Sie sind auf Seite 1von 9

Sleep, 19(4):318-326

© 1996 American Sleep Disorders Association and Sleep Research Society

Effects of Sleep Deprivation on Performance:


A Meta-Analysis

June J. Pilcher and Allen 1. Huffcutt

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


Department of Psychology, Bradley University, Peoria, Illinois. US.A.

Summary: To quantitatively describe the effects of sleep loss, we used meta-analysis, a technique relatively new
to the sleep research field, to mathematically summarize data from 19 original research studies. Results of our
analysis of 143 study coefficients and a total sample size of 1,932 suggest that overall sleep deprivation strongly
impairs human functioning. Moreover, we found that mood is more affected by sleep deprivation than either cog-
nitive or motor performance and that partial sleep deprivation has a more profound effect on functioning than either
long-term or short-term sleep deprivation. In general, these results indicate that the effects of sleep deprivation may
be underestimated in some narrative reviews, particularly those concerning the effects of partial sleep deprivation.
Key Words: Sleep deprivation-Partial sleep deprivation-Cognitive performance-Motor performance-Mood-
Meta-analysis.

Results of sleep research have traditionally been Meta-analytic reviews, because of their mathemati-
summarized through narrative reviews. In a narrative cal nature, tend to be fairly objective and consistent.
review, a prominent researcher examines the studies in Tn addition, meta-analysis has several statistical advan-
a given area and subjectively arrives at various con- tages. Since each individual study represents a sample
clusions. For example, Krueger (1) reviewed the ef- taken from a larger popUlation, sample results may not
fects of sleep deprivation on performance and con- always match those of the population (i.e. a sampling
cluded that sleep deprivation results in decreased re- error). Mathematically averaging across studies mini-
action times, less vigilance, an increase in perceptual mizes the influence of sampling error since the high
and cognitive distortions and changes in affect. and low random deviations tend to balance out. More-
An alternative approach to summarizing primary re- over, since some studies may be based on a relatively
search has emerged over the last couple of decades. small sample, problems with low power are avoided
Denoted as "meta-analysis," it is based on a mathe- since no formal significance testing is done at the in-
matical rather than a subjective combination of studies dividual study level. (Effectually, all individual sam-
(see 2,3). Like narrative reviews, meta-analytic re- ples are combined into one large sample which should
views can contribute meaningful information and con- be largely representative of the general population.)
clusions. Specifically, a meta-analysis can provide de- While meta-analysis has been gaining popularity in
tails regarding the strength and consistency of an ex- other fields, such as personnel management, clinical
perimental effect and the conditions or factors which psychology and education (e.g. 4-6), it has yet to gain
moderate the effect. widespread acceptance in the sleep research commu-
Each approach to summarizing research has its own nity. To date, only four articles have reported meta-
unique strengths. Narrative reviews are typically done analyses of primary sleep studies. Benca et al. (7) re-
by someone who has considerable experience and ex- viewed sleep patterns in psychiatric disorders. Hudson
pertise in that area, someone who is in a very good et al. (8) looked at polysomnographic measures in
position to understand the details and intricacies of good and bad sleep. Knowles and MacLean (9) as-
each study. Thus, more subjective factors, such as the sessed age-related changes in sleep. Lastly, Koslowsky
quality of each study, can easily be taken into consid- and Babkoff (10) examined the effect of total sleep
eration when forming conclusions. deprivation on work-paced and self-paced task perfor-
mance.
Accepted for publication January 1996. Moreover, consistent with the pattern observed in
Address correspondence and reprint requests to June J. Pilcher,
Ph.D., Department of Psychology, Bradley University, Peoria, IL other fields, some of the first meta-analyses to appear
61625, U.S.A. have been somewhat limited in scope. In regard to
318
SLEEP DEPRIVATION 319

sleep deprivation, there are a number of potentially addition, our analysis utilizes data from both partial
important moderator variables which could be taken and total sleep deprivation studies.
into account. For example, there are three types of
measures commonly used to assess the effects of sleep
deprivation: cognitive performance, motor perfor- METHOD
mance and mood. And, there may be additional vari-
Location of study data
ables operating within each of these measures which
may further change the effects of deprivation on func- The data for our meta-analyses were located by ex-
tioning. tensive literature searches on the computerized data-
Some evidence does, in fact, suggest that perfor- bases PsychLit (1987-1993) and Med On-Line (1986-

