Directory UMM :Data Elmu:jurnal:B:Biological Psichatry:Vol48.Issue10.2000:

Does Obstructive Sleep Apnea Confound Sleep
Architecture Findings in Subjects with
Depressive Symptoms?
Wayne A. Bardwell, Polly Moore, Sonia Ancoli-Israel, and Joel E. Dimsdale
Background: Compared with normal subjects, depressed
patients have shorter rapid eye movement sleep latency
(REML), increased REM and decreased slow wave sleep
as a percentage of total sleep time (REM%, SWS%), and
longer sleep latency (SL). Obstructive sleep apnea (OSA)
patients experience longer REML, decreased REM% and
SWS%, and shorter SL. We examined the interplay of
depressive symptoms, OSA, and sleep architecture.
Methods: Subjects (n 5 106) were studied with polysomnography. OSA was defined as a Respiratory Disturbance
Index $15. Subjects were divided into Hi/Lo groups using
a Center for Epidemiological Studies—Depression
(CES-D) score of 16.
Results: OSA patients had shorter SL than non-OSA
patients (14.5 vs. 26.8 min, p , .001); Hi CES-D subjects
showed a trend toward longer SL than Lo CES-D subjects
(23.7 vs. 17.5 min, p 5 .079). Significant OSA 3 CES-D
interactions emerged, however, for REM% (p 5 .040) and

SL (p 5 .002): OSA/Hi CES-D subjects had higher REM%
than OSA/Lo CES-D subjects (19.3% vs. 14.3%, p 5
.021); non–OSA/Hi CES-D subjects had SL (35.3 min) 2–3
times as long as other subjects (p 5 .002–.012).
Conclusions: Because of the high prevalence of OSA and
depression, findings suggest that OSA must be considered
in studies of mood and sleep architecture. Conversely,
depressive symptoms must be considered in studies of OSA
and sleep architecture. Biol Psychiatry 2000;48:
1001–1009 © 2000 Society of Biological Psychiatry
Key Words: Depression, mood, obstructive sleep apnea,
sleep architecture, REM, sleep latency

Introduction

S

leep disruption has long been recognized as accompanying depression. When compared with the nondepressed, the most frequently reported alterations in sleep
From the Department of Psychiatry, University of California San Diego (WAB,
PM, SA-I, JED) and Veterans Affairs San Diego Healthcare System (WAB,

SA-I), San Diego, California.
Address reprint requests to Wayne A. Bardwell, Ph.D., UCSD, Dept. of Psychiatry,
9500 Gilman Drive 0804, La Jolla CA 92093-0804.
Received January 27, 2000; revised March 15, 2000; accepted March 20, 2000.

© 2000 Society of Biological Psychiatry

architecture in depressed patients include reduced rapid
eye movement sleep latency (REML), increased REM as a
percentage of total sleep time (REM%), reduced percentage of slow wave sleep (Stages 3 and 4) (SWS%), and
increased time to sleep onset (also known as sleep latency)(Benca et al 1992; Gillin et al 1993). During acute
illness, some form of sleep abnormality is present in about
90% of depressed patients (Reynolds and Kupfer 1987). In
addition, mild levels of obstructive sleep apnea (OSA)
have been observed in 15% of inpatients with affective
disorders (Reynolds et al 1982).
Table 1 summarizes research documenting alterations in
sleep architecture in patients with varying levels of depressive symptoms. The mean REM% in these studies
ranged from 17% to 27%, averaging 22%. REML ranged
from 46 to 107 min, averaging 67 min. The mean sleep

latency recorded in these studies ranged from 11 to 36
min, averaging 24 min. By comparison, normative data
based on a sample of subjects free of sleep or alertness
complaints and without major medical or psychiatric
diagnoses (Doghramji et al 1997) reveals that REM%
averaged 19.27%; REML averaged 106.6 min; SWS%
averaged 16.49%; and sleep latency averaged 10.9 min. It
is generally accepted that the degree of sleep disruption
does not necessarily reflect the severity of the depression;
even comparatively mild depressive symptomatology may
produce alterations in polysomnographically recorded
sleep (Buysse et al 1997; Hauri et al 1974).
An intriguing aspect that has been given little attention
in the psychiatric literature is the possibility that coexisting sleep disorders may confound these relationships
between depressive symptoms and sleep architecture. In
this article, we examined how OSA affected the relationships between sleep architecture and depressive symptoms. Affecting up to 9% of middle-aged adults and higher
percentages in the elderly, OSA is a devastating illness
(Ancoli-Israel et al 1991; Jennum and Sjol 1992; Jeong
and Dimsdale 1989; Young et al 1993) that leaves patients
exhausted from sleep deprivation.

