Lung--.

«0.9. .
L. I 0‘ .
.4 sfimfifi

. v : $.11th

. I .u ..
: Isl! 1..
I. l ...

 

1 WWW.)
. .- Immuwfl». turn “I.

'1‘.-.

.. 5.55...
-\.|

35.535.595.525.

.
.l..\

55.55

 

 

THESIS

ll llllllllllll

LIBRARY
Michigan State
University

 

 

 

 

This is to certify that the
thesis entitled
THE RELATIONSHIP BETWEEN SOCIOECONOMIC STATUS
AND PREVALENCE OF BEHAVIORAL RISK FACTORS FOR
STROKE AMONG OLDER ADULTS

presented by

Trudy Ellen Day
has been accepted towards fulfillment

of the requirements for

._M._S_.__degree in M. n

Major professor

Date /2//0/?8

MS U is an Affirmative Action/Equal Opportunity Institution

 

PLACE IN REI'URN Box to remove this checkout from your record.
To AVOID FINIS return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

tttttt

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

W W14

THE RELATIONSHIP BETWEEN SOCIOECONOMIC STATUS AND
PREVALENCE OF BEHAVIORAL RISK FACTORS FOR STROKE AMONG OLDER
ADULTS
By

Trudy Ellen Day

A THESIS

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

MASTER OF SCIENCE
College of Nursing

1998

ABSTRACT
THE RELATIONSHIP BETWEEN SOCIOECONOMIC STATUS AND
PREVALENCE OF BEHAVIORAL RISK FACTORS FOR STROKE AMONG OLDER
ADULTS
By

Trudy Ellen Day

Recognition of socioeconomic status (SES) as a predisposing factor for engaging
in unhealthy lifestyle behaviors could potentially yield major dividends in stroke
prevention Directing interventions that change high risk lifestyle behaviors at those
persons at greatest risk would make the most emcient and efl‘ective use of scarce health
care resources. This study’s purpose was to determine the relationship between SES and
prevalence of behavioral risk factors for stroke among older adults. Williams’ (1990)
theory on SES and health outcomes provided the framework for this study. This non-
experimental descriptive study was conducted using data from the Northern Michigan
Community Health Assessment (1995) survey which contained questions on income,
education, and behavioral risk factors—hypertension (H'I'N), smoking, alcohol
consumption, blood cholesterol, obesity, and physical inactivity. Subjects aged 55 and
older were included in this secondary data analysis; a stratified simple random sample
totaling 2,710 older adults. Prevalence of behavioral risk factors by income and education
was determined using frequency counts which show a higher prevalence of HTN, current
smoker, obesity, and inactivity among participants in the lowest income and education
level. Chi-Square test revealed that there were significant differences in prevalence of

some behavioral risk factors for income and education.

ACKNOWLEDGMENTS

This thesis would not be possible without the invaluable assistance of my
committee chairperson, Sharon King. I am grateful that she graciously accepted the
position of chairperson and aided me in identifying additional committee members. I
would like to express my sincere appreciation for her expertise in thesis writing and
research, her proficiency with electronic technology, and her swifi feedback after
reviewing drafts of this thesis. The positive feedback I received fi'om Sharon was really
what encouraged me to continue though the research process and made me truly proud of
my work.

I wish to extend my thankfulness to Patty Peck and Laura Struble, committee
members, for the time they devoted to reading my thesis proposal and final drafi and their
suggestions for change. Their recommendations served to elevate the quality of my work.

I am equally indebted to Robert Sloan and Minda Latham for their knowledge of
and assistance with the research process; and more importantly for enabling me to access
the survey data. I do appreciate the time expended answering all my questions on the
methods used to conduct the primary research study.

Lastly, many, many thanks to Gary and Molly who stood by me during this work
and supported me emotionally. I am fortunate to have such loyal supporters. Please share

in my delight at the completion of this huge undertaking.

iii

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... vi
LIST OF FIGURES ....................................................................................................... vii
INTRODUCTION ........................................................................................................... 1
PROBLEM STATEMENT .............................................................................................. 3
THEORETICAL FRAMEWORK .................................................................................... 4
REVIEW OF THE LITERATURE .................................................................................. 8
Conceptual Definitions of the Variables .............................................................. 10
Socioeconomic Status (8138) .................................................................. 10

Behavioral Risk Factors (BRF) ............................................................... 12

SES and Single Behavioral Risk Factors ............................................................. 15

SES and Multiple Behavioral Risk Factors ......................................................... 19
METHODS ................................................................................................................... 21
Sample .............................................................................................................. 21

Field Procedures ................................................................................................ 22

Data Collection Procedures and Recording ......................................................... 22
Prctection of Human Rights ............................................................................... 24
Operational Definitions of the Variables ............................................................. 24

SES ........................................................................................................ 24

Behavioral Risk Factors (BRF) ............................................................... 25

Limitations ......................................................................................................... 26
Instrumentation .................................................................................................. 28
Scoring and Data Summarizing Procedures ........................................................ 29
Research Design ................................................................................................ 30

Data Processing and Statistical Analysis ............................................................. 30
RESULTS ..................................................................................................................... 31
Demographics .................................................................................................... 31
Prevalence of Behavioral Risk Factors ................................................................ 33
Significance in Difference of Proportions ............................................................ 37
Interpretation of Findings ................................................................................... 39
DISCUSSION ............................................................................................................... 42
Limitations ......................................................................................................... 42
Implications for Advanced Nursing Practice and Primary Care ............................ 43
Recommendations for Further Research .............................................................. 48
Summary ............................................................................................................ 49

iv

APPENDIX A - UCRIHS Approval Letter .................................................................... 51

BIBLIOGRAPHY ......................................................................................................... 54

LIST OF TABLES

Table 1 - Distribution of Study Population by Income and Education .............................. 32

Table 2 - Prevalence of Individual Behavioral Risk Factors
by Income and Education ................................................................................. 34

Table 3 - Prevalence of Total Number of Behavioral Risk Factors
by Income and Education ................................................................................. 35

Table 4 - Chi-Square for Individual and Total Behavioral Risk Factors
by Income and Education ................................................................................ 38

LIST OF FIGURES

Figure l - A Paradigm for Research on Socioeconomic Status and Health
Williams, 1990, p.81) ..................................................................................... 5

Figure 2 - Schematic Representation of Williams’ (1990) Paradigm With
Study Variables ............................................................................................... 9

Figure 3 - Schematic Representation of Williams’ (1990) Paradigm With
Study Variables, Revised ............................................................................... 41

INTRODUCTION

Stroke is major cause of disability and the third leading cause of death in the
United States and Michigan (U .8. Preventive Services Task Force, 1996; Michigan
Department of Public Health, 1993). It is a major public health problem and the most
common life threatening neurologic disease that affects older adults. With an estimated
prevalence of 3 million stroke survivors, this disease places enormous burdens on family
members and caretakers, often necessitating skilled care in an institutional setting. The
direct and indirect costs of stroke in the United States have been estimated at $30 billion
annually (U .8. Preventive Services Task Force, 1996).

Many of the leading causes of death are associated with unhealthy lifestyles and
high risk behaviors. For this reason, Healthy People 2000 (U .S. Department of Health and
Human Services, 1990) established a national strategy to improve the health of all
Americans. Its purpose is to promote health and prevent disease through changes in
lifestyle and enviromnental factors. To accomplish this, broad public health goals with
specific objectives and separate priority areas that provide direction for the 10-year-drive
were established. Stroke was identified as one of several priority areas to target since
many stroke risk factors are related to lifestyle, and therefore are modifiable. Healthy
Michigan 2000 (Michigan Department of Public Health, 1993) also included stroke as an
area to target with a goal to reduce the number of stroke deaths.

Essential to reducing the incidence and prevalence of stroke is the identification of
causative factors. Risk factors for stroke in relation to unhealthy lifestyle and behaviors are

hypertension, smoking, alcohol intake, elevated blood cholesterol, obesity, and physical

inactivity (Kelly-Hayes, 1991; Haheirrr, Holrne, Hjermann & Leren, 1993). Considerable
evidence has accumulated demonstrating that maximizing healthy behaviors and
minimizing risky lifestyles can reduce the incidence of stroke (Dunbabin & Sandercock,
1990; Wolf, 1990; Matcher, McCrory, Barnett & Feussner, 1994). It has been postulated
that the decline in stroke mortality in the early 19703 was attributable to increased control
of hypertension by use of antihypertensive medications. In addition, other lifestyle factors
probably influenced the decline both directly and indirectly, for example, reduction in use
of cigarettes and reduction in sodium and alcohol intake on blood pressure respectively
(Casper, ng, Strogatz, Davis & Tyroler, 1992; Jacobs, Mcgovem & Blackburn, 1992).
This supports that risk factors can be eliminated or modified which is the basis for primary
and secondary prevention (Wolf, D’Agostino, Belanger & Karmel, 1991).

Equally important for reducing the incidence of stroke is the identification of
predisposing factors for engaging in unhealthy lifestyles and high risk behaviors (W ister,
1996). AS noted in Healthy People 2000, “Health disparities between poor people and
those with higher incomes are almost universal for all dimensions of health.” (U. S.
Department of Health and Human Services, 1990, p. 29) Socioeconomic status (SES) has
been described as one of the strongest and most consistent predictors of a person’s
mortality and morbidity (kaleby, Jatulis, Frank & F ortmann, 1992). Mster (1996)
found that SES affects health behaviors in relatively important ways but this depends on
the measure, specific behavior and the age group. Most studies examine associations
between a Single measure of SES and a single disease risk factor, or compare the mortality
and morbidity experience of different SES groups. (Jefl‘ery & French, 1996; Matthews,

Kelsey, Meilahn, Kuller & ng, 1989;Jacobsen & Thelle, 1988). Recognition of

SES— specifically, income and education as a predisposing factor for high risk behaviors
could potentially yield major dividends in disease prevention. A reduction in risk factors at
any age can have a positive impact on health outcomes.

More research on identifying persons prone to engaging in unhealthy behaviors,
could aid health care practitioners in impacting stroke incidence. Determining the extent to
which SES is related to high risk behavior could assist with the identification of
interventions that change lifestyle behavior and enhance the health of older adults.
Understanding the association between measures for SES and a set of risk factors for
stroke, risk factor reduction can be launched by essential plamring, services, and
education Educational and behavioral interventions, provided in the primary care setting
and directed at those persons at greatest risk, would make the most eficient and efl‘ective
use of scarce health care resources.

PROBLEM STATEMENT

Examining the association between income, education and a set of risk factors for
stroke-namely, hypertension, smoking, excessive alcohol consumption, elevated blood
cholesterol, obesity, and physical inactivity could conceivably aid in identifying the stroke-
prone person. Study of the prevalence of these risk factors in different socioeconomic
groups among older adults may lead to the possibility of focusing preventive measures on
areas where maximum effect may be gained. Therefore, the purpose of this study is to
answer the question: “Is there a sigmficant relationship between SES and prevalence of
behavioral risk factors for stroke mnong older adults in Northern Michigan?”

