The influence of physical activity, sedentary behavior on health-related quality of life among the general population of children and adolescents: A systematic review

1The associations between physical activity (PA), sedentary behavior (SB) and physical and mental health among children and adolescents have been well established. Systematic reviews and primary studies of PA and health have indicated that children and adolescents who engaged in increased levels of physical activities had better physical and mental health and psychosocial well-being than those in an inactive lifestyle. Promoting PA among children and adolescents has been demonstrated to benefit a number of disease conditions, including obesity, coronary heart disease and other health problems. Sedentary behavior characterized often as screen-based media use behaviors including watching television (TV), using computers/smartphones and playing video games are associated with various negative health consequences. The adverse consequences resulted from sedentary behaviors include an increased risk of obesity, cardiovascular disease and all-cause mortality, and a range of impaired psychological health. Sedentary behavior also contributes to a delay of cognitive development and a decrease in academic achievement of children and youth.

2Health-related quality of life (HRQOL) has been increasingly used as a health outcome among children and adolescents to assess their physical and social functioning, mental health and well-being, and to evaluate population-based intervention programs. HRQOL is a multidimensional construct that covers physical, psychological, and social health and hence represents overall health of an individual. Assessment of HRQOL among children and adolescents is important in identifying subgroups with poor health status and in guiding effective intervention strategies to improving health of the younger population. The association between PA and HRQOL in children and adolescents has been mainly investigated among those with chronic disease conditions such as obesity, asthma and cancer. These studies have reported that children and adolescents who undertake an active lifestyle experience better HRQOL than their peers who engage in an inactive lifestyle. In the general population, the relationship between PA and HRQOL has been well investigated in adults relative to children and youth (e.g.,school or population-based samples). Moreover, much less is known about the relationship between sedentary behavior and HRQOL. In the past decade, we have found accumulating studies that examined the effect of PA and sedentary behavior on HRQOL among populations of children and adolescents. To our best knowledge, no systematic review to date has been published to evaluate the relationship between PA, sedentary behavior and HRQOL among the general popluation of relatively healthy children and adolescents. Particularly, it is essential to explore how PA and sedentary behavior influence different aspects of physical, psychological and social functioning of HRQOL among children and adolescents, and whether a dose-response relation exists between PA levels, time spent on sedentary behaviors and HRQOL. Such information will help to provide an evidence-base for public health policy to invest in school-based health promotion programs in order to enhance health and quality of life among children and adolescents.

3The purpose of the present study was to 1) review and synthesize the existing literature that investigated associations between PA, sedentary behavior and HRQOL in the general population of children and adolescents; 2) provide evidence-based recommendations for guiding school-based health behavior intervention programs to enhance HRQOL among children and adolescents.

4A computer search was performed for English language literature using databases of MEDLINE (1946-Week 2, January, 2017), EMBASE (1974-Week 2, January, 2017), PSYCINFO (1987-Week 2, January, 2017). MeSH headings and keywords used included ‘physical activity’, ‘exercise’, ‘sedentary behavior’, ‘screen time’, ‘television’, ‘computers’, ‘video games’, ‘lifestyle’, ‘health-related quality of life’, ‘quality of life’, ‘health status’,‘children’, ‘adolescents’. From the relevant articles, we searched the PubMed related articles and manually examined the reference lists of the existing literature to retrieve other eligible studies.

5The electronic search was conducted by a single researcher (XYW). The reviewer screened the citations and abstracts and selected the eligible articles based on the inclusion and exclusion criterion. In case there were studies that the primary reviewer was uncertain whether a paper was eligible for the review, the full text articles were obtained. The full-text articles of all potentially eligible studies were retrieved, and then reviewed by the two reviewers (XYW, LHH) separately for inclusion criteria. Disagreements regarding the eligibility of the studies for inclusion were resolved by discussion among all the researchers.

