Determinants and impact of sleep duration in children and ... - Nature

3 downloads 48 Views 137KB Size Report
Jul 30, 2008 - B Hitze1, A Bosy-Westphal1, F Bielfeldt1, U Settler1, S Plachta-Danielzik1, M Pfeuffer2, ..... (Mercer et al., 1998). Sleep duration was further ...
European Journal of Clinical Nutrition (2009) 63, 739–746

& 2009 Macmillan Publishers Limited All rights reserved 0954-3007/09 $32.00 www.nature.com/ejcn

ORIGINAL ARTICLE

Determinants and impact of sleep duration in children and adolescents: data of the Kiel Obesity Prevention Study B Hitze1, A Bosy-Westphal1, F Bielfeldt1, U Settler1, S Plachta-Danielzik1, M Pfeuffer2, ¨ nig3 and MJ Mu ¨ ller1 J Schrezenmeir2, H Mo 1

Institut fu¨r Humanerna¨hrung und Lebensmittelkunde, Christian-Albrechts Universita¨t Kiel, Kiel, Germany; 2Max Rubner-Institut, Kiel, Germany and 3Klinik fu¨r Allgemeine Innere Medizin, Universita¨tsklinikum Schleswig-Holstein, Kiel, Germany

Background/Objectives: This study investigates determinants of sleep duration and its impact on nutritional status, resting energy expenditure (REE), cardiometabolic risk factors and hormones in children/adolescents. Subjects/Methods: In 207 girls and 207 boys (13.0±3.4 (6.1–19.9) years) body mass index standard deviation score (BMI SDS), waist circumference (WC) z-score, body composition (air-displacement plethysmography), REE (ventilated hood system; n ¼ 312) and cardiometabolic risk factors/hormones (n ¼ 250) were assessed. Greater than 90th percentile of BMI/WC references was defined as overweight/overwaist. Sleep duration, media consumption (TV watching/computer use), physical activity, dietary habits, parental BMI, socio-economic status and early infancy were assessed by questionnaire. Short sleep was defined as o10 h per day for children o10 years and otherwise o9 h per day. Results: Total 15.9% participants were overweight, mean sleep duration was 8.9±1.3 h per day. Age explained most variance in sleep (girls: 57.0%; boys: 41.2%) besides a high nutrition quality score (girls: 0.9%) and a low media consumption (boys: 1.3%). Sleep was inversely associated with BMI SDS/WC z-score (girls: r ¼ 0.17/0.19, Po0.05; boys: r ¼ 0.21/0.20, Po0.01), which was strengthened after adjusting for confounders. Short vs long sleep was associated with 5.5-/2.3-fold higher risks for obesity/overwaist (girls). After adjusting for age, REE (adjusted for fat-free mass) was positively associated with sleep in boys (r ¼ 0.16, Po0.05). Independently of age and WC z-score, short sleep was associated with lower adiponectin levels in boys (11.7 vs 14.4 mg/ml, Po0.05); leptin levels were inversely related to sleep in girls (r ¼ 0.23, Po0.05). Homoeostasis model assessment–insulin resistance (r ¼ 0.20, Po0.05) and insulin levels (r ¼ 0.20, Po0.05) were associated with sleep (girls), which depended on WC z-score. Conclusions: Age mostly determined sleep. Short sleep was related to a higher BMI SDS/WC z-score (girls/boys), a lower REE (boys), higher leptin (girls) and lower adiponectin levels (boys).

European Journal of Clinical Nutrition (2009) 63, 739–746; doi:10.1038/ejcn.2008.41; published online 30 July 2008 Keywords: sleep duration; overweight; REE; cardiometabolic risk factors; hormones

Introduction Childhood overweight is a major public health concern of complex aetiology and short sleep duration was considered Correspondence: Professor Dr med MJ Mu¨ller, Institute of Human Nutrition and Food Science, Christian-Albrechts University Kiel, Du¨sternbrooker Weg 17-19, Kiel D-24105, Germany. E-mail: [email protected] Contributors: Writing of the manuscript, BH and MJM; study design, AB-W and MJM; data collection, BH, AB-W and FB; data analysis, BH, AB-W, US, MP, JS, HM; discussion of data, BH, AB-W, SP-D and MJM. All contributors helped with the revision of the paper. Received 19 March 2008; revised 4 June 2008; accepted 30 June 2008; published online 30 July 2008

as a determinant. Already Locard et al. (1992) showed that short sleep was associated with overweight in children. In a recent meta-analysis a 58% higher risk for overweight in children/adolescents with short vs long sleep was described (Chen et al., 2008). However, determinants of short sleep in children/adolescents are not well defined. Besides age, which was inversely associated with sleep duration (Iglowstein et al., 2003), sleep could be further determined by lifestyle. Exercise compared to sedentary activities may account for better/longer sleep, as sleep duration was positively associated with physical activity and inversely associated with television viewing (von Kries et al., 2002). In the same study, eating snacks while watching television was related to sleep

Determinants and impact of short sleep B Hitze et al

740 duration, whereas caloric intake was not. Moreover, a low socio-economic status (SES) as another potential determinant was related to short sleep (Dollman et al., 2007). Short sleep influences both sides of energy balance, which is explained by several determinants such as increased sympathetic activity, elevated cortisol and ghrelin levels, decreased leptin levels and insulin resistance (Spiegel et al., 1999, 2004a). Spiegel et al. (2004b) showed that sleep deprivation alters feelings of hunger/appetite especially for high-fat/carbohydrate foods in men. To our knowledge, the impact of sleep deprivation on resting energy expenditure (REE) has not been investigated in humans so far. However, in rats metabolism rose to 166% of baseline during sleep deprivation (Koban and Swinson, 2005). When compared to adults, the association between sleep and cardiometabolic risk factors/hormones is yet not well defined in children. However, preliminary data obtained in a small population of children showed that sleep duration was inversely associated with insulin resistance (Flint et al., 2007). This study aims to investigate the determinants of sleep duration and its impact on nutritional status, REE, cardiometabolic risk factors and hormones in children/adolescents.

