The relationship between family-marital functioning and chronic illness

Material Information

The relationship between family-marital functioning and chronic illness
Atkins, Howard Gray, 1945-
Copyright Date:
Physical Description:
xi, 121 leaves : ; 28cm.


Subjects / Keywords:
Chronic conditions ( jstor )
Control groups ( jstor )
Correlations ( jstor )
Diseases ( jstor )
Hospitals ( jstor )
Immediacy ( jstor )
Life events ( jstor )
Psychological stress ( jstor )
Questionnaires ( jstor )
Rites of passage ( jstor )
Chronic diseases -- Psychological aspects ( lcsh )
Diseases -- Causes and theories of causation ( lcsh )
Dissertations, Academic -- Psychology -- UF ( lcsh )
Psychology thesis Ph. D ( lcsh )
bibliography ( marcgt )
non-fiction ( marcgt )


Thesis--University of Florida.
Bibliography: leaves 117-120.
Additional Physical Form:
Also available on World Wide Web
General Note:
General Note:
Statement of Responsibility:
by H. Gray Atkins, Jr.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
000164346 ( alephbibnum )
02784534 ( oclc )
AAT0709 ( notis )


This item has the following downloads:

Full Text



H. Gray Atkins, Jr.

A Dissertation Presented to the Graduate Council of
The University of Florida
In Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy



To Dorothy and Howard Atkins


I want to express my heartfelt appreciation to my

wife, Sally, who endured the day-to-day ordeal of this

dissertation for the past three years. During that time

she remained giving of her support and encouragement, even

during the worst of times. Her consistent belief in my

capabilities was a source of great strength to me and add-

ed a very meaningful dimension to our relationship.

I want to thank Dr. Ben Barger for always being

there when I needed him and for sharing his many personal

and professional resources with me. The enrichment I de-

rived from our association will undoubtedly transcend the

boundaries of our limited time together. Thanks also to

Dr. Mark Goldstein, who offered his remarkable creativity

and enthusiasm in addition to much-needed material assist-

ance, and to Dr. William C. Thomas, Jr., who made the re-

sources of the Gainesville, Florida, Veterans Administration

Hospital available to me.

Further acknowledgment goes to Dr. Gerald Stein for

his timely medical consultation and to David Smolen for

executing the necessary computer operations. All computer

runs were made at the Northeast Florida Regional Data Cen-

ter, Gainesville, Florida. Finally, I want to give special

thanks to Carol Pait and Bill Gasparini for invaluable

assistance in contacting subjects and collecting data.







A Conceptualization of Stress 2
Early Psychophysiological Studies 3
Personality Versus Environmental
Approaches 8
Life Crises Related to Specific
Disease Entities 10
Life Change Related to General
Incidence of Stress 16
Life Change and Severity/
Immediacy of Illness 30
The Present Investigation 34

Data Collection Instruments 39
The Sample 49
Procedural Format 53
Operational Specifications 56
Data Analysis 57

Demographic Comparisons Between
Groups 61
Demographic Characteristics of
the Samples 64
Comparisons Associated with
Hypotheses 69

Comparison with Previous Life
Change Studies 80
Family-Marital Variables in
Relationship to Illness 81

CONTENTS (Continued)


IV Relationship Between Other Life
Change Variables and Illness 86
Other Findings of Interest 91
Interrelationships Among Illness
Variables 96
Conclusions 97






1 Between-Group Comparisons of Age and
Education 63

2 A Comparison of Health History Totals 65

3 A Comparison of Marital History Information 65

4 A Comparison of Employment Information 67

5 A Classification of Subjects According to
Diagnostic Category 67

6 Correlations Between SRE Life Change Category
Scores and SIRS Values 70

7 Correlation Between Locke-Wallace SMAT/
Conventionality Scores and SIRS Values 70

8 A Comparison of SRE Life Change Category
Scores 72

9 A Comparison of Locke-Wallace SMAT/Conven-
tionality Scores 72

10 Correlations Between SRE Life Change Category
Scores and Immediacy of Illness Onset Data 73

11 Correlation Between Locke-Wallace SMAT/
Conventionality Scores and Immediacy of
Illness Onset Data 73

12 Correlations Between SRE Life Change Category
Scores and Locke-Wallace SMAT/Convention-
ality Scores 75

13 Correlations Between Age and Locke-Wallace
SMAT/Conventionality Scores 76

14 Correlations Between SIRS Scores and Pre-
morbid LCU Totals 76

15 Correlations Between Health History Totals
and SRE/SMAT Scores 77

16 Point-Biserial Correlations Between SRE/SMAT
Variables and Hospitalization 77


LIST OF TABLES (Continued)


17 Correlations Between Subjects' Age and
Other Variables of Interest 78

18 Intercorrelations Among Three Illness
Variables 78


Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy



H. Gray Atkins, Jr.

August, 1975

Chairman: Ben Barger
Major Department: Psychology

Research has shown that stressful environmental con-

ditions are often associated with subsequent alterations in

human.physiological functioning. More recently, interrelat-

edness between incidence of life change events (one form of

environmental stress) and subsequent onset of illness has

been demonstrated. The present study sought to determine

relationships between the incidence of life change within

discrete areas of living (i.e., family-marital, personal-

environmental, and occupational-financial) and measures of

subsequent chronic illness episodes (i.e., severity of ill-

ness and immediacy of onset). The focus of the study was

upon family and marital factors in relation to illness;

therefore, a measure of subjects' perceived marital adjust-

ment was included.

It was predicted that a high incidence of life change

within the family-marital area during two years prior to

illness would be directly related to the severity of the ill-

ness and inversely related to the immediacy of illness on-

set; likewise, that a reported level of marital adjustment

among the chronically ill would be inversely related to ill-

ness severity and directly related to immediacy of onset.

It was also predicted that both reported marital adjustment

levels and incidence of family-marital life change would re-

flect more disruption within a chronically ill group in rela-

tion to a comparable group of healthy individuals.

Fifty-one chronically ill inpatients comprised the

hospitalized group; 36 healthy participants formed the con-

trol group. The Schedule of Recent Experiences was used to

assess the incidence of life change; the Locke-Wallace Short

Marital Adjustment Test (SMAT), to measure subjects' per-

ceived marital adjustment.

It was found that neither incidence of family-marital

life change nor marital adjustment was related to severity

of illness. However, family-marital life change was related

to immediacy of illness onset. Also, a higher incidence of

family-marital change as well as lower levels of reported

marital adjustment were found among members of the hospital-

ized group than among members of the control group. The

incidence of personal-environmental life change was found

to be the most significant life change factor in the study,

being related to both severity and immediacy of illness as

well as occurring at a higher rate within the hospitalized

group than within the control group. Incidence of life

change was found to decrease as a function of increased age

among healthy participants, whereas no such trend toward

life change stabilization was found among chronically ill



The focus of medical research in recent years has

increasingly centered upon the interplay between social-

psychological factors and illness. A notable growth in the

psychosomatic literature reflects increasing collaboration

between physicians and behavioral scientists in developing

a more comprehensive view of the illness process. A cen-

tral concept in this expanded inquiry into illness phenomena

has been that of psychophysiological stress within the con-

text of ongoing life change events. In the present study,

one particular source of psychophysiological stress has been

selected for emphasis.

Reviews and discussion presented in Chapter I will

focus upon the following research issues: (1) the relation-

ship between stress and alterations in physiological func-

tioning; (2) subsequent relationships between psychophysio-

logical stress and illness; (3) the association between life

crises and illness; (4) the delineation of life change, as

one dimension of the life crisis concept, and its relation-

ship to illness; and (5) relationships between life change

events and specific aspects of illness. Following a review

of the literature, the rationale for the present study, with

its focus on the association between family-marital life

change and illness, will be presented. The specification

of hypothesized outcomes related to these issues will con-

clude this chapter.

A Conceptualization of Stress

The relationship of stress to illness and physiolog-

ical dysfunction has been a topic of interest to researchers

for several decades. Much early research in this area

originated in the laboratories of Harold G. Wolff at Cornell

University and provided convincing arguments that "stress-

ful" life events, by evoking psychophysiologic reactions,

played an important causative role in the natural history

of many diseases (Holmes & Rahe, 1967).

Unfortunately, as important as the concept of stress

is to the behavioral sciences, it has successfully eluded

precise definition. For example, Dodge and Martin (1970)

listed intellectual stress, emotional stress, physiological

stress, psychological stress, and social stress as compris-

ing the nosology of stress. They also noted that the locus

of stress may reside in some type of feeling or emotion in-

ternal in origin or in some external force or situation

eliciting an apprehensive response from an individual.

For purposes of the present study, stress has been

employed as a descriptive term which defines specific physio-

logical reactions, as opposed to antecedent causes of these

reactions, a perspective consistent with usage within the

medical and behavioral sciences. Stress is defined as "the

state of apprehension and tension with associated physiolog-

ical changes that accompanies the individual's attempt to

adapt to any stimulus condition" (Dodge & Martin, 1970, p.

32). Having the virtue of succinctness, this conceptualiza-

tion of stress also emphasizes the interrelationship between

sociological, psychological, and phyiological precipitants,

all of which play a distinctive part in the stress process.

This definition of stress is in general accord with studies

to be revied and has been incorporated into the conceptual

foundation of the present study.

Early Psychophysiological Studies

Most of the background research which sought to link

stress to illness was conducted in the 1930s and 1940s and

tended to be psychophysiological in nature; that is, the

immediate effect of environmental events and psychological

sequelae (emotional arousal) upon various physiological

measurements constituted the methodology of choice. While

physiological measurements varied according to medical syn-

dromes of interest, the environmental events selected to

elicit physiological alterations were of two types: (1)

experimentally induced events conducted in controlled labo-

ratory settings and (2) naturally occurring events arising

spontaneously in a subject's environment. Representative

studies from both categories will be reported here.

Controlled Laboratory Studies

As early as 1939, researchers were investigating an

apparent relationship between stressful laboratory situations

and physiological alterations or disease. Mittelmann and

Wolff (1939), for example, studied the relationship between

stressful affective states and skin temperature changes,

and found that drops in skin temperature during stressful

periods occurred in most subjects. These temperature drops

were attributed to an interplay between emotional stress

and environmental temperature.

Several years later, these experimenters (Mittelman

& Wolff, 1942) extended their efforts to patients with pep-

tic ulcers and sought to ascertain changes in gastroduodenal

function that might be associated with various emotional

situations. During stressful interview situations, gastric

secretion and stomach motility increased in all subjects,

with peptic ulcer patients experiencing an intensification

of symptoms. In subsequent supportive interviews, all sub-

jects experienced a decrease in secretions and motility.

Observed alterations in gastroduodenal functioning were

attributed to the presence or absence of emotion-arousing

situations and were considered likely forerunners of tissue


A third experiment (Mittelmann & Wolff, 1943) invaded

the sanctity of psychoanalytic interviews in order to explore

the relationship between a flow of psychological events


observed during interview and the concomitant bodily states

of five patients. Skin temperature changes were monitored

in five analysands during some 83 interviews, and various

emotional states in the patients were simultaneously noted.

A consistent relationship between finger temperature and a

large variety of emotional reactions was reported.

Grace and Graham (1952) undertook to document an

interplay between specific sets of attitudes and 12 physio-

logical symptoms or diseases, e.g., urticaria, cold hands.

On the basis of interview information obtained from subjects,

the following conclusion was offered:

It was found that each of these conditions was
associated with a particular, completely con-
scious attitude toward the precipitating situa-
tion. There were, in other words, physiological
changes specific to each attitude.
(Grace & Graham, 1952, p. 250)

This landmark work provided the impetus for a research move-

ment based on the "specificity of attitude" hypothesis.

Stern, Winokur, Graham, and Graham (1961), for exam-

ple, made a further inquiry into attitude specificity.

Healthy subjects were told under hypnosis to assume one or

another of three attitudes which had been found in previous

studies to be associated with hives, Raynaud's disease, and

essential hypertension. Three of five predicted physiologi-

cal measurement differences were found significant. Differ-

ent attitudes were found to produce different but predicta-

ble physiological changes which parallel in healthy subjects

the pathological symptoms of subjects with the above three

medical conditions.