Two hypotheses will be tested in this study. The hypotheses are: (1) higher

prevalence of behavioral risk factors-hypertension, smoking, excessive alcohol

consumption, elevated blood cholesterol, obesity, and physical inactivity is associated with
lower income; and (2) higher prevalence of behavioral risk factors—hypertension,
smoking, excessive alcohol consumption, elevated blood cholesterol, obesity, and physical
inactivity is associated with lower educational attainment.

THEORETICAL FRAMEWORK

The fiamework for this study will be explained using erliams’ (1990) theory on
SES and health. Williams (1990) provides an explanatory framework for predicting health
outcomes. The relationship of the study variables to the fi'amework will be described.

Williams (1990) posits SES as an important determinant of health status in his
Paradigm for Research on Socioeconomic Status and Health. According to Williams
(1990), social status shapes individual values and behaviors, consequently linking social
structure and health outcomes. This perspective predicts that because “social structures
Shape individual values and behaviors, SES differentials in morbidity and mortality are due
at least in part to conditions in life that derive from an individual’s structural position” (p.
81).

In this paradigm, as illustrated in Figure 1, several factors are linked to health
outcomes. Psychosocial factors refers to lifestyle characteristics and living conditions that
are viewed not as individual characteristics but as the patterned response (response to
stress, a way of coping) of social groups to the realities and constraints of the external
environment. Such factors include: (1) health practices, as in behaviors like nutrition and
smoking; (2) social ties, referring to social integration and support; (3) perceptions of
control, including mastery; and (4) stress in family, occupational and or residential

environment. This paradigm suggests that variations in the type and quantity of stress at

.22 .83 52.535 5.5. a... 2:5 35:88.98 .3 £885. 5 55:... < ._ 2.5...—

 

 

 

.82 can .82 .92 25.3
85 .8. E38."— can—mason

 

 

 

 

 

 

5 u
_
_

\r

\r\

   

 

\\

 

 

 

 

 

 

 

 

_
_
_
5
5
_
_

/1 I.

 

 

 

 

 

 

_
_
_
.
_
_
_
_
_

 

 

53:858. gauges .58: Spam 3.
.258 .3 S38? 3

a: 38m a

38.53. 4283 age 8898 £8: a
238m 388%?—

 

 

 

work or in the home are linked to socioeconomic position. Correspondingly, smoking,
alcohol use, health enhancing attitudes, and social ties depends on the social and economic
structures and arrangements of society (Williams, 1990). High levels of social support
may minimize the need to use unhealthy coping strategies, such as smoking and alcohol
use, to modify stress (V enters, 1989).

Another factor linked to health outcomes in this paradigm is SES. Williams (1990)
postulates that SES is predictive of health outcomes. An adequate theoretical
understanding of the relationship between social stratification and health status must
include the role played by psychosocial factors. These factors are concomitants of social
status and they represent the pathways through which the effects of social stratification are
mediated to individuals. Nelson (1994) points out that social stratification results in the
concentration of the poor in low-status, physically demanding, and less rewarding jobs in
which they also experience greater risks of unemployment, exposure to occupational
hazards, and increased psychological stress.

Medical care, another factor within this paradigm, alludes to access to and
utilization of health care. Psychosocial factors and medical care are viewed as linked to
social status and can be mediators of the association between SES and health outcomes.
Both of these factors are posited to affect health outcomes either by direct additive
relationships as illustrated in Figure 1 (indicated by a thin, solid line), or by interactive
relationships with social status (indicated by a dotted line). In other words, lower SES
persons not only may receive more exposure to psychosocial risk factors and deficits in
medical care but also may be more vulnerable to them (Kessler, 1979; Williams, 1990).

The relationship between these two explanatory factors-medical care and psychosocial

factors may also be reciprocal. Medical care, especially preventative care, can influence
psychosocial factors directly (indicated by a thin, solid line); and psychosocial factors can
afl‘ect patterns of health care utilization (Williams, 1990).

Some researchers argue that social factors are even more important determinants
of health status than the availability of medical care and health knowledge. Disease states
are more clearly related to social phenomena such as stress and industrialization (Cassel,
1976). Each additional dollar spent on education, formal or health education, improves
health status more than each additional dollar Spent on medical care. Health education
achieves only limited success however, and is more effective in producing behavior change
in higher SES persons (Slater & Carlton, 1985). It has been noted that more educated
persons are more aware of health risks and more likely to initiate interventions to reduce
these risks (Cockerharn, Lueschen, Kunz & Spaeth, 1986). Adopting health enhancing
behavior is not simply the result of greater health knowledge. Formal education has
pervasive efi‘ects on health apart from health knowledge. Persons with less formal
education face different stnrctural constraints, have more stress and fewer resources to
cope with this stress. Risk factors for distant health outcomes may be the basic survival
strategies of day-to-day existence for low SES persons (Venters, 1989).

Additional factors linked to health outcomes within this paradigm are demographic
factors and biomedical factors. Demographic factor includes age, race and sex. This factor
has a direct additive relationship to all other factors within the paradigm. The strength of
the relationship; however, is less (indicated by a thin, solid line) than those factors linked
with a thick, solid line. Biomedical factors refers to genetics or make-up of the body, and a

direct additive relationship exists with psychosocial factors, health outcomes, and SES.

This relationship is weak as indicated by a thin, solid line (Williams, 1990).

Williams’ (1990) organizational framework provides an opportunity for examining
the relationship between the study variables. Figure 2 illustrates the variables in this study
(depicted in bold, lowercase letters) and their connections to health outcomes within this
paradigm. The intent of this research is to focus on the relationship between SES and the
prevalence of six behavioral risk factors for stroke among older adults. These variables are
identified in the fiamework under socioeconomic status and psychosocial factors. The
predictive study variables include income and education, and are identified under SES.
The outcome variables hypertension, smoking, excessive alcohol consumption, elevated
blood cholesterol, obesity, and physical inactivity are identified under health practices
within psychosocial factors. Stroke risk is the health outcome of interest.

In summary, this paradigm accepts explicitly that health status is the result of
complex causes; it includes controls for sociodemographic factors, such as age and sex,
and for early environmental, genetic, and biomedical factors. Moreover, it suggests that
psychosocial factors such as health behavior, stress, and the resources to cope with stress,
can account for a substantial part of the association between social status and health
outcomes. lastly, this paradigrn infers that primary prevention be directed toward changes
in social and economic environment rather than a person’s behavior (Williams, 1990).

REVIEW OF THE LITERATURE

The purpose of this Study is to determine the extent to which SES is related to the

prevalence of behavioral risk factors for stroke. The literature will be reviewed to

ascertain what is already known about the study variables. First, research focusing on

£23.; Sam 5..: 3.5.5.. 8.3: .255...» a 8:38.23. 8.523 .N 2.5....

 

 

«E.— oxehn .
”3803.5 .2855

f

 

 

 

 

 

 

 

 

$2.8... 3...... -

$8.... -

358835 ecu-30.0 ..

newt—8:239 =58... ..

memo—ea: -

noun—6.3%.— -
8388a .23: 2
£38..— Emogonozmm

 

 

55:63.0 158.85 -
2:35 ..

538m 0588080505

 

conceptual definitions and measurement of SES and behavioral risk factors as they relate
to stroke risk will be discussed. Second, research concentrating on SES and single
behavioral risk factors, as well as SES and multiple behavioral risk factors, will be
examined.

C l D [i . . E l I! . l l

W. Whenused in this study, education and income are
the independent variables used as the primary measure of SES. Education is the highest
grade or year of school completed (kaleby, Fortman & Barrett, 1990). Income is annual
income from all sources before taxes (Winkleby et al., 1990).

A variety of approaches to the conceptualization and measurements of SES have
been taken in research (Feinstein, 1993, Kaplan & Keil, 1993). Social class, social status,
occupation, employment status, income, and education are some examples. Income and
education are more objective measures and are therefore typically used more often
(Nelson, 1994). Education is the most widely used measure of SES in epiderniologic
studies. Kaplan and Keil (1993) examined SES measures in an extensive literature review
and concluded that of the studies of chronic diseases published in the Americanlmmalgf
Epidemiology in 1982 and 1985 in which SES was used, education was used 45% of the
time as a measure of SES. The thought is that this is primarily because questions about
education in surveys have relatively low nonresponse rates due to the simplicity of the
question to answer. Another thought is education usually remains constant throughout
adulthood and as a result is unlikely influenced by poor health among adults, unlike
income and occupation (F einstein, 1993).

Although education is very useful as a measure, it is not without some limitations.

10

For example, there are large cohort difl‘erences in level of education, so that the behavioral
and social correlates of a given level of education may vary depending on a person’s age
(Liberatos, Link & Kelsey, 1988). The value of education as a measure of SES may also
differ for certain subgroups such as Afiican-Americans. For African-Americans and
women, the link between education and income is weaker than it is for Caucasian men.
These sub-groups receive less economic return from their education than do Caucasian
men (Kaplan & Keil, 1993). The usefulness of education as a predictive variable in certain
populations may be lessened due to an increasing homogeneity in educational level.
Additionally, education doesn’t reflect acquired experience or continuing education—for
example, certification programs, conferences and seminars. The percent of U. S.
population with at least a high school education was approximately 75% in 1986,
reflecting a decreased stratification of the population based on educational level (Nelson,
1994). Lastly, although illness in adulthood cannot influence level of education completed
before adulthood, it is not implausible that poor health as a child could influence the
amount or quality of education ultimately obtained (kaleby, et al., 1989).

Measures of income are obviously an important measure of SES. Income provides
access to services and goods including medical care and quality education that may protect
against disease. However, lower income may reflect the influence of poorer health
(Nelson, 1994). In addition, the measurement of income level is complicated. Family or
individual income can be measured; income can be compared to poverty levels; family
income can be adjusted for family size; sources of income other than wages can be
included, such as food stamps or Medicare; and income levels may vary over time. It may

also be important to measure aflluence which includes total assets. Income was reported in

11

only 15% of the research articles published in the WWW in
the 1980s that had measures of SES. This could be due to the perceived sensitivity of

survey respondents on income related questions. Nonresponse rates for income related
questions on surveys averages 9% to 10%. Respondents need to be reassured about the
confidentiality of the data they provide to reduce this mimber (Kaplan & Keil, 1993).

mm For this study, BRF are the dependent variables
and refer to self-reported: (1) high blood pressure; (2) smoker or heavy smoker, (3) heavy
drinker or binge drinker, (4) elevated blood cholesterol; (5) overweight; and (6) inactivity
(Community Health Assessment, 1995; Stein, Lederman & Shea, 1993).

