2005 The Association Between Obesity and Short Sleep Duration A Population Based Study

Prévia do material em texto

Journal of Clinical Sleep Medicine, Vol. 1, No. 4, 2005
Over the past decades, the rise in obesity in our society has been paralleled by a reduction in sleep time. The possible 
direct links between these 2 escalating health problems of sleep 
loss and obesity have only recently been investigated. The 
National Health and Nutritional Examination Survey found that, 
in 1960, an estimated 13.4% of adult men and women were obese 
(body mass index [BMI] ≥ 30), whereas, by 2000, the proportion 
of obese adults had increased to 30.5 %,1 contributing to more 
than 300,000 premature deaths2 and more than $90 billion in 
direct healthcare costs each year in the United States alone.3 The 
well-known morbidity of obesity includes elevations in systolic 
and diastolic blood pressure, serum total cholesterol, high-
density lipoprotein-c, low-density lipoprotein-c, triglycerides, 
apolipoprotein A-I, apolipoprotein B, and fasting blood sugar 
levels.4 Obesity is also associated with increases in inflammatory 
markers5 and is strongly linked to increased risks of coronary heart 
disease and type 2 diabetes.6 The high morbidity associated with 
obesity emphasizes the importance of identifying the multiple 
and diverse factors that may contribute to its increasingly high 
prevalence in the population. 
 In addition to the rise in obesity, chronic voluntary sleep 
curtailment below 8 hours per night has also become commonplace, 
with the average numbers of hours of sleep during the weekdays 
being 6.9 ± 1.3 for young adults.7 Normal (average) sleep duration 
has fallen from 8.5 hours per night in 19598,9 to 7.3 hours per night 
in 2002.10 In terms of morbidity associated with shorter sleep time, 
reduction of sleep to 6 hours or less per night has been shown to 
result in decreased alertness and cognitive performance even after 
a single night.11 Sustained bouts of sleep loss have been shown to 
have cumulative effects.12 Sleep loss also contributes to a variety 
of metabolic, hormone, and immunologic changes. For example, 
several nights of sleep restriction can impair glucose tolerance and 
decrease insulin sensitivity13 and is also associated with changes in 
thyroid activity and in the activity of the hypothalamic-pituitary-
adrenal axis.14 Elevated sympathovagal balance, abnormal leptin 
and growth hormone secretion, and related increases in appetite 
associated with sleep loss have also been reported.13 These 
metabolic findings have led some investigators to hypothesize a 
direct link between sleep loss and obesity.15-19
 Recently, there has been accumulating evidence suggesting a 
relationship between obesity and reduced sleep duration. Prior 
studies have reported increased BMI as a correlate of short sleep 
duration. Kripke and coworkers15 reported a negative associa-
tion between sleep duration and BMI in men and a U-shaped re-
lationship between sleep duration and BMI in women. Hassler 
The Association Between Obesity and Short Sleep Duration: A Population-Based 
Study
Meeta Singh, M.D.1; Christopher L. Drake, Ph.D.1,2; Timothy Roehrs, Ph.D.1,2; David W. Hudgel, M.D.1; and Thomas Roth, Ph.D.1,2
1Henry Ford Hospital Sleep Disorders and Research Center and 2Department of Psychiatry and Behavioral Neurosciences, Wayne State College of 
Medicine, Detroit, MI
Disclosure Statement
Drs. Singh, Drake, Roehrs, Hudgel, and Roth have indicated no financial 
conflicts of interest.
Submitted for publication March 21, 2005
Accepted for publication June 2, 2005
Address correspondence to: Meeta Singh, M.D., 2799 W. Grand Blvd., CFP 
3, Sleep Disorders Center, Detroit, MI 48202; Tel: (313) 401-6421; Fax: (313) 
916-5167; Email: msingh2@hfhs.org 
SCIENTIFIC INVESTIGATIONS
357
Study Objectives: To assess the association between nightly total sleep 
time (TST) and obesity in an epidemiologic sample of metropolitan De-
troit, Michigan. 
Methods: Data were collected through telephone interviews completed 
using a population-based sample of 3158. The self-reported average 
nightly TST during the 2 weeks before the interview was used to divide 
the sample into 6 groups (hours per night of sleep; ≤ 5, > 5 ≤ 6, > 6 ≤ 7, 
> 7 ≤ 8, > 8 ≤ 9, > 9). Obesity was defined as a body mass index ≥ 30. 
Health and demographic variables were also assessed. 
