Hypoxemia vs Sleep Fragmentation as
Cause of Excessive Daytime Sleepiness in
Obstructive Sleep Apnea*
Henri G. Colt, M.D.;t Helmut Haas, M.D., F.C.C.R;+ and
Gerald B. Rich, M.D.§
To determine the effects of intermittent hypoxemia on
daytime sleepiness in the clinical setting of obstructive
sleep apnea syndrome, we enrolled seven patients in a
prospective, randomized, crossover study. We had two
experimental conditions with NCPAP treatment as follow:
(1) to correct apneas, sleep fragmentation, and hyoxemia;
and (2) to correct apneas and sleep fragmentation and at
the same time, induce intermittent hypoxemia. The out-
come variable, daytime sleepiness, was measured objec-
tively with the multiple sleep latency test following comple-
tion of baseline and each treatment condition. Compared
with sleep latencies in the untreated condition, both exper-
imental treatment arms prolonged sleep latencies (p<O.05).
We found DO statistically significant differences between
mean MSLT scores obtained after NCPAP treatment under
E xcessive daytime sleepiness is a symptom common
to patients with severe obstructive sleep apnea.
The consequences ofEDS are significant deterioration
of the psychosocial and cognitive functions of individ-
uals with OSA.l In addition to its impact on quality of
life, EDS increases the risk for industrial and motor
vehicle accidents, 2 decreases work capaci~ and fre-
quently leads to adverse psychosocial consequences. 3
Excessive sleepiness is a common cause for consulta-
tion in centers for sleep disorders." Apnealhypopnea-
related EDS is usually reversible with appropriate
nasal continuous positive airway pressure treatment.5-7
The mechanisms underlying EDS are unclear. The
following two hypotheses have been proposed to
explain the sleepiness associated with the obstructive
sleep apnea syndrome: sleep fragmentation8-l0 and
hypoxemia. ll - l2 Because both sleep fragmentation and
hypoxemia are statistically related to EDS in sleep
apnea patients, separating the effects of one from the
other on daytime sleepiness requires independent
study ofeach condition. With this in mind, we studied
sleep apnea-related EDS before and after NCPAP
*From the Section of Pulmonary and Critical Care Medicine,
Department of Veterans Affairs Medical Center, and the Depart-
ment of Medicine, Oregon Health Sciences University, Portland.
This study was supported in part by the American Lung Association
of Oregon.
tFonner Fellow. Currently at the Centre Laser du CHU Sud,
Marseille, France.
*Associate Professor.
§Clinical Assistant Professor of Neurology.
Manuscript received December 3; revision accepted March 18.
1542
hypoxemic and oonhypoxemic conditions. In summary, two
nights of iDtermittent noctumaI hypoxemia during NCPAP
treatment for OSAS did not diminish the objective improve-
ment in daytime somnolence seen with NCPAP treatment
in the absence of noctumaI hypoxemia. Results lend further
support to the hypothesis relating excessive daytime sleep-
iness to sleep fragmentation. (Chat 1991; lOO:l54J-48)
AlII = apnea-hypopnea index; BMI = body mass indeX; EDS =
=
excesdve daytime sleepiness; EOG electrooculogram;
MCST=mean continuous sleep time; MSLT=multip)e sleep
=
latency test; NCPAP nasal continuous positive airway J)res-
sure; OSA = obstructive sleep apnea; OSAS = obstructive sleep
apnea syndrome; PLM = periodic leg movements; PSG = polY-
=
sonmogram; BAI respiratory arousal index; SE sleep edi-=
ciency; TST = total sleep time
treatment designed to either eliminate OSA episodes,
sleep fragmentation, and hypoxemia or eliminate OSA
episodes and sleep fragmentation and at the same
time produce repetitive arterial oxygen desaturations.
We used the multiple sleep latency test (MSLT) scores
as the outcome variable to measure physiologic sleep
tendency before and after each treatment condition.
METHODS
Study Population
Seven male patients from 31 to 68 years of age were enrolled in
this prospective, randomized, crossover study. Clinical features and
pulmonary function data are summarized in Table 1. Individuals
were selected for screening from a list of patients waiting to be
studied at the Portland VA Medical Center Sleep Laboratory.
Patients completed a questionnaire detailing the history and symp-
toms of their sleep disturbances and were interviewed by the same
investigator (HGC). Criteria for exclusion from the study were as
follow: (1) age greater than 70 years; (2) a history of insomnia,
seizure disorder, stroke, congestive heart failure, angina pectoris,
or unstable cardiac rhythm; (3) discovery of narcolepsy or other
reasons for daytime sleepiness besides OSA; (4) arterial CO2 tension
greater than 50 mm Hg. Of 40 patients interviewed, only 11 were
enrolled in the study. All were in stable condition and were not
receiving sedatives or psychotropic medications. Diuretic, antihy-
pertensive, and bronchodilator drugs were allowed prior to poly-
somnography if clinically indicated. The research protocol was
approved by the institutional subcommittee on human studies. Four
patients were subsequently excluded: one had an abnonnal noctur-
nal EEG and required neuropsychiatric evaluation, one showed no
evidence of EDS on the MSLT after the second-night baseline
polysomnogram, one left town without notice before study began,
and one could not tolerate NCPAP because it produced otalgia.
ExcessNe Daytime Sleepiness in OSA (Colt, Haas, RIch)
 Table l-Baeline ArteritJl Blood G68 and Ventilatory Function Data

