. For example, Houston (1990,

p.21) argued that the difference in the conceptualization of

time by African Americans can best be understood "within the

religious ontology of African life and viewed as a vestige

of African culture that has survived the obliterating

effects of Western influence." He based this assertion on

research conducted by Mbiti (1969) who interpreted that

African time consists of phenomena that have already

occurred (past), those that are currently taking place

(present), and those that are to transpire immediately

(immediate future). In essence, African time is viewed as

bidimensional--past and present--with a slight degree of

future. Houston (1990, p. 21) argued that, for the African,

the distant future is "an almost incomprehensible concept,"

because the events symbolizing its existence have not

occurred. These African time perceptions seem salient among

many African Americans, although there is some individual

variation with regard to the degree of influence these

African cultural traditions have on the behavior of modern

African Americans (Anderson, 1989). Based upon the










preceding information, the following hypotheses were

constructed:

H7a: Controlling for poverty status, African American

subjects will be more present oriented and less

future oriented than White American subjects on

the two experiential domains of hypertension.

H7b: Controlling for poverty status, African Americans

and White Americans will not differ in temporal

orientation on the nonexperiential domain of

hypertension.

As described earlier in this chapter, the HBM proposes

that modifying factors influence health actions primarily

through their influences on health perceptions. As such,

demographic and sociopsychological factors are not believed

to have any direct influences on health action. The

following hypothesis was based on this HBM proposition:

H8: The addition of demographic and sociopsychological

variables (i.e., ethnicity, poverty status, and

temporal orientation) into the logistic regression

model will not significantly change the

probabilities of either level of drug use behavior

once the effects of the health perceptions have

already been taken into account.














CHAPTER 4
METHODOLOGY


This section delineates the methods and procedures that

were used in this study in order to meet the research

objectives. They include sample selection, data collection,

nonrespondent bias, study variables, instrument development

and validation, data analysis, and limitations.


Sample Selection


The pool of subjects for this research was individuals

with hypertension residing in the northern and central

northern counties in the state of Florida. The northern

section of Florida was particularly relevant to this

research since it is contained in the southeastern region

commonly referred to as the "Stroke Belt" region of the

country (Siegel, et al., 1992). Moreover, these northern

Florida counties have aggregated stroke mortality rates

higher than that of other states contained in the "Stroke

Belt" region. In this research, stroke mortality rates were

used as a marker for hypertension incidence in the study

counties. From this population, a random sample of subjects

were selected according to the following inclusion criteria:










1) The individual must have been diagnosed by a

physician as having high blood pressure;

2) The individual patient must have had high blood

pressure for at least one year;

3) The individual must have been prescribed at least

one antihypertensive medication; and

4) The individual must not have experienced nor

currently be experiencing any complications (e.g.,

kidney disease, stroke, heart attack, blindness)

of uncontrolled hypertension.

These inclusion criteria were developed after extensive

discussions about key methodological and conceptual factors

relevant to this research. For example, the fourth

criterion was generated after an examination of the

perceived susceptibility to consequences of uncontrolled

hypertension construct, a primary variable in this research.

Among individuals who are experiencing or who have already

experienced these consequences, this construct would

probably be meaningless. The sample was stratified post hoc

based upon certain preselected categories:

1) Poverty status as indicated by age of head of

household, household size, and household income.

Respondents were classified as either above,

within or below the poverty threshold according to

guidelines developed by the U.S. Department of










Commerce criteria for 1993 (Bureau of the Census,

1994).

2) Ethnic group membership as indicated by self-

identification. The sample included approximately

equal numbers of two selected groups, African

American and White American individuals with

hypertension.

3) Geographic locale as indicated by residence in

either metropolitan or nonmetropolitan counties as

defined by the US Department of Commerce (Bureau

of the Census, 1993).


Data Collection Procedures


The research was non-experimental employing a cross

sectional design. Data collection was conducted via

telephone interview methods by a local marketing research

firm. It is believed that this procedure did not bias the

sample significantly since approximately 98% of US

households have telephones (National Center for Health

Statistics, 1987). The telephone methodology was chosen for

data collection because of the nature of the dependent

variable being studied. Given the conceptualization of druz

use behavior in this study, it was important to include

those individuals who have been medically diagnosed as

having hypertension and prescribed an antihypertensive

medication, but who are perhaps not currently engaging in










follow-up physician visits or not taking prescribed

antihypertensive medications. It is for the aforementioned

reasons that the study sample was not solicited from

physicians' practices or phar-acy practices because these

methods would exclude the types of individuals described

above.

Data were collected over a period of apprc.imately

three weeks, from January 30 to February 16, 19i. A random

sample of five thousand subjects residing in the northern

and north central counties of Florida was purchased from a

sampling house. Individuals were contacted by phone by

trained interviewers between the evening hours of five and

nine on Monday through Friday. A comprehensive log of the

results of each telephone contact was kept (answering

machine, no answer, busy, refusal, etc.). Four attempts

were made to contact those individuals who were not

contacted initially. Each interviewer utilized the

computer-assisted telephone interview (CATI) system for data

entry. The CATI system enabled interviewers to directly

enter responses into the computer for immediate creation of

the data base.

All interviews began with the identification of the

interviewers including their names, where they were calling

from, and what the calls were about. The next step was to

ascertain whether or not an adult within the household had

hypertension and, if appropriate, ask to speak with that








58

individual. Potential respondents were then given a brief

introduction about the study and an assurance of

confidentiality. Participation in this study was voluntary

and verbal consent was obtained from each study parti:cpant.