Results: The overall prevalence of obesity was 24.8% and was signifi-
cantly higher in individuals with lower amounts of TST. Compared with 
those with 7 to 8 hours of TST, individuals obtaining 5 hours or less and 
more than 5 but 6 hours or less of TST had significantly increased odds 
of being obese, after controlling for age, sex, loud snoring, hypertension, 
diabetes, arthritis, and alcohol intake (odds ratio = 1.7, 95% confidence 
interval = 1.3-2.3 and odds ratio = 1.4, 95% confidence interval = 1.1-
1.8, respectively). A low TST was a significant predictor of a high body 
mass index. Furthermore, a low TST was also a significant predictor 
of diabetes, prior to controlling for body mass index. Interestingly, the 
prevalence of reduced habitual TST (≤ 5 hours) was higher in African 
Americans, in comparison with Caucasians (18.7% vs 7.4%; p < .001).
Conclusions: Our population-based data suggest that short sleep dura-
tion is associated with elevated prevalence of obesity and adds to the 
growing body of evidence supporting this relationship.
Keywords: Obesity, short sleep duration, diabetes, self reported sleep 
times
Citation: Singh M; Drake CL; Roehrs T et al. The Association between 
obesity and short sleep duration: a population-based study. J Clin Sleep 
Med 2005;1(4):357-363.
Journal of Clinical Sleep Medicine, Vol. 1, No. 4, 2005 358
M Singh, CL Drake, T Roehrs et al
and colleagues, 16 showed a similar association in young adults. 
Other data suggest that the risk for obesity is higher for each hour 
of reduced sleep time below 8 hours.17 Even more disconcerting 
are data illustrating that similar associations between short sleep 
duration and obesity are present in children.18,19 Several of these 
studies investigating the relationship between sleep duration and 
obesity have certain limitations. Some were done using nonrepre-
sentative samples.16-19 In other cases, no adjustments were made 
for the presence of comorbid diseases such as hypertension, dia-
betes, arthritis, or sleep apnea, making the results difficult to in-
terpret given the known association between these comorbid dis-
eases and both sleep and obesity.15-18
 In order to remedy some of the limitations of previous studies, 
we assessed the relationship between daily total sleep time (TST) 
and obesity in a representative sample. Data were drawn from 
the population of the tricounty metropolitan Detroit area, which 
includes Oakland, Wayne, and Macomb Counties. Data were also 
controlled for several relevant comorbid conditions. An additional 
factor assessed in the present study, which has not been explored 
previously, is the relationship between obesity and TST in dif-
ferent racial groups. While there are inherent limitations when 
performing epidemiologic investigations, improved controls, rep-
lication, and the convergence of multiple lines of evidence will 
help to improve our understanding of the link between sleep and 
obesity. Given the evidence for a plausible mechanism relating 
reduced sleep time and obesity,20 we hypothesized that individu-
als who obtain less TST would have an increased prevalence of 
obesity, even after adjusting for important health-related and de-
mographic variables.
METHODS
Participants
 Individuals participating in this study were assessed in 
conjunction with a larger ongoing epidemiologic study investigating 
the prevalence of daytime sleepiness. Participants were drawn from 
the general population of tricounty Detroit area using random digit 
dialing techniques. For eligibility, the calling address had to bea 
residence and the participant an adult between the ages of 18 and 
65 years.
 A random probability selection procedure was used to determine 
the sex of the target adult. If 2 or 3 adults within the target sex were 
present in a household, a random probability selection procedure 
(oldest/second, oldest/youngest) was used to determine the target 
respondent. If 4 or more adults with the target sex were present 
in the household, last birthday method was used to determine the 
target respondent. In order to maintain an unbiased sample, only 
individuals who could not answer the questionnaire due to sensory 
or mental impairment were excluded from the sample. From 4682 
eligible participants, 3283 interviews were obtained (response 
rate, calculated by the number of interviews that were conducted 
relative to the number of eligible participants, was 70.1%). After 
discarding individuals who were missing data on 1 of the variables 
of interest, the final sample (N = 3158) included 1589 men 
(50.3%) and 1569 women (49.7%), with a mean age of 41.6 ± 12.6 
years. The demographic details of the sample, including race and 
socioeconomic status, are shown in Table 1a and 1b and are nearly 
identical to the 2000 census data for the area.21 The institutional 
review board approved all procedures, and informed consent was 
obtained from all participants. Individuals were paid $25.00 for 
study participation.
Procedures
 Participants completed a 20-minute telephone interview, which 
included questions related to sleep and health habits, along with 
general information regarding medical and psychiatric status. 
