Association between Sleep Duration and Body Composition in Girls Ten to Eighteen Years of Age: A Population-Based Study

Abstract

Background:

This study aimed to evaluate the relationship between sleep duration and body composition as measured by dual-energy X-ray absorptiometry (DXA).

Methods:

Based on data obtained from the Korea National Health and Nutrition Examination Survey (2010), 303 girls were divided into three groups by sleep duration: very short, short, and recommended.

Results:

By analysis of covariance, compared with the “very short” group, the “recommended” group had reduced adjusted mean DXA-assessed total mass (TM, 46.46 kg vs. 51.36 kg, p = 0.012), fat mass (FM, 14.38 kg vs. 17.55 kg, p = 0.002), and fat mass percentage (FMP, 30.66% vs. 33.15%, p = 0.017) in the whole body; TM (20.85 kg vs. 23.63 kg, p = 0.007), FM (5.82 kg vs. 7.64 kg, p = 0.001), and FMP (27.48% vs. 30.70%, p = 0.009) in the trunk; TM (4.59 kg vs. 5.15 kg, p = 0.006), FM (1.62 kg vs. 2.05 kg, p = 0.001), and FMP (27.48% vs. 30.70%, p = 0.019) in the arms; and TM (16.75 kg vs. 18.23 kg, p = 0.042) and FM (6.09 kg vs. 6.97 kg, p = 0.018) in the legs. By multiple linear regression analysis, sleep duration in hours had a significant negative association with DXA-assessed TM (β = −1.221, p = 0.016) and FM (β = −0.760, p = 0.006) in the whole body; FM (β = −0.014, p = 0.026) in the head; TM (β = −0.699, p = 0.010), FM (β = −0.454, p = 0.003), and FMP (β = −0.714, p = 0.030) in the trunk; TM (β = −0.152, p = 0.005) and FM (β = −0.101, p = 0.004) in the arms; and FM (β = −0.191, p = 0.045) in the legs.

Conclusions:

Our results suggest that shorter sleep duration is related to higher FM, but not lean mass, especially in the trunk.

Background

The prevalence of overweight/obesity during childhood and adolescence has risen substantially over the past several decades.¹ Approximately one-fifth of children and adolescents in Korea are classified as overweight or obese.² Child and adolescent overweight/obesity has become a major concern for public health worldwide. As the prevalence of childhood obesity increases, a variety of obesity-related comorbidities could threaten the personal health of affected children and adolescents and burden national health care systems, causing socioeconomic impacts.³ Child and adolescent obesity is closely related to adult obesity and considered an independent risk factor for the development of adult morbidities.⁴ Early identification of childhood and adolescent obesity and its related factors is recognized as essential for reducing personal and societal burdens. Age-specific intervention can prevent obesity-related morbidity in adulthood.

It has been recognized that sleep plays a role in energy homeostasis by controlling diurnal rhythms.⁵ Appropriate sleep duration varies with age. Previous studies have found that sleep duration tends to decrease with age during childhood and adolescence.⁶ Age-appropriate sleep duration plays an essential role in health. A series of epidemiological studies has indicated that increases in the prevalence of obesity tend to be related to sleep duration.^7–9 Most studies have suggested that short sleep duration and/or poor sleep quality are related to overweight/obesity in children,⁸ adolescents,⁸ and adults.⁹ Previous studies have generally used anthropometric measurements, including BMI, to represent body composition, especially obesity.^7–9 BMI can provide a rational estimate of adiposity in children and adolescents and is considered a clinical standard of overweight and obesity for pediatric populations ≥2 years of age.¹⁰ However, anthropometric measurements such as BMI may provide relatively limited information compared to systematic information related to body composition, which can be obtained with two-dimensional projections using dual-energy X-ray absorptiometry (DXA).¹¹

This population-based study aimed to evaluate the relationship between sleep duration and body composition, the latter of which was determined using anthropometric measurements and DXA data in girls 10–18 years of age included in nationally representative Korean data. To investigate these relationships, sleep duration was analyzed not only as a categorical variable grouped into three levels but also as a continuous variable measured in hours. In addition, this study evaluated whether these relationships were statistically significant after adjusting for possible confounders.

Methods

Subjects

Data obtained from the 2010 Korea National Health and Nutrition Examination Survey (KNHANES) were analyzed in this study. The KNHANES, which is a cross-sectional and nationally representative survey, is conducted regularly by the Division of Chronic Disease Surveillance, Korean Centers for Disease Control and Prevention (KCDC),¹² is composed of a health questionnaire, health examination, and nutritional assessment, and uses a stratified and multistage probability sampling design to select household units for inclusion. The details of the KNHANES have been described previously.¹³ A total of 7043 participants were included in the 2010 KNHANES. Of these participants, 448 subjects 10–18 years of age were included in the preliminary analyses. Girls 10–18 years of age were included in this analysis because in a previous study, the relationship between shorter sleep duration and obesity was dependent on sex and more significant in girls than in boys.¹⁴ Girls who did not have complete DXA data were excluded (n = 10). Female participants who did not have complete anthropometric or laboratory measurements, or complete health questionnaire data were excluded (n = 128). Because low-density lipoprotein cholesterol (LDL-C) levels were determined with Friedewald's equation,¹⁵ participants with triglyceride (TG) ≥400 mg/dL were also excluded (n = 5). We excluded two girls with a sleep duration >10 hours (∼12 hours) because their data could have resulted in a lack of statistical significance (n = 2). Finally, a total of 303 girls were included in the final analyses. The database is available to the public at the KNHANES website (http://knhanes.cdc.go.kr). The study protocols of the 2010 KNHANES were approved by the Institutional Review Boards of the KCDC. Informed consent was provided by all KNHANES subjects. All methods used in the KNHANES were performed in accordance with relevant guidelines and regulations. This study was exempted by the institutional review board of Hallym University Dongtan Sacred Heart Hospital from the need to obtain participant consent because the data were analyzed anonymously.