Written informed consent was not required since an exe-ption

was granted by the Health Science Center Institutional

Review Board (IRB) at the University of Florida.

Once the individual had verbally consented to becoming

a potential respondent, the interviewer began the screening

process in order to verify that the participant met the

criteria for enrollment. This was done through the

application of the "screener" developed for this study (See

Appendix D). The screener included all of the

aforementioned study criteria as well as questions on ethnic

identity, county and city of residence. When respondents

met all of the screening criteria, the application of the

study instrument immediately followed. When respondents did

not meet at least one of the screening criteria, they were

thanked for their willingness to participate and the

contacts were immediately terminated. This process

continued until 300 individuals were obtained who both

qualified and agreed to participate.

Each interview took an average of 20 minutes to

complete. Each respondent was asked a series of questions

contained in the survey instrument developed to measure the

various constructs: 1) Ethnicity, 2) Poverty Status, 3)










Temporal Orientation, 4) Health and Treatment Perceptions,

and 5) Drug Use Behavior (See Appendix D for survey).

At least 300 respondents were needed in this study in

order to evaluate ths research hypotheses. This sample size

estimate was based cn achieving a statistical power of 0.80,

a Type I error rate equal to 0.05, and an effect size of 2%

for individual effects in a model explaining 23% of the

variance in total (Keppel, 1991). In other words, at an

alpha level of 0.05, there was an 80% chance of detecting

differences on the estimated smallest effect size of 2%.

Moreover, this sample size was estimated on the assumption

that the study model would explain at least 23% of the

variance in drug use behavior as suggested by the

literature.


Nonrespondent Bias


A total of 2442 contacts were made during the data

collection period. This total included completed

interviews, ineligible contacts, and all refusals.

Individuals who had hypertension and refused to participate

were logged as high blood pressure refusals. These

individuals comprised the nonrespondent group of interest in

this research. Of the 753 total hypertension contacts, 66

people were logged as refusals which resulted in a refusal

rate of 8% among hypertensives. This refusal rate was










calculated by dividing the number of high blood pressure

refusals by the total number of contacts with hypertensives.


Study Variables


Dependent Variables


Althc-gh the four perception elements of the Health

Belief Model (HBM) have been studied extensively, there are

no current standardized measurements of these variables.

Accordingly, the items for HBM model variables in this study

were developed through the use of pilot interviews along

with information in the literature.

Perceived severity of hypertension

Perceived severity is the degree to which an individual

believes hypertension to be a serious disease. Perceived

severity was measured in terms of the seriousness of

hypertension, the fear of having hypertension, and the

limitation that hypertension imposes on social activities.

Each of the three dimensions was measured on a 5-point

Likert-type scale anchored by 1 "Strongly Agree" and 5

"Strongly Disagree" with 1 representing the strongest belief

on each dimension.

Perceived susceptibility to consequences of hypertension

Perceived susceptibility is the degree to which an

individual perceives her/himself to be vulnerable to

consequences of uncontrolled hypertension. Perceived

susceptibility was measured by the extent to which










individuals believed it is probable that they would

experience a stroke, heart attack, or kidney problems and by

their estimate of how much at risk they were to having a

stroke, heart attack or kidney problems. Susceptibility

scores were summed over two items requiring a response to a

5-point Likert-type scale anchored by 1 "Strongly Agree" and

S"Strongly Disagree." A score range of 2 to 10 was

possible with 2 representing the highest belief on perceived

susceptibility.

Perceived benefits of antihypertensive medication (Rx)

Perceived benefits of antihypertensive medication (Rx)

represented the degree to which respondents believed their

medication to be effective in controlling high blood

pressure and preventing adverse consequences. Perceived

benefits of Rx were measured in terms of the medication's

ability to control high blood pressure, to prevent strokes,

heart attacks, and kidney disease, and to ease one's mind

about having high blood pressure. Benefits of Rx scores

were summed over three items, each on a 5-point Likert-type

scale anchored by 1 "Strongly Agree" and 5 "Strongly

Disagree." A score range of 3 to 15 was possible with 3

representing the highest belief on perceived benefits of Rx.

Perceived costs of antihypertensive medication (Rx)

Perceived costs of antihypertensive medication (Rx)

represented the degree to which an individual believed there

are barriers (or costs) associated with the use of










antihypertensive medication. Perceived costs of Rx were

measured in terms of paying for the medication (financial),

forgetting to take the medication, problem obtaining

refills, and experiencing side effects from the medication.

Each of the four dimensions was measured on a 5-point

Likert-type scale anchored by 1 "Strongly Agree" and 5

"Strongly Disagree" with 1 representing the strongest belief

on each dimension.

Perceived benefits of home remedies (HR)

Perceived benefits of home remedies (HR) represented

the degree to which an individual believed home remedies

were effective in controlling high blood pressure and

preventing adverse consequences. Perceived benefits of HR

were measured by beliefs about home remedies' effectiveness

in controlling high blood pressure and by their ability to

keep the body and blood balanced. Benefits of HR scores

were. summed over two items requiring a response to a 5-point

Likert-type scale anchored by 1 "Strongly Agree" and 5

"Strongly Disagree. A score range of 2 to 10 was possible

with 2 representing the highest belief on perceived benefits

of HR.

Perceived-costs of home remedies (HR)

Perceived costs of home remedies (HR) represented the

degree to which an individual believed there are barriers

(or costs) associated with the use of home remedies.