Subjects and methods Study design and population Subjects were recruited by notice-board postings, writing to families who attended the Kiel Obesity Prevention Study (KOPS; Danielzik et al., 2004) and propaganda of participants. After exclusion of two children aged o6 years and four subjects with incomplete data, 207 girls and 207 boys (6.1–19.9 (13.0±3.4) years) remained for analysis. In subcohort analyses blood samples were taken (n ¼ 250) and REE was measured (n ¼ 312). Subjects were healthy and did not take any medication known to influence body composition, cardiometabolic risk factors or REE. The ethic committee of Christian-Albrechts-University Kiel approved the study. Written informed consent was obtained from each child/ adolescent and their legal guardian.

Anthropometric measurements and body composition analysis After an overnight fast, height was measured to the nearest 0.5 cm against a stadiometer (Seca, Model 220, Hamburg, Germany). Body weight was measured to the nearest gram using the digital scale coupled to the BodPod system (Body Composition System; Life Measurement Instruments, Concord, CA, USA). Body mass index (BMI) was calculated as weight (kg)/height (m2). German references were used to calculate BMI SDS (standard deviation score) and to define overweight/obesity (Kromeyer-Hauschild et al., 2001). The upper age limit of 18.5 years for these references resulted in 396 children/adolescents for this analysis. For adults, WHO (1995) definitions were used. European Journal of Clinical Nutrition

Waist circumference (WC) was measured to the nearest 0.5 cm midway between the lowest rib and the iliac crest with subjects dressed in underwear and respiring minimal. Greater than 90th age-/sex-specific percentile (McCarthy et al., 2001) was used to define overwaist. Applying these references, a z-score was calculated: (xm)/s (measured WC (x), group mean (m), standard deviation (s)). The upper age limit for these references was 17 years, resulting in 352 children/adolescents for this analysis. Blood samples were taken from 218 subjects. Body composition (fat mass, FM and fat-free mass, FFM) was assessed by air-displacement plethysmography (BodPod) and child-specific corrections were applied as described elsewhere (Bosy-Westphal et al., 2005). To define overfat, 490th percentile of FM references (McCarthy et al., 2006) was used, which had an upper age limit of 18.9 years resulting in 399 subjects for this analysis.

Assessment of REE In 312 subjects a valid measure of REE was obtained using a ventilated hood system (Vmax-model 29n, SensorMedics; Viasys Healthcare, Bilthoven, the Netherlands), which was described elsewhere (Bader et al., 2005). REE was adjusted for FFM (REEadjFFM) according to Ravussin and Bogardus (1989). A total of 298 subjects had measures of thyroid hormones and were included in our analysis.

Cardiometabolic risk factors and hormones Blood pressure was measured with a manual sphygmomanometer. Lipid profile and glucose levels were assessed enzymatically by Konelab 20i Analyzer (Konelab, Espoo, Finnland). The intra-assay coefficients of variation (CVs) were o1.2% (total cholesterol), o3.5% (high-density lipoprotein (HDL) cholesterol), o2.7% (low-density lipoprotein (LDL) cholesterol), o2.5% (triglycerides) and o2.2% (glucose). Radioimmunoassays (RIAs) were used to assess plasma insulin (Adaltis, Freiburg, Germany; CVo5.4%), serum leptin and adiponectin concentrations (Linco Research, St Charles, MO, USA). Intra-/inter-assay CVs were 3–8 and 4–6% (leptin), 2–6 and 7–9% (adiponectin). Leptin levels were referred to kg FM. Serum concentrations of thyroid-stimulating hormone (TSH; Brahms, Henningsdorf, Germany), free T3 (fT3) and T4 (fT4) (DiaSorin, Dietzenbach, Germany) were also analysed by RIA (intra-/inter-assay CVs: 2.5/5.7% (TSH), 4.6/6.5% (fT3) and 2.4/6.8% (fT4)). Insulin resistance was calculated by homoeostasis model assessment: HOMA-IR ¼ (glucose (mmol/l)  insulin (mU/ ml))/22.5 (Matthews et al., 1985).

Assessment of sleep duration and confounding factors Subjects filled out a questionnaire. Children o11 years got help from their parents, and those above 11 years completed it by themselves.

