The authors have declared that no competing interests exist.
Conceived and designed the experiments: CP MP. Performed the experiments: CP FL AB CGAN MP. Analyzed the data: CP FL AB CGAN MP. Contributed reagents/materials/analysis tools: CP MP. Wrote the paper: CP MP.
To assess relationship between obesity and chronic shorter sleep duration in children and to determine if lack of sleep represents an independent determinant of childhood Body Mass Index.
This cross-sectional study was conducted in all children enrolled in the fifth class (approximately 10 years of age) of all public primary schools in Catanzaro (Southern Italy). The overall response rate was 62% resulting in 542 participating children. Parents completed a questionnaire with information on their demographics and socio-economic characteristics, their health status, characteristics of their child birth and health status. The sleeping habits were investigated in the 3 months preceding the consultation and parents were asked to indicate hours of bedtime and wake-up of their children. Multivariate linear regression analysis was performed to examine the association between child BMI and chronic lack of sleep.
36.7% of the children surveyed were overweight or obese. A quarter of children did not routinely play sports and many of them spent more than an hour a day watching TV (60.7%) and using videogames or computer (51.1%). Widespread dietary habits were inadequate, especially concerning vegetables and fruit intake with more than 95% of children who consumed insufficient amounts. The average duration of sleep was equal to 9.4 (SD = ±0.6) hours, and the short-sleepers accounted for 38.9% of the total sample. The results of multivariate analysis showed a significant 0.77 Kg/m2 increase of BMI for children classified as short compared to normal sleepers (95%CI = 0.16–1.38, p = 0.01).
Chronic lack of sleep appears to be associated to higher BMI even in middle childhood and strongly suggests that public health strategies, focused on promoting healthy lifestyles should include an innovative approach to ensure an adequate duration of sleep at night especially in children, alongside more traditional approaches.
In recent decades childhood prevalence of overweight and obesity increased worldwide from 4.2% in 1990 to 6.7% in 2010, and this trend is expected to reach 9.1% in 2020
As well as obesity and overweight, a further “epidemic” has spread in recent years, following the progressive chronic sleep deprivation in the population. Data from the National Sleep Foundation show that, from 1998 to 2005, the number of USA adults that sleep an average of 8 hours per night has decreased from 35% to 26%
Despite several public health interventions that have been carried out for many years to try to reduce the prevalence of obesity, the phenomenon is still growing. Therefore, it is necessary to identify innovative approaches, based on a clear understanding of the emerging determinants to develop preventive strategies that should complement those already established.
Relationship between obesity and chronic shorter sleep duration has undoubted relevance in terms of public health implications, since the latter represents a potentially modifiable risk factor, and can be a target of prevention campaigns for the promotion of healthy lifestyles. To date, the link between the two conditions, even though it finds support in the results of several studies, is still controversial especially in the intermediate age. We focused our interests on this specific age group and carried out a study in a sample of Southern Italian children around 10 years old, with the aim to assess the role of characteristics of sleep as an independent determinant of BMI.
The study was approved by the Institutional Ethical Committee (‘Mater Domini’ Hospital of Catanzaro, Italy, study number 2010/8).