Table 1A—Sociodemographic Characteristics of the Study Sample 
and Comparative Data From the United States Census*
Sociodemographic 
Characteristic
Sample 
(N, 3158)
Tricounty
(N, 4,043,467)
United States 
Census
(N, 281,421,906)
Sex, %
Men 50.3 48.5 49.1
Women 49.7 51.5 50.9
Socioeconomic 
status, in thousands†
< $10 6.4 9.5 9.5
$10 < $15 4.4 6.3 6.3
$15 < $25 11.1 12.8 12.8
$25 < $35 11.3 12.8 12.8
$35 < $50 14.5 16.5 16.5
$50 < $75 22.5 19.5 19.5
> $75 29.8 22.4 22.4
Race/Ethnicity
Caucasian 68.9 75.1 75.1
African American 24.9 12.3 12.3
Asian/Pacific 
Islander
1.9 3.7 3.7
Native American 0.9 0.9 0.9
Other/refused 3.4 7.9 7.9
Age, y
18-24 10.9 21.0 21.0
25-34 21.4 21.4 21.4
35-44 24.9 24.3 24.3
45-54 24.9 20.2 20.2
55-64 17.8 13.0 13.0
*Data are presented as percentages. Data for the Tricounty and United 
States are from the 2000 census. †Fewer than 5% of individuals refused 
to answer any 1 of the questions; data for socioeconomic status are 
presented in percentage of “households.” ‡United States Census and 
Tricounty census data from this category included individuals ages 
15-17 years, which accounted for the increased percentages in these 
categories.
Table 1B—Health Characteristics of the Sample and Comparative Data 
from the United States Census
Health Characteristic Sample (N =3158)
United States Census
(N = 281,421,906)
Hypertension 25.1 25.0
Diabetes 6.0 6.4
Data are presented as percentages. Data for the United States National 
Center for Health Care Statistics are from the following sources: diabe-
tes, Summary Health Statistics for United States Adults, 2001, and hy-
pertension, Health, United States, 2002, table 68; data for the tricounty 
area not available.
Journal of Clinical Sleep Medicine, Vol. 1, No. 4, 2005
Table 2—Prevalence of Different Demographic and Health-Related Variables in the 6 Groups Based on Total Sleep Time 
Total Sleep Time, (in hours)
Demographic and Health-Related Variables ≤ 5 >5 ≤ 6 > 6 ≤ 7 >7 ≤ 8 > 8 ≤ 9 > 9
Total sample, no. (%) 330 (10.4) 606 (19.2) 989 (31.1) 857 (27.1) 264 (8.4) 112 (3.5)
Age, y, mean±SD 42.1±11.5 41.0±12.0 42.2±12.15 42.1±13.2 39.4±14.0 38.4±14.4
Sex, women 53.3 50.3 45.9 49.9 53.4 58.0
Race
White 48.4 64.4 74.0 74.3 73.2 56.8
Black 44.0 29.9 20.5 18.3 19.7 35.6
Other 7.6 5.7 5.5 7.4 7.1 7.6
Loud snoring
Yes 9.1 7.1 4.2 4.8 5.3 6.3
No 90.9 92.9 95.8 95.2 94.7 93.8
Diabetes
Yes 9.4 5.4 5.0 6.0 4.9 10.7
No 90.6 94.6 95.0 94.0 95.1 89.3
Hypertension 
Yes 33.6 26.4 23.8 22.8 21.6 31.3
No 66.4 73.6 76.2 77.2 78.4 68.8
Drinks per week, mean±SD 3.9±9.0 3.9±9.8 3.0±7.0 3.0±7.0 3.4±6.4 2.5±5.2
Colitis
Yes 3.4 4.5 2.5 2.9 1.5 1.8
No 96.6 95.5 97.5 97.1 98.5 98.2
Asthma
Yes 16.4 15.2 11.2 11.0 11.7 18.8
No 83.6 84.8 88.8 89.0 88.3 81.3
Epilepsy
Yes 4.2 2.3 2.0 1.6 3.4 6.3
No 95.8 97.7 98.0 98.4 96.6 93.8
Cancer
Yes 5.5 4.1 4.3 4.8 4.9 8.0
No 94.5 95.9 95.7 95.2 95.1 92.0
Thyroid problems 
Yes 8.8 8.9 5.7 7.7 6.8 15.2
No 91.2 91.1 94.3 92.3 93.2 84.8
Migraines
Yes 32.7 26.2 21.9 20.9 20.5 25.5
No 67.3 73.8 78.1 79.1 79.5 74.5
Depression
Yes 23.0 17.1 13.5 15.0 15.2 26.8
No 77.0 82.9 86.5 85.0 84.8 73.2
Heart disease
Yes 9.4 5.1 5.5 6.0 8.0 12.6
No 90.6 94.9 94.5 94.0 92.0 97.4
Stroke 
Yes 4.5 1.3 0.9 0.8 1.5 2.7
No 95.5 98.7 99.1 99.2 98.5 97.3
Neurologic problems 
Yes 7.6 4.1 3.7 2.7 2.3 8.0
No 92.4 95.9 96.3 97.3 97.7 92.0
Emphysema 
Yes 17.0 8.1 7.3 6.9 10.3 9.8
No 83.0 91.9 92.7 93.1 89.7 90.2
Ulcers 
Yes 16.1 11.4 9.8 6.2 16.1 12.6
No 83.9 88.6 90.2 93.8 93.9 87.4
Arthritis 
Yes 36.7 26.0 23.1 20.8 17.4 34.2
No 63.3 74.0 76.9 79.2 82.6 65.8
Other condition
Yes 31.5 28.5 27.6 27.2 31.1 33.9
No 68.5 71.5 72.4 72.8 68.9 66.1
Data are presented as percentages unless otherwise noted.
