Effect of catch-up sleep on obesity in Korean adolescents: a nationwide cross-sectional study

Data from the 7th Korea National Health and Nutrition Examination Survey (KNHANES VII), conducted between 2016 and 2018, were analyzed. KNHANES has been conducted periodically since 1998 by the Korea Centers for Disease Control and Prevention and is a large cross-sectional representative survey of the health and nutritional status of the South Korean population. Informed consent was obtained from all participants for the survey, and for this study, data from adolescents who were middle school students and high school students were included. The institutional review board of Kangwon National University Hospital approved the study protocol (approval number KNUH-2022-10-001).

Of the 1,335 middle school students and high school students from KNHANES VII, those who had metabolic and endocrine diseases diagnosed by a doctor (N = 1) and those without records of anthropometric data (N = 3) were excluded. Participants were considered to be outliers if they: (i) had a calculated sleep duration on weekdays exceeding 14 h (N = 1) or <4 h (N = 9); (ii) had a calculated sleep duration on weekend exceeding 15 h (N = 1) or <4 h (N = 3); (iii) went to bed after 4 a.m. (N = 0); or (iv) woke up after 10 a.m. (N = 11) on weekdays, and were subsequently excluded. As such, the final number of study participants was 1,306 (Figure 1).

The KNHANES VII survey included anthropometric data for each participant, including age, sex, height, weight, and body mass index (BMI). A portable stadiometer was used to measure height to the nearest 0.1 cm, and a digital scale was used to measure weight to the nearest 0.1 kg, with the participants wearing light clothing and no shoes. BMI was calculated as weight (in kilograms) divided by the square of height (in meters), and standard scores (z-scores) for BMI were obtained for the same age and sex from the 2017 Korean Children and Adolescents Growth Chart. Underweight, overweight, and obesity were defined as a BMI at the 5th percentile or below, between the 85th and 95th percentiles, and at the 95th percentile or above, respectively, for the same age and sex. The four categories of weight status were then combined to create dichotomous variables (underweight/normal weight and overweight/obesity).

Adolescents were asked to report their sleep habits while at home on weekdays and weekends. The questionnaires used to assess sleep were developed for individuals aged 12 years and older by the Korea Centers for Disease Control and Prevention (). The variables of interest were: (i) sleep duration, (ii) bedtime, (iii) wake time, and (iv) sleep mid-time. Bedtime was defined as the time that the participants fell asleep at the end of the day, and wake time was defined as the time when they awoke from this sleep (i.e., daytime naps were not included). The values for bedtime and wake time were converted into minutes to calculate sleep duration and sleep mid-time, in addition to analyzing their effect on obesity as continuous variables. Bedtime and wake time were calculated from midnight; for example, when a participant went to bed at 10 pm and woke up at 7 a.m., it would be stated as bedtime at 22 × 60 = 1,320 min and waking up at 7 × 60 = 420 min. However, if the bedtime was after midnight, it was calculated by adding 24 h; that is, a person who began to sleep at 1 a.m. would have bedtime stated as (24 + 1) × 60 = 1,500 min. Total sleep duration (TSD) was calculated by considering bedtime and wake time. The TSD for the entire week was a weighted mean of sleep durations on weekdays and weekends, calculated as:Participants were classified into four groups as follows: (i) less than 6 h, (ii) 6–7 h, (iii) 7–8 h, and (iv) 8 h or more according to TSD per weekday. Supplementary Figure S1 ( × + × ) 5 2 7 TSD on weekdays TSD on weekends /

Weekend sleep extension (WK-E) was defined as the difference in sleep duration between weekdays and weekends and was calculated as TSD on weekend minus TSD on weekday. Participants were then classified into four groups based on WK-E as follows: (i) less than 1 h, (ii) 1–2 h, (iii) 2–3 h, and (iv) 3 h or more. To evaluate the effect of weekday and weekend sleep on TSD, the participants were classified into one of four sleep duration groups, using the median values for TSD on weekdays and weekends: (i) short sleep on both weekdays and weekends (SWD/SWK), (ii) long sleep on weekdays and short sleep on weekends (LWD/SWK), (iii) short sleep on weekdays and long sleep on weekends (SWD/LWK), and (iv) long sleep on both weekdays and weekends (LWD/LWK).

The covariate data collected from the lifestyle questionnaire included sex, physical activity, sedentary time, food intake, and household income. Physical activity was determined by the number of days the participant exercised at an intensity, which resulted in heavy breathing for >60 min in a week, both during and outside of school hours. The participants also reported sedentary time as the time (h and min) spent sitting or lying down each day during the past week. Food intake was assessed by categorizing foods into grains, bean products, potatoes, meat, fish, seaweed, fruit, vegetables, milk products, and beverages. Participants were required to recall the amount of each food type eaten during the previous 2 days. Nutritionists assessed this food intake and converted it into weight in grams and calories consumed. The cumulative income of the family was partially adjusted based on the number of family members and was divided into quintiles according to the average monthly household income.

Statistical analyses were performed using SPSS (Version 26.0; IBM Corp., Armonk, NY, USA) and SAS version 9.4 (SAS Inc., Cary, NC, USA). All tests were two-tailed, with a significance level set at 0.05. SPSS and SAS survey procedures were used to describe the complex sampling design in KNHANES. Sample weights were assigned to participants to represent the adolescent population of South Korea from 2016 to 2018 for all analyses. Sample weights were generated by accounting for the complex sample design, which consisted of non-response rates of the target population, multistage stratification, and posterior stratification. The normality of data distribution of variables was examined using univariate and histogram methods, and frequency tests were used to assess the relationships between variables.

