Factors influencing metabolic control in adolescents with insulin-dependent diabetes mellitus

MISSING IMAGE

Material Information

Title:
Factors influencing metabolic control in adolescents with insulin-dependent diabetes mellitus
Physical Description:
viii, 133 leaves : ill. ; 29 cm.
Language:
English
Creator:
Reynolds, Lynn Ann, 1959-
Publication Date:

Subjects

Subjects / Keywords:
Research   ( mesh )
Diabetes Mellitus, Type I -- drug therapy   ( mesh )
Diabetes Mellitus, Type I -- diet therapy   ( mesh )
Diabetes Mellitus, Type I -- Adolescent   ( mesh )
Patient Compliance -- psychology   ( mesh )
Patient Compliance -- Adolescent   ( mesh )
Department of Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Health Related Professions -- Department of Clinical and Health Psychology -- UF   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 121-132).
Statement of Responsibility:
by Lynn Ann Reynolds.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001708489
oclc - 26208799
notis - AJC0764
System ID:
AA00009068:00001

Full Text


















FACTORS INFLUENCING METABOLIC CONTROL
IN ADOLESCENTS WITH
INSULIN-DEPENDENT DIABETES MELLITUS











BY

LYNN ANN REYNOLDS


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


UNIVERSITY OF FLORIDA


1991





















In loving memory of my grandfathers,

Hamilton L. McNichol and Paul E. Reynolds















ACKNOWLEDGEMENTS

This project could not have been accomplished without

the combined efforts of many individuals. I would first

like to express my admiration and appreciation to my

chairperson, Dr. Suzanne Bennett Johnson, who has truly been

a mentor in every sense of the word, as a teacher, a

counselor, and a friend. Her patience, encouragement, and

endless support made it possible for me to complete this

project.

I would also like to thank my committee members, Dr.

Janet Silverstein, Dr. Stephen Boggs, Dr. Constance Shehan,

and Dr. Hugh Davis (to whom I feel special appreciation for

his important influence on my graduate training), for their

contributions to this endeavor. Dr. Gary Geffken also

deserves special gratitude for his input to this project,

and as the supervisor of my therapy with several adolescents

with IDDM.

I would like to thank Michael Kelly for his patience in

enduring endless computer questions, and for his hospitality

as I was completing this project. Michael Nurick, Lucretia

Mann, Dr. Deborah Ader, David Goodwin, and David Saliwanchik

also have my appreciation for their hospitality during that

memorable summer in Gainesville.


iii










The essential ingredients of a research project are the

participants and the research assistants. My thanks go to

both: to the adolescents and families who willingly gave

their time to participate, and to my research assistants,

who contributed to every stage of this project, from data

collection to typing references for the final draft.

Special thanks go to Martha McCallum, R.N., whose help

recruiting participants was essential to the success of this

project. Financial support for this project came from the

Geoffrey Clark-Ryan Memorial Research Award in Pediatric

Psychology; the founders and contributors to the fund have

my greatest appreciation.

My family, especially my parents, have my continuing

love and gratitude for their support (emotional and

financial!), encouragement, and constant belief in my

abilities.

Finally, I want to express my appreciation to Dr. Alan

C. Homans, whose tolerance, generosity, and nuturing enabled

me to successfully complete this endeavor. He has my

deepest gratitude and my endless love for his involvement in

this project and, more importantly, for his presence in my

life.
















TABLE OF CONTENTS

page
ACKNOWLEDGEMENTS ........................................ iii

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

INTRODUCTION AND REVIEW OF THE LITERATURE................ 1

Adherence .......................................... 2
Stress ............................................. 3
Pubertal Status.................................... 4
Psychological Adjustment and Family Functioning.... 5

METHODS ................................................. 39

Participants........................................ 39
Measures ........................................... 40
Procedures ......................................... 50

RESULTS ................................................. 53

Adjustment Measures.................................. 53
Stress Measures..................................... 56
Family Measures...................................... 58
Adherence Measures.................................. 75
Metabolic Control................................... 78
Multiple Regression Analyses ........... ............ 78

DISCUSSION .............................................. 102

REFERENCES.................... .......................... 121

BIOGRAPHICAL SKETCH ..................................... 133















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

FACTORS INFLUENCING METABOLIC CONTROL
IN ADOLESCENTS WITH
INSULIN-DEPENDENT DIABETES MELLITUS

By

Lynn Ann Reynolds

December 1991

Chairperson: Dr. Suzanne Bennett Johnson
Major Department: Clinical and Health Psychology


Adolescence has been identified as a period of life

when metabolic control is likely to be unstable for persons

with insulin-dependent diabetes mellitus (IDDM). A variety

of explanations have been proposed. They include poor

adherence to the treatment regimen, stress, pubertal effects

on physiology, poor psychological adjustment, and family

factors. This study examined the validity of a model of the

associations among these factors and metabolic control in

adolescents with IDDM.

In addition to attempting to delineate a more

comprehensive model of these associations, this

investigation differed from other studies in its use of

standardized questionnaire and behavioral measures from

multiple respondents to assess psychosocial factors; 24-hour

recall methodology, on multiple occasions with both










adolescent and parent respondents, to assess adherence; a

measure of pubertal development; and indices of both glucose

and lipid metabolism to assess diabetes control.

Hierarchical multiple regression techniques were used

to examine relationships among these variables. Fifty-six

adolescents with a mean age of 14.4 years participated in

the study. Due to sample size constraints, three sections

of the proposed model were separately tested.

Results were consistent with previous research in

finding significant relationships between family relations

and psychological adjustment. An unexpected finding was the

positive relationship between maternal stress and adolescent

social competence.

Adolescent psychological adjustment was found to be

associated with three measure of adherence to the treatment

regimen: insulin injections, consumption of concentrated

sweets, and exercise. Family behavior was associated with

exercise and calorie consumption. Consistent with previous

research, age was related to several measure of adherence.

Models of diabetes control relationships found gender

to be significantly associated with all measures of

metabolic control; girls were in poorer control than boys.

Few associations between measures of adherence and diabetes

control were found. Stronger relationships between

adherence behaviors and diabetes control were found for

indices of lipid metabolism than of glucose metabolism. For










most, these associations appeared to differ depending on

gender. This suggests a need for future research to attempt

to identify underlying mechanisms, including hormonal

changes, that may be influencing adherence-control

relationships differentially between girls and boys.


viii















INTRODUCTION AND REVIEW OF THE LITERATURE


Adolescence is the stage of development characterized

by physiological, emotional, and cognitive changes.

Separation from the family unit increases as the adolescent

begins to function more independently. These changes can

lead to stress and turmoil, both intrapsychically and

interpersonally, as maturation from childhood to adulthood

occurs (Hoette, 1983). The addition of a chronic health

problem, such as insulin-dependent diabetes mellitus (IDDM),

can increase the problems of adjustment associated with

adolescence. For a person with IDDM, maintaining good

diabetes control may also become increasingly difficult

(Amiel, Sherwin, Simonson, Lauritano, & Tamborlane, 1986;

Blethen, Sargeant, Whitlow, & Santiago, 1981; Cacciari,

Salardi, Ballardini, Righetti, Zucchini, & Natali, 1985;

Mann & Johnston, 1982).

Diabetes is a chronic illness that results from

insufficient insulin production by the pancreas; insulin is

necessary for the efficient metabolism of glucose. The

disease is the most common endocrine disorder of childhood

(Tarnow & Silverman, 1981-1982). Management of IDDM is

complex and demanding for the patient and family. Careful

attention must be paid to diet and exercise to aid in the










control of blood glucose levels. Daily insulin injections

must be administered, and frequent blood glucose testing is

encouraged in an effort to maintain patients in good

diabetes control (i.e., maintenance of the patient's blood

glucose levels within the normal range; Fisher, Delamater,

Bertelson, & Kirkley, 1982). Maintaining good control is

important in order to avoid the short-term complications of

diabetes: hypoglycemia, hyperglycemia, and ketoacidosis.

Equally important is the effect good control is thought to

have on the onset and seriousness of the long-term

complications associated with this disease: heart disease,

kidney failure, blindness, neuropathy, and other vascular

changes (Cahill, Etzwiler, & Freinkel, 1976; D'Antonio et

al., 1989; Leslie & Sperling, 1986; Tchobroutsky, 1978).

Both the clinical literature (Tattersall & Lowe, 1981)

and empirical studies (Amiel, Sherwin, Simonson, Lauritano,

& Tamborlane, 1986; Blethen, Sargeant, Whitlow, & Santiago,

1981; Cacciari et al., 1985; Mann & Johnston, 1982) have

identified adolescence as a period of life when metabolic

control is likely to be unstable. A variety of explanations

have been proposed. They include poor adherence to the

treatment regimen, stress, pubertal effects on physiology,

poor psychological adjustment, and family factors.

Adherence

The role of adherence to the diabetes treatment regimen

in maintaining good glycemic control has frequently been

identified in the clinical literature but has not been










subjected to rigorous empirical investigation. Of the

studies which have been reported, some suggest a link

between adherence behavior and diabetes control, at least in

adolescent patients (Christensen, Terry, Wyatt, Pichert, &

Lorenz, 1983; Hanson, Henggeler, & Burghen, 1987a, 1987c;

Johnson, Freund, Silverstein, Hansen, Malone, 1990;

Schafer, Glasgow, McCaul, & Dreher, 1983), while others have

failed to find a significant relationship (Hanson,

Henggeler, & Burghen, 1987b; Schafer, McCaul, & Glasgow,

1986). Other studies provide evidence that adolescents with

diabetes, as compared with their younger and older

counterparts, are less adherent to the diabetes regimen

(Allen, Tennen, McGrade, Affleck, & Ratzan, 1983; Bobrow,

AvRuskin, & Siller, 1985; Christensen et al., 1983;

Jacobson, Hauser, Lavori, et al., 1990; Jacobson, Hauser,

Wolfsdorf, et al., 1986; Johnson, Silverstein, Rosenbloom,

Carter, & Cunningham, 1986).

Stress

Psychological stress is a second factor which is

thought to be related to glycemic control in patients with

diabetes. Data suggests that adolescents, in general,

experience an increase in normal life stresses. Several

studies cite the peak incidence of life stressors as

occurring during middle adolescence (Coddington, 1972;

Johnson, 1982; Newcomb, Huba, & Bentler, 1986).

Research has demonstrated that parameters commonly used

to measure diabetes control (i.e., ketones, glucose, free










fatty acids) are influenced by life stress. Reviews of the

literature indicate that the impact of stressful life events

affects both physical health and psychological adjustment in

both normal and patient populations (Compas, 1987; Johnson,

1986). Some studies with IDDM children and adolescents have

supported this relationship (Barglow, Hatcher, Edidin, &

Sloan-Rossiter, 1984; Brand, Johnson, & Johnson, 1986; Chase

& Jackson, 1981; Hanson & Pichert, 1986; Tarnow &

Silverman, 1981-1982) while other studies have failed to

find a significant association (Delamater, Kurtz, & Bubb,

1984; Delamater et al., 1985; Gilbert, Johnson, Silverstein,

& Malone, 1989).

Pubertal Status

More recently, investigators have recognized that

physiological changes during adolescence are important in

the metabolic control of diabetes. The most notable

physiological change is an increased resistance to insulin

which occurs during puberty in both normal children and in

children with diabetes. Amiel et al. (1986) reported an

impairment in insulin-stimulated glucose metabolism in

adolescent patients at Tanner Stages II-IV (i.e., the middle

stages of pubertal development, marked by breast development

and pubic hair growth in girls, and genital development and

pubic hair growth in boys) but not in those at Tanner Stage

I (pre-pubertal stage of adolescent development). Pubertal

(Tanner II-IV) children, as compared with prepubertal

(Tanner I) children, also had significantly greater






5


glycosylated hemoglobin values. Glycosylated hemoglobin is

a measure of metabolic control which reflects the percentage

of glucose molecules that attach to the body's hemoglobin

over the course of 60 days; higher levels are indicative

poorer metabolic control. Other studies have noted the same

pattern of better metabolic control in prepubertal as

compared with pubertal children, although the mechanism for

the observed effect was not identified (Anderson, Miller,

Auslander, & Santiago, 1981; Kaar, Akerblom, Huttunen, Knip,

& Sakkinen, 1984).

Psychological Adjustment and Family Functioning

Two additional variables which have received research

attention in patients with diabetes are psychological

adjustment and family functioning. These psychosocial

factors have generally been hypothesized to be related to

metabolic control indirectly, by interfering with adherence

to the medical regimen, although a direct relationship of

stress on glucose homeostatis has also been suggested

(Barcai, 1970; Newbrough, Simpkins, & Maurer, 1985; Schafer

et al., 1983; Simonds, 1979; Tarnow & Silverman, 1981-1982).

A review of the literature by Johnson (1985) organized

relevant studies into these three areas: a) the

relationship between the child's adjustment and physical

health, b) the relationship between family characteristics

and the child's health, and c) the relationship between

family factors and the child's adjustment. The same

organization will be used here.










Psychological Adjustment and Physical Health

Early investigators focused their research on looking

for a personality specific to persons with diabetes. A

thorough review of the literature by Dunn and Turtle (1981)

concluded that there is no evidence of a characteristic

personality that distinguishes patients with diabetes from

any other chronically ill group or from the normal

population. However, children in poorer diabetes control

have been consistently found to have more emotional and

behavioral problems than their well-controlled counterparts.

In an early study reported by Simonds (1977), 40

children (aged 6-18 years) in good metabolic control,

matched on age, sex, and duration of diabetes, with 40

children in poor metabolic control, were interviewed by an

examiner who was blind to the children' diabetes status.

In addition, each child's mother completed a behavioral

checklist of emotional and behavioral characteristics

regarding her child. Psychometric data on the questionnaire

was not reported. Metabolic control was measured by

physician ratings based on medical history, urine test

records, physical examination, and urine and blood test

results at the time of the clinic visit. Results indicated

that while there was minimal psychiatric disturbance in any

of the children, the children in poor diabetes control

reported more interpersonal conflicts than those in good

control. Their mothers also reported significantly more

behavioral and emotional problems in those children.










In a similar study, Gath, Smith, and Baum (1980)

assessed the emotional and educational status of a group of

76 IDDM children (aged 5-16 years) in order to investigate

the interaction of those factors with diabetes control.

