| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Psychiatry (E.O.B., A.N.V., S.L.C., A.K.) and Health Evaluation Sciences (H.-M.L.), Sleep Research and Treatment Center, Pennsylvania State University College of Medicine, Hershey, Pennsylvania 17033; and Department of Psychiatry (A.V.-B.), Autonomous University, 28003 Madrid, Spain
Address all correspondence and requests for reprints to: E. O. Bixler, Ph.D., Sleep Research and Treatment Center, Department of Psychiatry MC:H073, Pennsylvania State University, College of Medicine, 500 University Drive, Hershey, Pennsylvania 17033. E-mail: eob1{at}psu.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: The purpose of this study was to assess the association between the complaint of EDS and sleep apnea, considering a wide range of possible risk factors in a population sample.
Design and Setting: We examined this question in the Penn State cohort (a random sample of 16,583 men and women from central Pennsylvania, ranging in age from 20 to 100 yr). A random subset of this cohort (n = 1,741) was further evaluated for one night in the sleep laboratory.
Main Outcome Measure: The main measure was a complaint of EDS.
Results: The final logistic regression model indicated depression was the most significant risk factor for EDS followed by body mass index, age, typical sleep duration, diabetes, smoking, and finally sleep apnea. The strength of the association with EDS decreased with increasing age, whereas the association of depression with EDS was stronger in the young. EDS is more prevalent in the young (<30 yr), suggesting the presence of unmet sleep needs and depression, and in the very old (>75 yr), suggesting increasing medical illness and health problems. EDS was associated with a reduced report of typical sleep duration without any association with objective polysomnographic measures.
Conclusions: It appears that the presence of EDS is more strongly associated with depression and metabolic factors than with sleep-disordered breathing or sleep disruption per se. Our findings suggest that patients with a complaint of EDS should be thoroughly assessed for depression and obesity/diabetes independent of whether sleep-disordered breathing is present.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The literature reports a wide variety of symptoms in addition to sleep disturbance that are associated with EDS, indicating that the mechanism of EDS is multifactorial. For example, EDS has been shown to be associated with obesity in the absence of SDB (3, 4). Recent literature suggests that it may also be associated with the metabolic syndrome (e.g. obesity, diabetes, insulin resistance) (5). Furthermore, EDS has been reported to be more common in women (2, 6, 7, 8) and especially in those with mental health disorders and even more so with depression (7, 9, 10, 11, 12, 13). Considering alternative factors for EDS may explain why continuous positive airway pressure at times fails to improve EDS, especially in milder cases of sleep apnea (14).
The purpose of this study was to assess the relative association between the complaint of EDS and SDB, considering a wide range of possible etiologic factors in a population sample.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In the second phase of this study, a subsample of 741 men and 1000 women randomly selected from those subjects previously interviewed by telephone were studied in our sleep laboratory. The response rate for this phase was 67.8 and 65.8% for men and women, respectively. We contrasted those subjects who were recorded in the laboratory with those who were selected but not recorded in terms of age, body mass index (BMI), and prevalence of sleep disorders. There were no significant differences between these two groups on any of these variables. Each subject selected for laboratory evaluation completed a comprehensive sleep history and physical examination. All subjects were evaluated for one night in the sleep laboratory in sound-attenuated, light- and temperature-controlled rooms. During this evaluation, each subject was continuously monitored for 8 h using 16-channel polygraphs including electroencephalogram, electrooculogram, and electromyogram. The sleep records were subsequently scored independently according to standardized criteria (19). Respiration was monitored throughout the night by use of thermocouples at the nose and mouth and thoracic strain gauges. All-night recordings of hemoglobin oxygen saturation (SaO2) were obtained with an oximeter attached to the finger.
As part of the initial interview (phase I), we assessed for the presence of all sleep disorders. The presence of EDS was established based on a moderate or severe rating on either of the following two questions: "do you feel drowsy or sleepy most of the day but manage to stay awake?" and "do you have any irresistible sleep attacks during the day?" In addition, we ascertained whether the respondent was currently treated for allergies, asthma, depression (including suicidal thoughts or attempts), diabetes, or hypertension. In the subsample of 1741, diabetes was defined as being treated for diabetes or having a fasting blood sugar greater than 126 from blood drawn the morning after the subjects slept in the sleep laboratory. Additional information was obtained during the polysomnographic evaluation (phase II) including history of smoking (current use of any type of tobacco product) and alcohol use (more than two alcoholic drinks per day) and objective sleep data including sleep apnea assessment. For the purposes of this study, sleep apnea was defined as an obstructive apnea or hypopnea index 15 or greater (OHI
15). In addition, for those subjects who participated in polysomnographic phase of this study, we assessed subjectively: "how many hours do you usually sleep at night?" and "how many hours of sleep would you like to get?"
