Health-Related Quality of Life and Metabolic Control in Children With Type 1 Diabetes: A prospective cohort study

The methodology of the baseline study has been described in detail elsewhere (1). Briefly, the baseline sample was representative of 5- to 18-year-old children attending the Diabetes Outpatient Clinic at the Royal Children’s Hospital in Melbourne, Australia (population 3.4 million), between May and August 1998. Parents of all children completed questionnaires written in English in the clinic waiting room; in addition, children aged 12–18 years were invited to participate in the self-report arm of the study. One hundred forty-one children aged 5.4–18.9 years (12.3 ± 3.6 [mean ± SD]) participated in the baseline study. Of these, 70 had parent-proxy reports only, 58 had both parent-proxy reports and adolescent self-reports, and 13 had adolescent self-reports only (90% parent response and 92% adolescent response).

The follow-up study was conducted between November 1999 and April 2001. Of the original 141 subjects, 24 were ineligible for the follow-up study (13 were aged ≥19 years and a further 11 were no longer attending the Royal Children’s Hospital Diabetes Clinic). Parents of the remaining 117 subjects were invited to participate in the follow-up study, as were the 53 eligible adolescents who had provided self-reports in the baseline study. Most questionnaires were completed in the waiting area before scheduled appointments, with the remainder posted to the children’s homes and returned directly to the researchers.

Both the baseline and follow-up studies received ethics approval from the Royal Children’s Hospital Ethics in Human Research Committee.

At both baseline and follow-up, parents completed the 50-item Australian authorized adaptation of the Child Health Questionnaire (CHQ) PF-50 (12). The CHQ PF-50, which was originally developed in the U.S. (13), is a standardized HRQOL measure with 4 single-item and 11 multi-item subscales assessing the domains of physical health and well-being, psychosocial health and well-being, and the impact of the child’s health on parents and family (Table 1). All items are based on 4-week recall, except the single-item Change-in-Health scale, which inquires about change in the child’s health over the past 12 months and has a possible range of 0 to 4. All multi-item scale scores are transformed to a range of 0 to 100, with lower scores indicating worse health and higher scores indicating better health. In addition, two summary scores (Physical and Psychosocial) are derived from the CHQ PF-50.

Adolescents aged 12–18 years at baseline completed the 80-item self-report version of the Australian authorized adaptation of the CHQ CF-80 at both baseline and follow-up (14). The CHQ CF-80 is analogous to the CHQ PF-50 and derives 10 multi-item and 4 single-item scales but no summary scores (Table 1). In the baseline study, both the CHQ PF-50 and the CHQ CF-80 demonstrated good psychometric properties (1).

At each clinical visit, HbA_1c levels are assessed using the Bayer DCA 2000 immunoagglutination method (Calabria, Barcelona, Spain). HbA_1c levels are recorded as a percentage of total hemoglobin. The mean HbA_1c level for the baseline study sample was 8.1%, which was somewhat lower than the mean levels of 8.6% reported in other Australian youth samples (15,16) and in an amalgam of 18 international pediatric diabetes centers (17). In the baseline study, we used the sample’s 75th percentile (HbA_1c level 8.8%) to divide the 25% of children with the poorest control from the other children in the sample (1). The 75th percentile represents a practical cut point that allows for sufficient numbers of children in both groups to perform statistical comparisons. As clear international criteria still remain to be established, we again used an HbA_1c level >8.8% to define children with poor metabolic control in this follow-up study.

CHQ PF-50 scale and summary scores and CHQ CF-80 scale scores were calculated according to the manual (13). Independent sample t tests were used to compare characteristics of children retained and those lost to follow-up, mean CHQ scores for children with HbA_1c levels >8.8% and ≤8.8% at follow-up, and mean change in HbA_1c for children with baseline HbA_1c levels >8.8% and ≤8.8%. Paired sample t tests were used to compare mean baseline and follow-up CHQ scores and HbA_1c levels. Pearson correlation coefficients were calculated for baseline and follow-up CHQ scores and HbA_1c levels. McNemar’s χ² statistic was used to compare the proportions of children with HbA_1c levels >8.8% at baseline and follow-up. Univariate linear regression models assessed whether HbA_1c at follow-up or change in HbA_1c could be predicted from HbA_1c at baseline or from CHQ scores (adjusting for baseline HbA_1c).