Information was obtained through interviews at the clinic

and questionnaires (one which was described as "widely

used") completed by the school, which assessed attitude,

academic performance and behavior of the child. Diabetes

control was clinically assessed using daily records of urine

analysis, symptoms of hypo- and hyperglycemia, and a record

of the child's growth. Psychiatric disorders were not more

common in diabetic children than in healthy controls.

However, among the children with diabetes, a correlation was

found between poor diabetes control and the presence of

psychiatric disorders. Poor metabolic control was also

found to be associated with the presence of adverse

psychosocial factors in the family background, although

these adverse factors were not described.

A more recent study of 31 IDDM adolescents (aged 12-19

years) investigated the relationship between personality

factors, assessed using standardized personality measures,

and metabolic control, measured by glycosylated hemoglobin

(Lane et al., 1988). Measures of trait anger, anxiety, and

curiosity, as well as extraversion, neuroticism,

impulsivity, and sociability, were obtained. Negative

correlations between metabolic control and measures of

extraversion (r = -.36) and trait curiosity (r = -.42) were










reported; poorer control was associated with lower

curiosity, more introversion, and poorer sociability.

Adherence was measured as the number of recorded blood

glucose tests per day; no relationship was found between

adherence to blood glucose testing and glycosylated

hemoglobin values.

Simonds, Goldstein, Walker, and Rawlings (1981) also

investigated the relationship between psychological and

personality variables and metabolic control, as measured by

HbAlc (i.e., a measure of glycosylated hemoglobin; HbA1 =

HbAla + HbAlb + HbAlc). A level of 9.5% was chosen as the

cutoff for adequate vs. inadequate control. Subjects in the

study were 52 adolescents, aged 13-19 years, with 5 or more

years duration of IDDM. Anxiety, locus of control, self-

concept, and several personality variables were assessed

with standardized self- and mother-report questionnaires.

Unlike previous studies, no significant differences were

found between high and low HbAlc levels for any of the

psychological variables examined. However, female subjects

were significantly more likely than males to have high HbAlc

values.

While it has been assumed that psychological factors

may influence metabolic control indirectly, through an

association with adherence to the diabetes regimen, few

investigators have studied this relationship. Jacobson and

colleagues (1987, 1990) recently reported on results of a

four-year longitudinal study in which one of the










relationships assessed was the association between

psychological adjustment and adherence behaviors. Sixty-one

children, aged 9-16 years, with recent onset IDDM were

studied over the four-year period. Self-esteem, locus of

control, ego defense mechanisms, adaptive strengths,

behavioral symptoms and social functioning, and self-

reported attitudes, feelings, and behaviors directed

primarily towards diabetes, were measured using standardized

self- and parental report instruments, with reported

reliability and validity, and semistructured clinical

interviews. Psychometric data on questionnaire measures for

the study sample were not reported. These variables were

examined in relationship to clinical indices of adherence

provided by ratings from the health provider. At both 9-

and 18-month follow-ups, higher self-esteem, diabetes

adjustment, social functioning, and fewer parental reports

of behavioral symptoms were correlated with higher levels of

adherence. Adolescents (aged 13-15 years) were less

adherent than pre-adolescents (aged 9-12 years). Higher

initial adjustment was significantly associated with better

adherence over the four-year study, as were the use of more

mature defenses and greater adaptive capacity. No variables

were found to be associated with change in adherence.

Results of analyses examining the relationships of metabolic

control to psychosocial variables and to adherence were not

reported in these studies. However, the authors reported










that they have found that these aspects of patient

functioning also predict metabolic control.

Several problems are evident with the research to

date on the relationship between psychological adjustment

and metabolic control. Measurement problems abound, both in

assessing adjustment and in determining diabetes control.

Often psychometric data on the measures used to examine

psychological adjustment are not reported. When they are

reported, authors often refer to norms from earlier

literature on questionnaire development, and do not report

psychometric data on their study sample. Measures of

metabolic control range from relatively subjective physician

ratings to objective laboratory results. Adherence, when

measured at all, is also often poorly assessed with measures

lacking in reliability and validity.

Frequently studies have failed to report the influence

of relevant variables, including age, gender, disease

duration, and socioeconomic status, on their findings. Most

studies have been correlational and cross-sectional,

precluding any statement about the direction of observed

effects. The mechanism by which psychological adjustment

may be related to health outcome has not been elucidated; it

may that adjustment influences metabolic control through an

influence on adherence. However, this association has not

been carefully investigated. Only Jacobson et al. (1987,

1990) have examined adolescent adjustment longitudinally,

and while they have looked at its relationship to adherence,










they have not reported on the relationship of adherence to

metabolic control.

In spite of these problems, some consistencies in the

literature have been reported. The research has generally

supported a relationship between psychological adjustment,

measured as emotional, behavioral, and social functioning,

and metabolic control. Better adjustment has been found to

be associated with better diabetes control. While studies

have not examined the relationship by which adjustment is

thought to be related to control, the investigation of

adjustment-adherence relationships is the first step in that

direction.

Family Factors and Physical Health

Several reviews have been published which describe the

literature relating family factors to the metabolic control

of diabetes (Anderson & Auslander, 1980; Fisher et al.,

1982; Hauser & Solomon, 1985; Johnson, 1980; Wishner &

O'Brien, 1978). In her review, Johnson (1980) described

five specific family patterns which are seen by clinicians

as particularly detrimental to patients' health: a)

overanxious patterns; b) overindulgent patterns c)

overcontrolling patterns d) patterns of resentment and

rejection and e) disinterest and neglect. Citing data from

the Simonds (1976-77) study, she suggested that good control

may be associated with unusually healthy or well-integrated

families; even "normal" family conflicts may be related to

poor control in some adolescents with diabetes.










Minuchin and his colleagues described a

psychophysiological model of the relationship between

parental or family patterns and a child's diabetes condition

(Baker, Minuchin, Milman, Leibman, & Todd, 1975; Minuchin,

Rosman, & Baker, 1978). They suggested that psychological

factors may influence diabetes in two ways: 1) emotional or

psychological disturbances may result in behavior problems

and non-adherent behavior which can have metabolic

consequences, or 2) emotional disturbances may cause

metabolic instability directly through psychophysiological

mechanisms. This second effect was termed "psychosomatic";

and the role of the family in producing psychosomatic

diabetes has been the subject of much interest and study by

Minuchin and colleagues. They defined a psychosomatic

family by four characteristics, which are: 1) enmeshment to

such an extent that the individual identities and roles of

family members are unclear, 2) overprotectiveness toward all

family members, 3) rigidity in maintaining the status quo,

and 4) lack of conflict resolution. They hypothesized that

the sick child plays a role in the family's attempts to

avoid conflict. When family conflict, which is unavoidable,

does occur, it leads to emotional and physiological arousal.

This is described as the "turn-on" phase. In a

psychosomatic family, the "turn-off" phase (i.e., the return

to baseline homeostatic levels) is hindered by the family's

attempts to avoid conflict; the result is a lack of conflict

resolution.










Newbrough, Simpkins, and Maurer (1985) reviewed the

literature on family factors as they affect diabetes

management and metabolic control in children with IDDM.

They also hypothesized that psychosocial factors in families

of children with diabetes have an impact both on management

of the diabetes (adherence) as well as on metabolic

stability. Five aspects of family that have been identified

in the literature as affecting blood glucose levels were

reviewed: 1) parent and child characteristics, including

parental and child self-esteem, and parental involvement in

the treatment regimen; 2) family group relationships,

including family conflict, stability, cohesion,

communication, and marital satisfaction; 3) parenting

style; 4) adaptation to the new life-style, including

acceptance of diabetes; and 5) relationship with the

community, including good social support. The authors

suggested a developmental model for understanding diabetes

management in terms of the stages of family development and

adjustment to the illness. They presented a framework for

organizing information on family functioning and its

relationship to the control of diabetes in children.

A number of investigators have examined the

relationship between metabolic control and various aspects

of the family. An early study by Swift, Seidman, & Stein

(1967) of 50 IDDM children (aged 7-17 years) obtained

information through semistructured interview and

standardized measures, conducted with both the children and










their parents. Better metabolic control, as rated by

diabetes specialists, were found to be related to the

following family characteristics: 1) few conflicts at home,

2) a low level of stress in the relationship between the

parent and the child with diabetes, 3) satisfactory home

adjustment by the child with diabetes and 4) an absence of

economic problems. Also, the child with diabetes tended not

to be the oldest in the family.

One of the early, and few, longitudinal studies

investigating family factors and metabolic control was

reported by Koski and Kumento (1977). They described a 5-

year follow-up study of 60 children, aged 10-21 years, with

diabetes. Information was obtained through ophthalmologic,

pediatric, and psychiatric evaluations. Metabolic control

was assessed by clinical judgement based on growth, fasting

blood glucose, 24-hour urine glucose, acetonuria, and number

of hospitalizations. Eight children were identified as

having poorly controlled diabetes; families of these

children were described as chaotic and having severe

conflicts. In cases where diabetes control had worsened

over time, a stressful life event had occurred. Eleven

children were identified as having well-controlled diabetes.

Five aspects were found to be related to good control: 1)

the family composition was stable, 2) clear, distinct

boundaries between generations were recognized by all family

members, 3) family members were realistic and cooperative in

implementing the treatment plan, 4) marital conflict was










low, 5) and both parents were present or a competent single

parent headed the family.

Cerreto and Mendlowitz (1983) used five standard

measures of family functioning and HbAlc to investigate the

relationship between family functioning and metabolic

control. Subjects were 84 children and adolescents with

diabetes (aged 7-17 years) and their parents. Family

measures were completed by the parents. Correlations

between family measures were low. Only the dimensions of

Control and Cohesiveness differentiated children in good

metabolic control from those in poor control. Children from

families with more rules and procedures demonstrated better

metabolic control.

Waller et al. (1986) reported on the development of a

disease-specific family support scale used to identify

family behaviors that correlate with metabolic control in

children and adolescents with IDDM. Subjects were 42

children and adolescents, aged 7-17 years, who were

attending a summer camp and completed the Diabetes-Specific

Family Behavior Scale (DSFB). Internal consistency, test-

retest reliability, and parent-child agreement for the scale

were adequate. The group of items which measured the

dimension of "guidance and control" correlated with

metabolic control, as measured by HbAlc, for the younger age

group (7-12) while "warmth and caring" items correlated with

metabolic control for both adolescents and children.

However, one item--"My parent gets angry with me when I make










a slip in taking care of my diabetes"--had an inverse

relationship for adolescents as compared with children. It

was associated with better control in the younger group and

worse control in the older group. The dimension of problem-

solving was not found to have a relationship with metabolic

control for either age group.

Schafer and her colleagues completed two studies that

looked at the relationship of family support, and additional

family factors, to metabolic control. They hypothesized

that family factors affect control indirectly through an

impact on adherence behaviors. The first study investigated

the relationship between family support, perceived family

environment, "barriers" to adherence, adherence, and

metabolic control (Schafer et al., 1983). Subjects were 34

adolescents with IDDM (age 12-14 years). Adherence was

assessed with a self-report measure; psychometric data on

the measure was not reported. HbA1 was used to assess

metabolic control. Three of seven adherence measures (i.e.,

extent to which diet was followed, care in measuring insulin

doses, number of daily glucose tests) were significantly

related to HbA1 levels; the measures were unrelated to each

other, indicating that adherence to one area of the IDDM

regimen is not highly related to compliance in other areas.

Family measures, as assessed by validated, standardized

questionnaires, were not highly related to metabolic

control. However, high family conflict was related to poor

adherence to glucose testing, while increased "barriers" to










adherence appeared to be most highly related to poor

adherence to diet and insulin injections. In general,

specific measures of psychosocial family variables were

better predictors of adherence than were more global

measures.

In their second study, Schafer et al. (1986)

investigated the relationship between family support, as

measured by the Diabetes Family Behavior Checklist (DBFC);

adherence, assessed by self-report, self-monitoring, and 24-

hour dietary recall interviews; and metabolic control,

measured with HbA1 values. Fifty-four adults and 18

adolescents (aged 12-18) with IDDM took part in this study.

Reliable differences between adolescents and adults were

found. Adolescents were in poorer metabolic control, and

reported more negative interactions with family members,

than did adults. However, negative family interactions did

not predict adherence or metabolic control for adolescents.

Adherence and metabolic control were not found to be related

in this study; the relationship between family factors and

metabolic control was not assessed.

In a study reported by Marrero, Lau, Golden, Kershnar,

and Myers (1982), participants were 40 adolescents aged

13-18 years, with IDDM. Information was obtained through

clinical interview and a "parent behavior description scale"

(no psychometric data was reported). Metabolic control was

rated based on clinical history. Results indicated that

perceived paternal behavior, rather than maternal behavior,










was related to metabolic control. Specifically, paternal

behavior which was seen as dominant and controlling was

associated with poor control, while paternal behavior which

was perceived as supportive and encouraging of automony was

associated with good control. There were no differences

reported between groups of adolescents in good versus poor

control in their perceptions of maternal behavior.

One recent study (Kovacs, Kass, Schnell, Goldston,

Marsh, 1989) of 85 children and early adolescents (aged 8-13

years) followed longitudinally over a 6-year span found no

relationship between measures of family functioning and

metabolic control. Overall quality of family life was

assessed with a standardized parental-report measure; the

authors reported good internal consistency and mother-father

agreement for the sample. Marital satisfaction was assessed

less reliably, using four items derived from two different

psychosocial measures. HbA1 and weight-adjusted insulin

dosage were measures of metabolic control. In contrast to

other studies, no associations between family measures and

metabolic control were found whether or not demographic

variables, including age, gender, race, and SES, were

included in the analyses.

While most studies of the relationship between

family functioning and metabolic control used interview and

self-report measures of family interaction, Baker et al.

(1975) employed a behavioral family task interview. This

was a method of observing family members interacting with










one another; structured family tasks were assigned, then

videotaped and rated. Tapes were scored for over-

protectiveness, enmeshment, rigidity, and lack of conflict

resolution. A family diagnostic interview was also used to

assess the involvement of the child in family conflict.

During the interview, an attempt was made to intensify

spouse conflict. At that point, the child was instructed to

enter the room and a measurement of free fatty acids in the

child during the interview was obtained. Of ten families

studied, six were found to fit the psychosomatic group, as

previously described. Treatment with a beta adrenergic

blockade propranololl) was found to be most effective in

children in which excessive arousal was the problem.