The design of this study included oversampling of those at higher risk for sleep disordered breathing to increase the precision of the risk estimates. Because of this, numeric weights, which have been previously reported (15, 16), were developed for the analysis to obtain prevalence estimates for the original target population. Furthermore, as we began evaluating the sample of women, it appeared that the mean BMI (kilograms per square meter) was markedly too high, compared with the national population. Thus, a poststratification population control weight was also established based on the Third National Health and Nutrition Examination Survey data (20).
Univariate analyses of these data were initially conducted to compare those with and without a complaint of EDS with respect to various outcomes using weighted t tests or 
2 tests. Means ± SEs, odds ratios (ORs) ± 95% confidence intervals (CIs) and the P values of testing the difference between two groups are reported. This analysis was completed within the larger sample (phase I) whenever possible. This initial analysis was followed by a weighted multivariate logistic regression in the laboratory sample (phase II). To assess the contribution of OHI15 relative to all other available risk factors, we performed a stepwise selection process with OHI15 always in the model. The initial set of covariates also included age, logarithm-transformed BMI [log(BMI)], diabetes, depression, smoking, alcohol use, gender, allergy, asthma, hypertension, alcohol, thyroid function measures [levels of T4, thyroid uptake (TUptake), and the ratio between TUptake and free T4 (TU-FT4)], subjective typical sleep duration, and objective polysomnographic data [percent stage 1 (% 1), number of wakes (Nwakes), number of sleep stage changes (Nstgch), percent slow wave (%SW), and percent rapid eye movement (%REM)]. Based on the findings from the main effect model, we further considered a second model with all possible two-way interactions, as appropriate.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
). The age-specific prevalence for EDS observed in this sample is presented in Fig. 1, which shows that a linear decline was present between the ages of 30 and 75 yr. The prevalence was higher for ages younger and older than this age range.
|
|
reports the BMI-specific prevalence, which shows a dramatic increase in the prevalence of EDS around the threshold of BMI that is considered to be overweight (BMI = 28). In our sample there were strong associations observed between EDS and reports of treatment for diabetes [OR = 1.9 (1.6, 2.4), P < 0.001] (Fig. 3) and between EDS and those reporting current treatment for depression [OR = 3.3 (2.9, 3.8), P < 0.0001] (Fig. 4).
|
|
|
displays the age-specific % ST for those without any complaints of sleep disturbance by gender. As age increased, there was a decline in % ST, which is analogous to the decline observed for EDS (Fig. 1
). A similar pattern was observed in those with a complaint of EDS. To investigate further the association of EDS with possible sleep disturbance, we also assessed the association with % 1, Nwakes, and Nstgch. The association of each of these estimates of sleep fragmentation for those subjects with EDS, compared with those without EDS, was as follows: % 1 = 10.8 ± 0.7 vs. 10.9 ± 0.2% (P = 0.894); Nwakes = 9.7 ± 0.5 vs. 10.8 ± 0.2% (P = 0.037); and Nstgch = 77.2 ± 2.3 vs. 75.2 ± 0.6 (P = 0.402). The subjective report of the typical and desired sleep duration reported by subjects with EDS, compared with those without EDS, was: typical sleep duration = 6.5 ± 0.1 vs. 6.9 ± 0.03 h (P = 0.007); and sleep desired = 7.6 ± 0.1 vs. 7.6 ± 0.03 h (P = 0.651). We also assessed for differences in sleep stage distribution in terms of %REM and %SW sleep. The sleep stage distribution comparing those reporting EDS with those not reporting EDS was: %REM = 17.1 ± 0.5 vs. 17.0 ± 0.2 (P = 0.464); and %SW = 4.1 ± 0.5 vs. 3.9 ± 0.2% (P = 0.641). Finally, EDS has been commonly associated with SDB. The prevalence of EDS within those with OHI 15 was 18.3%, compared with 10.7% within those without any SDB [OR = 1.9 (1.0, 3.4), P = 0.038]. The association of EDS with SDB was also evaluated using the minimum SaO2 observed during the 8-h polysomnogram. The minimum SaO2 for those with EDS, compared with those without EDS, was 90.8 ± 0.3 vs. 91.0 ± 0.2 (P = 0.57, respectively).