6In this review, we aimed to collect studies that focused on examinations of the associations between physical activity, sedentary behavior and HRQOL among the population of healthy children and adolescents. The inclusion criteria were as follows: (1) The outcome of interest: Studies used one or more multi-dimensional HRQOL measures, and the outome has to be indicated as quality of life (QOL) or health-related quality of life. (2) Population of interest: The genereal population of children and adolescents aged between 3–18 years, including school-aged, or community-based children and adolescents. The general population of children and adolescents refers to all children and adolescents in communities or schools in a geographic region or a country who are relatively healthy in comparison with those children and adolescents with specific diseases (e.g., patients with diabetes, obesity or cerebral palsy, etc.). For longitudinal and intervention studies with a follow-up age of greater than 18 years, the age of children and youth had to be 3 to 18 years at baseline when at least one exposure of PA and sedentary behavior was measured. (3) Types of studies: Cross-sectional, longitudinal and health intervention studies. The intervention studies for quality of life were those studies that targeted to promoting PA and reducing sedentary behavior among children and adolescents. (4) Measure of the exposure: PA is defined as any bodily movement worked with muscles that requires energy expenditure. Sedentary behavior is any waking behavior while sitting, lying, reclining or standing with low energy expenditure. Both subjective and objective measures of PA and sedentary behavior were included.

7The exclusion criteria were: (1) Studies that examined associations between PA, sedentary behavior and HRQOL among children and adolescents with specific chronic disease conditions (e.g.,diabetes, asthma, obesity). (2) Studies that assessed associations between PA, sedentary behavior and HRQOL among adults. (3) Studies that used only a single item of self-rated/self-perceived health as a marker of HRQOL. (4) Studies that used combined indicators of PA, sedentary behavior and other variables (e.g.,diet factors,sleep) that were generated from cluster analysis. (5) Reviews, meta-analyses, study protocols, comments, letters, case reports and guidelines.

8A single researcher (XYW) abstracted data from all eligible full text articles that included study publication year, primary author, country, study design, sample, PA and sedentary behavior assessments, HRQOL measures, statistical methods and main findings. Another reviewer (LHH) checked the abstracted information for completion and accuracy.

9We used the Newcastle-Ottawa Scale to assess the study quality. This scale is a checklist with eight items that consists of three quality components: selection, comparability and outcome. Each item can be scored as one or two points and summed up to a total score, ranging from 0 to 9, with a higher score indicating low risk of bias or better quality. Previous research categorized the risk of bias score of individual studies into high, moderate and low risk of bias. We grouped the summed scores into high (0–4), moderate (5–6), and low (7–9) risk of bias.

10The main findings were synthesized using descriptive tables for qualitative comparisons. The study characteristics, the assessment of the exposure and the outcome, the association between PA, sedentary behavior and HRQOL, and the risk of bias were presented for each included study in the evidence table. For studies that utilized similar measures of the exposure (e.g., PA and screen time) and the outomes (e.g., the PedsQL 4.0 total score), we performed meta-analyses for the overall association between PA, screen time and HRQOL. Pooled estimates were obtained for the differences in total HRQOL scores and their 95% confidence intervals between different levels of physical activity and sedentary time. A forest plot was generated for the overall effect across the studies. We used the random-effects model in meta-analyses to account for heterogeneity across the studies. The tau-square (Tau2) value and I2 (squared) index was used to test the degree of heterogeneity. Review Manager Software 5.2 (The Cochrane Collaboration, Copenhagen Denmark) was used for the meta-analysis. Reporting of this review was guided by the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement.

11A total of 6,071 relevant citations (MEDLINE: 3,108, EMBASE: 2,545, PSYCINFO: 418) were identified through electronic database searching and screened for eligibility. Additional 22 articles were identified through searching PubMed-related articles and reference lists from existing relevant studies. After the title and abstract review, and removing duplicate references, 49 full text articles were retrieved for detailed examinations. Among these studies, 18 were excluded due to inappropriate outcomes (e.g., anxiety,self-esteem), inappropriate exposure (e.g., cluster of PA,screen-based behavior and diets), adult participants or a review. In total, 31 studies were included in the final synthesis. Selection process of the studies was displayed in the PRISMA flow diagram (Fig 1).

12Table 1 shows the characteristics of the included studies. In total, 79,046 participants were included in the statistical analysis in the primary studies with a sampe size ranging from 152 (Dalton et al., 2010) to 16,560 (Galán et al., 2013) students. The included studies came from 15 countries. Seven studies were conducted in Australia and four studies were from Spain. The remaining 20 studies were conducted in the following countries: United States (n = 3), Japan (n = 3), Germany (n = 2), Switzerland(n = 2), Canada (n = 2), UK (n = 1), Brazil (n = 1), Italy (n = 1), France (n = 1), Norway (n = 1), China (n = 1), Malaysia (n = 1), Iran (n = 1).

