Abstract
Only a few studies have investigated the metabolic consequences of social jetlag. Therefore, we examined the association of social jetlag with the metabolic syndrome and type 2 diabetes mellitus in a population-based cohort. We used cross-sectional data from the New Hoorn Study cohort (n = 1585, 47% men, age 60.8 ± 6 years). Social jetlag was calculated as the difference in midpoint sleep (in hours) between weekdays and weekend days. Poisson and linear regression models were used to study the associations, and age was regarded as a possible effect modifier. We adjusted for sex, employment status, education, smoking, physical activity, sleep duration, and body mass index. In the total population, we only observed an association between social jetlag and the metabolic syndrome, with prevalence ratios adjusted for sex, employment status, and educational levels of 1.64 (95% CI 1.1-2.4), for participants with >2 h social jetlag, compared with participants with <1 h social jetlag. However, we observed an interaction effect of median age (<61 years). In older participants (≥61 years), no significant associations were observed between social jetlag status, the metabolic syndrome, and diabetes or prediabetes. In the younger group (<61 years), the adjusted prevalence ratios were 1.29 (95% CI 0.9-1.9) and 2.13 (95% CI 1.3-3.4) for the metabolic syndrome and 1.39 (95% CI 1.1-1.9) and 1.75 (95% CI 1.2-2.5) for diabetes/prediabetes, for participants with 1-2 h and >2 h social jetlag, compared with participants with <1 h social jetlag. In conclusion, in our population-based cohort, social jetlag was associated with a 2-fold increased risk of the metabolic syndrome and diabetes/prediabetes, especially in younger (<61 years) participants.
Keywords: social jetlag, metabolic syndrome, type 2 diabetes mellitus, age, population-based
Despite vast efforts toward prevention, the metabolic syndrome and type 2 diabetes mellitus are still highly prevalent in developed countries (International Diabetes Federation [IDF], 2015). Age, obesity, and lifestyle are well-known risk factors (IDF, 2015), and recently, disturbance of the circadian rhythm was added to this list (Qian and Scheer, 2016). Shift work and time zone travel are examples of disruption of our circadian rhythm, with light and activity during the circadian night and sleep during the circadian day (Roden et al., 1993), which are associated with a 2-fold higher risk of both the metabolic syndrome and type 2 diabetes mellitus (Scheer et al., 2009; Gonnissen et al., 2012; Gonnissen et al., 2013; Gan et al., 2015).
A less known but more chronic disruption of the circadian rhythm is social jetlag. Social jetlag represents the discrepancy between circadian and social clocks; people often use alarm clocks and/or medication to align their sleep and wake times with social obligations (e.g., work and school schedules) rather than with their internally regulated sleep-wake times (Wittmann et al., 2006; Roenneberg et al., 2012). Social jetlag is highly prevalent; 69% of adults reported at least 1 h of social jetlag (Wittmann et al., 2006; Roenneberg and Merrow, 2007; Roenneberg et al., 2012; Rutters et al., 2014), and it is suggested to disturb physiological processes, such as blood pressure and glucose metabolism.
To date, only a few studies have investigated the metabolic consequences of social jetlag. Social jetlag was associated with a higher body mass index (BMI) in an overweight population (Roenneberg et al., 2012; Parsons et al., 2015) as well as increased heart rate and cortisol levels in the general population (Kantermann et al., 2013; Rutters et al., 2014). To date, only 2 smaller cohorts (n < 1000) showed an increased prevalence of the metabolic syndrome with social jetlag (Parsons et al., 2015; Wong et al., 2015) among participants younger than 50 years. In addition, a study of the association between social jetlag and type 2 diabetes mellitus is still lacking. Therefore, the aim of this study was to examine the association of social jetlag with the metabolic syndrome and type 2 diabetes mellitus in a population-based cohort. As the prevalence of social jetlag decreases with age (Roenneberg et al., 2003), while the prevalence of the metabolic syndrome and diabetes or prediabetes increases with age, we assessed whether age was an effect modifier in this cohort of people aged 50 and older.
Methods
Study Population
The New Hoorn Study (NHS) is a population-based cohort that is representative of the general Dutch population. As depicted in Figure 1, from 2006 to 2007, a population-based survey of glucose tolerance was carried out in the Dutch city of Hoorn, which is a medium-sized town of about 70,000 residents with a mixed rural and urban population. The eligible population of the NHS consisted of 6180 men and women aged 40 to 75 years randomly selected from the municipal registry. Of the eligible participants, 45% agreed to participate, resulting in the NHS cohort of 2807 participants (van ‘t Riet et al., 2010).
Figure 1.
Flow chart of the study population of the New Hoorn Study cohort.
Between January 2013 and December 2015, 1734 participants agreed to participate in follow-up. As the questionnaire on social jetlag was introduced in 2013-2015, for the current analysis we use only data from the 2013-2015 measurement. Of the 1734 participants, we excluded those performing shift work (n = 94) and those with no information regarding sleep-related measures (n = 47) or metabolic syndrome, diabetes, or prediabetes (n = 8), leaving 1585 participants for analysis. The Ethics Committee of the VU University Medical Centre approved the New Hoorn Study, and written informed consent was obtained from all participants.