|
To assess the relative significance of the variables shown to be associated with EDS, we used a multivariate analysis using logistic regression. The standardized effect sizes (beta coefficients divided by the corresponding SEs), P values, and ORs based on the final logistic regression model are reported. Table 2 shows the results of the main effects model, which indicate that a report of being treated for depression is the most significant risk factor for the complaint of EDS, followed by BMI, age, subjective estimate of typical sleep duration, diabetes, smoking, and finally sleep apnea. For continuous variables, the OR is reported for 1 and 2 SD change from the mean. Note that the complaint of EDS decreased with increasing age, and the quadratic terms for both age and log(BMI) were not significant. More importantly, the strength of the association between OHI>15 and EDS, after adjusting for other covariates did not change but became nonsignificant, compared with the univariate result. We also assessed for OHI>5 and OHI>30 to evaluate whether the relationship was sensitive to the threshold for OHI selected. The OR observed when OHI>5 was used increased, whereas the OR for OHI>30 decreased. In neither case did the internal relationship with the other variables change nor did OHI become significant in the model. In addition, measures of sleep disturbance (% 1, Nwakes, and Nstgch), sleep stage distribution (%SW and %REM), gender, allergy, asthma, hypertension, and alcohol use were not independently associated with EDS. Furthermore, no effect of any medication use was independently associated with EDS. Finally, thyroid function measures were also not independently associated with EDS.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
We observed a robust relationship between EDS and BMI. Specifically, the BMI-specific prevalence of EDS remained constant until the overweight threshold (BMI = 28) was reached. Beyond this BMI threshold, the prevalence of EDS increased in an exponential manner. In the logistic regression model of this study, BMI was independently associated with EDS with an effect size second only to depression. This finding is consistent with previous studies that have observed a strong association between BMI and EDS in the absence of SDB (3, 4). In addition, the strong association between EDS and diabetes adds further support for the association of EDS with the metabolic syndrome (5). This finding is clinically significant because it would suggest that diabetes should be considered whenever a complaint of EDS is present. Specifically, we observed that obesity (BMI) and diabetes were both independently associated with EDS. This suggests that a simple fasting blood sugar test should be considered appropriate in the presence of an EDS complaint.
The third strongest independent factor in our multivariate analysis was age. The prevalence of EDS decreased in a linear fashion with increasing age between the age limits of 30 and 75 yr. The prevalence of EDS was higher in the young and the very old, the former most likely a result of increased unmet sleep needs and the latter associated with increased health problems and medical illness. This U-shaped distribution is consistent with what has previously been reported (21). As anticipated, sleep efficiency also declined with increasing age. This relationship argues strongly against the model that EDS is caused solely by decreased nocturnal sleep efficiency because the prevalence of both EDS and objective measures of sleep efficiency declined with age. Further support for this model is the finding that the administration of an arousing hormone (i.e. CRH) after sleep onset caused a greater degree of sleep disturbance in middle age, compared with young men (22). All of these findings are more consistent with the model that with increasing age the homeostatic sleep mechanism is weakening, affecting both nighttime sleep efficiency and daytime sleep propensity. Finally, evaluating the interaction between age and depression with EDS suggests that the strongest association between depression and EDS was in the young. This finding is consistent with clinical observations that depression is more commonly associated with hypersomnia in the young, compared with the old (23). From a practical standpoint, these findings combined suggest that EDS in the young is not just the result of increased unmet sleep need but also possibly of depression that should be appropriately evaluated and managed. Finally, from a population perspective such as this sample, age-related changes in sleep patterns or self-induced sleep restriction appear not to be as important for EDS as depression and/or the metabolic syndrome.
The fourth strongest independent factor in our EDS model was a subjective estimate of typical sleep duration. This finding is in the absence of any objective indication of sleep loss, which included estimates of total sleep time, sleep stage distribution, and sleep fragmentation. There have been several studies that have shown that obesity is associated with reduced nocturnal sleep (3, 13, 24, 25, 26, 27, 28). Because obesity controlling for SDB has been shown to be associated with EDS, several studies have attempted to evaluate nocturnal sleep (both objective and subjective) with a complaint of EDS (3, 6, 7, 13, 29, 30). These studies have consistently reported an association between subjective but not objective sleep duration and a complaint of EDS. Two studies, one in subjects without SDB and the other in subjects with SDB, have evaluated the association between EDS based on objective polysomnographic data for nocturnal sleep and Multiple Sleep Latency Test for daytime sleepiness (3, 29). The study in subjects without SDB found that nocturnal polysomnographic data were associated with obesity, i.e. obese subjects slept less than nonobese. Both studies, whether the subjects had SDB (29) or not (3), reported that within the obese subjects, the degree of objective daytime sleepiness was positively associated with the amount of nocturnal sleep, i.e. obese subjects with more nocturnal sleep had more daytime sleepiness. This finding argues against a simple cause-and-effect link between objectively measured nocturnal sleep loss and daytime sleepiness. It is also commonly assumed that sleep fragmentation is the primary mechanism for EDS. In this study we observed that polysomnographic measures of sleep disturbance (e.g. %1, Nwakes, Nstgch) did not contribute to the risk for EDS, which is consistent with previously reported findings (29, 31). Based on these findings it is possible to speculate an alternative mechanism for the association between a complaint of EDS and a subjective but not objective report of nocturnal sleep loss. In some individuals it is possible that the presence of a sensation of daytime sleepiness may lead them to believe that it is caused be nocturnal sleep loss because they are not aware of other possible alternatives.
