Diabetes Problem Solving by Youths with Type 1 Diabetes and their
Caregivers: Measurement, Validation, and Longitudinal Associations
with Glycemic Control
Tim Wysocki,1PHD, Ronald Iannotti,2PHD, Jill Weissberg-Benchell,3PHD, Lori Laffel,4MD,
Korey Hood,4PHD, Barbara Anderson,5PHD, Rusan Chen,6PHD, for the Family Management
of Childhood Diabetes Steering Committee
1Nemours Children’s Clinic,2National Institute of Child Health and Human Development,3Children’s Memorial
Hospital,4Joslin Diabetes Center,5Texas Children’s Hospital, and6Georgetown University
(T1DM) and adult caregivers in correcting glycemic fluctuations.
Interview (DPSI), a structured interview, was validated during a pilot study of a behavioral intervention. DPSI
data and measures of diabetes management were obtained at baseline from 114 youths (ages 9–14.5) and 109
caregivers. Glycosylated hemoglobin (HbA1c) was measured quarterly over 9 months.
confirmed the psychometric adequacy of the DPSI. For caregivers, but not youths, low DPSI scores
(indicating poor problem-solving skills) were significantly associated with worse HbA1cover 9 months.
Conclusions The DPSI has clinical and research utility as a measure of diabetes problem-solving skills.
Identification and targeted remediation of caregivers’ deficient diabetes problem-solving skills or promotion of
youths’ utilization of these skills could possibly enhance glycemic control in youths with T1DM.
This article introduces a new measure of problem-solving skills of youths with type 1 diabetes
Methods The Diabetes Problem Solving
Key words adolescents; children; diabetes; problem solving.
Youths with type 1 diabetes mellitus (T1DM) who can
can enjoy a more flexible lifestyle, while minimizing
their risks of serious long-term complications (Diabetes
Control and Complications Trial Research Group, 1994).
Living with T1DM requires youths and their parents/
caregivers to recognize unwanted fluctuations in blood
glucose levels and to respond promptly and effectively to
these events (Chase, 2006). Chronically, high blood
glucose (hyperglycemia) raises the risk of both short-term
complications (e.g., diabetic ketoacidosis) and long-term
complications (e.g., cardiovascular disease, retinopathy,
nephropathy, and neuropathy). Episodes of low blood
glucose (hypoglycemia) may be associated with cognitive
decrements (Hershey, Lillie, Sadler, & White, 2002),
blood glucose control
increased risk of injury, embarrassment, and disruption
of normal activities. The extent to which extreme
excursions of blood glucose can be either prevented or
corrected promptly may be a key factor affecting
emotional adjustment to diabetes, self-efficacy about its
management, family conflict and symptoms of depres-
sion, andanxiety among
Delamater, & Santiago, 1990). Consequently, the profi-
ciency of youths and parents in responding to unwanted
blood glucose fluctuations is likely to be a critical
determinant not only of medical outcomes of diabetes,
but also of its psychosocial outcomes. Experience gained
in the Diabetes Control and Complications Trial (DCCT;
Diabetes Control & Complications Trial Research Group,
1993) suggests anecdotally that patients who acquired
strong diabetes problem-solving skills tended to report
All correspondence concerning this article should be addressed to Tim Wysocki, PhD, Center for Pediatric
Psychology Research, Nemours Children’s Clinic, 807 Children’s Way, Jacksonville, FL 32207-8426, USA.
Journal of Pediatric Psychology 33(8) pp. 875–884, 2008
Advance Access publication March 17, 2008
Journal of Pediatric Psychology vol. 33 no. 8 ? The Author 2008. Published by Oxford University Press on behalf of the Society of Pediatric Psychology.
All rights reserved. For permissions, please e-mail: email@example.com
more flexible lifestyles and more favorable diabetes-related
quality of life compared with patients who lacked
these skills.The American
(Mensing et al., 2005) includes diabetes problem-solving
skills among its required curricular elements in its
standards for recognition of diabetes education programs.
Problem solving is an element of executive function-
ing and requires analysis of the problem, generation of
possible solutions, evaluation of the risks and benefits of
those solutions, and analysis of the outcomes (Bagner,
Williams, Geffken, Silverstein, & Storch, 2007; Cook,
Alkens, Berry, & McNabb, 2001). Reliable and valid
measurement of these skills is a prerequisite to conduct-
ing sound research on this topic and to applying that
research to clinical practice (Johnson, 1984, 1995). There
have been several tests of diabetes knowledge for youths
and parents published in the past (Eastman, Johnson,
Silverstein, Spillar, & McCallum, 1983; Harkavy, et al.,
1983; Johnson, et al., 1982; La Greca, Follansbee, &
Skyler, 1990; Wysocki, et al., 1996), but these instru-
ments typically demand relatively low levels of cognitive
operations such as recognition or recall of facts, rather
than more sophisticated cognitive skills such as applica-
tion, analysis, synthesis, or evaluation (Bloom, 1984;
Wysocki, 2000). Effective diabetes problem solving may
be mediated by these higher cognitive functions.