The literature is mixed regarding the role played by
psychological factors in OSA (for review, see Bardwell et
0006-3223/00/$20.00
PII S0006-3223(00)00887-8

1002

Table 1. Values for Sleep Latency, Rapid Eye Movement (REM) Latency and REM% in Clinical Depression: A Sample of the Literature
Subjects

Hubain et al 1998

300 (198 F, 102 M)

Thase et al 1998

Definition of depression

Sleep latency

REM latency

REM%

Inpatients, drug free,
HRSD $ 20

NR

F: 74 (60)
M: 52 (37)

F: 17 (7)
M: 18 (7)

78 (? f, ? M)

Outpatients, drug free,
HRSD $ 14

Grp 1:26 (18)
Grp 2:14 (9)

Grp 1:63 (17)
Grp 2:75 (22)

Grp 1:23 (5)
Grp 2:23 (5)

Reynolds et al 1990

302 (151 F, 151 M)

Outpatients, drug free,
HRSD $ 15

F: 23 (27)
M: 21 (21)

F: 64 (30)

M: 65 (33)

F: 23 (6)
M: 22 (6)

Reynolds et al 1984

25 M

KDS $ 15

NR

Grp 1:85 (46)
Grp 2:55 (28)

NR

Inpatients, drug free,
HRSD $ 30

36 (3)a

46 (4)a

25 (1)a

34 (20 F, 14 M)

Coble et al 1979

12 (8 F, 4 M)

Nondelusional inpatients,
drug free, HRSD $
30

35 (8)

48 (7)

NR

Gillin et al 1978

9 (5 F, 4 M)

Inpatients, moderately to
severely depressed

29 (16)

58 (27)

27 (2)

Hauri et al 1974

14 patients
(10 F, 4 M) and

4 control subjects

Drug-free remitted
patients

Patients: 24 (23)
Control subjects: 6 (3)

Patients: 90 (25)
Control subjects:
76 (20)

Patients: 21 (6)
Control subjects:
21 (4)

Controlled for various confounding factors
(such as age, weight loss, severity of
depression, etc). Correlations were found
with increased waking and decreased

SWS, but not with REM latency.
Two groups of depressed patients were
classified on the basis of “abnormal” vs.
“normal” pretreatment sleep study
measures. Certain sleep abnormalities were
stable across time (including REMlatency) and after treatment with cognitive
behavior therapy; whereas other measures
improved (such as sleep efficiency).
Gender-related differences in sleep variables
were found for slow-wave sleep measures
but not for REM latency.
Grp 1 5 KDS $ 15; Grp 2 5 KDS , 15.
40% of male OSA patients met criteria for
affective disorder or alcohol abuse.
Followed patients with various dosages of
tricyclic antidepressant, and found that
treatment response was associated with
changes in REM latency.
Sleep parameters were examined over time
in patients on placebo for 5 weeks. Sleep
parameters remained fairly unchanged
across time.
Patients’ sleep was studied prior to treatment
with a tricyclic antidepressant, during
treatment and during withdrawal from
treatment. Patients with clinical
improvement showed a significant REM
rebound effect during drug withdrawal, but
the patients who failed to improve did not.
Some sleep abnormalities found to persist in
patients who had been remitted for longer
than six months. NREM sleep
abnormalities (longer sleep latency,
decreased SWS) seemed to persist longer
than certain of the REM abnormalities.

Values given are means (SD), except where a indicates (SEM). Sleep latency values are in minutes. REM latency values are given in minutes, but may have been defined slightly differently in different studies.
F, female; M, male; HRSD, Hamilton Rating Scale—Depression; NR, not reported; SWS, slow wave sleep; KDS, Kupfer–Detre Scale; OSA, obstructive sleep apnea; NREM, non-REM.

W.A. Bardwell et al

Kupfer et al 1981

Findings

BIOL PSYCHIATRY
2000;48:1001–1009

Reference

Depression, Sleep Architecture, OSA

al 1999). Nonetheless, many researchers have found associations between OSA and overall psychological distress
and, more specifically, clinical depression (Cheshire et al
1992; Guilleminault et al 1977; Millman et al 1989;
Reynolds et al 1984) or increased levels of depressive
symptoms (Borak et al 1994; Derderian et al 1988;
Edinger et al 1994; Engleman et al 1994; Flemons and
Tsai 1997; Kales et al 1985; Platon and Sierra 1992). For
example, in a study of 25 male patients with sleep apnea,
40% met criteria for an affective disorder or alcohol abuse
(Reynolds et al 1984).
Sleep architecture differences in OSA patients contrasted with normal subjects (i.e., compared to Doghramji
et al [1997] normative values) have included shorter sleep
latency (Fry et al 1989; Hanzel et al 1991; Kass et al 1996;
Series et al 1992; Smith et al 1985), increased REML
(Mendelson 1992, 1995; Series et al 1992; Stewart et al
1992), decreased REM% (Colt et al 1991; Chervin and
Aldrich 1998; Espinoza et al 1987; Hanzel et al 1991;
Mendelson 1992, 1995; Oksenberg et al 1997; Roehrs et al
1989; Series et al 1992; Smith et al 1985; Stewart et al
1992), and decreased SWS% (Weitzman et al 1980; White
1992). With the exception of SWS%, these findings are
opposite to those observed in depressed patients.
Table 2 summarizes research documenting sleep architecture in OSA patients. The mean REML in these studies
ranged from 63 to 175 min, averaging 121 min. OSA
patients also tend to experience a lower REM% than
normal subjects do. This may simply be a consequence of
the impaired sleep continuity and fragmentation of sleep
resulting from repetitive airway obstruction. The mean
REM% in these studies ranged from 6% to 20%, averaging
14% of total sleep time. In addition, sleep latency in OSA
patients tends to be shorter than in normal subjects—
probably because of the accumulated sleep debt. The mean
sleep latency recorded in these studies ranged from 4 to 24
min, averaging 10 min.
Because the presence of OSA may not be routinely
determined in studies of sleep and mood, findings in this
area of research may be confounded. Conversely, depressive symptoms may complicate studies of sleep architecture in OSA patients. To our knowledge, no previous
studies have examined the potential interaction effect of
OSA and depressive symptoms on sleep architecture. We
wondered, for example, if the contrasting sleep effects
seen in OSA and in depressive symptoms would offset
each other or if some other pattern might emerge in
patients exhibiting both high levels of depressive symptoms and OSA.
Most previous research into affective disorders and
sleep architecture has focused on patients with diagnosed
mood disorders. Because of the growing interest in the
impact of subclinical levels of depressive symptoms, we