Perceived costs of HR were measured in terms of the lack of









63
physicians' acceptance of using HR, the lack of comfort with

discussing HR use with their physicians, and the lack of

efficacy of HRs compared to prescription medication. Each

of the three dimensions of costs of HR was measured on a 5-

point Likert-type scale anchored by 1 "Strongly Agree" and 5

"Strongly Disagree" with 1 representing the strongest belief

on each dimension.

Drug use behavior

Drug use behavior is an individual's treatment response

to a diagnosis of hypertension or high blood pressure. A

treatment response could involve the use of prescribed

antihypertensive medications, as well as the use of home

remedies. Individuals exhibiting these health behaviors may

be classified as: (1) compliant or noncompliant with

prescribed antihypertensive therapy; or (2) users and

nonusers of home remedies.

Compliant respondents were those individuals who

reportedly took their medications every day as medically

prescribed during the 30-day reference period. Noncompliant

respondents were those individuals who reported one day or

more of not taking their medication as prescribed within the

reference period of 30 days. Compliant and noncompliant

categories were measured by the question, "In the last 30

days, how many days have you taken your high blood pressure

medication exactly as your doctor prescribed?" Those who

answered 30 were classified as compliant and those who










answered 29 or less were classified as noncompliant. The

compliance measure was originally designed to be a

continuous measure, however, not unlike many self-reported

compliance measures in the literature, the compliance scores

in this study -ere highly skewed, indicating a

dichotomizaticn of the variable. The cutpoint chosen for

the dichotomy is similar to that used in other compliance

studies (Hershey et al., 1980).

Respondents were also categorized on the basis of

whether or not they used home remedies to treat their

hypertension. Some examples of home remedies include

vinegar, garlic, aloe vera, and vitamins. The use of home

remedies was measured by the question: "Besides prescription

medication, do you use anything else to treat your high

blood pressure?" Those who answered yes were classified as

users of home remedies and those who answered no were

classified as nonusers of home remedies. Users of home

remedies were also asked to name the home remedy they used.

This second question was asked to verify that a home remedy

was actually being used.


Independent Variables

Ethnicity

Ethnicity is a socially defined state of belonging to a

particular subgroup and seen as belonging to that subgroup

by others. Ethnicity was measured by self-report on a

single item scale requiring one of the following responses:










1) Non-Hispanic Black or African American, 2) Non-Hispanic

White or Caucasian, 3) Hispanic, 4) Asian, or 5) Other.

Only respondents in categories 1 and 2 were eligible for the

study.

Poverty status

Poverty status is a measure of the financial well-being

of an individual in U.S. society. Poverty status was

measured by an algorithmic combination of age of household

head, household size, and household income. The composite

of these three items were then used to classify individuals

as above, within or below the poverty threshold level as

defined by the U.S. Department of Commerce criteria for 1993

(Bureau of the Census, 1994). Estimated poverty thresholds

are single numerical estimates of income rounded to the

nearest dollar. Because it was believed that respondents

would not be able to give such refined estimates of their

incomes, ranges of $1000 were created such that estimated

poverty thresholds were captured within these ranges. For

example, if the estimated poverty threshold was $7517, the

created range was $7000 to $8000. An individual, for which

this range was applicable, would be classified as above the

poverty threshold if her income was greater than $8000,

within the threshold if her income was between $7000 and

$8000, or below the threshold if her income was less than

$7000. Poverty status was chosen to measure financial well-








66

being because this designation not only reflects income, but

also age of household head and size of household.

Temporal orientation

Temporal orientation is defined as the time perspective

an individual holds with regard to hypertension and its

management. Temporal orientation is conceptualized as

having three distinct dimensions: past, present, and future.

However, previous research suggests that most Americans

primarily hold either present or future orientations

(Gonzalez and Zimbardo, 1985). Hence, only present and

future dimensions were measured in this research effort.

An eleven item hypertension temporal orientation scale

was developed based upon data in the literature and data

from the pilot interviews described below. Construct

validation procedures of the temporal orientation scale

revealed three primary domains of hypertension: a

nonexperiential domain, an experiential disease domain, and

an experiential treatment domain (See Results chapter). The

nonexperiential domain, measured by five items, represented

issues that are potential consequences of both hypertension

and hypertension management. The experiential disease

domain, measured by three items, characterized concerns that

are related to the day-to-day dealing with hypertension.

The experiential treatment domain, measured by three items,

represented day-to-day concerns involving hypertension

management (the medication). Each item required a response










to a 5-point Likert-type scale ranging from 1 "Strongly

Agree" to 5 "Strongly Disagree." Items within each subscale

were summed and then averaged resulting in a possible score

range of 1 to 5. Item scores were recorded such that

movement towards the low end (1) of each subscale

represented a more future orientation and movement towards

the high end (5) represented a more present orientation.


Instrument Development and Validation


Pilot Interviews


The initial phase of this project involved the

development and validation of the survey instrument to

measure: 1) health-specific temporal orientation of patients

in regard to hypertension and its management, 2) health and

treatment perceptions of patients in regard to hypertension

and antihypertensive therapy, and 3) drug use behavior of

hypertensive patients with regard to the usage of both

prescribed antihypertensive medications and home remedies

(see Appendix C).

