Sleep, 20(12): II 97-1200
© 1997 American Sleep Disorders Association and Sleep Research Society
Sleep EEG and Computerized Analyses
Comparative Effects of Sleep on a Standard Mattress to an
Experimental Foam Surface on Sleep Architecture and
CAP Rates
Martin B. Scharf, Robin Stover, Michael McDannold, Herbert Kaye and David V. Berkowitz
Tri-State Sleep Disorders Center and Center for Research in Sleep Disorders, Inc" Cincinnati, Ohio, U.S.A.
Summary: The comparative effects of sleep patterns and rates of cyclic alternating patterns (CAP rate) in a high
quality innerspring mattress were compared to those on a unique foam support mattress in 10 normal subjects,
Results showed no differences in sleep stages, number of wakes, or total sleep time between the two conditions.
CAP rates were significantly reduced on the foam surface. CAP rate was sensitive to the first-night effect on both
surfaces, but was blunted on the foam mattress. Key Words: Cyclical alternating patterns (CAPs)-Sleep qualitySleep surface.
The role that the bed itself plays in contributing
to good sleep has been explored in the sleep laboratory (1-4). According to Sullivan, "the type of
bed in which a person sleeps can affect a night's rest
if the sleeper feels unaccustomed to the bed". He
concludes that "comfort and ease of mind have more
effect on sleep than do innersprings versus a waterbed" (2). Sleep laboratory studies have been unable
to identify differences in sleep quality or sleep process associated with bed surfaces (3,4). Even in a
study carried out evaluating sleep in a fluidized bead
bed, in which patients literally were supported by
and slept on a cloud of powdery beads, no differences were seen in sleep stages or sleep consolidation (4). Conclusions from these results were that the
bed "surface" did not significantly affect the sleep
of the subjects (4).
Recently, the microanalysis of the sleep electroencephalogram (EEG) has resulted in the identification
of a cyclic pattern of alternating states of arousal
("CAPS") that seems to correlate more strongly with
subjective estimates of sleep quality (5-9). Terzano et
al. have shown that statistically significant improvements in subjective sleep quality correlate with reAccepted for publication August 1997.
Address correspondence and reprint requests to Martin B. Scharf,
Ph.D., Tri-State Sleep Disorders Center, Cincinnati, OH 45246,
U.S.A.
duced CAP rates in insomniacs receiving a hypnotic
and that elevated CAP rates correlate with reduced
sleep quality in normal sleepers whose sleep was disturbed by noise (7,9). Similarly, we have reported a
number of clinical conditions where poor sleep was
associated with elevated CAP rates and effective treatment associated with reduced CAP rates (10-12).
Changes in CAP rate are not detected by standard
scoring methods of the sleep EEG as described by
Rechtschaffen and Kales (13) and therefore have been
missed until now. Thus, the failures to detect differences in objective correlates of sleep quality may have
simply reflected the insensitivity of previous measurements.
We recently conducted a crossover comparison of
two sleep surfaces in normal sleepers. The purpose of
the study was to compare the sleep obtained on a high
quality innerspring hospital mattress with that obtained
on an experimental foam surface utilizing polysomno graphically measured sleep patterns, with a focus on
rates of cyclic alternating pattern sequences.
METHODS
This study was organized as a single center twoway crossover design utilizing 10 normal sleepers who
had no reported history of significant daytime impairment, a subjective reported sleep latency over the pre-
1197
1198
M. B. SCHARF ET AL.
vious month of less than 30 minutes, and a typical TABLE 1. Comparative effects of sleep on a standard
subjectively reported customary total sleep time of 6 mattress to an experimental foam sUrface on sleep architecture and CAP rates
hours of more. All subjects had been previously evalSD
Parameter
Old
SD
New
uated polysomnographically and found to be free of
sleep apnea, periodic leg movements in sleep, and dif- % Sleep efficiency
5.49
92.44
4.82
92.37
3.33
5.71
4.65
2.42
ficulty falling and staying asleep. All subjects were % Stage 1
7.82
8.14
61.14
59.98
% Stage 2
healthy, were taking no medications, and provided in- % Stage 3/4
13.03
5.26
12.94
5.79
formed consent.
