Children's activity and diet behaviours in the summer holidays versus school year

Summary Background Evidence shows children gain more weight during the summer holidays versus the school year. Objectives To examine within‐child differences in activity and diet behaviours during the summer holidays versus the school year. Methods Children (mean age 9.4 years; 37% male) wore accelerometers (GENEActiv; n = 133), reported activities (Multimedia Activity Recall for Children and Adolescents; n = 133) and parents reported child diet (n = 133) at five timepoints over 2 years capturing school and summer holiday values. Mixed‐effects models were used to compare school and summer holiday behaviours. Results Children spent less time in moderate‐ to vigorous‐physical activity (−12 min/day; p = 0.001) and sleep (−12 min/day; p < 0.001) and more time sedentary (+27 min/day; p < 0.001) during summer holidays versus the school year. Screentime (+70 min/day; p < 0.001), domestic/social activities (+43 min/day; p = <0.001), self‐care (+24 min/day; p < 0.001), passive transport (+22 min/day; p = 0.001) and quiet time (+16 min/day; p = 0.012) were higher during the summer holidays, compensating for less time in school‐related activities (−164 min/day; p < 0.001). Diet quality was lower (−4 points; p < 0.001) and children consumed fewer serves of fruit (−0.4 serves; p < 0.001) during the summer holidays versus the school year. Conclusions Children are displaying poorer activity and diet behaviours during the summer holidays, which may contribute to accelerated weight gain over the holiday period.

(SMD = À0.25) and more screen time (SMD = À0.48, reverse-scored) on less structured days. Sleep timing (SMD = À1.05) and diet quality (SMD = À0.29) were also less healthy on less structured days. 5 Most included studies 5 examined differences in obesogenic behaviours on weekend days versus weekdays (k = 287), with just a few (k = 9) [6][7][8][9][10][11][12][13] comparing obesogenic behaviours during school holidays and school days, and only one specifically compared obesogenic behaviours in the summer holidays with school days. 14  To our knowledge, only two studies have examined all obesogenic behaviours (activity behaviours and diet) in a single study using a within-subject design. 17,18 Both found US children (predominantly low income, minority) spent more time sedentary (10-55 min/day) and slept longer (15 min/day in both studies) each day during the summer holidays compared with the school year. Results for physical activity and diet were mixed. Brazendale et al. 17 reported children consumed more sugar-based foods (6 days vs. 3.5 days/week) but MVPA did not differ during summer holidays compared with the school year. In contrast, Weaver et al. 18 reported children engaged in less light-intensity physical activity (LPA) (À42 min/day) and MVPA (À11 min/day) during summer holidays versus school but found no dietary differences. 18 These inconsistencies may be due to differences in activity behaviour measurement (Fitbit vs. ActiGraph accelerometry), different participant pools or different measurement periods (i.e., 9-day vs. continuous assessment).
While those studies 17,18 provide some insight into children's obesogenic behaviours in summer holidays versus the school year, several limitations should be noted. The findings from studies of low-income, minority children may not be generalizable to the population and the small sample sizes for children (n = 30) and schools (n = 1-3) limit the ability to draw confident conclusions. Further, studies have predominantly been conducted in the United States (75% or 3/4). Data from other regions are needed to understand the impact of summer holidays on children's obesogenic behaviours around the globe. The purpose of this study was to examine differences in obesogenic behaviours during the summer holidays versus the school year in a sample of primary-school-aged Australian children using a withinsubject study design.

| Study sample and design
This study used data from the Life on Holidays project, for which the full methods have been previously published. 19 Life on Holidays is a longitudinal cohort study that examines rates of change in fitness and fatness during the summer holidays and school year, and how rates of change in these outcomes relate to changes in activity and diet. 19 This study presents activity and diet behaviour data from summer holidays and school years. Twenty-four primary schools (21% uptake) located in different socioeconomic areas across metropolitan Adelaide, Australia, were recruited. From these schools, 381 participants (43% uptake) were recruited, though 23 did not commence baseline assessments (n = 1 moved school; n = 1 ineligible; n = 21 unknown reason),

| Measurements
Measurements commenced in Term 1 of 2019 (wave 1) and Term 1 of 2020 (wave 2). Activity and diet were measured at five timepoints ). This resulted in missing weight, fitness and accelerometry data for these children at these timepoints.   27 The ASA24 is an online tool for collecting 24-h dietary recall data using the seven-pass method (i.e., meal-based quick list, meal gap review, detail pass, forgotten foods, final review, last chance, usual intake) with digital photographic measures to aid portion size estimation. 28,29 Recalls were administered via phone by a trained research assistant. Parent/carers completed the recalls as proxy, with the child present where possible, recalling food and drinks consumed over the previous 24 h. Food group intake, energy and nutrient intake was estimated using the Australian Food Supplement and Nutrient Database (AUSNUT) 2011-2013. 30 Diet quality was assessed using the Dietary Guidelines Index for Children and Adolescents (DGI-CA), which provides a measure of adherence to the Australian Dietary Guidelines by children and adolescents. 31 Eleven indicators, which reflect diet variety, adequacy, quality and moderation, are combined to provide a score from 0 to 100, with lower scores indicating poorer dietary guideline compliance.

