Post on 14-May-2023
doi: 10.1111/j.1365-2869.2008.00660.x
Executive functions and cognitive subprocesses in patients with
obstructive sleep apnoea
STEFAN IE L I S 1 , S TEPHAN KR IEGER 1 , DOROTHEE HENN IG 1 ,
CHR I ST IAN R ODER 2 , P ETER K IRSCH 1 , WERNER SEEGER 3 ,
B ERND GALLHOFER 1 and R ICHARD SCHULZ 3
1Department of Psychiatry, Justus-Liebig University Giessen, Giessen, Germany, 2Department of Psychiatry, Erasmus MC Rotterdam,
Rotterdam, The Netherlands and 3Department of Internal Medicine, Justus-Liebig University Giessen, Giessen, Germany
Accepted in revised form 4 March 2008; received 11 September 2007
SUMMARY In recent years, special interest has been focused on impairments of executive functions
in patients with obstructive sleep apnoea syndrome (OSAS). However, the majority of
studies have not clearly separated deficits in executive functions from impairments in
other cognitive processes involved in task solving. In the present study, working memory
(WM) functions of 20 patients with OSAS were compared with those of 10 age-, sex- and
education-matched healthy subjects. Cognitive functions were measured four times a
day; each of these measurements was accompanied by an assessment of subjective and
objective daytime sleepiness. To separate dysfunctions of WM from those of
additionally involved processes, n-back tasks were applied embedded in a reaction-
time-decomposition approach. Deficits in n-back tasks could be observed in OSAS
patients in accuracy and reaction times. However, the slowing could already be observed
in simple reaction time tasks. The drop in 1-back accuracy in the morning was related to
daytime sleepiness. During the afternoon, accuracy of OSAS patients dropped in 2-back
tasks, an effect which correlated neither with sleepiness nor with the extent of sleep
apnoea or oxygen desaturation. In conclusion, our data reflect a complex perspective
upon cognitive deficits in OSAS. Cross-group differences in processing time on the
higher level WM task appeared to be attributable to slowing at a more elementary
cognitive processing level. In contrast, reduced accuracy during the WM task in the
OSAS group could not be explained by deficits in more elementary cognitive processes.
k e y w o r d s cognition, daytime sleepiness, executive function, obstructive sleep
apnoea, working memory
INTRODUCTION
The existence of cognitive impairments in patients with
obstructive sleep apnoea syndrome (OSAS) is well established.
Nevertheless, there is still an ongoing debate about the exact
nature of the cognitive dysfunction underlying the impair-
ments observable in various psychological and neuropsycho-
logical test procedures (Fulda and Schulz, 2003; Verstraeten,
2007). It is also unclear to which extent performance deficits
can be linked to daytime sleepiness caused by chronically
fragmented sleep or to recurrent hypoxaemia caused by
nocturnal apnoea episodes. The purpose of the present study
is to help determine the specific cognitive processes impaired in
OSAS patients and to contribute to the understanding of the
factors that influence these deficits.
Impaired performance in OSAS patients has been reported
in tests of various cognitive functions such as attention ⁄ vig-ilance, memory, psychomotor performance and executive
functioning (Aloia et al., 2004). In recent years, special interest
has focussed on impairments of executive functions because of
their connection to prefrontal cortex (PFC) activity (Beebe and
Gozal, 2002; Saunamaki and Jehkonen, 2007). The PFC is a
Correspondence: Dr. Stefanie Lis, Centre for Psychiatry, Justus-Liebig-
University Giessen, Am Steg22, 35385 Giessen, Germany. Tel.: 0049-
641-9945775; fax: 0049-641-9945789; e-mail: Stefanie.Lis@
psychiat.med.uni-giessen.de
J. Sleep Res. (2008) 17, 271–280 Sleep disordered breathing
� 2008 European Sleep Research Society 271
brain area that may be especially affected by chemical and
structural central nervous system cellular injury resulting from
sleep disruption and blood gas abnormalities and known to
prevent sleep-related restorative processes.
Results of studies on executive functioning in OSAS are
heterogeneous, i.e. some of these studies here suggested
executive dysfunction (Bedard et al., 1991; Feuerstein et al.,
1997; Naegele et al., 1995) whereas others have not (Kim
et al., 1997; Lee et al., 1999; Redline et al., 1997; Verstraeten
et al., 2004). In many studies which indicate executive
dysfunctions, performance deficits are reported not only in
executive tests but in all examined cognitive abilities, i.e. in all
test procedures applied (e.g. Bedard et al., 1991; Feuerstein
et al., 1997; Naegele et al., 1995). This could be interpreted as
a global cognitive impairment which affects all domains of
cognitive functioning. Alternatively, it might point to a
selectively disturbed more elementary cognitive subfunction
involved in all test procedures. Which of these alternative
interpretations is correct is difficult to determine because most
of the usually applied (neuro-) psychometric tests have been
developed primarily to discriminate between subjects. They do
not allow an estimation of the relative contribution of single
cognitive functions to the total outcome. Therefore, they are
not suited to identify the dysfunction of a single cognitive
function as the cause underlying the impaired performance
related to a specific disorder (Krieger et al., 2001). It also holds
true for the usage of test batteries that measure different
cognitive functions by using different tests. Generally, these
tests differ not only in regard to their target function but also
with respect to difficulty, reliability and additionally involved
cognitive subfunctions. This limits the validity of direct
comparisons of performance levels across subtests (see Chap-
man and Chapman, 1978; Goldberg and Gold, 1995). Only few
tests exist which try to isolate different functions, e.g. the Digit
Span Test (see Verstraeten and Cluydts, 2004). When these
were applied, no disturbance of executive functions in OSAS
patients could be observed (Verstraeten et al., 2004).