In a prospective study of 43 newly diagnosed diabetic

children (age 7-17 years), the relationship between

metabolic control and family factors described as family

competence, adaptational capacity, and emotional

supportiveness, were examined (Baker, Rosman, Sargent,

Nogueira, & Stanley, 1982; Sargent, Rosman, Baker, Nogueira,

& Stanley, 1985). Measures of family functioning were

obtained with a structured family interview, a standardized

self-report inventory, and ratings by a diabetes educator.

Each interview was taped and rated by blind raters for the

following qualities of family interaction: emotional

supportiveness, availability, role flexibility,

communication and decision making. These measures were used

to predict HbA1 at 12-18 months and 36-48 months post










diagnosis. Results indicated that positive parental and

parent-child emotional support, and family competence and

effectiveness were strongly related to good metabolic

control at both 12-18 and 36-48 month follow-ups.

Positive parent-child and sibling interaction were related

to good metabolic control only at the 36-48 month follow-up.

The authors concluded that factors relating to family

emotional supportiveness, competence in managing the child's

behavior, and effectiveness were significant predictors of

the degree of metabolic control achieved in a child with

diabetes and that the influence of parent-child and sibling

interaction on metabolic control increased over time.

Several studies attempted to investigate the link

between family factors and adherence. Shouval, Ber, &

Galatzer (1982) studied the relationship between perceived

family environment and adherence to the medical regimen.

Subjects were 97 Israeli children and adolescents with IDDM

(aged 10-20). Adherence was assessed based on subjective

reports from the psychosocial staff of the medical unit.

Family environment was assessed with two measures, including

a frequently-used standardized scale (the Family Environment

Scale), which the authors reported had poorer internal

consistency for their sample than has been previously

reported. Results indicated that patients who were more

adherent described their family atmosphere as supportive and

organized. Patients' metabolic control was not assessed in

this study.










Kurtz and Delamater (1984) described a study in which

they looked at family interaction patterns among groups of

chronically-ill mother-adolescent dyads, including 15 IDDM

pairs (mean age=14.7). The relationship between family

functioning and adherence was also assessed. The family

measures used were specific to parent-adolescent

interactions and conflict (i.e., Issues Checklist, Conflict

Behavior Questionnaire). Adolescents with diabetes

additionally completed disease-specific measures of family

conflict and family support. Adherence with glucose testing

was measured by recovering all chemstrips used in a 14-day

period; adherence to other aspects of the diabetes treatment

regimen was not assessed. Metabolic control was measured

with HbA1 assays. Results indicated that for adolescents

with diabetes, supportive family behavior and calm

discussions, rather than angry arguments, were associated

with better adherence to glucose testing. Positive

attributions regarding the behavior of the other and the

interaction of the mother-adolescent dyad were also highly

related to good adherence. The relationship between family

factors and HbA1 was not reported; however, adherence and

HbA1 were significantly correlated.

The relationship between mother-daughter behavioral

interaction and adherence among IDDM adolescents was

examined in a study by Bobrow et al. (1985). Participants

were 50 adolescent girls, aged 12-17 years, and their

mothers. Following an interview with the adolescent to










identify conflict issues, the mother and daughter were

brought together to discuss the three most salient issues

for five minutes each. After the third discussion, each

dyad was then asked to discuss their feelings about diabetes

together. Adherence was assessed with a Likert-type

questionnaire (interview) completed by the mother, the

adolescent, and the physician. Each discussion was recorded

and then scored with two interaction scoring systems. The

first method provided statement-by-statement ratings of type

of interaction (conventional, assertive, speculative, and

confrontive) and the second provided global ratings of

empathy, expressiveness, clarity, permeability,

responsibility, closeness, goal-directed negotiation, mood

and tone, and conflict. While the authors expressed caution

in interpretation of results due to small sample size and

lack of norms for their questionnaires, they did report

several interesting findings. Adolescents who were less

adherent were involved in more emotionally charged

interactions, were more directly confrontive, and were less

efficient at negotiating issues with their mothers. In

their statements about themselves on questionnaires,

adolescents confirmed the observations that poor adherers,

more than good adherers, had difficulty discussing feelings,

problems, and concerns with their mothers.

Most recently, Hauser and colleagues (1990) reported on

the results of their four-year longitudinal project in which

the relationship between family functioning and adherence










was one of the areas examined. The data analyzed was

obtained from 52 youngsters, aged 9-16 years, with IDDM and

their parents. Family functioning was assessed with a

frequently-used standardized paper and pencil measure, with

reported reliability and validity (the Family Environment

Scale), and adherence was measured by ratings from the

health provider. Separate ratings for diet, glucose

testing, and insulin injections were obtained. Results of

the study indicated that family conflict, as perceived by

both parents and patient, cohesion, as perceived by the

patient, and organization, as perceived by the parent, were

significantly correlated with a composite score of

adherence, as well as with individual aspects of adherence,

over the first year of the study. Children in families in

which conflict was low, and cohesion and organization were

high, demonstrated better adherence to the diabetes regimen.

Families who described themselves as initially more cohesive

also had children who showed improved adherence over time,

as well as demonstrating good adherence overall, for the

four years of the study. Initial family conflict was also

associated with overall adherence; poorer adherence was

noted in youngsters whose family conflict was high. No

measure of metabolic control was reported in this study.

As with the literature investigating the relationship

between psychological adjustment and metabolic control,

several problems are evident with the research on the

relationship between family factors and health outcome in










adolescents with diabetes. Once again, measurement problems

are prevalent. Family variables have been assessed with a

variety of different measures, including self-report

questionnaires and behavioral observations of family

interactions which makes it difficult to compare results

across studies. Again, psychometric data are frequently not

reported, or are reported only from the original studies on

questionnaire development and not from the study sample.

Metabolic control and adherence are often poorly

assessed. Measures of metabolic control have ranged from

the relatively subjective to the very objective, although in

recent studies, glycosylated hemoglobin seems to be the

standard measure of diabetes control. Measurement of

adherence is often limited to one aspect of the treatment

regimen, or is assessed as a unitary rather than a

multidimensional construct. Again, the influence of

relevant variables, including age, gender, disease duration,

and socioeconomic status, has not been examined. Mediating

variables are often not specified (i.e., adherence, stress),

or even if they are specified, they may not be assessed.

As with the psychological adjustment-health outcome

research, most studies investigating the relationship

between family factors and metabolic control have been

correlational and cross-sectional, precluding firm

conclusions about causation. While a conflictual,

unsupportive family environment may negatively influence

metabolic control, it is just as likely that living with an










adolescent in poor health may provoke family conflict.

Current research does not effectively address this issue.

Again it is Jacobson and colleagues (Hauser et al. (1990)

who have followed adolescents longitudinally, in an attempt

to examine the relationship between family factors and

health outcome, and again, they have looked only at

adherence, and not at metabolic control, as a health outcome

measure.

The results on studies investigating the relationship

between family factors and metabolic control have shown some

consistency, however, and suggest that several family

variables may be important. In particular, family conflict,

cohesion, organization, adaptability, supportiveness, and

parents' marital satisfaction were consistently found to be

related to metabolic control, and less consistently, to

adherence. The relationship between family variables,

adherence, and metabolic control needs further

investigation.

Psychological Adjustment and Family Factors

Although not specific to diabetes, Pless, Roghmann, and

Haggerty (1972), in a classic study, examined the

relationship between child adjustment, family functioning,

and chronic illness in a large sample of youngsters, aged 6-

11 years, in Monroe County, NY. The sample consisted of 209

children with chronic illness and 113 healthy controls.

Information was obtained using semistructured household

interviews, symptom checklists of child behavior completed










by a parent, information obtained from the child's teacher,

and the child's self-report on a standardized self-esteem

inventory. Each child then received an overall mental-

health adjustment index, based on these measures, and each

family was rated on a family-functioning index, based on the

household interview. Results indicated that both family

functioning and physical health seemed to contribute to a

child's psychological adjustment; youngsters in poor health

and those from poorly functioning families showed more

adjustment problems. The highest incidence of psychological

disturbance occurred in children who were ill and lived in

dysfunctional families, with older children showing the most

disturbance. The authors conclude that chronically ill

children who live in poorly functioning families may be

especially at risk for developing social and emotional

problems.

Hauser and his colleagues recently completed a four-

year longitudinal study of adolescents with IDDM and their

families. One of the first reports of their research

described the cross-sectional findings from the first year

of the study; the relationship between family functioning

and adjustment was examined with 30 IDDM adolescents aged 9-

17 (Hauser, Jacobson, Wertlieb, Brink, & Wentworth, 1985).

Using a standardized self-report measure of family social

environment (the Family Environment Scale; FES), family

factors were compared to adjustment to diabetes and

perceived competence, also assessed with frequently used










standardized self-report measures. Psychometric data on

these measures for the study sample were not reported.

Results from the study indicated that, controlling for age

and SES, family emphases on independence, participation in

social/recreational activities, and organization were

significantly associated with the patient's perceived

competence and diabetes adjustment. Aspects of adolescent

diabetes adjustment were also predicted by family

achievement orientations; better adjustment was related to a

family emphasis on achievement.

In an extension of their first study, the authors

(Hauser, Jacobson, Wertlieb, Wolfsdorf, et al. 1985)

reported results which used a larger sample (52 IDDM

adolescents, mean age=12.82), followed the parent's

perspectives separately from child's, and included

additional individual indices of self-esteem. In addition,

multiple regression analyses, controlling for age, gender,

and SES, were performed to identify the special contribution

of family orientations to individual patient

characteristics. They again found an association between

specific dimensions of the family on the Family Environment

Scale, this time as described by the parent, and the child's

perception of competence, as measured by a well-validated

self-report instrument. The most important family

dimensions were organization and independence; higher self-

esteem was associated with greater family organization and

an emphasis on independence. The child's adjustment to










diabetes was also significantly related to parental

perceptions of the family, better adjustment was associated

with greater family cohesion and an emphasis on

intellectual/cultural activities.

The most recent report from Hauser and his colleagues

(Wertlieb, Hauser, & Jacobson, 1986) included 46 children,

aged 9 to 16 years, studied longitudinally from the time of

diagnosis. Family environment and its relationship to

behavior symptomatology was examined. A comparison of 29

children with recent acute illness served as a comparison

group. The results of the study indicated that higher

levels of internalizing and externalizing child behavior

symptoms were reported by the mothers of children with

diabetes; however, the differences were nonsignificant when

social class was controlled. A range of family environment

variables was found to be related with behavioral symptoms.

For both groups, there was an positive association between

behavior symptomatology and levels of family conflict,

although more so for children with diabetes. For children

with diabetes, social and recreational family activities and

clear routines and organization were related to fewer

behavior problems. A moral/religious emphasis correlated

positively with internalizing symptoms in youngsters with

diabetes, but not in the acute illness group.

Few investigators have examined the relationship

between psychological adjustment and family functioning in

adolescents with diabetes. The studies described here










address many of the methodological issues plaguing the

literature on adjustment/health outcome and family

factors/health outcome relationships. They used

standardized measures and multiple respondents, and

controlled for the effects of variables including gender,

age, and SES. Again, however, psychometric data on the

assessment measures for the study samples are lacking, and

data is limited to self-report questionnaires. In spite of

these limitations, results from these studies have shown a

relationship between psychological adjustment and family

functioning. In general, various family dimensions have

been related to better psychological adjustment, as measured

by behavior symptomatology, self-esteem, and adjustment to

diabetes. Additional research is needed to identify

consistent relationships between specific family factors and

adjustment variables.

Multiple Psychosocial Factors and Physical Health

Recently, several investigators have examined the

relationship between multiple psychosocial variables,

including childhood adjustment, life stress, and family

functioning, and physical health, in adolescents with

diabetes. White, Kolman, Wexler, Polin, & Winter, (1984),

in a retrospective review of 30 children and adolescents

(ages not given) with recurrent diabetic ketoacidosis (DKA),

reported that only a minority of episodes of DKAs (i.e.; a

complication of diabetes caused by insufficient insulin,

which is marked by high blood glucose levels, a large amount










of urine ketones, pH inbalance, and symptoms of polydypsia,

polyuria, vomiting, abdominal pain, and rapid breathing)

were found to be related to intercurrent illness or poor (as

subjectively rated) compliance. Most of the subjects lived

in very dysfunctional families; problems included chronic

unresolved interpersonal conflict, inadequate parenting,

lack of a father in the home, financial stress, and lack of

family involvement with diabetes. Additionally, many of the

children displayed behavioral and personality problems which

seemed to have existed prior to the onset of IDDM. The

authors concluded that emotional or psychological stress can

act alone or together with poor compliance to make the child

more susceptible to DKA episodes. The retrospective nature

of this study requires that these conclusions be accepted

cautiously. However, these results are consistent with

other research findings.

Anderson et al. (1981), in a more well-controlled

study, examined the family environments of 58 adolescents

(aged 11-19) with IDDM. Metabolic control was measured

using HbAlc levels; three categories of control were derived

based on the distribution of HbAlc values. Family

environment was assessed with a frequently-used standardized

questionnaire (i.e., the Family Environment Scale).

Structured interviews were also conducted with adolescents

and parents separately. Psychometric data on these measures

were not reported for the study sample. Consistent with

previous findings, a trend was noted for middle-adolescent










(14-16 years) girls to have HbAlc values higher, indicating

poorer control, than either younger (11-13 years) or older

adolescent (17-19 years) girls, which the authors attributed

to physiologic changes of puberty, psychological or social

factors, or an interaction unique to girls in middle

adolescence that may have a detrimental influence on their

metabolic control. Middle adolescent girls also had higher

HbAlc values than did middle adolescent boys. Other results

included higher scores on total self-concept and less

anxiety in well-controlled adolescents. Parents, who were

predominantly mothers, of those in good control reported

greater family cohesion, less conflict, and more

encouragement of independence. Parents of children in poor

control indicated that diabetes had had a significantly

negative impact on their child's school experience and

social functioning.

A longitudinal study of 84 adolescents and adults with

IDDM (age 13-41 years), who were followed at 6-week

intervals for 36 weeks, investigated the relationship

between personality dimensions, emotional problems, life

stress, and metabolic control (Mazze, Lucido, & Shamoon,

1984). Psychological variables were measured by

standardized self-report instruments. Metabolic control was

assessed by HbAI assays. Gender, age, and SES, were

controlled in analyses. Personality variables were not

found to be related to glycemic control, while anxiety,

depression, and quality of life variables (ie, amount of










daily stress) were found to be significantly related to

metabolic control. Poorer control was related to

significantly greater anxiety, depression, and problems of

daily living than good control. Over the course of the

study, improvement in control was associated with

improvement in anxiety, depression, and quality of life,

while worsening control tended to be related with increased

anxiety, depression, and problems of daily living. These

findings seem to suggest that better metabolic control leads

to better adjustment; however, the authors expressed caution

in making that conclusion and encouraged further studies.