Determinants and impact of short sleep B Hitze et al

741 Sleep duration on weekdays was asked by time bar reaching half-hourly from o6 to 412 h per day and was classified into ‘short’/‘long’ according to Chen et al. (2008) applying the following cutoffs: 10 h per day for children o10 years and otherwise 9 h per day. A stratification of short sleep into ‘very short’ (o9 h per day for children o10 years and otherwise o8 h per day; n ¼ 47, girls; n ¼ 37, boys) and ‘short’ (9–10 h per day for children o10 years and otherwise 8–9 h per day) was conducted. Activity and inactivity (media consumption) were assessed by information about the membership in a sports club and time spent daily watching TV or using computers. Dietary habits were recorded by a validated Food Frequency Questionnaire. Five healthy items (whole-meal products, milk products, fruits, vegetables and potatoes, and fish) and five risk-related items (white bread, meat products, soft drinks, fast food and sweets) were analysed according to their consumption frequency (several times a week and daily vs once a week or less). A nutrition quality score was calculated with 52 points as the highest possible score (Mast et al., 1998). Hence, a low score was characterized by low consumptions of healthy and high consumptions of risk-related items and vice versa. A mean nutrition quality score of 31.8±4.4 (range: 17–43) was obtained. The highest educational level of parents was used for classification in three SES groups (low, middle and high). A subcohort (n ¼ 125) attended the Family Path Study as part of KOPS (Bosy-Westphal et al., 2006), where parental BMI was measured. Otherwise parental weight and height were self-reported, which were shown to be highly correlated with measured values (McAdams et al., 2007). Birth weight and weight at the age of 2 were recorded from children’s examination booklets (measurements took place in an official routine after birth and between the 20.5th and 29.5th months). Weight SDS at birth and around 2 years was calculated using German references taking into account children’s age as exact as possible (Kromeyer-Hauschild et al., 2001). D-Weight SDS was calculated by subtracting birth weight SDS from weight SDS at around 2 years. Because of incomplete data of 23 subjects, this analysis could be obtained in 391 subjects. For further analyses, birth weight was adjusted for gestational age (n ¼ 409). Moreover, mothers were asked about the duration of breastfeeding.

Statistical analysis Analyses were performed using SPSS 13.0 for Windows (Chicago, IL, USA). Descriptive statistics were given as median (interquartile range; IQR) or mean (95% confidence interval (CI)). Mann–Whitney U-test was used to compare independent samples. w2-Test was applied to analyse differences in frequency distributions. Pearson’s correlation was performed to demonstrate the relationship between two variables. To analyse an association while considering covariates, partial correlation was adopted. Comparison of means with regard to covariates was tested by general linear

model (analysis of covariance, ANCOVA; Bonferroni post hoc test). When calculating odds ratios (OR) for the association between ‘short’ sleep and overweight/obesity, overwaist and overfat, ‘long’ sleep was the reference. To explain the variance in sleep duration, multiple step-wise regression analyses were performed with the following independent variables: age, physical activity, media consumption, nutrition quality score, SES and change in weight SDS (birth till 2 years). To explain the variance in BMI SDS/WC z-score parental BMI, SES, birth weight, change in weight SDS (birth till 2 years), duration of breastfeeding, sleep duration, physical activity, media consumption and nutrition quality score were used as independent variables. Regression analysis was adopted to adjust for confounders. Normal distribution was tested by Kolmogorov–Smirnov test. Parameters that showed no normal distribution were log10-transformed for correlation/regression analysis. A P-value o0.05 (two sided) was considered to be statistically significant.

Results Characterization of the study population Boys had a higher body weight, height and FFM as well as a lower per cent FM and WC z-score compared to girls (Table 1). Determinants of sleep duration The variance in sleep duration was explained by age (57.0%, girls; 41.2%, boys) with an additional effect of lifestyle. A high nutrition quality score (girls) and a low media consumption (boys) could explain further 0.9/1.3%. ‘Short’ vs ‘long’ sleepers had lower physical activities (girls) and a higher media consumption (girls/boys; Table 2). However, adjustment for age weakened this association (sleep duration vs media consumption in girls: r ¼ 0.08 and boys: r ¼ 0.12; P40.05). In brief, 4.9/80.6% of girls with ‘short’ vs 0/93.3% of girls with ‘long’ sleep ate fast food/sweets several times a weak or daily (Po0.05/Po0.01) and 30.3% of boys with ‘short’ compared to 13.3% with ‘long’ sleep consumed soft drinks at a frequency of several times a weak or daily (Po0.01). Sleep duration and nutritional status Children/adolescents with ‘short’ compared to those with ‘long’ sleep were older; deductive the differences in weight, height, BMI, WC, FFM and FM could be explained by age (Table 3). However, ‘short’ vs ‘long’ sleep was associated with a higher BMI SDS (girls/boys) and WC z-score (girls). Moreover, ‘very short’ compared to ‘long’ sleepers had a higher BMI SDS (girls: 0.55 vs 0.02; boys: 0.42 vs 0.05; Po0.05) and WC z-score (girls: 1.5 vs 0.7; boys: 0.81 vs 0.42; Po0.05), whereas children/adolescents with ‘short’ sleep did not differ from those with ‘very short’ or ‘long’ sleep, which was probably due to small sample sizes. European Journal of Clinical Nutrition

Determinants and impact of short sleep B Hitze et al

742 Short sleep duration explained between 3.6 and 5.2% of the variance in BMI SDS and WC z-score (Table 4). Moreover, sleep duration had an inverse relationship with BMI SDS (girls: r ¼ 0.17, Po0.05; boys: r ¼ 0.21, Po0.01) and WC z-score (girls: r ¼ 0.19, Po0.05; boys: r ¼ 0.20, Po0.01), which was strengthened after adjusting for confounders (Table 4) for BMI SDS (girls: r ¼ 0.27, Po0.001; boys: r ¼ 0.25, Po0.01) and WC z-score (girls: r ¼ 0.30, Po0.001, boys: r ¼ 0.21; Po0.01). Using ‘long’ sleep as a reference and adjusting for confounders (Table 4), girls with ‘short’ sleep had increased risks for being obese (OR (95% CI) ¼ 5.5 (1.3–23.5)) and

overwaist (2.3 (1.2–4.6)). However, sleep duration did not influence the risk of being overfat.