This cross-sectional study was conducted on the children enrolled in the fifth class (approximately 10 years of age) of all public primary schools in Catanzaro (Southern Italy) during a one year period (2010–2011). The involvement of the subjects in the study took place with the written consent of their parents. Parents completed a questionnaire with information on their demographics and socio-economic characteristics, their health status (height, weight, occurrence of chronic diseases, smoking habits), characteristics of their child birth (gestational age, birth weight, breastfeeding) and health status (occurrence of diseases, surgical interventions, chronic use of drugs). For the assessment of children’s sleep characteristics, according to the procedure of Iglowstein and colleagues
All students participating in the study completed a questionnaire administered by face-to-face trained interviewers focusing on health behaviours related to obesity, such as physical activity, diet, time spent watching TV and playing videogames or using computer. Frequency of sports activities was investigated as the number of times per week spent practicing sports or getting exercise in the previous 12 months; school-related activities such as physical education were included in the count. Food habits were recorded as self reported of food groups referring to the food pyramid and taking into account both school and home dietary habits. Appropriate number of daily servings of food groups was calculated according to the Nutrition Guidelines provided by the Italian Institute of Nutrition
Examinations were conducted at school by trained examiners and then calibrated. Height and weight were recorded on bare-footed subjects, and waist circumference was measured at the umbilicus level over the T-shirt.BMI was calculated by dividing weight (in kg) by the square of height (in m). Subjects were categorized as “overweight” or “obese”, using the tables of BMI percentiles developed for 6 to 20 years of age subjects by the Italian Pediatrics Society of Diabetology and Endocrinology (SIEDP)
Data were summarized using frequencies and percentages for categorical data and mean and standard deviations for continuous data. In the primary analysis, we performed univariate regression models to examine the association between BMI and several explanatory variables, that are established on proposed risk factors of childhood obesity. Then we developed a multivariate linear regression analysis. Child BMI was the outcome variable in all models and the independent variables included were the following: gender (0 = male, 1 = female), age (continuous), maternal BMI (0 = low/normal weight; 1 = overweight/obesity), parental education’s highest level (continuous, in years), breastfeeding at least until the 3rd month (0 = no, 1 = yes), birthweight (continuous, in kilograms); daily servings of each food group (cereal: continuous; vegetable and fruit: continuous; milk and derived: continuous; meat and fish: continuous; snack: continuous), watching TV (three categories: 1 = <1 hour/day; 2 = 1–3 hours/day; 3 = >3 hours/day), playing videogames or computer use (four categories: 0 = never; 1 = <1 hour/day; 2 = 1–3 hours/day; 3 = >3 hours/day), physical activity (three categories: 0 = never; 1 = 1–3 days/week; 2 = >3 days/week). As regards to sleep duration, it was investigated as nighttime sleep duration (0 = normal sleepers, 1 = short sleepers), nighttime sleep duration (continuous) and as the difference between weekend and weekday sleep duration (continuous). We did not include pubertal status among the variables included in the model because only 3.7% of children had onset puberty. Regression coefficient (β), standard deviations (SD) and 95% confidence intervals (CIs) were calculated.
Finally, we also modeled overweight/obesity as the outcome variable of the logistic regression analysis, and results are presented as odds ratios (ORs) and 95%CIs. All reported P values are two-tailed. The data were analyzed using the Stata software program, version 11 (Stata Corporation. College Station, TX).
The overall response rate was 62% resulting in 542 participating children. The main characteristics of the examined sample are reported in
Univariate model | Multivariate model |
|||
Variable | N (% ) or Mean±DS | BMI |
Regression coefficient; SE ( |
