Journal of Pediatric Psychology, Vol. 15, No. 4, 1990, pp.

477-492

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
Assessing Family Sharing of Diabetes Responsibilities1
Barbara J. Anderson2
Joslin Diabetes Center
Wendy F. Auslander, Kenneth C. Jung, J. Phillip Miller,
and Julio V. Santiago
Washington University

Received December 15, 1989; accepted January 9, 1990

This study examined sharing of diabetes responsibilities between mothers and

their diabetic children and the relationship between patterns of mother-child
sharing of responsibility for diabetes tasks and demographic variables, ad-
herence, and metabolic functioning in children with insulin-dependent dia-
betes mellitus (IDDM). A factor analysis of the Diabetes Family
Responsibility Questionnaire (DFRQ), a 17-item questionnaire developed for
the present study, resulted in a meaningful three-factor solution. Factors in-
cluded responsibilities related to regimen tasks, General Health Maintenance,
and Social Presentation of Diabetes. Analyses indicated that the DFRQ had
adequate internal consistency and concurrent validity. One hundred and
twenty-one children with IDDM, 6-21 years of age, and their mothers com-
pleted the DFRQ. Glycosylated hemoglobin (HbAlc) was used to index the
child's level of metabolic control. Results of multiple regression analyses in-
dicated that the child's age, disease duration, and sex are significant predic-
tors of mother and child patterns of sharing diabetes responsibilities.
Disagreements between mothers and children in perceptions of who is as-
suming responsibility and adherence level were significant predictors of
HbAlc. Results indicated that children assume increasing responsibility with

'This study was supported by grant AM20579 from the National Institute of Arthritis,
Metabolism, and Digestive Diseases, National Institutes of Health, to the Diabetes Research
and Training Center and the Edward MaJlinckrodt Department of Pediatrics, Washington
University School of Medicine, St. Louis, Missouri. This paper was presented in part to the
Annual Meeting of the American Diabetes Association, San Antonio, Texas, June 1983.
2
All correspondence should be sent to Barbara J. Anderson, Mental Health Unit, Joslin Dia-
betes Center, One Joslin Place, Boston, Massachusetts 02215.
477
0146-8693/9O/080O-O477S06.00/O © 1990 Plenum Publishing Corporation
478 Anderson, Auslander, Jung, Miller, and Santiago

increasing age. Clinicians should not assume that mothers and children com-
municate about the sharing of diabetes responsiblities in the family or about
changes in expectations of who is responsible as children develop. To foster
better control and adherence in diabetic children, members of the health care
team can help to identify diabetes tasks for which no one in the family takes
responsibility.

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
KEY WORDS: diabetes; family; mother-child dyad; responsibility; treatment regimen; adherence.

Adherence to the diabetic treatment regimen in children with insulin-

dependent diabetes mellitus (IDDM) requires the cooperation and contribu-
tion of the family. Continual adjustments may be made in family roles and
responsibilities to adapt to the demands of the treatment regimen on the fa-
mily. Of particular importance is the sharing and transfer of responsibility
that must occur between parent and child as the child grows older. From
8-12 years of age, the child participates gradually in diabetes care, sharing
in many of the diabetes tasks. By age 13 years, most youths can perform
all the regimen-specific tasks; however, the diabetic teen-ager continues to
need parental supervision and support to ensure adherence and to enhance
problem-solving skills of management (Follansbee, 1989; Johnson, 1985; Par-
tridge, Garner, Thompson, & Cherry, 1972; Sargent, 1985).
Although numerous studies have focused on relationships among fam-
ily functioning, adherence and metabolic control in diabetic children, and
the additional demands that the regimen places on the family (Anderson,
Miller, Auslander, & Santiago, 1981; Hanson, Henggeler, & Burghen, 1987;
Shafer, McCaul, & Glasgow, 1986), surprisingly few investigations have
described the sharing of diabetes-related responsibilities in families. Studies
that have examined responsibility in diabetic children have consistently found
that older children assume greater responsibility for tasks of the treatment
regimen (Allen, Tennen, McGrade, Affleck, & Ratzan, 1983; Rubin, Young-
Hyman, & Peyrot, 1989). In addition, according to studies by La Greca (1982)
and Allen et al. (1983) children who assumed greater responsibility for
regimen-specific tasks were in poorer metabolic control than children who
assumed less responsibility. Findings by Ingersoll, Orr, Herrold, and Golden
(1986) indicated that during the adolescent years, when responsibility for tasks
was transferred from parent to child, the child did not assume responsibility
as expected by the parent. Allen et al. (1983) compared parents' and chil-
dren's perceptions of sharing responsibilities and found no significant differ-
ences in the mean responsibility scores of parents as a group and children
as a group, except in the area of daily eating patterns. However, no study
to date has examined the level of agreement of sharing diabetes-related respon-
siblities between parent and child within the same family. Thus, the aims
of the present study are as follows:
Family Sharing of Diabetes Responsibilities 479

1. To introduce a clinically useful research tool for children with dia-

betes, age 6-21 years, and their parent(s): the Diabetes Family Responsibili-
ty Questionnaire (DFRQ), which measures individual family member's
perceptions of who takes responsibility for a broad range of diabetes
tasks —and to demonstrate acceptable psychometric properties for this in-
strument.