Grey, Genel, and Tamborlane (1980) investigated

the relationship between psychosocial adjustment, self-

esteem, family functioning, and metabolic control in 20

children and young adolescents with diabetes (age 7-13

years) and their families. The measures used were a

standardized interview and paper and pencil inventories.

Psychometric data on psychosocial measures were not

reported. Diabetes control was assessed using a 24-hour

urinary glucose excretion measure. Better psychosocial

adjustment was significantly correlated with higher parental

and child self-esteem and with optimal family functioning,

and these measures significantly discriminated between the

maladjusted group and the adjusted group. The maladjusted

group also had significantly greater 24-hour urinary glucose

excretion, indicating poorer control, compared with the










adjusted group. This study did not directly relate the

measure of family functioning to metabolic control.

In a study designed to examine developmental

differences in the relationships among a variety of

psychosocial factors, adherence, and metabolic control in

children and adolescents with diabetes, several significant

results were obtained (Burns, Green, & Chase, 1986).

Participants in the study were 72 youngsters, aged 8-16

years, and their parents. Information was obtained by

mother, father, and child report; data from parents was

examined both separately and combined. Measures for the

patient included diabetes knowledge, locus of control,

diabetes adjustment, self-esteem, social competence and

behavior problems, perceived competence, and adherence, and

for the parents, depression, anxiety, family functioning,

daily life stress, and parental discipline. These patient

and family measures were assessed through structured

interview, standardized and normed self- and parental-

report instruments, and research measures. Psychometric

data were not reported. Due to the number of measures,

tests of age-based sets of correlations were made; only when

those were significant were individual correlations within

sets examined. Results showed that for all ages combined,

there was a trend for the number of mother-reported behavior

problems to correlate positively with HbAl. For middle

adolescents, (aged 11-13), greater parental laxity and fewer

family rules were associated with worse control. No










correlation was noted between daily life stress and

metabolic control. Better adherence seemed to be associated

with better metabolic control, although the measure of

adherence was not described.

Hanson and colleagues described results from a large

research project which also investigated the relationship

between several psychosocial variables, adherence, and

metabolic control. Subjects in this study were 104

adolescents with IDDM (mean age=14.5) and their mothers

(Hanson, Henggeler, & Burghen, 1987c). The variables

assessed were life stress, social competence, and family

support. HbAlc was used as a measure of metabolic control

and self-report and observational methods were used to

measure adherence. Validity and reliability of the

adherence measure were not reported. Several standardized

paper and pencil measures were used to assess psychosocial

factors. Both stress and adherence were found to be related

to metabolic control; higher stress and poorer adherence

were associated with poorer metabolic control. Stress was

not significantly related to adherence. Social competence

significantly buffered the negative association between

stress and metabolic control. Adolescents with poor social

competence were in poor control under conditions of high

stress, while the metabolic control of adolescents with good

social competence was not influenced by stress. Parental

support was significantly related to adherence, and

adolescent age was significantly related to parental










support, with younger adolescents receiving more support

than older adolescents. The authors concluded that

adolescent age was indirectly linked to adherence through

parental support.

Subjects in a second report by these authors (Hanson,

Henggeler, & Burghen, 1987a) were 93 IDDM adolescents (mean

age=14.4) and their families. The psychosocial variables

assessed in this study were family knowledge about IDDM,

parents' marital satisfaction, family cohesion and

adaptability, family support of the diabetes regimen, life

stress, and the adolescent's social competence. Adherence

and metabolic control were also measured, using self-report

and observational items, and HbAlc values, respectively, as

in the previous study. Psychosocial measures were obtained

from both the adolescent and the parents, using validated

paper and pencil questionnaires. Measures were combined,

where necessary, to provide a single global index of seven

conceptual domains: family knowledge about IDDM, family

relations, chronic stress, social competence, adherence,

metabolic control, and adolescent age. Multiple regression

analyses showed that good adherence and low stress were

predictive of good metabolic control, while high family

knowledge about IDDM, positive family relations, and young

adolescent age were directly associated with good adherence.

Finally, in a study designed to investigate race and

sex differences in metabolic control, Hanson and colleagues

(1987b) again looked at the relationship between










psychosocial variables, adherence, and metabolic control.

Subjects in the study were 27 black and 27 white adolescents

with IDDM who were similar in age (mean=14.7 years), age at

diagnosis, and social class. The black female group had

worse metabolic control, as measured by HbAlc, than each of

other groups. In order to determine which of several

psychosocial variables were associated with their poor

metabolic control, multivariate analyses of variance were

performed. The variables of interest were knowledge of

diabetes, perceived competence, coping patterns, family

adaptability and cohesion, family support, life stress,

maternal social support, and satisfaction with the health

care system, obtained by self-report measures. Adherence

was measured by self-report and observation. Results from

this study, unlike those previously reported by these

authors, found no significant correlations between any of

these factors and metabolic control or between adherence and

control. The relationship between adherence and

psychosocial variables was not reported.

As has been indicated, previous research has suffered

from a variety of methodological problems including reliance

on self-report or interview data, measures of metabolic

control which vary from study to study, failure to report

reliability and validity of measures, and failure to equate

comparison groups on variables such as age, pubertal

development, duration of disease, and socioeconomic status.

While many studies have examined individual relationships










between psychosocial variables and health outcome measures,

few have attempted to examine the relationships among all

significant parameters. Of those which have, valid and

reliable measurement of relevant variables, particularly of

adherence, has been lacking.

The purpose of the current research study was to

examine the relationships among psychological adjustment,

family functioning, adherence, and metabolic control in

adolescents with IDDM and their families. This study

attempted to correct for problems of previous studies by

using multiple measures of adjustment, family functioning,

adherence, and metabolic control which are reliable and

well-validated. Multiple respondents were used (i.e.,

mother, father, adolescent), and observational, as well as

questionnaire data, was collected.

A thorough review of the literature suggested a

possible model of the relationships between the various

factors associated with metabolic control (See Figure 1).

Three factors appeared to be directly related to metabolic

control; they were stress, adherence behaviors, and the

physiological changes of puberty. Two factors,

psychological adjustment and family functioning, appeared to

be indirectly related to glycemic control, through the

intervening variable of adherence, although some family

variables (i.e., conflict) may fall more easily under the

heading of stress. At least initially, however, family

variables were hypothesized to affect metabolic control









indirectly. Finally, family factors also seemed to be

related to the psychological adjustment of the child.

This study investigated the validity of this model.


Psychological
Adjustment





Family Factors


Stress




Adherence 3




Puberty


Metabolic
Control


Figure 1
Model of Factors Influencing Metabolic Control















METHODS

Participants

Participants in the study were 56 adolescents with

insulin-dependent diabetes mellitus (IDDM), and their

parentss. Fifty-five mothers and 41 fathers participated

in the study. Adolescents' mean (+/-SD) age was 14.4 +/-

1.7 years (range = 12 to 18 years) and their mean duration

(+/-SD) of IDDM was 6.2 +/- 3.6 years (range = 2-16 years).

Fifty-seven percent of the adolescents were male, and all

but one adolescent were white. Most (79.6%) of the

adolescents were from two-parent households. The sample was

predominantly middle class, as determined by the

Hollingshead (1957) two-factor index of socioeconomic status

(Class I = 9.4%, Class II = 18.9%, Class III = 28.3%, Class

IV = 35.8%, Class V = 7.5%).

Adolescents with diabetes of less than 2 years were

excluded from the study because these patients may have been

producing some endogenous insulin. Also excluded were

adolescents from families who did not own telephones (needed

for data collection purposes), adolescents with other

significant medical problems in addition to diabetes, and

adolescents from families where more than one child had

diabetes or another significant medical problem.

Approximately seventy percent of adolescents and families










contacted agreed to participate in the study. The most

common reasons for refusal included lack of time and prior

participation in a research project.

Measures

General Information

A general information questionnaire was given to obtain

basic demographic information about the adolescent and

his/her family, including age, gender, family structure, and

socioeconomic status.

Adjustment Measures

Child Behavior Checklist. The Child Behavior Checklist

(CBCL; Achenbach & Edelbrock, 1983) was given to assess

behavior problems or symptoms as reported by parent or child

(A teacher version is also available). The CBCL is a 118-

item list of behaviors which are rated as 0, 1, or 2 in

terms of severity. The instrument generates a total symptom

count (number of problems checked), a severity score (sum of

ratings) and subscale scores for "Internalizing"

symptomatology and "Externalizing" symptomatology. The

subscales have been developed through prior, large scale

factor analyses; broad-band factors--internalizing and

externalizing--hold up across age, sex, and version of

checklist, while narrow-band factors vary according to

sample. Internalizing symptoms include behaviors such as

depression and withdrawal. Externalizing symptoms include

impulsive behavior, conduct problems, and aggressive

behavior.










In addition to symptoms, twenty items are used to

assess a range of social competencies, including school,

peer-related, and social activities. Extensive data on the

reliability and validity of the parent, teacher, and youth

report forms have been presented (Achenbach & Edelbrock,

1983, 1987).

Stress Measures

Hassles Scale. (Kanner, Coyne, Schaefer, & Lazarus,

1981). This scale was developed to assess daily sources of

stress. The scale consists of 117 "hassles" such as

problems with money, school or work difficulties, and social

concerns. Respondents determined which items were hassles

to them during the preceding month, and then indicated the

frequency and severity of the hassle on 3-point scales.

Frequency and severity ratings were thus obtained from the

instrument.

Two versions of the Hassles Scale were used in this

study. The original version, developed by Kanner et al.

(1981) was used with parents, while adolescents completed a

modified version which rewords certain items to increase

understanding, and replaces irrelevant items with more

relevant "hassles". The Hassles Scale has been shown to be

reliable and valid in studies with adults (Kanner et al.,

1981; Cox et al., 1984).

Life Events Checklist. (LEC; Johnson & McCutcheon,

1980). This measure was designed to assess major life

stress; it consists of 46 life events (plus 4 spaces










provided for indicating significant events experienced but

not included on the scale) which were chosen to represent

life changes frequently experienced by children and

adolescents. Respondents were asked to indicate which of

the events they have experienced in the past year, and to

rate each of these events as good or bad. A positive life

change score is obtained by summing the events rated as

desirable; events rated as undesirable are summed to obtain

a negative life change score. Reliability and validity of

this measure have been demonstrated (Brand & Johnson, 1982;

Gad & Johnson, 1980; Greene, Walker, Hickson, & Thompson,

1985; Johnson, 1982, 1986; Johnson & McCutcheon, 1980).

Social Readiustment Rating Scale. (Holmes & Rahe,

1967). This measure was also designed to assess major life

stress; 54 items are listed which describe significant life

events relevant to adults, such as births, deaths, or major

changes in employment or marital status. Respondents

determined which events they had experienced during the past

year, and rated each of the events as positive or negative.

Adequate reliability and validity have been reported for

this measure (Grant, Sweetwood, Gerst, & Yager, 1978; Lei &

Skinner, 1980).

Family Measures

Family Environment Scale. (FES; Moos & Moos, 1981,

1983). This instrument was used to assess the social

environment of families as perceived by individual family

members. It is a 90-item true-false test which yields










standard scores on 10 subscales. The subscales are divided

into three dimensions. The Personal Development (Personal

Growth) dimension includes five subscales: Independence,

Achievement Orientation, Intellectual-Cultural Orientation,

Activity-Recreational Orientation, and Moral-Religious

Orientation. The Relationship dimension, which provides a

description of interpersonal relationships among family

members, consists of the Cohesion, Expressiveness, and

Conflict subscales. The Organization and Control subscales

make up the System Maintenance dimension, which measures the

basic structure or "system" of the family.

Adequate reliability and validity have been reported

for the FES (Moos & Moos, 1981). Average subscale

intercorrelations are approximately .20; these indicate that

the scales are measuring distinct, though somewhat related,

facets of the family social environment.

Family Behavior Checklist. (Glasgow, McCaul, &

Schafer, 1985). This instrument was designed to assess the

frequency of both supportive and nonsupportive behaviors

directed towards persons with diabetes by family members.

The measure is specific to behaviors related to the diabetic

regimen. It consists of 16 items which are rated for

frequency on a 5-point Likert scale; from these a positive

(supportive) score and a negative (nonsupportive) score can

be obtained. The authors have provided data on the

reliability and validity of this measure. Although the

authors have urged caution in use of this instrument with










adolescents (Schafer et al., 1986), other investigators have

demonstrated its utility with this population (Hanson et

al., 1987b; Kurtz & Delamater, 1984).

Conflict Behavior Questionnaire. (Prinz, Foster, Kent,

& O'Leary, 1979). The Conflict Behavior Questionnaire was

designed to obtain evaluations of parent and adolescent

communication. It consists of 75 dichotomous (yes-no) items

assessing communication-conflict behavior. For each family

member, the CBQ provides two types of information:

1)appraisal of the other member's behavior and 2)appraisal

of the dyadic interaction. High scores represent negative

appraisals and low scores represent positive appraisals.

Data on the reliability and validity of the CBQ have been

presented (Foster & Robin, 1988; Prinz et al., 1979; Robin &

Foster, 1989; Robin & Weiss, 1980).

Issues Checklist. (Prinz et al., 1979). The Issues

Checklist was designed to assess the frequency and intensity

of discussions associated with specific family issues.

Parents and adolescents are asked to recall disagreements

about 44 specific issues such as curfew, chores, smoking,

and drugs. For each topic, the respondent indicates whether

the issue had been discussed during the previous four weeks;

if affirmative, the respondent rates the intensity of the

discussions on a 5-point scale ranging from calm to angry,

and estimates how often the topic was raised.

The Issues Checklist yields three scores for each

family member: 1) the quantity of issues discussed, 2) the










mean anger-intensity level of the issues endorsed, and 3)

the weighted average of the frequency and anger-intensity

level of the issues endorsed. High anger-intensity and

weighted frequency by anger-intensity scores are indicative

of angry arguments while low scores are indicative of calm

discussions. Several studies have presented data on the

reliability and validity of this measure (Foster, Prinz, &

O'Leary, 1983; Prinz et al., 1979; Robin & Foster, 1984;

Robin & Weiss, 1980).