The questionnaire development process began by

conducting in-depth interviews with a convenience sample of

seven hypertensive patients. Six respondents resided

locally and one respondent lived in another state. Subjects

were solicited such that the sample included both African

American and White American hypertensives. Participants

included three black females, two white females, and two








68

white males with an age range of forty to sixty years. All

respondents had diagnosed high blood pressure for at least

ten years with the exception of one white male who had been

diagnosed with high blood pressure a little over one year

prior to the interview. Each respondent received a monetary

incentive for participation. Administrators at local

clinics and private practices and churches were contacted

and asked to help identify and recruit patients who met the

relevant criteria. In addition, they were asked to either

distribute or prominently display a statement describing the

project so that eligible patients were informed. Patients

who were willing to participate were asked to contact the

researcher directly so that appointments could be set.

In-depth interviews were conducted with each respondent

in order to develop items to be included in the three

scales: 1) Health Temporal Orientation, 2) Health and

Treatment Perceptions, and 3) Drug Use Behavior. Most

interviews took 30 to 45 minutes to complete with the

exception of one interview which lasted 75 minutes. The

respondents were asked open-ended questions about their

health beliefs (including temporal orientation) and their

decisions about drug use. Probes were used to tap into

other relevant dimensions of each construct and to clarify

and elaborate when needed. All interviews were tape

recorded. The informants' responses were then content

analyzed and coded such that the identified domains could be










captured in scale items. Scale items were also generated

from data in the literature.


Pretest of Instrument


In an effort to make survey questions relevant and

understandable to hypertensive patients, an expert panel of

researchers and practitioners in the area of patient

compliance and patient care, respectively, were asked to

assess the instrument for content and face validity. Based

upon the expert panel's recommendations, questionnaire items

were then modified as necessary such that the "language" of

the questionnaire, the completeness of items, and the

response categories were deemed adequate. The modified

instrument was then reassessed by panel members and was

considered appropriate for pretest evaluation.

The study instrument was then pre-tested on a different

group of 20 hypertensive patients. These respondents also

received monetary compensation in exchange for their

participation. Ten interviews in the pretest were conducted

face-to-face and the remaining ten interviews were conducted

over the telephone. The instrument was pretested over the

telephone for the last ten interviews because this was going

to be the method of data collection for the main study.

After each of the first ten interviews, short debriefing

sessions were conducted with subjects to assess

questionnaire clarity and completeness.










Consent forms were distributed to local pretest

participants for written permission to have their pharmacy

medication profiles examined. Pharmacy refill records of

the 2 months prior to the interview served as an indicator

of the validity of patient self-reported prescribed

medication practices.

A total of 45 items was selected to represent the

independent measures in this study. Items were grouped into

scales according to the constructs they were intended to

measure. Statistical procedures-were employed to assess the

reliability of the instrument. The internal consistency of

each scale was calculated to obtain reliability estimates.

Reliability of the scales were assessed through the use of

Cronbach's coefficient alpha statistic. The coefficient

alphas obtained provided an estimate of how consistent

subjects performed across items measuring the same

construct. A high value of coefficient alpha indicates a

consistent performance of respondents across items and that

the performance is generalizable to other potential items

pertaining to the same content domain (Crocker and Algina,

1986). Although what is considered "low" for alpha depends

on the purpose of the research (Churchill, 1979), a

reliability estimate of 0.50 or 0.60 is considered

acceptable (Nunnally, 1978).

An item analysis was also obtained in order to

determine the final set of items for the survey instrument.








71

The decision to retain or delete an item from its respective

subscale was base: upon the following information: (a) an

interitem correlation matrix, (b) a corrected item to total

correlation coefficient, and (c) the coefficient alpha

estimate if the i-em was deleted from the scale (Ferketich,

1991).

The interiter correlation matrix provides information

about how a certain item relates to other items in the

scale. Generally, interitem correlations of 0.30 or higher

are desirable (Ferketich, 1991).

Corrected item to total correlations involve the

correlation of item score with the total score of the

remaining items ir. the scale under examination. Corrected

item to total correlations were calculated to adjust for

spurious values that are obtained when the item scores

contribute to the total scores (Crocker and Algina, 1986).

Corrected item to total correlations are especially relevant

when there is a srall number of items in the scale

(Ferketich, 1991). A more recent rule of thumb for

corrected correlations is that they should be 0.50 or

greater (Bearden et al., 1989), however, corrected

correlations above 0.30 have been considered sufficient

(Nunnally, 1978).

The revised coefficient alpha reveals the change in

alpha if the item was dropped from the scale. If there is a

substantial improvement in alpha when the item is deleted,










this is considered as some support for dropping the item.

However, the revised alpha is most informative when it is

used in combination with the aforementioned aspects of item

analysis.


Pretest Results


Of the twenty pretest participants, sixty percent

(N=12) were African American and the remaining forty percent

(N=8) were White American. Most (N=15) of the respondents

were female. The mean age of respondents was 61.8 years and

they had been diagnosed with hypertension for an average of

16.8 years. Fifty percent (N=10) of the respondents had

less than a high school education, while the remaining half

were at least high school graduates. Moreover, one half

(N=10) of the respondents had annual household incomes of

less than $10,000 and the other half had incomes of $10,000

or greater.

Psychometric properties of pretest scales

A primary criterion for assessing the quality of each

scale was coefficient alpha. If scale items intended to

measure a construct indeed came from the same domain,

responses to these items should be highly intercorrelated,

i.e., internally consistent. The original and revised

reliabilities of study scales are reported in Table 4-1.