We are aware of only three such tools that have been
developed and validated: Johnson’s Test of Diabetes
Awareness and Reasoning Test developed by Heidgerken
et al. (2007), and the Diabetes Problem-Solving Measure
for Adolescents (Cook et al., 2001). The first two
instruments included several diabetes problem-solving
scenarios in which children and parents are asked to
identify which of several alternative solutions is most
appropriate in each situation. The Test of Diabetes
monitoring of blood glucose, the use of insulin pumps,
the recent introduction of new insulin types, and the
intensification of diabetes management subsequent to the
DCCT and so its applicability to modern therapy for
T1DM is limited. The Diabetes Awareness and Reasoning
Test includes modern diabetes regimens but requires
recognition of a correct solution rather than generating
appropriate solutions or providing a rationale for selection
of the most appropriate solution. These problem-solving
items do not represent a unique construct; they have the
same psychometric properties as the knowledge items
and have been incorporated into a single total score along
with items assessing general knowledge about insulin,
al., 1982), theDiabetes
widespreaduse of self-
nutrition, hyperglycemia/hypoglycemia, pump use, and
school issues. The Diabetes Problem-Solving Measure for
Adolescents most closely meets the criteria for a problem-
solving task (vignettes are presented and solutions are
spontaneously generated by the respondent), but it was
developed and validated with a cross-sectional sample of
older adolescents(ages 13–17).
primary purpose of this article is to introduce a new
measure of diabetes problem-solving skills suitable for
preadolescents and early adolescents and their adult
caregivers, and to present information on its psycho-
metric properties using longitudinal data.
youths to achieve and maintain acceptable glycemic
control, while deficiencies in those skills should be
purpose of the work reported here was to evaluate the
extent to which baseline measurements of adult care-
givers’ and youths’ diabetes problem-solving skills were
associated with glycosylated hemoglobin levels measured
prospectively over 9 months.
Participants in this study were enrolled in a multi-site
pilot and feasibility study that was preliminary to a larger
randomized controlled trial of a clinic-based, family-
focused intervention designed to optimize family adapta-
tion to childhood diabetes during late childhood and
and feasibility study were to evaluate the feasibility and
acceptability of the measurement and intervention proto-
cols and to refine those protocols for the larger
randomized controlled trial to follow. All parents or
other legal caregivers had signed institutionally approved
informed consent or parental permission forms and all
youths had assented to participation in the study using
each center’s approved procedures for doing so. Eligibility
criteria for children and adolescents were: age 9.0
through 14.5 years; duration of T1DM of 12 months or
longer; established diabetes care at the enrolling center;
absence of other chronic systemic diseases; grade-
appropriate reading skills in English; not enrolled in
self-contained special education; and no history of
psychiatric hospitalization within the prior 6 months.
Enrollment criteria for parents/caregivers and families
included at least 5th grade reading fluency in English;
absence of diagnosis of psychosis, substance use disorder,
major depression, or bipolar disorder; no history of
objectives ofthe pilot
Wysocki et al.
psychiatric hospitalization in prior 6 months; and work-
ing telephone service.
The full sample for the pilot and feasibility study
included 122 children with T1DM and their caregivers, of
whom 114 youths and 109 adult caregivers were
interviewed using the DPSI as described subsequently.
Missing data were attributable primarily to malfunctions
of the recording equipment or to inaudible recordings
that could not be transcribed satisfactorily. Demographic
characteristics of the 114 youths and 109 parents whose
data were analyzed for this article are summarized in
Table I. There were no statistically significant differences
at baseline between the intervention and control groups
on these demographic characteristics.
Families enrolled at one of four pediatric diabetes centers
located in the Northeastern, Southeastern, Southwestern,
and Midwestern United States. Participants were enrolled
in a pilot randomized trial designed to establish the
feasibility of the intervention and measurement protocols
prior to a larger randomized controlled trial (Nansel et al.,
manuscript under review). Families
assigned, stratified by age (<12 or ?12-years old) and
most recent glycosylated hemoglobin level (<8.3 or
?8.3%), to either the Intervention or Control conditions.
All study families received the same medical care for
diabetes during the study that they would have received if
not enrolled in the study. At all sites this consisted of
clinic visits approximately once each 3 months with a
pediatric endocrinologist and a diabetes nurse and, as
needed, a dietitian, social worker, or psychologist.
