Docstoc

Age-Related Increase in Awakenin

Document Sample
Age-Related Increase in Awakenin Powered By Docstoc
					 SLEEP AND AGING

Age-Related Increase in Awakenings: Impaired Consolidation of NonREM Sleep
at All Circadian Phases
Derk-Jan Dijk PhD, Jeanne F. Duffy MBA, PhD, Charles A. Czeisler PhD, MD

Division of Sleep Medicine, Department of Medicine, Harvard Medical School, Brigham and Women’s Hospital, Boston, MA



  Study Objectives: (1) To assess the circadian and sleep-dependent reg-             Measurements and Results: Circadian and sleep-dependent regulation
  ulation of the frequency and duration of awakenings in young and older             of the frequency and duration of awakenings and of sleep structure pre-
  people; (2) to determine whether age-related deterioration of sleep con-           ceding awakenings were assessed in 482 sleep episodes (9h 20 min
  solidation is related to an increase in the frequency or duration of awak-         each). The circadian modulation of wakefulness within sleep episodes
  enings; (3) to determine whether pre-awakening sleep structure is prefer-          was primarily related to a variation in the duration of awakenings. In con-
  entially enriched by REM sleep or nonREM sleep and (4) to determine                trast, the age-related reduction of sleep consolidation was primarily relat-
  whether sleep structure prior to awakenings is affected by age.                    ed to an increase in the frequency of awakenings. Whereas in both young
  Design: Between age-group comparison of sleep consolidation and sleep              and older subjects pre-awakening sleep contained more REM sleep than
  structure preceding awakenings.                                                    overall sleep, this enrichment of REM sleep (i.e., the gating of wakeful-
  Setting: Environmental Scheduling Facility, General Clinical Research              ness by REM sleep) was diminished in older people. In older people, pre-
  Center.                                                                            awakening sleep contained more nonREM sleep and stage two sleep in
  Participants: Eleven healthy young men (21—30 years) and 13 older                  particular, than in young people.
  healthy men (n=9) and women (n=4) (64—74 years).                                   Conclusions: At all circadian phases, the age-related reduction of sleep
  Interventions: Forced desynchrony between the sleep-wake cycle and cir-            consolidation is primarily related to a reduction in the consolidation of
  cadian rhythms by scheduling of the rest-activity cycle to 28-h for 21—25          nonREM sleep.
  cycles.


INTRODUCTION                                                                         od while they were living in a time-free laboratory environment
                                                                                     for approximately one month. Under these conditions the intrin-
AGING IS ASSOCIATED WITH MAJOR CHANGES IN                                            sic period of the human circadian pacemaker, as indexed by the
SLEEP STRUCTURE, SLEEP QUALITY AND SLEEP TIM-                                        plasma melatonin and body temperature rhythm, was on average
ING. Decreases in stage 4 or EEG slow-wave activity are among                        24.18 h in both young and older adults.13 These data and data
the most pronounced and first age-related changes in sleep struc-                    obtained in adolescents living under similar conditions14 chal-
ture to occur. Earlier timing of sleep and wakefulness and                           lenge the notion that a systematic age-related reduction of the
reduced nocturnal sleep consolidation and according to some                          intrinsic period of the human circadian pacemaker is a major
studies, a reduction in REM sleep, occur later in the aging pro-                     determinant of age-related changes in sleep timing and sleep con-
cess. Sleep disorders such as sleep apnea and nocturnal                              solidation.
myoclonus and other health problems can contribute to these                             Analysis of the subjective perception of the duration of final
changes in sleep structure and quality in older people. These                        awakening in this forced desynchrony protocol revealed that,
changes in sleep quality and sleep timing may lead to sleep com-                     even in the absence of knowledge about either clock time or cir-
plaints and the associated use of hypnotics in older people.1-5                      cadian phase, older subjects report earlier awakening than young
Two neurobiological processes, circadian rhythmicity and sleep                       subjects, especially when the scheduled wake time occurs at or
homeostasis, play a key role in the regulation of sleep timing and                   closely after the nadir of the temperature cycle.15 Analyses of
sleep structure.6-12 It has been hypothesized that changes in these                  polysomnographic recordings of these sleep episodes have
two processes contribute to the age-related changes in sleep tim-                    demonstrated that in both young and older people the propensity
ing and sleep consolidation which occur in healthy people with-                      for wakefulness depends on a non-additive interaction of the time
out sleep complaints or sleep disorders. We recently quantified                      elapsed since start of sleep episode (i.e., the sleep homeostatic
the contribution of the circadian process in a forced desynchrony                    process) and circadian phase.16 Wakefulness in older people is at
protocol in which the sleep-wake cycles of healthy older men and                     higher levels throughout the sleep episode at all circadian phases
women and healthy young men were scheduled to a 28-hour peri-                        and this increase in wakefulness as sleep progresses occurs at a
                                                                                     more rapid rate than in young people. Furthermore, the interac-
                                                                                     tion between the circadian and homeostatic regulation of sleep is
Accepted for publication March 2001                                                  changed such that there is an age-related phase advance of wake
Address correspondence to: Derk-Jan Dijk, PhD, Centre for Chronobiology,             propensity relative to the endogenous circadian component of the
School of Biomedical and Life Sciences, University of Surrey, Guildford, GU2         body temperature and plasma melatonin rhythms. These analyses
7XH, United Kingdom; Tel: + 44 1483 259341; Fax: +44 870 1371590;                    as well as other experiments,17-20 suggest that an age-related
E-mail: d.j.dijk@surrey.ac.uk
SLEEP, Vol. 24, No. 5, 2001                                                    565                                 Age-Related Increase in Awakenings—Dijk et al
                  1

                  5


                 10                                      τ=24:16                                         τ=24:11

                 15


                 20
           Day




                 25


                 30
                                                                 1144                                                1458
                                                               (23 y.o.)                                           (67 y.o.)

                      0   6     12     18    0     6      12    18 0 0       6         12     18    0     6      12    18      0
                                                                  Time of Day

Figure 1—Double raster plot of scheduled sleep episodes (open bars) and awakenings (black bars within open bars) within these scheduled sleep episodes in a 23
y.o male (subject 1144) and a 67 y.o male (subject 1458) during the forced desynchrony protocol. In subject 1458, PSG data were not available for sleep episode 10.
The progression of the fitted maximum of the plasma melatonin rhythm and the fitted nadir of the core body temperature rhythm are indicated by the dashed and solid
line respectively. Tau is the average of the intrinsic period estimated for the melatonin and core body temperature rhythm. Tau =24:16 for 1144 and 24:11 for 1458.

change in the sleep-dependent (i.e., homeostatic) component of                       to awaken is modulated by the ultradian alternation of nonREM
sleep consolidation in interaction with a reduced circadian drive                    and REM sleep, such that most spontaneous awakenings occur
for sleep in the early morning hours underlies the increase in                       from REM sleep,23-27 although not all earlier studies had reported
sleep disruption and early morning awakening in older people.16                      this.28,29 On the basis of this phenomenon and the observation
However, the neurophysiological processes underlying age-relat-                      that the circadian modulation of REM sleep crests in the morning
ed changes in the sleep-dependent and the circadian contribution                     hours, it may be predicted that the age-related increase in “early
to sleep consolidation have not been fully identified                                morning” awakening is primarily associated with an increase in
   In our analyses of the age-related changes in the interaction of                  the number of awakenings from REM sleep. On the other hand,
the circadian and homeostatic regulation of sleep, the deteriora-                    the pronounced age-related reduction of electroencephalographic
tion of sleep consolidation was quantified by computing the per-                     slow waves,30-32 and sleep spindles,33,34 which can be considered
centage of time awake within scheduled sleep episodes without                        inhibitory phenomena protecting sleep from arousal, may lead to
distinguishing between the frequency and duration of awaken-                         the prediction that deterioration of nonREM sleep consolidation
ings. Early analyses of nocturnal sleep consolidation in older                       is a primary mediator of the age-related deterioration of sleep
people sleeping at their habitual bedtimes21;22 have suggested that                  efficiency.
aging impacts primarily the frequency of spontaneous awaken-                              To further analyze the processes underlying the marked age-
ings although the duration of spontaneous awakenings may also                        related deterioration of sleep consolidation, we conducted an
be affected. This implies that aging affects the consolidation of                    analysis of the frequency and duration of awakenings as well as
sleep rather than the ability to return to sleep, although after                     an analysis of the association between sleep structure and the
experimentally induced awakenings the return to sleep may be                         probability to awaken from sleep for sleep episodes that were
prolonged in older people.21 How aging affects the process of                        scheduled to occur at virtually all phases of the endogenous cir-
awakening and the duration of awakenings when sleep occurs at                        cadian cycle from our previously published data.16 Specific
other than habitual times (i.e., at other circadian phases) is cur-                  objectives of this analysis were: (1) to assess the circadian and
rently not known.                                                                    sleep-dependent regulation of the frequency and duration of
   In our previous analyses of the sleep-dependent modulation of                     awakenings in young and older people; (2) to determine whether
the propensity to awaken, time since start of sleep episode was                      age-related deterioration of sleep consolidation is related to an
the only independent variable and the contribution of sleep stage                    increase in the frequency or duration of awakenings; (3) to deter-
and/or the nonREM-REM cycle, (i.e., ultradian modulation of the                      mine whether pre-awakening sleep structure is preferentially
propensity to awaken) were not taken into consideration. Thus,                       enriched by REM sleep or nonREM sleep; and (4) to determine
currently it is not known whether the age-related deterioration of                   whether sleep structure prior to awakenings is affected by age. A
sleep consolidation is related to nonREM sleep consolidation,                        preliminary report of these analyses has been published in
REM sleep consolidation or sleep consolidation independent of                        abstract form.35
sleep stage. Previous studies have indicated that the probability

