Weight Status, Quality of Life, and Self-concept in African American, Hispanic, and White Fifth-grade Children
This study examined the association between weight status and quality of life (QOL) in fifth-grade African American, Hispanic, and white children and the potential mediation of this relationship by self-concept. A sample was recruited from fifth-grade public school students in three sites, of whom 599 were African American (40%), Hispanic (34%), or white (26%). During a home interview, physical and psychosocial QOL and global and body-specific self-concept were measured. Measured height and weight were used to calculate BMI. In this sample, 57% were classified by BMI as not overweight, 17%, overweight, and 26%, obese. Although there was no significant interaction between weight classification and race/ethnicity for QOL, obese children reported significantly lower psychosocial but not physical QOL than those classified as not overweight. There was a significant association between BMI (measured continuously) and psychosocial QOL, but only 2% of the variance was accounted for. Both global self-concept and body dissatisfaction independently mediated significant portions of the association between BMI and psychosocial QOL. Being obese in childhood may have negative psychosocial effects.
The prevalence and severity of obesity in children have increased dramatically in the past 20 years. As defined with cutpoints from the International Obesity Task Force (1) for the sex- and age-specific distributions of BMI, classifications applied to children are “not overweight” (BMI < 85th percentile), “overweight” (85th ≤ BMI < 95th percentile), and “obese” (BMI ≥ 95th percentile). In a U.S. national study of children aged 4 to 12 years, 22% of African Americans, 22% of Hispanics, and 12% of non-Hispanic whites were overweight (2).
In children, being obese is associated with increased blood pressure, elevated total cholesterol, abnormal lipoprotein ratios, hyperinsulinemia, and type 2 diabetes (3,4). In addition, obese children are more likely to become obese adults (5,6) and thus may face various long-term complications of obesity. The most prevalent and immediate consequence from obesity in childhood, however, may be its negative impact on quality of life (QOL) (3,7,8). As applied to children, QOL is a comprehensive, multidimensional construct that encompasses physical and psychosocial functioning (9). Physical QOL incorporates the ability to move about and the presence of general ailments (e.g., pains and aches), while psychosocial QOL combines emotional well-being, social integration, and role functioning (e.g., participation in school) (10).
One prominent early study reported that a sample of obese children (BMI > 95th percentile) experienced significantly reduced QOL compared with children in the general population (11). Indeed, their QOL was similar to that of children receiving treatment for cancer. The sample, however, consisted of 106 children from a wide age range (5–18 years), all of whom were referred for treatment at a tertiary care center for being severely obese (BMI M = 34.7). In addition, the sample included a high percentage of Hispanics (60%). These findings have been replicated in a study of a clinic-referred sample of significantly obese (BMI M = 42) African American and white adolescents (12). Although these findings are not necessarily generalizable, they raise the question of whether QOL is associated with weight status among children in general.
Population-based studies in Australia (7,13) have not only suggested there is such a relationship but also that the relationship between BMI and QOL in children is less striking than that found in clinic-referred samples. Generally, overweight children had only slightly lower QOL than not-overweight children, but obese children had much lower QOL than not-overweight children. Yet, the latter difference was less dramatic than the corresponding difference reported for children referred for treatment for obesity in other studies (11,12).
Taking these findings as a whole, at an elevated BMI children appear to be at risk for experiencing a reduced QOL. Still, we have little information on whether this finding applies across racial/ethnic groups in U.S. community samples. In perhaps the only study to examine QOL and weight status across racial or ethnic groups in a general sample of children, a random community sample of 5,530 3–18-year-olds was recruited, which included 46% Hispanic and 54% white (14). Compared to those in other weight classifications, children who were obese had significantly reduced psychosocial but not physical QOL. There was no difference in QOL between Hispanics and whites, but the interaction between weight class and ethnicity was not reported.