To our knowledge this is the first time that smoking has been raised as a risk factor for EDS. Previous data have indicated that smoking disturbs sleep (32). One could speculate that the EDS reported the next day could be the result of nocturnal sleep disturbance caused by the smoking; however, as we have previously noted, there was not a strong association between EDS and sleep disturbance in our general population sample. An alternative hypothesis is that the stimulant effect of nicotine is used by those with a complaint of EDS to self-treat their problem. Recognition that EDS/fatigue may be a risk factor for smoking might lead to more effective strategies to reducing smoking and its adverse effects.
The final variable that was present in the logistic regression model was sleep apnea (OHI15). This variable did not make a significant contribution to the model (P = 0.266). This is consistent with the reported weak association between apnea/hypopnea index and EDS in epidemiologic samples (2, 33). This finding may explain to some extent why continuous positive airway pressure at times fails to improve EDS in patients with sleep apnea (14).
In this study gender was not associated with EDS. Previous studies have reported a gender-related association with EDS (2, 6, 7, 8), whereas others have not (13, 33, 34, 35). This inconsistency may in part reflect the way daytime sleepiness was ascertained. We also assessed whether a hypothyroid condition might contribute to the presence of EDS in our model. We assessed for TUptake, T4, and TU-FT4. None of these factors contributed to the model. We did not have a measure of TSH, so we cannot rule out that a subclinical hypothyroid condition might have contributed to the presence of EDS.
Finally, there are weaknesses with this study that need to be considered. First, this study is cross-sectional, and thus, causation cannot be directly tested. For example, the changes observed associated with increasing age could be considered a reflection of a survivor cohort. However, the data demonstrating that middle-aged men, compared with young men, were more susceptible to a CRH challenge in terms of sleep disturbance strongly support the model that the homeostatic sleep mechanism is weakening with age affecting both night time sleep efficiency and daytime sleep propensity (22). Furthermore, the subjective method of ascertaining the complaint of EDS could be considered another weakness. However, a similar decline in sleepiness with increasing age has been reported in objective measures of daytime sleepiness using the Multiple Sleep Latency Test (36, 37, 38, 39). The ascertainment of sleep apnea may have been underestimated because we used only thermocouples to assess airflow, which is consistent with other large-scale epidemiologic samples (2, 33). To assess the possible impact of this, we evaluated the model at various OHI thresholds ranging from OHI>5 to OHI>30. In general the OR for OHI became larger as the threshold was lowered; however, the internal relationship among the other relevant variables did not change and OHI never became significant. The objective nocturnal sleep data reported in this study were based on a single night. Thus, we cannot rule out the possible first-night effect. Because a measure of TSH was not available, the possibility that a subclinical hypothyroid may play a small contributing role cannot be ruled out. Finally, in the Penn State cohort, we were not able to differentiate between type 1 and type 2 diabetes; however, type 1 is relatively rare in adults; thus, it can be safely assumed that the overwhelming majority of subjects reporting diabetes have type 2.
In summary, these data indicate that when diagnosing a case with a complaint of EDS, sleep disturbance (e.g. due to sleep apnea) should not be considered the only cause. It appears that EDS is more strongly associated with mood factors (e.g. depression) as well as metabolic factors (obesity and/or diabetes), i.e. the metabolic syndrome. EDS appears to be more prevalent in the very young, suggesting unmet sleep needs and/or depression. EDS is also more prevalent in the very old, most likely associated with increasing medical illnesses and health issues. Our findings indicate that patients with a complaint of EDS should be adequately assessed for depression, obesity, and/or diabetes both in the presence or absence of sleep apnea and then treated appropriately.
|
Footnotes |
|---|
First Published Online June 7, 2005
Abbreviations: BMI, Body mass index; CI, confidence interval; EDS, excessive daytime sleepiness; log(BMI), logarithm-transformed BMI; Nstgch, number of sleep stage changes; Nwakes, number of wakes; OHI15, obstructive apnea or hypopnea index 15 or greater; OR, odds ratio; % 1, percent stage 1; %REM, percent rapid eye movement; SaO2, oxygen saturation; SDB, sleep-disordered breathing; % ST, percent sleep time; %SW, percent slow wave; TU-FT4, ratio between TUptake and free T4; TUptake, thyroid uptake.
