pii: sp-00174-14 http://dx.doi.org/10.5665/sleep.

4488

NATURAL HISTORY OF EXCESSIVE DAYTIME SLEEPINESS

Natural History of Excessive Daytime Sleepiness: Role of Obesity, Weight Loss,

Depression, and Sleep Propensity
Julio Fernandez-Mendoza, PhD1; Alexandros N. Vgontzas, MD1; Ilia Kritikou, MD1; Susan L. Calhoun, PhD1; Duanping Liao, MD, PhD2; Edward O. Bixler, PhD1
1
Sleep Research and Treatment Center, Department of Psychiatry, Pennsylvania State University College of Medicine, Hershey, PA; 2Department of
Public Health Sciences, Pennsylvania State University College of Medicine, Hershey, PA

Study Objectives: Excessive daytime sleepiness (EDS) is highly prevalent in the general population and is associated with occupational and
public safety hazards. However, no study has examined the clinical and polysomnographic (PSG) predictors of the natural history of EDS.
Design: Representative longitudinal study.
Setting: Sleep laboratory.
Participants: From a random, general population sample of 1,741 individuals of the Penn State Adult Cohort, 1,395 were followed up after 7.5
years.
Measurements and Results: Full medical evaluation and 1-night PSG at baseline and standardized telephone interview at follow-up. The
incidence of EDS was 8.2%, while its persistence and remission were 38% and 62%, respectively. Obesity and weight gain were associated with
the incidence and persistence of EDS, while weight loss was associated with its remission. Significant interactions between depression and PSG
parameters on incident EDS showed that, in depressed individuals, incident EDS was associated with sleep disturbances, while in non-depressed
individuals, incident EDS was associated with increased physiologic sleep propensity. Diabetes, allergy/asthma, anemia, and sleep complaints
also predicted the natural history of EDS.
Conclusions: Obesity, a disorder of epidemic proportions, is a major risk factor for the incidence and chronicity of excessive daytime sleepiness
(EDS), while weight loss is associated with its remission. Interestingly, objective sleep disturbances predict incident EDS in depressed individuals,
whereas physiologic sleep propensity predicts incident EDS in those without depression. Weight management and treatment of depression and
sleep disorders should be part of public health policies.
Keywords: daytime sleepiness, depression, incidence, obesity, persistence, polysomnography
Citation: Fernandez-Mendoza J, Vgontzas AN, Kritikou I, Calhoun SL, Liao D, Bixler EO. Natural history of excessive daytime sleepiness: role of
obesity, weight loss, depression, and sleep propensity. SLEEP 2015;38(3):351–360.

