Physical activity and sleep changes among children during the COVID-19 pandemic

The study investigates how COVID-19-related disruptions to structured daily routines (in-person school and extracurricular activities) affected objectively measured physical activity (PA) and sleep among children with obesity. Childhood obesity is a significant public health problem, and PA and sleep are key modifiable behaviors. The structured days hypothesis suggests school-day structure supports healthier behaviors (e.g., built-in activity, consistent sleep/wake times). The pandemic’s sudden closures created prolonged periods of unstructured time, potentially undermining these behaviors. This study aims to quantify changes in step counts and sleep timing/duration before, during, and after pandemic-related closures, providing evidence specific to children with pre-existing obesity using longitudinal wearable data.

Prior research during the pandemic consistently reported decreased PA, increased sedentary time, and altered sleep in children, though many studies relied on self-reported data or short-term accelerometry. Reported PA declines ranged from 10 to 90 minutes/day, with reductions on both weekdays and weekends. Sleep changes included delayed bedtimes (~0.65 h), shifts in sleep midpoint (>2 h), reduced sleep efficiency (~2.09%), and high prevalence of sleep disturbances. Objectively measured studies (accelerometers and commercial wearables) showed declines in MVPA (e.g., −9.4 to −15.3 min/day) and step counts (e.g., −21% to −24% in special populations) over short windows (3 days to 6 weeks). Few studies focused specifically on children with obesity, and those often used subjective measures. There is limited research on step counts per se, making comparability challenging; however, reductions in steps of the magnitude observed here align with 7–15 minutes/day reductions in activity reported elsewhere. Sleep studies noted delayed sleep timing and sometimes increased duration; however, findings varied by context and methodology. This study addresses gaps by using longitudinal, objective, wrist-worn wearable data in a cohort with obesity across extended pre-, during-, and post-closure periods.

Design: Secondary data analysis of the Hearts & Parks (H&P) crossover randomized controlled trial (ClinicalTrials.gov NCT03339440), a clinic–community intervention to reduce pediatric obesity in Durham, NC. Participants were enrolled Feb 8, 2018–Mar 10, 2020; analysis window restricted to Mar 1, 2019–Jun 30, 2021 to align pre- and during-closure durations. Exposure periods: Pre-closure (Mar 1, 2019–Mar 14, 2020), During-closure (Mar 15, 2020–Mar 31, 2021), Post-closure (Apr 1, 2021–Jun 30, 2021), defined by NC stay-at-home orders and Durham school reopening. Participants: From 260 trial enrollees, inclusion for this analysis required sufficient valid wearable data across periods. After data processing, 94 children (55.3% female; median age 9.7 y) contributed to PA analyses: pre n=93, during n=53, post n=8. Wearable and measures: Water-resistant Garmin VivoFit 3 wristbands collected 15-minute epoch step counts and mean motion intensity (MMI; 0–7). Parents/participants synced weekly via Garmin Connect; data aggregated by Pattern Health. Wear-time determination: Because zeros were not reported, non-wear was inferred using MMI < 1. Valid day defined as > 41% wear (≥ 40 of 96 epochs; ≥ 10 h). Participants required > 60 valid days overall for inclusion. Only valid days were analyzed. PA metrics: Daily step counts summed across epochs with MMI ≥ 1, averaged per participant per month, then averaged across participants for each period. School-time PA analyzed for weekdays 7:00–16:00 during non-summer months (excluding June–August). Summer school-time analyses compared summers 2019 vs 2020; a post-closure summer subset reported descriptively. Sleep metrics: Garmin’s proprietary sleep epochs (in seconds) used to derive bedtime (earliest sleep epoch between 18:00 same day and 08:00 next day), waketime (latest sleep end in that window), and total sleep duration (sum of sleep epochs per day). Monthly per-participant averages were computed, then averaged across participants. Bedtime and waketime rounded to nearest 15 minutes. Summer values reported separately. Statistical analysis: Mann–Whitney U tests compared pre-closure vs matched during-closure months for average daily steps and sleep duration. Post-closure data (n=8; limited months) were descriptive only. Analyses conducted in Python 3.7.4 (Jupyter; Seaborn) within Duke PACE.