Because of these methodological problems, Beebe and Gozal
(2002) emphasized in the Journal of Sleep Research that
�strong research in this area involves dissociating effects on
executive function tests (which are expected to be significant)
from those obtained on tests of more basic skills (which are
expected to be negligible)’ (page 6).
A method meeting this requirement is the subtraction
method first introduced by Donders (1868). It involves a series
of measurement arrangements. Each setting differs from a
companion setting only in that it requires one additional
cognitive process for task solving. It is assumed that each
subprocess requires time and thus increases processing time.
This permits an estimation of the time demands of the
additionally involved process by simply subtracting the reac-
tion times (RTs) between pairs of tasks. This RT decompo-
sition approach is based upon the assumption of the existence
of discrete and distinguishable subprocesses within a serial
structure of processing. Although this assumption has led to
criticism by some authors in the past, this model emerged as a
parsimonious and heuristically useful approach towards cog-
nitive processing and the identification of selectively disturbed
cognitive subfunctions e.g. in psychiatric disorders (Krieger
et al., 2001, 2005).
The first aim of the present study was to prove whether this
approach might contribute to the understanding of executive
dysfunctions in OSAS. For this purpose, working memory
(WM) processes as an example for one domain of executive
functions were investigated with n-back tasks (Gevins et al.,
1990; Verstraeten and Cluydts, 2004). The n-back tasks require
subjects to compare the present stimulus with the one
presented n-stimuli back. They are assumed to involve WM
functions by maintaining stimulus information until the
response can be executed and allow for a parametric modu-
lation of WM load by increasing n. Thereby, demands upon
maintaining, monitoring and updating the WM buffer, i.e.
holding the temporal structure of the task relevant information
online, increase. The n-back WM paradigm has in recent years
emerged as one of the most highly used methods to study the
pathophysiology of WM dysfunction in psychiatric disorders
like schizophrenia (Glahn et al., 2005). The involvement of
dorsolateral PFC area was shown in several brain imaging
studies (Owen et al., 2005).
We embedded n-back tasks in a RT decomposition
approach to allow the differentiation of deficits in WM from
those in elementary cognitive subprocesses such as sensory
transduction, pattern integration, stimulus discrimination and
response selection as well as motor preparation and execution
(Massaro, 1990; Sanders, 1980). If cognitive dysfunctions in
OSAS patients can be attributed to WM deficits, a poorer
performance relative to controls is expected on n-back tasks
compared with control tasks which do not require WM
processes for task solving. Statistically, this should become
obvious in variance analytical approaches as interaction effects
involving the factors �group� (i.e. OSAS patients and control
subjects) and �type of task� (i.e. tasks requiring ⁄not requiringWM functions) (see Krieger et al., 2001; Verstraeten, 2007).
An increase of WM demands is expected to lead to an
accentuation of the performance deficit in OSAS patients
compared with healthy subjects.
The second aim of the study was to analyse whether
performance deficits of OSAS patients can be linked to
increased daytime sleepiness. Most studies analysing the
relationship between daytime sleepiness and cognitive altera-
tions in OSAS assess sleepiness by use of questionnaires like
the Stanford Sleepiness Scale (SSS, Hoddes et al., 1972) or the
Epworth Sleepiness Scale (Johns, 1991). It has however also
been shown that subjective sleepiness is only modestly corre-
lated with objective state sleepiness as measured by the
Multiple Sleep Latency Test (MSLT) or the Maintenance of
Wakefulness Test (Banks et al., 2004; Sangal et al., 1999). We
therefore assessed cognitive functioning not only together with
a measure of subjective sleepiness but added an MSLT as
measure of objective sleepiness. As it is known that sleepiness
as well as cognitive functioning show circadian variations
(Blatter et al., 2005), we measured cognitive functions at
272 S. Lis et al.
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
different times of the day to assess whether changes
in sleepiness were accompanied by changes in cognitive
performance.
METHODS
Subjects
Twenty patients (19 males, 1 female) with OSAS were recruited
for the study. All of them had been consecutively admitted to
the sleep laboratory of the Medical Clinic II of the Justus
Liebig University Hospital Giessen, Germany and had been
investigated by full-night attended polysomnography (PSG).
All data were visually analysed according to standard criteria.
OSAS was diagnosed if the respiratory disturbance index
(RDI) exceeded 10 per hour of sleep in the presence of sleep-
related symptoms (i.e. snoring, witnessed apnoeas, excessive
daytime sleepiness). Mean RDI was 57.9 (± 20.2) and mean
nocturnal oxygen saturation was 91.3 ± 3.6. They had no
other diagnosable sleep disorder and had no previous treat-
ment for sleep apnoea syndrome. Mean age of the patients was
53.4 ± 10.5 years.
Ten healthy subjects (8 male, 2 female) were recruited as
control group. Sleep disorders were excluded by means of a
PSG. Mean age of the healthy controls (HC) was comparable
with the age of the patients group (53.6 ± 10.5 years, inde-
pendent t-test: t = 0.04, P = 0.969). Both groups were
matched for education.