SLEEP, Vol. 24, No. 5, 2001                                                    566                                 Age-Related Increase in Awakenings—Dijk et al
Table 1—Sleep structure in young (n=11) and older (n=13) people during forced desynchrony

                                                               Young Average                             Young SE            Older Average                    Older SE         P
TRT (min)                                                      560.0                                     0.5                 559.2                            0.7              0.3917
TST (min)                                                      466.9                                     10.7                371.2                            10.9             0.0001
Sleep Efficiency (%)                                           83.4                                      1.9                 66.4                             1.9              0.0001
Sleep Latency (min)                                            5.5                                       0.7                 6.1                              0.7              0.5356
REM Latency (min)                                              56.0                                      4.5                 38.3                             4.4              0.0107
Stage One (%)                                                  8.9                                       1.0                 13.3                             1.1              0.0094
Stage Two (%)                                                  51.6                                      1.2                 49.3                             1.7              0.2921
Stage Three (%)                                                5.9                                       0.7                 10.5                             1.4              0.0107
Stage Four (%)                                                 12.4                                      1.5                 4.6                              1.6              0.0016
REM Sleep (%)                                                  21.2                                      1.0                 22.2                             0.8              0.4269

TRT: total recording time (time between lights off and lights on). TST: Total sleep time. P: significance level from unpaired t-tests.



                                                                                         All Awakenings
                                                                                             sgninekawA llA                  Awakenings from Sleep > 2 min
                                                                                                                              nim 2 > peelS morf sgninekawA

                                                                               2.5
                                     Frequency (per hour)




                                                                               5.2                            A                                                BB
                                                                               1 1              Older         A
                         ycneuqerF
                         )ruoh rep(




                                                                               0.5
                                                                               5.0             redlO
                                                                               0.2
                                                                              2.0
                                                                               0.1
                                                                              1.0
                                                                              0.05 Young
                                                                              50.0   gnuoY
                                                                              0.02
                                                                              20.0
                                                                              0.01
                                                                              10.0

                                                                              100
                                                                              001
                                                                                                                                                               D
                                                     Relative Frequency (%)
                            ycneuqerF evitaleR




                                                                                                              C                                                D
                                                                               50
                                                                              05
                                                                                                              C
                                                                               25
                                                                              52
                                  )%(




                                                                               10
                                                                               01
                                                                               55
                                                                                2    2
                                                                                1    1
                                                                                 0.5 1
                                                                                1 5.0     2 4     8 16 32 64 >64
                                                                                         46 >46 23 61 8 4 2                  0.5 1
                                                                                                                             1 5.0    22 44    8 16 32 64 >64
                                                                                                                                               61 8   46 > 46 23
                                                                                            sgninekawA fo noitaruD                   sgninekawA fo noitaruD
                                                                                Duration of Awakenings (minutes)
                                                                                                  )setunim(                 Duration of Awakenings (minutes)
                                                                                                                                           )setunim(



Figure 2—Frequency and duration of awakenings in older (solid symbols) and young (open symbols) subjects. The frequency of all awakenings (A) and awakenings
from consolidated sleep (B) is plotted against their duration. Frequency is expressed as number of awakenings per hour of recording time. The relative frequency of
all awakenings (C) and awakenings from consolidated sleep (D) is plotted against their duration. Data are plotted at the upper limits of the bins of duration. Vertical
bars represent 1 SEM.



METHODS                                                                                                                    usual time cues. In this experiment, the circadian rhythms of core
                                                                                                                           body temperature and plasma melatonin during the forced desyn-
Subjects and Protocol                                                                                                      chrony segment of the protocol exhibited an average period of
    Details of the subject recruitment procedures and the approx-                                                          24.18 h in both the young and older group.13 Scheduled sleep
imately month long laboratory protocol have been described else-                                                           episodes (9h 20min each) thereby occurred at many circadian
where.16 All subjects gave written informed consent form prior to                                                          phases (see Figure 1).
admission to the laboratory. The protocol, consent, and adver-
tisements were approved by the Human Research Committee at                                                                 Sleep Recording and Analyses
Brigham and Women’s Hospital. Subjects were young adult (11                                                                   All 482 sleep episodes included in the present analysis were
men, age range: 21-30, mean 23.7 years) and older (age range:64-                                                           recorded polysomnographically and scored in 30-second epochs
74, mean 67.4 years; men: n=9, women: n=4) healthy volunteers                                                              according to established criteria, by a RPSGT (ER) or investiga-
without sleep complaints. They participated in a forced desyn-                                                             tor (DJD). Scorers were not blind to the age of the subject but, by
chrony protocol in which, after a baseline segment, the rest-activ-                                                        the nature of the experiment, were “blind” to circadian phase.
ity and associated dim light-dark cycle was scheduled to a 28-                                                             Each 30-second epoch was assigned a circadian phase value,
hour period while the subjects lived in an environment free of

SLEEP, Vol. 24, No. 5, 2001                                                                                          567                               Age-Related Increase in Awakenings—Dijk et al
derived from non-orthogonal spectral analysis of the core body               definition of awakenings, in some analyses the effects of age, cir-
temperature data,13,36 and a time-elapsed-since start-of-sleep-              cadian phase and time since start of sleep episode were assessed
episode value. This allowed analysis of both sleep-dependent and             on the basis of all awakenings as well as after exclusion of awak-
circadian modulation of awakenings. As in our previous publica-              enings of one minute or shorter.
tions on sleep during forced desynchrony, sleep episodes follow-                All means and standard errors represent the averages across
ing the sleep deprivation associated with constant routines                  subjects and the between subject standard error (i.e., n represents
(which in general were scheduled after three baseline sleep                  the number of subjects rather than the number of sleep episodes
episodes, and at the end of the forced deynchrony protocol) were             or the number of awakenings).
not included in the analyses.
    To illustrate the interaction between sleep-dependent and cir-           Statistics
cadian regulation of awakenings, sleep episodes were analyzed
per third of scheduled episode (Figures 4, 7, and 8). This resolu-           Frequency of Awakening
tion was sufficient to illustrate some of the sleep-dependent
                                                                                 The distributions of awakenings were analyzed in two-steps.
aspects while at the same time allowing most subjects to con-
                                                                             Awakenings were first sorted according to their duration and
tribute to each average at every circadian bin.
                                                                             assigned to bins, 0.5, 1, 2, 4, 8, 16,32, 64 and >64 minutes. The
     Analyses were carried out with software written in Turbo
                                                                             bin width was selected on the basis of an exploratory analysis that
Pascal for Windows and the SAS statistical package was used
                                                                             revealed a linear relationship between the log of frequency and
for all statistical tests.
                                                                             bin width when bin width was defined as described above. For
                                                                             each subject, the frequency of awakenings in these bins (with the
Analyses of Awakenings                                                       bin >64 excluded) was log transformed. The distributions for
    The frequency and duration of awakenings were first assessed             each subject were then quantified by determining the slope and
for each subject separately. The duration of awakenings was                  intercept in a regression procedure. Inspection of the individual
determined by the number of consecutive 30-second epochs of                  regression coefficients indicated that the transformation was suc-
wakefulness. Awakenings were defined in two ways: (1 Any 30-                 cessful (young subjects range of r2: 0.43 to 0.88; older subjects
second epoch of wakefulness following any epoch of sleep                     range of r2: 0.72 to 0.99). Age effects on the distributions were
(including Stage 1); and (2) Any 30-second epoch of wakefulness              then analyzed by nonparametric ANOVA on the intercepts and
following more than two minutes of consolidated sleep. These                 slopes of these individual distributions.
latter awakenings are referred to as “awakenings from consoli-
                                                                             Duration and Sleep Structure Preceding Awakenings and
dated sleep.” This analysis was carried out because it may be
argued that an analysis of all awakenings without a minimum                  Circadian and Sleep-Dependent Modulation:
sleep duration preceding an awakening is confounded by high                     Repeated measures ANOVAs were conducted in all cases in
frequency alterations between sleep and wakefulness (i.e., the               which all subjects contributed to each level. In cases in which this
transition from sleep to wakefulness and the transition from                 was not possible because of missing values, we have indicated
wakefulness to sleep).                                                       this. Ratio measures were log transformed prior to statistical
    The distribution of the duration of awakenings was quantified            tests. When repeated measures ANOVA were used, the Huynh-
by sorting the awakenings in bins. The upper limits of these bins            Feldt correction for sphericity was applied but the original
were 0.5, 1, 2, 4, 8, 16, 32, and 64 minutes. Awakenings longer              degrees of freedom are reported.
than 64 minutes were assigned to the >64 minutes bin. The fre-
quency of awakenings was expressed as number of awakenings                   RESULTS
per hour of recording time between sleep onset and lights-on.
Relative frequencies of awakenings were computed by express-                 Sleep Structure in Young and Older People During Forced
ing the frequency of awakenings in a bin as a percentage of the              Desynchrony
total number of awakenings.
    To assess the association between sleep structure and the prob-             Overall sleep structure is summarized in Table 1. Total sleep
ability to awaken, we computed sleep structure during time inter-            time and sleep efficiency, as well as stage 4 sleep, were marked-
vals immediately preceding an awakening. This approach                       ly reduced in the older subjects. Stage 1 sleep was significantly
appears to better reflect the neurophysiology of vigilance state             enhanced and stage 2 sleep and REM sleep were not significant-
control than the strict statistical transition probability approach.         ly different between the age groups, although the latency to REM
For instance, consider two three-minute sequences of 30-second               sleep was significantly shorter in older people. A detailed
epochs of sleep stages: (1) REM, REM, REM, REM, Stage 1, W;                  description of the circadian and sleep-dependent modulation of
and (2) stage 2, stage 2, stage2, stage 2, Stage 1, W. Considering           sleep structure in young in older people in this study has been
only the last 30-second epoch preceding W would lead to a clas-              reported elsewhere.16
sification of both awakenings as awakenings from stage 1, even
though it seems more reasonable to classify the first awakening              Distribution of the Duration of Awakenings
as an awakening from stage REM and the second awakening as                       In both young and older subjects, the shortest awakenings
an awakening from stage 2. Because we consider awakening to                  (0.5min) were most frequent (Figure 2). The frequency of awak-
be a gradual process rather than an abrupt event, we quantified              enings declined progressively as the duration became longer,
the time course of sleep structure leading up to an awakening. To            even though the bin width of awakenings increased exponential-
investigate the robustness of our findings against the arbitrary