Diabetes regimens at baseline were: 44 youths (36%)
on insulin pumps, 32 (26%) on ‘‘basal-bolus’’ multiple
daily injection regimens and 46 (38%) on conventional,
fixed-dose insulin injection regimens with no differences
between the Intervention and Control groups. Self-
management education was strongly emphasized and
targets of treatment were to maintain HbA1Cas close to
normal as possible, while minimizing the occurrence of
families received up to three sessions (M¼2.85) of a
family-focused, low-intensity behavioral intervention deliv-
ered during quarterly routine diabetes clinic visits over a
6-month study period. Intervention families were taught
by a specially trained Health Advisor (with a BA or MA in
a behavioral science or related field) to apply a basic
problem-solving strategy to daily problems in family
management of childhood diabetes. Families received
education, assistance with negotiation of an intervention
plan, and supplemental handouts. Clinic encounters were
followed by telephone follow-ups to evaluate and refine
the intervention plan and to prepare for the subsequent
clinic visit. These circumstances were maintained for a
period encompassing three successive diabetes clinic
visits for each study family, or ?6 months. Each family
also had one follow-up visit ?9 months after randomiza-
tion, at which time the final blood sample for a
glycosylated hemoglobin assay was collected. Although
treatment outcome measures were obtained, demonstra-
tion of a treatment effect favoring the experimental
intervention was not expected during the pilot and
feasibility study because of its relatively short duration,
low intervention dosage, modest sample size, and
preliminary intervention content and materials.
Biomedical data were collected by medical record review
during or just after diabetes clinic visits. All question-
naires and interviews were collected at assessments
Table I. Demographic Characteristics of Parents (n¼109) and Youths
(n¼114) who Contributed Data for this Report
Duration of diabetes (Years)
Relationship to child with diabetes
Household annual income (%)
Highest educational level (%)
High school graduate
Diabetes Problem Solving
conducted in the home by field interviewers who were
unaware of the participants’ group assignments or of
the adequacy of the child’s diabetes management.
The measures described subsequently were collected at
baseline prior to the family’s randomization and following
the last clinic visit during the study period. Measures
beyond those described here were administered, but only
those listed below were analyzed for the purposes of the
Diabetes Family Responsibility Questionnaire (DFRQ)
This is a 17-item questionnaire on which caregivers or
children with T1DM rated the degree to which respon-
sibility for each diabetes management task is a Parent
Responsibility, a Shared Responsibility, or a Child
Responsibility (Anderson, Auslander, Jung, Miller, &
levels of child responsibility for diabetes management.
Acceptable internal consistency, test–retest reliability and
parent–child agreement have been reported consistently
in a number of studies that have used this instrument.
Greater parent–child discordance in scores on this
measure has been associated with higher HbA1Clevels.
Alpha coefficient for the present sample was .67 for
caregivers and .73 for youths.
Diabetes Self Management Profile (DSMP)
This is a 24-item structured interview for the assessment
of diabetes treatment adherence and self-management
designed for administration to caregivers or youths ?11
years of age (Diabetes Research in Children Network,
2005; Harris et al., 2000). Separate forms have been
validated for patients treated on conventional, fixed-dose
insulin regimens and for patients on flexible insulin
regimens (i.e., insulin pumps or ‘‘basal-bolus’’ injection
regimens) in which insulin bolus doses are adjusted
proactively based on carbohydrate counting and dosage
correction factors that account for prevailing glucose
levels. Internal consistency (a-coefficients) for the present
sample’s total scores was .66 for the Youth Conventional
Regimen form, .67 for the Youth Flexible Regimen form,
.71 for the Parent Conventional Regimen form, and .76
for the Parent Flexible Regimen form. The caregivers’ total
score correlated significantly with youths’ HbA1C levels
Glycosylated Hemoglobin (HbA1C)
This blood test estimates average glycemic concentration
over the prior 2–3 months (Chase, 2006). Blood samples
were obtained quarterly by fingerstick and shipped to
Joslin Diabetes Center for processing using the Tosoh
High Performance Liquid Chromatography method. This
schedule yielded three HbA1C measurements during
the 6-month intervention phase of the study and one
follow-up measurement at 9 months.
An initial version of the DPSI consisted of separate
structured interviews of parents/caregivers and children in
which they were faced with four realistic diabetes vignettes.
A collection of 12 vignettes was developed in consultation
with pediatric endocrinologists and diabetes educators at
each study center. Each interview addressed four vignettes,
one each concerned with Prevention of Hypoglycemia,
Prevention of Hyperglycemia, Correction of Hypoglycemia,
and Correction of Hyperglycemia. In order to minimize
observational reactivity, parents and youths were adminis-
tered different randomly selected sets of vignettes at each
evaluation. Participants were also administered different
sets of vignettes at each successive evaluation to minimize
practice effects. Participants were asked this series of
questions about each vignette:
? What is the diabetes problem here?