INTRODUCTION population sample using both subjective and PSG data. We hy-
Excessive daytime sleepiness (EDS) is a pervasive problem pothesized that psychiatric, cardiometabolic, and sleep disor-
that clinicians encounter very frequently in their practice. Popu- ders are the main predictors of the incidence and persistence of
lation-wise, EDS is highly prevalent, affecting up to 30% of the EDS, and that different mechanisms account for such associa-
general public, and is associated with significant personal and tions. Specifically, we hypothesized that (1) obesity and weight
occupational sequelae and public safety hazards.1–7 However, its gain are strong predictors of the incidence and persistence of
etiology and pathophysiology is still poorly understood and little EDS, whereas weight loss is associated with its remission; and
is known about how to treat EDS in specific patient populations. (2) depression and increased physiologic sleep propensity are
Cross-sectional studies have shown that EDS is associated independent risk factors for EDS, suggesting the presence of
with psychiatric, cardiometabolic, and sleep disorders, particu- two distinct pathophysiologic pathways leading to EDS.
larly depression, obesity, and sleep apnea.8–18 Despite its high
prevalence and significant medical correlates, the natural his- METHODS
tory of EDS in the general population remains to be documented,
and consequently, very little is known about the determinants Population
of the incidence, persistence, and remission of EDS.1,19,20 In fact, The data presented here were collected as part of the Penn
no study to date has explored the role of sleep, psychiatric, and State Adult Cohort, a population-based study of sleep dis-
medical disorders in the natural history of EDS using physi- orders that used a 2-phase protocol to randomly select adult
ologic sleep data, i.e., polysomnography (PSG). (age ≥ 20 years) men and women using standard sampling
The aim of this study was to longitudinally examine the pre- methods.21–23 A comprehensive presentation of the design and
dictors of the natural history of EDS in a large, random, general sampling methodology has been presented elsewhere.10,24–36 In
brief, phase I included 16,583 men and women (response rates
of 73.5% and 74.1%, respectively) and phase II included 1,741
Submitted for publication March, 2014 randomly selected men and women studied in the sleep labo-
Submitted in final revised form October, 2014 ratory (response rates of 67.8% and 65.8%, respectively). Of
Accepted for publication November, 2014 those, 1,395 were followed up after an average of 7.5 years with
Address correspondence to: Julio Fernandez-Mendoza, PhD, Sleep Re- a standardized telephone interview (response rates of 79.7%
search & Treatment Center, Department of Psychiatry, College of Medi- for the 1,741 and 90.9% for the 1,526 alive).26–30 After giving a
cine, Pennsylvania State University, 500 University Drive H073, Hershey, complete description of the study to the subjects, written and
PA 17033; Tel: (717) 531-6385; Fax: (717) 531-6491; Email: jfernandez- verbal informed consents were obtained at baseline and follow-
mendoza@hmc.psu.edu up, respectively. The whole study procedure was approved by
SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 351 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 Key Measurements
All subjects completed a comprehensive sleep history and
physical examination and were evaluated for one night in the
sleep laboratory (8 h fixed-time period) in sound-attenuated,
light- and temperature-controlled rooms using 16-channel PSG.
Sleep recordings were scored according to standard criteria
and evaluated for parameters of sleep continuity (sleep onset
latency, total sleep time, etc.) as well as sleep architecture (e.g.,
percentage of sleep stages such as REM).37 Increased physi-
ologic sleep propensity was defined as a sleep onset latency ≤ 8
minutes.38 Respiration was monitored throughout the night by
use of thermocouples at the nose and mouth and thoracic strain
gauges, and all-night recordings of hemoglobin oxygen satu-
ration (SpO2) were obtained with an oximeter attached to the
finger. Sleep apnea was defined as an apnea-hypopnea index
(AHI) ≥ 15.24,25 Body mass index was based on measured
height (cm) and weight (kg) during the subjects’ sleep labora-
tory visit. Obesity was defined as a BMI ≥ 30 kg/m 2. We also
ascertained the presence of sleep, physical and mental health
problems, and substance use using a standardized question-
naire. The presence of sleep difficulty was established on 3
levels of severity: insomnia, poor sleep, and normal sleep, as
defined elsewhere.26–36 We ascertained whether the respondent
was currently being treated for physical (ulcer, anemia, allergy/
asthma, etc.) and mental (e.g., depression, alcohol abuse, drug
abuse) health problems.26,28 Hypertension and diabetes were
also defined based on blood pressure measured in the evening
and fasting blood glucose measured in the morning during the
sleep laboratory visit, respectively.32,33 Finally, we ascertained
participants’ daily consumption of caffeine, tobacco, and
alcohol.26–28
Follow-up measures taken through telephone interview in-
cluded the standardized questionnaire that subjects completed
at baseline during their sleep laboratory visit,26–29 which in-
cluded self-reported height and weight data.30 Sleep-related
Figure 1—Participants’ flow in the study.
questions were used to establish the presence of EDS at follow-
up as mentioned above.
the university’s institutional review board. Figure 1 shows the
participant flow in the study. Statistical Analyses
The design of this study included oversampling of those at
Definition of Incident, Persistent, and Remitted EDS higher risk for sleep apnea and women with markedly higher
Commensurate with our previous cross-sectional study,10 levels of BMI to increase the precision of the risk estimates.
the presence of EDS was established based on a moderate-to- A sampling weight was developed so that the estimates could
severe self-report of daytime drowsiness or sleepiness occur- be inferred to the general population, including the use of the
ring most of the day (“Do you feel drowsy or sleepy most of NHANES III laboratory data as the standard to be representa-
the day but manage to stay awake?”) and/or irresistible sleep tive of the national population, as reported elsewhere.10,21,24,55
attacks (“Do you have any irresistible sleep attacks during Thus, all analyses were adjusted for the sampling weight, in-
the day?”). Each question was answered on a 4-point Likert cluding the incidence rate (proportion of cases developing EDS
scale (0 = none, 1 = mild, 2 = moderate, and 3 = severe), and at follow-up among those without EDS at baseline) and persis-
the presence of EDS was defined as a moderate or severe re- tence or remission rates (proportion of cases with or without
port to either of the 2 questions. Of the 1,395 subjects who EDS at follow-up among those with EDS at baseline, respec-
were followed-up, 222 reported EDS at baseline and 1,173 did tively). Descriptive data (means, standard deviations, and
not. Subjects were classified according to their baseline and proportions) of all variables were calculated for the entire pop-
follow-up status into: no EDS (i.e., individuals without EDS ulation, as well as stratified by EDS status. Bivariate logistic
at baseline and at follow-up; n = 1,035), incident EDS (i.e., in- regression was used to assess the odds ratios (OR) associated
dividuals without EDS at baseline and with EDS at follow-up; with each baseline risk factor with the incidence or persistence
n = 138), remitted EDS (i.e., individuals with EDS at baseline of EDS. Multivariable logistic regression models were used to
and without EDS at follow-up; n = 147), and persistent EDS assess the independent association and relative significance of
(i.e., individuals with EDS at baseline and at follow-up; n = 75). each baseline risk factor with the incidence or persistence of
SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 352 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 Table 1—Sociodemographic and behavioral characteristics at baseline.
All No EDS Incident EDS Remitted EDS Persistent EDS
(n = 1,395) (n = 1,035) (n = 138) (n = 147) (n = 75)
Gender
Female, % 49.4 49.7 33.3 55.4 71.4 †
Male, % 50.6 50.3 66.7 ** 45.6 28.6
Race
Caucasian, % 93.2 94.5 86.7 88.2 81.0
Non-Caucasian, % 6.8 5.5 13.3 ** 11.8 19.0
Age, years 49.2 (13.1) 49.5 (12.9) 49.5 (15.6) 46.2 (13.0) 44.7 (9.7)
≤ 30, % 7.8 6.7 16.3 * 13.2 7.1
31–49, % 47.0 47.2 36.5 47.1 66.7
50–64, % 31.4 32.3 24.0 32.4 23.8
≥ 65, % 13.9 13.8 23.1 * 7.4 2.4
Caffeine, cups/day 2.3 (2.9) 2.3 (2.7) 2.1 (2.5) 2.7 (3.8) 2.7 (4.6)
None, % 33.6 33.2 38.1 30.4 38.1
1–2 cups/day, % 31.2 31.1 29.5 33.3 33.3
≥ 3 cups/day, % 35.2 35.6 32.4 36.2 28.6
Tobacco, cigarettes/day 4.0 (10.4) 3.8 (10.4) 4.0 (9.3) 5.5 (11.6) 6.2 (10.3)
None, % 78.8 79.7 76.9 73.5 66.7
1–20 cigarettes/day, % 9.8 10.0 9.6 7.4 9.5
≥ 20 cigarettes/day, % 11.4 10.3 13.5 19.1 23.8
Alcohol, drinks/day 1.2 (5.9) 1.3 (6.3) 1.3 (3.3) 0.7 (2.2) 0.5 (1.2)
None, % 73.2 72.1 76.9 80.9 81.0
1 drink/day, % 10.7 11.3 6.7 7.4 7.1
≥ 2 drinks/day, % 16.2 16.6 16.3 11.8 11.9

Data are mean (standard deviation), unless otherwise stated. All data are adjusted for sampling weight, except absolute number of cases across groups.
†
P < 0.10, * P < 0.05, ** P < 0.01.