SLEEP, Vol. 24, No. 5, 2001                                            568                             Age-Related Increase in Awakenings—Dijk et al
                                                                                  Circadian
                                                                                 naidacriC                                         Sleep Dependent
                                                                                                                                tnednepeD-peelS




                             )emit gnidrocer fo ruoh rep(
                                  (per hour recording time)
                                  Frequency o Awakenings
                              sgninekawA fo ycneuqerF
                                                               4
                                                               4        Older
                                                                        redlO                           A              Older                               B
                                                                                                    A                   redlO                          C
                                                               3
                                                               3                                                                                 2ls


                                                               2
                                                               2                                                       Young
                                                                                                                        gnuoY
                                                               1
                                                               1
                                                                       Young
                                                                         gnuoY
                                                               0
                                                               0
                           sgninekawA fo noitaruD
                         Duration of Awakenings




                                                                                                    B   C                                              D   D
                                                              15
                                                              51
                                 )setunim(
                                (minutes)




                                                              10
                                                              01

                                                              55

                                                              00
                                                                   0
                                                                   0   90 180 270 0
                                                                       0 072 081 09     90 180 270
                                                                                        072 081 09                 0
                                                                                                                   0      90
                                                                                                                          09     180 270 360 450 540
                                                                                                                                 072 081    045 054 063

                                                                        Circadian Phase (Degrees)
                                                                            esahP naidacriC                            Time in Sleep Episode (minutes)
                                                                                                                               edosipE peelS ni emiT
                                                                               )seerged(                                             )setunim(



Figure 3—Circadian and sleep-dependent modulation of the frequency and duration of awakenings from consolidated sleep in older (solid symbols) and young sub-
jects (open symbols). The circadian modulation of the frequency (A) and duration (C) of awakenings from consolidated sleep is plotted with respect to circadian phase
as derived from the core body temperature rhythm, with a resolution of 45 degrees, (i.e., 3 circadian hours. Circadian phase 0 degrees corresponds to the fitted nadir
of the temperature rhythm) which under entrained conditions occurs at approximately 5:15 h in older people and 6:00 h in young people (15). Data are double-plot-
ted at the midpoints of the circadian bin. The sleep-dependent modulation of the frequency (B) and duration (D) of awakenings from consolidated sleep is plotted with
respect to time in sleep episode, with a resolution of 112 minutes (i.e., fifths of the sleep episode). Data are plotted at the midpoints of the bins. Vertical bars repre-
sent 1 SEM.



ly. This pattern was similar regardless of whether awakenings                                                     Circadian and Sleep-Dependent Modulation of the Frequency
were defined as “all awakenings” or as “awakenings from con-                                                      and Duration of Awakenings
solidated sleep.” The frequency of awakenings was higher in
older people for all durations of awakenings. This was reflected                                                     The frequency of awakenings, which exhibited a modest but
in a significant effect of the factor Age (as assessed by Wilcoxon                                                significant circadian modulation (Circadian Phase: F7,154=8.19;
test, p< 0.002) on the intercepts derived from the regression anal-                                               p=0.0001), was much higher in older people than in young peo-
yses on the transformed distributions of all awakenings and                                                       ple, at all circadian phases (Fig 3a) (Age: F1,22=18.68; p=0.0003).
awakenings from consolidated sleep.                                                                               The age-related increase in the frequency of awakening was most
    The slopes of the regression were not different between the                                                   pronounced between 270 and 45 degrees (i.e., at the circadian
age groups for either all awakenings (p=0.7943) or awakenings                                                     phase interval which under entrained conditions corresponds to
from consolidated sleep (p=0.9769). In accordance with the                                                        approximately 24:00 h to 9:00 hours in older people (Age*
results from this analysis, the relative frequency distributions of                                               Circadian Phase: F7,154=6.74; p=0.0001).
awakenings were quite similar for the young and older subjects.                                                      Analyses of the circadian modulation of the duration of awak-
This was confirmed by the results of a repeated measures                                                          ening revealed a prominent circadian modulation (Fig.
ANOVA on the relative frequency of awakenings which revealed                                                      3C)(Circadian Phase: F7,154=12.78; p=0.0001). In both young and
a significant effect of bin (F8,176=260.13; p<0.0001) and no sig-                                                 older people longest durations were observed at 225 degrees (i.e.,
nificant effect of Age (F1,22=0.20, p=0.662). The interaction                                                     nine hours before the temperature minimum) and shortest dura-
between these two factors (F8,176=2.56;p=0.1096) was also not                                                     tions at 0 degrees (i.e., at the minimum of the endogenous circa-
significant even though the relative frequency of the shortest                                                    dian component of the core body temperature rhythm). The dura-
awakenings appeared slightly higher in young people and the rel-                                                  tion of awakenings was near identical in young and older subjects
ative frequency of awakenings of medium duration was some-                                                        at all circadian phases (Age: F1,22=0.02; p=0.8924;
what higher in older people The intra-subject standard deviation                                                  Age*Circadian Phase: F7,154=0.73; p=0.5249). Similar results
of the frequency of awakenings from consolidated sleep ranged                                                     were obtained after computing the geometric mean of the dura-
from 0.40 to 1.37 awakenings per hour in the young subjects and                                                   tions (Circadian Phase: F7,154=13.36; p=0.0001; (Age: F1,22=3.35;
0.49 to 1.28 in the older subjects.                                                                               p=0.0809; Age*Circadian Phase: F7,154=0.80; p=0.5428).
                                                                                                                     The frequency of awakenings remained near constant in the
                                                                                                                  older people and exhibited a small increase over the course of
                                                                                                                  sleep in the young subjects (Fig. 3B) (Time: F4,88=4.56;

SLEEP, Vol. 24, No. 5, 2001                                                                                 569                                 Age-Related Increase in Awakenings—Dijk et al
                                                                                                    Frequency
                                                                                                   ycneuqerF                                                                                        Duration
                                                                                                                                                                                                      noitaruD

                                                                                                                III
                                                                                                                III                                                                                                     III
                                                                                                                                                                                                                          III




                          )emit gnidrocer fo ruoh rep( sgninekawA fo ycneuqerF
                                                                                                                                                                                    30
                                                                                                                                                                                    03
                                                                                 4
                                                                                 4



                             Frequency o Awakenings (per hour recording time)
                                                                                                                                                                                    20
                                                                                                                                                                                    02
                                                                                 2




                                                                                                                          )setunim( sgninekawA fo noitaruD
                                                                                 2
                                                                                                                                                                                    10
                                                                                                                                                                                    01




                                                                                                                                                 Duration of Awakenings (minutes)
                                                                                 0
                                                                                 0                                                                                                   0
                                                                                                                                                                                     0

                                                                                          Older                 II                                                                  30
                                                                                                                                                                                    03                               II
                                                                                 4         redlO                    II                                                                                                  II
                                                                                 4
                                                                                                                                                                                    20
                                                                                                                                                                                    02
                                                                                 2
                                                                                 2
                                                                                                                                                                                    10
                                                                                                                                                                                    01
                                                                                          Young
                                                                                           gnuoY
                                                                                 0
                                                                                 0                                                                                                   0
                                                                                                                                                                                     0


                                                                                                                I     I                                                             30
                                                                                                                                                                                    03                              I
                                                                                 4
                                                                                 4                                                                                                                                      I

                                                                                                                                                                                    20
                                                                                                                                                                                    02
                                                                                 2
                                                                                 2
                                                                                                                                                                                    10
                                                                                                                                                                                    01

                                                                                 0
                                                                                 0                                                                                                   0
                                                                                                                                                                                     0
                                                                                     0 90 180 270 0 90 180 270 0
                                                                                     0   0 072 081 09     0 072 081 09                                                                   0 90 180 270 0 90 180 270 0
                                                                                                                                                                                         0   0 072 081 09      0 072 081 09
                                                                                          Circadian Phase (degrees)
                                                                                         )seerged( esahP naidacriC                                                                            Circadian Phase (degrees)
                                                                                                                                                                                             )seerged( esahP naidacriC



Figure 4—Interaction of the circadian and sleep-dependent modulation of the frequency (left panels) and duration (right panels) of awakenings from consolidated
sleep in older (solid symbols) and young (open symbols) subjects. The circadian modulation of the duration and frequency of awakenings was computed separately
for the first (I) second (II) and last (III) third of sleep episodes. Data are plotted as in Fig. 3.