Of the 117 eligible subjects, 83 consented to participate in the follow-up study (71% response rate; 55% male children). Seventy-six subjects (43 male and 33 female; mean age 12.8 years at follow-up) had parent reports at both time points, and 29 (15 male and 14 female; mean age 15.6 years) had adolescent self-reports at both time points. Eleven subjects moved from the 5–11 years to the 12–18 years age-group over the course of the study. On average, subjects who participated in the follow-up study had lower mean HbA_1c levels at baseline than eligible subjects lost to follow-up (7.8 vs. 8.6%; P = 0.006), were younger (11.1 vs. 13.0 years; P = 0.01), and had been diagnosed more recently (3.6 vs. 5.3 years since diagnosis; P = 0.015).

Parent-reported child HRQOL at follow-up was not significantly different from their baseline HRQOL (Table 2). The correlation between baseline and follow-up scores were ≥0.40 on all CHQ scales and summary scores, except the Role/Social Limitations-Emotional/Behavioral (r = 0.12, P = 0.29) and the Role/Social Limitations-Physical scales (r = 0.33, P = 0.004).

Unlike parents, whose reports of child HRQOL were quite stable from baseline to follow-up, some domains of adolescent-reported HRQOL changed over time (Table 3). Small improvements were reported on the Bodily Pain (t = −2.14, P = 0.04) and General Health Perceptions (t = −3.79, P = 0.001) scales, along with a moderate worsening of Behavior scale scores (t = 2.21, P = 0.04). Most strikingly, Family Activities scale scores improved markedly (t = −4.88, P < 0.001) over the 2 years. Baseline and follow-up scores correlated strongly (r ≥ 0.40) for all scales except Physical Functioning (r = 0.34, P = 0.08), Role/Social Limitations-Behavioral (r = −0.01, P = 0.97), and Family Activities (r = 0.27, P = 0.17).

Mean HbA_1c levels for the 83 children rose significantly over the 2 years from baseline to follow-up (7.8 vs. 8.5%; t = 6.42, P < 0.001), with a mean rise in HbA_1c of 0.75 ± 1.1 (mean ± SD; range −2.4 to 2.6) and a strong correlation between baseline and follow-up HbA_1c levels (r = 0.57). Nearly twice as many children had HbA_1c levels >8.8% at follow-up as at baseline (41 vs. 22%; P < 0.001) (Table 4). Whereas mean HbA_1c increased by 0.97% in children whose HbA_1c was initially ≤8.8%, it fell only slightly (−0.36%) in children with initially high baseline HbA_1c (P < 0.001). Lower HbA_1c levels at baseline therefore predicted increases in HbA_1c levels (r² = 0.25, P < 0.001), accounting for 25% of the variance in HbA_1c change between baseline and follow-up.

As in the baseline study, children with HbA_1c >8.8% (n = 31) had significantly poorer Psychosocial Summary scores than children with HbA_1c ≤8.8% (n = 44) (mean 46.1 vs. 50.9, P = 0.03). They therefore also scored more poorly on most of the psychosocially related subscale scores: Behavior (mean 66.4 vs. 75.7; P = 0.02), Mental Health (mean 74.7 vs. 81.7; P = 0.02), Family Activities (mean 68.9 vs. 78.6; P = 0.04), and Family Cohesion (mean 67.5 vs. 77.5; P = 0.04). None of the adolescent self-reported scales indicated differences in health according to HbA_1c levels >8.8 or ≤8.8%. However, neither the baseline Psychosocial Summary score nor any of its subscales predicted HbA_1c at follow-up or change in HbA_1c between baseline and follow-up.

Again replicating the baseline study, Physical Summary scores did not differ significantly between children with HbA_1c >8.8% and those with HbA_1c ≤8.8% (mean 48.5 vs. 50.2, P = 0.45). However, the baseline Physical Summary score predicted both follow-up HbA_1c level (β = −0.04, 95% CI: −0.06 to −0.01) and change in HbA_1c (−0.04, −0.06 to −0.01) after accounting for baseline HbA_1c. In other words, on average every 10 additional points on the baseline Physical Summary score equated to a 0.4% decrease in HbA_1c at follow-up. Similarly, adolescent self-reports on the Physical Functioning scale at baseline predicted follow-up HbA_1c and change in HbA_1c (−0.07, −0.13 to −0.01 and −0.07, −0.13 to −0.01, respectively) after accounting for baseline HbA_1c level.