Sleep duration and REE After adjusting for age, REEadjFFM was positively associated with sleep duration in boys (r ¼ 0.16, Po0.05), but not in girls (r ¼ 0.05, P40.05). In multiple step-wise regression analyses, variance in REEadjFFM was explained by higher fT3 levels in girls (5.6%). In boys lower TSH levels (4.1%) and a higher sleep duration (3.5%) explained variance in REEadjFFM, whereas age and fT4 levels did not. However, REEadjFFM did not differ between ‘short’ and ‘long’ sleepers (Table 5).

Table 1 Characterization of the study population Girls (n ¼ 207) Age (years) Weight (kg) Height (m) BMI (kg/m2) BMI SDSa Prevalence of overweight (%)b Prevalence of obesity (%)b Waist circumference (cm) Waist circumference z-scorec Body fat (%) Fat-free mass (kg) Sleep duration (h per day)

12.7 46.9 1.57 19.2 0.18

(10.5–15.6) (36.8–59.7) (1.44–1.66) (16.9–22.0) (0.45–0.86) 4.3

Sleep duration, cardiometabolic risk factors and hormones After adjusting for age, ‘short’ vs ‘long’ sleep was related to lower adiponectin levels in boys (Table 5), which was independent of WC z-score. The relationship between sleep duration and cardiometabolic risk factors/hormones adjusted for age is shown in Table 6. Although in boys no association could be found, sleep duration in girls was inversely associated with leptin levels. After adjusting for WC z-score, insulin levels and HOMA-IR were no longer associated with sleep in girls.

Boys (n ¼ 207) 13.1 51.2 1.63 19.3 0.28

(10.5–15.8) (37.4–67.4)* (1.45–1.76)*** (17.0–21.9) (0.53–0.94) 7.7

10.6

9.2

67.1 (61.0–74.8) 0.88 (0.22–1.8)

70.0 (62.0–76.6) 0.48 (0.01–1.1)**

19.0 (13.3–27.3) 38.6 (29.4–45.3) 9.0 (8.0–10.0)

12.9 (8.3–19.9)*** 42.4 (31.5–59.7)*** 9.0 (8.0–10.0)

Discussion As to the determinants of sleep duration, sleep was mostly determined by age with minor but additional effects of lifestyle. Sleep duration was inversely associated with BMI SDS/WC z-score in both genders; but ‘short’ sleep was associated with higher risks for being obese/overwaist in girls only. In boys, REEadjFFM was positively associated with sleep. ‘Short’ sleep was further related to lower adiponectin

Abbreviations: BMI, body mass index; SDS, standard deviation score. *Po0.05; **Po0.01; ***Po0.001: difference between girls and boys. Mann–Whitney U-test; median (IQR) as well as w2-test; %. a n ¼ 199 for girls and n ¼ 197 for boys. b Defined by Kromeyer-Hauschild et al. (2001) and WHO (1995). c n ¼ 179 for girls and n ¼ 173 for boys.

Table 2 Lifestyle factors and parameters of parental influence as well as early infancy according to sleep durationa Sleep duration in girlsa

Physically active (% ) Media consumption (min per day) Nutrition quality score Sleep duration (h per day) BMImother (kg/m2) BMIfather (kg/m2) Low socio-economic status (%) Birth weightadj (g)b D-Weight SDS (birth till 2 years)c Duration of breastfeeding (week)

Sleep duration in boysa

‘Short’ (n ¼ 103)

‘Long’ (n ¼ 104)

‘Short’ (n ¼ 109)

‘Long’ (n ¼ 98)

69.9 (90.0–180.0) (30.0–36.0) (7.5–8.5) (20.9–28.6) (23.7–28.1) 8.7 3266 (3046–3466) 0.16 (0.60–1.0) 24.0 (12.0–40.0)

85.6* (51.3–120.0)*** (30.0–35.0) (9.0–10.0)*** (22.3–26.9) (24.2–28.5) 7.8 3363 (3118–3719) 0.05 (0.88–0.68) 28.0 (14.5–40.0)

74.8 (120.0–255.0) (28.0–34.0) (7.5–8.5) (22.3–26.7) (24.3–28.3) 10.2 3580 (3314–3846) 0.02 (0.67–0.63) 24.0 (12.0–36.0)

85.6 (60.0–150.0)*** (28.8–35.0) (9.0–10.5)*** (22.1–26.9) (23.6–27.5) 10.4 3635 (3399–3966) 0.002 (0.67–0.78) 24.0 (13.8–36.0)

120.0 34.0 8.0 24.1 25.4

75.0 32.5 9.9 24.1 25.5

180.0 31.0 8.0 24.1 26.0

Abbreviations: BMI, body mass index; SDS, standard deviation score. *Po0.05; ***Po0.001: difference between ‘short’ and ‘long’ sleep duration. Mann–Whitney U-test; median (IQR) as well as w2-test; %. a Cutoffs for sleep duration: 10 h per day for children aged o10 years and 9 h per day for children/adolescents aged X10 years. b Adjusted for gestational age; n ¼ 103/102 (girls) and n ¼ 106/98 (boys). c n ¼ 98/98 (girls) and n ¼ 103/92 (boys).