Age, Weight, 8MI, PaOj , PaCO., . Seo l , FEV 1.o, FEV/FVC,

Patient yr kg kglm l mmHg mm~~. 'II FEV.,L ,.PRD %PRD
•• "i ? ..
~

1 31 147 52 72 45 90 2.62 71 103

2 38 161 46 75 41 95 4.61 114 104
3 41 150 45 85 40 fT1 3.68 95 107
4 58 125 38 64 47 92 2.42 71 96
5 59 73 26 65 41 95 3.26 116 102
6 65 126 37 63 46 89 1.67 47 65
7 68 95 29 94 39 96 2.59 91 101
Mean 51 125 39 74 43 93 2.fT1
SD 14 32 8.7

*% PRD, percent of predicted normal value .16

Study Protocol
Patients were hospitalized for a total of six nights (seven days).
Resting pulmonary function tests and arterial blood gas analysis
were obtained. After ascertaining the severity of sleep apnea by
means of two baseline polysomnograms, patients were studied
under two experimental conditions: condition 1 corrected apneas,
sleep fragmentation and hypoxemia by treating obstructive sleep
apnea with NCPAP; condition 2 corrected apneas and sleep
fragmentation with NCPAP but induced repeated desaturations by
the intermittent addition of 100 percent nitrogen gas to air. The
order of treatment conditions was determined from a table of
random numbers. A washout period of three nights separated the
two experimental conditions. Thus, polysomnography was per-
formed on six nights with baseline study and each experimental
condition requiring two all-night polysomnograms in order to avoid
the first night effect. Severity of daytime sleepiness was measured
by MSLT on the day following the second night of the baseline
polysomnogram and the second night of each experimental condi-
tion.
Procedures
Polysomnography: Standard nocturnal polysomnogram included
continuous monitoring of central and occipital electroencephalo-
grams, electrooculogram, submental and anterior tibialis electro-
myograms, and electrocardiogram using conventional leads. Airflow
was monitored with oral and nasal thermistors, and respiratory
effort was measured by respiratory inductance plethysmography
with transducers placed around the abdomen and chest. Oxyhe-
moglobin saturation was continuously recorded with a pulse oxim-
eter set on its fastest response. All variables were continuously
recorded by a sleep system (CNS).
Measures of Sleep \fJriables: Nocturnal polysomnograms were
compute~scoredfor sleep stages using sleep system software version
4.30. The default analysis of the raw data was reviewed and corrected
in preparation for the final analysis by the system. Amplifiers were
used for signal conditioning. The amplified signals were input to
the sleep system for data acquisition-processing and analysis.
Sleep onset was measured from the onset of the test until two
consecutive nonwake epochs of sleep were recorded. All staging
was performed in 3O-s epochs. The methodology employed by the
sleep system allowed the human scorer to set levels for individual
subjects discriminating awake from sleep in addition to levels for
slow wave sleep and stage REM using the criteria outlined by
Rechtschaffen and Kales. 13
Oxyhemoglobin desaturations were detected if there was a local
drop in blood oxygen saturation equal to or greater than 4 percent
lasting 8 s or more. Desaturations below 30 percent were categorized
as artifact in addition to any desaturation where the rate of the fall
exceeded 10 percent per 2 s.
Apnea detection required a signal amplification lower than 3
