Health-related quality of life in overweight German children and adolescents: do treatment-seeking youth have lower quality of life levels? Comparison of a clinical sample with the general population using a multilevel model approach

A clinical sample of overweight youth participating in the ‘Obeldicks light’ intervention study was compared with a population sample of similar age from the nationally representative German Health Interview and Examination Survey (KiGGS) [26]. Comparisons between these samples were possible because identical instruments were used to collect the data of interest.

Table 1 describes both samples in terms of sociodemographic and anthropometric measures, as well as perceived weight status. There were no differences between the samples with regard to socioeconomic status or proportion of mothers as proxy-raters. However, the proportions of females and adolescents were higher in the treatment sample than in the population sample, while the proportion of immigrants was significantly lower. The treatment sample was also significantly younger on average, and their parents self-reported significantly higher BMIs. As expected, patients had a higher BMI z-score than the general population, and all were overweight. With nearly all of the patients’ weight rated as ‘(far) too fat’, the perceived weight also significantly differed between the samples. Because there was almost no variation in this group perceived weight was not differentiated within the treatment sample.

The mean HRQoL scores of girls and boys in the compared groups and the entire KiGGS sample are displayed in Table 2.

Worth remarking, HRQoL differences between overweight and normal-weight children within the perceived weight categories seemed rather small. Furthermore, the differences between obese and overweight youth who were rated as ‘too fat’ were generally small in relation to the differences between perceived weight categories. These patterns were similar between boys and girls.

Table 3 shows the results of the final three-level model of parent-reported HRQoL scores. The main results are the fixed effects for the group comparison of patients with objective/subjective weight groups from the population sample. The detailed interpretation of the results of Tables 3 and 4 is described in Additional file 1.

BMI z-score of the child was tested to control for differences between groups of the same weight category but showed no additional predictive value when added as covariate. Because the parental BMI had no significant effect on any of the HRQoL subscales, it was also not included in the final model. The only significant two-way interactions were found for child gender by proxy-rater on two subscales. With regard to age, non-linear effects were additionally explored but were found to be insignificant.

In general, mothers reported lower HRQoL scores for the subscales of physical well-being, family, and friends but higher scores for girls’ school-functioning than other proxies, and HRQoL scores varied with sociodemographic variables.

There were significant group effects on all HRQoL subscales. Various comparisons of the treatment sample with the general population demonstrated lower scores for the patients. The patients had a significantly lower physical and emotional well-being than normal-weight and overweight children rated ‘proper weight’ by parents, but they had similar values to those of normal-weight and other overweight children rated as ‘too fat’. With regard to the self-esteem and friends scales, patients scored significantly lower than all other groups, and the largest differences were observed between patients and children with a parent-perceived ‘proper weight’. Differences in the family and school scores were less pronounced between groups, and the only group from which the patients differed was the group with the highest HRQoL scores, namely the children with objective as well as perceived normal-weight.

The effect sizes for significant comparisons of the patients with other overweight youth rated ‘too fat’ were small (d=0.31-0.34), while small to moderate effect sizes (d=0.20-0.65) were found for comparisons with normal-weight youth rated as having a ‘proper weight’.

The deviance of the described final model was significantly lower than that of the null model (χ² _(df=62)=1712.303, p<0.001) and of a model without the group predictors (χ² _(df=31)=411.177, p<0.001).

When compared with overweight youth in the general population (independent of perceived weight), the patients exhibited significantly lower scores for parent-rated emotional well-being, self-esteem, and friends (see Figure 2). The parent-rated HRQoL of patients appeared to be impaired in all dimensions when compared with the HRQoL of normal-weight youth. However, compared with children and adolescents in the general population rated as ‘too fat’ (independent of objective weight), patients were only rated lower on the self-esteem and friends subscales.

The results of the hierarchical model of self-reported HRQoL in adolescents (11–16 years of age) are shown in Table 4. As in the parents’ model, BMI z-score showed no additional predictive value and was, therefore, not included in the final model.

There were significant group interaction effects for gender on physical well-being and for immigrant status on emotional well-being. All other interactions and non-linear age effects were not significant. Immigrants within the treatment sample self-reported markedly higher emotional well-being than did immigrants of all comparison groups, whereas no such group differences were observed in youth of German descent. However, this interaction was only apparent in comparisons with the treatment sample, and the number of immigrants in this sample (N=8) seems too small to draw general conclusions from this result. We therefore decided not to include this interaction effect in the final model.