Definition of Sleep Duration

Sleep duration was assessed in the KNHANES using the question, “How many hours of sleep do you usually get in a day on average?” The original units of hours/day were retained. The National Sleep Foundation (NSF) and the CDC provide age-specific recommendations for healthy sleep duration (https://cdc.gov/sleep). The NSF and CDC recommend a sleep duration of 9–11 hours in children 6–13 years of age and 8–10 hours in adolescents 14–17 years of age (recommended sleep duration). These authorities suggest a minimum sleep duration of 7–8 hours in school-aged children 6–13 years of age and 7 hours in males and females 14–17 years of age as conditionally appropriate (short sleep duration). In contrast, a sleep duration of <7 hours in boys and girls 6–13 years of age and a sleep duration of <7 hours in teenagers 14–17 years of age are not recommended by these sources (very short sleep duration). Subjects were divided into three groups based on sleep duration as recommended by the NSF and the CDC: (i) very short, (ii) short, and (iii) recommended.

Measurements

Anthropometric assessments, including height, weight, waist circumference (WC), and systolic and diastolic blood pressure (SBP and DBP, respectively; mmHg), were performed by a trained expert using standard methods. BMI was calculated as the weight/square of the height (kg/m²). The standard deviation scores (SDSs) for height, weight, WC, and BMI were determined using LMS (lambda for the skewness, mu for the median, and sigma for the generalized coefficient of variation) methods based on the 2007 Korean reference data.¹⁶ According to the KNHANES protocol, blood samples are randomly obtained from the antecubital vein year-round during the daytime after the boys and girls had fasted for at least 8 hours. The collected blood samples were immediately processed, refrigerated, and transported to a central laboratory (NeoDin Medical Institute, Seoul, Korea) for analysis within 24 hours. Blood biochemistry tests were performed using a Hitachi 7600 automatic analyzer (Hitachi, Tokyo, Japan). LDL-C (mg/dL) levels were calculated according to a previously described method.¹⁵

Body Composition Analyses and Precision Study

The DXA scans in the KNHANES were performed between 2008 and 2011 using a QDR Discovery fan-beam densitometer (Hologic, Inc., Bedford, MA). Well-trained and qualified technicians conducted standardized daily quality control of the DXA instruments. The DXA results were analyzed using Hologic Discovery software (version 13.1). All DXA data, including bone mineral content (g), total mass (TM), lean mass (LM, g), fat mass (FM, g), and fat mass percentage (FMP, FM/TM × 100, %), were obtained from the KNHANES according to individualized demographic information. This study used DXA-assessed parameters, including TM, LM, FM, and FMP in the whole body, head, trunk, arms, and legs, for the analyses. The coefficient of variation (%) of the DXA data was 2.01% for bone mineral content, 2.05% for FM, 1.46% for LM, and 0.79% for FM, all of which were within the limits proposed by the International Society for Clinical Densitometry 2013 conference.¹⁷

Collection of General Characteristics

Alcohol drinking, smoking, physical activity, household income, and residence were included as lifestyle-related parameters. Alcohol drinkers were divided into two groups (drinkers vs. nondrinkers). Smokers were classified into two groups (smokers vs. nonsmokers). Physical activity was categorized into two groups (yes vs. no). Household income was reported in quartiles, and participants were divided into two groups (first quartile vs. ≥second quartile). The place of residence was classified into two groups (urban vs. rural).

Statistical Analyses

In this study, statistical analyses were performed using R version 3.5.1 (The R Foundation for Statistical Computing, Vienna, Austria). Continuous and categorical variables are presented as the means ± standard deviations and percentages, respectively. The differences were analyzed using analysis of variance for continuous variables and the chi-squared test for categorical variables. The adjusted means and standard errors (SEs) of BMI SDS, WC SDS, and the DXA-assessed parameters, including TM, LM, FM, and FMP in the whole body, head, trunk, arms, and legs, were determined for each of the three sleep duration groups using analysis of covariance (ANCOVA) after adjustment for possible confounders. In ANCOVA model 1, the adjusted means and SE of BMI SDS and WC SDS were estimated after controlling for age, alcohol consumption, smoking, physical activity, household income, and residence. In ANCOVA model 2, the adjusted means and SE of BMI SDS and WC SDS were assessed after controlling for age, SBP, DBP, glucose, total cholesterol (T-C), high-density lipoprotein cholesterol (HDL-C), TG, LDL-C, alcohol drinking, smoking, physical activity, household income, and residence. The adjusted means of DXA-assessed variables in models 3 and 4 were determined after controlling for confounders that were used in models 1 and 2 plus BMI SDS. The pairwise differences between sleep duration groups were tested for significance using post hoc tests with the Bonferroni correction in each ANCOVA model. To evaluate sleep duration and parameters of DXA, simple and multiple linear regression analyses were conducted among the three groups according to sleep duration and DXA-assessed parameters. In model 1, simple linear regression analyses were generated after no adjustment. Multiple linear regression analyses were conducted after controlling for age and BMI SDS in the partially adjusted model 2. In model 3, multiple linear regression analyses were performed after adjustment for age, BMI SDS, SBP, DBP, glucose, T-C, HDL-C, TG, LDL-C, alcohol drinking, smoking, physical activity, household income, and residence. Finally, multiple linear regression analyses were performed for hourly sleep duration and DXA-assessed parameters after adjustment for the confounding factors in models 1, 2, and 3 described above. The corresponding standardized regression coefficient (β) and SE were estimated. All significant differences were determined based on a p value <0.05 using a two-tailed method.

Results

Clinical Characteristics of the Study Population According to Sleep Duration Groups

The clinical characteristics of the study population separated into the three sleep duration groups are shown in Table 1. Girls with the recommended sleep duration tended to have a higher mean age and a lower mean weight SDS (p = 0.021), BMI SDS (p = 0.038), SBP (p = 0.022), and DBP (p = 0.005).

