Group Title: relationship between psychosocial factors and response to medical treatment in chronically ill adolescent patients
Title: The relationship between psychosocial factors and response to medical treatment in chronically ill adolescent patients
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00098835/00001
 Material Information
Title: The relationship between psychosocial factors and response to medical treatment in chronically ill adolescent patients
Physical Description: viii, 147 leaves : ill. ; 28 cm.
Language: English
Creator: Reiss, John Gilbert, 1949-
Copyright Date: 1984
 Subjects
Subject: Chronically ill children -- Family relationships   ( lcsh )
Youth -- Diseases -- Psychological aspects   ( lcsh )
Chronic diseases -- Psychological aspects   ( lcsh )
Counselor Education thesis Ph. D
Dissertations, Academic -- Counselor Education -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis (Ph. D.)--University of Florida, 1984.
Bibliography: Bibliography: leaves 135-146.
General Note: Typescript.
General Note: Vita.
Statement of Responsibility: by John Gilbert Reiss.
 Record Information
Bibliographic ID: UF00098835
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000491174
oclc - 11960294
notis - ACQ9676

Downloads

This item has the following downloads:

PDF ( 8 MBs ) ( PDF )


Full Text










THE RELATIONSHIP BETWEEN PSYCHOSOCIAL FACTORS AND
RESPONSE TO MEDICAL TREATMENT IN CHRONICALLY ILL ADOLESCEW VATIE~IS










BY

JOHN GILBERT REISS
















A DISSERTATION PRESENT) 'I THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY




UNIVERSITY OF FLORIDA




























To

Beverly

and

Molly

















ACKNOWLEDGMENTS


This study would have not been possible without the support,

encouragement, guidance, and cooperation of many faculty members,

physicians, and allied health professionals. I would like to thank Dr.

Franz Epting for introducing me to General Systems Theory, and Dr. Harry

Grater for encouraging me to dare to conduct research on Family Systems.

I would like to acknowledge Dr. Shea Kosch who gave freely of her time

and ideas, and who helped formulate the study. I would also like to

thank Dr. David Suchman who through word and deed never let me forget

that all things really are interconnected. Dr. Jaquelyn Resnick not

only helped to transform my theoretical fantasy into a concrete study,

but she also used her talents as a therapist to help me through the

darkest days of dissertation doldrums. Without the guidance,

structure, and patience of Dr. Joe Wittmer, I might never have

finished.



Thanks is also due to the many pediatricians at Shands Teaching

Hospital and Clinics, who trusted me to work with their patients. I

would like to especially thank Drs. Joel Andres and Donald George, who

actively encouraged their patients to participate in the study; and

Dr. John Graham-Pole, whose ongoing support, encouragement, and interest










kept me going. Dr. Graham-Pole also allowed me to experience his magic

touch with children with cancer.

I would like to acknowledge Dr. Michael Resoick, Director of

Children's Developmental Services. He was a boss who remembered what it

was like tackling too big a project; he reminded me about priorities and

made my job flexible enough to allow me to finish.

I would also like to acknowledge Dr. Randy Carter who guided me

through the complexities of discriminant analysis.

To Beverly Posa, my partner in life and love, I cannot express my

full measure of appreciation.

To my nine-month old daughter, Molly, thanks for reminding me about

the true wonders of the world.

To the families, who took the time to help me during their own time

of need my thanks, respect, and best wishes.












TABLE OF CONTENTS

VAG;9

ACKNOWLEDGMENTS ................................ ... jii

ABSTRACT......................... ............... vii
CHAPTER
I. INTRODUCTION.......................................

Rationale for the Study .........................2
Statement of the Problem.........................5
Definition of Terms .................... ......... 6
Organization of the Remainder of the Study......8

II. REVIEW OF THE LITERATURE............................. 9

Introduction ................................... 9
General Systems Theory Paradigm................10
Closed Systems ..........................11
Open Systems............................12
Health and Disease ....................14
The Biomedical Model ...........................17
The Psychosomatic Model....................... 18
Personality Characteristics.............18
Psychodynamic theory...............18
Psychophysiological theory..........19
Summary............................ 20
Psychosocial Factors....................21
Quality of life....................21
Quantity of life change............23
Social support ..................... 27
Family membership.................30
Summary............................32
The Family-Systems Model.......................35
Summary....................................... 43

III. METHODOLOGY.........................................45
Subjects....................................... 46
Hypotheses.................................... 47
Instrumentation ................................ 49
Family Adaptation and Cohesion
Evaluation Scales..................49
Family Functioning Index................52
Family APGAR............................ 54
Schedule of Recent Events...............55
Life Events Record ......................56
A Short Scale for the Evaluation of
Social Support..................... 57
Physician's Form for Rating Level of
Response to Medical Treatment......59
Procedures .................................... 60









CHAPTER PAGE

IV. RESULTS............................................ 62

Data Transformations ...........................62
Rating of Level of Response to Medical
Treatment................................,67
Sample Characteristics........................,68
Disease Group Characteristics..................,70
Distinguishing Among the Three Levels of
Medical Response by Using All the
Predictor Variables.......................79
Relationship Between Level of Medical
Response and Family Functioning...........88
Relationship Between Level of Medical
Response and Quantity of Life Change.....91
Relationship Between Social Support and
Quality of Response to Medican Treatment.92
Interrelationship Among Life Stress and
Social Support and the Quality of
Response to Medical Treatment.............83

V. DISCUSSION............................................94

Discussion of Results ..........................98
Limitations.................................. 100
Recommendations for Further Study.............104
Summary ....................................... 106

APPENDICES

A FAMILY ADAPTABILITY AND COHESION EVALUATION SCALES...109

B FAMILY FUNCTIONING INDEX............................. Ill

C FAMILY APGAR.......................................... 115

D SCHEDULE OF RECENT EVENTS............................116

E LIFE EVENTS RECORD. .................................. 118

F A SHORT SCALE FOR THE EVALUATION OF SOCIAL SUPPORT...120

G PHYSICIANS FORM FOR RATING QUALITY OF RESPONSE
TO MEDICAL TREATMENT.................................123

H LETTER TO RESEARCH FAMILIES ..........................124

I INFORMED CONSENT FORM.................. ..............125

J RESULTS OF ANOVA'S FOR GENDER AND RACE MAIN EFFECTS..127

REFERENCE NOTES..............................................134

BIBLIOGRAPHY.................................................135

BIOGRAPHICAL SKETCH. ......................................... 147
















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



THE RELATIONSHIP BETWEEN PSYCHOSOCIAL FACTORS AND
RESPONSE TO MEDICAL TREATMENT IN CHRONICALLY ILL ADOLESCENT PATIENTS

By

John Gilbert Reiss

April, 1984

Chair: P. Joseph Wittmer
Co-chair: Jaqueline Resnick
Major Department: Counselor Education

The present study was an attempt to examine the relationship

between family and psychosocial factors and the quality of response

of chronically ill adolescents to medical treatment. Previous

research has generally supported the thesis that the development and

course of physical illness is related to the following psychosocial

factors: family functioning and structure, life stress, and social

support. The primary purpose of the present study was to determine

if, by assessing these three factors, it was possible to

differentiate among chronically ill adolescents whose response to

medical treatment was better than expected, those whose response was

about as expected, and those whose response was worse than expected.

The secondary purpose of this study was to determine if "worse

response" adolescents were from dysfunctional families, and/or had









experienced high levels of stress, and if social support moderates

the adverse effects life stress has on health.

Data was obtained from families each having an offspring (age

14-19) with one of the following four types of chronic disorders;

pulmonary (N=21), gastroenterological (NN13), cancer (N=11), and

juvenile rheumatoid arthritis (N=3). Parents from each family were

administered the Family Adaptation and Cohesion Evaluation Scales

(FACES), the Family Functioning Index (FFI), the Family APGAR

(APGAR), the Schedule of Recent Events, and A Short Scale for the

Evaluation of Social Support (ASSESS). Adolescent patients were

administered FACES, APGAR, ASSESS, and the Life Events Record.

Results indicated that it was possible, using a discriminant

analysis, to distinguish among adolescents in the sample from the

three medical response groups. However, the jackknife validation

procedure indicated that given a new sample population, the

discriminant function derived from adolescents' data would identify

members of the "as expected" response group, but would not

differentiate members of the "worse" or "better" response groups.

The validation procedure indicated that the discriminate functions

derived from mothers' and fathers' data would not differentiate among

any of the response groups.

The results did not support the hypotheses that medical response

is associated with family functioning, that life stress is associated

with poor medical response or that social support moderates the

adverse effects stress has on health.












CHAPTER I
INTRODUCTION



Determining an accurate prognosis and providing optimal care for

chronically ill children has long been a significant problem for the

medical profession (Engel, 1962; Apley and MacKeith, 1973; Weakland,

1977). Recent studies conducted within the disciplines of

epidemiology, psychology, sociology, and anthropology, as well as

within the primary care disciplines in medicine, indicate that the

family unit and the social environment play a significant role in

both the onset of childhood disease and the response of the child to

medical treatment (Cassel, 1976; Schmidt, 1978). However, these

factors are often not taken into account in diagnosis and treatment

(Jaffe, 1978; Schmidt, 1978). This is due, in part, to two factors:

(a) the lack of a general framework for integrating existing medical

knowledge with the new data on family and psychosocial factors in

disease (Brody and Sobel, 1979), and, (b) the lack of a reliable,

efficient, and integrated instrumentation for obtaining clinically

relevant data on family and psychosocial factors (Pless and

Satterwhite, 1975).

This study addresses these two problems in the following manner.

First, this study presents evidence which demonstrates that General

Systems Theory provides a framework for the comprehensive study of

physical illness. Second, the relevant literature is reviewed and










critiqued from the perspective of General Systems Theory, and the

significant psychosocial variables are identified. However, the

basic purpose of this study is to determine the utility of a set of

psychosocial assessment tools in differentiating among chronically

ill adolescents patients whose response to medical treatment is

better than expected, those whose response is about as expected, and

those whose response is worse than expected.



Rationale for the Study



Theories of disease have changed a great deal over the centuries

and differ across cultures, being determined by prevailing views of

human nature and the relationship of humankind to the cosmos (Dubos,

1965). At times emphasis has been placed on the whole person and

her/his relationship to the physical, psychological, and social

environment, while at others the focus has been on fragments of human

nature, such as the mind or the component parts of the "body

machine". Within the former perspective, disease is seen as a

process which is inseparable from the person-environment interactive

system. Within the latter ontologicalal") view, disease is

conceptualized as a specific entity which is essentially unrelated to

a person's personality, bodily constitution, style of life, or

environment (Dubos, 1965).

In prescientific medicine, the ontological doctrine took the

form of demonological concepts with disease being regarded as the

result of malevolent influences of taboo violations, sorcery,

vengeful ghosts, hostile ancestors or animal spirits (Dubos, 1965).











In modern times, the ontological doctrine is still influential,

Patients are prone to blame their illnesses on something they

"caught", they ate, or that happened to them, or to account for their

disease in terms of punishment. Further, the dominant theoretical

model of modern scientific medicine, the biomedical model, is

compatible with the ontological perspective. This model assumes that

all aspects of human illness are the result of specifiable chemical

and/or physical influences. Physicians find this model attractive,

since it allows them to see the "cause" of all disease as something

which can be changed physically, through surgery, or attacked and

destroyed through chemical interventions (Brody and Sobel, 1979).

However, within this perspective, consideration of the mind and the

personal and psychosocial dimensions of illness are neglected or are

changed into biochemical terms (Engel, 1977).

In the past two decades, the significance of personal and

psychosocial factors in human health and disease has been clearly

demonstrated. Considerable evidence has been accumulated which

indicates that feelings of helplessness, hoplessness, and unresolved

grief, generate or aggravate many illnesses (Engel, 1962; Schmale,

1958; Engel and Schmale, 1967; Wolff, 1968. Other research indicates

that the stress involved in adjusting to a rapidly changing social

environment may lead to or exacerbate a variety of physical disorders

(Cohen, 1979; Rahe, 1972). Apley et al. (1977) estimates that

psychosocial factors play a part in 45 per cent of all hospital

admissions of children, and are the chief reason in another 15 per

cent.










While research on the relationship between family interaction

and illness is limited, this factor is considered by some medical

researchers to be of preeminent importance in understanding the

disease process (Meissner, 1966, 1974; Grolnick, 1972; Apley and

MacKeith, 1973; Minuchin et al., 1975). According to Schnidt (1978)

knowing what is "going on" in the family is as important as detailing

the individual's symptoms. It is his belief that medical care could

be both more humane and more effective, in terms of outcome and cost,

if the providers of that care would consider the complex interactions

that occur between the individual patient and her/his psychosocial

environment. Further, a variety of physical disorders, including

anorexia nervosa (Palazzoli, 1974; Minuchin et al., 1979),

superlabile diabetes mellitis (Minuchin et al., 1979), intractable

asthma (Liebman et al., 1974; White, 1979), and non-organic abdominal

pain (White, 1979; Apley and Hale, 1973) have been successfully

treated through family therapy.

While the research on the relationship between family function-

ing and the onset and course of disease has yielded promising

results, it has failed to stimulate other investigators to enter the

field. Weakland (1977), a prominent family systems theorist, has

identified the area of physical illness and disease as a "neglected

edge" of family systems research. Family Process, the flagship

journal of family systems research, published only two articles

related to physical illness in its first ten years (1965-1975) of

existence. There have been only three empirical research studies

published in this journal on this topic since Weakland's (1977)

article suggesting the need for further research. The developments










in psychosocial and family systems research have also failed to

influence established researchers in the medical field. The major

medical journals concerned with psychosomatic medicine, the Journal

of Psychosomatic Medicine, Psychosomatics and Psychosomatic Medicine,

together contain only one article which adopts a family systems,

rather than an individualistic or dyadic orientation to the

psychological and psychosocial factors in disease.

Thus, while there is considerable evidence which indicates that

family and other psychosocial factors play a significant role in the

onset and course of physical illness, these factors have remained, on

the whole, outside the main channels of medical thinking and

experimentation.



Statement of the Problem



This study investigates the relationship between family

interaction, life stress, and social support, and the response of

chronically ill adolescents to standard medical treatment. More

specifically, this study attempts to answer the following questions:

1. Is it possible to differentiate among adolescent patients

whose response to nmdical treatment is better than expected, those

whose response is about as expected, and those whose response is

worse than expected by using data on family interaction, life change

and social support.

2. Is the quality of response of chronically ill adolescents to

medical treatment related to the quality of family interaction?










3. Is the quality of response of chronically ill adolescents to

medical treatment related to the quantity of life stress experienced

by the adolescent and/or other family members?

4. Is the quality of response of chronically ill adolescents to

medical treatment related to the level of social support experienced

by the child and/or other family members?



