|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
FAMILY FACTORS, ADHERENCE, AND METABOLIC CONTROL INT YOUTH WITH
TYPE 1 DIABETES.
DANNY C. DUKE
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
Danny C. Duke
To my family, whose sacrifices made my education possible.
I would like to express my gratitude to my mentor and chair, Dr. Gary R. Geffken. Also,
my appreciation goes to Adam Lewin and Laura Bimbo, for their assistance with data collection;
and to Dr. Janet Silverstein, Division Chief of Pediatric Endocrinology, for allowing us to recruit
families from the pediatric diabetes clinic. Finally, thanks are due to my committee members, Dr.
Sheila Eyberg, Dr. Michael Perri, and Dr. William Perlstein.
TABLE OF CONTENTS
ACKNOWLEDGMENT S .............. .................... iv
LI ST OF T ABLE S ............_ ..... ..__ .............. vii..
LIST OF FIGURES ............_...... .__ ..............viii...
AB STRAC T ................ .............. ix
1 INTRODUCTION ................. ...............1.......... ......
Overview of Diabetes .............. ..... .... ........ .. ........ ... ...................
Treatment and the Diabetes Control and Complications Trial (DCCT) ................... ....2
Measures of Adherence ................ .............. .................. ...............4
Measures of Metabolic Control ................... ................... .. ........ ....... ........7
Relationships between Family Factors, Adherence, and Metabolic Control ................7
Diabetes Specific Family Measures............... ..........................1
Externalizing, Diabetes Specific Family Functioning, Adherence and Metabolic
Control .............. ...............14....
Sum m ary ................. ...............14.......... ......
2 RE SEARCH QUE ST IONS .............. ............... 17....
Question 1: Family Factors Predicting Metabolic Control ................... .. ................ ..17
Question 2: Adherence Mediating between Diabetes-Specific Family Factors and
M etabolic Control. ............. ....... .......... .. .. ... .. ........1
Question 3: Family factors mediating between child externalizing and metabolic
control. ............. ............. .... .. .. .... .........1
Question 4: Adherence mediating child externalizing behaviors and metabolic
control? ................ ...............18.................
3 M ETHODS .............. ...............22....
Participants and Settings............... ...............22
Procedure .............. ...............22....
M measures ................... ........... ...............23.......
Demographic Questionnaire ................. ...............23.................
Diabetes-Specific Family Measures ................. ...............23................
Diabetes Family Behavior Scale (DFB S) ................. ................ ....._.23
Diabetes Family Behavior Checklist (DFBC). ................. ............. .......24
Diabetes Family Responsibility Questionnaire (DFRQ) ................... ...........24
Child Behavior............... .... .. .............2
Child Behavior Checklist .............. ...............25....
Measurement of Adherence .....__ ................. ........._._ ......... 2
Measure of Metabolic Control .....__ ................. .........__ ........ 2
4 RE SULT S .............. ...............27....
5 DI SCUS SSION ................. ...............44...............
Interpretation of Results ................. ..... ... .............4
Limitations, Implications, and Future Directions .....__ ................ ............._.._47
REFERENCES .............. ...............50....
BIOGRAPHICAL SKETCH ............. .............56......
LIST OF TABLES
4-1 Demographic Characteristics. ............. ...............36.....
4-2 Means and Standard Deviations for Age, HbAlc, Duration of Diabetes, and Income........37
4-3 Intercorrelations between HbAlc, Diabetes Family Measures, Adherence, and
Externalizing. ............. ...............38.....
4-4 Hierarchical Regression Analysis Predicting HbAlc from Diabetes-Related Family
Factors and Adherence. .............. ...............39....
4-5 Hierarchical Regression Analysis Predicting HbAlc from Parent and Child Report of
Adherence ........... ..... .._ ...............40...
4-6 Mediation Regression Analysis Predicting HbAlc: Adherence Mediating the
Relationship between Family Factors (DFBC) and Metabolic Control (HbAlc): Final
Block of the Regression. ............. ...............41.....
4-7 Mediation Regression Analysis Predicting HbAlc: Family Factors (DFBC) Mediating
the Relationship between Child Extemnalizing and Metabolic Control (HbAlc): Final
Block of the Regression. ............. ...............42.....
4-8 Mediation Regression Analysis Predicting HbAlc : Adherence Mediating the
Relationship between Child Extemnalizing and Metabolic Control (HbAlc): Final Block
of the Regression. ................. ...............43.............
LIST OF FIGURES
2-1 Theoretical model of adherence mediating family factors and metabolic control. ..............19
2-2 Theoretical model of family factors mediating externalizing and metabolic control. .........20
2-3 Theoretical model of adherence mediating externalizing and metabolic control. ..............21
4-1 The relationship between family factors (DFBC) and metabolic control (HbAlc)
partially mediated by adherence (Parent DSMP). .............. ...............33....
4-2 The relationship between externalizing and metabolic control (HbAlc) fully mediated
by family factors (DFBC). ............. ...............34.....
4-3 The relationship between externalizing and metabolic control (HbAlc) fully mediated
by adherence (Parent DSMP). .............. ...............35....
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
FAMILY FACTORS, ADHERENCE, AND METABOLIC CONTROL INT YOUTH
WITH TYPE 1 DIABETES
Danny C. Duke
Chair: Gary Geffken
Considerable research has examined the relationship of family variables and
adherence to metabolic control in pediatric populations with type 1 diabetes, with most
studies finding that these constructs were only weakly related to metabolic control.
Children and adolescents with type 1 diabetes are often noncompliant with their
prescribed treatment regimens. Clinical evidence suggests that poor adherence leads to
poor metabolic control and conversely that compliance with a diabetes treatment regimen
predicts good metabolic control. Inadequate metabolic control leads to physiological
complications (e.g., retinopathy, neuropathy, nephropathy) at an earlier age. Our study
examined the ability of a combined battery of diabetes-specific family functioning
measures (e.g., family responsibility, parental support, guidance, warmth, and criticism
specific to diabetes care) to predict metabolic control (health status). One hundred twenty
children and parents, who presented in a pediatric diabetes clinic, completed four
measures of diabetes-specific family functioning, and separate adherence interviews.
Overall, combined with demographic variables, these measures predicted approximately
44% of the variance in metabolic control. Examination of parent and child reports of
adherence showed that, while controlling for demographic variables, parent report
accounted for 27.4% of the variance in metabolic control, while child report did not add
significantly to the variance accounted for in metabolic control beyond parent report.
Adherence was found to partially mediate the relationship between child metabolic
control and perceptions of parental nagging, criticizing, and arguing about diabetes
specific tasks. Family variables were found to partially mediate the relationship between
externalizing and metabolic control and adherence was found to fully mediate the
relationship between child externalizing and metabolic control. Our data suggest that
reports of adherence behaviors are strongly related to a child's health status independently
and through important pathways of family functioning. These data suggest that an
increased and comprehensive focus on assessing and monitoring family behaviors related
to adherence is critical for optimizing health-status outcomes.
Overview of Diabetes
The incidence of diabetes in children and adolescents is increasing worldwide
(Diabetes Control and Complications Trial Research Group, 1986; EURODIAB ACE
Study Group, 2000; Onkamo et al., 1999). This disease adds significantly to public
health care costs and profoundly affects individual quality of life, having treatment
demands that are often complex and unrelenting. Diabetes is one of the most common
chronic diseases in school-aged children, affecting about 151,000 young people in the
United States, or about 1 in every 400 to 500 youth under 20 years of age. With the peak
incidence of type 1 diabetes occurring at puberty (Feingold & Funk, 1997), the
prevalence of type 1 diabetes in the U. S. is increasing with approximately 30,000 new
cases of type 1 diabetes diagnosed annually (Engelgau & Geiss, 2000).
Pancreatic beta cells are insulin-producing bodies within the pancreas that are
responsible for producing the insulin necessary for the body's metabolism of glucose.
Type 1 diabetes is thought to result from an autoimmune process that leads to the
destruction of the insulin producing beta cells. The typically acute onset develops over a
period of a few days to weeks. The percentage of beta cell destruction ranges from 60 to
80% at the time of clinical symptom presentation (National Diabetes Data Group, 1995;
Notkins & Lernmark, 2001). Without insulin, cells are unable to metabolize sugars
present in the bloodstream. As a means of compensating for the body's lack of insulin
production, individuals with diabetes require insulin via subcutaneous inj sections.
Additionally, maintaining near normal blood-glucose levels requires intensive and often-
complex management, including frequent blood-glucose monitoring and regular exercise,
and dietary and insulin adjustments based on blood-glucose levels (American Diabetes
Treatment and the Diabetes Control and Complications Trial (DCCT)
To avoid periods of elevated blood sugar (hyperglycemia), the American Diabetes
Association (ADA) guidelines recommend that individuals with type 1 diabetes maintain
blood glucose levels of 80 tol20 mg/dL. Hyperglycemia can lead to long-term
complications, including retinopathy, hypertension, neuropathy, nephropathy, and
cardiovascular disease (ADA, 2003). Profound hyperglycemia associated with
insufficient insulin can result in diabetic ketoacidosis (DKA), a lethal condition if not
promptly treated with insulin and fluids (Feingold & Funk, 1997). The Diabetes Control
and Complications Trial (DCCT) identified an empirical relationship between intensive
insulin therapy and reduced complications (Diabetes Control and Complications Trial
Research Group, 1987; 1993; 1994). This randomized, controlled clinical trial showed
that for patients with type 1 diabetes, the risk of the development and progression of
retinopathy, nephropathy, and neuropathy is reduced 50 to 75% by intensive treatment
regimens when compared to conventional treatment regimens. For example, Zhang,
Krzentowski, Albert, & Lefebvre (2001) found that, among patients in the DCCT (ages
13 to 39) with mean HbAlc < 6.87% (good control), 90% remained free of retinopathy.