In another study of attitude specificity, Graham,

Stern, and Winokur (1958) recruited subjects to participate

in hypnotic sessions during which two different attitudes

were suggested, i.e., attitudes associated with Raynaud's

disease and with hives. A simultaneous measurement of skin

temperature response was made. The experimental attitudes,

when suggested under hypnosis, were found to produce the

predicted skin temperature responses.

More precisely designed experiments were undertaken

by Ax (1953) and Schachter (1957). Both researchers exposed

subjects to apparently accidental situations intended to

arouse fear or anger. The fear-arousing situation resulted

from an apparent breakdown of the laboratory apparatus that

threatened harm to the subject; the anger-inducing situation

was created by insolent behavior on the part of a collabora-

tor posing as a laboratory technician. There were, on the

whole, differences in physiologic responses to these two


Natural Environment Studies

Another series of studies regarding physiological

alterations in response to stress focused on stressful situ-

ations found to occur in the natural environment. Studies

using experimental manipulations of "real life" situations

have been few, undoubtedly because of inherent difficulties

in design. Essentially, such experiments have exposed the

subject to disturbing events as they actually occurred

rather than to abstractions of such events via hypnotic sug-

gestion, interview, etc.

Holmes, Goodell, Wolf,and Wolff (1950) attempted to

document the effect on nasal functioning of unpleasant af-

fect stemming from stressful life events. Much of the data

emanated from subjective reports of life experiences and a

subsequent interference by the experimenters. The data con-

cerned the affective reaction that accompanied such experi-

ences. Nasal dysfunction was represented as an attempt by

the body to shut out noxious stimuli which, in this research,

consisted primarily of unpleasant emotional states arising

from difficult life experiences.

Engel, Reichsman, and Segal (1956) conducted investi-

gations of gastric activity in the presence of different

environmental conditions. In observations of a baby girl

with a gastric fistula, it was found that marked changes in

gastric activity could be induced either by changing the

person interacting with the girl or by changing the inter-

active behavior. For example, when the girl was with a

familiar observer toward whom she showed obvious signs of

affection, her stomach secreted substantially more acid

than when she was in the presence of a stranger. Also,

Wertlake, Wilcox, Haley, and Peterson (1958) found that the

serum cholesterol level in medical students was higher on

examination days than on nonexamination days.

Wolff (1963) conducted a more in-depth study of life

situations, personality features, and the migraine symdrome.

A thorough compilation of life histories encompassing a

number of important developmental and personality variables

was conducted. Wolff concluded that stressful situations

in and of themselves did not necessarily evoke physiological

malfunctions. Instead, he contended, the psychological

makeup of the individual must be conducive to stress induc-

tion occurring in the context of difficult life situations.

Personality Versus Environmental Approaches

Two directions have been taken in research reported

in the literature. The first direction has been oriented

toward the role of personality structures in relation to

illness susceptibility; the other, toward the contribution

of environmental events to the onset of illness. In essence,

a choice has been offered between internal, largely intra-

psychic personality variables and external, environmental

events when approaching the study of stress and its relation-

ship to illness. In the present investigation, the latter

course of inquiry was selected as offering the most promise

for identifying clear-cut, quantifiable stress factors that

may be associated with illness.

Previous research efforts toward relating different

aspects of psychological stress to changes in health have

been at best equivocal. Many of the studies reported above

(e.g., Mittelmann & Wolff, 1939, 1942, 1943; Grace & Graham,

1952; Stern et al., 1961) were based upon extensive case-by-

case detailing of subjects' developmental histories and

psychological makeups. Apparently, the psychological vari-

ables related to stress are difficult to identify and quan-

tify. Indeed, all these studies are somewhat deficient in-

sofar as methodology is concerned. To wit, many of these

studies were based on anecdotal data or on case histories,

and with the exception of the study by Stern et al. (1961),

statistical analyses of the data are either nonexistent or

markedly inadequate.

The anecdotal approach to data collection in these

studies may well reflect the methodological zeitgeist of

the times, but this approach also reflects the difficulties

to be encountered in developing a valid, comprehensive formu-

lation concerning the psychological status of illness-prone

individuals. For example, Holmes et al. (1950) presented

extensive case histories and psychological status reports

in an attempt to link affective reactions and personality

types to nasal disorders. In this myriad of documentation,

some similarities in background and psychological makeup

did appear to exist among persons experiencing disturbances

in nasal functioning, but another, more important, conclu-

sion is inescapable: Among individuals experiencing nasal

symptomatology, each maintains a unique background of

deprivation, frustration, dissatisfaction, etc. Inasmuch as

each individual's psychophysiological system is unique in

many ways, the stress potential of a given affective state

may be understood only within the context of this system.

Hence, one individual's experience of anger and resentment

may culminate in acute hyperemia and hypersecretion in the

nasal passageways whereas a similar emotional response of

the same intensity in another individual may result in no

symptom formation.*

In short, the process whereby personality character-

istics interface with physical illness must seemingly be

investigated within the framework of each individual's psycho-

physiological system; however, this approach tends to dis-

courage the collection of data sufficient for definitive

statistical analyses to be conducted. Therefore, due to the

burdensome sampling procedures and the relative inaccessi-

bility of intrinsic psychological states and processes, the

personality approach to stress as related to illness was

discarded in favor of an approach which focuses upon environ-

mental events.

Life Crises Related to Specific Disease Entities

While the foregoing discussion has highlighted psycho-

logical research into stress and illness, a series of studies

*It is noteworthy that Thomas Holmes has abandoned his
personality orientation to illness and is currently a lead-
ing investigator of environmental factors related to illness.

having a somewhat different focus has emerged in the past 25

years. These studies have sought to relate the concept of

life crises, i.e., stress-inducing situations that persist

over time, to specific types of disease. These studies dif-

fer from the previously cited research in that they have

drawn heavily from psychosocial life histories and histori-

cal medical data.

The major portion of life crisis research began around

1950. During this time, Hans Selye advanced a well document-

ed theoretical link between stress and subsequent deteriora-

tion of body tissues, organs, and systems.

The General Adaptation Syndrome

Selye and associates (Selye, 1956; Selye & Fortier,

1949) advanced the General Adaptation Syndrome (GAS) in an

effort to provide a physiological account of the body's at-

tempts to cope with a stressful condition or event. As de-

tailed by Selye and Fortier (1949), the GAS was seen as a

three-stage model of biological mobilization consisting of

an alarm reaction stage, a stage of resistance, and a stage

of exhaustion. Selye maintained that this process centered

around activity of the pituitary-adrenal system which pro-

duces important alterations in the morphology and function

of the nervous system.

The pituitary-adrenal system was portrayed by Selye

as the main endocrine regulator of adaptive processes,

functioning as the mobilizer of bodily resources meeting the

threat of a stressor agent. He contended that if the process

of adaptation continued unabated, significant tissue damage

would ensue, in some cases leading to death. Selye's early

research focused primarily upon the effect of physical

trauma, e.g., anoxia, hemorrhage, burns, upon the organism;

however, he later advocated as well the potency of alarming

psychogenic stimuli, including "a game of tennis or even a

passionate kiss" (Selye, 1956, p. 53).

The advent of Selye's general adaptation theory ap-

pears to have given credibility as well as impetus to a

psychosomatic movement blossoming in the early 1950s. Re-

searchers were no longer cast in the role of speculators

regarding stress and its contribution to disease; therefore

they were able to inquire into sources of stress more dili-

gently, without the necessity of couching conclusions as

tentatively as their predecessors. The following studies

are representative of the research into relationships between

life crises and illness.

Life Crisis Studies

Research in life crisis and illness has varied with

respect to organ systems and areas of psychosocial function-

ing involved. For example, Holmes, Treuting, and Wolff

(1951) explored the relationship between life situations,

emotions, and nasal disease, finding that a life situation

engendering conflict and anxiety enhanced the magnitude and

intensity of ongoing nasal hyperfunction and heightened in-

dividual susceptibility to further nasal inflammation.

In a study of hospitalized male leukemia patients,

Greene (1954) found that, in all cases, both symptoms and

diagnosis of the disease occurred while the patient was

having to adjust to stressful life situations. Such stress,

according to Greene, was related primarily to loss or sepa-

ration from significant persons in the patients' lives.

In a similar study of female leukemia patients,

Greene, Young, and Swisher (1956) found that all patients

under study had experienced some type of loss, separation,

or threat of separation during four years prior to the on-

set of disease. "One of the multiple conditions determin-

ing development of lymphoma and leukemia in adults may be

separation from a key object or goal with ensuing depres-

sion" (Greene et al., 1956, p. 303).

Weiss, Dlin, Rollin, Fischer, and Bepler (1957) com-

pared patients with coronary occlusion to a matched control

group and found that nearly half of the coronary patients

experienced gradually mounting emotional stress over a period

of months or years prior to onset of the occlusion. No evi-

dence of stress buildup was observed in the control group.

Hawkins, Davies, and Holmes (1957) found that a group

of sanatorium employees who became ill with tuberculosis had

experienced "a concentration of disturbances, such as domestic

strife, residential and occupational changes, and personal

crises" during two years preceding the onset of disease.

These findings contrasted significantly with a control

group of healthy employees. Thus, stress-inducing environ-

mental conditions were conducive to lowered resistance ren-

dering the individual more susceptible to tubercular infec-


Greene and Miller (1958) investigated the relation-

ship between various types of loss, separation, or threats

of separation and onset of leukemia in children and adoles-

cents. Losses included the birth of a younger sibling,

changes of school or residence, and the death, separation,

or the threat thereof, of parents and grandparents. One or

more of these types of losses occurred in 94% of these

young patients, with half of the losses occurring during a

six-month period prior to onset of the disease.

A study of coronary heart disease by Russek and

Zohman (1958) compared coronary patients and a correspond-

ing control group to determine etiological factors that con-

tributed to the disease. In addition to hereditary and

dietary factors it was found that "severe emotional strain,"

of occupational origin, was present in 91% of the coronary

patients, compared with a 20% rate of incidence among con-

trol group members.

Another study of coronary dysfunction (Fischer, Dlin,

Winters, Hagner, and Weiss, 1962) revealed that gradually

mounting tension, acute emotional stress, and "conscious

stress" were present to a significant degree in coronary

occlusion. Three years prior to the onset of occlusion, an

unusually large number of stressful environmental events

were reported by most patients.

Jacobs, Spilken, and Norman (1969) conducted a study

of upper respiratory infection in college students. The

infected group and a healthy control group were given a

series of questionnaires to determine, among other things,

the incidence rate of distressing life changes during the

year prior to infection. The infected group had experienced

a significantly larger number of personal crises or failures

during the year prior to illness than had the control group.

A further analysis revealed that 63% of crisis events re-

ported by the infected group occurred during a two-month in-

terval prior to the onset of illness.

A recent and important study by Dohrenwend (1973) of

stressful life events sought to determine the effect of the

desirability of a given event upon the "stressfulness" of

that event. A psychological symptom checklist as well as

a measure of life change events experienced during the pre-

vious year were administered to a cross-section of respond-

ents. Life change responses were scored in two ways. First,

events were assigned dichotomous weights according to rela-

tive desirability or undesirability. The other scoring pro-

cedure consisted of attaching a magnitude of readjustment

value derived by Holmes and Rahe (1967) to each event. Both

measures of life change were correlated with psychological

symptom formation reported by subjects. The results revealed

that while both measures of life change were significantly

related to the occurrence of symptoms, readjustment values

were more highly correlated with symptom formation than were

desirability weights. "Change, rather than undesirability,

is the characteristic of life events that should be measured

for the more accurate assessment of their stressfulness"

(Dohrenwend, 1973, p. 174).