BIOL PSYCHIATRY
2000;48:1001–1009

1003

wondered if these previously reported findings would hold
when dividing patients into high and low depressive
symptom groups, regardless of diagnosis. In this study we
examined the relationships between sleep latency, REML,
and REM% versus OSA diagnosis and level of depressive
symptoms. We wondered if a diagnosis of OSA would
impact these sleep architecture variables in subjects with
high levels of depressive symptoms, and if depressive
symptoms would impact them in subjects with OSA.

Methods and Materials
Subjects (n 5 106) included 57 men and 10 women with OSA
and 27 men and 12 women without OSA. None of the subjects
had any other sleep disorder. Subjects ranged from 32 to 64 years
of age (Table 3) and were recruited by advertising and word of
mouth to participate in our research on sympathetic nervous
system physiology in subjects with and without OSA. To qualify,
subjects had to be between 100% and 150% of ideal body weight
as determined by Metropolitan Life Insurance tables (Metropolitan Life Foundation 1983). Although OSA is more common
among the obese, subjects with greater than 150% of ideal body
weight were excluded because of the possibility of confounding
by other conditions associated with obesity. Subjects were also
excluded if they had major medical illnesses other than OSA.
Subjects were studied after we had obtained written informed
consent. The protocol was approved by the University of
California, San Diego, Human Subjects Committee.
Polysomnographic sleep was monitored all night in the Clinical Research Center, and included standard central and occipital
electroencephalogram (EEG); bilateral electrooculogram (EOG);
submental electromyogram (EMG); airflow, thoracic, and abdominal excursions with Respitrace respiratory inductive plethysmography (Non-Invasive Monitoring Systems, Miami Beach,
FL); and bilateral tibialis EMG. Tracings were scored according
to the criteria of Rechtshaffen and Kales (1968) and the number
of apnea/hypopnea events was recorded. The majority of subjects
had solely obstructive type events; only a few subjects showed
evidence of central apneas.
Subjects were classified as having sleep apnea if their Respiratory Disturbance Index (RDI) was 15 or more: in OSA subjects
(n 5 67), RDI ranged from 15 to 142 (mean 5 50.7, SD 5 28.7);
for non-OSA subjects (n 5 39), RDI ranged from 1 to 14
(mean 5 5.6, SD 5 3.7). Sleep latency was defined as the time
from lights out to the first epoch of stage 2 sleep. REM latency
was defined as time from the first epoch of stage 2 sleep to the
first epoch of REM sleep. REM% and SWS% were calculated by
dividing min of REM by total sleep time, and min of stage 3 plus
stage 4 sleep by total sleep time, respectively.
Subjects completed the Center for Epidemiological Studies—
Depression (CES-D) scale (Radloff 1997), a frequently used
20-item self-report measure of depressive symptoms. When used
with medically ill patients, depression rating scales are sometimes influenced by symptoms of their illness. For example,
mood-related observations in OSA patients could be confounded
by the fatigue/sleepiness they experience; however, the CES-D
primarily taps the cognitive/affective symptoms of depression

Reference

Subjects

OSA
criterion

Sleep latency,
nighttime

REM latency

REM%

NR

11.2 (2.3)

NR

9.6 (2.1)

Hanzel et al 1991

7 M, 5 F

NR

4.1 (1.1)

NR

17 (2)

Fry et al 1989

30 M, 3 F

RDI $ 10

7.5 (6.2)

62.8 (20.2)

19.8 (6.0)

Smith et al 1985

13 M, 2 F

RDI $ 10

6.9 (1.5)

NR

Approx. 10a

Mendelson 1992

169 M, 23 F

RDI $ 5

11.7 (1.5)

134 (6.5)

13.8b (unknown)

Mendelson 1995

430 M, 88 F

RDI $ 5

13.8 (4.6)

131.0 (84.1)

14.4b (unknown)

Oksenberg et al 1997

574 patients in 2 groups
(G1, G2)

RDI $ 10

G1: 12.9 (4.2)
G2: 11.7 (3.7)

G1: 89.6 (46.6)
G2: 104 (63.7)

G1: 19 (7.4)
G2: 17 (8.5)

Series et al 1992

8M

NR

8.7 (3.0)

149.6 (24.9)

Stewart et al 1992

8M

NR

24 (3.3)

175 (32)

13.4b (unknown)

Kass et al 1996

34 (15 M, 19 F)

RDI , 10

8.3 (3.4)

NR

ns

Colt et al 1991

7M

NR

NR

NR

6 (5)

Roehrs et al 1989

415 M, 51 F

NR

NR

NR

12.1 (6.4)