However, in patients with a mean HbAlc > 9.49% (poor control), complications
developed in 57% of the patients. In a study of 195 adolescents, participants who were
assigned an intensive therapy regimen (administration of three or more inj sections per day
or use of an insulin pump, adjustment of insulin based on blood-glucose levels, at least
four blood-glucose tests per day, diet, and exercise) showed delayed onset and slower
progression of retinopathy compared to individuals in a conventional therapy (one or two
insulin inj sections and self monitoring of blood glucose at least twice per day).
Additionally, intensive therapy also reduced the risk of the development of
microalbuminuria, which is a precursor to the development of diabetes-related
nephropathy (Diabetes Control and Complications Research Group, 1994).
To prevent long-term complications, the DCCT research group recommended
intensive insulin and monitoring therapy for adults and adolescents as young as 13 years
old (Diabetes Control and Research Group, 1993, 1994). While intensive therapy reduces
the risk of diabetes-related complications and high HbAlc levels, it is a complex and
demanding regimen for children and families. Physicians typically prescribe complex
daily regimens that include insulin inj sections, blood-glucose monitoring, diet and
exercise. It is expected that these behavioral adjustments be integrated into the patient' s
daily routine. Data indicate that children and adolescents are often noncompliant with
many of these recommendations. A nine-year follow-up study by Kovacs and colleagues
(1992) established that 45% of adolescents engage in some significant form of
noncompliance during adolescence. These findings were before the publication of DCCT
research, before intensive insulin therapy was routinely prescribed. As children and
adolescent medical regimens become more intense and complex, understanding
adherence becomes an increasingly important facet of treatment.
Past research has often failed to account for the behavioral contributions of the
child to reciprocal parent-child interactions necessary to administer such complex
treatment regimens. There are many problematic child behaviors that may contribute to
poor regimen adherence. Achenbach (1991) categorized childhood problems into the two
broad dimensions of internalizing and externalizing. Internalizing encompasses behavior
problems characterized by anxiety, depression, and social withdrawal, while externalizing
behavior problems are characterized by aggression and antisocial behavior. Limited
research has examined relationships between child psychopathology, family functioning
and metabolic control. Work by Northam et al. (2005) found that in a sample of
adolescents with diabetes, those with high blood sugars had higher levels of externalizing
behaviors than those with good and low blood sugars. Conversely, Cohen et al. (2003)
found that lack of internalizing behaviors significantly predicted adherence in an
economically disadvantaged population.
Measures of Adherence
Haynes (1979) defined compliance as the extent to which a person' s behavior
coincides with medical or health advice. Johnson (1993) maintains that defining
adherence (or compliance) with the childhood diabetes regimen is difficult due to the
multitude and complexity of the regimen behaviors. This has hindered the development
of an accurate and reliable measure of adherence for youth with diabetes.
Adherence to diabetes treatment regimens is multidimensional, with research
demonstrating that a patient' s behavior with regard to one aspect of the diabetes regimen
is not predictive of the same patient' s behavior in relation to other components of the
regimen (Glasgow, McCaul, & Schafer, 1987; Johnson et al 1986; Kavanagh, Gooley, &
Wilson, 1993). Adding to the obfuscation, there is not universal agreement on explicit
standards for measuring adherence. Methods range on a continuum from direct to indirect
(La Greca, 1990). Intuitively, the most accurate method of assessment would be
observation of diabetes care behavior. However, the behavioral observational approach is
impractical for the clinical scientist practitioner since the complexity of the diabetes
treatment regimen would require intense scrutiny of numerous behaviors over extended
periods of observation. Indirect measures of adherence behavior offer an alternative
means of assessment. One approach involves obtaining patient self-reports or parental
reports of adherence (Anderson, Auslander, Jung, Miller, & Santiago, 1990). La Greca
(1990) cautions about generalizations from self-reports due to extensive variability in
their detail and comprehensiveness. Johnson, Silverstein, Rosenbloom, Carter, and
Cunningham (1986) developed a 24-hour recall interview to quantify adherence
behaviors. Interviewers ask patients and parents, over a series of three 20-minute
structured phone interviews, to describe diabetes-management behaviors during the
previous 24-hour period. Statistical analyses resulted in indices of adherence. Broadly
speaking these indices covered different aspects of glucose testing, insulin administration,
diet, and exercise constructs. Significant correlations between parent and child report was
documented in separate analyses (Freund, Johnson, Silverstein, & Thomas, 1991;
Johnson et al., 1986; Spevack, Johnson, & Riley, 1991). Advantages of the interview
include a minimization of recall errors, improved reliability, and the ability to capture a
non-reactive example of the child' s adherence. However, the 24-hour recall interview is
difficult to implement in a clinical setting, due to its reliance on multiple interviews (3
child, 3 parent) and complex scoring system (McNabb, 1997). The resources required to
conduct interviews and analyze data (e.g., time and manpower) are not typically available
to healthcare providers. Additionally, despite the comprehensiveness of the 24-hour recall
interview, strong associations between adherence and HbAlc were not supported in a
longitudinal analysis of approximately 200 children using the measure (Johnson, Kelly,
Henretta, Cunningham, Tomer, & Silverstein, 1992).
The Self-Care Adherence Inventory (SCAI) (Hanson, Henggeler, & Burghen
1987), a semi-structured interview that assesses adherence to type 1 diabetes treatment
regimens, provides an alternative to the 24-hour recall interview (Hanson et al., 1987;
Hanson, DeGuire, Schinkel, Kolterman, Goodman, & Buckingham, 1996). The interview
is administered to the patient by an individual familiar with the requirements of the
diabetes regimen. The SCAI content areas include glucose testing, dietary behaviors,
insulin adjustment, and hypoglycemia preparedness. The authors found that the SCAI
related to HbAlc on three separate administrations
(partial r = -0.28, -0.25, and -0.20, p < 0.001).
The SCAI was further refined and the Diabetes Self Management Profile (DSMP)
was developed by Harris and colleagues (2000). The DSMP consists of structuredd
interviews of children/adolescents and parents with diabetes, consisting of 23 questions,
that assess five areas of diabetes management, including: insulin administration/dose
adjustment, blood-glucose monitoring, exercise, diet, and management of hypoglycemia.
The DSMP is designed to access diabetes self-management over the preceding three
months. Investigators found good internal consistency and inter-observer agreement
(Harris et al., 2000). The predictive validity (r = -0.28, p < 0.01) for metabolic control
with the DSMP was consistent with data from the SCAI, however, the measure was
shown to only account for 7.8% of the variance in metabolic control (HbAlc) (Harris et
Measures of Metabolic Control
While adherence is a behavioral construct, metabolic control is a biological assay
of health status. The widely accepted standard of measurement for metabolic control is
the glycated hemoglobin Alc test (GHB/HbAlc). HbAlc is a measure of glucose bound
to specific hemoglobin molecules within red blood cells. The amount of glucose bound to
such hemoglobin molecules is directly proportional to the concentration of glucose in the
blood. Therefore, HbAlc provides an estimate of the average blood glucose
concentration. Since red blood cells live for approximately 90 days this provides an
estimate over the preceding 8 tol2 weeks. The DCCT observed an average HbAlc of
7.2% for individuals in the intensive treatment group compared to 9.0% in the
conventional treatment group, indicating that individuals in the intensive insulin therapy
group were under better metabolic control.
Relationships between Family Factors, Adherence, and Metabolic Control
Despite the Eindings from the DCCT, most studies fail to aind high or even
moderate correlations between regimen adherence and diabetes metabolic control
(Glascow et al., 1987; Hanson et al., 1987; Hanson et al., 1996; Harris et al., 2000;
Johnson et al, 1986, 1992, 1993, 1994; Kavanagh et al., 1993; Schafer, McCaul, &
Glasgow, 1986; Spevack et al., 1991). An explanation for this discrepancy might result
from examination of the instrument used to measure adherence. Most studies of
adherence have measured an individual's behavior in relation to an ideal treatment
regimen, and have not examined variations from individual prescriptions (La Greca,
1990). Additionally, if a patient' s prescribed treatment regimen was ineffective, perfect
adherence would not be expected to relate to good metabolic control (Johnson, 1992).