For physical well-being, male patients reported significantly lower scores than all of the general population groups, while female patients had similar scores to those of females perceived as having a ‘proper weight’ and higher scores than other girls who perceived themselves as ‘too fat’ (see Table 2 for descriptive means; the same trend was found for emotional well-being but did not reach statistical significance). Emotional well-being scores were generally lower in the groups that felt ‘too fat’, with patients reporting similar values to those of other groups that felt ‘too fat’. The patient scores only significantly differed from those of the overweight/proper weight group.

With regard to school and familial well-being, patients differed mainly from the groups who felt they were having a ‘proper weight’, with patients reporting a lower HRQoL. Patients had significantly lower scores on the friends scale than all of the other groups, with the exception of the overweight/too fat (p<0.10) and obese/too fat (p>0.10) groups. Contrary to the parent-reported results, treatment-seeking youth had better self-esteem scores than did other youth in the population sample who felt ‘too fat’; in contrast, no difference was observed when comparing patients to groups who felt they were of the ‘proper weight’.

In general, the significant effects were of low to moderate size (d=0.22-0.54), although large effects were found for the comparison between male patients and males in the population sample who felt they were at the ‘proper weight’ (d=0.81 and 0.95, respectively).

The deviance of the described model was significantly lower than that of the null model (χ² _(df=56)=1247.40, p<0.001) and of a model without group predictors (χ² _(df=35)=316.43, p<0.001).

Additional comparisons of patients with populations groups in terms of objective or subjective weight are illustrated in Figure 3. They revealed that treatment-seeking youth, contrary to their parents, rated their self-esteem significantly higher than overweight adolescents in the general population. In terms of physical well-being, treatment-seeking boys reported lower scores than normal-weight and overweight boys, while treatment-seeking girls reported significantly higher scores than other overweight girls. Male patients also rated their physical well-being significantly lower than other adolescents who felt ‘too fat’, while girls rated it higher. Both genders reported a higher self-esteem in the treatment-seeking sample than in the general population group with a perceived excess weight. The only dimension that was different between patients of both genders and other youth who felt ‘too fat’ was ‘friends’. The scores for the friends dimension were also significantly reduced in patients compared with the normal-weight youth, while compared with overweight youth the difference did not reach significance (p<0.10).

For conclusions about HRQoL differences due to treatment-seeking status patients should be compared with otherwise similar populations. In our overweight treatment-seeking sample, perceived weight was rated as ‘(much) too fat’ for nearly all participants. Compared with the group of overweight youth in the general population rated ‘too fat’, controlling for sociodemographic differences, treatment-seeking status in our sample was associated with impairments in only some HRQoL domains. Parents and adolescents differed in these perceived HRQoL limitations, and in some HRQoL domains patterns of results differed between treatment-seeking girls and boys.

Other studies that compared overweight/obese clinical samples with overweight/obese youth in the general population found more marked impairments in HRQoL than those observed here either using weight-specific HRQoL instruments [16, 17] or, in extremely obese patients, using non-weight-specific HRQoL measures [18]. In another study using the generic PedsQL instrument, in contrast, HRQoL differences were not significant between a clinical sample and a community sample of obese children and adolescents after controlling for gender, age, and BMI z-score [14].

However, clinical samples were mainly compared with normal-weight controls. In our study differences between the treatment sample and the general population were generally higher for comparisons of patients with normal-weight youth than for comparisons between the treatment sample and overweight youth. Hughes et al. [39] found very similar results to our findings in obese treatment-seeking children compared with healthy controls with significant impairments in all HRQoL domains, while self-reported scores were less impaired and school and emotional well-being were the least affected. In another study on overweight and obese patients seeking outpatient training [40], self-reported HRQoL scores were only reduced in the ‘friends’ dimension compared with normal-weight students. On the other hand, a study with severely obese patients [41] found more pronounced HRQoL differences compared with normal-weight controls with large significant effects. Knöpfli et al. [42] also demonstrated significant impairments in all studied HRQoL dimensions in severely obese patients, but a direct comparison with a healthy sample was missing.

