Introduction

Physical activity (PA) and its associated benefits to health are now well established. Examples of these benefits include skeletal health, obesity prevention (coupled with dietary intervention), psychological health and esteem, and prevention of Cardiovascular disease (CVD) risk factors [1–5].

The American College of Sports Medicine (ACSM) and the American Heart Association (AHA) have explicitly quantified the amounts and intensities of PA that should be accumulated for optimum health. During a week, adults are recommended to participate in aerobic PA (of a moderate intensity level) for at least 30 min, on at least 5 days, or 20 min of vigorous intensity cardiovascular exercise on 3 days (or a combination of the two) alongside eight to ten strength-based exercises twice per week [6]. In the UK, the Department of Health recommends that children take part in 60 min of PA per day on each day of the week and at least twice per week the activity should be muscle strength and flexibility based [7].

A positive association between PA and a higher perceived health-related quality of life (HRQoL) in adults has also been documented [8]. Some evidence exists that children who regularly participate in PA are more likely to report higher quality of life than those who had never participated [9]. However, PA and QoL studies in children are often confined to populations with a chronic condition or specific health problem. For example, Shoup and colleagues reported that physical, psycho-social and total quality of life scores were significantly lower in obese children compared to overweight children [10]. Further studies by Schwimmer, Friedlander and Pinhaus-Hamiel and colleagues all examined overweight compared to normal weight children and found normal weight children reporting higher quality of life scores [11–13].

Few studies are available that examine whether English secondary school pupils who meet the recommended guidelines for PA show any difference in self-reported QoL to those children not attaining the recommendations. Therefore, the main aim of this cross-sectional survey was to explore the relationship between self-reported physical activity and QoL (as measured by the PedsQL and EQ-5D) in English school children aged 11–15. Secondary objectives were to investigate the relationship between self-reported dietary intake, Body Mass Index (BMI) and QoL.

Methods

Design and setting

Four comprehensive secondary schools were matched according to characteristics described in the Office for Standards in Education (Ofsted) reports; two were in the northwest of England, two in the southwest. The schools were selected on the basis of a close match in examination results, percentage of children on free school meals and percentage of children with special educational needs (SEN). The participating schools were part of a cross-sectional study examining PA, diet and QOL. Questionnaires were completed in class in the presence of a teacher and the same questionnaires were completed twice, once in the summer term and once in the winter term.

University of Sheffield research ethics committee approval was obtained for this study, and the Local Education Authority was consulted in order to gain initial contact with the secondary schools involved. Consultation with the heads of the secondary schools followed this. Initially, the study details were circulated in a school newspaper, which every parent receives, at each school. After this process, the whole school populations were given an information letter to take home with a consent slip to be returned by a parent or guardian. As children were of secondary school age, it was thought they could sign consent to fill out a survey on the day of the study, if a parental slip had not been returned [14].

Participants

Two thousand eight hundred and fifty-eight pupils aged 11–15 in four secondary schools in England (2 in the northwest (NW) and 2 in the southwest (SW) region) were sent a letter with consent slip attached explaining the survey study and invited to participate in an anonymous survey on two occasions (once in the winter and again in the summer). The participating children had to give written consent to take part in the study. Eight hundred and sixty-nine children (869/2858 or 30%) responded to the winter survey and had valid self-reported physical activity data, and 35% (1000/2858) responded to the summer survey, an overall response rate of 33% (1869/5716) (see Table 1). Of these, 1,771 also completed the PedsQL and had valid QoL and physical activity outcome data which were analysed in this study.

Table 1 Survey response rates and school characteristics
Full size table

There were no differences between the children who completed the QoL assessments and declined to complete the QoL assessment, on self-reported physical activity, BMI, fruit intake, fat intake, sex and receipt of free school meals. The only significant differences were that those who did not complete the QoL assessments were more likely to be younger (mean of 12.2 vs. 13.2 years of age); more likely to be at school in the SW (6.8 vs. 4.4%) and of white ethnicity (6.6 vs. 3.1%). As this was an anonymous survey, no information was collected on the characteristics of the non-respondents to the questionnaire; therefore, a comparison of respondents to non-respondents cannot be made. Also, since this was an anonymous survey, we have no information on how many children completed the survey twice once in the winter and again in the summer.

Measures

Self-reported demographic information was collected from the pupils such as age, sex, ethnicity, entitlement to free school meals, height and weight.

Physical activity

The self-completed Western Australian Child and Adolescent Physical Activity and Nutrition Survey (CAPANS) questionnaire was used to assess physical activity [15]. The CAPANS consists of 24 questions and was first successfully used in a sample of 2274 children aged 7–16 in 2003 [16]. The CAPANS asks children to select the type of physical activity (from a comprehensive list of activities), they usually do in a typical week; and then to record the number of times they did the activity and the time spent on that activity. The total time spent on physical activity per week was calculated by totaling the time children spent in moderate or vigorous activity per day and then dividing this figure by seven to give an average for the week. The UK government recommends children are physically active at a moderate intensity for 60 min per day [7]. Therefore, the physical activity data were further classified into whether or not children were meeting the recommendations or not.