Table 1.

Clinical Characteristics of the Study Population According to the Three Sleep Duration Groups (n = 303)

	Sleep duration^a			p
	Very short	Short	Recommended
	n = 115	n = 127	n = 61
Age (years)	13.17 ± 2.54	12.98 ± 2.45	14.92 ± 2.17	<0.001
Height SDS	1.63 ± 1.90	1.72 ± 1.55	1.16 ± 1.89	0.156
Weight SDS	1.64 ± 1.12	1.61 ± 1.10	1.17 ± 1.19	0.021
BMI SDS	1.22 ± 0.95	1.16 ± 0.94	0.88 ± 0.99	0.038
WC SDS	1.40 ± 1.00	1.38 ± 0.98	1.27 ± 1.03	0.403
SBP (mmHg)	121.79 ± 17.99	119.49 ± 18.72	115.18 ± 15.01	0.022
DBP (mmHg)	77.32 ± 10.63	75.57 ± 11.38	72.46 ± 9.58	0.005
Glucose (mg/dL)	96.48 ± 14.33	93.37 ± 13.84	91.20 ± 7.78	0.008
T-C (mg/dL)	184.22 ± 36.84	183.46 ± 42.18	181.31 ± 37.17	0.655
HDL-C (mg/dL)	50.45 ± 12.11	48.65 ± 10.30	46.97 ± 10.90	0.043
TGs (mg/dL)	112.44 ± 64.71	118.97 ± 75.2	124.39 ± 68.16	0.265
LDL-C (mg/dL)	111.27 ± 31.97	111.05 ± 37.32	109.39 ± 33.20	0.753
DXA
Whole body
Total mass (kg)	50.68 ± 12.59	48.27 ± 9.89	48.55 ± 9.11	0.141
Lean mass (kg)	33.60 ± 6.55	32.51 ± 5.39	32.97 ± 5.54	0.275
Fat mass (kg)	17.07 ± 6.90	15.76 ± 5.58	15.76 ± 4.54	0.106
Fat mass percentage (%)	31.71 ± 6.10	31.92 ± 5.96	31.11 ± 4.83	0.420
Head
Total mass (kg)	4.33 ± 0.42	4.25 ± 0.40	4.31 ± 0.39	0.595
Lean mass (kg)	3.44 ± 0.34	3.40 ± 0.33	3.44 ± 0.33	0.777
Fat mass (kg)	0.88 ± 0.14	0.86 ± 0.13	0.87 ± 0.12	0.352
Fat mass percentage (%)	20.38 ± 2.27	20.14 ± 2.24	20.18 ± 2.22	0.491
Trunk
Total mass (kg)	23.29 ± 6.82	21.85 ± 5.23	21.97 ± 4.85	0.092
Lean mass (kg)	15.89 ± 3.37	15.30 ± 2.90	15.41 ± 2.95	0.229
Fat mass (kg)	7.39 ± 3.93	6.56 ± 2.95	6.56 ± 2.45	0.063
Fat mass percentage (%)	30.18 ± 7.54	29.97 ± 7.08	29.26 ± 5.59	0.301
Arms
Total mass (kg)	5.07 ± 1.33	4.78 ± 1.02	4.83 ± 0.99	0.110
Lean mass (kg)	3.09 ± 0.60	2.98 ± 0.50	3.01 ± 0.52	0.208
Fat mass (kg)	1.98 ± 0.87	2.98 ± 0.50	30.01 ± 0.52	0.137
Fat mass percentage (%)	37.54 ± 8.39	36.54 ± 8.82	37.14 ± 6.88	0.629
Legs
Total mass (kg)	18.00 ± 4.39	17.39 ± 3.54	17.44 ± 3.20	0.278
Lean mass (kg)	11.17 ± 2.47	10.84 ± 1.91	10.93 ± 2.00	0.375
Fat mass (kg)	6.82 ± 2.47	6.55 ± 2.01	6.51 ± 1.58	0.273
Fat mass percentage (%)	37.35 ± 5.68	37.07 ± 5.64	37.18 ± 4.76	0.798
Alcohol drinking (%)	41 (35.65%)	51 (40.16%)	21 (34.43%)	0.673
Smoking (%)	1 (0.87%)	2 (1.57%)	1 (1.64%)	0.865
Physical activity (%)	61 (53.04%)	49 (38.58%)	27 (44.26%)	0.077
Household income ≤1st quartile (%)	16 (13.91%)	14 (11.02%)	12 (19.67%)	0.275
Residence (rural)	16 (13.91%)	20 (15.75%)	12 (19.67%)	0.609
Sleep duration (hours)	5.64 ± 0.60	7.24 ± 0.43	8.51 ± 0.72	<0.001

Data are expressed as the means ± SDs for continuous variables and percentages (%) for categorical variables.

Sleep duration was classified into three groups: (i) very short, (ii) short, and (iii) recommended.

(i) A very short sleep duration was defined as <7 hours in subjects 6–17 years of age.

(ii) A short sleep duration was defined as 7–8 hours in children 6–13 of age and 7 hours in adolescents 14–17 years of age.

(iii) The recommended sleep duration was defined as 9–11 hours in girls 6–13 years of age and 8–10 hours in those 14–17 years of age.

The percentages shown for alcohol drinking, smoking, physical activity, and household income ≤1st quartile indicate the number of subjects with alcohol drinking, smoking, physical activity, and household income ≤1st quartile divided by the total study population in the three sleep duration groups.

SDS, standard deviation score; BMI, body mass index; WC, waist circumference; SBP, systolic blood pressure; DBP, diastolic blood pressure; T-C, total cholesterol; HDL-C, high-density lipoprotein cholesterol; TG, triglyceride; LDL-C, low-density lipoprotein cholesterol; DXA, dual-energy X-ray absorptiometry; SD, standard deviation.