359
Obesity and Sleep Duration
Journal of Clinical Sleep Medicine, Vol. 1, No. 4, 2005
Sleep duration was assessed in the following manner: participants 
reported their habitual TST (per 24 hours per day) separately 
for weekends and weekdays. These questions referred to the 2-
week period immediately prior to the study. A weighted average 
of weekend and weekday reports was calculated ([5 × weekday 
total sleep time + 2 × weekend total sleep time] / 7) to determine 
average daily TST.
 Self-report height and weight were used to determine the BMI. 
BMI was calculated by using the formula, weight in kilograms 
divided by the square of the height in meters. BMI was used 
both as a dichotomous (obesity, defined as BMI ≥ 30)22 and a 
continuous variable to facilitate comparisons with previous 
literature. Hypertension and diabetes were determined using 
questions regarding any history of these disorders. Because 
obstructive sleep apnea is often comorbid with obesity, in the 
absence of nocturnal polysomnography, snoring was used as 
a surrogate measure. Specifically, participants responded to 
a question using a Likert scale: Would you say your snoring is 
quiet, moderate, loud, or very loud? We assessed the presence of 
depression by using the Diagnostic Interview Schedule.23 Alcohol 
use was assessed by combining the following 2 questions; number 
of days used per week and number of drinks per occasion. Alcohol 
drinks per week was then calculated by multiplying the 2 numbers 
from these questions. Any self-reported history of heart disease, 
thyroid problems, cancer, stomach ulcers, colitis, arthritis, 
migraines, asthma, chronic bronchitis or emphysema, stroke, 
seizures, or other neurologic problems was also assessed. Finally, 
other problems were determined by asking about the presence 
of additional health conditions that may have been present. An 
average nightly TST was used to divide the sample into 6 groups 
(≤ 5, > 5 ≤ 6, > 6 ≤ 7, > 7 ≤ 8, > 8 ≤ 9, > 9 hours). The reference 
group for the analysis of the study population was 7 to 8 hours of 
TST per night, as this included both the mean and the median of 
the population sample. The prevalence of demographic and health 
variables in each TST groups is presented in Table 2.
Analysis
 The association of sleep duration with obesity, BMI, and dia-
betes was assessed, after adjustment fora number of demographic 
and health-related measures. Three separate analysis were per-
formed with BMI, obesity, and diabetes as dependent measures. 
In each analysis, the following demographic and health variables 
were used as covariates: age, sex, race, socioeconomic status, 
hypertension, diabetes, loud snoring, drinks per week, colitis, 
asthma, epilepsy, cancer, thyroid problems, migraines, depres-
sion, heart disease, stroke, “neurologic conditions,” emphysema, 
ulcers, arthritis and “other condition.” Only significant covariates 
were included in each of the final analysis. For the first analysis, 
a multiple logistic regression model was used to predict obesity 
based on the 6 TST groups. For the second analyses, BMI, as a 
continuous measure, was predicted based on each of the 6 TST 
groups, using a multiple linear regression model. In the third 
analysis, a multiple logistic regression model was used to predict 
the presence of diabetes based on each of the 6 TST groups. Fi-
nally, because BMI is an important diabetes-related measure, it 
was added to the third analysis (multiple logistic model) to deter-
mine its potential influence on the relationship between diabetes 
and TST.