EDS. We calculated the adjusted odds ratios (AOR) and their Incident EDS
95% confidence intervals (95% CI) from the multivariable The incidence of EDS was 8.2%. Male gender (OR = 2.0,
regression models to estimate the odds associated with each 95% CI = 1.3–3.0, P < 0.01), non-Caucasian race (OR = 2.6,
risk factor. Given the known clustering of clinical risk factors, 95% CI = 1.4–4.8, P < 0.01), and young and older age were as-
particularly obesity, sleep apnea, diabetes, and hypertension, a sociated with incident EDS (see Table 1). Figure 2 depicts the
stepwise procedure examined which of these were the stron- U-shaped relationship between age and the incidence of EDS.
gest predictors of incident EDS, using a backward conditional Consistently, age was modeled using fine-grained, clinically
model followed by forced entry of sociodemographic factors meaningful categories in multivariable analyses, i.e., young
(see supplemental material). Furthermore, based on previous adults (≤ 30 y), middle-age (31–64 y, which served as the refer-
clinical studies,8,9 we further tested the interaction between de- ence group), and older adults (≥ 65 y).
pression and PSG parameters on incident EDS in multivariable As shown in Table 2, depression (OR = 2.8, 95% CI = 1.8–
logistic regression models adjusting for potential confounding 4.5, P = 0.00002), sleep apnea (OR = 2.2, 95% CI = 1.1–4.4,
factors associated with depression, PSG parameters, or inci- P = 0.031), obesity (OR = 2.1, 95% CI = 1.4–3.1, P = 0.001),
dent EDS (i.e., gender, race, age, obesity, hypertension, dia- and diabetes (OR = 2.0, 95% CI = 1.2–3.3, P = 0.008) were
betes, allergy/asthma, sleep difficulties, and sleep apnea). The significantly associated with incident EDS, whereas hyper-
no EDS group served as the reference for incident EDS, while tension (OR = 1.5, 95% CI = 1.0–2.2, P = 0.063), allergy/
the remitted EDS group served as the reference for persistent asthma (OR = 1.5, 95% CI = 1.0–2.3, P = 0.085), and mi-
EDS. Finally, given the association of obesity with EDS,12–15 graine (OR = 1.7, 95% CI = 1.0–2.9, P = 0.073) were margin-
we examined the role of weight gain and weight loss with the ally associated. Other medical conditions (e.g., stroke, cancer,
natural history of EDS. All analyses were conducted with colitis, epilepsy, kidney/bladder, thyroidism, arthritis, and
SPSS version 21.0 for Windows. alcohol or drug abuse) were not associated with incident
EDS. A multivariable backward regression model showed
RESULTS that depression (AOR = 2.6, 95% CI = 1.6–4.2, P = 0.0002),
The demographic and behavioral characteristics of the obesity (AOR = 1.8, 95% CI = 1.2–2.8, P = 0.011), diabetes
overall study sample and stratified by the natural history of (AOR = 1.7, 95% CI = 1.0–3.0, P = 0.046), and allergy/asthma
EDS are presented in Table 1. (AOR = 1.5, 95% CI = 1.0–2.4, P = 0.057) were the strongest
SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 353 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 CI = 0.1–8.1, P = 0.993). Please see also Figure S1 (supple-

STXDGUDWLF  mental material).

Persistent EDS
 The persistence and remission rates of EDS were 38%
and 62%, respectively. Ulcer (OR = 3.6, 95% CI = 1.1–11.1,
 P = 0.030), anemia (OR = 2.9, 95% CI = 1.1–7.8, P = 0.040),
('6LQFLGHQFH

and sleep difficulties (OR = 2.0, 95% CI = 1.1–3.6, P = 0.020)


were significantly associated with persistent EDS, whereas
 depression (OR = 2.1, 95% CI = 1.0–4.6, P = 0.063), hyperten-
sion (OR = 2.1, 95% CI = 1.0–4.6, P = 0.064), allergy/asthma
 (OR = 2.0, 95% CI = 0.9–4.5, P = 0.096), female gender
(OR = 2.0, 95% CI = 0.9–4.6, P = 0.095), and BMI (2 kg/m 2

increase OR = 1.1, 95% CI = 1.0–1.2, P = 0.125) were only

marginally associated with persistent EDS. PSG parameters
     • did not differ between remitted and persistent EDS, while
reporting short or long sleep duration were associated with
persistent EDS (see Table 3). A multivariable backward regres-
Figure 2—Age-specific incidence of EDS. sion model showed that anemia (AOR = 3.4, 95% CI = 1.1–10.0,
P = 0.031), sleep difficulties (AOR = 3.0, 95% CI = 1.1–8.1,
P = 0.029), and BMI (2 kg/m 2 AOR = 1.1, 95% CI = 1.0–1.3,
clinical predictors of incident EDS (see Table S1, supple- P = 0.088) were the strongest predictors of persistent EDS,
mental material). whereas ulcer, depression, allergy/asthma, and hypertension
Table 3 shows that several subjective and objective sleep pa- did not enter the model when sleep difficulties were included
rameters were associated with the incidence of EDS. Reporting (see Table S2, supplemental material).
short (≤ 5 h) or long (> 8 h) sleep duration were significantly
associated with incident EDS (OR = 2.7, 95% CI = 1.4–5.0, Weight Gain and Weight Loss
P = 0.002 and OR = 2.2, 95% CI = 1.0–4.8, P = 0.044, respec- Given the association of baseline obesity with EDS, we fur-
tively), while reporting sleep difficulties was not significantly ther explored whether weight gain and weight loss, as mea-
associated with incident EDS (OR = 1.3, 95% CI = 0.8–2.1 for sured by ΔBMI, were associated with the natural history of
poor sleep and OR = 1.5, 95% CI = 0.7–3.2 for insomnia, P- EDS. As shown in Table 4, individuals with incident or per-
linear = 0.166). Snoring was significantly associated with in- sistent EDS gained significantly more weight as compared to
cident EDS, particularly in those with milder levels of sleep those without EDS (ΔBMI 2.5 ± 3.5 and 2.8 ± 2.8 vs. 1.3 ± 3.2
apnea (OR = 1.9, 95% CI = 1.2–2.9, P = 0.004). Furthermore, kg/m2, respectively), even after adjusting for baseline obe-
PSG markers of sleep fragmentation, such as increased wake sity (ΔBMI 2.4 ± 3.6, P = 0.001 and 2.7 ± 4.1, P = 0.006 vs.
(P = 0.008) and stage 1 (P = 0.005) and decreased stage 2 1.4 ± 2.9 kg/m2, respectively). Interestingly, individuals with
(P = 0.037) and slow wave sleep (P = 0.083), as well as PSG remitted EDS gained significantly less weight as compared to
markers of increased physiologic sleep propensity (i.e., sleep individuals without EDS (ΔBMI 0.4 ± 2.7 vs. 1.3 ± 3.2 kg/m2,
onset latency ≤ 8 min, OR = 2.7, 95% CI = 1.6–4.7, P = 0.0002) P = 0.027), which indicated that an important proportion of
were associated with incident EDS. remitted EDS cases lost weight over time (see Table 4). In fact,
Importantly, we found significant interactions between de- weight loss (i.e., ΔBMI ≤ −2.0 kg/m 2) was associated with re-
pression and PSG parameters of sleep onset latency (P = 0.001), mitted EDS (OR = 2.5, 95% CI = 1.2–5.1, P = 0.013), whereas
total sleep time (P = 0.004), sleep efficiency (P = 0.005), total weight gain (i.e., ΔBMI ≥ + 2.0 kg/m 2) was associated with
wake time (P = 0.007), percent of stage 1 (P = 0.010), and incident (OR = 2.8, 95% CI = 1.8–4.3, P = 0.00006) and persis-
percent of REM sleep (P = 0.002) on incident EDS, even after tent EDS (OR = 2.7, 95% CI = 1.4–5.2, P = 0.003). These find-
adjusting for gender, race, age, obesity, hypertension, dia- ings were similar when other indices of weight gain/loss (i.e.,
betes, allergy/asthma, sleep difficulties, and sleep apnea. As absolute or percent change in body weight) were examined (see
shown in Figure 3, the association of PSG parameters with Table 4) and when analyses were further adjusted for baseline
incident EDS differed in depressed and non-depressed indi- sleep duration (ΔBMI = 1.4, 2.4, 2.9, and 0.5 kg/m2 for none,
viduals. Longer sleep latency, higher wake time and percent incident, persistent, and remitted EDS, respectively).
of stage 1, and lower sleep efficiency and percent of REM
sleep were associated with incident EDS in depressed indi- DISCUSSION
viduals. In contrast, longer sleep duration, higher sleep ef- The main findings of this population-based, longitudinal
ficiency, shorter sleep latency, and lower wake time, were study are that (1) obesity, weight gain, and weight loss are as-
associated with incident EDS in non-depressed individuals. sociated with the incidence, persistence, and remission of EDS,
In fact, increased physiologic sleep propensity (sleep onset la- respectively; and (2) objective sleep disturbance predicts the
tency ≤ 8 min) was significantly associated with incident EDS incidence of EDS in individuals with depression, while physi-
in those without depression (AOR = 3.1, 95% CI = 1.8–5.6, ologic sleep propensity predicts the incidence of EDS in in-
P = 0.0001) but not in those with depression (AOR = 0.9, 95% dividuals without depression. These findings underscore that
SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 354 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 Table 2—Clinical characteristics at baseline.
All No EDS Incident EDS Remitted EDS Persistent EDS
(n = 1,395) (n = 1,035) (n = 138) (n = 147) (n = 75)
Physical Health Problems
No, % 35.9 38.5 27.9 22.1 7.1
Yes, % 64.1 61.5 72.1 * 77.9 92.9 *
Allergy/asthma
No, % 72.9 74.4 66.7 70.3 53.7
Yes, % 27.1 25.6 33.3 † 29.7 46.3 †
Anemia
No, % 91.5 92.3 92.4 88.2 72.1
Yes, % 8.5 7.7 7.6 11.8 27.9 *
Diabetes
No, % 86.2 88.1 79.0 72.1 73.8
Yes, % 13.8 11.9 21.0 ** 27.9 26.2
Hypertension
No, % 64.4 66.2 57.1 58.8 40.5
Yes, % 35.6 33.8 42.9 * 41.2 59.5 †
Migraine
No, % 87.2 89.2 83.7 73.5 61.9
Yes, % 12.8 10.8 16.3 † 26.5 38.1
Obesity
No, % 74.7 77.8 62.9 54.4 50.0
Yes, % 25.3 22.2 37.1 ** 45.6 50.0
Sleep apnea
No, % 94.8 95.3 90.4 94.2 92.9
Yes, % 5.2 4.7 9.6 * 5.8 7.1
Ulcer
No, % 95.0 96.0 93.3 92.6 76.2
Yes, % 5.0 4.0 6.7 7.4 23.8 *
Mental Health Problems
No, % 80.6 84.6 70.5 54.4 38.1
Yes, % 19.4 15.4 29.5 ** 45.6 61.9 †
Depression
No, % 84.6 88.7 73.3 58.8 40.5
Yes, % 15.4 11.3 26.7 ** 41.2 59.5 †