The mean values of sleep dimensions and other variables were compared between the underweight/normal weight group and the overweight/obesity group using complex samples general linear model (CSGLM) for continuous variables and Rao–Scott χ² test for categorical data. The potential association between each continuous sleep variable and BMI z-score was investigated using CSGLM. Additionally, the relative risk of overweight/obesity conferred by each sleep variable was assessed using Complex Samples Poisson Regression Analysis. The mean BMI z-score according to TSD on weekdays and WK-E were also investigated using CSGLM. The mean BMI z-score and the risk of overweight/obesity among the four sleep duration groups were compared using CSGLM and Complex Samples Poisson Regression Analysis, respectively. Trends were tested to assess the decremental mean BMI z-score across the four groups, and the relative risk for the overweight/obesity group was calculated using the SWD/LWK group as the reference. Various models were developed for CSGLM and Complex Samples Poisson Regression Analysis. Model 1 was developed without adjustments for covariates. Model 2 was adjusted for physical activity, sedentary time, household income, total food intake (kJ), and frequency of breakfast intake. All data were reported as mean ± standard error (SE) or n (%).

Results of the linear regression models are summarized in Table 2. The sequential regression models for each sleep variable showed a significant negative correlation with the BMI z-score, as did the mean TSD for weekdays, weekends and whole week, particularly on weekends, even after adjusting for specific variables. Due to the positive correlations between sleep duration and wake-up time (weekdays, β = 0.707, p < 0.001; weekends, β = 0.061, p < 0.001), there was also a negative correlation between wake-up time and BMI z-score on both weekdays and weekends (Table 2). Moreover, the longer WK-E was negatively correlated with BMI z-score (Table 2).

Table 2 also presents the Poisson regression analysis between sleep variables and the relative risk of overweight/obesity. Importantly, longer TSD on weekends and longer WK-E decreased the relative risk of overweight/obesity. Each 30 min increase in sleep over the weekend reduced the relative risk of overweight/obesity by a factor of 0.39, regardless of covariates. In addition, as the WK-E increased by 30 min, the risk of overweight/obesity decreased by a factor of 0.93, following adjustment of the covariates. These results suggest that longer TSD on weekends and getting more sleep on weekends than weekdays may be associated with a reduction in the mean BMI z-score and relative risk of overweight/obesity.

To evaluate whether there was a difference in the effects of TSD between weekdays and weekends on overweight/obesity, the participants were divided into four sleep duration groups (SWD/SWK, LWD/SWK, SWD/LWK, and LWD/LWK) and compared. Supplementary Table S1 shows the characteristics of the four groups, while Figure 2 shows the mean BMI z-score and relative risk of overweight/obesity. Those who slept less on weekends had a higher BMI z-score than those who slept longer on weekends, and the mean BMI z-score was the highest in SWD/SWK, while the lowest in LWD/LWK (Figure 2). Moreover, despite LWD/SWK sleeping more during the whole week than SWD/LWK (7.6 h and 7.0 h, respectively, Supplementary Table S1), the mean BMI z-score was significantly higher in LWD/SWK than that in SWD/LWK (Figure 2). After adjusting the variables, the participants in the SWD/SWK group were 2.87 and 1.91 times more likely to show overweight/obesity than those in the SWD/LWK group and LWD/LWK, respectively (Figure 2). These results demonstrated that longer sleep on weekends is important to reduce the mean BMI z-score and the relative risk of overweight/obesity.

We evaluated how many sleep hour were required on weekends to reduce the relative risk of overweight/obesity. Participants with WK-E less than 2 h had an overweight/obesity risk 1.53 times [95% confidence interval (CI): 1.12–2.10] higher than participants with WK-E of 2 h or more (p = 0.008). To evaluate how much WK-E was required, the TSD on weekdays was classified into four groups: (i) less than 6 h, (ii) 6–7 h, (iii) 7–8 h, and (iv) more than 8 h. Similarly, WK-E was classified into four groups: (i) less than 1 h, (ii) 1–2 h, (iii) 2–3 h, and (iv) more than 3 h. The results are shown in Figure 3. The shorter the TSD on weekdays, the higher was the mean BMI z-score (Figure 3A). In participants with a TSD of less than 6 h on weekdays, BMI z-scores were less than 0 only when WK-E was more than 3 h (Figure 3B). Furthermore, when the WK-E was less than 3 h, the overweight/obesity relative risk increased by 1.93 times (95% CI, 1.07–3.48) in participants whose TSD on weekdays was less than 6 h. The BMI z-score was 0 or less when WK-E was more than 2 h for 6–7 h of TSD on weekdays and 1 h for 7–8 h of TSD on weekdays. Based on this, the shorter duration of sleep during weekdays was associated with a need for longer duration of sleep on weekends to achieve a BMI z-score of 0 or less. Interestingly, participants who slept more than 8 h on weekdays had near-zero BMI z-scores, regardless of their weekend sleep extension (Figure 3B). These results suggest that if the sleep duration is less than 8 h on weekdays, an additional hour of weekend sleep may be necessary for each hour of weekday sleep lost to reduce the BMI z-score.