percent of the full scale deflection with a duration of at least 10 s.
To detect the effort component of a mixed or central event, it was
required that the signal amplitude be less than 5 percent full scale
deflection, also with a duration of at least 10 s. Hypopneas were
detected if there was at least a 50 percent reduction in airflow
relative to a local baseline coupled with a desaturation of at least 4
percent. Hypopnea duration requirement also was at least 10 s.
The EEG arousals were detected in relation to respiratory events
and periodic leg movements in addition to those occurring sponta-
neously. For arousals related to respiratory events and periodic leg
movements, a baselin~ period of 10 s was used prior to the event in
question followed by an active comparison period of 10 s after the
event lasting 10 s. In the case of PLM, an additional active period
of 7 s was used just prior to the leg movement. For an arousal to be
scored, a 75 percent increase in the power of alpha plus beta, a 75
percent increase in total power, or the presence of a K-eomplex in
the active period was required. Spontaneous arousals were detected
using a 15-s sliding window with a baseline and active comparison
time segment. A 200 percent increase in alpha plus beta power or
a 200 percent increase in total power was required. Overlapping
arousals were eliminated in the analysis.
Experimental Conditions: Condition 1 corrected nocturnal hy-
poxemia and sleep fragmentation by using standard procedures of
NCPAP therapy with additional monitoring of pressures at the
sampling port of the NCPAP mask with a companion manometer.
A NCPAP system (Sleep Easy II, Respironics, Inc, Monroeville,
PA) was used. Pressures were adjusted with a remote valve through
a range of 0 to 18 em ~O pressure.
Condition 2 produced intermittent nocturnal hypoxemia while
correcting sleep fragmentation and apneic events. The NCPAP
application was identical in both experimental conditions. For
condition 2, nitrogen gas was administered intermittently through
an input port located 5 em below the exhalation valve on the NCPAP
mask. In order to achieve oxygen desaturations similar to those
which occurred spontaneously during baseline polysomnography, a
suitable flow rate was determined by sleep technicians who manu-
ally controlled the duration and quantity of nitrogen gas delivered
by means of a solenoid valve. Nitrogen administration ceased when
the oximeter indicated the desired desaturation. One sampling port
on the NCPAP mask was reserved for 100 percent oxygen admin-
istration in the event of an unexpectedly severe or protracted
hypoxemia. System oxygen concentrations were continuously mon-
itored with an oxygen analyzer (Instrumentation Laboratories,
Lexington, Mass).
Measurements of Excessive Daytime Sleepiness: The MSLTs were
performed in the sleep laboratory the day following the second
night baseline PSG and after the second night PSG of each
experimental condition. After being instructed to lie on either a
bed or in a reclining chair, patients were asked to go to sleep
without NCP~ Tests were performed at 10:00 AM, noon, 2:00 PM,
and 4:00 PM. Standard EEG, EOG, EMG, and ECG were contin-
uously recorded (Gould MultiChannel Recorder 28OOS, Cleveland).
Patients were monitored by videocamera and microphone and were
observed on television screen. The MSLTs were hand-scored. The

