To examine the associations between sleep problems and major weight gain during a 5- to 7-year follow-up among middle-aged women and men.
The Helsinki Health Study prospective cohort baseline survey data from 2000 to 2002 (n=8960, response rate 67%) among 40- to 60-year-old municipal employees and follow-up survey data from 2007 (n=7332, response rate 83%) were used. Logistic regression analysis was used to examine the association between the four-item Jenkins Sleep Questionnaire and major weight gain of 5 kg or more over a 5- to 7-year follow-up.
Half of the participants reported at least occasional sleep problems, whereas 13% of women and 17% of men reported no such problems at baseline. The frequency of sleep problems varied by item. Frequent sleep problems were reported by 20% of women and 17% of men. Major weight gain was reported by 25% of women and 24% of men. Trouble falling asleep (odds ratio (OR) 1.65; 95% confidence interval (CI) 1.22, 2.22), waking up several times per night (OR 1.49; 95% CI 1.22, 1.81) and trouble staying asleep (OR 1.41; 95% CI 1.13, 1.75) were associated with major weight gain during the follow-up in women but not in men. In contrast, waking up tired was unassociated with weight gain. The summary measure of the four items was also associated with weight gain in women. Adjusting for baseline body mass index, physical health, health behaviour, marital status, education, work arrangements and sleep duration had only minor effects on the above associations. Adjusting for common mental disorders at baseline, the associations were attenuated but remained for trouble falling asleep, waking up several times per night and trouble staying asleep. Occasional sleep problems were also associated with weight gain.
Sleep problems likely contribute to weight gain. To prevent major weight gain and obesity, sleep problems need to be taken into account.
Obesity has become a major public health problem during recent decades throughout the world.1 In Finland, the prevalence of obesity was 21% among both men and women in 2007.2 Sleep problems are equally prevalent and a recent Finnish study reported a slight increase in sleep problems over recent decades, especially among employed middle-aged people.3 The associations of sleep duration with body mass index (BMI) and weight gain have been examined in several studies,4, 5, 6, 7, 8, 9, 10 but studies on the associations of sleep problems with BMI and weight gain in particular are scarce. It has been suggested that sleep problems are associated with obesity.11, 12 Cross-sectional studies further suggest that objectively measured sleep problems are associated with BMI6, 12 and a higher risk of obesity.12 Few studies have examined the associations between sleep problems and subsequent weight gain, using a longitudinal design, and the evidence so far is inconsistent.13, 14, 15 In one study, men reporting sleep problems had a larger increase in BMI over a 10-year follow-up than their counterparts without such problems.14 However, other longitudinal studies found no associations between sleep problems and change in BMI.13, 15 Sleep problems and obesity are risk factors for many diseases.1, 16 Further evidence is needed for the association of sleep problems with weight gain and the determinants of this association to prevent obesity.
We lack large-scale longitudinal cohort studies examining whether sleep problems are associated with weight gain when other pertinent risk factors have been taken into account. The aim of this study was to examine the association of sleep problems and major weight gain over a 5- to 7-year follow-up among middle-aged women and men. A further aim was to examine whether the associations studied are affected by age, BMI, physical health, health behaviour, marital status, education, work arrangements, sleep duration and common mental disorders measured at baseline.
The data were derived from the Helsinki Health Study cohort mail questionnaire surveys among 40- to 60-year-old employees of the City of Helsinki, Finland. At baseline (2000–2002), 8960 participants responded to the survey (80% women, response rate 67%).17 The follow-up survey was conducted in 2007 among all respondents to the baseline survey (n=7332, response rate 83%). We excluded those who were pregnant (n=25) or had missing data on weight or height (n=169), very large differences (85, 94, 113 kg) between baseline and follow-up weight (n=3) and 4 cm or larger differences between baseline and follow-up height (n=113). The final sample with full data included 7022 respondents (5723 women and 1299 men).