Diet

The Block food intake screener [17] was used to assess diet and the intake of fat, fibre, fruit and vegetables. The responses to the screener can then be used to estimate the amount of fat, fruit and vegetables a child is consuming in their diet. Fat and fruit intake for each child was then further classified as achieving the optimal of fats (<35% of calories consumed per day) or fruit and vegetable (at least 5 portions per day) or not.

Body mass index (BMI)

The pupils estimated their own height and weight which was then used to calculate each child’s BMI. UK-specific BMI reference values and cut-off points were then used to classify each child as having normal weight or being overweight or obese [18, 19].

Quality of life

The Pediatric Quality of Life Inventory (PedsQL version 4.0) and European Quality of Life 5 Dimension measure (EQ-5D) were used to assess QoL [20, 21]. The 23-item PedsQL instrument is designed to measure QoL in children aged 4–18 and includes four QoL scales of (1) Physical Functioning (PF- 8 items), (2) Emotional Functioning (EF–5 items), (3) Social Functioning (SF- 5 items) and (4) School Functioning (Sch F 5 items). Two further scales can be created a Total scale and a Psycho-social health summary score. The Psychosocial health summary score is computed as the sum of the items over the number of items answered in the Emotional, Social and School functioning scales. The Total scale score is computed as the sum of all the items over the number of items answered on all of the scales. Responses to the items are scored and transformed to a 0–100 scale, so that a higher score indicates better QoL [21].

The six-item EQ-5D (previously referred to as the EuroQol) is a generic quality of life instrument, designed to assess health outcomes. We used the youth version, EQ-5D-Y, which has been especially adapted for children [22, 23]. It was divided into two sections; section one addresses mobility, self-care, usual activities, pain/discomfort, and anxiety/depression, which are each assessed by a single question on a three-point ordinal scale (no problems, some problems, extreme problems). An EQ-5D ‘health state’ is defined by selecting one level from each dimension. A total of 243 health states are thus defined. Values or preference weights for a sample of these health states were obtained from a general community sample using a time-trade-off (TTO) technique [24]. Estimates for all health states were extrapolated from this sample by statistical regression modelling. The EQ-5D preference-based measure can be regarded as a continuous outcome scored on a −0.59 to 1.00 scale, with 1.00 indicating ‘full health’ and 0 representing dead [24]. The negative EQ-5D scores represent certain health states valued as worse than death. The sixth item consists of a 100-point Visual Analogue, which asks responders to rate their overall health today on 0 (worst possible health) to 100 (best possible health) scale.

Statistical analyses

We used statistical methods to analyse the QoL outcome data as described in Walters [25]. The association between the QoL outcomes and BMI and minutes of physical activity per day was examined using Pearson correlation coefficients. Two independent sample t-tests were used to compare mean QoL scores between those children meeting or not meeting the recommended guidelines for physical activity; fat dietary consumption; fruit and vegetable dietary consumption and weight status. Finally, a multiple linear regression analysis was used to compare QoL outcomes between the above groups and allow for the potential confounding factors of age, sex (male vs. female), ethnicity (white vs. non-white), receipt of free school meals (yes/no), and area (NW/SW). Ninety-five per cent confidence intervals for the mean difference in QoL scores between the groups are reported for the unadjusted and adjusted analyses. This was an anonymous survey and we have no information about whether or not the children completed the survey twice once in the winter and again in the summer. Therefore, the majority of the statistical analyses were performed and reported separately by season. To interpret the mean differences, we assumed a minimal important difference (MID) of 4.5 points for the PedSQL dimensions [26] and 0.07 for the EQ-5D utility score [27]. A P-value of less than 0.05 was regarded as statistically significant. SPSS version 14.0 was used for analysis of the data.

Results

Table 2 shows the demographic, QoL and PA levels of the responders to the survey. Just over half (51.7%) of the recruited children were boys, and the average age of the participants was 13.2 years (SD 1.2). Forty per cent of the participants were non-white. In addition, approximately, 25% of the participants were meeting the recommended guidelines for physical activity, and 23.5% of participants were classified as overweight/obese.

Table 2 Demographic characteristics of sample by season
Full size table

Table 3 shows the correlations between BMI, PA and QoL. The correlations between QoL and PA; and QoL and BMI suggested a very weak relationship (r < 0.20). The strongest correlations were for the intra-dimension correlations of the PedsQL.