7.56
20.74
5.53
21.37
% REM
10.58
8.45
24.57
17.93
Subjects were randomly assigned to one of two % CAP NREM epochs
13.03
11.3
11.62
9.91
latency minutes
treatment groups. Group 1 spent three consecutive Sleep
50.23
27.07
103.8
REM latency minutes
92.98
nights on the experimental foam surface first and then
CAp, cyclic alternating pattern; SD, standard deviation; REM, rapthree consecutive nights on the innerspring mattress. id eye movement; NREM, non-rapid eye movement.
The other group slept on the mattress types in reverse
order. Patients were prepared for polysomnographic
monitoring and retired at their reported customary bed- VA yielded no significant differences between beds,
time. The two 3-night sessions took place during two nights, or bed-by-night effects. There was a clear adconsecutive weeks. Each session took place on the aptation effect with regard to rapid eye movement
same three nights of the week. Sleep architecture was (REM) sleep in both bed conditions, with REM minhand scored according to the criteria of Rechtschaffen utes being characteristically suppressed on the first
and Kales by a scorer blinded to the treatment condi- night compared to night 3. REM minutes increased
tions (13). Cyclic alternating patterns were scored ac- from 76.95 ± 29.12 on night 1 to 94.15 ± 29.16 on
cording to a modified criterion of Terzano et al. (5). night 3 with the innerspring mattress and from 76.95
Instead of scoring CAPS in all non-rapid eye move- ± 41.58 to 99.85 ± 36.40 minutes on the new surface
ment (NREM) sleep stages, only stages 1 and 2 of (p < 0.05).
This suppression was accompanied by an increase
NREM sleep were scored for CAPS. Standard central
in
REM latency in both conditions. In general, other
EEG lead rates were used for scoring the CAPS, with
than
these effects, the MANOVA for these parameters
the scorer similarly blinded as to condition. The exyielded
no significant F scores, although the bed-byperimental surface was a foam support mattress with
night
effect
approached significance at p < 0.084 for
a four-zone design, creating a sleep surface claimed to
REM
latency.
None of the other measures (e.g. total
provide low interface pressure and exceptional mussleep
time,
number
of wake times, or duration of wake
culoskeletal support for the sleeper.
yielded
significant
differences between beds or
time)
The variables for the 10 subjects were first defined
between
nights.
using a statistical data description program, which inThe most marked effects were observed in terms of
cluded the mean, standard deviation, standard error of
CAPS.
CAP rate was reduced across the three nights
the mean, variance, kurtosis, skewness, and range.
in
each
condition, with each of the three nights in the
Given that a specific variable was not significantly
bed being lower than the CAP rate on
experimental
skewed or kurtotic and that it had similar variances
the
innerspring
bed. The bed effect was highly signifover nights, that variable was subjected to a two by
icant
at
p
<
0.001,
night effect was strong at p < 0.05,
three within subject multivariate analysis of variance
and
the
bed-by-night
evaluation showed no effect, in
(MANOVA) for two sleeping surfaces and three nights
was
a
decrease
from night 1 to night 3 in
that
there
on each surface. The surfaces were labeled "old" and
both
conditions.
The
comparison
between the effects
"new", and the nights on each surface were labeled,
on
CAPS
for
the
first
night
approached
statistical sigsequentially, night 1, night 2, and night 3. Old refers
nificance
at
p
<
0.067,
suggesting
that
the
disruption
to the innerspring mattress and new to the foam matassociated
with
the
first
night
is
blunted
on
the expertress. Post hoc analyses were carried out using a t staComparisons
of
CAP
rate
on
night
3 for
imental
bed.
tistic for matched samples, with a Bonferonni correcthe
two
bed
conditions
were
significantly
different,
tion for related data.
with improved sleep consolidation evident in the experimental bed.
