key: cord-0736795-nwctu9co
authors: Wright, Kenneth P.; Linton, Sabrina K.; Withrow, Dana; Casiraghi, Leandro; Lanza, Shannon M.; Iglesia, Horacio de la; Vetter, Celine; Depner, Christopher M.
title: Sleep in University Students Prior to and During COVID-19 Stay-at-Home Orders
date: 2020-06-10
journal: Curr Biol
DOI: 10.1016/j.cub.2020.06.022
sha: a3626f8d3979f82cf9762e5ae028fccfae2f9ff4
doc_id: 736795
cord_uid: nwctu9co

Sleep health has multiple dimensions including duration, regularity, timing, and quality [1-4]. The Coronavirus 2019 (COVID-19) outbreak led to Stay-at-Home orders and Social Distancing Requirements in countries throughout the world to limit the spread of COVID-19. We investigated sleep behaviors prior to and during Stay-at-Home orders in 139 university students (aged 22.2 ± 1.7 years old [±SD]) while respectively taking the same classes in-person and remotely. During Stay-at-Home, nightly time in bed devoted to sleep (TIB, a proxy for sleep duration with regard to public health recommendations [5]) increased by ∼30 min during weekdays and by ∼24 mins on weekends and regularity of sleep timing improved by ∼12 min. Sleep timing became later by ∼50 min during weekdays and ∼25 min on weekends, and thus the difference between weekend and weekday sleep timing decreased - hence reducing the amount of social jetlag [6,7]. Further, we find individual differences in the change of TIB devoted to sleep such that students with shorter TIB at baseline before the first COVID-19 cases emerged locally had larger increases in weekday and weekend TIB during Stay-at-Home. The percentage of participants that reported 7 h or more sleep per night, the minimum recommended sleep duration for adults to maintain health [5] - including immune health - increased from 84% to 92% for weekdays during Stay-at-Home versus baseline. Understanding the factors underlying such changes in sleep health behaviors could help inform public health recommendations with the goal of improving sleep health during and following the Stay-at-Home orders of the COVID-19 pandemic.

The COVID-19 pandemic has led to unprecedented changes in human behavior worldwide. We conducted an observational study to investigate changes in multiple dimensions of sleep health behaviors during the COVID-19 pandemic by comparing baseline sleep log data collected from January 29 to February 4, 2020 (before the COVID-19 outbreak spread across North America), to sleep log data collected in the same university students from April 22 to April 29, 2020, when the Stay-at-Home/Safer-at Home order was in effect. Baseline data collection was planned but not initially designed as a control condition for Stay-at-Home data collection. We used daily sleep logs to assess bedtimes and waketimes across each study week. Classes at the University of Colorado Boulder officially switched from in-person teaching to remote learning on March 16, 2020. Thirteen participants subsequently moved out of the local Mountain Time Zone (7 moved one time zone west, 5 moved one time zone east, and one moved two time zones east). Because students continued remote learning with classes scheduled according to Mountain Time, the sleep logs for all participants were analyzed according to Mountain Time. Of the 139 students, 98 were female and 6 did not report their sex. Institutional review board approval and written consent were obtained.

Outcomes included daily, weekday, and weekend TIB devoted to sleep, bedtimes, waketimes, and sleep midpoints-middle of the reported sleep opportunity-and regularity of sleep timing. Regularity was quantified by the standard deviations of bedtimes, sleep midpoint times and waketimes of each individual with lower scores indicating more regular sleep schedules. We also computed social jetlag-the difference between sleep midpoint on weekends versus weekdays [7]-and the percentage of individuals reporting ≥7h sleep per night.

Three dimensions of sleep health behaviors significantly changed during Stay-at-Home (Table  S1 , supplemental information): (1) TIB devoted to sleep increased on weekdays (Baseline=7.9±1.0h, Stay-at-Home=8.4±1.1h, p<0.0001) and weekends (8. 4±1.5h, 8.8±1.2h, p<0.05) during Stay-at-Home (Figure 1 panel A)-in fact, TIB increased every day of the week (p<0.05) except for Saturday (p=0.29; see supplemental information), and more participants reported the recommended 7h TIB [6] on weekdays during Stay-at-Home (92%) versus baseline (84%) (McNemar's χ 2 , p<0.001). Furthermore, participants with shorter TIB devoted to sleep at baseline, before the COVID-19 outbreak, increased their TIB more during Stay-at-Home ( Figure. 1B & 1C; Weekday r=0.51, p<0.000001; weekend r=0.70, p <0.000001); (2) Regularity of sleep timing, determined by the standard deviations of bedtime (1.2±0.6h, 1.0±0.6h, p<0.05), sleep midpoint (1.0±0.5, 0.8±0.5h), and waketime (1.2±0.5h, 1.0±0.5h, p<0.05) improved during Stay-at-Home-this is considered a positive change in behavior as prior findings show that irregular sleep schedules are associated with poor health and performance outcomes [4] [5] ;