Chervin and Aldrich
1998

814 M, 332 F

RDI $ 10

NR

NR

14.2 (6.2)

13.4 (1.7)

Compared a 1-night infusion of aminophylline to
a placebo infusion night. Aminophylline caused
a decrease in central events, but no change in
obstructive events and also disturbed sleep
efficiency.
Compared the effect of fluoxetine to protryptiline.
Both drugs decreased REM time and number of
REM-related respiratory events.
Periodic limb movements in OSA were measured
at baseline, and with two CPAP recordings.
The therapeutic effects of weight loss on OSA
were examined.
OSA patients were compared to patients with
central sleep apnea and pathologic snorers
(only the OSA data are reported in this table).
OSA patients were subjectively sleepier.
Hypertension was strongly associated with
weight and also with RDI.
Compared OSA patients to patients with
subclinical sleep-disordered breathing.
OSA patients divided into two groups based on
whether or not the RDI was worsened 3 2
when positioned supine.
Administered a short-acting SWS-enhancing drug,
g-hydroxy-butyrate, to see if overall RDI could
be improved.
Apnea patients were treated for one week with a
nonsteroidal antiandrogen.
REM RDI correlated with sleepiness measures in
patients with mild OSA but worsening during
REM.
No statistically significant differences in MSLT
scores after CPAP treatment under hypoxemic
or nonhypoxemic conditions.
Found that the best predictor of MSLT score was
the degree of sleep fragmentation during the
previous night (i.e., the respiratory arousal
index).
Concluded that apneic events during both REM
and non-REM sleep contribute to sleepiness.

Values given are means (SEM).
Sleep latency values are in minutes. REM latency values are given in minutes, but may have been defined slightly differently in different studies.
OSA, obstructive sleep apnea; M, male; NR, not reported; F, female; RDI, Respiratory Disturbance Index; CPAP, continuous positive airway pressure; SWS, slow wave sleep; MSLT, multiple sleep latency test.
a
Value not specified but estimated from figure.
b
Value was calculated from given information of total sleep time, or TST (min), and REM sleep (min): (REM min/TST min) 3 100 to give % REM.

W.A. Bardwell et al

10 M

Findings
BIOL PSYCHIATRY
2000;48:1001–1009

Espinoza et al 1987

1004

Table 2. Values for Sleep Latency, Rapid Eye Movement (REM) Latency and REM% in Obstructive Sleep Apnea: A Sample of the Literature

Depression, Sleep Architecture, OSA

BIOL PSYCHIATRY
2000;48:1001–1009

1005

Table 3. Demographic Variables (Mean 6 SD)

Agea,e
Body mass indexa,e
REM latency
REM as % of total sleep timeb,d
SWS as % of total sleep time
Sleep latency (min)a,b,e
Respiratory Disturbance Indexa,e
CES-D Scorec,e

No apnea
Lo CES-D
(n 5 23)

No apnea
Hi CES-D
(n 5 16)

Apnea
Lo CES-D
(n 5 45)

Apnea
Hi CES-D
(n 5 22)

46 6 6.4
27.0 6 3.6
80.7 6 15.7
18 6 7.6
10.0 6 9.4
18.3 6 14.0
5.7 6 3.6
7.0 6 4.1

43 6 5.6
28.1 6 3.5
120.0 6 19.5
16 6 8.2
9.5 6 8.0
35.3 6 31.7
5.5 6 3.9
23.3 6 6.2

50 6 7.7
28.9 6 3.7
120.7 6 11.2
15 6 7.7
4.6 6 6.0
16.8 6 12.3
52.0 6 29.6
6.0 6 4.2

48 6 8.0
31.1 6 5.4
106.0 6 15.7
19 6 9.4
9.6 6 9.5
12.1 6 13.1
48.0 6 27.8
26.3 6 8.2

n 5 106. CES-D, Center for Epidemiological Studies—Depression; REM, rapid eye movement sleep; SWS, slow wave sleep (stages 3 1 4).
Significant main effect for apnea diagnosis.
b
Significant apnea diagnosis 3 CES-D Hi/Lo interaction.
c
Significant main effect for CES-D.
d
p , .05.
e
p , .01.
a

and has been shown to be useful in chronically ill groups
experiencing fatigue (e.g., human immunodeficiency virus
[HIV], cancer) (Cockram et al 1999; Hann et al 1999). In a
variety of populations, a large percentage of subjects with a score
of $16 have been shown to meet diagnostic criteria for dysthymia or major depression (Murphy 1982; Schulberg et al 1985).
Using this commonly accepted cut-off score, our subjects were
divided into Hi/Lo depressive symptom groups. For the Lo
CES-D group, scores ranged from 0 to 15 (mean 5 6.2, SD 5
4.2); for the Hi CES-D group, scores ranged from 16 to 49
(mean 5 25.0, SD 5 7.2).
Data were analyzed using SPSS 9.0 software (SPSS, Chicago)
multiple analysis of covariance (MANCOVA) routines (using
age as a covariate): REML, REM%, SWS%, and sleep latency as
the dependent variables, and OSA/non-OSA and Hi/Lo CES-D
as the independent variables. Stepwise regression analysis was
also employed, using the independent variables in their continuous form. An interaction term was created by first standardizing
and then multiplying RDI and CES-D scores. In step one, age
was controlled via forced entry; in step two, RDI and CES-D
were entered; in step three, the interaction term was entered.