All subjects were right-handed (Annett, 1967). All partici-
pants gave informed consent prior to participating in the study.
Experimental procedure
Cognitive testing was performed together with an MSLT at the
day following diagnostic PSG. For this purpose, subjects spent
1 day in the sleep laboratory of the Psychiatric University
Hospital of Justus-Liebig-University. They had to solve a set of
cognitive tasks at four time-points (at 10:00, 12:00, 15:00 and
17:00 hours). Each performance was followed by a nap
following standardMSLT procedures (Carskadon et al., 1986).
Stimuli and responses
Each of the four cognitive testing sessions consisted of a set of
six tasks. In two sessions, visual stimuli were used, while in the
other sessions acoustical stimuli were presented. Visual as well
as acoustical stimuli were applied because modality of input is
assumed to affect WM processes (Baddeley, 1986). A recent
study confirmed that modality of input affects WM network
configurations not only regarding sensory cortices but includ-
ing PFC area already in simple WM tasks (Protzner and
McIntosh, 2007).
Stimulus modality was pseudo-randomly assigned to the
four time-points of testing, i.e. each was administered one time
in the morning sessions (10:00 or 12:00 hours) and another
time in the afternoon sessions (15:00 or 17:00 hours).
In each task, two stimulus types were presented in pseudo
random order. A total of 60 stimuli were presented with 50%
probability of occurrence for each stimulus type. Visual
stimulus types were triangles and squares. These were pre-
sented on a computer screen (screen size: 17¢, stimulus
duration: 50 ms). As acoustical stimulus types high (800 Hz)
and low (1200 Hz) tones were applied through earphones.
Subjects were seated in a comfortable chair in front of a
computer screen. In all tasks, they had to indicate their
response by moving a cursor as fast as possible from a starting
button to a corresponding target button using pen movements
on a graphic tablet. The starting button and target buttons
were displayed on the computer screen.
Trials were self-paced with subjects signalling the start of a
trial by positioning the cursor at the starting button.
Responses had to be initiated within 4 s after stimulus onset;
slower RTs were processed as errors. Mean trial duration was
4870 ms (± 630 ms) with a mean interval of 3980 ms
(± 495 ms) between response and subsequent stimulus.
Cognitive tasks
To assess WM functions, n-back tasks were used with two
versions of WM load (1-back, 2-back). They were presented
within a hierarchy of tasks with increasing processing demands
(Donders, 1868; Krieger et al., 2001), i.e. together with a
simple reaction task (SRT), a stimulus discrimination task
(SDT), a choice reaction task (CRT) and a vigilance version of
the CRT (CRT-vig). All tasks were presented at random order.
Figure 1 gives a schematic illustration of the six types of
tasks applied.
In the SRT, subjects had to guide the cursor immediately
after stimulus presentation towards the target button. This
task measures the four elementary cognitive subprocesses:
sensory transduction, pattern integration, motor preparation
and motor execution (Sanders, 1980).
In the SDT, subjects were asked to respond to one
stimulus type but to ignore the other, which required a
discrimination process to differentiate between the two
stimuli types.
In the CRT, two buttons were presented. They were labelled
�triangle� and �square� for visual stimuli and �high� and �low� foracoustical stimuli. The stimulus determined the target button.
Here, a response selection stage became essential (Massaro,
1990).
The CRT-Vig equalled the CRT with the exception of
altered probabilities for the two stimulus types: triangles and
high tones were presented with 15% probability, squares and
low tones with 85% probability.
In the n-back-task, the two target buttons were labelled
�same� and �different�, requiring subjects to compare the
stimulus presented either 1-trial (1-back) or 2-trials (2-back)
back with the currently presented one. Same and different
judgements were required with equal probability. The n-back
task is assumed to involve WM functions. With increasing n
the WM load is increased.
Executive functions in OSAS 273
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
Subjective and objective sleepiness
After each cognitive testing session, sleepiness was assessed.
Subjects rated their subjective sleepiness by means of the SSS.
This was immediately followed by a maximum 20 min
standard PSG recording to assess objective sleepiness. Subjects
lay down in a darkened room and were instructed to fall asleep
as fast as possible. The recording was terminated 20 min after
lights-out if there did not occur sleep, or after three consecutive
epochs of stage 1 sleep or after the first epoch of another sleep
stage.
Measurement variables and statistical analysis
Cognitive performance
Dependent variables were the percentage of correct responses
in each task as well as the RT (time from stimulus onset until
reaching the target button, time resolution 5 ms). For each
subject and task the median of the RT distribution was
determined and used in further analysis. To avoid confounding
RT measures with error recovery and escape strategies, only
correct responses were analysed.
A 2 · 2 · 2 · 6 anova with the independent factor �group�(�OSAS�, �HC�) and the repeated measurement factors �time of
cognitive testing� (�morning�, �afternoon�), �modality� (visual,
acoustical stimuli) and �task� (SRT, SDT, CRT, CRT-vig, 1-
back, 2-back) was conducted to analyse each of the dependent
variables. The degrees of freedom in the anova were corrected
according to Greenhouse and Geisser (1959). When appropri-
ate, post hoc t-tests (two-tailed) were applied. Because of the
explorative character of the study, no adjustment of the alpha
level was carried out to avoid inflation of type II error.