Study Design
Participants were asked to refrain from eating and drinking (except drinking water) from 20:00 h the night before the visit and from drinking alcohol and smoking from 17:00 h the day before the visit. Participants who did not follow these instructions were asked to reschedule their visit. Before the visit, participants were asked to fill out questionnaires about their physical activities, medical history, quality of life, diet, and sleeping behavior. During the visit, we determined anthropometrics and blood pressure, collected a fasting blood sample, administered questionnaires, and noted self-reported medication. All visits were scheduled on weekdays (van ‘t Riet et al., 2010).
Social Jetlag Status
Social jetlag status was determined using an adapted version of the Munich ChronoType Questionnaire (MCTQ) (Roenneberg et al., 2003), which questions typical bedtime and wake times on weekdays and weekend/free days (Reutrakul et al., 2013). This questionnaire has shown good validity and reliability, with actual sleep time on weekdays and weekend/free days correlating highly (r ~ 1, p < 0.001) (Roenneberg et al., 2003; Roenneberg et al., 2012). Social jetlag was measured as the difference in hours in midpoint of sleep between workdays and free days (Wittmann et al., 2006). For example, when a person sleeps from 22:00 h until 06:00 h on weekdays, the midpoint is 02:00 h, and when one sleeps from 00:00 h until 10:00 h on weekend/free days, the midpoint is 05:00 h, which results in a 3-h social jetlag. In our population, social jetlag ranged from 0 to 4.5 h, which resulted in a skewed distribution. After log transformation, the distribution was still skewed. We therefore stratified the participants per hour of social jetlag, resulting in 3 groups: no social jetlag (<1 h), 1-2 h social jetlag, or >2 h social jetlag, as previously done by Roenneberg et al. (2012) and Rutters et al. (2014). We could not make 4 groups, as only 0.4% of the population had >3 h social jetlag.
Metabolic Syndrome
Metabolic syndrome status (yes/no) was defined according to the Adult Treatment Panel III of the National Cholesterol Education Program. That is, participants had to meet 3 or more of the following criteria: fasting plasma glucose levels ≥6.1 mmol/L, high-density lipoprotein (HDL) cholesterol levels <1.0 mmol/L in men or <1.3 mmol/L in women, triglyceride levels ≥1.7 mmol/L, waist circumference ≥102 cm in men or ≥88 cm in women, hypertension defined by ≥130/85 mmHg or use of hypertension medication (Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults, 2001).
Glucose levels (mmol/L) were determined from a fasting blood sample, using the glucose dehydrogenase method (Merck, Darmstad, Germany). Levels of HDL cholesterol (HDL-C, mmol/L) and triglycerides (mmol/L) were determined using enzymatic techniques (Boehringer Mannheim, Mannheim, Germany). Waist circumference (cm) was measured according to a standardized procedure (Seidell et al., 1988). Sitting blood pressure (mmHg) was measured 3 times on the right arm with a random-zero sphygmomanometer (Hawksley-Gelman, Lancing, UK), and the average between the 2 measures with the smallest difference between them was used in the statistical analyses.
Prediabetes and Type 2 Diabetes Mellitus
Prediabetes or type 2 diabetes mellitus status (yes/no) was classified according to the criteria of the WHO Consultation of 2011; prediabetes was defined as fasting plasma glucose levels ≥6.1 mmol/L and HbA1c levels ≥6.0% or ≥44 mmol/mol, and type 2 diabetes mellitus was defined as fasting plasma glucose levels ≥6.5 mmol/L, HbA1c levels ≥6.5% or ≥ 48 mmol/mol, and/or treatment for diabetes, including medication and diet (Colagiuri, 2011). Due to budget restraints, HbA1c levels were measured only in those with a higher risk for diabetes (n = 881), as determined using the DIRECT prediction model (Rauh et al., 2017).
Covariates
Demographic factors were determined by questionnaires. Educational level was measured in 8 categories based on highest attainment; we stratified these categories into low educational level containing categories 1 and 2 (no education, primary school), middle educational level containing categories 3 to 5 (secondary education), and high educational level containing categories 6 to 8 (tertiary education). Job status was based on current status: having a job versus no job, the latter including unemployed people, housewives, and retirees. Smoking habits (current and previous smoking: yes/no) and physical activity were assessed using questionnaires. Low physical activity (yes/no) was defined as fewer than 5 days of 30 min of physical activity, the Dutch exercise norm. Average sleep duration (hours) was calculated as the difference between sleep onset until wake time from the MCTQ for weekdays and weekend days. Finally, weight and height were measured with participants wearing light clothes only, and we calculated BMI as weight divided by height squared (kg/m2).
Statistical Analysis
First, to assess the participants’ characteristics per social jetlag status, we determined the characteristics as percentages for dichotomous variables, as means and standard deviations for normally distributed continuous variables, and as medians and interquartile ranges for nonnormal distributions. To address missing values for confounders, we used imputation via mean substitution. To assess possible bias, we used t tests to compare participant characteristics for participants with and without follow-up as well as for participants with and without social jetlag data.
Second, we assessed the association between social jetlag and the metabolic syndrome as well as type 2 diabetes mellitus. As former and the latter are dichotomous outcome variables, we wanted to use logistic regression. However, as the high prevalence of the metabolic syndrome would have led to overestimation of relative risks, we used modified Poisson regression models (Zou, 2004; Knol et al., 2012). We reported the association between social jetlag and the metabolic syndrome as well as type 2 diabetes mellitus as prevalence ratios and 95% confidence intervals (CI) and p for trend.
