The authors have declared that no competing interests exist.
Conceived and designed the experiments: KIR G. Csako G. Cizza. Performed the experiments: EAL MSM ABC G. Cizza. Analyzed the data: EAL XZ KIR MSM ABC LdJ G. Csako G. Cizza. Wrote the paper: EAL KIR MSM LdJ G. Csako G. Cizza.
Short sleep duration and decreased sleep quality are emerging risk factors for obesity and its associated morbidities. Chronotype, an attribute that reflects individual preferences in the timing of sleep and other behaviors, is a continuum from morningness to eveningness. The importance of chronotype in relation to obesity is mostly unknown. Evening types tend to have unhealthy eating habits and suffer from psychological problems more frequently than Morning types, thus we hypothesized that eveningness may affect health parameters in a cohort of obese individuals reporting sleeping less than 6.5 hours per night.
Baseline data from obese (BMI: 38.5±6.4 kg/m2) and short sleeping (5.8±0.8 h/night by actigraphy) participants (
Eveningness was associated with eating later and a tendency towards fewer and larger meals and lower HDL-cholesterol levels. In addition, Evening types had more sleep apnea and higher stress hormones. Thus, eveningness in obese, chronically sleep-deprived individuals compounds the cardiovascular risk associated with obesity.
The prevalence of obesity, 36% among adults in the US, has reached epidemic proportions
Individual preference for sleep timing and timing of other behaviors is a stable trait, referred to as one’s chronotype. A large twin study demonstrated that approximately 50% of the chronotype features are heritable
In “modern” regimented societies sleep timing, especially during working days, is influenced by social norms, a phenomenon known as “social jet lag” and we are often forced to be awake against our preferred times. Evening types tend to get less and more shallow, non-refreshing sleep during working days
Circadian rhythms are oscillations with a period of approximately 24 h that are generated in the suprachiasmatic nucleus of the hypothalamus. These endogenous rhythms orchestrate most physiological functions and are synchronized to the environment mainly by sunlight, physical activity, feeding and sleep
In this study, we examined the relationship of chronotype with sleep, food intake, endocrine and metabolic parameters in a cohort of obese subjects who reported sleeping less than 6.5 h per night.
This analysis is part of the Sleep Extension Study, a prospective, randomized, controlled study of obese (BMI 30–55 kg/m2) men and premenopausal women, 18 to 50 years old, reporting sleeping less than 6.5 h per night. Details of this study have been previously reported
A 24 h urine collection was started during the first afternoon of the Overnight Randomization Visit. At 8∶00 the following day, after an overnight fast, blood samples, vital signs, and anthropometric measures were obtained. Participants completed the Pittsburgh Sleep Quality Index, Epworth Sleepiness Scale, and the Horne and Ostberg questionnaire
The study was conducted at the NIH Clinical Center in Bethesda, MD, USA, after approval by the Institutional Review Board of the NIDDK, and is listed in ClinicalTrials.gov (identifier: NCT00261898). Each participant signed a written informed consent form.
The Horne and Ostberg questionnaire inquires about preferred sleep time and daily performance (score range: 16–86)
Sleep duration was assessed by a wrist actigraphy monitors (Actiwatch-64, Mini Mitter/Respironics/Philips, Bend, OR) that participants (
The Pittsburgh Sleep Quality Index and the Epworth Sleepiness Scale are questionnaires that assess sleep quality and daytime sleepiness (score ranges: 0–21 and 0–24 with abnormal thresholds of 5 and 10, respectively)
A portable screening device (Apnea Risk Evaluation System, Advanced Brain Monitoring Inc, Carlsbad, CA) provided an estimate of the respiratory disturbance index (RDI): the number of apneas and hypopneas per hour of sleep
Height was measured to the nearest centimeter using a wall-mounted stadiometer (SECA 242, SECA North America East, Hanover, MD) and weight was measured using a stand-on-scale in a hospital gown to the nearest 1/10th of a kg (SR555 SR Scales, SR Instruments, INC, Tonawanda, NY). Neck and waist circumference measurements were done using a non-stretch measuring tape in triplicate to the nearest mm. Neck circumference was measured at the minimal circumference with the head in the Frankfort Horizontal Plane and waist circumference was measured at the uppermost border of the iliac crest.