Life Change Related to General Incidence of Stress

Retrospective Studies of Life Change and Illness Onset

As demonstrated above, a considerable amount of re-

search has been undertaken to demonstrate the relationship

between life events and onset of specific diseases. How-

ever, more recent studies have shown that life change is

related to a wide variety of illnesses and other physiolog-

ical dysfunctions. For example, Hinkle, Christenson, Kane,

Ostfeld, Thetford, and Wolff (1958) reported that, in gen-

eral, persons who find their life situations least satisfac-

tory and most demanding have the highest incidence of ill-

ness. A group of 20 persons of Chinese origin were studied,

10 having had a high illness rate throughout their lifetimes

and 10, a low rate. Without knowing the illness rates, a

psychiatrist was able to distinguish members of each group

significantly better than chance on the basis of whether

they saw childhood environment as satisfactory (the low

illness group) or unsatisfactory (the high illness group).

Individual illnesses were not distributed at random during

the lifetime of subjects, but often occurred in clusters.

These clusters of illness emerged during times when subjects

reported difficulty in adapting to stressful life situations.

Rahe, Meyer, Smith, Kjaer, and Holmes (1964) sought

to demonstrate that many, if not all, diseases have their

onset during such clusters of social stress. Seven patient

samples, five of whom manifested distinct medical syndromes,

and two control groups were studied. The five medical enti-

ties included tuberculosis, cardiac disorders, hernia, skin

disease, and pregnancy. The instrument used to gather in-

formation concerning each individual's history of life change

was the Schedule of Recent Experiences (SRE), devised by

Hawkins et al. (1957). The SRE is composed of 42 life

events "empirically derived" from the authors' clinical ex-

periences and represents a broad spectrum of subjects' re-

cent life changes. Representative events were drawn from

the areas of family constellation, marriage, occupation,

financial status, residence, group and peer relationships,

education, religion, and recreation. All items are equally

weighted with regard to the degree of associated stress.

In a separate section, major social readjustments are listed

according to year of occurrence over a 10-year period.

From a sample of tuberculosis sanatorium employees

who developed tuberculosis on the job, SRE data were com-

pared to an individually matched control sample of healthy

employees. "The temporal pattern of social stresses experi-

enced in the ten-year period prior to illness was the dif-

ferentiating feature between the two groups. The tubercu-

losis group showed a skewing of social stresses into the

final two of the ten premorbid years" (Rahe et al., 1964,

p. 42). The difference between the tuberculous group and

the control group with regard to amount of reported social

stress was significant beyond the .02 level. Similarly, life

change data from a sample of tuberculous outpatients and a

group of cardiac patients were compared to a control group

of similar, but healthy subjects and to one another. Both

disease groups demonstrated clustering of social stresses

in the final two years prior to onset of disease. The life

change differences between either patient sample or the re-

spective control groups were significant beyond the .05


Skin disease patients reported that between 25% and

67% of all changes in social status experienced in the 10

years prior to illness onset were encountered in the final

2 years. Data from patients with inguinal hernia indicated

that if these subjects were to experience changes of a per-

sonal nature in the 10 years preceding onset of symptoms,

the probability was two to four times as great that onset

would occur in the final two-year period than at any other

time. Unwed mothers also experienced a steady rise in fre-

quency of social stress up to the year of delivery. The

clustering of changes in social status as measured by the

SRE during the final two premorbid years was termed the

"psychosocial life crisis." No uniform method of present-

ing data was undertaken, however, and no statistical pro-

cedures were conducted with the skin disease, hernia, and

pregnancy groups.

Holmes and Rahe (1967) developed a rating scale

whereby not only the number and types of stressful life

events could be measured but also the magnitude of required

readjustment, i.e., the subjective impact, that each of

these events carried for the individual. This scale was

developed to introduce greater precision into the quantita-

tive study of the relationship between life change and phys-

ical illness. The sample was composed of 394 subjects, each

of whom completed a pencil-and-paper form of the Social Re-

adjustment Rating Questionnaire. The questionnaire included

the 42 life events listed on the SRE and an additional item.

Each subject was asked to rate these life events according

to relative degree of readjustment necessary.

To provide a reference point on the questionnaire,

the life event of "marriage" was assigned an arbitrary value

of 500, and each subject was asked to rate other events in

terms of required readjustment according to the degree of

departure from that anchor value. Results were converted

to a ratio scale in which each of the 43 mean scores were

reduced by a factor of .10. This procedure produced a scale

ranging from a mean value of 11 for a minor violation of

the law to a mean value of 100 for the death of a spouse.

All remaining studies reported here employed the life change

magnitudes in conjunction with the SRE.

Rahe and Holmes (1966a) conducted a pilot study of

the relationship between life crisis events and onset of

illness. Eighty-eight resident physicians completed the

SRE and a questionnaire asking for a list of all major

health changes by year of occurrence for the previous 10

years. Resident physicians were chosen as subjects because

of presume sophistication in matters of health and disease.

The items subscribed to in the SRE were summed for each year,

and total life change units (LCU) were derived and plotted

for each subject for the decade under study.

A total of 88 diseases or changes in health status

were reported by 96 subjects for a 10-year period prior to

the study. Of the 96 subjects, 89 (93%) reported that health

changes occurred within a two-year period following the oc-

currence of a life change cluster which totaled at least 150

LCU per subject/year. The term life crisis was defined as

"any clustering of life change events whose individual values

summed to 150 LCU or more in one year." Chi square was

employed to test the significance of the 93% association

between reported health changes and life change events, and

the association was found to be significant beyond the .001


A linear relationship was claimed between the magni-

tude of life crisis and the risk of health change. Three

ranges of scores were used to delineate mild (150-199 LCU),

moderate (200-299 LCU), and major (300+ LCU) life crises.

For subjects with a mild life crisis, 37% had an associated

health change. For those within the moderate range, 51%

reported an associated health change, with a 79% health

change occurrence rate found among those in the major life

crisis category.

The statistical analysis of data reported by Rahe

and Holmes is at best disappointing. A chi-square procedure

for frequency data was limited to defining the probability

of events occurring beyond expected (chance) occurrence.

No stronger measure of the magnitude of association between

reported life change events and subsequent changes in health

was attempted.

A study by Thurlow (1971) of 165 employees in an

Ontario brewery sought to demonstrate that an individual's

illness experience is related to life situation. A number

of tests and questionnaires, including the SRE, were admin-

istered to each individual in the sample, with illness data

obtained from company health service records. Information

was compiled for each of five years preceding data collec-

tion. A two-year followup was also conducted in which 111

of the original sample participated.

From the original 42 items of the SRE, 38 were select-

ed for use in the study. Of these 38 items, 20 were judged

to be objective, i.e., were externally verifiable (e.g.,

change in residence), while 17 were placed in a subjective

category, i.e., were judged by the subject to have occurred

(e.g., change of eating habits). An item concerning major

health change was scored in a separate category. A total

score derived from the sum of the objective and subjective

categories was also compiled. Six illness parameters for

the five-year observation period were intercorrelated, with

"number of illnesses" and "total days off" being selected to

represent relatively minor and relatively major illness

experiences, respectively.

In computing the correlation between SRE score cate-

gories and the two illness parameters described above,

Thurlow found that the objective category of SRE scores was

not significantly correlated with either retrospective (pre-

ceding five years) or prospective (following two years) ill-

ness ratings. The strongest relationship existed between

five-year total SRE subjective scores and number of illness-

es retrospectively and days off prospectively.

Thurlow then introduced a regression analysis, in

both retrospective and prospective directions, employing

the two illness parameters as dependent variables and a multi-

tude of questionnaire data (including the various components

of the SRE) as independent variables. In the prospective

regression analysis, neither of the SRE category scores ac-

counted for a significant amount of illness variance. How-

ever, in the retrospective analysis, the subjective category

of the SRE was found to be a significant predictor of both

illness parameters. On the basis of these findings, Thurlow


Correlations of these [two] illness parameters
with the questionnaire data suggested that the
subjects' social experience, as measured by the
SRE, was associated with relatively minor ill-
ness of preceding years and with relatively major
illness of subsequent years. Minor illness would
thus seem to be related to ongoing social change,
which may in turn affect future health in terms
of more serious illness (reflected in "number of
days off") .. .Furthermore, the subjective
impression of social change for the preceding
five-year period was more highly related to sub-
sequent illness than the impression of social
change for the preceding year. This suggests
that the "overall viewpoint" of the rater when
he completed the questionnaire is a more useful
predictor of illness than is a year-by-year
analysis of the changes experienced by him. This
interpretation is at variance with previously
cited experience in the use of the SRE question-
(Thurlow, 1971, pp. 84-85)

While Thurlow's conclusions may be of some heuristic value,

the validity and comprehensiveness of the illness parameters

used must be questioned. While minor and major illnesses

may be related statistically to "number of illnesses" and

"number of days off," respectively, this definition of illness

appears to be more in the interest of convenience in gather-

ing data than of accurate portrayal of subjects' true health


Rahe, McKean, and Arthur (1967) conducted a retrospec-

tive examination of 50 health records of Navy and Marine

Corps personnel who had received a discharge from the serv-

ice for health reasons. Mean yearly illness episodes were

calculated for each of the 50 subjects, and cluster years

of illness were defined in accordance with procedures estab-

lished by Hinkle and associates (Hinkle et al., 1958; Hinkle,

Redmont, Plummer, and Wolff, 1960). A slightly modified

form of the SRE, the Life Change Units Scale, was developed

to fit life change characteristics of the sample. The Life

Change Units Scale contains 41 different categories of life

change, ranging from 11 LCU for a minor infraction of the

law to 100 LCU for the death of a spouse.

The health record of each subject was inspected for

information about significant life changes which could be

fitted into one of the 41 life change categories. An import-

ant exception to this procedure was the exemption of all

life changes that appeared to be symptomatic of illness or

a result of illness; thus a spurious elevation of the cor-

relation between life change and illness was circumvented.

Three separate mean LCU totals were computed for sub-

jects under study: (1) the mean LCU value per year of active

duty for the entire sample; (2) the mean LCU total for the

year prior to a single illness episode or clustering of ill-

ness episodes of a minor nature; and (3) the mean LCU total

for the year prior to a single illness or clustering of ill-

ness episodes of greatest severity. Interestingly enough,

the mean LCU total for the year prior to both a minor and

major health change differed from the mean LCU value per

year beyond the .001 level of significance. This finding

is to be contrasted with Thurlow's (1971) conclusion that

no significant relationship between life change and major

illness episodes existed. Indeed, Rahe et al. (1967) found

that the mean LCU total for the year prior to a major ill-

ness was significantly greater than for the year prior to

a minor illness. However, an undisclosed proportion of ill-

ness episodes reported in this study were psychiatric in

nature. Objections to including this diagnostic category

will be discussed in a later section.

A recurring issue within the field of psychosomatic

research centers around the relationship between the illness

experience and various life events seen to occur shortly be-

fore and/or after the experience. A number of theoreticians

(e.g., Wolff, 1963; Weiss & English, 1957; Graham & Steven-

son, 1963) have debated whether life changes occurring around

the time of an illness operate as cause or effect of that

illness. Research by Rahe and Arthur (1968) addressed itself

to this issue with some interesting results. A sample of

2,900 naval officers and enlisted men completed the SRE and

a health questionnaire spanning the previous four years. LCU

and illness data were examined at six-month intervals for

the four-year period; each six-month interval in which an

illness experience was indicated was designated as an illness

period. Similarly, six-month periods free of health change

prior to or following an illness episode were designated as

pre- and postillness intervals.

As in other studies (Rahe et al., 1964, 1967; Hinkle

et al., 1958; Rahe & Holmes, 1966a), the magnitude of life

change was observed to increase significantly over six-month

intervals until the time of illness onset. Further analysis,

however, revealed a reversed trend of life change subsequent

to an illness episode, producing a nearly symmetrical pic-

ture of life change surrounding the illness period. "Life

changes resulting from illness experience are virtually equal

in timing and intensity to those life changes having a

causal influence on illness" (Rahe & Arthur, 1968, p. 344).

