Trends and Determinants of Time in Bed in Geneva, Switzerland

Abstract

Study Objectives:

There is limited information regarding sleep duration and determinants in Switzerland. We aimed to assess the trends and determinants of time in bed as a proxy for sleep duration in the Swiss canton of Geneva.

Methods:

Data from repeated, independent cross-sectional representative samples of adults (≥ 18 years) of the Geneva population were collected between 2005 and 2011. Self-reported time in bed, education, monthly income, and nationality were assessed by questionnaire.

Results:

Data from 3,853 participants (50% women, 51.7 ± 10.9 years) were analyzed. No significant trend was observed between 2005 and 2011 regarding time in bed or the prevalence of short (≤ 6 h/day) and long (> 9 h/day) time in bed. Elderly participants reported a longer time in bed (year-adjusted mean ± standard error: 7.67 ± 0.02, 7.82 ± 0.03, and 8.41 ± 0.04 h/day for 35-50, 50-65, and 65+ years, respectively, p < 0.001), while shorter time in bed was reported by non-Swiss participants (7.77 ± 0.03 vs. 7.92 ± 0.03 h/day for Swiss nationals, p < 0.001), participants with higher education (7.92 ± 0.02 for non-university vs. 7.74 ± 0.03 h/day for university, p < 0.001) or higher income (8.10 ± 0.04, 7.84 ± 0.03, and 7.70 ± 0.03 h/day for < 5,000 SFr; 5,000-9,500 SFr, and > 9,500 SFr, respectively, p < 0.001). Multivariable-adjusted polytomous logistic regression showed short and long time in bed to be positively associated with obesity and negatively associated with income.

Conclusion:

In a Swiss adult population, sleep duration as assessed by time in bed did not change significantly between 2005 and 2011. Both clinical and socioeconomic factors influence time in bed.

Citation:

Gubelmann C, Guessous I, Theler JM, Haba-Rubio J, Gaspoz JM, Marques-Vidal P. Trends and determinants of time in bed in Geneva, Switzerland. J Clin Sleep Med 2014;10(10):1129-1135.

Short sleep duration has been associated with obesity, cardiovascular disease, and overall mortality.^1–3 A decrease in sleep duration in the general population has been reported in some studies but not in others.^4,5 A common assertion is that there is a trend toward fewer hours of sleep in most industrialized countries.^6,7 Still, a recent review showed that self-reported sleep duration increased in seven countries and decreased in six others, and that the absolute changes were very small: increases ranged from 0.1 to 1.7 min per night each year, while decreases ranged from 0.1 to 0.6 min per night each year.⁸ Another study conducted in adults from 10 industrialized countries showed an increase in the percentage of long (> 9 h/day) but not of short (≤ 6 h/day) sleepers.⁹ For instance, in the USA, the percentage of short sleepers fell from 11.7% in 1985 to 9.2% in 2007, while the corresponding values for long sleepers were 26.3% and 37.5%; in the Netherlands, the percentage of short sleepers increased from 0.4% in 1975 to 0.8% in 2005, while the corresponding values for long sleepers were 22.2% and 25.7%.⁹

BRIEF SUMMARY

Current Knowledge/Study Rationale: There is conflicting data whether sleep duration has decreased in the general population. We aimed at assessing trends and socioeconomic determinants of time in bed (as a proxy for sleep duration) in a representative sample of an adult population of a Swiss canton.

Study Impact: This study adds further weight to the proponents that time in bed did not decrease in the general population. It also adds new information regarding the importance of socioeconomic traits in time in bed.

Many socioeconomic factors such as low income, low educational level, or belonging to a minority ethnic group are positively associated with short sleep.^10,11 A study conducted in the USA showed over one hour difference in sleep duration between whites and African/Caribbean immigrants.¹² Similarly, low household income people present more insomnia-related symptoms than others.¹³ Switzerland is a European country characterized with a high socioeconomic status,¹⁴ but little is known regarding how sleep has evolved and of its associations with socioeconomic status and obesity. Hence, we analyzed seven-year (2005-2011) trends in time in bed (as a proxy for sleep duration) and in the prevalence of short and long time in bed in a representative population sample. The clinical and socioeconomic determinants of time in bed were also assessed.