The revised coefficient alpha estimates ranged from 0.32 to











0.87. Based upon acceptable levels of 0.50 or 0.60, only

the costs of Rx scale had low reliability.


Table 4-1
Original and Revised Reliabilities (a) of Pretest Scales


Scale Original Original Revised Revised
# of a = # of a =
items items

Perception Variables
Severity 4 0.58 3 0.61
Susceptibility 4 0.63 3 0.66
Benefits of Rx 4 0.64 3 0.79
Costs of Rx 5 0.47 4 0.32
Benefits of HR 4 0.79 3 0.87
Costs of HR 4 0.55 3 0.53

Modifying Factors
Ethnicity 1 NA 1 NA
Poverty Status 1 NA 1 NA
Temporal Orientation
Nonexperiential 7 0.38 5 0.64
Experiential 9 0.67 6 0.73

Drug Use Behavior
Compliance with Rx 1 NA 1 NA
Use of HR 1 NA 1 NA


Note: Rx = prescription medication, HR =
number, NA = not applicable.


home remedies, # =










Modifications to the scales were made based upon both

individual item characteristics and theoretical importance.

In addition, instrument length was considered so that

telephone interviews could be kept at approximately twenty

minutes.

Individual scale items were evaluated in terms of their

interitem correlations, item to total correlations, and

revised coefficient alphas when an item was deleted from the

scale. The decision to delete or retain an item was

determined on both psychometric and conceptual grounds.

Caution was taken not to eliminate items only on its

psychometric characteristics because of the relatively small

pretest sample size. The evaluation process of each scale

is described below.

Severity scale. The reliability of this four item

scale was 0.58 (See Table 4-1). Item Q3, "High blood

pressure cannot kill you," was deleted from the scale based

upon poor interitem and corrected item to total correlations

(See Appendix A). The reliability of the scale increased to

0.61 after the deletion of the item Q3.

Susceptibility scale. This four item scale exhibited a

reliability of 0.63 (See Table 4-1). Item Q9, "If I do not

take care of myself, there is a good chance that I will

suffer from serious health problems caused by my high blood

pressure," had low interitem and corrected item to total











correlations (See Appendix A). This item was deleted from

the scale and the revised alpha for the scale was 0.66.

Benefits of Rx scale. The reliability of t:is scale

was 0.64 (See Table 4-1). The second item (Q54) :f this

scale, "My high blood pressure would not get worse if

stopped taking my blood pressure medication," was very

confusing to respondents. It had been negatively worded in

order to decrease any response set bias. However, there was

no variance on this item among pretest subjects and,

therefore, the correlations of item Q54 were zer: (See

Appendix A). Item Q54 was deleted from the scale and the

revised alpha after the deletion was 0.79 (See Table 4-1).

Costs of Rx scale. The reliability of this scale was

0.47 (See Table 4-1). Responses to item Q60, "S~de effects

prevent me from taking my blood pressure medication as

prescribed," did not vary and, again, correlations were zero

(See Appendix A). However, item Q60 was retained for the

main study because it was believed that this ite- would

perform better in a larger and more heterogenous sample.

Although item Q61, "I do not have trouble keeping up with

how often I am suppose to take my blood pressure

medication," performed well, respondents had trouble

remembering the entire statement during pretest telephone

interviews and thus it was deleted from the scale. The

revised alpha for this scale was 0.32. Although the

reliability was low in terms of an acceptable alpha level of










0.50 or 0.60, these items were chosen for the main study

because they have been shown in the literature to be

important barriers to using prescription medication properly

(See Chapter 2).

Benefits of HR scale. The reliability of this scale

was 0.79 (See Table 4-1). The only poorly performing item

was item Q64, "Folk remedies provide me with immediate

relief since I can feel them working in my body," which had

low interitem and low corrected iter to total correlations

(See Appendix A). This item was deleted and the revised

alpha for this scale was 0.87 (See Table 4-1).

Costs of HR scale. This scale exhibited a reliability

of 0.55 (See Table 4-1). Item Q66, "Folk remedies do not

really work over the long run," was deleted based upon poor

interitem and poor corrected item tc total correlations (See

Appendix A). Although items Q67, "I believe folk remedies

are not well accepted by my doctor," and Q68, "I feel that I

can discuss using folk remedies with ny doctor or other

health care professionals," had poor correlations, they were

retained because of their theoretical and practical

importance. However, item Q67 was reworded because

respondents found it difficult to answer in its original

format. The revised alpha for the scale was 0.53 (See Table

4-1).

Nonexperiential domain scale. This scale had seven

items originally. The reliability of this scale was 0.38










(See Table 4-1). Iters Q15, Q23 and Q27 were treatment

specific and items Q1I, Q19, Q21 and Q25 were specific to

hypertension. Items Q21, "I think about potential problems

that could occur later with my high blood pressure," and

Q25, "I think about h.. my high blood pressure might affect

me in the future," were deleted because of poor interitem

correlations (See Appendix A). The correlations on item

Q27, "I see taking blc:d pressure medication as an

investment in my future health," were zero since there was

no variance in responses on this item. However, upon

further evaluation, item Q27 was retained since it was

believed that it would perform better in the larger study

and more heterogenous population in terms of hypertension

temporal orientation. Therefore, five items (3 treatment

specific and 2 hypertension specific) remained for the final

study. The revised coefficient alpha of this scale was 0.64

(See Table 4-1).