p=0.0046). The frequency of awakenings was higher in the older                                                                                                                      24 subjects for whom data were available in all 24 cells of the
subjects, in particular in the beginning of the sleep episode (Age:                                                                                                                 3*8 matrix) yielded a significant effect of Age (F1,22=19.36;
F1,22=17.56; p=0.0004; Age* Time: F4,88=21.09; p=0.0001).                                                                                                                           p<0.0002), a significant effect of Time (F2,44= 5.75; p=0.0060), a
Analyses of the sleep-dependent duration of awakenings demon-                                                                                                                       significant effect of Circadian Phase (F7,154=5.79;p=0.0001), a
strated that duration increased as sleep progressed in young and                                                                                                                    significant interaction between Age and Time (F2,44=19.72;
older subjects (Fig 3D) (F4,88=16.10; p<0.0001) and that the                                                                                                                        p<0.0001), a significant interaction between Age and Circadian
sleep-dependent increase was not different between the age                                                                                                                          Phase (F7,154=5.79; p=0.0001) as well as a significant three way
groups (Age*Time: F 4,88=0.45; p=0.7230). After computing the                                                                                                                       interaction (F14,308=2.74; p=0.0016).
geometric mean of the durations similar results were obtained                                                                                                                            The amplitude of the circadian modulation of the duration of
(Time: F4,88=15.68; p<0.0001; Age*Time: F4,88=1.11; p=0.3521),                                                                                                                      awakenings increased as sleep progressed in both young and
although the effects of Age now approached significance (Age:                                                                                                                       older people, and longest duration were observed at 270 degrees
F1,22=4.02;p=0.0575).                                                                                                                                                               (Fig. 4 right panel). In the first and second third of the sleep
    The main effects of circadian phase and time since start of                                                                                                                     episode the durations of awakenings and the waveform of circa-
sleep episode on the frequency and duration were qualitatively                                                                                                                      dian modulation were near identical in the two age groups. In the
similar when analyzed for “all awakenings” (data not shown) and                                                                                                                     final third of the sleep episode, the durations of awakenings were
“awakenings from consolidated sleep.”                                                                                                                                               somewhat longer in the older subjects, in particular in the inter-
     The interaction between the sleep-dependent and circadian                                                                                                                      val of 0 to 135 degrees. Whereas in the young subjects the circa-
modulation of the frequency and duration of awakenings is illus-                                                                                                                    dian wave-form remained unimodal in this final third of the sleep
trated in the left panel of Figure 4. In the first third of the sleep                                                                                                               episode, the wave-form of the duration of awakenings in the
episode, neither the frequency nor the duration of awakenings                                                                                                                       older people exhibited a bimodal pattern, with local minima at
exhibited a pronounced circadian modulation. In the second and                                                                                                                      180 and 315 degrees. A repeated measures ANOVA (on 19 sub-
final third of the sleep episode, the circadian modulation of the                                                                                                                   jects for whom data were available in all 24 cells of the 3*8
frequency of awakenings became more pronounced, in particular                                                                                                                       matrix) yielded no significant effect of Age (F1,17=2.10;
in the older subjects. The frequency of awakenings was higher in                                                                                                                    p=0.1657), a significant effect of Time (F2,34=19.47; p=0.0001), a
the older subjects in all thirds of the sleep episode, and in the sec-                                                                                                              significant effect of Circadian Phase (F7,119=4.95;p=0.0026), a
ond and final third of the sleep episode in particular in the inter-                                                                                                                significant interaction between Age and Time (F2,34=4.84;
val of 270 to 45 degrees. Thus, a repeated measures ANOVA (on                                                                                                                       p=0.0141); no significant interaction between Age and Circadian

SLEEP, Vol. 24, No. 5, 2001                                                                                                                                  570                                                 Age-Related Increase in Awakenings—Dijk et al
                                                                                                   sleep,” with the exception of the bimodal pattern of the duration
                                                                                                   of awakenings in the older people in the final third of the sleep
            )%( enO egatS                                                                          episode which was not observed when ‘all awakenings’ were
                                               20
       A                      Stage One (%)    02                                                  considered.
      A                                        10
                                                                                                   Sleep Structure Prior to Awakenings from Consolidated
                                               01

                                               00                                                  Sleep
                                               60
                                               06
            )%( owT egatS




                                                                                                       The dynamics of sleep structure in the 15 minutes prior to
                              Stage Two (%)




                                               50
                                               05                                                  awakenings from consolidated sleep was computed for young
      B
       B                                       40
                                               04                                                  and older subjects and compared to overall sleep structure
                                                                                                   (Figure 5). In this figure, all data are expressed as percentage of
                                               30
                                               03
                                                                                                   TST, and wakefulness in the 15 minutes preceding an awakening
                                               15
                                               51                                                  is not represented in this figure. Stage 1 sleep in the interval
            )%( eerhT egatS
                              Stage Three (5




                                                                                                   between 15 and 2 minutes prior to awakening was close to the
                                               10
                                               01                                                  average for overall sleep structure in the older subjects, and was
       C                                        5
                                                5
                                                                                                   slightly higher than its overall average in the young subjects. No
      C
                                                                                                   significant difference in the amount of stage 1 sleep preceding an
                                                0
                                                0                                                  awakening between the age groups was observed (Effect of Age:
                                                                                                   F1,22=015; p=0.7036). Stage 1 sleep did exhibit a consistent
            )%( ruoF egatS




                                               10                                                  change in the course of the pre-awakening interval (Effect of
                              State Four (%)




                                               01
                                                                                                   Time: F29,638=2.44; p=0.0449). In the last two minutes prior to
       D                                        5
      D
                                                5                                                  awakening, stage 1 sleep was higher in the young than in the
                                                                                                   older people, and this was reflected in the significant interaction
                                                0
                                                0
                                                                                                   between the factor Age and Time (Time * Age: F29,638= 8.40;
           )%( peels MERnoN




                                               80
                                               08                                                  p<0.0001). Stage 2 sleep declined over the 15 minute interval
                              NonREM (%)




                                                                                                   prior to awakenings (Time: F29,638=30.25; p<0.0001) and more so
                                               60
                                               06                                 Older
                                                                                  redlO            in the young subjects (Age:F1,22=4.75; p=0.0403) (Time* Age:
      EE                                                                                           F29,638=17.59; p<0.0001) and was below the overall percentage of
                                               40
                                               04
                                                                                     Young
                                                                                     gnuoY
                                                                                                   stage 2 sleep, especially in the young subjects. Stage 3 sleep dur-
                                               60
                                               06                                                  ing the pre-awakening interval was higher in older subjects (Age
                              REM sleep (%)
           )%( peels MER




                                                                                                   F1,22=14.60; p=0.0009), declined over the 15 minutes prior to
                                               40                                                  awakening in both young and older subjects (Time: F29,638=9.61;
                                               04
      F                                                                                            p<0.0001), (Time*Age: F29,638=1.84; p=0.1532) and in the pre-
       F                                                                                           awakening interval was below the overall percentage of this sleep
                                               20
                                               02
                                                                                                   stage. Stage 4 exhibited a modest decline prior to awakening
                                                    -14 -12 -10 -8 -6 -4 -2
                                                    4- 6- 8- 01- 21- 41-   2-    0
                                                                                 0                 (Time: F29,638=4.58; p=0.0159) with no evidence for a different
                                                             Time (minutes)
                                                      )setunim( emiT                               time course in the two age groups (Time * Age: F29,638=1.19;
                                                                                Awakening
                                                                           gninekawA               p=0.3131). In the young subjects the percentage of stage 4 prior
                                                                                                   to awakening was much lower than the overall values of this
Figure 5—Pre-awakening sleep structure. Stage 1, 2, 3, 4, nonREM sleep, and                        sleep stage. In the older subjects the percentage of stage 4prior to
REM sleep in the 15 minutes prior to awakenings from consolidated sleep in                         awakenings was only moderately lower than the overall levels of
older (solid symbols) and young (open symbols) subjects. The average percent-                      stage 4, which was markedly reduced compared to the values
age (of TST) of the sleep stages computed across all sleep episodes is indicat-
                                                                                                   observed in the young subjects such that in the pre-awakening
ed by the horizontal solid lines for the older and dashed lines for the young sub-
jects. The between-subject SEM for these average percentages are indicated by                      interval no significant effect of age was observed (F1,22=0.97;
the vertical bars at Time = -0.5 minutes.                                                          p=0.3364).
                                                                                                       The percentage of nonREM sleep (stages 2, 3, and 4 com-
                                                                                                   bined) gradually declined as the awakening approached for both
Phase (F7,119=1.75; p=0.1589); and no significant three way                                        the young and older subjects (Time F29,638= 51.18; p<0.0001).
interaction (F14,238=1.21; p=0.3022). When the geometric means                                     This decline was steeper in the young than in the older subjects
of the durations were computed a significant effect of Age                                         (Time * Age: F29,638=14.80; p<0.0001)). In the last minutes prior
(F1,17=4.87; p=0.0413); a significant effect of Time (F2,34=18.61;                                 to awakening the percentage of nonREM sleep was lower than
p=0.0001); and a signficant effect of Circadian Phase                                              the overall levels in both the young and older subjects and this
(F7,119=5.72; p=0.0004) were observed. The interaction between                                     difference was larger in the young subjects.
Age and Time was significant (F2,34=3.94; p=0.0290); The other                                          The percentage of REM sleep gradually increased as the
interactions were not significant.                                                                 awakening approached (Time: F29,638=38.97; p<0.0001). This
   The interaction of the circadian and sleep-dependent modula-                                    increase was steeper in the young than in the older subjects (Time
tion was qualitatively similar when analyzed for “all awaken-                                      * Age: F29,638=8.34; p=0.0002). In the last minutes prior to awak-
ings” (data not shown) and “awakenings from consolidated                                           ening the percentage of REM sleep was approximately 45% and