 
RESULTS
 The sociodemographic characteristics of the study sample and 
comparative data from the United States census are shown in Ta-
ble 1. Table 2 shows the demographic and health characteristics 
of our sample, separately for each of the 6 TST groups. The over-
all prevalence of obesity in the sample was 24.8%, and the mean 
BMI was 27.2 kg/m2 ; both were significantly higher in individu-
als with lower TST (p < .05). In the subsample of residents in the 
360
M Singh, CL Drake, T Roehrs et al
Table 3—Results of Logistic Regression Predicting Obesity Prevalence (Body Mass Index ≥ 30) Adjusting for Significant Covariates
Variables Unadjusted Odds Std p value Adjusted Odds 95% Confidence Interval
Ratio Error Ratio Lower Upper
Age .001 .004 .732 1.001 .994 1.009
Sex .021 .090 .818 1.021 .855 1.219
Race
Caucasian .000 
African American† .452 .101 .000 1.571 1.289 1.914
Other race† .256 .180 .155 1.292 .908 1.840
Hypertension .764 .099 .000 2.147 1.770 2.605
Diabetes 1.053 .165 .000 2.866 2.076 3.958
Alcohol intake -.020 .007 .003 .980 .967 .993
Snoring .922 .168 .000 2.515 1.811 3.493
Arthritis .402 .103 .000 1.495 1.222 1.830
Total sleep time 
> 7 ≤ 8 .001
≤5* .529 .153 .001 1.697 1.257 2.290
> 5 ≤ 6* .316 .130 .015 1.371 1.062 1.770
> 6 ≤ 7* .133 .118 .259 1.142 .907 1.439
> 8 ≤9* -.084 .185 .649 .919 .639 1.321
> 9* -.003 .250 .990 .997 .611 1.626
Constant -1.915 .231 .000 .147 
*Compared with reference group total sleep time (in hours) >7 ≤ 8.
†Compared with the reference group Caucasian.
Variables not in the equation are asthma, epilepsy, cancer, thyroid problems, migraines, colitis, depression, heart disease, stroke, other neurologic 
conditions, emphysema, ulcers, and other condition.
Journal of Clinical Sleep Medicine, Vol. 1, No. 4, 2005
city of Detroit, the prevalence of obesity was much higher (34%), 
consistent with previous epidemiologic data. Obesity, BMI, and 
diabetes analyses are described separately below.
Obesity
 Table 3 shows the results of the logistic regression model pre-
dicting obesity prevalence, adjusting for significant covariates. 
Compared to those obtaining 7 to 8 hours of TST, individuals ob-
taining 5 hours or less of TST, and those obtaining 5 to 6 hours 
or less of TST, had significantly increased odds of being obese, 
after controlling for age, sex, socio-economic status, loud snor-
ing, hypertension, diabetes, arthritis, and alcohol intake (odds ra-
tio [OR]=1.7, 95% confidence interval [CI]= 1.3 -2.3; OR = 1.4, 
95% CI=1.1-1.8 respectively)( Figure 1 and Table 3). This model 
accounted for 13% of the variance in obesity.
Body Mass Index
 Table 4 shows the results of the multiple linear regression model 
predicting BMI as a continuous variable adjusting for significant 
covariates. Lower TST (≤ 6 and < 5 ≥ 6 hours per night) was 
associated with an elevated BMI (Figure 2 and table 4).
Diabetes
 Lower TST was associated with a higher prevalence of diabetes 
(p < .05). However, after adjusting for BMI, TST was no longer a 
significant predictor of diabetes in the multiple logistic regression 
model.
 Interestingly, 18.7% of the African American population slept 
for 5 hours of less, as compared with 7.4% of the Caucasian popu-
lation (p < .001). Of the African Americans, 33.0% were obese, 
versus 21.7% of the Caucasians (p < .001). Coincidently, with 
this relationship, 8.2% of the African American population was 
diabetic, as compared with 5.8% of the Caucasian population (p < 
.001).
 Although the 7- to 8-hour TST group was used as the reference 
category and there were no significant differences found between 
that group and the 9-hours-or-greater TST group, visual inspection 
of Figures 1 and 2 reveals the possibility of a quadratic relation-
ship between TST and BMI or obesity. A posthoc trend analysis 
revealed linear trends (p < .01) but did not show any significant 
quadratic trends after controlling for covariates (p > .05)
DISCUSSION
 The present study assessed the association between nightly 
TST and obesity in a population-based sample. The overall 
prevalence of obesity was 25% and was significantly higher in 
individuals with lower TSTs. Furthermore, there was a clear 
dose-response relationship between sleep duration and BMI after 
adjusting for significant covariates, including age, sex, race, loud 
snoring, hypertension, diabetes, alcohol intake, and arthritis. Our 
findings therefore, support the hypothesis that reduced TST is as-
sociated with an increased prevalence of obesity. This is consis-
tent with findings from previous studies showing associations be-
tween sleep duration and BMI or obesity in children and younger 
adults.15-19 Sleep duration was a statistically significant correlate 
of obesity, but any potential causal role for sleep in contribut-
ing to weight regulation cannot be determined with the current 
study methods. While other modifiable variables such as caloric 
intake and physical activity likely account for a greater amount of 
variance in obesity24 than does TST, future studies are needed to 
determine the relative strength of these variables in relation to one 
another.