Diabetes Issues Checklist. (Johnson, 1986). The

Diabetes Issues Checklist is a modified version of the

Issues Checklist; it was designed to assess the frequency

and intensity of discussions associated with issues

specifically related to diabetes. Similar to to the Issues

Checklist, parents and adolescents were asked to recall

disagreements about 40 issues specific to diabetes, such as

diet, exercise, and injections. The same procedure as for

the Issues Checklist was then followed; the frequency and

intensity of the discussions regarding each relevant issue

were rated and three scores were generated.

Behavioral Interaction Task. A sample of each

family's problem-solving communication behavior was

collected by asking a parent and the adolescent to discuss

and attempt to resolve two problems for 10 minutes each.

One item each from the Issues Checklist and Diabetes Issues

Checklist was selected for discussion; specific guidelines

for choosing the most conflictual item were followed. When










possible, the parent who acted as primary caretaker

participated in the problem-solving discussions.

Each problem-solving discussion was tape-recorded and

then rated using the Interaction Behavior Code (IBC; Prinz &

Kent, 1978). Several coders, who were blind to family

status, listened to an entire discussion, then completed

dichotomous (yes-no) or trichotomous (no-a little-a lot)

ratings of 31 discrete problem-solving communication

behaviors. Also completed were Likert ratings of degree of

insult, friendliness, communication, and problem resolution.

From these scores, eight summary scores were obtained:

Negative Parental Behavior, Positive Parental Behavior,

Negative Adolescent Behavior, Positive Adolescent Behavior,

Degree of Resolution, Insult, Friendliness, and General

Problem-Solving Effectiveness. Reliability and validity

with this coding method have been demonstrated (Foster et

al., 1983; Prinz & Kent, 1978; Prinz, Rosenblum, & O'Leary,

1978).

Dyadic Adjustment Scale. (Spanier, 1976). This

measure was developed to assess the quality of marriage and

other similar dyads. It consists of 32 items which assess

four components of dyadic adjustment--dyadic satisfaction,

dyadic cohesion, dyadic consensus, and affectional

expression. Higher scores are indicative of better marital

adjustment. Evidence of content, criterion-related, and

construct validity, as well as of high scale reliability has

been reported (Spanier, 1976).










Adherence Measures

Twenty-four Hour Recall Interview. (Johnson et al.,

1986). This measure is a modified version of the 24-hour

recall interview used to assess dietary intake. Information

regarding the adolescent's usual daily diabetes management

behaviors was obtained by asking the patient to recall the

day's events in tQmporal sequence, beginning with the time

he/she woke up in the morning and ending with retiring to

bed. All diabetes relevant behaviors are recorded by the

interviewer. Three interviews were conducted with each

respondent to obtain a more representative sample of the

patient's usual behavior, two focusing on weekday activities

and one on weekend behavior. Both the adolescent and the

primary caretaker were interviewed separately regarding the

adolescent's behaviors.

From the data obtained on the interview forms, thirteen

separate adherence measures were calculated as described by

Johnson et al. (1986). All measures were constructed so

that a range of scores was possible, with higher scores

indicating relative noncompliance and scores close to zero

indicating relative compliance. Previous factor-analytic

work suggested that these 13 measures should be grouped into

six adherence factors (Johnson et al., 1986; Johnson, Tomer,

Cunningham, & Henretta, 1990).

Factor scores were calculated by first standardizing

the relevant adherence measures to the 1982 sample described

by Johnson et al. (1986), on which the original factor-










analytic work was based, and then averaging the standardized

scores, to correct for those factors which required

combining several adherence measures based on different

measurement scales.

The Injection factor was composed of four measures:

injection regularity (the degree to which injections are

given at the same time every day), injection interval (the

degree to which the time between injections approaches

ideal), injection-meal timing (the degree to which

injections are given 30 to 60 minutes before eating), and

regularity of injection-meal timing (the degree to which the

time between injection and eating is consistent across

days).

The Exercise factor was composed of three measures:

exercise frequency (how often the adolescent exercised),

exercise duration (how long the adolescent exercised), and

exercise type (the strenuousness of the adolescent's

exercise).

The Diet Type factor was composed of two measures:

percentage of calories from carbohydrates (in relation to

the 60% ideal recommended by the American Diabetes

Association; Nuttall & Brunzall, 1979) and percentage of

calories from fat (in relation to the 25% ideal recommended

by the American Diabetes Association; Nuttall & Brunzall,

1979).

The Testing/Eating Frequency factor was composed of two

measures: testing frequencing (how often the adolescent










conducted a glucose test on a daily basis) and eating

frequency (how often a adolescent ate on a daily basis).

The Calories Consumed factor consisted of the

adolescent's ideal total number of daily calories (based on

age, sex, and height) subtracted from the adolescent's

reported daily calorie consumption.

The Concentrated Sweets factor consisted of the average

number of concentrated sweet exchange units eaten on a daily

basis (40 calories of any concentrated sweet was equivalent

to one concentrated sweet exchange unit).

Data on the reliability and validity of these measures

have been presented (Freund, Johnson, Silverstein, & Thomas,

1991; Johnson et al., 1986; Johnson, Freund, Silverstein,

Hansen, & Malone, 1990; Johnson, Tomer, Cunningham, &

Henretta, 1990; Reynolds, Johnson, & Silverstein, 1990).

Pubertal Status

Pubertal status was assessed by obtaining a physician's

judgement as to the Tanner stage (Tanner, 1962; Marshall &

Tanner, 1969, 1970) of pubertal development which the

adolescent had reached. In girls, breast development and

pubic hair growth were assessed and rated into one of five

stages ranging from Tanner I (i.e., prepubertal) to Tanner V

(mature adult). In boys, genitalia (i.e., testes, scrotum,

and penis) and pubic hair growth were similarly rated into

one of five stages.










Metabolic control

Metabolic control was assessed in several ways.

Glycosylated hemoglobin (Hemoglobin Alc) values, which

reflect the percentage of glucose molecules that attach to

the body's hemoglobin over the course of 60 days, and which

are generally considered the best overall measure of

metabolic control, were obtained. This measure provides an

index of average blood glucose levels over the past 2 to 3

months (Ziel & Davidson, 1987). HbAlc was assayed using

column chromatography (BIO-RAD). Normal values in our

laboratory range from 3.5% to 6.1%.

Triglyceride and cholesterol levels were also used as

measures of metabolic control, as they are often elevated in

youngsters with poorly controlled diabetes (Glasgow, August,

& Hung, 1981; Lopes-Virella, Wohltmann, Loadholt, & Buse,

1981; Peterson, Koenig, Jones, Saudek, & Cerami, 1977;

Sosenko, Breslow, Miettinen, & Gabbay, 1980). Triglyceride

levels were determined by SmithKline Laboratories using the

Technicon enzymatic method; normal adult values range from

33 to 111 mg/dl. Cholesterol levels were also assayed by

SmithKline Laboratories using the Technicon enzymatic

method; normal adult values range from 120 to 200 mg/dl.

Procedures

Participants in the study were recruited through the

Pediatric Endocrinology Clinic at Shands Hospital and

through the 1989 and 1990 sessions of Florida's Camp for

Children and Youth with Diabetes. Participants and their










parents were initially contacted by telephone, approximately

one month prior to their clinic visit or to the beginning of

camp, to explain the study and to obtain permission to

participate. Once verbal consent was obtained, the

participants were mailed a packet containing an informed

consent form and several questionnaires, to be completed by

each adolescent and both parents, when possible: a General

Information Form, the Child Behavior Checklist (both Youth

Report and Parent Report versions), the Family Environment

Scale, the Diabetes Family Behavior Checklist, and the Life

Events Checklist (adolescent) and Social Readjustment Rating

Scale (parents). In two-parent households, the Dyadic

Adjustment Scale was also included. Participants were asked

to return the completed forms in pre-addressed, stamped

envelopes included with their packets.

During the month prior to the clinic visit (or the

beginning of camp), each adolescent and the primary

caretaker were telephoned to obtain three 24-hour recall

interviews. Approximately one week before the clinic visit

(or the beginning of camp), a second packet of

questionnaires was sent which included: the Issues

Checklist, Diabetes Issues Checklist, Conflict Behavior

Questionnaire, and Hassles Scale, for each adolescent and

his/her parents) to complete, relative to the previous four

weeks. During the clinic visit, a blood sample was drawn,

in order to obtain HbAlc, triglycerides, and cholesterol

levels, and relevant information, including Tanner stage,










insulin dosage, and duration of diabetes, were obtained from

the physician. All adolescents were asked to bring in

recorded chemstrip results for the past four weeks, as well;

however, compliance with that request was minimal. In

addition, the adolescent and one parent (preferably the

primary caregiver) participated in a behavioral interaction

task. A teacher report form of the Child Behavior Checklist,

and a questionnaire regarding school absences and course

grades, were initially to one teacher of each adolescent,

after obtaining a release of information from the adolescent

and parents. However, the return rate was so low that the

attempt to obtain school information was abandoned.















RESULTS

Adjustment Measures

Adjustment was measured with the parent report and

adolescent self-report versions of the Child Behavior

Checklist. Relative to the normative samples, adolescent

adjustment in this sample fell within the normal range

(Achenbach & Edelbrock, 1983). Table 1 presents means and

standard deviations of social competence and behavior

problem scores as obtained from mothers, fathers, and

adolescents.

Agreement among respondents was assessed by using

Pearson product moment correlations and repeated measures

analyses of variance (ANOVAS). Significant correlations

were found between mothers' and fathers' T-scores of total

competence (r=.50, p<.001, n=40), total behavior problems

(r=.72, p<.0001, n=40), total internalizing problems (r=.70,

p<.0001, n=40), and total externalizing problems (r=.74,

p<.0001, n=40). Significant correlations between mothers'

and adolescents' T-scores were found for total behavior

problems (r=.29, p<.03, n=54) and total externalizing

problems (r=.41, p<.002, n=54). Only one significant

correlation was found between fathers' and adolescents' T-

scores: total externalizing problems (r=.33, p<.04, n=40).









Table 1
Means and Standard Deviations of CBCL T-Scoresb

Mothers Fathers Adolescents
(n=55) (n=41) (n=55)

CBCLa M (SD) M (SD) M (SD)

Total
Social Competence 43.6 (20.0) 43.1 (18.2) 54.0 (12.2)

Total
Behavior Problems 56.8 (9.6) 56.2 (10.4) 46.3 (10.9)

Total Internalizing
Problems 56.6 (8.5) 55.2 (9.7) 47.3 (11.0)

Total Externalizing
Problems 55.7 (8.2) 55.6 (9.0) 45.7 (10.3)
aEBCL = Child Behavior Checklist
For normalized T-scores, Mean = 50 and SD = 10.


Table 2
Means and Standard Deviations of Life Events and Daily Hassles
Scores

Mothers Fathers Adolescents
(n=53) (n=36,38) (n=56,51)

M (SD) M (SD) M (SD)

Positive
Life Events 3.5 (2.4) 2.9 (2.1) 3.8 (2.0)

Negative
Life Events 1.9 (2.6) 1.3 (1.7) 2.8 (2.6)

Total
Life Events 5.6 (3.6) 4.5 (3.0) 6.7 (3.8)
[KR20] [.67] [.62] [.65]

Quantity of
Hassles 21.2 (16.4) 16.5 (12.1) 25.3 (24.2)
[KR20] [.96] [.94] [.99]

Severity of
Hassles 1.5 (0.4) 1.5 (0.5) 1.5 (0.5)
[Co-Alpha] [.97] [.89] [.96]

Severity by Frequency
of Hassles 2.5 (1.4) 2.5 (1.7) 2.8 (1.6)

Note. n's are variable as some forms were not completed by all.










There were significant respondent effects for T-scores of

total social competence, F (2,35)=8.86, p<.0008, total

behavior problems, F (2,37)=16.81, p<.0001, total

internalizing problems, F (2,37)=12.08, p<.0001, and total

externalizing problems, F (2,37)=16.19, p<.0001.

Adolescents reported significantly greater social competence

than did mothers' (mean difference=11.14, t=3.58, p<.0008)

and fathers' (mean difference=11.70, t=3.62, p<.0009), while

mothers' and fathers' scores were not significantly

different from each other. A similar pattern was found for

total problem scores and externalizing scores: adolescents

reported significantly fewer total problem scores and

externalizing scores than did mothers (mean difference for

total=-10.30, t=-6.07, p<.0001; mean difference for

externalizing=-9.74, t=-6.91, p<.0001), and fathers (mean

difference for total=-8.23, t=-3.68, p<.0007; mean

difference for externalizing=-8.69, t=-4.46, p<.0001), while

mothers and fathers did not differ. Mothers reported

significantly more internalizing problems than did

adolescents (mean difference=-9.17, t=-4.92, p<.0001) or

fathers (mean difference=2.72, t=2.35, p<.02). Fathers also

reported significantly more internalizing problems than did

adolescents (mean difference=-6.59, t=-2.91, p<.006).

Across measures of adjustment, adolescents reported

greater social competence and fewer problems than did their

parents.










Stress Measures

Major life events were measured with the Life Events

Checklist for adolescents, and the Social Readjustment

Rating Scale for parents. Daily stresses were measured with

the Hassles Scale. Internal consistencies for Total Life

Event scores on the Life Events Checklist and on the Social

Readjustment Rating Scale were adequate (KR20 for

adolescents = .65, KR20 for parents = .65; see Table 2).

Internal consistencies for the Quantity and Severity scores

of the Hassles Scales were excellent (overall, KR20 for

Quantity of Hassles = .96, and coefficient alpha for

Severity of Hassles = .96; see Table 2).

Relative to a normative sample, adolescents in this

sample reported major life event scores within the normal

range (Brand & Johnson, 1982; Johnson & McCutchen, 1980).

No normative data is available on adolescents' daily hassles

scores. Parents in this sample obtained major life event

scores and daily hassles scores within the normal range

(Grant et al., 1978; Kanner et al., 1981). Table 2 presents

means and standard deviations of positive, negative, and

total life events, and quantity, severity, and severity by

frequency of daily hassles, as obtained from mothers,

fathers, and adolescents.

Agreement among respondents was assessed by using

Pearson product moment correlations and repeated measures

analyses of variance (ANOVA's). Significant correlations

were found between mothers' and fathers' scores of positive










(r=.53, p<.001, n=35), negative (r=.38, p<.02, n=35), and

total (r=.46, p<.006, n=34) life events. Significant

correlations were also found between mothers' and fathers'

scores of quantity (r=.34, p<.05, n=36) and severity by

frequency (r=.37, p<.03, n=35) of daily hassles. Trends for

agreement were found between mothers' and adolescents'

negative (r=.25, p<.07, n=53) and total (r=.24, r<.09, n=52)

life events. Mothers' and adolescents' scores of quantity

of daily hassles were significantly correlated (r=.31,

p<.03, n=49). No significant correlations were found

between fathers' and adolescents' life event scores or daily

hassles scores. (However, the correlation between fathers'

and adolescents' total life events scores was r=.27, p<.11,

n=35.)