SLEEP, Vol. 24, No. 5, 2001                                                                  571                             Age-Related Increase in Awakenings—Dijk et al
                                                                                        REM sleep prior to awakenings
                                                                                        sgninekawa ot roirp peels MER
                                                                                                                    Young
                                                                                                                    gnuoY




                               )TST fo %( peels MER
                                          REM sleep (% of TST)
                                                                     50
                                                                     05




                                                                     25
                                                                     52
                                                                                                            Older
                                                                                                            redlO



                                                                                          nonREM sleep prior to awakenings
                                                                                       sgninekawa ot roirp peels MERnon
                               )TST fo %( peels MERnon




                                                                     75
                                           nonREM sleep (% of TST)




                                                                     57
                                                                                                            Older
                                                                                                            redlO


                                                                     50
                                                                     05


                                                                                         Young
                                                                                          gnuoY

                                                                     25
                                                                     52
                                                                          0
                                                                          0   45 90 135 180 225 270 315 0 45 90 135 180 225 270 315
                                                                              54   513 072 522 081 531 09    0   54   513 072 522 081 531 09
                                                                                             Circadian Phase (Degrees)
                                                                                            )seergeD( esahP naidacriC

Figure 6—Circadian modulation of the time course of nonREM and REM sleep during the 15 minutes preceding awakenings that were longer than 1 minute and
occurred from consolidated sleep in older (solid symbols) and young (open symbols) subjects. Awakenings were binned according to circadian phase at which they
occurred with a resolution of 45 degrees. Data are double plotted at the midpoints of the circadian bins. Each curve (consisting of three points) represents nonREM
or REM sleep averaged for 0-5, 5-10 or 10-15 min prior to an awakening. SE represent the between subject SE. The data indicate that the percentage of nonREM
sleep decreases and the percentage of REM sleep increases as the awakening approaches.



35 % for the young and older subjects respectively. In both age                                             after exclusion of short awakenings (i.e., awakenings of a dura-
groups these values were higher than the overall percentages of                                             tion up to one minute). This analysis thereby investigates whether
REM sleep which were 21% and 22% for the young and older                                                    the age-related differences in pre-awakening sleep structure, as
subjects respectively. Pre-awakening sleep structure was thus                                               observed in Fig. 5, could be solely due to the marked increase in
more enriched with REM sleep in young subjects than in older                                                the frequency of short duration awakenings in older subjects. The
subjects (Age: F1,22=7.54; p=0.0118).                                                                       overall effect of Age on sleep structure prior to awakenings
                                                                                                            longer than one minute was significant for both nonREM
Circadian Modulation of Sleep Structure Prior to Awakenings                                                 (F1,22=10.87; p<00033) and REM sleep (F1,22=7.10; p<0.0142).
from Consolidated Sleep                                                                                     In both young and older subjects, nonREM (Figure 6 lower
                                                                                                            panel) and REM sleep (Fig. 6 upper panel) prior to awakenings
     The effect of circadian phase on the time course of nonREM                                             longer than 1 minute varied with circadian phase (Circadian
and REM sleep in the 15 minutes prior to awakenings was ana-                                                Phase NonREM sleep F7,154=3.390; p<0.0010; REM sleep:
lyzed. The overall effect of Age on sleep structure prior to awak-                                          F7,154=3.21; p<0.0087). At all circadian phases nonREM sleep
enings was significant for both nonREM (F1,22=6.36; p<0.019)                                                decreased (Time: F2,44=58.46; p<0.0001) and REM sleep
and REM sleep (F1,22=8.50; p<0.008). NonREM and REM sleep                                                   increased (Time: F2,44=40.27; p<0.0001) prior to awakenings in
prior to awakenings varied with circadian phase (Circadian                                                  both age groups (Fig. 6 ). This time course was also observed
Phase: NonREM sleep F7,154=8.30; p<0.001; REM sleep:                                                        when analyzed separately per third of sleep episode. The decline
F7,154=11.12; p<0.0001) and with Time (nonREM: F2,44=73.81;                                                 of nonREM sleep was steeper in young subjects (Time*Age:
p<0.0001. REM : F2,44=60.96; p<0.0001). Significant interac-                                                F2,44=17.10; p<0.0001) and the increase of REM sleep was shal-
tions between Time and Age were observed for nonREM sleep                                                   lower in older subjects (Time*Age: F2,44=6.79; p=0.0069).
(Time*Age: F2,44=19.10; p<0.0001) and REM sleep (F2,44=9.31;                                                Consequently, the largest differences between young and older
p<0.0019).                                                                                                  people were observed close to the awakenings (i.e., in the last
   To investigate whether these effects were robust against the                                             two five-minute intervals preceding the awakening). The per-
definitions of awakenings, an identical analysis was conducted                                              centage of nonREM sleep in the 15 minutes prior to awakenings

SLEEP, Vol. 24, No. 5, 2001                                                                           572                             Age-Related Increase in Awakenings—Dijk et al
                                                                                                                                    Older
                                                                                                                                   redlO                    gnuoYYoung


                                                                                                                         60
                                                                                                                         06                                                   60
                                                                                                                                                                              06

                                                                                                                         40
                                                                                                                         04                                                   40
                                                                                                                                                                              04
                                                                        III
                                                                         III
                                                                                                                         20
                                                                                                                         02                                                   20
                                                                                                                                                                              02
                                                                                                                                     2sowtdlo                 2sowtdlo
                                                                                                                          0
                                                                                                                          0
                                                                                                                                                                              0
                                                                                                                                                                              0


                                                                                 )TST fo %( owT egatS
                            edosipE peelS fo drihT

                                                                                                                                                                 Total


                                                                                                  Stage Two (% 0f TST)
                                               Third of Sleep Episode

                                                                                                                                                                 latoT
                                                                                                                         60
                                                                                                                         06                                                   60
                                                                                                                                                                              06


                                                                        II                                               40
                                                                                                                         04                                                   40
                                                                                                                                                                              04
                                                                         II
                                                                                                                         20
                                                                                                                         02                                                   20
                                                                                                                                                                              02
                                                                                                                                                      Pre-awakening
                                                                                                                                                      gninekawa-erP
                                                                                                                          0
                                                                                                                          0                                                   0
                                                                                                                                                                              0

                                                                                                                         60
                                                                                                                         06                                                   60
                                                                                                                                                                              06

                                                                                                                         40
                                                                                                                         04                                                   40
                                                                                                                                                                              04
                                                                         I   I
                                                                                                                         20
                                                                                                                         02                                                   20
                                                                                                                                                                              02

                                                                                                                          0
                                                                                                                          0                                                   0
                                                                                                                                                                              0
                                                                                                                              0 120 240 0 120 240 0 0 120 240 0 120 240 0
                                                                                                                              0 042 021 0 042 021 0   0 042 021 0 042 021 0
                                                                                                                                        Circadian Phase (Degrees)
                                                                                                                                     )seergeD( esahP naidacriC

Figure 7—Interaction of the circadian and sleep-dependent modulation of stage 2 sleep during the 10 minute interval prior to awakenings longer than 1 minute from
consolidated sleep, compared to overall sleep structure. Data are separated per third of the sleep episode. Left hand panels: Older subjects; open symbols stage 2
independent of awakening, i.e., total sleep; solid symbols stage 2 in the 10-min interval preceding awakenings. Right hand panels: Young subjects; open symbols
stage 2 independent of awakening; solid symbols stage 2 in the 10-min interval preceding awakenings. Vertical bars represent 1 SEM.


was higher in older than in young subjects at all circadian phas-                                                                                       prior to awakenings longer than one minute was compared to the
es, but this age-related difference in the time course prior to                                                                                         circadian and sleep-dependent modulation of these sleep stages
awakening was modulated by circadian phase such that the dif-                                                                                           independent of awakenings in the young and older subjects
ference appeared most pronounced in the interval of -45 to 45                                                                                           (Figures 7 and 8).
degrees.                                                                                                                                                     The percentage of stage 2 increased as sleep progressed in
                                                                                                                                                        both young and older subjects at all circadian phases (Fig. 7).
Interaction of Circadian and Sleep-Dependent Modulation of                                                                                              Stage 2 in the 10 minutes prior to awakenings also increased as
Sleep Structure Prior to Awakenings                                                                                                                     sleep progressed. Stage 2 in the 10 minutes prior to an awaken-
                                                                                                                                                        ing was lower than the overall percentage of stage 2 at nearly all
   The analysis presented above shows that the largest age-relat-                                                                                       circadian phases. These lower percentages of stage 2 in the pre-
ed differences are present in the last 10 minutes prior to awaken-                                                                                      awakening interval were observed in both the young and older
ings and that differences in pre-awakening sleep structure                                                                                              subjects. However, the reduction of stage 2 appeared larger for
(nonREM and REM sleep) are observed at all circadian phases.                                                                                            the young subjects. This was quantified by computing the ratio of
These age-related differences are observed when all awakenings                                                                                          the percentage of stage 2 during the pre-awakening interval and
from consolidated sleep or those awakenings that are longer than                                                                                        the overall percentage of stage 2. If sleep structure prior to awak-
one minute are considered. Next we analyzed the interaction of                                                                                          ening were a random sample of overall sleep structure, (i.e., if
the circadian and sleep-dependent modulation of sleep structure                                                                                         awakenings occur independent of sleep stage) the ratio would be
prior to awakenings. This analysis was conducted during the 10-                                                                                         on average one. In fact, the ratio was less than one in both young
minute interval preceding awakenings from consolidated sleep.                                                                                           and older subjects in all but one of the 24 combinations of circa-
To further investigate the age-related differences in pre-awaken-                                                                                       dian phase and third of sleep episode. However, in older subjects
ing sleep structure the interaction between circadian and sleep-                                                                                        the ratio was closer to one at virtually all circadian phases and
dependent modulation was analyzed for REM, nonREM, and                                                                                                  thirds of the sleep episode. The statistical significance of the
Stage 2 sleep. These analyses were conducted for all awakenings                                                                                         main effect of age in a 3-way ANOVA with factors age, Circadian
and for awakenings longer than one minute. The interaction of                                                                                           phase and third of sleep episode confirmed this observation
circadian and sleep-dependent modulation of stage 2 sleep,                                                                                              (F1,430=24.75; p<0.0001). In younger subjects this ratio was par-
nonREM sleep and REM sleep during the 10-minute interval                                                                                                ticularly low when the second third of the sleep episode occurred