Glucose, total cholesterol, and triglycerides were determined with enzymatic methods high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) with direct homogeneous methods in fasting serum specimens on an automated analyzer (Dimension Vista 1500, Siemens Health Diagnostics, Deerfield, IL). Plasma ACTH, serum cortisol, and insulin levels were assessed with chemiluminescence immunoassays (Immulite 2500, Siemens). Urinary free cortisol and catecholamines were measured in 24 h collections with liquid chromatography-tandem mass spectrometry, and high-performance liquid chromatography, respectively.
Participants recorded food intake for three consecutive days, preferably two weekdays and one weekend day. At the Screening Visit, they received written and verbal instructions on recording food intake and were instructed to begin record everything they ate or drank beginning when they woke up on the first day until they woke up then next morning. At the Randomization Visit, dietitians and health technicians reviewed the food records with the participants utilizing three dimensional food models. Information on timing and location of meals was also obtained. The data were analyzed using the Nutrition Data System for Research software (versions 2007–2010, Nutrition Coordinating Center (NCC), University of Minnesota, Minneapolis, MN)
Normality of variables was examined by Q-Q normality plots. Mean and standard deviation (SD) were calculated for variables with a normal distribution; median and interquartile range were used for variables with a skewed distribution (RDI, plasma ACTH, serum cortisol, HDL-C, LDL-C, triglycerides, insulin, 24 h-urinary norepinephrine, 24 h-urinary epinephrine, urinary free cortisol, and food intake after 20∶00). Logarithmic transformation was applied for skewed variables for further analysis. Morning types were compared to Evening types by independent
Relationships between chronotype score and outcome variables were assessed by linear regression, corrected for gender, age, and anthropometric measures, as appropriates. To assess possible predictors of chronotype score, multivariate forward stepwise analyses (
Approximately two-thirds of participants were Morning types and they were about three years older than Evening types (
Morning chronotype(score 50–86) |
Evening chronotype(score 16–49) |
||
Chronotype score (range 16–86) | 59.1±6.8 | 42.7±5.0 |
|
Age (years) | 41.7±5.9 | 38.6±7.8 |
|
Female | 76% | 80% | 0.692 |
African-American/White/Other races ( |
54/38/9 | 72/28/0 | 0.054 |
BMI (kg/m2) | 38.2±6.3 | 39.1±6.6 | 0.470 |
Waist circumference (cm) | 113.0±13.6 | 114.7±11.5 | 0.510 |
Neck circumference (cm) | 38.8±3.8 | 39.6±3.8 | 0.340 |
Morning resting heart rate (beats/min) | 68.4±10.1 | 74.0±10.1 |
|
Values are means ± SD,
median and interquartile range for skewed variables, or %. Skewed variables were log-transformed prior to analysis.
Fisher exact test including all races;
Fisher exact test excluding “Other races”.
As expected for short sleepers, 29% of the participants experienced excessive daytime sleepiness and 84% experienced abnormal sleep quality, as assessed by the Epworth Sleepiness Scale and the Pittsburgh Sleep Quality Index, respectively (
Morning chronotype(score 50–86) |
Evening chronotype(score 16–49) |
||
Pittsburgh Sleep Quality Index score | 8.0±2.7 | 8.6±2.9 | 0.263 |
Epworth Sleepiness Scale score | 8.4±4.5 | 7.6±4.5 | 0.387 |
|
|||
Working days | 361±56 |
403±51 |
|
Non-working days | 420±63 | 421±39 | 0.947 |
|
|||
Working days | 339±52 |
346±64 | 0.576 |
Non-working days | 386±61 |
379±60 | 0.625 |
|
|||
Working days | 81.5±5.7 | 81.3±8.2 | 0.914 |
Non-working days | 81.1±7.5 | 81.0±7.8 | 0.964 |
|
|||
RDI (events/h) |
5 (2–12) | 10 (7–16) |
|
RDI >5 (events/h) | 47% | 81% |
|
Values are means ± SD,
median and interquartile range for skewed variables, or %. Skewed variables were log-transformed prior to analysis.
Fisher exact test including all races;
Fisher exact test excluding “Other races”.