There were significant respondent effects for positive,

F (2,33)=3.35, p<.05, negative, F (2,33)=3.70, p<.04, and

total life events, F (2,32)=6.25, p<.005. Adolescents

reported significantly more positive (mean difference=1.14,

t=2.61, p<.01), negative (mean difference=1.25, t=2.41,

p<.02), and total life events (mean difference=2.35, t=3.24,

p<.003) than did fathers. Mothers reported significantly

more total life events (mean difference=1.15, t=2.06, p<.05)

than did fathers.

A significant respondent effect was found for the

severity by frequency score of daily hassles, F (2,30)=3.36,

p<.05. Further analysis revealed only a trend for

difference between adolescents and mothers (mean










difference=.53, t=1.72, p<.09), and no differences between

adolescents and fathers, or mothers and fathers.

Adolescents reported more life events, both positive

and negative, and more hassles, than their parents. Mothers

reported more life events than fathers.

Family Measures

Family Environment Scale

Family social environment was measured with the Family

Environment Scale. Internal consistencies for subscales of

this measure for this sample were extremely poor, ranging

from KR20 = -.13 for Control to KR20) = .34 for

Independence, for the combined sample of adolescents and

parents (see Table 3 for KR20s for individual respondents).

Relative to the normative sample, the scores of individual

family members in this sample fell within the normal range

on this measure, with the exception of the subscale of

Independence, where scores fell significantly below average

(Moos & Moos, 1981). Table 3 presents means and standard

deviations of subscale scores as obtained from mothers,

fathers, and adolescents.

Agreement among respondents was assessed by using

Pearson product moment correlations and repeated measures

analyses of variance (ANOVAS). Significant correlations

were found between mothers' and fathers' subscale scores for

Cohesion (r=.39, p<.02, n=39), Conflict (r=.35, p<.03,

n=39), Intellectual/Cultural (r=.39, p<.01, n=39), and









Table 3
Means and Standard Deviations of FES Subscale Scores


Mothers
(n=55)

M (SD)


Cohesion
[KR20]

Expressiveness
[KR20]F

Conflict
[KR20]

Independence
[KR20]

Achievement
Orientation
[KR20]

Intellectual/
Cultural
[KR20]

Activity/
Recreational
[KR20]

Moral/Religious
[KR20]

Organization
[KR20]


Control
[KR20]


5.85 (0.97)
[-.24]

5.49 (1.30)
[-.06]

3.0 (1.17)
[-.11]

3.76 (1.39)
[.36]


4.56 (1.21)
[.15]


4.92 (1.37)
[-.009]


4.84 (1.30)
[-.16]

4.93 (1.09)
[ .18]

4.91 (1.31)
[ .04]

4.91 (1.31)
[-.13]


Fathers
(n=40)

M (SD)


5.65 (1.27)
[.09]

5.15 (1.35)
[-.13]

3.25 (1.13)
[ .05]

4.13 (1.47)
[ .27]


4.20 (1.38)
[.19]


5.23 (1.37)
[-.02]


4.75 (1.53)
[ .18]

5.03 (1.19)
[-.04]

4.43 (1.60)
[.22]

4.43 (1.60)
[ .05]


FESa


5.41 (1.44)
[ -.21]


aFES = Family Environment Scale


Adolescents
(n=56)

M (SD)


5.66 (1.56)
[.31]

4.68 (1.34)
[-.09]

3.46 (1.35)
[-.24]

3.98 (1.66)
[.40]


4.96 (1.33)
[.21]


4.50 (1.33)
[-.101


5.13 (1.39)
[.001]

4.98 (1.42)
[.15]

5.41 (1.44)
[.19]










Moral/Religious (4=.47, p<.003, n=39). They showed trends

for agreement for Expressiveness (r=.27, p<.09, n=39),

Achievement Orientation (r=.29, p<.08, n=39),

Activity/Recreational (r=.27, p<.09, n=39), and Organization

(r=.30, p<.06, n=39). Significant correlations between

mothers' and adolescents' subscale scores were found for

Moral/Religious (r=.46, p<.0004, n=55) and Organization

(r=.34, p<.01, n=55), with trends for Independence (r=.23,

p<.10, n=55) and Activity/Recreational (r=.24, p<.08, n=55).

Only one significant correlation was found between fathers'

and adolescents' subscale scores: Moral/Religious (r=.33,

p<.04, n=40). However, there were trends for Independence

(r=.28, p<.08, n=40) and Intellectual/Cultural (r=-.27,

p<.09, n=40).

There were significant respondent effects for

Expressiveness, F (2,37)=4.21, p<.02, Achievement

Orientation, F (2,37)=7.63, p<.002, and Organization, F

(2,37)=4.07, p<.03, subscale scores. Adolescents reported

lower Expressiveness than did their mothers (mean

difference=-.74, t=-3.28, p<.002), higher Achievement

Orientation than did their fathers (mean difference=1.0,

t=3.70, p<.0007), and greater Organization than did their

mothers (mean difference=.48, t=2.2, p<.03) and their

fathers (mean difference=.79, t=2.6, p<.01).

Mothers and fathers reported similar perceptions of

their family environment. Adolescents, in contrast to their

parents, viewed their families as less open to emotional










expression, more oriented towards achievement, and more

organized and structured.

Diabetes Family Behavior Checklist

Family supportive behaviors were measured with the

Diabetes Family Behavior Checklist. Internal consistencies

for subscale score with this measure were adequate (overall,

coefficient alpha for positive scores = .77, coefficient

alpha for negative scores = .67; see Table 4). Relative to

normative samples, the scores of individual family members'

in this sample fell within the normal range on this measure

(Glasgow et al., 1985; Schafer et al., 1983; McCaul,

Glasgow, & Schafer, 1987). Table 4 presents means and

standard deviations of subscale scores as obtained from

mothers, fathers, and adolescents.

Agreement among respondents was assessed by using

Pearson product moment correlations and repeated measures

analyses of variance (ANOVAS). A significant correlation

was found between mothers' and fathers' subscale scores for

negative behaviors (r=.42, p<.007, n=40). They showed a

trend for agreement for positive behaviors (r=.26, p<.10,

n=40). Significant correlations between mothers' and

adolescents' subscale scores were found for both negative

(r=.55, p<.0001, n=53) and positive (r=.57, p<.0001, n=53)

behaviors. Fathers' and adolescents' subscale scores for

negative behaviors were not significantly correlated (r=.25,

p<.12, n=39). However, there was a significant correlation

between fathers' and adolescents' subscale scores for

















4-)






41


U) II








ro.0
U) E -M
OO0

t (0


0 0 tN


00 0 0

0) 00 11 m ro 11



o 3' '.o
(N1







(N O v



CN (




(N r


m
VII
i- -

a)
O C
















Z4
a) 0 II



0 C
(1) Q) t
.-i> ~- 11
0

















CO
U
031


(N co



0 0 1 11 0
0-I 0 II 0 II



(N i


'0 C--
.- (N




(N.* co *
0)tM 0 r-
If) C.-

(N -


oo0 0 a
*H H -iH H
>



4- -W
) U a) U
> -H > --
-H 4-4 -H-|44
J 4-i4 4-i44
-H O) o a) 0
U 0 0)0 0
0 U ( U 3
a4 Z-; U


o 00






o
Co






0
a>)

0
q-I

0





edo


(n

(N

m LO
oD r- 11
(N c

(N


cO


0


LO0"







Cf)
'.0





0


(-


rC-



co
o-










(N
C--


(~"0

m (.


To
C2 04

Q O
4-4 4-1 4-4
0 4-4 0 C
0 0
,-H >, -H

U 4- -H -H
-H- -H- ) r- U) U 4-4
(d -.1) (U0 C W -4
SN- z 3 a) :-3 .a)-

^ u z a) En u)
< H M 4H H -


m
C--

co


(N1








co

o
(N

0



' 100 o
or (7


LO













-


>4
r-4
.r-4


uA .4

O 0


C
*H
E-4 ul

0
-1


LO















U)
c-

















0
1'










U4
(0 (1

-C-ia

















4-)

S- ---- II

OO0 --













) II

0 0


0 0Q o
.-J II




40







E E
lq M






W)Q i
4 i
-u '- n
LOs


cool

0


LO




0
P--


in
r-1
If-0
LO)


4-4
SOH


S03 -
4-0 0) 0


OD












0










0 U

i r--

C U)




0U)(
Ha


-.
























(U)

a)




-H




-r4 0)
C: -i. -H


4-W
4J




0 H
a) C
U C
0 U)

0 4-1- (





0 0
*H U) H






> 0a) -


0 3 U
rX4 U)
U ) i U

a) u)
4 11 a)
a 4) 4 -10




S0 H
0 U H 0


m m u


*a
0


on










ND


--
iu)

n *
* i


n0


CN
) *










-


r---

o*-










positive behaviors (r=.42, p<.008, n=39). Adolescents'

viewed their parents as similar. Their subscale scores

rating their mother's behaviors were significantly

correlated with those rating their father's behaviors for

both negative (r=.79, p<.0001, n=41) and positive (r=.87,

p<.0001, n=41) behaviors.

There were significant respondent effects for both

negative, F(3,35)=6.99, p<.0008, and positive,

F(3,35)=.0001, behaviors. Adolescents reported

significantly fewer negative (mean difference=-1.72, t=

-2.48, p<.02), and positive (mean difference=-2.3, t=-3.18,

p<.003) behaviors than did their mothers, and mothers

reported significantly more negative (mean difference=2.28,

t=2.87, p<.007) and positive (mean difference=4.49, t=4.18,

p<.0002) behaviors than did fathers. Adolescents' viewed

their mothers as exhibiting significantly more negative

(mean difference = 1.80, t=3.60, p<.0009) and positive (mean

difference=2.59, t=4.76, p<.0001) behaviors than their

fathers.

While adolescents acknowledged that their mothers

exhibit more supportive and unsupportive behaviors than

their fathers, they did not view their mothers as exhibiting

as many positive and negative behaviors as their mothers

themselves report. Fathers did report fewer supportive and

fewer unsupportive behaviors than did mothers.










Conflict Behavior Questionnaire

Evaluation of parent and adolescent communication was

assessed with the Conflict Behavior Questionnaire. Internal

consistencies for this measure with this sample were

adequate to good (for adolescents, KR20 for appraisal of

parent = .86, KR20 for appraisal of dyad = .79; for parents,

KR20 for appraisal of adolescent = .88, KR20 for appraisal

of dyad = .76; see Table 4). Relative to the normative

sample, the scores of individual family members' in this

sample fell within normal limits on this measure (Robin &

Foster, 1989). Table 4 presents means and standard

deviations of subscale scores as obtained from mothers,

fathers, and adolescents. Higher scores are indicative of

more negative parent/adolescent interactions.

Agreement among respondents was assessed by using

Pearson product moment correlations and repeated measures

analyses of variance (ANOVAS). Significant correlations

were found between mothers' and fathers' subscale scores for

appraisal of adolescent (r=.60, p<.0001, n=39) and

appraisal of dyad (r=.38, p<.02, n=39). Significant

correlations between mothers' and adolescents' subscale

scores were also found for appraisal of other (r=..28,

p<.04, n=53) and appraisal of dyad (r=.51, p<.0001, n=53).

Fathers' and adolescents' subscale scores for appraisal of

other (r=.41, p<.009, n=39) and for appraisal of dyad

(r=.48, p<.002, n=39) were significantly correlated.

Adolescents viewed their interactions with their parents










similarly. Their subscale scores rating their mothers were

significantly correlated with those rating their fathers for

both appraisal of other (r=.48, p<.001, n=41) and appraisal

of dyad (r=.33, p<.03, n=41).

There were no significant respondent effects for either

the appraisal of other, F(3,35)=.48, p<.70, or the appraisal

of dyad, F(3,35)=2.30, p<.09, subscale scores.

Adolescents and their parents were alike in their

perceptions of their communication with each other. Mothers

and fathers were perceived by adolescents as behaving

similarly in their interactions.

Issues Checklist

Aspects of family discussions associated with specific

issues were assessed with the Issues Checklist and the

Diabetes Issues Checklist. Internal consistencies for the

Issues Checklist with this sample were excellent (overall,

KR20 for Quantity of Issues = .92, Coefficient Alpha for

Intensity of Issues = .91; see Table 4). Relative to the

normative sample, the scores of individual family members'

in this sample fell within one standard deviation of the

nonclinic sample on this measure (Robin & Foster, 1989).

Only for the quantity of issues endorsed by fathers was the

mean score closer to the clinic sample mean (M=18.38,

SD=5.05) than for the nonclinic sample mean (M=11.6,

SD=4.6). Table 4 presents means and standard deviations of

subscale scores as obtained from mothers, fathers, and

adolescents.










Agreement among respondents was assessed by using

Pearson product moment correlations and repeated measures

analyses of variance (ANOVAS). A significant correlation

was found between mothers' and fathers' score for quantity

of issues (r=.53, p<.0005, n=39). They showed a trend for

agreement for intensity of issues (r=.30, p<.07, n=39).

Significant correlations between mothers' and adolescents'

scores were found for both quantity (r=.55, p<.0001, n=53)

and intensity (r=.36, p<.007, n=53) of issues. Fathers' and

adolescents' scores for quantity of issues were not

significantly correlated (r=.30, p<.06, n=39). However,

there were significant correlations between fathers' and

adolescents' scores for intensity (r=.43, p<.007, n=38) and

intensity by frequency (r=.53, p<.0006, n=38) of issues.

Adolescents perceived their interactions with their parents

as similar. Their scores rating their mothers were

significantly correlated with those rating their fathers for

quantity (r=.69, p<.0001, n=41), intensity (r=.86, p<.0001,

n=40), and intensity by frequency (r=.62, p<.0001, n=40) of

issues.

There was a significant respondent effect for quantity

of issues, F(3,35)=17.43, p<.0001. Adolescents reported

significantly fewer issues than their mothers (mean

difference=-5.09, t=-3.82, p<.0004) and their fathers (mean

difference=-7.38, t=-4.08, p<.0002). Adolescents' reports

of the quantity of issues discussed with their mothers were

significantly greater than their reports of those discussed










with their fathers, as well (mean difference=5.10, t=3.99,

p<.0003).