Definition of Terms



Adaptability: The ability of a marital or family system to change

its power structure, role relationships, and relationship rules

in response to situational and developmental stress.

Adaptation: A dynamic balance between the processes of homeostasis

and mirphogenesis.

Causality, circular: The property of living systems in which

information is processed. For example, information moves from A

to B; from B' to C; from C' to D; from D'to A; from A' to B'' ;

from B'' to C' etc. Each link in modified by the interaction,

and the interaction involves a feedback loop (D' to A).

Causality, linear: A property of closed systems, in which a fixed

quantity of energy is distributed through the system causing

a fixed energy output. For example, energy moves along a chain

from A to B; from B to C; and from C to D.

Closed system: a non-living system.

Cohesion: The emotional bonding family members have with one another

and the degree of individual autonomy a person experiences in

the family.










Disease: The failure of a living system to respond adaptively to

environmental challenges.

Enmeshment: A property of family interaction in vhich there is a

high degree of responsiveness to, involvement with, and inter-

dependence on family relationships; a lack of personal privacy;

poorly differentiated interpersonal perception, and "excessive

"togetherness" and sharing.

Equifinality: The ability of living systems to reach the same final

state from different initial conditions.

General Systems Theory: A paradigm developed specifically for the

study of living organisms (systems).

Health: The ability of a system to respond adaptively to a wide

variety of environmental challenges.

Hierarchical organization: General Systems Theory principle Niich

holds that living systems are organized along ordered and highly

structured lines, with clearly identifiable differential levels

of conplexity that relate in logical fashion one to another.

Homeostasis: The ability of living systems to maintain a dynamic

steady state.

Isomorph: A principle of dynamic interaction or interrelationship

which in characteristic of living systems in general.

Mathematico-reductionistic paradigm: The underlying assumptions of

scientific method; the assumptions are that all phenomena can be

(a) reduced into causal elements; (b) adequately described

in terms of mathematical equations and laws; and (c) understood

in terms of linear causality.

Morphogenesis: The ability of living systems to grow and change.











Omnipotentiality: The ability of living systems to reach different

final states from the same initial conditions.

Open system: A living system.

Overprotectiveness: A property of family interaction in which there

is a high degree of concern for family members' welfare,

Paradigm: The set underlying assumptions of a method of inquiry,

Rigidity: A property of family interaction in which there is a

heavy commitment to maintaining the status quo.

System: A set of units or elements standing in some consistent

relationship or international stance with each other.

Wholeness: General Systems Theory principle which holds that the

behavior of a living system cannot be fully understood apart

from its context or environment; nor can it be totally explained

in terms of the behavior of its component parts.



Organization of the Remainder of the Study



The remainder of this study is organized into four chapters.

The second chapter is a review of the related literature. Topics

covered in this section include the basic principles of General

Systems Theory and a review of research on the cause, course, and

effects of physical illness based on the biomedical, psychosomatic,

and family-systems models of disease. The third chapter presents the

research methodology. The fourth chapter contains the results of the

study. In the fifth chapter the study is summarized, the results and

implications are discussed, and suggestions are made for further

research.












CHAPTER II
REVIEW OF THE LITERATURE



Introduction



Kuhn (1970), a leading authority on the history of science,

states that all scientific inquiry is conducted within a specifiable

scientific paradigm. This paradigm, or disciplinary matrix, is the

set of underlying assumptions which determine how the phenomena in

question are to be viewed and studied, what questions are asked and

how they are posed, the possible methods by which the questions can

be answered, the preferred models, analogies, and metaphors, and what

will be accepted as an explanation. Kuhn also states that most

researchers fail to identify the paradigm which underlies their

inquiry.

The paradigm adopted by this investigator is General Systems

Theory. In the first section of this review, the underlying

assumptions and basic principles of General Systems Theory are

outlined, and a model of disease, based on this paradigm, is

presented. In the following sections, literature concerning the

cause, course, and effects of physical illness based on the

biomedical, psychosomatic, and family-systems models of disease is

discussed. This literature is also critiqued from the perspective of

General Systems Theory.











General Systems Theory Paradigm



General Systems Theory was developed in the 1920s and 1930s ag

a reaction against the then dominant mathematico-reductionistic

paradigm of scientific research. The basic assumptions of the

nathematico-reductionistic paradigm are that all phenomena can be

(a) reduced or broken down into essential isolatable causal chains,

elements, or units, (b) adequately described in terms of mathematical

equations, (c) adequately described in terms of precise mathematical

laws, which hold invariably true under specifiable "standard

conditions", and (d) understood in terms of linear causality

(Bertalanffy, 1968; Steinglas, 1978; Wood, 1974). The method of

inquiry employed in this paradigm is the analytic method. In

simplest terms, the analytic method can be described as follows: the

experimenter holds all factors constant but two, the independent

variable (IV) and the dependent variable (DV), then systematically

varies the IV and observes the effect of this systematic variation on

the DV. By means of manipulations of this sort, the experimenter

seeks to observe situations in a controlled manner, obtain clear,

unambiguous results, and thereby determine the true nature of the

phenomena under study (Giorgi, 1973).

Historically, this method has been most successfully employed by

the natural or "hard" sciences (physics, chemistry, etc.) in which

the phenomena under study can be carefully and closely controlled.

In the life or "soft" sciences (psychology, sociology, biology, etc.)

the phenomena under study do not lend themselves to rigorous control.

Sophisticated research designs and statistical methods of data










analysis have therefore been developed to compensate for this lack of

rigorous control (Giorgi, 1973).

Bertalanffy (1952, 1967, 1968, 1972) holds that the mathematicp-

reductionistic paradigm and the analytic method are inadequate for

the study of living systems, since such systems are destroyed when

broken down into component parts. Furthermore, he proposes a new

model or paradigm for the study of organic, living systems; one which

focuses on the general overriding principles (isomorphs) which

characterize these systems. A detailed description of this new

paradigm, known as General Systems Theory follows.

Within General Systems Theory, phenomena are conceptualized in

terms of systems or "sets of units or elements standing in some

consistent relationship or interactional stance with each other"

(Bertalanffy, 1968, p. 38). All systems can be classified as either

"open" or "closed".



Closed Systems

The behavior of all closed systems has the following

characteristics: (a) they follow the Second Law of Thermodynamics

(i.e., proceed toward a state of maximum entropy, a time independent

state of equilibrium and disorder), (b) the final state is completely

determined by its initial conditions, and any change in these

conditions causes a totally predictable change in the end state, (c)

all reactions are completely reversible (i.e., a reversal results in

a return to the initial conditions), and (d) they can be completely

isolated from the environment, and do not need to exchange energy

(e.g., information, heat, etc.) with the environment in order to










exist and persist. Closed systems are in accordance with the basic

assumptions of the mathematico-reductionistic paradigm, and are

subject to study by means of the analytic method, All closed system'

are, by definition, non-living.



Open Systems

The behavior of open systems is fundamentally different from

that of closed systems, and can be understood only in terms of the

following principles of dynamic interaction and interrelationship

(isomorphs).

1. Systems follow the principles of hierarchical organization

and wholeness. Systems are organized, one to another, into a series

of hierarchical levels. Every system is itself composed of component

subsystems of smaller scale, and is, in turn, a component of a larger

system. In closed systems, the behavior of suprasystems can be

directly inferred from the combined behavior of subsystems. In open

systems, each system within the hierarchy constitutes a functional

whole and has unique properties. Thus, an open system cannot be

adequately understood or totally explained in terms of the behavior

of its component parts. The basic character of an open system

transcends its components, and belongs to a higher order of

abstraction. Similarly, no single element or group of elements

within an open system can act independently.

2. Open systems can reach the same final state from different

initial conditions. This is the principle of equifinality. In

addition, different final states can be reached from the same initial

conditions. This is the principle of omnipotentiality. From the










General Systems Theory perspective, the historical chain of events

which may have preceded the present state of affairs is not seen as

being especially important in understanding a phenomena. Rather, the

focus is on mutual or circular causality, i.e., on critical elements

and on the contemporary relationships between these elements.

3. Open systems are able to maintain a dynamic stability of

subsystem properties or relationships within a fixed set of reference

points. This steady state is maintained despite the continuous flow

of both matter and energy through the system. As was demonstrated by

Cannon (1939), organisms, in order to survive, maintain an internal

dynamic steady state of critical biological functions, such as

temperature, and electrolyte concentration. This process, which may

involve the modification of the external, as well as the internal

environment, is known as homeostasis. When this process involves a

modification of the external environment, it is often referred to as

assimilation (Piaget, 1971; Piaget and Inhelder, 1969; French, 1979).

4. Open systems are able to maintain sufficient closeness among

subsystems and components to enable them to interact and to resist

forces which could disrupt the system as a whole (i.e., homeostasis).

This is the principle of cohesion.

5. Open systems have the ability to develop a higher order of

complexity (i.e.. to grow and change); to increase hierarchical

organization and complexity of structure. This process, known as

morphogenesis, involves the ability of a system to shift its

fundamental reference points or parameters with respect to which an

organism maintains its homeostatic balance (French, 1979). It is










analogous to the concept of accommodation (Piaget, 1971; Piaget and

Inhelder, 1969).

6. Optimally functioning open systems achieve a state of

adaptation, a dynamic balance between the processes of homeostasis

and morphogenesis and are therefore capable of maintaining themselves

within a wide range of environmental conditions. Open systems which

follow the principles of homeostasis and morphogenesis are living

systems.



Health and Disease

Based on the General Systems Theory model of living systems,

Brody and Sobel (1979) propose that "health" is the "ability of a

system (for example cell, organism, family, society) to respond

adaptively to a wide variety of environmental challenges (for

example, physical, chemical, infectious, psychological, social)"

(p. 93). Thus, from the General Systems Theory perspective, health

is a positive process, and is not merely the absence of the signs and

symptoms of disease. This definition is not restricted to biological

fitness or somatic well being, but rather, involves a consideration

of the broader environmental, socio-cultural, and behavioral

determinants of health. Further, health is seen as a dynamically

changing state; encounters with environmental forces result in either

a lower level of health, a restoration of equilibrium, or a

growth-enhancing response.

Brody and Sobel (1979) propose that "disease" is the failure

of a living system to respond adaptively to environmental challenges.

Since all levels within a living system are interconnected, it is











expected, within the General Systems Theory paradigm, that a

pathological disruption is not limited to one level of a system, but

rather,



the disruption will tend to spread up and down in
the hierarchy. For example, in diabetes, genetic and
environmental factors interact to produce an initial
disruption at the biochemical level that can lead to
pathological changes in cellular function and a
disruption of organ systems (for example, kidney and
eye). Such changes are likely to disrupt the
individual's behavior and may strain the family as
well as produce a potential resource drain of the
community. A disruption can also travel downward
through the hierarchy, as when economic or natural
disasters produce societal disruptions creating
upheavals in community and family function that, in
turn, precipitate a variety of psychosomatic or
sociosomatic symptoms among individuals.
Therefore, from a systems view diseases are not
regarded as discrete entities localized in one organ
or tissue but as patterns of disruptions manifested
at various levels of the system at various times.
Patterns may differ in regard to where the
disruption arises, which hierarchical levels are
most affected, the type of environmental force that
initiated the disturbance, and so on....(Brody and
Sobel, 1979, p. 94)



From within the General Systems Theory paradigm, there are two

complementary ways of intervening in a system's pathological process

(Brody and Sobel, 1979). The first approach involves active invasive

therapeutic interventions, either chemical or surgical. In systems

terms, this approach involves a "disruption from the environment

designed to oppose a specific disease-disruption, as when antibiotics

are used to treat bacterial infections. The difference between a

therapeutic disruption and a disease-producing disruption lies in the

value of the ...(expected) outcome of each" (Sobel and Brody, 1979,

p. 95).










The second therapeutic approach is aimed at strengthening the

natural ability of an organism to adapt. In systems terms, this

approach involves attempts to improve the information flow in the

system in order to accommodate disruptions and facilitate the

restoration of equilibrium. Since disease most often involves

multiple levels, disrupting the person and the social group, multiple

interventions directed at different levels can be therapeutic.

Improving feedback and communication among family members, through

family therapy, may stabilize the hierarchy at that level, rendering

the family system more capable of handling challenges and resisting

disruption, and potentially bringing about an improvement in the

physical condition of a symptomatic family member. The work of

Simonton and Simonton (1975) with cancer patients illustrates this

approach. Standard biological therapies (radiotherapy, chemotherapy,

and surgery) are combined with adjunctive support at the person level

(various meditation and relaxation exercises) as well at the family

level (group work and counseling). "While diseases may represent

patterns of disruption affecting many hierarchical levels, a therapy

aimed at just one level may be highly efficacious because it can

affect other levels via the interconnected patterns of information

flow" (Brody and Sobel, 1979, p. 96).

In the following three sections, literature concerning the

cause, course, and effects of physical illness, as based on the

biomedical, the psychosomatic, and the family systems models of

disease, is presented and critiqued from the perspective of the

General Systems Theory paradigm.










The Biomedical Model



The biomedical model, which is based on the mathematcqo

reductionistic paradigm, holds that all disease processes can be

fully accounted for in terms of deviations from the norm of a

specifiable set of measurable biochemical variables (Weil, 1973;

Engel, 1977). Within this model, disease in understood to be a

discrete "thing" which is separable from its host and is capable of

existing independently of it. This model proposes that all

infectious illnesses are caused by bacteria and viruses, whose

appearance correlates closely with other physical manifestations of

illness (the "germ theory"). Further, it is held that the specific

bacterial or viral cause of all illnesses can be identified through

the analytic method. Since the biomedical model defines and

identifies illness exclusively in terms of specific somatic and

biochemical variables, it excludes social, psychological, and

behavioral factors from the explanation of illness (Engel, 1977).

From the perspective of General Systems Theory, the biomedical

model is conceptually inadequate, since it proposes a closed systems

model to describe disease processes even though these processes

behave like open, living systems. The open systems character of

disease processes is illustrated by the fact that, rather than

following the rules of simple linear causality, most pathological

states, as they naturally occur, are the consequence of numerous

factors acting simultaneously (Dubos, 1965). Further, in accordance

with the open systems principle of omnipotentiality, noxious agents

can express themselves in a great variety of different pathological










states. In accordance with the open systems principle of

equifinality, different agents can elicit similar reactions.

Finally, in accordance with the open systems principles of wholeness

and hierarchical organization, a disease cannot be separated from its

host; such a separation, itself, constitutes a pathological state

(Dubos, 1965; Engel, 1977; Weil, 1973; Weiner, 1977; Brody and Sobel,

1979).