All data are adjusted for sampling weight, except absolute number of cases across groups. Diabetes = fasting blood sugar ≥ 126 mg/dL or being treated for
diabetes. Hypertension = systolic ≥ 140 mm Hg and/or diastolic ≥ 90 mm Hg blood pressure or use of antihypertensive medication. Obesity = body mass
index ≥ 30 kg/m2. Sleep apnea = apnea-hypopnea index ≥ 15 events/hour of sleep. Depression = lifetime history of treatment for depression, including a
history of suicidal thoughts or attempts. † P < 0.10, * P < 0.05, ** P < 0.01.

obesity, depression, and sleep disorders should be our priority Role of Obesity, Weight Gain, and Weight Loss
in terms of public health policies and early clinical interven- In line with previous cross-sectional studies,10,12–15 obesity
tions for EDS. and diabetes were independent risk factors for EDS. An im-
portant and novel finding in our study was that weight gain
Incidence, Persistence, and Remission of EDS was a strong predictor of the incidence and persistence of EDS
As stated above, up to 30% of the general population com- even in non-obese individuals. A recent longitudinal study
plains of EDS, depending on the definition used.1,2 For example, showed that EDS was not a significant predictor of the inci-
8% to 10% of the general population has moderate-to-severe dence of obesity.30 Together, these studies indicate that EDS is
complaints of EDS.1,10 Using this latter definition, we found that likely to be a consequence of metabolic aberrations associated
the incidence of EDS is about 8% and that it is very likely to remit with obesity rather than a premorbid characteristic of obese
(62%), with only about 38% persisting with the complaint after subjects. From a mechanistic standpoint, these data further
a long follow-up. This scenario is similar to that of “poor sleep,” suggest that metabolic aberrations and chronic low-grade in-
which is more likely to remit, while “insomnia” is a highly per- flammation, conditions common in obesity and diabetes, may
sistent disorder.27,28 Thus, this study further suggests that EDS is play a role in the pathogenesis of EDS and fatigue13–15,39 in a
a symptom rather than a disorder per se (i.e., hypersomnia).2,10,11 large proportion of individuals from the general population.
SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 355 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 Table 3—Subjective and objective sleep characteristics at baseline.
All No EDS Incident EDS Remitted EDS Persistent EDS
(n = 1,395) (n = 1,035) (n = 138) (n = 147) (n = 75)
Subjective Sleep Difficulty
Normal sleep, % 72.0 76.2 70.5 35.3 16.7
Poor sleep, % 20.7 17.8 21.0 50.0 54.8
Insomnia, % 7.3 6.0 8.6 14.7 28.6 *
Subjective Sleep Duration, h 6.8 (1.3) 6.8 (1.2) 6.6 (1.7) 6.5 (1.5) 6.6 (1.8)
> 8 hours, % 6.1 5.5 10.1 * 4.8 19.4 *
7–8 hours, % 53.0 55.1 44.9 44.4 22.6
5–7 hours, % 30.9 31.1 27.0 34.9 29.0
≤ 5 hours, % 9.9 8.3 18.0 * 15.9 29.0 *
Sleep Onset Latency, min 33.1 (32.6) 32.7 (30.8) 32.5 (33.0) 32.0 (36.2) 39.3 (48.2)
≥ 30 minutes, % 41.1 41.8 39.0 33.3 38.1
8–30 minutes, % 45.1 45.6 35.2 49.3 50.0
≤ 8 minutes, % 13.8 12.6 25.7 ** 17.4 11.9
WASO, min 91.7 (56.7) 92.5 (56.5) 83.3 (55.9) 87.4 (53.3) 97.0 (58.6)
Awakenings, # 11.3 (7.0) 11.3 (6.9) 13.2 (8.1) ** 10.4 (7.2) 8.7 (4.8)
Total Wake Time, min 124.9 (71.6) 125.2 (70.8) 116.2 (72.8) 119.4 (69.7) 136.2 (72.4)
Total Sleep Time, h 5.9 (1.2) 5.8 (1.2) 6.1 (1.2) † 5.9 (1.3) 5.6 (1.4)
7–8 hours, % 16.5 15.7 23.1 22.1 11.9
6–7 hours, % 36.9 37.1 40.4 26.5 38.1
5–6 hours, % 23.4 24.2 11.5 27.9 21.4
≤ 5 hours, % 23.3 23.0 25.0 23.5 28.6
Sleep Efficiency, % 73.8 (14.9) 73.7 (14.8) 75.8 (15.0) 74.8 (14.6) 70.9 (16.3)
Stage 1, % 11.3 (9.6) 11.1 (9.3) 13.8 (12.3) ** 10.1 (8.6) 12.7 (10.8)
Stage 2, % 69.1 (9.3) 69.4 (9.3) 67.4 (9.1) * 68.4 (9.1) 67.3 (11.9)
SWS , % 3.0 (5.2) 2.9 (5.1) 2.0 (3.9) † 4.7 (7.7) 3.6 (4.7)
REM, % 16.6 (7.0) 16.6 (6.9) 16.7 (7.7) 16.9 (5.3) 16.4 (9.6)
Apnea-Hypopnea Index 2.5 (7.5) 2.3 (7.1) 3.6 (9.2) 1.5 (6.2) 4.7 (12.2)
Minimum SpO2 92.6 (6.2) 92.7 (6.2) 92.4 (5.7) 93.1 (5.1) 92.2 (7.8)
Snoring
No, % 70.7 74.4 61.5 35.3 47.6
Yes, % 29.3 25.6 38.5 ** 64.7 52.4