Prospective Studies of Life Change and Illness Onset

Several studies have been conducted in which life

change data have been used to predict subsequent changes in

health. One such prospective study of life change and ill-

ness onset was conducted by Rahe and Holmes (1966b). The

same resident physicians who had served in a previous study

(Rahe & Holmes, 1966a) were contacted for an update on their

health status during an eight-month interval following the

original study. A response was received from 84 residents,

and the new data were analyzed for associations with the

previously gathered life change data.

Thirty-two major health changes were reported during

the follow-up period; 31 of these changes followed a life

crisis of at least 150 LCU (97%). Subjects were divided

into high, moderate, or low risk groups based on the origi-

nal LCU scores, approximating the mild, moderate, and major

life crisis groups of the original study.

With the subjects at risk for eight months in the
present study, 49% of subjects with a major life
crisis experienced health changes. For those sub-
jects with a moderate and a mild life crisis, 25%
and 9%, respectively, experienced health changes.
(Rahe & Holmes, 1966b, p. 3)

When data from the two studies by Rahe and Holmes (1966a,

1966b) were plotted, the slope of the two lines was similar.

Differences in health change percentages among major, moder-

ate, mile life crisis categories in the prospective study

were statistically significant beyond the .01 level.

Rahe (1968) conducted another prospective study in

which he employed life change data obtained from nearly

2,500 enlisted men and officers aboard three U. S. Navy

cruisers, to predict that population's future illness rates

of minor and major illness. The SRE was administered to

each individual in the sample, with LCU totals being compiled

for six months preceding the study. At the end of a six-

month cruise period, each man's medical record was reviewed

for illness entries. Subjects were rank-ordered according

to each man's LCU total for only the six-month interval im-

mediately prior to the cruise. The upper 30% of this group

was designated as a high risk group, whereas the lower 30%

was designated as a low risk group. Rahe found that the

high risk group reported a significantly greater number of

illnesses than the low risk group for each of the six months

studied. Differences were significant beyond the .01 level

during the third month and beyond .001 for the remaining

five months.

One difficulty with the shipboard studies conducted

by Rahe and his associates lies in the possibility that

some crew members did not report illnesses. Rahe (1972) con-

tended that this is a particularly valid question for older,

senior enlisted men who could retire to quarters with an

illness without having to report it. A study by Cline and

Chosey (1972) of 134 military academy cadets dealt effec-

tively with the problem of illness reporting. A prospective

study of life change and future illness was conducted with

a sample required to report for illness histories and phys-

ical examinations two weeks following the beginning of train-

ing at the academy.

A correlation of .22 was found between previous six-

month LCU scores and number of illnesses experienced during

the first two weeks of training, a correlation significant

at the .05 level. For approximately half the subjects,

health change data were collected at four-month intervals

during the following year. Correlations obtained between

these subjects' LCU scores and total number of health changes

were .34 at four months (significant at .01), .30 at eight

months (.05), and .37 for the entire year, excluding the

initial two weeks (significant beyond .01).

Concurrent Research into Life Change and Illness Onset

An interesting study by Holmes and Holmes (1970)

examined the concurrence of daily life changes with corre-

sponding daily health changes. A Schedule of Daily Experi-

ences (SDE) was derived from the SRE in order that the 42

life change items might be recorded on a daily basis. At

the bottom of the schedule, each subject was instructed to

record all day-to-day health changes however minor. The

sample consisted of 55 students and staff at the University

of Washington Medical School, with subjects' participation

in recording daily events ranging from two to nine weeks.

The amount of time during which subjects recorded life and

health changes totaled more than 1,300 man-days. During

this period, over 1,200 health changes were reported, only

one of which required professional attention (broken tooth).

These health changes were described as the "signs and symp-

toms of everyday life, which reflect in varying degrees

each individual's life style" (Holmes & Holmes, 1970, p. 122)

Statistical analysis of the data revealed a signifi-

cant relationship (.001) between the magnitude of daily life

change and subsequent health changes. A significant clus-

tering effect was found for high LCU totals on the day be-

fore, day of, and day after the occurrence of a symptom. A

similar clustering of low amounts of life change was found

surrounding days for which no symptoms were reported.

Life Change and Severity/Immediacy of Illness

Life Change and Illness Severity

A positive relationship between the occurrence of

life change and the onset of illness has been documented in

the foregoing discussion. Several other studies have demon-

strated that the magnitude of life change events is also

related to the severity of subsequent illness episodes. For

example, Rahe et al. (1967) sought to determine whether life

change history was related to illness severity. On the

basis of life change histories and medical records of 200

medically retired servicemen, it was found that "severe ill-

nesses and clusterings of severe illnesses were preceded by

LCU totals significantly higher than those totals preceding

minor illnesses" and was concluded that "it appears that

death, rather than coming on unpredictably in life, may well

follow a major life crisis" (Rahe et al., 1967, p. 365).

Wyler, Masuda, and Holmes (1971) conducted an investi-

gation which explored the relationship between severity of

illness and quantity of life change that patients had under-

gone during two premorbid years. The sample consisted of

232 hospitalized patients manifesting 42 different disease

entities. Life change information was obtained by adminis-

tering the SRE to each patient. Subjects were asked to in-

dicate which changes had occurred during time periods of

six months, one year, and two years prior to hospitalization.

LCU totals were computed for each subject for these three

time periods. The measure of illness severity for each dis-

ease was obtained by assigning the illness severity ratings

derived by Wyler, Masuda, and Holmes (1968). The procedure

for assessing illness severity will be described in more

detail in Chapter II.

Several findings are of interest here. To begin with,

a significant positive relationship between life change and

illness severity was found. For the time periods of six

months, one year, and two years prior to illness onset, the

correlations between these two variables were .30, .32, and

.35, respectively, each significant beyond .005. When the

diseases were divided into acute and chronic categories,

only the latter showed a significant positive correlation

(.65) in all time periods. Acute illnesses correlated nega-

tively but nonsignificantly in each time period. Black mem-

bers of the sample tended to report larger LCU totals than

white members while simultaneously exhibiting illnesses of

less severity. With the removal of this minority group,

correlations between life change and illness severity were

elevated to .38, .41, and .48 for the six-month, one-year,

and two-year periods, respectively. Regarding this apparent


it may well hold true that the more homogenous
the sample becomes, with respect to socio-economic
class, the stronger the relationship between
seriousness of illness and quantity of life change.
. It is possible that different ethnic and/or
socio-economic groups have different basal levels
of life change.
(Wyler et al., 1971, p. 119)

These findings that acute illness severity was not related

to magnitude of life change and that minority group LCU

totals were consistently higher than totals for whites had

important implications for the sampling procedure of the

present study; they will be discussed in greater detail in

Chapter II.

Life Change and Immediacy of Illness

The notion that people who experience a large amount

of stressful life change over a short period of time succumb

to illness more rapidly is an interesting, yet largely un-

tested proposition. Rahe (1968), studying Navy cruiser per-

sonnel, found that individuals in the highest two LCU deciles

(based on prospective SRE administration) developed nearly

twice as many first illnesses during the first follow-up

month as individuals in the two lowest LCU deciles. During

the second follow-up month, 60% more first illnesses were

reported by the same high LCU group than by the low group.

Preliminary results reported by Rahe suggested that an

immediacy of illness concept would serve as a useful measure

of the illness response to stressful life change.

Effect of Changes in Family Life Patterns on Illness

The usefulness of the SRE as a measure of life change

has been amply demonstrated, at least insofar as applications

to illness incidence are concerned. However, virtually no

attention has been given to relationships between various

life change areas and illness. For example, while marital

and family variables were apparently regarded by developers

of the SRE and related life change measures as quite import-

ant, almost no research has been undertaken to determine

the relative contribution of family-related stress to ill-

ness. The SRE includes 15 familial or marital life change

events out of a total of 42 events. The magnitude of the

LCU ratings which accompany these items subsumes 47% of the

total possible LCU score. With such important consequences

for health being attributed to the vicissitudes of family

and marital functioning, the dearth of research in this area

is surprising.

One study undertaken in this area (Sheldon & Hooper,

1969) was far from definitive. An intensive study of 26

couples during the first year of marriage was conducted by

collecting marital adjustment and health status information.

The sample generally scored in the well-adjusted range of

marriage, with only two couples falling in a'poorly adjusted"

category. A comparison of the six most highly adjusted with

the six least adjusted couples revealed that health measures

for both sexes differed significantly (.05) between the two

groups, with poorly adjusted marriages exhibiting a larger

incidence of poor health. Use of the Cornell Medical Index

(a relatively subjective illness measure), coupled with the

fact that the sample was small, drawn from a university popu-

lation, and married for one year or less, tended to narrow

the conclusiveness of the study. Yet a relationship between

marital functioning and illness has been demonstrated,support-

ing further exploration in the present study.

The Present Investigation

The preceding review of the literature has presented

the evolution of research from the early studies of stress-

inducing stimuli and altered physiological functioning. Two

key concepts in these studies, i.e., stress-inducing stimuli

and physiological alteration, have been preserved in subse-

quent expansion and elaboration of research in this area to

the point that life change events have been found to be relat-

ed to illness onset. The present study was conceived with

the intent of exploring associations between various areas

of life change and illness with a focus upon family-marital


The selection of family-marital functioning as the

primary area of emphasis was based upon two considerations,

both empirical in nature. The first of these considerations

was the observed preponderance of familial and marital life

change events on the SRE (see above). The second considera-

tion arose from the author's experiences with both medical

and psychiatric patients at the Veterans Administration Hos-

pital at Gainesville, Florida. In this regard it was ob-

served that patients with multiple hospitalizations often

reported disruptive or unsatisfactory family-marital condi-

tions at home. These observations engendered a curiosity

if a relationship could be found between the status of

family-marital environments and subsequent illness episodes

requiring hospitalization. This curiosity was subsequently

incorporated into a research paradigm founded upon the as-

sumption that family-marital variables are in fact related

to illness.

The design of the present study constituted an exten-

sion and modification of previously cited research methodol-

ogy. The concept of life change was selected as the primary

vehicle of assessing stress-inducing stimuli because of its

demonstrated efficacy as a measurement technique. Insofar

as family-marital functioning was concerned, two variables

were chosen: (1) life change rates within the familial and

marital areas of the SRE and (2) marital adjustment. The

rationale for use of the family-marital areas of the SRE

has been discussed above. However, a second measure of

family-marital functioning was desired in order to provide

a more encompassing view of this area. While results of the

study by Sheldon and Hooper (1969) were limited in conclu-

siveness, the observation that marital adjustment is related

to health status prompted the inclusion of this measure in

the present study.

Two measures of physiological alteration in the pres-

sent study were derived from the life change/illness litera-

ture. The first of these, severity of illness, was seen as

an important aspect of the illness process. As noted above,

the viability of this variable was demonstrated by Wyler et

al. (1971), and the measurement process was clearly spelled

out in an earlier study by the same group (Wyler et al.,

1968). The other measure of physiological alteration, im-

mediacy of illness, has not been previously studied. Based

on findings by Rahe (1968), however, it was felt that this

concept merited further exploration; hence immediacy was

selected as the second illness variable. With regard to

acute illness, the notable lack of significant findings re-

ported by Wyler et al. (1971) suggested that few, if any,

productive findings would emanate from the inclusion of this

illness category in the study. Hence, hospitalized patients

were limited to those experiencing chronic illness.

One notable omission in the previous research has

been the use of control groups, a mainstay in the behavioral

sciences for many years. If properly employed, the use of

control groups lends considerable strength to conclusions

about experimental group results. Most of the previously

cited studies of life crisis/change have produced findings

which cannot be related to healthy patterns of functioning.

The determination of healthy patterns of life change was

considered essential to the present study; therefore steps

were taken to insure that a group of healthy individuals

comparable to hospitalized subjects was available for pur-

poses of comparing and contrasting results (see Chapter II

for further discussion).

Experimental Hypotheses

Prior to data collection, eight experimental hypoth-

eses were formulated which reflect the emphasis upon family-

marital variables along several dimensions. One set of

hypotheses focuses upon family-marital functioning as re-

lated to both severity and immediacy of illness. Another

set of hypotheses predicts differences in family-marital

functioning between a healthy group of individuals and a

hospitalized group. A third set of hypotheses deals with

predicted relationships between the two measures of family-

marital functioning in both healthy and hospitalized groups.