Sleepiness
Dependent variables were the score of the SSS for subjective
sleepiness and the sleep onset latency in the MSLT for
objective sleepiness. Sleep onset latency was defined as the
elapsed time from light-out to the first epoch scored as sleep.
SRT React to all stimuli
Triangle Square Triangle Square Triangle Square Triangle Square Triangle Square
SDT React to triangles only
CRT React to triangles with „triangle“, to squares with „square“
1-back task Compare the current stimulus with that presented 1-back
Trial 1 Trial 2 Trial 3 Trial 4 Trial 5
different same different same different same different same different same
2-back task Compare the current stimulus with that presented 2-back
CRT-vig React to triangles with „triangle“, to squares with „square“
Triangle Square Triangle Square Triangle Square Triangle Square Triangle Square
different same different same different same different same different same
Figure 1. Schematic illustrations of the six task of the reaction time decomposition approach for the visual stimulus modality.
274 S. Lis et al.
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
Online staging of the PSG recordings were confirmed by an
experienced psychologist who was blind for group and time of
measurement.
A first statistical analysis was performed separately for SSS
and sleep onset latency with a 2 · 4 anova with the indepen-
dent factor �group� (�OSAS�, �HC�) and the repeated measure-
ment factors �time of sleepiness rating� (10:00, 12:00, 15:00 and
17:00 hours) to analyse differences in sleepiness between
groups over the day. The degrees of freedom in the anova
were corrected according to Greenhouse & Geisser.
To analyse differences in individual sleepiness at the time of
the cognitive testing, the single sleepiness scores were assigned
to each of the cognitive testing sessions and analysed with an
anova design corresponding to the one used for the cognitive
performance, i.e. a 2 · 2 · 2 · 6 anova.
Relation between cognitive performance, sleepiness and OSAS
characteristics
Interdependence between cognitive performance and sleepiness
was analysed using Pearson�s correlation coefficient. Sleepiness
parameters were used as covariates in the anova design,
respectively, subdesigns to test whether group differences go
beyond effects that can be explained by differences in
subjective and ⁄or objective sleepiness scores.
To investigate whether alterations in cognitive processing of
OSAS patients could be linked to RDI or hypoxaemia,
Spearman�s rank-order correlation was calculated between
RDI and mean oxygen saturation and those cognitive para-
meters that show alterations in OSAS patients� performance.
RESULTS
Subjective and objective sleepiness
Figure 2 depicts subjective and objective sleepiness for four
times of the day. No differences in SSS score emerged between
OSAS patients and their HC.
In contrast, sleep onset latencies in MSLT were shorter in
OSAS patients than in HC (�group�: F(1,28) = 5.09,
P = 0.032). Although the extent of this difference between
groups seemed to depend on the time of day with the most
pronounced difference at 12:00 hours, this could not be
confirmed statistically [interaction �group� · �time of day�;F(2,64) = 1.54, P = 0.220].
When sleepiness was related to the single measurements of
cognitive performance, significant differences between groups
emerged neither for the SSS score nor for the sleep onset
latencies that depended on whether they are visual or acoustical
stimuli were presented (Fig. 3). For both groups, sleep onset
latencies were increased in the afternoon compared with the
morning measurements [time: F(1,28) = 4.17, P = 0.05].
Cognitive testing
Results of the statistical analysis are given for both accuracy of
task solving and RTs in Table 1.
Accuracy of task solving
Obstructive sleep apnoea syndrome patients showed a lower
accuracy depending on type of task and time of day (Fig. 4,
time · task · group; F = 4.62, P = 0.009); accuracy only
decreased when WM processes were required to solve the
min
0
1 2 3 4 5 6
7 8 9
10 11 12 13
14 15 16 17 18 19 20
0
1
2
3
4
5
10 am 12 am 3 pm 5 pm
10 Healthy controls 20 OSAS
Sleep onset latency (MSLT)
SSS
Time
10 am 12 am 3 pm 5 pm Time
(a)
(b)
Figure 2. Mean and SE of daytime sleepiness for obstructive sleep
apnoea syndrome patients and their healthy controls over the day. a)
Subjective sleepiness (Stanford Sleepiness Scale score). b) Objective
sleepiness (sleep onset latency in multiple sleep latency test).
Executive functions in OSAS 275
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
tasks. In the morning sessions, demands upon WM in 1-back
tasks compared with demands of the CRT task were associated
with a drop in accuracy that was more pronounced in OSAS
patients than in HC (t = 3.37, P = 0.003). In the afternoon
sessions, an increase in WM load caused a decrease in
performance in OSAS patients, which was more marked than
in HC (t = 2.11, P = 0.044). OSAS patients solved the 2-
back tasks in the afternoon less accurately than in the morning
(t = 3.23, P = 0.004).
Effects were independent of stimulus modality; mean accu-
racies are therefore reported combined for visual and acous-
tical stimuli (Fig. 4).