? Why is this a problem? What would happen if he/she
? Tell me all the ways this problem could be fixed.
? How would you fix this problem?
? How would that solution work?
? How would you know if you really fixed the problem?
Interviews were audio-recorded and subsequently
transcribed verbatim for coding. Raters who were blind
to the intervention assignment and demographic char-
acteristics of the study families scored each transcript
using detailed vignette-specific coding rules (available
from the first author). Each response to a given question
was rated as a ‘‘0’’ if a respondent offered no answer or
gave an incorrect response (e.g., ‘‘I don’t know’’) a ‘‘1’’ if
a respondent offered a partially correct or incomplete
response (e.g., the respondent would treat a presumed
low blood glucose without first completing a blood
glucose check) or a ‘‘2’’ if a respondent gave a correct
response with supporting details or evidence of meta-
cognition (e.g., ‘‘After taking extra insulin, I would
re-check my blood glucose about once an hour until
I was sure it was back to normal’’). All transcribed
interviews were rated independently by two raters to
permit assessment of inter-rater agreement (see Results
After completion of interviews with the first 67
families who entered the study, concern about the overall
Wysocki et al.
length of the baseline assessment, which included
collectionof multiple measures,
evaluation of the need to administer all four types of
scenarios at each interview. The interviewers reported that
engagement of many children and some caregivers tended
to wane when four scenarios were presented. There were
also a number of instances in which both caregivers
and children had difficulty discerning the diabetes
problem that was implicit in the ‘‘Prevention’’ vignettes.
Consequently, in addition to dropping certain other
measures from the assessment protocol, a decision was
made to delete the two ‘‘Prevention’’ vignettes, and to
retain the two ‘‘Correction’’ vignettes from the DPSI. The
remaining 47 caregiver–child dyads who were interviewed
at baseline, and all interview participants at End of Study,
were presented with two vignettes, consisting of one
Correction of Hypoglycemia and one Correction of
Hyperglycemia vignette for their analysis and resolution;
for all participants, only scores on these two types of
vignettes were analyzed and reported in this paper.
A DPSI total score, consisting of the number of points
credited for each of the six structured questions
(maximum of 12 points) averaged across the two
vignettes (possible range 0–12), was generated for each
participant. Caregivers and children were always inter-
viewed in separate rooms about different vignettes to
minimize potential reactivity. Table II shows the three
Correction of Hypoglycemia and three Correction of
Hyperglycemia vignettes that were utilized in this study.
Since the various DPSI vignettes differed in mean
scores following the baseline administration of the
interview, it was decided to devise a weighted scoring
procedure to equate the difficulty levels of the various
scenarios. Weights were calculated separately for children
and caregivers. First, mean scores for each vignette were
calculated using both baseline and end of study data.
A mean for each type of vignette (correction of low and
correction of high) was then calculated using all six
vignettes within that type. The ratio of the mean score of
each vignette within that type to the mean for that type
of vignette was then calculated. The inverse of this ratio
was used to correct individual scores on the correspond-
ing vignette. All DPSI scores entering the analyses
reported below were adjusted using this method.
Various statistical analyses permitted evaluation of the
measurement properties of the DPSI. These included
calculation of descriptive statistics for each vignette
separately for caregivers and youths at baseline and
end of study; internal consistency for each vignette
and for the DPSI total scale at baseline and end of
and caregiver and child DPSI total scores at baseline
and end of study; comparison of and associations
between caregiver and child DPSI total scores at baseline
and end of study; inter-rater agreement at baseline
and end of study; and convergent validity as measured
by associationswith other
family diabetes management and diabetes outcomes.
Between-group effects on DPSI scores were examined
by comparing change in DPSI total scores from baseline
to end of study for the intervention group and the
control group. Finally, Mixed Effects models (Hedeker
& Gibbons, 2006) were used to examine the longitu-
dinal association between DPSI scores and glycemic
control with repeated HbA1C levels as the outcome
variable. Compared with conventional repeated measures
analyses of variance (RM-ANOVA), the mixed effects
models approach confers several statistical and inter-
capacity to control simultaneously for possible effects
of intervention groups and clinical sites on HbA1C,
of the longitudinal
Table II. Vignettes Utilized in the DPSI
Correction of low blood sugar (Hypoglycemia) (CL) vignettes:
Tim played basketball before lunch at school. Later he was waiting in line for lunch. He started feeling dizzy, hot, and shaky.
Colin is at school and starts to feel shaky and weak in the middle of math class.
Lori had played soccer all morning. Just after she ate a big lunch, she started feeling really shaky, sweaty, hot, and like she might faint.