Adolescents reported fewer issues than did both their

mothers and their fathers. They reported more issues with

their mothers than with their fathers, although this

difference was not reported by their parents. Adolescents

and parents reported similar intensity and frequency in

their discussions of issues.

Diabetes Issues Checklist

As with the Issues Checklist, internal consistencies

for this measure with this sample were excellent (overall,

KR20 for Quantity of Diabetes Issues = .94, Coefficient

Alpha for Intensity of Diabetes Issues = .94; see Table 4).

No normative data on this measure is currently available.

Relative to the normative sample for the Issues Checklist,

the scores of individual family members' in this sample fell

within one standard deviation of the nonclinic sample, with

the exception of the adolescents' reports of quantity of

diabetes issues discussed with both mother and father.

These scores were significantly less than those from the

normative sample for the Issues Checklist (Robin & Foster,

1989). Table 4 presents means and standard deviations of

subscale scores as obtained from mothers, fathers, and

adolescents.

Agreement among respondents was assessed by using

Pearson product moment correlations and repeated measures

analyses of variance (ANOVAS). Significant correlations










were found between mothers' and fathers' scores for quantity

(r=.55, p<.0003, n=39) and intensity (r=.53, p<.002, n=32)

of diabetes issues. A significant correlation between

mothers' and adolescents' scores were found for quantity of

diabetes issues (r=.49, p<.0002, n=52). Trends for

agreement were found for fathers' and adolescents' scores

for quantity (r=.31, p=.06, n=39) and intensity by frequency

(r=.38, p<.06, n=25) of diabetes issues. Again, adolescents

viewed their interactions with their parents as similar.

Their scores rating their mothers were significantly

correlated with those rating their fathers for quantity

(r=.78, p<.0001, n=41), intensity (r=.45, p<.02, n=27), and

intensity by frequency (r=.68, p<.0001, n=27) of diabetes

issues.

There was a significant respondent effect for quantity

of diabetes issues, F(3,35)=11.69, p<.0001. Adolescents

reported significantly fewer issues than their mothers (mean

difference=-6.19, t=-4.21, p<.0001) and than their fathers

(mean difference=-4.36, t=-2.31, p<.03), and mothers

reported significantly more issues than fathers (mean

difference =5.13, t=2.97, p<.005) in their report of

quantity of diabetes issues. Adolescents' reports of the

quantity of diabetes issues discussed with their mothers

were significantly greater than their reports of those

discussed with their fathers, as well (mean difference=3.27,

t=2.98, p<.005).










In comparing scores on the Issues Checklist with those

on the Diabetes Issues Checklist, when the respondents were

mothers, significant correlations were found for quantity

(r=.80, p<.001, n=57), intensity (r=.62, p<.0001, n=55), and

intensity by frequency (r=.82, p<.0001, n=55) of issues.

Mothers reported that general issues were significantly

greater in quantity (mean difference=2.59, t=2.80, p<.007),

intensity (mean difference=.25, t=2.95, p<.005), and

intensity by frequency (mean difference=4.68, t=3.39,

p<.001) than were diabetes issues.

When the respondents were fathers, a similar pattern

was found. Significant correlations between scores on the

Issues Checklist and the Diabetes Issues Checklist were

found for quantity (r=.67, p<.0001, n=41), intensity (r=.62,

p<.0001, n=35), and intensity by frequency (r=.70, p<.0001,

n=35) of issues. Fathers also reported significantly

greater quantity (mean difference=5.35, t=4.18, p<.0002),

intensity (mean difference=.41, t=4.21, p<.0002), and

intensity by frequency (mean difference=4.11, t=2.77,

p<.009) of general issues than diabetes issues.

When adolescents were the respondents, significant

correlations between scores on the Issues Checklist and the

Diabetes Issues Checklist were found for quantity (r=.60,

p<.0001, n=55), intensity (r=.83, p<.0001, n=53), and

intensity by frequency (r=.72, p<.0001, n=53) of issues with

mothers, and for quantity (r=.83, p<.0001, n=43), intensity

(r=.39, p<.04, n=29), and intensity by frequency (r=.59,










p<.005, n=29) of issues with fathers. Adolescents reported

significantly more general issues than diabetes issues both

with mothers (mean difference=3.96, t=3.37, p<.001) and with

fathers (mean difference=2.40, t=2.76, p<.009), but did not

report significant differences in intensity (mother: mean

difference=.09, t=1.33, p<.19; father: mean difference=.08,

t=.43, p<.67) or intensity by frequency (mother: mean

difference=1.62, t=1.46, p<.15; father: mean difference=.83,

t=.39, p<.70) of issues with either parent.

As with general issues, adolescents reported fewer

diabetes issues than did both their mothers and their

fathers. They reported more diabetes issues with their

mothers than with their fathers; this difference was also

reported by their parents. Adolescents and parents reported

similar intensity and frequency in their discussions of

diabetes issues.

Both adolescents and parents reported fewer diabetes

issues than general issues. Parents reported that

discussions about general issues happened more frequently

and were greater in intensity than diabetes issues, as well.

Behavioral Interaction Task

For this measure, the mean of four raters scores was

used to represent the family's communication. Interobserver

agreement was estimated using the Spearman-Brown correction

for multiple raters (Winer, 1971). Reliabilities on summary

scores for this sample ranged from .79 for Positive

Adolescent Behavior, Degree of Resolution, and Friendliness,









Table 5
Means, Standard Deviations, and Interrater Reliabilites for
Behavioral Interaction Task Summary Scores

Summary Score General Issues Diabetes Issues
(n=46) (n=46)

M (SD) M (SD)
[Spearman-Brown] [Spearman-Brown]
Positive Adolescent .240 (.134) .234 (.144)
Behavior [.79] [.81]

Negative Adolescent .119 (.083) .088 (.059)
Behavior [.81] [.83]

Positive Parental .413 (.212) .431 (.196)
Behavior [.84] [.76]

Negative Parental .117 (.100) .071 (.046)
Behavior [.84] [.63]

Degree of Resolution 3.22 (1.13) 2.80 (1.21)
[.79] [.87]
Insult 1.55 (0.69) 1.23 (0.35)
[.83] [.35]
Friendliness 2.57 (0.78) 2.81 (0.77)
[.79] [.74]
Problem-Solving 3.22 (1.05) 2.95 (1.15)
Effectiveness [.81] [.83]










to .84 for Positive Parental Behavior and Negative Parental

Behavior, for general issues, and from .35 for Insult to .87

for Degree of Resolution, for diabetes issues (see Table 5).

Summary scores for behavioral interactions about

general issues for this sample fell closer to the mean

scores for a clinic-referred sample than to those for a

nonclinic sample for all scores except Friendliness, Degree

of Resolution, and Problem-Solving Effectiveness, which fell

within the normative range (Prinz et al., 1979; Prinz &

Kent, 1978). Summary scores for behavioral interactions

about diabetes issues for this sample fell within the

normative range, with the exception of Positive Adolescent

Behavior and Positive Maternal Behavior, which were closer

to the mean for a clinic-referred sample. Table 5 presents

means and standard deviations of summary scores for both

general and diabetes-related behavioral interactions.

Higher summary scores indicate a higher rate of the

identified behavior, with the exception of higher scores

indicating poorer Degree of Resolution and poorer Problem-

Solving Effectiveness.

In comparing scores from the behavioral interactions

about general issues to those about diabetes issues,

significant correlations were found for all scores, and

ranged from r=.48, p<.0008, n=45, for Negative Parental

Behavior to r=.83, p<.0001, n=44, for Positive Parental

Behavior. There were significant differences between

behavioral interactions about general issues and those about










diabetes issues for five of the eight summary scores.

Discussions about general issues, as compared to diabetes

issues, had higher Negative Parental Behavior (mean

difference=.04, t=3.32, p<.002), higher Negative Adolescent

Behavior (mean difference=.02, t=2.63, p<.01), higher Insult

(mean difference=.27, t=3.32, p<.002), less Friendliness

(mean difference=-.20, t=-2.66, p<.01) and poorer Degree of

Resolution (mean difference=.33, t=2.17, p<.04). There was

a trend for discussions about general issues to show poorer

Problem-solving Effectiveness than those about diabetes

issues, as well (mean difference=.18, t=1.73, p<.09).

Behavioral interactions were consistent with adolescent

and parental reports about general and diabetes-related

issues. Behavioral interactions about general issues, in

contrast to those about diabetes issues, had higher negative

behavior from both adolescent and parent, more insults, and

poorer resolution and effectiveness of problem-solving.

Dyadic Adjustment Scale

Marital satisfaction and adjustment was assessed with

the Dyadic Adjustment Scale. Internal consistencies of this

measure with this sample were excellent (Coefficient alphas

for mothers = .95, for fathers = .93, for parents combined =

.94). The mean total scores obtained by this sample were

111.8 (SD=16.8) for mothers and 115.3 (SD=14.1) for fathers,

which, relative to the normative sample, fell within the

normal range (Spanier, 1976).










Agreement between respondents was assessed by using

Pearson product moment correlations and paired comparison T-

tests. A significant correlation was found between mothers'

and fathers' scores of dyadic adjustment (r=.48, p<.003,

n=37). There was not a significant difference between

mothers' and fathers' scores (mean difference=1.14, t=.46,

p<.65).

Adherence Measures

Adherence with the daily diabetes regimen was assessed

with 24-hour recall interviews. Relative to the normative

sample, adherence measures in this sample fell within one

standard deviation of the norm, with the exception of the

measures of Carbohydrate Consumption and Fat Consumption.

Scores on these two measures were significantly below

average, indicating better adherence in those areas. Table

6 presents means and standard deviations of adherence scores

as obtained from parents, adolescents, and the combined

sample. Higher scores on these measures were indicative of

poorer adherence.

Agreement between respondents was assessed by using

Pearson product moment correlations and paired comparison T-

tests. Significant correlations were found between parents'

and adolescents' scores for Exercise Frequency, Exercise

Type, Injection Regularity, Injection Interval, Injection

Meal Timing, Eating Frequency, Carbohydrate Consumption in

Relation to Ideal, Fat Consumption in Relation to Ideal,









Table 6
Means and Standard Deviations


of Adherence Scores


Parents
(n=50,54)


Adolescents
(n=55)


Combined
(n=54,55)


Adherence Measure


Exercise Frequency 29.7 (20.4)
[per day] [4.2]

Exercise Duration .12 (.22)
[minutes] [22.53]

Exercise Type .972 (.017)
[kilocalories/minute] [.029]

Injection Regularity .59 (.53)
[minutes] [35.4]

Injection Interval 1.36 (1.13)
[minutes] [81.6]

Injection-Meal .82 (.54)
Timing [minutes] [10.8]

Regularity of
Injection-Meal 24.3 (46.5)
Timing [minutes]

Eating Frequency 20.0 (12.9)
[per day] [4.8]

Glucose Testing 65.5 (25.8)
Frequency [per day] [1.4]

% Calories: 14.9 (6.5)
Carbohydrate [45.1]

% Calories: 12.9 (5.9)
Fat [37.9]

Calories Consumed -1187 (676)
(compared to ideal)

Concentrated Sweets 1.1 (1.5)
(per day)


33.9 (20.1)
[4.0]

.10 (.19)
[24.32]

.969 (.017)
[.033]

.71 (.56)
[42.6]

1.34 (1.06)
[80.4]

.80 (.46)
[12.0]


20.9 (20.7)


26.6 (12.5)
[4.4]

62.6 (25.3)
[1.5]

13.0 (6.9)
[47.1 ]

13.2 (7.3)
[38.2]

-625 (812)


1.3 (1.9)


61.0 (21.7)
[2.3]

.07 (.07)
[23.83]

.970 (.017)
[.032]

.67 (.51)
[40.2]

1.26 (1.04)
[75.6]

.79 (.44)
[12.6]


20.1 (18.2)


22.9 (11.5)
[4.6]

58.3 (25.9)
[1.7]

12.9 (6.8)
[47.1]

14.0 (7.2)
[39.0]

-437 (751)


1.7 (2.2)


Note.Values in brackets provide the reader with an interpretation
of each adherence measure using a familiar measurement
scale.


M (SD)


M (SD)


M (SD)









Table 7
Parent-Adolescent Correlations for Adherence Scores

Adherence Score r (n) Adherence Score r (n)
P < p <
Exercise Frequency .73 (54) Eating Frequency .63 (54)
(.0001) (.0001)
Exercise Duration .12 (54) Concentrated Sweets .64 (54)
(.38) (.0001)
Exercise Type .64 (54) *% Calories: Carbs .48 (54)
(.0001) (.0002)
Injection Regularity .71 (51) % Calories: Fat .56 (54)
(.0001) (.0001)

Injection Interval .73 (51) Calories Consumed .28 (54)
(.0001) (.04)
Injection-Meal .73 (53) Testing Frequency .81 (54)
Timing (.0001) (.0001)

Regularity of
Injection-Meal .25 (51) *% Calories: Carbohydrate
Timing (.08)










Calorie Consumption in Relation to Ideal, and Testing

Frequency (see Table 7).

There were significant differences between parents' and

adolescents' scores on four of the thirteen measures.

Parents reported greater Exercise Frequency (mean

difference=5.11, t=2.60, p<.01), better Injection Regularity

(mean difference=.12, t=2.00, p<.05), greater Eating

Frequency (mean difference=6.83, t=4.58, p<.0001), and fewer

fewer Calories Consumed (mean difference=577.32, t=4.63,

p<.0001). There were trends for differences on two of the

measures: Parents reported less strenuous Exercise Type

(mean difference= -.0004, t=-1.94, p<.06) and greater

deviation of Carbohydrate Consumption in Relation to Ideal

(mean difference=-1.69, t=-1.81, p<.08). With the exception

of the difference in Calorie Consumption, these differences

are of questionable clinical significance.

Metabolic Control

HbAlc averaged 9.5% (SD=1.9%), triglyceride levels

averaged 71 mg/dl (SD=32 mg/dl), and cholesterol levels

averaged 173 mg/dl (SD=31 mg/dl) for the adolescents in this

sample. HbAlc was significantly correlated with both

triglycerides (r=.30, p<.04, n=47) and cholesterol (r=.34,

p<.02, n=44). Triglycerides and cholesterol were

significantly correlated (r=.50, p<.0004, n=45).