It may be inferred that treatment-seeking status in overweight or obese children and adolescents is associated with impaired HRQoL, but associations seem more manifest in obese than in overweight samples and may to a great extent be attributable to higher or extreme excess of weight in clinical samples. Furthermore, the perceived weight or body satisfaction of comparison groups was seldom taken into account.

Our results show that HRQoL differences between the patients and the general population were mainly due to differences between patients and other youth rated as ‘proper weight’. In other words, a large portion of HRQoL impairments were due to a perceived, but not objective, excess in weight.

We did not find any other study that compared a clinical sample with population groups differentiated according to perceived weight. However, our result that HRQoL impairments in overweight youth are mainly due to perceived instead of objective excess in weight supports the findings of Kurth and Ellert [22] regarding self-reported HRQoL of obese German youth. Correspondingly, a large U.S. study [23] found that the effects of obesity on psychological well-being were completely mediated by body dissatisfaction. Similar results were reported for female adolescents [25], where differences in psychosocial variables according to weight status lost significance after controlling for body satisfaction. In Dutch adolescents [24], the associations were found to be reversed, with obese adolescents reporting better psychological well-being than normal-weight counterparts after controlling for perceived weight.

Previous studies found that physical well-being was usually significantly impaired in clinical samples, but most previous studies utilized obese samples. Our result that treatment-seeking status was only associated with lower physical well-being in boys, while girls had even higher scores than the general population, is a novel result that merits additional examination. These differences may especially concern moderately overweight youth who have less adverse physical impacts than individuals with greater excess weight who, therefore, have a higher risk of compromised physical well-being. It can be suspected that high physical well-being in overweight girls may encourage them to initiate life-style modifications that include regular exercising, whereas our results indicate that low physical well-being in boys may be an important constituent of the motivation to initiate weight loss treatment.

To our knowledge, our study is also the first to find higher self-reported self-esteem scores in treatment-seeking children and adolescents than in the general population. Although the decision to seek treatment in this age group presumably is considerably determined by the parents, and parent-perceived low self-esteem seems to be conducive to it, a child’s high self-esteem might encourage this decision. Another reason for higher self-esteem scores in treatment-seekers might be that more involved parenting fosters both children’s self-esteem and parents’ readiness to participate in a treatment program that requires commitment by parents. It was also supposed that overweight youth may hesitate to acknowledge negative impacts resulting from their weight [10].

Among the social HRQoL domains, ‘friends’ was the domain that was most affected by treatment-seeking status in our sample, whereas familial well-being was less consistently reduced. Other studies reported significant impairments in the PedsQL social dimension in treatment-seekers [18, 39, 41, 43] or in the overweight/obese in general [10, 13, 14], while studies that distinguished between impairments in the domains of family and friends mainly found impairments in the friends but not family domain in treatment-seeking youth [40].

Impairments in the friends dimension are likely due to negative stereotypes and stigmatization of overweight children and adolescents by peers [5]. Warschburger [7] suggested that, as a result of such peer stereotypes and stigmatization, obese children have fewer opportunities to develop social competence and supportive relationships, and thus, social well-being is negatively affected.

In summary, a universal pattern of impairments associated with treatment-seeking status in overweight youth cannot be deduced due to the limited number of studies available to date. However, impaired social well-being seems to be consistently affected in treatment-seekers independent of degree of overweight or source of information (self- vs. proxy-report).

We found greater proxy-reported HRQoL impairments than self-reported impairments in this study, and this is in line with other studies on treatment-seeking youth [14, 39, 41]. Tsiros and colleagues [10] concluded that parents in general report lower HRQoL in overweight/obese adolescents than the adolescents themselves. Contrary to their result, we did not find proxy-reported scores to be consistently lower than self-reported scores in overweight or obese adolescents, but this was partly true for the treatment sample. This larger association of treatment-seeking status with parent-reported values can be explained by the vital role that parents play in health decision-making in this age group [40, 44].

Concerning HRQoL associations with sociodemographic variables, our results are in accordance with those reported in the literature [10, 37].

In general, associations between treatment-seeking status and weight were very similar for both genders and did not differ according to SES. For self-reported physical well-being, however, a significant interaction with gender was encountered. Thus, the decision to seek treatment for weight reduction appears to be partially motivated by different factors between boys and girls.