13Most of these studies were cross-sectional studies (n = 21). Six were longitudinal studies, and three studies evaluated the effect of school-based PA programs on HRQOL in school children and adolescents using clustered-randomized controlled trial (RCT) design. One additional study examined both cross-sectional and longitudinal associations between physical activity, sedentary behavior and HRQOL among adolescents.

14Of the included studies, 16 studies assessed the association between both physical activity and sedentary behavior and HRQOL among children and youth. A total of 13 studies examined the association between physical activity only and HRQOL among children and youth. One study analyzed the long-term effect of youth physical activity (measured at age of 7–18 years) on their HRQOL 22 years later in adulthood. One study investigated the association between electronic media use (television, video game, computer and telephone) and HRQOL among adolescents.

15Most included studies used self-report of physical activity (Table 1). Two observational studies used parent-report of physical activity for children. One Canadian study used a composite measure of PA that was based on the combination of child’s self-report and parent-report questions. One study collected PA data objectively using accelerometers. Physical activity by self- or parent-report was measured as hours/minutes a day, or number of times a week or number of days over a week or a month (Table 1). In two Japanese studies, frequency of PA was measured on a four-point scale: very often, often, seldom, almost never.

16Sedentary behavior was measured in most studies by number of hours or minutes spent daily in watching television, playing video games, using computers and telephones. Average time spent daily on sedentary activities was calculated in most of the studies. Some studies also calculated number of hours spent daily for television viewing, playing video games or using computers, respectively. One study assessed sedentary time by accelerometer.

17With respect to the HRQOL instrument, the Paediatric Quality of Life Inventory 4.0 Generic Core Scales (PedsQL 4.0) was used in 12 studies (Table 2). Three studies applied the EQ-5D-Y, and three studies utilized the KIDSCREEN-10. Other HRQOL measures included the Japanese COOP charts (n = 3), the Child Health Questionnaire (CHQ-PF50) (n = 2), the Child Health Utility 9D (CHU9D) (n = 2), the Child Health and Illness Profile-Child Edition (CHIP-CE) (n = 2), German KINDL-R (n = 2), KIDSCREEN-27 (n = 1), the Duke Health Profile (DHP) (n = 1), the Medical Outcomes Study Short Form 36 (SF-36) (n = 1), and the WHOQOL-Bref (n = 1) (Table 2). Most of the HRQOL outcomes were reported as continuous variables (e.g., PedsQL total and subscale scores, KIDSCREEN-10 index, CHU9D). Some studies also used subscales of the HRQOL measures (e.g., the EQ-5D-Y dimensions) as categorical outcome variables in the analysis.

18The average score of risk of bias among all included studies was 5.71 (SD = 1.47). The risk of bias score ranged between 3 and 8 for the individual studies. A total of 20 studies were scored as moderate risk of bias (n = 14) and high risk of bias (n = 6). Eleven studies were rated as low risk of bias (Table 2). The common reasons for the low quality included small sample size, lack of control for important confounders, inadequate statistical methods, and non-response bias. The overal evidence of the study quality was low since most of the studies were based on cross-sectional design with only a few longitudinal studies and RCTs.

19Table 2 presents the fingdings for the association between PA, sedentary behavior and HRQOL. Of the 21 cross-sectional studies that examined the relationship between PA and HRQOL among children and adolescents, 19 studies found that children and adolescents in a higher level of PA compared with those in the lower level of PA had significantly higher HRQOL. Finne et al.(2013) found in a representative sample of German children and adolescents aged 11–17 years that a higher frequency of PA was related to higher HRQOL with small to moderate effect sizes for daily PA relative to no regular PA in both boys and girls. Gopinath et al.(2012) reported that adolescents (n = 1,094) in the highest tertile of total PA had significantly 3.15 points higher in total PedsQL scores compared with those in the lowest tertile of total PA. Similarly, adolescents in the highest tertile relative to the lowest tertile of total PA had higher scores in the physical summary (5.8 points difference) and social (4.18 points difference) domains. Lacy et al.(2011) found among a large sample of Australian adolescents (n = 3,040) that higher levels of PA were associated with higher HRQOL for both boys and girls, and the relationship remained after adjusting for weight status.