Regression coefficient; SE ( |
261 (48.1) | 19.36±3.3 | – | ||
281 (51.9) | 19.53±3.5 | −0.17; 0.29 ( |
−0.34; 0.31 ( |
|
9.9±0.4 | 19.54±3.4 |
−0.40; 0.32 ( |
−0.24; 0.35 ( |
|
13.2±3.6 | 19.54±3.5 |
−0.005; 0.04 ( |
0.02; 0.04 ( |
|
354 (69.6) | 19.03±3.1 | – | ||
155 (30.4) | 20.5±3.8 | 1.42;0.32 ( |
1.4; 0.33 ( |
|
401 (76) | 19.46±3.4 | Reference category | ||
127 (24) | 19.47±3.4 | −0.01; 0.35 ( |
−0.02; 0.36 ( |
|
3.2±0.5 | 21.3±4.1 |
0.69; 0.28 ( |
0.59; 0.29 ( |
|
4.1±1.5 | 19.1±3.3 | 0.28; 0.10 ( |
0.19; 0.10 |
|
2.3±1 | 18.41±3 | 0.09; 0.14 ( |
0.11, 0.15 ( |
|
1.6±0.7 | 19.48±3.4 | 0.14; 0.20 ( |
0.16; 0.22 ( |
|
1.1±0.5 | 19.57±3.5 | 0.49; 0.29 ( |
0.46; 0.31 ( |
|
1.4±1.1 | 19.65±3.5 | −0.26; 0.13 ( |
−0.29; 0.14 ( |
|
−0.21; 0.27 ( |
||||
213 (39.3) | 19.49±3.4 | |||
308 (56.8) | 19.44±3.4 | −0.04; 0.30 ( |
||
21 (3.9) | 19.25±3 | −0.23; 0.78 ( |
||
0.20; 0.21 ( |
||||
116 (21.4) | 19.41±3.8 | 0.003; 0.38 ( |
||
277 (51.1) | 19.40±3.3 | |||
141 (26) | 19.58±3.4 | 0.18; 0.35 ( |
||
8 (1.5) | 19.49±3.4 | 0.09; 1.22 ( |
||
−0.44; 0.25 ( |
||||
136 (25.1) | 19.61±3.8 | |||
315 (58.1) | 19.66±3.3 | 0.06; 0.35 ( |
||
91 (16.8) | 18.5±3 | −1.1; 0.46 ( |
||
330 (61.1) | 19.1±3.3 | – | ||
210 (38.9) | 20.01±3.5 | 0.92; 0.30 ( |
0.77; 0.31 ( |
N = 496 subjects after exclusion of participants with missing data.
The mean and standard deviation (SD) of BMI of the sample is 19.45±3.4.
referred to the age of 10 years.
the number that do not add to 542 are due to missing data for the variable.
referred to the parental education of 13 years.
referred to birth weight ≥4 kilograms.
referred to recommended daily servings.
The average duration of sleep was equal to 9.4 (SD = ±0.6) hours, and short sleepers accounted for 38.9% of the total sample. Almost 60% of the children went to sleep between about 9∶30 p.m. and 10∶00 p.m. on weeknights. Moreover, children were very likely to change their sleep pattern on weekends with a mean of hours slept during weekend nights of 10.2±0.9 compared to 9.3±0.6 hours in the weekday nights (data not shown).
The results of the univariate analysis indicated that children higher BMI was significantly associated with maternal overweight and obesity (β = 1.42, 95%CI = 0.78–2.04, p<0.001) and higher birth weight (β = 0.69, 95%CI = 0.14–1.24, p = 0.01). Highlights of the analysis of the eating behaviours and physical activity were: each additional daily serving in the cereal group was significantly associated with a 0.28 (95%CI = 0.08–0.47, p = 0.005) increase in BMI, while an inverse association was noted with snacks intake (β = −0.26, 95%CI = −0.51–0.01, p = 0.04) and physical activity performed at least three times a week (β = −1.1, 95%CI = −1.1–0.2, p = 0.02). Regarding sleep duration, univariate regression model has shown a statistically significant higher BMI (β = 0.92, 95%CI = 0.34–1.5, p = 0.002) in short compared to normal sleepers.
This statistically significant association was confirmed testing the sleep duration as continuous variable (β = −0.59, 95%CI = −1.09–0.09, p = 0.02) as showed in the
When the multivariate regression analysis was performed, the results basically did not change. Indeed, a significant 1.4 Kg/m2 increase of BMI was found for children whose mother was overweight or obese (95%CI = 0.75–2.06, p<0.001) and BMI raised of 0.59 Kg/m2 for each additional kilogram of child’s birth weight (95%CI = 0.02–1.16, p = 0.04). Furthermore, higher BMI was significantly associated with a lower daily consumption of snacks (β = −0.29, 95%CI = −0.56–0.02, p = 0.04). Adjustment for explanatory variables did not substantially change the relation between children BMI and sleep pattern that was associated with a 0.77 Kg/m2 increase in short compared to normal sleepers (95%CI = 0.16–1.38, p = 0.01). When we introduced in the model the difference between weekend and weekday sleep duration as an additional variable, that may represent the degree of chronic lack of sleep sustained by the children (the more sleep is restricted during the week, the larger the weekend 'recovery' sleep response may be), thus trying to provide clues to the role of sleep restriction as opposed to absolute sleep duration as predictive of BMI, we did not find any significant statistical relationship (β = −0.09, 95%CI = −0.41–0.22, p = 0.55) (data not shown).