Currently, there is no specific, standardized set of self-care behaviors applicable to all
children with diabetes. It is difficult to measure congruity of behaviors that are not
uniformly defined in operational terms (McNabb, 1997). Johnson (1993) maintains that
only weak relationships exist between metabolic control and behavior. Nevertheless,
clinical evidence indicates adherence is a significant predictor of metabolic control. For
example, data from a residential treatment program for youth in poor metabolic control
demonstrate a significant mean reduction in HbAlc when the treatment regimen is
delivered in a controlled environment (Geffken, Lewis, Johnson, Silverstein,
Rosenbloom, & Monaco, 1997). Glascow et al. (1987) suggested that the relationship
between adherence and metabolic control is not always straightforward, and regimen
adherence must be examined in the context of other factors. Anderson and Coyne (1993)
argue that examination of adherence in chronically ill children must include the parent' s
involvement in treatment tasks. When considering the complexity of diabetes
management, it is intuitive to consider that familial relationships and parental behaviors
will relate to child health status.
The relationships between family functioning, diabetes adherence and metabolic
control have been the focus of several analyses (Anderson et al., 1990; Lewin et al.,
2006; Liss, Waller, Kennard, McIntire, Capra, Stephens, 1998; McKelvey et al., 1993;
Schafer et al., 1986; Waller et al., 1986). For example, Hauser and colleagues (1990)
found that children's perceptions of family conflict, as measured by the Family
Environmental Scale (FES), were the strongest predictor of poor adherence with multiple
diabetes treatment regimen components (r = 0.50,
p < 0.001). In addition, the study suggested that child and parent perception of family
cohesion was related to both improved adherence, and comprehensive levels of patient
adherence. Likewise, a study found that child report of parent-child conflict (measured by
the Parent-Child Scales) contributes to the unique variance in predicting both adherence
(r = 0.50, p < 0.001) and metabolic control (r = 0.31, p < 0.05) (Miller-Johnson, Emery,
Marvin, Clarke, Lovinger, & Martin, 1994). Similarly, Hanson, De Guire, Schinkle,
Henggeler, and Burghen (1992) found that both disease-specific and general family
factors related to treatment adherence. Another study by Leonard et al, (2002) using the
Youth Self-report, found that individuals who reported higher levels of attention
problems, and higher levels of aggressive and delinquent behaviors reported worse
metabolic control as measured by HbAlc. However, there is some controversy in the
literature regarding the optimal method for measuring family interaction. Several studies
indicate that general measures of family functioning (general family stress, conflict,
support, etc.), such as the Family Environmental Scale (FES) and the Family Assessment
Measure (Skinner, Steinhauser, & Santa-Barbara, 1983), do not correlate with metabolic
control (Hauser et al., 1990; Liss et al., 1998; Schafer, Glasgow, McCaul, & Dreher,
It is possible that certain weaknesses in family functioning are distinct from
general family function, being limited to the issues and concerns specific to dealing with
the demands of diabetes. Measures of family functioning that are specifically designed
for families of children with diabetes (e.g., family support specifically related to diabetes
care) may be more sensitive in detecting relationships between family functioning and
metabolic control. While the FES and other general measures of family interaction may
assess characteristics such as family organization, the DFB S includes assessments of
diabetes specific family dimensions, e.g. organization specific to diabetes management.
For example, Liss et al. (1998) found that children hospitalized with DKA and their
parents reported less diabetes family support than did a clinic control group. However,
the Family Assessment Measure (a non-diabetes-specific measure of family functioning)
detected no differences in family functioning between the DKA and clinic control groups.
Additionally, Schafer et al. (1983) found that diabetes-specific family measures were
more predictive of adherence than were more general measures of family interaction
(mean correlation between general measures of family interaction and adherence M~<
0. 10; mean correlation between diabetes-specific measures and adherence M~= 0. 18. The
Diabetes-Specific Family Behavior Scale (DFBS; Waller et al., 1986), the Diabetes
Family Behavior Checklist (DFBC; Schafer et al., 1983), and the Diabetes Family
Responsibility Questionnaire (DFRQ; Anderson et al., 1990) are three measures of
diabetes-specific family support. There follows a description of these three measures
specifically developed to assess diabetes-specific family interaction.
Diabetes Specific Family Measures.
Waller and colleagues (1986) developed the Diabetes-Specific FamFFFFFFFF~~~~~~~~~ily Behavior
Scale of child/adolescent report of diabetes-specific family behaviors. The study
investigated whether the child' s overall perception of family diabetes-specific support
predicted metabolic control. The scale developed by Waller and colleagues (1986) was
based on the assumption that diabetes specific family behaviors consisted of three distinct
child perceived dimensions: parental warmth/caring, parental guidance/control about
diabetes, and diabetes related problem solving. Children and adolescents, ages 7 to 17,
completed the measure with results indicating that, for children under age 13 (n = 20),
there was a significant association (r = 0.50, p < 0.001) between the child's perception of
family-guidance/control and metabolic control (HbAlc), the same finding was not
present for older adolescents. Examples of DFB S items that correlated significantly with
metabolic control were as follows:
Whether the parent watches while the child tests for blood sugar (r = 0.55)
Whether the parent writes down blood sugar tests (r = 0.52)
Whether the child has someone in the family to talk to about diabetes (r = 0.55)
The correlation between metabolic control and warmth/caring was 0.36, p < 0.03 across
the entire age span of 7 to 17 years. Waller et al (1986) found the problem solving scale
did not account for a significant portion of the variance in metabolic control. Further
examination of the DFB S by McKelvey and colleagues (1993) replicated the correlation
between family guidance/control and diabetes metabolic control. However, the child
report of guidance/control accounted for only 1.4% of the variance in HbAlc and
problem-solving was again not related to metabolic control.
The Diabetes FamFFFFFFFF~~~~~~~~~ily Responsibility Questionnaire (Anderson et al. 1990) is a
questionnaire that measures mother and child perceptions of division of responsibility for
diabetes tasks related to treatment. In this study the measure was administered to both
children/adolescents (age 6-21) and their mothers. Responses for each item could be
classified into one of three possible patterns from the dyad scores: (1) perfect agreement
(both mother and child agree as to how the responsibility is shared), (2) overlap
(disagreement where both mother and child claim responsibility), and (3) no
responsibility (neither mother nor child report responsibility for a task). Factor analysis
revealed a three-factor solution representing these domains: general health management
tasks, regimen tasks, and social presentation of diabetes. The investigators found that
mother-child dyad scores indicating that no one takes responsibility for diabetes
management, and lower adherence scores based on parental report explained 13.4% of
the variance in HbAlc. Parent-child dyad scores indicating that neither the parent nor the
child takes responsibility for diabetes management correlated significantly with
metabolic control (r = 0.32, p < 0.001). A study by Wysocki and colleagues found that
children and adolescents reporting increased diabetes management responsibilities
relative their developmental level on a composite index of the DFRQ and the Diabetes
Independence Survey demonstrated less treatment adherence and marginally worse
metabolic control (Wysocki, Linschlid, Taylor, Yeates, Hough, & Naglieri, 1996).
Schafer and colleagues (1983) developed the Diabetes Family Behavior Checklist
(DFBC) to assess patient and family member report of supportive (positive) and
unsupportive (negative) diabetes-specific family behaviors. Administration of the DFBC
to 34 adolescents demonstrated significant relationships between negative DFBC scores
and adolescent's self-reported adherence to diet, care in measuring insulin, and frequency
of glucose testing.
However in a subsequent administration of the DFBC by Schafer and colleagues
(1986) wherein adolescents (ages 12 to 19) and adults (ages 20 and older) completed the
measures, negative family behaviors (related to glucose testing, insulin inj sections, and
diet) were inversely correlated with adherence in adults, but neither positive nor negative
DFBC scores were significantly related to adherence in adolescents. Reports of
unsupportive DFBC scores (criticism, nagging, argumentative statements) were
marginally correlated (r = 0.24, p < 0. 10) with metabolic control in adults but not in
adolescents. However, it may be that this study lacked statistical power to detect such a
relationship, as only 18 adolescents were included in this sample. Nevertheless, Hanson
and colleagues (1987) suggested that the measure was the best instrument available for
assessing parental support of the diabetes treatment regimen, despite the need for larger
sample sizes to demonstrate its predictive validity.
In a study evaluating a parent-adolescent teamwork approach to diabetes
management, families in the intervention group reported significantly less conflict
(measured by the DFBC) and were in better metabolic control (Anderson, Ho, Brackett,
& Laffel, 1999). In addition, Schafer et al (1983) demonstrated a significant relationship
between negative DFBC scores and adolescent's self-reported adherence.
Across all three diabetes-specific measures of family functioning were child
perception of parental support or lack of support of the diabetes treatment regimen.
McKelvey et al (1993) noted appropriate parent supervision of diabetes care tasks, as
measured in the DFRQ, may be closely related to the diabetes regimen guidance/control
dimension of the DFBS. The assessment of supportive and unsupportive parent behaviors
in the DFBC focuses on the same dimension of parent involvement in the disease
treatment. Children who perceive their parents as supportive of their diabetes care have
better metabolic control; children who perceive parents as negative, critical, and
unsupportive of their diabetes management are in poorer metabolic control, as are
families where no one takes responsibility for diabetes management. While the
relationships of the DFBC, DFRQ, and DFBS to metabolic control are small, in several
cases, the diabetes-specific family factors explain larger percentages of the variance in
metabolic control than adherence measures (SCAI, DSMP, 24-hour recall). Lewin et al
(2006) suggest that when examined concurrently, these measures may delineate a
construct of diabetes specific family factors related to adherence. Diabetes-specific
family functioning is a critical construct to evaluate to optimize metabolic control and
adherence outcomes (Anderson et al., 1997).