Correction of high blood sugar (Hyperglycemia) (CH) vignettes:
Amanda was supposed to go to soccer practice after school, but the coach cancelled it at the last minute. She ate lunch, but didn’t take any insulin
because she thought she would be playing soccer for about 2hrs.
Emma was at dinner at her great aunt’s house. Before dinner, she guessed that she would be eating 75g of carbohydrates and she took her premeal
insulin based on that guess. On the way home, she asked her parents about this and they thought she had eaten more like 100g of carbs.
Mary Ann went to the pool with her friends. She thought she was going to be swimming so she took less insulin than usual before her lunch. But,
when she arrived at the pool, it was closed.
Diabetes Problem Solving
relationship among DPSI and HbA1C during the entire
periodof observation rather
discrete time points, the use of DPSI raw scores
rather than derived cutoff scores as is implicit in
RM-ANOVA, and the capacity to retain cases for
analysis if there are missing HbA1C values instead of
excluding those cases as in RM-ANOVA. All available
HbA1Cvalues at 0, 3, 6, and 9 months were included
in the analysis. Average DPSI scores were used as a
between-subject predictor in the model. Intervention
effect (coded 1 and 0 for treatment and control groups,
respectively) and site (as a random effect) were included
in the model for controlling possible intervention and
between-site effects on HbA1C.
The Intervention and Control groups did not differ
significantly in mean HbA1C, or total scores on the
Responsibility Questionnaire, or DPSI at any measure-
ment point. Since there was no treatment effect on DPSI
scores or the primary diabetes outcomes that were
measured, the Intervention and Control groups were
DPSI Descriptive Statistics
Mean?SD caregiver scores on the DPSI (maximum¼12)
at baseline were 7.1?1.7 and child scores were
6.5?1.7, a statistically significant difference (paired
samples t¼3.19; p<.01). At baseline, there were no
significant differences in DPSI total scores for either
parents or youths who were administered four vignettes
versus those who were administered two vignettes. At 6
months, the mean score for caregivers increased to
7.6?1.4 (t¼2.25; p<.05), while that for youths
increased to 7.0?1.5 (t¼2.92; p<.01). The statistically
significant difference between caregiver and child scores
persisted at 6 months (t¼?3.60; p<.001).
Internal consistency of the DPSI (Cronbach’s a-coeffi-
cient) was .59 for caregivers and .67 for children at
baseline and .53 and .51, respectively, at 6 months.
Item-total correlations were
both caregivers and children at baseline (M¼.44;
range¼.19–0.54) and 6 months (M¼.36; range¼.13–
0.45) and statistically significant. Stability of DPSI scores
over time was modest, with Pearson correlations between
baseline and 6 months of .33 (p<.001) for children and
.35 (p<.001) for caregivers. Caregiver scores were not
significantly correlated with child scores at baseline
(r¼.17) or at 6 months (r¼.17). Intra-class correlations
were calculated to assess inter-rater reliability. Based on
independent coding of all vignettes, the intra-class
correlation was .82 for the baseline assessments and
.80 for the 6-month assessments.
At baseline, children’s scores correlated significantly with
child age (r¼.18; p<.05) but not with duration of
diabetes. Caregivers’ scores correlated significantly with
HbA1c(r¼?.31; p<.002), while youths’ scores did not.
With child age controlled statistically, caregiver DPSI
scores correlated significantly with degree of child
responsibility for diabetes management (DFRQ) (r¼.23;
p<.002). Higher parental DPSI scores predicted more
child responsibility and lower baseline HbA1C.
Longitudinal Associations between DPSI Scores
and Glycosylated Hemoglobin
Two mixed effects models were used to evaluate the
association between DPSI scores and HbA1C, one for
parent and one for child DPSI scores, respectively.
In both models, the effect for treatment group was not
a significant predictor of HbA1C. To facilitate graphical
presentation of the longitudinal effects of interest,
participants were divided into categories based on
whether their baseline DPSI scores fell into the lowest,
(Low), middle (Medium), or highest (High) third of the
DPSI score distribution. Figure 1 shows that parent DPSI
scores were significantly associated with HbA1C across
time [F (1, 537)¼11.55, p<0.001], with weaker
parental problem-solving skills related to poorer glycemic
control in youths (b¼?0.25).
Figure 2 shows the corresponding results for youths
in the Low, Medium, and High DPSI groups. In contrast
to the findings for caregivers, youths with High DPSI
scores appeared to have slightly higher mean HbA1C
(range 8.8–9.0%) during the study than those in the
Medium and Low DPSI groups. However, the child DPSI
scores were not statistically associated with HbA1C
levels across time [F (1, 547)¼0.41, p¼0.52). Further
analysis showed that this latter finding was evident for
both the younger and older halves of the sample and that
these sub-groups did not differ significantly from each
Wysocki et al.