In contrast to the more basic tasks SRT, SDT, CRT and
CRT-vig, the percentage of correct responses in the WM
tasks was relatively low. Because of difference in perfor-
mance, an additional analysis of the type of error was
performed for explorative purposes in the WM tasks. Trials
with responses to the incorrect stimuli were separated from
those during which no response occurred, i.e. the subjects
might have missed the stimulus. The percentages of these two
types of errors of the total number of trials are presented in
Fig. 5. Statistical analysis (2 · 2 · 2 · 2 · 2 anova with
factors group, modality, WM load, time of day and type of
error) revealed different proportions of trials with missed
responses between OSAS patients and their HC depending on
the WM load of the n-back task. The percentage of trials
without overt response increased only if OSAS patients
solved tasks with high WM load (F = 4.93, P = 0.035). In
general, fewer trials without overt responses could be
observed than trials with a reaction towards the incorrect
target button (F = 19.66, P = 0.0001). There was no differ-
ence between types of errors regarding time of day or
modality of stimuli.
Accuracy of task solving and sleepiness
In the morning, a correlation between sleepiness and drop in
accuracy was observed after introduction of WM processes:
the decrease in performance became more pronounced with
Table 1 Results of the 2 · 2 · 2 · 6 anova for percentage of correct
solutions and reaction times
% Correct Reaction time
F d.f. P F d.f. P
Group 2.16 1, 28 0.153 4.83 1.28 0.036
Stimulus modality
(mod)
0.78 1, 28 0.384 3.81 1.28 0.061
Mod · group 1.39 1, 28 0.248 0.95 1.28 0.337
Time 0.71 1, 28 0.405 7.53 1.28 0.010
Time · group 0.84 1, 28 0.774 0.12 1.28 0.733
Task 95.20 2, 63 <0.0001 150.01 1.42 <0.0001
Task · group 1.98 2, 63 0.142 0.49 1.42 0.562
Mod · time 0.38 1, 28 0.543 1.67 1.28 0.206
Mod · time · group 0.67 1, 28 0.421 0.72 1.28 0.403
Mod · task 1.48 2, 78 0.228 1.16 1.53 0.319
Mod · task · group 0.28 2, 78 0.830 0.70 1.53 0.494
Time · task 1.19 2, 68 0.314 3.57 1.54 0.036
Time · task · group 4.62 2, 68 0.009 2.13 1.54 0.129
Mod · time · task 0.65 2, 78 0.578 0.34 2.61 0.736
Mod · time · task ·group
0.71 2, 78 0.543 0.97 2.61 0.389
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Visualstimuli
10 Healthy controls
20 OSAS
SSS
Sleep onset latency (MSLT)
Morning Afternoon
Acousticalstimuli
Visualstimuli
Acousticalstimuli
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Visualstimuli
Acousticalstimuli
Visualstimuli
Acousticalstimuli
min
Morning Afternoon
(a)
(b)
Figure 3. Mean and SE of daytime sleepiness for obstructive sleep
apnoea syndrome patients and their healthy controls linked to cogni-
tive testing sessions. a) Subjective sleepiness (Stanford Sleepiness Scale
score). b) Objective sleepiness (sleep onset latency in multiple sleep
latency test).
276 S. Lis et al.
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
shorter sleep latency (r = 0.429, P = 0.018) and as a trend
with increase in subjectively rated tiredness (r = )0.309,P = 0.096). After including sleep onset latency and SSS score
as covariates, no significant difference between OSAS and HC
could be found anymore [comparison of groups for
CPCMT1)CPCRT (morning): F(1,26) = 1.97, P = 0.172]. In
the afternoon, no relation between sleepiness and drop in
accuracy of task solving with increasing WM load emerged.
This was the case for the SSS score as well as for the sleep
onset latency. Group differences were not due to individual
sleepiness at this particular time of the day [comparison of
groups for CPCMT2)CPCMT1 (afternoon) with SSS and sleep
onset latency as covariate: F(1,26) = 6.90, P = 0.014].
Accuracy of task solving, RDI and oxygen saturation
No correlations were observed.
%
0 50
60
70
80
90
100
%
0 50
60
70
80
90
100
SRT 1-back
10 Healthy controls 20 OSAS
Afternoon
Morning
2-back CRT-vig CRT SDT
SRT 1-back 2-back CRT-vig CRT SDT
Figure 4. Mean and SE of the percentage of correct solutions for
obstructive sleep apnoea syndrome patients and their healthy controls
in the six tasks during task solving in morning and afternoon test
sessions summarized over stimulus modality.
0
5
10
15
20
25
30
Incorrect response
0
5
10
15
20
25
30
1-back 2-back
20 OSAS
1-back 2-back
%%
No response
10 Healthy controls
Figure 5. Mean and SE of the percentage of different types of errors
for obstructive sleep apnoea syndrome patients and their healthy
controls in the n-back tasks summarized over the time of the day and
stimulus modality.
Executive functions in OSAS 277
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
Reaction time
Obstructive sleep apnoea syndrome patients needed more time
for task solving than HC (see Fig. 6, Table 1: �group�,F = 4.83, P = 0.036). Differences between groups in mean
RT seem to be less pronounced in the n-back tasks. Yet, this
could not be confirmed statistically. Differences between
groups were independent not only on the type of task but
also of the time of day and the stimulus modality (see Table 1).
Reaction time and sleepiness
Reaction time was clearly correlated with the subjective rating
of sleepiness (r = 0.550, P = 0.002). With increasing SSS
score, processing time increased. As a statistical trend, a
comparable effect can be reported for objective sleepiness. A
shortening of the sleep onset latency was accompanied by an
increase of the time required to solve the tasks (r = )0.353,P = 0.056).