European Journal of Clinical Nutrition

90.0 31.0 10.0 23.5 25.3

Determinants and impact of short sleep B Hitze et al

743 Table 3 Age and parameters of nutritional status according to sleep durationa Sleep duration in girlsa ‘Short’ (n ¼ 103) Age (years) Weight (kg) Height (m) BMI (kg/m2) BMI-SDSb Prevalence of overweight(%)c Prevalence of obesity (%)c Waist circumference (cm) Waist circumference z-scored Body fat (%) Fat free mass (kg)

15.3 55.9 1.65 20.8 0.3

72.3 1.2 22.4 44.7

(12.9–17.1) (46.9–65.7) (1.56–1.69) (18.7–23.0) (0.2–0.9) 2.9 14.6 (66.7–78.7) (0.5–2.4) (16.7–28.8) (39.5–48.3)

Sleep duration in boysa

‘Long’ (n ¼ 104) 10.9 39.1 1.48 17.6 0.02

62.8 0.7 15.8 32.0

(9.2–12.4)*** (30.8–46.6)*** (1.37–1.58)*** (15.9–19.8)*** (0.6–0.8)* 5.8 6.7 (57.6–68.5)*** (0.01–1.6)** (10.4–25.7)*** (25.9–38.2)***

‘Short’ (n ¼ 109) 15.0 63.5 1.73 21.0 0.4

72.7 0.6 12.9 55.2

(13.0–17.1) (48.9–76.1) (1.60–1.80) (18.5–23.7) (0.3–1.1) 8.3 12.8 (68.1–79.9) (0.08–1.7) (8.5–20.2) (40.6–64.4)

‘Long’ (n ¼ 98) 11.5 40.2 1.52 17.5 0.05

64.7 0.4 12.7 34.9

(9.3–13.0)*** (31.2–50.4)*** (1.39–1.62)*** (15.8–19.8)*** (0.7–0.7)** 7.1 5.1 (58.4–71.0)*** (0.1–1.0) (7.6–19.3) (27.9–41.6)***

Abbreviations: BMI, body mass index; SDS, standard deviation score. *Po0.05; **Po0.01; ***Po0.001: difference between ‘short’ and ‘long’ sleep duration. Mann–Whitney U-test; median (IQR) as well as w2-test; %. a Cutoffs for sleep duration: 10 h per day for children aged o10 years and 9 h per day for children/adolescents aged X10 years. b n ¼ 96/103 for girls and n ¼ 99/98 for boys. c Defined by Kromeyer-Hauschild et al. (2001) and WHO (1995). d n ¼ 76/103 for girls and n ¼ 81/92 for boys.

Table 4 Results of multiple step-wise regression analyses to explain the variance in BMI SDS and waist circumference z-score Independent variables

BMImother (kg/m2) BMIfather (kg/m2) Birth weightadj (g)a D-Weight SDS (birth till 2 years) Duration of breastfeeding (week) Sleep duration (h per day) Total explained variance

BMI SDS

Waist circumference z-score

Girls (n ¼ 190)

Boys (n ¼ 186)

Girls (n ¼ 171)

19.1 10.6 3.1 2.1

4.7 14.4 5.2 5.8 3.9 3.6 37.6

7.0 12.9 4.3 6.5

4.4 39.3

5.2 35.9

Boys (n ¼ 163)

12.8 4.0 6.6 4.4 3.7 31.5

Abbreviations: BMI, body mass index; SDS, standard deviation score. Excluded variables: socio-economic status, physical activity, media consumption and nutrition quality score. % of explained variance. a Adjusted for gestational age.

(boys) and higher leptin levels (girls), whereas the inverse relationship between sleep and insulin levels/HOMA-IR in girls depended on WC.

Determinants of sleep duration Concordant with Iglowstein et al. (2003) sleep duration was mostly determined by age. Thus, older children suffer from a greater sleep deprivation, as sleep need does not decrease (Mercer et al., 1998). Sleep duration was further determined by a healthier diet (girls) and low media consumption (boys). Moreover, fast food (girls) and soft drinks (boys) were more often consumed by ‘short’ sleepers, whereas for sweets the opposite was true (girls). Anymore, girls with ‘long’ compared to ‘short’ sleep were more physically active (Table 2). Whereas Benefice et al. (2004) could not find an association between sleep and physical activity, von Kries et al.

(2002) described that short sleep in children is associated with increased inactivity and reduced participation in organized sports, suggesting that physical activity contributes to better/longer sleep. In turn, sleep deprivation affects physical activity by fatigue (Patel and Hu, 2008). Short sleep was shown to increase eating (Sivak, 2006) and to alter feelings of hunger/appetite (Spiegel et al., 2004b). However, von Kries et al. (2002) found no association between sleep and caloric intake, whereas eating snacks while watching television was related to short sleep. Thus, further studies are needed, which should keep in mind agedependent effects, as the association between sleep and media consumption was mediated through age.

Sleep duration and nutritional status The inverse relationship between sleep duration and BMI SDS/WC z-score is concordant with previous studies European Journal of Clinical Nutrition

Determinants and impact of short sleep B Hitze et al

744 Table 5 Resting energy expenditure adjusted for FFM, cardiometabolic risk factors and hormones according to sleep durationa Sleep duration in girlsa ‘Short’ (n ¼ 55) REEadjFFM (kcal per day)b RRsys (mm Hg) RRdias (mm Hg) Triglycerides (mg/100 ml) Total cholesterol (mg/100 ml) LDL-C (mg/100 ml) HDL-C (mg/100 ml) Glucose (mg/100 ml) Insulin (mU/ml) HOMA-IR ((mmol/l)  (mU/ml)) Leptin (ng/ml) Leptin/FM ((ng/ml)/kg) Adiponectin (mg/ml) TSH (mU/L)b fT3 (pg/ml)b fT4 (pg/ml)b

1317.0 113.2 71.4 79.3 161.7 85.3 60.8 90.2 12.5 2.8 12.8 1.0 14.6 2.8 4.9 14.2

Sleep duration in boysa

‘Long’ (n ¼ 67)

(1283.4–1350.6) (109.7–116.7) (68.8–74.0) (68.8–89.8) (152.4–171.0) (77.6–93.0) (56.9–64.6) (88.2–92.2) (11.1–13.9) (2.4–3.1) (9.7–15.9) (0.9–1.2) (12.8–16.3) (1.6–4.0) (4.6–5.1) (13.6–14.8)