The four-item Jenkins Sleep Questionnaire18 was used to measure sleep problems during the previous 4 weeks. Sleep problems were obtained from the baseline questionnaire. The following questions were asked: ‘How often in the past 4 weeks did you have the following symptoms? (1) Trouble falling asleep, (2) Waking up several times per night, (3) Trouble staying asleep and (4) Waking up after the usual amount of sleep, feeling tired and worn out?’. The six response alternatives were: not at all, 1–3 nights, 4–7 nights, 8–14 nights, 15–21 nights and 22–28 nights. The responses to each of the items were recoded on a 1- to 4-point scale. If the response was ‘not at all’, this signified no sleep problems. If the response was 1–3 nights, the respondent was classified as having rare sleep problems, and if 4–7 or 8–14 nights the respondent was classified as having occasional sleep problems. Those with more sleep problems were classified as having frequent sleep problems. In addition to separate items, a summary measure including all four items was examined. If the response to any of the four items was 15–21 or 22–28 nights, the respondent was classified as having frequent sleep problems. Intermediate groups consisted of those with rare or occasional sleep problems, whereas those reporting ‘not at all’ to all the items served as the reference group with ‘no sleep problems’. Analyses included those with responses to at least three of the four items.
The BMI was based on self-reported weight and height, which were obtained from the baseline questionnaire. Weight gain was calculated from the difference between baseline and follow-up weight. Major weight gain was defined as one of at least 5 kg during the follow-up, based on the World Health Organization (WHO) recommendation for avoiding weight gain greater than 5 kg during adult life.19
All covariates were obtained from the baseline questionnaire. Physical activity was measured by approximate metabolic equivalent tasks (METs),20 and was classified as inactive (under 14 MET-hours per week), active moderate (14–30 MET-hours per week), very active moderate (30–50) and conditioning (over 50).21 Smoking was divided into current smokers and nonsmokers. Drinking was measured by daily consumption of beer, wine and hard liquor and classified into four categories: not at all, less than four portions per week, 4–16 portions per week and 16 or more portions per week on average.22 Educational level included basic, intermediate and higher education. Marital status was classified into three categories: unmarried, cohabiting/married and divorced/widowed. Work arrangements included regular daytime work, shift work without night shifts and shift work with night shifts. Common mental disorders were measured by the 12-item General Health Questionnaire (GHQ-12).23 The reference time for each item was past 4 weeks. The summary score varied from 0 to 12 and was dichotomized into those without (scores 0–2) and those with (scores 3–12) common mental disorders. Sleep duration was measured by a question asking: ‘On average, how many hours do you sleep each weekday?’ The six response alternatives were: 5 h or less, 6, 7, 8, 9 and 10 h or more. The responses were then classified into three categories: 6 h or less, 7–8 and 9 h or more.
First, the age-adjusted prevalence percentages for weight gain by sleep problems were calculated. Next, logistic regression analysis was used to examine the associations between sleep problems and weight gain. The results are presented as odds ratios (ORs) and their 95% confidence intervals (CIs). Model 1 was adjusted for age, model 2 for age and baseline BMI as a continuous variable and model 3 additionally for physical activity, smoking and alcohol consumption. Model 4 included model 2 and was adjusted additionally for marital status, education and work arrangements. Model 5 included model 2 and was adjusted additionally for sleep duration. Model 6 included model 2 and was adjusted additionally for common mental disorders. The analyses were conducted, using the Statistical Analysis System (SAS) version 9.2 (SAS Institute, Cary, NC, USA) and the Statistical Package for the Social Sciences (SPSS) version 15.1 (SPSS (IBM), Chicago, IL, USA).
Half of the participants reported at least occasional sleep problems, whereas 13% of women and 17% of men reported no such problems at baseline. The frequency of sleep problems varied by item. At baseline, frequent sleep problems according to the summary measure were reported by 20% of women and 17% of men (Table 1).
One fourth of women and men gained 5 kg or more during the follow-up (Table 2). Women with frequent sleep problems more often gained 5 kg or more than those without sleep problems. About one-third of those with frequent sleep problems gained 5 kg or more during the follow-up, and the corresponding figure for those without sleep problems was 22%. Among men, weight gain with sleep problems was less consistent.