SLEEP, Vol. 24, No. 5, 2001                                                                                                                       573                              Age-Related Increase in Awakenings—Dijk et al
                                                                                                                                       Older
                                                                                                                                      redlO                       gnuoYYoung
                                                                                                                             80
                                                                                                                             08                                                    80
                                                                                                                                                                                   08

                                                                                                                             60
                                                                                                                             06                                                    60
                                                                                                                                                                                   06

                                                                                III
                                                                                III                                          40
                                                                                                                             04                                                    40
                                                                                                                                                                                   04

                                                                                                                             20
                                                                                                                             02                                                    20
                                                                                                                                                                                   02
                                                                                                                                        2sowtdlo                    2sowtdlo
                                                                                                                              0                                                    0
                                                                                                                                                            Pre-awakening


                                                                                    )TST fo %( peelS MER
                                                                                                                             0                                                     0
                              edosipE peelS fo drihT




                                                                                                      REM Sleep (% of TST)
                                                                                                                                                                gninekawa-erP
                                                                                                                             80
                                                       Third of Sleep Episode
                                                                                                                             08                                                    80
                                                                                                                                                                                   08

                                                                                                                             60
                                                                                                                             06                                                    60
                                                                                                                                                                                   06

                                                                                                                             40
                                                                                                                             04                                                    40
                                                                                                                                                                                   04
                                                                                II
                                                                                II
                                                                                                                             20
                                                                                                                             02                                                    20
                                                                                                                                                                                   02
                                                                                                                                                                    Total
                                                                                                                                                                     latoT
                                                                                                                             00                                                    0
                                                                                                                                                                                   0

                                                                                                                             80
                                                                                                                             08                                                    80
                                                                                                                                                                                   08

                                                                                                                             60
                                                                                                                             06                                                    60
                                                                                                                                                                                   06


                                                                                I
                                                                                I                                            40
                                                                                                                             04                                                    40
                                                                                                                                                                                   04

                                                                                                                             20
                                                                                                                             02                                                    20
                                                                                                                                                                                   02

                                                                                                                             00                                                    0
                                                                                                                                                                                   0
                                                                                                                                  0 120 240 0 120 240 0 0 120 240 0 120 240 0
                                                                                                                                  0 042 021 0 042 021 0   0 042 021 0 042 021 0
                                                                                                                                            Circadian Phase (Degrees)
                                                                                                                                           )seergeD( esahP naidacriC

Figure 8—Interaction of the circadian and sleep-dependent modulation of REM sleep during the 10 minute interval prior to awakenings longer than 1 minute from
consolidated sleep, compared to overall sleep structure. Data are separated per third of the sleep episode. Left hand panels: Older subjects; open symbols REM sleep
independent of awakening, i.e, total sleep; solid symbols REM sleep in the 10-min interval preceding awakenings. Right hand panels: Young subjects; open symbols
REM sleep independent of awakening; solid symbols REM sleep in the 10-min interval preceding awakenings. Vertical bars represent 1 SEM.


close to the temperature nadir.                                                                                                                                 awakening from sleep in the morning hours is facilitated by the
   When computed for nonREM sleep, (i.e., stages 2, 3 and 4                                                                                                     sleep-dependent dissipation of sleep pressure, the reduced circa-
combined) a very similar pattern emerged (except that a sleep-                                                                                                  dian drive for sleep, the circadian modulation of REM sleep
dependent increase in the percentage of nonREM sleep was not                                                                                                    (which exhibits its circadian crest in the morning hours shortly
observed) and the effect of Age was significant (F1,441=40.03;                                                                                                  after the core body temperature nadir), the sleep-dependent dis-
p<0.0001).                                                                                                                                                      inhibition of REM sleep, and the ultradian gating of the proba-
   REM sleep exhibited a circadian variation with highest values                                                                                                bility of awakening such that most awakenings occur from REM
close to the temperature nadir in both young and older subjects.                                                                                                sleep. The present analyses demonstrate that the marked deterio-
REM sleep increased as sleep progressed, especially in young                                                                                                    ration of sleep consolidation in older people is present at all cir-
subjects (Fig. 8). In the 10-minute interval preceding awaken-                                                                                                  cadian phases, is primarily related to an increase in the frequen-
ings, REM sleep was higher than the overall percentage of REM                                                                                                   cy of awakenings, although a small increase in their duration can-
sleep in all thirds of the sleep episode and at virtually all circadi-                                                                                          not be excluded, and is not related to a selective increase in the
an phases, in both young and older subjects. This enrichment of                                                                                                 probability to awaken from REM sleep. On the contrary, many
the pre-awakening interval with REM sleep was more pro-                                                                                                         more awakenings occurred from nonREM sleep in the older sub-
nounced in the young subjects (F1,449=65.11; p<0.0001).                                                                                                         jects. This suggests that age-related changes in nonREM sleep
   Similar patterns for nonREM, Stage Two and REM sleep were                                                                                                    consolidation are of primary importance in the age-related dete-
observed when all awakenings from consolidated sleep were                                                                                                       rioration of sleep consolidation.
included in the analysis and the effects of Age were significant
(p< 0.0001in) all cases.                                                                                                                                        Frequency and Duration of Awakenings
                                                                                                                                                                   The circadian modulation of wakefulness within scheduled
DISCUSSION
                                                                                                                                                                sleep episodes, which crests in the wake maintenance zone (i.e.,
   The alternation between the three vigilance states, (i.e., wake-                                                                                             approximately eight hours before the temperature nadir)37
fulness, nonREM sleep, and REM sleep) is regulated by the ultra-                                                                                                appears to be mediated primarily by a modulation of the duration
dian nonREM-REM sleep cycle, the circadian pacemaker and                                                                                                        of awakenings rather than the frequency of awakenings in both
NonREM and REM sleep homeostasis. Sleep consolidation and                                                                                                       young and older subjects. Longest duration awakenings were
the transition between sleep and wakefulness is achieved by a                                                                                                   observed at approximately 240 degrees, especially when the last
fine-tuned interaction of these processes. Thus, in young people                                                                                                third of the sleep episode occurred at this phase. This result is