20A dose-response relation between PA and HRQOL was reported by three studies with large samples. A significant linear trend between PA and HRQOL was observed in a dose-response manner for most subscales of the PedsQL 4.0 in the study by Finne et al.. Galán et al.(2013) reported a dose-response association between moderate to vigorous physical activity (MVPA) and HRQOL in a national sample of Spanish children and adolescents (n = 16,560, aged 11–18 years), where an increase in PA levels was related to additional improvements in HRQOL. One Japanese study (Chen et al.,2005b) showed a dose-response relation between PA and HRQOL, where children who engaged in lower frequency of PA had a higher odds of experiencing poor HRQOL.

21Three studies used the EQ-5D-Y, a recent developed HRQOL measure for children and youth between 8–18 years that has been previously validated. Petracci et al.(2013) used quantile regression in their study and found that children who exercised less than 2 hours a week had significantly lower scores of the EQ-5D-Y VAS relative to children who exercised more than 11 hours a week. Wu et al. (2012) observed in a population-based sample of Canadian grade five students (n = 3,421) that physically active children reported significantly less HRQOL problems (based on the adjusted odds ratio) relative to their peers who were physically inactive on four of the five EQ-5D-Y dimensions: ‘looking after myself’, ‘doing usual activities’, ‘having pain or discomfort’, and ‘feeling worried, sad or unhappy’. Children who were physically active had a significanlty higher VAS score (difference = 4.49 points) compared with those peers who were physically inactive. The observed relations between PA and HRQOL were independent of the potential confounding effect of the socio-demographic factors (gender, parental education level and family income), diet quality and weight status. Boyle et al. (2010) observed that children who achieved the recommendation of 60 minute PA per day had significantly better EQ-5D-Y VAS scores than those who did not achieve the recommendation.

22One study did not observe a significant relation between PA and HRQOL, and the study was rated as having a high risk of bias due to small sample and reliance only on the analysis of Pearson correlation. Another study observed a significant correlation between PA and HRQOL based on Pearson correlation, but not in multiple regression analysis.

23Of the seven studies that examined longitudinal associations between PA and HRQOL, six studies observed a significant effect among children and adolescents. Omorou et al.(2016) found that the cumulative level of high PA was associated with high HRQOL at 2-year follow up among adolescents (n = 1,445) in all four dimensions of the Duke Health Profile: physical, mental, social and general health. Vella,et al.(2014) reported that children who maintained participation in sports throughout the 2-year follow up had better HRQOL at follow-up than children who did not participate in sports, dropped out of sports, and commenced participation after the baseline.The magnitude of differences in total PedsQL score between sport participants and nonparticipants was approximately 5 units, greater than the minimum clinically meaningful difference of 4.5 units on PedsQL. Gopinath et al.(2012) examined temporal relations between PA and HRQOL in adolescents. The result showed that adolescents who remained in the highest level of PA over the 5 years of follow up compared with those in the lowest level of total PA had higher scores in the PedsQL total score (P = 0.04), physical summary (P = 0.0001), and social (P = 0.02) domains. Two Japanese cohort studies evaluated longitudinal relations between PA and HRQOL among children. Wang et al. (2008) found that children who were less active in early childhood (aged 3 years) had a greater odds (OR = 1.51, p = 0.016) of having lower QOL in their early adolescents (first year of junior high school study) compared with the peers who were active in early childhood. Chen et al. (2005a) reported in a cohort of 7,794 children aged 9–10 years that children with higher frequency of PA (‘very often’) relative to lower frequency of PA (‘seldom’ or ‘almost never’) at baseline survey (1999) experienced better QOL after 3 years of follow up (2002). In comparison with children who maintained higher PA as ‘often’, children who changed from ‘often’ to ‘seldom’ and who remained ‘seldom’ were more likely to have poor QOL (OR = 2.10, 95% CI:1.84–2.39; OR = 2.21, 95% CI:1.88–2.59). The longitudinal study in Canada did not find a sigificant effect of physical activity in youth on the HRQOL 22 years later in adulthood using a subgroup of participants aged 7–18 years at baseline from a Canadian population-based survey.