MATERIALS AND METHODS

Recruitment

The Bus Santé study is a cross-sectional, population-based study conducted in the Swiss Canton of Geneva. The study was approved by the Ethical Committee of the Canton of Geneva, and a complete description has been published elsewhere.¹⁵ Each year, a representative sample of approximately 1,000 adults (500 men and 500 women) living in the Canton of Geneva is drawn. The study is conducted in 3 locations (2 in the Geneva University Hospitals and one medical mobile unit). Canton Geneva has a high frequency (41%) of non-Swiss nationals, and 40% of Swiss residents are from other cantons (www.ge.ch/statistique/domaines/apercu.asp?dom=01_02_1).

Data Collected

Self-reported sleep duration was collected by questionnaire, where the participants indicated the usual time they went to bed and the usual time of waking up. All questionnaires were checked by trained collaborators. Sleep duration was computed as the time spent in bed, a method also used in other studies.^16–18 Indeed, time in bed is easier to report and has shown an adequate correlation with sleep time calculated by actigraphy.¹⁹ Conversely, no information was available regarding time spent in bed without sleeping, and no data on depression, anxiety, or sleep complaints was collected. Time in bed was further classified as short (≤ 6 h/ day) or long (> 9 h/day).^9,20 A second classification using different thresholds for short (≤ 7 h/day) and long (> 10 h/day) time in bed was also applied. The number of participants with a very short time in bed (< 5 h) was only 26 for the entire study period, and 44 after including participants with < 4 h of bed time. Hence, no category of very short time in bed was created.

Socioeconomic status included education (university level yes/no) and income, classified into < 5,000 SFr, 5,000-9,500 SFr, and > 9,500 SFr (to convert to US$, multiply by 1.137; to convert to European €, multiply by 0.8218). For comparison, a common definition of poverty in Switzerland corresponds to a monthly income < 2,243 CHF for a single adult or < 3,990 CHF for a couple with 2 children.²¹ Demographic data included Swiss nationality (yes/no). Smoking status was classified into never, former (irrespective of time since quitting), and current smokers. Body mass index (BMI) was calculated from measured weight and height. Hypertension was defined as systolic blood pressure ≥ 140 mm Hg and/or diastolic blood pressure ≥ 90 mm Hg and/or presence of antihypertensive drug treatment; dyslipidemia was defined as the presence of a hypolipidemic drug treatment; diabetes was defined as fasting glucose ≥ 7.0 mmol/L or non-fasting glucose ≥ 10 mmol/L and/or presence of an antidiabetic drug treatment. As participants were examined during the whole day, it was not possible to have them all in the fasting state.

Participants were excluded if they had missing data, if their self-reported time in bed was < 4 or > 12 h/day, or if their BMI was < 15 or > 49 kg/m².

Statistical Analysis

Statistical analyses were conducted using Stata version 12.0 for windows (Stata Corp, College Station, Texas, USA). Descriptive results were expressed as number of participants (percentage) or as average ± standard deviation. Bivariate analyses were performed using χ² or Fisher exact test for qualitative variables and Student t-test, analysis of variance, or Kruskal-Wallis test for quantitative variables. Trends were assessed using linear regression for quantitative data and logistic regression separately for short and long sleepers (vs. reference category) as reported previously.⁹ Determinants of short or long time in bed were assessed using polytomous logistic regression (mlogit command of Stata) using the “normal” group as reference; the results were expressed as relative risk ratios (RRR) and 95% confidence interval (CI). For quantitative data, multivariable analysis was performed using analysis of variance and the results were expressed as multivariable-adjusted mean ± standard error. Statistical significance was assessed for p < 0.05.

RESULTS

Characteristics of Included and Excluded Participants

Of the initial 4,195 participants, 295 (7.0%) were excluded because of missing data for income or educational level, 29 (0.7%) because of self-reported time in bed < 4 or > 12 h/day, and 18 (0.4%) because of a BMI < 15 or > 49 kg/m² (Figure 1). Compared to participants retained for analysis, excluded participants were older, more frequently women, and had a lower educational level, while no differences were found regarding smoking status, nationality, or income (Table S1, supplemental material).

Figure 1. Exclusion procedure.

The clinical characteristics according to gender and survey year of the 3,853 participants retained for analysis are summarized in Table 1. No change occurred during the study period regarding the distribution of gender, age group, smoking status, nationality, income category, or BMI group, while the percentage of participants with a university level tended to increase.

Table 1.

Characteristics of non-excluded participants according to survey year.