Hypothesis 1: Among hospitalized subjects, family-
marital sources of life change will be more highly
associated with near-future illness severity than
any other area of life change.

Hypothesis 2: Among hospitalized subjects, a sig-
nificant inverse relationship will be found between
marital adjustment and severity of near-future ill-

Hypothesis 3: Among hospitalized subjects, a sig-
nificantly greater magnitude of family-marital
life change will be found in comparison with a
control group of healthy subjects.

Hypothesis 4: Among hospitalized subjects, sig-
nificantly lower levels of marital adjustment will
be found in comparison with a control group of
healthy subjects.

Hypothesis 5: Among hospitalized subjects, a sig-
nificant inverse relationship will be found be-
tween magnitude of family-marital life change and
length of time between life crisis period and sub-
sequent onset of illness.

Hypothesis 6: Among hospitalized subjects, a sig-
nificant positive relationship will be found be-
tween marital adjustment levels and length of time
between life crisis period and subsequent onset of

Hypothesis 7: Among hospitalized subjects, a sig-
nificant inverse relationship will be found be-
tween marital adjustment levels and magnitude of
family-marital life change.

Hypothesis 8: Among healthy subjects, a signifi-
cant inverse relationship will be found between
marital adjustment levels and magnitude of family-
marital life change.

Chapter II will outline the means whereby the validity of

these hypothesized outcomes was determined.


The methodological sequence of data collection and

analysis in the present study was relatively straightforward.

The following discussion will describe the procedures em-

ployed in collecting data and in subsequently determining

the validity of the eight experimental hypotheses. Discus-

sion will focus upon various procedural aspects of the

study, including the instruments employed in data collection,

an operational restatement of the experimental hypotheses,

the criteria and procedures applied in procuring the sample,

and the statistical procedures employed in analyzing the


Data Collection Instruments

Three instruments were used in the present study to

generate the necessary data upon which the hypotheses could

be verified or rejected. These instruments provided quanti-

tative data in the following areas: (1) overall patterns of

life change; (2) family-marital patterns of change; (3)

marital adjustment; (4) severity of illness; and (5) illness


Overall Patterns of Life Change

An overall assessment of life change for each partici-

pant in the study was obtained through administration of the

Schedule of Recent Experiences (SRE). As described in Chap-

ter I, the SRE is a systematic inventory of 42 stressful

life events which necessitate greater or lesser amounts of

adaptation for most individuals. The form of the SRE used

in the present study incorporated corresponding LCU weights

derived by Holmes and Rahe (1967). (See Appendix I).

The 42 life change events have recently (Rahe, 1972)

been grouped into four major areas of life adjustment:

family, personal, work, and financial (see Appendix II).

In order to facilitate data analyses (to be described below),

Rahe's work and financial areas were combined into a single

category designated as the occupational-financial category.

In order to provide a more descriptive nomenclature for the

remaining two categories, items listed within the family

area by Rahe were designated as the family-marital category,

and items included within Rahe's personal area were desig-

nated as the personal-environmental category. The only

change in Rahe's item arrangement was the removal of the

item "sexual difficulties" from his personal area into the

family-marital category of the present study. This change

was based on the fact that subjects in the present study

were married; therefore it was assumed that sexual diffi-

culties occurred within the context of marital relationship.

The revised categories and corresponding item listings are

presented in Appendix II.

Two additional SRE categories were introduced into

this study, each embodying a different method of tabulating

LCU values. The combined total category of life change re-

flects the LCU total for all endorsed events taken over all

six-month intervals indicated by the subject. This scoring

method is identical to the method used in life change stud-

ies cited in Chapter I and has been used in this study as

an index of total readjustment impact over time. However,

the combined total category potentially contains an over-

balance of LCU values, and hence this category may present

a distorted view of required social readjustment. For exam-

ple, if a subject were to endorse "a change in the number

of family get-togethers" as having occurred during each of

four six-month intervals prior to testing, his LCU score

for that event would total 60 rather than a single occur-

rence value of 15. In using the combined total scoring

method, previous researchers have assumed that a constant

amount of social readjustment is required regardless of

the total time interval during which the change occurred.

The event total scoring method was devised for use

as an alternative tabulation procedure. The underlying

premise for invoking this procedure is that only a finite

amount of readjustment is required for a given event, re-

gardless of the duration of that event. Scores within the

event total category were thus derived by summing LCU values

for each endorsed event regardless of recurrence during sub-

sequent six-month intervals. Subsequent to data collection,

these two scoring categories were compared in terms of re-

spective associations with other variables of interest in

the study (see Chapter III).

Subjects were asked to complete the SRE for the two

years prior to administration of the questionnaire. This

two-year period was partitioned into four six-month inter-

vals, and each subject was asked to indicate the interval

or intervals during which each life change event occurred.

Reliability estimates for SRE usage have ranged from

as low as .26 (Thurlow, 1971) to .90 (McDonald et al., 1972).

Rahe (1974) attributed this large variation to five factors:

(1) time interval between administrations of the question-

naire; (2) educational level, and probable intelligence

level, of subjects; (3) time interval over which subjects'

recent life changes may be measured; (4) wording and format

of various life event questions; and (5) sequence in which

life event questions appear on the questionnaire. Regarding

these sources of variation, Rahe (1974) elaborated on time


When the time interval between questionnaire admin-
istrations was two weeks, the test-retest correla-
tion was .90; when the interval was eight months,
the correlations ranged between .64 to .74; a ten-
month interval gave correlations of .26.
(Rahe, 1974, p. 16)

Commenting on educational factors, Rahe noted:

Highest correlations were obtained from graduate
students in psychology (.90) and physicians (.64
to .74). Intermediate correlations were obtained
from military enlisted men (.55 to .61). The ex-
tremely low correlation of .26 was obtained from
brewery workers.
(Rahe, 1974, p. 16)

Rahe noted that when subjects reported life changes for

yearly intervals rather than six-month intervals, reliabil-

ity increased. He further indicated that questions with

modifiers, e.g., major or a lot more, and questions with

intricate formats were less reliably answered than questions

presented without qualifiers or more simply. In discussing

item sequence on the SRE, Rahe pointed out that "since many

of the life change questions proved to be highly intercor-

related, test-retest reliability was seen to be enhanced by

[ordering] the questions by interrelated clusters than by

LCU score" (Rahe, 1974, p. 16).

While a test-retest reliability procedure would have

undoubtedly enhanced the methodological soundness of the

present study, the logistic requirements of that task were

deemed excessive. It was estimated that reliability values

in the present study would approximate those found among

military enlistees cited by Rahe. Although Rahe's (1974)

comments on the factors limiting the reliability of the SRE

are well taken, these same factors are likewise delimiting

to other social measurement scales, many of which boast

nonetheless substantial reliability coefficients. Despite

the seemingly marginal nature of the SRE's reliability, no

other demonstrated measure of life change was available at

the time of this study. Proceeding on the premise that SRE

data were considerably better than no data, the issue was

set aside until conclusions could be ascertained from the

results (see Chapter IV).

Few estimates of the validity of the SRE have been

offered. Rahe (1974) contended that when life change infor-

mation has been obtained both by interview and by question-

naire, "the interviewers have been invariably impressed

that the information obtained by questionnaire is a valid

although a conservative estimate of subjects' recent life

change experience" (Rahe, 1974, pp. 14-15). Citing unpub-

lished research in which 140 patients completed the SRE

while their spouses completed a separate questionnaire as

if the spouse were the patient, Rahe (1974) showed that,

despite the fact that many spouses did not know all of the

mate's recent life changes, interpair correlations ran be-

tween .50 and .75 over the one- to two-year period immediate-

ly prior to study. Other unpublished research cited by Rahe

(1974) indicated that when life events were carefully dated

by interviewers and confined to events that both husband

and wife knew about, interpair correlations ran as high as

.78. Rahe concluded that "the SRE questionnaire is a mod-

erately valid measure, but less valid and more conservative

than an interview" (Rahe, 1974, p. 15).

Measurement of Family-Marital Functioning

Two measurement procedures were employed in the assess-

ment of family-marital functioning. The first, a more quan-

titative measure, consisted of the family-marital items on

the SRE. A second measure, the Locke-Wallace Short Marital

Adjustment Test (SMAT), provided a more qualitative perspec-

tive with regard to marital functioning. A description of

both measures follows.

As mentioned above, the family-marital category of

life change was derived from Rahe's (1972) classification

system of all SRE life event items. Fifteen items pertain-

ing to stressful family and marital change were included in

this category, accounting for 47.8% of the possible LCU

total. The personal-environmental and occupational-financial

categories accounted for 27.7% and 24.5% of the remaining

possible LCU total, respectively.

The Locke-Wallace SMAT was developed by Locke and

Wallace (1959) to provide a shortened form for assessing

marital adjustment without appreciable loss in reliability

or validity from longer adjustment tests (e.g., Burgess-

Wallin Marital Success Schedule, Locke Marital Adjustment

Test). The SMAT contains 15 items, with possible scores

ranging from 2 to 158 points. Locke and Wallace concurrent-

ly reported a reliability and validity study with the SMAT

in which couples receiving marital counseling as well as

those judged exceptionally well adjusted were asked to com-

plete the test. The reliability was found to be .90 as com-

puted by the split-halves technique. An arbitrary cutoff

point of 100 was found to differentiate significantly be-

tween the two groups.

Not atypically, the sample employed by these research-

ers to test the reliability and validity of the SMAT was pre-

dominantly white, middle class, and well educated. In con-

trast, the sample used in the present study reflected more

modest socioeconomic and educational attainment. Hence,

the Edmonds correction scale, consisting of 15 marriage-

related items, was incorporated into the SMAT to correct

for serendipitous response tendencies often observed in this

population. The SMAT with the incorporated Edmonds items

is shown in Appendix III.

The Seriousness of Illness Rating Scale (SIRS) was

developed by Wyler et al. (1968) to provide a quantitative

measure of illness severity. A total of 258 medical and

nonmedical respondents were asked to rate 126 diseases as

to relative seriousness. An arbitrary value of 500 was as-

signed to peptic ulcer, with respondents quantitatively

rating the remaining diseases as more or less serious. No

appreciable difference in severity ratings was found between

medical and nonmedical samples; hence the data were pooled

to obtain a total sample mean for each disease. Each

hospitalized subject was assigned an SIRS score on the basis

of primary diagnosis according to the total sample means de-

scribed above. The SIRS disease items and the respective

mean severity ratings are listed in Appendix IV.

Disease items on the SIRS were also employed in con-

structing the Health History Inventory (HHI), a measure of

different illness episodes during five years prior to the

present study. A number of the more common, less serious

disease items on the SIRS (corns, hiccups, bad breath) were

deleted in devising the inventory, leaving only more infre-

quent and medically significant illnesses for inclusion. A

Health History Inventory score was computed for each subject

by assigning SIRS values to each disease item endorsed and

then summing these values into a single total. The Health

History Inventory is shown in Appendix V.

Immediacy of Illness Onset

Having characterized the concept of immediacy of ill-

ness onset as a promising, innovative variable which might

serve to broaden the scope of illness measurement, the less

glamorous task of operationalizing this variable was under-

taken. Immediacy of illness as alluded to by Rahe (1968)

consisted of the elapsed time between the period of great-

est life change (the "life change crisis period") and the

time at which a subsequent illness was reported. Rahe de-

scribed this elapsed time interval in terms of months, but

the use of such a large time unit was seen as inexact and

therefore unsuitable for the present study. Instead, the

number of days which elapsed between the midpoint of the

six-month life change crisis period and subsequent onset of

illness was selected as the measurement modality. Unfor-

tunately, determination of a precise date of onset proved

to be a deceptively simple process.