Sleep duration differed on working
By self-reported diaries, Morning types reported sleeping significantly less than Evening types on working days, a difference that disappeared during non-working days (
Horizontal bars represent sleep time as measured by actigraphy. Bedtime (mean±SD) and risetime (mean±SD) on all days combined and separately on non-working days and on work days in Morning (grey bars) and Evening types (black bars). The mid-sleep times (MSTs) (mean±SD) are depicted by open diamonds and compared between chronotypes by
When examining the three day food recall diary, no differences between chronotypes were observed in total energy intake, portion size, or number of eating occasions, albeit during working days Morning types tended to have smaller portions and to eat more frequently than Evening types (
Plot of chronotype scores versus the number of eating occasions (A), the portion size (B), and the amount of caloric intake after 20∶00 (C) with a trend line, respectively. N = 113 for all. R2, slope, and
Working day | Non-working day | |||||
Morning type | Evening type | Morning type | Evening type | |||
Total food intake (kcal) | 2129±631 | 2276±815 | 0.373 | 2383±928 | 2378±883 | 0.922 |
Portion size (kcal) | 461±177 | 545±219 | 0.065 | 599±273 | 622±380 | 0.751 |
Eating occasions per day ( |
4.9±1.5 | 4.4±1.5 | 0.175 | 4.2±1.2 | 4.3±1.6 | 0.568 |
First eating occasion (h:min) | 7∶17±1∶31 | 8∶38±1∶52 |
|
8∶56±2∶30 | 9∶59±2∶32 | 0.075 |
Food intake after 20∶00 (kcal) | 299±354 | 677±460 |
|
327±354 | 537±480 |
|
Food intake after 20∶00 (% of total caloric intake) | 14±15 | 30±18 |
|
14±15 | 24±20 |
|
Values are means ± SD. Intake parameters of chronotypes were compared with
Moving from morningness towards eveningness scores was associated with an increase in BMI, larger neck circumference, and lower HDL-C levels (
Plot of Chronotype scores versus BMI (A), neck circumference (B), and HDL-C (C) with a trend line, respectively. N = 119 for all. R2 slope, and
Morning types had higher plasma ACTH and 24 h-urinary epinephrine concentrations and tended to have higher levels of 24 h urinary norepinephrine than Evening types (
Morning chronotype(score 50–86) |
Evening chronotype(score 16–49) |
||
|
|||
Morning plasma ACTH (pg/ml)# | 17 (12–24) | 21 (16–32) |
|
Morning serum cortisol (µg/dL)# | 9 (6–12) | 10 (6–14) | 0.530 |
Urinary free cortisol (µg/24 h)# | 17 (12–24) | 19 (10–28) | 0.885 |
Urinary norepinephrine (µg/24 h)# | 39 (28–56) | 45 (37–61) | 0.052 |
Urinary epinephrine (µg/24 h)# | 3 (2–5) | 4 (3–7) |
|
|
|||
Insulin (mU/l)#, |
9 (7–14) | 10 (6–16) | 0.674 |
Glucose (mg/dL) |
88.6±10.5 | 90.1±9.6 | 0.465 |
Total cholesterol (mg/dL) |
178.9±36.7 | 177.1±35.5 | 0.806 |
Triglycerides (mg/dL)#, |
91 (62–133) | 75 (55–106) | 0.214 |
HDL-C (mg/dL)#, |
48 (42–58) | 49 (41–52) | 0.513 |
LDL-C (mg/dL)#, |
106 (91–125) | 116 (89–134) | 0.532 |
Values are means ± SD or #median and interquartile range for skewed variables.
Measured in fasting morning plasma samples.
Progression towards eveningness was associated with higher plasma ACTH levels, higher 24 h-urinary epinephrine and norepinephrine levels, and a higher heart rate (
Chronotype scores on the horizontal axis are regressed against plasma ACTH (
RDI was directly related to plasma ACTH (regression coefficient = 0.132;
A gender- and age-corrected multivariate forward stepwise model was fitted with predictors that related to chronotype score at
Step 0 model:Gender+Age | Step 1 model:Gender+Age+Urinary norepinephrine | Step 2 model:Gender+Age+Urinary norepinephrine+RDI | |
|
R2 = 0.019, |
R2 = 0.259, |
R2 = 0.339, |
Gender | β = −0.038, |
β = 0.021, |
β = 0.072, |
Age | β = 0.137, |
β = 0.273, |
β = 0.302, |
Urinary norepinephrine (µg/24 h) | N.A. | β = −0.531, |
β = −0.512, |
RDI (events/hour) | N.A. | N.A. | β = −0.290, |
This model depicts a gender- and age-corrected model, generated by multivariate forward stepwise analysis. R2 and
At Step 1, urinary 24 h norepinephrine entered the model and, together with age and gender, accounted for approximately 26% of the variability in chronotype score. At Step 2, the addition of RDI increased the variability in chronotype score accounted for by the model to 34%. Thus, the resulting final model retained two significant predictors, 24 h urinary norepinephrine and RDI, which were both strongly related in an inverse fashion to chronotype score.