This article introduced the DPSI and provided an
evaluation of several psychometric properties of this
measure. These contributions begin to fill a gap in the
available measurement tools that have been validated for
use in this clinical population by providing a method of
quantifying skills of youths with T1DM and their adult
caregivers in responding to unwanted glycemic fluctua-
tions. To the extent that these skills may represent a key
mediator of behavioral effects on diabetes outcomes, the
DPSI provides a valuable tool for further research on
these relationships. The measure yielded approximately
normal score distributions for both caregivers and youths,
and demonstrated sufficient variability to enable analyses
of the measure’s statistical associations with other
variables of interest. The reliability of the measure, as
assessed by indices of internal consistency, item-total
correlations, inter-rater agreement, and test–retest relia-
bility, was marginally acceptable. Since the DPSI items
targeting correction of hypoglycemia and hyperglycemia
focus on distinct diabetes management skills, it is not
consistency and item-total correlations were not particu-
larly high. Evidence of the validity of the DPSI was
provided by significant correlations between youths’ age
and DPSI scores, between caregivers’ DPSI scores and
concurrent HbA1Cmeasurements, and, with youths’ age
controlled, scores on the DFRQ.
While the analyses that were performed yielded only
modest support for the psychometric properties of this
structured interview, the mixed effects modeling analyses
that were completed suggest that this is a promising
direction for psychological assessment in pediatric T1DM
and that further refinement of the instrument is certainly
analysis showed that low DPSI scores among caregivers
were particularly predictive of poor glycemic control over
the ensuing 9 months. Thus, families in which caregivers
lack sufficient skill for responding to and managing blood
glucose fluctuations may be at special risk for unaccep-
table diabetes outcomes
quite durable over time. This observation suggests that
there may be a threshold for problem-solving skills in
parents necessary for adequate diabetes management.
Consequently, efforts to identify caregivers with deficient
diabetes problem-solving skills and to provide them with
targeted education may be particularly beneficial in terms
of ultimate effects on their children’s diabetes outcomes.
The corresponding relationship between youths’
DPSI scores and HbA1Clevels revealed minimal evidence
of associations similar to those found with caregivers.
Youths’ diabetes problem-solving skills were related to
their current HbA1Clevels only after the contributions of
pertinent parental behaviors were accounted for statisti-
unrelated to their subsequent levels of glycemic control.
On the surface, this effect would seem to be somewhat
counter-intuitive. However, the psychological and educa-
tional research literature on pediatric diabetes is replete
with reports of no relationship between diabetes knowl-
edge orskillsand measures
or treatment adherence (see Johnson, 1984, 1995
Figure 1. Youths’ mean (?1 SEM) HbA1C(%) and caregivers’ mean
scores on the DPSI.
Figure 2. Youths’ mean (?1 SEM) HbA1C(%) and youths’ mean
scores on the DPSI.
Diabetes Problem Solving
for reviews). There are several plausible explanations for
why caregivers’ diabetes problem-solving skills would be
more strongly associated with diabetes outcomes than
would youths’ skills.
Thomas, Peterson, and Goldstein (1997) reported
that, although older youths demonstrated more sophisti-
cated diabetes problem-solving skills in social situations,
compared with younger children they were more likely to
avoid utilization of their diabetes problem-solving skills in
favor of behaviors that are perceived by them as more
likely to yield peer affiliation and acceptance. Thus,
adolescents who face social dilemmas pitting optimal
diabetes management against peer affiliation and accep-
tance will tend to behave in accord with the latter
priority. Similarly, Wysocki, Hough, Ward, Allen, and
Murgai (1992) found that active use of self-monitored
blood glucose data for treatment decisions was associated
significantly with parental diabetes knowledge, but not
with youth knowledge. A second possible explanation is
that most youths in this age range may continue to rely
heavily on parental involvement in decision making
regarding treatment adjustments in response to blood
glucose monitoring results. If youths do rely more heavily
on their caregivers’ diabetes problem-solving skills than
on their own skills, it is reasonable to expect that youths’
skills will account for minimal variance in diabetes
outcomes and that youths whose caregivers have deficient
skills will tend to struggle with diabetes management.
Another possible explanation for this pattern of findings
is that, since caregivers’ DPSI scores were significantly
higher than youths’ scores, it is possible that few youths
had sufficiently well-developed diabetes problem-solving
skills to equip them to make active, appropriate treatment
decisions in a timely manner without parental support or
guidance. As with caregivers, there may be a minimum
threshold for problem-solving skills to be effective. Youths
with DPSI scores in the upper tertile had a mean score of
7.66, which overlaps that achieved by caregivers in the
middle (M¼6.86) and highest (M¼8.43) tertiles of the
caregiver distribution. Since children of these caregivers
achieved similar, better HbA1Clevels compared to those
in the lowest tertile, it seems implausible that similar
DPSI scores obtained by youths would not also equip
them to maintain similar levels of glycemic control.