Results
Table 3 shows values for demographic, sleep, and mood
variables by Apnea/CES-D group. OSA subjects were
slightly older (49.3 vs. 45.1 years, p 5 .007), so age was
used as a covariate in subsequent analyses. As expected,
OSA subjects were heavier (29.7 vs. 27.4 body mass
index, p 5 .004), and had significantly greater RDI (50.7
vs. 5.6, p , .001) than non-OSA subjects. Also, subjects
in the Hi CES-D group reported significantly more depressive symptoms than Lo CES-D subjects (25.0 vs. 6.2, p ,
.001). OSA and non-OSA subjects did not differ significantly on the CES-D (16.3 vs. 14.8, p 5 .190); and Hi and
Lo CES-D subjects did not differ significantly in terms of
RDI (25.8 vs. 29.0, p 5 .509).
In the MANCOVA, REML, REM%, SWS%, and sleep

latency were the dependent variables, and OSA (yes/no)
and CES-D (Hi/Lo) were the independent variables. No
significant main effects emerged for OSA (yes/no) for
REML (109.6 vs. 109.8 min, p 5 .991), REM% (17.1%
vs. 16.6%, p 5 .784), or SWS% (7.0% vs. 9.6%, p 5
.148). OSA subjects showed significantly shorter sleep
latency than non-OSA subjects, however (14.5 vs. 26.8
min, p 5 .001). Similarly, no significant main effects
emerged for CES-D Hi/Lo for REML (119.8 vs. 99.6 min,
p 5 .203), REM% (17.5% vs. 16.3%, p 5 .495), or SWS%
(9.6% vs. 7.0%, p 5 .133); however, Hi CES-D subjects
showed a trend toward longer sleep latency than Lo
CES-D subjects, (23.7 vs. 17.5 min, p 5 .079).
The OSA 3 CES-D interaction was not significant for
REML (p 5 .066) or SWS% (p 5 .141); however,
significant OSA 3 CES-D interactions emerged for both
sleep latency (p 5 .002; Figure 1) and REM% (p 5 .040;
Figure 2). Post hoc analyses revealed that the sleep latency
in non–OSA/Hi CES-D subjects was twice as long as
non–OSA/Lo CES-D subjects (35.3 vs. 18.3 min, p 5
.012) and OSA/Lo CES-D subjects (35.3 vs. 16.8 min, p 5

Figure 1. Sleep latency vs. obstructive sleep apnea (OSA) 3
Center for Epidemiological Studies—Depression (CES-D) interaction (p 5 .002).

1006

BIOL PSYCHIATRY
2000;48:1001–1009

Figure 2. Percentage of rapid eye movement sleep (REM) vs.
obstructive sleep apnea (OSA) 3 Center for Epidemiological
Studies—Depression (CES-D) interaction (p 5 .040).

.010), and nearly three times as long as OSA/Hi CES-D
subjects (35.3 vs. 12.1 min, p 5 .002). Sleep latency
differences between OSA/Lo CES-D and OSA/Hi CES-D
and between OSA/Lo CES-D and non–OSA/Lo CES-D
subjects were not significant.
Regarding the significant interaction for REM%, posthoc analyses revealed that OSA/Hi CES-D subjects experienced higher REM% than OSA/Lo CES-D subjects
(19.4% vs. 14.8%, p 5 .021); however, non-OSA subjects
did not differ in terms of REM% regardless of CES-D
level; Hi CES-D subjects did not differ in terms of REM%
regardless of apnea diagnosis; and Lo CES-D subjects did
not differ in terms of REM% regardless of apnea
diagnosis.
Data were also examined in stepwise regression analyses using continuous RDI and CES-D scores and the
RDI 3 CES-D interaction. Age was controlled by forced
entry. Using sleep latency as the dependent variable, only
the interaction term entered the model (b 5 24.462, b 5
2.260, p 5 .008). Using SWS% as the dependent variable,
RDI (b 5 2.001, b 5 2.352, p , .001) and CES-D (b 5
.002, b 5 .265, p 5 .004) entered the model. Using
REM% or REML as the dependent variables, none of the
independent variables entered the model.

Discussion
It is well documented that mood disorders can have a
detrimental effect on quality and quantity of sleep. Several
researchers have observed sleep architecture differences
between patients with varying levels of depressive symptoms. These differences have generally, but not always,
included a cluster of abnormalities, such as shorter REML,
increased REM%, diminished SWS%, and prolonged
sleep latency; however, no single aspect of sleep architecture can be considered a definitive biological marker for
depressive disorders.