In this cohort of sleep-deprived obese subjects, eveningness was associated with an unhealthy eating pattern, characterized by eating later on both working and non-working days and a trend towards a decreased number of eating occasions with larger portion sizes. Evening types were also more likely to have sleep apnea than Morning types, had higher levels of stress hormones, and had a higher resting heart rate. In addition, moving from morningness toward eveningness scores was associated with an increase in BMI and a decrease in HDL-C.
Eating less frequently has been reported to be associated with higher BMI and increased risk of weight gain
In our sample, eveningness was associated with lower HDL-C levels, and a trend towards higher BMI and neck circumference. In a study of lean participants, Evening types tended to have increased BMI
In our study, in which the actual sleeping schedule was left to the individual subject, we found no differences in sleep quality and sleepiness scores between chronotypes. This is in accord with a well-conducted polysomnography study of 12 Morning types and 12 Evening types, tested in the sleep laboratory according to their habitual sleep schedule in which there were no differences in subjective sleep quality, as well as in total sleep time by EEG
In a large population-based, Finnish study, the FINRISK 2007 survey, sleep duration did not differ between chronotypes
In our study sleep duration by actigraphy did not differ between chronotypes either, but Evening types reported sleeping approximately 40 min longer during working days. Further, self-reported sleep duration in both chronotypes was longer than sleep duration by actigraphy and this difference was present on both working and non-working days. We cannot account for the discrepancy between sleep duration by diary and actigraphy. It may take longer for Evening types to fall asleep because they feel less sleepy and may display more active behavior before going to bed. As an internal validation of self-reported measures, participants reported sleeping longer during non-working than during working days. Of note, most of the epidemiological studies on the relationship between sleep and weight relied on self-reported measures
The longer sleep latency period observed in our study for the Evening type may also be due to another related contributory factor, the quality of light. During the waking hours Evening types are less exposed to light, especially in the 100 to 500 lux range, which is the average intensity of interior lights
In our sample, mean sleep efficiency was similarly suboptimal (81%,) in Morning and Evening types. A German actigraphy study conducted in a large sample of young students reported a somewhat better sleep efficiency in Morning types (88%) than Evening types (84%). It is possible that we have observed a “floor” effect in our sample that may have zeroed and obscured chronotype differences in sleep efficiency.
The intriguing possibility that assortative mating, defined as the nonrandom mating of individuals with respect to phenotype and cultural factors, may play a role in chronotype. This has been proposed in a report of a positive relationship in a German sample of 84 couples
Compared to Morning types, Evening types had 20–30% higher 24 h urinary levels of epinephrine and higher morning plasma ACTH levels, indicating an activation of the sympatho-adrenal system. As further evidence, the 24 h urinary norepinephrine was also borderline significantly elevated. In agreement with a previous report
The mean chronotype score, 53.7 seemed to be slightly skewed towards eveningness; for example, a study of 2,526 New-Zealand adults of averagely 40 years of age reported a mean chronotype score of 58.1
As indicated by twin studies, there is a clear genetic component to morningness-eveningness, which accounts for more than 50% of its total variance
Some study limitations should be noted. The cross-sectional nature of this report did not allow establishment of causality. Hormone determinations at a single time-point may have missed endocrine differences between chronotypes in circadian rhythms. Endocrine rhythms shift by two to three hours depending on chronotype
In summary, in our sample of obese, sleep-deprived subjects, eveningness was associated with a trend towards higher stress hormones, higher resting heart rate, more sleep apnea, lower HDL-C levels, and a trend towards higher BMI. High BMI and low HDL-C levels predict cardiovascular morbidity
We would like to thank the following colleagues (alphabetical order) for their scientific advice and critical suggestions in the development and conduct of the study protocol: Karim Calis, Janet Gershengorn, Gregor Hasler, Emmanuel Mignot, Susan Redline, Terry Phillips, Nancy Sebring, Duncan Wallace, Robert Wesley, Elizabeth Wright. We would also like to thank present and past members of the study team: Peter Bailey, Laide Bello, Meredith Coyle, Paula Marincola, Patrick Michaels, Svetlana Primma, Angela Ramer, Rebecca Romero, Megan Sabo, Tanner Slayden, Sara Torvik, Sam Zuber, Elizabeth Widen, and Lyda Williams. The bioinformatics support of Frank Pierce (Esprit Health) is gratefully acknowledged. Finally we are grateful to all of our enthusiastic study participants.