Data are mean (standard deviation), except otherwise stated. All data are adjusted for sampling weight, except absolute number of cases across groups.
REM, rapid eye movement sleep; SWS, slow wave sleep; WASO, wake time after sleep onset. † P < 0.10, * P < 0.05, ** P < 0.01.

Moreover, weight loss was a significant predictor of the remis- A novel and important finding of this longitudinal study was
sion of EDS in the present study. These data not only reinforce that PSG characteristics were differentially associated with the
the etiological link between obesity/weight gain and the devel- incidence of EDS in individuals with and without depression.
opment and chronicity of EDS but also suggest that focusing Specifically, we found that in individuals with depression, inci-
on modifiable behavioral and metabolic factors associated dent EDS was associated with PSG sleep disturbances, such as
with weight gain and weight loss will lead to higher success in increased sleep onset latency. These longitudinal, population-
reducing the prevalence and severity of EDS. It is important to based findings are consistent with previous clinical studies
keep in mind that weight gain or loss is likely to be associated showing that the complaint of EDS in psychiatric disorders is
with other factors (e.g., changes in diet and physical activity) more likely associated with a state of physiologic hyperarousal
that may be the actual cause of incident or remitted EDS. For rather than with physiologic sleep propensity (i.e., hypoarousal
example, increased physical activity may lead to both weight of central origin).8,9,11
loss and reduced EDS or weight change may (partially or fully)
mediate the association of physical activity with EDS. Future Role of Physiologic Sleep Propensity
experimental and clinical studies should examine the under- In contrast, the incidence of EDS in individuals without de-
lying mechanisms in energy balance leading to weight loss and pression was associated with increased physiologic sleep pro-
EDS remission. pensity. Previous studies have shown that increased daytime
sleep propensity is associated with shorter nighttime sleep
Role of Depression and Objective Sleep Disturbances onset latency and longer sleep duration5,40 as well as with sig-
We found a strong association between depression and EDS, nificant genetic influences.41–45 Our data suggest the existence
which is consistent with previous cross-sectional studies.8–12,18–20 of a phenotype of individuals without psychiatric disorders who
SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 356 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 1RW'HSUHVVHG 'HSUHVVHG 1RW'HSUHVVHG 'HSUHVVHG
 

 

6OHHS/DWHQF\PLQXWHV
6OHHS(IILFLHQF\

 

 

 

 

 
1R('6 ,QFLGHQW('6 1R('6 ,QFLGHQW('6

1RW'HSUHVVHG 'HSUHVVHG 1RW'HSUHVVHG 'HSUHVVHG

 


7RWDO6OHHS7LPHPLQXWHV

7RWDO:DNH7LPHPLQXWHV 



 






 
1R('6 ,QFLGHQW('6 1R('6 ,QFLGHQW('6

1RW'HSUHVVHG 'HSUHVVHG 1RW'HSUHVVHG 'HSUHVVHG

 





6WDJH5(0

6WDJH


 






 
1R('6 ,QFLGHQW('6 1R('6 ,QFLGHQW('6

Figure 3—Graphic representation of the interaction between depression and polysomnographic characteristics on incident EDS. Depressed individuals
with incident EDS showed sleep disturbances at baseline (longer sleep onset latency, shorter sleep duration, lower sleep efficiency, more time awake and
in stage 1, and less time in REM sleep), while non-depressed individuals with incident EDS showed increased physiologic sleep propensity (shorter sleep
latency, longer sleep duration, higher sleep efficiency, and less time awake). Data adjusted for gender, race, age, obesity, diabetes, hypertension, allergy/
asthma, sleep difficulty, and sleep apnea.

SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 357 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 Table 4—Association of weight gain and weight loss with the natural history of EDS.
No EDS Incident EDS Remitted EDS Persistent EDS
(n = 1,035) (n = 138) (n = 147) (n = 75)
Weight at baseline, kg 76.6 (17.1) 81.3 (15.9) ** 82.4 (12.3) 85.1 (15.8)
Weight at follow-up, kg 80.0 (19.3) 87.6 (18.1) ** 83.2 (12.7) 92.0 (16.6) *
Δweight, kg 3.4 (9.1) 6.3 (10.8) ** 0.8 (7.1) 6.9 (8.0) **
≤ −5 kg weight loss, % 8.7 10.7 17.9 7.1
No change, % 60.6 37.9 58.2 52.4
≥ +5 kg weight gain, % 30.7 51.5 23.9 40.5
Δweight, % 4.7 (11.3) 8.6 (13.5) ** 1.6 (7.6) 9.1 (10.3) **
≤ −1% weight loss, % 26.4 25.0 29.4 11.9
No change, % 49.1 26.0 51.5 54.8
≥ +10% weight gain, % 24.4 49.0 19.1 33.3
BMI at baseline, kg/m2 26.0 (4.8) 27.4 (4.3) ** 28.5 (3.8) 30.3 (5.6)
BMI at follow-up, kg/m2 27.3 (5.5) 29.9 (5.2) ** 28.9 (4.1) 33.1 (6.1) **
ΔBMI, kg/m2 1.3 (3.2) 2.5 (3.5) ** 0.4 (2.7) 2.8 (2.8) **
≤ −2 kg/m2 weight loss, % 6.7 11.5 16.7 7.1
No change, % 58.6 33.7 57.6 35.7
≥ +2 kg/m2 weight gain, % 34.8 54.8 25.8 57.1