CHEST I 100 I 8 I DECEMBER, 1991 1543

 Table 2-~ and HypopnetUI Auociated with three-way comparisons were significant, posthoc paired Student's
Each Experimental Condition· t-tests were used to compare groups. A probability value of less
than 0.05 was accepted as statistically significant.
Paired Differences
Student's t-Tests, RESULTS
Means SD p Values
Baseline data including body weight, body mass
02 6.3 11.9 index, and results of spirometry are shown in Table 1.
0.707
01 4.9 4.5
For all 7 patients, the BMI exceeded the upper limit
0A2 1.6 2.6
OAI 1.4 1.3
0.706 of accepted normal range (25). Two patients were
M2 23.9 57.0 overweight (BMI 25 to 30), and the remaining five
0.325
Ml 0.7 1.9 were obese (BMI >30). Although the patients were
MA2 12.0 30.4 mildly hypoxemic at rest for their age, only patient 6
0.349
MAl 0.3 0.8
had severe ventilatory dysfunction by spirometric
C2 117.0 111.8
CI 21.9 47.9
0.102 standards16 and arterial blood gas values.
CA2 35.1 39.6 Mean MSLT scores and sleep stage measurements
0.075
CAl 2.4 4.0 are summarized in Table 3. At the beginning of the
H2 67.7 88.3 study, patients were excessively sleepy as indicated by
0.078
HI 8.7 17.0
a mean sleep latency of 2.7 min. Sleep was primarily
HA2 20.0 26.7
HAl 3.7 8.2
0.072 stage 2 and stage 1 with very little stage REM. The
time spent in all sleep stages except stage 4 improved
*Cumulative data from two polysomnograms for each of two
as the mean sleep latency increased to 5.2 min
experimental conditions. 0, obstructive apnea; M, mixed apnea;
C, central apnea; A, arousal; H, hypopnea; 1, NCPAP treatment (p=O.09) after NCPAP application (without addition
without induced hypoxemia; 2, NCPAP treatment with induced of nitrogen gas) for two consecutive nights. When
hypoxemia. results of NCPAP with and without the induction of
hypoxemia were compared, no statistically significant
differences were noted in the percentage of sleep for
MSLT reader was blinded to the patient's identity, clinical history, different sleep stages (except for a small decrease in
and order of randomization. We used a modification of the standard stage 2 under condition 2), TST, or mean latency to
technique for measuring daytime sleepiness. If Each MSLT was
sleep onset. Sleep efficiency increased from 81 percent
terminated when sleep onset was reached (defined by three
consecutive 3O-s epochs of stage 1 sleep or any single epoch of stage
at baseline to 91 percent in condition 1 and to 84
2, 3, 4, or REM sleep), or if sleep was not attained within 20 min. percent in condition 2. The differences in sleep
A value of 20 min was given if no sleep occurred; otherwise, the efficiency between the two experimental conditions
time value was that from lights out to the first epoch of sleep onset. did not reach statistical significance.
The mean continuous sleep latency for the day was calculated by
Results of measures of nocturnal Sa02 are summa-
averaging the individual latency scores of the four MSLTs. Severe
EDS was defined as a mean latency to sleep onset of less than or
rized in Table 4. Patients had significant nocturnal
equal to 5 min. 15 hypoxemia during baseline polysomnography with 395
(mean) desaturations below Sa02 of 90 percent. An
Statistics average of 119 min were spent below SaO! of 85
The data presented are from second night polysomnograms percent. These measures improved strikingly with the
obtained at baseline and each of the two experimental conditions. application of NCPAP alone. When nitrogen was
The data in Table 2 are based on two successive all-night polysom-
nograms obtained for each experimental condition. Analysis of
added to the NCPAP mask, the degree of nocturnal
variance for repeated measures was used to test for significant hypoxemia obtained was similar to that which occurred
differences among mean scores in the three conditions. When the spontaneously during baseline studies. This is dem-

Table 3-Sleep Stage MetJBUrementB and MSLT at Baeline and Each Erperimental Condition·
Post-hoc p Values
Baseline Condition 1 Condition 2
Variable Study CPAP CPAP+N2 BvsCI BvsC2 Cl vs C2

TST, min 343 (56) 385 (37) 363 (45) 0.01 ns ns

% Stage 1 11 (5) 6 (5) 6 (3) 0.006 0.006 ns
% Stage 2 78 (9) 69 (9) 61 (8) 0.02 0.0006 0.04
% Stage 3 1 (3) 7 (6) 10 (4) 0.04 0.004 its
'*'
Stage 4 0.4 (.1) 0.3 (1) 0.1 (.2) ns ns ns
% Stage REM 6 (5) 17 (8) 23 (7) 0.02 0.0005 ns
'MSLT,
*' SE min 81 (10)
2.7 (1.7)
91 (6)
5.2 (2.1)
84 (8)
6.2 (3.6)
0.01
0.045
ns
0.008
os
ns

*Results of second night study expressed as mean with SD in parentheses.

1544 Excessive Daytime Sleepiness in OSA (Colt, Haas, Rich)

 Table 4-Sleep Related Oxygenation Meamres at Baseline and Each Experimental Condition·

Post-hoc p Values
Baseline Condition 1 Condition 2
Shldy CPAI' CpAp+N2 B vsCI B vs C2 Cl vs C2

No. desat 458 (257) 7 (9) 205 (184) 0.0002 0.01 0.03
Av duration of desat <90%. s 22 (4) 19 (25) 35 (12) ns ns ns
No. desat <90% 395 (281) 4 (8) 190 (179) 0.0009 0.03 0.05
TIme <90%. min 163 (167) 38 (95) 175 (139) 0.003 ns 0.001
TIme <85%. min 119 (171) 0.4 (1) 119 (143) 0.04 ns 0.04
*Results of second night study expressed as mean with SD in parentheses.

