Excess body weight has become a public health burden in both developing and developed countries(Reference Ebbeling, Pawlak and Ludwig1). In Canada, 25·7 % of children and adolescents are overweight or obese, and 8·6 % obese(2). Excess body weight has been widely acknowledged to contribute to various chronic diseases, resulting in diminished life expectancy(Reference Reilly and Kelly3–Reference Ball and McCargar5). Overweight or obesity in children and adolescents has also negative consequences for self-esteem, psychosocial health and cognitive development(Reference Wang and Veugelers6–Reference Li, Dai and Jackson8).
Unhealthy diet, characterized by increased intakes of fat and sugar and inadequate intakes of fruits, vegetables and whole grains(Reference Florence, Asbridge and Veugelers9), as well as insufficient physical activity (PA) have been identified as two fundamental factors leading to overweight and obesity(Reference Swinburn, Caterson and Seidell10, Reference Janssen, Katzmarzyk and Boyce11). Most childhood obesity strategies therefore include the combination of promotion of healthy eating and active living(Reference Brown and Summerbell12, Reference Veugelers and Fitzgerald13). Such approaches have also been shown to benefit self-esteem and academic performance(Reference Wang and Veugelers6, Reference Florence, Asbridge and Veugelers9, Reference Veugelers and Fitzgerald13, Reference Rampersaud, Pereira and Girard14).
The importance of excess body weight for impaired health-related quality of life (HRQOL) in children and adolescents has been documented in both clinical and population-based studies(Reference Hughes, Farewell and Harris15–Reference Swallen, Reither and Haas22). However, only a few studies have looked at the importance of the factors underlying excess body weight – being diet quality and PA – for HRQOL. Moreover, the few that have were conducted were mostly among children and adolescents with chronic diseases or specific health conditions(Reference Shoup, Gattshall and Dandamudi19, Reference Yackobovitch-Gavan, Nagelberg and Phillip23, Reference Hamiwka, Cantell and Crawford24). Very few studies on diet quality, PA and weight status in relation to HRQOL in children used representative population-based samples(Reference Chen, Sekine and Hamanishi25–Reference Chen, Sekine and Hamanishi26). Although such studies are important in identifying the undesired dietary and activity patterns and in designing effective intervention strategies, no such studies have been conducted in Canada. The purpose of the present study was therefore to establish the associations of diet quality, PA and weight status with HRQOL among children in Canada.
Methods
The survey
The Raising Healthy Eating and Active Living Kids in Alberta (REAL Kids Alberta) survey was developed to evaluate Alberta Health and Wellness initiatives that promote healthy body weights among children and adolescents. The survey was conducted in 2008 among grade 5 students who are primarily 10 to 11 years old. The survey employed a one-stage stratified random sampling design. The sampling frame includes all elementary schools in the province with the exception of private schools (4·7 % of all Alberta children), francophone schools (0·6 %), on-reserve federal schools (2·0 %), charter schools (1·7 %) and colony schools (0·8 %)(27), leaving primarily public and Catholic schools in the sampling frame. Schools were stratified into three geographies: (i) urban, i.e. Calgary and Edmonton; (ii) cities, i.e. other municipalities with more than 40 000 residents; and (iii) rural, i.e. municipalities with fewer than 40 000 residents. Schools were randomly selected within each of these three strata to achieve a balanced number of schools and students in each stratum.
Of the 184 invited schools, 148 (80·4 %) participated in the study. All grade 5 students (n 5594) attending these schools received an envelope with a consent form and a survey to take home for their parent/guardian(s) to complete. A total of 3645 students returned the forms and had received parental consent to participate in the study. In total, 3421 students (61·2 % of all students) completed the survey when trained assistants visited their schools to administer the surveys and to measure heights and weights. The surveys included questions on nutrition, PA and HRQOL measured by the youth version of the EQ-5D (EQ-5D-Y)(Reference Wille, Badia and Bonsel28). The questionnaires, both for students and parents, are posted online (www.REALKidsAlberta.ca).
Assessments
Diet quality assessment
The Harvard Food Frequency Questionnaire for Youth and Adolescents (YAQ) is a validated food frequency instrument that is suitable for grade 5 students(Reference Rockett, Wolf and Colditz29, Reference Rockett, Breitenbach and Frazier30). The YAQ provides detailed information on the frequency and kinds of foods that children and adolescents consume(Reference Rockett, Wolf and Colditz29). On the basis of students’ responses to the YAQ and Canadian Nutrient Files(31), we calculated intakes of nutrients and energy for each participant. On the basis of these intakes we determined the diet quality using the Diet Quality Index–International (DQI-I) composite measure. The DQI-I encompasses variety, adequacy, moderation and overall balance of the diet(Reference Kim, Haines and Siega-Riz32). We divided the DQI-I scores into tertiles for the purpose of our analysis.