The Psychosomatic Model



In this section, research conducted under the psychosomatic

model of disease is discussed. The psychosomatic model of disease

holds chat mind and body are an inseparable and integrated whole, and

that psychological and/or social, as well as biological factors, are

significant in the development, course, and outcome of physical

disorders (Lipowski, 1975). The studies are divided into two broad

categories; those which focus on the identification of personality

characteristics associated with specific illnesses or with illness in

general, and those which correlate the incidents and course of

disease with conditions of, and changes in, the social environment.



Personality Characteristics

The studies in this category are divided, according to their

theoretical orientation, into the following two sections:

psychodynamic and psychophysiological.

Psychodynamic theory. Exemplary of the psychodynamic approach

is the work of Alexander (1950). This researcher sought to identify










predisposing factors involved in the initiation and maintenance of

disease by analyzing clinical data produced in the course of

psychoanalytic treatment and/or the study of patients with cllrqnic

organic ailments in which emotional conflict was thought to play an

etiological role. Based on this data, Alexander proposed that the

following three factors are involved in the onset of certain

psychosomatic disorders: (a) a specific psychodynamic constellation

or unconscious conflict (the "visceral neurosis"), (b) a specific

"onset situation" which activated the unconscious conflict, and (c) a

constitutional (genetic) vulnerability of a specific tissue or organ

system, which was designated the "X" factor. Alexander held that

disease developed only when all three factors were present and active

in the appropriate combination. Alexander's observations of patients

have been supported as valid descriptive findings by other

investigators (Mirsky, 1958; Weiner, 1970; Dongier et al., 1956;

Wallerstein et al., 1965). However, there is no clear, consistent,

empirical evidence to support Alexander's contention that the

psychodynamic factors which he identified play a primary causative

role in the onset and course of the seven disorders which he

investigated (Reiser, 1975; Weiner, 1977; Wittkower, 1974).

Psychophysiological theory, This approach to the study of

somatic illness was developed by Wolff (1968) and his colleagues.

These researchers focused on personality features and behaviors that

were directly observable or measurable and that pertained primarily

to conscious layers of a patient's personality and life experience.

These researchers made psychological observations simultaneously with

measurements of the physiological functioning of affected organ











systems. Based on their multi-method studies, Wolff proposed thai

illness is the consequence of a patient's perception of environmental

situations as threatening to life itself or to emotional security,

In the face of the received threat, the patient is hypothesized to

protect and defend her/himself with an "organismic" response. The

specific organ involved in the defensive response was said to be

determined by the nature of the stress, and by the nature of the

organ's functions.

Wolff proposed that the perception of threat is associated with

an increase in risk for becoming ill with some kind of disease.

Grace (1950) and Graham et al. (1962) expanded this formulation,

proposing the "specificity of attitude" hypothesis. This hypothesis

states that there is an association between a given disease and a

specific attitude toward the life events) which first precipitates

and later exacerbates the illness; that the attitude is different for

each disease, and all persons with a given disease have the same

attitude (Graham et al., 1962). Attitude is defined by these

theorists in terms of how the person perceives her/his position in

the situation, and what, if any action s/he wishes to take.

Summary. The linear cause and effect models proposed by the

psychodynamic and psychophysiological theorists have been widely

criticized as being conceptually inadequate and methodologically

flawed (Reiser, 1975; Lipowski, 1977; Mirsky, 1957; Weiner, 1977;

Engel, 1960). Because of a lack of predictor variables for disease,

researchers were not able to select a relevant subject population

prior to the onset of disease, and therefore were not able to conduct

prospetiv ud hopecive ropoetive staiudie, tha role of










personality factors and associated physiological functioning in the

etiology and course of disease cannot be demonstrated empirically

(Weiner, 1977; Reiser, 1975).

Some studies conducted by these researchers demonstrated that

patients with certain disorders resemble each other more than they

resemble members of the population as a whole, or patients with other

types of disorders. However, given even detailed accounts of a

patient's personality, experts have not been able to predict with any

degree of confidence and reliability, what disease, if any, a patient

might have (Engel, 1955).

From the perspective of the General Systems Theory paradigm, the

models proposed by the psychodynamic and psychophysiological

theorists are conceptually inadequate since they clearly are not in

accordance with the open systems principles of omnipotentiality,

equifinality, hierarchical organization, and wholeness.



Psychosocial Factors

Research studies which examine the relationship between the

psychosocial environment and the onset and course of illness can be

divided into four broad categories: those which focus on specific

traumatic life events and the quality of life, those which evaluate

the quantity of life change, those which focus on social support, and

those which look at family membership.

Quality of life. The most prominent theory in this category of

psychosocial research is that of "object loss". This theory, which

has been most clearly articulated by Engel and Schmale (Engel, 1968;

Engel and Schmale, 1967; Schmale, 1972), holds that feelings of










bereavement, depression, helplessness, and hopelessness, which occur

in persons who experience actual, threatened, or symbolic loss, are

often associated with an attitude of "giving up". This attitude is

hypothesized to be associated with a basic biological response state

("conservation withdrawal"), which acts in a non-specific manner to

render an organism less resistive to existing somatic predispositions

for illness or to external pathogenic factors.

Some researchers who have tested this theory of object loss

have focused on feelings of hopelessness and helplessness.

Representative of this line of research is the series of predictive

studies (Schmale and Iker, 1966, 1971) which followed patients who

were given diagnostic cone biopsies because of repeated evidence of

suspicious cells, but who were asymptomatic for cervical cancer.

Patients who reported real or apparent loss and/or feelings of

hopelessness were found to be significantly more likely to contract

cervical cancer.

Other researchers have focused on the impact of specific loss

events. For example, in studies on the impact of the death of a

spouse, it was found that widows retrospectively report a significant

increase in minor physical illness, when compared with similar

individuals who had not lost a spouse (Maddison and Viola, 1968;

Parks et al., 1969; Parks and Brown, 1972). Other studies report an

increase in mortality among widows and widowers in the six month

period following the death of their spouse (Ekblom, 1963; Young et

al., 1963; Jacobs and Ostfeld, 1977; Rowland, 1977).

Loss and separation have also been found to be associated with

the onset of lung cancer (Kissen, 1967), rheumatoid arthritis (Engel,











1969), and ulcerative colitis (Engel, 1955). However, studies of

American soldiers during World War II and concentration camp victims

(Wolff, 1968), populations under military occupation (Malmarps,

1950), occupants of London during the "blitz" (Glover, 1940), and

Hungarian refugees (Hinkle et al., 1958) report finding no

significant relationship between loss or separation and morbidity or

mortality. Thus, the findings of this line of research are

inconclusive. This suggests that object loss in and of itself is

neither a necessary nor a sufficient condition for illness onset;

that loss may play a role in some cases of disease and death; and

that the effects of loss may be moderated by other factors (Rowland,

1977; Cohen, 1979).

Quantity of life change. This theory holds that life change per

se, regardless of the desirability of the change, is associated with

illness onset and exacerbation. The most prominent life change model

of disease is that formulated by Holmes and Rahe (1967a). This model

proposes that life events cause an increase in physiological activity

which, over time, has a wearing effect on the body, lowers body

resistance, and enhances the probability that a disease will occur.

Thus, a direct link between life change and illness onset is

hypothesized.

In order to test their theory, Holmes and Rahe (1967a) first

developed the Social Readjustment Rating Scale. Through this

instrument, they determined the relative amount of psychological

readjustment (intensity and length of time) necessary to adjust to

each of 43 life events (e.g., divorce, death of spouse, change job).

In their research on the connection between life stress and disease,










Holmes and Rahe used the Schedule of Recent Events (SRE) (Holmes and

Rahe, 1967a), which contains the same 43 life events. On the SRE,

subjects are asked to document the occurrence of the life event item

over a specific period of time (usually 6 months). By adding the

life change value of each life event, as determined through the

Social Readjustment Rating Scale, a quantitative score, in life

change units (LCU's), can be determined for each subject.

Research employing the SRE in the study of a variety of

populations and diseases indicates that high life change scores

(scores over 450) are associated with changes in health. Individuals

with the highest scores have been found to demonstrate the most signs

and symptoms, with even minor health changes being closely related to

events requiring adaptive behavior (Petrich and Holmes, 1977). The

following studies are representative of this line of research.

Jacobs and Charles (1980) in a study of children with leukemia,

and Heisel (1972) in a study of children with juvenile rheumatoid

arthritis, found that, for the year prior to disease onset, these

chronically ill children had significantly higher LCU scores than

physically healthy comparison groups. Prospective and retrospective

controlled studies of deaths from myocardial infarction (heart

attack) have shown that those patients who died had significantly

higher LCIJ scores in the 6 months prior to infarction compared to

those who survived (Rahe and Lind, 1971; Theorell and Rahe, 1972,

1975; Theorell et al., 1975). Stevenson, Nabseth, and Masuda (cited

in Masuda and Holmes, 1978) found patients with duodenal ulcers had

high LCU scores prior to needing surgery; and four years after

surgery, patients with higher postoperative LCU scores had











significantly more residual symptoms than those with lower LCU

scores. Allen (cited in Masuda and Holmes, 1978) in i study of

patients with pulmonary tuberculosis, found that those patients

suffering a relapse had significantly higher scores than those who

did not suffer a relapse. One half of the relapse group had LCU

scores over 450 (indicating a major life crisis). In a series of

prospective studies of 5000 Navy personnel (Rahe, 1968, 1972, 1974;

Ruhin et al., 1969; Rahe et al., 1970), it was found that those men

with the highest LCU scores for the 6 months preceding a sea cruise

were found to seek significantly more medical care than those men

with the lowest LCU scores.

When first proposed, the SRE and the life stress theory of

disease gained great popularity among researchers in psychosomatics.

Recently, however, the model has been criticized on methodological

and theoretical grounds. On theoretical grounds, Cleary (1974) has

questioned whether LCU values accurately represent the pathogenic

significance of life events, and whether the effects are additive.

Further, as different life events produce different physiological

responses, Cleary questions the validity of a unidimensional life

event scale. While Holmes and Rahe (1967b) suggest that all life

events, whether positive or negative, increase the probability of

disease, Vinokur and Seizer (1975) have found that the undesirable

events of the SRE are the most strongly correlated with the onset of

illness symptoms. Cohen (1979) notes that while significant results

have often been found between LCU scores and illness onset, the

magnitude of the relationship has often been small. In a Navy study

(Rahe, 1974), for example, the correlation was low (r-0.12). While










this is significant (p<.05) in this large sample (S's=5000), the LCU

scores accounted for less than 2% of the total variance. In some

studies (Rahe et al., 1970; Rahe and Arthur, 1978), the correlation

between illness and demographic and occupational factors was higher

than that between illness and LCU scores.

Based on the fact that some people become ill or are

hospitalized when no discernable changes in their lives have

occurred, while others undergo many severely stressful events without

developing any illness, Wershow and Reinhart (1974) conclude that the

life stress model is incomplete. These authors suggest that coping

factors, such as coping style and social support, play a significant

role in moderating the effects of life stress.

In response to these criticisms, Rahe (1974) modified the

original life change model, which posited a direct link between the

quantity of life change events and the probability of disease onset.

The new model proposes that there is a sequence of several moderating

factors such as past experience, social support and other psycho-

social defenses, coping style and illness behavior which act to

increase or decrease the impact of a given life event.

Thus, based on the results of these studies and critiques, and

in accordance with the revised life change model, life change events

appear to play a role in the occurrence of many cases of disease.

However, high life change is neither a necessary nor a sufficient

condition for illness onset, and the effect of change may be modified

by other factors (Rowland, 1977; Cobb, 1976; Dean and Lin, 1977;

Murowski et al., 1978).










Social support. This model holds that a low level of social

support is associated with a higher incidence of disease, while a

high level of social support has a moderating effect on the impact

of stressful life events and is associated with a lower incidence of

disease.

A positive relationship between low levels of social support and

increased somatic symptomology has been reported by several studies.

In an epidemiological study of psychosomatic symptomology, Schwab et

al. (1979) found that, compared to asymptomatic individuals, persons

with psychosomatic complaints had more friends and relatives nearby,

but were much less likely to utilize their support system by sharing

problems or by asking for help in times of crisis. These researchers

concluded that a relative lack of a meaningful support system is a

common characteristic of the psychosomatically ill.

In a study of the relationship between social support and

mortality, Berkman and Syme (1979) found that persons who lacked

social and community ties, as measured by the Social Network Index

(Berkman, 1977; see Appendix F), showed a higher rate of mortality

than those with greater social ties. The age-adjusted relative risk

for those most isolated compared to those with the most extensive

ties was 2.3 for men (p<.001) and 2.8 for women (p<.001). A low

level of social support has also been found to be associated with the

incidence of specific disorders, such as tuberculosis (Jackson, 1954;

Holmes, 1957), coronary heart disease in Chinese-Americans (Marmot

and Syme, 1976), cardiovascular disease in Italian-Americans (Bruhn

et al., 1969; Wolf, 1976), and ulcers in unemployed men (Gore, 1978).

Nuckolls et al. (1972) studied the relationship between










social stress, psychosocial assets (social support) and medical

complications experienced during pregnancy. Data was obtained on a

group of white married women of similar age and social class, all of

whom were pregnant for the first time, and delivered at the same

hospital. It was found that women with high life stress scores and

low social support experienced significantly more complications than

both women with high life stress and high social support, and women

with low life stress scores (regardless of level of social support.)

In a related study, De Araujo et al. (1973) examined the

association between psychosocial assets, life change, and dosage of

adrenocorticosteroids required to control chronic intrinsic asthma.

They found a negative rank order correlation between social support,

as measured by the Berle Index (Berle et al., 1952), and steroid

dosage (r=.564, p<.001). There was no direct relationship between

life stress scores, as measured by the SRE, and steroid dosage.

However, when the life change and social assets scores were combined,

it was found that patients with high social support scores invariably

required smaller doses of steroids regardless of their LCU scores.

Patients with low social support and high LCU scores required

significantly higher doses than those patients with low social support

and low LCII scores (p<.01).

The results of these studies consistently support the hypothesis

that social support acts as a buffer against, or moderator of, the

adverse effects of stress. However, there is a methodological

problem with this line of research; the conceptualization and

measurement of social support used in these studies is not consis-

tent. For example, the Berle Index, which was used in the De Araujo









et al. (1972) study, combines into a single score demographic and

medical information, data on the patient's interpretation of family

and interpersonal relationships, and the physician's judgment of

the patient's past performance, personality structure, and attitudes

toward illness. This measure has been criticized as being ambiguous.

as measuring social status rather than interpersonal support, and as

relying on the subjective judgments of the physician and patient

(Murowski et al., 1978). TAPPS, the measure developed by Nuckolls et

al. (1972) also combines information from several areas into a single

score; this instrument tapped the areas of self-concept, attitude

toward marriage and extended family, social resources, and attitudes

toward the pregnancy.