Table 5-Respiratory Measures During Sleep at Baseline and Each Experimental Condition·

Post-hoc p Values
Baseline Condition 1 Condition 2
Study CPAI' CpAp+N2 BvsCI B vs C2 Cl vs C2

No. apnealhypopnea events 427 (202) 18 (36) 127 (133) 0.‫סס‬OO 0.0004 ns
Apnealhypopnea. h 73 (26) 3.2 (7) 21 (20) 0.‫סס‬OO 0.0001 os
No. respir arousals 220 (105) 1.7 (2) 47 (50) 0.‫סס‬OO 0.0001 liS
Respir arousal. h 37 (13) 0.2 (.3) 8 (8) 0.‫סס‬OO 0.‫סס‬OO liS
Total No. arousals. h 50 (13) 7.3 (5) 14 (5) 0.‫סס‬OO 0.‫סס‬OO os
MCST. min 1.3 (.3) 12.3 (8.3) 4.22 (1.5) 0.0007 ns 0.006

*Results of second night study expressed as mean with SD in parentheses.

onstrated by the lack of statistically significant differ- abnormal sleep architecture with sleep fragmentation
ences between baseline and experimental condition 2 and nocturnal hypoxemia. Two nights of treatment
in regard to the number of minutes spent below Sa02 with NCPAP greatly improved the sleep architecture
of90 percent and 85 percent, and in the mean duration and mean MSLT scores. When nitrogen was admin-
of each desaturation. istered, the severity of hypoxemia was similar to that
Measures of respiratory disturbances during sleep which existed during baseline studies.
are shown in Table 5. Patients had a baseline apnea- Results of MSLTs are represented in Figure 1. Bar
hypopnea index of 73 and a respiratory arousal index graphs are in order of randomization. The order of
of 37. Except for the mean continuous sleep time study after baseline PSG proceeded from experimen-
under condition 2, all measured variables improved tal condition 1 to condition 2 for patients 1 and 4, and
significantly after NCPAP application. When nitrogen from condition 2 to condition 1 for all others. The
gas was added to NCPAp, the severity of respiratory results of post-hoc Student's (-tests (Table 3) demon-
disturbance increased somewhat compared with strate significant differences in mean MSLT scores be-
NCPAP alone, but the differences were not statistically tween baseline and NCPAP alone (p = 0.045) and be-
significant except for the MCST. tween baseline and NCPAP with nitrogen (p=0.OO8).
The combined results entered in Tables 4 and 5 Mean MSLT scores were similar, however, for exper-
show that before NCPAP treatment, patients had imental conditions 1 and 2, and the differences be-
14 tween them were not statistically significant.
o Baseflne
12 • Condition 1 (NCPAP alone) DISCUSSION
!:l Condition 2 (NCPAP and N2)
10
The mechanism underlying EDS in patients with
sleep apnea is unclear. The following two hypotheses
~
:i 8
have been offered to explain daytime sleepiness: (1)
sleep fragmentation; and (2) hypoxemia. Because sleep
.S
~ 6 fragmentation and hypoxemia are correlated, both
C/)
~
seem to contribute to daytime sleepiness in patients
4
with obstructive sleep apnea. 17 • 19 We studied the
effects of intermittent nocturnal hypoxemia on phys-
2
iologic sleepiness. To our knowledge, this is the first
o study designed to produce hypoxemia, concurrently
2 3 4 5 6 7 correct sleep apnea and sleep fragmentation with
Patients NCPAp' and measure the response variable objec-
FIGURE 1. Results of MSLTs. tively. The main finding of our study is that intermit-