24Of the three cluster-randomized controlled trials, one study assessed the effect of a school-community program on HRQOL in adolescent girls in Australia, and two cluster-RCTs evaluated the effect of a school-based PA program on HRQOL in school children in Switzerland. Casey et al. (2014) found that the intervention program positively influenced quality of life of adolescent girls. The other two RCTs did not show improved physical QOL among the children, and only a little positive influence (p<0.05) of the program was observed on psychosocial QOL in first grade students.

25Of the included studies, 17 studies assessed the association between sedentary behavior and HRQOL among children and adolescents, including 13 cross-sectional studies and 4 longitudinal studies (Table 2). Most of the cross-sectional studies reported a significant association between sedentary behavior and poor HRQOL. The findings for the association are consistent across different types of sedentary behaviors, such as television viewing, using computers, playing video games, reading and doing homework and screen time measured by an accelerometer. Two cross-sectional studies did not observe a significant association between screen time and HRQOL, and these studies had relatively small samples (n = 156 and n = 371) that may compromise adequate statistical power. Finne et al.(2013) reported a dose-response relation between time spent on screen-based media use and HRQOL.

26Similarly to the effect of PA on the domains of HRQOL, sedentary behavior was linked to the multiple domains of HRQOL in childhood and adolescence. Children and adolescents who spent more time in sedentary activities reported lower HRQOL in physical, mental and psychosocial health, school functioning, and general health domains. Chen G,et al. (2014) and Xu,et al. (2014) found that longer time spent on TV viewing, playing computers or video games, or doing homework was associated with a lower utility score measured by the Child Health Utility 9D.

27All four studies with the longitudinal design found consistently that more time spent on sedentary activities (television viewing, use of computers or video games, telephones) correlated with reduced HRQOL. Omorou et al.and Gopinath et al. observed that greater screen time during follow-up was related to lower scores in several domains of QOL including physical, psychosocial, mental, emotional and school domains. In addition, Chen X, et al.(2005a) reported that there is a dose-response relation between screen time and HRQOL, where children who engaged in longer screen time were more likely to have poor HRQOL at follow up.

28Meta-analysis was performed for five observational studies using the PedsQL measure. There was a significant difference in total PedsQL scores (mean difference = 3.86, 95% CI: 2.44–5.27, P<0.01) between inactive and active children and adolescents (Fig 2). A higher level of PA was associated with an increased PedsQL total score.

29Pooled analysis for the four subgroups of the two studies indicated a significant overall difference in total PedsQL scores (mean difference = 2.71, 95% CI: 1.59–3.83, P<0.01) between sedentary (>2 hours/day) and non-sedentary (≤2 hours/day) children and adolescents, suggesting that more time of being sedentary is related to worse HRQOL (Fig 3).

30This review synthesized the relationship between physical activity, sedentary behavior and health-related quality of life among the general population of healthy children and adolescents. We found the evidence that elevated levels of physical activity are associated with higher HRQOL and sedentary behavior is inversely related to HRQOL among children and adolescents. Physical activity and sedentary behavior have significant effects on multiple physical, mental and psychosocial domains of HRQOL.

31The results from cross-sectional studies revealed that children and adolescents who participated in higher levels of physical activities had better HRQOL. The association between PA and HRQOL is consistent irrespective of weight status, age, sex, and socio-economic characteristics. The findings are also in agreement with the previous observation in general adult populations showing that PA has a positive influence on HRQOL.

32We found some evidence in this review for a dose-response effect between PA and HRQOL. For example, Finne et al. reported a linear trend between frequency of PA and PedsQL total and subscale scores, indicating a dose-response relation in both genders. Galán revealed an increasing dose-response relationship between moderate to vigorous physical activity and HRQOL in both genders of adolescents, with an observation of both linear and quadratic trends. The dose-response association between frequency of PA and HRQOL was also showed in a Japanese cross-sectional study in children. Yet more studies are needed to confirm the association and whether the relationship of HRQOL with PA is linear or nonlinear since the finding was primarily based on cross-sectional studies and previous studies have also reported nonlinear associations between PA and HRQOL among adults.

33The relationship between physical activity and HRQOL was observed for both total QOL score and subscale scores including physical, psychological and social subscales. These results lend a support to a number of previous studies demonstrating that children and adolescents who maintain an active lifestyle exhibit better physical and better psychosocial health.