1362.3 111.2 69.8 80.3 165.7 90.8 58.7 89.4 10.7 2.4 10.2 0.9 14.6 4.1 4.8 13.9

(1326.1–1398.5) (108.1–114.2) (67.6–72.1) (70.9–89.6) (157.4–173.9) (84.0–97.6) (55.2–62.1) (87.7–91.2) (9.4–12.0) (2.1–2.7) (7.4–13.0) (0.8–1.0) (13.0–16.1) (2.7–5.4) (4.6–5.1) (13.2–14.5)

‘Short’ (n ¼ 46) 1587.7 114.1 70.1 79.0 155.9 83.6 56.5 93.0 12.5 2.9 5.5 0.6 11.7 3.0 5.0 14.9

(1550.9–1624.5) (110.5–117.7) (67.9–72.4) (68.5–89.5) (148.1–163.8) (76.6–90.6) (52.2–60.8) (91.1–94.9) (10.6–14.4) (2.4–3.3) (3.6–7.3) (0.5–0.8) (9.8–13.5) (2.7–3.4) (4.8–5.2) (14.3–15.5)

‘Long’ (n ¼ 50) 1597.4 114.9 69.8 70.2 160.5 83.9 61.4 93.6 10.7 2.5 5.1 0.6 14.4 2.8 4.9 14.2

(1558.7–1636.2) (111.6–118.3) (67.7–71.9) (60.1–80.3) (153.0–168.0) (77.3–90.5) (57.4–65.5) (91.7–95.4) (8.8–12.5) (2.1–2.9) (3.3–6.8) (0.5–0.7) (12.6–16.2)* (2.5–3.2) (4.7–5.1) (13.6–14.9)

Abbreviations: FM, fat mass; HDL-C, high-density lipoprotein cholesterol; HOMA-IR, homeostasis model assessment of insulin resistance; LDL-C, low-density lipoprotein cholesterol; REEadjFFM, resting energy expenditure adjusted for fat free mass; RRdias, diastolic blood pressure; RRsys, systolic blood pressure. *Po0.05: difference between ‘short’ and ‘long’ sleep duration. ANCOVA (adjusted for age); mean (95% CI). a Cutoffs for sleep duration: 10 h per day for children aged o10 years and 9 h per day for children/adolescents aged X10 years. b n ¼ 79/70 (girls) and n ¼ 78/71 (boys).

Table 6 Relationship between sleep duration and cardiometabolic risk factors as well as hormones

logRRsys

(mm Hg) (mm Hg) logTriglycerides (mg/100 ml) Total cholesterol (mg/100 ml) LDL-C (mg/100 ml) HDL-C (mg/100 ml) Glucose (mg/100 ml) logInsulin (mU/ml) logHOMA-IR ((mmol/l)  (mU/ml)) logLeptin (ng/ml) logLeptin/FM ((ng/ml)/kg) logAdiponectin (mg/ml) logRRdias

Sleep duration in girls (n ¼ 122)

Sleep duration in boys (n ¼ 96)

0.03 0.12 0.01 0.10 0.11 0.01 0.08 0.20* 0.20* 0.20* 0.23* 0.03

0.06 0.06 0.08 0.10 0.13 0.15 0.01 0.07 0.06 0.02 0.00 0.08

Abbreviations: FM, fat mass; HDL-C, high-density lipoprotein cholesterol; HOMA-IR, homeostasis model assessment of insulin resistance; LDL-C, lowdensity lipoprotein cholesterol; RRdias, diastolic blood pressure; RRsys, systolic blood pressure. *Po0.05. Partial correlation adjusted for age.

(von Kries et al., 2002; Lumeng et al., 2007; Chaput and Tremblay, 2007). Contrary to von Kries et al. (2002), we could not find an association with per cent FM independent of age (Table 3), which might be due to missing references to create age-/sex-dependent z-scores. Contrary to our results, Sekine et al. (2002), Chaput et al. (2006) and Chen et al. (2008) described that boys compared to girls were more affected by short sleep. The explanation for this gender difference remains unclear. Sleep may European Journal of Clinical Nutrition

influence weight gain differently in girls and boys (Knutson, 2005a) and a greater sleep deprivation may be needed in girls (Eisenmann et al., 2006). However, ‘short’ compared to ‘long’ sleep was related to increased risks for being obese/overwaist in girls only, suggesting a greater influence of short sleep in girls.

Sleep duration and REE An inverse association between sleep duration and REE was observed in rats, where sleep deprivation increased energy expenditure to 66% (Koban and Swinson, 2005). Moreover, sleep deprivation increased plasma cortisol levels and sympathetic activity in men suggesting stress-induced metabolism (Spiegel et al., 2004a). However, it remains unclear if short sleep increases energy needs to keep the organism awake and/or if energy expenditure is increased adaptive to an increased caloric intake. In our study, sleep duration was positively associated with REEadjFFM in boys, which could result in a positive energy balance.