Other instruments have focused on more discrete components of

social support. The Social Network Index (Berkman, 1977), the

instrument employed by Berkman and Syme (1979), assesses marital

status, number of and frequency of contact with friends and

relatives, and group membership and participation. Other researchers

have developed measures of social support which assessed subjects'

confidants and acquaintances (Miller, Ingham, and Davidson, 1976),

availability of helpful others in coping with problems (Medalie and

Goldbourt, 1976), values similarity (Brim, 1974), and degree of

satisfaction with available support (Sarason et al., 1981).

Murowski et at. (1978), in a critical review of the measurement

methods developed to evaluate social support, propose that

researchers either have tended to use too broad a conceptualization

of, or have focused on discrete components of social support. These

researchers propose that the concept of social support, when used in










the study of illness, should be limited to the characteristics of

interpersonal relationships, and should not include socio-economic

factors or material assets per se. Further, they propose that the

measurement of social support should include an inventory of Qtose

persons and institutions which provide interpersonal support, a

measure of patterns of social affiliation, and an assessment of

satisfaction with available support. They conclude that there is

presently no adequate instrument to measure social support as it is

related to disease etiology and coping with disease.

While the research on the relationship between social support

and illness is limited by measurement and conceptual problems, the

studies conducted to date strongly suggest that social support is

protective of health. Further, while life stress appears to play a

role in the development and course of some illness, the combination

of factors of low social support and high life stress appear to be a

better predictor of illness than either factor alone.

Family membership. This model holds that factors such as family

stress, family adjustment. and interpersonal relationships within the

family have a significant effect on disease course and onset.

In a study which focused on stress within the family, Meyer and

Haggerty (1962) followed 100 members of 16 families for a year,

periodically taking throat cultures for beta streptococci, and

clinically evaluating illness. It was found that acute family

crises, including accidents, illness or death, divorce, and job loss,

were four times more common in the two-week period proceeding

strepococcal infections and illness than in the two-week period

following illness onset.










In an extensive seven year study of 223 adult medical and

surgical patients, Duff and Hollingshead (1968) examined, among other

things, the interrelations between disease onset and family

adjustment. It was found that 47% of patients' illnesses were linked

to unsatisfactory family relationships, and that a significant per

centage of these patients came from severely maladjusted or

moderately adjusted families. This study also found that two-thirds

of the patient's physicians had no awareness of the connection

between the patient's illness and the family situation. Apley (1959),

Apley and MacKeith (1973), Kellner (1963), Peachey (1963), and

Hopkins (1959) also report data which support the hypothesis that

poor family adjustment and high family stress are significantly

correlated with somatic symptomalogy in family members.

Other researchers have focused on dyadic relationships within

the family. Many of the early psychosomatic studies, as based on

psychoanalytic theory, focused on the interaction of the mother-child

dyad. Typical of this research is Forrer's (1960) case study in

which it was proposed that an infant developed two different

dermatological lesions in "psychosomatic compliance" with unconscious

conflicts which the mother experienced in her own psycho-sexual

development. This research has generally been refuted (Reiser, 1975;

Lipowski, 1977) as being limited by its theoretical orientation, as

ignoring the role of the father and other family members and as

suffering from numerous methodological flaws.

More recently, researchers have focused on the relationship

between physical illness and the dyadic relationship between husbands

and wives. Typical of this line of research is the work of Cobb










et al. (1969). In his study of the intrafanilial transmission of

rheumatoid arthritis, it was found that arthritic women were married

to men with peptic ulcers with a frequency well above chance. Based

on data from extensive interviews and medical histories, Cobb et al.

proposed that the development and course of the two disorders was

best understood as a part of the interpersonal relationship between

the members of the couple. It was suggested that these couples

develop a relationship because of the wife's tendency to be

controlling and the husband's need to be controlled. When

difficulties arise in the marriage, the resulting marital hostility

contributes to rheiamtoid arthritis in the wife via resentment and

depression, and to the peptic ulcer in the husband via unmet needs

for emotional support.

In a related study, Henker (1964) looked at recurrent

psychosomatic illness in 37 couples treated in groups over a four

year period. He found that exacerbation of symptoms coincided to a

significant degree with periods of increased marital tension, and

concluded that the onset of the somatic symptoms was caused by the

tension within the marital dyad.

Summary. The psychosomatic models of disease, as elaborated by

researchers focusing on personality characteristics, psychosocial

variables, and family membership, have been criticized as being

inadequate and highly inferential (Reiser, 1975; Weiner, 1977; Brody

and Sobel, 1979; Minuchin et al., 1978). By employing the

psychosomatic model, which holds that mind and body constitute a

functional unity, these researchers sought an alternative to the

restrictive biomedical model. However, these investigators utilized










the same model of linear causality and reductionistic methods of

analysis that were used to develop and apply the germ theory. By

adopting this metholodogical approach, they focused on a single

factor or simple combination of factors, while ignoring dynamic

interrelationships among personality, psychophysiological, and

environmental variables, and proposed various linear models in which

disease is understood to be contained within the individual (see

Figure 1). Further, since these theorists lacked a common conceptual

framework for psychological and physiological variables, they were

able to demonstrate covariance between factors, but not the causality

they sought to prove. While these studies show a correlation between

illness and various life events, social support, personality, and

family membership variables, these findings in and of themselves

prove nothing about time sequence and causality, as understood in a

linear sequence model (Reiser, 1975).

From the General Systems Theory perspective, the various

psychosomatic models are conceptually inadequate, since they are not

in accordance with the principles of wholeness, hierarchical organi-

zation, omnipotentiality, equifinality, and circular causality, as

inherent in all open living systems. Grolnick (1972), in his systems

oriented review of research on family-related factors of illness,

proposes that it is simplistic to assume a linear sequence of events,

such as marital tension-psychosomatic exacerbation or psychosomatic

exacerbation-marital tension. According to Grolnick, "marital

tension" is a system at a different and hierarchically higher level

than "somatic processes"; the former is most appropriately understood


















Life Stress






Emotions.
Personality Emt s Defenses and Coping
Mechanisms


Autonomic
Nervous
System


Musculo-
Skeletal
Nervous


Other
Physiological and
Biochemical
Systems


Disease


Figure 1.
Linear Model of Disease
(Minuchin et al., 1978)


Endocrine
System











to be the context within which the somatic symptoms occur rather than

the direct cause of the symptoms.



Family Systems Model



The family systems model of disease holds that the unit of

analysis to which many disease processes can be most meaningfully

related is the family system; and that the patient, through her/his

symptoms, manifests pathology which is inherent in the family system.

Thus, this model holds that disease does not originate or reside

solely within the individual (Meissner, 1974; Brody and Sobel, 1979;

Minuchin et al., 1978).

The most comprehensive research, using the family systems model

of disease, has been conducted by Minuchin et al. (1978). This

project involved the intensive study of two groups of families; one

having children with chronic conditions under poor medical control,

the second having children with chronic conditions under good

control. In the first group were children with anorexia nervosa,

intractable asthma, and superlabile diabetes. In the second group

were normal diabetic children, and diabetic children whose illness

was under good control but who had significant behavioral problems.

Families were assessed by means of a family task interview, a

structured interview, and long term family therapy. As part of the

structured interview, a direct measure of the physiological effects

of parental conflicts on a child's disease was made. The

physiological measure used was blood concentration of free fatty

acids (FFA). FFA serves as a measure of emotional arousal in the










general population (Bogdonoff and Nichols, 1964) and signals the

advent of ketoacidosis (i.e., the state of poor control of diabetes)

(Baker et al., 1974).

The results of this study indicate that the three types of "poor

medical control" (PMC) families were similar to each other, and that

they differed from the "good medical control" (GMC) families in

several ways. Compared to the GMC families, the PMC families tended

to be enmeshed, i.e., to be more responsive to, involved with, and

interdependent on family relationships; to be more intrusive on

other's communication; to have less differentiated perceptions of

oneself and of other family members; and to have weak family

subsystem boundaries. The PMC families tended to be more over-

protective than the GMC families. The former displayed significantly

more nurturant-protective and protectiveness-eliciting behaviors.

PMC families were found to avoid and diffuse conflict more

frequently. Families with normal diabetic children agreed and

disagreed more, and considered more alternatives in completing the

family tasks. The behavior problem families tended to diffuse

conflict, but were able to express conflict more openly that the PMC

families.

The results of the analysis of the physiological data showed

significant results for the three diabetic groups only. The

superlabile diabetic group was found to differ from the other two

groups in two respects. First, the PMC children had a rise in FFA

levels while viewing parental conflict. The other two groups showed

a slight decline in FFA levels. Second, following the resolution of

the parental conflict, the FFA levels in the superlabile group










remained elevated while the levels in the two control groups moved

toward the baseline levels. While previous medical studies showed no

intrinsic physiological differences among the children in these three

groups, this experiment showed the superlabile group to have an

exaggerated "turn on" and an impaired "turn off" physiological

response to family conflict.

The physiological results also indicated that the PMC children

played a role in maintaining family stability homeostasiss). FFA

levels of the diabetic children were plotted against those of the

parent whose arousal was highest during the interview. In the

superlabile group, it was found that the parent showed a decrease in

FFA level when the child was brought into the conflict situation.

These changes in FFA levels were not found in the other two groups.

Thus, while the superlabile child's stress was increased and her/his

medical condition was exacerbated, the parent's stress was

alleviated.

As part of this study, the symptoms of the PMC patients were

treated by means of family therapy. All the children with

siperlabile diabetes had either a good or excellent level of control

following therapy. Prior to therapy, all of the children with

intractable asthma were on steroid therapy, were experiencing

prolonged and severe asthma attacks, and were missing school for

weeks at a time. Following therapy, 80% of the patients were having

only occasional, mild attacks, were not on steroid therapy, and were

not missing any school. The remaining cases showed moderate

improvement. Of the anorectics who were treated through family










therapy, 88% were completely recovered, 6% were unimproved, and 6%

relapsed after apparent successful treatment.

Based on this multi-variable multi-method experiment, Min4phin

et al. (1979) proposed the following model of psychosomatic illness

in children (see Figure 2). The symptomatic child is physiologically

vulnerable, i.e., a specific organic dysfunction in present. The

family has four organization or functional characteristics:

eneshment, overprotectiveness, a lack of conflict resolution, and

rigidity. The symptomatic child plays an important role in the

family's pattern of conflict avoidance, and this role is an important

source of reinforcement for the child's symptoms.

In contrast to the models of disease previously discussed, which

hypothesize that specific disease symptoms are related to a given

family constellation or a simple etiological factor, this model

posits that there are general types of family processes which

encourage somatization and other dysfunctions, and that there are a

cluster of related, interactive factors involved in the disease

process. Causality in this model is circular: certain types of

family organizations are related to the development and maintenance

of somatic: symptoms in children, and the child's somatic symptoms

pliy a major role in maintaining the stability of the family's

interaction and organization.

This open system model of illness, as proposed by Minuchin et

al., is supported by the findings of several other studies. For

example, Nye (1957) found that students from broken homes had fewer

psychosomatic symptoms than did students from unhappy but unbroken

homes. Nye interpreted this data as supporting the hypothesis that




















Family
Extrafamilial___- Organization
Stresses and
Functioning '





Vulnerable
Child


Physiological,
Endocrine,
> and
Biochemical
Mediating Mechanisms
















Figure 2.
Open Systems Model of Disease
(Minuchin et al., 1978)


Symptomatic
Child
.-,I











somatic symptoms are related to a high level of family cohesion

(enmeshment), and the suppression of differences and open conflict.

Stewart (1962) found that illness is related to the suppression pf

aggressive and non-conforming feelings. In this long term

prospective study relating subsequent disease to social and emotional

adjustment, those persons presenting psychosomatic symptoms were

found to show significantly better family and social adjustment than

did those showing behavioral maladjustment.

In a study of families having a child with ulcerative colitis,

Jackson and Yalom (1966) found that arguments and emotional comments

were avoided and that there was a lack of tender affectionate

interaction between the parents. Members of these families had a

restricted number of roles within the family group. Communication

was found to be exceedingly indirect. Many of the siblings of the

symptomatic child were found to display symptoms of behavioral and/or

psychological problems. Parents often thought of the symptomatic

child as the least nervous and the most stable of the children, and

questioned the possible connection between emotional distress and

ulcerative colitis. Finally, the parents were restrictive, keeping

the children within the family circle. While they commented on the

children's lack of socialization, they did little or nothing about

it.

Research on the relationship between family characteristics and

level of control of diabetes also supports the family systems model.

Koski and Kumento (1977) found poor control in diabetic children to

be associated with unresolved family conflicts, a strong parent-child

coalition, diffuse generational boundaries, social isolation of the










family and a lack of social support, denial of health and

psychological problems on the part of the parents, and a focus on

child problems rather than marital problems. Excellent control was

associated with a stable family life, intact boundaries between

generations, a realistic and responsible attitude toward diabetic

care, and flexible problem solving. Siminds (1977) found an

unusually low divorce rate in families of well-controlled patients

compared to poor-controlled and non-diabetic comparison groups.

Johnson (1980) interprets the results of this study as indicating

that good control may be associated with unusually healthy or well-

integrated families. Steinhausser et al. (1977) found that well-

controlled patients reported their mothers to be highly supportive at

disease onset, and to be less supportive over time. The opposite

pattern was reported by patients with poor control. Other family

patterns found to be associated with poor control include high levels

of anxiety, overindulgence, overcontrol, resentment and rejection,

and disinterest and neglect (Bruch, 1973; Katz, 1957; Khurana and

White, 1970; Kravitz et al., 1971; Starr, 1955).

Several studies have found an association between family factors

and the presence or exacerbation of symptoms of asthma. It has been

noted in clinical reports and in controlled studies that about 40% of

asthmatic children lose their symptoms immediately upon separation

from their families through hospitalization (Coolidge, 1956; Peshkin

and Abramson, 1959; Puncell et al., 1969) or attending boarding

school (Bastians and Groen, 1955).










Research also supports the open systems model hypothesis that

somatic symptoms can be treated by means of family therapy. For

example, Lask and Matthews (1979) followed a group of children with

moderate to severe chronic asthma. All children received regular

medical care from a physician. In addition, children in the experi-

mental group attended six one-hour family therapy sessions during a

four-month treatment period. Results indicate that the experimental

group showed significant improvement in their symptoms, while the

control group did not show any improvement. Similarly, White et al.

(1978) report that, during a two year study, family therapy was used

successfully in improving the level of control of children's asthma.

No empirical results were reported in this study.