CHEST I 100 I 6 I DECEMBER. 1991 1545

tently induced hypoxemia per se does not diminish
the beneficial effect of NCPAP on EDS as measured
by MSLT. Thus, our results do not support the
hypothesis which states that apnea-related hypoxemia
by itself leads to daytime sleepiness.
Several investigators have studied the effects of
nocturnal hypoxemia and sleep fragmentation on
daytime somnolence in patients with OSA.lG-12.lS.l7-l9
Orr et al ll compared four hypersomnolent with four
asymptomatic sleep apnea patients. Both groups of
patients had a similar number of apneas and mean
CO2 tension of 39 mm Hg during sleep. The only
difference between the two groups was the lower
arterial oxygen tension on waking (54 vs 80 mm Hg)
and during sleep (47 vs 70 mm Hg) in the symptomatic
group. The authors concluded EDS is produced by
hypoxemia. Because sleepiness was judged by subjec-
tive complaints which do not always accurately reflect
the physiologic state of sleepiness, iD their conclusion
is not definitive. Sink et all! studied 37 elderly patients
with OSA and compared two subgroups of patients,
one reporting excessive sleepiness, the other being
asymptomatic. Based on discriminant analysis of oxy-
gen saturation during sleep and the number of apneas
and hypopneas, the two patient groups were discrim-
inated by two hypoxemia variables (Sa02 <90 percent
and SaO. <80 percent). The authors concluded that
somnolent patients are characterized by more severe
oxygen desaturations. Because sleepiness was judged
by subjective complaints, and data analysis did not
include the total number of arousals, their conclusion
cannot be accepted as definitive. Guilleminault et al lS
studied 100 patients with obstructive sleep apnea
syndrome. The authors measured daytime sleepiness
objectively and applied regression and cluster analyses
to assess the determinants ofEDS. This study did not
find any statistically significant correlates between
polysomnographic measures of sleep (amount of stage
1 sleep, respiratory disturbances, and oxygen satura-
tion) and daytime sleepiness. The authors concluded
the best predictors of EDS are those reflecting sleep
disturbances and cc sleep structure anomalies." Roehrs
et al lO recently studied 466 patients with OSAS by
means of polysomnography followed by MSLT in an
attempt to find predictors of EDS. Correlation and
multiple regression analyses were performed on meas-
ures ofhypoxemia and sleep fragmentation. The MSLT
score correlated best with the RAJ which measures
the number of arousals from sleep associated with
respiratory disturbances. Measures of hypoxemia
added little or no independent information towards
explaining the variants in EDS not accounted for by
measures of sleep fragmentation. The authors con-
cluded that sleep fragmentation is an important de-
terminant of daytime sleepiness in patients with
apnea. In our study; intermittent hypoxemia during
1548
sleep resembling that observed in the untreated sleep
apnea state did not by itself cause increased daytime
sleepiness when apneas were eliminated by NCPAE
This supports previous work relating EDS to various
measures of sleep disruption rather than to oxygen
desaturation per se.
Our research protocol was not designed to address
the question of whether or not severe waking hypoxe-
mia or extreme and repetitive oxygen desaturations
during sleep might lead to EDS. That low oxygen
tensions present during both day and night separated
those patients with OSA who had subjective sleepiness
from those that did not,11 suggests a possible role for
hypoxemia in the causation of EDS in a subgroup of
apnea patients. We have no way of knowing if the
outcome of our study would have been different had
we selected patients with more severe daytime hy-
poxemia. An unexpected finding was the occurrence
ofhypopneas and central apneas in response to boluses
of nitrogen gas given through the input port of the
NCPAP mask (Table 5). The majority of these respira-
tory events were not associated with arousals or
reduced sleep latencies. It is noteworthy that induced
progressive eucapnic hypoxemia as low as 70 percent
Sa02 (Po2 =37.0 mm Hg) is a poor arousal stimulus in
healthy adults, both in NREM and REM sleep. 21
Although our data show a trend towards more apneasl
hypopneas with the induction of intermittent hypoxe-
mia than without it, the differences in apneas, hypo-
pneas, or arousals between the two treatment condi-
tions did not achieve statistical significance. Further
studies along these lines would be of interest to
elucidate the mechanism and significance of these
observations. A recent paper by Orr et alii reports no
apparent relationship between chronic daytime hy-
poxemia and subjective and objective daytime sleepi-
ness in patients with COPD. Clearly, hypoxemia by