Adjusted Body Composition Means According to Sleep Duration Groups

The adjusted means of BMI SDS, WC SDS, and DXA-assessed parameters according to the three sleep duration groups are presented in Table 2. The subjects with a recommended sleep duration had lower adjusted mean values for DXA-assessed TM (46.46 kg vs. 51.36 kg, p = 0.014), FM (14.38 kg vs. 17.55 kg, p = 0.002), and FMP (30.66% vs. 33.15%, p = 0.019) in the whole body; TM (20.85 kg vs. 23.63 kg, p = 0.008), FM (5.82 kg vs. 7.64 kg, p = 0.001), and FMP (27.48% vs. 30.70%, p = 0.010) in the trunk; TM (4.59 kg vs. 5.15 kg, p = 0.006), FM (1.62 kg vs. 2.05 kg, p = 0.001), and FMP (27.48% vs. 30.70%, p = 0.021) in the arms; and TM (16.75 kg vs. 18.23 kg, p = 0.048) and FM (6.09 kg vs. 6.97 kg, p = 0.019) in the legs than were found in those with very short sleep durations after adjustment for possible confounding factors.

Table 2.

Adjusted Means of BMI and Waist Circumference Standard Deviation Scores and Dual-Energy X-ray Absorptiometry-Assessed Parameters in the Whole Body or the Head, Trunk, Arms, and Legs According to Sleep Duration Groups among Korean Girls 10–18 Years of Age (n = 303)

	Sleep duration^a
	Model 1			Model 2
	Very short	Short	Recommended	Very short	Short	Recommended
	n = 115	n = 127	n = 61	n = 115	n = 127	n = 61
BMI SDS	1.18 ± 0.08	1.09 ± 0.08	1.10 ± 0.12	1.14 ± 0.07	1.08 ± 0.07	1.18 ± 0.11
WC SDS	1.42 ± 0.09	1.31 ± 0.09	1.32 ± 0.13	1.37 ± 0.09	1.31 ± 0.07	1.38 ± 0.11
	Model 3			Model 4
DXA
Whole body
Total mass (kg)	51. 14 ± 0.94	49.05 ± 0.90	46.06 ± 1.34	51. 36 ± 0.96	48.66 ± 0.91	46.46 ± 1.37^*
Lean mass (kg)	33.79 ± 0.53	32.85 ± 0.50	31.72 ± 0.75	33.81 ± 0.54	32.67 ± 0.51	32.07 ± 0.77
Fat mass (kg)	17.34 ± 0.51	16.20 ± 0.49	14.34 ± 0.73^*	17.55 ± 0.52	15.99 ± 0.49	14.38 ± 0.74^*
Fat mass percentage (%)	32.95 ± 0.51	32.30 ± 0.49	30.89 ± 0.73	33.15 ± 0.52	32.23 ± 0.49	30.66 ± 0.74^*
Head
Total mass (kg)	4.34 ± 0.04	4.27 ± 0.04	4.25 ± 0.05	4.35 ± 0.04	4.25 ± 0.04	4.26 ± 0.05
Lean mass (kg)	3.45 ± 0.03	3.41 ± 0.03	3.40 ± 0.04	3.46 ± 0.03	3.39 ± 0.03	3.41 ± 0.04
Fat mass (kg)	0.89 ± 0.01	0.86 ± 0.01	0.85 ± 0.02	0.89 ± 0.01	0.86 ± 0.01	0.85 ± 0.02
Fat mass percentage (%)	20.43 ± 0.21	20.19 ± 0.20	20.00 ± 0.30	20.45 ± 0.22	20.16 ± 0.20	20.03 ± 0.31
Trunk
Total mass (kg)	23.54 ± 0.51	22.27 ± 0.48	20.63 ± 0.72^*	23.63 ± 0.52	22.08 ± 0.49	20.85 ± 0.73^*
Lean mass (kg)	16.00 ± 0.28	15.48 ± 0.27	14.84 ± 0.39	16.00 ± 0.29	15.39 ± 0.27	15.03 ± 0.41
Fat mass (kg)	7.54 ± 0.28	6.79 ± 0.27^**	5.79 ± 0.40^*	7.64 ± 0.29	6.69 ± 0.27	5.82 ± 0.41^*
Fat mass percentage (%)	30.47 ± 0.62	29.45 ± 0.59	27.70 ± 0.87^*	30.70 ± 0.62	29.35 ± 0.59	27.48 ± 0.89^*
Arms
Total mass (kg)	5.12 ± 0.10	4.86 ± 0.01	4.57 ± 0.14^*	5.15 ± 0.10	4.82 ± 0.01	4.59 ± 0.14^*
Lean mass (kg)	3.11 ± 0.05	3.00 ± 0.05	2.93 ± 0.07	3.10 ± 0.05	2.99 ± 0.05	2.97 ± 0.08
Fat mass (kg)	2.01 ± 0.07	1.86 ± 0.06	1.64 ± 0.09^*	2.05 ± 0.07	1.83 ± 0.06	1.62 ± 0.10^*
Fat mass percentage (%)	37.87 ± 0.73	37.10 ± 0.70	35.35 ± 1.03	38.30 ± 0.73	36.97 ± 0.69	34.83 ± 1.04^*
Legs
Total mass (kg)	18.14 ± 0.34	17.66 ± 0.32	16.62 ± 0.48^*	18.23 ± 0.34	17.51 ± 0.33	16.75 ± 0.49^*
Lean mass (kg)	11.24 ± 0.19	10.97 ± 0.19	10.56 ± 0.27	11.26 ± 0.20	10.90 ± 0.19	10.67 ± 0.28
Fat mass (kg)	6.91 ± 0.18	6.69 ± 0.17	6.06 ± 0.25^*	6.97 ± 0.18	6.61 ± 0.17	6.09 ± 0.26^*
Fat mass percentage (%)	37.52 ± 0.50	37.32 ± 0.48	36.35 ± 0.71	37.67 ± 0.51	37.28 ± 0.48	36.14 ± 0.73

Data are presented as the means ± SEs.

