1. Introduction

In 2020, coronavirus disease 2019 (COVID-19) became a global pandemic, and people’s lifestyle has since been dramatically changed to prevent contagious infections. In particular, school-aged adolescents had a long period of homeschooling via digital devices, since schools were closed during the COVID pandemic under government regulations. The combined effect of changes in lifestyle behaviors; confinement to the home through government restrictions on travel; and elevated depression, anxiety, and stress associated with the current COVID-19 pandemic, may have significant negative impacts on sleep.[1] This has been especially evident in healthcare workers, who may be required to work longer shifts in highly stressful environments.[2,3]

Recent evidence has shown that the COVID-19 pandemic induced changes in sleep habits among adults, especially in places where lockdown was adopted. Despite an increase in time in bed, poorer sleep quality was reported among adults.[4] Adolescents typically have a preference to sleep late. A delayed sleep phase,[5] use of electronic devices, and social life are associated with adolescents’ sleep behavior. A common consequence is sleep restriction and poor school performance.[6] During the COVID-19 pandemic, adolescents have had more flexibility with their schedules, which should help align with their sleep preferences. Online classes begin later than the usual in-person classes and there is no time spent commuting to school. The impact of the pandemic on sleep habits among adolescents has not been adequately characterized and compared with the period previous to the pandemic in adolescents. So, our study aim is to find the effect of COVID-19-related sleep behavior changes using school-based self-reported data from a nationally representative Korean adolescent population.

2. Methods 2.1. Study participants

We used cross-sectional data from the Korea Youth Risk Behavior Web-Based Survey (KYRBWS) which were using adjusted weighted values sampling method to represent Korean adolescents. In 2019, the KYRBWS was investigated from 3 June through 12 July: this period before the COVID-19 pandemic. In the 2020 KYRBWS, data were investigated from 3 August through 13 November; this period enables the mid-pandemic situation.

The aim of KYRBWS survey is to assess Korean adolescent`s health status and health behavior in order to provide basic data for Korean adolescents. KYRBWS survey was composed of a complex survey design with selection probabilities and post-stratification. All adolescents about 400-middle school and 400-high school students with the exception of those with absenteeism and dyslexia were eligible to participate. Since 2015, the ethics approval for the KYRBWS was waived Institutional Review Board by the KCDC Institutional Review Board. The participants investigated the self-administered questionnaires online in the school’s computer room and all participants gathered informed consent. Among the 112,251 total KYRBWS participants (57,303 in 2019; 54,948 in 2020: the following were excluded from this study participants without information sleep time (n = 14,125). Finally, 98,126 participants (51,651 in 2019 (before the COVID-19 pandemic lockdown); 46,475 in 2020 (before the COVID-19 pandemic lockdown) 12 through 18 years old were included in this study.

2.2. Socioeconomic and demographic factors

The self-administered questionnaires were used to provide information on socioeconomic factors (age, sex, school type [middle, high, or vocational school; southern, girls only, or coeducation school], family income, residence area, family structure, and academic achievement), health behavior factors (smoking status, alcohol drinking, regular exercise, sexual experience, and illicit drug use), psychological factors (self-rated health status, self-rated stress status, depression, suicidal ideation, suicidal plan, suicidal attempt) and co-morbidity (asthma, allergic rhinitis, atopy).

Regular exercise was determined as ≥3 times/week activity in the past 7 days.[7] Illegal drug experience was assessed by experiencing any of the following drugs (glue, butane gas, stimulants, marijuana, amphetamine, heroin, high-dose cold medicine, or anxiolytics for mood elevation, hallucinations, or diet excessively).[8] Smoking and alcohol use was defined as smoking cigarettes or drinking alcohol on >1 day over the last month.[9] Depression was defined using the Korean version of the World Health Organization Composite International Depression was defined using the following question which has been validated for health surveys such as our cross-sectional design[10]: “Have you experienced more than 2 consecutive weeks where you felt sad, blue, or depressed during the last year?.” Co-morbidity (asthma, allergic rhinitis, atopy) was defined as co-morbidity if participants have been diagnosed by a physician.