Prior to data collection it was presumed that infor-

mation necessary for determining a relatively precise date

of onset for all hospitalized subjects would be contained

in medical records. Soon after data collection began, how-

ever, this presumption was reassessed in light of the vir-

tual absence of such information in the records. It was

then decided that, in addition to medical record checks, the

patients themselves would be queried as to the date when

symptoms first became noticeable. After several efforts in

this direction, it became evident why the medical records

lacked such information: Many of the patients interviewed

were unable to pinpoint, or even estimate, the time at which

symptoms began to appear or, as was often the case, to re-


Almost all patients for whom a date of onset could

not be determined fell within one or more of the following

categories: (1) too ill, (2) too medicated, and (3) too

physically or emotionally depleted. In a few instances the

symptoms associated with illness were neither precipitous

in appearance nor well defined, thereby precluding any accu-

rate and reliable determination of onset date.* In some

cases the date of onset could be determined from medical

records. Eventually, however, only 30 of the 51 patients

in the study were assigned what was felt to be a reasonably

accurate date of onset.

The Sample

Two populations were sampled in the present study.

The first sample was drawn from a population of hospitalized

patients; the second, from a population of healthy individ-


The Hospitalized Sample

The hospitalized sample was obtained from the Veter-

ans Administration Hospital in Gainesville, Florida, during

the Spring Quarter, 1974. Patients were selected on the

basis of sex, race, diagnostic classification, age and

marital status characteristics. Each of these criteria will

be discussed briefly.

Sex.--Due to the nature of the population receiving

treatment at the Gainesville veterans hospital, all members

of the sample were male. As such, the sample was one of

convenience inasmuch as the overwhelming majority of patients

*In answering the author's query as to date of onset,
one patient responded, with some irritation, that his symp-
toms were his doctor's problem, not his.

at the VA Hospital are male. However, the selection of an

all-male sample was conservative in nature, at least if the

findings of Holmes and Holmes (1970) are valid for the popu-

lation presently under study. Females in that study were

more symptomatic in relation to day-to-day life change than

were males.

Race.--Based on the findings of Wyler et al. (1971),

black Americans appear to differ with regard to baseline

life change patterns, suggesting that LCU ratings derived

by Holmes and Rahe (1967) do not hold up reliably for blacks.

Since the issue of racial differences was not resolved in

the literature and exceeded the scope of the present study,

only white American-born subjects were selected for partici-

pation in this study.

Diagnosis.--As noted earlier, Wyler et al. (1971)

found a significant relationship between life change and

illness among the chronically ill, while no such relation-

ship was found among those diagnosed as acutely ill. Hence,

for purposes of the current study, only those diagnosed as

chronically ill were included in the sample. It was found

that the definition of chronic illness has been difficult

to specify. Some direction was offered by Dodge and Martin

(1970), who identified some general characteristics of


In using the term "chronic diseases" we adhere to
the principle that the term indicates a long dura-
tion of the disease process in contrast to that
process associated with the acute diseases. Beyond

this stipulating it is difficult to maintain a
sharp differentiation between the two groups of
diseases as we use them. If we employ the criteria
of infectiousness or communicability, it can be
noted that some chronic diseases, like the acute
diseases, involve known pathogenic microorganisms
and are communicable. If we point out that chronic
diseases usually leave residual disabilities and
generally require extensive periods of recovery
or rehabilitation, undoubtedly a few acute diseases
can be found with these characteristics. However,
it must be noted that these factors when associ-
ated with chronic diseases are generally, but not
in every case, more of a certainty, more extensive,
severe, and more often nonreversible.
(Dodge & Martin, 1970, p. 78)

A number of the disease items listed on the SIRS were

clearly acute in nature, e.g., sunburn, shark bite, chicken

pox, etc. All disease items on the SIRS were reviewed by

the physician consultant to the study, who employed the

criteria for chronicity cited above in determining which

SIRS items could be clearly designated as chronic.*

Age.--The relationship between life change and illness

has heretofore been found in other studies to be generally

constant across all age groups. However, because of the

chronic nature of the diagnostic criteria, the minimum age

for inclusion in the study was 25; the ceiling age was set

at 65. Both age criteria were somewhat arbitrary in nature.

However, the inclusive age range defined by these two boundary

*Several items which fall under the general rubric of
psychiatric disorders were found among the SIRS listings.
Previous life crisis research has implied little, if any,
distinction between such classical psychiatric syndromes as
manic-depression or schizophrenia and physiological disease
entities. Indeed, the evidence linking psychotic disorders
to chemical and metabolic alterations in the body is con-

ages was expected to demonstrate a generally uncomplicated

picture of chronic illness as related to life change.

Marital status.--In order to obtain adequate informa-

tion regarding family-marital functioning, it was necessary

to establish marital criteria for inclusion in the study.

One year of marital contact was considered minimal. The

criteria were as follows; All members of the hospitalized

sample were to have been married for at least one year dur-

ing the two-year period prior to data collection. Thus in-

cluded were persons who had become divorced or legally sepa-

rated and persons whose wives had died during the year prior

to the study. Also included were individuals who had been

newly married for at least one year prior to the study.

Participants were not required to have been married for the

full two-year period prior to the study because some of the

most heavily weighted SRE items included death of spouse,

separation, or marriage.

The Control Group

The rationale in forming a control group was to pro-

vide a means of comparison between life change/marital

vincing. However, the undifferentiated inclusion of psychi-
atric disorders with illness stemming from physiological
pathology presupposes an etiologic and qualitative similar-
ity which, to the author's knowledge, has not been conclu-
sively demonstrated. Therefore, the present study focused
upon cases of chronic illness in which the presenting symp-
tomatology was judged to emanate primarily from physiological

adjustment patterns between hospitalized and healthy popula-

tions. The control group employed in the present investiga-

tion was matched with the hospitalized group insofar as age,

sex, race, marital status, and education were concerned.

That is, sample selection was made to insure that group dif-

ferences between the two samples would not differ signifi-

cantly with regard to these five dimensions. In addition,

all control group participants were required to be eligible

for treatment at the Gainesville VA Hospital and to live

within the hospital's catchment area. Only individuals who

had not been hospitalized during five years prior to the

study were designated as eligible to participate. Data from

the control group were collected during the same period that

data were received from the hospitalized group. The control

group was intended to consist of approximately the same num-

ber of healthy veterans as the hospitalized group. Because

the battery of questionnaires required moderate ability in

reading comprehension, subjects in both samples were required

to have at least a tenth-grade education.

Procedural Format

The following section will outline the procedures em-

ployed in identifying subjects for the sample and collecting


Data Collection Within the Hospitalized Group

In order to be selected to participate in the study,

each patient was required to meet the demographic criteria

described above. The physician consultant to the study

assisted in screening prospective subjects to verify com-

pliance with the criteria for inclusion. Screening was

assomplished by periodic medical record reviews on each of

the medicine and surgery wards in the hospital.

When an eligible patient was identified, he was con-

tacted by a member of the research team who briefly described

the nature of the research project. Each prospective sub-

ject was assured that all information would remain confiden-

tial. Patients agreeing to participate were asked to com-

plete an information sheet including the patient's age and

educational attainment, admitting diagnosis, date of admis-

sion, and marital information (see Appendix VI).

If the prospective patient met the minimal education-

al requirements for participation, he was then given a bat-

tery of questionnaires consisting of a consent agreement,

Health History Inventory, SRE, and SMAT to complete at his

convenience. Each participant was requested to avoid col-

laboration with others in completing the questionnaires. At

the time of the patient's discharge, the patient's medical

record was reviewed to insure that the primary diagnosis was

not changed during hospitalization. If the primary discharge

diagnosis differed from the primary admitting diagnosis, the

former was used in determining illness severity. In a few

instances the primary discharge diagnosis differed from the

admitting diagnosis and did not appear among the chronic

disease items designated on the SIRS. In these instances

the patient was dropped from the study.

Data Collection Within the Control Group

Participation in the control group was contingent

upon meeting the demographic criteria employed in selecting

hospitalized subjects, i.e., criteria pertaining to age,

sex, race, marital status, and education. The questionnaires

administered to control group participants were identical to

those given to the hospitalized group. All participants

were assured of anonymity.

Two sources of data were employed to assemble the

control group. The first source consisted of VA-affiliated

veterans' groups in Ocala, Orlando, and Daytona Beach, Flori-

da. The writer attended meetings at each site and delivered

a short presentation to each group, describing the nature of

the study and the criteria for participation. All eligible

members were then requested to support the project by com-

pleting the battery of questionnaires following the meeting.

Attendance at these meetings netted a total of 36 subjects

for the control group.

The second source of control group data was an adver-

tisement in the Gainesville Sun, the major local daily news-

paper, briefly describing the study and the requirements for

participation and inviting interested persons to respond. A

telephone number was provided for further inquiries. Tele-

phone respondents were given a more detailed description of

the study and were asked to complete a questionnaire in per-

son at the VA Hospital. Respondents agreeing to participate

were given an appointment at their convenience. Subjects

appearing for appointments were ushered to an unoccupied

office where instructions for completing the questionnaire

were given. A total of 20 subjects participated in this


Operational Specifications

In order to adequately test the experimental hypoth-

eses advanced at the conclusion of Chapter I, a more precise

operational formulation of each hypothesis was required to

insure that all hypotheses could be evaluated through use of

the above-mentioned instruments and measurement techniques,

as well as to facilitate subsequent interpretation of the

results. Hence the eight hypotheses were restated in the

following operational terms:

Hypothesis 1: Among members of the hospitalized
sample, a higher positive correlation will be
found between LCU values within the family-marital
category of the SRE and corresponding SIRS values
in comparison with correlations between the re-
maining SRE categories and SIRS values.

Hypothesis 2: Among members of the hospitalized
sample, a significant inverse correlation will be
found between SMlAT scores and corresponding SIRS

Hypothesis 3: Among members of the hospitalized
sample, significantly larger LCU values will be

found within the family-marital category of the
SRE in comparison with the same values among mem-
bers of the control sample.

Hypothesis 4: Among members of the hospitalized
sample, significantly lower SMAT scores will be
found in comparison with the same scores among
members of the control sample.

Hypothesis 5: Among members of the hospitalized
sample, a significant inverse correlation will be
found between LCU values within the family-marital
category of the SRE and corresponding immediacy of
onset values expressed in number of days.

Hypothesis 6: Among members of the hospitalized
sample, a significant positive correlation will
be found between SMAT scores and corresponding
immediacy of onset values expressed in number of

Hypothesis 7: Among members of the hospitalized
sample, a significant inverse correlation will
be found between SMAT scores and corresponding
LCU values wtihin the family-marital category of
the SRE.

Hypothesis 8: Among members of the control sample,
a significant inverse correlation will be found
between SMAT scores and corresponding LCU values
within the family-marital category of the SRE.

The directionality of predicted results in the above

hypotheses required a one-tailed test of statistical signi-

ficance. The .05 level of significance was selected as the

criterion value for acceptance or rejection of each hypoth-


Data Analysis

Nine separate statistical analyses were planned in

the present study. The first eight were conducted in con-

junction with each of the eight experimental hypotheses.

The ninth analysis compared differences between hospitalized

and control groups regarding age and education. The data de-

rived from both SRE and SMAT were ordinal in nature; thus,

nonparametric statistical analyses of this data were indi-

cated. The analysis of demographic data between the two

groups were based on equal-interval data, thus permitting

the use of parametric statistical procedures. All statisti-

cal computations were performed by computer.

Hypotheses 1 and 2

The Spearman rank correlation procedure was employed

to test Hypotheses 1 and 2. To test Hypothesis 1, five cor-

relations were computed, one for each of the SRE categories

and corresponding SIRS scores. Within each of the five cor-

relational analyses, subjects were ranked in ascending order

according to the magnitude of mean LCU scores. Each subject

was also assigned an ascending SIRS ranking based on the

magnitude of the SIRS score associated with his diagnosis.

Differences between the five SRE category rankings and the

corresponding SIRS rankings were computed in deriving the

respective correlation coefficients. A similar analysis was

performed in conjunction with Hypothesis 2, with SMAT and

SIRS values being rank ordered and compared in generating the

correlation coefficient between marital adjustment and ill-

ness severity.