Physical activity assessment
Students and their parent/guardian(s) responded to questions on: (i) travel to and from school; (ii) time spent to get to and from school; (iii) frequency of child's activities outside school hours; (iv) activities at morning and lunch recess in the past 7 d; and (v) frequency of involvement in sports and physical activities in the past 7 d. These questions, containing twenty-nine items, were largely adopted from the Physical Activity Questionnaire for Children (PAQ-C) which has been demonstrated to be valid and have high reliability(Reference Crocker, Bailey and Faulkner33, Reference Kowalski, Crocker and Faulkner34). We derived a composite score ranging from 0 to 5 based on the score given to the twenty-nine items.
Overweight and obesity assessment
Standing height was measured to the nearest 0·1 cm without shoes and body weight to the nearest 0·1 kg on calibrated digital scales. BMI was calculated by dividing weight (in kilograms) by the square of height (in metres). Body weight was categorized as normal weight, overweight and obese using the BMI cut-off points for children and adolescents by the International Obesity Taskforce(Reference Cole, Bellizzi and Flegal35). These cut-offs are based on adult definitions of overweight (25 kg/m2 or more) and obesity (30 kg/m2 or more), adjusted to specific age and gender groups for children.
Outcome measures
HRQOL was assessed by the EQ-5D-Y (youth) where the language of the EQ-5D instrument for adults is modified so that children can better understand it. The HRQOL instrument consists of a five-dimensional descriptive system asking whether children have (i) no problems, (ii) some problems or (iii) a lot of problems with: (i) walking; (ii) looking after myself; (iii) doing usual activities; (iv) having pain or discomfort; and (v) feeling worried, sad or unhappy, respectively(Reference Wille, Badia and Bonsel28). The instrument also includes a Visual Analogue Scale (VAS) which is anchored at 100 (best imaginable health) and 0 (worst imaginable health) to capture self-rated values of health status in children. The EQ-5D-Y has been validated for several languages and countries(Reference Ravens-Sieberer, Wille and Badia36). The main advantages of the instrument are that it is short and simple, can be completed within 10 min by children, and can be used to estimate a single index score to be analysed subsequently in economic evaluation studies(Reference Wille, Badia and Bonsel28).
Analytical methods
We applied the χ 2 test to examine differences in the prevalence of reported health problems for each of the five EQ-5D-Y dimensions by the observed predictors. As very few students reported ‘a lot of problems’, we combined this with ‘some problems’ to create a dichotomous outcome (no problems v. with any problems). We described generic HRQOL by different groups of diet quality, PA and weight status as measured by the EQ-5D-Y VAS score. We applied multilevel multivariable linear regression to assess the association of diet quality, PA and body weight with the generic HRQOL. We applied multilevel multivariable logistic regression to examine the effect of diet quality, PA and body weight for the EQ-5D-Y dimensions. These regression models accommodated the hierarchical data structure in that student observations are nested within their schools. The regression analyses were adjusted for the confounding influence of gender, place of residence, household income and parental education.
The EQ-5D-Y descriptive system was fully completed by 3406 students (99·6 %) and 3379 students (98·8 %) answered the VAS. These missing outcomes were not considered in the analyses. Of all participating students, 3340 parents completed a survey on educational attainment, household income, place of residency (urban, town, rural) and their child's PA. Missing values for education and income were considered as separate categories in the analysis but the estimates are not presented. All analyses were weighted to accommodate the design effect such that all estimates pertain to the population of grade 5 students in Alberta. Data were analysed using the STATA statistical software package version 11·0 (StataCorp, College Station, TX, USA). The study programme was approved by Health Research Ethics Board of the University of Alberta.
Results
Students who were physically inactive reported significantly more HRQOL problems relative to their peers who were physically active on four of the five dimensions: ‘looking after myself’, ‘doing usual activities’, ‘having pain or discomfort’, and ‘feeling worried, sad or unhappy’. Compared with the normal weight group, obese students had significantly more HRQOL problems on the ‘looking after myself’ and ‘feeling worried, sad or unhappy’ dimensions. Furthermore, across diet quality tertiles, statistically significant differences were reported with respect to ‘having pain or discomfort’ (Table 1). Mean HRQOL score for students in the highest tertile of diet quality, with physically active lifestyle and with healthy weight was 82·2, 84·2 and 81·5, respectively (Table 1).
Prevalence of problems in the EQ-5D-Y dimensions and mean VAS score by diet quality, physical activity and weight status: grade 5 students aged 10–11 years (n 3421), Alberta, Canada, 2008
VAS, Visual Analogue Scale; DQI-I, Diet Quality Index–International.