W.A. Bardwell et al

OSA is a relatively common sleep disorder and patients
with OSA often experience higher levels of depressive
symptoms than people without a sleep disorder. Beyond
such accepted markers as RDI, arousals, and oxygen
desaturation, researchers have also documented longer
REML, decreased REM%, diminished SWS%, and shorter
sleep latency in OSA patients. As in the depression
research, findings have not been consistent across studies
for these variables.
Because depression and OSA have often been shown to
have opposite effects on REML, REM%, and sleep latency,
we were interested in examining how the depressive symptoms 3 OSA interaction would impact sleep architecture.
Also, because of the burgeoning interest in the impact of
subclinical levels of depressive symptoms on health and
behavior, we wondered if these sleep architecture differences
would hold when comparing subjects reporting a range of
depressive symptoms—regardless of the presence of a diagnosed mood disorder. We hypothesized that OSA may
confound sleep architecture observations when comparing
groups of subjects with comparatively high and low levels of
depressive symptoms. Conversely, we also hypothesized that
high levels of depressive symptoms may confound sleep
architecture differences found when comparing subjects with
and without OSA.
In our sample of 106 subjects, we were able to replicate
some of the findings from the sleep-in-depression literature as well as the sleep-in-OSA literature. Our OSA
subjects experienced significantly shorter sleep latency—
about half as long as that in subjects without OSA. There
was also a trend for subjects who reported high levels of
depressive symptoms to experience sleep latency one third
longer than subjects reporting low levels of depressive
symptoms. We did not observe significant differences
between OSA and non-OSA subjects nor between Hi
CES-D and Lo CES-D subjects for REML, REM%, or
SWS%; however, the interaction of depressive symptoms
and OSA diagnosis produced results that were statistically
and clinically significant.
We had speculated that the REML-reducing effects of
depression and the REML-increasing effects of OSA
might offset each other in OSA patients who had depressive symptoms. Reynolds et al (1982, 1984) have reported
REML findings that differ from our results. In that study,
the REML in patients with OSA and depression was
significantly longer than in patients having only OSA.
They postulated that the association between the REM
sleep abnormalities of depression and OSA might be due
to the fact that both disorders increase with age. Our larger
sample size allowed us to test the depression 3 OSA
interaction and to control for the effects of age. These
factors, along with the use of different scales for rating
depression, may help explain our divergent findings.

Depression, Sleep Architecture, OSA

Depressive symptoms and OSA diagnosis interacted to
produce SL and REM% patterns that were not consistent
with previously reported findings. Non-OSA subjects who
endorsed high levels of depressive symptoms experienced
the longest sleep latency—two to three times as long as all
other subjects. Apparently, in non-OSA subjects, the
presence of high levels of depressive symptoms significantly prolongs sleep latency. In OSA subjects, however,
sleep latency seems to be resistant to the effects of
depressive symptoms—remaining comparatively low regardless of the level of depressive symptoms reported. In
OSA subjects reporting high levels of depressive symptoms, the fatigue associated with OSA may override the
longer sleep latency that often accompanies high levels of
depressive symptoms.
Regarding REM%, OSA subjects reporting high levels
of depressive symptoms experienced REM% nearly one
third higher than those OSA subjects with low levels of
depressive symptoms. Thus, the presence of comparatively high levels of depressive symptoms may override
the diminution of REM sleep often associated with OSA.
In subjects without OSA, however, REM% did not differ
in terms of level of depressive symptoms.
We did not find significant main effects or interaction
effects with REML as the dependent variable. Some
readers may ask about first-night effects, particularly in
terms of REM sleep variables. Although this may have had
an impact on our data, it is typical of the studies reported
in our review to rely on single-night recordings.
We wondered if subjects in our sample might have
experienced only minor depression rather than more serious levels of dysphoric mood and, if so, if this might help
explain our results. To test this hypothesis, we increased
the CES-D cutoff to 20; however, this change in cutoff did
not significantly affect the findings reported here.
Our findings suggest that the presence of comparatively
high levels of depressive symptoms—not simply the
presence or absence of a diagnosed clinical mood disorder—must be considered in studies of sleep architecture in
OSA patients. Conversely, the presence of OSA must be
considered in studies of sleep architecture in subjects with
mood disorders or with high levels of depressive symptoms. The interaction effects we observed may account for
some of the differences in findings reported separately in
previous studies of sleep architecture in depressed patients
and OSA patients.

Limitations
There are certain limitations of this study that could
impact the generalizability of our findings. Regarding
potential bias in recruitment, we sought both subjects with
and without symptoms suggestive of sleep apnea. It is

BIOL PSYCHIATRY
2000;48:1001–1009

1007

possible that persons with particularly prominent apnea
symptoms were more likely to seek inclusion in the study,
thus skewing our sample toward those with more severe
illness; however, the RDI in our participants ranged from
0 to 152, averaging 50 in the apneic group.
Some of our findings varied depending on the statistical
methodology employed. Using sleep latency as the dependent variable, our findings regarding the RDI 3 CES-D
interaction were consistent when using either dichotomous
or continuous forms of this independent variable. Our
findings regarding REM% could not be duplicated, however, when using continuous independent variables. Replication in another data set is required.
It is possible that the fatigue/sleepiness experienced by
OSA patients confounded our evaluation of mood symptoms. As mentioned previously, the CES-D tends to assess
cognitive/affective domains of depression and has been
used successfully in other chronically ill groups experiencing fatigue. Nevertheless, we examined the relationship between CES-D and Epworth Sleepiness Scale (Johns
1991) scores in our OSA subjects. Results were not
significant, suggesting that CES-D scores are not confounded by sleepiness/fatigue.

This work was supported by National Institutes of Health Grants Nos.
HL44915, AG02711, and RR00827.