As the prevalence of obese children differs among racial/ethnic groups (2), it is important to examine the relationship between weight status and QOL in interaction with race/ethnicity. Moreover, we need to know more precisely the weight levels where QOL is reduced as studies published to date have not shown whether the reduction occurs only at significantly elevated BMI or whether there is a gradient across the range of BMI. Thus far, published studies have compared only weight classifications rather than considering weight status (BMI) as a continuous measure.
In addition, the mechanisms underlying an association between weight status and QOL have not been addressed. One possibility is a negative effect of weight status on children's self-concept. Some researchers have conceived of self-concept as a single, overarching global construct (15); applying this framework, numerous studies have documented a low self-concept in overweight children (16,17,18). Other theories hold that self-concept in children is a multifaceted construct with several unique subcomponents (e.g., scholastic competence, social acceptance, satisfaction with physical appearance) (e.g., 19,20). Within this latter framework, one would hypothesize that overweight and obese children would experience some dissatisfaction with their bodies (21). Even so, a link between self-concept (in either general or specific dimensions) and QOL has, to our knowledge, not been examined empirically in the context of weight status in children. Therefore, we hypothesize that a negative self-concept results in the experience of reduced QOL, and we postulate that self-concept mediates a relationship between weight status and QOL.
Thus, the present study aims to address gaps in knowledge about the relationship of weight status and QOL among fifth graders in the U.S., a group that is just about to enter adolescence. Three hypotheses were examined: (i) obese children experience a lower QOL than not-overweight children regardless of major racial/ethnic group (African Americans, Hispanics, non-Hispanic whites), (ii) there is an inverse relationship between BMI and QOL across the range of BMI in a community sample of children in the U.S., and (iii) the relationship between BMI and QOL is mediated by dissatisfaction with one's body and one's general self-concept.
Methods and Procedures
Data used for this report were collected from the initial cross-sectional study of Healthy Passages, completed from May to September 2003 at three research sites: the University of Alabama at Birmingham, the University of California at Los Angeles/RAND, and the University of Texas, Houston. Institutional review boards at the sites and the Centers for Disease Control and Prevention (CDC), which funded the research, approved the study. A description of the full study is available (22).
The study population included all fifth-grade students in three geographic areas around the research sites who were enrolled in public schools that had at least 25 fifth graders. Each research site used a two-stage probability sampling procedure. In the first stage, schools were randomly selected with constraints to ensure that the three major racial/ethnic groups were well represented in the final sample. In the second stage, information about the study was distributed to each student in all the fifth-grade classrooms in the selected schools (n = 21) for them to take home to their parents or primary caregivers (hereafter referred to as parents), which yielded a potential pool of 1,848 students. Children were excluded if they were not attending a regular academic classroom or they or their primary caregiver could not complete interviews in English or Spanish. A total of 1,059 (57%) students and their parents returned written permission to be contacted about the study. Not all eligible families were fully pursued because of a limited time frame, but of the 871 (82% of those giving permission) families that were, 650 (75%) completed both the parent and child interviews required for the current analysis. The 7% of the children who were not identified by their parents as being non-Hispanic African American, Hispanic, or non-Hispanic white were eliminated from the current analysis. Of the 599 constituting the analysis sample, 40% were African American, 34% Hispanic, and 26% white, and child age M = 10.80 (s.d. = 0.62). Additional sample demographics are provided in Table 1. There were no significant differences in the racial/ethnic composition of this sample and the sampled population. We were unable to compare the analysis sample with the sampled population on additional characteristics because the active informed consent procedure approved by the IRB did not allow us to collect any information on nonparticipants.
Table 1. Demographics of the sample: Healthy Passages initial cross-sectional study
The procedures were approved by the Institutional Review Boards at five institutions. Two trained interviewers visited the child and parent at the home or an agreed-upon community location. All written materials could be presented in either English or Spanish. All measurements took place in an area sufficiently private to prevent other household members from seeing the results. Computer-assisted interviews were conducted with the child and parent separately. Each child and parent received a gift certificate for their time.