Suicidal ideation, suicidal plan, and suicidal attempts were assessed if they have ideation, plans, and attempts in last year. This indicator is a well-documented predictor of suicide behaviors that have been previously well-validated in another study.[11]

2.3. Sleep

Self-reported waking time and bedtime were investigated for each participant as weekday and weekend sleep duration. Average sleep duration was calculated as (5 × weekday sleep duration + 2 × weekend sleep duration)/7. Sleep time was divided into 5 subcategories: ≤5, 6, 7, 8, and ≥9 hours. Weekend catch-up sleep was calculated from the weekend sleep duration minus weekday sleep duration. Bedtime delay was calculated from weekday bedtimes minus weekend bedtimes. Long weekend catch-up sleep was calculated from weekend sleep duration minus weekday sleep duration of >2 hours.[12,13]

The average sleep time for participant’s weekdays and weekends was calculated based on the participant’s responses to the Korean version of the Munich Chronotype Questionnaire: “On a weekday (or school days) and on a weekend (school-free days), at what time do you usually go to bed for sleep and at what time do you usually get up?.” To define sleep chronotypes, we calculated the midpoint of sleep on school-free days (MSF), the midpoint of sleep on school days (MSW), sleep duration on school days (SDW), and sleep duration on school-free days (SDF) using the above sleep questions. Midpoint of MSF corrected for sleep extension on school-free days (MSFsc) was used as the chronotype indicator.[14] MSFsc was calculated using the following equation.[15] MSFsc = MSF − ([SDF − (SDW × 5 + SDF × 2)/7)/2. When SDF was shorter than or equal to SDW, MSFsc was the same as MSF. Chronotype was categorized as quintiles of the MSFsc: early chronotype (Q1, lowest MSFsc), intermediate chronotype (second [Q2], third [Q3], and fourth [Q4] quintiles), and late chronotype (Q5, highest MSFsc).[16] Social jetlag was calculated from the equation: social jetlag = MSF − MSW. Sleep satisfaction was defined by answering: “Have you ever had enough sleep to recover from fatigue in the last week?” The answer options using a 5-point Likert scale include more than enough, enough, moderate, not enough, and less than enough. Sleep quality was reclassified into 3 groups: enough (more than enough), moderate, and not enough (less than enough to not enough).

2.4. Data analysis

The general characteristics between the before COVID-19 pandemic lockdown and during COVID-19 pandemic lockdown in Korean adolescents were compared with chi-square test with complex sampling Rao–Scott correction, to represent the entire population, as this study was designed to use weighted values. Multiple logistic regression analysis with complex sampling adjusted for age, sex, school type, family income, residence area, family structure, and academic achievement, health behavior factors (smoking status, alcohol drinking, regular exercise, sexual experience, and illicit drug use, self-rated health status, self-rated stress status, depression, suicidal ideation, suicidal plan, suicidal attempt and co-morbidity (asthma, allergic rhinitis, atopy). P values < 0.05 were considered statistically significant. All data were analyzed using SPSS for Windows (version 21.0; SPSS Inc., Chicago, IL).

3. Results

The general characteristics of the included population are presented in Table 1. In 2020 (during the COVID-19 pandemic lockdown), the population was higher in age (P values < 0.001, Table 1). Health hazard behavior such as smoking, alcohol drinking, substance use, and sexual experience was significantly decreased in 2020 (during the COVID-19 pandemic lockdown) than in 2019 (before the COVID-19 pandemic lockdown). On the contrary, regular exercise decreased after the COVID-19 pandemic lockdown. Psychosomatic change after the COVID-19 pandemic lockdown showed positive results. School violence, depression, suicidal idea, suicidal plans, and suicidal attempts were significantly lower in 2020 (during the COVID-19 pandemic lockdown) compared to those in 2019 (before the COVID-19 pandemic lockdown), respectively (P < .001, Table 1). The differences in the reporting of physician-diagnosed allergic diseases between 2019 and 2020 were analyzed. The reporting of asthma was lower in 2020 (6.2%) than that in 2019 (7.1%, P < .001). The reporting of allergic rhinitis in 2020 was 35.0%, which was also lower than that of 35.4% reported in 2019, P < .001).

