[Paper Review] COVID-19 Related Mobility Reduction: Heterogenous Effects on Sleep and Physical Activity Rhythms

Ju Lynn Ong, T.Y. Lau+8 more|arXiv (Cornell University)|Jun 3, 2020

Sleep and related disorders15 references18 citations

TL;DR

This study analyzes wearable device data from 1,824 working adults during COVID-19 mobility restrictions, revealing heterogeneous impacts on sleep and physical activity rhythms. Despite a 38% average drop in physical activity, sleep duration increased by 24 minutes on weekdays with improved sleep efficiency and resting heart rate, while a subgroup of younger, single, less active individuals experienced the largest declines, highlighting distinct behavioral responses to lockdown.

ABSTRACT

Mobility restrictions imposed to suppress coronavirus transmission can alter physical activity (PA) and sleep patterns. Characterization of response heterogeneity and their underlying reasons may assist in tailoring customized interventions. We obtained wearable data covering baseline, incremental movement restriction and lockdown periods from 1824 city-dwelling, working adults aged 21 to 40 years, incorporating 206,381 nights of sleep and 334,038 days of PA. Four distinct rest activity rhythms (RARs) were identified using k-means clustering of participants' temporally distributed step counts. Hierarchical clustering of the proportion of time spent in each of these RAR revealed 4 groups who expressed different mixtures of RAR profiles before and during the lockdown. Substantial but asymmetric delays in bedtime and waketime resulted in a 24 min increase in weekday sleep duration with no loss in sleep efficiency. Resting heart rate declined 2 bpm. PA dropped an average of 38%. 4 groups with different compositions of RAR profiles were found. Three were better able to maintain PA and weekday/weekend differentiation during lockdown. The least active group comprising 51 percent of the sample, were younger and predominantly singles. Habitually less active already, this group showed the greatest reduction in PA during lockdown with little weekday/weekend differences. Among different mobility restrictions, removal of habitual social cues by lockdown had the largest effect on PA and sleep. Sleep and resting heart rate unexpectedly improved. RAR evaluation uncovered heterogeneity of responses to lockdown and can identify characteristics of persons at risk of decline in health and wellbeing.

Motivation & Objective

• To understand how mobility restrictions during the COVID-19 pandemic affected sleep and physical activity rhythms in working adults.
• To identify subgroups with distinct behavioral responses to lockdown using clustering of rest-activity rhythms (RARs).
• To examine the role of social cues and mobility reduction in shaping changes in sleep and physical activity patterns.
• To uncover predictors of vulnerability to health-related behavioral decline during pandemic-related restrictions.

Proposed method

• K-means clustering was applied to temporally distributed step counts to identify four distinct rest-activity rhythm (RAR) profiles among participants.
• Hierarchical clustering was used to group individuals based on their mixture of RAR profiles before and during lockdown.
• Wearable device data covering baseline, incremental movement restriction, and full lockdown periods were collected across 206,381 nights of sleep and 334,038 days of physical activity.
• Sleep duration, sleep efficiency, resting heart rate, and physical activity levels were quantified and compared across time periods and subgroups.
• Statistical analysis assessed changes in sleep and activity patterns, with a focus on weekday versus weekend differences and subgroup heterogeneity.

Experimental results

Research questions

• RQ1 How did COVID-19-related mobility restrictions affect sleep duration and sleep efficiency in working adults?
• RQ2 What are the distinct rest-activity rhythm (RAR) profiles observed in urban working adults during lockdown?
• RQ3 How did physical activity levels change across different subgroups, and what factors predicted greater declines?
• RQ4 To what extent did the removal of habitual social cues during lockdown influence sleep and activity patterns?
• RQ5 Which demographic and behavioral characteristics predicted worse outcomes in sleep and physical activity during mobility restrictions?

Key findings

• Weekday sleep duration increased by 24 minutes during lockdown, with no decline in sleep efficiency, indicating improved sleep consolidation.
• Resting heart rate decreased by 2 beats per minute, suggesting improved autonomic regulation during the lockdown period.
• Physical activity levels dropped by an average of 38% across the cohort, with the most significant reductions observed in a subgroup comprising 51% of participants who were younger and predominantly single.
• Three of the four identified RAR groups maintained better differentiation between weekday and weekend physical activity patterns, while the least active group showed minimal distinction and the largest decline.
• The removal of habitual social cues during lockdown had the largest impact on physical activity and sleep patterns, more so than other mobility restriction levels.
• Heterogeneous responses were uncovered through RAR evaluation, identifying a vulnerable subgroup at risk of declining health and wellbeing due to reduced physical activity and disrupted rhythms.