No significant difference between OSAS and HC emerged
when taking sleep onset latency and SSS score as covariates
into account [comparison of groups for total RT:
F(1,26) = 1.99, P = 0.170].
Reaction time and RDI and oxygen saturation
No correlations were observed.
DISCUSSION
The purpose of the present study was to contribute to the
understanding of cognitive dysfunction in OSAS. We used a
RT decomposition approach to analyse executive functioning.
Specifically, we investigated whether impaired performance in
tasks involving executive functions can be attributed to deficits
in executive functions themselves or to impairments in more
basic cognitive subprocesses additionally involved in task
solving. With the help of subjective and objective measure-
ments of daytime sleepiness, we investigated the impact of
sleepiness upon the pattern and extent of the observed
cognitive deficit.
Our data demonstrate prolonged processing times of OSAS
patients in n-back tasks. However, these were observed not
only when WM functions were involved but also in simple RT
tasks. This implies that the slowing of OSAS patients in n-back
tasks cannot be attributed to deficits in executive functions.
With increasing complexity of the tasks, an increase in
processing times could be observed. This was equal in both
groups, i.e. the extent of slowing in the OSAS group remained
almost constant across tasks. Thus, the slowing in OSAS
cannot be explained by a general slowing affecting all kinds of
cognitive subprocesses required for task solving. Such an
unspecific deficit should have resulted in an accentuation of
group differences with task complexity increasing proportion-
ally with processing times over tasks. In contrast, the deficits of
OSAS patients can be traced back to impairments of elemen-
tary cognitive functions, i.e. sensory transduction, feature
integration or motor preparation and execution which are
required even in simple RT tasks. Further studies are necessary
to attribute the deficit to one of these processes by systemat-
ically varying the demands upon those processes e.g. with
increasing demands on perceptive and ⁄or motor processes
without changing the cognitive architecture of the task
(Sanders, 1980).
In addition to slowing, a reduced accuracy of task solving
could be observed in OSAS patients. Contrary to increased
processing time, this deficit emerged only if WM functions
were involved in solving n-back tasks. Impairments differed
depending on the imposed WM load and the time of day at
which the cognitive testing took place. In the morning
sessions, deficits became obvious with low WM demands,
i.e. if the n-back task required the match between a present
stimulus and that 1-trial back in sequence. In the afternoon, a
higher WM load was necessary (2-back condition) to reveal
deficits in the performance of OSAS patients. Thomas et al.
(2005) in a brain imaging study, showed reduced performance
in untreated sleep apnoea patients on a 2-back WM task. This
was accompanied by reduced brain activation in the anterior
cingulate, dorsolateral prefrontal and posterior parietal cor-
tices. As in our study, this effect was independent of the
extent of night-time hypoxaemia. Both hypoxaemic and non-
hypoxaemic patients showed similar patterns of activation.
These results are in accordance with an increasing number of
studies that have not found a relation between cognitive
deficits and hypoxaemia (Aloia et al., 2003; Cohen-Zion
et al., 2004).
Modality of stimulus input is assumed to affect WM
processes and the related activity of PFC regions (Baddeley,
1986 Protzner andMcIntosh, 2007). However, no differences in
ms
0
400
500
600
700
800
900
1000
1100
1200
1300
1400
SRT SDT CRT CRT-vig 1-back 2-back
10 Healthy controls 20 OSAS
Figure 6. Mean and SE of the reaction times for obstructive sleep
apnoea syndrome patients and their healthy controls in the six tasks
summarized over the time of the day and stimulus modality.
278 S. Lis et al.
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
WM performance between OSAS patients and controls could
be observed that were influenced by stimulus modality.
In general, OSAS patients showed higher daytime sleepiness
compared with their HCs. Our data show that the slowing of
the OSAS patients could be related to this higher degree of
sleepiness. As with processing times, the drop in accuracy in
tasks with low demands on WM processes during the morning
sessions could be linked to an increased sleepiness in the
patient group. This is in accordance with several studies in
which OSAS patients� cognitive performance was very similar
to the cognitive decline typically found after sleep loss
(Durmer and Dinges, 2005). It might support the statement
that sleepiness is the primary factor in a parsimonious
explanation for the deficits in sleep apnoea without having to
assume prefrontal brain damage (Verstraeten and Cluydts,
2004). However, it should be considered that a correlation
between cognitive alterations and sleepiness is necessary but
not sufficient to demonstrate a causal sequence, i.e there are
limits to the inferences that can be drawn from cross-sectional
data.
Our data point to the fact that subjective and objective
parameters of sleepiness seem to tap different aspects of
tiredness. While the slowing is strongly related to the subjec-
tively assessed sleepiness, the loss in accuracy in tasks
involving executive functions in the morning seems to be
connected to objective sleepiness. Although these results
should be interpreted cautiously because of the small sample
sizes and the quite high sleepiness of the HCs in the present
study, they might point to the importance of considering both
aspects of sleepiness. Contrarily, the deficits in tasks with
higher WM demands in the afternoon persisted after taking
individual sleepiness into account. In a recent study, Blatter
et al. (2005) showed that after sleep deprivation, deficits in
executive tasks increased with the duration of the time spent
awake. However, this holds true only in tasks with high
difficulties. This might point to the fact that factors linked to
sleep loss other than the measurable sleepiness might contrib-
ute to deficits which become obvious only with increasing task
demands.