Model 1: Adjusted means of the BMI SDS and WC SDS were determined after adjustment for age, alcohol drinking, smoking, physical activity, household income, and residence according to the three sleep duration groups.

Model 2: Adjusted means of parameters of clinical variables were determined after adjustment for age, SBP, DBP, glucose, T-C, HDL-C, TG, LDL-C, alcohol drinking, smoking, physical activity, household income, and residence according to the three sleep duration groups.

Model 3: Adjusted means of DXA-assessed parameters were determined after adjustment for age, BMI SDS, alcohol drinking, smoking, physical activity, household income, and residence according to the three sleep duration groups.

Model 4: Adjusted means of DXA-assessed parameters were determined after adjustment for age, BMI SDS, SBP, DBP, glucose, T-C, HDL-C, TG, LDL-C, alcohol drinking, smoking, physical activity, household income, and residence according to the three sleep-duration groups.

Sleep duration was classified into three groups: (i) very short, (ii) short, and (iii) recommended.

(i) A very short sleep duration was defined as <7 hours in subjects 6–17 years of age.

(ii) A short sleep duration was defined as 7–8 hours in children 6–13 years of age and 7 hours in adolescents 14–17 years of age.

(iii) The recommended sleep duration was defined as 9–11 hours in girls 6–13 years of age and 8–10 hours in those 14–17 years of age.

p < 0.05 between the group with very short sleep durations and the group with recommended sleep durations using ANCOVA with Bonferroni-corrected post hoc tests.

p < 0.05 between the group with very short sleep durations and the group with short sleep durations using ANCOVA with Bonferroni-corrected post hoc tests.

ANCOVA, analysis of covariance; SE, standard error.

Multiple Linear Regression Analyses between Three Sleep Duration Groups and DXA Parameters

Table 3 shows the results of simple and multiple linear regression analyses between sleep duration and DXA-assessed parameters. No significant correlation was observed between the three groups separated by sleep duration and DXA-assessed parameters in the simple linear regression analyses. However, significant associations were found in partially adjusted and fully adjusted multiple linear regression analyses. The sleep duration groups were inversely associated with DXA-assessed TM (β = −2.493, p = 0.003), FM (β = −1.579, p = 0.001), and FMP (β = −1.240, p = 0.006) in the whole body; FM (β = −0.022, p = 0.035) in the head; TM (β = −1.416, p = 0.002), FM (β = −0.913, p < 0.001), and FMP (β = −1.636, p = 0.003) in the trunk; TM (β = −0.288, p = 0.001), FM (β = −0.214, p < 0.001), and FMP (β = −1.739, p = 0.007) in the arms; and TM (β = −0.735, p = 0.014) and FM (β = −0.430, p = 0.007) in the legs after adjustment for the possible confounders included in model 3.

Table 3.

Multiple Linear Regression Analyses of Hourly Sleep Duration and Parameters Assessed with Dual-Energy X-Ray Absorptiometry in the Whole Body, Head, Trunk, Arms, and Legs among Korean Girls 10–18 Years of Age (n = 303)

	Sleep duration (groups)^a
	Model 1			Model 2			Model 3
	β	SE	p	β	SE	p	β	SE	p
DXA
Whole body
Total mass (kg)	−1.244	0.842	0.141	−2.464	0.800	0.002	−2.493	0.830	0.003
Lean mass (kg)	−0.499	0.456	0.275	−1.020	0.446	0.023	−0.914	0.468	0.052
Fat mass (kg)	−0.746	0.460	0.106	−1.443	0.434	0.001	−1.579	0.448	0.001
Fat mass percentage (%)	−0.363	0.450	0.420	−0.966	0.433	0.027	−1.240	0.447	0.006
Head
Total mass (kg)	−0.017	0.032	0.595	−0.050	0.031	0.110	−0.054	0.032	0.094
Lean mass (kg)	−0.007	0.026	0.777	−0.030	0.026	0.250	−0.032	0.027	0.233
Fat mass (kg)	−0.010	0.010	0.352	−0.020	0.010	0.044	−0.022	0.011	0.035
Fat mass percentage (%)	−0.120	0.174	0.491	−0.219	0.177	0.217	−0.241	0.187	0.199
Trunk
Total mass (kg)	−0.764	0.451	0.092	−1.424	0.428	0.001	−1.416	0.444	0.002
Lean mass (kg)	−0.290	0.240	0.229	−0.571	0.235	0.016	−0.504	0.246	0.041
Fat mass (kg)	−0.474	0.254	0.063	−0.853	0.240	<0.001	−0.913	0.247	<0.001
Fat mass percentage (%)	−0.566	0.546	0.301	−1.329	0.522	0.011	−1.636	0.529	0.003
Arms
Total mass (kg)	−0.142	0.089	0.110	−0.273	0.084	0.001	−0.288	0.088	0.001
Lean mass (kg)	−0.053	0.042	0.208	−0.092	0.042	0.031	−0.074	0.044	0.097
Fat mass (kg)	−0.089	0.060	0.137	−0.181	0.056	0.001	−0.214	0.058	<0.001
Fat mass percentage (%)	−0.311	0.643	0.629	−1.179	0.617	0.057	−1.739	0.633	0.007
Legs
Total mass (kg)	−0.322	0.296	0.278	−0.716	0.284	0.012	−0.735	0.296	0.014
Lean mass (kg)	−0.148	0.167	0.375	−0.328	0.164	0.046	−0.305	0.172	0.077
Fat mass (kg)	−0.173	0.158	0.273	−0.389	0.151	0.011	−0.430	0.157	0.007
Fat mass percentage (%)	−0.109	0.425	0.798	−0.520	0.422	0.219	−0.731	0.441	0.099

Model 1: A simple linear regression analysis of the three groups separated according to sleep duration and parameters obtained using DXA was conducted after no adjustment.