Family therapy has also been found to be an effective

intervention in the treatment of recurrent, non-organic pain in

children. Recurrent pain is pain which occurs over a considerable

period (months or years) and is severe enough to affect a child's

appearance and/or activities (Apley et al., 1977). In a review of

the limited literature on the various types of recurrent pain, Apley

et al. found significant similarities between the different kinds of

recurrent pains (i.e., in different anatomical locations), between

the children with recurrent pains, and between the families of these

children. They concluded that all types of recurrent pain in

children should be conceptualized as a single disorder; that this

disorder is an expression of emotional stress, and that it is an

integral part of a family pattern of interaction. They suggest that,

as a rule, these children should be treated through a comprehensive

family oriented approach. Of the three studies which have evaluated










the effectiveness of family therapy in the treatment of recurrent

pain, all reported significant positive results (Apley and Hale,

1973; Berger et al., 1977; White et al., 1978).

The literature also indicates that, in the treatment pf

anorexia, family therapy, and individual therapy which focuses on

contemporary family dynamics, is an effective method of treatment.

Consistently positive results have been reported Bruch (1973), Barcai

(1971), Palazzoli (1974), and Minuchin et al., (1978). Vigersky

(1977), in his review of the research on the treatment of anorexia,

concludes that the family approach is the treatment of choice, being

significantly more effective than psychoanalytic or behavioral

methods.



Summary



The following conclusions about disease processes can be

summarized from this literature review.

1. Manifest disease is not caused by any single, isolatable

factor or event, but rather, is associated with the interaction of

physiological, social, life event, and familial factors. This is

supported by the fact that no one factor has been found which is

associated with all cases of a given disease, that all persons

experiencing a given factor do not manifest the disease, and that all

persons with a given disease do not respond equally to a given

intervention. This is in accordance with the General Systems Theory

principles of hierarchical organization, wholeness, omnipotentiality,

and equifinality.










2. The disease process can be conceptualized as a disorder or

an extreme variation in the complex regulation processes of an

organism or as the inability to respond successfully to environmental

changes. This model of disease is in accordance with the General

Systems Theory principles, as outlined by Miller et al. (1976) that

dysfunctional systems are characterized by a disturbance in

adaptation (i.e., extreme morphogenesis or homeostasis) and by a

disturbance in cohesion (i.e., being either enmeshed or disengaged).

3. Given manifest disease, those adolescents who are less

responsive to medical treatment will have experienced more stressful

life events and/or will have low social support and/or will be a

member of a less functional family system.













CHAPTER III
METHODOLOGY




The primary purpose of this study was to determine if, by

assessing psychosocial factors, it were possible to differentiate

adolescent patients whose chronic condition was in good medical

control and were doing as well or better than expected from those who

were in poor medical control and were not doing as well as expected.

The literature reviewed indicates that the development and course of

physical illness is related to the following three psycho-social

factors: (a) family structure and functioning, (b) life stress, and

(c) social support.

The secondary purposes of this study were to further test the

family systems model of illness in children as proposed by Minuchin

et al. (1979) which hypothesizes that chronically ill adolescents

who are in poor medical control are members of dysfunctional

families; and to test the psychosocial model of illness as proposed

by Cobb (1976), Dean and Lin (1977), Kaplan et al. (1977), Nuckolls

et al. (1972), and others, which hypothesizes that social support

has a moderating effect on the adverse health effects of life

stress.










Subjects



The subjects of this study were the member of 48 families each

having an offspring (age 14-19) with one of the following four types

of chronic disorders: pulmonary (asthma, cystic fibrosis) (Ne21),

gastroenterological (ulcerative colitis, Crohn's disease) (NN13),

cancer (N=11), and juvenile rheumatoid arthritis (N-3). All of the

adolescent patients included in the study had received medical care

for this condition through a specialty pediatrics outpatient clinic

at Shands Teaching Hospital and Clinics for a minimum of six months

prior to participation in the study.

During the data collection stage of the research 86 families

were contacted and asked to participate in the study. Of these, two

refused to participate when contacted. Thirty-four families which

were contacted did not return questionnaires. Questionnaires were

returned by 53 families. Of these, 23 returned questionnaires from

the patient, mother, and father; 24 returned information from the

patient and mother; one returned questionnaires from the patient and

father; and five returned questionnaires from the patient only. Of

the 53 families which returned data ("return group") five were

dropped from the study because the level of medical response was not

established. Of the 34 families which were contacted but did not

return questionnaires ("no-return" group), five were dropped because

the level of medical response was not established.










Hypotheses



The following hypotheses will be tested in this study.

I. HO: There will be no linear or quadratic combination

of the variables of family functioning, life stress, and

social support which will statistically distinguish between

adolescents in very good medical control, adolescents in good

medical control and adolescents in poor medical control.

1. HO: The level of functioning of families with a

chronically ill adolescent member in poor medical control will

not be significantly different from the level of functioning

of families with a chronically ill adolescent member in very

good or good medical control.

Hal: Families of adolescents in poor medical

control will be more likely to be functioning at the

extremes of the Circumplex Model than will families of

adolescents in very good or good medical control.

Ha2: Families of adolescents in poor medical

control will score significantly lower on the Family

Functioning Index than will families of adolescents in

very good or good medical control.

Ha3: Families of adolescents in poor medical

control will have more extreme scores on the Family

APGAR than will families of adolescents in very good

or good medical control.

III. HO: The quantity of recent life change experienced

by families of adolescents in poor medical control will not be










statistically different from the quantity of recent life

change experienced by families of adolescents in very good or

good medical control.

Ha: Families of adolescents in poor medical control

will have experienced more recent life change than will

families of adolescents in very good or good medical

control.

IV. HO: The quantity of social support experienced by

families of adolescents in poor medical control will not be

statistically different from the quantity of social support

experienced by families of adolescents in very good or good

medical control.

Ha: Families of adolescents in poor medical

control will have experienced significantly less social

support than will families of adolescents in good

medical control.

V. HO: The quantity of social support experienced by

the families of adolescents in very good or good medical

control (GMC families) which have experienced a high level of

life change will not be significantly different from the

quantity of social support experienced by the families of

adolescents in poor medical control (PMC families) which have

experienced a high level of life change.

Ha: PMC families which have experienced high life

change will have experienced significantly less social

support than will GMC families which have experienced a

high level of life change.










Instrumentation



Family Adaptation and Cohesion Evaluation Scales (FACES)

FACES (Olson et al., 1982) is a 60-item self-report instruwmap

designed as a tool for use by family therapists for diagnosing family

problem behaviors and for setting treatment goals (see Appendix A).

This instrument is a shortened and improved version of the original

edition of FACES (Olson et al., 1979a), which contained 11III items.

This assessment tool, which measures family functioning along the

dimes ions of "adaptability" and "cohesion", is based on the

Circumplex Model of family functioning (Olson et al., 1979b).

This nodel, which is derived from General Systems Theory and is

based on a review of the literature in the entire field of family

behavior, proposes that adaptability and cohesion are the two most

salient dimensions for describing family systems. "Cohesion" is

defined as: "the emotional bonding members have for one another and

the degree of individual autonomy a person experiences in the family"

(Olson et al., 1979b, p. 5). "Adaptability" is defined as: "the

ability of a marital or family system to change its power structure,

role relationships, and relationship rules in response to situational

and developmental stress" (Olson et al., 1979b, p.12). In the

Circumplex Model, these two independent dimensions are combined in

such a way that families can be classified according to where they

fall on both. By dividing each dimension into four levels: very low,

low to moderate, moderate to high, and very high, a 4 X 4 matrix is

formed defining 16 types of family functioning (see Figure 3).














COHESION


SEPARPA
(14-Aerately Low)


Mtmcmn
(Moderately High)


(VeESHiD
(Very High)


Figure 3.
Sixteen Types of Marital and Faaily Systems Derived front the Circumplex Model
(Olson et al., 1979a)


DISENGAGED
(Very Low)


CFOXBLEC
(Moderately
High)

STRUClTRED
(Moderately
low)

RIGID
(Very Low)


Chaotically Chaotically Chaotically Chaotically
Disergaged Separated Connected Ermeshed


Flexibly Flexibly Flexibly Flexibly
Disengaged Separated Connected* Eaeshed


Structurally Structurally Structurally Structurally
Disengaged Separated Connected Eneshed


Rigidly Rigidly Rigidly Rigidly
Disegaged Separated Connected Enmeshed










According to the Circumplex Model, the healthiest families are those

which fall in the moderate ranges of both dimensions. These four

types are designated by an asterisk (*), The unhealthiest families

are those at the extremes on both dimensions (those underlined in the

four corners). Between these two are the eight types of families

which are moderate on one dimension but extreme on the other.

FACES is comprised of statements concerning various aspects of

family interaction and functioning. Each family member independently

completes the questionnaire by indicating on a scale from one to five

the degree to which each statement is felt to be true of her/his

family. A "1" indicates that the statement is felt to be true of the

family "almost never", uhile a "5" means it is true of the family

"almost always".

Two primary scores are obtained, one for "cohesion" and one for

"adaptability". The range for cohesion scores is from 27 to 135.

The range for adaptability scores is from 23 to 115.

Analysis of data from 1000 families indicates that the internal

consistency reliabilities for the total scores for adaptability and

cohesion are high (r=.79 and r=.92 respectively). A factor analysis

and item analysis are now being conducted by the authors.

As the revised edition of FACES is new, no studies have yet been

reported using this version. However, the Circumplex Model, upon

which FACES is based, does appear to have empirical validity in terms

of differentiating families under stress and in setting treatment

goals for family therapy (Olson et al., 1982; Olson et al., 1979b).

Further, the validity of the original version of FACES in the study

of disease is supported by a study conducted by Lewis (1981) on










factors affecting the psychosocial adjustment in chronically ill

children and in their parents. Lewis found a significant relation-

ship (p<.001) between extreme FACES scores and the number of behavior

problems reported in the children. Lewis also found that children in

families with extreme FACES scores tended to have a lower self-

concept (p=.059) than children in families with moderate FACES

scores.

This instrument was administered to parents and the adolescent

patient. This measure was selected because it is derived from

General Systems Theory, and it is specifically designed to assess

family adaptability and cohesion. The importance of these two

dimensions of family functioning in illness outcome has been shown in

the work of Minuchin et al. (1979).



Family Functioning Index (FFI)

The FFI (Pless and Satterwhite, 1973) is a 15-item self-report

instrument designed as a diagnostic tool for physicians to identify

families with chronically ill children in need of special interven-

tion services (see Appendix B). The unitary dimension of family

functioning is measured by assessing the following areas: marital

satisfaction, frequency of disagreements, communication, problem

solving, and feelings of happiness and closeness.

FFI has an interobserver reliability of r=.72 and a test-retest

reliability of r=.83. Interobserver reliability was determined by

comparing independently obtained FFI scores of husbands and wives

(Pless and Satterwhite, 1973). Test-retest reliability was deter-

mined by a five year follow-up study (Satterwhice et al., 1976).










Validity of the instrument has been determined in several ways.

FFI scores of registrants at family service agencies were compared to

those of a random sample. Mean scores for agency families (X-l,l)

were significantly lower than those for the random sample (YX!25.4,

p<.001). Case workers, using a five-point rating scale designed to

reflect the content of the FFI, also rated these families. The

correlation between FFI scores of wives and case worker ratings were

significant (r=.48, p<.01). The correlation between FFI scores of

husbands and case worker ratings was also significant (r=.35, E<.05)

(Pless and Satterwhite, 1973). In a separate study, lay counselors

working with families with chronically ill children rated these

families on the five-point scale. Correlation between these ratings

and the mothers' FFI scores was r=.39 (p<.01) (Pless and Satterwhite,

1975). Low FFI scores have been associated with more behavioral

problems and lower self-esteem in children (Pless et al., 1972) and

non-compliant behavior among children with renal transplants (Kosch,

1978).

An augmented version of the FFI (Johnson, 1980) was used in this

study. In this version five questions were added to the original

instrument. These questions take into account "self" behaviors,

whereas the original instrument assessed only "spouse" behaviors.

This version yielded both an original version score and an augmented

version score.

This instrument was administered to the parents only. The FFI

was selected because it is the only established self-report measure

of family functioning designed specifically for use with physically

ill children.










Family APGAR (APGAR)

The Family APGAR (Smilkstein, 1978) is a five-item self-report

questionnaire designed as a diagnostic tool for physicians to measure

global family functioning, and to identify patients with family

difficulties (see Appendix C). Each of the five questions is

designed to measure a family member's satisfaction with a different

component of family functioning. The areas are adaptability,

partnership, growth, affection, and resolve.

Inter-item correlations for the Family APGAR range from r'.24 to

r=.67. Split-half reliability is estimated at r=.93. Inter-observer

reliability, determined by comparing independently obtained Family

APGAR scores of husbands and wives, was found to be r=.67 (Good et

al., 1979).

Validity of the measure has been determined by comparing Family

APGAR scores of clinical and non-clinical families, by comparing

APGAR scores with FFI scores, and by correlating APGAR scores with

therapist ratings of clinical families. Clinical families were found

to score significantly lower than non-clinical families on overall

index scores (p<.001), and on four of the five items (y<.001 for

items #1,#2, and #3; p<.Ol for item #4). No difference was found on

item #5, thich assessed satisfaction with the amount of time spent

with the family (Smilkstein, 1978). Following this study, item #5

was changed to reflect the quality, rather than the quantity of the

time commitment of the family (Smilkstein, 1980). Validity data is

not available for the revised form. The APGAR has been found to

correlate with FFI scores (r=.80, y<.01). The APGAR has also been










found to correlate with therapist ratings of a clinical group of

families (r=.64, _<.01).

Smilkstein (personal communication, 1980) has found that this

instrument does not reliably detect "psychosomatic families in

pathological equilibrium", but does detect "psychosomatic families in

which a member is attempting to break away". Specific data in regard

to these findings are not available. This instrument will be

administered to parents and the adolescent patient. This instrument

was selected because it is designed to assess the relationship

between family functioning and medical outcome. This instrument was

administered to parents and the adolescent patient.



Schedule of Recent Events (SRE)

The SRE (Holmes and Rahe, 1967b) is a 43-item self-report

instrument designed to assist social scientists in the study of the

relationship between social and life events and the onset and course

of physical illness (see Appendix D). Subjects indicate whether or

not they have experienced any of 43 described life events during the

previous year. Each life event has been assigned a life change unit

value (LCU), based on the judged magnitude of change in adjustment

required by the life event. An individual's score is the arithmetic

sum of the LCU values of the events experienced during the previous

12 months. Very high scores (450 or above) indicate a major life

crisis. High scores (300-450) indicate a major life change.

Moderate scores (150-300) indicate a minor life change.

Data indicate estimates of test-retest reliability of the SRE to

be from .26 to .90, and to average around .60. Higher reliability










scores have been found with more intelligent and educated subjects,

and over shorter periods of time (r=.90 over two weeks; r=.26 over 10

months) (Rahe, 1974).