BMI calculations were based on weight and standing height obtained according to standard anthropometric protocols (23) by trained and certified interviewers. Height was measured with the participant in bare feet or socks. Two independent measurements were taken for each participant, and if the measurements differed by ≥0.5 cm for height or ≥0.2 kg for weight, a third measurement was taken. Standing height was measured to the nearest millimeter using a portable stadiometer. Weight was measured to the nearest 0.1 kg using a Tanita electronic digital scale. Calibration of the scale was checked regularly. The two measurements of height and weight that were closest in agreement were averaged and employed to calculate BMI using the Quetelet index (weight (kg)/height (meters)2). Sex-specific ratios of children's weight for height by age (months) were calculated according to CDC guidelines (24), and classifications were constructed such that BMI < 85th percentile was not overweight, 85th ≤ BMI < 95th percentile was considered overweight, and BMI ≥ 95th percentile defined obese.
QOL was measured with the PedsQL Child Self-Report Age 8–12 form (10) administered to the child. This instrument is a 23-item well-validated scale that provides scores for total QOL as well as physical (8 items) and psychosocial (15 items) domains. The latter domain is constituted by subscales for emotional, social, and school functioning. The child is asked how much of a problem various physical activities, feelings, social situations, and school activities have been in the past month. Responses are provided on five-point scales (0 = never a problem, 4 = almost always a problem). To reduce the number of statistical tests of related variables while still distinguishing between two major domains of QOL, we restricted our analyses to the physical and psychosocial scale scores. Internal consistency α = 0.73 and 0.84, respectively, for these scales in this sample.
Body dissatisfaction was measured by presenting the child with a set of drawings of seven same-sex children representing graduated sizes, from thin to overweight (25). The children were asked first to choose which body they thought a boy or girl of their age should look like. They were next asked to choose which of the seven bodies looked most like them. The absolute difference between the ordered sizes of the two bodies chosen was calculated and classified into none, 1, and 2 or more. Numerous studies have provided support for this being a valid measure of body dissatisfaction (see reviews by, e.g., 26,27,28). Estimates of children's own body size have been shown to be accurate (29). Indeed, the correlations in the current study of 0.63 for boys and 0.71 for girls between the drawing they chose as looking most like their own body and their BMI suggest that the children demonstrated acceptable accuracy in representing their own body size. These correlations also provide present criterion-referenced validity for that component of measuring body dissatisfaction. In addition, test–retest reliability for the figure selection task for children aged as young as 8 years has been found to be high (30,31).
General self-concept was measured with the six-item global self-worth subscale of the Self-Perception Profile (32). On each item, children are asked to identify which contrasting description best fits them (e.g., some kids like the kind of person they are, other kids often wish they were someone else) and how much (sort of true, really true). Scores on this scale are highly stable up to 3 years (r = 0.61) (33). Because it is a short scale, the internal consistency α = 0.66 in this sample is considered satisfactory and comparable to those reported elsewhere for children of elementary school age (34,35).
Means for study variables were compared for sex, race/ethnicity, and their interactions using factorial ANOVAs. Mean differences in physical and psychosocial QOL between weight classifications were examined with one-way ANOVAs. Hierarchical multiple regression analysis was used to examine the relationship between BMI and QOL, with BMI as a continuous variable, and in conjunction with body dissatisfaction and self worth. When repeated with log-transformed variables to correct skewness in indicated variables, results were identical to those for raw variables. The results from analysis of raw variables are presented here. Which demographic covariates (e.g., the child's age and sex, family income, parental education) of QOL should be controlled in the hypotheses tests were identified first with univariate correlations and multiple regressions. Because only parental education had a significant association with any of the QOL variables, it was entered as a control variable in all regression analyses. Whether body dissatisfaction and general self-concept mediate a relationship between BMI and QOL was tested according to procedures outlined elsewhere (36,37). Accordingly, four conditions must be met for the mediation hypothesis to be supported: (i) BMI is related to QOL, (ii) BMI is related to body dissatisfaction and general self-concept, (iii) body dissatisfaction and general self-concept are related to QOL, and (iv) the relationship between BMI and QOL is reduced when body dissatisfaction and general self-concept are entered into a regression equation before BMI.