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


mance varies on different types of cognitive tasks. In 1993) and in the extensive Sleep Research bibliogra-
a comprehensive survey of the sleep deprivation lit- phy (1986-1993). These searches resulted in refer-
erature, Johnson (11) concluded that the results from ences for studies published between 1984 and 1992.
studies using accuracy as the performance variable de- We then extended our search by locating sleep depri-
pended on the type of cognitive measure (e.g. logical vation studies that were referenced in these studies. A
reasoning, mental addition, visual search tasks, word total of 56 primary articles were located which inves-
memory tasks). In addition, other narrative reviewers tigated the effects of sleep deprivation on performance.
(e.g. 12) have suggested that the length and pacing of
cognitive tasks may affect performance.
Similarly, with motor performance measures, the Decision rules
use of different tasks could affect results. In a review
of the effect of sleep loss on exercise, Martin (l3) We established the following criteria for inclusiOll
concluded that the effect of sleep deprivation depends in our analysis. First, enough information had to be
on the type and length of the motor task. For example, provided to allow computation of an effect size statis-
while several studies suggest that exercise is not ad- tic (explained in the next section) for each dependent
versely affected by sleep deprivation (e.g. 14-17), oth- measure. In most cases this meant direct reporting of
the means (or a clear enough graph such that the
ers report that performance on certain endurance tasks
means could be estimated) and standard deviations for
is decremented (e.g. 18). In an extensive review of the
the sleep-deprived and non-sleep-deprived groups. In
sleep deprivation and exercise performance literature,
cases where results were expressed only as a t or a 1
VanHelder and Radomski (19) concluded that sleep
df, an effect size statistic was computed through a sta-
deprivation up to 72 hours does not affect muscle
tistical transformation (see 2).
strength or reaction but does decrease time to exhaus-
Second, the study had to involve short-term total
tion.
sleep deprivation (:::;45 hours), long-term total sleep
One factor that may affect all three measures is the
deprivation (>45 hours), or partial sleep deprivation
length of sleep deprivation. Naitoh (20), for example,
(sleep period of <5 hours in a 24-hour period). The
found that sleep deprivation of less than 46 hours is
determination of short-term and long-term sleep de-
usually too short to have a substantial effect on either privation followed the criteria set by Koslowsky and
cognitive or motor tasks. Other researchers, however, Babkoff (10). The decision to use a sleep duration of
have reported performance decrements at sleep loss <5 hours as the criteria for partial sleep deprivation
durations of less than 45 hours (e.g. 21). While mood was made after reviewing a number of studies which
appears to be decremented by sleep deprivation (e.g. allowed the subjects to sleep for short periods of time
12,22-24), it is unclear whether different types of de- in a 24-hour period and then selecting a number that
privation differentially impact mood. reflected a natural cutoff point.
The purpose of this study is to use the meta-analytic Third, the study had to use either a cognitive per-
technique to provide a comprehensive, quantitative formance task, a motor performance task or a mood
analysis of the effects of sleep deprivation on func- scale as the dependent measure. In particular, cognitive
tioning. Our work extends that of Koslowsky and Bab- performance tasks (e.g. logical reasoning tasks, mental
koff (10) in that we evaluate a number of additional addition tasks, Torrance tests) had to be either :::;6 min-
moderator variables. Specifically, we categorize and utes in duration or :=:: 10 minutes in duration. Motor
separately analyze the measures as being either mood performance tasks (e.g. serial reaction time, treadmill
assessment, motor task performance or cognitive task walking, manual dexterity tasks) had to be either 53
performance. Then, we further differentiate both the minutes in duration or :=::8 minutes in duration. The
motor and cognitive task performance categories ac- above times for cognitive and motor tasks were based
cording to the length and complexity of the tasks. In on our review of a large number of studies and ap-
Sleep, Vol. 19, No.4, 1996
320 1. J. PILCHER AND A. /. HUFFCUTT