Model 2: A multiple linear regression analysis of the three groups according to sleep duration and parameters obtained using DXA was conducted after adjustment for age and BMI SDS.

Model 3: A multiple linear regression analysis of the three groups according to sleep duration and parameters obtained using DXA was conducted after adjustment for age, BMI SDS, SBP, DBP, glucose, T-C, HDL-C, TG, LDL-C, alcohol drinking, smoking, physical activity, household income, and residence.

Sleep duration was classified into three groups: (i) very short, (ii) short, and (iii) recommended.

Multiple Linear Regression Analyses between Hourly Sleep Duration and DXA Parameters

The results of simple and multiple linear regression analyses between sleep duration and DXA-assessed parameters are shown in Table 4. In the unadjusted simple linear regression analyses, hourly sleep duration was inversely associated with DXA-assessed FM in the trunk (β = −0.309, p = 0.046) and LM in the arms (β = −0.051, p = 0.046). Hourly sleep duration was significantly inversely associated with DXA-assessed TM (β = −1.221, p = 0.016) and FM (β = −0.760, p = 0.006) in the whole body; FM (β = −0.014, p = 0.026) in the head; TM (β = −0.699, p = 0.010), FM (β = −0.454, p = 0.003), and FMP (β = −0.714, p = 0.030) in the trunk; TM (β = −0.152, p = 0.005) and FM (β = −0.101, p = 0.004) in the arms; and FM (β = −0.191, p = 0.045) in the legs in the fully adjusted multiple linear regression analyses.

Table 4.

Multiple Linear Regression Analyses of Hourly Sleep Duration and Parameters Obtained Using Dual-Energy X-Ray Absorptiometry in the Whole Body, Head, Trunk, Arms, and Legs among Korean Girls 10–18 Years of Age (n = 303)

	Sleep duration (hours)
	Model 1			Model 2			Model 3
	β	SE	p	β	SE	p	β	SE	p
DXA
Whole body
Total mass (kg)	−0.910	0.512	0.076	−1.155	0.481	0.017	−1.221	0.502	0.016
Lean mass (kg)	−0.411	0.277	0.139	−0.510	0.267	0.057	−0.462	0.282	0.103
Fat mass (kg)	−0.499	0.280	0.076	−0.645	0.262	0.014	−0.760	0.272	0.006
Fat mass percentage (%)	−0.216	0.274	0.432	−0.346	0.260	0.185	−0.525	0.271	0.054
Head
Total mass (kg)	−0.017	0.019	0.389	−0.023	0.019	0.212	−0.026	0.020	0.183
Lean mass (kg)	−0.006	0.016	0.696	−0.010	0.015	0.503	−0.012	0.016	0.462
Fat mass (kg)	−0.010	0.006	0.092	−0.013	0.006	0.031	−0.014	0.006	0.026
Fat mass percentage (%)	−0.154	0.106	0.148	−0.183	0.105	0.084	−0.194	0.112	0.086
Trunk
Total mass (kg)	−0.529	0.274	0.055	−0.666	0.258	0.010	−0.699	0.269	0.010
Lean mass (kg)	−0.220	0.146	0.132	−0.277	0.141	0.050	−0.245	0.148	0.100
Fat mass (kg)	−0.309	0.154	0.046	−0.389	0.145	0.008	−0.454	0.150	0.003
Fat mass percentage (%)	−0.331	0.332	0.321	−1.329	0.522	0.011	−0.714	0.326	0.030
Arms
Total mass (kg)	−0.110	0.054	0.042	−0.137	0.051	0.007	−0.152	0.053	0.005
Lean mass (kg)	−0.051	0.026	0.046	−0.060	0.025	0.019	−0.050	0.027	0.060
Fat mass (kg)	−0.059	0.036	0.107	−0.078	0.034	0.022	−0.101	0.035	0.004
Fat mass percentage (%)	−0.122	0.391	0.756	−0.295	0.371	0.427	−0.643	0.385	0.096
Legs
Total mass (kg)	−0.254	0.180	0.159	−0.329	0.170	0.055	−1.286	0.769	0.054
Lean mass (kg)	−0.133	0.101	0.190	−0.164	0.098	0.096	−0.155	0.104	0.137
Fat mass (kg)	−0.121	0.096	0.208	−0.165	0.091	0.070	−0.191	0.095	0.045
Fat mass percentage (%)	−0.050	0.259	0.845	−0.149	0.253	0.556	−0.260	0.267	0.330

Model 1: A simple linear regression analysis of hourly sleep duration and parameters obtained using DXA was conducted after no adjustment.

Model 2: A multiple linear regression analysis of hourly sleep duration and parameters obtained using DXA was conducted after adjustment for age and BMI SDS.

Model 3: A multiple linear regression analysis of hourly sleep duration and parameters obtained using DXA was conducted after adjustment for age, BMI SDS, SBP, DBP, glucose, T-C, HDL-C, TG, LDL-C, alcohol drinking, smoking, physical activity, household income, and residence.

Discussion

This nationally representative population-based study shows that the group with the recommended sleep duration had a lower adjusted mean FM in the whole body, trunk, arms, and legs and FMP in the whole body, trunk, and arms after adjustment for confounders in ANCOVA. In multiple linear regression analyses, hourly sleep duration was negatively associated with DXA-assessed TM and FM, but not LM in the whole body, trunk, and arms after adjustment for confounders. In particular, the relationship between hourly sleep duration and FMP was significant only in the trunk.