Another possible explanation might be that youths
with extremely stable glycemic control may have fewer
opportunities to engage in and practice problem solving
than those with less stable glycemic control. If true,
this could dilute a possible association between youths’
problem-solving skills and indices of glycemic control.
While all of these possible interpretations of our
findings are interesting and plausible, it remains for
future research todetermine
The present study has a number of limitations that
should be taken into account when interpreting these
results. Foremost among these is that several psycho-
metric properties of the DPSI proved to be rather
marginal. While the present findings reveal some promise
for a measure of this type, further refinement of the
measure appears warranted, perhaps including a more
extensive collection of diabetes vignettes and empirically
driven retention of those that prove to be most strongly
associated with diabetes management behaviors and out-
comes. The present study evaluated diabetes problem-
solving skills in children as young as 9 years of age,
but perhaps the findings suggest that these skills do
Supplementation of vocal presentation of vignettes with
visual aids could possibly enhance youths’ comprehen-
sion of the diabetes problems. Finally, it is possible that a
revised scoring system that enables more fine-grained
quantification of problem-solving skills could result in a
measure that is more consistently associated with other
The primary clinical implication of the findings
reported here is that youths with T1DM from families
in which the primary diabetes caregiver has deficient
diabetes problem-solving skills may be at elevated risk of
poor glycemic control. If confirmed by further research,
this observation implies that active efforts to identify
these caregivers may be fruitful if these families can either
be provided with effective remedial education targeting
these specific skills or provided with additional consulta-
tion and support to enable them to compensate for these
skill deficiencies. An additional clinical implication of the
present findings derives from the suggestion that youths
may not adequately utilize the diabetes problem-solving
skills they have acquired. The absence of an association
between youths’ DPSI scores and either their measured
adherence (DSMP scores) or glycemic control (HbA1C)
suggests that interventions that promote youths’ utiliza-
tion of problem solving skills in either naturalistic or
realistically simulated circumstances (Gross, Heimann,
Schimmel, 1985) may be particularly valuable. Taken as
a whole, the present findings support the targeting of
diabetes problem-solving skills in behavioral and psycho-
logical interventions that seek to facilitate effective family
management of pediatric T1DM.
ifany are valid
Wysocki et al.
This research was supported by the intramural research
program of the National Institutes of Health, National
Institute of Child Health and Human Development. The
authors would like to acknowledge the assistance of
Madiha Tahseen and Lauren Gase in development of the
coding system and coding of the vignettes.
Conflicts of interest: None declared.
Received December 20, 2007; revisions received February
28, 2008; accepted March 2, 2008
The following institutions and investigators comprised the
steering committee of the Family Management of Diabetes
National Institute of Child Health and Human
Morton, EdD, Tonja R. Nansel, PhD, and Ronald J.
Joslin Diabetes Center, Boston, Massachusetts: Lori
Laffel, MD MPH, Korey Hood, PhD. Contract N01-HD-
Nemours Children’s Clinic, Jacksonville, Florida: Tim
Wysocki, PhD, Amanda Lochrie, PhD. Contract N01-HD-
Texas Children’s Hospital, Houston, Texas: Barbara
Anderson, PhD. Contract N01-HD-4-3362.
Children’s Memorial Hospital, Chicago, Illinois: Jill
James Bell Associates, Arlington, Virginia; Cheryl
McDonnell, PhD, MaryAnn D’Elio, MS. Contract N01-
More detailed psychometric data are available from
the first authors upon request.
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.
Bagner, D. M., Williams, L. B., Geffken, G. R.,
Silverstein, J. H., & Storch, E. A. (2007). Type 1
diabetes in youth: The relationship between adher-
ence and executive functioning. Children’s Healthcare,
Bloom, B. (1984). Taxonomy of educational objectives,
Book I: Cognitive domain. White Plains, NY:
Chase, H. P. (2006). Understanding diabetes (11th ed.).
Denver, Colorado: Children’s Diabetes Foundation
Cook, S., Alkens, J. E., Berry, C. A., & McNabb, W. L.
(2001). Development of the diabetes problem-solving
measure for adolescents. The Diabetes Educator, 27,
Diabetes Control and Complications Trial Research
Group. (1993). The role of the trial coordinator in
the Diabetes Control and Complications Trial. The
Diabetes Educator, 19, 509–512.
Diabetes Control and Complications Trial Research
Group. (1994). Effect of intensive treatment on
the development and progression of long term
complications in adolescents with insulin-dependent
diabetes mellitus. Journal of Pediatrics, 125, 177–188.