34Longitudinal observational studies support a positive asscoiation between PA and HRQOL. Six studies found a significant longitudinal or temporal association between PA and HRQOL, and most of these studies analysed large samples of children and adolesents. A dose-response effect of PA on HRQOL was observed in a cohort of children with a three years follow up. However, the HRQOL data were not available at baseline in some cohort studies, and thus limiting the ability to examine changes in QOL outcomes in relation to changes in PA and sedentary behavior. Further, it did not permit to account for the impact of baseline HRQOL on the HRQOL at follow up. Despite the longitudinal and some intervention studies demonstrated that PA predicts future HRQOL, we can not conclude a causal effect of PA on QOL given that the number of longitudinal and RCT studies is small and one study also found that HRQOL predicts PA later.

35The studies that assessed the effect of sedentary behavior on HRQOL support the finding that longer sedentary time is connected with decreased HRQOL in children and adolescents. The association between sedentary behavior and HRQOL is independent of the PA, gender, age and body weight among children and adolescents. Watching TV, use of computers and playing video games for more than two hours a day are significantly associated with lower HRQOL. The findings from the longitudinal studies showed that children and adolescents who spent greater time on sedentary activities during follow-up experienced worse HRQOL at follow up, suggesting a negative effect of sedentary behavior for future HRQOL. A dose-response relation between screen time and adverse QOL was reported in three observational studies. Children and youth who spent excessive amount of time on sedentary behaviors reported lower HRQOL in physical, mental and psychosocial health domains. For example, Finne et al., 2013 reported that there was a dose-response relation for most subscales of the KINDL-R except for the family domain among girls, and a dose-response relation for physical well-being and school domains among boys. The evidence of the adverse impact of sedentary behavior on mental/psychosocial health is consistent with a number of previous studies that focused on the outcomes of mental health in school-aged children and adolescents.

36Due to the methodological heterogeneities in the measurement of PA, sedentary behavior and HRQOl, we highlighted the results in this review from qualitative synthesis. We performed meta-analyses for five studies that used the same QOL measure of PedsQL, and similar assessment methods for PA and sedentary behavior. The findings from the pooled analysis support a beneficial effect of PA and a detrimental effect of sedentary behaviors for HRQOL among children and youth. Future research would warrant to examine the magnitude of the difference in QOL from meta-analysis by inclusion of more studies that utilize similar measures of PA, sedentary behavior and HRQOL. Specifically, consistent measures for physical activity and sedentary behavior across studies are required to assess intensity and frequency of the health-related behaviors to facilitate the inter-group comparisons.

37The strength of the associations between HRQOL and PA or sedentary behavior in this study was largely based on judgement from statistical significance of the difference in QOL outcomes. When comparing the difference in HRQOL values across groups by PA and sedentary behavior, it is also important to examine the extent of the score difference (e.g.,effect size) in order to determine the minimally important difference (MID) of QOL scores as the MID signifies a meaningful difference that has clinical and practical importance for public health and health interventions. Based on Cohen’s criteria, effect size for HRQOL is defined as small (0.2), moderate (0.5), and large (0.8) in the magnitude for differences between groups. In our review, the MID or effect size for group comparisons in QOL was discussed or reported in eleven studies. The MID criterion varies with HRQOL instruments, and is more frequently used in clinical settings to compare HRQOL among patients with chronic disease conditions than in healthy populations. For example, the study by Omorou et al. indicated that the difference in HRQOL scores between active and inactive living adolescents during two years of follow-up was close to or greater than 5 points (considered as clinically meaningful difference) for all the four dimensions of the Duke Health Profile: physical, mental, social and general health. Vella,et al.(2014) reported that children who maintained participation in sports throughout a 2-year follow up period had an approximately 5 units higher total PedsQL score at follow-up than children who did not participate in sports. A 4.5-unit difference in HRQOL between group comparisons was considered as the MID value for the PedsQL.