(3) Sleep timing in general was later during Stay-at-Home versus baseline ( Figure 1A ). Specifically, average weekly bedtime (0018h±1h12min, 0048h±1h36min, p<0.001) and waketime (0824h±1h0min, 0906h±1h42min, p<0.000001) as well as weekday bedtime (p<0.001), and weekday and weekend waketimes (p<0.01) were later during Stay-at-Home. Weekend bedtimes were not statistically different (p=0.17). Furthermore, sleep midpoint was later for weekdays (0406h±1h0min, 0454h±1h36min, p<0.000001), weekends (0506h±1h18min, 0530h±1h42min, p<0.005), and for all days (p<0.05) during Stay-at-Home (Figures 1A & S1, supplemental information). Generally, later sleep timing is associated with poor health outcomes in adults [2] [3] 8] . Social jetlag was reduced during Stay-at-Home (0.9±1.0h, 0.6±0.9h, p<0.01); this is considered a positive change as findings from prior research shows that larger social jetlag is associated with poor health outcomes [7] [8] . Relationships between sunset and bedtime, and sunrise and waketime (Figure 1 , panel A) were similar to that previously observed [9] and if anything, sleep duration would be expected to be longest in the winter [10]; thus, although we did not measure light exposure, it is unlikely that seasonal changes in timing of sunset and sunrise contributed strongly to the observed changes in sleep.

Insufficient sleep duration, irregular and late sleep timing, and social jetlag are common in modern society. These poor sleep health behaviors contribute to and worsen major health problems, including cardiovascular disease, obesity, diabetes, mood disorders, and immune disorders [6] . Our findings provide further evidence that these poor sleep behaviors are modifiable. Additional research using objective measures such as wearable technology validated against polysomnography and markers of circadian phase are needed to better understand our observational data, to determine if our findings apply to the general population, and to identify which factors during Stay-at-Home orders-including public health recommendations and changes in social/work constraints-contribute to changed sleep health behaviors. Further, additional research is needed to assess the impact of experimental manipulation of sleep health behaviors on daytime function, wellbeing and health outcomes. Tuesday TIB could not be examined as we did not assess waketime on Wednesday. Percentage of students that reported 7h or more TIB per night was higher for Stay-at-Home on the weekend (87%, 93%) and on Wednesday (77%, 92%), Thursday (75%, 86%), Friday (76%, 89%), Saturday (81%, 87%), Sunday (69%, 83%), but was less on Monday (80%, 70%) (all McNemar's χ 2 , p<0.001). Individual differences show a large range of TIB on the weekdays (C) and weekend (D). Further, participants with shorter TIB devoted to sleep at baseline during the weekdays and weekend tended to increase their TIB during Stay-at-Home; whereas participants with longer TIB at baseline during the weekdays tended to either increase or decrease their TIB during Stayat-Home and during the weekend tended to decrease their TIB during Stay-at-Home. While some people reduced their TIB during Stay-at-Home, the overwhelming majority still maintained 7h or more of TIB (i.e., of those that reported 7h or more TIB at baseline only ~2% on weekdays and ~6% on weekends reported less that 7h TIB during Stay-at-Home Exploratory factor analyses show three separate dimensions of sleep health during the public health Stay-at-Home order: TIB devoted to sleep, sleep regularity, and sleep timing. Specifically, exploratory factor analysis of the change in sleep behaviors during Stay-at-Home shows sleep behaviors have a simple-to-understand factor structure with high loadings of sleep behaviors on a particular factor (in red bolded text) and negligible loadings on other factors (in black text): Factor 1-sleep timing, Factor 2-sleep time regularity, Factor 3-time in bed devoted to sleep.