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
2. To describe patterns of agreement and disagreement between mothers
and children with diabetes concerning their perceptions of the division of
responsibility across a broad range of diabetes management tasks, as mea-
sured by a mother-child dyadic score based on responses of each on the
DFRQ.
3. To describe the relationships among demographic variables, DFRQ
mother-child scores, a global measure of adherence, and metabolic control
using bivariate and multivariate analytic techniques.

METHOD

Subjects

The sample consisted of 121 children diagnosed with IDDM, who were
receiving medical treatment at Children's Hospital of Washington Universi-
ty Medical Center, and their mothers. The families represented consecutive
admissions over a 1 lA- year period for an inpatient research hospitalization
(described below). All children were diagnosed with IDDM for at least 1 year
and were living at home. Table I reports the characteristics of the sample.

Table I. Sample Characteristics

n %
Sex
Male 56 46.3
Female 65 53.7
Race
White 107 84.4
Black 12 9.9
Other 2 1.7
No. of parents at home
2 105 86.8
1 16 13.2
M SD Range
Child's age (years) 13.3 3.3 6-21
Disease duration 5.5 3.5 1-15
SES (Hollingshead) 30.9 15.2 11-61
HbAlc (%) 11.6 2.7 6.3-19.3
480 Anderson, Auslander, Jung, Miller, and Santiago

The family's socioeconomic status was determined by the Hoilingshead Index

(1975) which takes into account the parents' occupational and educational
levels, marital status, and number of wage earners in the family. Our sam-
ple represented a broad range of socioeconomic levels (scores ranged from
11 to 61), with the majority of families scoring in the middle socioeconomic

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
status (SES) group.

Procedure

Families were seen voluntarily while children were healthy inpatients

during a research hospitalization for the collection of longitudinal biomedi-
cal data at the Clinical Research Center of Washington University Medical
Center. Approximately 95% of the 450 families with diabetic children who
were cared for at Children's Hospital agreed to this annual week-long
hospitalization, during which time the child received comprehensive eye, kid-
ney, and nerve tests, and biomedical data were collected for a longitudinal
research data base. Children participated in this research hospitalization only
when they were healthy and not experiencing any acute diabetes-related
problems, such as diabetic ketoacidosis. In 95% of the families, the mother
was the parent who brought the child to the Clinical Research Center for
hospitalizations. Therefore, the mother was the parent who responded to
the interviews and questionnaires in this study. Informed consent for this
study was obtained independently for children and mothers. Questionnaire
and interviews were administered separately to mothers and children by differ-
ent trained interviewers who were blind to the child's level of metabolic con-
trol. Blood samples were also collected during this time to index the child's
metabolic control. The present study was part of a larger project investigat-
ing family adaptation to childhood diabetes (see Anderson et al. (1981).

Measures

Diabetes Family Responsibility Questionnaire

Mothers and children independently completed the DFRQ, a self-report

questionnaire developed for the present study. The initial DFRQ consisted
of 22 items that describe diabetes and general health-related situations or
tasks relevant to children and adolescents. The content for the items was de-
rived from interviews with health care provider and professional diabetes
educators, as well as from pilot clinical interviews with famines with diabetic
children aged 6-20 years. Our intent was to include items sampling a broad
range of tasks in the family that contribute to the child's diabetes manage-
ment, with tasks ranging from remembering injections to noticing changes
Family Sharing of Diabetes Responsibilities 481

in the child's general health. From these professional and patient sources,
22 task were initially identified. For each item, the respondent assigned a
1 if the parent (mother and/or father) was predominantly in charge of the
task, a 2 if the child and parent shared the responsibility, and 3 if the child
assumed primary responsibility for the task. Items were written as phrases