Externalizing, Diabetes Specific Family Functioning, Adherence and Metabolic
The Child Behavior Checklist (CBCL) was first published in 1978 (Achenbach &
Edelbrock, 1978). The CBCL has become a widely used standardized parent report
questionnaire for 4- to 18-year-olds that exhibits excellent psychometric properties
(Achenbach, 1991). The CBCL provides separate scores for Intemnalizing and
Externalizing. These broad band psychological problems have been identified repeatedly
in multivariate analyses (Achenbach, 1991) and reflect a distinction between depressed,
fearful, and inhibited over-controlled behaviors known as Internalizing, and aggressive,
oppositional, and antisocial under controlled behaviors known as Extemnalizing. There
exists very little work that has examined the relationships between child
psychopathology, family functioning, adherence and metabolic control. Our clinical
experience has led us to hypothesize that externalizing behavior has important
relationships to diabetes specific family functioning and adherence in explaining
metabolic control in youth with type 1 diabetes.
Recent research has found that more intensive diabetes regimens are related to
better metabolic control and that worse metabolic control leads to increased diabetes
related complications at an earlier age of onset in children and adolescents. Accordingly,
physicians have prescribed increasingly more demanding regimens. Clinicians believe
that the extent to which patients follow treatment recommendations is integral in
producing positive health outcomes. However, most existing psychological research only
weakly supports these assumptions. Prior research suggests the absence of a strong
relationship between behavior and health status in children and adolescents with type 1
diabetes. Methodological problems (e.g. the measurement of adherence, the complexity
and quality of the medical regimen, disease specific verses general measures of family
functioning, and the reliability of health-status indicators) are often cited as explanations
for these Eindings. Additionally, multiple studies fail to show strong relationships
between individual diabetes-specific family factors related to adherence and metabolic
McKelvey and colleagues (1993) suggested that incorporating the DFBC and
DFRQ scales with the DFBS might demonstrate a stronger connection between family
adherence and HbAlc. To date, only Lewin and colleagues (2006) have examined the
relationship between this combination of diabetes-specific family-related adherence
variables and metabolic control with encouraging results. Lewin et al (2006) found 34%
of the variance in metabolic control was accounted for by diabetes-specific family factors
after accounting for demographic variables. Further they hypothesized and successfully
demonstrated a mediation model wherein adherence partially mediated the relationship
between a composite of diabetes-specific family variables, and metabolic control.
Limited research has examined the contribution of children' s externalizing behavior to
metabolic control. The first aim of this investigation was to replicate the Eindings of
Lewin et al (2006) by accounting for a significant portion of the variance in metabolic
control through the use of diabetes-specific measures of family functioning related to
disease management. The second aim was to replicate a model wherein adherence
partially mediates the relationship between diabetes-specific family variables and
metabolic control. The third aim was to examine a proposed mediation model wherein
family factors mediate the relationship between externalizing and metabolic control. Our
fourth and final aim was to examine a proposed mediation model wherein adherence
mediates the relationship between externalizing and metabolic control.
Question 1: Family Factors Predicting Metabolic Control
Does a combination of diabetes-specific family functioning measures predict significant
variance in metabolic control in children and adolescents with type 1 diabetes?
Metabolic control will be predicted by the following adherence-related family
variables: both parent and child deny responsibility for diabetes care tasks (DFRQ); child
perception of the parent providing guidance/control over diabetes care (DFBS); child
perception of parental warmth and caring surrounding the diabetes treatment regimen
(DFBS), and child perceptions of parental nagging, critical, and argumentative behavior
related to diabetes management (DFBC).
Question 2: Adherence Mediating between Diabetes-Specific Family Factors and
Does adherence mediate the relationship between diabetes-specific family factors and
metabolic control in children and adolescents with type 1 diabetes?
Our study tested a model that proposes that diabetes-specific family factors affect
metabolic control through pathways of adherence. Adherence is expected to mediate the
relationship between diabetes-specific family factors and metabolic control (Figure 2-1).
Question 3: Family factors mediating between child externalizing and metabolic
Do diabetes-specific family factors mediate the relationship between externalizing and
Our study tested a model that proposes that children's externalizing behaviors
affect metabolic control through pathways of family factors. It is expected that child
externalizing behaviors will mediate the relationship between family factors and
Question 4: Adherence mediating child externalizing behaviors and metabolic
Does adherence mediate the relationship between child externalizing and metabolic
Finally, our study tested a model that proposes that child externalizing affects
metabolic control through pathways of adherence. It is expected that adherence will
mediate the relationship between externalizing behaviors and metabolic control (Figure
Figure 2-1. Theoretical model of adherence mediating family factors and metabolic
Figure 2-2. Theoretical model of family factors mediating externalizing and metabolic
B ehavi ors
Figure 2-3. Theoretical model of adherence mediating externalizing and metabolic
Participants and Settings
Participants were 120 children with type 1 diabetes and their parents, attending an
outpatient pediatric diabetes clinic affiliated with the University of Florida in Gainesville,
Florida. The sample was composed of 51 boys and 69 girls, ages 8.25 to 18.75 (M~=
13.92, SD = 2.71). The ethnic distribution was 72.5% Caucasian, 15.0% African
American, 10.0% Hispanic, and 2.5% representing other ethnic groups. More mothers
(78.3%) participated in the study compared to fathers (13.3%) and other parents (8.4%).
The mean family income was 46.6K with a standard deviation of 29. 1K.
Parents and children were recruited based on the following study inclusion
criteria: eight to eighteen years of age, a diagnosis of type 1 diabetes for at least 6
months, living with and accompanied by their primary caretaker, no evidence of mental
retardation. Of families approached, 82.8% agreed to participate. A signed informed
consent, approved by the Institutional Review Board of the University of Florida, was
obtained from each participant. Both parents and children completed self-report
questionnaires and one face-to-face structured interview while waiting for their scheduled
appointment. Trained graduate or undergraduate psychology students conducted
interviews individually with both caregiver and child of approximately 10-15 minutes in
length. Instructions were given to both for completing the self-report measures. Self-
report measures were typically completed within 30-40 minutes. Children and
adolescents were prompted to answer questions honestly and were reminded that no one
else in their family would see their answers. Blood samples for measuring HbAlc were
drawn by trained clinic staff as part of each patient' s routine visit.
All research assistants underwent extensive training with the first author in the
administration of these measures, which included: (1) attending an instructional meeting,
(2) observing 3 administrations of the measures, (3) administering the measures 3 times
with in vivo observation and supervision.
Parents were administered a brief demographics questionnaire to assess the
child's age, sex, race, age at onset of diabetes, family structure, parental education,
occupation and estimated income.
Diabetes-Specific Family Measures
Diabetes Family Behavior Scale (DFBS).
The DFB S is a measure of perceived family support for children and adolescents
with type 1 diabetes (Waller et al., 1986). This 60-item questionnaire has three subscales
(warmth/caring, guidance/control, and problem solving) that are used to determine the
child' s overall perception of family support. Scores from the guidance/control and
warmth/caring subscales correlated with metabolic control (.50 and .36, respectively), but
the problem-solving subscale did not (.08). Test-retest reliability coefficients for the
warmth/caring and guidance/control subscales are good (.79 and .83, respectively), but
not so for the problem-solving subscale (.51). Analyses indicate stronger reliability with
younger children (under age 13) that have been attributed to developmental differences
(McKelvey, Waller, North, et al, 1993). Children and adolescents completed the warmth
and caring, and the guidance and control scales of this measure. Overall, the DFB S has
good internal consistency (alpha coefficient of 0.82). Cronbach's a in our study was
acceptable (a = .74).
Diabetes Family Behavior Checklist (DFBC).
This instrument is a measure of both supportive and non-supportive behaviors by
the family and their effects on the child's diabetes self-care regimen (Schafer et al.,
1986). This forced-response, 16-item scale is designed for ages 12 to 64 and consists of a
summative response scale from 1 (never) to 5 (at least once a day). Summary scores are
generated for both the positive/supportive and negative/unsupportive domains. A positive
summary score can range from 9 to 45 and the negative summary scale can range from 7
to 35. Schafer and colleagues (1986) found that negative behaviors of family members
reported by the adults with diabetes were inversely correlated with adherence (glucose
testing, diet, and insulin injections). Parents and, children and adolescents completed this
measure. Cronbach's alpha coefficients range from .67 to .80 for the positive/supportive
scale and .74 to .82 for the negative/unsupportive scale (Schafer, personal
communication, 1998). Internal consistency for this sample was acceptable
(a = .64).
Diabetes Family Responsibility Questionnaire (DFRQ).