1. High blood pressure, as defined by Uphold and Graham (1994), is a systolic
blood pressure greater than 140 mm mercury or a diastolic blood pressure greater than

90 mm mercury. Hypertension is the most potent and common precursor of stroke. It has
long been recognized as an important risk factor for stroke. The American Heart
Association (1988) says that the risk of stroke rises directly with a person’s blood pressure
and both systolic and diastolic pressures are important. People with hypertension sufi‘er 2
to 4 times as many strokes as those with normal blood pressure.

The Behavioral Risk Factor Surveillance System (BRFS S), an ongoing surveillance
program developed and funded by the Centers for Disease Control and Prevention, is
designed to estimate the prevalence of risk factors for the major causes of death in the
US. Data from the BRFSS are a central component of federal and state activities designed
to monitor progress toward achieving the Healthy 2000 health objectives (Stein,
Lederman & Shea, 1993). The Behavioral Risk Factor Survey (BRFS) operationally

defines hypertension as “self-reported hypertension measured as ever being told by a

12

doctor or other health professional that one has high blood pressure; or self-reported
medical prescription for hypertension” (Jackson, Jatulis & Fortrnan, 1992, p.414). Another
definition is, on more than one occasion blood pressure was high or is taking blood
pressure medication (Stein, ct al., 1993).

2. Smoking promotes atherosclerosis and is the leading risk factor for stroke (U .8.
Preventive Services Task Force, 1996). Smoking is probably the most studied lifestyle
behavior afl‘ecting health. The BRFS defines smoking as self-reported smoking status—-
current cigarette use, cigarette consumption and former cigarette use (Stein et al., 1993;
Jackson et al., 1992). In the Nurses’Health Study of approximately 120,000 women who
were followed prospectively for 8 years for the development of diseases, including stroke,
current smoking was strongly associated with stroke. The relative risk increased directly
with the number of cigarettes smoked. Heavy smokers (defined as more than 40 cigarettes
per day) had nearly fourfold increased risk of stroke when compared with nonsmokers
(Colditz, Bonita & Stampfer, 1988). In a study of male cigarette smokers, Kleg and
Whelton (1987) found that up to 16,000 strokes per year in men could be attributed solely
to smoking. According to the National Center for Health Statistics data ( 1990), the age-
adjusted prevalence for current cigarette smoking is greatest for persons with less than 12
years of education, nearly 2 V2 times the rate for those with 16 or more years of education.

3. Heavy drinking is defined by 60 or more drinks consumed, on an average, in a
month (Stein et al., 1993; Community Health Assessment, 1995). Heavy alcohol intake
has been associated with strokes (Palomaki & Kaste, 1993). There is support for an
increased incidence of stroke with an increase in alcohol consumption in men.

Nondrinkers and heavy drinkers have a higher incidence of stroke, whereas moderate

13

users generally have a lower incidence (W annarnethee & Shaper, 1996; Hillbom & Kaste,
1982). Binge drinking is defined as five or more drinks consumed on an occasion at least
once in the past month (Stein, et al., 1993; Community Health Assessment, 1995).

4. Elevated blood cholesterol is defined in the BRFS as cholesterol level
200mg/dL or higher; derived from non-fasting venous samples measured in milligrams per
deciliter (Stein et al., 1993). A strong positive relationship exists between blood
cholesterol levels and risk of coronary heart disease; however, less reliable evidence exists
as a direct relationship between stroke incidence and cholesterol levels (Iso, Jacobs &
Wentworth, 1989). It can be rationalized, however, that lowering blood cholesterol levels
through diet modification and or drugs is an efi‘ective treatment for stroke prevention since
those people with coronary disease are at risk for stroke (Dyken, Wolf& Barnett, 1984).

5. Obesity measured by body mass index (BMI) which is defined as wt(kg)/ht in
meters squared. The mean BMI is calculated fi'om self-reported weight and height. Obese
is a BMI exceeding 27.8 kg/rn2 for men and 27.3 kg/m2 for women (Winkleby et al., 1990;
Stein et al., 1993; Community Health Assessment, 1995). Whether obesity contributes
independently to stroke is uncertain. Obesity does, however, aggravate hypertension, heart
disease and diabetes; all definite risk factors for stroke (I-Iaheim, Holrne, Hjermann &
Leren, 1993).

6. Inactivity is termed sedentary lifestyle and is defined as no exercise in past
month in the BRFS (Stein et al., 1993). Physical activity in the literature has been studied
using leisure time physical activity, job related activity, house hold physical activity, and
walking (Ford, et al., 1991).

Finally, in stroke, the importance of each specific risk factor depends on two

14

factors. First, whether the risk factor itself is a high-level risk that, in itself, places a person
in a higher probability range for stroke by the strength of its association with stroke.
Second, the number and combinations of risk factors present that join to make a high-risk
status for stroke for the person (Kelly-Hayes, 1991 ).

There is a considerable body of evidence for an association between SES and
mortality. The results showed an inverse relationship between level of education and
mortality (Salonen, 1982; Siegel, et al., 1987; Keil, Sutherland, Knapp & Tyroler, 1992).
Specifically, the National Health and Nutrition Examination Survey revealed that men
aged45 to64with0to 7years ofeducationhad 1.96 timestheriskofdeaththanthose
with 12 or more years of education. Those with 8 to 11 years of education had 1.6 times
the risk (Feldman, Makuc, Kleimnan & Coroni-Huntley, 1989). The results of the
National Longitudinal Mortality Study (Roget, Sorlie, Johnson & Scmitt, 1992) indicated
a strong inverse relationship between total family income and death. Caucasen males with
incomes ofless than $5,000 had amortality ratio 1.8 times greaterthanthat ofmenwith
incomes of $50,000 or more, and for women the ratio was 1.3. Even greater differences
were seen for subgroups like Afiican-Americans.

The literature has primarily addressed the influence of SES indicators on individual
behavioral risk factors, but some studies have looked at multiple risk factors. Probably the
most evidence for an inverse relationship between SES and risk factors has to do with
hypertension (Feinstein, 1993; Kaplan & Keil, 1993). The research supports that low SES
is related to both the prevalence and incidence of hypertension (Keil, Tyroler, Sandifer, &

Boyle, 1977; Keil, Sandifer, Loadholt & Boyle, 1981). A statistically significant inverse

15

relation between educational level and blood pressure was found among all Caucasians
subjects in the Chicago Heart Association Detection Project. This was also true for
younger Afiican-American men, but for older subjects and women the relationship was not
significant due to the small number of subjects in these groups (Dryer, Starnler, Shekelle &
Schoenberger, 1976).

A strong inverse relationship between level of current smoking and education is
found in the literature (Pierce , Fiore, Novotny, Hatziandreu & Davis, 1989).
Age- adjusted smoking prevalence was strongly and inversely related to educational levels
for both males and females in a study conducted in the Twin Cities area. Study participants
with less than a high school education were more likely to be current smokers than were
other education groups. The same pattern was also true for income; low income was
associated with the high prevalence of smoking in men and women (Iribarren, et al.,
1997). In a study that examined the effects of SES on patterns of exercise and smoking for
three age groups, Wister (1996) found that education and income exhibited statistically
significant relationships with smoking. Results depended upon the age group and the
specific behavior. Persons with post-secondary education were less likely to smoke than
those with lower education for the two age groups under age 65. Persons who completed
high school were more likely to smoke than persons with lower education for the 25 to 44
age group. However, this did not hold true for the 45 to 64, and 65 and over age groups.
In contrast, income appeared to predict smoking behavior in a more uniform pattern
across age groups. The middle income group was less likely to smoke than the lower
income group for all age groups. The high income group was less likely to smoke than the

lower income group, but only among persons aged 25 to 44 and those aged 45 to 64. No

16

relationship was supported for persons aged 65 and over. Mirand and Welte (1996)
sought to estimate prevalence and observe whether active or health oriented lifestyles
were related to heavy alcohol intake among the elderly. They found that the prevalence of
heavy drinking among the study participants was 6%. Adjusted analysis showed positive
associations between heavy drinking and being male. Negative relationships were observed
between heavy drinking and SES.

There are few published papers that specifically focus on the relationship between
total cholesterol and SES; most focus on high density lipoproteins (the “good guys”). A
study of 561 men and women, that had a 9 year follow-up, found that educational
achievement was positively related to high-density lipoprotein levels for both genders
(Donahue, Orchard, Kuller & Drash, 1985).

Reports of a study among a population of 59,566 female volunteers in the “Take
Off Pounds Sensibly” program in the US. and Canada conclude that there was an inverse
relation between obesity and SES (Oken, Hartz, Giefer & Rirnrn, 1977). Another study,
based on a longitudinal survey of a national probability sample of 10,039 subjects 16 to 24
years of age explored the relationship between SES and obesity. The researchers found
that during the course of 17 years, overweight women married less often, had lower
incomes and more of them had incomes below the poverty level, and completed fewer
years of school. Similar, but weaker, trends were found among men (Gortmaker, Must,
Penin, Sobel & Dietz, 1993).

Levels of physical activity have consistently been shown to vary by SES, although
not always in the same direction for leisure and on the job physical activity. Matthews,

Kelsey, Meilalm, Kuller & ng (1989) found that middle-aged women who had an

17

advanced degree expended 56% more energy per week in nonwork related physical
activities than women with a high school education or less. Similarly, those in the lowest
social class had an increased risk ofbeing inactive 4 1/2 times that ofsubjects in the highest
classes (Helrnert, Herman, Joeckel, Greiser & Madans, 1989). Mster (1996) examined the
efl‘ects of SES on patterns of exercise and smoking for three age groups: persons age 25
to 44, 45 to 64, and 65 and over. Results indicated that among the 25 to 44 age group,
persons having completed high school are 1.51 times as likely to be active compared with
the lower education group (some high school or less). Statistically significant but slightly
weaker associations between education and activity were found for the other two age
groups. Conversely, income was not found to be an important predictor of activity level
for any of the age groups. Two studies using SES measure that combined education and
income, had conflicting findings. A direct relationship between SES and level of both
occupational and leisure time physical activity was discovered with those in the lower class
being less active (Holrne, Helgeland, Hjermann, Leren & Lund-Larsen, 1981). In contrast,
lower SES men in Finland had higher levels of occupational physical activity than higher
SES men (Salonen, Slater, Tuomilehto & Raurarnaa, 1988). Ford et a1. (1991), using 574
subjects living in Pittsburgh, demonstrated in his study that higher SES status women
spent significantly more time each week in leisure time physical activity, job related
activity, and household physical activity than did lower SES women. Lower SES men
spent significantly more time each week walking and doing household chores, whereas
higher SES men tended to be more active in leisure time physical activity. These data

suggest important difi‘erences in activity among population subgroups.