Series of studies regarding the relationship between sleep duration and obesity have been conducted in children, adolescents, and adults. Most adult studies have revealed a U-shaped relationship between sleep duration and obesity, as well as metabolic syndrome and its components.¹⁸ The results obtained in children and adolescents indicate that sleep duration is negatively correlated with obesity,¹⁹ although a Canadian study demonstrated a U-shaped relationship between sleep duration and obesity.²⁰ However, most studies analyzed data based on anthropometric measurements or non-DXA instruments, such as bioelectric impedance. In our analyses, sleep duration was not significantly associated with anthropometric measures, but was significantly negatively associated with DXA-assessed FM. These results may suggest that sleep duration is significantly related to obesity even though this relationship is not recognized when using anthropometric assessments.

The recent guidelines of the Endocrine Society recommend using BMI as the standard diagnostic tool for overweight and/or obesity in children and adolescents ≥2 years of age.²¹ However, these guidelines also suggest that associations between BMI and obesity-related comorbidities vary according to race/ethnicity and that increased muscle mass increases BMI.²¹ Anthropometric or bioelectric impedance measurements do not provide systematic or precise information regarding components of the body's composition, such as visceral or subcutaneous adipose tissues.²² Compared to non-DXA measurements, measurements obtained using DXA have some advantages. For example, DXA provides information regarding the quantification of tissue volume or FM, LM, and bone mass. DXA is also recognized as a convenient, quick (5–13 min for each scan), and safe modality that can also be used in people of all ages.²³ In particular, information regarding regional fat distribution in the abdominal and/or gluteofemoral regions can be obtained using DXA and allows the analysis of differential risks from excess accumulated fat.²⁴ For example, higher amounts of gluteofemoral fat may exert a protective effect against cardiometabolic disease in postmenopausal women.²⁵ In children and adolescents, higher amounts of abdominal fat have been significantly associated with cardiometabolic risk factors in childhood, independent of BMI.²⁶ In these analyses, shorter sleep duration was related to higher FM, but not LM in our DXA analyses. Notably, the relationship between shorter sleep duration and fat accumulation was augmented in the trunk region. This finding may suggest that children and adolescents need to achieve the recommended sleep duration.

Some mechanisms have been suggested to explain the relationship between sleep duration and obesity. Shorter sleep duration may be related to increased appetite. Sleep restriction may result in decreased inhibition of orexigenic activity in the hypothalamus, which modulates appetite regulation.²⁷ A randomized crossover clinical study showed that subjects with a sleep duration of ∼4 hours/day exhibit increased total energy intake.²⁸ In the absence of appropriate physical activity, a sustained increased appetite may lead to obesity in the long term. Another possible mechanism may involve hormone disturbances. A study revealed that sleep restriction is related to the biphasic secretion of growth hormone (GH), which is characterized by a second large pulse that occurs before bedtime; however, a single nocturnal GH pulse was observed during early sleep in extended sleep durations in healthy nonobese participants.²⁹ A U.S. study suggested that a sleep duration of ∼4 hours/day for six nights is related to elevated evening cortisol levels.³⁰ A short sleep duration was associated with not only increased levels of ghrelin but also decreased levels of leptin in a population-based longitudinal study.³¹ These altered hormonal levels could be linked to obesity (especially central obesity). In our analyses, sleep duration was negatively associated with FM and FMP in the trunk, but was related to only FM and not FMP in other regions. These results support previous experimental findings.

This study has some limitations. First, our analyses were cross-sectional in nature, and causality could not be assessed. Second, there is a possibility of information bias because data for sleep duration were based on a self-report questionnaire that was reported by the children themselves or their parents in the Korean national survey. Although sleep durations reported by parents are thought to provide accurate information, it provided information regarding what hour their child went to bed, but did not show precisely at what hour they fell asleep. In fact, compared to measured sleep, self-reported sleep durations were overestimated by ∼30 minutes in a U.S. study.³² The observed correlation between self-reported sleep duration and objective sleep duration assessed by actigraphy was 78%–81% in an Australian study.³³ In addition, other sleep variables, such as sleep timing, sleep duration on workdays vs. non-workdays, sleep onset latency, waking after sleep onset, or sleep quality, could not be analyzed, although these variables are considered to be related to obesity in children. In particular, some studies that evaluated the relationship between sleep timing or bedtimes and obesity have suggested that children who go to bed after a certain hour at night may be at greater risk for obesity.^34,35 Finally, we did not analyze obesity-related genes and/or factors or conditions related to sleep disturbance, including sleep apnea.

Conclusions

This Korean nationally representative population-based study found negative linear associations between sleep duration groups and DXA-assessed FM and FMP in the whole body, trunk, arms, and legs after adjustment for confounders in ANCOVA. In multiple linear regression analyses, hourly sleep duration was negatively associated with DXA-assessed FM, but not LM in the whole body, head, trunk, arms, and legs. In addition, hourly sleep duration was negatively associated with DXA-assessed FMP only in the trunk and not in other regions after controlling for confounders. Our results suggest that shorter sleep duration may be related to increased FM, but not LM, especially in the trunk region.

Footnotes

Funding Information

This research was supported by a Hallym University Research Fund, 2014 (HRF-201407-015).

Author Disclosure Statement

No competing financial interests exist.

References

Lobstein

, Jackson-Leach

, Moodie

, et al. Child and adolescent obesity: Part of a bigger picture. Lancet, 2015; 385:2510–2520.

Kim

, Jung

, Hong

, et al. The role of overweight and obesity on bone health in Korean adolescents with a focus on lean and fat mass. J Korean Med Sci, 2017; 32:1633–1641.

Dietz

, Robinson

. Clinical practice. Overweight children and adolescents. N Engl J Med, 2005; 352:2100–2109.

Juonala

, Magnussen

, Berenson

, et al. Childhood adiposity, adult adiposity, and cardiovascular risk factors. N Engl J Med, 2011; 365:1876–1885.