SLEEP, Vol. 24, No. 5, 2001                                                                                                                               574                             Age-Related Increase in Awakenings—Dijk et al
reminiscent of the data obtained in squirrel monkeys10 and sug-             considered. The enrichment of pre-awakening sleep by REM
gests that the circadian pacemaker opposes the return to sleep at           sleep occurred at all circadian phases and not only at the crest of
this circadian phase which under entrained conditions corre-                the circadian REM sleep rhythm. This suggests that the increased
sponds to approximately 22:00 h in young adults sleeping                    probability to awaken from REM sleep is related to REM sleep
between 24:00 and 08:00. Because the frequency of awakenings                itself rather than to the circadian phase of highest REM sleep
exhibited only minor circadian modulation and no clear increase             propensity. The association between REM sleep and awakening
in the frequency of awakenings was observed in the morning                  in both young and older subjects is in accordance with previous
hours, it appears that there is no strong circadian “wake-up” sig-          observations and the proposed role of REM sleep as a gate to
nal in the morning hours. This is in accordance with analyses               wakefulness.23-27 However, the pre-awakening enrichment of
based on wakefulness expressed as percentage of total recording             sleep by REM sleep was less pronounced in the older subjects.
time.16 Spontaneous awakenings in the morning hours thus seem               Even the last five minutes of sleep preceding an awakening con-
to be mediated by the dissipation of homeostatic sleep pressure in          tained less REM sleep in the older subjects. This also holds for
conjunction with a reduction of the active promotion of sleep by            the last two minutes preceding an awakening.35 This latter obser-
the circadian pacemaker, which occurs after the nadir of the core           vation indicates that the attenuation of REM sleep during the
body temperature rhythm in young adults.                                    pre-awakening interval is unlikely to be related to an increase in
     The marked age-related increase in wakefulness within                  the frequency of awakenings from very short (e.g., two-minute,
scheduled sleep episodes was primarily related to an increase in            “episodes” of REM sleep in the older subjects).
the frequency of awakenings rather than their duration. The pre-
sent analyses demonstrate that the age-related increase in the fre-         Age-Related Reduction of the Protective Effect of NonREM
quency of awakenings is present at all circadian phases and                 Sleep on Arousal
throughout the sleep episode. This suggests that a change in the
sleep process contributes to the age-related deterioration of sleep            Not surprisingly, stage 4 sleep was markedly attenuated in the
consolidation, and that changes in this sleep process in conjunc-           interval preceding awakenings in the young subjects. This con-
tion with age-related changes in its interaction with the circadian         firms the notion the stage 4 sleep is “the deepest” stage of sleep
process (see below), lead to the age-related increase in the sus-           and that awakening is unlikely to occur from this stage. In the
ceptibility to circadian phase misalignment. The observation that           older subjects, overall percentage of stage 4 sleep was markedly
the frequency of awakenings is much higher in older people in               reduced, and even though stage 4 declined prior to awakening,
particular in the beginning of sleep episodes maybe surprising              the apparent protective effect of Stage 4 on the probability to
because older people complain primarily of early morning awak-              awaken was reduced in the older subjects. This may be related to
enings. However, we don’t know how the complaint of early                   the reduction of EEG slow-wave activity within stage 4 in older
morning awakening relates to the frequency and duration of                  people. In this view, it is the amplitude of slow-waves, which
awakenings. Furthermore, the frequency and duration of awak-                markedly declines with aging2 which reflects the inhibitory
enings in a given time interval are not independent. For instance,          impact of nonREM sleep. We therefore did not proceed to an
if the average duration of awakenings in an hourly interval is 30           adjustment of the amplitude criterion for scoring stage 3 and 4 in
minutes, it is impossible to observe a frequency of awakening               older people, as has been suggested and implemented in previous
greater than 2. Thus the increase in the duration of awakenings,            analyses of sleep in older people.38 In our analyses of sleep in
and the increase in the total duration of wakefulness (i.e., the            older people, stage 4 is reduced, but overall SWS (i.e., stage 3 +
product of frequency and duration) as sleep progresses (rather              stage 4) was not significantly reduced. This is related to the
than the frequency of awakenings) may be related to the subjec-             increase in the percentage of stage 3. Although this finding may
tive perception of awakenings. We have reported previously that             be somewhat surprising in view of the notion that SWS declines
awakening from sleep exhibits an internal circadian phase                   with age, a reduction of stage 3 has not been reported in all stud-
advance in older people (i.e., older people wake up at an earlier           ies on aging and the reduction in stage 4 is the most consistently
circadian phase of the body temperature and melatonin cycle).16             reported observation.2 The current analyses demonstrate that the
In other words, the age-related reduction in sleep consolidation            probability to awaken from SWS was increased in the older sub-
varies with circadian phase such that the largest difference                jects as compared to the young subjects at nearly all circadian
between young and older subjects is observed in the early morn-             phases. Whether this is related to the reduction of slow-wave
ing hours. In the current analyses this internal circadian phase            activity within SWS remains to be assessed.
advance is reflected primarily in the frequency of awakenings.                   Age-related changes in the probability to awaken from
Thus the difference between young and older subjects in the fre-            nonREM sleep were not limited to SWS but also included stage
quency of awakening was most pronounced around the nadir of                 2 sleep. Whereas in the young subjects the percentage of stage 2
the circadian body temperature rhythm (see Fig 3 and 4).                    sleep in the interval preceding awakenings was far below the
                                                                            overall percentage of stage 2 sleep, in the older subjects the per-
REM Sleep as a Gate to Wakefulness in Young and Older                       centage of stage 2 was almost similar to the overall level. This
People                                                                      indicates that stage 2 sleep is less protective of arousal from sleep
                                                                            in older people. A recent publication of awakening in older peo-
    The analyses of sleep structure preceding awakenings demon-             ple sleeping at habitual times reported a similar finding39
strated that the percentage of REM sleep increased prior to awak-           although in that study no direct comparison between older and
enings in both young and older subjects. This increase in REM               young subjects was available and circadian phase was not
sleep prior to awakenings was observed irrespective of whether              assessed. The conclusion from that study was that “an inability to
all awakenings or only awakenings longer than one minute were               sustain stage 2 may be a mechanism which contributes to the dif-

SLEEP, Vol. 24, No. 5, 2001                                           575                              Age-Related Increase in Awakenings—Dijk et al
ficulty of sleep maintenance in the elderly.” This conclusion is                 The marked age-related changes in the nonREM sleep process
very similar to our interpretation of the current data, which                 and the persistence of these changes at all circadian phases sug-
included sleep occurring at many circadian phases. Murphy and                 gest that circadian processes, or at least the outputs of the pace-
colleagues40 analyzed spontaneous awakenings in young and                     maker that are rhythmic, do not contribute markedly to these
older subjects living under conditions in which subjects were iso-            changes. It should be noted though that the age-related difference
lated from time cues. Although their data confirm the data pre-               in awakenings was modulated by circadian phase. Thus, more
sented here, they emphasized age-related changes in the REM                   awakenings from nonREM sleep were observed in older people
sleep process. They concluded that “age-associated changes in                 scheduled to sleep near the core body temperature nadir (Fig 6).
sleep continuity may render unnecessary the putative role of                  How age-related changes in this circadian modulation might
REM sleep in providing a ‘gate’ to wakefulness.” In our view, the             relate to other circadian processes such as body temperature and
age-related changes in sleep consolidation are primarily related              plasma melatonin remains unclear.
to a diminution of the protective role of nonREM sleep, includ-
ing stage 2 sleep.                                                            Implications for Models of Sleep Regulation
     Although in most discussions on aging and stage 2 sleep it is
assumed that the EEG in stage 2 sleep is similar in young and                    The concept of the regulation of sleep propensity by an inter-
older people, quantitative EEG analyses have demonstrated that                action of a circadian process and a sleep-dependent homeostatic
this is not the case. Both slow-wave activity as well as sleep spin-          process,6,9 has been inspiring and successful. However, new data
dle activity, both overall and within stage 2, are markedly                   sets and new analyses imply that this concept needs refinement
reduced in middle-aged and older people.32-34,41 The age-related              and modification. We previously demonstrated that a simple
reduction in sleep-spindle activity has been observed at all circa-           addition of the homeostatic and circadian processes cannot
dian phases and the amplitude of the circadian modulation of                  explain the observed changes in sleep propensity during desyn-
spindle activity is also attenuated in older people.34 Interestingly,         chrony of the sleep-dependent and circadian process.11 The pre-
the largest difference in pre-awakening stage 2 sleep between the             sent analyses add another complexity (i.e., the modulation of the
young and older people was observed when the last two-thirds of               propensity to awaken by sleep stage). In a recent version of the
the sleep episode occurred close to the temperature nadir. It is at           two-process model45 the decline of the homeostatic regulatory
this circadian phase that we previously observed the largest age-             variable S is proportional to slow-wave activity (SWA, EEG
related reduction in the magnitude of the circadian promotion of              power density 0.75-4.5 Hz) (i.e., S declines rapidly when SWA is
sleep spindle activity.34 Sleep of older people is more susceptible           high, such as during stage 4 sleep). This implies that during stage
to disruption by auditory stimuli, and it may be argued that this is          4 sleep the awakening threshold is approached rapidly. In this
related to diminished slow-wave activity and/or sleep spindle                 recent version of the two-process model, S is assumed to increase
activity within nonREM sleep, although no direct evidence for                 during REM sleep. This assumption was necessary to obtain an
this is available. Such increased susceptibility to internal and              adequate fit between computer simulation of the time course of
external stimuli may also explain the greater disruption of sleep             SWA and actual data obtained during extended sleep. Although,
when sleep is scheduled to occur at other than normal circadian               in view of the similarities of the EEG during REM sleep and
phase (i.e., during circadian phase misalignment). The current                wakefulness as well as the similarities in brain metabolism dur-
data show that in older people, circadian phase misalignment and              ing these two states this assumption is not unreasonable, it
associated reduction of sleep consolidation is primarily related to           implies that according to the model, the propensity to awaken
a reduced consolidation of nonREM sleep.                                      decreases as REM sleep progresses. The present and previous
   A screening for sleep apnea and nocturnal myoclonus while                  data indicate that the propensity to awaken is highest during
the older subjects were sleeping at their habitual bedtimes yield-            REM sleep and in particular when REM density is high.27 This
ed an average sleep apnea index of 2.8/h, stdev=2.4; and a peri-              appears contrary to the model predictions. Current models of
odic limb movements with EEG arousal index of =4.8, stdev=                    sleep regulation need to incorporate changes in the propensity to
2.9. However, we cannot exclude the possibility that subclinical              awaken with sleep stage and EEG content to adequately describe
levels of these sleep disorders contributed in part to the present            the alternations between sleep and wakefulness which are partic-
results. Furthermore, we cannot exclude the possibility that the              ularly frequent in older people.
absence of women in our sample of young subjects had an effect                CONCLUSION
on the reported differences between the young and older subjects.
This is because gender differences in sleep structure, and in par-               Sleep of older people is interrupted frequently by awakenings,
ticular differrences in SWS or slow-wave activity, have been                  and this deterioration of sleep continuity is present at all circadi-
reported for young and older subjects,42-44 and this topic, and in            an phases and appears to be related primarily to a reduction of the
particular the association between changes in nonREM sleep and                consolidation of nonREM sleep. It remains to be established
sleep consolidation, deserves further attention.                              whether age-related reductions in slow wave and/or sleep spindle
   The present findings were robust against different definitions             activity are associated with this age-related attenuation of the
of awakenings. Thus the same patterns emerged when the analy-                 nonREM sleep related inhibition of arousal from sleep.
ses were performed by including all awakenings, awakenings
from consolidated sleep only, or only awakenings longer than one              ACKNOWLEDGMENTS
or two minutes (data not shown). This implies that the observa-                  Supported by P01 AG 09975 to CAC and General Clinical
tions are not dependent on the precise definition of awakenings               Research Grant M01 RR02635. We thank Mr. Eymard Riel for
or on subtle differences in interpretation of EEG recordings.                 his polysomnographic assistance, the subject recruiters, the tech-