18

SES llll'lEl . ”3.”,-

There is a substantial amount of literature on the relationship between SES and
individual risk factors; whereas, very little literature is available on SES and multiple risk
factors. In a study examining associations between education and six behavioral risk
factors for disease, a highly significant relationship was found. Those participants with
lower education were at higher risk for disease. Increased education was related to lower
smoking, obesity, cholesterol levels, and hypertension prevalence. The findings were
remarkably significant (p < .01) for both sexes and in the young as well as older age
group, with the exception of cholesterol among males and the 50 to 74 year old age group
(Winkleby, et al., 1990).

In a study that examined correlations among education, income, and occupation
and a set of risk factors—smoking, cholesterol, blood pressure, positive correlations were
found among the three SES indicators. Correlations ranged fiom 0.23 to 0.67 with the
strongest being between education and occupation and the weakest being between
education and income. All correlations among the three indicators and risk factors were
found in an univariate analysis; when a forward regression analysis was used, only
education was found to be significantly and inversely related with the risk factors
(Winkleby, Jatulis, Frank & Fortmann, 1992).

The association between educational attainment and behavioral risk factors for
disease was studied among 541 middle-aged women in Pennsylvania. The study showed
that the less educated women had higher blood pressure and higher low density
lipoproteins (all p values < .01). The less educated women reported health behaviors

distinctly different from those more educated. The less education the women had, the

19

more often they reported being current or ever smokers of cigarettes and expending kw
kilocalories per week in physical exercise; and the less often they reported consmning
alcohol one day a week or more (all p values < .01). Interestingly, there were minimal
difi‘erences in diet among the women according to education (Matthews, Kelsey, Meilahn,
Kuller & Wing, 1989).

Cockerrnan, Lueschen, Kunz and Spaeth (1986) hypothesized that as SES
increases the more likely a person will be involved in forms of health-advancing behavior;
the poor will be less engaged in these behaviors than those of higher status. Income,
education, and occupation were the components used for measuring SES. Food habits,
participating in sports and activities, smoking status, and alcohol consumption were
indicators used to reflect health-advancing behavior. The researchers found that, in 401
adult residents of Illinois, income and occupation were significant with respect to food
habits; and education was significant with respect to sports and activities. Precisely, lower
income persons were more likely to select their food carefully, as were persons in higher
status occupations (but not necessarily with higher incomes). Thus, findings for groups
opposite ends of the social spectrum appear to coincide. Persons with higher education
and younger persons were significantly more likely to participate in sports and exercise
than those with limited education and older. For smoking status and alcohol consumption,
only sex was statistically significant in that men reported more drinking and smoking than
women. Overall, the results did not lend strong support to their hypothesis. Only
education was significant with respect to sports and exercise, unlike the other indicators
for SES and health-advancing behaviors.

In summary, there is an abundance of literature linking SES with behavioral risk

20

factors. In the few studies of multiple risk factors there has generally been an inverse
relationship between SES measures and hypertension, smoking, excessive alcohol intake,
blood cholesterol, and obesity. There is also evidence for an inverse relationship between
individual risk factors and SES. The literature is most abundant for the relations between
SES and hypertension or SES and smoking. The literature is more limited for SES and
obesity or SES and physical activities; and quite scarce for examining multiple measures
for SES and a set of behavioral risk factors in the older adult population The evidence for
a relation between cholesterol level and SES is inconsistent.
METHODS

Sample

Data for this study was derived from the 1995 Northern Michigan Community
Health Assessment (NMCHA) survey. The survey consisted of a stratified simple random
sample of 6,533 adults 18 years of age and older. The strata were the 21 northern counties
of the upper-lower peninsula of Michigan participating in the study. Random digit-dialing
method was used to generate the probability sample. The sampling frame was all
residential telephone numbers ringing into households in the 21 counties that fall into
known banks of working residential numbers. The frame was provided by Survey
Sampling Incorporated (SSI). The 881 method used all possible area code, exchange, and
working block combinations known to contain at least one working household number. As
a result, all telephone households in the geographic sampling flame were given equal
probability of selection within the limits of available data. A minimum of 300 surveys in
each county was required. When a residential number was successfully contacted, the

household members 18 years of age and older were listed according to BRFS procedures;

21

men and then women were listed by first name, nickname, or position within family (i.e.,
father, uncle) fiom oldest to youngest. Wrthin each sample household, an adult respondent
was randomly chosen from all resident adults 18 years of age or older. A computer-
generated random number was used to select among eligible respondents listed in a
household. Substitutions were not permitted (Community Health Assessment, 1995).

For this study, in which the focus is on older adults, only subjects aged 55 and
older were included in the secondary data analysis. This yielded a sample size of 2,710
(Community Health Assessment, 1995).
EieldBmcedures

In the primary research study, data were ascertained through a one-time telephone
interview conducted by Information Transfer Systems Incorporated (ITS) of Ann Arbor,
Michigan, from July to September 1995. ITS, a research firm contracted out by the North
Central Counsel of the Michigan Health and Hospital System, followed data collection
procedures that are consistent with the Michigan BRFS project (Remington, et al., 1988).
ITS used computer-assisted telephone interviewing which permits direct entry of the data
into a computer file, thus facilitating interviewer monitoring, data coding and entry, and
quality control procedures. Interviewers were ITS employees experienced in health
surveys and in the use of computer-assisted interviewing device, and were given
specialized training in the use of the survey (Community Health Assessment, 1995).
Dmfioflmfimmmdmming

When a phone number was called and an eligible respondent selected in the
primary research study, 1 of 3 outcomes were possible: (1) the interview was completed;

(2) the respondent refilsed to be interviewed or hung up during the interview; or (3) the

22

respondent selected was unavailable and an appointment was made for call-back When
the phone mnnber was working but the line was busy, a recorder or no one answered the
phone, the number was redialed. Repeated attempts were made to contact the selected
respondent. If the respondent was unavailable after multiple attempts, he/she was not
included in the sample. Ifthe number reached was identified as a nonresidential number or
outside the geographic area studied, the interviewer informed the respondent that he/she
was ineligible (Community Health Assessment, 1995).

Interviewers used computer-assisted telephoning interviewing devices that
displayed the interview instrument on a computer screen and permitted immediate coding
of responses by the interviewers, consistent with the state health department data
collection procedures. To maintain consistency, information and instructions were given to
the subjects using a script. All subjects were asked to respond to a series of questions
related to demographics, health behaviors and practices. The questions on the survey were
core questions taken from the Michigan BRFS. The core of questions includes one or
more questions in the area of general health status, access to health care, smoking, blood
cholesterol, diabetes, diet, obesity, hypertension, physical activity, alcohol use, seat belt
use, and health preventative practices (Community Health Assessment, 1995). All
questions and answers were coded in the primary research study.

For this study, in which the purpose is to examine the relationship between SES
and behavioral risk factors, only data related to hypertension, smoking, alcohol use, blood

cholesterol, obesity, physical activity, income and education was used in the secondary
analysis.

23

E . [H E. 1

In the initial study, the principal researcher protected the right of the respondents.
Subject participation was voluntary and subjects were free to withdraw from further
questioning at any time during the telephone interview. Informed, verbal consent was
obtained fi'om each respondent prior to questioning. Names (first name, nickname, or
position in family—father, uncle) were only used to randomly select respondents.
Confidentiality of subjects was maintained in that respondents were informed that their
answers would be kept strictly confidential and no information which identified them
would be released or published. Anonymity was maintained in that names were not
attached to the individual data collection files which were only accessible to data
collectors and researchers.

The investigator of the secondary data analysis also protected the rights of
subjects through strict adherence to standard criteria set forth by the Michigan State
University Committee on Research Involving Human Subjects (Appendix A).
Confidentiality and anonymity were preserved in that the investigator did not have access

to names and referred to subject’s only by their identification number assigned to the data

file.

 

SES, Reported household income fi'om all sources before taxes and highest grade
or year of school completed were used as the index of SES which is the independent
variable for this study. Income information was obtained fi'om a single questionnaire item
that requested total annual income from all sources before taxes in 1994 in the primary

research study (item QINC on the survey). Income was recorded in 1 of 5 classes: (1) less

24

. ‘H-l'

 

than $10,000; (2) $10,000 through $19,999; (3) $20,000 through $34,999; (4) $35,000
through $50,000; or (5) greater than $50,000. Education information was obtained fi'om a
single questionnaire item that asked highest grade or year of school completed in the
primary research study (item QEDUC on the survey). Education was documented in 1 of
4 categories: (1) less than high school; (2) high school graduate; (3) some college; or (4)
college graduate (Community Health Assessment, 1995).

W), Hypertension was assessed by a single questionnaire
item from the primary research survey that asked respondents if they had ever been told by
a health professional that they lmve high blood pressure (item Q24 on the survey).
Response categories were yes or no. When considered together, two items on the survey
were used to measure the variable smoker. Items Q39 and Q41 asked respondents to
report: (1) having smoked at least 100 cigarettes in entire life; or (2) smoking cigarettes
now, respectively. Response categories for both questions were yes and no. The variable
smoker was categorized as “former” and “current;” the former defined by respondents
who reported that they had smoked at least 100 cigarettes but no longer smoke, and the
latter defined by respondents who reported that they now smoke. Heavy smoker was
defined as current smoker who reported smoking 40 or more cigarettes per day. Item Q42
was used to measure heavy smoker. It asked respondents to state number of cigarettes
smoke per day on an average.

Heavy drinker was defined as consuming 60 or more drinks on an average in the
previous month; and binge drinker was defined as 5 or more drinks per occasion
consumed at least once in the previous month. Item QALC2 asked respondents how many

days per week or month they drank alcoholic beverages. Item QALC3 asked respondents

25

average number of drinks consumed on those days when they did drink. QALC4 asked
respondents how many times during the past month did they have five or more drinks on
an occasion. These three items when considered together were intended to reflect the
variable heavy/binge drinker.

Elevated blood cholesterol was defined by a single questionnaire item (item Q27
on the survey) that asked respondents if they had ever been told by a health professional
that their blood cholesterol was too high. Overweight was defined as BMI greater than or
equal to 27.8 kg/m2 for men and 27.3 kg/m2 for women. BMI will be calculated using item
QWGT and QTALL which asked respondents to state their weight and height,
respectively. The formula for calculating BMI was weight in kilograms divided by height
in meters squared. Inactivity was defined as 2 days per week, or less, of at least moderate
exercise, measured using item QMODEX which asked respondents to state average
number of days per week that they exercise (Community Health Assessment, 1995).

Limitations must be considered when using surveys, computer assisted
interviewing, and secondary data. First, telephone surveys may lead to biased population
estimates because of the under representation of certain groups. This bias may result from
difficulty in contacting certain groups; for example, persons who leave town for long
periods of time for work or leisure related reasons; and households without a telephone,
often the case in very low income households. Also, this bias may result from high refirsal
rates; for example, persons who are unwilling to be surveyed; and severely hearing
impaired persons (Remington et al., 1988). Supplementary, telephone surveys, or surveys

in general, use a forced-choice answer format. The most appropriate response may not be

26

available to the person being surveyed resulting in inaccurate or missing data.