Baseline data collection were conducted as part of a university class project and thus we took the opportunity to re-collect the same data during Stay-at-Home orders to describe changes in sleep behaviors in the student population tested. Thus, there is selection bias as we did not randomly sample our participants from the student population. Outcomes were analyzed with mixedmodel ANOVA with subject as a random factor and week (baseline versus Stay-at-Home) and sex as fixed factors. Sex was not significant and removed from the models. Change in sleep behaviors were analyzed with exploratory factor analysis, limited to factors with eigenvalues greater than one (Varimax rotation, normalized). McNemar's chi-square was used to test for a within-subject's analysis of participants obtaining 7h or more time in bed devoted to sleep (TIB) per night. Pearson correlations were performed between TIB at baseline and the change in TIB devoted to sleep between baseline and Stay-at-Home. Sensitivity analysis removed participants who moved to another time zone and results were the same. One hundred students had complete sleep log data on both visits and these participants contributed to the factor analyses and percent of people obtaining 7h or more time in bed devoted to sleep. Findings were similar if all participants were included in these analyses. Extreme outliers-three times the interquartile range-were identified for the waketime regularity variable and these participants were removed from the ANOVA analyses. Statistical analyses were conducted with Statistica 13 (StatSoft Inc.). Data reported are mean ± standard deviation.

We did not collect other relevant variables that may contribute to changes in sleep behaviors such as changes in people living in the home, extent of social isolation, being single or living with a partner/parents, changes in commute time to school and work demands, changes in napping or daytime sleepiness, changes in light exposure and physical activity, changes in electronic device and social media use, changes in caffeine and alcohol consumption, changes in stress and anxiety, and health status/exposure to COVID-19. The data presented in the paper are observational, the results are descriptive, and as noted, several relevant variables were not collected; thus, causal inferences should be made with caution. Further, the population of university students in Boulder Colorado, USA are not necessarily representative of the general student population nor the general population. For example, findings from prior studies show that the percentage of college students reporting less than 7h sleep per night was ~29% [S1], a higher percentage than found in our sample, and that average sleep per night based on actigraphy was ~6.8h per night [S2] . Further, according to public health databases, Boulder Colorado is ranked as one of the healthiest cities in the USA. With regards to sleep health specifically, data from the "500 Cities Project: Local Data for Better Health" [S3] a collaboration between the Centers for Disease Control (CDC), the Robert Wood John Foundation and the CDC Foundation, show that of the top 500 populated cities in the USA, Boulder Colorado has the lowest percentage of adults that obtain less than 7h sleep per night, at 24.5%. Thus, that we find improvements in TIB in this already healthy sleep population suggests it may be possible to achieve even larger percent increases of TIB devoted to sleep in other student and in other adult populations that have a higher prevalence of insufficient sleep duration.

Our findings also indicate lower levels of social jetlag in the student population studied here compared to findings from other studies. For example, ~one-third adults show 2h or more of social jetlag and ~69% 1h or more of social jetlag [S4] We find that during baseline, ~16% and ~44% of our study participants showed 2h or more and 1h or more of social jetlag, respectively; whereas during Stay-at-Home, ~7% and ~29% of our study participants showed 2h or more and 1h or more of social jetlag, respectively. Thus, the percentage of those with social jetlag for these cutoffs was significantly reduced during Stay-at-Home (McNemar's χ2, 2h p<0.0001, 1h p<0.01). We did not assess circadian phase or any questionnaire data to estimate chronotype. We provide however an analysis of sleep midpoint on weekends as a proxy for chronotype and show that sleep midpoint is correlated for baseline and Stay-at-Home weeks (r=0.60, p<0.001) such that those with later sleep midpoints at baseline still had later midpoints during Stay-at-Home; but there was no association between sleep midpoint and the change in sleep midpoint (r=-.14, p=0.16) suggesting that chronotype at baseline does not determine how much sleep midpoint changes under these unique circumstances.

K.P.W., S.B.L., L.C., S.M.L., H.O.D. designed the experiment; K.P.W., S.B.L., and S.M.L. collected data; all authors analyzed data/contributed to data analysis discussions and edited the paper; K.P.W. wrote the manuscript.

declare no competing interests. C.V., reports during the conduct of the study receiving research support from the NIH, was a scientific advisory board member of Circadian Light Therapy Inc. and Chronsulting, and served as a paid consultant to the US Department of Energy outside the submitted work. C.M.D. reports receiving research support from the NIH outside the submitted work. K.P.W. reports during the conduct of the study being a board member of the Sleep Research Society

Ethnoracial sleep disparities among college students living in dormitories in the United States: a nationally representative study

Irregular sleep and event schedules are associated with poorer self-reported well-being in US college students

Social jetlag and obesity