Validity of the SRE has been supported through correlation of

the scale with the PUP test, another measure of life events (=r.79)

(Hurst et al., 1978). Predictive validity has been demonstrated

through a variety of studies which have found significant

relationships between SRE scores and the subsequent onset of a

variety of illnesses including diabetes mellitus (Kimball, 1971),

tuberculosis (Holmes, 1954, 1957), cardiac disease (Rahe and Lind,

1971), and asthma (De Araujo et al., 1973).

This instrument was administered to parents only. This measure

was selected because of its previously demonstrated utility in the

study of the onset and course of a variety of medical conditions.



Life Events Record (LER)

The LER (Coddington, 1972b) is a 42-item self-report instrument

designed to assist social scientists in the study of the relationship

between social and life events and physical illness in adolescents

(see Appendix E). Subjects indicate whether or not they have

experienced each of 42 life events during the previous year. Each

life event has been assigned a life change unit value (LCU), based on

the judged magnitude of adjustment required by the life event. An

individual's score is the arithmetic sum of the LCU values of the

events experienced during the previous year. Based on a survey of

3620 randomly selected children, means and standard deviations have

been established for social adjustment required by age.










Test retest reliability of the LER has not been reported.

Predictive validity has been demonstrated through a variety of

studies which have found a significant relationship between qER

scores and the subsequent onset or exacerbation of a variety of

illnesses, including juvenile rheumatoid arthritis (Heisel, 1972) and

cancer (Jacobs and Charles, 1980).

This measure was administered to adolescent patients only. This

instrument was selected because it is specifically designed to

evaluate the relationship between life change events and the onset

and course of illness in children.



A Short Scale for the Evaluation of Social Support (ASSESS)

ASSESS (Cohen and Reiss, 1981), is a 15-item self-report

questionnaire designed to assess the quantity and quality of family

and community support available to individuals under stress (see

Appendix F). The following have been identified by one or more

researchers or theorists as central to the concept and measurement of

social support:

.Enduring interpersonal ties to people and/or institutions

that can be relied on to provide emotional support, help, reassur-

ance, and feedback in times of need (Caplan, 1974; Berkman, 1977).

2. Networks of relationships, i.e., how interactive a person's

social contacts are with each other (Kaplan et al., 1977).

3. The pattern of an individual's social affiliation (Murowski

et al., 1978).










4. The number of "available others" to whom one can turn in

times of need, and the degree of satisfaction with the available

support (Saranson et al., 1981).

5. Information leading an individual to believe that s/he iq

cared for, is esteemed and valued, and belongs to a network of

communication and mutual obligations (Cobb, 1976).

ASSESS was designed to measure these aspects of social support

in the following manner:

1. Enduring interpersonal ties are measured by ASSESS items

#1-#6. These items constitute the Berkman Social Network Index

(Berkman, 1977), which assesses the availability of a confidant

(spouse), contacts with close friends and relatives, church

membership, and group membership. Test-retest reliability data is

not available for this instrument. The predictive validity of this

instrument was demonstrated in the study of the relationship between

social support and mortality conducted by Berkman and Syme (1979)

discussed earlier. In this study it was found that the age-adjusted

relative rate of mortality for those scoring lowest on the Berkman

Index compared with those having the highest scores on the Berkman

Index was 2.3 for men (<.00l) and 2.8 for women (p<.O0O).

2. Network of relationships is measured by ASSESS item #7: "How

many of your friends are friends with each other?".

3. The pattern of affiliation is measured by ASSESS item #8,

which measures how often an individual sees, telephones, and writes

important friends and relatives.

4. The number of "available others" and degree of satisfaction

with support is measured by ASSESS items #9-#15. Items #9-#15 were










selected from the 27-item Social Support Questionnaire (Sarason et

al., 1981). Item selection followed Cobb's (1976) conceptualization

of social support as described above. Item #15 was designed to

specifically assess social support for medically related pToblems.

Each item asks the subject to identify the number of people t9 whom

they can turn and on whom they car rely in a specified circumstance.

Each item also asks the subject to indicate how satisfied s/he is

with the available social support. Satisfaction is rated on a

six-point Likert-type scale ranging from "very satisfied" to "very

dissatisfied". Test-retest correlations (over a four week interval)

for the Social Support Questionnaire are reported to be r-.90 for

"number of people" (N) scores, and r-.83 for satisfaction (S) scores.

The alpha coefficient of internal reliability for N and S scores are

reported to be .97 and .94 respectively.

In order to establish the test-retest reliability of ASSESS,

this instrument was given to volunteers at the Gainesville Florida

Suicide Prevention and Crisis Intervention Center. The second

administration was four weeks after the first. Thirty-eight

volunteers completed ASSESS twice. Test-retest reliability was found

to be .86.

This instrument was administered to parents and to the

adolescent patient.



Physician's Form for Rating Level of Response to Medical Treatment

The physician's form for rating the patients' level of response

to medical treatment is a one-item questionnaire designed










specifically for this study (see Appendix G). On this form,

physicians were asked to rate each patient's response to medical

treatment on a four-point Likert-type scale ranging from "very poor,

much worse than expected" to "very good, much better than expected."

The instructions stated that the rating should reflect the relative

quality of response the patient had made to medical intervention,

given the patient's disease. It was specifically stated that the

rating should not reflect the relative level of medical compliance or

the prognosis.



Procedures



Families were contacted by the investigator during an

adolescent's outpatient visit to Shands Teaching Hospital and

Clinics. Propsective subjects were told about the study and were

informed that their participation would not affect the medical

treatment received. They were also told that the information gained

from the questionnaires would be kept confidential. If they chose to

participate, family members were asked to read and sign the research

informed consent form (Appendix H), and to complete the appropriate

research questionnaire packets. Slightly different sets of

questionnaires were given to parents and patients. Parents were

administered FACES, the FFI, the Family APGAR, the Schedule of Recent

Events and ASSESS. Adolescent patients were administered FACES, the

Family APGAR, the Life Events Record, and ASSESS.

In order that subjects could complete the test battery in

privacy, space was set aside adjacent to the pediatric clinic waiting










area tor the completion of the questionnaire. Families were also

given a stamped, addressed envelope, for returning questionnaires if

they were not completed while waiting for the clinic appointment. If

both parents were not with the child at the clinic, a questionnaire

packet and consent form, and a stamped, addressed envelope was given

to the family for the absent parent. A follow-up phone call was made

three weeks later to all households which had not returned the

questionnaire. A follow-up letter (with response postcard) was sent

to all households which had not returned the questionnaires by six

weeks after the clinic visit. A copy of the letter and response post

card are contained in Appendix I.

At the conclusion of the data collection phase of the study

physicians were asked to complete the physician's form for rating the

level of response to medical treatment. Physicians were asked to

rate only those patients with whom they were familiar.













CHAPTER IV
RESULTS



The transformations performed on the data from this study are

described first in this chapter. Next, data concerning the

physicians' rating of the level of medical response are described and

data regarding the characteristics of the sample population are

presented. Finally, the results of hypothesis testing are presented

separately for each of the five major hypotheses.



Data Transformations



In order to effectively test the five hypotheses in this study,

four data transformations were performed. Each is described below.

The first transformation was done in order to be able to test

Hypothesis II-1, which, in general, stated that adolescents in this

study from the "worse" medical response group were more likely to

come from families which function at extreme levels on the Circumplex

Model dimensions of adaptability and cohesion than were adolescents

from the "as expected" or the "better" response groups. This

transformation calculated the value of the deviation of each

subjects' adaptability and cohesion score from the mean (after Lewis,

1981). This was done in the following manner: First, using scores

from the instrument Family Adaptation and Cohesion Evaluation Scales










(FACES), grand means were calculated for adaptability (ADP) and

cohesion (COH) for adolescents, mothers, and fathers separately.

These grand means are shown in Table 1.


Table 1. Grand Means and Standard Deviations of ADP and COH for
Adolescents, Mothers and Fathers Data

Standard
Variable N Grand Mean Deviation

Adolescents' scores 48
ADP 45.02 6.48
COH 58.41 9.63

Mothers' scores 42
ADP 46.59 5.86
COH 61.78 9.60

Fathers' scores 21
ADP 45.95 7.69
COH 64.00 9.56



Next, the deviation score was calculated for each subject on

each of the two dimensions. These scores were calculated by taking

the absolute value of the difference between a subject's score on a

dimension and the appropriate grand mean for that dimension. The

deviation score for adaptability (Dev ADP) can be represented in the

following way: Dev ADP = ADP ADP ?, where ADP is the subjects

adaptability score, and ADP is the appropriate grand mean for

adaptability. Similarly, the deviation score for cohesion (Dev COH)

can be represented as Dev COH = jCOH COH 1.

The second transformation was also performed in order to be able

to test Hypothesis II-1. This transformation calculated the distance

between each subjects position on the Circumplex Model and the

absolute center of the Circulplex Model (see Figure 3). This was










done in the following manner: First, using scores from the instrument

FACES, standard deviations were calculated for adaptability (Sd ADP)

and cohesion (Sd COH) for each family member group separately. These

standard deviations are presented in Table 1. Z-scores were then

calculated for each dimension. This calculation involved dividing

subjects' Dev ADP by the appropriate ADP standard deviation score,

and Dev COH by the appropriate COH standard deviation score. The

Z-score can be represented in the following way:

ADP Z-score Dev ADP / Sd ADP

Next the distance from the center of the Circumplex Model

intersect was calculated, in Z-score units. This score, hereafter

referred to as the FACES score, was calculated by taking the square

root of the sum of the Dev ADP Z-score squared and the Dev COH

Z-score squared. The FACES score can be represented in the following

way:

FACES = [(Dev ADP Z-score)2 + (Dev COH Z-score)2]1/2

The third transformation was performed in order to be able to

test Hypothesis 11-3, which, in general, stated that adolescents in

this study from the "worse" medical response group were more likely

to come from families which had mire extreme scores on the Family

APCAR instrument than adolescents from the other two response groups.

This transformation calculated the value of the deviation of each

subject's Family APGAR (APGAR) score from the mean. This was done in

a manner identical to that involved in deriving Dev ADP and Dev COH

scores, and yielded a Dev APGAR score for each subject. The grand

mean and standard deviation APGAR scores are shown in Table 2.











Table 2. Grand Means and Standard Deviations of Family APGAR
Scores for Adolescents, Mothers and Fathers

Standard
Variable N Grand Mean Deviation

Adolescents' APGAR 48 7.98 2.09
Mothers' APGAR 42 7.38 2.81
Fathers' APGAR 21 7.76 2.67



The fourth data transformation was performed in order to be able

to test Hypothesis V, which, in general, stated that adolescents in

this study who came from families which experienced high levels of

life change and low levels of social support were more likely to be

in the "worse" medical response group. This transformat ion

categorized subjects as to their relative level of life change and

social support. First, subjects were rank ordered according to their

life change score. Adolescents were rank ordered according to their

Life Events Record (STRESS) scores. Mothers and fathers were rank

ordered separately according to their Schedule of Recent Events

(STRESS) scores. A median split was then performed on the

distributions, and data from the centermost subject was discarded,

when necessary. The uppermost and lowermost halves of the

distributions were designated as high and low change, respectively.

Table 3 reveals the mean and standard deviation scores for STRESS for

each family member.

Next, subjects were rank ordered according to their scores on A

Short Scale of the Evaluation of Social Support (ASSESS), and a

median split was performed as with STRESS scores. Adolescents,

mothers and fathers were rank ordered separately. The uppermost and










lowermost halves of the distributions were designated as high and low

support respectively. Table 3 shows means and standard deviation

scores for ASSESS for each family member.

Finally, subjects who fell into both the high STRESS and low

ASSESS categories were classified as "at high risk", while subjects

who fell into any of the other three categories were classified as

"at low risk".


Table 3. Grand Means and Standard Deviations for
High and Low ASSESS and High and Low STRESS Scores for
Adolescents, Mothers and Fathers


High Stress

Standard


Low Stress

Standard


Variable Mean Deviation Mean Deviation

Adolescents 283.37 97.53 91.04 47.91
Mothers 263.35 104.12 96.90 42.15
Fathers 196.33 76.60 58.25 33.97


High Support Low Support

Adolescents 24.87 2.75 16.96 2.66
Mothers 28.21 2.69 19.40 3.93
Fathers 26.72 3.40 17.77 3.52



With these transformations, each adolescent had the following

ten scores: ASSESS, APGAR (Family APGAR), Dev APGAR, STRESS

(Life Events Record), COH (cohesion dimension of FACES), Dev COH,

ADP (adaptability dimension of FACES), Dev ADP, FACES (distance in

Z-score units from the intersect of the Circulplex Model), and "at

high risk" or "at low risk".

Each parent had the following twelve scores: ASSESS, APGAR

(Family APGAR), Dev APGAR, STRESS (Schedule of Recent Events), COH










(cohesion dimension of FACES), Dev COH, ADP (adaptability dimension

of FACES), Dev ADP, FACES (distance in Z-score units from the

intersect of the Circulplex Model), FFI (Family Functioning Index),

FFIA (Family Functioning Index Augmented form) and "at high risk" or

"at low risk".



Rating of Level of Response to Medical Treatment



The level of response to medical treatment was rated for each

patient by a pediatricians) familiar with the child's medical

history. All adolescent patients included in the study had received

care through a specialty pediatrics outpatient clinic at Shands

Teaching Hospital and Clinics for at least six months prior to

participation in the study.

At the conclusion of the data collection phase of the study,

physicians were asked to rate their patients on the following four-

point scale: (a) very poor, much worse that expected, (b) poor, worse

than expected, (c) fair, about as expected, and (d) good, better than

expected. Physicians could also indicate if they were unable to rate

the patient (see Appendix G for a copy of the rating form).

Adolescents with asthma and cystic fibrosis were rated by three

physicians from the Pulmonary Clinic. Of the 40 patients rated, 27

received the same rating from all physicians rating the case. Of

these cases, five were rated by only two physicians. Of the

remaining 18 cases, the patient was assigned the rating given by two

of the three physicians. Three from this group were given a










different rating by each of the physicians and were dropped from the

study.

Adolescents with cancer were rated by three physicians from chp

Hematology/Oncolocy Clinic. Of the 22 cancer patients, none were

rated by all three physicians. Two doctors rated two patients

apiece. Of these four patients, all received concordant ratings.

One physician rated 18 patients, and four were not rated. These four

were dropped from the study.

Patients with Crohn's disease and ulcerative colitis were rated

by two physicians from the Gastroenterology Clinic. Of the 18

patients rated, 11 received the same rating from both physicians. Of

the remaining seven, five were rated by only one physician. The two

subjects who received discordant ratings were dropped from the

study.