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
in simple and nontechnical language. The interviewer was present while
mothers and children completed the DFRQ. For any child who had difficulty
reading the DFRQ, the interviewer simply read the items and the response
options to the child, and the assessment was carried out in an interview
format.
The 22 items were subjected to a principal components analysis to de-
termine if the individual items could be collapsed into reliable multi-item sub-
scales. The parent and child report data were analyzed separately. The final
analysis of the parent report data resulted in a meaningful three-factor
solution— General Health Maintenance; Regimen Tasks; Social Presenta-
tion of Diabetes — using 17 of the 22 original items. Five of the items were
dropped from the analysis due to either high loadings on more than one fac-
tor or failing to load on a meaningful factor. The eigenvalue for General
Health was 5.10; for Regimen Tasks, 1.85; and for Social Presentation, 1.54.
This resulted in 39, 11, and 9%, of the total item variance explained by each
of the three factors, respectively. An oblique rotation resulted in moderate
to low correlations among the three factors: General Health and Regimen
Tasks (r = .41, p < .001), General Health and Social Presentation (r =
.26, p < .001), Regimen Tasks and Social Presentation (r = .21, p < .01).
The resulting three subscales have acceptable internal consistency, with
alpha's ranging from .69 (for the shortest scale, Social Presentation of Dia-
betes) to .85 (for the total DFRQ scale). The abbreviated items of the three
factors and their alpha coefficients are reported in Table II. (The complete
items of the DFRQ are displayed in the Appendix.) The corresponding struc-
ture did not emerge based on a principal components analysis of all child
report data nor from only those children in the sample 12 years and older.
Therefore, for children's responses, the 17 items were combined into an overall
scale (alpha = .84) for further analysis of the child DFRQ data.
Evidence of concurrent validity for the Regimen Task subscale and the
total DFRQ scale for mothers was found when correlations were performed
between mothers' scores on the DFRQ and mothers' scores on the Moos
Family Environment Scale (FES, Moos, 1986). The FES is a measure of fam-
ily functioning that was collected simultaneously as part of the larger research
project described earlier. Higher reports of independence as a priority for
individual family members as measured by the FES Independence Subscale
were associated with mothers' reports of the child assuming greater respon-
siblity for Regimen tasks on the DFRQ (r = .27, p < .01) and for the total
tasks of the DFRQ scale (r = .21, p < .05).
482 Anderson, Auslander, Jang, Miller, and Santiago

Table II. Factor Loadings and Cronbach's Alpha Coefficients of the DRFQ
(Mother's Report)
Factor loadings
General Health Regimen Tasks Social
General Health (a = .78)
Expiration' dates for

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
supplies .7065 -.0663 .0300
Foot care .6267 .1378 - .0833
Make MD appointments .5680 - .2363 .0410
Adjust insulin dose .5576 .3440 -.0994
Notice health changes .5185 .0387 .0788
Choose food when out .4646 .0359 .0176
Remember clinic .4288 .0576 .0868
Regimen Tasks (a = .79)
Rotate injection sites .0343 .7293 .0591
Give shots -.1418 .6988 .0226
Remember shots .1065 .6084 .1677
Carry Candy - .0552 .5976 -.0897
Notice reaction symptoms -.0044 .5158 -.2113
Remember Tests .3113 .4142 -.0609
Social Presentation (a = .69)
Tell friends .1171 .0866 .7559
Tell relatives .1702 -.2103 .7401
Explain school absences .0346 .0401 .6405
Tell teachers .2377 .0479 .4826
Total scale alpha = .85

To investigate further the relationship between mothers' responses and

children's responses in the same family, a dyadic mother-child score was con-
structed for each family. For each item, we examined the extent to which
mother and child agreed or disagreed in identifying the family member who
was assuming responsibility, and to what extent responsibility for the task was
assumed in the family. By considering mother and child responses together,
one of three possible response patterns was produced for each item: (a)
Mother and child agree precisely as to how responsibility for a task is shared
(Perfect Agreement pattern); (b) mother and child disagree, with each claim-
ing more responsibility than the other family member reports (Overlap pat-
tern); and (c) mother and child disagree, with each reporting that the other
family member takes more responsibility for the task (No One Takes Respon-
sibility pattern). This pattern also includes situation in which one member
reports they share responsibility and the other reports s/he has no responsi-
bility for the task. Disagreements of the second type, the Overlap pattern,
where both mother and child report taking responsibility for the task, indi-
cate that while mother and child do disagree, both family members report
taking responsibility, and, importantly, the task is being completed or co-
vered. Clinically, we were most interested in the third pattern, where "no
one takes responsibility," as this is the family pattern most likely to be re-
Family Sharing of Diabetes Responsibilities 483

lated to poor health outcomes, with neither mother nor child assuming respon-
sibility for the task and each reporting that the other person is performing
the task.
Our primary variable of interest, therefore, was a mother-child dyadic
score, that measured the degree to which no one in the family reported taking
responsibility across a range of diabetes management situations. This

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
mother-child score, constructed for further analyses, had a theoretical range
of 0-17, with a score of 0 meaning that there were no items for which the
mother-child dyad indicated no one takes responsibility, and a score of 17
meaning that the dyad indicated no one takes responsibility for all 17 DFRQ
items.