Next, logistic regression analyses were conducted to further examine the associations between sleep problems and weight gain (Table 3). After adjusting for age, women with frequent trouble falling asleep (OR 1.65; 95% CI 1.22, 2.22), women frequently waking up several times per night (OR 1.49; 95% CI 1.22, 1.81) or reporting trouble staying asleep (OR 1.41; 95% CI 1.13, 1.75) were more likely to have gained 5 kg or more than women without such sleep problems. Adjusting for sociodemographic factors, health behaviour, baseline BMI, work arrangements and sleep duration (models 2–5) had, but negligible, effects on the associations studied. Adjusting for baseline, common mental disorders (model 6) attenuated the associations, but each of the above sleep problems remained associated with major weight gain. However, there were no associations between waking up tired and major weight gain in any of the models.
The summary measure of all four sleep problems was associated with weight gain of 5 kg or more in women (OR 1.41; 95% CI 1.13, 1.75) after adjusting for age (Table 4). After adjusting for common mental disorders (model 6), the association was attenuated (OR 1.25; 95% CI 0.99, 1.58). Other covariates had but minor effects on the associations studied.
Similar analyses were conducted among men. However, no associations between sleep problems and weight gain could be confirmed (data not shown).
We examined the associations of sleep problems with weight gain in a cohort of middle-aged women and men over a 5- to 7-year follow-up. The main findings were as follows. Problems initiating and maintaining sleep were associated with major weight gain among women. Furthermore, a dose–response association was found, that is, the risk of weight gain was greater among those with frequent sleep problems than among those with occasional sleep problems. Nonrestorative sleep was, however, unassociated with weight gain. A summary of the four sleep problems also contributed to weight gain in women, but no associations could be confirmed among men. Owing to the smaller number of men, the power to detect effects is lower and might explain the apparent lack of associations in men.
Adjusting for baseline, BMI had a negligible effect on the associations studied. In addition, adjusting for physical health, health behaviour, marital status, education, work arrangements and sleep duration had only minor effects on the above associations. No previous study has been able to adjust for a similar broad variety of covariates.
In contrast to the other covariates, adjusting for common mental disorders somewhat attenuated the associations between sleep problems and weight gain. This suggests that there are links between sleep problems and mental health, but the interrelationships are likely to be complex.16, 24, 25
Previous longitudinal studies have not observed similar associations between sleep problems and weight changes13 or weight gain.15 In only one study on men14 was there an association between sleep problems and weight gain. In our study, nonrestorative sleep was unassociated with weight gain, but this has not been previously examined. However, a study by Ohayon26 shows that several factors, such as mental disorders, sleep deprivation and stressful life were associated with nonrestorative sleep. We observed that frequent problems in initiating sleep were somewhat more strongly associated with weight gain in women than other sleep problems. Problems maintaining sleep were also associated with weight gain, but the differences were minor. Janson et al.14 observed that sleep problems were associated with weight gain in men, but Owens et al.13 observed no associations between sleep problems and BMI or change in weight. Moreover, sleep problems have been measured by self-reports13, 14 based on physicians’ diagnoses15 or detecting sleep fragmentation.6, 12 Owing to methodological differences, findings from the various studies cannot be directly compared.
Some limitations need to be acknowledged. Weight was self-reported, both at baseline and follow-up. However, self-reported weight is not substantially biased compared with measured weight.27 Our results cannot be generalized to all middle-aged women, because we studied an occupational cohort of initially employed people from one, although a large, workplace. The nonresponse at baseline is a potential source of bias. However, our nonresponse analyses support the data in being broadly representative of the target population.28 The strengths of the study include a large, recently collected cohort and a follow-up design. Sleep problems were measured by a validated measure18 and our data included a broad variety of covariates. Weight and height were reported at two time points.
This study provides novel evidence suggesting that sleep problems contribute to major weight gain in women. Sleep problems have important implications for the prevention of weight gain and obesity.
The Helsinki Health Study is supported by grants from the Academy of Finland (No. 1129225, No. 1121748), and the Finnish Work Environment Fund (No. 107187, No. 107281). TL is supported by the Academy of Finland (Grants No. 133434, No. 130977), the Yrjö Jahnsson Foundation, and the Finnish Cultural Foundation. We also thank the City of Helsinki.