Third, because the prevalence of social jetlag decreases with age (Roenneberg et al., 2003) whereas the prevalence of the metabolic syndrome and diabetes/prediabetes increases with age, we assessed whether age was an effect modifier. A p value ≤ 0.10 was considered to be statistically significant for effect modification, in concordance with Twisk et al. (2010). In addition, we assessed possible confounding by sex, education, and employment status as well as possible confounding or mediating effects of smoking, physical activity, and BMI. We therefore corrected the association for these variables and reported 5 models, namely the unadjusted model (model 1); a model adjusted for sex, employment status, and educational level (model 2, the main model); a model adjusted for sex, employment status, educational level, smoking, and physical activity (model 3); a model adjusted for adjusted for sex, employment status, educational level, and BMI (model 4); and a model adjusted for adjusted for sex, employment status, educational level, and sleep duration (model 5).
Fourth, we assessed the association between social jetlag and the continuous parameters of the metabolic syndrome. We therefore used linear regression models and reported the unadjusted model and adjusted models as described above. Statistical analyses were performed with SPSS version 20.0 (SPSS Inc., Chicago, IL). Overall, a p value ≤ 0.05 was considered to be statistically significant.
Results
Our population-based cohort consisted of 1585 participants, of whom 47% were male, and had a mean age of 60.8 ± 6 years. Comparing participants at follow-up with those lost to follow-up, we observed no difference in sex (45% vs. 48%, p = 0.10), but participants without follow-up were slightly younger (53.0 vs. 53.6 years, p = 0.02) and less often had the metabolic syndrome (22% vs. 25%, p = 0.03) and more often had type 2 diabetes mellitus (10% vs. 4%, p = 0.01), compared with those included in our study. In addition, in those excluded due to having no social jetlag information (n = 47), we observed no significant difference in sex (47% vs. 58%, p = 0.12), age (60.6 vs. 61.9 years, p = 0.20), metabolic syndrome prevalence (19% vs. 18%, p = 0.78), or type 2 diabetes mellitus prevalence (11% vs. 19%, p = 0.07), compared with those with social jetlag information.
The baseline characteristics for the participants included in our analyses are described in Table 1, stratified for social jetlag. In our cohort, 61% of the participants reported not having social jetlag (<1 h social jetlag), 31% reported having 1-2 h social jetlag, and 8% reported having >2 h social jetlag. We observed differences between the social jetlag groups regarding age, those participants with 1-2 h and >2 h social jetlag being younger compared with participants without social jetlag (57 ± 5 years and 57 ± 5 years vs. 64 ± 6 years, p < 0.001). In addition, in the older participants (median age split, ≥61 years) only 24% and 19% had 1-2 h or >2 h social jetlag. As the prevalence of the metabolic syndrome and diabetes/prediabetes increases with age (stratified for median age, 61 years), 18% versus 22% and 26% versus 38%, respectively, we assessed whether age was an effect modifier.
Table 1.
Participant characteristics stratified for social jetlag (n = 1585).
| Social Jetlag <1 h | Social Jetlag 1-2 h | Social Jetlag >2 h | |
|---|---|---|---|
| Demographic variables | |||
| Number (%) of participants | 971 (61) | 497 (31) | 117 (8) |
| Sex, % men | 46 | 47 | 59 |
| Age, years | 64 ± 6 | 57 ± 5 | 57 ± 5 |
| Median age, % ≥61 years | 68 | 24 | 19 |
| Educational level, % low level | 6 | 3 | 2 |
| Job status, % with paid job | 27 | 78 | 87 |
| Behavioral variables | |||
| Social jetlag, min | 21.7 ± 17 | 78.9 ± 17 | 139.6 ± 24 |
| Average sleep duration, h | 8.2 ± 1 | 7.8 ± 1 | 7.8 ± 1 |
| Smoking, current/previous, % | 59 | 58 | 62 |
| Low physical activity, % | 31 | 38 | 38 |
| Metabolic variables | |||
| Fasting glucose, mmol/L | 5.9 ± 1 | 5.8 ± 1 | 5.9 ± 1 |
| High-density lipoprotein cholesterol, mmol/L | 1.57 ± 0.5 | 1.59 ± 0.5 | 1.53 ± 0.4 |
| Triglycerides, mmol/L | 1.35 [0.9-1.6] | 1.33 [0.8-1.6] | 1.37 [0.8-1.9] |
| Waist circumference, cm | 91.2 ± 11 | 90.9 ± 12 | 93.8 ± 11 |
| Systolic blood pressure, mmHg | 133 ± 19 | 127 ± 18 | 129 ± 17 |
| Diastolic blood pressure, mmHg | 79 ± 8 | 78 ± 9 | 80 ± 9 |
| Use hypertension medication, % | 33 | 23 | 23 |
| Body mass index, kg/m2 | 26.5 ± 4 | 26.4 ± 4 | 27.1 ± 4 |
| HbA1c, mmol/mol (n = 920) | 42.4 ± 6 | 41.5 ± 5 | 42.8 ± 6 |
| HbA1c, % (n = 920) | 6 ± 2.7 | 5.9 ± 2.6 | 6 ± 2.7 |
| Metabolic syndrome, % | 21 | 18 | 26 |
| Prediabetes and type 2 diabetes, %/% | 33/14 | 30/8 | 34/11 |
Characteristics are presented as percentage, mean ± standard deviation, or median [interquartile range].