itself is not a sufficient condition leading to EDS in
COPD patients; on the other hand, this does not
necessarily imply that hypoxemia can have no effect
on EDS of sleep apnea. The question can be raised
whether acute decreases in oxyhemoglobin saturation
rather than absolute levels of daytime or nocturnal
hypoxemia could be linked to the complaint of EDS.
Our results show that acute intermittent falls in Sa02
during sleep in the absence of significant sleep frag-
mentation do not lead to increased daytime sleepiness.
Ifhypoxemia alone does not invariably lead to daytime
sleepiness, it may do so indirectly by inducing sleep
fragmentation. Both oxygenation and sleep disruption
are statistically related to EDS.17 The degree of
hypoxemia is highly correlated with measures of sleep
fragmentation, especially with frequency of respira-
tory events rather than measures of sleep architec-
ture .18 These correlations assume a new meaning in
light ofrecently published data which show that airway
Excessive Daytime Sleepiness In OSA (Colt, Haas, Rich)
occlusion, hypoxia, or hypercapnia can produce
arousal from sleep at similar levels of ~-negative
esophageal pressure. 23 Even though arousal occurs at
different levels ofventilation and Sa02~~tlreventilatory
effort for each subject is similar at the point of arousal
no matter what the stimulus. If patients with sleep
apnea were to respond as did these healthy young
men, then it would follow that those patients mounting
a strong ventilatory effort in response to nocturnal
oxygen desaturation should experience more hypoxe-
mia-induced arousals, and therefore, EDS. We found
no statistically significant differences between meas-
ures of sleep architecture under normoxic and hypox-
emic NCPAP treatment conditions (Table 3), and thus,
are in agreement with previous work;18 on the other
hand, no statistically significant differences between
respiratory measures during sleep under normoxic
and hypoxemic NCPAP treatment conditions were
noted (Table 5). Because in our study acute repetitive
oxygen desaturations produced neither numerous
arousals nor interference with NCPAP treatment in
terms of its effect on EDS, we are unable to answer
the question of whether hypoxemia causes EDS indi-
rectly by means of sleep fragmentation. Although our
patients frequently expenenced desaturation below
85 percent (Table 4), an even more abrupt and severe
fall in Sa02 conceivably could have had a more
disruptive effect on sleep. Finally, if hypoxemia di-
rectly contributes to daytime sleepiness in patients
with OSA, then correction ofhypoxemia should lessen
daytime sleepiness. In fact, a recent study found no
improvement in the mean MSLT score after nasal
oxygen was given for one month to patients with mild
OSA.24 We believe, therefore, that nocturnal hypoxe-
mia alone does not produce measurable daytime
sleepiness, but its potential for causing arousals from
sleep and thereby to contribute to EDS in a subset of
obstructive apnea patients cannot be denied.
Our results might be criticized on the grounds that
insufficient time was given to each experimental
treatment condition to achieve its full impact as
regards EDS. Our protocol stipulated two successive
all-night PSGs with MSLTs on the following day for
baseline and each oftwo experimental conditions. This
length of time undoubtedly is suboptimal in terms of
the NCPAP treatment effect on sleepiness. However,
prolonged hospitalization was impractical, and treat-
ment at home would have required a fail-safe device
for intermittent induction of hypoxemia during sleep.
To our knowledge, no such FDA-approved device
exists. Since the beginnings of NCPAP treatment in
clinical medicine, physicians have noticed that one
night of treatment reverses OSA and its attendant
sleepiness. IS This impression has been confirmed by
recent studies,«~·7 Furthermore, a paper published
during our investigation found that MSLT scores
improved significantly after one night and further
improved significantly after 14 nights; there was no
additional improvement after 42 nights. 5 Certainly,
our data demonstrate that treatment with NCPAP for
two consecutive nights is adequate for the detection
of measurable improvement in MSLT scores. On the
other hand, the time course of potentially adverse
effects from nocturnal hypoxemia on daytime alertness
is unknown, and the adverse effects from hypoxemia
might be cumulative. Whether the improvements in
MSLT scores under hypoxemic NCPAP conditions
observed by us would have continued to parallel the
findings under normoxic NCPAP conditions over an
extended period of time is unknown.
We conclude that patients with OSA show similar
improvements in MSLT scores after two nights of
treatment with NCPAP whether or not they are