Experiential domain scale. This scale initially had

nine items: items Q14, Q22, Q24, Q26 and Q29 were treatment

specific and items Q16, Q18, Q20 and Q28 were hypertension

specific. The reliability of this scale was 0.67 (See Table

4-1). Items Q16, "I see no sense in thinking about possible

problems with my high blood pressure right now because they

may not happen," and Q24, "I take my blood pressure

medication to keep me healthy now rather than thinking about

the future," were deleted because of poor interitem











correlations (See Appendix A). The correlations for items

Q22, "I think it is a waste to take blood pressure

medication when I don't feel like my blood pressure is

high," and Q29, "If I start having problems with my blood

pressure medication, I will just stop taking it," were zero

because there was no variance in responses to these items

(See Appendix A). After further evaluation of both items in

terms of theoretical relevance, item Q22 was retained and

item Q29 was deleted. Moreover, it was believed that item

Q22 would perform better in a larger and more heterogenous

sample with respect to different hypertension temporal

orientations. Thus, six items (3 treatment specific and 3

hypertension specific) were retained for the main study.

The revised reliability of this scale was 0.73 (See Table 4-

1).

Overall, a total of 11 items was deleted from the

instrument based upon poor interitem correlations, poor

corrected item to total correlations, negative contributions

to the internal consistency of its respective scale, and

poor clarity among pretest hypertensive respondents.


Validity cf Self-Reported Medication Use


Overall, the pretest sample reported high compliance

with their medication regimens. In recognition of the

debate about how patients tend to over-report their degree

of compliance with medication regimens, an attempt was made










to validate this measure. After interviews were completed,

eleven of the local pretest participants consented to have

their pharmacy refill records examined, however, only 9

participant records were reviewed. In the other two cases,

pharmacists wanted verbal consent from the patients

themselves, but the patients were not available.

Nevertheless, the 9 reviewed records included both self-

reported compliant and noncompliant individuals.

Patient self-reported compliance in the 30-day

reference period prior to the interview was correlated with

refill records from the previous two months. From the

refill records, a compliance score was calculated by

dividing the number of units dispensed by the number of days

between refills. The self-reported mean compliance score

was 98.8 3.3 and the mean compliance score from refill

records was 98.3 11.7. The Pearson correlation

coefficient was 0.92 which indicated that patient self-

reports were accurate and valid. Considering the average

age (61.8 years) and length of having hypertension (16.8

years) for this sample, it was plausible that the patients'

medication taking practices had become part of their daily

routine. In fact, most participants indicated that they

habitually took their medications after brushing their teeth

or after eating breakfast. Furthermore, the validity of

these results has some support in the literature such that










compliance has been found to be positively related to age

and length of having the disease (Klein, 1988).


Data Analysis


The relationships between the independent variables and

dependent variables in this study were examined through the

use of logistic regression, analysis of variance (ANOVA),

and multiple regression procedures. Using multivariate

logistic regression procedures, the overall adequacy of the

proposed theoretical model was evaluated in. terms of the

goodness-of-fit of the proposed model in predicting both

compliance with Rx and use of HR. Interactive logistic

regression modeling was used to test research hypotheses and

to determine the most parsimonious model in predicting

compliance with Rx and use of HR. The description of tests

of proposed research hypotheses are outlined below.


Tests of Research Hypotheses


Health perceptions and drug use behavior

HI: The more susceptible individuals perceive themselves to

be to consequences of hypertension and perceive that

hypertension is a severe condition, the greater their

likelihood of being compliant with prescribed

antihypertensive medication and the greater their

likelihood of being users of home remedies.










H2a: Perceived benefits of prescribed antihypertensive

medication will increase the probability of being

compliant with prescribed antihypertensive medication

and perceived costs of antihypertensive medication will

decrease the probability of being compliant with

prescribed antihypertensive medication.

H2b: Perceived benefits of home remedies will increase the

probability of using home remedies and perceived costs

of home remedies will decrease the probability of using

home remedies.

H3a: Perceived benefits of prescribed antihypertensive

medication will decrease the probability of using home

remedies and perceived costs of prescribed

antihypertensive medication will increase the

probability of using home remedies.

H3b: Perceived benefits of home remedies will decrease the

probability of being compliant with prescribed

antihypertensive medication and perceived costs of home

remedies will increase the probability of being

compliant with prescribed antihypertensive medication.

This set of hypotheses (H1 to H3b) delineates the

hypothesized relationships between the health perception

variables and drug use behavior, i.e., compliance with Rx

and use of HR. Multivariate logistic regression procedures

were used to test each of these probabilistic hypotheses.

In these analyses, compliance with Rx and use of HR were










treated as separate criterion variables in two multivariate

models. The severity, susceptibility, benefits and costs

dimensions served as predictor variables in each of the

models. Multiple logistic regression models were used to

calculate adjusted regression coefficients, and odds ratios

(OR) and 95% confidence intervals (CI) were calculated from

these coefficients. Statistical significance was determined

at p < 0.05.

Ethnicity and health perceptions

H4a: Controlling for poverty status, the severity of

hypertension will be rated higher among African

Americans than among White Americans.

H4b: Controlling for poverty status, African Americans will

see themselves as less vulnerable to hypertension

consequences as compared to susceptibility beliefs

among White Americans.

H4c: Controlling for poverty status, African Americans will

rate the benefits of home remedies and the costs of

prescription medication higher than that of White

Americans.

H4d: Controlling for poverty status, African Americans will

rate the costs of home remedies and the benefits of

prescription medication lower than that of White

Americans.