Parents of children and adolescents with diabetes consistently report the health of their children as considerably poorer than the health of children and adolescents in the wider community (1), and this is remarkably stable over a 2-year period. In contrast, adolescents with diabetes consistently report that their health is very similar to adolescents without diabetes (1), but unlike their parents they reported that, with time, their health was having less impact on the family’s activities. Poorer parent-reported psychosocial health was consistently associated with poorer metabolic control at both time points.

Although poorer parent-reported psychosocial health and metabolic control seemed linked at any given time, poorer baseline psychosocial health as reported by parents did not predict a deterioration in metabolic control over the 2 years of the study. Somewhat surprisingly, it was baseline physical health (as reported by both parents and adolescents) that predicted changes in metabolic control over time. This might indicate that poorer physical health is a risk factor for deteriorating metabolic control. Alternatively, better physical health might actively protect against the usual deterioration in metabolic control seen over these years (18); perhaps better functioning in the domain of physical health enables children to participate more fully in the range of social and other activities open to them and thus protects against deteriorations in both metabolic control and psychosocial well-being. Interestingly, while the relationship between psychosocial health and concurrent metabolic control was only apparent in parent reports, the longitudinal relationship between physical well-being and later metabolic control held for both parent and adolescent reports—this was the only finding that was consistent for both proxy and self-reports.

Parents and young people often diverge in their reports of children’s health (19), even when the child is relatively healthy (20). The disparity between parent and adolescent reports of the health of children and adolescents with diabetes corresponds with similar trends seen for children with cancer (21). However, no previous studies have considered changes in HRQOL over time as viewed by parents and adolescents. Unlike the proxy views of parents, adolescents’ views of their own HRQOL changed somewhat over time. Adolescents may be more aware of, and thus able to report, small changes in their HRQOL that are less apparent to their parents. This highlights the importance of incorporating child self-report measures, when the age and cognitive ability of the child allows, in the assessment of HRQOL.

Partially by virtue of the inclusion criteria for the follow-up study, those subjects lost to follow-up tended to be older (and therefore less recently diagnosed) and had significantly higher HbA_1c levels at baseline than those children retained in the cohort. This reflects the logistics of the study, which precluded follow-up subjects who were outside the age range of the CHQ (≥19 years) and those who were no longer attending the clinic. While this may limit the generalizability to older children, it should not compromise the internal validity of the findings.

In our earlier research, we showed that the HRQOL of children and adolescents with diabetes is poor compared with the health of children in the wider community and that poorer metabolic control goes hand in hand with poorer psychosocial functioning (1). Furthermore, the CHQ can be readily incorporated into everyday diabetes clinic practice (1) and relevant CHQ scales correlate strongly with childhood behavior problems as measured against a behavioral “gold standard” (7). In this study, we have demonstrated that the parent-reported HRQOL of children and adolescents with diabetes appears relatively stable over a 2-year period and that better baseline physical functioning may protect against the deteriorating metabolic control apparent in the group as a whole. While the use of a disease-specific HRQOL instrument may have detected greater changes in diabetes-specific health issues over time, there are important benefits of looking at changes over time as detected by a generic instrument such as the CHQ. Generic instruments allow investigation of changes in aspects of health important to all children, regardless of the presence or type of health condition, and potentially allow the comparison of such changes between healthy children and children with different health conditions.

As these children make the transition to young adulthood, it is likely that their HRQOL will deteriorate due to acquired psychosocial comorbidities and comorbidities arising from other diabetes complications (22,23). Clearly, the relationship between HRQOL and metabolic control over time is complex and requires further study. Regardless of its predictive value, we believe that HRQOL is worthy of measurement and action in the clinical setting in its own right (7). Our challenge is to find ways of improving the quality of life of these children and to ensure that optimal HRQOL is maintained throughout childhood and adolescence.

The baseline study was funded by a grant from the Victorian Department of Human Services and the follow-up study by a Royal Children’s Hospital, Special Purposes Grant.

We thank Sophie Jerrim for assistance with data collection in the baseline study and Louisa Salmon, Rebecca Gebert, and Emma White for coordinating data collection in the follow-up study.