First, we assessed the associations between social jetlag status and prevalence of the metabolic syndrome and diabetes/prediabetes, as depicted in Table 2. We observed no significant association between social jetlag status and diabetes/prediabetes. Additionally, we observed that social jetlag was associated with an increased prevalence of metabolic syndrome. After adjustment for sex, employment status, and educational level, prevalence ratios of 1.15 (95% CI 0.9-1.5) were observed for participants with 1-2 h social jetlag and 1.64 (95% CI 1.1-2.4) for participants with >2 h social jetlag, compared with participants with <1 h social jetlag (p for trend = 0.04). We observed confounding effects in model 2 and models 3-5, which showed that the association was confounded or mediated by BMI.
Table 2.
Prevalence ratios (with 95% confidence intervals) of the metabolic syndrome and diabetes/prediabetes per social jetlag groups in the New Hoorn Study cohort (n = 1585).
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | ||
|---|---|---|---|---|---|---|
| Metabolic syndrome | Social jetlag <1 h | [reference] | [reference] | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.87 (0.7-1.1) | 1.15 (0.9-1.5) | 1.04 (0.8-1.4) | 1.03 (0.8-1.3) | 1.11 (0.9-1.5) | |
| Social jetlag >2 h | 1.23 (0.9-1.7) | 1.64 (1.1-2.4) | 1.51 (1.1-2.2) | 1.25 (0.9-1.8) | 1.62 (1.1-2.4) | |
| p for trend | 0.14 | 0.04 | 0.10 | 0.47 | 0.10 | |
| Diabetes/prediabetes | Social jetlag <1 h | [reference] | [reference] | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.89 (0.8-1.1) | 0.99 (0.8-1.2) | 0.99 (0.8-1.2) | 0.96 (0.8-1.2) | 0.99 (0.8-1.2) | |
| Social jetlag >2 h | 1.02 (0.8-1.3) | 1.11 (0.8-1.5) | 1.10 (0.8-1.5) | 0.98 (0.8-1.3) | 1.10 (0.8-1.5) | |
| p for trend | 0.76 | 0.75 | 0.77 | 0.92 | 0.75 |
Model 1: unadjusted. Model 2: adjusted for sex, employment status, and educational level. Model 3: adjusted for sex, employment status, educational level, smoking, and physical activity. Model 4: adjusted for sex, employment status, educational level, and body mass index. Model 5: adjusted for employment status, educational level, and sleep duration. Bold values indicate significant association.
We observed effect modification by age (p = 0.01 for metabolic syndrome and p = 0.02 for diabetes/prediabetes) and therefore stratified the analyses for median age (≥61 years). In the participants older than 61 years, no significant associations were observed between social jetlag status and prevalence of the metabolic syndrome and diabetes/prediabetes (Tables 3 and 4). However, in the younger group (<61 years), social jetlag was associated with an increased risk of metabolic syndrome and diabetes/prediabetes. After adjustment for sex, employment status, and educational level, prevalence ratios of 1.29 (95% CI 0.9-1.9) were observed for participants with 1-2 h social jetlag and 2.13 (95% CI 1.3-3.4) for participants with >2 h social jetlag, compared with participants with <1 h social jetlag (Table 3) for the metabolic syndrome. In addition, we observed prevalence ratios of 1.39 (95% CI 1.1-1.9) for participants with 1-2 h social jetlag and 1.75 (95% CI 1.2-2.5) for participants with >2 h social jetlag, compared with participants with <1 h social jetlag (Table 4) for diabetes/prediabetes. We observed confounding effects in model 2 and confounding/mediating effects in model 3-5, but the association remained significant.
Table 3.
Prevalence ratios (with 95% confidence intervals) of the metabolic syndrome per social jetlag groups stratified for age in the New Hoorn Study cohort (n = 1585).
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | ||
|---|---|---|---|---|---|---|
| Metabolic syndrome, <61 years | Social jetlag <1 h | [reference] | [reference] | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 1.13 (0.8-1.6) | 1.29 (0.9-1.9) | 1.15 (0.8-1.7) | 1.13 (0.8-1.6) | 1.22 (0.8-1.8) | |
| Social jetlag >2 h | 1.91 (1.3-2.9) | 2.13 (1.3-3.4) | 1.83 (1.1-2.9) | 1.55 (1.0-2.4) | 2.06 (1.3-3.3) | |
| p for trend | 0.01 | 0.01 | 0.04 | 0.13 | 0.01 | |
| Metabolic syndrome, ≥61 years | Social jetlag <1 h | [reference] | [reference] | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.86 (0.6-1.3) | 1.04 (0.7-1.6) | 0.94 (0.6-1.5) | 0.99 (0.6-1.5) | 1.04 (0.7-1.6) | |
| Social jetlag >2 h | 0.39 (0.1-1.5) | 0.56 (0.2-2.2) | 0.64 (0.2-2.5) | 0.51 (0.1-1.9) | 0.58 (0.2-2.2) | |
| p for trend | 0.29 | 0.70 | 0.79 | 0.61 | 0.70 |
Model 1: unadjusted. Model 2: adjusted for sex, employment status, and educational level. Model 3: adjusted for sex, employment status, educational level, smoking, and physical activity. Model 4: adjusted for sex, employment status, educational level, and body mass index. Model 5: adjusted for employment status, educational level, and sleep duration. Bold values indicate significant association.