38There are several strengths of this review. The present study is the first to review the associations between physical activity, sedentary behavior and HRQOL among studies using population-based samples of children and adolescents. We conducted a comprehensive literature search and did rigorous selection and assessments for the eligible studies using predefined inclusion and exclusion criteria. We included both PA and sedentary behavior as the predictor variables for HRQOL which allowed us to examine the independent effect of the two health-related behaviors on HRQOL. The majority of the included studies analysed data from large samples of children and adolescents. This offered good opportunities for the analysis to adjust for the effect of important confounders such as body weight, age, gender, parental education and household income. Thus the systematic review contributes to the existing literature for the associations between PA, sedentary behavior and HRQOL which are independent of body weight and socio-demographic characteristics of children and youth. Consistent associations between PA, sedentary behavior and HRQOL were reported across studies in different countries, suggesting that these results are robust. The outcome was measured by a variety of generic multi-dimensional HRQOL measures that have been validated for pediatric and youth populations, allowing for investigation of the effect of PA, sedentary behaviors on a single dimension of QOL as well as the overall QOL. Most primary studies applied multivariable regression analyses and some used multilevel regressions to account for hierarchical feature of school data, and thus providing appropriate and rigorous statistical methods for parameter estimates.

39Limitations of this review also deserve to be clarified. Most of the included studies are cross-sectional, and very few longitudinal and intervention studies are identified. Informative conclusion can not be made about causal relationship between PA, sedentary behavior and HRQOL. PA and sedentary behavior assessments in the included studies were largely based on child and youth self-report or parent-report, and thus may have affected the measurement error. The use of objective measures of PA (e.g., pedometer) and sedentary behavior (e.g., accelerometer, inclinometer or screen use monitor) is needed to make more accurate assessments of these behaviors. In addition, the citation and abstract screening for selection of the eligible studies was done by a single reviewer (XYW) rather than use of two reviewers. However, we made alternative efforts to ensure the study selection is intensive, rigorous, complete and accurate to meet the study goal. These work included the use of predefined inclusion and exclusion criteria, the evaluation of the full-text articles of potetial eligible studies by the two reviewers (XYW, LHH), and comprehensive screening for PubMed-related articles and reference lists of the relevant studies to identify other eligible studies.

40Children and adolescents spend more time engaging in sedentary activities than a decade ago due to the increasing use of screen-based electronic devices (e.g., smart phones, laptops) and widespread accessibility to the internet. Many countries have so far developed physical activity and/or sedentary behavior guidelines for school-age children and adolescents in order to improve their health. WHO suggests that children and youth aged 5–17 years old should accumulate at least 60 minutes of moderate to vigorous-intensity of physical activity everyday. Yet majority of the young people in many countries do not meet the recommendations of PA levels. HRQOL comprises multifaceted aspects of health. The importance of PA and sedentary behavior for QOL is that promoting PA and decreasing sedentary behaviors among young people may benefit not only to a specific health condition (e.g., obesity) but also to their mental health and overal health status. Therefore, it is worth to further investigate to determine the appropriate amount of PA, time spent in sedentary behaviors for HRQOL improvement. For exampple, whether a threshold exists for having no additional benefits on total or a dimension of HRQOL when PA or sedentary time exceeds a certain amount. As PA and sedentary behavior can be measured and expressed in different types (e.g., objective or subjective), different aspects (e.g., leisure time PA, TV or computer use) and different units (e.g., frequency, intensity and duration), future research is needed to investigate which types or aspects of physical activity and sedentary behavior would be the most significant contributors for QOL. In the meantime, to facilitate comparisons using quantitative synthesis of different studies, standardized measures as well as the objective measures of these health-related behavioral indicators for QOL are needed. Future studies are also needed to examine prospectively how changes in these behaviors affect changes in HRQOL among children and adolescents. This will provide useful information for evalations of the effectiveness health promotion programs targeting to modify unfavorable inactive and sedentary behaviors.

41The present review has found the evidence that a higher level of physical activity and less time spent on sedentary behavior are associated with increased health-related quality of life among the general population of children and adolescents. Future research is needed to identify potential causal mechanisms for these relationships. More longitudinal and cluster-randomized controlled trials are required to assess the dose-response effect of physical activity and sedentary behavior on health-related quality of life among children and adolescents. This will help justify school health intervention efforts promoting active lifestyle, reducing sedentary behaviors to enhance quality of life of the young population. The findings in this review may be used as evidence to inform primary prevention and public health policy for promoting the health of children and youth.