The Diabetes Family Responsibility Questionnaire is used to assess the family
sharing of responsibilities for diabetes treatment (Anderson et al., 1990). Both the parent
and child independently complete this measure consisting of reading 17 statements
concerning diabetes management tasks and indicating which family member has the
responsibility for the specific task. A parent-child dyadic score is calculated to determine
patterns of agreement and disagreement within the dyad. Factor analyses of this 17-item
measure indicate three domains of 1) general health management tasks, 2) regimen tasks,
and 3) social presentation of diabetes. Internal consistencies for the three subscales were
acceptable, ranging from .69 to .85 (Anderson et al., 1990). For this sample internal
consistency was acceptable (a = .75).
Child Behavior Checklist
Achenbach and Edelbrock first published the Child Behavior Checklist (CBCL) in
1978 and 1979. The CBCL has become a widely used standardized 118-item parent
report questionnaire for 4-18 years olds that exhibits excellent psychometric properties
(Achenbach, 1991). It was designed to assess the behavioral problems and social
competencies of children 4 to 18 years of age. The CBCL groups 20 competence items
into 11 Problem Scales (including eight Syndrome Scales) and four Competence Scales.
The CBCL also yields two broadband, higher order psychopathology scales, internalizing
and externalizing. The CBCL is widely used in both clinical and research settings
because of its demonstrated reliability and validity, ease of administration and scoring,
and applicability to clinical, nonclinical, and cross-cultural samples (Achenbach, 1991;
Cohen, Gotlieb, Kershner, & Wehrspann, 1985; Drotar, Stein, & Perrin, 1995; Sandberg,
Meyer-Bahlburg, &Yager, 1991). Cronbach's a in our study were strong (u = .94)
Measurement of Adherence
The DSMP is a refined version of the SCAI (Harris et al., 2000). The DSMP is a
structured interview, consisting of 23 questions, with an administration time of
approximately 15 to 20 minutes. Questions assess five areas of diabetes management,
including: insulin administration/dose adjustment, blood-glucose monitoring, exercise,
diet, and management of hypoglycemia. To minimize response bias, the interviewer
begins with a statement indicating that imperfect diabetes management is common.
Additionally, tasks for which nonadherence is more readily admitted (e.g. exercise) are
assessed first. Investigators found good internal consistency (Cronbach's alpha = 0.76)
and inter-observer agreement (94%). Parents and children were interviewed separately.
Internal consistency was found to be acceptable in our study for both parents (a = .86)
and children (a = .65).
Measure of Metabolic Control
During regularly scheduled appointments, patients in the Outpatient Diabetes
Clinic have their HbAlc routinely checked as part of their normal check-up. Blood
samples were analyzed using a Bayer DCA 2000+. HbAlc for individuals without
diabetes typically ranges between 4%-6.5% on this instrument.
Preliminary analyses were conducted to test for relations between demographic
variables and study variables (HbAlc, adherence, and family functioning) for purposes of
control in subsequent hierarchical multiple regression. HbAlc was correlated with child's
age (r = .23,
p < .01), duration with diabetes (r = .27, p < .01), and family income (r = -. 16, p = .09).
Child age was significantly correlated to the DFBC (r = .25, p < .01), to the DFBS
guidance and control subscale (r = -.57, p < .01), and the parent DSMP (r = -.37, p < .01).
Family income was found to be significantly correlated to both the child (r = .22, p < .05)
and parent DSMP (r = .21, p < .05). Duration of diabetes was significantly related to
HbAlc (r = .27, p < .01) and to parent DSMP (r = -.26, p < .01). All other demographic
variables were not significantly related to measures of diabetes-specific family
functioning, parent reports of child externalizing, or to reports of adherence.
Table 4-1 provides the sample characteristics on sex and ethnicity and Table 4-2
reports means, standard deviations, and ranges for child's age, HbAlc, duration with
diabetes, and income. The mean age and duration with diabetes for children in our study
was approximately 13.92 (SD = 2.7) and 4.80 (SD = 3.81) years, respectively. Mean
income was 46.0K (SD = 29.1K), and the mean HbAlc for the sample was 8.93% (SD =
1.99%), which is in the elevated range. No sex differences were found in HbAlc (t (120)
= 1.17, p = 0. 19, mean for boys: 9. 18% (SD = 1.77%); mean for girls: 8.75% (SD =
2. 13%). Likewise, no sex-differences were identified on measures of diabetes-specific
family functioning or self-report of adherence (oc = .05). Intercorrelations between
variables are found in table 3. All subsequent regression analyses were assessed for
collinearity by calculating tolerance and variance inflation factors (VIF) and no
significant collinearity was identified (Bowerman and O'Connell, 1990).
Study Aim 1: Regression of Metabolic Control (HbAlc) on Family Functioning and
To examine the relation between multiple measures of diabetes-specific family
factors and metabolic control, we conducted a hierarchical multiple linear regression. We
expected that when examined simultaneously, measures of adherence-related diabetes-
specific family factors would be predictive of significant variance in metabolic control.
To control for the influence of the child' s age, family income, and duration with diabetes,
we entered these variables into the regression equation in steps 1 and 2. To partition out
the variance shared by duration of diabetes and child's age these variables were
controlled in separate blocks, with child age and estimate of family income controlled in
the first block and duration of diabetes controlled in the second block. Combined, child
age and family income accounted for 6.7% of the variance in metabolic control, F (2,
102) = 3.69, p < .05. The unique variance accounted for by duration of diabetes was
5.3%, F (3, 101) = 4.62, p < .01. Considering that there were no a priori suppositions
regarding the magnitude of the relationship between each family functioning variable and
metabolic control, each family predictor was entered simultaneously into block 3 of the
regression. Combined, the four diabetes-specific family functioning variables (i.e.
parental warmth and caring (DFBS), guidance and control (DFBS), critical and negative
parenting (DFBC), and no-responsibility for the treatment regimen (DFRQ), explain an
additional 13.3% of the variance in metabolic control (HbAlc), F (7, 97) = 4.70, p < .01.
Lastly, child and parent report of adherence were entered into the model, which
combined accounted for 18.6% of the variance in metabolic control, F (9, 95) = 8.27, p <
.01. All together, the model predicted 43.9% of the variance in metabolic control
(HbAlc). The final model demonstrated significant contributions from child report of
critical, negative, unsupportive parents, and parent report of adherence (Table 4-4).
Study Aim 2: Does Adherence Mediate the Relationship Between Diabetes-specific
Family Factors and Metabolic Control?
To determine whether child report of adherence accounted for unique variance in
metabolic control beyond that of parent report, a regression equation was calculated
controlling for demographic variables in blocks one and two, with parent and child report
of adherence entered in the third and fourth blocks respectively. Parent report of
adherence explained 27.4% of the variance in metabolic control (HbAlc), F (4, 101) =
16.48, p < .01. Child report failed to account for significant variance in metabolic control
above and beyond that accounted for by parent report, explaining only .6% of the
variance in metabolic control. The complete model for both parent and child reports of
adherence, while controlling for demographic variables, accounted for 40. 1% of the
variance in metabolic control (See table 4-5). Since warmth and caring, guidance and
control, lack of responsibility, and child report of adherence, failed to add significantly to
the model in this and previous analyses, these variables were excluded from subsequent
Next, Baron and Kenny's (1986) guidelines for mediation were followed to test
the influence of the remaining diabetes-specific family factor (DFBC) on metabolic
control via adherence. The following criteria are necessary for mediation: (I) the
predictor (DFBC) should be significantly associated with the outcome (HbAlc), (II) the
predictor should be significantly associated with the mediator (adherence), (III) the
mediator should be associated with the outcome variable (metabolic control), and (IV)
lastly, the addition of the mediator to the full model should significantly reduce the
relationship between the predictor and outcome variable. Baron and Kenny's first through
third criteria were met in previous analyses, see table 3. To examine Baron and Kenny's
next criteria, family functioning (DFBC) was entered after demographic variables, and
parent report of adherence was entered into the final block of the regression. The family
functioning variable (DFBC) was found to account for 12.6% of variance in metabolic
control (HbAlc), while parent report of adherence (DSMP) added 18.6% to the model.
Finally, family functioning was entered into the model while controlling for adherence
and demographic variables. The final model accounted for 40.2% of the variance in
metabolic control (Table 4-6). To examine the significance of change in the path
coefficient, a Sobel's z-score (Sobel, 1988) was calculated, z = -6.70, p < .001, and was
found to be significant, meeting Baron and Kenny's criteria for mediation. The path
coefficient between family factors and metabolic control remained significant while
controlling for adherence, thereby indicating partial mediation. Standardized coefficient
of Family Factors on Metabolic Control equals .204 for direct and .201 for indirect effect
Study Aim 3: Regression Analysis to determine if Family Factors (DFBC) mediates
the relationship between Externalizing and Metabolic Control (HbAlc).