Then, we also modeled overweight/obesity as the outcome variable and we found a significant association with chronic lack of sleep both in univariate analysis (OR = 1.44; 95% CI = 1.01–2.06; p = 0.04) and after adjusting for children lifestyle habits (OR = 1.47; 95% CI = 1.01–2.12; p = 0.04) (data not shown).
The important new insights added by our study, concerning the association between short sleep and higher BMI, are related to the investigated age group and the objective classification of children's sleep duration by age-specific percentile curves.
Our study confirms, from a cross-sectional point of view and after adjustment for many factors influencing body weight, that chronic lack of sleep is associated with higher BMI in children around 10 years old. Several studies both cross-sectional
In regards to age, the data presented in this study provide useful insight on the relationship between insufficient sleep and BMI in a transitional stage of life, i.e. middle childhood towards adolescence. Indeed, numerous larger studies have been conducted in younger children
As regards to measurement of sleep duration, differences in definitions of short sleepers varied substantially across studies, ranging from less than 8 hours
As reported in previous studies using BMI as the outcome variable, the changes in BMI influenced by sleep duration do not seem so impressive as the ORs reported in studies using overweight/obesity as the outcome variable
Several hypothesis have been formulated on the basis of this association and it has been suggested that sleep exerts a control mechanism on the metabolic and endocrine systems and, mainly, on the ratio of ghrelin and leptin, of the profile of insulin, glucose
We conducted our analysis taking into account several known factors influencing BMI. Previous studies have shown that overweight/obesity in parents, high birth weight, physical inactivity and poor eating habits are significantly related to childhood overweight and obesity, and our results are consistent with these studies
Although this was not a main aim of our study, we found a high prevalence of short sleepers and, as expected, of overweight and obesity. Comparison with results of a survey conducted on the Italian population showed that our prevalence of overweight was aligned with that reported in the Calabria Region (26%) and higher than that in Italy (23%), while we found a lower prevalence of obesity, with a value of 7% versus 15% in Calabria and 11% in Italy
The mean total sleep duration reported in our study showed the presence of unhealthy sleeping habits in Italian children with 38.9% of children classified as short sleepers and a mean sleep duration of 9.4 (SD±0.6). Although comparisons with the results of previous survey on sleep behavior are difficult, since the measures adopted to evaluate sleep as well as data collection methods and sample characteristics were different, this result is consistent with values given by a multicentric study involving eight European countries on the distribution of sleep duration in children
As in most analogous research
Another limit of this study is that we have not investigated the qualitative aspects of sleep disorders and the causes of the chronic lack of sleep as, for example, sleep latency, anxiety, daytime sleepiness, but these were not included among our aims.
Finally, since the study was cross-sectional in design, there was no temporal separation between explanatory variables and outcomes; therefore cause and effect relationships could not be determined. However our data have contributed to the improvement of the description of the complex link between sleep and obesity.
In conclusion, our study showed that chronic lack of sleep appears to be associated with higher BMI even in middle childhood. Therefore, considering that, as in most western countries, we also found that overweight and obesity, unhealthy food and sleep habits were broadly diffused in our population, our results strongly suggest that public health strategies, focused on promoting healthy lifestyles should include an innovative approach to ensure an adequate duration of sleep at night especially in children, alongside more traditional approaches focused on balanced diet and regular physical activity. Routine paediatric evaluation of sleep pattern is also urged as well as the use of percentile curves to classify short, normal and long sleepers.