Table 1 - General characteristics of participants according to history of asthma. Before COVID-19 During COVID-19 P value (n = 51651) (n = 46475) Sex, n (%*) .120‡  Girl 25,061 (48.6) 22,318 (47.8)  Boy 26,590 (51.4) 24,157 (52.2) Age 14.9 ± 1.8 15.1 ± 1.8 <.001† School, n (%*) <.001‡  Middle school 26,222 (47.6) 24,144 (49.0)  Academic high school 20,334 (43.4) 17,812 (42.5)  Vocational high school 4684 (9.0) 4172 (8.5) School type, n (%*) .135‡  Southern school 8790 (17.1) 8129 (17.9)  Girl school 8752 (17.1) 7785 (16.5)  Coeducation 34,109 (65.8) 30,561 (65.6) Residence, n (%*) <.001‡  Rural 3198 (4.5) 2904 (4.7)  Urban 48,465 (95.5) 43,571 (95.3) Living, n (%*) <.001‡  Living without parents 2644 (4.4) 2221 (3.8)  Living with parents 49,007 (95.6) 44,254 (96.2) Family income, n (%*) <.001‡  Low 6675 (12.5) 6006 (12.4)  Medium 24,846 (47.9) 22,315 (47.5)  High 20,230 (39.6) 18,154 (40.1) Subjective academic achievement, n (%*) <.001‡  Low 16,101 (31.3) 14,931 (32.0)  Middle 15,625 (30.4) 14,131 (30.4)  High 19,925 (38.3) 17,413 (37.7) Smoking, n (%*) 2021 (6.0) 2566 (5.6) <.001‡ Alcohol, n (%*) 12,558 (24.6) 10,493 (22.7) <.001‡ Regular exercise, n (%*) 17,896 (33.9) 15,090 (31.5) <.001‡ Sexual experience, n (%*) 2782 (5.5) 1930 (4.2) <.001‡ Substance use, n (%*) 429 (0.8) 294 (0.6) <.001‡ Perceived stress, n (%*) <.001‡   Severe to very severe 9952 (19.0) 10,140 (21.5)   Moderate 21,244 (41.2) 20,856 (45.1)   None to mild 20,455 (39.8) 15,479 (33.4) Perceived health status, n (%*) .747‡  Healthy 26,273 (70.2) 32,185 (70.3)  Moderate 11,571 (22.6) 10,317 (22.4)  Bad 2707 (7.2) 3343 (7.33 School violence, n (%*) 1045 (2.0) 508 (1.1) <.001‡ Experiences of depressive mood for 2 or more continuous weeks, n (%*) 14,266 (27.8) 11,251 (24.1) <.001‡ Suicidal idea, n (%*) 6621 (12.8) 4789 (10.2) <.001‡ Suicidal plan, n (%*) 1919 (3.7) 1500 (3.2) <.001‡ Suicidal attempts, n (%*) 1447 (2.7) 828 (1.7) <.001‡ Allergic rhinitis, n (%*) 17,961 (35.4) 15,795 (35.0) <.001‡ Atopy, n (%*) 11,614 (22.6) 10,690 (23.3) <.001‡ Asthma, n (%*) 3579 (7.1) 2819 (6.2) <.001‡

COVID-19 = coronavirus disease 2019.

*Estimated mean or rate-adjusted recommended weighted value;

†linear regression analysis with complex sampling, significance at P < .05;

‡chi-square test with Rao–Scott correction, significance at P < .05.

Table 2 summarizes sleep parameters, which differed between the groups. During the COVID-19 pandemic lockdown, Korean adolescents slept less (≤5 hours: 26.6% vs 28.3%; P < .001), had a later weekend bedtime (≥1:00 am: 68.2% vs 71.5%), and late weekend wake time (≤7:00 am: 13.3% vs 10.7%) compared to before COVID-19 pandemic lockdown. Average sleep duration (434.7 ± 102.6 vs 428.2 ± 100.4 minutes; P < .001), weekday sleep duration (398.2 ± 103.2 vs 390.8 ± 103.7 minutes; P < .001), weekend sleep duration (525.7 ± 178.2 vs 521.5 ± 163.3 minutes; P < .001) were significantly lower in 2020 (during COVID-19 pandemic lockdown) compared to 2019 (before COVID-19 pandemic lockdown). On the contrary, weekend catch-up sleep duration (127.5 ± 173.7 vs 130.7 ± 162.4 minutes; P < .001) and weekend catch-up sleep >2 hours (42.1% vs 43.7%; P < .001) were significantly higher in 2020 (during COVID-19 pandemic lockdown) compared to 2019 (before COVID-19 pandemic lockdown). Mean sleep duration (434.7 ± 102.6 vs 428.2 ± 100.4 minutes; P < .001), MSF (262.9 ± 89.1 vs 260.8 ± 81.8 minutes; P < .001), MSW (262.9 ± 89.1 vs 260.8 ± 81.8 minutes; P < .001), SDW (398.2 ± 103.2 vs 390.8 ± 104.7 minutes; P < .001), and SDF (525.7 ± 178.3 vs 521.1 ± 121.5 minutes; P < .001) were significantly lower during COVID-19 pandemic lockdown, as was social jetlag (63.7 ± 86.8 vs 65.3 ± 81.2 minutes; P < .003). Late chronotype was significantly more common during the COVID-19 pandemic lockdown (17.1% vs 22.9 %, P < .001). Sleep satisfaction was significantly improved (sleep satisfaction enough: 21.0% vs 30.5%, P < .001) during the COVID-19 pandemic lockdown. Table 3 shows the adjusted odds ratio (OR) for the COVID-19 pandemic lockdown according to sleep duration and chronotype in the adolescent population. After adjusting for multiple confounding variables, short sleep duration (≦5 hours, OR 1.14; 95% confidence interval [CI] 91.10–1.19), 6 hours, OR 1.07; 95% CI 1.03–1.12) was significantly associated with COVID-19 pandemic lockdown compared to a sleep duration of 7 hours. After adjusting multiple confounding variables, long weekend catch-up sleep was significantly associated with COVID-19 pandemic lockdown (OR, 1.08; 95% CI, 1.06–1.11), and weekend “late owl” (weekend bedtime 1 ≥ am, OR, 1.27; 95% CI, 1.233–1.31) and weekend “early bird” (weekend wake time ≤ 7:00 am, OR, 1.29; 95% CI, 1.24–1.34) was significantly associated with an increased frequency of COVID-19 pandemic lockdown. After adjusting multiple confounding variables, the late chronotype was significantly associated with an increased frequency of the COVID-19 pandemic lockdown (OR, 1.43; 95% CI, 1.38–1.47) compared to the intermediate chronotype.