Baron and Kenny's (1986) guidelines for mediation were followed to test the
influence of child externalizing behaviors on metabolic control (HbAlc) via diabetes
specific family factors (DFBC). Child externalizing was entered into the regression, after
controlling for demographic variables, and was found to account for 5.6% of the variance
in metabolic control, F= 5.40,
p < .001. Next, family factors (DFBC) were entered into the model and were found to
account for 8.6% of the variance in metabolic control, F = 7. 11, p < .001. Finally, child
externalizing was entered into the model while controlling for family factors (DFBC) and
was found to account for a non-significant portion of the variance in metabolic control,
meeting criteria for full mediation (Table 4-7). To examine the significance of change in
the path coefficient, a Sobel's z-score (Sobel, 1988) was calculated, z = 2.67, p < .01, and
was found to be significant, meeting Baron and Kenny's criteria for mediation. The path
coefficient between externalizing and metabolic control became non-significant while
controlling for family factors, thereby indicating full mediation. The standardized
coefficients of externalizing on HbAlc equals .161 for direct, and .1 10 for indirect effect
Study Aim 4: Regression Analysis to Determine if Adherence (Parent DSMP)
Mediates the Relationship Between Externalizing Behaviors and Metabolic Control
Baron and Kenny's (1986) guidelines for mediation were followed to test the
influence of child externalizing behaviors on metabolic control via adherence. Child
externalizing was entered into the regression, after controlling for demographic variables,
and was found to account for 5.6% of the variance in metabolic control, F = 5.40, p < .01.
Next, parent reports of adherence (DSMP) were entered into the model and were found to
account for 22.2% of the variance in metabolic control, F = 13.24, p < .001. Finally,
externalizing was entered into the model while controlling for adherence and accounted
for a non-significant portion of the variance in metabolic control, indicating full
mediation (Table 4-8). To examine the significance of change in the path coefficient,
Sobel's z-score (Sobel, 1988) was calculated, z = 3.66,
p < .001, and was found to be significant, meeting Baron and Kenny's criteria for
mediation. Since the path coefficient between externalizing and metabolic control was
reduced to a non-significant effect while controlling for adherence, full mediation was
indicated. Standardized coefficient of DFBC on HbAlc equals .059 for direct and .212
for indirect effect (figure 4-6).
Type of Mediation ----- Partial
Sobel z-value ------------ -6.69526, p =.000001
Standardized coefficient of DFBC on HbAlc:
Figure 4-1. The relationship between family factors (DFBC) and metabolic control
(HbAlc) partially mediated by adherence (Parent DSMP).
Type of Mediation ---- Full
Sobel z-value ---------- 2.671389, p =.007554
Standardized coefficient of Externalizing on HbAlc:
Figure 4-2. The relationship between externalizing and metabolic control (HbAlc) fully
mediated by family factors (DFBC).
Type of Mediation ---- Full
Sobel z-value ----------- 3.66278, p =.000249
Standardized coefficient of Externalizing on HbAlc:
Figure 4-3. The relationship between externalizing and metabolic control (HbAlc) fully
mediated by adherence (Parent DSMP).
Table 4-1. Demographic characteristics
Table 4-2. Means and Standard Deviations for Age, HbAlc, Duration of Diabetes, and
Table 4-3. Intercorrelations among HbAlc, Diabetes Family Measures, Adherence, and
2 Child adherence report
3 Parent adherence report
4 Warmth and caring
5 Guidance and control
6 No responsibility
7 Critical parenting
8 Child externalizing
6 7 8
006 .405** .271**
)04 -.279* -.269**
)06 -.374** -.358**
49 -.324** -.320**
93* -.225* -.095
*p <. 05, **p< .01
Table 4-4. Hierarchical Regression Analysis Predicting HbAlc from Diabetes-Related
Family Factors and Adherence.
Step Variables R2 R2 ~
2 .121 .053 6.1
Duration of diabetes
3 .253 .133 4.3
Warmth and caring (DFBS)
Guidance and control (DFBS)
Lack of responsibility (DFRQ)
Critical parenting (DFBC)
4 .439 .186 15.
Child reported adherence (DSMP)
Parent reported adherence (DSMP)
All standardized regression coefficients are from the final block of the equation.
*p < .05, ** p < .01, *** p < .001
Table 4-5. Hierarchical Regression Analysis Predicting HbAlc from Parent and Child
Report of Adherence.
Step Variables R2 2R F1
1 .067 .067 3.69*
Child Age .066
2 .121 .053 6.10*
Duration of diabetes .124
3 .395 .274 45.79*
Parent report of adherence -.520
4 .401 .006 1.05
Child report of adherence
All standardized regression coefficients are from the final block of the equation.
*p < .05, ** p < .01, *** p < .001
Table 4-6. Mediation Regression Analysis Predicting HbAlc: Adherence Mediating the
Relationship between Family Factors (DFBC) and Metabolic Control
(HbAlc): Final Block of the Regression.
Step Variables R2 2R F1
1 .067 .067 3.69*
Child Age .030
2 .121 .053 6.10*
Duration of diabetes .126
3 .371 .274 45.79**
Parent report of adherence (DSMP) -.492**
4 .402 .035 6.20**
Critical parenting (DFBC) .206
*p < .05, ** p < .01
Table 4-7. Mediation Regression Analysis Predicting HbAlc: Family Factors (DFBC)
Mediating the Relationship Between Child Externalizing and Metabolic
Control (HbAlc): Final Block of the Regression.
Duration of diabetes
Family Factors (DFBC)
*p < .05, ** p < .01
Table 4-8. Mediation Regression Analysis Predicting HbAlc : Adherence Mediating the
Relationship Between Child Externalizing and Metabolic Control (HbAlc):
Final Block of the Regression.
Duration of diabetes
Parent report of adherence
*p < .05, ** p < .0
Interpretation of Results
The purpose of our study was to examine how behavioral variables including
adherence to the diabetes treatment regimen, child externalizing, and family functioning
specific to diabetes management related to metabolic control. First, we investigated the
relationship between a combination of diabetes-related family factors and metabolic
control. Regression analysis indicated that, taken together, diabetes-related family factors
accounted for 13.3% of the variance in metabolic control after controlling for
demographic variables. In addition, it is important to note that reports of adherence
explained significant variance in metabolic control. Our study replicated Lewin et al's
findings of diabetes related family factors and adherence significantly predicting
Another goal of our study was to replicate adherence mediating the relationship
between diabetes-specific family factors and metabolic control. The proposed mediating
model suggests that negative family functioning processes have a deleterious impact on
children's adherence behaviors and subsequent metabolic control. This model posits that
the child's perception of critical and negative parenting around the diabetes treatment
regimen is related to adherence to a treatment regimen. Overall, children who reported
more negative and critical relationships with their parents were in worse metabolic
control and such relationships are mediated by poor adherence. We believe it is likely
that there is a reciprocal relationship between critical parenting and poor diabetes
regimen adherence, with families becoming trapped in a coercive cycle or struggle for
control (Patterson, 1974). An adolescent's nonadherent behaviors may elicit parent
criticism, which in turn can lead to further adolescent rebellion toward adherence.
Regression analysis supported our third hypothesis that adherence mediates the
relationship between externalizing and metabolic control. The mediating model suggests
that children's externalizing behaviors have deleterious affects on children's adherence
behaviors and subsequent metabolic control. This model suggests that externalizing
behaviors interfere with good metabolic control and compliance with the prescribed
treatment regimen. Overall, children with higher reported externalizing behaviors were in
worse metabolic control. Parental difficulty dealing with externalizing behaviors such as
stubbornness and arguing, are unlikely to help with a child's adherence. Poor adherence
may fuel the parent and child battleground. Difficulties with adherence to complex and
demanding treatment regimens may exacerbate general child oppositionality and further
thwart any support parents might provide in management aimed at better metabolic
Finally, regression analysis supported our fourth hypothesis; that family factors
mediate the relationship between externalizing and metabolic control. The mediating
model suggests that externalizing is related to children's perceptions of critical and
negative parenting around diabetes and its relationship to poor metabolic control. This
model suggests that parent reported externalizing behaviors are linked to child reported
perceptions of negative parenting that includes criticizing, nagging, and arguing with the
child about diabetes related treatment tasks, leading to worsened metabolic control that
are shown, in aim three, to be largely through pathways of adherence. Overall children
with higher parent reported externalizing behaviors were in worse metabolic control.
Child externalizing behaviors may precipitate reciprocal negative behaviors from parents,
leading to increased family conflict that disrupts adherence processes and subsequent
metabolic control. It was shown that younger children who experienced parental warmth
had improved metabolic control (Davis et al, 2001), suggesting the reciprocal to be true.
Our study did not identify a significant association between child ratings of
parental guidance and control and metabolic control. However, given our sample mean
age of 13.92 years, this finding is not inconsistent with the extant literature, which
suggests that in adolescent populations, the relationship between parental
guidance/control and HbAlc is weaker (McKelvey et al., 1993; Waller et al., 1986). As
most adolescents strive for autonomy, parents may have less influence regarding whether
or not they attempt to provide guidance and control. This pattern suggests that while the
DFBS Guidance and Control Scale may have examined content independent from the
other scales, the construct was not related to metabolic control with this sample.
Additionally, our study did not find that the DFBS warmth and caring or responsibility
subscale was significantly related to metabolic control.