Table 4.
Prevalence ratios (with 95% confidence intervals) of diabetes/prediabetes per social jetlag groups stratified for age in the New Hoorn Study cohort (n = 1585).
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | ||
|---|---|---|---|---|---|---|
| Diabetes/prediabetes, <61 years | Social jetlag <1 h | [reference] | [reference] | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 1.35 (1.0-1.8) | 1.39 (1.1-1.9) | 1.40 (1.0-1.9) | 1.34 (1.0-1.8) | 1.39 (1.0-1.9) | |
| Social jetlag >2 h | 1.79 (1.3-2.5) | 1.75 (1.2-2.5) | 1.67 (1.1-2.5) | 1.50 (1.1-2.1) | 1.73 (1.2-2.5) | |
| p for trend | 0.01 | 0.01 | 0.02 | 0.05 | 0.01 | |
| Diabetes/prediabetes, ≥61 years | Social jetlag <1 h | [reference] | [reference] | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.94 (0.7-1.2) | 0.89 (0.7-1.2) | 0.88 (0.7-1.2) | 0.88 (0.7-1.2) | 0.89 (0.7-1.2) | |
| Social jetlag >2 h | 0.58 (0.3-1.3) | 0.54 (0.3-1.2) | 0.59 (0.3-1.2) | 0.50 (0.3-1.1) | 0.54 (0.3-1.2) | |
| p for trend | 0.36 | 0.21 | 0.28 | 0.11 | 0.21 |
Model 1: unadjusted. Model 2: adjusted for sex, employment status, and educational level. Model 3: adjusted for sex, employment status, educational level, smoking, and physical activity. Model 4: adjusted for sex, employment status, educational level, and body mass index. Model 5: adjusted for employment status, educational level, and sleep duration. Bold values indicate significant association.
We also assessed the association between social jetlag and the parameters of metabolic syndrome factors separately (Tables 5 and 6). In the participants ≥61 years old, after adjustment for sex, employment status, and educational level, no significant associations were observed between social jetlag status and the parameters of the metabolic syndrome. In the younger group (<61 years old), after correcting for sex, employment level, and educational level, we observed 0.33 mmol/L (95% CI 0.1-0.5) higher glucose levels as well as 2.94 cm (95% CI 0.2-5.7) higher waist circumference in participants with >2 h social jetlag, compared with those without (<1 h) social jetlag. The association between social jetlag, glucose levels, and waist circumference was confounded by employment status, educational level, and sleep duration, decreasing the association, but still being significant. No differences were observed with regard to the parameters of the metabolic syndrome in participants with 1-2 h social jetlag, compared with those without social jetlag.
Table 5.
Unstandardized betas (with 95% confidence intervals) for the association between social jetlag and parameters of the metabolic syndrome for participants <61 years old.
| Model 1 | Model 2 | Model 3 | ||
|---|---|---|---|---|
| Glucose, mmol/L | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.07 (–0.1 to 0.2) | 0.09 (–0.1 to 0.2) | 0.10 (–0.1 to 0.2) | |
| Social jetlag >2 h | 0.34 (0.1 to 0.5) | 0.33 (0.1 to 0.5) | 0.34 (0.1 to 0.5) | |
| High-density lipoprotein cholesterol, mmol/L | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | –0.02 (–0.1 to 0.1) | –0.01 (–0.1 to 0.1) | –0.01 (–0.1 to 0.1) | |
| Social jetlag >2 h | –0.08 (–0.2 to 0.1) | –0.02 (–0.2 to 0.1) | –0.02 (–0.1 to 0.1) | |
| Log triglycerides, mmol/L | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | –0.01 (–0.1 to 0.1) | –0.01 (–0.1 to 0.1) | –0.01 (–0.1 to 0.1) | |
| Social jetlag >2 h | 0.02 (–0.1 to 0.1) | 0.02 (–0.1 to 0.1) | 0.02 (–0.1 to 0.1) | |
| Waist circumference, cm | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.89 (–0.9 to 2.6) | 0.93 (–0.9 to 2.6) | 0.66 (–1.2 to 2.5) | |
| Social jetlag >2 h | 3.93 (1.2 to 6.7) | 2.94 (0.2 to 5.7) | 2.78 (0.1 to 5.6) | |
| Systolic blood pressure | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.65 (–1.9 to 3.3) | 0.73 (–2.0 to 3.5) | 0.57 (–2.2 to 3.4) | |
| Social jetlag >2 h | 2.17 (–1.8 to 6.2) | 1.79 (–2.4 to 5.9) | 1.69 (–2.5 to 5.9) |
Model 1: unadjusted. Model 2: adjusted for sex, employment status, and educational level. Model 3: adjusted for employment status, educational level, and sleep duration.
Table 6.