Sleep duration, cardiometabolic risk factors and hormones Contrary to studies in adults (Spiegel et al., 1999, 2005; Gottlieb et al., 2006), short sleep was not associated with blood pressure, plasma lipids or glucose (Tables 5 and 6), suggesting that sleep deprivation or overweight as a result of short sleep may exist longer to affect metabolic risk. In fact, sleep duration was not related to insulin/HOMA-IR in girls independently of WC (Table 6), suggesting that insulin

Determinants and impact of short sleep B Hitze et al

745 resistance follows overweight. These results are also confirmed by Verhulst et al. (2008). Contrary to studies in adults (Spiegel et al., 2004a; Chaput et al., 2007) short sleep was associated with higher leptin levels in girls (Table 6). However, this discrepancy could be explained by the inverse relationship between sleep duration and overweight, as leptin levels reflect energy stores. Moreover, leptin concentrations were higher in children with respect to FM compared to adults, hypothisizing that children develop leptin resistance beneficial for their energy needs (Hassink et al., 1996). Furthermore, high leptin levels were associated with future weight gain (Savoye et al., 2002). Concordant with Kotani et al. (2007), who showed a positive association between sleep duration and adiponectin concentrations in men, boys with ‘short’ vs ‘long’ sleep had lower adiponectin levels (Table 5), which was independent of age and WC z-score.

Study limitations Our study has three main limitations. Sleep duration was self-reported and not measured. However, Taheri et al. (2004) described, that self-reported sleep duration is highly correlated with polysomnographic measurements and both measures are stable. Second, age ranged from 6.1 to 19.9 years. As sleep duration decreased with age (Table 3), older children/adolescents were more sleep deprived. Puberty may act as a mediator of causal linkage between sleep and metabolic risk (Knutson, 2005b), as physiological changes occur during adolescence (Hannon et al., 2006). However, Flint et al. (2007) found no association between sleep duration and pubertal status. Third, because of our crosssectional study design, we cannot deduce causation. Short sleep may also be a consequence of overweight. For example, sleep-disordered breathing leading to impaired/shorter sleep was associated with overweight (Redline et al., 2007). However, we feel, that this idea may become true for very obese only. In conclusion, short sleep was associated with a higher BMI SDS and WC z-score. However, when compared to boys ‘short’ vs ‘long’ female sleepers had higher risks for being obese/overwaist only. The inverse relationship between sleep and insulin resistance in girls was probably mediated through the development of overweight, whereas ‘short’ sleep was related to higher leptin (girls) and lower adiponectin levels (boys) independently of WC z-score. REEadjFFM was positively associated with sleep in boys. Lifestyle could add to a positive energy balance, as ‘short’ compared to ‘long’ sleepers were less physically active (girls), had higher consumptions of soft drinks (boys) and fast food (girls), but lower consumptions of sweets (girls). Besides age as the major determinant, variance in sleep duration was explained by healthier dietary habits (girls) and a low media consumption (boys).

Acknowledgements ¨ r Bildung This work was supported by Bundesministerium fu und Forschung (project 6.1.2 Network Kiel: Dietary Fat and Metabolism).

References ¨ ller MJ (2005). Intra- and Bader N, Bosy-Westphal A, Dilba B, Mu interindividual variability of resting energy expenditure in healthy male subjects—biological and methodological variability of resting energy expenditure. Br J Nutr 94, 843–849. Benefice E, Garnier D, Ndiaye G (2004). Nutritional status, growth and sleep habits among Senegalese adolescent girls. Eur J Clin Nutr 58, 292–301. Bosy-Westphal A, Danielzik S, Becker C, Geisler C, Onur S, Korth O et al. (2005). Need for optimal body composition data analysis using air-displacement plethysmography in children and adolescents. J Nutr 135, 2257–2262. Bosy-Westphal A, Geisler C, Onur S, Korth O, Selberg O, Schrezenmeir J et al. (2006). Value of body fat mass vs. anthropometric obesity indices in the assessment of metabolic risk factors. Int J Obes 30, 475–483. Chaput JP, Brunet M, Tremblay A (2006). Relationship between short sleeping hours and childhood overweight/obesity: results from the ‘Que´bec en Forme’ Project. Int J Obes 30, 1080–1085. Chaput JP, Despre´s JP, Bouchard C, Tremblay A (2007). Short sleep duration is associated with reduced leptin levels and increased adiposity: results from the Que´bec Family Study. Obesity 15, 253–261. Chaput JP, Tremblay A (2007). Does short sleep duration favor abdominal adiposity in children? Int J Pediatr Obes 2, 188–191. Chen X, Beydoun MA, Wang Y (2008). Is sleep duration associated with childhood obesity? A systematic review and meta-analysis. Obesity 16, 265–274. ¨ ller MJ Danielzik S, Czerwinski-Mast M, Langna¨se K, Dilba B, Mu (2004). Parental overweight, socioeconomic status and high birth weight are the major determinants of overweight and obesity in 5–7y-old children: baseline data of the Kiel Obesity Prevention Study. Int J Obes Relat Disord 28, 1494–1502. Dollman J, Ridley K, Olds T, Lowe E (2007). Trend in the duration of school-day sleep among 10- to 15-year-old South Australians between 1985 and 2004. Acta Pediatr 96, 1011–1014. Eisenmann JC, Ekkekakis P, Holmes M (2006). Sleep duration and overweight among Australian children and adolescents. Acta Paediatr 95, 956–963. Flint J, Kothare SV, Zihlif M, Suarez E, Adams R, Legido A et al. (2007). Association between inadequate sleep and insulin resistance in obese children. J Pediatr 150, 364–369. Gottlieb DJ, Redline S, Nieto FJ, Baldwin CM, Newman AB, Resnick HE et al. (2006). Association of usual sleep duration with hypertension: the Sleep Heart Health Study. Sleep 29, 1009–1014. Hannon TS, Janosky J, Arslanian SA (2006). Longitudinal study of physiologic insulin resistance and metabolic changes of puberty. Pediatr Res 60, 759–763. Hassink SG, Sheslow DV, De Lancey E, Opentanova I, Considine RV, Caro JF (1996). Serum leptin concentrations in children with obesity: relationship to gender and development. Pediatrics 98, 201–203. Iglowstein I, Jenni OG, Molinari L, Largo RH (2003). Sleep duration from infancy to adolescence: reference values and generational trends. Pediatrics 111, 302–307. Knutson KL (2005a). Sex differences in the association between sleep and body mass index in adolescents. J Pediatr 147, 830–834.