Consistent with Lewin et al (2006), significant relationships were found between
children reporting more critical, negative, unsupportive relationships with their parents
(regarding their diabetes management) and worse metabolic control. Also, parent and
child reports of adherence were strongly associated with metabolic control. Child report
of adherence did not significantly add to the variance accounted for in metabolic control
by parent report. Parent report of adherence accounted for 27.4% of the variance in
metabolic control, above and beyond demographic variables. However, the relationship is
reduced to 18.6% when family factors are controlled. The complete model did account
for 43.9% of the variance in metabolic control while controlling for demographic
Limitations, Implications, and Future Directions
Before implications of our study and speculations on future research are
discussed, it is important to note limitations of our study. Due to the correlational nature
of the data, causal relationships cannot be implied. However, ethically, a controlled
analysis could not be implemented since children cannot be randomly assigned to
supportive and unsupportive families. Secondly, families reporting to this clinic are
largely of low economic status and sponsored by state-funded insurance (State of Florida
Children's Medical Services). Results may not generalize to higher SES families. Third,
while children are informed that no parent or physician will see their results and are
encouraged to be as accurate as possible, there is the potential for report-bias on the
questionnaires and interview. Finally, administration of general family functioning
measures along with diabetes-specific questionnaires would have allowed for more
comprehensive comparisons with other similar studies.
Our study identified that reports of adherence are significantly related to
metabolic control. The use of adherence interviews may an important method to identify
those patients who are at risk for poor metabolic control. Such a preventative approach
may lead to more timely interventions, avoiding some of the risk associated with poorly
controlled diabetes. A reduction in HbAlc by 1% was associated with a 15 to 30%
decrease in the risk of microvascular and neuropathic complications of diabetes,
highlighting the significance of this difference (Diabetes Control and Complications
Research Group, 1987; ADA, 2003).
It is not a new concept, in a clinical context, that child behaviors may have
important implications for medical treatment, however an evidence-based examination of
this concept through research is still important. Our study highlighted the importance of
considering children's externalizing behaviors as they relate to other family factors,
children's adherence to their treatment regimens and metabolic control.
Importantly, adherence interviews used in our study can be completed in a
clinical setting, require minimal training to score and administer, and need little time to
complete. These brief screeners can identify families for behavioral health interventions.
Additionally, as there is variation in the complexity of the prescribed treatment regimen
that is largely clinically determined by the patient's ability to manage such regimens, it is
logical to posit that screening for behavioral suitability to the more complex regimens
would increase the success of more complex regimens.
Our data suggest that family behavior is significantly related to a child's health
status, especially when multiple aspects of family functioning related to regimen
behaviors are considered in making the assessment. Therefore, when identifying barriers
to adherence, clinicians should assess child perceptions of parental negativity, and child
externalizing behaviors in addition to assessing disease management behaviors.
The results of our study should also be considered during interventions for
children with poor metabolic control. Specifically, behavioral family-systems such as that
developed by Wysocki et al, (2006) seem most beneficial for these problems. Instead of
focusing exclusively on improving specific adherence behaviors, therapy should address
instrumental processes, such as improving family communication patterns and reducing
factors that promote and maintain argumentative interaction patterns specific to diabetes
management. Parent training should facilitate parenting described by Baumrind (1991) as
authoritative rather than authoritarian, indulgent or neglectful. In addition to a solid
understanding of issues surrounding diabetes treatment, therapists require process skills
to optimally address adherence problems in the context of improving family dynamics
related to diabetes management.
It is appropriate to conclude our study by reviewing the powerful finding that
parent report of adherence explained 27.4% of the variance in metabolic control, after
controlling for demographic variables, without directly measuring diabetes knowledge.
Congruent with clinical observations, adherence was highly related to metabolic control,
highlighting the importance of clinical assessments of regimen-specific adherence
behaviors and family factors to optimize metabolic control.
Future research should further examine the family functioning, adherence, and
metabolic control relations, identified in the mediation model from our study. For
example, this model may contribute to intervention studies of these relationships. Given
that improved health status and reduced diabetes-related complications are the clinical
goals, the aim should be to develop family models of adherence that optimize metabolic
control. Future analyses should examine other factors that might relate to diabetes family
functioning. For example, parents who already perceive themselves as overwhelmed
might be less likely to be supportive of and responsible for their child's diabetes
treatment regimen. Other examples include parent and child, anxiety and depression.
Lastly, as interactive patterns of behavior are unlikely to be unidirectional, future studies
should examine potential reciprocal interaction patterns between family variables,
adherence behaviors, and metabolic control that are specific to diabetes.
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist / 4-18 and 1991 Profile.
Burlington: University of Vermont Department of Psychiatry.
Achenbach, T. M., Edelbrock, CS. (1978). The Child Behavior Profile. Journal of Consulting
and Clinical Psychology, 46, 478- 488.
American Diabetes Association. (2003). Standards of medical care for patients with diabetes
mellitus. Diabetes Care, 25, S33-S50.
Anderson, B. J., Auslander, W. F., Jung, K. C., Miller, J. P., & Santiago, J. V. (1990). Assessing
family sharing of diabetes responsibilities. Journal of Pediatric Psychology, 15, 477-492.
Anderson, B. J., & Coyne, J. C. (1993). Family context and compliance behavior in chronically
ill children. In N. A. Krasnegor, L. Epstein, S. B. Johnson, & S. J. Yaffer (Eds.),
Developmental Aspects of Health Compliance Behavior (pp. 77-89). Hillsdale: Lawrence
Anderson, B. J., Ho, J., Brackett, J., & Laffel, L. M. B. (1999). An office-based intervention to
maintain parent-adolescent teamwork in diabetes management: Impact on parent
involvement, family conflict, and subsequent glycemic control. Diabetes Care, 22, 713-
Anderson, B. J., & Laffel, L. M. B. (1997). Behavioral and psychosocial research with school-
aged children with type 1 diabetes. Diabetes Spectrum, 10, 281-285.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal of
Personality and Social Psychology, 52, 1173-1182.
Baumrind, D. (1991). The influence of parenting style on adolescent competence and substance
use. Journal of Early Adolescence, 11, 56-95.
Cohen, N.J., Gotlieb, H., Kershner, J., & Wehrspann, W. (1985). Concurrent validity of the
internalizing and externalizing profile patterns of the Achenbach Child Behavior Checklist.
Journal of Consulting and Clinical Psychology, 53, 724-728.
Davis, C.L., Delamater, A.M., Shaw, K.H., La Greca, A.M., Eidson, M.S., Perez, M.S.,
Rodriquez, J.E., & Nemery, R. 2001. Parenting styles, regimen adherence, and glycemic control
in 4- to 10-year-old children with diabetes. Journal of Pediatric Psychology. 26, 126-129.
Diabetes Control and Complications Trial Research Group. (1986). Diabetes control and
complications trial: Design and methodologic considerations for the feasibility phase.
Diabetes, 35, 530-545.
Diabetes Control and Complications Trial Research Group. (1987). Results of the Feasibility
Study. Diabetes Care, 10, 1-19.
Diabetes Control and Complications Trial Research Group. (1993). The effect of intensive
treatment of diabetes on the development and progression of long-term complications in
insulin-dependent diabetes mellitus. The New England Journal of Medicine, 329, 977-986.
Diabetes Control and Complications Trial Research Group. (1994). Effect of intensive diabetes
treatment on the development and progression of long-term complications in adolescents
with insulin-dependent diabetes mellitus: Diabetes control and complications trial. The
Journal of Pediatrics, 125, 177-188.
Drotar, D., Stein, R.E., & Perrin, E.C. (1995). Methodological issues in using the Child Behavior
Checklist and its related instruments in clinical psychology research. Journal of Clinical Child
Psychology, 24, 184-192.
Engelgau, M. M., & Geiss, L. S. (2000). The burden of diabetes mellitus. In J. L. Leahy, N. G.
Clark, & W. T. Cefalu (Eds.), Medical Management of Diabetes Mellitus (pp. 1-17). New York:
Marcel Dekker, Inc.
EURODIAB ACE Study Group. (2000). Variation and trends in incidence of childhood diabetes
in Europe. Lancet, 355, 873-876.
Feingold, R. R. & Funk, J. L. (1997). Disorders of the endocrine pancreas. In McPhee, S. J.,
Lingappa, V. R., Ganong, W. F., and Lange, J. D. (Eds.), Pathophysiology of disease (pp. 423-
469). Stanford, CT: Appleton and Lange.
Freund, A., Johnson, S.B., Silverstein, J., & Thomas, J. (1991). Assessing daily management of
childhood diabetes using 24-hr recall interviews: Reliability and stability. Health
Psychology, 10, 200-208.
Geffken, G.R., Lewis, C., Johnson, S.B., Silverstein, J.H., Rosenbloom, A.L., & Monaco, L.
(1997) Residential treatment for youngsters with difficult to manage Insulin Dependent
Diabetes Mellitus. Journal of Pediatric Endocrinology and Metabolism, 10, 517-527.
Glasgow, R. E., McCaul, K. D., & & Schafer, L. C. (1987). Self-care behaviors and glycemic
control in type 1 diabetes. Journal of Chronic Diseases, 40, 399-412.
Hanson, C. L., De Guire, M. J., Schinkle, A. M., Henggeler, S. W., & Burghen, G. A. (1992).