Received January 6, 2005.
Accepted May 16, 2005.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  I. Koutsourelakis, E. Perraki, N. T. Economou, P. Dimitrokalli, E. Vagiakis, C. Roussos, and S. Zakynthinos Predictors of residual sleepiness in adequately treated obstructive sleep apnoea patients Eur. Respir. J., September 1, 2009; 34(3): 687 - 693. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. R. Bonsignore and J. Eckel Metabolic aspects of obstructive sleep apnoea syndrome Eur. Respir. Rev., June 1, 2009; 18(112): 113 - 124. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J-L. Pepin, V. Viot-Blanc, P. Escourrou, J-L. Racineux, M. Sapene, P. Levy, B. Dervaux, X. Lenne, and A. Mallart Prevalence of residual excessive sleepiness in CPAP-treated sleep apnoea patients: the French multicentre study Eur. Respir. J., May 1, 2009; 33(5): 1062 - 1067. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. R. Chasens, S. M. Sereika, and L. E. Burke Daytime Sleepiness and Functional Outcomes in Older Adults With Diabetes The Diabetes Educator, May 1, 2009; 35(3): 455 - 464. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Roche, J-M. Gaspoz, V. Pichot, M. Picard-Kossovsky, D. Maudoux, A. Garcin, S. Celle, E. Sforza, J. C. Barthelemy, and on behalf of the PROOF and SYNAPSE Study Groups Association between C-reactive protein and unrecognised sleep-disordered breathing in the elderly Eur. Respir. J., April 1, 2009; 33(4): 797 - 803. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Yukawa, Y. Inoue, H. Yagyu, T. Hasegawa, Y. Komada, K. Namba, N. Nagai, S. Nemoto, E. Sano, M. Shibusawa, et al. Gender Differences in the Clinical Characteristics Among Japanese Patients With Obstructive Sleep Apnea Syndrome Chest, February 1, 2009; 135(2): 337 - 343. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Gozal and L. Kheirandish-Gozal Obesity and Excessive Daytime Sleepiness in Prepubertal Children With Obstructive Sleep Apnea Pediatrics, January 1, 2009; 123(1): 13 - 18. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. S M Ip Obstructive sleep apnoea, insulin resistance and sleepiness Thorax, November 1, 2008; 63(11): 939 - 940. [Full Text] [PDF]
|
|
|
|
|
|
B. Buyse and the participants of working group 2 Treatment effects of sleep apnoea: where are we now? Eur. Respir. Rev., December 1, 2007; 16(106): 146 - 168. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. N. Vgontzas, S. Pejovic, E. Zoumakis, H.-M. Lin, C. M. Bentley, E. O. Bixler, A. Sarrigiannidis, M. Basta, and G. P. Chrousos Hypothalamic-Pituitary-Adrenal Axis Activity in Obese Men with and without Sleep Apnea: Effects of Continuous Positive Airway Pressure Therapy J. Clin. Endocrinol. Metab., November 1, 2007; 92(11): 4199 - 4207. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Montserrat, F. Garcia-Rio, and F. Barbe Diagnostic and Therapeutic Approach to Nonsleepy Apnea Am. J. Respir. Crit. Care Med., July 1, 2007; 176(1): 6 - 9. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Ursavas, M. Karadag, Y. O. Ilcol, B. Burgazlioglu, I. Ercan, and R. O. Gozu Relationship Between Serum Substance P Levels and Daytime Sleepiness in Obstructive Sleep Apnea Syndrome Chest, May 1, 2007; 131(5): 1400 - 1405. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. T. McNicholas, M. R. Bonsignore, and the Management Committee of EU COST ACTION B26 Sleep apnoea as an independent risk factor for cardiovascular disease: current evidence, basic mechanisms and research priorities Eur. Respir. J., January 1, 2007; 29(1): 156 - 178. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Ryan, C. T. Taylor, and W. T. McNicholas Predictors of Elevated Nuclear Factor-{kappa}B-dependent Genes in Obstructive Sleep Apnea Syndrome Am. J. Respir. Crit. Care Med., October 1, 2006; 174(7): 824 - 830. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Steptoe, V. Peacey, and J. Wardle Sleep duration and health in young adults. Arch Intern Med, September 18, 2006; 166(16): 1689 - 1692. [Abstract] [Full Text] [PDF]
|
|
Read all eLetters
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