Table 3 Correlations between physical activity, BMI and quality of life
Full size table

Table 4 shows that there were no statistically significant differences between the two groups of children, those who achieved the physical guidelines and those who did not, on any of the dimensions of the PedsQL and the EQ-5D utility score. When a multiple linear regression model was applied, to adjust the comparison between the groups for age, gender, ethnicity, free school meals and area, there was also no significant difference between those achieving the PA recommendation and those who did not on any of the dimensions of the PedsQL and the EQ-5D utility score. The EQ-5D VAS scores for summer and winter showed those achieving the 60 min of PA per day recommendations reported significantly better scores than those who did not achieve the recommendations. However, the observed differences in EQ-5D VAS scores between the groups were less than four points suggesting that these differences are small in magnitude and may not be of any clinical or practical importance.

Table 4 Mean scores of QoL dimensions by physical activity status and season
Full size table

Table 5 shows the mean QoL scores by weight status. Statistically significant differences were observed between the normal and overweight/obese groups for the PF, EF, SF, PHSS and Total dimensions of the PedsQL and the EQ-5D VAS in summer and winter, with the normal weight group reporting better QoL. These differences remained after adjustment for covariates. However, the observed differences in PedsQL and EQ-5D VAS scores between the groups were generally between four and five points, around the MID of 4.5 points for the PedsQL, suggesting that these differences are potentially of some clinical or practical importance.

Table 5 Mean scores of QoL dimensions by weight status by season
Full size table

Table 6 shows the mean QoL scores by dietary fat consumption status. Statistically significant differences were observed between the optimal (< 35% of daily calorie intake in fats) and fat intake too high groups only for the EF dimension of the PedsQL in the winter survey and this difference remained after adjustment for covariates, with the optimal fat intake group reporting better QoL. However, the observed difference in EF scores between the two groups was 3.9 points, less than the MID of 4.5 points for the PedsQL, suggesting that this difference is small in magnitude and may not be of any clinical or practical importance.

Table 6 Mean scores of QoL dimensions by fat dietary consumption status by season
Full size table

Table 7 shows the mean QoL scores by dietary consumption of fruit and vegetables. Statistically significant differences were observed between the optimal consumption (5 or more portions of fruit and vegetables per day) and not optimal consumption groups for the EF, SF, PHSS and Total dimensions of the PedsQL in the winter survey. Those who achieved the optimal consumption reported poorer QoL and these differences remained after adjustment for covariates. In the summer survey, the pattern was less clear with the only statistically significant differences being observed between the optimal consumption (5 or more portions of fruit and vegetables per day) and not optimal consumption groups for the PedsQL SF and EQ-5D VAS dimension. Those who achieved the optimal consumption reported poorer QoL. However, all of these observed differences were small, less than the MID of 4.5 points for the PedsQL suggesting that these differences are small in magnitude and may not be of any clinical or practical importance.

Table 7 Mean scores of QoL dimensions by fruit and vegetable dietary consumption status and season
Full size table

Discussion

The correlations observed in this study indicate little or no relationship between self-reported QoL, BMI and moderate to vigorous PA. We also found no statistically significant differences between the two groups of children, who achieved the recommended PA guidelines and those who did not, on any of the dimensions of the PedsQL and the EQ-5D utility score. Only on the EQ-5D VAS score was there a statistically significant difference between the groups.

This difference on the EQ-5D VAS dimension is the only evidence from this study which agrees with several reports on adult physical activity and QoL [28–32] and studies on children examining obesity, QoL and PA [9, 10]. It may be that our measures of QoL (the PedsQL and EQ-5D) in this relatively healthy group were not sensitive enough to detect differences between the more active and less active children. Or it could be that at this younger age, the differences of being active or not may not yet have impacted on the pupils’ health [33]. Wendel-Vos and colleagues [33] found some cross-sectional associations between leisure time activity and physical components of QoL, whereas longitudinal associations were predominantly observed for mental components of QoL. This shows that there is possibly a beneficial effect of PA on QoL over a longer sustained period of time.

While physical activity showed little relationship to QoL and diet showed some relationships; children who self-reported a BMI of overweight to obese (according to UK cut-points [18]) had significantly lower QoL on both dimensions of the EQ-5D and every dimension of the PedsQL apart from the School Functioning dimension. These findings lend further support to an existing evidence base of overweight/obese children reporting lower quality of life. De Beer and colleagues found in their study of 31 obese adolescents to 62 normal weight 12- to 18-year-olds that the obese subjects reported significantly lower PedsQL dimension scores compared to the normal weight subjects [34]. Friedlander’s study, using a different measure of QoL in a younger (8–11 years) group of children also found similar results [12]. Of the 371 children involved in this study, the overweight children had increased odds of lower scores on various health-related quality of life dimensions. Similarly, Schwimmer and colleagues found that overweight children were five times more likely to report low QoL scores when compared to healthy weight children [11]. Several other studies have similarly shown that overweight or obese children and adolescents report lower QoL scores in at least some if not all dimensions compared to healthy weight subjects [13, 35–38].