Comparing Social Leamning and Family Systems Correlates of Adaptation in Youths with IDDM.
Journal of Pediatric Psychology, 17, 555-572.
Hanson, C. L., De Guire, M. J., Schinkel, A. M., Kolterman, O. G., Goodman, J. P., &
Buckingham, B. A. (1996). Self-care behaviors in insulin-dependent diabetes: evaluative tools
and their associations with glycemic control. Journal of Pediatric Psychology, 21, 467-482.
Hanson, C. L., Henggeler, S. W., & Burghen, G. A. (1987). Social competence and parental
support as mediators of the link between stress and metabolic control in adolescents with insulin-
dependent diabetes mellitus. Journal of Consulting and Clinical Psychology, 55, 529-533.
Harris M.A., Wysocki T., Sadler M., Wilkinson K., Harvey L.M., Buckloh L.M., Mauras N., &
White N.H. (2000). Validation of a structured interview for the assessment of diabetes self-
management. Diabetes Care, 23, 1301-1304.
Hauser, S. T., Jacobson, A. M., Lavori, P., Wolfsdorf, J. I., Herskowitz, R. D., Milley, J. E.,
Bliss, R., Wertlieb, D., & Stein, J. (1990). Adherence among children and adolescents with
insulin-dependent diabetes mellitus over a four-year longitudinal follow-up: II. Immediate
and long-term linkages with the family milieu. Journal of Pediatric Psychology, 15, 527-
Haynes, R. B. (1979). Introduction. In R. B. Haynes, D. W. Taylor, & D. L. Sackett (Eds.),
Compliance in health care (pp. 1-10). Baltimore: Johns Hopkins University Press.
Johnson, S. B. (1992). Methodical issues in diabetes research. Diabetes Care, 15, 1658-1667.
Johnson, S. B. (1993). Chronic diseases of childhood: Assessing compliance with complex
medical regimens. In N. A. Krasnegor, L. Epstein, S. B. Johnson, & S. J. Yaffer (Eds.),
Developmental Aspects of Health Compliance Behavior (pp. 77-89). Hillsdale: Lawrence
Johnson, S. B. (1994). Health behavior and health status: Concepts, methods, and applications.
Journal of Pediatric Psychology, 19, 129-141.
Johnson, S. B., Kelly, M., Henretta, J. C., Cunningham, W. R., Tomer, A., Johnson, S. B.,
Silverstein, J., Rosenbloom, A., Carter, R., & Cunningham, W. (1986). Assessing daily
management in childhood diabetes. Health Psychology, 9, 493-501.
Johnson, S. B., Kelly, M., Henretta, J. C., Cunningham, W. R., Tomer, A., & Silverstein, J. H.
(1992). A longitudinal analysis of adherence and health status in childhood diabetes.
Journal of Pediatric Psychology, 17, 537-553.
Kavanagh, D. J., Gooley, S., & Wilson, P. H. (1993). Prediction of adherence and control in
diabetes. Journal of Behavioral Medicine, 16, 509-522.
Kovacs, M., Goldston, D., Obrosky, S., & lyengar, S. (1992). Prevalence and predictors of
pervasive noncompliance with medical treatment among youths with insulin dependent
diabetes mellitus. Journal of the American Academy of Child and Adolescent Psychiatry,
La Greca, A. M. (1990). Issues in adherence with pediatric regimens. Journal of Pediatric
Psychology, 15, 423-436.
Leonard, B. J., Yuh-Pyng, J., Savik, K., Plumbo, P. M., & Christensen, R. (2002). Psychosocial
factors associated with levels of metabolic control in youth with type 1 diabetes. Journal of
Pediatric Nursing, 17, 28-37.
Lewin, A. B., Heidgerken, A. D., Geffken, G. R., Williams, L. B., Storch, E. A., Gelfand, K. M.,
& Silverstein, J. H. (2006). The relation between family factors and metabolic control: The
role of diabetes adherence. Journal of Pediatric Psychology, 3 1, 174-183
Liss, D. S., Waller, D. A., Kennard, B. D., McIntire, D., Capra, P., & Stephens, J. (1998).
Psychiatric illness and family support in children and adolescents with diabetic
ketoacidosis: A controlled study. Journal of the American Academy of Child and
Adolescent Psychiatry, 37, 536-544.
McKelvey, J., Waller, D. A., North, A. J., Marks, J. F., Schreiner, B., Travis, L. B., & Murphy, J.
N. III. (1993). Reliability and validity of the Diabetes Family Behavior Scale. The
Diabetes Educator, 19, 125-132.
McNabb, W. L. (1997). Adherence in diabetes: can we define it and can we measure it? Diabetes
Care, 20, 215-218.
Miller-Johnson, S., Emery, R. E., Marvin, R. S., Clarke, W., Lovinger, R., & Martin, M. (1994).
Parent-child relationships and the management of insulin-dependent diabetes mellitus. Journal of
Consulting and Clinical Psychology, 62, 603-610.
National Diabetes Data Group. (1995). Diabetes in America. 2d ed. Bethesda, Md.: National
Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; NIH
publication no. 95-1468.
Northam, E. A., Matthews, L. K., Anderson, P. J., Cameron, F. J., & Werther, G.A. (2005).
Psychiatric morbidity and health outcome in Type 1 diabetes perspectives from a prospective
longitudinal study. Diabetic Medicine, 22, 152-157
Notkins, A. L., & Lernmark, A. (2001). Autoimmune type 1 diabetes: resolved and unresolved
issues. Journal of Clinical Investigation 108, 1247-1252.
Onkamo, P., Vaananen, S., Karyonen, M., Tuomilehto, J. (1999). Worldwide increase in
incidence of Type 1 diabetes- the analysis of the data on published incidence trends.
Diabetologia, 42, 13 95-1403.
Patterson, G.R. (1974). The aggressive child: Victim and architect of a coercive system. In E. J.
Mash, L. A. Hamerlynck, & L. C. Handy (Eds.), Behavior modification and families (pp.267-
316). New York: Brunner/Mazel.
Sandberg, D.E., Meyer-Bahlburg, H.F., & Yager, T.J. (1991). The Child Behavior Checklist
nonclinical standardization samples: Should they be utilized as norms? Journal of the American
Academy of Child and Adolescent Psychiatry, 30, 124-134.
Schafer, L. C., Glasgow, R. E., McCaul, K. D., & Dreher, M. (1983). Adherence to IDDM
regimens: Relationship to psychosocial variables and metabolic control. Diabetes Care, 6, 493-
Schafer, L. C., McCaul, K. D., & Glasgow, R. E. (1986). Supportive and nonsupportive family
behaviors: Relationships to adherence and metabolic control in persons with type 1 diabetes.
Diabetes Care, 9, 179-185.
Skinner, H.A., Steinhauser, P.D., & Santa-Barbara, J. (1985). The Family Assessment Measure.
Canadian Journal of Community Mental Health, 2, 91-105.
Sobel, M. E. (1988). Direct and indirect effects in linear structural equation models. In J. S. Long
(Ed.), Common problems/proper solutions: Avoiding error in quantitative research (pp. 46-64).
Beverly Hills, CA: Sage.
Spevack, M., Johnson, S.B., & Riley, W. (1991). The effect of diabetes camp on adherence
behaviors and glycemic control. In J.H. Johnson & S.B. Johnson (Eds.), Advances in child health
psychology (pp. 285-292). Gainesville: University of Florida Press.
Waller, D. A., Chipman, J. J., Hardy, B. W., Hightower, M. S., North, A. J., Williams, S. B., &
Babick, A. J. (1986). Measuring diabetes-specific family support and its relation to metabolic
control: A preliminary report. Journal of the American Academy of Child Psychiatry, 25,
Wysocki, T., Linschlid, T. R., Taylor, A., Yeates, K. O., Hough, B. S., & Naglieri, J. A. (1996).
Deviation from developmentally appropriate self-care autonomy: Association with diabetes
outcomes. Diabetes Care, 19, 119-125.
Wysocki, T., Harris, M. A., Buckloh, L. M., Mertlich, D., Lochrie, A. S., Taylor, A., Sadler, M.,
Mauras, N., & White, N. H. (2006). Effects of Behavioral Family Systems Therapy for diabetes
on adolescents' family relationships, treatment adherence, and metabolic control. Journal
of Pediatric Psychology, 31, 928-938.
Zhang, Y., Krzentowski, G., Albert, A., & Lefebvre, P.J. (2001). Risk of developing retinopathy
in Diabetes Control and Complications Trial type 1 diabetic patients with good or poor
metabolic control. Diabetes Care, 66, 1275-1279.
Danny C. Duke was born on October 7, 1954 in Redlands, California. The oldest of
three children, he grew up living in Yucaipa and later in Placerville, California, where he
graduated from El Dorado High School in 1972. He owned a successful landscape
contracting business operating in the greater Sacramento, California area from 1973 until
2004. He has been married to his lovely wife (Vicki) since 1973 and they have two
children (Erin and Ryan). He earned his B.A degree in psychology from California State
University, Sacramento in 2002. He plans to complete his Ph.D. in Clinical Psychology
and to pursue his clinical and research interests in pediatric psychology and teaching.