This set of hypotheses (H4a to H4d) involve ethnic

group differences in health perceptions while controlling










for poverty status. Multivariate analysis of variance

(MANOVA) was employed to allow for simultaneous testing of

differences in health perceptions (severity, costs of Rx and

costs of HR) that comprised of multiple single item

measures. Analysis of variance (ANOVA) procedures were used

to compare mean perception scores of each ethnic group,

controlling for poverty status. In each test, the health

perceptions were treated as dependent variables in separate

ANOVA models. The t-statistic and its associated p-value

were assessed for statistical significance at a level of p <

0.05. Beta coefficients were analyzed to determine the

direction of the relationship. Least square means for the

ethnic groups were obtained to ascertain their relative

positions on the perception variables.

Hypertension temporal orientation and health perceptions

H5: Compared to people with more future orientations,

individuals who are more present oriented on each

domain of hypertension will see themselves as less

susceptible to hypertension consequences.

H6a: More future oriented individuals on each domain of

hypertension will give greater weight to the temporally

distant benefits of either prescribed antihypertensive

medication or home remedies as compared to the

presently salient costs of either treatment modality.

H6b: More present oriented individuals on each domain of

hypertension will give greater weight to the presently










salient costs of either prescribed antihypertensive

medication or home remedies as compared to the

temporally distant benefits of either treatment

modality.

This set of hypotheses (H5 to H6b) delineates the

hypothesized relationships between the hypertension temporal

orientation and health perception variables. Again, the

health perception variables served as criterion variables in

separate regression models. Perceived susceptibility,

perceived benefits, and perceived costs were regressed on

each domain of hypertension temporal orientation. An

analysis of standardized regression coefficients was

conducted to determine the significance at p < 0.05 and the

direction of the proposed relationships between hypertension

temporal orientation and the health perception variables.

For temporal orientation group comparisons, both MANOVA

and ANOVA procedures were employed. A number of planned

contrasts were conducted between temporal orientation groups

in evaluating health perceptions. The first contrast of the

FPP (i.e., subjects exhibiting future orientations on the

nonexperiential domain, present orientations on the

experiential disease domain, and present orientations on the

experiential treatment domain) and FFF groups compared

present versus future orientations on both experiential

domains of hypertension. The second contrast of the FFP and

FPF groups compared future/present versus present/future










orientations on the experiential disease and experiential

treatment domains, respectively. The next two contrasts

involved group comparisons when respondents' orientation

differed only on one or the other experiential domain, that

is, comparing the FPF group to the FFF group and the :FP

group to the FPP group. It turned out that there were no

significant differences in group means when respondents'

temporal orientation differed only on the experiential

disease domain. Therefore, respondents were pooled across

the experiential disease domain and only comparisons on the

experiential treatment domain are reported. This final

contrast compared present orientations (average of groups

FFP and FPP) to future orientations (average of groups FPF

and FFF) on the experiential treatment domain of

hypertension. No contrasts concerning the nonexperienrial

domain were conducted since all of the groups used in the

inferential analyses held a future orientation with respect

to the nonexperiential domain. Significant differences were

evaluated at p < 0.05. Least square means of temporal

orientation groups were also obtained to evaluate their

relative positions on the health perceptions.

Ethnicity and hypertension temporal orientation

H7a: Controlling for poverty status, African American

subjects will be more present oriented and less future

oriented than White American subjects on the two

experiential domains of hypertension.











H7b: Controlling for poverty status, African Americans and

White Americans will not differ in te-poral orientation

on the nonexperiential domain of hypertension.

These hypotheses involve ethnic group differences in

hypertension temporal orientation while cc."rolling for

poverty status. MANOVA was used to compare the two ethnic

groups on all three temporal orientation d:mains. Both

African Americans and White Americans were compared on their

mean temporal responses to each dcnain of -ypertension.

Separate ANOVAs were conducted to assess any significant

ethnic group differences (p < 0.05) of least square means on

hypertension temporal orientation.

Modifying factors and drug use behavior

H8: The addition of demographic and socicosychological

variables (i.e., ethnicity, poverty status, and

temporal orientation) into the logistic regression

model will not significantly change the probabilities

of either level of drug use behavior once the effects

of the health perceptions have already. been taken into

account.

To test whether including the modifying factors added

significantly to the explanatory capability of the proposed

model, two interactive stepwise logistic regressions were

run on each level of drug use behavior by forcing the health

perceptions in the model and allowing any of the modifying

factors to enter. This was an interactive process because










both individual variables and theoretical subsets of

variables were tested for entry into the model. Subsets of

variables were tested for entry because sometimes, even when

individual associations are weak, the variables as a group

could become important pre:;ctors of the outcome (Hosmer and

Lemeshow, 1989).

Variable or group entry was based upon the significance

of its scoretest. The sccretest associated with a single

variable or group of variables represents how well that

variable or group of variatles discriminate between the two

levels of the dependent variable (Hosmer and Lemeshow,

1989). In a sense, a scoretest can be viewed similarly to

the incremental R2 in linear regression. In logistic

regression, the significance of the scoretest is evaluated

in terms of its ability to reduce the deviance of a model in

which it is not a part. Acain, in a linear regression

framework, deviance is similar to the sum of squared errors

(SSE), i.e, unexplained variation in the dependent variable.

If the scoretest was statistically significant (p < 0.05),

the variable or group of variables were allowed to enter the

multivariate model.