The overall BMI M = 21.38 in the sample corresponds to the 69th percentile on the national growth charts (24). Furthermore, 57% of the sample was classified as not overweight, 17% as at risk for overweight, and 26% as obese. Mean scores on BMI, physical and psychosocial QOL, and body dissatisfaction and general self-concept are presented in Table 2 for the sample as a whole.
Table 2. Means (and standard deviations) for study variables for overall sample and by race/ethnicity
No significant main effects were found for child's sex or for interaction between child's sex and either race/ethnicity or weight classification for any of the main variables. Even so, as can be seen in Table 2, there were significant racial/ethnic differences for all main variables. BMI for whites was significantly lower than for African Americans and Hispanics, who did not differ from one another. Psychosocial QOL was significantly higher for whites than for both African Americans and Hispanics, who again did not differ. Whites had significantly higher physical QOL than Hispanics; neither group differed from African Americans in this dimension. Body dissatisfaction was significantly lower among whites than Hispanics; again, neither group differed from African Americans. Finally, for global self-concept, whites were significantly higher than African Americans, who in turn were significantly higher than Hispanics.
Association between weight status and QOL
The interaction between weight classification and race/ethnicity on either physical or psychosocial QOL was not significant. As can be seen in Table 3, there was a significant main effect from weight classification on psychosocial but not physical QOL. Post hoc comparisons identified a difference in psychosocial QOL only between the not-overweight and obese classes; the overweight class did not differ significantly from the other weight classes. There were also significant main effects on body dissatisfaction and global self-concept (see Table 3), again only with differences between the not-overweight and obese classes. As can be seen from Table 4, Analysis A, BMI was significantly associated with psychosocial but not physical QOL, when controlling for parental education. Even so, only 2% of the variance in psychosocial QOL was accounted for by BMI beyond that attributed to parental education.
Table 3. Means (and standard deviations) for study variables by weight classification
Table 4. Regressions of quality of life
Mediation by body dissatisfaction and global self-concept
Based on the analyses presented above, condition (i) for meeting the mediation hypothesis was met because BMI was associated with psychosocial QOL. The analyses indicate that mediation cannot occur for physical QOL because it is not associated with BMI. Condition (ii) is also met because analyses showed that BMI was significantly associated (change in R2 = 0.12, P < 0.001) with both body dissatisfaction (β = 0.26, P < 0.001) and global self-concept (β = −0.10, P < 0.05). Condition (iii) was met as well (Table 3, Analysis B). Body dissatisfaction and global self-concept jointly accounted for a significant 14% of the variance in psychosocial QOL beyond that accounted by parental education. In addition, both body dissatisfaction (β = −0.10) and global self-concept (β = 0.35) contributed uniquely to psychosocial QOL.
To address condition (iv) for mediation, the regression coefficients for BMI obtained in Analysis A and Analysis B were compared. When body dissatisfaction and global self-concept were not included in the regression of psychosocial QOL, the contribution of BMI was significant whereas it was not when these variables were included (β = −0.13, P < 0.01 and β = −0.06, P = n.s., respectively). Thus, the contribution of BMI to QOL was reduced, essentially to zero (change in R2 = 0.00) when there was accounting for body dissatisfaction and global self-concept. The Sobel test (38) indicated that significant portions of the relationships between BMI and psychosocial (Z = 3.10, P < 0.01) QOL were mediated by body dissatisfaction and global self-concept.
The results indicate that obese fifth-grade children experience a significantly lower psychosocial QOL than children who are not overweight regardless of racial/ethnic group. Physical QOL, however, does not differ by weight status. This finding of reduced psychosocial QOL in obese children is largely consistent with previous findings. Obese 8–12-year-old children in a population-based study in Australia (7) self-reported both their psychosocial and physical QOL to be at levels lower than those of not-overweight children. The QOL reported by Australian children was slightly higher (better) in each weight classification than what was reported by the current U.S. sample. The much larger representation in the present study of racial or ethnic minority children, who report a significantly lower QOL on average than white children, might explain this difference.