Unstandardized betas (with 95% confidence intervals) for the association between social jetlag and parameters of the metabolic syndrome for participants ≥61 years old.
| Model 1 | Model 2 | Model 3 | ||
|---|---|---|---|---|
| Glucose, mmol/L | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | –0.19 (–0.4 to 0.1) | –0.19 (–0.4 to 0.1) | –0.19 (–0.4 to 0.1) | |
| Social jetlag >2 h | –0.41 (–0.8 to 0.1) | –0.44 (–0.9 to 0.1) | –0.43 (–0.8 to −0.1) | |
| High-density lipoprotein cholesterol, mmol/L | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.05 (–0.1 to 0.1) | 0.03 (–0.1 to 0.1) | 0.03 (–0.1 to 0.1) | |
| Social jetlag >2 h | –0.05 (–0.2 to 0.1) | –0.03 (–0.2 to 0.1) | –0.03 (–0.2 to 0.1) | |
| Log triglycerides, mmol/L | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | –0.01 (–0.1 to 0.1) | –0.01 (–0.1 to 0.1) | –0.01 (–0.1 to 0.1) | |
| Social jetlag >2 h | –0.04 (–0.1 to 0.1) | –0.03 (–0.1 to 0.1) | –0.03 (–0.1 to 0.1) | |
| Waist circumference, cm | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | 0.93 (–0.9 to 2.7) | 0.37 (–1.8 to 2.5) | 0.37 (–1.8 to 2.6) | |
| Social jetlag >2 h | 2.94 (0.2 to 5.7) | 1.92 (–2.45 to 6.3) | 1.92 (–2.5 to 6.2) | |
| Systolic blood pressure | Social jetlag <1 h | [reference] | [reference] | [reference] |
| Social jetlag 1-2 h | –5.83 (–9.6 to −2.1) | –4.04 (–8.1 to 0.1) | –4.01 (–8.1 to 0.1) | |
| Social jetlag >2 h | –1.58 (–9.7 to 6.5) | –0.11 (–8.3 to 8.2) | –0.1 (–8.4 to 8.1) |
Model 1: unadjusted. Model 2: adjusted for sex, employment status, and educational level. Model 3: adjusted for employment status, educational level, and sleep duration.
Discussion
The aim of this study was to examine the association of social jetlag with the metabolic syndrome and type 2 diabetes mellitus, in a population-based cohort. In our total cohort, we observed an association between social jetlag and the metabolic syndrome only for participants with >2 h social jetlag, compared with participants with <1 h social jetlag. However, we observed an interaction effect of median age (<61 years). We observed that having >2 h social jetlag was associated with an approximate 2-fold increased risk of the metabolic syndrome and diabetes mellitus or prediabetes, compared with participants without social jetlag in the younger (<61 years) group. This association was mainly driven by the increased glucose levels and higher waist circumference associated with >2 h social jetlag. In contrast, in the older (≥61 years) group we observed no significant associations between social jetlag status, parameters of metabolic syndrome, and diabetes/prediabetes.
We showed an association between social jetlag and a higher prevalence of diabetes/prediabetes. These results are congruent with earlier studies that showed similar directionality of the association between social jetlag and diabetes markers (Parsons et al., 2015; Wong et al., 2015). However, in contrast to Wong et al. (2015), we did not observe a positive association between social jetlag and triglycerides. This contradiction to the only other study that investigated the association between social jetlag and metabolic syndrome parameters might be caused by differences in study population. Wong et al. included a younger group of more homogenous participants: 500 healthy, working participants, aged 30 to 54 years, while our population was population-based and thus more heterogeneous.
Second, our analysis showed effect modification for age. This modification is due to the lower prevalence of social jetlag in our older (≥61 years) participants: only 24% and 19% of them had 1-2 h or >2 h social jetlag, respectively, while the prevalence of the metabolic syndrome and diabetes/prediabetes increases in this group. The lower prevalence of social jetlag in the older participants confirms the results from Roenneberg et al. (2012), who showed a decrease in social jetlag with age. In the younger (<61 years) participants, we observed similar effect sizes and direction of the association between social jetlag and metabolic syndrome, when compared with literature studying even younger (<50 years) populations (Parsons et al., 2015; Wong et al., 2015). However, no such association was observed in our older (≥61 years) participants. This lack of association might be due to social jetlag being driven by work obligations, which decline after age 61 years due to retirement. The absence of social clocks might decrease social jetlag in these older participants, creating a group of people who had social jetlag for more than 40 years and other people who never had social jetlag, resulting in a mixed group in which the associations cannot be determined.
Overall, the findings from the current study suggest the association between social jetlag and metabolic syndrome to be driven by higher glucose and waist circumference. This is congruent with our hypotheses on how social jetlag can lead to adverse health outcomes. First, circadian misalignment is suggested to disrupt the endogenous circadian rhythm controlled by the suprachiasmatic nucleus (SCN) (la Fleur et al., 2001), which controls the hypothalamus pituitary adrenal (HPA) axis and thus alters its regulation (la Fleur et al., 2001). In concordance with the Bjorntorp hypothesis, disturbance of the HPA axis is associated with the development of visceral obesity and the metabolic syndrome (Bjorntorp and Rosmond, 2000; Gonnissen et al., 2012). Second, the SCN also controls the peripheral clocks, which in turn control glucose homeostasis. By disturbing the SCN and peripheral clocks, circadian misalignment also alters glucose homeostasis (Van Cauter et al., 1991; la Fleur et al., 2001). Overall, these data suggested that the effect of social jetlag is driven by visceral obesity and glucose homeostasis rather than cardiovascular parameters.