 The relationship between short sleep duration and obesity may 
seem somewhat inconsistent with animal data, in which shorter 
sleep duration is associated with significant weight loss. For ex-
ample, chronic total sleep deprivation in the rat produces weight 
loss, increased catabolic activity, and eventual death.25 This cata-
bolic process that results from total sleep deprivation is differ-
ent from the largely anabolic effects that partial sleep deprivation 
causes through the metabolic and endocrine alterations shown in 
humans.13
 Although, the current epidemiologic data do not address any 
possible causal relationship between reduced TST and increased 
obesity, there is some support for a potential mechanistic involve-
ment of sleep loss in weight regulation. For example, the hor-
mone leptin has been implicated in the regulation of food intake 
and metabolism26 and is modified by changes in sleep-wake activ-
ity. Circulating levels of leptin have a distinct circadian rhythm, 
with the nocturnal rise serving to suppress appetite during over-
night periods of fasting and sleep.27 Sleep deprivation is known to 
suppress leptin levels and may prevent the suppression of appe-
tite.28 Serum cortisol concentrations also show an increase toward 
the evening in conditions of total and partial sleep deprivation.13 
Cortisol has a lipogenic property, which may contribute to the 
weight gain in chronic sleep restriction. Shortened sleep duration 
is also associated with decreasedglucose tolerance and increased 
sympathetic tone, both of which are well-recognized risk factors 
for the development of insulin resistance, obesity, and hyperten-
sion.14 Finally, nocturnal growth hormone secretion is dependent 
on sleep, and, in the absence of sleep, growth hormone secretion 
is markedly decreased.29 Growth hormone is essential for main-
taining lipolysis during the night29 and, thus, might be another 
factor contributing to development of obesity in individuals with 
reduced sleep. Although sleep duration may be a modifiable risk 
factor for obesity, its role in terms of chronic and acute sleep 
loss clearly needs further controlled experimentation in humans, 
with a focus on a potential mechanistic relationship. Addition-
ally, studies should be performed to determine the relationship of 
short sleep duration to obesity, relative to interactions with other 
known contributing factors, such as nutrition and exercise.
 Our data revealed that short sleep duration was significantly 
361
Obesity and Sleep Duration
0
5
10
15
20
25
30
35
40
≤ 5 >5 ≤ 6 >6 ≤ 7 >7 ≤ 8 >8 ≤ 9 >9
P
re
va
le
nc
e 
of
 O
be
si
ty
 (%
)
Total Sleep Time Groups (hours)
n= 330 
n= 606 
n= 989
n= 857
n= 264
n= 112
*
*
 
Figure 1—Prevalence of obesity in each group, based on total sleep 
time (unadjusted). *Indicates significant differences between groups 
(vs >7 ≤ 8 hours per night).
Journal of Clinical Sleep Medicine, Vol. 1, No. 4, 2005
associated with diabetes. However, after controlling for BMI, this 
association was attenuated and was no longer significant. These 
data are consistent with the findings of Ayas et al,30 suggesting 
that the association between sleep loss and diabetes may be medi-
ated by weight gain. Interestingly, our data showed that the preva-
lence of short sleep time (ie, 5 hours or less per night) was higher 
in African Americans, as compared with other racial groups. This 
finding was coincident with higher rates of obesity and diabetes 
in African Americans, as compared with Caucasians, as has been 
previously reported.1 Future studies are needed to assess what the 
relative risk short sleep duration has for obesity in various de-
mographic populations, in conjunction with other modifiable risk 
factors such as calorie intake and exercise.
 Several limitations of our study warrant mention. First, infor-
mation about sleep and other health parameters was self-reported: 
self-reported sleep duration may not have accurately reflected 
TST or its stability over time. However, self-reported sleep dura-
tion has been investigated in other studies and has been shown to 
closely approximate objective measures of sleep duration.31 Ob-
jective measurement in the sleep laboratory also has certain limi-
tations. Namely, first-night effects are often present, and the ex-
pense and time associated with polysomnograms are prohibitive 
in most large-scale epidemiologic studies and office-based clini-
cal practice. Finally, despite controlling for numerous health-re-
lated and demographic variables, we cannot rule out the presence 
of unrecognized confounders that may not have been accounted 
for (for example, physical activity level). However, considering 
the clear dose-response relationship found, it is doubtful that such 
variables would fully account for the observed effects. 