Diabetes Research in Children Network (DirecNet) Study
Group. (2005). Diabetes self management profile for
flexible insulin regimens: Cross-sectional and long-
itudinal analysis of psychometric properties in
a pediatric sample. Diabetes Care, 28(8), 2034–2035.
Eastman, B. G., Johnson, S. B., Silverstein, J.,
Spillar, R. P., & McCallum, M. (1983).
Understanding of hypo- and hyperglycemia by
youngsters with diabetes and their parents. Journal of
Pediatric Psychology, 8, 229–243.
Gross, A. M., Heimann, L., Shapiro, R., & Schultz, R. M.
(1983). Children with diabetes: Social skills training
and HbA1Clevels. Behavior Modification, 7, 151–163.
Harkavy, J., Johnson, S. B., Silverstein, J., Spillar, R.,
McCallum, M., & Rosenbloom, A. (1983). Who
learns what at diabetes summer camp? Journal of
Pediatric Psychology, 8, 143–153.
Harris, M. A., Wysocki, T., Sadler, M., Wilkinson, K.,
Harvey, L. M., Buckloh, L., et al. (2000). Validation of a
structured interview for the assessment of diabetes self
management. Diabetes Care, 23, 1301–1304.
Hedeker, D., & Gibbons, R. D. (2006). Longitudinal
data analysis. New Jersey: John Wiley & Sons, Inc.
Heidgerken, A. D., Merlo, L., Williams, L. B.,
Lewin, A. B., Gelfand, K., Malasanos, T., et al.
(2007). Diabetes awareness and reasoning test:
A preliminary analysis of development and psycho-
metrics. Children’s Healthcare, 36, 117–136.
Hershey, T., Lillie, R., Sadler, M., & White, N. H. (2002).
The effect of severe hypoglycemia on cognition in
Diabetes Problem Solving
children and adolescents with type 1 diabetes Download full-text
mellitus. Diabetes Nutrition and Metabolism, 15,
Johnson, S. B. (1984). Knowledge, attitudes and
behavior: Correlates of health in childhood diabetes.
Clinical Psychology Review, 5, 545–564.
Johnson, S. B. (1995). Insulin-dependent diabetes
mellitus in childhood. In M. C. Roberts (Ed.),
Handbook of pediatric psychology (pp. 263–285). New
York: Guilford Press.
Johnson, S. B., Pollack, R. T., Silverstein, J.,
Rosenbloom, A., Spillar, R., McCallum, M., et al.
(1982). Cognitive and behavioral knowledge about
insulin-dependent diabetes among children and
parents. Pediatrics, 69, 708–713.
Kaplan, R. M., Chadwick, M. W., & Schimmel, L. E.
(1985). Social learning intervention to improve
metabolic control in Type 1 diabetes mellitus: Pilot
experiment results. Diabetes Care, 8, 152–155.
Kuttner, M., Delamater, A. M., & Santiago, J. V. (1990).
Learned helplessness in diabetic youth. Journal of
Pediatric Psychology, 15, 581–594.
LaGreca, A. M., Follansbee, D. M., & Skyler, J. S. (1990).
Developmental and behavioral aspects of diabetes
management in youngsters. Children’s Health Care,
Mensing, C., Boucher, J., Cypress, M., Weinger, K.,
Mulcahy, K., Barta, P., et al. (2005). National standards
for diabetes self-management education. Diabetes Care,
28(Suppl 1), S72–S79.
Nansel, T., Anderson, B. J., Laffel, L., Simons-Morton, B.,
Weissberg-Benchell, J., Wysocki, T., et al. (manu-
script under review). A multi-site trial of a clinic-
integrated intervention for promoting family man-
agement of childhood type 1 diabetes: Feasibility and
Thomas, A. M., Peterson, L., & Goldstein, D. (1997).
Problem solving and diabetes regimen adherence by
children and adolescents with IDDM in social
pressure situations: A reflection of normal develop-
ment. Journal of Pediatric Psychology, 22, 541–561.
Wysocki, T. (2000). Effective utilization of self-monitored
blood glucose data: Cognitive and behavioral pre-
requisites. In B. Bercu, & B. T. Stabler (Eds.),
Symposium on therapeutic outcome of endocrine
disorders: Efficacy, innovation and quality of life
(pp. 151–160). New York: Springer.
Wysocki, T., Hough, B., Ward, K. M., Allen, A.,
& Murgai, N. (1992). Use of blood glucose data by
families of children and adolescents with IDDM.
Diabetes Care, 15(8), 1041–1044.
Wysocki, T., Taylor, A., Hough, B. S., Linscheid, T. R.,
Yeates, K. O., & Naglieri, J. A. (1996). Deviation
from developmentally appropriate self-care auton-
omy: Association with diabetes outcomes. Diabetes
Care, 19(2), 119–125.
Wysocki et al.