Data are mean (standard deviation), except otherwise stated. All data are adjusted for sampling weight, except absolute number of cases across groups.
BMI, body mass index. * P < 0.05, ** P < 0.01.

have increased physiologic sleep propensity and significant risk multivariable model after adjusting for obesity. These data re-
of developing a disorder of EDS (e.g., hypersomnia of central inforce the hypothesis that metabolic and inflammatory aber-
origin). It is not clear, however, why these individuals with rations may play a major role in the pathogenesis of EDS, even
physiologic sleep propensity develop the complaint of EDS later in individuals with sleep apnea.39 Furthermore, allergy/asthma,
on in life. Several putative mechanisms include (1) impaired anemia, ulcer, and hypertension were associated with the inci-
homeostatic sleep/arousal mechanisms with increasing age, (2) dence or persistence of EDS. The association of allergy/asthma
fatigue-mediating factors (e.g., inflammatory) that increase with incident EDS was weaker than that of metabolic and mood
with age, and/or (3) increased personal, occupational, and so- disorders, while the link between ulcer, allergy/asthma, depres-
cial demands that do not allow extended nighttime sleep or brief sion, and hypertension with the persistence of EDS was the
daytime naps to alleviate physiologic sleep pressure. presence of complaints of sleep difficulties, which may be the
It is important to note that many patients with hypersomnia final common path for the persistence of the complaint of EDS
of central origin present with depressive symptoms or even in individuals with those conditions. It remains to be tested
clinical diagnosis of depression.11 In this instance, the differ- whether treatment of underlying sleep difficulties improves
ential diagnosis of hypersomnia of central origin vs. hyper- both the medical disorder and associated EDS or whether treat-
somnia related to depression becomes a challenge in the clinic ment of the underlying medical disorder improves sleep and
if it is only based on clinical history and subjective reports. associated EDS. These data reinforce the inclusion of blood,
Our longitudinal data confirm and expand our previous find- respiratory, pain, and sleep disorders, including unmet sleep
ings9 that objective measures of EDS may be necessary to con- needs, as part of the thorough assessment of patients with EDS.
firm or rule-out the presence of increased physiologic sleep
propensity, particularly in depressed individuals. Limitations
Some limitations should be taken into account when in-
Role of Aging, Sleep, and Other Medical Conditions terpreting our results. The objective sleep in this study was
The incidence of EDS was higher in the young and in the based on one night of PSG, which may not be representative
old; the former most likely as a result of increased unmet sleep of the subjects’ typical or optimal sleep and may be affected
needs and the latter associated with increased health problems by the “first night effect.” Nevertheless, large epidemiological
and medical illness,1,2,10 which was evidenced by the associa- studies show high consistency in sleep parameters (e.g., av-
tion of reported insufficient and long sleep with incident EDS. erage sleep duration of about 6 h) independent of whether sleep
Snoring, particularly in those with milder levels of sleep apnea, is recorded at home with PSG,46 for three consecutive nights
was significantly associated with incident EDS, which indicates with actigraphy,47 or in the sleep laboratory with PSG.32–36 The
that these individuals are likely to progress into more severe consistency among these three large epidemiological studies
forms of sleep apnea, most likely through further weight gain, supports the validity of our findings. In the present study, we
and develop clinical symptoms of the disorder, i.e., EDS. Inter- did not ascertain EDS using standardized questionnaires or
estingly, moderate-to-severe sleep apnea at baseline was sig- physiologic measures of daytime sleep propensity, i.e., mul-
nificantly associated with EDS; however, it did not enter the tiple sleep latency test. Despite the limitations of using two
SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 358 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 items to ascertain the complaint of EDS, these items specifi- 2. Ohayon MM. Determining the level of sleepiness in the American
cally referred to two core symptoms used in the diagnosis of population and its correlates. J Psychiatr Res 2012;46:422–7.
3. Foley D, Monjan A, Masaki K, et al. Daytime sleepiness is associated
hypersomnia (sleepiness or drowsiness and irresistible sleep with 3-year incident dementia and cognitive decline in older Japanese-
attacks) and capture the complaints typically expressed by pa- American men. J Am Geriatr Soc 2001;49:1628–32.
tients.38 Importantly, the prevalence estimate in the Penn State 4. Jaussent I, Bouyer J, Ancelin ML, et al. Excessive sleepiness is
Cohort using this definition of EDS10 is consistent with that of predictive of cognitive decline in the elderly. Sleep 2012;35:1201–07.
5. Drake C, Roehrs T, Breslau N, et al. The 10-year risk of verified motor
other population-based studies, where the complaint of EDS is vehicle crashes in relation to physiologic sleepiness. Sleep 2010;33:745–
defined based on severity criteria.1,2 Nevertheless, the potential 52.
effect of measurement characteristics, measurement error, and 6. Ward KL, Hillman DR, James A, et al. Excessive daytime sleepiness
regression to the mean in our natural history estimates cannot increases the risk of motor vehicle crash in obstructive sleep apnea. J
Clin Sleep Med 2013;9:1013–21.
be entirely ruled out. The study had one long follow-up time 7. Stevenson MR, Elkington J, Sharwood L, et al. The role of sleepiness,
point, which precludes an assessment of a potential waxing sleep disorders, and the work environment on heavy-vehicle crashes in
and waning pattern and short-term, stronger associations with 2 Australian States. Am J Epidemiol 2014;179:594–601.
baseline predictors. Similarly, the lack of measures of habitual 8. Nofzinger EA, Thase ME, Reynolds CF 3rd, et al. Hypersomnia in
bipolar depression: a comparison with narcolepsy using the multiple
sleep duration at follow-up precludes examining the trajectory sleep latency test. Am J Psychiatry 1991;148:1177–81.
of sleep duration as it relates to the natural history of EDS, 9. Vgontzas AN, Bixler EO, Kales A, Criley C, Vela-Bueno A. Differences
particularly its incidence and persistence. Another potential in nocturnal and daytime sleep between primary and psychiatric
limitation is that we had to rely on self-reported data for our hypersomnia: diagnostic and treatment implications. Psychosom Med
2000;62:220–6.
weight gain/loss analyses. However, we have previously shown 10. Bixler EO, Vgontzas AN, Lin HM, Calhoun SL, Vela-Bueno A, Kales
that objectively measured and self-reported weight data highly A. Excessive daytime sleepiness in a general population sample: the role
correlated in our cohort,30 which provides further validity to of sleep apnea, age, obesity, diabetes, and depression. J Clin Endocrinol
the present findings. Metab 2005;90:4510–15.
11. Dauvilliers Y, Lopez R, Ohayon M, Bayard S. Hypersomnia and
depressive symptoms: methodological and clinical aspects. BMC Med
Implications for Clinical Practice and Public Health Policy 2013;11:78.
In summary, our data indicate that obesity and weight gain 12. Vashum KP, McEvoy MA, Hancock SJ, et al. Prevalence of and
play a key role in the development and chronicity of EDS, associations with excessive daytime sleepiness in an Australian older
population. Asia Pac J Public Health 2013 Oct 4. [Epub ahead of print].
whereas weight loss is associated with its remission. Further- 13. Vgontzas AN, Papanicolaou DA, Bixler EO, Kales A, Tyson K, Chrousos
more, EDS in individuals with depression is associated with GP. Elevation of plasma cytokines in disorders of excessive daytime
objective sleep disturbances, while increased physiologic sleepiness: role of sleep disturbance and obesity. J Clin Endocrinol
sleep propensity is an independent predictor of EDS. These Metab 1997;82:1313–6.
14. Vgontzas AN, Bixler EO, Chrousos GP. Obesity-related sleepiness and
data suggest potentially different etiologic mechanisms for fatigue: the role of the stress system and cytokines. Ann N Y Acad Sci
the complaint of EDS in the presence of psychiatric disorders, 2006;1083:329–44.
i.e., physiologic hyperarousal vs. central nervous system hy- 15. Panossian LA, Veasey SC. Daytime sleepiness in obesity: mechanisms
poarousal. Future clinical studies should test whether (1) the beyond obstructive sleep apnea--a review. Sleep 2012;35:605–15.
16. Gottlieb DJ, Whitney CW, Bonekat WH, et al. Relation of sleepiness
complaint of EDS in depression may respond better to treat- to respiratory disturbance index: the Sleep Heart Health Study. Am J
ments targeting mood and nighttime sleep disturbances rather Respir Crit Care Med 1999;159:502–7.
than pharmacological agents increasing physiologic alertness, 17. Morrell MJ, Finn L, McMillan A, Peppard PE. The impact of ageing
and (2) whether extending sleep (e.g., regular sleep schedules, and sex on the association between sleepiness and sleep disordered
breathing. Eur Respir J 2012;40:386–93.
brief daytime naps) and/or enhancing alertness through phar- 18. Basta M, Lin HM, Pejovic S, Sarrigiannidis A, Bixler E, Vgontzas AN.
macological agents (stimulants) prevent EDS in individuals Lack of regular exercise, depression, and degree of apnea are predictors
with physiologic sleep propensity. From a public health policy of excessive daytime sleepiness in patients with sleep apnea: sex
perspective, this study suggests that obesity, depression, and differences. J Clin Sleep Med 2008;4:19–25.
19. Hasler G, Buysse DJ, Gamma A, et al. Excessive daytime sleepiness in
sleep disorders should be a priority in the prevention and re- young adults: a 20-year prospective community study. J Clin Psychiatry
duction of EDS in the general population. 2005;66:521–9.
20. Quan SF, Katz R, Olson J, et al. Factors associated with incidence and
ACKNOWLEDGMENTS persistence of symptoms of disturbed sleep in an elderly cohort: the
Cardiovascular Health Study. Am J Med Sci 2005;329:163–72.
The work was performed at the Sleep Research & Treatment 21. ten Have TR, Bixler EO. Modelling population heterogeneity in
Center, Penn State Milton S. Hershey Medical Center, and the sensitivity and specificity of a multi-stage screen for obstructive sleep
staff is especially commended for their efforts. apnoea. Stat Med 1997;16:1995–2008.
22. Waksberg J. Sampling methods for random digit dialing. J Am Stat
Assoc 1978;73:40–6.
DISCLOSURE STATEMENT 23. Kish L. Survey sampling. New York: John Wiley & Sons, Inc, 1965.
This was not an industry supported study. This research is 24. Bixler EO, Vgontzas AN, Ten Have T, Tyson K, Kales A. Effects of age
funded in part by the National Institutes of Health grants RO1 on sleep apnea in men: I. Prevalence and severity. Am J Respir Crit Care
HL51931, RO1 HL40916, and RO1 HL64415. The authors have Med 1998;157:144–8.
25. Bixler EO, Vgontzas AN, Lin HM, et al. Prevalence of sleep-disordered
indicated no financial conflicts of interest. breathing in women: effects of gender. Am J Respir Crit Care Med
2001;163:608–13.
REFERENCES 26. Singareddy R, Vgontzas AN, Fernandez-Mendoza J, et al. Risk factors
1. Ohayon MM. From wakefulness to excessive sleepiness: what we know for incident chronic insomnia: a general population prospective study.
and still need to know. Sleep Med Rev 2008; 12:129–41. Sleep Med 2012;13:346–53.

SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 359 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 27. Vgontzas AN, Fernandez-Mendoza J, Bixler EO, et al. Persistent 38. American Psychiatric Association. Diagnostic and statistical manual
insomnia: the role of objective short sleep duration and mental health. of mental disorders, 5th ed. Arlington, VA: American Psychiatric
Sleep 2012;35:61–8. Association, 2013.
28. Fernandez-Mendoza J, Vgontzas AN, Bixler EO, et al. Clinical and 39. Vgontzas AN. Does obesity play a major role in the pathogenesis of sleep
polysomnographic predictors of the natural history of poor sleep in the apnoea and its associated manifestations via inflammation, visceral
general population. Sleep 2012;35:689–97. adiposity, and insulin resistance? Arch Physiol Biochem 2008;114:211–
29. Fernandez-Mendoza J, Vgontzas AN, Liao D, et al. Insomnia with 23.
objective short sleep duration and incident hypertension: the Penn State 40. Chervin RD, Kraemer HC, Guilleminault C. Correlates of sleep
Cohort. Hypertension 2012;60:929–35. latency on the multiple sleep latency test in a clinical population.
30. Vgontzas AN, Fernandez-Mendoza J, Miksiewicz T, et al. Unveiling the Electroencephalogr Clin Neurophysiol 1995;95:147–53.
longitudinal association between short sleep duration and the incidence 41. Mignot E, Young T, Lin L, Finn L. Nocturnal sleep and daytime
of obesity: the Penn State Cohort. Int J Obes (Lond) 2014;38:825–32. sleepiness in normal subjects with HLA-DQB1*0602. Sleep
31. Bixler EO, Vgontzas AN, Lin HM, Vela-Bueno A, Kales A. Insomnia in 1999;22:347–52.
central Pennsylvania. J Psychosom Res 2002;53:589–92. 42. Ojeda DA, Niño CL, López-León S, Camargo A, Adan A, Forero DA.
32. Vgontzas AN, Liao D, Bixler EO, Chrousos GP, Vela-Bueno A. A functional polymorphism in the promoter region of MAOA gene is
Insomnia with objective short sleep duration is associated with a high associated with daytime sleepiness in healthy subjects. J Neurol Sci
risk for hypertension. Sleep 2009;32:491–7. 2014;337:176–9.
33. Vgontzas AN, Liao D, Pejovic S, Calhoun S, Karataraki M, Bixler EO. 43. Evans DS, Snitker S, Wu SH, et al. Habitual sleep/wake patterns in the
Insomnia with objective short sleep duration is associated with type 2 Old Order Amish: heritability and association with non-genetic factors.
diabetes: a population-based study. Diabetes Care 2009;32:1–6. Sleep 2011;34:661–9.
34. Fernandez-Mendoza J, Calhoun S, Bixler EO, et al. Insomnia 44. Goel N, Banks S, Mignot E, Dinges DF. DQB1*0602 predicts
with objective short sleep duration is associated with deficits in interindividual differences in physiologic sleep, sleepiness, and fatigue.
neuropsychological performance: a general population study. Sleep Neurology 2010;75:1509–19.
2010;33:459–65. 45. Goel N, Banks S, Lin L, Mignot E, Dinges DF. Catechol-O-
35. Fernandez-Mendoza J, Calhoun SL, Bixler EO, et al. Sleep methyltransferase Val158Met polymorphism associates with individual
misperception and chronic insomnia in the general population: the role differences in sleep physiologic responses to chronic sleep loss. PLoS
of objective sleep duration and psychological profiles. Psychosom Med One 2011;6:e29283.
2011;73:88–97. 46. Silva GE, Goodwin JL, Sherrill DL, et al. Relationship between
36. Vgontzas AN, Liao D, Pejovic S, et al. Insomnia with short sleep reported and measured sleep times: the sleep heart health study (SHHS).
duration and mortality: the Penn State Cohort. Sleep 2010;33:1159–64. J Clin Sleep Med 2007;3:622–30.
37. Rechtschaffen A, Kales A. A manual of standardized terminology, 47. Lauderdale DS, Knutson KL, Yan LL, et al. Objectively measured sleep
techniques and scoring system for sleep stages of human subjects. characteristics among early-middle-aged adults: the CARDIA study.
Bethesda: National Institutes of Health, 1968. Am J Epidemiol 2006;164:5–16.