The χ 2 test was used to obtain the P values where weighted percentages of students with problems in different dimensions are presented.
*The EQ-5D-Y VAS score ranged from 0 to 100, where 100 is best imaginable health.
Table 2 shows multivariate-adjusted associations of HRQOL with diet quality, PA and body weight status. The VAS value was statistically significantly higher for students who were physically active, normal weight and in the highest DQI-I tertile relative to students who were not physically active, overweight or obese and in the lowest DQI-I tertile.
Associations of diet quality, physical activity, body weight status and sociodemographic factors with VAS scoreFootnote *: grade 5 students aged 10–11 years (n 3421), Alberta, Canada, 2008
VAS, Visual Analogue Scale; DQI-I, Diet Quality Index–International.
The regression analysis was mutually adjusted for variables in the table. All estimates were weighted to represent population estimates.
* The EQ-5D-Y VAS score ranged from 0 to 100, where 100 is best imaginable health.
Table 3 presents the adjusted odds ratio of reporting problems on the EQ-5D-Y dimensions. Diet quality, body weight status and PA significantly affected one, two and four of the five dimensions, respectively, after accounting for the effect of sociodemographic variables. The results are very similar to the unadjusted results in Table 1.
Odds ratio of reporting problems in the EQ-5D-Y dimensions by diet quality, physical activity, weight status and sociodemographic factors: grade 5 students aged 10–11 years (n 3421), Alberta, Canada, 2008
DQI-I, Diet Quality Index–International.
All analyses were mutually adjusted for variables in the table. All estimates were weighted to represent population estimates.
Discussion
The present study reveals that diet quality, PA and body weight are associated with HRQOL in grade 5 students. These associations were independent of gender and sociodemographic factors. The study further reveals an association of diet quality with the VAS score whereby children with better diet quality reported better HRQOL. Students who were physically inactive, overweight or obese had reportedly a lower HRQOL.
The relationship between PA and HRQOL has been well described in adults relative to younger populations. An association of higher HRQOL scores with higher PA levels has been consistently documented in healthy adults(Reference Bize, Johnson and Plotnikoff37). Our observations that physically active children have significantly higher HRQOL scores than those in the inactive group support the previous findings in both adult(Reference Bize, Johnson and Plotnikoff37) and the few child and adolescent studies(Reference Chen, Sekine and Hamanishi25, Reference Chen, Sekine and Hamanishi26, Reference Sánchez-López, Salcedo-Aguilar and Solera-Martínez38, Reference Boyle, Jones and Walters39). A systematic review of HRQOL among children and adolescents reported that excess body weight had a moderate to strong negative influence on HRQOL, whereas the role of psychosocial, emotional and school functioning on HRQOL had been inconsistent(Reference Tsiros, Olds and Buckley40). Our observation in a large population-based sample of grade 5 students confirms this relationship of excess body weight with lower quality of life. We also showed that children from parents who received less education had lower quality of life. Identifying determinants for different aspects of the HRQOL is essential to developing public health intervention strategies and targets. Our study revealed that PA has a significant impact on each of the five EQ-5D-Y dimensions except ‘walking’. This is consistent with the few previous studies that have demonstrated that physically active children exhibit better physical and psychological quality of life(Reference Chen, Sekine and Hamanishi26), better self-esteem(Reference Strauss, Rodzilsky and Burack41) and better psychosocial quality of life(Reference Shoup, Gattshall and Dandamudi19, Reference Strauss, Rodzilsky and Burack41). We observed that overweight and obese children were reportedly more worried, sad or unhappy, which seems consistent with HRQOL studies reporting that obesity is associated with impaired psychosocial functioning(Reference Friedlander, Larkin and Rosen17, Reference Williams, Wake and Hesketh18, Reference Varni, Limbers and Burwinkle21, Reference Swallen, Reither and Haas22), lower physical functioning(Reference Tsiros, Olds and Buckley40), lower emotional functioning(Reference Schwimmer, Burwinkle and Varni16, Reference Swallen, Reither and Haas22) and lower self-esteem(Reference Friedlander, Larkin and Rosen17, Reference Swallen, Reither and Haas22).
Relative to studies in other countries using the EQ-5D-Y, children in Alberta reported a higher prevalence of health problems in the dimension of pain or discomfort (46·0 %). High percentage of any health problems in pain or discomfort (43·6 %) was also presented in a general population sample of adults in Alberta using the EQ-5D(Reference Johnson and Pickard42). A possible explanation for this finding in our study may be that response to the EQ-5D-Y descriptive system could be culturally different across different countries or in geographic areas within a country. Further analysis using the EQ-5D-Y in other provinces in Canada and in other countries may help to ascertain the origin of this finding.