The effect of ethnic background on HRQoL has not been well studied. Other studies that considered ethnic background in comparisons with clinical samples reported conflicting results [18, 39, 45, 46]. The number of immigrants in our sample was too small to generalize our finding of higher emotional well-being in treatment-seeking immigrants; however, because differential effects have been found in the past, it may be assumed that treatment-seeking status is associated with specific HRQoL patterns in ethnic minorities. Minority youth may have specific needs that might differ in particular ethnic groups, and should be further examined by future research.

Some strengths of our study should be highlighted. In contrast to previous studies, we focused on moderately overweight but not obese children and adolescents and included the perspectives of both adolescents and parents with regard to HRQoL. Because interventions for this target group are still missing, knowledge is limited. This group is particularly important because of the advantages in preventing rather than treating severe excess in weight. A further strength is the large representative sample that we were able to use for our comparison. To our knowledge, this is the first study to include not only objective but also subjective weight classifications in comparisons of a treatment sample with the general population. Finally, employing a multilevel approach to the data analysis, we allowed for clustering effects of the population sample and used all of the available data without having to exclude cases with single missing values.

However, some limitations should also be considered. First, the clinical sample was relatively small (especially for self-reports by adolescents) and consisted of participants of one treatment program in one area of Germany. Even if we controlled for deviations in sociodemographic characteristics, further unknown differences in the characteristics of the sample group may have influenced the results. The generalizability of our findings is therefore indeterminate, particularly in terms of whether our results can be generalized to treatment-seeking immigrants. Unequal group sizes are also a statistical concern, although inevitable when looking at naturally occurring groups in survey research. However, the employed hierarchical regression analysis approach allows for unequal group sizes. The reliability of some HRQoL subscales was unsatisfactory (especially ‘friends’), which may have reduced the accuracy of results. HRQoL scores were also not perfectly normally distributed. However, based on the computed robust standard errors and re-evaluation of results using transformed values, the distribution of scores should not have had considerable effects on our results. Finally, the measure of perceived weight status was rather crude. With a more sophisticated measure of perceived weight or body image, additional insights into HRQoL effects of subjective weight perceptions should be investigated in future studies.

Our results have some practical implications for interventions addressing overweight children and adolescents.

In terms of treatment motivation for both parents and youth, social well-being (such as the anticipation of making friends in a training program or better acceptance by peers after losing weight) seems to play a crucial role. This motivation should therefore be adequately addressed by interventions.

Concerning gender differences in our results, especially in patients’ physical well-being, different treatment motives in girls and boys should be considered by gender-specific intervention content and marketing of interventions for overweight youth. Potentially varying needs of ethnic minority youth should also be taken into consideration when planning interventions, and this may involve preparatory exploration of targeted minority groups.

Because of the crucial influence parents may have on treatment decisions for their offspring, professionals planning or implementing treatments targeted at overweight children and adolescents should also take seriously the parent perceived well-being and problems of the child and, therefore, both perspectives should be assessed if possible [44]. Even if an intervention primarily aims at a better subjective well-being of the child, responding to issues that only parents perceive as problematic may be important to assure compliance with treatment and parental assistance in achieving life-style changes for the child.

The fact that perceived weight or body satisfaction seems to impact HRQoL more than objective weight status suggests that more attention should be paid to body self-perceptions in prevention and treatment programs for overweight in children and adolescents [23]. Health professionals should be aware of possible negative consequences for well-being, when informing overweight children and their families about excess weight of the child, while at the same time awareness of overweight may be required to facilitate treatment decisions. A dilemma arises as to how overweight youth can be engaged for interventions without further impairing their well-being [22]. On the one hand, it seems preferable to prevent further weight gain in moderately overweight youth by early treatment. On the other hand, in this target group impaired well-being due to perceived heavy weight actually may be the worst immediate consequence because manifest symptoms are rather unlikely in modestly overweight youngsters. It might be advantageous, therefore, to concentrate on counseling the parents first about overweight of the child and about possible negative effects of body dissatisfaction, since this may bring about treatment motivation without affecting the well-being of the child too much. Furthermore, programs for overweight children should include this topic and ensure that, besides weight reduction, self-acceptance and HRQoL are adequately addressed.