Further studies are required to confirm our data with larger
sample sizes, enabling the analysis of different characteristics
of the patients such as intelligence (Alchanatis et al., 2005),
individual vulnerability to sleep loss (Durmer and Dinges,
2005), age, sex or severity of illness (Beebe, 2005). Because no
influence of the modality of stimulus input upon the deficits in
OSAS patients could be observed, it seems to be advantageous
to use only one modality of stimulus input in future studies.
This would allow for a finer grained analysis of changes of
individual cognitive performance in relation to changes of
sleepiness over the day. In the present study, this was
prevented because acoustical and visual stimuli were presented
in separate sessions and in balanced order. To avoid mixing
cognitive performance based upon tasks with varying stimulus
modalities in the analysis, data had to be compressed into two
periods, i.e. the morning and afternoon session. As our data
show, within these periods sleepiness might be quite
different for the two times of the day that are combined in
the analysis.
In summary, our data confirm the existence of deficits in
tasks of executive functioning in OSAS patients. Using a RT
decomposition approach, however, we could attribute cross-
group differences in RT on the high level WM task to slowing
at a more basic processing level. In contrast, reduced accuracy
in the WM task in the OSAS group could not be explained by
deficits in more elementary cognitive processes. The majority
of the deficits observed in WM tasks in OSAS patients could
be shown to co-vary with subjective or objective sleepiness.
Moreover, OSAS patients exhibited deficits in tasks with high
demands on executive functions that became apparent only
over the course of the day, indicating the importance of
considering circadian variations or the duration of time spent
awake when measuring cognitive dysfunctions in patients with
sleep disorders. In conclusion, our data point to a pattern of
intact and impaired cognitive subfunctions and thus to the
complexity of deficits of executive functions in OSAS.
REFERENCES
Alchanatis, M., Zias, N., Deligiorgis, N., Amfilochiou, A., Dionellis,
G. and Orphanidou, D. Sleep apnea-related cognitive deficits and
intelligence: an implication of cognitive reserve theory. J. Sleep Res.,
2005, 14: 69–75.
Aloia, M. S., Ilniczky, N., Di Dio, P., Perlis, M. L., Greenblatt, D. W.
and Giles, D. E. Neuropsychological changes and treatment
compliance in older adults with sleep apnea. J. Psychosom. Res.,
2003, 54: 71–76.
Aloia, M. S., Arnedt, J. T., Davis, J. D., Riggs, R. L. and Byrd, D.
Neuropsychological sequelae of obstructive sleep apnea-hypopnea
syndrome: a critical review. J. Int. Neuropsychol. Soc., 2004, 10: 772–
785.
Annett, M. The binomial distribution of right, mixed and left
handedness. J. Exp. Psychol., 1967, 19: 327–333.
Baddeley, A. Working Memory. Oxford University Press, Oxford,
1986.
Banks, S., Barnes, M., Tarquinio, N., Pierce, R. J., Lack, L. C. and
Mcevoy, R. D. Factors associated with maintenance of wakefulness
test mean sleep latency in patients with mild to moderate obstructive
sleep apnoea and normal subjects. J. Sleep Res., 2004, 13: 71–78.
Bedard, M. A., Montplaisir, J., Richer, F., Rouleau, I. and Malo, J.
Obstructive sleep apnea syndrome: pathogenesis of neuropsycho-
logical deficits. J. Clin. Exp. Neuropsychol., 1991, 13: 950–964.
Beebe, D. W. Neurobehavioral effects of obstructive sleep apnea: an
overview and heuristic model. Curr. Opin. Pulm. Med., 2005, 11:
494–500.
Beebe, D. W. and Gozal, D. Obstructive sleep apnea and the
prefrontal cortex: towards a comprehensive model linking nocturnal
upper airway obstruction to daytime cognitive and behavioral
deficits. J. Sleep Res., 2002, 11: 1–16.
Blatter, K., Opwis, K., Munch, M., Wirz-Justice, A. and Cajochen, C.
Sleep loss-related decrements in planning performance in
healthy elderly depend on task difficulty. J. Sleep Res., 2005, 14:
409–417.
Carskadon, M. A., Dement, W. C., Mitler, M. M., Roth, T.,
Westbrook, P. R. and Keenan, S. Guidelines for the multiple sleep
latency test (MSLT): a standard measure of sleepiness. Sleep, 1986,
9: 519–524.
Chapman, L. J. and Chapman, J. P. The measurement of differential
deficit. J. Psychiatr. Res., 1978, 14: 303–311.
Executive functions in OSAS 279
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280
Cohen-Zion, M., Stepnowsky, C., Johnson, S., Marler, M., Dimsdale,
J. E. and Ancoli-Israel, S. Cognitive changes and sleep disordered
breathing in elderly: differences in race. J. Psychosom. Res., 2004, 56:
549–553.
Donders, F. C. Die Schnelligkeit psychischer Prozesse. Reicherts¢s &
Dubois Archiv fur Anatomie, Physiologie und Wissenschaftliche
Medizin, 1868, 000: 657–681.
Durmer, J. S. and Dinges, D. F. Neurocognitive consequences of sleep
deprivation. Semin. Neurol., 2005, 25: 117–129.