Interviews

Structured interviews, which were part of the larger research project,

and the self-report DFRQ were conducted separately with mothers and chil-
dren. Demographic information such as child's age, sex, disease duration,
race, family SES, and family structure were obtained during the parent in-
terview. Both maternal and child perspectives on overall adherence to the
treatment regimen were obtained through an identical item on parent and
child interviews. Mothers and children were asked to rate the child's level
of overall adherence to treatment on a 3-point scale, where 1 indicated higher
adherence levels than 3 (i.e., lower scores reflect higher levels of adherence).
For all of these pediatric diabetes patients at Children's Hospital, the
prescribed treatment regimen consisted of (a) taking two insulin injections
per day; (b) testing blood glucose three times each day; (c) following an in-
dividualized meal plan; and (d) regular exercise. The problem with reliably
measuring adherence to the diabetic treatment regimen, a multivariate con-
struct, has been discussed by Glasgow, Wilson, and McCaul (1985). In the
absence of a more reliable and accurate method at the time of the present
study, overall adherence to diabetes treatment was assessed with a single in-
terview item.

Metabolic Control

At the time of the study, our laboratory utilized glycosylated hemoglobin

(HbAlc) to assess the child's metabolic control. HbAlc provides an estimate
of blood glucose control over the previous 6-8 week period and has been
shown to be a valid and reliable measure of the child's metabolic function-
ing (Daneman, Wolfson, Becker, & Drash, 1981). Higher HbAlc values are
associated with poorer metabolic control. The mean HbAlc value for 290
diabetic children at our clinic with a disease duration of at least 1 year was
484 Anderson, Auslander, Jung, Miller, and Santiago

11.9% (S£> = ± 2.7). Mean values for nondiabetic individuals of the same
age from our laboratory were reported as 5.9% ± 1.3 (Davis, McDonald,
& Jarett, 1978).

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
RESULTS

Correlations: Demographic Variables and DFRQ Scores

As stated earlier, one of the primary aims of this study was to develop
a research tool for measuring individual family member's perceptions of who
takes responsibility for the tasks of diabetes treatment, which would also
be clinically useful in working with the family system around adherence and
support issues. Thus, in order to translate the individual responses from
DFRQs of mother and child into a dyadic score that would provide some
indication of how the mother-child system was operating with respect to
responsibility for diabetes management tasks, we constructd a mother-child
score as described earlier. The mean dyadic mother-child score for the sam-
ple across the total DFRQ scale was 2.3 (SD = 1.8, with a range of 0-9).
Table III shows the frequencies of disagreement scores of the pattern, "No
one takes responsibility"; 16% of the sample did not reveal any items of this
disagreement pattern, that is, they were in Perfect Agreement or reported
Overlap in assuming responsibility for all items. On the other extreme, 12.6%
of the mother-child dyads reported "No one takes responsibility" on 5 to
9 of the DFRQ items, indicating that each member of the dyad attributed
more responsibility for the task to the other family member. Further exami-
nation of the items on which mother-child pairs responded with a "No one
takes responsibility" pattern indicated that 70% of the families who reported
this pattern did so for one or more items on the Regimen Tasks subscale.
the majority of the sample, 75%, reported 1 to 3 items for which "No one
takes responsibility."

Table III. Mother-Child Scores on Diabetes Family Responsibility

Questionaire Indicating "No One Takes Responsibility"
No. of items (out of 17) indicating
"No one takes responsibility" Frequency of dyads %
0 19 16.0
1 27 22.7
2 28 23.5
3 16 13.4
4 14 11.8
5-9 15 12.6
Family Sharing of Diabetes Responsibilities 485

Pearson correlations were computed in order to determine the relation-

ships among demographic variables, maternal scores on the DFRQ subscales,
mother and child total scale scores, mother-child dyadic scores, overall ad-
herence ratings, and the child's level of metabolic control. Degrees of free-
dom equaled 120 for all variables except DFRQ variables, for which degrees
of freedom ranged between 100 and 120. Table IV displays a correlation