Trends in Time in Bed

Trends in reported time in bed, short and long time in bed are reported in Table 2. No significant change in reported time in bed was observed. Similarly, no significant trend was found for the prevalence of short (≤ 6 h/day), or long (> 9 h/day, p = 0.25) time in bed. When the thresholds for short and long time in bed were increased to ≤ 7 and > 10 h/day, respectively, a significant decrease in the prevalence of short time in bed was found (Table 2).

Table 2.

Trends in reported time in bed and in the prevalence of short and long time in bed for the Geneva canton, 2005-2011.

Determinants of Time in Bed

The associations between time in bed or prevalence of short and long time in bed and the participants' clinical and socioeconomic characteristics are summarized in Table 3. After adjusting for survey year, women, elderly, never smokers, and Swiss nationals reported longer time in bed, while participants with university level or with high income reported shorter time in bed.

Table 3.

Reported time in bed and prevalence of short and long time in bed according to the characteristics of the participants, Geneva canton.

Women, elderly participants, and Swiss nationals had a higher prevalence of long time in bed and a lower prevalence of short time in bed. Obese participants had a higher prevalence of both short and long time in bed. Participants with a university degree or a high income had a lower prevalence of long time in bed. These findings were replicated when the thresholds to define short and long time in bed were increased or after multivariable adjustment (Table 4). Further analysis of all participants with available data did not change the results (Table S2, supplemental material).

Table 4.

Multivariable analysis of determinants of reported time in bed and prevalence of short and long time in bed, Geneva canton.

DISCUSSION

To our knowledge, this is one of the few studies assessing trends and determinants of time in bed in a representative sample of the Swiss canton of Geneva. Our results indicate that time in bed did not change during the study period (2005-2011) and that several clinical and socioeconomic factors significantly influence time in bed. The results add to the current knowledge of the association between time in bed and socioeconomic and demographic factors.

Trends in Time in Bed

Neither total time in bed nor the proportion of participants reporting short and long time in bed changed during the period 2005-2011. This finding is in agreement with some studies^8,22 but not with others.^4,23 A Finnish study found 18 minutes' decrease in self-reported sleep duration in a 33-year time interval, but the proportion of short sleep remained stable.⁴ It is possible that our study spanned too few years to observe significant changes; the ongoing data collection of the Bus Santé study will allow a better assessment within several years.

Determinants of Time in Bed

The effect of gender on sleep duration is still a matter of debate.^11,13,24 In this study, women reported longer time in bed, reported less frequently a short time in bed, and more frequently a long time in bed than men. Our results thus suggest that women tend to report longer time in bed than men, although objectively measured sleep data are needed to confirm our findings.

Contrary to other studies,^25,26 elderly participants reported longer time in bed; they also had a lower frequency of short time in bed and a higher frequency of long time in bed. A lower prevalence of insufficient sleep among elderly subjects had also been reported in the behavioral risk factor surveillance system.²⁷ A possible explanation is that retired people stay longer in bed without sleeping, thus overestimating their true sleep time. Indeed, changes in quality, quantity, and architecture of the sleep appear during aging so that elderly are less able to initiate and maintain sleep.²⁸

There is an ongoing debate on the association between obesity and short or long sleep.²⁹ Short sleep duration might be associated to endocrine and metabolic changes,³⁰ which in turn favor the development of obesity. In this study, no association was found between obesity and reported time in bed. Conversely, obesity was positively associated with both short and long time in bed irrespective of the thresholds used. This U-shaped association has been described previously for other cardiovascular risk factors, such as hypertension, type 2 diabetes, and also for cardiovascular disease and overall mortality.^31–34 It is possible that these associations between obesity and sleep might be due to different mechanisms: longer sleep durations might be a marker of low physical activity (which is associated with obesity), while short sleep durations might influence obesity via endocrine and metabolic changes.³⁰ Overall, our results suggest a U-shaped association between obesity and time in bed.

Smoking status has been associated with both short^11,24 and long²⁴ sleep. In this study, current smokers reported shorter time in bed and had a higher likelihood of reporting short time in bed. Possible explanations are the impact of nicotine on sleep duration³⁵ or that smoking is a surrogate marker of stress, which has also been associated with short sleep time.²⁴

Low educational level has been associated with lower sleep duration^13,36; another study found low educational level to be negatively associated with both short and long sleep.²⁶ In this study, participants with a university degree had a lower likelihood of reporting long time in bed, but this association was no longer significant when the threshold to define long time in bed was increased. Overall, our results suggest that educational level does not influence time in bed in the Geneva population.