Multiple Regression Analyses

Three conceptual domains (i.e., psychological

adjustment, stress, and family relations) were hypothesized










to be important in the proposed model of the associations

among psychosocial variables, adherence, and diabetes

control. Pubertal development was an additional factor

considered in the model. Due to the relatively small sample

size, and the large number of potential predictor variables,

psychosocial variables within each conceptual domain were

combined to provide global measures of important constructs.

Prior to combining variables, variables with poor

reliability and those with significant correlations to other

subscale scores from the same measure were dropped. These

included Family Environment Scale subscale scores because of

their extremely poor internal consistencies; Negative

Parental Behavior and Insult from the diabetes-related

Behavioral Interaction Task due to poor interrater

reliability; and the Intensity by Frequency scores from the

Issues Checklist (IC) and the Diabetes Issues Checklist

(DIC), and the Appraisal of Other score from the Conflict

Behavior Questionnaire (CBQ) because of their significant

correlations with other subscales scores from those

measures. Father scores were not included based on results

from correlation analyses and repeated measure ANOVAs;

mothers and adolescents were correlated on most measures

while fathers and adolescents were not. Also, there were

fewer actual differences between mother and adolescent

scores as compared with father and adolescent scores.

Sample size considerations also precluded the use of father










scores, as there were fewer fathers than mothers and

adolescents in the study.

Subscale scores from each measure were standardized and

combined, prior to combining across measures, thus ensuring

equal weighting for each measure. For adjustment and family

relations measures, mother and adolescent scores were

standardized and combined to yield aggregate scores. Mother

and adolescent scores were not combined for the stress

measures, as each score represented the respondent's own

life stress.

The conceptual domain of adjustment consisted of two

scores: social competence, from the Child Behavior

Checklist, and behavior problems, from the same measure.

Two scores were also produced for stress: adolescent

stress and maternal stress. These scores were the

combination of the Quantity score from the Hassles Scale and

the Total Score from the Life Events Checklist, for

adolescents, or the Social Readjustment Rating Scale, for

mothers.

Family relations consisted of two scores: positive

family relations and negative family relations. The

positive family relations score included these measures:

positive subscale score from the Diabetes Family Behavior

Checklist (DFBC), Positive Adolescent Behavior, Positive

Parental Behavior, and Friendliness (from both general and

diabetes-related discussions) from the Behavioral

Interaction Task, and the Dyadic Adjustment Scale (DAS)










score. The negative family relations score included these

measures: negative subscale score from the DFBC, Negative

Adolescent Behavior, Negative Parental Behavior, Degree of

Resolution, Insult, and Problem-Solving Effectiveness, from

the general discussions, and Negative Adolescent Behavior,

Degree of Resolution, and Problem-Solving Effectiveness,

from the diabetes-related discussions, Quantity and

Intensity scores from the IC and from the DIC, and Appraisal

of Dyad score from the CBQ.

Hierarchical multiple-regression techniques (Cohen &

Cohen, 1983) were used to examine relationships among

measures of adjustment, stress, family, adherence, and

health outcome. The first step of this approach is to test

an initial set of predictor variables, such as adolescent's

gender, age, disease duration, Tanner Stage, and

socioeconomic status (SES). Next, the psychosocial

variables are added to the initial model, and the resulting

R2 is statistically compared to the initial model's R2 to

determine whether there has been a significant increase in

variance accounted for. The hierarchical progression

continues from tests of simple main effects to tests of more

complex models with interaction terms. Only when a more

complex model accounts for significant variance beyond that

provided by a preceding, simpler model is it retained. In

order to ensure a fair comparison between hierarchical

models, each hierarchical progression is based on the same

data set. Due to sample size constraints, three sections of










the proposed model were separately tested: adjustment

relationships (see Figure 2), adherence relationships (see

Figure 3), and diabetes control relationships (see Figure

4). Only adjusted R2 values are reported as they correct

for the small sample size.

Adjustment

Due to sample size constraints, separate series of

hierarchical regressions were used to examine predictors of

social competence and of behavior problems (see Figure 2).

Gender, age, disease duration, Tanner stage, and SES were

first entered into the analyses. None were significantly

predictive for either adjustment score, so all were removed.

Next, family scores were entered together, followed by the

addition of stress scores, both separately from family

scores, and then simultaneously with them. For social

competence, positive family behavior was a significant

predictor (R2 =.03), F(1,52) = 2.80, p < .10; higher

positive family behavior was associated with better social

competence. The mother stress measure contributed

additional significant predictive power as a main effect,

t(51) = 3.72, p < .0005, increasing the model's R2 from .03

to .22, F(2,51) = 8.68, p < .0006; unexpectedly, higher

mother stress was associated with better social competence

(See Table 8). For behavior problems, a significant main

effect was found for negative family behavior (R2 = .21),

F(1,54) = 15.32, p < .0003; as expected, higher negative













Adolescent
Characteristics
Age
Gender
Tanner Stage
Disease Duration
SES


Stress
Maternal Stress
Adolescent Stress


_Social
Competence


Family Factors
Positive Family Behavior
Negative Family Behavior


Expected Relationship
Tested (but not necessarily
expected) Relationship

Figure 2
Model of Adjustment Relationships












Stress
Maternal Stress
Adolescent Stress -


Psychological Adjustment


Social Competence
Total Behavior Problems ~ ~

Injection
Family Factors Adherence
Positive Family Behavior
Negative Family Behavior /


Adolescent
Characteristics
Age
Gender
Tanner Stage
Disease Duration
SES


Expected Relationship
-- Tested (but not necessarily
expected) Relationship

Figure 3
Model of Adherence Relationships


I


\
s








Stress


Psychological Adjustment
Social Competence
Total Behavior Problems -------


Family Factors
Positive Family Behavior -----
Negative Family Behavior

Puberty


-- - -----~p


Tanner Stage
(Pubertal or Not Pubertal)


Adherence
Injection
Exercise
Diet Type
Testing/Eating Frequency
Calories Consumed
Concentrated Sweets
(each entered separately)


--- Expected Relationship
Tested (but not necessarily
expected) Relationship

Figure 4
Model of Metabolic Control Relationships


^


D









Table 8
Best Prediction Models for Adjustment:
Social Competence and Behavior Problems


Adjusted
p< R-


Predictor


F(df)


Social Competence

Intercept
Positive Family
Behavior
Maternal Stress

Behavior Problems

Intercept
Negative Family
Behavior


49.27

5.42
6.85


30.60

2.04
3.72


51.43 52.19


.0001 .22 8.68 (2,51)

.05
.0005


.0001 .21 15.32 (1,54)


7.60 3.91 .0003


.0006


.0003










family behavior was associated with more behavior problems

(See Table 8). No other predictors emerged.

Adherence

Each adherence measure was examined in a separate

series of hierarchical regressions. Gender, age, disease

duration, Tanner Stage, and SES, were entered initially.

Family relation scores were entered next, followed by stress

scores, and finally, adjustment scores (see Figure 3).

Again, these scores were entered separately (i.e., from

scores from other domains) and then simultaneously. Because

it seemed likely that diabetes-specific family measures

would be stronger predictors to adherence behaviors than

general and diabetes-specific family measures combined, all

series of regressions were also run with family relation

scores computed with only diabetes-specific family measures

(i.e., Diabetes Issues Checklist, Diabetes Family Behavior

Checklist, Behavior Interaction Task for diabetes issues).

However, as no other significant associations with these

measures emerged, only results from regressions using the

combined family relation scores are reported.

Infection. Both adjustment scores emerged as

significant predictors of the Injection factor, (social

competence, t(51) = 1.82, p < .07; behavior problems, t(51)

= 1.81, p < .08); however, the variance accounted for was

small (R2 = .08), F(2,51) = 3.21, p < .05. Better social

competence and more behavior problems seemed to be

associated with poorer injection adherence.









Table 9
Best Prediction Models for Adherence Measures


Adjusted
p< RR


Predictor

Injection


F(df)


Intercept
Behavior Problems
Social Competence

Calories Consumed

Intercept
Negative Family
Behavior


-1 .53
.02
.01


-2.37
1 .82
1.81


-.72 -5.00

-.55 11.95


.02
.07
.08


.0001


.08 3.21 (2,51)


.05 3.80 (1,53)


Exercise


Intercept
Age
Externalizing
Behavior Problems
Positive Family
Behavior


Diet Type

Intercept
Age


1 .59
-.27

.05

.47


1 .35
-.20


.99
-2.99

2.72

1.95


.98
-2.08


.33
.004

.009

.06


.33
.04


.21 5.93 (3,51) .002


.06 4.33 (1,53)


Testing/Eating
Frequency


Intercept
Age


-1 .39
.13


-1 .49
2.03


.14
.05


.06 4.12 (1,49)


Concentrated Sweets


Intercept
Gender
Behavior Problems
Social Competence


-2.20 -2.37
-.56 -2.26
.03 2.05
.03 2.9


Note.In these regressions, boys = 1, girls = 2; for Tanner
Stage, not pubertal = 1, pubertal = 2.
For all adherence measures except Exercise, lower scores
indicate greater adherence. For Exercise, higher scores
indicate greater adherence.


.05


.06


.05


.02
.03
.05
.005










Negative family behavior also emerged, separately, as a

significant predictor (R2 = .05), F (1,52) = 3.93, p < .05;

more negative family behavior was associated with poorer

injection adherence. Social competence contributed

additional significant predictive power to the model, t(51)

= 1.77, p < .08, increasing the model's R2 to .09, F (2,51)

= 3.6, p < .03; as before, better social competence was

associated with poorer adherence.

When behavior problems and negative family behavior

were both entered into the model, neither emerged as a

significant predictor; this effect was due to the strong

correlation between the two measures. Based on the previous

finding of negative family behavior as a predictor of

behavior problems, the model including behavior problems and

social competence was retained as the best model (see Table

9).

Calories Consumed. Negative family behavior emerged as

a significant predictor of Calories Consumed (R2 = .05),

F(1,53) = 3.80, p < .06; more negative family behavior was

associated with undereating (see Table 9). No other

predictors emerged.

Exercise. Age emerged as a significant predictor (R2 =

.11), F(1,53) = 7.35, p < .009; older adolescents were less

compliant with exercise than younger adolescents (see Table

10). Behavior problems contributed additional significant

predictive power as a main effect, t(51) = 1.96, p < .06,

increasing the model's R2 from .11 to .15, F(2,52) = 5.80, p










< .005; more behavior problems were associated with more

exercise. Positive family relations also contributed

significant predictive power, t(51) = 1.86, p < .07, further

increasing the model's R2 to .19, F(3,51) = 5.2, p < .003;

greater positive family relations were associated with

better exercise adherence. Because the behavior problems

score consists of measures of internalizing behaviors and

externalizing behaviors, it was unclear which was

responsible for the significant effects. To investigate

this issue, hierarchical analyses were rerun with

internalizing behaviors and externalizing behaviors

considered in separate analyses. Only externalizing

behaviors offered significant predictive power, t(52) = 2.3,

p < .03, increasing the model's R2 to .17, F(2,52) = 6.63, p

< .003; more externalizing behaviors were associated with

more exercise. Positive family relations again contributed

significant predictive power, t(51) = 1.95, p < .06, further

increasing the model's R2 to .21, F(3,51) = 5.9, p < .002

(see Table 9).

Diet Type. Only age emerged as a significant predictor

of Diet Type (R2 = .06), F(1,53) = 4.33, p < .04;

unexpectedly, older adolescents were more compliant than

younger adolescents (see Table 9). In an attempt to

understand this finding, adolescents in the sample were

divided into three groups by ages (i.e., 12-13, 14-15, and

16-17 years). The means of the two adherence measures which









Table 10
Means and Standard Deviation of Exercise, Diet Type, and
Testing/Eating Frequency Adherence Scores by Age Group


12-13 years
(n=28)


14-15 years
(n=15)


16-17 years
(n=13)


Adherence Measure

Exercise

Exercise Frequency
[per day]

Exercise Duration
[minutes]


M (SD)


52.6 (22.7)
[2.8]

.05 (.03)
[27.72]


Exercise Type .966 (.018)
[kilocalories/minute] [.036]


M (SD)


64.8 (17.5)
[2.1 ]

.07 (.04)
[20.23]

.968 (.012)
[.033]


M (SD)


75.9 (14.7)
[1.4]

.11 (.12)
[19.28]

.980 (.015)
[.020]


Diet Type


% Calories:
Carbohydrate

% Calories:
Fat


13.5 (5.6)
[46.5]

14.4 (5.2)
[39.4]


16.5 (6.3)
[43.5]

7.5 (8.7)
[42.5]


7.2 (6.7)
[52.8]

8.8 (6.6)
[33.8]


Testing/Eating Frequency


Eating Frequency
[per day]

Glucose Testing
Frequency [per day]


22.0 (11.3)
[4.7]

52.4 (22.7)
[1.91


22.6 (11.4)
[4.6]

58.9 (25.5)
[1.6]


25.5 (12.4)
[4.5]

71.5 (30.3)
[1.1]


Note.Values in brackets provide the reader with an interpretation
of each adherence measure using a familiar measurement
scale.










make up Diet Type (i.e., percentage of calories from

carbohydrates, percentage of calories from fat) were

examined for each age group (see Table 10). Middle

adolescents were the least compliant to low-fat, high-

carbohydrate diets, while the oldest adolescents were the

most compliant.

Testing/Eating Frequency. Age emerged as a significant

predictor for Testing/Eating frequency (R2 = .06), F(1,49) =

4.12, p < .05; older adolescents were less adherent than

their younger counterparts. Tanner Stage contributed

additional significant predictive power as a main effect,

t(48) = -.52, p < .05, increasing the model's R2 to .11,

F(2,48) = 4.17, p < .02. Tanner Stage alone explained as

much variance as age and Tanner Stage together (R2 = .12),

F(1,49) = .007. Tanner Stage was dichotomized as pubertall"

(i.e., Tanner Stages II, III, and IV) and "not pubertal"

(i.e., Tanner Stages I and V). It appears that pubertal

adolescents were more adherent than adolescents who were not

pubertal. Further investigation reveals this inconsistent

result to be due to the significant number of Tanner Stage V

adolescents (i.e., 17 of 22) in the "not pubertal" group.

These adolescents tended to be older (correlation of age and

Tanner Stage, r=.71, p<.0001, n=52), and, as previously

stated, older adolescents were less adherent than younger

adolescents (see Table 10). Because the effect of Tanner

Stage was apparently actually due to age, the model in which