Table 2 - Descriptive statistics of sleep variables. Before COVID-19 During COVID-19 P value (n = 51651) (n = 46475) Sleep duration, average, min 434.7 ± 102.6 428.2 ± 100.4 <.001† Sleep time, n (%*) <.001‡   ≤ 5h 12,894 (26.6) 12,399 (28.3)   6h 11,255 (22.6) 10,676 (23.6)   7h 10,152 (19.4) 9235 (19.4)   8h 9397 (17.2) 7776 (15.8)   ≥9h 7953 (14.2) 6449 (12.9)  Weekday   Sleep duration, min 398.2 ± 103.2 390.8 ± 104.7 <.001†   Bedtime, n (%*) <.001‡   ≤21:00 769 (1.2) 721 (1.4)   22:00 4001 (6.9) 3316 (6.2)   23:00 9546 (17.0) 7428 (14.9)   24:00 11,465 (22.0) 9761 (20.5)   1:00 12,676 (25.2) 11,206 (24.7)   ≥2:00 13,294 (27.7) 14,043 (32.3)  Wake time, n (%*) <.001‡   ≦5:00 1645 (3.0) 1282 (2.6)   6:00 14,956 (28.5) 12,504 (26.3)   7:00 30,146 (60.0) 27,095 (58.3)   8:00 4004 (8.5) 5594 (12.8)   ≥9:00 0 (0) 0 (0) Weekend day  Sleep duration, min 525.7 ± 178.2 521.5 ± 163.6 <.001†   Bedtime, n (%*) <.001‡   ≤21:00 743 (1.3) 651 (1.3)   22:00 1881 (3.2) 1433 (2.8)   23:00 5703 (10.2) 4239 (8.5)   24:00 9061 (17.1) 7153 (14.9)   1:00 10,570 (20.7) 8605 (18.7)   ≥2:00 23,613 (47.5) 24,393 (53.8)  Wake time, n (%*) <.001‡   ≤5:00 586 (1.1) 449 (1.0)   6:00 1682 (3.1) 1200 (2.5)   7:00 4840 (9.1) 3397 (7.2)   8:00 10,252 (19.8) 7807 (16.8)   ≥9:00 34,291 (66.9 33,622 (72.5) Weekend-weekday difference  Sleep time difference 127.5 ± 173.7 130.7 ± 162.4 <.001†  Weekend catchup sleep, ≥2 h, n (%*) 21,467 (42.1) 20,137 (43.7) <.001‡  Bedtime delay 49.5 ± 124.7 55.4 ± 119.0 <.001†  Wake time delay 163.8 ± 137.1 179.5 ± 139.0 <.001† Midpoint of sleep on school-free days (MSF) 262.9 ± 89.1 260.8 ± 81.8 <.001† Midpoint of sleep on school days (MSW), 199.1 ± 51.6 195.4 ± 52.3 <.001† Sleep duration on school days (SDW) 398.2 ± 103.2 390.8 ± 104.7 <.001† Sleep duration on school-free days (SDF) 525.7 ± 178.3 521.5 ± 121.5 <.001† Chronotype <.001‡  Q1: early 11,222 (20.9) 8335 (17.4)  Q2: intermediate 11,021 (21.4) 8726 (18.9)  Q3: intermediate 10,494 (20.7) 8973 (19.6)  Q4: intermediate 10,052 (19.9) 9719 (21.2)  Q5: late 8862 (17.1) 10,722 (22.9) Social jetlag, min 63.7 ± 86.8 65.3 ± 81.2 <.001† Sleep satisfaction, n (%*) <.001‡  Enough 11,201 (21.0) 14,321 (30.5)  Moderate 16,864 (32.5) 16,015 (34.3)  Unenough 22,586 (46.5) 16,139 (35.2)