Sleep, 20(4):258-266
© 1997 American Sleep Disorders Association and Sleep Research Society
.j
The Alpha Attenuation Test: Assessing Excessive Daytime
Sleepiness in Narcolepsy-Cataplexy
Christi E. D. Alloway, Robert D. Ogilvie and *Colin M. Shapiro
Brock University, St. Catharines, and *University of Toronto, The Toronto Hospital, Toronto, Ontario, Canada
Summary: Daytime sleep tendency was assessed in 10 drug-free patients with narcolepsy-cataplexy and 10 normals matched for age and gender. Following nocturnal polysomnography, the alpha attenuation test (AAT) and the
mUltiple sleep latency test (MSLT) were administered during five sessions occurring at 2-hour intervals beginning
at 0900 and 1000 hours, respectively. For the AAT, participants were polysomnographically recorded for 8 minutes
while seated in an illuminated room with their eyes alternately opened and closed. Power spectral analyses of
electroencephalograph (EEG) activity at 02-Al (10 second epochs) were calculated using fast Fourier transformations (FFT) within the alpha frequency range (8-12 Hz) to obtain ratios of mean eyes-closed to mean eyes-open
alpha power (i.e. the alpha attenuation coefficient, AAC). The narcoleptics were sleepier than the normals as
indicated by a significantly smaller mean AAC and a significantly shorter mean latency to stage I on the MSLT.
These findings suggest that the AAT may provide a quick and practical objective assessment of the excessive
daytime sleepiness (EDS) associated with narcolepsy. Key Words: Narcolepsy-Alpha attenuation test-Excessive
daytime sleepiness-Physiological sleep tendency-EEG power spectral analysis.
The multiple sleep latency test (MSLT) (1) is the
traditional measure for objectively assessing excessive
daytime sleepiness (EDS) in individuals with narcolepsy. Physiological sleep tendency is assessed by
measuring the speed at which individuals fall asleep
on multiple occasions while lying with their eyes
closed in a darkened room (2)-that is, the speed at
which the electroencephalograph (EEG) moves from
the alpha activity (8-12 Hz) of relaxed eyes-closed
wakefulness to the theta activity (4-8 Hz) of the first
stage of sleep. A mean sleep latency of 5 minutes or
less on the MSLT is considered to represent a pathological degree of sleepiness and in conjunction with
the presence of at least two sleep onset rapid eye
movement (REM) periods, is considered diagnostic of
narcolepsy (3). However, the validity of the MSLT
may be questioned due to the confounding of sleepiness with the learned ability to fall asleep (4). Furthermore, the efficacy of the MSLT is limited by a
floor effect (near zero latencies) in clinical populations
(5) and the failure of EDS patients to show significant
improvement in MSLT latencies following subjectively effective pharmacological treatment (6). In addition,
Accepted for publication January 1997.
Address correspondence and reprint requests to: Christi E. D. Alloway, Psychology Department, Queen's University, Kingston, Ontario K7L 3N6, Canada.
the cooperation of the patient in the attempt to fall
asleep is essential to the validity of the MSLT, and
patients wishing to avoid falling asleep may engage in
a variety of behaviors to distort the MSLT results. Furthermore, the MSLT is time-consuming. The MSLT
also relies on the presence of a polysomnographer to
visually sleep score the EEG concurrently with each
nap attempt, introducing a subjective component to an
objective measure of sleepiness (7). These shortcomings suggest the need for alternate means of detecting
pathological degrees of sleepiness.
The technique of EEG power spectral analysis provides a method of quantifying fluctuations in sleepiness and alertness and has the potential to become an
alternative to sleep latency testing in the assessment
of EDS. Studies (8-9) have demonstrated that as individuals move from alertness toward sleepiness, EEG
power in the alpha frequency range decreases when
eyes are closed but increases when eyes are open.
Based on these findings, the alpha attenuation test
(AAT) (10) was developed as a new method of quantifying variations in physiological sleepiness. During
the AAT, individuals are asked to repeatedly vary their
eyelid position from open to closed while seated in an
illuminated room. In a recent validation study, Stampi
et al. (7) investigated the ratio of eyes-closed to eyesopen EEG alpha power (referred to as the alpha atten-
258
ALPHA AITENUATION TEST IN NARCOLEPSY
~\
I.,
~'
I
(
'tl-
uation coefficient, AAC) to determine its efficacy as
an objective measure of physiological sleepiness.
Stampi et al. (7) deprived 10 normal sleepers of sleep
for 40 hours. Every 2 hours (for a total of 18 sessions)
participants were administered the following battery of
tests: subjective sleepiness measures and performance
tests (starting on the hour), the AAT (at 20 minutes
past the hour), and the MSLT (at 40 minutes past the
hour) (7). The AAC was found to be sensitive to increasing sleepiness following one night of sleep loss
such that participants were significantly less alert (i.e.
had lower AACs) on the 2nd day of testing compared
to the I st day (7). Furthermore, the AAC correlated
significantly with the MSLT in eight out of 10 participants (i.e. r > 0.72 for four participants, r > 0.62 for
three participants and r = 0.53 for one participant),
and these correlations were higher than the correlations
between the MSLT and the subjective sleepiness and
performance measures, suggesting that the AAT provides a valid assessment of physiological sleepiness in
sleepy normals (7).
To our knowledge, the AAT has yet to be evaluated
in patients with narcolepsy. The present study is the
first to investigate the utility of the AAT in distinguishing narcoleptics from normals [however, preliminary
findings have been reported in abstract form (11)]. On
the basis of EEG power spectral studies of sleep-deprived normals (7,9) it was predicted that the ratio of
eyes-closed to eyes-open alpha power (i.e. the AAC)
would be lower in narcoleptics than normal controls
and that, compared to controls, narcoleptics would
demonstrate greater eyes-open alpha power and less
eyes-closed alpha power.
METHODS
(~
Participants
Five female and five male patients diagnosed with
narcolepsy-cataplexy and 10 normal sleepers matched
for age and gender were studied. Narcoleptics were
drug-free at the time of testing. Narcoleptics ranged in
age from 29 to 62 years [mean = 44.3 years, standard
deviation (SO) = 11.9 years], and controls ranged
from 28 to 56 years (mean = 42.7 years, SO = 10.6
years). All participants gave their written informed
consent and received an honorarium for their participation in the study.
A screening process ensured that narcoleptics met
the American Sleep Disorders Association's diagnostic
criteria for narcolepsy (3) and that normal controls did
indeed report normal sleep habits (i.e. the normals reported that they had no difficulty in falling or remaining asleep at night, typically slept between 6 and 8.5
hours a night, and were alert during the daytime). All
259
narcoleptics complained of daytime sleepiness and reported taking naps daily or several times weekly. Normals reported that they were generally alert during the
day. All narcoleptics had a history of cataplexy, six
reported experiencing sleep paralysis and hypnagogic
hallucinations at least once a month, and two reported
incidents of hypnagogic hallucinations that occurred
1-5 times during their lifetime. No normals reported
a history of cataplexy, although two normals reported
experiencing incidents of sleep paralysis and four reported incidents of hypnagogic hallucinations occurring 1-5 times during their lifetime. All narcoleptics
reported multiple nocturnal awakenings. Four normals
reported no nocturnal awakenings, and six reported
awakening 1-3 times during the night.
Participants were also given a sleep diary in which
they recorded their sleep and wake patterns for the 7
days preceding testing. A t test showed that narcoleptics and normals did not differ with respect to the selfreported mean duration of nocturnal sleep. The selfreported mean duration of nocturnal sleep ranged from
4.7 to 10.8 hours for the narcoleptics (mean = 7.9
hours, SD = 2.1 hours) and from 6.2 to 8.6 hours for
the normal controls (mean = 7.4 hours, SD = 0.9
hours).
Symptoms related to depression (which may elicit
sleep onset REM periods) were assessed with the Beck
Depression Inventory (BOI) (12), which consists of 21
statements related to depression. Participants rank each
statement according to the degree to which it is experienced [i.e. from neutral (0) to maximal severity
(3)]. BDI scores may range from 0 to 63. A t test
showed that narcoleptics and normals did not differ
with respect to the number of depression-related symptoms they reported on the BDl. The scores obtained
on the BDI for the narcoleptics ranged from 1.0 to 25.0
(mean = 8.6, SO = 7.47) and for the normals ranged
from 1.0 to 16.0 (mean = 4.5, SO = 5.0).
Prior to (but not during) testing, nine of the 10 narcoleptics were taking medications for their symptoms.
Six narcoleptics were taking central nervous system
stimulants (five took methylphenidate hydrochloride,
one took dexamphetamine sulfate) for their excessive
daytime sleepiness, three were taking sleeping pills
(immovane) to improve their nocturnal sleep, and two
were taking REM sleep suppressants (clomipramine)
to reduce their cataplexy. Two of the narcoleptics taking methylphenidate hydrochloride were withdrawn
for 2 days prior to testing, and the seven remaining
medicated narcoleptics were withdrawn from their
medications for 7 days prior to testing.
All of the normal participants and nine narcoleptics
reported consuming caffeinated beverages. On average, normals consumed 2.5 cups of tea/coffee/pop per
day (SO = 1.0). Narcoleptics consumed significantly
Sleep, Vol. 20, No.4, 1997
260
C. E. D. ALLOWAY ET AL.
more cups of tea/coffee/pop per day (mean == 5.0, SD
== 3.7) [t(18) == 2.13, P = 0.047]. Five normals and
three narcoleptics reported consuming alcoholic beverages. On average, normals consumed 1.8 alcoholic
drinks per week (SD = 2.7). This did not differ from
the average number of alcoholic drinks consumed by
narcoleptics (mean = 1.5 drinks weekly, SD = 3.0).
None of the normals reported smoking cigarettes or
pipes, whereas four narcoleptics smoked an average of
29 cigarettes per day. During testing, these participants
were permitted to smoke after each MSLT session. All
participants agreed to refrain from alcohol, caffeine,
and sleep-related medications during the 24-hours preceding testing.
Procedure
Participants reported to the sleep lab at 2130 hours
for orientation and electrode application. Electrodes
were positioned to enable the monitoring of Cz (central) and 02 (right occipital) EEG [referenced to A2
(right mastoid)], submental electromyogram (EMG)
and outer canthi electrooculogram (EOG) activity. A
16-channel Neurofax polygraph (Nihon Kohden, Irvine, CA) was used to amplify the polysomnographic
recordings. Time constants were set at 0.3 for EEG
and EOG recordings and at 0.03 for EMG recordings.
High-frequency filters were set at 30 Hz for EEG and
EOG and at 70 Hz for EMG. Recording sensitivity was
7 fL V /mm for EEG and EOG and 1-3 fL V/mm for
EMG. Polysomnographic data from the polygraph
were acquired on paper and on computer using the
software program Microcomputer Quantitative Electrophysiology (Imaging Research Inc., under continued development at Brock University, St. Catharines,
Ontario, Canada).
Participants retired for the night at 2300 hours and
slept undisturbed until 0800 hours. The MSLT was
administered according to the guidelines for clinical
use outlined in the report from the American Sleep
Disorders Association (13). Following the nocturnal
polysomnography, nap opportunities were given at
1000, 1200, 1400, 1600, and 1800 hours. At the start
of each session of the MSLT, participants were instructed to try to fall asleep while lying quietly with
their eyes closed in a darkened room. During the
MSLT session, EEG, EOG, and EMG were recorded
on paper and computer according to the parameters
outlined above and were scored in 30 second epochs
using the criteria of Rechtschaffen and Kales (14).
Sleep onset was defined as the first three consecutive
minutes of stages 1, 2, or REM sleep. Each MSLT
session was terminated either 15 minutes following the
initial onset of sleep or after 20 minutes in bed with
no sleep. A mixed two-factor 2 X 5 (group-by-session)
Sleep, Vol. 20, No.4, 1997
analysis of variance (ANOVA) was performed on the
latency to stage 1 sleep with session extracted as the
within-subjects factor.
The 8-minute version of the AAT was administered
at 0900, 1100, 1300, 1500, and 1700 hours. Participants were polysomnographically recorded while seated in an illuminated room within 3 feet of a wall upon
which an "X" made of black tape had been placed at
eye level. Participants were instructed t~ sit quietly
with their eyes open and focus on the black tape on
the wall. Following 1 minute of artifact-free recording
with eyes open, participants were instructed to sit quietly with their eyes closed for 1 minute. Each eyesopen and eyes-closed session was repeated three more
times so that a total of eight I-minute samples of artifact-free EEG were obtained. The MSLT and AAT
were scheduled 1 hour apart to reduce the likelihood
that participating in one task would influence the other.
For the AAT, EEG data were digitized using a sampling rate of 102.4 Hz with a digitizer sensitivity of
16 bits for ± 2.83 volts. Power spectral analyses of
eyes-open and eyes-closed EEG at 02 were calculated
using FFT on 10 second epochs within the alpha frequency band (8-12 Hz) using a bin size of 0.1 Hz.
The bins were averaged across the 8-12-Hz frequency
range to produce an estimate of alpha power in
squared microvolts. The ratio of mean eyes-closed to
mean eyes-open alpha power (i.e. the AAC) was then
calculated. A mixed two-factor 2 X 5 (group-by-session) ANa VA was performed on the AAC, and a
mixed three-factor 2 X 2 X 5 (group-by-eyelid-position-by-AAT-session) ANOV A was performed on
mean alpha power. In both analyses, session was extracted as the within-subjects factor.
Subjective sleepiness was assessed using the Stanford sleepiness scale (SSS) (15) and the visual analogue sleepiness scale (VASS) (16). The SSS consists
of seven statements describing progressive changes in
subjective sleepiness, ranging from" 1. Feeling active
and vital; alert; wide awake" to "7. Almost in reverie;
sleep onset soon; lost struggle to remain awake". Participants are asked to choose the statement that best
represents their present state of sleepiness. The V ASS
is a horizontal line approximately 100 mm in length,
labeled "Very Alert" on the left and "Very Sleepy"
on the right. Participants are asked to draw a vertical
mark on the line at the point corresponding to their
present state of sleepiness. V ASS scores are obtained
by measuring the distance of the mark from the left
end of the line (higher scores being associated with
more intense feelings of sleepiness). The SSS and
V ASS were administered every 1;2 hour during the daytime testing, including immediately prior to and following each MSLT and AAT session. To compare
overall differences in subjective sleepiness between
ALPHA AITENUATION TEST IN NARCOLEPSY
I
TABLE 1.
261
Raw data for AAT and MSLT
Narcoleptics
Session I
Session 2
Age
EO
EC
SOL
EO
EC
62
57
30
59
33
47
45
39
43
29
2.23
0.90
1.44
2.96
1.04
2.32
1.01
2.14
2.14
1.32
1.77
1.70
3.28
2.54
2.17
1.97
2.13
2.22
1.51
3.93
1.5
11.7
14.3
2.9
2.7
1.0
2.1
0.7
2.0
6.2
1.77
1.15
1.56
2.62
1.07
2.44
0.96
242
2.06
1.47
1.30
1.27
3.29
2.25
2.79
2.39
1.41
2.44
1.61
3.75
Session 4
Session 3
Session 5
SOL
EO
EC
SOL
EO
EC
SOL
EO
EC
SOL
1.0
9.4
2.2
2.1
2.4
0.5
1.0
1.2
0.3
7.5
1.43
1.10
1.35
2.31
1.09
1.19
1.22
2.64
1.36
1.78
1.42
1.44
2.85
2.43
2.78
1.59
1.14
2.40
1.15
4.78
1.7
3.3
6.7
0.5
3.3
4.0
0.8
1.2
1.7
5.3
1.38
1.19
1.91
2.79
1.34
2.71
1.19
2.45
2.43
1.81
1.35
1.29
2.52
2.50
2.94
4.28
2.17
2.45
1.50
3.35
0.7
0.7
2.7
0.7
4.1
0.8
4.8
6.9
1.0
3.2
1.19
0.93
1.82
2.61
1.15
3.14
1.35
2.77
2.39
2.08
1.08
1.27
2.26
2.10
3.04
4.78
1.69
2.90
1.75
4.92
0.8
2.0
4.6
2.9
8.3
5.0
3.9
4.2
1.2
5.8
Normals
Session 1
Session 2
Session 3
Session 5
Session 4
Age
EO
EC
SOL
EO
EC
SOL
EO
EC
SOL
EO
EC
SOL
EO
EC
SOL
56
50
45
39
52
56
38
29
35 a
28
1.80
2.09
1.08
0.93
0.94
1.08
2.66
1.75
1.32
1.20
7.84
3.46
1.24
4.55
5.21
2.31
3.83
2.34
4.48
3.29
8.0
1.8
20.0
5.7
20.0
2.8
20.0
6.9
1.7
12.7
2.19
3.04
0.79
0.85
0.97
1.02
3.24
1.81
1.62
1.55
6.30
3.78
0.91
4.98
5.56
1.80
3.91
2.64
4.76
3.18
9.3
5.5
5.4
8.5
6.5
8.2
11.1
6.8
4.0
5.5
2.16
3.06
0.75
1.69
1.21
1.02
3.27
1.79
1.95
1.37
4.98
3.65
0.91
5.61
5.56
1.97
3.90
2.53
7.49
3.46
17.2
4.1
11.2
8.7
8.2
2.7
12.2
5.0
3.7
6.7
2.48
3.10
0.82
1.92
1.15
0.94
3.28
1.71
2.33
1.78
6.69
2.90
1.00
5.71
4.89
1.88
3.45
2.66
6.09
3.72
8.9
1.6
9.6
7.5
10.4
2.9
20.0
7.5
20.0
20.0
2.67
2.65
0.89
1.61
1.13
0.81
3.27
1.44
2.44
1.54
4.22
3.17
1.09
5.76
5.45
1.54
3.78
2.79
6.64
4.58
8.8
6.7
9.9
20.0
20.0
3.2
20.0
17.2
2.2
20.0
AAT, alpha attenuation test; MSLT, multiple sleep latency test; EO, mean eyes-open alpha power for AAT session in ILy2; EC, mean
eyes-closed alpha power for AAT session in ILy2; SOL, latency to stage I for MSLT session in minutes.
a R.K. case study (see Discussion).
normals and narcoleptics, mean scores on the SSS and
VASS were calculated within each participant, and t
tests were used to assess group differences. t tests were
also used to assess group differences in SSS and V ASS
ratings obtained prior to and following each MSLT and
AAT session. The relationship among the subjective
sleepiness ratings and the MSLT and AAT was examined using Pearson correlation coefficients.
RESULTS
It
I....... Normal-Narcolepticl
3.5,-------=======-------.
3
AAT
2.5
"N
:J:
N
2
~
~ 1.5
'"
Raw data for each AAT and MSLT session are presented in Table 1.
----------------....---
'v
0.5
r.
O+-----~----~------+_----~----~r---~
900
1100
1300
1500
1700
TIME OF DAY (hr)
FIG. 1. Mean alpha (8-12 Hz) attenuation coefficient (AAC) on
the alpha attenuation test in narcoleptics and normals as a function
of time of day.
The analysis of the AAC showed no significant
group-by-session interaction, but significant main effects of group [F(l,I8) = 4.97, p = 0.039] and session
[F(4,72) = 3.28, P = 0.016] were found (see Fig. 1).
The ratio of eyes-closed to eyes-open alpha power
(AAC) ranged from 0.6 to 3.0 for the narcoleptics, and
from 0.9 to 5.9 for the normals. As predicted, collapsing across session, the mean AAC was significantly
lower for the narcoieptics (mean = lA, SD = 0.7)
compared to the normals (mean = 2.5, SD = 104).
Collapsing across group, the mean AAC was lowest
at the 1500-hours session.
The analysis of mean alpha power revealed a significant group by eyelid position interaction [F(l,I8)
6.52, P = 0.020] (see Fig. 2). Eyes-closed alpha
Sleep, Vol. 20, No.4, 1997
C. E. D. ALLOWAY ET AL.
262
I...... Normal-Narcolepticl
14r----------=======~~~--------,
",
12
C 10
i
----------------
-
~.. ~.. ~ ..~ ..~ ..'1. -' ~. -':' ~. '~:~""'; .•~ .... ", .•• ~." .~•• ~.:
>~
~,...
8
6
w
Cl
ct
Iii
4
2
o+-----~----~----_+----_+----~----~
900
1100
1300
1500
1700
~-------
O+-----~----~----_+-----+----~----~
1000
power ranged from 1.1 fJ. y2 to 4.9 fJ. y2 for the narcoleptics and from 0.9 fJ. y2 to 7.8 fJ. y2 for the normals.
Eyes-open alpha power ranged from 0.9 fJ. y2 to 3.1
fJ. y2 for the narcoleptics and from 0.7 fJ. y2 to 3.3 fJ. y2
for the normals. As predicted, during the eyes-closed
condition, mean alpha power was significantly lower
in the narcoleptics (mean = 2.4 fJ. y2, SO = 1.0 fJ. y2)
than the normals (3.9 fJ.Y2, SO = 1.8 fJ.Y2) [t(18) =
2.5, P = 0.022]. However, contrary to predictions, during the eyes-open condition, mean alpha power did not
differ between the groups (mean = 1.7 fJ. y2, SO = 0.6
fJ. y2 for the narcoleptics and mean = 1.7 fJ. y2, SO =
0.8 fJ.Y2 for the normals) [t(18) = 0]. No significant
group-by-session, eyelid position-by-session, or groupby-eyelid position-by-session interactions were found
for mean alpha power. The main effect of group (i.e.
collapsing across eyelid position and session) did not
reach significance; however, there was a trend for
mean alpha power to be lower in narcoleptics (2.0
fJ.Y2) compared to normals (2.8 fJ.Y2) [F(1,18) = 4.01,
P = 0.060]. There was a significant main effect for
eyelid position [F(1,18) = 19.76, p = 0.001] such that
collapsing across group and session, mean eyes-closed
alpha power (3.2 fJ. y2) was greater than mean eyesopen alpha power (1.7 fJ. y2). No session main effect
was found for mean alpha power.
MSLT
The analysis of latency to stage 1 sleep on the
MSLT showed no significant group-by-session interaction, but a significant main effect of group was
found [F(1,18) = 15.11, P = 0.001] (see Fig. 3). LaSleep, Vol. 20, No.4, 1997
1400
1600
1800
TIME OF DAY (hr)
TIME OF DAY (hr)
FIG. 2. Mean eyes-open and mean eyes-closed alpha (8-12 Hz)
power in f.L V2 on the alpha attenuation test in narcoleptics and normals as a function of time of day.
1200
FIG. 3. Mean latency in minutes to stage I sleep on the multiple
sleep latency test in narcoleptics and normals as a function of time
of day.
tency to stage 1 ranged from 0.5 to 15.5 minutes for
the narcoleptics and from 1.5 to 20.0 minutes for the
normals. Mean latency to stage 1 was significantly
shorter for narcoleptics (mean = 3.5 minutes, SO =
3.2 min) than normals (mean = 10.3 minutes, SO =
6.4 minutes). The session main effect did not reach
significance [F(4,72) = 2.45, p = 0.054].
Narcoleptics experienced a total of 26 REM-containing naps (i.e. sleep onset REM periods) and 24
non-REM (NREM) sleep naps on the MSLT. Eleven
of the narcoleptic REM naps contained stages 1, 2, and
REM sleep and 15 contained just stage 1 and REM
sleep. Twenty-three of the 24 narcoleptic NREM naps
contained both stage 1 and stage 2 sleep. A total of
six REM naps occurred at the 1200 and 1400 hours
naps, five occurred at the 1000 and 1600 hours naps,
and four occurred at the 1800 hours nap. Two narcoleptics had REM sleep on all five naps, five other narcoleptics had two or more REM containing naps, and
three had only one nap containing REM sleep. Of
these three narcoleptics, one had a sleep onset REM
period during the nocturnal sleep.
Normals experienced a total of 44 NREM sleep naps
and six naps with no sleep onset. Thirty of the 44
normal NREM naps contained stage 1 and stage 2
sleep, whereas 14 contained only stage 1 sleep. One
of the normal participants (R.K.) had two naps that
contained REM sleep: one at 1400 hours and the other
at 1600 hours.
AAT and MSLT
The relationship between the AAT and the MSLT
was investigated by calculating for each participant the
fl
.,..
ALPHA ATTENUATION TEST IN NARCOLEPSY
;
mean AAC and mean latency to stage 1 on the MSLT
and then correlating mean AAC and mean MSLT latency for narcoleptics and normals using Pearson correlation coefficients. Mean AAC correlated significantly with mean latency to stage 1 on the MSLT for
narcoleptics (r = 0.75, P = 0.006) but not for normals
(r = 0.15, P = 0.342).
SSS and VASS
Overall, mean SSS scores tended to be higher for
narcoleptics (mean = 3.2, SO = 1.0) than normals
(mean = 2.4, SO = 0.6) [t(18) = 2.09, p = 0.051].
Similarly, mean VASS scores tended to be higher for
narcoleptics (mean = 40.7, SO = 19.7) than normals
(mean = 25.6, SO = 12.0) [t(18) = 2.06, p = 0.054].
t tests were used to compare narcoleptic and normal
subjective sleepiness ratings obtained just prior to and
immediately following each AAT and MSLT session.
For these analyses, the alpha level was set at 0.01 in
order to allow for the fact that more than 20 t tests
were computed. As such, p values between 0.01 and
0.05 are reported as indicating trends only. In regard
to the pre- and post-AAT mean subjective sleepiness
ratings, narcoleptics rated themselves as sleepier than
normals on the VASS administered following the AAT
session at 1500 hours (narcoleptics: mean = 51.2, SO
= 29.9; normals: mean = 25.6, SO = 17.9) [t(18) =
2.32, p = 0.032]. However, no other differences were
observed in subjective sleepiness ratings obtained preand post-AAT. In regard to the pre- and post-MSLT
subjective sleepiness ratings, narcoleptics rated themselves as sleepier than normals on the pre-MSLT (1000
hours) SSS (narcoleptics: mean = 3.7, SO = 1.8; normals: mean = 2.3, SO = 1.1) [t(18) = 2.15, P =
0.045], the pre-MSLT (1200 hours) VASS (narcoleptics: mean = 38.8, SO = 25.9; normals: mean = 16.5,
SO = 11.1) [t(18) = 2.50, P = 0.022], the pre-MSLT
(1400 hours) VASS (narcoleptics: mean = 38.7, SO
= 24.2; normals: mean = 19.1, SO = 11.7) [t(18) =
2.31, P = 0.033], the pre-MSLT (1600 hours) SSS
(narcoleptics: mean = 3.4, SO = 1.3; normals: mean
= 2.0, SO = 0.8) [t(18) = 2.94, P = 0.009], the preMSLT (1600 hours) VASS (narcoleptics: mean =
44.8, SO = 28.4; normals: mean = 19.0, SO = 15.0)
[t(18) = 2.54, p = 0.021], the post-MSLT (1600 hours)
SSS (narcoleptics: mean = 3.9, SO = 1.5; normals:
mean = 2.6, SO = 0.8) [t(18) = 2.36, P = 0.030], the
post-MSLT (1600 hours) VASS (narcoleptics: mean =
52.4, SO = 25.2; normals: mean = 27.7, SO = 14.4)
[t(18) = 2.69, P = 0.015], the pre-MSLT (1800 hours)
SSS (narcoleptics: mean = 3.0, SO = 1.6; normals:
mean = 1.8, SO = 0.8) [t(18) = 2.17, P = 0.044], and
the post-MSLT (1800 hours) SSS (narcoleptics: mean
263
= 3.4, SO = 0.8; normals: mean = 2.5, SO = 0.9)
[t(18) = 2.38, P = 0.029].
To examine the relationship between the subjective
sleepiness measures and the AAT, mean scores for the
SSS and VASS administered prior to and following
each AAT session were first calculated for each participant, then Pearson correlation coefficients were calculated between mean AAC and pre- and post-AAT
mean SSS and V ASS for narcoleptics and normals (i.e.
a total of four correlations were calculated for both
narcoleptics and normals). None of these correlations
reached significance for narcoleptics. For normals,
there was a significant correlation between mean AAC
and post-AAT mean VASS (r = -0.61, p = 0.032).
The relationship of subjective sleepiness measures
with the MSLT was investigated in a similar manner.
For each participant, mean latency to stage I on the
MSLT and pre- and post-MSLT mean SSS and V ASS
were calculated, then Pearson correlation coefficients
were calculated between mean latency and pre- and
post-MSLT mean SSS and VASS for narcoleptics and
normals. None of these correlations reached significance for narcoleptics. For normals there was a significant correlation between mean latency to stage 1
on the MSLT and pre-MSLT mean SSS (r = 0.74, p
= 0.007).
DISCUSSION
The present study is the first to demonstrate that the
AAT can be used to distinguish a clinical population
of excessively sleepy individuals such as narcoleptics
from normal controls. As predicted, the ratio of mean
eyes-closed to mean eyes-open alpha power (i.e. the
AAC) was significantly smaller for narcoleptics than
normals, suggesting that increased physiological sleepiness is associated with lower AACs. These findings
are consistent with those of Stampi et al. (7) who observed a decrease in AACs in normals throughout 40
hours of sleep deprivation. Studies of experimentally
sleep-deprived normals and shiftworkers (9) have
demonstrated that, during maximal sleepiness, alpha
power is lower during eyes-closed conditions than during eyes-open conditions, and during maximal alertness, alpha power is higher during eyes-closed conditions than during eyes-open conditions.
In the present study, it was predicted that narcoleptics would demonstrate lower mean eyes-closed alpha
power and higher mean eyes-open alpha power than
normals. However, it was found that mean eyes-closed
alpha power was significantly reduced in narcoleptics
compared to normals; whereas mean eyes-open alpha
power did not differ between narcoleptics and normals.
It appears that when the eyes were open, the illumination and the task of focusing on a target on the wall
Sleep, Vol. 20, No.4, 1997
264
C. E. D. ALLOWAY ET AL.
may have acted as alerting stimuli for the narcoleptics,
enabling the suppression of the latent physiological
sleepiness that was observed once they closed their
eyes. Thus, the significantly reduced AAC in narcoleptics was mediated almost entirely by the eyesclosed condition of the AAT.
The decreased AAC and increased eyes-closed alpha power in narcolepsy mimics the effects of shiftwork and experimentally induced sleep deprivation in
normals. However, it must be stressed that in the present study no evidence was found to suggest that the
narcoleptics were sleep deprived. Mean nocturnal
sleep length did not differ significantly between normals and narcoleptics for either the sleep diary that
was completed for 7 days prior testing or the nocturnal
polysomnography carried out immediately preceding
the daytime testing. Furthermore, the nocturnal polysomnography showed no significant differences between narcoleptics and normals in sleep efficiency or
in the percentage of time spent in stages 2, 3, 4, and
REM sleep. The only significant differences found between narcoleptics and normal sleepers during the nocturnal polysomnography were that latencies to stage 1
and REM sleep were shorter, and the percentage of
time spent in stage 1 was higher for narcoleptics. Thus,
there was no evidence to suggest that the reduced
AACs observed during seated daytime wakefulness in
narcoleptics could be accounted for by the effects of
nocturnal-sleep deprivation. In the absence of sleep deprivation in the narcoleptics, some neurophysiological
mechanism is producing an elevation in the physiological need for sleep in narcoleptics.
The AAC correlated significantly with latency to
stage 1 on the MSLT in narcoleptics but not in normals. This result is contrary to the findings of Stampi
et al. (7) who demonstrated that the AAC correlated
with latency to sleep onset on the MSLT in eight out
of 10 sleep-deprived normal participants. However,
Stampi et al. (7) measured latency to sleep onset within minutes of measuring eyes-open and eyes-closed alpha power; whereas in the present study, more than 45
minutes separated the two measurements, and neither
the normals nor the narcoleptics were experimentally
deprived of sleep. We propose that the AAT and MSLT
produced relatively independent measures of sleepiness in our nonsleep-deprived normal participants because they were scheduled to begin on alternate hours
throughout the day. These tests were not scheduled to
run consecutively to avoid both sleepiness priming and
sleep inertia effects. It was desirable that participants
not become sleepy and fall asleep faster on the MSLT
because they had just finished sitting quietly while
staring at the wall or closing their eyes and vice versa.
Subjective sleepiness ratings obtained just prior to
or following four of the AAT sessions did not differ
Sleep, Vol. 20, No.4, 1997
significantly between narcoleptics and normals, but
following the AAT session at 1500 hours, narcoleptics
tended to rate themselves as sleepier than normals on
the V ASS (i.e. p = 0.032). In contrast, although sleepiness ratings obtained just prior to or following the
MSLT sessions were more frequently higher for narcoleptics than normals, all but one of these differences
failed to reach the 0.01 level of significance. The tendency for subjective sleepiness ratings centered around
the MSLT, but not the AAT, to differentiate narcoleptics and normals may have been due in part to the
environment in which the subjective sleepiness ratings
were obtained. For the AAT, sleepiness ratings were
obtained while participants were seated in a chair. By
contrast, for the MSLT, sleepiness ratings were obtained while participants were lying down in bed, in
anticipation of or just following a nap opportunity.
Thus, in situations that promote sleepiness (i.e. lying
down for the MSLT vs. sitting in a chair for the AAT),
subjective sleepiness ratings tended to be higher for
narcoleptics than normals. What is noteworthy is that
the AAT successfully differentiated the narcoleptics
and normals in the absence of group differences in
subjective sleepiness ratings.
One of the normal participants in the present study
(RK.) presents an interesting case study demonstrating
the merit of the AAT in the situation of false-positive
MSLT results. RK., age 35 years, is a normal sleeper
who reported sleeping an average of 6.9 hours each
night during the week before testing and obtained 8.7
hours of sleep the night before testing. RK. reported
no need to nap during the day and no history of excessive daytime sleepiness, cataplexy, or sleep paralysis, although he did report having unusual visual or
auditory experiences (i.e. hypnagogic hallucinations)
one to five times during his lifetime. He was not taking
any medications, nor did he demonstrate symptoms of
depression (his score was 2 out of 63 on the BDI).
RK. experienced two sleep onset REM periods on the
MSLT and his mean latency to stage 1 sleep was 3.6
minutes, suggesting that he met the diagnostic criteria
for narcolepsy. [The occurrence of sleep onset REM
periods in otherwise normal sleepers has been documented by Rosenthal et al. (17) who reported that 15%
(i.e. 11 of 73) of their drug-free normal sleepers
(asymptomatic for narcolepsy) experienced two or
more sleep onset REM periods on the MSLT.] However, RK. subjectively rated himself as alert throughout the day (mean SSS = 1.4, mean VASS = 10.2).
Moreover, his AAT results gave no indication of excessive sleepiness. On the contrary, they suggested an
above-average degree of physiological alertness.
RK. 's mean AAC (3.1) was higher than the average
AAC for normals (2.5) as was his mean eyes-closed
alpha power (5.9 j.1V2 vs. 3.9 j.1V2). His mean eyes-
~)
265
ALPHA ATTENUATION TEST IN NARCOLEPSY
,""
open alpha power (1.9 IL V2) was comparable with that
obtained in narcoleptics and normals (1.7 IL V2). Thus,
the present study has documented the ability of the
AAT to confirm the presence of physiological alertness
in a normal participant matching the MSLT diagnostic
criteria for narcolepsy.
The AAT may be instrumental in the clinical assessment of excessive daytime sleepiness. It is quick
and simple to administer and is free of the serious
limitations associated with the assessment of sleepiness via the MSLT-namely, a floor effect in excessively sleepy populations (5), the confounding of
sleepiness with the ability to fall asleep (i.e. "sleepability") (4), and the reliance on the presence of a polysomnographer to sleep score the EEG record "on
line" during each nap opportunity (7). Furthermore,
the AAT is an ideal measure for assessing sleepiness
in field studies or in actual work environments because
it is nonintrusive (i.e. it does not necessitate a sleepconducive environment, thereby inducing less sleepiness than the MSLT) (7). Although the MSLT is perhaps the best measure for documenting the occurrence
of sleep onset REM periods (18), it clearly has the
potential to produce misleading evaluations of physiological sleepiness (due to the sleepability confound
and floor effects). The inclusion of the AAT in the
clinical assessment of patients with sleep-related complaints appears warranted. The AAT may easily be accommodated into the MSLT paradigm by scheduling
these two tests to occur on alternate hours throughout
the day as was done in the present study.
We recommend that future research investigate the
efficacy of the AAT in the evaluation of pharmacological-treatment efficacy for excessive daytime sleepiness. Given that studies of subjectively effective stimulant medications in narcoleptics have failed to demonstrate a reduction in sleepiness on the MSLT (6), it
would be advantageous for future studies to investigate
the ability of the AAT to detect variations in sleepiness/alertness following clinically effective pharmacological treatment. Future research is also needed to
test the ability of the AAT to assess sleepiness in clinical populations other than narcoleptics. For example,
it is of interest whether the AAT would identify increased physiological sleepiness in sleep onset insomniacs (who would have trouble falling asleep on the
MSLT) and in other patients complaining of excessive
daytime sleepiness (e.g. patients with central and obstructive sleep apnea).
Some researchers have suggested that the AAT may
be limited in its ability to assess sleepiness in highalpha producers. Stampi et al. (19) reported in their
study of sleep-deprived normals that in individuals
who produced extremely high AACs (labeled "high
alpha producers"), the AAC tended not to correlate
with latency-to-sleep onset on the MSLT. In the study,
the MSLT was administered 5 minutes following the
administration of the AAT. Stampi et al. (19) reported
that although the high alpha producers demonstrated a
high level of alertness on the AAT (i.e. high AAC),
this degree of alertness was not related to latency to
sleep onset on the MSLT. That is, presumably, these
high alpha producers were still able to fall asleep within the 20-minute nap opportunity provided by the
MSLT. We propose that these findings may reflect
more on the sleepability confound of the MSLT rather
than on a specific limitation of the AAT. Heitmann et
al. (20), in their study of shiftworkers working a nightshift, reported that the AAC correlated best with subjective sleepiness measures in individuals producing
"medium"-level AACs. This suggests that the AAT
may be less sensitive to subjective sleepiness in individuals who demonstrate extreme levels of alertness or
sleepiness on their AACs. However, in the present
study, subjective sleepiness ratings administered prior
to and following each AAT correlated with the AAC
in just four participants (two narcoleptics and two normals), and no apparent pattern was observed for persons having higher or lower level AACs. We suggest
that the AAT is a measure of the underlying or latent
physiological sleepiness, whereas subjective sleepiness
ratings tend to reflect manifest sleepiness, which is influenced by moment-to-moment fluctuations in alerting stimuli within the environment (2); thus the lack
of correlation among these measures in some individuals may not present a serious consequence.
To conclude, the present study is the first to demonstrate that normal sleepers can be distinguished from
excessively sleepy narcoleptics on the basis of EEG
power spectral analysis of alpha power during seated
wakefulness.
REFERENCES
1. Carskadon MA, Dement We. Sleepiness and sleep state on a
90-minute schedule. Psychophysiology 1977;14:127-33.
2. Carskadon MA, Dement WC. The multiple sleep latency test:
what does it measure? Sleep 1982;5:S67-72.
3. Diagnostic Classification Steering Committee, Thorpy MJ,
Chairman. International classification of sleep disorders: diagnostic and coding manual. Rochester, MN: American Sleep Disorders Association, 1990.
4. Broughton RJ. Qualitatively different states of sleepiness. In:
Broughton RJ, Ogilvie RD, eds. Sleep, arousal and peiformance. Boston: Birkhauser, 1992:45-59.
5. Sugarman JL, Walsh JK. Physiological sleep tendency and ability to maintain alertness at night. Sleep 1989;12:106-12.
6. Hartse KM. Roth T, Zorick FJ. Daytime sleepiness and daytime
wakefulness: the effect of instruction. Sleep 1982;5:SI07-18.
7. Stampi C, Stone P, Michimori A. A new quantitative method
for assessing sleepiness: the alpha attenuation test. Work &
Stress 1995;9:368-76.
8. Akerstedt T, Torsvall L, Gillberg M. Sleepiness in shiftwork. A
review with emphasis on continuous monitoring of EEG and
EOG. Chronobiollnt 1987;4:129-40.
Sleep, Vol. 20, No.4, 1997
266
C. E. D. ALLOWAY ET AL.
9. Akerstedt T, Torsvall L, Gillberg M. Sleepiness in laboratory
and field experiments. In: Koella Wp, Ruther E, Schulz H, eds.
Sleep 1984. Stuttgart: Gustav Fischer Verlag, 1985:125-6.
10. Stampi C, Stone P, Michimori A. The alpha attenuation test: a
new quantitative method for assessing sleepiness and its relationship to the MSLT. Sleep Res 1993;22: 115.
11. Alloway CED, Ogilvie RD, Shapiro CM. The alpha attenuation
test: assessing physiological sleepiness in narcolepsy. Sleep Res
1995;24:187.
12. Beck AT. Depression: Causes and treatment. Philadelphia, PA:
University of Pennsylvania Press, 1967.
13. Thorpy MJ. Report from the American Sleep Disorders Association: the clinical use of the multiple sleep latency test. Sleep
1992;15:268-76.
14. Rechtschaffen A, Kales A. A manual of standardized terminology, techniques, and scoring system for sleep stages of human
subjects. Los Angeles: Brain Research Institute, 1968.
15. Hoddes E, Dement W, Zarcone V. The development and use of
16.
17.
18.
19.
20.
the Stanford sleepiness scale (SSS). Psychophysiology 1971;9:
150.
Folstein MF, Luria R. Reliability, validity and clinical application of the visual analogue mood scale. Psychol Med 1973;3:
479-86.
Rosenthal L, Bishop C, Helmus T, Roehrs TA, Brouillard L,
Roth T. The frequency of multiple sleep onset REM periods
among subjects with no EDS. Sleep Res 1995;24:331.
Mitler MM, Van den Hoed J, Carskadon MA, Richardson G,
Park R, Guilleminault C, Dement We. REM sleep episodes during the multiple sleep latency test in narcoleptic patients. Electroencephalogr Clin Neurophysiol 1979;46:479-81.
Stampi C, Michimori A, Aguirre A. Comparison between three
versions of the alpha attenuation test for the objective assessment of sleepiness. Sleep Res 1995;24:499.
Heitmann A, Stampi C, Anandan S. The alpha attenuation test:
its first application for objective assessment of sleepiness in field
studies. Sleep Res 1995;24:100.
-I)
Sleep, Vol. 20, No.4, 1997