| Anthropometric assessment
Height and weight were obtained using a Seca 213 stadiometer (Seca, Hamburg, Germany) and InBody BIA scales (InBody USA, Cerritos, CFA). Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in metres (kg/m 2 ) and converted to BMI z-scores using the World Health Organization child growth standards. 32 The International Obesity Taskforce Criteria were used to categorize children as thin, normal weight, overweight or obese. 33,34 2.6 | Socioeconomic position Socioeconomic position (SEP) was reported by parents in a demographic questionnaire at baseline. Both parents reported their occupation, household income and highest education level. From these, a composite SEP z-score was derived and categorized as low, middle and high SEP, based on the procedure outlined in Gibbings et al. 35
Only children with valid data from at least two in-school timepoints and the summer holiday timepoint, as well as complete SEP, sex, BMI category and pubertal status data were included for analyses. Mixedeffects models were used to assess differences in activity and diet between the summer holidays and school years. Analyses were adjusted for participants nested in schools and in waves. Interactions between time (school vs. summer holidays) and subgroups (i.e., SEP, sex, pubertal status and BMI) were also explored. For significant interactions, modelled differences between school and summer holidays for each subgroup were calculated. Results from Stata's equality of standard deviation (variance) test using the sdtest command indicated that the assumption of homogeneity of variance was violatedvariance was greater during holidays than during the in-school periods. Stata's residuals (independent, by(timepoint)) command was used to accommodate the heteroscedasticity induced by the timepoint variable. The alpha level (0.05) was adjusted for multiple comparisons using Holm-Bonferroni adjustment, with adjustments made separately for MARCA, GENEActiv and diet assessments. No adjustments for multiple comparisons were carried out for the subgroup analyses (i.e., SEP, sex, pubertal status and BMI). Statistical significance was set at p < 0.05.

| RESULTS
There were different subsets of children included in each analyses.
One hundred and sixty nine children had complete demographic data AND at least one of the following: complete MARCA (n = 133); complete GENE (n = 133) and/or complete diet data (n = 128). Table 1 provides the demographic characteristics of participants. Additionally, characteristics of participants excluded from analyses (due to missing data for accelerometry, MARCA, diet or demographic data) are presented. Compared with included children, excluded children were similar in age and BMI category, but a greater proportion were male, and from lower SEP households, and slightly more were classified as early pubertal and slightly fewer as mid + pubertal. Table 2 shows the mean school and summer holiday values for activity and diet behaviours, along with the results of the mixed-effects model. Females reported more sleep (modelled difference = +6 min/day), had lower GENEActiv counts (modelled difference = À29 405 counts/day) and consumed less fruit (modelled difference = À0.9 serves/day) in the summer holidays compared with the school year. In contrast, males reported less sleep (modelled difference = À34 min/ day), had lower GENEActiv counts (modelled difference = À54 452 counts/day) and consumed less fruit (modelled difference = À0.2 serves/day) in the summer holidays compared with the school year.
The interaction between time and pubertal status was significant for carbohydrate intake (40; 95% CI: 6, 73). Early pubertal children consumed more carbohydrates (modelled difference = +21 g/day), whereas pre-pubertal children consumed less carbohydrates (modelled difference = À19 g/day) in the summer holidays compared with the school year.
T A B L E 1 Demographic characteristics of participants. more time in domestic/social activities, self-care, passive transport and quiet time. There were few differences in activity and diet behaviours in the summer holidays across sociodemographic characteristics.

| Implications
The findings of the current study align with previous studies that have found children spent more time on screens during the summer holidays compared with the school year. 14, [16][17][18]36 This finding is expected given the presence of the 6-h school day, which limits recre-   18,36 or that children consumed more fruit during the summer holidays. 17 The higher diet quality during school time is consistent with common school nutrition policy promoting low energy density foods such as fruit and vegetables. compared with days they do not attend. [40][41][42] The SDH also provides explanation for the observed changes in children's movement behaviours during home quarantine/lockdown imposed by the COVID-19 pandemic, with school closures inadvertently demonstrating the importance of structure in shaping children's movement behaviours and weight-related outcomes. 43

| Strengths and limitations
This is the first Australian study to compare children's activity and diet patterns in the summer holidays and school year. We recruited a representative sample of children across socioeconomic groups, and we collected data on all activity and diet behaviours of the same children in both the school year and summer holidays, allowing within-person comparisons. We also utilized valid and reliable measurements of activity and diet.
This study has limitations that should be considered when interpreting the findings. Data were missing because the COVID-19 pandemic commenced approximately halfway through data collection, with some schools not allowing further data collection. The COVID-19 pandemic also likely contributed to participant dropout, which was considerable. The included and excluded sample were broadly similar in age and BMI status, but differed in terms of SEP, pubertal status and sex. Another limitation was the lack of information on the context of children's days during summer vacation, such as whether they attended structured programming. Based on the SDH, children who attended structured programming such as vacation care may not experience unhealthy changes in obesogenic behaviours on those days. Future studies should track children's attendance at structured summer programming. The narrow age range of participants (age 9-11 years) and narrow geographical location (all participants were recruited from one Australian city) were further limitations.

| CONCLUSION
In summary, this study provides some evidence that children engage in less healthful activity and diet behaviours in the summer holidays, compared with the school year. These findings are consistent with the SDH, suggesting that the school day may regulate children's obesogenic behaviours and the absence of school may cause increased unhealthy behaviours. We recommend that screen time be a target for intervention in the summer holidays.

ACKNOWLEDGMENTS
The Life on Holidays study was funded by the National Health and