exposed to intermittent nocturnal hypoxemia. A cross-
over NCPAP trial with and without nocturnal hypoxe-
mia carried out for several weeks would answer the
question of whether or not prolonged nocturnal hy-
poxemia alone leads to EDS in patients with sleep
apnea. Under present conditions, such a study would
be difficult to perform because of a lack of suitable
technical equipment for use outside the hospital or
excessive financial cost if done in hospital.
ACKNOWLEDGMENTS: We thank Larry Lauritzen and Steve
Munoz for their technical assistance, Dr. David H. Hickam for
statistical analysis, and Theresa Ciullo for secretarial assistance.
REFERENCES
1 Roth T, Roehrs TA, Conway WA. Behavioral morbidity ofapnea.
Semin Res Med 1988; 9:554-59
2 Findley LJ, Unverzagt ME, Suratt PM. Automobile accidents
involving patients with obstructive sleep apnea. Am Rev Respir
Dis 1988; 138:337-40
3 Kales A, Caldwell AD, Cadieux RJ, Vela-Bueno A, Ruch LG,
Mayes SD. Severe obstructive sleep apnea-II: associated psy-
chopathology and psychosocial consequences. J Chron Dis 1985;
38:427-34
4 Guilleminault C, Dement WC. Sleep apnea syndromes and
related sleep disorders. In: Williams RL, Karacan I, Moore CA,
eds. Sleep disorders: diagnosis and treatment. New York: John
Wile~ 1988:47-71
5 Lamphere J, Roehrs 'I: Wittig R, Zorick F, Conway WA, Roth L
Recovery ofalertness after CPAP in apnea. Chest 1989; 96:1364-
67
6 Rajagopal KR, Bennett LL, Dillard TA, Tellis CJ, Tenholder
MF. Overnight nasal CPAP improves hypersomnolence in sleep
apnea. Chest 1986; 90:172-76
7 Wittig RM, Conway WA, Zorick F, Sicklesteel J, Roehrs T, Roth
T. CPAP: reduction in daytime sleepiness after one night's use
(Abstract). Sleep Res 1987; 16:459
8 Roth T, Hartse KM, Zorick F, Conway W Multiple naps and
the evaluation of daytime sleepiness in patients with upper
airway sleep apnea. Sleep 1980; 3:425-39
9 Stepanski E, Lamphere J, Badia ~ Zorick F, Roth 1: Sleep
fragmentation and daytime sleepiness. Sleep 1984; 7:18-26
10 Roehrs T, Zorick F, Wittig R, Conway ~ 80th T. Predictors of
objective level of daytime sleepiness in patients with sleep-
related breathing disorders. Chest 1989; 95:1202-06
CHEST I 100 I 6 I DECEMBER. 1991 1547
11 Orr WC, Martin RJ, Imes NK, Rogers RM, Stahl ML. Hyper-
somnolent and nonhypersomnolent patients with upper airway
obstruction during sleep. Chest 1979; 75:418-22
12 Sink J, Bliwise DL, Dement WC. Self-reported excessive
daytime somnolence and impaired respiration in sleep. Chest
1986; 90:177-80
13 Bechtscbaft'en A, ICales A, eels. A manual of standardized
termin~ techniques and scoring system for sleep stages of
human subjects. Los Angeles: Brain Information ServicelBrain
Besearch Institute, University of California at Los Angeles, 1968
14 Carsbdon MA, Dement WC, Mittler MM, Roth T, Westbrook
PR, Keenan S. Guidelines for the multiple sleep latency test
(MSLT): a standard measure of sleepiness. Sleep 1986; 9:519-24
15 GuilIemiDault C, Partinen M, Quera-Salva MA, Hayes B,
Dement WC, Nino-Murcia G. Determinants of daytime sleep-
iness in obstructive sleep apnea. Chest 1988; 94:32-37
16 Morris JF, Koski A, Johnson LC. Spiromebic standards for
healthy nonsmolcingadults. Am Rev Respir Dis 1971; 103:57-67
17 TlIDDII RM, Shaforenko R, Hajdukovic RM, Mittler MM. Sleep
apnea syndrome: quantitative studies of nighttime measures and
daytime alertness (Abstract). Sleep Res 1985; 14:222
18 Rosenthal L, Moyles T, Zorick F, Conway ~ Sicldesteel J, Both
1: Hypoxemia and sleep fragmentation in sleep apnea [abstract].
Sleep Res 1988; 17:241
Plan to Attend ACCP's
58th Annual SCientific Assembly
cA
....-.--......~ --~
Chicago
OCtober 25-29, 1992
1141
19 Poceta JS, Jeong DU, HO S, TImms RM, Mittler MM.
Hypoxemia as a determinate of daytime sleepiness in obstructive
sleep apnea. Annual Meeting Abstracts, APPS, Minneapolis,
MN.I990:17
20 Zorick F, Roehrs T, Conway ~ Fujita 5, Wittig R, Roth 1:
EfFects ~ uvulopalatopharyngoplasty on the daytime sleepiness
associated with sleep apnea syndrome. Bull Eur Physiopath
Respir 1983; 19:600-03
21 Berthon-Jones M, Sullivan CE. ~ntilatory and arousal re-
sponses to hypoxia in sleeping humans. Am Rev Respir Dis
1982; 125:632-39
22 Orr WC, Sbamma-<>thman Z, Levin D, Othman J, Rundell
OH. Persistent hypoxemia and excessive daytime sleepiness in
chronic obstructive pulmonary disease (COPD). Chest 1990;
'i11:583-85
23 Gleeson Ie, Zwillich ~ White DE The influence of increasing
ventilatory eftOrt on arousal from sleep. Am Rev Bespir Dis
1990; 142:295-300
24 Phillips BA, Schmitt FA, Berry DTR, Lamb DC, Amin M,
Cook YR. Treatment of obstructive sleep apnea. Chest 1990;
98:325-30
25 Sullivan CE, Issa FG, Berthon-Jones M, Eves L. Reversal of
obstructive sleep apnoea by continuous positive airway pressure
applied through the nares. Lancet 1981; 1:862-65