Li-itations


This cross sectional study represented one point in

time and did not reflect possible changes in individual

perceptions, beliefs, and behaviors over time. In fact, it










is anticipated that if an individual experiences a major

complication of uncontrolled hypertension (e.g., stroke,

heart attack, kidney disease), this experience would have

marked influences on her/his entire belief system with

regard to hypertension, hypertension outcomes, and treatment

behaviors. Fur-ner research will include a longitudinal

study of these respondents to assess how their beliefs and

perceptions change with variations in their health statuses.

Another limitation of this study involved the survey

instrument. Study participants may have been constrained in

that all possible alternatives may not have been represented

in survey response categories. A similar limitation deals

with the respondents' model of illness or hypertension

causality. Although individual beliefs about causes of

hypertension have been associated with a variety of

treatment responses (Heurtin-Roberts and Reisin, 1991), it

is unknown if these models of illness are direct

determinants of treatment behaviors. Future research should

seek to assess a causal connection between illness models

and treatment behaviors.

Finally, the single item measures in the severity,

costs of Rx and costs of HR scales introduced difficulty in

establishing the reliability and validity of these measures.

However, confirmation of several of the research hypotheses

did provide some indication of the validity of the single

item measures in this study.















CHAPTER 5
RESULTS


Sample Description


A total of 300 h:pertensive individuals residing in the

northern section of Fl:rida participated in this study. The

characteristics of the sample as a group and by ethnicity

are listed in Table 5-1. Fifty-six percent and 69% of the

respondents were African American and female, respectively.

These numbers are consistent with the extant hypertension

prevalence whereby it is most widespread among African

American women (National Institutes of Health, 1993). The

mean age of respondents was 60 years and they had

hypertension an average of 14.6 years with African Americans

having hypertension slightly longer than White Americans.

Most participants were above the poverty level and had at

least a high school education. African Americans were

poorer and had relatively less education than the White

Americans in this sample. Eighty percent of the African

American respondents lived in metropolitan counties and 58%

of the White American respondents resided in nonmetropolitan

counties.











Table 5-1
Sample Description


Characteristic Overall Black White
(N=300) (N=167) (N=133)
Freq (%) Freq (%) Freq (%)

Ethnicity 167 (56) 133 (44)

Gender
Female 206 (69) 127 (76) 79 (59)
Male 94 (31) 40 (24) 54 (41)

Education
Less than HS 95 (32) 55 (33) 35 (26)
HS Graduate 88 (29) 46 (28) 42 (32)
Greater than HS 116 (39) 66 (39) 56 (42)

Poverty Status
Below 75 (25) 54 (32) 21 (16)
Within Threshold 38 (13) 31 (19) 7 (5)
Above 168 (56) 43 (43) 96 (72)
Unknown 19 (6) 10 (6) 9 (7)

Residence
Metropolitan 190 (63) 134 (80) 56 (42)
Nonmetropolitan 110 (37) 33 (20) 77 (58)

Age 60.3 13.1 60.4 13.4 60.2 12.7
1st Diagnosis 14.6 10.6 15.5 11.1 13.5 9.9


Reported as Mean SD

Note: Freq = frequency, SD = standard deviation











Evaluation of Nonrespondent Bias


Geographic residence was the only known characteristic

of those who refused to participate. First, a comparison

was made to determine if refusals were dependent on -hether

an individual was from a metropolitan or nonretropolitan

county. A Chi-square test of independence revealed no

significant differences between refusers and nonrefusers

(Chi-square = 0.10 p > 0.70). Second, study participants

and refusers were compared across metropolitan and

nonmetropolitan counties. Again, no significant differences

were found between the two groups (Chi-square = 0.42, p >

0.50). It is not known with certainty that refusers would

have responded like the participants, however, based on the

information available, there was no reason to suspect: there

were any particular differences between the groups.


Evaluation of Final Instrument


Thirty-four items were chosen to represent the

independent and dependent variables in this study. These

items were initially grouped on the basis of the pretest

results. For variables with multiple items, the reliability

of each scale was calculated using coefficient alpha.

Interitem correlations and corrected item to total

correlations were also obtained for scale items (See

Appendix B). Generally, items that exhibited poor interitem

and corrected item to total correlations and lacked positive










contributions to the internal consistency of their assigned

scale were subsequently dropped from the scale and new

coefficient alphas were calculated. This procedure resulted

in the removal of 2 items, susceptibility item QC, "I do not

think that I will experience any major problems in the

future that are caused by my high blood pressure." and

benefits of HR item HRBEN2, "My high blood pressure would

not get worse if I stopped using home remedies." In both

cases, the removal of each problematic item substantially

improved the reliability of its respective scale. There

were some exceptions to this procedure whereby scales with

low reliabilities were decomposed into single item measures

and these are discussed below.

The reliabilities of the final study scales are shown

in Table 5-2. Coefficient alpha estimates for perception

scales ranged from 0.12 for the severity scale tc 0.76 for

the benefits of HR scale. Given acceptable coefficient

alpha estimates of 0.50 or 0.60 (Nunnally, 1978), three of

the 6 perception scales exhibited poor teliabilizies.

The severity, costs of Rx, and costs of HR scales had

low reliability estimates with corresponding low interitem

corrected item to total correlations (See Appendix B). This

was not surprising given the nature of these three

constructs. Constructs can be characterized as being either

formative or reflective (Bollen, 1989). A formative

construct, having no real existence on its own, is defined