Hypotheses 3 and 4

The family-marital LCU totals from the SRE and the

SMAT scores were compared between subjects in the hospital-

ized and control groups to test Hypotheses 3 and 4. Statis-

tical analysis of the data was performed by using the Mann-

Whitney test. Family-marital and SMAT scores from the com-

bined samples were listed in rank order according to respec-

tive magnitudes. The U statistic was then computed, and the

level of significance for the differences determined.

Hypotheses 5 and 6

In conjunction with Hypothesis 5, the Spearman rank

correlation procedure was employed in computing the correla-

tion between scores from the five SRE categories and the

corresponding immediacy of illness (II) values. Mean LCU

scores were used to rank all subjects in ascending order.

Each subject was also assigned an ascending II rank accord-

ing to the number of elapsed days between the life crisis

period and illness onset. Differences between the five SRE

category rankings and the corresponding II rankings were

employed in computing five correlation coefficients. The

analysis performed in conjunction with Hypothesis 6 was simi-

lar, with SMAT and II scores being ranked and compared in

computing the correlation coefficient.

Hypotheses 7 and 8

The Spearman correlation procedure was again employed

in analyses associated with Hypotheses 7 and 8. In this in-

stance family-marital LCU values and SMAT scores were rank

ordered in both the hospitalized and control groups. Corre-

lation coefficients for both samples were derived from the

computed differences in rank values.

Testing for Group Differences

A t test was employed to test for significant differ-

ences in age and education between the hospitalized and con-

trol groups.

Chapter III will present the results of the foregoing

analyses as well as the demographic characteristics of both




The results of the study will be presented in three

sections. The first section will present the results of

age and educational comparisons between hospitalized and

control samples; the second section, the demographic char-

acteristics for both samples; and the third section, the

results of analyses performed in conjunction with the eight

experimental hypotheses.

Demographic Comparisons Between Groups

The data collection process netted a total of 54 hos-

pitalized group members and 56 control group participants.

Compliance with criteria regarding sex, race, marital status,

and minimal educational attainment was accomplished through

discriminative sample selection procedures.* A t test was

conducted with the hospitalized and control groups, analyzing

*Sex and race criteria were checked during the medical
record review sessions, with noneligible patients being
screened prior to contact. Information regarding educa-
tional background and marital status, however, was often
omitted from medical records, necessitating personal con-
tact with patients in question. A surprisingly large number
of patients contacted for further educational and marital
information did not meet minimum requirements for inclusion.
In such cases the patient was thanked for his cooperation,
and the contact was terminated.

age and educational data to determine if significant differ-

ences between the two groups existed with regard to either

variable. When full-sample hospitalized and control age

data were compared, it was found that differences in age

between the two groups approximated the .05 level of signi-

ficance (see Table 1). The analysis of full-sample educa-

tional data between the two groups revealed very significant

differences, with educational attainment in the control

group eclipsing that of the hospitalized group beyond the

.0001 level of significance (see Table 1).

Because of large educational discrepancies between

the two groups, upper-grade-level cutoff values were imposed

upon both groups in an effort to minimize group differences.

A cutoff level of 15 years of schooling minimized education-

al differences between the two samples; hence all members

from both samples with more than 15 years of education were

dropped, resulting in a loss of 3 members from the hospital-

ized group while 20 members from the original control sample

were eliminated.* The resultant age and educational levels

in the two groups (numbering 51 in the hospitalized group

and 36 in the control group) were again compared by t test.

Results of the two comparisons are shown in Table 1. The

t value for the intergroup comparison of age was -.0007,

*Most of the control group members who were eliminated
by the grade-level cutoff were respondents to the local
newspaper ad. For the most part, these individuals were
young military veterans with bachelor's degrees who had
returned to the University of Florida to pursue graduate



Full Sample


n mean
















1.59 NS

3.60 .0001

Dropped Sample

n mean















-.0007 NS

-.1208 NS

t P



t p

while the comparison of educational level between the two

groups yielded a t value of -.1208. Both values were notably


One further comparison was undertaken to determine if

the two newly formed groups differed significantly with re-

gard to illness history. Health history totals were thus

compared between the two groups through the use of the Mann-

Whitney U test. As may be seen in Table 2, the health his-

tory totals reported by the hospitalized group significant-

ly exceeded those reported by members of the control group,

with the statistical significance of the differences falling

well beyond the .001 level.

Demographic Characteristics of the Samples

The most notable demographic characteristics of both

the hospitalized and control groups were compiled (see

Tables 3, 4, and 5, and Figure 1). These data pertain to

family and marital history, employment status, and age dis-

tribution within the two samples as well as to diagnostic

classification within the hospitalized group.

An analysis of Table 3 reveals that, in general, par-

ticipants in the hospitalized group experienced a more dis-

rupted, disorganized family and marital history than did

control group counterparts. For example, 42% of hospital-

ized group veterans had been divorced at least one time; 22%

of control group members reported a previous divorce. Of



n z p
Health History

Hospital 51 5.541 .001

Control 34



Hospitalized Control
Group Group

Subjects reporting
previous divorces

Wives' previous divorces
(subjects' reports)

Subjects reporting sepa-
rations from previous

Subjects reporting sepa-
rations from present

Subjects reporting divorces
among parents

42% 22%

the hospitalized group, 6% reported having been legally sepa-

rated from the present wives, and 30% had been separated from

previous wives. In contrast, none of the veterans in the

control group had been legally separated from current spouses,

while 11% reported separations from previous wives. Interest-

ingly enough, veterans in both groups reported an identical

percentage of previous divorces among present wives (36%).

Of the hospitalized group, 21% reported at least one divorce

between parents, compared with a 16% parental divorce rate

reported by control group members.

Both groups reported a substantial rate of unemploy-

ment (see Table 4). Hospitalized group members reported a

47% rate of unemployment; control group members reported a

36% unemployment rate. Among those who reported being un-

employed, hospitalized subjects had been out of work for an

average of 44.2 months, whereas subjects in the control

group had been unemployed for an average of 12.8 months. An

analysis of age distribution within the two groups (see

Figure 1) reveals a somewhat similar pattern of age clusters.

Within the hospitalized group, 45% of all participants fell

within the 42-49 age bracket, while 50% of control group mem-

bers were encompassed within the 50-57 age interval. Above

age 40 were 88% of the hospitalized and 77% of the control

group. Fourteen chronic disease entities were represented

in the hospitalized group (see Table 5), with cancer, arterio-

sclerosis, peptic ulcer, and kidney disease accounting for

the major sources of affliction.


Hospitalized Control
Group Group

Subjects reporting unem- 47% 36%
ployment at time of study

Average elapsed time since
last employment among un- 44.2 12.8
employed subjects months months


Number of Percent of
Diagnostic Subjects Subjects

Cancer 9 17%
Hardening of the arteries 7 13%
Peptic ulcer 6 12%
Uremia 6 12%
Leukemia 5 10%
High blood pressure 4 8%
Chest pain 4 8%
Heart failure 3 6%
Cirrhosis 2 4%
Arthritis 1 2%
Diarrhea 1 2%
Pancreatitis 1 2%
Bronchitis 1 2%
Diabetes 1 2%

22-25 26-29 30-33 34-37 38-41 42-45 46-49 50-53 54-57 58-61 62-65 66-69

Distribution of age
(in four-year increments)

Hospitalized Group

F 9
r 8
e 7
q 6
u 5
e 4
n 3
c 2
y 1-

22-25 26-29 30-33 34-37 38-41 42-45 46-49 50-53 54-57 58-61 62-65 66-69

Distribution of age
(in four-year increments)
Control Group

Figure 1. Distribution of age (Hospitalized and Control

Comparisons Associated with Hypotheses

Hypothesis 1 predicted that, in the hospitalized

group, the correlation between SRE family-marital LCU val-

ues and SIRS values would be larger than corresponding cor-

relations between the remaining four life change categories

and illness severity. A Spearman correlation coefficient

was computed between each of the five SRE category LCU

scores and corresponding SIRS values within the hospital-

zed group (see Table 6). Inasmuch as the correlation be-

tween family-marital life change scores and severity of ill-

ness values was the lowest of the five computed correlations,

Hypothesis 1 was not confirmed.

Hypothesis 2 predicted that an inverse correlation

would be found between Locke-Wallace SMAT scores and cor-

responding SIRS values for the hospitalized sample. To test

the hypothesis, the Spearman correlation was computed be-

tween SMAT scores and SIRS values. The correlation coeffi-

cient between these two variables was found to be -.076,

which was statistically nonsignificant (see Table 7). Hence

Hypothesis 2 was not confirmed.

Hypothesis 3 predicted that a larger amount of family-

marital life change would be found in the hospitalized

group than in the control group. In order to test this as-

sumption, the Mann-Whitney U test was employed to compare

family-marital LCU totals between the two groups (see Table

8). Family-marital totals in the hospitalized group were



(Hospitalized Group)

Family- Personal- Occupational- Combined Event
Marital Environmental Financial Total Total
n 51 51 51 51 51
r .149 .312 .192 .255 .197

p NS .025 NS .05 NS



(Hospitalized Group)

Locke-Wallace Score
SIRS SMAT Conventionality

n 51 51
r -.076 -.030


indeed found to exceed totals within the control group. The

computed z score was significant at the .067 level. While

the difference was found to approach statistical signifi-

cance, the .05 level was not attained.

Hypothesis 4 predicted that subjects in the hospital-

ized group would report lower Locke-Wallace SMAT scores than

subjects in the control group. The Mann-Whitney U test was

again employed to test for significant differences between

SMAT scores within the two groups. The z score was found

to be significant at the .044 level (see Table 9) thereby

confirming the hypothesis.

Hypothesis 5 predicted that, within the hospitalized

group, an inverse correlation would be found between family-

marital LCU scores and immediacy of illness onset values.

To test the hypothesis, the Spearman correlation was com-

puted between these two variables (see Table 10). A correla-

tion of -.311 was found between the two variables, signifi-

cant beyond the .05 level, which confirmed the hypothesis.

Hypothesis 6 predicted that, among hospitalized sub-

jects, a positive correlation would be found between marital

adjustment scores and immediacy of illness onset. The Spear-

man correlation procedure was again employed to determine

the correlation between Locke-Wallace SMAT scores and imme-

diacy of illness onset values. The SMAT scores were found

to correlate .089 with the immediacy of onset values, which

was not statistically significant (see Table 11). Therefore

Hypothesis 6 was not confirmed.



Hospital Control
n n z p
SRE Category
Family-marital 51 34 1.501 .067
Personal- 51 34 1.565 .057
Occupational- 51 34 -.355 .364
Combined total 51 34 2.131 .012
Event total 51 34 2.848 .002



Hospital Control
n n z p
Locke-Wallace Score

SMAT 51 36 -1.707 .044
Conventionality 51 36 1.440 .075



(Hospitalized Group)

Family- Personal- Occupational- Combined Event
Immediacy Marital Environmental Financial Total Total
of Illness ---
n 30 30 30 30 30
r -.311 .261 -.309 .007 .085

p .025 .05 .025 NS NS



(Hospitalized Group)

Locke-Wallace Score
Immediacy of SMAT Conventionality
Illness Onset

n 30 30
r .089 .179


Hypothesis 7 predicted that, within the hospitalized

group, an inverse correlation would be found between Locke-

Wallace SMAT scores and family-marital LCU scores. The Spear-

man correlation coefficient was computed between the two

variables to test this assumption. A correlation of -.369

was found significant beyond the .001 level (see Table 12).

Thus, Hypothesis 7 was confirmed.

Hypothesis 8 predicted that an inverse correlation

would also be found between Locke-Wallace SMAT scores and

family-marital LCU scores in the control group. The Spear-

man correlation was found to be -.201 between the two vari-

ables (see Table 12), which was nonsignificant. Predicted

results not having been obtained, Hypothesis 8 was not con-


A more detailed discussion of these results, as well

as a discussion of other findings of interest in the study,

will be presented in Chapter IV.