Patients with juvenile arthritis were rated by two physicians

from the Infectious Diseases/Immunology Clinic. Of the eight

patients rated, four received the same rating from both physicians,

and three were rated by only one physician. The one patient who

received discordant ratings was dropped from the study.



Sample Characteristics



Prior to the testing of the hypotheses, analyses were conducted

to determine if questionnaires were returned by a representative

sample of the families contacted, and if there were significant

differences between the families from the four different disease

groups.










Following the protocol described in the procedures section, 86

families were contacted and asked to participate in the study. Of

these, two refused to participate when contacted, and 33 agreed to

participate but returned no questionnaires. Of the 53 families which

returned at least one questionnaire, 24 returned questionnaires from

the patient, mother, and father; 23 returned information from the

patient and mother; one returned questionnaires from the patient and

father; and five returned questionnaires from the patient only. Of

these 53 families which returned data ("return group"), five were

dropped from the study because the level of medical response was not

established through the physicians' ratings. Of the 33 families

which were contacted but did not return questionnaires ("no-return"

group), five were also dropped because the level of medical response

was not established through the physicians' ratings.

No significant differences were found between the return and

no-return groups in regard to their proportion of males and females

(X2(l)=.8, y>.05), or blacks and whites (X2(1)=.67, p>.05). Also

no significant differences were found between the two groups in

regard to their proportion from each of the four disease groups

(X2(3)=4.0, p>.05), or from each of the three levels of medical

control (X2(2)=2.19, p>.05). An ANOVA revealed no significant

difference between the mean age of adolescents from each group

(F(1,76)=1.25, y>.05).









Table 4. Characteristics of the Return and No-Return Groups by
Gender, Race, Level of Medical Response, and Age


Female
Male
Total.

Black
White
Total

Cancer
Gastro.
Arthri tis
Pulmonary
Total

Worse
As Expected
Better
Total

Mean Age
Sd Age
Age range


Return

21
27
48

9
39
48

11
13
3
21
48

9
24
15
48

16.16
2.01
13-19


No-Return

10
20
30

8
22
30

7
3
4
16
30

10
13
7
30

15.83
1.48
13-19


Total

31
47
78

17
61
78

18
16
7
37
78

19
37
22
78


Disease Group Characteristics


Several analyses were done to determine if there were any

differences between the disease groups in regard to demographic, or

psychosocial (predictor) variables.

For these analyses only, the alpha level was set at 2<.10. This

liberal alpha level, was used in order to guard against making a Type

II error, i.e., concluding that there was no difference between the

disease groups on a predictor vaviable when, in fact, there was a

difference. It was important to guard against Type II errors because


- ~-----`-----










the low number of subjects in each disease group made it necessary to

use the sample as a whole (collapsing across disease groups) to test

the study hypotheses.

Comparisons using both return group and no-return group families

indicated no significant differences among the disease group in

regard to the proportion of males and females (X2(3)"5.3, p>.10).

No significant difference was found among the mean ages of the

adolescents from each of the four disease groups (F(3,74)'1.45,

j>.10). A significant difference was found in regard to the distri-

bution of blacks and whites among the disease groups (X2(3)-9.94,

y<.05). Significantly fewer blacks were in the gastroenterology

group than expected (X2(1)=9.94, p<.05). A significant difference

was also found among the four disease groups in regard to the

proportion of patients rated into each of the three levels of medical

response (X2(6)=11.91, P<.10). The proportion of juvenile

rheumatoid arthritis patients rated as worse than expected ("worse")

was significantly higher than expected (X2(1)=3.1, p<.10); the

proportion of gastroenterology patients rated as better than expected

("better") was also significantly higher than expected (X2(1)2.7,

<.10).









Table 5. Characteristics of the Three Levels of Medical Response
Groups by Gender, Race, Levels of Medical Response, and Age


Juvenile
Cancer GaOtro. Arthritis Pulmonary Total

Female 9 10 2 26 47
Male 9 6 5 11 31
Total 18 16 7 37 78

Black 2 0 3 12 17
White 16 16 4 25 61
Total 18 16 7 37 78

Worse 3 2 4 10 19
As Expected 8 6 2 21 37
Better 7 8 1 6 22
Total 18 16 7 37 78

Mean Age 16.81 L5.38 16.33 16.28
Sd Age 1.83 1.60 1.52 2.32
Age range 13-18 13-19 14-19 13-19



A three way ANOVA (disease group x gender x race) was then

conducted on each of the 34 predictor variables (10 variables from

each adolescent and 12 from each parent). Results revealed

significant main effects for disease along three of the predictor

variables. The results of these ANOVA's are contained in Table 6

(adolescents' data), Table 7 (mothers' data) and Table 8 (fathers'

data).

A significant difference was found among the four disease groups

on patients' ASSESS scores. Duncan's multiple range test revealed

that adolescents with cancer reported significantly higher levels of

social support than adolescents from the other three disease groups

(y<.10). A significant difference was also found on patients' FACES

scores. Duncan's multiple range test indicated that adolescents with








Table 6


Means, Standard Deviations, and F Scores For Disease Group Main Effects for
Adolescents' Scores (Across Race and Gerder)

Juvenile
Cancer Gastroenterology Arthritis Pulmonary
(N. I i. )3) (N3) (=21)

Standard Standard Standard Standard Ft p
Variable Mean Deviation Mean Deviation Mean Deviation Mean Deviation Value Value

ASSESS 25.00 3.03 19.23 4.26 20.66 2.08 19.85 5.13 4.23 .010***

STRESS 196.36 105.98 178.69 116.20 112.00 62.69 198.4 198.42 .47 .704

APGAR 8.54 1.75 8.00 2.34 7.66 .57 7.71 2.28 .37 .773

Dev APCAR 1.46 .75 1.85 .96 .34 .21 1.90 1.01 1.88 .147

ADP 46.54 4.20 46.00 7.62 42.66 3.21 43.95 7.08 .59 .626

Dev ADP 2.81 3.39 5.45 5.19 2.35 3.21 5.71 4.13 1.53 .221

CH 59.63 5.37 58.46 10.15 59.00 3.46 57.66 11.80 .10 .961

Dev OOH 4.32 3.14 7.96 5.85 2.86 .47 8.65 7.82 1.72 .177

FACES .68 .54 1.23 .92 .52 .40 1.40 .80 2.85 .049**


t N48, df=3
**W significant at p=.01
** significant at p_.05








Table 7


4ans, Standard Deviations, and F Scores For Disease Group Main Effects for
Mothers' Scores (Across Eace and Sex)

Juvenile
Cancer Gastroenterology Arthritis Pulmnary
1i-'O)- UNi 3J -(3) (F15)

Standard Standard Standard Standard F p
Variable Mean Deviation Mean Deviation Mean Deviation Mean Deviation Value Value

ASSESS 24.11 6.39 20.53 4.44 21.33 5.50 25.93 5.13 2.94 .046**

STRESS 211.70 121.74 172.69 71.78 48.00 38.11 180.81 134.90 .50 .687

APGAR 8.30 2.66 6.84 3.46 8.33 2.88 8.06 1.87 .73 .540

Dev AGAIR 1.89 .89 2.71 1.02 2.41 1.34 1.46 .62 1.29 .291

AEP 45.30 4.73 45.69 7.04 45.33 3.51 48.37 5.79 .76 .525

Dev AEP 3.50 3.27 5.80 3.73 2.85 1.55 4.60 3.79 1.00 .402

C1H 62.30 7.49 59.00 11.85 62.66 6.42 63.50 9.50 .52 .673

Dev COH 5.54 4.72 9.17 7.59 4.58 3.31 7.84 5.28 .98 .413

FACES .93 .58 1.51 .77 .71 .34 1.27 .60 2.06 .123

FFI 26.00(a) 6.54 19.90(b) 7.72 23.00 11.31 26.55(c) 6.28 1.47tt .250

FFIA 31.00(a) 7.09 24.30(b) 9.08 27.50 13.43 31.77{c) 7.47 1.3771 .279


t Ni41, df=3 ttN=28, df=3
*( significant at p(.05
(a) ?N6 (b) NW10 (c) N9







Table 8

Means, Standard Deviations, and F Scores For Disease Group Main Effects for
Fathers' Scores (Across Race and Gerder)

Juvenile
Cancer Gastroenterology Arthritis Pulmonary
(N4) (N=7) (N=3) (N=7)

Standard Standard Stardard Standard Ft p
Variable Mean Deviation Mean iation MEatn an Deviation Mean Deviation VaJie Value

ASSESS 26.00 5.00 21.00 7.09 26.00 3.60 21.57 4.82 .98 .429

STRESS 115.50 98.56 124.00 83.92 47.00 24.24 142.14 104.59 .08 .970

APGAR 6.25 4.50 7.85 2.73 8.33 2.08 7.14 2.41 .37 .777

Dev APGAR 3.25 1.16 2.40 1.23 1.87 .93 1.80 .84 .70 .568

AlP 46.25 9.42 43.14 8.61 48.33 11.93 47.57 4.03 .46 .715

Dev ADP 6.27 6.04 5.97 6.45 8.31 6.86 3.03 2.93 1.20 .343

OCI 63.00 10.70 60.28 11.78 67.00 10.14 67.00 6.50 .73 .550

Dev COH 7.50 6.45 9.14 7.58 7.00 6.55 5.00 4.86 .49 .697

FACES 1.22 .87 1.29 1.06 1.31 1.10 .74 .49 .68 .579

FF 24.75 8.53 21.71(a) 6.79 23.00 5.00 27.20(a) 1.64 .83tt .500

FFIA 28.00 10.29 26.14(a) 7.79 28.66 3.51 33.20(a) 1.92 .97tt .35


t N=21, df=3 ttN=19, df=3
** significant at p=.05
(a) N=6










pulmonary disease had more extreme scores on the Circumplex Model

than adolescents with cancer or arthritis. The third significant

difference was found between the disease groups on mothers' ASSESS

scores. Duncan's multiple range test revealed that mothers of

pulmonary patients reported higher levels of social support than

mothers of gastroenterology patients (P<.10).

While the ANOVA did not indicate a significant difference on

patients' Dev COH scores (7=.177), Duncan's test revealed that

adolescents with pulmonary and gastroenterological diseases had more

extreme scores on the dimension of cohesion than adolescents with

cancer and arthritis ((<.10). While the ANOVA did not indicate a

significant difference on patients' Dev APGAR scores (*.l147),

Duncan's test indicated adolescents with arthritis had less extreme

family functioning scores than adolescents with pulmonary and

gastroenterological diseases (p<.10). Finally, while the ANOVA did

not indicate a significant disease main effect for other's FACES

scores (~=.12), Duncan's test indicated that, on the Circumplex

Model, the functioning of families with an adolescent with pulmonary

disease is more extreme than that of families with an adolescent with

arthritis (p<.10).

Results for the gender analysis revealed main effects for three

predictor variables. Fathers of male patients reported significantly

less extreme scores on the dimension of adaptability (Dev ADP) than

fathers of female patients (F(1,15)=9.91, E<.01). Fathers of male

patients also reported significantly less extreme (FACES) scores on

the the Circumplex Model (F(1,15)=4.89, E<.05). Mothers of male

patients reported significantly lower social support scores than










mothers of female patients (F(1,35)-3.84, y<.05). Means and standard

deviations for significant variables for the main effect of gender

are presented in Table 9. Means, standard deviations and F scores

for all variables for the main effect of gender are eontained in

Table 25 (adolescents' data), Table 26 (mothers' data) and Table 27

(fathers' data) (see Appendix J).

Results for the race analysis revealed main effects for two

predictor variables. Black patients achieved significantly lower

life change scores (F(1,42)=4.23, R<.01), and had significantly less

extreme family functioning (APGAR) scores (F(1,42)-5.19, <.05), than

did white patients. Means and standard deviations for significant

variables for the main effects of race are presented in Table 9.

Means, standard deviations and F scores for all variables for the

main effect of race are contained in Table 28 (adolescents' data),

Table 29 (mothers' data) and Table 30 (fathers' data) (see

Appendix J).










Table 9. Means and Standard Deviations for Race and Gender
Main Effects of Significant Predictor Variables


Black White

Standard Standard
Variable Mean Deviation Mean Deviation

Adolescents' scores
STRESS 124.66 73.27 201.05 128.68
Dev APGAR 1.00 .85 1.85 1.23

Females Males

Standard Standard
Mean Deviation Mean Deviation

Mothers' scores
ASSESS 25.05 5.65 22.37 5.37

Fathers' scores
Dev ADP 8.91 5.42 2.18 2.72
FACES 1.55 .83 .69 .65


Because significant differences were


found on some of the


predictor variables with respect to disease, gender, and race,

analyses employed to test the hypotheses of this study used residual

scores rather than raw scores. Residual scores were calculated in

the following manner. First means were calculated for each predictor

variable for each disease, gender, and racial subgroup for

adolescents, mothers, and fathers separately. These calculations

yielded the following type of scores: mean APGAR score for all

adolescents with pulmonary disease, mean APGAR score for all white

adolescents, and mean APGAR score for all female adolescents. The

residual scores were then calculated by subtracting from each

subject's score for each variable the following three values: the










mean variable score for the subject's disease group, the mean

variable score for the subject's gender group, and the mean variable

score for the subject's racial group. (The grand mean of residual

scores for a given variable is, by definition, zero.) In the

following text and tables, residual scores are denoted with an "R/"

proceeding the variable name. For example, the residual score for

the variable ADP is denoted as "R/ADP".



Distinguishing Among the Three Levels of Medical Response
By Using All the Predictor Variables



In order to test Hypothesis I three discriminant analyses were

conducted. Hypothesis I stated, in general, that there was no linear

or quadratic combination of predictor variables which would

distinguish between the three levels of medical response at a rate

significantly greater than chance. Each of the three analyses used

data from a different family member. For the two analyses using

parental data, the Family Functioning Index scores were deleted in

order to maximize the number of subjects used to calculate the

discriminant function.

The prior probabilities of response level membership used in the

discriminant analyses were the proportions of response level

membership found in the whole subject pool (return and no-return

groups combined). These proportions were .25 from the "worse" group,

.47 from the "as expected" group, and .28 from the "better" group.

The proportion of patients which would be expected to be classified

by chance from each of the three medical response levels, into each

of the three response levels is shown in Table 10.









Table 10. Proportions of Patients Which Would Be Classified by
Chance into a 3 x 3 Classification Table


From Into Response Level
Response Level

Worse As Expected Better Total

Worse .0625'" .1175' .0700' .2500

As Expected .1175 .2209'" .1316' .4700

Better .0700 .1316' .0784" .2800

Total .2500 .4700 .2800

SCorrect classification
Incorrect classification


The proportion of subjects which was expected to be correctly

classified by chance is equal to the sum of proportions of patients

from each of the subgroups which would be classified by chance into

the correct subgroup. Therefore, the proportion of all patients

which was expected to be classified correctly by chance equals .3618

(.0625 + .2209 + .0784).