The lack of a differentiation in physical QOL between obese and not-overweight children in the current sample is counter to the expectation that obese children would be more likely to experience difficulties in the physical realm. Being obese might make it more difficult for children to run and participate in sports and more likely for them to have reduced energy, which are examples of indicators of physical QOL (10). It could be argued that the extremely obese children, who would experience a more reduced physical QOL, choose not to participate in this study. Yet we note that the current sample consisted of 5.3% with a BMI in the 99th percentile. However, previous research has been inconsistent in reporting a reduced physical QOL among obese children (e.g., 7,8,10). One difference from previous studies is that the current sample contained only young children (10–11 years old), whereas these other studies have included older children into adolescence. It may be that any physical effects of being obese have not yet developed by age 10 or 11. Examination of the relationship between weight classification and physical QOL differentiated by age in future studies could clarify this issue.
Inspecting the mean QOL in each weight status group, including the severely obese groups examined in a previous study using the same QOL measure (11), suggests there may be a weight status gradient in the QOL of children. In the present study, however, when we examined QOL in relation to BMI (expressed as a continuum) in a community sample, the association, while significant, was not strong. BMI accounted for only 2% of the variance in psychosocial QOL, and parental education, a marker for general socioeconomic status, was a stronger predictor than BMI. Consequently, although there appears to be a general trend toward reduced psychosocial QOL as children exhibit unhealthier weight, especially in the obese range, there is considerable variation in any given weight class, and other factors might influence QOL more strongly.
This study also aimed to examine the role of children's self-concept in the relationship between weight status and QOL. Both global self-concept and body dissatisfaction independently mediated the association between BMI and psychosocial but not physical QOL. The mediating roles of the two measures differed, however. Not surprisingly, we found that BMI is more strongly related to body dissatisfaction than to global self-concept, while global self-concept is more strongly related to psychosocial QOL than is body dissatisfaction. That is, based on these findings, the relatively smaller effect an increased BMI may have on global self-concept in children has a larger effect on psychosocial QOL. Alternatively, although an increased BMI may have a large effect on children's dissatisfaction with their bodies, it may also have a less general effect, as measured by psychosocial QOL. Of course, this is a cross-sectional correlational study, which severely limits discussion of cause and effect.
Our study has other limitations that could be addressed in future research. First, a minority of the identified population was enrolled in this study, raising issues of sampling bias. Whereas a representative distribution of race/ethnic status could be achieved, the analysis sample may have differed in other respect. Second, this study covered only fifth graders. Studying these associations longitudinally would be illuminating. It would be even more informative if QOL could be examined in a setting in which BMI is experimentally manipulated, such as in studies evaluating randomized treatments for overweight in children. In our study, only children's self-report of QOL was obtained. Corroborating such self-report with proxy reports, such as was done in one of the Australian studies (7), would be valuable. Finally, the relatively small size of the sample may have hindered our ability to find interactions between weight status and racial/ethnic group membership on QOL.
Additional dimensions of self-concept should be considered in future research to further understand the effects on QOL of being obese in childhood. Social dimensions of self-concept may be particularly important to examine given the stigmatizing effect of being obese in the United States (26,39,40). Stigma associated with obesity start as early as preschool (41) and can be expected to increase as children progress into adolescence. These realities underscore the importance of exploring opportunities to address increasing BMI early in life through interventions with at-risk children as well as universal prevention programs. Such efforts, while possibly improving physical health in the long run, may well have more immediate effects on psychosocial functioning, general well-being, and adjustment of children.
The Healthy Passages Study is funded by the Centers for Disease Control and Prevention, Prevention Research Centers (Cooperative Agreements U48DP000046, U48DP000057, and U48DP00056). The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC. The contributions made to this research by study participants in the Birmingham, Houston, and Los Angeles areas, other Healthy Passages investigators, field teams at each site, and the CDC Division of Adolescent and School Health are gratefully acknowledged.
The authors declared no conflict of interest.