European Journal of Clinical Nutrition

Determinants and impact of short sleep B Hitze et al

746 Knutson KL (2005b). The association between pubertal status and sleep duration and quality among a nationally representative sample of U.S. adolescents. Am J Hum Biol 17, 418–424. Koban M, Swinson KL (2005). Chronic REM-sleep deprivation of rats elevates metabolic rate and increases UCP1 gene expression in brown adipose tissue. Am J Physiol Endocrinol Metab 289, 68–74. Kotani K, Sakane N, Saiga K, Kato M, Ishida K, Kato Y et al. (2007). Serum adiponectin levels and lifestyle factors in Japanese men. Heart Vessels 22, 291–296. Kromeyer-Hauschild K, Wabitsch M, Kunze D, Geller F, Gei HC, Hesse ¨ r den Body-mass-Index fu ¨ r das KindesV et al. (2001). Perzentile fu und Jugendalter unter Heranziehung verschiedener deutscher Stichproben. Monatsschrift Kinderheilkunde 149, 807–818. Lumeng JC, Somashekar D, Appugliese D, Kaciroti N, Corwyn RF, Bradley RH (2007). Shorter sleep duration is associated with increased risk for being overweight at ages 9 to 12 years. Pediatrics 120, 1020–1029. Locard E, Mamelle N, Billete A, Miginiac M, Munoz F, Rey S (1992). Risk factors of obesity in a five year old population. Parental versus environmental factors. Int J Obes Relat Metab Disord 16, 721–729. ¨ ller MJ (1998). Erna¨hrungsverhalten Mast M, Kortzinger I, Mu und Erna¨hrungszustand 5-7-ja¨hriger Kinder in Kiel. Aktuelle Erna¨hrungsmedizin 23, 282–288. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC (1985). Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28, 412–419. McAdams MA, Van Dam RM, Hu FB (2007). Comparison of selfreported and measured BMI as correlates of disease markers in US adults. Obesity 15, 188–196. McCarthy HD, Cole TJ, Fry T, Jebb SA, Prentice AM (2006). Body fat reference curves for children. Int J Obes 30, 598–602. McCarthy HD, Jarrett KV, Crawley HF (2001). The development of waist circumference percentiles in British children aged 5.0–16.9y. Eur J Clin Nutr 55, 902–907. Mercer PW, Merritt SL, Cowell JM (1998). Differences in reported sleep need among adolescents. J Adolesc Health 23, 259–263. Patel SR, Hu FB (2008). Short sleep duration and weight gain: a systematic review. Obesity 16, 643–653. Ravussin E, Bogardus C (1989). Relationship of genetics, age and physical fitness to daily energy expenditure and fuel utilization. Am J Nutr 49, 968–975.

European Journal of Clinical Nutrition

Redline S, Storfer-Isser A, Rosen CL, Johnson NL, Kirchner HL, Emancipator J et al. (2007). Association between metabolic syndrome and sleep-disordered breathing in adolescents. Am J Respir Crit Care Med 176, 401–408. Savoye M, Dziura J, Castle J, DiPietro L, Tamborlane WV, Caprio S (2002). Importance of plasma leptin in predicting future weight gain in obese children: a two-and-a-half-year longitudinal study. Int J Obes 26, 942–946. Sekine M, Yamagami T, Handa K, Saito T, Nanri S, Kawaminami K et al. (2002). A dose–response relationship between short sleeping hours and childhood obesity: results of the Toyama Birth Cohort Study. Child Care Health Dev 28, 163–170. Sivak M (2006). Sleeping more as a way to lose weight. Obesity Rev 7, 295–296. Spiegel K, Knutson K, Leprould R, Tasali E, Van Cauter E (2005). Sleep loss: a novel risk factor for insulin resistance and type 2 diabetes. J Appl Physiol 99, 2008–2019. Spiegel K, Leproult R, L’hermite-Baleriaux M, Copinschi G, Penev PD, Van Cauter E (2004a). Leptin levels are dependent on sleep duration: relationships with sympathovagal balance, carbohydrate regulation, cortisol, and thyrotropin. J Clin Endocrinol Metab 89, 5762–5771. Spiegel K, Leproult R, Van Cauter E (1999). Impact of sleep dept on metabolic and endocrine function. Lancet 354, 1435–1439. Spiegel K, Tasali E, Penev P, van Cauter E (2004b). Brief communication: sleep curtailment in healthy young men is associated with decreased leptin levels, elevated ghrelin levels, and increased hunger and appetite. Ann Intern Med 141, 846–850. Taheri S, Lin L, Austin D, Young T, Mignot E (2004). Short sleep duration is associated with reduced leptin, elevated ghrelin, and increased body mass index. PLoS Med 1, e62. Verhulst SL, Schrauwen N, Haentjens D, Rooman RP, Van Gaal L, De Backer WA et al. (2008). Sleep duration and metabolic dysregulation in overweight children and adolescents. Arch Dis Child 93, 89–90. von Kries R, Toschke AM, Wurmser H, Sauerwald T, Koletzko B (2002). Reduced risk for overweight and obesity in 5-and 6-y-old children by duration of sleep—a cross-sectional study. Int J Obes 26, 710–716. WHO (1995). Physical Status: The Use and Interpretation of Anthropometry Technical Report Series WHO: Geneva.