SLEEP, Vol. 24, No. 5, 2001                                             576                              Age-Related Increase in Awakenings—Dijk et al
nical staff of the Circadian, Neuroendocrine and Sleep Disorders                    22. Bixler EO, Kales A, Jacoby JA, Soldatos CR, Vela-Bueno A.
Section and the staff of the General Clinical Research Center for                   Nocturnal sleep and wakefulness: effects of age and sex in normal sleep-
protocol execution. We thank Dr. Shelley Hurwitz, (GCRC-                            ers. Int J Neurosci 1984;23:33-42.
statistician) for advice on statistical procedures. The first author                23. Weitzman ED, Czeisler CA, Zimmerman JC, Ronda JM. The tim-
                                                                                    ing of REM sleep and its relation to spontaneous awakening during tem-
thanks Laura and Luce Jemmi for access to their home-office.
                                                                                    poral isolation in man. Sleep Res 1980;9:280.
REFERENCES                                                                          24. Lavie P, Oksenberg A, Zomer J. “It’s time, you must wake up now.”
                                                                                    Percept Mot Skills 1979;49:447-450.
1. Miles LE, Dement WC. Sleep and aging. Sleep 1980;3:119-220.                      25. Schulz H, Massetani R, Fagioli I, Salzarulo P. Spontaneous awak-
2. Bliwise DL. Review: Sleep in normal aging and dementia. Sleep                    ening from sleep in infants. Electroenceph clin Neurophysiol
1993;16:40-81.                                                                      1985;61:267-271.
3. Blackman MR. Age-related alterations in sleep quality and neuroen-               26. Campbell SS. Spontaneous termination of ad libitum sleep episodes
docrine function: interrelationships and implications. JAMA 2000;879-               with special reference to REM sleep. Electroenceph clin Neurophysiol
881.                                                                                1985;60:237-242.
4. Prinz PN. Sleep and sleep disorders in older adults. J Clin                      27. Barbato G, Barker C, Bender C, Giesen HA, Wehr TA. Extended
Neurophysiol 1995;12:139-146.                                                       sleep in humans in 14 hour nights (LD 10:14): relationship between
5. Van Cauter E, Leproult R, Plat L. Age-related changes in slow wave               REM density and spontaneous awakening. Electroencephalogr clin
sleep and REM sleep and relationship with growth hormone and cortisol               Neurophysiol 1994;90:291-297.
levels in healthy men. JAMA 2000; 284:861-868.                                      28. Webb WB. The spontaneous ending of sleep. Percept Mot Skills
6. Borbély AA. A two process model of sleep regulation. Hum                         1978;46:984-986.
Neurobiol 1982;1:195-204.                                                           29. Webb WB, Campbell S, Hendlin R. The termination of extended
7. Czeisler CA, Weitzman ED, Moore-Ede MC, Zimmerman JC, Knauer                     sleep. Biol Psychol 1980;11:45-48.
RS. Human sleep: its duration and organization depend on its circadian              30. Feinberg I, Koresko RL, Heller N. EEG sleep patterns as a function
phase. Science 1980;210:1264-1267.                                                  of normal and pathological aging in man. J Psychiatr Res 1967;5:107-
8. Zulley J, Wever R, Aschoff J. The dependence of onset and duration               144.
of sleep on the circadian rhythm of rectal temperature. Pflügers Arch               31. Ehlers CL, Kupfer DJ. Effects of age on delta and REM sleep
1981;391:314-318.                                                                   parameters. Electroenceph clin Neurophysiol. 1989;72:118-125.
9. Daan S, Beersma DGM, Borbély AA. Timing of human sleep:                          32. Dijk D-J, Beersma DGM, Van den Hoofdakker RH. All night spec-
Recovery process gated by a circadian pacemaker. Am J Physiol                       tral analysis of EEG sleep in young adult and middle aged male subjects.
1984;246:R161-R178.                                                                 Neurobiol Aging 1989;10:677-682.
10. Edgar DM, Dement WC, Fuller CA. Effect of SCN lesions on sleep                  33. Guazzelli M, et al. Sleep spindles in normal elderly: comparison
in squirrel monkeys: evidence for opponent processes in sleep-wake reg-             with young adult patterns and relation to nocturnal awakening, cognitive
ulation. J Neurosci 1993;13:1065-1079.                                              function and brain atrophy. Electroenceph clin Neurophysiol
11. Dijk D-J, Czeisler CA. Contribution of the circadian pacemaker and              1986;63:526-539.
the sleep homeostat to sleep propensity, sleep structure, electroen-                34. Wei HG, Riel E, Czeisler CA, Dijk D-J. Attenuated amplitude of
cephalographic slow waves and sleep spindle activity in humans. J                   circadian and sleep-dependent modulation of electroencephalographic
Neurosci 1995;15:3526-3538.                                                         sleep spindle characteristics in elderly human subjcts. Neurosci Lett
12. Dijk D-J. Physiology of sleep homeostasis and its circadian regula-             1999;260:29-32.
tion. In: Schwartz WJ, ed. sleep science: Integrating Basic Research and            35 Dijk D-J, Czeisler CA. REM sleep as a gate to wakefulness during
Clinical Practice. Basel: Karger; 1997:10-33.                                       forced desynchrony in young and older people. Sleep 1998;21:S298.
13. Czeisler CA, Duffy JF, Shanahan TL, et al. Stability, precision, and            36. Brown EN, Czeisler CA. The statistical analysis of circadian phase
near-24-hour period of the human circadian pacemaker. Science                       and amplitude in constant-routine core-temperature data. J Biol
1999;284:2177-2181.                                                                 Rhythms 1992;7:177-202.
14. Carskadon MA, Labyak SE, Acebo C, Seifer R. Intrinsic circadian                 37. Dijk D-J, Czeisler CA. Paradoxical timing of the circadian rhythm
period of adolescent humans measured in conditions of forced desyn-                 of sleep propensity serves to consolidate sleep and wakefulness in
chrony. Neurosci Lett 1999;260:129-132.                                             humans. Neurosci Lett 1994;166:63-68.
15. Duffy JF, Dijk D-J, Klerman EB, Czeisler CA. Later endogenous cir-              38. Webb WB. The measurement and characteristics of sleep in older
cadian temperature nadir relative to an earlier waketime in older people.           persons. Neurobiol Aging 1982;3:311-319.
Am J Physiol 1998;275:R1478-R1487.                                                  39. Salzarulo P, Fagioli I, Lombardo P, et al. Sleep stages preceding
16. Dijk D-J, Duffy JF, Riel E, Shanahan TL, Czeisler CA. Ageing and                spontaneous awakenings in the elderly. Sleep Res Online 1999;2:73-77.
the circadian and homeostatic regulation of human sleep during forced               40. Murphy PJ, Rogers NL, Campbell SS. Age differences in the spon-
desynchrony of rest, melatonin and temperature rhythms. J Physiol                   taneous termination of sleep. J Sleep Res 2000;9:27-34.
(Lond) 1999;516:611-627.                                                            41. Landolt H-P, Dijk D-J, Achermann P, Borbély AA. Effect of age on
17. Campbell SS, Terman M, Lewy AJ, Dijk D-J, Eastman CI, Boulos Z.                 the sleep EEG: slow-wave activity and spindle frequency activity in
Light treatment for sleep disorders: consensus report. V. Age-related dis-          young and middle-aged men. Brain Res 1996;738:205-212.
turbances. J Biol Rhythms 1995;10:151-154.                                          42. Dijk D-J, Beersma DGM, Bloem GM. Sex differences in the sleep
18. Haimov I, Lavie P. Circadian characteristics of sleep propensity                EEG of young adults: visual scoring and spectral analysis. Sleep
function in healthy elderly: a comparison with young adults. Sleep                  1989;12:500-507.
1997; 20: 294-300.                                                                  43. Hume KI, Van F, Watson A. A field study of age and gender differ-
19. Carrier J, Monk TH, Reynolds CF 3rd, Buysse DJ, Kupfer DJ. Are                  ences in habitual adult sleep. J Sleep Res 1998;7:85-94.
age differences in sleep due to phase differences in the output of the cir-         44. Armitage R, Hoffmann R, Trivedi M, Rush AJ. Slow-wave activity
cadian timing system? Chronobiol Int 1999;16:79-91.                                 in NREM sleep: sex and age effects in depressed outpatients and healthy
20. Dijk D-J, Duffy JF. Circadian regulation of human sleep and age-                controls. Psychiatry Res 2000; 95:201-213.
related changes in its timing, consolidation and EEG characteristics.               45. Achermann P, Dijk D-J, Brunner DP, Borbély AA. A model of
Ann Med 1999;31:130-140.                                                            human sleep homeostasis based on EEG slow-wave activity:
21. Webb WB, Campbell SS. Awakenings and the return to sleep in an                  Quantitative comparison of data and simulations. Brain Res Bull
older population. Sleep 1980;3:41-46.                                               1993;31:97-113.
SLEEP, Vol. 24, No. 5, 2001                                                   577                               Age-Related Increase in Awakenings—Dijk et al

				
DOCUMENT INFO
Shared By:
Stats:
views:29
posted:3/18/2011
language:English
pages:13