- Sample: 94 participants included for PA (55% female; median age 9.7 y). Pre n=93, during n=53, post n=8. - Overall daily steps: • Pre-closure: mean 8810 ± 453 steps/day (range 8239–9521). • During-closure: mean 7155 ± 669 steps/day (range 6354–8711); decline of 1655 steps (≈19%); p < 0.05. • Post-closure: mean 8763 ± 325 steps/day. - School-time (weekdays 7:00–16:00; non-summer months): • Pre-closure: 5441 ± 229 steps (≈61% of daily steps during school hours). • During-closure: 3468 ± 542 steps; decline of 1973 steps; p < 0.05. • Post-closure: 4616 ± 39 steps. - Summer school-time comparison (weekday 7:00–16:00): • Pre-closure summer 2019: 4888 steps (n=44). • During-closure summer 2020: 3351 steps (n=41); decline of 1537 steps; p < 0.05. • Post-closure summer: 4487 steps (n=4). - Sleep timing and duration (overall averages): • Bedtime: pre 10:45 pm ± 25 min; during 11:45 pm ± 32 min (summer: pre 11:30 pm ± 17 min; during 12:15 am ± 17 min). • Waketime: pre 7:15 am ± 16 min; during 8:00 am ± 11 min (summer: pre 7:30 am ± 10 min; during 8:00 am ± 3 min). • Sleep duration: decreased from 8.1 ± 0.17 h pre to 7.9 ± 0.22 h during (−12 min; −2.5%); p = 0.01. - Trends rebounded toward pre-closure with the resumption of in-person schooling/activities.

Objectively measured data show that COVID-19-related closures were associated with substantial reductions in PA, especially during typical school hours, and meaningful delays in sleep timing with a small decrease in sleep duration among children with obesity. The strong temporal alignment of declines with school closures and the rebound with reopening supports the role of structured school-day routines (e.g., active commuting, PE/recess, classroom transitions) in enabling daily movement and maintaining regular sleep schedules. Findings are consistent with global literature reporting reduced MVPA and delayed sleep during the pandemic, though prior work often relied on self-report or brief accelerometry windows. Given the critical role of PA and sufficient, regular sleep in pediatric obesity treatment and broader cardiometabolic and cognitive outcomes, the observed changes likely had adverse health implications for this vulnerable population. While causal inference is limited in this observational analysis, results underscore the importance of school- and community-based structures in sustaining healthy behaviors and inform preparedness for future disruptions.

Using longitudinal, objective wearable data, this study demonstrates that children with obesity experienced a ~19% reduction in daily steps, large delays in sleep onset and wake times, and a modest reduction in sleep duration during COVID-19 school/activity closures, with partial recovery upon reopening. The work highlights the centrality of structured daily routines for maintaining PA and sleep health in this population and provides objective evidence to guide parents, clinicians, educators, and policymakers. Future research should: (1) include larger and more diverse cohorts across regions; (2) extend follow-up to assess long-term trajectories post-reopening; (3) evaluate causal pathways linking remote schooling and reduced structure to behavior change; and (4) test interventions and policy strategies that preserve or replicate healthy structure during disruptions.

- Generalizability: Single-region cohort (Durham, NC) of children with obesity enrolled in an intervention trial; results may not generalize to all populations. - Study design: Secondary analysis; observational associations cannot establish causality. - Sample sizes by period: Reduced wear-compliant sample during-closure (n=53) and small post-closure group (n=8) limit precision of subgroup and post-closure estimates. - Wearable data constraints: Garmin’s proprietary algorithms; zeros not reported, necessitating MMI-based non-wear inference; potential misclassification remains. - Sleep measurement: Proprietary sleep detection may be affected by movement or watch wear; some nights had multiple sleep epochs. - Seasonality and schooling schedules: Although month-matched comparisons were used, residual confounding by unmeasured factors (e.g., varying local restrictions, household factors) is possible. - Intervention exposure: Some participants were in the intervention arm of H&P, which may influence behaviors independent of closures.