Second, despite the advantages of immediate data entry with computer assisted
interviewing, key board errors can occur. Verification of data entry is not possible when
data is entered into the computer during the interview. Using paper forms rather than a
computer with subsequent key entry would permit detection of obvious errors at the
keying stage (Shea, et al., 1991).

Third, because the researcher used data collected from a prior research study,
questions may be worded or phrased differently than what would have been preferred by
the researcher. For example, income was grouped into categories which can be
problematic with regards to comparability across households of difi‘erent sizes. Family
size must also be included since the impact of a given income is vastly different for a
family of 2 compared with a family of 10. Liberatos et al. (1988) recommends using the
poverty index developed by the federal government, which adjusts for family size and cost
of living, as a base. Income categories that are multiples of the poverty level can be
created. The categorical approach is more common since many people feel less
uncomfortable if they place themselves into groups. Another example, college graduate
was one category choice for measuring education. This category fails to delineate type of
college-2 year or 4 year. Type of college, number of years or private verses public; type
of degree earned; and major field of concentration are educational differences that in turn
can have an impact on subsequent income and occupational attaimnent.

Furthermore, the researcher does not have the opportunity to ask additional
questions that could prove useful in the secondary data analysis. For example, it is not

possible to ascertain from the survey questions measuring the variable smoker, number of

27

years smoked, total amount smoked or “pack years” and smoking cessation date.
Additionally, it is not possible to establish from the survey question measuring the variable
hypertension as to the respondent’s status-controlled or uncontrolled. These factors need
consideration since they afl‘ect rate of risk.

Lastly, the results of this study can only be generalizable to the geographic area
sampled. The survey was conducted in the summer months which included the seasonal
older adults. These older adults are typically more aflluent; thus, the proportion of persons
in the lower income groups could potentially be small.

Instrumentation

The NMCHA survey was administered. The questions for this survey were taken
from the BRFS questionnaire which has widespread usage. It is used by more than 40
state health departments in conjunction with the Centers for Disease Control to measure
state specific prevalence and time trends of a number of health related behaviors. A core
of questions on this questionnaire was derived from other national surveys such as the
National Health and Nutrition Examination Surveys and the National Health Interview
Survey. The questionnaire was used because of its reliability and validity; and because it
concentrates on behavior rather than knowledge or attitudes (Shea, Stein, Lantiqua &
Basch, 1991).

The BRFS has good test-retest reliability. Stein et al. (1993) administered the
survey twice over 21 to 44 days to a representative sample of 210 Massachusetts
residents. The prevalence of risk factors was consistent between the two administrations.
None of the differences between the two administrations were significant (p > .05).

Employing Kappa and Pearson’s r, reliability coeflicients were generally high with most

28

coeficients above .70 and many above .80. For the demographic factor variables, there
were no statistically significant differences in the variables across administrations. For
education the coeflicient was .80; and for income the coefiicient was .75.

Shea et al. (1991) administered the BRFS twice over 10 to 21 days in New York
State. Again, the prevalence of each risk factor was consistent between administrations.
Reliability coeflicients for the six behavioral risk factors that will be used in this study
ranged fi'om .65 to .85. The variable “blood pressure measured in past 2 years” was low
but this was not due to discordance in responses between administrations. One respondent
was excluded from the variable “current smoker” because on examination of the data, an
error was found. A value of 4 was used in the first entry and 74 in the second; presumably

a keying error.

 

In the primary research study, responses were recorded according to the scoring
key. Respondents who answered that they did not know or refused to answer a question
were coded as such. In the secondary analysis, respondents were excluded fi'om the total
number for each behavioral risk factor and demographic information if they gave a “did
not knouf’ response or refirsed to respond. This approach to item non-response assumes
that the proportion of respondents at risk is the same for those with missing information as
for those who offer responses to the survey item. This approach also means that the total
sample size reported for each questionnaire item will vary slightly. This procedure for
handling missing data is different than in the BRFS in that any missing data (respondent
answered don’t know or refirsed to answer) for a behavioral risk factor item is viewed as

that risk factor being absent.

29

Community Health Assessment (1995) data was stored in Wmdows; file is entitled
nccreg.sav. In the secondary analysis, data was analyzed using the Statistical Package for
the Social Sciences (SPSS). Frequency counts for each item were obtained and then
reported in percentiles.

Remhllesisn

This study was a non-experimental descriptive approach selected to examine the
relationship between SES and prevalence of behavioral risk factors for stroke among older
adults in Northern Michigan. SES, comprising income and education, was the independent
variable and 6 behavioral risk factors were the dependent variables.

A few statistical procedures were used to analyze the data. First, descriptive
statistics were used to summarize the demographic characteristics-particularly, age and
gender by income and education Demographic data that was used included income and
education which was analyzed categorically by the prevalence of each behavioral risk
factor (individual) and the total number of behavioral risk behaviors (aggregated). A
tabular presentation (cross-tabulation) of data for prevalence of individual behavioral risk
factors for stroke by income and education is in percentage form. Additionally, a tabular
presentation of data for prevalence of number of behavioral risk factors by income and
education is in percentage form. Distribution of the study population is stratified by age
ranges and reported by fiequency to include both men and women by income and
education.

Second, to test the difference in proportions (prevalence) of individual and total

number of behavioral risk factors on income and education level, chi-square statistic was

30

applied to the cross-tabulations. Chi-square is used for testing group differences when the
dependent variable(s) is measured on a nominal scale (Polit & Hungler, 1995). The chi-
square statistic was used to establish significance of different proportions (prevalence).
Statistical significance was set at the .05 level.

Two hypotheses were tested in this study. The hypotheses are: (1) higher
prevalence of behavioral risk factors—hypertension (HTN), smoking, excessive alcohol
consumption, elevated blood cholesterol, obesity, and physical inactivity is associated with
lower income; and (2) higher prevalence of behavioral risk factors—HTN, smoking,
excessive alcohol consumption, elevated blood cholesterol, obesity, and physical inactivity
is associated with lower educational attaimnent.

RESULTS

The results of this study are reported in three parts: (1) distribution of study
population; (2) prevalence of individual and multiple behavioral risk factors; and (3)
significance of difi‘erences in proportions of individual and multiple behavioral risk factors.
Wins

The sample consisted of 2,7 10 adults age 55 and older. Of the 2,710 subjects
participating in the study, 1,895 subjects responded to the income level question and 2,701
subjects responded to the education level question. Although there were more subjects in
the $20,000 to 34,999 income level and HS graduate education level, all other levels were
well represented (Table 1). Income level was less than $10,000 among 15.8% of the study
population, $10,000 to $19,999 among 26%, $20,000 to $34,999 among 29.7%, $3 5,000

to $50,000 among 16.7%, and greater than $50,000 among 11.8%. Education level was

31

 

 

 

 

 

 

 

 

25.8: 88 can u 5 8v 8” A32 u 5 as new a2: u 5 3m Be 45.2.
Gems: E on S 8 v: S 8 920 memdoo
$2: an S 3 we 8 m5 2 $348 88m
€1.95 £2 5; K as o: 8” e: 3520 mm
$5.8 as v: S as m: NS 3 E235 mmmu

_ 22:09:”.— _
A38: 32 can u 5 m8 «2 a: u 5 8—. m: a2 u 5 ma. an 45.2.
@535 «mm m a 3 a. 8 as Soda A
@035 E. R S 9. 8 3 E 25.8-3
@035 8m 2. as N: ”2 a2 a: 833%
£5.05 «a. as S .5 ea 3 on amaze;
@325 SM 3 8 a R 8 2 Sod; v
3.5 43.3 onB a zmz a 5:53 a 2m: 1. 529$ a 52 a fl HEB:— a,

+3 3.3 31?. mm”:

 

_ 035.

32

less than high school (HS) for 23.6% of the subjects participating in the study, HS
graduate for 40.1%, some college for 18.8%, and college graduate for 17.5%.
E 1 EE 1 . l E. l E

The prevalence of individual behavioral risk factors varied by SES and specific
behavior. The prevalence of individual behavioral risk factors by income and education is
displayed in Table 2. HTN was the most frequently occurring individual risk behavior for
the three lower income levels and for all the education levels except college graduate.
HTN was most prevalent among subjects with an income level less than $10,000 (51.7%)
and education level less than HS (49.3%); and least prevalent among subjects falling into
the $3 5,000 to $50,000 income level (42.4%) and college graduate education level
(41.8%). Similarly, current smoker, obesity, and inactivity were behaviors most common
in the less than $10,000 income level (21.7%, 42.8%, and 30.8%, respectively) and the
less than HS education level (19.2%, 42.8%, and 22.4%, respectively). Furthermore, these
behaviors were least common in the $50,000 income level for current smoker (11.2%) and
$3 5,000 to $50,000 income level for obesity and inactivity (34.9% and 14.2%,
respectively) and least common in the college graduate education level ( 10.4%, 30.7%,
and 14.3%, respectively).

Conversely, heavy smoker, binge and heavy drinker, and elevated cholesterol were
behavioral risk factors more common in the greater than $5 0,000 income level (16.0%,
11.7% and 9% and 43.6%, respectively) and college graduate education level (17.0%,
8.1%, 9.7% and 39.8%, respectively) when compared to the lowest income level and the
least education level. This unexpected finding could reflect the high price of alcohol,

cigarettes, and certain foods high in cholesterol, so that those with the lowest

33

 

 

 

 

 

 

 

 

m4; 5.0m mdm 6o mm 95 1: Q? Q3 20 mam—4.50
9w— oém 6; me me $2 5.2 m9. vsv ”yams—400 mEOm
n.w_ Nam Q; Wm ad #2 5.2 men 0.2. mH<DQ<MG mm
Yum adv vsm m.m as an NE osm mdv mm 25.: mmmq
_,1 ZOP<DDGE
v.3 v.3 ode od 6: 0.3 N: adv 6? 8063A
NE of". ndv 02 v.2 Q: QS ofiv v.9. ooodm..mmm
QB Fun mdv om ad Num— m.m~ _.~v owe oomiflomm
w.: 0.3 mam me fix m6 m3 how «.5. 95.265
mdm adv adv ha man Wm— 55 man 53 ooodSV
a. .x. .x. .\. .\. .\. _ ”ESE
bio: cue:— bie: «eat—.0 eunuch
E0447: HmHm—O AOKHHmHAOflU AOMGOi IHMO—zm ZEN mam
92:05— v—m—d SOEEHE 4555an—
.. 27:... 7‘11 , ... ... .a

a 03:.