SLEEP,
Downloaded Vol. 38, No. 3, 2015
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 360 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018
 SUPPLEMENTAL MATERIAL

Table S1—Multivariable regression models of clinical factors predicting Table S2—Multivariable regression models of clinical factors predicting
incident excessive daytime sleepiness. persistent excessive daytime sleepiness.
Incident EDS Persistent EDS
Model 1 Model 2 Model 1 Model 2
OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI)
R2 0.06 0.13 R2 0.15 0.19
Male – 2.0 (1.2–3.3) ** Female – 1.4 (0.6–3.5)
Non-Caucasian – 2.8 (1.4–5.4) ** Race – 2.4 (0.6–9.3)
Age ≤ 30 years – 4.2 (2.3–7.8) ** Age ≤ 30 years – 0.5 (0.1–2.7)
Age ≥ 65 years – 2.2 (1.3–3.8) ** Age ≥ 65 years – 0.3 (0.1–4.3)
Allergy/asthma 1.5 (1.0–2.4) † 1.6 (1.0–2.6) * Allergy/asthma 1.7 (0.7–4.2) –
Diabetes 1.7 (1.0–3.0) * 1.4 (0.8–2.5) Anemia 3.4 (1.1–10.0) * 3.2 (1.0–10.3) *
Hypertension 1.1 (0.7–1.8) – Hypertension 1.6 (0.7–3.9) –
Migraine 1.2 (0.7–2.2) – BMI (per 2 kg/m2 1.1 (0.9–1.2) † 1.1 (1.0–1.2)
Obesity 1.8 (1.2–2.8) ** 2.0 (1.2–3.2) ** increase)
Sleep apnea 1.7 (0.8–3.7) – Ulcer 2.6 (0.8–9.0) –
Depression 2.6 (1.6–4.2) ** 3.1 (1.8–5.2) ** Depression 1.4 (0.6–3.6) –
Sleep difficulties 3.0 (1.1–8.1) * 2.9 (1.0–8.3) *
Model 1 = multivariable backward conditional model of clinical risk
factors. Model 2 = model 1 plus forced entry of sociodemographic Model 1 = multivariable backward conditional model of clinical risk
factors. OR, odds ratio; 95% CI, 95% confidence interval. † P < 0.10, factors. Model 2 = model plus forced entry of sociodemographic factors.
* P < 0.05, ** P < 0.01. OR, odds ratio; 95% CI, 95% confidence interval. † P < 0.10, * P < 0.05.

Not Depressed Depressed

0.25

0.20
EDS Incidence

0.15

0.10

0.05

0.00
≤ 8 min 9–29 min ≥ 30 min

Sleep Onset Latency

Figure S1—Interaction between depression and polysomnographic

sleep onset latency on incident excessive daytime sleepiness. Depressed
individuals had a higher incidence of EDS (17%) than non-depressed
individuals (7%). Individuals with physiologic sleep propensity (≤ 8 min)
had a higher incidence of EDS (15%) than those with normal (9–29 min)
or increased (≥ 30 min) sleep latency (7% and 8%, respectively). A
significant interaction between these two factors showed that depressed
individuals with increased sleep latency had a higher incidence of
EDS (22%) than depressed individuals with normal sleep latency or
physiologic sleep propensity (13.4% and 11.0%, respectively), whereas
in contrast non-depressed individuals with physiologic sleep propensity
had a higher incidence of EDS (16.3%) than non-depressed individuals
with normal or increased sleep latency at baseline (5.6% and 5.7%,
respectively). Data adjusted for gender, race, age, obesity, hypertension,
diabetes, allergy/asthma, sleep difficulties, and sleep apnea.

SLEEP,
Downloaded Vol. 38, No. 3, 2015 360A
from https://academic.oup.com/sleep/article-abstract/38/3/351/2416855 Natural History of Hypersomnia—Fernandez-Mendoza et al.
by guest
on 30 July 2018