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
matrix of these variables. Our data show that age is strongly associated with
scores of mothers and children on the total DFRQ. Older children assume
greater responsibility in the General Health, and Regimen Tasks domains
(p < .001) and Social Presentation domain (p < .05) assessed by maternal
questionnaire. Older children also assumed greater responsibility as measured
by the total DFRQ for mothers and children (p < .001). Age was also sig-
nificantly related to mother-child scores indicating "No one takes responsi-
bility"; higher levels of this type of disagreement between mother and children
were found among younger children (p < .01). As expected, age was as-
sociated with longer disease duration {p < .001). Both mother and child
reports of child's overall adherence were related to the child's age (p < .01),
with older ages associated with lower levels of adherence (higher scores indi-
cate lower levels of adherence). Finally, age was also associated with HbAlc
levels, with older children (adolescents) in poorer metabolic control than
younger children (p < 05).
In addition to age, associations were found between the child's sex and
mothers' responses on the DFRQ. Girls were reported to take more respon-
sibility than boys on tasks on the Social Presentation of Diabetes subscale
(p < .05).
Higher mother-child dyadic scores indicating "No one takes responsi-
bility" were significantly related to better adherence to treatment as reported
by mothers (p < .05). Stated in another way, when mothers and children
each reported that the other member of the dyad was assuming more respon-
sibility for the task, mothers perceived higher levels of adherence. When con-
trolling for age, mother-child dyadic scores approached significance (p =
.09, j3 = - .158). Also, the children's report of overall adherence was sig-
nificantly related to their total DFRQ score, indicating the children who per-
ceived that they were taking more responsiblity also reported higher levels
of adherence (p < .05). Utilizing a general measure of overall adherence,
significant relationships were found between HbAlc and adherence levels
as reported by mothers (p < .001) and children (p < .01), with lower levels
of adherence relating to poorer metabolic control.

Multiple Regression Analyses

In order to identify significant predictors of DFRQ scores and HbAlc,

multiple regression analyses were performed using simultaneous entry of
 Table IV. Correlation Matrix of Predictor and Outcome Variables
1 2 3 4 5 6 7 8 9 10 11
1.
Child's Age -
2.
Sex: 0 = Male, 1 = female -.08 —
3.
Disease duration .40' .02 —
4.
General Health (M report) .73' .04 .18 —
5.
Regimen Tasks (M report) .57' .03 .07 .56' —
6.
Social presentation (M report) .19° .20° .16 .35' .24* _
7.
Total scale (M report) .68' .10 .19° .85' .83' .62' —
8.
Total scale (C report) .74' .05 .23° .69' .58' .36' .70'
9.
Mother-child dyadic score -.28' .10 -.13 -.07 .05 .21° .07 -.50'
("No one takes responsibility") o
10. Overall Adherence (M report): .26* -.17 .17 .00 -.01 -.16 -.04 .14 -.23° - -B
1 = all the time, >
1 = sometimes, ~
3 = hardly ever §
11. Overall adherence (C report) .29" .02 .16 .17 .08 -.02 -.08 .20" -.06 .13 - q
12. HbAlc .18° -.04 .12 .13 .14 -.12 .10 .06 .10 .32' .24* "^
^ ^ _ ^ _ ^ _ _ _ ^ _ ^ ^ — ^ — — ^ _ ^ _ _ _ _ (3
"p < .05, two-tailed. g
b
Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014

p < .01, two-tailed. "_

c
p < .001, two-tailed. E
ST
•i
to
D
O.
Family Sharing of Diabetes Responsibilities 487

predictor variables. This technique was utilized so that the unique variance
of each predictor variable could be determined, while controlling for the ef-
fects of the other variables in the model. The unique variance is defined as
the proportion of R2 that is attributed to the predictor variable as if it was
entered last in the model (Pedhazur, 1982). Separate models were generated
to predict mother and child scores on the DFRQ and HbAlc. Subjects were

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
dropped from the multiple regression analysis if they had any missing items
on the DFRQ, thus explaining the varying degrees of freedom across the
models. Only variables that were significant beyond the .05 level in the bivar-
iate analysis, with the exception of the mother-child dyadic scores, were in-
cluded in the multivariate analysis. This is a standard method of data
reduction, since in most cases variable that are not significant at the bivari-
ate level are not signficant at the multivariate level.