 To summarize, our population-based data suggest that short 
sleep duration is associated with elevated prevalence of obesity 
and adds to the growing body of evidence supporting this rela-
tionship. The association between short sleep duration and obesi-
ty found in the current study is in concordance with findings from 
other cross-sectional and longitudinal studies that have investigat-
ed this association in different social and cultural settings.15-19 We 
have shown that this association persists even after adjustments 
for age, sex, race, diabetes, hypertension, loud snoring, arthritis, 
and alcoholic drink intake. Furthermore, our data replicate pre-
vious findings using a population-based sample. Further studies 
are needed to better elucidate the possible mechanisms that may 
underlie this association and to determine whether sleep duration 
is a modifiable risk factor for obesity and its effect on long-term 
health.
REFERENCES
1. Flegal KM, Carroll MD, Ogden CL, Johnson CL. Prevalence 
and trends in obesity among US adults, 1999-2000. JAMA 
2002;288:1723-7.
2. Allison DB, Fontaine KR, Manson JE, Stevens J, VanItallie TB. 
Annual deaths attributable to obesity in the United States. JAMA 
1999;282:1530-8.
3. Colditz GA. Economic costs of obesity and inactivity. Med Sci 
Sports Exerc 1999;31:S663-7.
4. Ashton WD, Nanchahal K, Wood DA. Body mass index and 
metabolic risk factors for coronary heart disease in women. Eur 
Heart J 2001;22:46-55.
5. Visser M, Bouter LM, McQuillan GM, Wener MH, Harris TB. 
Elevated C-reactive protein levels in overweight and obese adults. 
JAMA 1999;282:2131-5.
6. Must A, Spadano J, Coakley EH, Field AE, Colditz G, Dietz WH. 
The disease burden associated with overweight and obesity. JAMA 
1999;282:1523-9.
7. Breslau N, Roth T, Rosenthal L, Andreski P. Daytime sleepiness: 
an epidemiological study of young adults. Am J Public Health 
1997;87:1649-53.
8. Hammond EC. Some preliminary findings on physical complaints 
from a prospective study of 1,064,004 men and women. Am J Public 
Health 1964;54:11-24.
9. Kripke DF, Simons RN, Garfinkel L, Hammond EC. Short and long 
sleep and sleeping pills. Is increased mortality associated? Arch 
Gen Psychiatry. 1979;36:103-16.
10. Walsleben JA, Kapur VK, Newman AB, et al. Sleep and reported 
daytime sleepiness in normal subjects: the Sleep Heart Health 
Study. Sleep 2004;27:293-8.
11. Drake CL, Roehrs, T, Burduvali, E, Bonahoom, A, Rosekind, M, 
Roth, T. Effects of rapid versus slow accumulation of eight hours of 
sleep loss. Psychophysiology. 2001;38:979-987.
12. Van Dongen HP, Maislin G, Mullington JM, Dinges DF. The 
cumulative cost of additional wakefulness: dose-response effects 
on neurobehavioral functions and sleep physiology from chronic 
sleep restriction and total sleep deprivation. Sleep 2003;26:117-26.
13. Spiegel K, Leproult R, Van Cauter E. Impact of sleep debt on 
metabolic and endocrine function. Lancet 1999;354:1435-9.
362
M Singh, CL Drake, T Roehrs et al
Table 4—Results of Multiple Linear Regression Predicting Body Mass 
Index as a Continuous Variable, Adjusting for Significant Covariates
Variables Unstandardized
Coefficients
T p value
 β Std Error 
Age 0.01 0.009 1.576 .115
Sex 0.619 0.211 2.93 .003
Race 0.693 0.174 3.976 .000
Socioeconomic 
status
-0.103 0.051 -2.040 .041
Hypertension -2.3 0.250 -9.258 .000
Diabetes -4.01 0.442 -9.098 .000
Arthritis -.914 0.254 -3.600 .000
Loud snoring 3.11 0.442 -4.420 .000
Drinks/wk -0.03 0.013 -2.811 .005
TST groups -.234 0.053 -4.420 .000
Constant 46.08 1.506 30.605 .000
TST refers to total sleep time.
 r2 = .140 (adjusted into r2 = .137); nonsignificant variables that were 
not entered in the equation are colitis, asthma, epilepsy, cancer, thyroid 
problems, colitis, migraines, depression, heart disease, stroke, other 
neurologic conditions, emphysema, ulcers, and other condition.