RESULTS
All 10 subjects completed the crossover study.
There were two males and eight females, mean age
32.6 ± 4.9. The parameters of sleep recorded from the
innerspring hospital mattress compared to the experimental mattress are detailed in Table 1. The MANOSleep. Vol. 20. No. 12. 1997
DISCUSSION
This study is the first to evaluate the comparative
effects of different sleep environments using the CAP
rate as a measurement. The results are striking in their
1199
COMPARATIVE EFFECTS OF SLEEP
consistency and extend our previous findings with this
sleep micro architectural parameter (10-12). We previously found that nonapneic snorers had higher CAP
rates in sleep than did non snorers and that the CAP
rate in snorers decreases when wearing an external nasal dilator (Breathe Right, CNS, Minneapolis, MN)
during sleep (10). Similarly, we found the CAP rate to
be elevated in peri- and postmenopausal women who
were experiencing hot flashes and decreased with effective pharmacotherapy (11). We have also seen reductions in CAP rates in elderly insomniacs receiving
temazepam, and Terzano and Parrino reported reductions in CAP rates in younger insomniacs receiving
zolpidem (7).
Our modification of the CAP scoring system described by Terzano and Parrino, as indicated previously, involved scoring CAP only in stages 1 and 2 sleep.
Stage 1 represented approximately 5% of the total
sleep time, so that the overwhelming portion of the
data reflects stage 2 CAP rates. Discussions with Terzano and Parrino confirmed our impression that there
does not appear to be a sleep stage-specific difference
in CAP rate and that scoring CAPS in these two sleep
stages provides a representative CAP rate for total
NREM sleep.
The current study also extended our appreciation of
the changes in sleep that occur during the adaptation
process and showed that CAP rates may be an additional factor that is sensitive to the "first-night effect"
(14,15). However, our results not only detected a difference in the degree of sleep disruption between the
experimental mattress and the spring mattress in terms
of CAP rate, but clearly showed an attenuation of the
adaptation effect with the experimental surface.
The first-night effect is often used as a model for
transient insomnia (16). The attenuated effect on CAP
rate with the experimental mattress on the adaptation
night suggests that this may be an important parameter
that could aid in the evaluation of hypnotic drugs and
other treatments for transient insomnia. Further, the
differences in CAP rates were maintained on the third
night. Long-term studies with subjective data are needed to evaluate the potential perseveration of this effect.
While thousands of studies have been conducted detailing sleep architecture in healthy subjects and the
sleep of patients with over 80 different sleep-related
disorders, a data-based characterization of "good"
sleep has never been established. While patients speak
of "good sleep quality", "sound sleep", "deep sleep",
or "refreshing sleep", much of our understanding
about sleep is based on comparisons of noncomplaining individuals to those with specific sleep complaints
or pathologies. As a result, in a pattern strikingly similar to other areas of medicine, we seem to know more
about sleep illness than sleep wellness.
Intuitively, one can say that a good night's sleep is
one that leaves the sleeper refreshed and energized
upon awakening. However, evidence is mounting that
duration is not the only factor that contributes to refreshed mornings-that the degree of sleep consolidation is also critical. Studies of patients with sleep
apnea and periodic leg movements in sleep suggest
that frequent nighttime arousals and awakenings are
related to the daytime complaints of fatigue and sleepiness (17,18). While duration and consolidation clearly
seem to contribute to sleep quality, other factors such
as room temperature and bed surface have been considered as individualized issues (1,2).
Does the improvement in CAP rates, however, truly
reflect improvement in sleep quality? One of the
unique strengths of polysomnographic data is its consistency, reliability, and reproducibility as well as the
general standardization of recording techniques. As a
result, sleep laboratory studies generally do not require
the larger numbers of subjects needed for subjective
clinical trials. However, while the n's in these studies
may be large enough to detect CAP rate changes, they
are not large enough to adequately evaluate subjective
assessments. While much needs to be done to verify
the relationship between CAP rates and sleep quality,
the potential of this parameter in identifying sleep
quality is exciting.