Predictors of DFRQ Scores

Several demographic variables (age, duration, sex) were entered in

regression models to predict maternal responses on the DFRQ. For both the
General Health and Regimen Tasks subscales, age and duration remained
significant predictors of mothers' scores, while sex did not remain signifi-
cant. Therefore, sex was dropped from the final model. Overall, age 0 =
.802) and duration (/3 = .164) explained 56% of the variance in the General-
Health subscale and age remained the strongest predictor, F(2, 107) = 67.7,
p < .0001. No two-way or three-way (sex, age, duration) interactions were
found among the predictor variables.
A similar model emerged in predicting Regimen tasks, explaining 36%
of the variance, F(2, 112) = 30.89. p < .0001. Age remained the strongest
predictor (/3 = .647) and there were no interactions present.
The model to predict Social Presentation differed from the models for
the other two subscales. Sex of the child 03 = .197) was the only significant
predictor, with girls assuming greater responsibility than boys, and explained
only 4% of the variance in the dependent variable, F(l, 114) = 4.6, p < .05.
Models were also generated to predict the mother and child total DFRQ
scores. Using age, sex, and disease duration as predictor variables, only age
and sex remained signficant predictors of both parent and child total scores.
The final model, with sex and age as predictors, explained 49% of the vari-
ance in parent scores, F(2, 103) = 49.2, p < .0001, and 56% of the vari-
ance in children's scores, F(2, 100) = 64.7, p < .0001. No interactions were
found among the independent variables. Overall, age, disease duration, and
sex were significant predictors of scores on DFRQ reported by mothers and
children.
488 Anderson, Auslander, Jung, Miller, and Santiago

Predictors of Metabolic Control

The model to predict HbAlc included those variables that were signifi-
cant at the bivariate level (child's age and adherence). However, because of
the significant relationship found between parent report of adherence and
mother-child scores on the DFRQ indicating "No one takes responsibility,"

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
mother-child scores on the DFRQ were also included in the model. Separate
models were generated, one that included parent report of adherence and
one with child report of adherence.
Interestingly, when controlling for mother report of adherence 03 =
.319) and mother-child scores 03 = .223), age 03 = .117)) did not signifi-
cantly predit the child's metabolic control. In addition, when controlling for
adherence levels, mother-child scores emerged as a significant predictor of
HbAlc. The final model, then, while explaining only 13.4% of the variance
of HbAlc, indicated that higher levels of mother-child scores indicating "No
one takes responsibility; 03 = .183), and lower adherence levels 03 = .360)
significantly contribute to poorer metabolic control, F(2, 112) = 8.56, p <
.001. The same pattern was found using child report of adherence, however,
the unique variance contributed by mother-child scores only approached sig-
nificance (p = .07). No significant interactions were found among any of
the independent variables to predict metabolic control.

DISCUSSION

The first aim of this study was to develop a research instrument with
adequate psychometric properties that has clinical applications in working
with families around issues of family sharing of responsibility for diabetes
treatment tasks. We produced a 17-item scale that, based on responses of
mothers in our sample, factored into three meaningful subscales or domains
of responsibility: General Health Maintenanace tasks, Regimen tasks, and
Social Presentation of Diabetes. By considering mother and child scores
together, we were able to define a pattern that indicated "No one takes respon-
sibility." We were most interested in the level of disagreement, as a higher
level of disagreement would most likely relate to adherence and control
problems. For example, when family members believe that the regimen task
is being covered by another family member, they feel less responsible for
the implementation of this task. For tasks such as "remembering shots" and
"remembering blood tests" when no one assumes responsibility the task is
likely not being completed, without awareness of this by other members of
the family. These are very high-risk mistakes from a diabetes care perspective.
One important clinical use for this tool with families is to assess and
discuss who takes responsibility in the family for a task such as remember-
Family Sharing of Diabetes Responsibilities 489

ing shots, so that disagreements can be brought out into the open and clari-
fied. Work by Ingersoll et al. (1986) has documented that when parents
delegate responsibility for insulin adjustment to older adolescents, adoles-
cents do not necessarily assume the primary responsibility for this task. Clear-
ly, across many areas of diabetes management, parents decrease their
supervision with the expectation that the adolescent will both assume respon-

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
sibility and increase participation in the regimen task. Similarly to Ingersoll
et al., our findings indicate that in some families communication may need
to be strengthened with respect to who has the responsibility for the task.
Age was strongly related to mothers' and children's responses on the
DRFQ. It is important to point out that this research was a cross-sectional
study and did not provide longitudinal data. However, within this cross-
sectional data base, older children assumed greater responsibility across all
three subscales: Regimen Tasks; Social Presentation of Diabetes; and General
Health Maintenance. Allen et al. (1983), La Greca (1982), and Rubin et al.
(1989) have also reported that both parents and diabetic children attribute
increasing responsibility for self-care to the child with increasing age. This
is consistent with the general developmental trend for parents to delegate
increasing levels of responsibility for decision-making across a wide range
of areas to their children, beginning in the early adolescent years (Cooper,
Grotevant, & Condon, 1983).
Age was also related to mother-child "No one takes rsponsibility" scores,
higher levels of this type of diagreement were found between mother and
younger children. This may reflect the lack of concordance between adult
and child perspectives normally found in younger children (Piaget, 1932).
In addition, transitions in diabetes responsbilities from parent to child gener-
ally increase as the child approaches adolescence (La Greca, 1988). There
may be an increased need, therefore, for ongoing communication around
diabetes expectations in the family as children approach the early adolescent
years. Greater numbers of "No one takes responsibility" responses between
preadolescent diabetic children and their mothers may indicate that commu-
nication in these families about expectations for responsibility has lagged,
precisely at the time that transitions in diabetes responsibilities are accerler-
ating. In many families the focus on diabetes management may also decrease
after the family stabilizes following a crisis period such as at diagnosis or
an acute diabetes-related medical emergency (Etzwiler & Sines, 1962). Par-
ents with a diabetic child may not focus on the reality that the onset of adoles-
cence presents a new crisis in terms of expectations for self-care and new
patterns of family responsibility for covering the tasks demanded by diabetes.
Young adolescents have ongoing needs for support and some degree of paren-
tal involvement in diabetes management tasks (Ingersoll et al., 1986).
In our sample, age was also significantly associated with level of over-
all adherence, with lower levels of adherence reported in older children. Other
investigators have also documented that adherence deteriorates with increasing
490 Anderson, Auslander, Jung, Miller, and Santiago