In the literature, several pathways have been proposed that would mediate the association between social jetlag and metabolic syndrome. For example, changes in circadian rhythm deregulate appetite and physical activity (Spiegel et al., 1999). Aside from altering health behaviors, changes in the circadian rhythm could also affect metabolic syndrome via several physiologic pathways, such as hyperactivation of the HPA axis and autonomic nervous system (Lucassen et al., 2012). We do not have data to study this; however, proxy measurements of these systems (i.e., blood pressure) were not associated with social jetlag. A recent experimental study by Rao et al. (2015) showed a role for nonesterified fatty acid (NEFA) levels in the decrease of whole body insulin sensitivity after sleep alterations. In general, it is believed that all these different pathways add up the metabolic syndrome. However, future studies need to include measures of these potential pathways.
Our study has some limitations that warrant discussion. First, our data were cross-sectional, which makes it difficult to determine cause, effect, and the possible mediating effect of risk factors. Second, we used data from a follow-up measurement that was not completely representative of the baseline cohort: Participants at the earlier visit were slightly older and had a higher prevalence of the metabolic syndrome and a lower prevalence of type 2 diabetes. Third, we used an adapted version of the Munich ChronoType Questionnaire from 2003 to assess social jetlag (Roenneberg et al., 2003; Wittmann et al., 2006). This version questions sleep and wake times on weekdays and weekend/free days and does not question use of an alarm clock on weekend/free days, which might lead to an underestimation of social jetlag, while our current estimates of sleep and wake times on weekend/free days would be earlier compared with days that entail unrestricted sleep and wake times. Fourth, we did not measure information on dietary intake, which could have affected or mediated the association between social jetlag and metabolic outcomes. Fifth, we did not measure HbA1c levels in the whole cohort, but only in those with an increased risk of having developed diabetes (55%), which might result in an underestimation of the prevalence of diabetes/prediabetes. Strengths of our study are that we diagnosed metabolic syndrome and diabetes/prediabetes, instead of using self-reported measures, and that we collected the data in a large, middle-aged, population-based cohort with comprehensive information on relevant confounders.
From our current study, we conclude that in our population-based cohort, social jetlag was associated with a 2-fold increased risk of the metabolic syndrome and diabetes mellitus or prediabetes, especially in younger (<61 years) participants. These results confirm that even small changes of circadian misalignment are associated with adverse health outcomes, such as metabolic syndrome and even diabetes/prediabetes. Possible interventions targeting social jetlag, including simple behavioral modifications such as keeping a regular sleep-wake schedule, could therefore have beneficial effects on the development of the metabolic syndrome and type 2 diabetes mellitus, especially in younger (<61 years) people. However, more in vivo research in prospective cohorts is necessary to confirm these findings.
Acknowledgments
We appreciate the cooperation of the participants and research assistants who were involved in the study. We thank Tootje Hoovers and Jolanda Bosman for the organization of the study. A.D.M.K, S.P.R., J.W.B., and F.R. researched data and wrote the manuscript. E.v.t.R., L.G., A.A.v.d.H., and P.J.E. reviewed and edited the manuscript. J.W.B., J.M.D., G.N., and F.R. conceived and designed the study and reviewed and edited the manuscript. A.D.M.K. and F.R. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Footnotes
Conflict of Interest Statement: The author(s) have no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
References
- Bjorntorp P, Rosmond R. (2000) Obesity and cortisol. Nutrition 16:924-936. [DOI] [PubMed] [Google Scholar]
- Colagiuri S. (2011) Glycated haemoglobin (HbA1c) for the diagnosis of diabetes mellitus—practical implications. Diabetes Res Clin Pract 93:312-313. [DOI] [PubMed] [Google Scholar]
- Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (2001) Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 285:2486-2497. [DOI] [PubMed] [Google Scholar]
- Gan Y, Yang C, Tong X, Sun H, Cong Y, Yin X, Li L, Cao S, Dong X, Gong Y, et al. (2015) Shift work and diabetes mellitus: a meta-analysis of observational studies. Occup Environ Med 72:72-78. [DOI] [PubMed] [Google Scholar]
- Gonnissen HK, Mazuy C, Rutters F, Martens EA, Adam TC, Westerterp-Plantenga MS. (2013) Sleep architecture when sleeping at an unusual circadian time and associations with insulin sensitivity. PLoS One 8:e72877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gonnissen HK, Rutters F, Mazuy C, Martens EA, Adam TC, Westerterp-Plantenga MS. (2012) Effect of a phase advance and phase delay of the 24-h cycle on energy metabolism, appetite, and related hormones. Am J Clin Nutr 96:689-697. [DOI] [PubMed] [Google Scholar]