References
Ancoli-Israel S, Kripke DF, Klauber MR, Mason WJ, Fell R,
Kaplan O (1991): Sleep disordered breathing in communitydwelling elderly. Sleep 14:486 – 495.
Bardwell WA, Berry CC, Ancoli-Israel S, Dimsdale JE (1999):
Psychological correlates of sleep apnea. J Psychosom Res
47:583–596.
Benca RM, Obermeyer WH, Thisted RA, Gillin JC (1992): Sleep
and psychiatric disorders: A meta-analysis. Arch Gen Psychiatry 49:651– 668.
Borak J, Cieslicki J, Szelenberger W, Wilczak-Szadkowska H,
Koziej M, Zielinski J (1994): Psychopathological characteristics of the consequences of obstructive sleep apnea prior to
and three months after CPAP. Psychiatr Pol 28:33– 44.
Buysse DJ, Frank E, Lowe KK, Cherry CR, Kupfer DJ (1997):
Electroencephalographic sleep correlates of episode and vulnerability to recurrence in depression. Biol Psychiatry 41:
406 – 418.
Chervin RD, Aldrich MS (1998): The relation between multiple
sleep latency test findings and the frequency of apneic events
in REM and non-REM sleep. Chest 113:980 –984.
Cheshire K, Engleman H, Deary I, Shapiro C, Douglas NJ
(1992): Factors impairing daytime performance in patients
with sleep apnea/hypopnea syndrome. Arch Intern Med 152:
538 –541.
Coble PA, Kupfer DJ, Spiker DG, Neil JF, McPartland RJ

1008

BIOL PSYCHIATRY
2000;48:1001–1009

(1979): EEG sleep in primary depression: A longitudinal
placebo study. J Affect Disord 1:131–138.
Cockram A, Judd FK, Mijch A, Norman T (1999): The evaluation of depression in inpatients with HIV disease. Aust N Z
J Psychiatry 33:344 –352.
Colt HG, Haas H, Rich GB (1991) Hypoxemia vs. sleep
fragmentation as cause of excessive daytime sleepiness in
obstructive sleep apnea. Chest 100:1542–1548.
Derderian SS, Bridenbaugh RH, Rajagopal KR (1998): Neuropsychologic symptoms in obstructive sleep apnea improve
after treatment with nasal continuous positive airway pressure. Chest 94:1023–1027.
Doghramji K, Mitler MM, Sangal RB, Shapiro C, Taylor S,
Walsleben J, et al (1997): A normative study of the maintenance of wakefulness test (MWT). Electroencephalogr Clin
Neurophysiol 103:554 –562.
Edinger J, Carwile S, Miller P, Hope V, Mayti C (1994):
Psychological status, syndromatic measures, and compliance
with nasal CPAP therapy for sleep apnea. Percept Mot Skills
78:1116 –1118.
Engleman HM, Martin SE, Deary IJ, Douglas NJ (1994): Effect
of continuous positive airway pressure treatment on daytime
functioning in sleep apnea/hypopnea syndrome. Lancet 343:
572–575.
Espinoza H, Antic R, Thornton AT, McEvoy RD (1987): The
effects of aminophylline on sleep and sleep-disordered
breathing in patients with obstructive sleep apnea syndrome.
Am Rev Respir Dis 136:80 – 84.
Flemons WW, Tsai W (1997): Quality of life consequences of
sleep-disordered breathing. J Allergy Clin Immunol 99:S750 –
S756.
Fry JM, DiPhillipo MA, Pressman MR (1989): Periodic leg
movements in sleep following treatment of obstructive sleep
apnea with nasal continuous positive airway pressure. Chest
96:89 –91.
Gillin JC, Dow BM, Thompson P, Parry B, Tandon R, Benca R
(1993): Sleep in depression and other psychiatric disorders.
Clin Neurosci 1:90 –96.
Gillin JC, Wyatt RJ, Fram D, Snyder F (1978): The relationship
between changes in REM sleep and clinical improvement in
depressed patients treated with amitryptiline. Psychopharmacology 59:267–272.
Guilleminault C, Eldridge FL, Tilkian A, Simmons FB, Dement
WC (1977): Sleep apnea syndrome due to upper airway obstruction: A review of 25 cases. Arch Intern Med 137:296 –300.
Hann D, Winter K, Jacobsen P (1999): Measurement of depressive symptoms in cancer patients: Evaluation of the Center
for Epidemiological Studies Depression Scale (CES-D).
J Psychosom Res 46:437– 443.
Hanzel DA, Proia NG, Hudgel DW (1991): Response of obstructive sleep apnea to fluoxetine and protryptiline. Chest 100:
416 – 421.
Hauri P, Chernik D, Hawkins D, Mendels J (1974): Sleep of
depressed patients in remission. Arch Gen Psychiatry 31:
386 –391.
Hubain PP, Staner L, Dramaix M, Kerkhofs M, Papadimitriou G,
Mendlewicz J, et al (1998): The dexamethasone suppression
test and sleep electroencephalogram in nonbipolar major
depressed inpatients: A multivariate analysis. Biol Psychiatry
43:220 –229.