age in diabetic youngsters (Allen et al., 1983; La Greca, 1988). Similarly,

in the multiple regression analyses, age was the strongest predictor of mothers'
responsibility scores on all three subscales and of mothers' and children's
scores on the total DRFQ. When the effects of disease duration were con-
trolled for, age and sex remained significant predictors of both mother and
child total DRFQ scores, explaining almost 50% of the variance in mothers'

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
scores and 56% of the variance in children's scores. This indicated that both
mothers and children reported higher responsibility levels with increasing age,
and reported higher responsibility levels for girls than for boys. Girls ma-
ture physically at a more rapid rate than boys, and parents frequently have
higher behavioral expectations for their more developed female children
(Brooks-Gunn & Zahaykevich, 1989). These naturally occurring gender differ-
ences in parental behavioral expectations may explain why parents and chil-
dren in our study reported giving more responsibility to girls than to boys.
Finally, we found that age significantly correlated with metabolic con-
trol, with older children (i.e., adolescents) in poorer metabolic control than
younger children. Other researchers have also reported that metabolic con-
trol, as indexed by HbAlc, higer levels of mother-child "No one takes respon-
sibility" scores, and lower levels of maternal report of adherence significantly
contributed to poorer metabolic control. However, this model explained only
13.4% of the variance in HbAlc. These findings indicate that the child's lev-
el of metabolic functioning is a complex composite of multiple factors. Yet
mother and child perceptions of responsibility need to be considered when
attempting to understand contributions to the child's metabolic control. In
explaining research findings indicating that adolescents who had the most
responsibility were in the poorest control, La Greca (1988) suggested that
"children responsible for their own care may not have been executing the
glucose monitoring or insulin measurement tasks effectively" (p. 145).
Similarly, our finding that higher mother-child "No one takes responsibili-
ty" scores predict higher HbAlc may indicate that in addition to effective-
ness of task administration, we need to ask of all family members: "Who
is responsible for the basic execution of the task?"
These findings emphasize the importance of ongoing family communi-
cation around expectations of responsibility for diabetes management tasks
with children. The health care team can work together with family members
to help clarify and confirm who is able to assume responsibility for diabetes
tasks and how responsibilities are best shared to insure adherence to the treat-
ment plan.

REFERENCES

Allen, D. A. Tennen, H., McGrade, B. J., Affleck, G., & Ratzan, S. (1983). Parent and child
perceptions of the management of juvenile diabetes. Journal of Pediatric Psychology,
8, 129-141.
Family Sharing of Diabetes Responsibilities 491

Anderson, B. J., Miller, J. P., Auslander, W. F., & Santiago, J. V. (1981). Family characteris-
tics of diabetic adolescents: relationship to metabolic control. Diabetes Care, 4, 586-594.
Brooks-Gunn, J., & Zahaykevich, M. (1989) Parent-daughter relationships in early adolescence:
A developmental perspective. In K. Kreppner & R. M. Lerner (Eds.), Family systems
and life span development (pp. 223-246). Hillsdale, NJ: Erlbaum.
Cooper, C. R., Grotevant, H. D., & Condon, S. M. (1983). Individuality and connectedness
in the family as a context for adolescent identity formation and role-taking skill. In H.