34

 

TOTAL NUMBER OF BEHAVIORAL RISK FACTORS

 

 

 

 

 

SES 0 1-2 3-4 > 4
[goons 7. % v. %
< $10,000 5.0 23.0 59.3 12.7
81049999 7.7 26.8 50.4 15.0
$20-34,999 8.3 21.8 54.5 15.3
ass-50,000 8.2 21.8 56.5 13.6
> $50,000 7.6 22.4 57.4 12.6
[EDUCATION

LESS THAN HS 7.5 24.1 53.3 15.1
HS GRADUATE 8.9 24.0 53.7 13.5
SOME COLLEGE 8.1 24.8 55.2 12.0
COLLEGE GRAD 9.3 24.6 55.3 10.8

 

 

35

incomes may find these items too expensive to purchase. The mean age of the individuals
with the highest income and education level is unknown. Typically, income drops after
retirement and older individuals have less than a college graduate education level.
Therefore, it’s plausible that these individuals are still in the workforce and job stressors
are responsible for this increase in prevalence of smoking and drinking.

Finally, the prevalence of former smoker was positively related to both income and
education level; prevalence increased as income and education increased. The prevalence
of multiple risk factors varied depending on the SES indicator and total number of
behavioral risk factors (Table 3). For all of the income and education levels, subjects
reported more often a total of three to four risk factors than any other total mrmber of
behavioral risk factors. Additionally, for all of the income and education levels, subjects
more often reported a total of greater than four risk behaviors than zero risk behaviors.
Among subjects with zero risk factors, fewer were in the lowest income and educational
level (5.0% and 7.5%, respectively) as compared to all other levels; and more were in the
highest educational level (9.3%) as compared to all other education levels. The prevalence
of zero risk factors increased as education level increased; although surprisingly,
prevalence was less for some college education level (8.1%) than for HS graduate level
(8.9%). The prevalence of 1 to 2 risk factors was, as expected, more likely for the lower
income levels (23.0% for less than $10,000 and 26.8% for $10,000 to
$19,999) than the higher income levels (21.8% for $35,000 to $50,000 and 22.4% for
greater than $50,000); and unexpectedly, less likely for the lower education levels than the
higher education levels. Surprisingly, the prevalence of three to four risk factors increased

as education and income level increased except in the lowest income level. Furthermore,

36

 

the end of the income spectrum appear to coincide, the prevalence of greater than four
risk factors was similar for the lowest (12.7%) and highest (12.6%) income levels. As
expected, the prevalence of greater than four risk factors increased as the education level
decreased.

Thus, the first hypothesis that higher prevalence of behavioral risk factors--
hypertension (HTN), smoking, excessive alcohol consumption, elevated blood cholesterol,
obesity, and physical inactivity is associated with lower income is supported for only some L
of the risk behaviors. Namely, hypertension, current smoker, obesity, and inactivity were

the risk behaviors more prevalent among the lower income level. Likewise, the second

 

hypothesis that higher prevalence of behavioral risk factors-HTN, smoking, excessive
alcohol consumption, elevated blood cholesterol, obesity, and physical inactivity is
associated with lower educational attainment is partially supported. The prevalent
behavioral risk factors are the same behaviors more prevalent among subjects in the lower
income level (HTN, current smoker, obesity, and inactivity).

5° '6 . 1m [E .

Chi-Square test was used to examine the difference in proportions (prevalence) of
individual and total number of behavioral risk factors for income and education. The chi-
square statistic was used to establish significance in different proportions (prevalence).

Statistical significance was set at the .05 level. The results of the chi-square for both
individual and total number behavioral risk factors are presented in Table 4. The purpose
of this study was to answer the question: “Is there a sigmficant relationship between SES
and prevalence of behavioral risk factors for stroke among older adults in Northern

Michigan?” The results indicate that there were significant differences in prevalence of

37

 

 

 

 

 

 

 

no. w Am
fl. 8. 8. mm. 8. on. 2. 8. 8. 2. $95525
2.. 8. an. S. 8. 8. mu. 8. 8. 2. _ ”—2002:
is: 6mm... 93:39:30 .33....
ufibh 85.37: amaze Sausage—6 42504:, mac—3% 2.5 mam

 

EDOmAflU SO >g§m
...u. .. ..u._.o... .. gr." ,,..1. “yoga?“ no ..u n. .W... o . w“... 1..

v 033.

38

some individual behavioral risk factors, but not multiple risk factors, for income and
education. There were significant differences in prevalence of former and current smoker
(p = .00, p = .01, respectively), binge and heavy drinker (p = .00, p = .00), and inactivity
(p = .00) for income level. Similarly, there were significant differences in prevalence of
former and wrrent smoker (p = .00, p = .00), heavy drinker (p = .02), obesity (p = .00),
and inactivity (p = .00) for education level. Lastly, differences in prevalence of total
mrmber of behavioral risk factors for income (p = .46) and education (p = .72) were not
significant.

I . EE' 1'

Conclusions can be drawn fiom the interpretation of the findings of this study with

 

respect to the existing literature and the conceptual framework used in this study. There is
a substantial amount of literature on the relationship between SES and individual risk
factors; whereas, very little literature is available on SES and multiple risk factors. The
literature is even less limited for study population of older adults. By and large, the
existing literature supported a higher risk for disease with a decrease in education.
Decreased education was related to a higher prevalence of smoking, obesity, hypertension,
and inactivity (Winkleby, et al.,1990; Winkleby, et al., 1992; Matthews, et al., 1989; and
Pierce, et al., 1989). Prevalence did vary; however, depending on the specific measure
used for SES, the specific behavior, and the age group. Additionally, gender was found to
be a factor afi‘ecting prevalence with men reporting more drinking and smoking
(Cockerman, et al., 1986; Mirand & Welte, 1996). The findings in this study were
consistent with previous literature in that the less educated had a higher prevalence of

hypertension, current smoking, obesity, and inactivity. The lowest income level also had a

39

higher prevalence of hypertension, current smoking, obesity, and inactivity which is
consistent with some literature (Wister, 1996; Gortmaker, et al., 1993; Helrnert, et al.,
1989). Findings in this study were inconsistent with an earlier study that found a negative
relationship between heavy drinking and SES (Mirand & Welte, 1996). A positive
relationship between binge drinking and income, and heavy drinking, income and
education was found in this study. Incidently, the risk factor former smoker is more
prevalent among the higher income level and higher education level. Although this
behavior is considered a risk factor, the rate of risk declines with smoking cessation;
hence, subjects in the lower income and education level are at greater risk as former
smoker is least prevalent for these levels. This study suggests that among those subjects
with a total number of greater than four risk factors for stroke, the rate of risk increases as
the education level decreases as prevalence of greater than four risk behaviors is higher
with less education.

The Paradigm on Research on Socioeconomic Status and Health (Williams, 1990)
was used as a fiarnework for this study. Williams’ (1990) paradigm provides an excellent
conceptual framework for the variables in this study. The purpose of this study was to
determine the extent to which SES is related to the prevalence of behavioral risk factors
for stroke. The hypothesized relationship between SES and the health practices listed
under psychosocial factors in the fiamework, was partially supported in this research. The
prevalence of some of the health practices; precisely, hypertension, current smoker,
obesity, and inactivity are clearly related (a negative relationship) to SES. The prevalence
of former smoker, and binge and heavy drinking are also clearly related (positive

relationship) to SES. It’s noteworthy that the strength of the relationship between SES

4O

 

.833: .8369 83m .53 35:... 885 .EaaB .6 eoeeeoaaom 65.52% .n 8:3...

 

 

as. 8.2.... -
”88830 530m

 

2.2.
25 26:82

 

 

,

 

 

 

‘

 

 

 

_
_
_
_
_
F
a
_
_
_
_
_

 

 

£2.35 32%.... -
3.26 -

.8022er 3.55% ..
539.538 258.: .
nae—ea: -
5.33.2.3 ..
80¢er 530m

2

£8on Eoomoaommm

 

sea—8381.58 ..
2:85 -
“mafia 03888268

|

 

 

 

 

\

41

and some of the health practices studied in this research was significant as is suggested by
the thick, solid line in the framework. Other health practices studied in this research were
found to have a strong association; however, the association was not significant.

For this study, medical care (see Figure 1) was not included in the schematic
representation of Williams" (1990) paradigm (see Figure 2) as access and utilization of
health care were not variables under study. Although this research did not focus on
medical care, it is the one area within Williams’ (1990) paradigm where Advanced
Practice Nurses can impact health outcomes through a direct, or an interactive,

relationship between medical care and psychosocial factors. Subsequently, the schematic

 

representation of Williams’ (1990) paradigm was revised (Figure 3) to include medical

care linked directly to psychosocial factors (indicated by a thin, solid line) and as an

intermediary between SES and health outcomes (indicated by a dotted line).
DISCUSSION

I . . .

Additional limitations to this study must be acknowledged. First, as with all self-
report survey data, over reporting or under reporting of individual behaviors may occur.
Recalling of the past, an event or behavior, can be dificult. This is especially true if the
events occurred a long time ago, or if they require the recall of separate events or a lot of
detailed information as in the survey used in this study (Converse & Presser, 1986).

Second, no response bias must be considered when interpreting the findings as
they relate to income level. 0f the 2,710 subjects participating in the study, 1,895 subjects
responded to the income level question; consequently, there was missing income data on

815 subjects. This can produce misleading results since it is unknown if these

42

nonresponders were of a lower income level.

Third, although probability sampling was used in the primary research study,
sampling bias must be considered when drawing conclusions in the secondary data
analysis. In the primary research study, the 21 counties were the strata; therefore, it was
decided to include a minimum of 300 surveys fi'om each county. In the secondary research
study, the attributes of interest were income and education which were not identified
strata in the sample selection process of the primary research study. Consequently, the
middle income group was over represented. The representativeness of the final sample

could be enlmnced had the primary researcher determined the proportion of participants

 

needed for the various income and education levels. The primary researcher could have
asked prospective subjects questions pertaining to income and education before
determining eligibility.

Lastly, the literature suggests that risk for stroke increases as the total number of
high risk behaviors increases; nonetheless, the inclusion of former smoker in determining
prevalence of total number of behavioral risk factors for stroke potentially presents
deceiving results. Although this behavior is considered a risk factor for stroke, the rate of
risk declines with smoking cessation. No attempt was nude to rank the relative danger
posed by each of the behavioral risk factors in this study; however, this would conceivably
provide more reliable results on stroke risk rather than considering number only. Former

smoker could then be deemed “less dangerous” than the other risk behaviors.

 

This study presents several implications for nurses in advanced practice in primary

care settings who are providing health care for older adults in difl‘erent SES groups. The

43

Advanced Practice Nurse (APN) can enhance positive health outcomes through
application of Williams’ (1990) paradigm which provides a framework for impacting
health practices; consequently, health outcomes.

Initially, a distinct understanding of the APN role and responsibilities is a
prerequisite in successfirl implementation of interventions for promoting positive health
outcomes by the APN. The expert practitioner role provides the base for the integration of
role characteristics of the APN.