- International Diabetes Federation (IDF) (2015) IDF Diabetes Atlas, 7th ed. IDF website. https://www.idf.org/
- Kantermann T, Duboutay F, Haubruge D, Kerkhofs M, Schmidt-Trucksass A, Skene DJ. (2013) Atherosclerotic risk and social jetlag in rotating shift-workers: first evidence from a pilot study. Work 46:273-282. [DOI] [PubMed] [Google Scholar]
- Knol MJ, Le Cessie S, Algra A, Vandenbroucke JP, Groenwold RH. (2012) Overestimation of risk ratios by odds ratios in trials and cohort studies: alternatives to logistic regression. CMAJ 184:895-899. [DOI] [PMC free article] [PubMed] [Google Scholar]
- la Fleur SE, Kalsbeek A, Wortel J, Fekkes ML, Buijs RM. (2001) A daily rhythm in glucose tolerance: a role for the suprachiasmatic nucleus. Diabetes 50:1237-1243. [DOI] [PubMed] [Google Scholar]
- Lucassen EA, Rother KI, Cizza G. (2012) Interacting epidemics? Sleep curtailment, insulin resistance, and obesity. Ann N Y Acad Sci 1264:110-134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Parsons MJ, Moffitt TE, Gregory AM, Goldman-Mellor S, Nolan PM, Poulton R, Caspi A. (2015) Social jetlag, obesity and metabolic disorder: investigation in a cohort study. Int J Obes (Lond) 39:842-848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Qian J, Scheer FA. (2016) Circadian system and glucose metabolism: implications for physiology and disease. Trends Endocrinol Metab 27:282-293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rao MN, Neylan TC, Grunfeld C, Mulligan K, Schambelan M, Schwarz JM. (2015) Subchronic sleep restriction causes tissue-specific insulin resistance. J Clin Endocrinol Metab 100:1664-1671. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rauh SP, Heymans MW, Koopman AD, Nijpels G, Stehouwer CD, Thorand B, Rathmann W, Meisinger C, Peters A, de Las Heras Gala T, et al. (2017) Predicting glycated hemoglobin levels in the non-diabetic general population: development and validation of the DIRECT-DETECT prediction model—a DIRECT study. PLoS One 12:e0171816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reutrakul S, Hood M, Crowley S, Morgan M, Teodori M, Knutson K, van Cauter E. (2013) Chronotype is independently associated with glycemic control in type 2 diabetes. Diabetes Care 36:2523-2529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Roden M, Koller M, Pirich K, Vierhapper H, Waldhauser F. (1993) The circadian melatonin and cortisol secretion pattern in permanent night shift workers. Am J Physiol 265:R261-R267. [DOI] [PubMed] [Google Scholar]
- Roenneberg T, Allebrandt KV, Merrow M, Vetter C. (2012) Social jetlag and obesity. Curr Biol 22:939-943. [DOI] [PubMed] [Google Scholar]
- Roenneberg T, Merrow M. (2007) Entrainment of the human circadian clock. Cold Spring Harb Symp Quant Biol 72:293-299. [DOI] [PubMed] [Google Scholar]
- Roenneberg T, Wirz-Justice A, Merrow M. (2003) Life between clocks: daily temporal patterns of human chronotypes. J Biol Rhythms 18:80-90. [DOI] [PubMed] [Google Scholar]
- Rutters F, Lemmens SG, Adam TC, Bremmer MA, Elders PJ, Nijpels G, Dekker JM. (2014) Is social jetlag associated with an adverse endocrine, behavioral, and cardiovascular risk profile? J Biol Rhythms 29:377-383. [DOI] [PubMed] [Google Scholar]
- Scheer FA, Hilton MF, Mantzoros CS, Shea SA. (2009) Adverse metabolic and cardiovascular consequences of circadian misalignment. Proc Natl Acad Sci U S A 106:4453-4458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Seidell JC, Oosterlee A, Deurenberg P, Hautvast JG, Ruijs JH. (1988) Abdominal fat depots measured with computed tomography: effects of degree of obesity, sex, and age. Eur J Clin Nutr 42:805-815. [PubMed] [Google Scholar]
- Spiegel K, Leproult R, Van Cauter E. (1999) Impact of sleep debt on metabolic and endocrine function. Lancet 354:1435-1439. [DOI] [PubMed] [Google Scholar]
- Twisk J. (2010) Introduction into Applied Biostatistics. Amsterdam: Elsevier Gezondheidszorg. [Google Scholar]
- Van Cauter E, Blackman JD, Roland D, Spire JP, Refetoff S, Polonsky KS. (1991) Modulation of glucose regulation and insulin secretion by circadian rhythmicity and sleep. J Clin Invest 88:934-942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- van ‘t Riet E, Alssema M, Rijkelijkhuizen JM, Kostense PJ, Nijpels G, Dekker JM. (2010) Relationship between A1C and glucose levels in the general Dutch population: the New Hoorn Study. Diabetes Care 33:61-66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wittmann M, Dinich J, Merrow M, Roenneberg T. (2006) Social jetlag: misalignment of biological and social time. Chronobiol Int 23:497-509. [DOI] [PubMed] [Google Scholar]
- Wong PM, Hasler BP, Kamarck TW, Muldoon MF, Manuck SB. (2015) Social jetlag, chronotype, and cardiometabolic risk. J Clin Endocrinol Metab 100:4612-4620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zou G. (2004) A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol 159:702-706. [DOI] [PubMed] [Google Scholar]