Habitual sleep patterns and the distribution of body mass index: cross-sectional findings among Swedish men and women

Highlights

• Little is known about habitual sleep in relation to the entire distribution of BMI.

• We compared BMI distributions between adults with different habitual sleep patterns.

• Short, long or poor sleepers showed larger BMI values at the upper end of the distribution.

• The middle part of the BMI distributions did not differ across sleep patterns.

• Unfavorable sleep patterns were associated with higher BMI values in a limited subset of people.

Abstract

Objective

To compare distributions of body mass index (BMI) between individuals with different habitual sleep patterns.

Methods

We performed cross-sectional analyses of 40,197 Swedish adults (64% women), who reported sleep duration and quality, weight, height, and possible confounding factors in 1997. Using quantile regression, we estimated associations between sleep patterns and selected percentiles of the distribution of BMI.

Results

While the medians were similar, larger adjusted values of BMI were estimated in the upper part of the distribution among men and women with short sleep (≤5 h) compared with medium-length sleep (6–8 h). For example, in men, the 90th percentile of BMI was 0.80 kg/m² (95% confidence interval: 0.17–1.43 kg/m²) higher among short sleepers. In women, long sleepers (≥9 h) also showed larger values in the upper part of the BMI distribution; the 90th percentile was 1.23 kg/m² (0.42–2.04 kg/m²) higher than in medium-length sleepers. In male long sleepers, smaller values were estimated in the lower part of the BMI distribution; the 10th percentile was 0.84 kg/m² lower (0.35–1.32 kg/m²) than in medium-length sleepers. The 90th percentile of BMI in women with poor-quality compared with good-quality sleep was larger by 0.82 kg/m² (0.47–1.16 kg/m²); the 10th percentile was smaller by 0.17 kg/m² (0.02–0.32 kg/m²).

Conclusions

Short, long or poor-quality sleepers showed larger, or smaller, values at the tails of the BMI distribution, but similar medians. Hence, unfavorable sleep patterns and BMI were associated only in a subset of this study population.

Introduction

Between 1980 and 2008, worldwide obesity prevalence nearly doubled from 5% to 10% in men and from 8% to 14% in women [1]. Alongside the obesity epidemic, sleep loss also has become more prevalent, as reflected in the increased percentage of people sleeping ≤6 h [2]. Hence, research into the possible link between sleep loss and obesity is mounting.

Several cross-sectional studies have reported an association between self-reported short sleep duration and higher body mass index (BMI, weight in kilograms divided by height in meters squared) or increased obesity risk [3], [4]. Whether sleep duration is related to weight gain, however, is not clear. Prospective studies, from which inferences on the direction of possible causality may be drawn, have yielded mixed results [5]. The literature on sleep quality or sleep disturbances in relation to body weight is less extensive, but studies have demonstrated that worse self-reported sleep quality or sleep disturbances are positively linked to obesity and weight gain [6], [7], [8].

Most of the previous studies analyzed the association between sleep and body weight measures using linear or logistic regression models to estimate mean BMI or the probability of obesity, respectively. Associations of sleep patterns with other parts of the BMI distribution have rarely been explored. Quantile regression, which allows for evaluation of predictors across the entire distribution of an outcome variable, such as BMI, has only been used in two studies on the relationship between sleep and body weight in adult populations. Both studies indicated that the distribution of BMI differed according to level of habitual sleep duration, so that shorter sleep was associated with higher values of BMI at the upper part of the distribution, particularly in men [9], [10]. Better sleep quality was associated with lower values of BMI at the lower part of the BMI distribution in women only [10]. Limited evidence thus suggests that the association with short or poor sleep varies across the distribution of BMI. Similar studies have not been conducted in European populations.

Analyses taking account of the entire distribution of BMI have the potential of providing a more comprehensive picture of the nature of the association between sleep and body weight. They also offer an opportunity to identify subgroups of the population that could be especially susceptible to the putative effects of poor sleep habits. Those subgroups would likely not be identified by approaches analyzing the influence of sleep habits on mean BMI [11]. Furthermore, the vast majority of studies within this research area have focused on sleep duration, typically using one sleep question. There are considerably fewer data on sleep quality and other potentially relevant dimensions of sleep in relation to body weight.

Our aim was to compare the entire distributions of BMI between people with different habitual sleep patterns using quantile regression as analytical approach in a large sample of Swedish men and women. Self-reported sleep assessments were wide-ranging and included sleep duration separately for weekdays and weekends, sleep quality, restorative power of sleep, and daytime sleepiness.