REFERENCES
1. Webb WB. Cultural aspects of sleep. In: Carskadon M, ed. Encyclopedia of sleep and dreaming. New York: Macmillan, 1993:
155-7.
2. Sullivan J. Beds. In: Carskadon M, ed. Encyclopedia of sleep
and dreaming. New York: Macmillan, 1993:66-8.
3. Rosekind M, Phillips R, Rappaport J, Babcock D, Dement We.
Effects of waterbed surface on sleep: a pilot study. Sleep Res
1976;5:32.
4. Bixler E, Scharf M, Leo L, Kales A. Hypnotic drugs and performance: a review of theoretical and methodological considerations. In: Kagan F, Harwood T, Rickels K, Rudzik A, Sorer
H, eds. Hypnotics: methods of development and evaluation.
New York: Spectrum Publications, 1975: 175-98.
5. Terzano MG, Parrino L, Spaggiari MC. The cyclic alternating
pattern sequences in the dynamic organization of sleep. Electroencephalogr Clin Neurophysiol 1988;69:437-47.
6. Terzano MG, Parrino L, Fioriti G, Farolfi A, Orofiamma B, Sauvanet Jp. Cyclic alternating pattern. A new approach to the pharmacology of sleep disorders. Neurophysiol Clin 1988;18:44757.
7. Terzano MG, Parrino L. Evaluation of EEG cyclic alternating
pattern during sleep in insomniacs and controls under placebo
and acute treatment with zolpidem. Sleep 1992;15:64-70.
8. Terzano MG, Parrino L. Clinical applications of cyclic alternating pattern. Physiol Behav 1993;54:807-13.
9. Terzano MG, Parrino L, Fioriti G, Orofiamma B, Depoortere H.
Modifications of sleep structure induced by increasing levels of
acoustic perturbation in normal subjects. Electroencephalogr
Clin Neurophysiol 1990;76:29-38.
10. Scharf MD, McDannold MD, Zaretsky NT, Hux BA, Brannen
DE, Berkowitz DY. Cyclic alternating pattern sequence in nonSleep, Vol. 20, No. 12. 1997
1200
M. B. SCHARF ET AL.
apneic snorers with and without nasal dilation. ENT 1996;75:
752-4.
11. Scharf MB, McDannold MD, Stover R, Zaretsky N, Berkowitz
DY. Effects of estrogen replacement therapy on rates of cyclic
alternating patterns and hot flash events during sleep in postmenopausal women. Clin Therap 1997; 19:304-11.
12. Reyna R, McDannold MD, Brannen DE, Berkowitz DV, Scharf
MB. CAP rates in normal and non-apneic snorers. In: Chase M,
ed. Sleep Research 1994.
13. Rechtschaffen A, Kales A, eds. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Los Angeles, CA: UCLA Brain Information ServicelBrain Research Institute, 1968.
Sleep, Vol. 20, No. 12, 1997
14. Agnew HW Jr, Webb WB, Williams RL. The first night effect:
an EEG study of sleep. Psychophycology 1966;2:263-6.
15. Scharf MB, Bixler EO, Kales A. Readaptation to the sleep laboratory in insomnia subjects. Psychophysiology 1975;12:69-73.
16. Roth T, Vogel G, Sterling W. Effects of temazepam on transient
insomnia. Sleep Res 1987;16:123.
17. Krieger 1. Obstructive sleep apnea: clinical manifestations and
pathophysiology. In: Thorpy M, ed. Handbook of sleep disor-.
ders. New York: Marcel Dekker, 1990:259-84.
18. Coccagna G. Restless legs syndrome/periodic leg movements in
sleep. In: Thorpy M, ed. Handbook of sleep disorders. New
York: Marcel Dekker, 1990:457-78.