Asanassessor,theAPNcanobtaindataonpersonalandsocialhistory,andpast
medical and family history. Social history would include assessing financial status and
education level. Assessing medical and family history would be particularly useful in
determining risk factors for disease. Additionally, the APN needs to assess individual and
family attitudes and behaviors toward health promotion and disease prevention. The nurse
can collaborate with the client and family in assessing, improving, enhancing, and
evaluating their current health practices. Identification of client risk taking behaviors,
providing assistance in decision making to client and family regarding choice of lifestyle,
and reinforcement for positive health behavior practices of the client is needed.

Health promotion plans for clients are frequently designed with upper SES
individuals in mind. For example, recommendations to increase exercise presumes that the
client has access to a safe environment where he or she can walk. As a clinician, the APN
needs to provide preventative services aggressively targeted at lower SES groups in an
effort to reduce disparities in prevalence of risk factors between difl‘erent SES groups. The
provision of preventative care will need to be sensitive to SES-related influences on health

and health behaviors. Recommendations for adherence and treatment to prevention

44

regimes need to be tailored to the client’s life circumstances. For example, when
suggesting exercise, it will be important to explore with clients when and where this will
be accomplished, both helping them to overcome real or perceived barriers and to provide
opportunities of which they may not be aware. APNs needs to be aware that as SES
declines the need for outreach to clients increases because lower SES groups may not seek
help voluntarily. Taking part in community health initiatives, such as free clinics, and the
development of self-help support groups is a critical role for the APN.

Some of the literature suggests that the lower SES group engages in high risk
behaviors in an attempt to cope with their stressful life environmental conditions (V enters,
1989; Pesznecker, 1984). Providing concrete resources for problem solving,
strengthening and developing social support systems for low income groups can be a
mediating factor between SES and health outcome and is therefore an appropriate primary
prevention role for the APN. The APN as a counselor, can enhance the coping abilities of
clients and ofi‘er social support to clients and families in an efi‘ort to mitigate some of life
strains. An idea for an outreach program, is the development of a self-help support group
initiated and facilitated by the APN initially, but that eventually could be continued by the
group participants themselves. By sharing similar experiences, concrete information about
resources, and participating in assisting others with problem solutions, the group members
not only receive but also provide support to each other. Consequently, clients of lower
SES may be less inclined to utilize ineffective coping strategies such as high risk behaviors
as smoking and drinking.

As a leader, the primary responsibility of the APN is to the client served, whether

the client is an individual, family, group, or community. Since health cannot be separated

45

from its environment as indicated in this study, it is essential that APNs become involved
in all aspects of planning for health to maximize the health potential of older adults. This
involvement needs to include attention to policy decisions and political action which can
be mediating factors between SES and health outcomes. Policy affects the broader aspect
of environment, the socioeconomic conditions of homes and communities, as well as the
health care delivery environment. Public policy decisions regarding firnding for health and
social programs are critical in providing resources and opportunities for low income
people. Health care providers have tremendous potential power to influence decision
making. The APN needs to get to know politically influential people in the community.
Legislators are influenced by the information they receive and the source of that
information. It is the responsibility of the APN to become informed. As well-informed,
empowered professionals, APNs play a significant role in supporting appropriate
legislative initiatives that promote and protect the health of the public and ensuring that
preventative health care for low SES groups is given priority. For the most part; however,
nurses do not envision themselves as agents for change in state and national politics.
Therefore, unquestionably, preparation for this role must begin during nursing education
and must be supported by professional nursing organizations and agencies.

Successful intervention to reduce the prevalence of behavioral risk factors for
stroke associated with lower SES will need to be broad based, addressing not only specific
risk factors but also societal conditions that lead to the adoption and maintenance of high
risk behaviors. The APN in the educator role needs to provide health education that
facilitates an individual’s ability to improve his or her personal living conditions, to make

informed decisions about personal, family, and community health practices and to utilize

46

health services appropriately. Often health care education programs are organized by
middle SES health care providers based on their own values and attitudes. Even though
the programs may be implemented with the lower SES in mind, the special difliculties and
differences in how the poor perceive the health care system frequently are not understood
or considered. Thus, APNs can solicit the views of the targeted audience (lower SES) and
include them in planning and implementation of educational programs. The APN needs to
explore new and appropriate techniques and methods to deliver more effective messages h
regarding healthy lifestyle behaviors found to prevent acute and chronic disease, decrease h

disability, and enhance wellness. It is not merely information distribution, an activity used

 

to increase awareness; rather, it involves guiding persons through stages of problem i
solving and decisions making. Finally, education should be developed to help health care
providers understand the magnitude of the problem of the SES-high risk behavior
relationship and the factors that underlie it.

Finding that a variety of unhealthy behaviors chrster not only in the lower, but in
the higher SES group, indicates the need for behavioral modification. As a change agent,
the APN has the opportunity and the means to modify behavior. Counseling,
reinforcement, reminders, and community programs are examples of some efi‘ective
strategies for bringing about positive alterations in an individual’s health behavior.

As a researcher, the APN can encourage families to participate in research studies
in an efl‘ort to firel the progress of future research and to promote positive health
outcomes. All professionals who work in primary care and care for older adults should
spread the word about the need to consider participating in research studies. APNs

because of there close proximity with patients are in an ideal position for raising the issue

47

and encouragm' g and supporting interested clients and their family members.
8 I . E 'E l B I

Research that would go beyond merely demonstrating associations with social
status is needed. There should be a focus on understanding the behavioral, social,
biological, and physiological mediators that link SES and high risk behaviors.
Identification and delineation of the mechanisms and processes that link SES to
psychosocial factors is necessary. Attention should be paid to aspects of study design and
measurement that are critical to determining whether the association is independent of
other factors.

There is a considerable need to better understand the links between economic
policy, health care coverage, unemployment, and other economic phenomena and the
prevention and prevalence of high risk behaviors. Measures of SES should be included in
all research on high risk behaviors including cross sectional and longitudinal studies. The
data should be analyzed categorically by age. It would be useful to know if some
unhealthy behaviors are clustered into certain age groups. Findings may suggest that there
are important stages within adulthood for attempting to alter health related behaviors.

Secondary analysis of existing data sets that contain potential information of SES
and high risk behaviors for stroke should be encouraged. Further attention should be
devoted to the effect of SES on risk behaviors throughout the life span with a particular
focus on clarifying the relative role of SES at various life stages. There is a need to focus
on psychosocial and personality factors associated with different SES groups and their
relations with high risk behaviors. Moreover, chronic stress associated with work and

other daily aspects of life and its role in high risk behaviors needs firrther investigation.

48

Perhaps pilot studies might be undertaken to determine whether modifications of SES are
efl‘ective for prevention of high risk behaviors. This may be an arduous task to complete
but it could potentially yield significant findings. More observational studies of health
practices in relation to stroke risk and intervention studies aimed at modifying behavior
related to stroke risk are needed. Lastly, attempts should be made to understand the
protective factors that enable some people to stay healthy by avoiding high risk behaviors
despite an increased risk due to lower SES.
Summary

The evidence in support of the importance of SES to health status has been
documented through several studies which have found a direct relationship between
individual’s SES and their health. Results of this study indicate that SES afl‘ects health
behaviors in relatively important ways but this depends on the measure used for SES and
the specific high risk behavior. The findings in this study, in part support the existing
literature and the hypothesis that higher prevalence of behavioral risk factors is associated
with lower income and lower educational attainment. Health care providers in primary
care are in an ideal position to effect change in individuals’ behavior by modifying
environmental risk factors-income and education—to ultimately promote positive health

outcomes.

49

APPENDIX

50

APPENDIX A

UCRIHS Approval Letter

51

 

quuumuwufim|

MtNmmSutUMnmm
246 Mammal Building
haanuJflmum
unmade

517355-21“)
FAX:517/432-1171

Imumhwflwflflufl
DashWUuwnmmfl
fiaunzhknn

stauflmwwafin
auwmunnflnmumr

    
  
  

MICHIGAN STATE

U is I \l E ii S I 1‘ Y

 

April 29, 1998

TO: Sharon King _
A110 Life Cienees Bldg.
College of NurSing

RE: IRE“: 98-295
TITLE: THE RELATIONSHIP BETWEEN SES AND PREVALENCE OF
BEHAVIORAL RISK FACTORS FOR STROKE AMONG OLDER
ADULTS
REVISION REQUESTED: N/A
CATEGORY: l-E
APPROVAL DATE: 04/28/98

The University Committee on Research Involving Human Subjects'(UCRIHsl
review of this project is complete. I am pleased to adVise that the
rights and welfare of the human subjects appear to be adequately
rotected and methods to obtain informed consent are apprepriate.
Egrefore, the UCRIHS approved this prOJect and any reVisions listed
a ve.

RENEHAL: UCRIHS approval is valid for one calendar year, beginning with
the approval date shown above. Investigators planning to
continue a project be and ene year must use the green renewal
form (enclosed with t e original agproval letter or when a
project is renewed) to seek u te certification. There is a
maXimum of four.such expedite renewals ssible. Investigators
wishing to continue a prOJect beyond the time need to submit it
again or complete reView.

REVISIONS: UCRIHS must review an changes in rocedures involving human
subjects, rior to in tiation of t e change. If this is done at
the time o renewal, please use the reen renewal form. To
revise an approved protocol at an 0 her time during the year
send your written request to the CRIHS Chair, requesting revised
approval and referencing the project's IRB I and title. Include
in our request a description of the change and any revised
ins ruments, consent forms or advertisements that are applicable.

PROBLEMS]

CHANGES: Should either of the followin arise during the course of the
work, investigators must noti UCRIHS promptly: (1) roblems
(unexpected side effects comp aints. e c.) involving uman
eubjects or (2) changes the research environment or new
information indicating greater risk to the human sub ects than
existed when the protocol was previously reviewed an approved.

If we can be of any future help, lease do not hesitate to contact us
at (517)355-2180 or FAX (517l4 2- 171.

sincerely, [Li

id 3. Wright, Ph :
CRIBS Chair

DEW:bed

cc: Trudy E. Day

52

BIBLIOGRAPHY

53

BIBLIOGRAPHY

American Heart Association, (1988). 81W. Dallas: American Heart Association.

Casper, M. ,ng, W. ,Strogatz, D. ,Davis, C. E., & Tyroler, H. A (1992).
Antihypertensive treatment and U. S. trends in stroke mortality 1962 to 1980

Amenmlcumalnfhrblicflealfldl 1600-1606

Cassel, J. (1976). The contribution of the social environment to host resistance. Ammcan
InumalnfEnideminlmJM, 107-123

Cockerman, W. ,Lueschen, G., Kunz, G. & Spaeth, J. L. (1986). Social stratification and
selfmanagemem of health WW, 1-14

Colditz, G. A, Bonita, R, & Starnpfer, M. J. (1988). Cigarette smoking and risk of stroke
in middle-aged women .NenLEnglandJnumalnfMedrmneallfi. 937-941

Community Health Assessment. (1995) WW

I L'II'HIU 1’31 .Hlllll! l.‘..Ia I "'1'. Id 0 .||.\

Wanna, Ann Arbor: Information Transfer Systems, Inc.

