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