W.A. Bardwell et al

Jennum P, Sjol A (1992): Epidemiology of snoring and obstructive sleep apnea in a Danish population, age 30-60. J Sleep
Res 1:240 –244.
Jeong DU, Dimsdale JE (1989): Sleep apnea and essential
hypertension: A critical review of the epidemiological evidence for co-morbidity. Clin Exp Hypertens 11:1301–1323.
Johns MW (1991): A new method for measuring daytime
sleepiness: The Epworth sleepiness scale. Sleep 14:540 –545.
Kales A, Caldwell AB, Cadieux RJ, Vela-Bueno A, Ruch LG,
Mayes SD (1985): Severe obstructive sleep apnea—II: Associated psychopathology and psychosocial consequences.
J Chronic Dis 38:427– 434.
Kass JE, Akers SM, Bartter TC, Pratter MR (1996): Rapid-eyemovement-specific sleep-disordered breathing: A possible
cause of excessive daytime sleepiness. Am J Respir Crit Care
Med 154:167–169.
Kupfer DJ, Spiker DG, Coble PA, Neil JF, Ulrich R, Shaw DH
(1981): Sleep and treatment prediction in endogenous depression. Am J Psychiatry 138:429 – 434.
Mendelson WB (1992): Sleepiness and hypertension in obstructive sleep apnea. Chest 101:903–909.
Mendelson WB (1995): The relationship of sleepiness and blood
pressure to respiratory variables in obstructive sleep apnea.
Chest 108:966 –972.
Metropolitan Life Foundation (1983): Metropolitan height and
weight tables. Stat Bull (Metropolitan Life Insurance) 64:1.
Millman RP, Fogel BS, McNamara ME, Carlisle CC (1989):
Depression as a manifestation of obstructive sleep apnea:
Reversal with nasal continuous positive airway pressure.
J Clin Psychiatry 50:348 –351.
Murphy J (1982): Psychiatric Instrument Development for Primary Care Research: Patient Self-Report Questionnaire,
Report on Contract 80M01428101D. Bethesda, MD: National
Institute of Mental Health.
Oksenberg A, Silverberg DS, Arons E, Radwan H (1997):
Positional vs. nonpositional obstructive sleep apnea patients:
Anthropomorphic, nocturnal polysomnographic and multiple
sleep latency test data. Chest 112:629 – 639.
Platon MJ, Sierra JE (1992): Changes in psychopathological
symptoms in sleep apnea patients after treatment with nasal
continuous positive airway pressure. Int J Neurosci 62:173–
195.
Radloff LS (1997): The CES-D Scale: A self-report depression
scale for research in the general population. Appl Psychol
Meas 1:385– 401.
Rechtshaffen A, Kales A (1968): A Manual of Standardized
Terminology: Techniques and Scoring System for Sleep
Stages of Human Subjects. Los Angeles: UCLA Brain Information Service/Brain Research Institute.
Reynolds CF, Coble PA, Spiker DG, Neil JF, Holzer BC, Kupfer
DJ (1982): Prevalence of sleep apnea and nocturnal myoclonus in major affective disorders: Clinical and polysomnographic findings. J Nerv Ment Dis 170:565–567.
Reynolds CF, Kupfer D, McEachran A, Taska L, Sewitch D,
Coble P (1984): Depressive psychopathology in male sleep
apneics. J Clin Psychiatry 45:287–289.
Reynolds CF, Kupfer DJ (1987): Sleep research in affective
illness: State of the art circa 1987. Sleep 10:199 –215.
Reynolds CF, Kupfer DJ, Thase ME, Frank E, Jarrett DB, Coble

Depression, Sleep Architecture, OSA

PA, et al (1990): Sleep, gender, and depression: An analysis
of gender effects on the electroencephalographic sleep of 302
depressed outpatients. Biol Psychiatry 28:673– 684.
Roehrs T, Zorick F, Wittig R, Conway W, Roth T (1989):
Predictors of objective level of daytime sleepiness with
sleep-related breathing disorders. Chest 95:1202–1206.
Schulberg HC, Saul M, McClelland M, Ganguli M, Christy W,
Frank R (1985): Assessing depression in primary medical and
psychiatric practices. Arch Gen Psychiatry 42:1164 –1170.
Series F, Series I, Cormier Y (1992): Effects of enhancing
slow-wave sleep by gamma-hydroxybutyrate on obstructive
sleep apnea. Am Rev Respir Dis 145:1378 –1383.
Smith PL, Gold AR, Meyers DA, Haponik EF, Bleecker ER
(1985): Weight loss in mildly to moderately obese patients
with obstructive sleep apnea. Ann Intern Med 103:850 –
855.
Stewart DA, Grunstein RR, Berthon-Jones M, Handelsman

BIOL PSYCHIATRY
2000;48:1001–1009

1009

DJ, Sullivan CE (1992): Androgen blockade does not
affect sleep-disordered breathing or chemosensitivity in
men with obstructive sleep apnea. Am Rev Respir Dis
146:1389 –1393.
Thase ME, Fasiczka AL, Berman SR, Simons AD, Reynolds CF
(1998): Electroencephalographic sleep profiles before and
after cognitive behavior therapy of depression. Arch Gen
Psychiatry 55:138 –144.
Weitzman E, Kahn E, Pollack CP (1980): Quantitative analysis
of sleep and sleep apnea before and after tracheostomy in
patients with hypersomnia-sleep apnea syndrome. Sleep
3:407– 423.
White DP (1992): Obstructive sleep apnea. Hosp Pract (Off Ed)
27(5A):57– 84.
Young T, Palta M, Dempsey J, Skatrud J, Weber S, Badr S
(1993): The occurrence of sleep-disordered breathing among
middle-aged adults. N Engl J Med 328:1230 –1235.