Mullington

, Chan

, Van Dongen

, et al. Sleep loss reduces diurnal rhythm amplitude of leptin in healthy men. J Neuroendocrinol, 2003; 15:851–854.

Hirshkowitz

, Whiton

, Albert

, et al. National Sleep Foundation's sleep time duration recommendations: Methodology and results summary. Sleep Health, 2015; 1:40–43.

Knutson

. Sleep duration and cardiometabolic risk: A review of the epidemiologic evidence. Best Pract Res Clin Endocrinol Metab, 2010; 24:731–743.

Sekine

, Yamagami

, Handa

, et al. A dose–response relationship between short sleeping hours and childhood obesity: Results of the Toyama Birth Cohort Study. Child Care Health Dev, 2002; 28:163–170.

Choi

, Lee

, Park

, et al. Relationship between sleep duration and the metabolic syndrome: Korean National Health and Nutrition Survey 2001. Int J Obes (Lond), 2008; 32:1091–1097.

10.

Freedman

, Sherry

. The validity of BMI as an indicator of body fatness and risk among children. Pediatrics, 2009; 124 Suppl 1:S23–S34.

11.

Prado

, Heymsfield

. Lean tissue imaging: A new era for nutritional assessment and intervention. J Parenter Enteral Nutr, 2014; 38:940–953.

12.

, Jeong

, Shim

. Association of the urinary sodium to urinary specific gravity ratio with metabolic syndrome in Korean children and adolescents: The Korea National Health and Nutrition Examination Survey 2010–2013. PLoS One, 2017; 12:e0189934.

13.

Kweon

, Kim

, Jang

, et al. Data resource profile: The Korea National Health and Nutrition Examination Survey (KNHANES). Int J Epidemiol, 2014; 43:69–77.

14.

Seo

, Shim

. Association of sleep duration with obesity and cardiometabolic risk factors in children and adolescents: A population-based study. Sci Rep, 2019; 9:9463.

15.

Friedewald

, Levy

, Fredrickson

. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem, 1972; 18:499–502.

16.

Moon

, Lee

, Nam

, et al. 2007 Korean national growth charts: Review of developmental process and an outlook. Korean J Pediatr, 2008; 51:1–25.

17.

Hangartner

, Warner

, Braillon

, et al. The official positions of the International Society for Clinical Densitometry: Acquisition of dual-energy X-ray absorptiometry body composition and considerations regarding analysis and repeatability of measures. J Clin Densitom, 2013; 16:520–536.

18.

Koren

, Dumin

, Gozal

. Role of sleep quality in the metabolic syndrome. Curr Diab Rep, 2016; 9:281–310.

19.

Felso

, Lohner

, Hollody

, et al. Relationship between sleep duration and childhood obesity: Systematic review including the potential underlying mechanisms. Nutr Metab Cardiovasc Dis, 2017; 27:751–761.

20.

Chaput

, Lambert

, Gray-Donald

, et al. Short sleep duration is independently associated with overweight and obesity in Quebec children. Can J Public Health, 2011; 102:369–374.

21.

Styne

, Arslanian

, Connor

, et al. Pediatric obesity-assessment, treatment, and prevention: An endocrine society clinical practice guideline. J Clin Endocrinol Metab, 2017; 102:709–757.

22.

Muller

, Lagerpusch

, Enderle

, et al. Beyond the body mass index: Tracking body composition in the pathogenesis of obesity and the metabolic syndrome. Obes Rev, 2012; 13 Suppl 2:6–13.

23.

Wells

, Haroun

, Williams

, et al. Evaluation of DXA against the four-component model of body composition in obese children and adolescents aged 5–21 years. Int J Obes (Lond), 2010; 34:649–655.

24.

Wiklund

, Toss

, Weinehall

, et al. Abdominal and gynoid fat mass are associated with cardiovascular risk factors in men and women. J Clin Endocrinol Metab, 2008; 93:4360–4366.

25.

Van Pelt

, Jankowski

, Gozansky

, et al. Lower-body adiposity and metabolic protection in postmenopausal women. J Clin Endocrinol Metab, 2005; 90:4573–4578.

26.

Gishti

, Gaillard

, Durmus

, et al. BMI, total and abdominal fat distribution, and cardiovascular risk factors in school-age children. Pediatr Res, 2015; 77:710–718.

27.

Gautier

, Del Parigi

, Chen

, et al. Effect of satiation on brain activity in obese and lean women. Obes Res, 2001; 9:676–684.

28.

Brondel

, Romer

, Nougues

, Touyarou

, Davenne

. Acute partial sleep deprivation increases food intake in healthy men. Am J Clin Nutr, 2010; 91:1550–1559.

29.

Spiegel

, Leproult

, Colecchia

, et al. Adaptation of the 24-h growth hormone profile to a state of sleep debt. Am J Physiol Regul Integr Comp Physiol, 2000; 279:R874–R883.

30.

Spiegel

, Leproult

, Van Cauter

. Impact of sleep debt on metabolic and endocrine function. Lancet, 1999; 354:1435–1439.

31.

Taheri

, Lin

, Austin

, et al. Short sleep duration is associated with reduced leptin, elevated ghrelin, and increased body mass index. PLoS Med, 2004; 1:e62.

32.

Lauderdale

, Knutson

, Yan

, et al. Self-reported and measured sleep duration: How similar are they?. Epidemiology, 2008; 19:838–845.

33.

Girschik

, Fritschi

, Heyworth

, et al. Validation of self-reported sleep against actigraphy. J Epidemiol, 2012; 22:462–468.

34.

Miller

, Kaciroti

, Lebourgeois

, et al. Sleep timing moderates the concurrent sleep duration-body mass index association in low-income preschool-age children. Acad Pediatr, 2014; 14:207–213.

35.

Anderson

, Andridge

, Whitaker

. Bedtime in preschool-aged children and risk for adolescent obesity. J Pediatr, 2016; 176:17–22.