Feuerstein, C., Naegele, B., Pepin, J. L. and Levy, P. Frontal lobe-
related cognitive functions in patients with sleep apnea syndrome
before and after treatment. Acta Neurol. Belg., 1997, 97: 96–107.
Fulda, S. and Schulz, H. Cognititive dysfunction in sleep-related
breathing disorders: a meta-analysis. Sleep Res. Online, 2003, 5: 19–
51.
Gevins, A. S., Bressler, S. L., Cutillo, B. A., Illes, J., Miller, J. C.,
Stern, J. and Jex, H. R. Effects of prolonged mental work on
functional brain topography. Electroencephalogr. Clin. Neurophys-
iol., 1990, 76: 339–350.
Glahn, D. C., Ragland, J. D., Abramoff, A., Barrett, J., Laird, A. R.,
Bearden, C. E. and Velligan, D. I. Beyond hypofrontality: a
quantitative meta-analysis of functional neuroimaging studies of
working memory in schizophrenia. Hum. Brain Mapp., 2005, 25: 60–
69.
Goldberg, T. E. and Gold, J. M. Neurocognitive deficits in schizo-
phrenia. In: S. R. Hirsch and D. R. Weinberger (Eds) Schizophrenia.
Blackwell Science Ltd, Oxford, 1995: 146–162.
Greenhouse, S. W. and Geisser, S. On methods in the analysis of
profile data. Psychometrika, 1959, 24: 95–112.
Hoddes, E., Dement, W. C. and Zarcone, V. The development and use
of the Stanford Sleepiness Scale (SSS). Psychophysiology, 1972, 10:
431–436.
Johns, M. W. A new method for measuring daytime sleepiness: the
Epworth Sleepiness Scale. Sleep, 1991, 14: 540–545.
Kim, H. C., Young, T., Matthews, C. G., Weber, S. M., Woodward,
A. R. and Palta, M. Sleep-disordered breathing and neuropsycho-
logical deficits. A population-based study. Am. J. Respir. Crit. Care
Med., 1997, 156: 1813–1819.
Krieger, S., Lis, S. and Gallhofer, B. Cognitive subprocesses and
schizophrenia. A. Reaction-time decomposition. Acta Psychiatr.
Scand. Suppl., 2001, 104: 18–27.
Krieger, S., Lis, S., Cetin, T., Gallhofer, B. and Meyer-Lindenberg, A.
Executive function and cognitive subprocesses in first-episode, drug-
naive schizophrenia: an analysis of N-back performance. Am. J.
Psychiatry, 2005, 162: 1206–1208.
Lee, M. M., Strauss, M. E., Adams, N. and Redline, S. Executive
functions in persons with sleep apnea. Sleep Breath, 1999, 3: 13–16.
Massaro, D. W. An information processing analysis of perception and
action. In: O. Neumann and W. Prinz (Eds) Relationships Between
Perception and Action. Springer, Berlin, 1990: 133–166.
Naegele, B., Thouvard, V., Pepin, J. L., Levy, P., Bonnet, C., Perret, J.
E., Pellat, J. and Feuerstein, C. Deficits of cognitive executive
functions in patients with sleep apnea syndrome. Sleep, 1995, 18: 43–
52.
Owen, A. M., Mcmillan, K. M., Laird, A. R. and Bullmore, E. N-back
working memory paradigm: a meta-analysis of normative functional
neuroimaging studies. Hum. Brain Mapp., 2005, 25: 46–59.
Protzner, A. B. and McIntosh, A. R. The interplay of stimulus
modality and response latency in neural network organization for
simple working memory tasks. J. Neurosci., 2007, 27: 3187–3197.
Redline, S., Strauss, M. E., Adams, N., Winters, M., Roebuck, T.,
Spry, K., Rosenberg, C. and Adams, K. Neuropsychological
function in mild sleep-disordered breathing. Sleep, 1997, 20: 160–
167.
Sanders, A. F. Stage analysis of reaction processes. In: G. E. Stelmach
and J. Requin (Eds) Tutorials in Motor Behaviour. Elsevier,
Amsterdam, 1980: 331–354.
Sangal, R. B., Sangal, J. M. and Belisle, C. Subjective and objective
indices of sleepiness (ESS and MWT) are not equally useful in
patients with sleep apnea. Clin. Electroencephalogr., 1999, 30: 73–75.
Saunamaki, T. and Jehkonen, M. A review of executive functions in
obstructive sleep apnea syndrome. Acta Neurol. Scand., 2007, 115:
1–11.
Thomas, R. J., Rosen, B. R., Stern, C. E., Weiss, J. W. and Kwong, K.
K. Functional imaging of working memory in obstructive sleep-
disordered breathing. J. Appl. Physiol., 2005, 98: 2226–2234.
Verstraeten, E. Neurocognitive effects of obstructive sleep apnea
syndrome. Curr. Neurol. Neurosci. Rep., 2007, 7: 161–166.
Verstraeten, E. and Cluydts, R. Executive control of attention in sleep
apnea patients: theoretical concepts and methodological consider-
ations. Sleep Med. Rev., 2004, 8: 257–267.
Verstraeten, E., Cluydts, R., Pevernagie, D. and Hoffmann, G.
Executive function in sleep apnea: controlling for attentional
capacity in assessing executive attention. Sleep, 2004, 27: 685–693.
280 S. Lis et al.
� 2008 European Sleep Research Society, J. Sleep Res., 17, 271–280