25
25.5
26
26.5
27
27.5
28
28.5
29
29.5
30
≤5 >5≤6 >6≤ 7 >7≤ 8 >8≤9 >9
Total Sleep Time Groups (hours)
B
od
y 
M
as
s 
In
de
x 
(k
g/
m
2)
*
*
n= 330 
n= 606 
n= 989
n= 857
n= 264
n= 112
 
Figure 2—Mean body mass index ± SEM in each group, based on to-
tal sleep time (unadjusted). *Indicates significant differences between 
groups (vs >7 ≤ 8 hours per night).
Journal of Clinical Sleep Medicine, Vol. 1, No. 4, 2005 363
Obesity and Sleep Duration14. Scheen AJ, Byrne MM, Plat L, Leproult R, Van Cauter E. 
Relationships between sleep quality and glucose regulation in 
normal humans. Am J Physiol 1996;271:E261-70.
15. Kripke DF, Garfinkel L, Wingard DL, Klauber MR, Marler MR. 
Mortality associated with sleep duration and insomnia. Arch Gen 
Psychiatry 2002;59:131-6.
16. Hasler G, Buysse DJ, Klaghofer R, et al. The association between 
short sleep duration and obesity in young adults: a 13-year 
prospective study. Sleep 2004;27:661-6.
17. Vioque J, Torres A, Quiles J. Time spent watching television, sleep 
duration and obesity in adults living in Valencia, Spain. Int J Obes 
Relat Metab Disord 2000;24:1683-8.
18. Sekine M, Yamagami T, Handa K, et al. A dose-response relationship 
between short sleeping hours and childhood obesity: results of the 
Toyama Birth Cohort Study. Child Care Health Dev 2002;28:163-
70.
19. Von Kries R, Toschke AM, Wurmser H, Sauerwald T, Koletzko B. 
Reduced risk for overweight and obesity in 5- and 6-y-old children 
by duration of sleep--a cross-sectional study. Int J Obes Relat Metab 
Disord 2002;26:710-6.
20. Spiegel K, Tasali E, Penev P, Van Cauter E. Brief communication: 
Sleep curtailment in healthy young men is associated with decreased 
leptin levels, elevated ghrelin levels, and increased hunger and 
appetite. Ann Intern Med 2004;141:846-50.
21. United States Census Bureau. (US Government). 2000 Census of 
Population and Housing; 2005.
22. Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CW, Jr. Body-
mass index and mortality in a prospective cohort of U.S. adults N 
Engl J Med. 1999;341:1097-105.
23. Helzer JE, Robins LN, McEvoy LT, et al. A comparison of 
clinical and diagnostic interview schedule diagnoses. Physician 
reexamination of lay-interviewed cases in the general population. 
Arch Gen Psychiatry 1985;42:657-66.
24. McTigue KM, Harris R, Hemphill B, et al. Screening and 
interventions for obesity in adults: summary of the evidence for the 
U.S. Preventive Services Task Force. Ann Intern Med. 2003;139:933-
49.
25. Rechtschaffen A, Bergmann BM, Everson CA, Kushida CA, 
Gilliland MA. Sleep deprivation in the rat: X. Integration and 
discussion of the findings. Sleep 1989;12:68-87.
26. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman 
JM. Positional cloning of the mouse obese gene and its human 
homologue. Nature 1994;372:425-32.
27. Saad MF, Riad-Gabriel MG, Khan A, et al. Diurnal and ultradian 
rhythmicity of plasma leptin: effects of gender and adiposity. J Clin 
Endocrinol Metab 1998;83:453-9.
28. Mullington JM, Chan JL, Van Dongen HP, et al. Sleep loss reduces 
diurnal rhythm amplitude of leptin in healthy men. J Neuroendocrinol 
2003;15:851-4.
29. Boyle PJ, Avogaro A, Smith L, et al. Role of GH in regulating 
nocturnal rates of lipolysis and plasma mevalonate levels in normal 
and diabetic humans. Am J Physiol 1992;263:E168-72.
30. Ayas NT, White DP, Al-Delaimy WK, et al. A prospective study 
of self-reported sleep duration and incident diabetes in women. 
Diabetes Care 2003;26:380-4.
31. Lockley SW, Skene DJ, Arendt J. Comparison between subjective 
and actigraphic measurement of sleep and sleep rhythms. J Sleep 
Res 1999;8:175-83.