The discriminant function calculated from adolescents' scores

classified 41 of the 48 subjects correctly (85.41%). This proportion

of correct classification was significantly greater than the

proportion of correct classification expected by chance (Z=7.09,

O<.005). Chi square analyses were then conducted to determine if,

for each of the three classification subgroups, the proportion of

patients classified correctly was significantly greater than the

proportion expected to be classified correctly by chance. The

proportion of patients classified correctly was significantly greater

than chance for adolescents from the "worse" group (X2(1)"10.96,










<.005); for adolescents from the "as expected" group (X2(1).6,65,

p<.01); and for patients from the "better" group (X2(1)-20.31,

p<.005). The classification table for this analysis is shown in

Table 11.


Table II. Classification Table from Discriminant Analysis
Based on Adolescents' Data


From Response Into Response Level
Level

Worse As Expected Better Total

Worse 9'" 0' 0 9
(3.00)1 (5.64) (3.36) (12.00)

As Expected 0. 19" 5' 24
(5.64) (10.60) (6.31) (22.56)

Better 1' 1' 13" 15
(3.36) (6.31) (3.76) (13.44)

Total 10 20 18 48
(12.00) (22.56) (13.44) (48)

Correct classifications
Incorrect classifications
t Numbers in parentheses are the number expected to be classified in
the cell by chance.


The discriminant function calculated from mothers' data

classified 27 of 41 patients correctly (65.85%). The proportion of

patients correctly classified was significantly greater that the

proportion which would be expected to be correctly classified by

chance (Z=3.95, p<.005). Chi square analyses were also conducted,

separately, on each of the three response level subgroups. Results

of these analyses indicate that the proportion of patients correctly

classified was significantly greater than chance for patients from

the "as expected" group (X2(1)-5.32, j<.05); and for patients from










the "better" group (X2(1)=5.71, y<.05). The proportion of patients

from the "worse" group classified correctly was not significantly

greater than chance (X2(1)-.07, y>.05). The classification table

for this analysis is shown in Table 12.


Table 12. Classification Table from Discriminant Analysis
Based on Mothers' Data


From Response Into Response Level
Level

Worse As Expected Better Total

Worse 3' 2 2' 7
(2.56)t (4.82) (2.87) (10.25)

As Expected 2' 16" 2" 20
(4.82) (9.06) (5.39) (19.27)

Better 2' 4' 8'" 14
(2.87) (5.39) (3.21) (11.48)

Total 7 22 12 41
(10.25) (19.27) (11.48) (41)

Correct classifications
Incorrect classifications
i Numbers in parentheses are the number expected to be classified in
the cell by chance.


The discriminant function calculated from fathers' data

classified 16 of 20 patients correctly (85.00%). The proportion of

patients correctly classified was significantly greater than the

proportion of patients which would be expected to be correctly

classified by chance (Z=4.45, (<.005). Chi square analyses were also

conducted, separately, on each response level subgroup. Results of

these analyses indicate that the proportion of patients classified

correctly was significantly greater than chance for adolescents from

the "better" group (X2(1)-15.48, p<.005); and for adolescents from










the "worse" group (X2(1)=8.45, y<.005). The proportion correctly

classified was not significantly greater than chance for patients

from the "as expected" group (X2(l)-.04, y>.05). The classifica-

tion summary table for this analysis is shown in Table 13.


Table 13. Classification Table from Discriminant Analysis
Based on Fathers' Data


From Response Into Response Level
Level

Worse As Expected Better Total

Worse 5" 2' 0' 7
(1.25)t (2.35) (1.40) (5.00)

As Expected l* 4" 0 5
(2.35) (4.42) (2.63) (9.40)

Better 1 0' 7" 8
(1.40) (2.63) (1.57) (5.60)

Total 7 6 7 20
(5.00) (9.40) (5.60) (20)

Correct classifications
Incorrect classifications
t Numbers in parentheses are the number expected to be classified in
the cell by chance.


All three of the discriminant functions correctly classified the

patients from the subject pool at a rate significantly greater than

chance. However, these proportions may be an overestimate of the

porportion of patients from a different subject pool which would be

expected to be correctly classified. The possible overestimation is

due to the number of subjects used to calculate the discriminate

function given the number of variables used in the function, and the

number of classification groups. The number of subjects required to

develop an unbiased discriminant function is calculated using the









formula: S = 50 + 10( x + c 1), where x is the number of variables

used, and c is the number of classification groups.

In order to estimate the proportion of subjects from a new

subject pool which would be correctly classified by the three

discriminant functions derived above, a jackknife validation

procedure was performed. In this procedure, data regarding ore

patient is deleted from the data used to calculate a discriminant

function. This function is then used to classify the deleted

patient. This patient is then returned to the data pool, and another

is deleted and classified. This process is repeated until each

patient is, in turn, deleted and classified.

The discriminant function calculated from adolescents' data

classified 22 of the 48 patients correctly (45.58%). This proportion

of correct classification was not significantly greater than the

proportion of correct classification expected by chance (Z-1.36,

P>.05). The proportion correctly classified was significantly

greater than chance for patients from the "as expected" group

(X2(l)=3.88, p<.05). The proportion correctly classified was not

significantly greater than chance for patients from the 'better"

group (X2(1I).01, p>.05). The proportion correctly classified from

the "worse" group was not significantly less than expected by chance,

(X2(1)=.66, p>.05). The summary classification table is presented

in Table 14.









Table 14. Classification Table from Jackknife Discriminant Analysis
Based on Adolescents' Data


From Response Into Response Level
Level

Worse As Expected Better Total

Worse 1" 7 1' 9
(3.00)t (5.64) (3.36) (12.00)

As Expected 1 17" 6" 24
(5.64) (10.60) (6.32) (22,56)

Better 2* 9' 4'" 15
(3.36) (6.32) (3.76) (13.44)

Total 4 33 11 48
(12.00) (22.56) (13.44) (48)

Correct classifications
Incorrect classifications
t Numbers in parentheses are the number expected to be classified in
the cell by chance.


Using the jackknife procedure, the discriminant analysis

calculated from mothers' data classified 15 of the 41 patients

correctly (36.58%). This proportion of correct classifications is

not sign ficantly greater than the proportion of correct

classifications expected by chance (Z=.050, y>.05). Chi square

analyses were also conducted, separately, on each response level

subgroup. Results of these analyses indicate that the proportion of

adolescents correctly classified was not significantly greater than

chance for patients from the "better" group (X2(1)=.192, p>.05);

or for patients from the "as expected group" (X2=.202, y>.05). The

proportion of patients from the "worse" group correctly classified

was not significantly less than expected by chance, (X2(1)1.6,











y>.05). The classification summary table for this analysis is

reported in Table 15.


Table 15. Classification Table from Jackknife Discriminant Analysis
Based on Mothers' Data


From Response Into Response Level
Level

Worse As Expected Better Total

Worse 0" 4' 3' 7
(2.56)t (4.81) (2.87) (10,25)

As Expected 6 11" 4' 21
(4.81) (9.05) (5.39) (19.27)

Better 3' 6' 4*" 13
(2.87) (5.39) (3.21) (11.48)

Total 9 21 11 41
(10.25) (19.27) (11.48) (41)

Correct classifications
SIncorrect classifications
t Numbers in parentheses are the number expected to be classified in
the cell by chance.


Using the jackknife procedure, the discriminant analysis using

fathers' data correctly classified seven of the 20 patients correctly

(35.00%). This proportion of correct classifications was not

significantly greater than the proportion of correct classifications

expected by chance (Z=.14, p>.05). Chi square analyses were also

conducted, separately, on each response level subgroup. Results of

these analyses indicate that the proportion of patients classified

correctly was not significantly greater than chance for patients from

the "worse" group (X2(1)=1.25, p>.05), or from the "better" group

(X2(l)=.60, R>.05). The proportion of patients correctly

classified from the "as expected" group was not significantly leoa











than that expected by chance, (X2(1)-1.78, y>.05). A summary of

the classifications from this analysis is contained in Table 16,


Table 16. Classification


From Response
Level



Worse


As Expected


Better


Total


8
(5.00)


Table from Jackknife Discriminaet Analysis
Based on Fathers' Data


Into Response Level


Worse As Ex ected


6
(9.40)


Better


6
(5.60)


Total

7
(12.00)

5
(9.40)

8
(5.60)

20
(20)


Correct classifications
SIncorrect classifications
t Numbers in parentheses are
the cell by chance.


the number expected to be classified in


Based on the data from the three discriminant analyses,

Hypothesis I-1 is accepted at the .05 level. However, the results of

the jackknife validation procedure indicated that these significant

findings cannot be generalized to other subject populations. That

is, the discriminant functions may not differentiate between the

three response groups at a rate greater than chance (at the .05

level) for subjects from a new sample group.


3" 3' 1'
(1.25)t (2.35) (1.40)

2' 1"' 2-
(2.35) (4.42) (2.63)

3' 2' 3'
(1.40) (2.63) (1.57)


__ _______ _Ili__ _











Relationship Between Level of Medical Response and Family Functioning



Hypothesis II-l, in general, stated that adolescents whose

response to medical treatment was worse than expected come from

families that function at the extremes of the Circumplex Modpl, To

test this hypothesis ANOVA's were performed which examined the

differences among the three medical response level groups in the mean

R/Dev COH, R/Dev ADP, and R/FACES scores. Table 17 shows the mean

scores for each variable.


Table 17.


Means and F Scores for ANOVA Examining
Differences Among Levels of Response to Treatment on
R/Dev ADP, R/Dev COH and R/FACES


Level of Response to Treatment

Worse As Expected Better
F
Variable Mean Mean Mean Value Value

Adolescents' scores
R Dev ADP .45 .36 -.84 .47 .629
R/Dev COH -.65 -.16 .65 .15 .875
R/FACES .03 .02 -.06 .10 .909

Mothers' scores
R/Dev ADP .80 -.49 .33 .47 .630
R/Dev COH 1.12 -1.40 1.55 1.41 .255
R/FACES .18 -.17 .17 1.76 .185

Fathers' scores
R/Dev ADP -.70 -.47 .97 .40 .674
R/Dev COH -2.83 1.36 1.45 1.41 .270
R/FACES -.29 .04 .22 1.05 .369


Results of

based on scores


the ANOVA's for each of

from each of the three


the three dimensions, as

family members, were not


significant. Therefore Hypothesis II-I is rejected at the .05 level.











Hypothesis 11-2 stated, in general, that adolescents whose

response to medical treatment was worse than expected come from

families which score lower on the Family Functioning Index, In order

to test this hypothesis ANOVA's were performed which examined the

differences in mean FFI and FFIA scores among the three medical

response groups. Table 18 summarizes the mean scores and F and p

values for each variable.


Table 18. Means and F scores for ANOVA Examining
Differences Among Levels of Response to Treatment on
R/FFI and R/FFIA


Level of Response to Treatment

Worse As Expected Better
F p
Variable Mean Mean Mean Value Value

Mothers' scores
R/FFI 1.92 -1.20 .33 .37 .692
R/FFIA 2.00 -1.33 .42 .32 .728

Fathers' scores
R/FFI -.47 2.99 -1.95 1.71 .211
R/FFIA -1.00 2.91 -1.55 1.02 .381


Results of the ANOVA's for each of

on scores from each of the parents, were

11-2 is rejected at the .05 level.


the two variables, as based

not significant. Hypothesis


Hypothesis 11-3 stated, in general, that patients whose response

to nmdical treatment was worse than expected come from families which

score significantly lower and/or have more extreme scores on the

Family APGAR. To test this hypothesis ANOVA's were performed which

examined the differences in mean R/APGAR and R/Dev APGAR scores










among the three medical response groups. The findings of these

analyses are summarized in Table 19.


Table 19.


Means and F Scores for ANOVA Examining
Differences Among Levels of Response to Treapmept on
R/APGAR and R/Dev APGAR


Level of Response to Treatment

Worse As Expected Better

Variable Mean Mean Mean Value Value

Adolescents' scores
R/APGAR -.70 -.12 .63 1.28 .287
R/Dev APGAR -.39 .08 .10 .75 .479

Mothers' scores
R/APGAR -.49 .83 -1.00 .35 .704
R/Dev APGAR .80 -.49 .33 2.65 .082

Fathers' scores
R/APGAR -.50 1.90 -.98 .12 .886
R/Dev APGAR -.70 -.47 .97 .17 .846


Results of the ANOVA's for each of

on scores from each of the three


significant.


Hypothesis 11-3 is rejected


the two variables, as based

family members, were not

at the .05 level.


As all of the three alternate hypotheses were rejected at the

.05 level, Hypothesis II (null) is accepted at the .05 level.


f











Relationship Between Level of Response to Medical Treatment and
Quantity of Life Change



Hypothesis III-I stated, in general, that those adolescents

whose response to medical treatment was worse than expected come from

families which have experienced higher levels of recent life change.

To test this hypothesis, ANOVA's were performed which examined the

differences in mean R/STRESS scores among the three medical response

groups. The results of these analyses are summarized in Table 20.


Table 20.


Means and F Scores for ANOVA Examining
Differences Among Levels of Response to Treatment on
R/STRESS


Level of Response to Treatment

Worse As Expected Better
F
Variable Mean Mean Mean Value Value

Adolescents' score
R/STRESS 31.66 2.38 -22.80 1.15 .326

Mothers' score
R/STRESS -43.20 29.60 -21.69 .01 .989

Fathers' score
R/STRESS 3.08 -4.17 1.99 .63 .538



Results of the three ANOVA's are not significant. Therefore

Hypothesis III-1 is rejected at the .05 level, and Hypothesis III

(null) is accepted at the .05 level.










Relationship Between Social Support and Quality of Response to
Medical Treatment



Hypothesis IV-1 stated, in general, that adolescents whose

response to medical treatment is worse than expected come from

families which experience lower levels of social support. To test

this hypothesis ANOVA's were performed which examined the difference

in mean R/ASSESS scores among the three medical response groups. The

results of these analyses are summarized in Table 21.


Table 21.


Means and F Scores for ANOVA Examining
Differences Among Levels of Response to Treatment on
R/ASSESS


Level of Response to Treatment

Worse As Expected Better
F
Variable Mean Mean Mean Value Value

Adolescents' scores
R/ASSESS -1.68 .74 -.17 1.15 .326

Mothers' scores
R/ASSESS -.92 -.47 1.14 .63 .538

Fathers' scores
R/ASSESS -1.59 1.20 .64 .57 .575



Results of the three ANOVA's are not significant. Therefore

Hypothesis IV-1 is rejected at the .05 Level, and Hypothesis IV

(null) is accepted at the .05 level.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs