Effects of sleep fragmentation on the arousability to resistive loading in NREM and REM sleep in normal men.

Abstract

In healthy subjects, arousability to inspiratory resistive loading is greater during rapid eye movement (REM) sleep compared with non-REM (NREM) sleep but is poorest in REM sleep in patients with sleep apnea. We therefore examined the hypothesis that sleep fragmentation impairs arousability, especially from REM sleep.
Two blocks of 3 polysomnographies (separated by at least 1 week) were performed randomly. An inspiratory-loaded night followed either 2 undisturbed control nights (LN(C)) or 2 acoustically fragmented nights (LN(F)) SETTING: Sleep laboratory.
Sixteen healthy men aged 20 to 29 years.
In both loaded nights, an inspiratory resistive load was added via a valved facemask every 2 minutes during sleep and turned off either when arousal occurred or after 2 minutes.
During LN(F), arousability remained significantly greater in REM sleep (71% aroused within 2 minutes) compared with stage 2 (29%) or stage 3/4 (16%) sleep. After sleep fragmentation, arousability was decreased in stage 2 sleep (LN(F): 29%; LN(C): 38%; p < .05) and low in early REM sleep, increasing across the night (p < .01). In stage 3/4 sleep, neither an attenuation nor a change across the night was seen after sleep fragmentation.
Mild sleep fragmentation is already sufficient to attenuate arousability in stage 2 sleep and to decrease arousability in early, compared with late, REM sleep. This means that sleep fragmentation affects the arousal response to increasing resistance and that the effects are different in stage 2 and REM sleep. The biologic reason for this increase in the arousal response in REM sleep across the night is not clear.

Full text

Page 1

SLEEP, Vol. 29, No. 4, 2006
SLEEP DISORDERED BREATHING
525
INTRODUCTION
AROUSAL IS BELIEVED TO BE AN ESSENTIAL PRO-
TECTIVE MECHANISM AGAINST LIFE-THREATENING
ASPHYXIA DURING SLEEP AND CONTRIBUTES TO day-
time sleepiness and blood pressure elevation1,2 in patients with
obstructive sleep apnea-hypopnea syndrome (OSAHS). Apneas
are mostly, although not exclusively, terminated by arousals.3,4 A
major consequence of repetitive arousals is sleep fragmentation,
which is marked in patients with OSAHS. Sleep fragmentation
causes increases in the arousal threshold to acoustic stimuli and is
associated with subsequent daytime sleepiness.5,6
Many stimuli can cause arousal. Although auditory arousal in
the absence of apnea and hypoxemia is associated with significant
nighttime blood pressure elevation,7 Brooks et al have shown that
only respiratory arousals lead to sustained daytime hypertension
in dogs after 3 months.1 As mechanical respiratory stimuli seem
to be most relevant in the context of OSAHS, inspiratory resistive
loading was used in this study to determine the arousal thresh-
old.
Rapid eye movement (REM) sleep has important influences
on the arousal responses to respiratory stimuli.8,9 In patients with
OSAHS, breathing abnormalities and hypoxemia are worst during
REM sleep.10 In healthy adults, breathing during REM sleep is of-
ten irregular.11 Morrell et al have shown, in healthy humans, that
“REM sleep is associated with partial respiratory load compensa-
tion suggesting that exacerbation of sleep disordered breathing in
REM (compared to non-REM [NREM]) sleep is unlikely to be
secondary to an inability to overcome increases in upper airway
resistance.”12
Much less is known about arousability and the effects of sleep
fragmentation on arousal responses from REM sleep, compared
with NREM sleep. One study in patients with OSAHS found a
more-pronounced effect of acoustic sleep fragmentation on the
arousal response during REM sleep, compared with NREM
sleep,13 while another study in dogs did not come to the same con-
clusions.14 In light of this controversial information and the im-
portant role in patients with OSAHS, we were therefore interested
in defining the effect of sleep fragmentation, per se, on arousal
responses in REM sleep in healthy humans. A clear understand-
ing of arousal responses before and after sleep fragmentation in
healthy humans will help to interpret arousal responses in patients
with OSAHS. Only male subjects were recruited because OSAHS
is more common in men than women15,16 and because normal men
are more vulnerable to load-induced hypoventilation than are
women.17
We hypothesized that, firstly, the arousals induced by inspi-
ratory resistive loading after sleep fragmentation remain more
frequent and faster in REM sleep, compared with NREM sleep,
in healthy subjects and that, secondly, the arousal response to in-
spiratory resistive loading is decreased after sleep fragmentation
compared with the response after normal sleep, both in REM and
NREM sleep. In addition, we wanted to examine whether sleep
Effects of Sleep Fragmentation on the Arousability to Resistive Loading in NREM
and REM Sleep in Normal Men
Jeannette Zavodny1; Corinne Roth, PhD1; Claudio L. Bassetti, MD2; Johannes Mathis, MD1; Neil J. Douglas, MD3; Matthias Gugger, MD1
1Sleep Medicine Center, Inselspital, University of Berne, Berne, Switzerland; 2Department of Neurology, University Hospital Zurich, Zurich, Switzer-
land; 3Department of Respiratory Medicine, University of Edinburgh, Edinburgh, UK
Sleep Fragmentation and Arousability—Zavodny et al
Disclosure Statement
This was not an industry supported study. Dr. Bassetti has participated in
a speaking engagement supported by ResMed. Dr. Douglas has received
research support from ResMed Ltd; and chairs for the international medical
advisory board of ResMed. Drs. Zavodny, Roth, Mathis, and Gugger have
indicated no conflicts of interest.
Submitted for publication May 2005
Accepted for publication December 2005
Address correspondence to: Professor Matthias Gugger, MD, Respiratory
and Sleep Medicine, Inselspital, 3010 Berne, Switzerland; Tel: +41 31 632 21
11; Fax: +41 31 632 98 33; E-mail: Matthias.Gugger@insel.ch
Study Objective: In healthy subjects, arousability to inspiratory resistive
loading is greater during rapid eye movement (REM) sleep compared with
non-REM (NREM) sleep but is poorest in REM sleep in patients with sleep
apnea. We therefore examined the hypothesis that sleep fragmentation
impairs arousability, especially from REM sleep.
Design: Two blocks of 3 polysomnographies (separated by at least 1
week) were performed randomly. An inspiratory-loaded night followed ei-
ther 2 undisturbed control nights (LNC) or 2 acoustically fragmented nights
(LNF).
Setting: Sleep laboratory.
Participants: Sixteen healthy men aged 20 to 29 years.
Interventions: In both loaded nights, an inspiratory resistive load was
added via a valved facemask every 2 minutes during sleep and turned off
either when arousal occurred or after 2 minutes.
Measurements and Results: During LNF, arousability remained signifi-
cantly greater in REM sleep (71% aroused within 2 minutes) compared
with stage 2 (29%) or stage 3/4 (16%) sleep. After sleep fragmentation,
arousability was decreased in stage 2 sleep (LNF: 29%; LNC: 38%; p <
.05) and low in early REM sleep, increasing across the night (p < .01). In
stage 3/4 sleep, neither an attenuation nor a change across the night was
seen after sleep fragmentation.
Conclusions: Mild sleep fragmentation is already sufficient to attenuate
arousability in stage 2 sleep and to decrease arousability in early, com-
pared with late, REM sleep. This means that sleep fragmentation affects
the arousal response to increasing resistance and that the effects are dif-
ferent in stage 2 and REM sleep. The biologic reason for this increase in
the arousal response in REM sleep across the night is not clear.
Keywords: Arousal, inspiratory resistive loading, sleep apnea, sleep
stage dependent
Citation: Zavodny J; Roth C; Bassetti CL et al. Effects of sleep fragmen-
tation on the arousability to resistive loading in NREM and REM sleep in
normal men. SLEEP 2006;29(4): 525-532.

Page 2

SLEEP, Vol. 29, No. 4, 2006
fragmentation affects the sleep-time–dependent evolution of the
arousal response within a sleep stage.
METHODS
Subjects
Sixteen healthy men aged 24 ± 0.7 years (mean ± SEM) were
studied. Subjects were recruited from a student population by ad-
vertisement on notice boards in our hospital and in the Technical
College in Biel. Only subjects with no sleep complaints assessed
by our sleep questionnaire were accepted. All were nonobese
(body mass index: 22 ± 0.3 kg/m2), nonsmokers, and regular noc-
turnal sleepers without any sleep disturbance or habitual snor-
ing, and none were taking any medication. They had an Epworth
Sleepiness Scale score of 6 ± 0.8. A screening night served to
exclude any subject with sleep-disordered breathing, and only
subjects with a respiratory disturbance index (RDI) less than 10
qualified for the study. All subjects gave written informed consent
to participate in the study and were paid for participation. Ethical
permission for the study was obtained from the ethics committee
of the Faculty of Medicine of the University and the Canton of
Berne.
Study Protocol

The study protocol is summarized in Figure 1. The screening
night, preceding the start of the study by at least 3 days, served
to familiarize subjects with the equipment. Subjects wore a face-
mask for acclimatization, but no load was applied. They had an
RDI of 2.9 ± 0.5 and a sleep efficiency of 88% ± 2.6%. Following
the screening night, subjects spent 6 study nights in the sleep lab-
oratory, and all 7 nights were studied using standard polysomnog-
raphy. The study nights were divided into 2 blocks of 3 nights, the
blocks separated from each other by at least a week. The subjects
were randomly allocated to start with either 2 undisturbed con-
trol nights or 2 acoustically fragmented nights. During the third
night of each block (LN: loaded night), an inspiratory resistive
load was applied via a facemask. All studies were performed be-
tween 10:00 PM and 7:00 AM, and the study time was matched to
the usual time in bed of each subject. After the study nights, the
subjects left the sleep laboratory and continued their normal daily
routine activities. They had to refrain from caffeinated beverages
and from alcohol 1 day before and throughout the 3 days of each
study block apart from 1 cup of coffee in the morning. Naps were
not allowed, and compliance with this instruction was verified
by continuous recording of wrist activity (Actiwatch Cambridge
Neurotechnology, Cambridge, UK).
The instrumentation was restricted to the minimum required
for investigating the questions outlined in the introduction be-
cause the sleep-disturbing effects of equipment often impair data
collection, particularly in REM sleep. To reduce the sleep-disturb-
ing effect of the performed load applications, the occlusion of the
inspiratory airflow was not complete. Only a single level of resis-
tance was used to decrease the number of possible comparisons
and to be able to compare the results between the different sleep
stages.
All measurements were recorded with a computerized record-
ing system (Embla; Flaga hf. Medical Devices, Reykjavik, Ice-
land). Sleep state was recorded using the standard placements
of electroencephalogram (C4/A1, O1/A2), two frontal electro-
encephalogram channels (F3/A2, F4/A1), left and right electro-
oculograms, and submental electromyogram. Nasal airflow was
monitored using a pressure transducer airflow sensor (PTAF2;
Pro-Tech Services Inc., Woodinville, WA, USA) and nasal prong
devices (Pro-Tech Services Inc.), thoracic and abdominal re-
spiratory effort using a piezo-electric sensor (Ultima DL Effort
Sensor; Braebon Medical Corporation, Kanata, ON, Canada).
Arterial oxygen saturation was recorded using pulse oximetry
(Oximeter Flex Sensor 8000J; Nonin Medical Inc., Minneapolis,
MN, USA). During LN, airflow was monitored via a sealed face-
mask (modified full facemask; ResMed Ltd, North Ryde, NSW,
Australia) with built-in inspiratory and expiratory valves serving
to apply the inspiratory resistance. The mask dead space was ap-
proximately 60 mL, depending on individual face configuration.
A leak detector, consisting of a perforated polyethylene catheter
connected to a capnometer (OxiCap 4700; Ohmeda, Louisville,
CO, USA), was attached to the circumference of the facemask
cushion and was set at maximum gain.18
Sleep stages and awakenings were scored for 30-second ep-
ochs using the criteria of Rechtschaffen and Kales.19 Arousal was
defined as an episode lasting 3 seconds or longer, according to the
criteria of the American Sleep Disorders Association.20 Figure 2
shows an example of an arousal during REM sleep. Apneas and
hypopneas were defined as an event lasting 10 seconds or lon-
ger according to the recommendations of the American Academy
of Sleep Medicine.21 Flow-limitation events were defined as any
series of 2 or more breaths (lasting at least 10 seconds) that had
a flattened or nonsinusoidal appearance on the inspiratory nasal-
cannula flow signal and ended abruptly with a return to breaths
with sinusoidal shape.22
Sleep Fragmentation
Sleep was fragmented on 2 consecutive nights because, in a
Screening
night

2
fragmented
nights

2 control
nights

Loaded
night
Loaded
night
2 control
nights

2
fragmented
nights

Loaded
night
Loaded
night
Figure 1—Study design showing a screening night followed by 2
blocks (allocated in random order) of 3 study nights.
526
Sleep Fragmentation and Arousability—Zavodny et al

Page 3

SLEEP, Vol. 29, No. 4, 2006
preliminary study, 1 night of sleep fragmentation combined with
partial sleep deprivation might have been insufficient to affect
arousal.23 Therefore, every 2 minutes (interseries interval) from
the onset of undisturbed stage 2, stage 3/4, or REM sleep, tone
series of 1000 Hz generated by an audiometer (GSI 16; Grason-
Stadler Inc., Littleton, MA) were presented via 2 loudspeakers
positioned beside the subject’s head until an arousal response was
obtained. The duration and volume of tones to produce arousal
were increased because normal subjects may acclimatize to arous-
ing stimuli.6,24-26 The first tone was applied for 5 seconds at 60 dB.
If no arousal was noted within 15 seconds (intertone interval), a
second tone was sent at 60 dB, but this time lasting 10 seconds.
If no response was observed, tone duration was kept constant (10
seconds), and sound intensity was augmented in 10-dB incre-
ments every 15 seconds until 90 dB. The next 3 tones were set
at a maximal decibel level of 95 dB. The first 8 tones were fre-
quency-modulated sine waves (warble tones). The ninth tone was
a narrow band noise at 90 dB and was repeated twice if no arousal
occurred. The intertone intervals were 15 seconds, and the inter-
series intervals were 2 minutes of undisturbed stage 2, stage 3/4,
or REM sleep. Each tone series was terminated by the technicians
upon signs of electroencephalographic arousal or after a total of
11 tones were presented.
Inspiratory Resistance
In accordance with previous studies,23,27,28 sustained resistive
loading was accomplished by the addition of a 25-cm H2O per
liter per second linear resistive load to the inspiratory line by turn-
ing a 3-way tap in the inspiratory line. The tap was turned only
during expiration to avoid disruption of the breathing cycle. The
tap was in a different room from the subject, and careful attention
was paid to avoid any noise or contact of the tubing that might
have disturbed sleep. The resistance was added every 2 minutes
from the onset of undisturbed stage 2, stage 3/4, or REM sleep.
After spontaneous sleep-stage changes, the resistance was not ap-
plied for at least 1 minute. In case of an occasional respiratory
event, the resistance was not applied for at least 1 minute after
termination of the event. If no arousal or awakening (> 15 sec-
onds) occurred during the resistive breathing, the resistance was
switched off after 2 minutes and again not reapplied for at least
2 minutes. If arousal or awakening occurred, the resistance was
switched off. “Time to arousal” was defined as the time between
the start of loading and the occurrence of an arousal within the
intervention. If no arousal occurred within the 2 minutes of load-
ing, it was defined as “nonarousal”.
Control Periods
Control periods of 2 minutes were selected on the control
nights to look at arousals during them. These periods were ret-
rospectively marked in the recordings during stage 2, stage 3/4,
and REM sleep, according to the inspiratory loading protocol and
regardless of arousal during these control periods. The number of
arousals occurring within these periods was compared with the
number of arousals occurring within 2 minutes of the application
of the inspiratory resistance.
Data Analysis
The results are presented as mean ± SEM. The data were first
analyzed with a 2-way analysis of variance for repeated measures
(rANOVA) using the SAS General Linear Model procedure with
Greenhouse-Geisser correction. In case of significant effects, con-
trasts were further tested by the 2-tailed Wilcoxon signed-rank
test and the Friedman test, respectively. All tests were paired. The
pooled numbers of arousals and “nonarousals” were compared
between the different sleep stages within a loaded night and be-
tween loaded nights (all-night data and per sleep stage) with 2 ×
2 contingency tables and χ2 tests of independence. Results with a
p value < .05 were considered significant.
The total number of arousals includes all detected arousals.
They include arousals in which retrospective analysis showed
that the stimulus was not applied strictly according to the study
protocol, as well as additional arousals within the intervention
interval. These arousals, which were 2% of all arousals, were not
included in any further analysis.
Sleep variables, as well as number of arousals of the 2 control
nights, were averaged to compare the control night (CN) mean
with the loaded and fragmented nights, respectively. In order to
take into account the numerous “nonarousals” when analyzing
“time to arousal”, “time to arousal” was ranked, and the highest
rank was assigned in event of a “nonarousal”.
To represent the sleep-time–dependent evolution of the arousal
response to inspiratory loading, the data obtained on the loaded
nights were broken into thirds of the night, and the first third was
compared with the last third of the night, as in a study of Philip
et al.26 To avoid an uneven distribution of the data in the differ-
ent sleep stages (since sleep stages are not equally distributed
throughout the night), we compared the results obtained with
the first third of load applications to the last third in each sleep
stage.
RESULTS
Effectiveness of Sleep Fragmentation
To fragment sleep, 67.6 ± 3.6 tone series with an average of 3.9
± 0.4 tones were applied in the first fragmented night (FN1), of
which 66.0 ± 3.9 (97% ± 1%) resulted in arousals. In the second
fragmented night (FN2) there were 69.1 ± 2.9 tone series with a
mean of 6.0 ± 0.4 tones and 61.2 ± 3.9 (88% ± 3%) induced an
arousal. The number of tone series was similar in both fragmented
nights, but the number of tones per series was significantly high-
er in FN2 than in FN1 (p < .0001). In FN2, significantly fewer
C4/A1
O1/A2
EMG
EOG
Flow
Loaded
Figure 2—Example of a 30-s epoch of rapid eye movement sleep
showing a load-induced arousal. Recording of central and occipital
electroencephalogram (C4/A1, O1/A2), left and right electrooculo-
gram (EOG), submental electromyogram (EMG), and airflow moni-
tored via facemask.
527
Sleep Fragmentation and Arousability—Zavodny et al

Page 4

SLEEP, Vol. 29, No. 4, 2006
arousals were induced than during FN1 (p < .01). Sleep variables
and the total number of arousals per hour slept (arousal index)
on fragmented nights are compared with the CN mean and pre-
sented in Table 1. Compared with CN, sleep fragmentation was
successful: the arousal index was significantly increased on both
fragmented nights (FN1: p < .0001, FN2: p < .01). Sleep variables
were altered by sleep fragmentation compared with CN (Table 1).
Sleep efficiency was significantly decreased in FN1 (p < .01) but
remained unchanged in FN2. The percentage of stage 1 was sig-
nificantly higher in both fragmented nights (FN1: p < .0001, FN2:
p < .05). Wake after sleep onset increased significantly in FN1
(p < .01) but remained unchanged in FN2. Percentage of stage 2
sleep increased significantly in FN1 (p < .01) but remained un-
changed in FN2. The percentage of slow-wave sleep was signifi-
cantly lower on both fragmented nights (FN1: p < .0001, FN2: p <
.01). The percentage of REM sleep decreased significantly in FN1
(p < .01) but remained unchanged in FN2. Sleep latency remained
unchanged in FN1 but was significantly lower in FN2 (p < .01).
REM sleep latency increased significantly in FN1 (p < .01) but
remained unchanged in FN2.
Loaded Nights
Sleep Variables
Sleep variables on both loaded nights (LNF: after sleep frag-
mentation, LNC: after normal sleep) are compared with the CN
mean and presented in Table 1. Compared with CN, the percent-
age of stage 1 sleep was significantly increased on both LN (LNC:
p < .05; LNF: p < .05). During LNC, sleep efficiency was signifi-
cantly decreased (p < .01), compared with CN: the percentage of
REM sleep (p < .01) was significantly decreased, and wake after
sleep onset was significantly increased (p < .0001). REM sleep
latency increased significantly in LNC (p < .05) but remained un-
changed in LNF, compared with CN.
Number of Load Applications
In LNF significantly more (p < .01) load applications (78.4 ±
2.3) were accomplished than in LNC (70.3 ± 2.3). In LNF, 26.6
(34%) ± 2.1 load applications resulted in arousals. In LNC, 27.9
(40%) ± 1.8 load applications induced an arousal. The number of
load applications and induced arousals per sleep stage are sum-
marized in Table 2.
Number of Arousals
The total number of arousals and the number of spontaneous
arousals per hour slept on both LN and CN are presented in Table
1. Inspiratory loading caused no change in either the total number
of arousals or in the number of arousals per hour slept on both
LN. The number of spontaneous arousals per hour slept on both
Table 1—Sleep Variables and Arousals on Fragmented, Control, and Loaded Nights

Sleep stage percentage, %
1
2
3/4
REM
TST, min
Movement time, min
WASO, min
Sleep efficiency, %
Sleep latency, min
REM sleep latency, min
Arousals, no./h of sleep
Total
Spontaneous
FN1

FN2

CN

LNF LNC
18.8 ± 1.5*
61.3 ± 1.4*
6.3 ± 1.2*
13.6 ± 1.3*
383.7 ± 17.0*
5.9 ± 0.8
72.5 ± 14.9*
81.2 ± 3.3*
10.6 ± 2.4
191.2 ± 21.2*

16.9 ± 1.2*
5.9 ± 1.0
11.8 ± 0.6*
53.6 ± 1.6
13.9 ± 1.5*
20.8 ± 1.3
443.3 ± 8.6
7.3 ± 1.2
23.9 ± 5.6
92.4 ± 1.4
5.4 ± 1.3*
94.9 ± 12.3

13.2 ± 0.9*
4.6 ± 0.5*
9.2 ± 0.6
51.0 ± 1.3
20.1 ± 1.4
19.7 ± 0.9
438.9 ± 5.8
7.3 ± 0.8
18.8 ± 3.6
92.1 ± 1.0
11.3 ± 1.8
91.1 ± 8.5

10.0 ± 0.9
7.5 ± 0.8
11.1 ± 0.7*
50.0 ± 1.3
19.9 ± 1.7
19.0 ± 1.1
433.3 ± 6.9
6.7 ± 0.8
25.2 ± 4.0
91.2 ± 1.2
9.8 ± 1.9
80.2 ± 7.7
13.3 ± 1.2*
51.1 ± 1.5
17.7 ± 1.0
18.0 ± 1.1*
412.7 ± 9.6*
6.2 ± 0.6
46.7 ± 8.4*
86.6 ± 1.9*
11.1 ± 1.8
125.6 ± 18.2*
11.2 ± 0.7
7.4 ± 0.6
11.7 ± 0.8
7.4 ± 0.6
Values are presented as means ± SEM (n = 16). FN1 refers to first fragmented night; FN2, second fragmented night; CN, mean of the 2 control
nights; LNF, loaded night after FN; LNC, loaded night after CN; TST, total sleep time: non-rapid eye movement sleep stages 1, 2, 3, 4, and rapid
eye movement (REM) sleep between sleep onset and lights on; WASO, wake after sleep onset; sleep efficiency, total sleep time as a percentage of
bedtime; sleep latency, time from lights off to the first epoch of stage 2, 3, 4, or REM sleep; REM sleep latency, time from sleep onset to first epoch
of REM sleep.
*FN1, FN2, LNF, and LNC respectively vs CN (p < .05; 2-tailed paired Wilcoxon signed-rank test).
Table 2—Number of Load Applications and Induced Arousals per Sleep Stage
Sleep stage

REM
2
3/4
LNF
Arousal
11.4 ± 1.1
12.0 ± 1.3
2.9 ± 0.5
LNC
Arousal
10.5 ± 1.0
14.7 ± 1.8
2.4 ± 0.3
Loading
16.6 ± 1.5
42.2 ± 1.7
19.3 ± 1.7*
Nonarousal
5.1 ± 0.9
30.2 ± 1.8*
16.4 ± 1.7*
Loading
15.2 ± 1.3
38.9 ± 2.2
15.6 ± 0.8
Nonarousal
4.7 ± 0.9
24.2 ± 2.0
13.3 ± 0.7
Values are shown as means ± SEM (n=16). Number of load applications and induced arousals per sleep stage for both loaded nights. LNF refers to
loaded night after 2 fragmented nights; LNC, loaded night after 2 control nights; REM, rapid eye movement.
*Significant difference between LNF and LNC (p < .05; 2-tailed paired Wilcoxon signed-rank test).
528
Sleep Fragmentation and Arousability—Zavodny et al

Page 5

SLEEP, Vol. 29, No. 4, 2006
LN was the same as on CN.
The number of arousals occurring within 2 minutes of the ap-
plication of the inspiratory resistance as a percentage of the total
number of load applications are shown in Figure 3. The compari-
son between stage 2, stage 3/4, and REM sleep showed a signifi-
cant difference between these sleep stages in both loaded nights
(LNC: p < .0001; LNF: p < .0001). There was a significant higher
percentage of arousals in REM sleep compared with stage 2 sleep
(LNC: p < .0001; LNF: p < .0001) and compared with stage 3/4
sleep (LNC: p < .0001; LNF: p < .0001). There was a significantly
higher percentage of arousals in LNC compared with LNF (all-
night data: 40% to 34%; p < .05). Looking at individual sleep
stages, there was a significantly higher percentage of arousals
only in stage 2 sleep (p < .05, 1-tailed paired Wilcoxon signed-
rank test).
In order to check the results (obtained from the individual
arousal frequencies in each sleep stage) for any bias caused by
doing a disproportionate number of tests in each subject in each
sleep stage, we also looked at the pooled data from all subjects.
The analysis of the pooled data showed the same changes. There
was a significantly higher percentage of arousals in REM sleep
(LNC: 69%; LNF: 69%) compared with stage 2 sleep (LNC: 38%,
p < .0001; LNF: 28%, p < .0001) and compared with stage 3/4
sleep (LNC: 15%, p < .0001; LNF: 15%, p < .0001). There was a
significantly higher percentage of arousals in LNC, compared with
LNF (all-night data: 40% to 34%; p < .005). Looking at individual
sleep stages, there was a significantly higher percentage of arous-
als only in stage 2 sleep (p < .0003).
During CN, there was a significantly higher percentage of
arousals within control periods in REM sleep (28%) compared
with stage 2 sleep (19%; p < .05) and compared with stage 3/4
sleep (14%; p < .001). During both loaded nights, the percentage
of arousals in REM sleep (LNC: p < .0001; LNF: p < .0001) and
stage 2 sleep (LNC: p < .0003; LNF: p < .01) was significantly
higher than in CN, whereas, in stage 3/4 sleep, there was no dif-
ference between the loaded nights and CN.
Time to Arousal
During both loaded nights, the probability of arousal in stage 2
and REM sleep was highest within the first 30 seconds after the
start of loading, whereas, in stage 3/4 sleep, the arousal frequency
was equally distributed in 30-second time bins within 2 minutes
of loading (Figure 4). During CN, the arousal frequency was
equally distributed in 30-second time bins within control periods
in the NREM sleep stages 2 and 3/4 and REM sleep (Figure 4).
There was no difference in the “time to arousal” between the
loaded nights. Figure 5 shows significant differences in the “time
to arousal” between REM sleep, stage 2 sleep, and stage 3/4 sleep
for both nights (LNC: p < .0001; LNF: p < .0001). The “time to
arousal” was significantly shorter in REM sleep compared with
stage 2 sleep (LNC: p < .0003; LNF: p < .0001) and stage 3/4 sleep
(LNC: p < .0001; LNF: p < .0001), respectively.
Considering only the “time to arousal” of the arousals within
2 minutes of the application of the inspiratory load (not ranked
and “nonarousals” not included), the subjects aroused faster from
REM sleep (LNC: 36 seconds; LNF: 43 seconds) than from stage
2 sleep (LNC: 45 seconds; LNF: 49 seconds) and from stage 3/4
sleep (LNC: 55 seconds; LNF: 54 seconds) in both loaded nights.
First Third Versus Last Third of Load Applications
Figure 6 presents the percentage of induced arousals on each
loaded night in the first third and in the last third of load applica-
tions for stage 2, stage 3/4, and REM sleep. On LNF, from the first
to the last third, percentage of arousals increased significantly in
REM sleep (p < .01), while decreasing significantly in stage 2
sleep (p < .05). On LNC, comparing the first and the last third,
stage 2 showed a decrease, as on LNF, but did not reach statisti-
cal significance (p = .06), while there was no significant change
in REM sleep from the first to the last third. From the first to the
last third, stage 3/4 sleep remained unchanged on both loaded
nights. However, there was a significant difference between stage
2, stage 3/4, and REM sleep in the first (LNC: p < .0001; LNF: p
< .0001) and the last (LNC: p < .0001; LNF: p < .0001) third on
both loaded nights. There was no significant difference between
the loaded nights for the first and the last third in each sleep stage.
REMStage 2Stage 3/4
0
20
40
60
80
100
‡
*** LNC, LNF
LNC
LNF
Induced arousals [%]
Figure 3—Number of arousals within 2 minutes of load application as
a percentage of the total number of load applications in the different
sleep stages (rapid eye movement [REM], stage 2, stage 3/4) for both
loaded nights (LNC, loaded night after 2 control nights; LNF, loaded
night after 2 fragmented nights). Values are shown as means ± SEM
(n=16).
‡ Significant difference between LNC and LNF (p < .05; 1-tailed paired
Wilcoxon signed-rank test).
***Significant difference between sleep stages (p < .0001; paired
Friedman test performed for each night).
LN
Time to arousal [s]
0-3031-60 61-90 91-120
0
10
20
30
40
50
60
CN
0-3031-60 61-90 91-120
REM
Stage 2
Stage 3/4
Induced arousals [%]
Figure 4—Distribution of all induced arousals during LN (pooled
data from both loaded nights) and CN (pooled data from both control
nights). Shown is the percentage of induced arousals within 30-s time
bins in the different sleep stages (rapid eye movement [REM], 2, and
3/4).
529
Sleep Fragmentation and Arousability—Zavodny et al