Lower income has been associated with poor sleep quality³⁷ and a higher prevalence of sleep complaints.²⁰ In this study, income was negatively associated with reported time in bed in both bivariate and multivariable analyses. Interestingly, participants with an income > 9,500 SFr had lower prevalence of short and long time in bed, but no difference was found when the threshold to define short time in bed was set at ≤ 7 hours/day. Our results do not confirm previous findings from a Finnish study,¹³ where a lower income was associated with short sleep duration, but are in agreement with a Chinese study,²⁶ where higher income was inversely associated with both short and long sleep duration. It is possible that participants with a higher income have longer working hours or adopt a lifestyle that leads to shorter time in bed.³⁸ After further adjusting for cardiovascular risk factors, a similar trend between income and time in bed was observed, i.e., participants in the highest income group tended to report less time in bed, although some RRRs were no longer significant (Table S3, supplemental material). Still, these results should be considered with some caution due to possible health care renunciations in the lowest income group^39,40 and to misclassification for some cardiovascular risk factors due to non-fasting status.

There are few studies focusing on sleep duration or sleep quality among migrants.⁴¹ In this study, Swiss nationals reported longer time in bed than foreigners; Swiss nationals also had a lower likelihood of reporting short time in bed, irrespective of the threshold applied. These findings are in agreement with two studies conducted in the USA where minority ethnic groups (African Americans) had shorter sleep duration,^10,12 although this statement has been questioned.²⁰ Possible explanations include living in a noisy environment, shift work, or higher work demands.⁴¹ Still, as there was no information regarding the living environment of the participants, further studies are needed to identify the factors associated with shorter reported time in bed among migrants in Switzerland.

In this study, most factors associated with short time in bed were non-modifiable. Still, participants reporting short time in bed with a high income could be sensitized to the fact that poor sleep is associated with annual productivity losses of almost 2,000 US$,⁴² and that increasing sleep duration could actually lead to a further increase in their wealth.

Study Limitations

This study has several limitations. First, time in bed was self-reported, and a reporting bias cannot be excluded, such as participants with a higher income tending to report shorter periods in bed as a long time in bed might not be socially acceptable. It has also been shown that the correlation between objectively and subjectively measured sleep duration is rather weak,⁴³ and direct data on sleep quantity and quality were not available. Second, time in bed (and not sleep duration) was used; still, time in bed seems to be an adequate proxy for sleep duration as reasonable correlation between sleep time calculated by actigraphy and reports for bedtimes and wake times has been shown.¹⁹ Other studies also assessed sleep duration using the same methodology^16–18; this method might also be less prone to bias, as it does not require mental calculations (contrary to calculating sleep duration) and no normative values exist.⁴⁴ Furthermore, as the evaluation method for time in bed was the same for all participants and did not change during the study period, it is unlikely that our results regarding the lack of trend or the determinants of time in bed are biased. Finally, no data were collected on sleepiness, insomnia, or obstructive sleep apnea, so it was not possible to assess their prevalence or to associate time in bed with other sleep related symptoms.

CONCLUSION

Between 2005 and 2011, no changes were found regarding time in bed (as a proxy for sleep duration) or the prevalence of short or long sleepers in the adult Geneva population. Several clinical (gender, age, body mass index), behavioral (smoking), and socioeconomic (income, migrant status) factors significantly influence time in bed.

DISCLOSURE STATEMENT

This was not an industry supported study. The Bus Santé study is funded by the Geneva University Hospitals through the General Directorate of Health (Canton of Geneva). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Dr. Guessous is supported by a grant from the Swiss Foundation for Science, (33CM30-124087). Dr. Marques-Vidal received a grant unrelated to this study from the Swiss Foundation for Science. The authors have indicated no financial conflicts of interest.

SUPPLEMENTAL MATERIAL

Table S1

Characteristics of included and excluded participants.

Table S2

Multivariable analysis of determinants of reported time in bed and prevalence of short and long time in bed, Geneva canton, all participants with available data.

Table S3

Multivariable analysis of determinants of prevalence of short and long time in bed, Geneva canton, adjusting for cardiovascular risk factors.