Converse, J..,M &Presser, S. (1986). - ' . '
Questinnnaire Newbury Park, California. Sage.

Donahue, RP. ,Orchard, T. J., Kuller, L. H., & Drash, AL. (1985). Lipids and
lipoproteins rn a young adult population: the Beaver county lipid study. American,
InnmalnffinidemialmJZZ, 458-467

Dryer, AR, Starnler, J. Shekelle, R. B., & Schoenberger, J. (1976). The relationship of
education to blood pressure: Findings on 40,000 employed Chicagoans.

CW, 987-992

Dunbabin, D.W., & Sandercock, PG. (1990). Preventing stroke by the modification of
risk factors. W02), Suppl. IV: 36-39.

 

Dyken, M.L., Wolf, P.A., & Barnett, H.J.M. (1984). Risk factors in stroke: A statement
for physicians by the subcommittee on risk factors in older women. W
1 105-11 13.

Feinstein, J .S. (1993). The relationship between socioeconomic status and health: A
review of the literature. W0), 279-321.

54

Feldman, J. J. Makuc, D. M., Kleinman, J. C, & Coroni-Huntley, J. (1989) National

trends In educational difl‘erentials 1n mortality. Amencanlmrmalnffimdemmlma
122, 919-933.

Ford, E. S. ,Merritt, RK., Health, G. W. ,Powell, K. E. ,Washburn, RA, Kriska, A, &
Haile, G. (1991). Physical activity behaviors in lower and higher socioeconomic

status populations. AmedeanlmnnalnffipidemialoqulUZ), 1246-1256.

Gortmaker, S.L., Must, A, Perrin, J.M., Sobel, A.M., & Dietz, W.H. (1993). Social
and economic consequences of overweight in adolescence and young adulthood.

NearEnghndloumalnfMedicinelZQ 1008-1012-

Haheim, L., Holrne, 1., Hjermann, 1., & Leren, P. (1993). Risk factors for stroke incidence
and mortality: A 12-year follow-up of the 0510 study. W10), 1484-1489.

Helrnert, U., Herman, B., Joeckel, K.H., Greiser, E., & Madans, J. (1989). Social class
and risk factors for coronary heart disease in the Federal Republic of Germany:
Results of the baseline survey of the German cardiovascular prevention study.

WM 37-42

Hillbom, M. & Kaste, M. (1982). Alcohol intoxication: A risk factor for primary
subarachnoid hemorrhage. W, 706-709.

Holrne, 1., Helgeland, A, Hjermann, 1., Leren, P., & Lund-Larsen, PG. (1981).
Physical activity at work and at leisure in relation to coronary risk factors and
social class: A 4-year mortality follow-up. 209, 277-283.

Iriberren, C., Russell, R. V. ,,Luepker R. ,McGovern, P. G. ,Arnett, D. K. & Blackburn, H.
(1997). Twelve-Year trends in Cardiovascular disease risk factors in the Minnesota

heart survey AmhiyeslnternalMedicineJfl, 873-881

Iso, H, Jacobs, D. R, & Wentworth, D. (1989). Serum cholesterol levels and six-year
mortality fi'om stroke in 350, 977 men screened for the multiple risk factor

intervention trial WWW, 904-910

Jacobs, D. R, McGovern, P. G. & Blackburn, H. (1992). The U. S. declines in stroke
mortality. What does ecological analysis tell us? WM
W,1596-1599.

Jackson, C. ,Jatulis, D. E. ,& Fortrnarr, S. P. (1992). The behavioral risk factor survey and
the Stanford five-city project survey. A comparison of cardiovascular risk behavior

estimates W6) 412-416

Jacobsen, B. K. ,& Thelle, D. S. (1988). Risk factors for coronary heart disease and level of
education Americanloumalnffipidemiologx, 923-932

55

Jefl‘ery, RW. & French, S. A (1996). Socioeconomic status and weight control practices

among 20 to 45 year old women Ammlamflnfhrhhmmhhm?) 1005-
1010

Kaplan, G.A, & Keil, J.E. (1993). Socioeconomic factors and cardiovascular disease: A
review ofthe literature. WM), 1973-1998.

Keil, J. E., Sutheriand, S. E. ,Knapp, R G, & Tyroler, H A (1992) Does equal
socioeconomic status in black and white men equal risk of mortality? Ameriean,

W2, 1133- 1136

Keil, J. E ,,Tyroler H. A, Sandifer, S. H., & Boyle, E. (1977). Hypertension: Efl'ects of
social class and racial admixture the results of a cohort study m the black

pontilation of Charleston, South Carolina Amefimnloumalnfhrhfieflealthfl,
634-639

Keil, J.E., Sandifer, S. H., Loadholt, C. B., & Boyle, E. (1981). Skin colorand education
efi‘ects on blood pressure. WWW], 532-534.

Kelly-Hayes, M. (1991). A preventative approach to stroke. Wmeeeflflmh
W4). 931-942.

Kleg, M. J., & Whelton, P. K, (1987). Risk of stroke in male cigarette smokers. Neg
M. 628-634

Kessler, R C. (1979). Stress, social status, and psychological distress. Jenmalefflmlth,
Winona), 259-272

Liberatos, P. ,Link, B. G., & Kelsey, J. L. (1988). The measurement of social class in
epidemiology Epidemialesielleumsfl, 87-121

Matcher, D. ,McCrory, D. ,Bamett, H ,& Feussner, J. (1994). Medical treatment for
stroke Prevention Ammmfiofleamfflhimmlm 1) 41- 53

Matthews K. A, Kelsey, S. F. ,Meilalrn, E. N., Kuller, L H. & W'mg, RR (1989).
Educational attainment and behavioral and biological risk factors for coronary

heart disease In middle aged women Amefieanlmrmalnffipidemiolmnflo),
1 132-1 144.

Michigan Department of Public Health (1993). W Michigan
Department of Health.

Mirand, AL. ,& Welte, J. W. (1996). Alcohol consumption among the elderly m a general

POPuletion, Ene County New York Ameneanlaumalnfhrhheflealthfiflfl
978-984.

56

National Center for Health Statistics (1990). Henlthflnitedfitnes. Publ No. (PHS) 91-
1232. Hyattsville, MD:US Government Printing Ofice.

Nelson, M A (1994) Economic impoverishment as a health risk: Methodologic and
conceptual issues W96) 1 12

Oken, B. ,Hartz, A, Giefer, E. ,& Rimm, AA (1977). Relation between socioeconomic
status and obesity changes in 9046 women. BREW 447-453.

Palomaki, H., & Kaste, M. (1993). Regular light-to-moderate intake of alcohol and the
risk of ischemic stroke: Is there a beneficial effect? W12), 1828-1832.

Peszencker, BL. (1984). The poor". A population at risk. WW,
237-249).

Pierce, J. P., Fiore, M. C. ,Novotny, T. E. ,Hatziandreu E. J., & Davis, RM. (1989).
Trends in cigarette smoking in the United States: Projections to the year 2000.,
WWW, 61-65.

Polit, D.F., & Hungler, B.P. (1995). t w 1.‘ ' :
Philadelphia: J .B. Lippincott.

 

Remington, PL, Smith, M.Y., Williamson, D.F., Anda, RF., Gentry, E.M., & Hogelin,
GO. (1988). Design, characteristics, and usefiilness of state-based behavioral risk
factor surveillance: 1981-87. WM), 366-375.

Roget, E., Sorlie, P. D. ,Johnson, N. J., & Scmitt, (1992) A mortality study of 1. 3 million
persons by demographic, social, and economic factors: 1979-1985 follow-up.
WW NIH Publication No.92-3297,1-5.

Salonen, J. T. (1982). Socioeconomic status and risk of cancer, cerebral stroke, and death
due to coronary heart disease and any disease: A longitudinal study in eastern

Finland. Iaumalnffimdemiolosicalflnmmunityflmlthilfi, 294-297

Salonen, J. T., Slater, J. S. ,,Tuornilehto J. ,Rauramaa, R (1988). leisure time and
occupational physical activity: Risk of death from ischemic heart disease.

Amefimlmrmalnffipidmiolosyilu 87-94

Shea, S. Stein, A D., Lantiqua, R, & Basch, C E. (1991). Reliability of the behavioral

risk factor survey in a triethnic population. WWW
131(5) 489-500.

Siegel, D., Kuller, L., Lazarus, N.B., Black, D., Feigal, D., Hughes, G., Schoenberger,
J. A, & Hulley, S. B (1987). Predictors of cardiovascular events and mortality in
the systolic hypertension in the elderly program pilot project. Ameneanleumalef
W, 385-399

57

Slater, C. ,& Carlton, B. (1985). Behavior, lifestyle, and socioeconomic variables as
determinants of health status: Implications for health policy development.

AmedmloumalnffirsxenfixeMedieineJfi), 25-33

Stein, A. D. ,Lederman, RI, & Shea, S. (1993) The behavioral risk factor surveillance
system questionnaire: Its reliability in a statewide sample. Amerieanlenmalef
WU” 1763-1772

Uphold, CR. & Graham. M-V- (1994). WWW).
Gainesville, Florida: Barmarrae Books.

US. Department of Health and Human Services, Public Health Service. Henlthyfieenle,
20%). National Health Promotion and Disease Prevention Objectives, Publ. No.
(PHS) 91-50213. Washington, DC Government Printing Office, 1990.

U. S. Preventive Services Task Force (1996). Screening for asymptomatic carotid artery

stenosis WWW Baltimore:
Williams & Wilkins.

Venters, M. H. (1989). F amrly-onented prevention of cardiovascular disease: A social
epidemiological approach. W“) 309-314

Wannarnethee, S.G., & Shaper, AG. (1996). Patterns of alcohol intake and risk of stroke
in middle-aged British men. Wfi), 1033-1039.

Williams, D. R. (1990) Socioeconomic difl‘erentials rn health: A review and redirection.
W0) 81-99

Winkleby, M. A Fortman, S. P., & Barrett, D. C. (1990) Social class disparities in risk
factors for disease: Eight-year prevalence patterns by level of education.
WW1) 1- 12

kaleby, M., Jatulis, D. ,Frank, E. ,& Fortman, S. (1992). Socioeconomic status and
health: How education, income, and occupation contribute to risk factors for

cardiovascular disease .Ameficanlaumalnfhrblieflealtuzw) 816-820

Wister, A. V. (1996). The effects of socioeconomic status on exercise and smoking.

WWW 467-488

Wolf, P.A, D’Agostino, R.B., Belanger, AJ., & Kannel, W.B. (1991). Probability of
stroke: A risk profile fi'om the Framingham study. W0), 312-318.

Wolf, RA (1990). An overview of the epidemiology of stroke. W102), Suppl H:
4-6.

58

"llllllllllllilli