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
Grotevant & C. Cooper (Eds.), Adolescent development in the family (pp. 43-59). San
Francisco: Jossey-Bass.
Daneman, D., Wolfson, D. H., Becker, D. J., & Drash, A. L. (1981). Factors affecting glyco-
sylated hemoglobin values in children with insulin-dependent diabetes. Journal of Pedi-
atrics, 99, 847-853.
Davis, J., McDonald, J., & Jaret, L. (1978). A high-performance liquid chromatography method
for hemoglobin Ale. Diabetes, 27, 102-107.
Etzwiler, D. D., & Sines, L. K. (1962). Juvenile diabetes and its managment; Family, social
and academic implications. Journal of the American Medical Association, 181, 304-308.
Follansbee, D. S. (1989). Assuming responsibility for diabetes management: What age? What
price? Diabetes Educator, 15, 347-352.
Glasgow, R., Wilson, W., & McCaul, K. (1985). Regimen adherence: A problematic construct
in diabetes research. Diabetes Care, 8, 300-301.
Hanson, C. L., Henggeler, S. W., & Burghen, G. A. (1987). Model of associations between
psychosocial variables and health-outcome measure of adolescents with IDDM. Diabetes
Care, 10, 752-758.
Hollingshead, A. B. (1975). Four-factor index of social status. Unpublished manuscript, Yale
University, Department of Sociology, New Haven.
Ingersoll, G., Orr, D., Herrold, A., & Golden, M. (1986). Cognitive maturity and self-
management among adolescents with insulin-dependent diabetes mellitus. Journal of Pedi-
atrics, 108, 620-623.
Johnson, S. B. (1985). The family and child with chronic illness. In D. C. Turk & R. D. Kerns
(Eds.), Health, illness and families (pp. 220-254). New York: Wiley.
La Greca, A. M. (1982). Behavioral aspects of diabetes management in children and adoles-
cents. Diabetes, 52(Suppl. 2), 47.
La Greca, A. M. (1988). Children with diabetes and their families: Coping and disease manage-
ment. In T. Field, P. McCabe, & N. Schneiderman (Eds.), Stress and coping across de-
velopment (pp. 139-159). Hillsdale, NJ: Erlbaum.
Moos, R. H. (1986). Family environment scale manual. Palo Alto, CA: Consulting Psycholo-
gists Press.
Partridge, J. W., Garner, A. M., Thompson, C. W., & Cherry, T. (1972). Attitudes of adoles-
cents toward their diabetes. American Journal of Diseases, of Childhood, 124, 226-229.
Pedhazur, E. J. (1982). Multiple regression in behavioral research.. New York: Holt, Rinehart
& Winston.
Piaget, J. (1932). The moral judgment of the child. Glencoe, IL: Free Press.
Rubin, R., Young-Hyman, D., & Peyrot, M. (1989). Parent-child responsibility and conflict
in diabetes care (Abstract). Diabetes, 38, (Suppl. 2), 28.
Sargent, J. (1985). Juvenile diabetes mellitus and the family. In P. I. Ahmed & N. Ahmed (Eds.),
Coping with juvenile diabetes (pp. 205-233). Springfield, IL: Charles C Thomas.
Schafer, L. C , McCaul, K. D., & Glasgow, R. E. (1986). Supportive and nonsupportive fami-
ly behaviors: Relationships to adherence and metabolic control in persons with type I
diabetes. Diabetes Care, 9, 179-185.
492 Anderson, Auslander, Jung, Miller, and Santiago

APPENDIX

Diabetes Family Responsibility Questionnaire (DFRQ)

Below are different tasks or situations that relate to diabetes management

Downloaded from http://jpepsy.oxfordjournals.org/ at University of Tennessee ? Knoxville Libraries on December 22, 2014
in your family. Choose one number from the three statements that best
describes the way each task or situation is handled in your family.

1 = Parent(s) take or initiate responsibility for this almost all of the time.
2 = Parent(s) and child share responsibility for this about equally.
3 = Child takes or initiates responsibility for this almost all of the time.
Situations or tasks:
1. Remembering day of clinic appointment. (GH)*
2. Telling teachers about diabetes. (S)
3. Remembering to take morning or evening injection. (R)
4. Making appointments with dentists and other doctors. (GH)
5. Telling relatives about diabetes. (S)
6. Taking more or less insulin according to results of blood
sugar or urine tests. (GH)
7. Noticing differences in health, such as weight changes or
signs of an infection. (GH)
8. Telling friends about diabetes. (S)
9. Noticing the early signs of an insulin reaction. (R)
10. Giving insulin injections. (R)
11. Deciding what should be eaten when family has meals out
(restaurants, friend's home). (GH)
12. Examining feet and making sure shoes fit properly. (GH)
13. Carrying some form of sugar in case of an insulin reaction.
(R)
14. Explaining absences from school to teachers or other school
personnel. (S)
15. Rotating injection sites. (R)
16. Checking expiration dates on medical supplies (GH)
17. Remembering times when blood sugar or urine should be
tested. (R)

*GH = General Health domain; R = Regimen domain; S = Social Presen-

tation domain.