In terms of diet and QoL, this study found that those eating more fat reported a significantly worse emotional functioning than healthy eaters although conversely those whose fruit intake was optimal reported that their QoL was significantly worse than those eating less fruit and vegetables. However, both effects were small in magnitude and may not be of any clinical or practical importance. There are several studies which would also support the idea of a ‘healthy’ diet supplementing a higher QoL. However, most of these studies have been conducted in populations with specific diseases or conditions. In a trial investigating diet and its implications on hypertension, a controlled diet of the recommended intakes of vegetables and fat improved participant’s perception of their quality of life [39]. Hassan and colleagues in the examination of the BRFSS (Behavioural Risk Factor Surveillance System) data, in which 182,372 US adults participated, also reported that better diet supported by exercise in the overweight and obese participants was associated with better QoL [40].

The study has several strengths and limitations. We had a large sample of over 1,700 children self-reporting QoL and physical activity. We believe that the participants in this study represented an ethnically diverse cross section of the secondary school population that is broadly similar to many comprehensive secondary schools in England. The data were collected over the same period of time in all schools which should account for any differences in activity due to holiday periods or seasonality.

The generalisability of this study, to other schools and areas in England, is likely to have been affected by the low response rate of 33% (1869/5716). This study involved only four schools in two regions and is not a random sample of pupils or schools; so therefore, the results must be interpreted cautiously and cannot be wholly representative of other schools in the NW and SW or indeed England. The low response rate may have potentially caused a bias in the estimated differences in QoL between the various groups. Unfortunately, as this was an anonymous survey, no information was collected on the characteristics of the non-respondents to the questionnaire; therefore, a comparison of respondents to non-respondents cannot be made. Also, since this was an anonymous survey, we have no information on how many children completed the survey twice once in the winter and again in the summer. So, we cannot rule out that some bias may have been introduced into the sample. However, 23.5% of our survey participants were classified as obese which is similar to previous estimates for English school children aged 11–15 of 21%; although only 25% of our sample met the physical activity guidelines compared to around 52% reported nationally [41]. Our sample appeared to have similar levels for receipt of free school meals (18 vs. 16.5%) compared to school roll information [42].

We believe that the responders, to our survey, are more likely to be a well-motivated group of students, who are more likely to report higher levels of QoL, physical activity and better levels of diet and lower levels of BMI (due to overestimating their height and underestimating their weight). If this is so, then we believe that the results and observed differences between the groups are potentially likely to be smaller than the true differences as we have a self-selected sample of students who eat and exercise well and generally have a good QoL.

The cross-sectional design is less robust than a longitudinal study. So, it must be clearly acknowledged that the data represent a snap shot of information on physical activity and QoL. The diet questionnaire although validated and piloted by the authors has mainly been used in an adult non-UK population, which again may have an effect.

Pragmatically, we used self-report methods rather than objective measures such as pedometers or accelerometers to estimate physical activity. This may have resulted in an overestimate of activity particularly if this was a well-motivated group of students. However, the use of objective measures, such as pedometers, is not without problems as the use of these tools tends to alter the behaviour of people being observed and again may result in an over estimate of activity [43]. In general, the potential ‘bias’ of self-reporting survey methods for diet or PA is of concern to any researcher and the over-reporting in activity or under-reporting fat intake maybe a particular worry in studies of children [43]. However, in previous studies of children and PA, the self-report survey shows some promise of being a quality research instrument with a young population [44] and remains the most widely used measurement tool [45]. For practical reasons (since we had a large sample), we used the self-reported CAPANS instrument to assess PA, which has been shown to be a reliable and valid measure in children [16]. In several studies, self-reported BMI in children has shown that students tend to underestimate their BMI. Those students who are overweight or obese tend to underestimate their BMI to a greater extent than normal weight students. However, further studies have found that differences between self-reported and measured height and weight in young people were not statistically significant and there was reasonable agreement between actual and self-reported measurements [14, 46]. In other studies, which found differences, they reported that over 90% of adolescent participants estimated weight and height were in the correct BMI (Overweight/obese or Normal) classification group [47, 48].

Conclusion

Those children aged 11–15 achieving the recommended 60 min of moderate to vigorous physical activity per day had similar QoL (as assessed by the PedsQL) to those who did not achieve the recommended physical activity guidelines. In this sample, those reporting a normal BMI had better QoL outcomes on both the EQ-5D and PedsQL measures (apart from the School Functioning dimension) than overweight/obese children, thus confirming previous studies. Overall, this study showed mixed results for pupils achieving the recommended targets for physical activity and diet and their relationship with QoL. Hence, further study into PA and diet and their effects on QoL is needed.