(Hospitalized Group)

Family- Personal-
Marital Environmental











SRE Category

Occupational- Combined
Financial Total










(Control Group)






Family- Personal-
Marital Environmental











SRE Category

Occupational- Combined
Financial Total































Age Sn


n 5

r .4

p .0

n 3'

r .2


Locke-Wallace Scores
AT Conventionality















(Hospitalized Group)


SIRS Scores








LCU Totals










SRE Life Change Categories

Health Family- Personal-
History Marital Environmental


Combined Event Conven-
Total Total tional







50 50 50 50

.250 .202 .175 .107

.05 NS NS NS




33 33

-.163 .051





SRE Life Change Categories

lace Score

ization/Non-Family- Personal- Occupational Combined Event Conven
hospital- Marital Environmental Financial Total Total tional

n 85 85 85 85 85 87 87

rpb .22 .46 .05 .32 .35 .19 .13
pb 2 .

ace Score






















P .025

.0001 NS

.005 .001 .05 NS



Health history totals
Family-marital LCU totals
LCU totals
LCU totals
Combined total LCU totals
Event total LCU totals

Subjects' Age
Hospital Control
n r p n r p

50 .255 .05 35 .320 .05
51 .114 NS 34 -.235 NS

51 -.005 NS 34 -.502 .01

51 -.184 NS 34 -.386 .025

51 -.185 NS
51 -.227 NS

34 -.448 .01
34 -.469 .01



Variables Correlated
SIRS-Immediacy of illness onset

SIRS-Health history totals

Immediacy of illness onset-
Health history totals

Correlation Coefficient





Results of the analyses presented in Chapter III

merit further discussion and, where possible, interpreta-

tion. The present chapter will elaborate on these results

with an emphasis on synthesizing the results into a more

descriptive framework. Although not directly related to

the hypothesized relationships or differences in the present

study, other data analyses were undertaken in order to iden-

tify noteworthy findings which were not of primary concern

at the outset of the study. A number of these results,

tabulated in Tables 6 through 18, are central to a more

comprehensive understanding of the etiology and demography

of chronic illness; therefore, these results will be refer-

red to as needed during the course of discussion.

Minimizing Group Differences

The establishment of educational attainment as the

criterion variable for minimizing both educational and age

differences between the hospitalized and control samples

was effective. The two groups were similar in terms of

level of educational attainment and age characteristics

(mean and standard deviation) following the elimination

procedure. As expected, the two groups did differ with re-

spect to health history totals; however, the magnitude of

difference was larger than anticipated, suggesting that mem-

bers of the hospitalized group had been beset by multiple

major medical difficulties during the five-year period prior

to the study while members of the control group were rela-

tively illness-free during that time. Results of the above

three comparisons indicate that the two groups were nearly

identical as to age and educational background data, yet

quite different insofar as prior illness rates are concerned.

The fact that the hospitalized group was markedly illness-

prone while the controls were relatively free of illness

tends to strengthen a number of the conclusions offered


Comparison with Previous Life Change Studies

Several findings in the present study support import-

ant conclusions arising out of previous research. For exam-

ple, veterans' combined total SRE scores (i.e., the total of

all life change events regardless of recurrence) in the hos-

pitalized sample were significantly higher than scores of

counterparts in the control group (see Table 8). Coupled

with the finding that hospitalized veterans' combined total

The combined total SRE score was computed in the same
manner as in all previous studies employing the SRE and is
therefore useful in comparing SRE results in the present
investigation to those in prior studies.

scores were significantly related to admission to the hospi-

tal (see Table 16), these higher scores lend further support

to previous researchers' conclusions that life change and

illness onset are significantly related. Furthermore, com-

bined total scores among hospitalized participants were

found to be correlated significantly with illness severity

(see Table 6), supporting the conclusions of Rahe et al.

(1967) that life change history is related to illness sever-

ity. Finally, as in the study by Wyler et al. (1971), the

two six-month periods immediately prior to hospitalization

were incrementally and significantly correlated with sever-

ity of illness (see Table 14). Hence a sense of continuity

exists between current results and previous research find-

ings. However, the modifications incorporated into the

present design offered a number of additional findings which

provide further insight into the illness process.

Family-Marital Variables in Relationship to Illness

The primary focus in the present study centered upon

family-marital functioning as related to illness. Indeed

all eight experimental hypotheses sought either to compare

veterans' family-marital functioning between the hospital-

ized and control groups or to establish a relationship be-

tween family-marital life changes or marital adjustment and

severity or immediacy of illness. The fact that three of

these hypotheses were confirmed, while another closely

approximated confirmation, suggests that a given individual's

family-marital experience is important to a broader under-

standing of illness etiology.

The first two hypotheses focused on the relationship

between veterans' perceptions of family-marital experiences

and the severity of subsequent illness. As noted earlier

(see Chapter I), Hypothesis 1 sought to explore the quanti-

tative aspects of this relationship through the measurement

of family-marital life change, i.e., an inventory of event

history which is relatively objective in nature. In con-

trast, Hypothesis 2 invoked the Locke-Wallace SMAT in an

effort to provide a qualitative perspective on subjects'

marital functioning through an assessment of subjects' per-

ceptions of spouse and the marital experience itself. The

lack of statistical significance associated with either

analysis leads to the conclusion that, for the population

under study, neither marital adjustment as perceived by

individuals in the study nor incidence of family-marital

life change was related to severity of illness to any appreci-

able degree.

Despite the fact that husbands' incidence of family-

marital life change and perceptions of marital adjustment

did not correlate significantly with illness severity, these

two measures did differ between the hospitalized and control

groups. As predicted by Hypothesis 3, family-marital life

change occurred more frequently among hospitalized group

participants relative to change rates among control group

members during the two-year period prior to study. While

this difference was not statistically significant, it did

approach significance. Marital adjustment levels were found

to be significantly lower among hospitalized group members

than among control group members.

Also of interest were results of a comparison of con-

ventionality scores between the two groups (see Table 9).

The two groups did not differ significantly in the degree

to which members responded with an eye toward social conven-

tion. However, hospitalized veterans tended to respond to

the SMAT in a more conventional manner than did members of

the control group, but this trend was not of sufficient

magnitude to support the conclusion that SMAT scores within

the hospitalized group were significantly overly convention-

alized in relation to control group scores. Based on these

between-group comparisons of life change and marital adjust-

ment, it may be concluded that substantial differences did

exist between the two groups in terms of family-marital


Hypotheses 5 and 6 predicted that, within the hospi-

talized group, a significant relationship would be found

between each of the two family-marital variables and the

immediacy with which illness onset occurred. As far as

Hypothesis 5 is concerned, veterans' reports of family-

marital life change were significantly correlated with

immediacy of onset. Hence, when the incidence of family-

marital change for a given individual was high, the time

span between the six-month life change crisis period and

illness onset tended to be shorter. On the other hand, the

relationship between marital adjustment and immediacy of

illness onset among hospitalized members, postulated by

Hypothesis 6 to be significant, was not found significant.

Perceived marital adjustment was not found to corre-

late significantly with either of the two illness variables.

Indeed, among the four correlations between the two meas-

ures of family-marital functioning and the dual parameters

of severity of illness and immediacy of illness onset, only

one (family-marital life change-immediacy of illness) emerged

as statistically significant.

The final two hypotheses predicted that higher rates

of veterans' family-marital life changes during the two-

year premorbid period would be associated with reports of

lower marital adjustment from both the hospitalized and con-

trol groups. As mentioned in Chapter II, family-marital

life change and marital adjustment as measured by the SMAT

were seen as essentially discrete variables within the

broader context of marital and family functioning. Hypoth-

esis 7 sought to determine the extent to which these two

areas of functioning were interrelated among hospitalized

participants. It was found that these areas were indeed

related and to a significant extent. However, the same

correlation undertaken with control group data showed no

significant relationship between veterans' marital adjust-

ment and family-marital life change. Hence a high incidence

of family-marital life change was associated with generally

lower levels of marital adjustment among the chronically

ill, whereas comparable rates of change among the healthy

participants were unrelated to marital adjustment.

Although previous illness data, as measured by health

history totals, were not designated as an illness variable

prior to the study, the relationship between these totals

and both family-marital life change and marital adjustment

are worthy of mention at this point. Within the hospital-

ized group, for example, the magnitude of veterans' family-

marital life change was found to be significantly related

to Health History Inventory totals; however, the same rela-

tionship among control group participants was nonsignifi-

cant (see Table 15).* Veterans' reports of marital adjust-

ment levels were not related to the incidence of prior ill-

ness in either group (see Table 15).

These results coupled with results of the experi-

mental hypotheses suggest the following picture of family-

marital functioning as related to the chronic illness proc-

ess: Of the two variables employed to measure family-marital

functioning, the incidence of family-marital life change

*An admitted difficulty exists with this type of retro-
spective analysis. The family-marital category scores em-
ployed in the present study were obtained by summing all
life change events within this category over the entire two-

among veterans was found to be significantly related to both

the immediacy of subsequent illness onset and the incidence

of previous illnesses. Despite nonsignificant relationships

between marital adjustment and illness parameters, hospital-

ized group members reported significantly lower levels of

marital adjustment. Also, family-marital life change magni-

tudes were found to be substantially higher in relation to

controls. These two variables were significantly interre-

lated in the hopsitalized group, whereas no significant re-

lationship was found between these variables among control

group participants. These findings do not support much in

the way of speculation as to what, if any role veterans'

marital adjustment played in relation to various illness

variables. (Correlation coefficients were derived through

the point-biserial correlation procedure). On the other

hand, the incidence of family-marital life change emerged

as a significant correlate vis-a-vis the illness process.

Relationship Between Other Life Change Variables and Illness

Although the results outlined above concerning the

characteristics of family-marital functioning proved somewhat

year period prior to study; however, the Health History Inven-
tory extends to five years prior to the study, measuring the
number of different illnesses without regard to frequency,
duration, or date of onset. Therefore, the correlation be-
ween family-marital life change and health history must pre-
suppose a constant life change rate within both samples for
the entire five-year period. Because of the time limitation
built into the SRE, this premise cannot be tested, and hence
conclusions based on these correlations must be regarded as

illuminating, the findings associated with veterans' personal-

environmental life change appear to be most useful in terms

of relating life change to illness onset. For example, among

five correlations between each of the life change categories

and illness severity (see Table 6), the incidence of veterans'

personal-environmental change contributed the most statisti-

cally significant relationship. Also, when LCU scores within

each of the three specific life change categories (family-

marital, personal-environmental, and occupational-financial)

were compared between the two groups (Table 8), the larger

magnitude of change reported by hospitalized subjects in

the personal-environmental category attained a closer prox-

imity to statistical significance.

The direction of the relationship between veterans'

personal-environmental change and the immediacy of illness

onset was unexpected (see Table 10) in that higher rates of

reported change within this category were associated with

a more extended time span between the crisis period and the

onset of illness. This finding ran counter to the supposi-

tion that higher rates of change within any of the cate-

gories would be followed by a more rapid onset of illness.

One other relationship of note was found: between the in-

cidence of personal-environmental change and the extent of

previous illness (see Table 15). Personal-environmental

change among the chronically ill was linked to illness his-

tories, such that an individual reporting a wide diversity

of previous illness tended to report a correspondingly high

rate of life change in this area of functioning. The same

relationship among control group participants was not found

to be significant. Moreover, the incidence of personal-

environmental change among individuals in the hospitalized

group was more highly related to illness history than re-

ported change within the remaining four SRE life change

categories (see Table 15).

To summarize, life change in the personal-environmental

category has emerged as an important discriminating variable

in that high rates of change in personal habits, residence,

school, and living conditions during the two-year period

prior to study appeared to characterize the chronically ill

members of the study relative to control group counterparts,

a difference which closely approached statistical signifi-

cance. Furthermore, change of this nature among hospital-

ized subjects was significantly related to severity of the

ongoing illness and retrospectively related to the onset of

previous illnesses of other types. In contrast, healthy

individuals were characterized by relatively low rates of

personal-environmental change, and to the extent that such

change did occur within this group, it was not retrospective-

ly related to previous illness onset.

The Role of Occupational-Financial Life Change

Occupational-financial life change among veterans in