In addition, it is also important to examine the magnitude of the differences to estimate minimally important differences (MID) in HRQOL scores between comparison groups(Reference Jaeschke, Singer and Guyatt43). MID for the EQ-5D index and the EQ-5D VAS have been previously estimated for some disease conditions(Reference Pickard, Neary and Cella44, Reference Walters and Brazier45). We have not identified any study demonstrating a MID value for the EQ-5D-Y VAS or index. Estimation of MID in HRQOL scores requires a variety of approaches, both distribution-based and anchor-based(Reference Lydick and Epstein46), and a rigorous examination of various factors that may affect the degree of minimal differences(Reference Norman, Sloan and Wyrwich47). Future research is warranted to investigate the magnitude and direction of differences/changes in HRQOL to establish MID cut-off points for the EQ-5D-Y for the general population of children and adolescents.
In the present study we did not estimate an index score for EQ-5D-Y as we were interested in quality of life that was measured and described by children themselves. Since no EQ-5D-Y tariff is available for use in younger populations, several previous studies in quality of life assessment in children and adolescents using the EQ-5D or EQ-5D-Y have reported on utility indices generated from the existing US or UK EQ-5D tariffs(Reference Wu, Ohinmaa and Veugelers48, Reference Willems, Joore and Nieman49). There is a debate about the applicability of the existing social tariffs for adults to children(Reference Ravens-Sieberer, Wille and Badia36). Current research interest in the field is to establish a child-specific value set for the EQ-5D-Y for use in population health research and economic evaluation studies(Reference Ravens-Sieberer, Wille and Badia36).
The present study is the first to reveal the associations of diet quality, PA and body weight with HRQOL among preteen children. Specifically, the study contributes to the evidence of positive associations between diet quality, PA and HRQOL in schoolchildren, independently of weight status and sociodemographic characteristics. These findings suggest that school-based programmes promoting healthy eating and active living may not only help to prevent children from becoming overweight, but may also benefit their HRQOL regardless of weight status. The differences in HRQOL outcomes by diet quality specifically suggest the importance of nutrition programmes focusing on improving diet quality among children in the development of school health promotion. One Canadian study has shown that nutrition programmes that are based on comprehensive school health exhibit a greater positive effect on students’ diets, PA and overweight reduction than a single nutrition programme(Reference Veugelers and Fitzgerald13). More research is needed to examine whether such comprehensive school health approaches that integrate nutrition education, nutrition policy, healthy food services, environmental support and various PA strategies into a whole school model will result in an improvement of HRQOL among children. This may justify broader investments in school programmes to the benefits of health and quality of life among children(Reference Veugelers and Fitzgerald13, Reference Deschesnes, Martin and Hill50).
Major strengths of the present study include the use of a large population-based sample of students, the use of objective measurement of height and weight, the adjustment for sociodemographic factors in the analysis, the use of a validated generic multidimensional HRQOL measure for children, and the application of multilevel regression to account for hierarchical data structure and with weighted analysis to accommodate the survey design effect. Limitations of the study should also be clarified. The observed associations of diet quality, PA and body weight with HRQOL could not be inferred as causality based on the cross-sectional survey design. Since participation in the survey was voluntary, selection bias may have occurred due to possible differences in characteristics between the participants and the non-participants. Our study was conducted in a sample of grade 5 students, which limits the generalizability of the results to other age groups of children. PA and diet assessments in the current study were based on measurement of self-report, and may have been affected by measurement error. The use of objective measures of PA (e.g. pedometers) would allow for more accurate evaluation of PA of students, although this may pose challenges in financial and resource support in large-scale population-based studies(Reference Simen-Kapeu and Veugelers51).
Conclusions
The present study demonstrated the importance of diet quality, PA and body weight status for HRQOL which will help justify broader implementation of school health programmes that promote healthy eating and active living, as these programmes will help reduce the burden of childhood obesity and improve quality of life.
Acknowledgements
This research was funded through a contract with Alberta Health and Wellness. The present analysis was further funded through a Canada Research Chair in Population Health and Alberta Innovates Health Solutions Scholarship to P.J.V. All interpretations and opinions are those of the authors. The authors declare that they have no conflict of interests. All authors have participated in the planning of the analysis and writing of the article. Analysis was done by X.Y.W. The authors thank all grade 5 students, parents and schools for their participation in the REAL Kids Alberta evaluation; all of the research assistants and regional health promotion coordinators for the execution of the data collection; and Connie Lu and Dr Stefan Kuhle for data management, validation and quality control.