TABLE 1. Coding characteristics respectively, for the experimental group and Nc and Sc
are the sample size and standard deviation for the con-
Type of sleep deprivation
A. Short-term sleep deprivation (:545 hours) trol group.
B. Long-term sleep deprivation (>45 hours) . .
C. Partial sleep deprivation «5 hours sleep III a 24-hour penod
(NE)SE2 + (Nd Sc 2
Sw = (2)
Type of dependent measure NE + Nc
A. Cognitive task performance Careful attention was paid to the sign of the effect
B. Motor task performance
C. Mood assessment size during coding, since d values mathematically can
Type of task
be positive or negative. Studies were uniformly coded
A. Simple task such that a negative d represented situations where the

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


B. Complex task experimental (i.e. sleep-deprived) group did worse on
Duration of task the dependent measure than the control group, while
A. Short (motor, :53 minutes; cognitive, :56 minutes) a positive d represented situations where the experi-
B. Long (motor, >8 minutes; cognitive, 210 minutes) mental group did better than the control group.
In total we were able to code 143 d values repre-
senting 1:932 subjects from the 19 primary studies.
peared to represent natural separations that could dif-
(Since most sleep deprivation studies used more than
ferentiate short from long durations.
one measure of performance, these data are not totally
Finally, if subjects performed multiple tasks of the
independent of each other. Such a situation is common
same type, all data had to be reported. If a study re-
in meta-analysis.) This subject pool represents a broad
ported results only for positive or statistically signifi-
range of subjects including both genders and a wide
cant effects and not for negative or nonsignificant ef-
age range. The 19 primary studies were divided ap-
fects, then it was dropped.
proximately equally between the three sleep depriva-
Of the 56 primary studies located, 37 were rejected
tion categories: four in short-term sleep deprivation
because they did not meet the above criteria (a com-
(37,42,43,45), six in long-term sleep deprivation
plete bibliography of the studies not used in this meta-
(33,35,46,47,50,51), three in both short-term and long-
analysis is available on request). Of these 37 articles,
term sleep deprivation (34,36,38) and six in partial
29 did not present data in a fashion that would allow
sleep deprivation (39-41,44,48,49). Such a data set is
us to compute an effect size (e.g. 21,25-28), seven did
large and provides the opportunity to do meaningful
not use a design that met our qualifications for type of
analyses.
task or type of deprivation (e.g. 29-31) and one pre-
The reliability of the coding process was assessed
sented only positive results (32). We attempted to di-
by having two independent researchers code all 19
rectly contact the authors of these articles for more
studies. The correlation between the raters was 0.60
information, but a majority either could not be con-
for type of sleep deprivation, 1.00 for type of depen-
tacted or could not locate the data in question.
dent measure, 0.94 for type of task, 0.78 for task du-
ration, 0.93 for sample size and 0.98 for effect size.
Coding of study information The lowest interrater reliability was seen for type of
sleep deprivation. This was due to the initial analysis
The final data set was formed from the remaining not clearly distinguishing between short-term and par-
19 primary journal articles (33-51). A special coding
tial sleep deprivation. With the second coding, this dis-
sheet was developed to capture information from these
tinction was clarified, which changed some of the pre-
studies. The information coded is listed in Table 1.
viously coded short-term deprivations to partial sleep
Effect size statistics, which indicate how many stan-
deprivations. All disagreements across all coding cri-
dard deviation units the experimental group was dif-
teria were investigated by the two researchers and re-
ferent from the control group, were computed using
solved. These results indicate that information could
the methodology outlined by Hunter and Schmidt (2).
be coded reliably from the studies.
The formula for the effect size statistic, d, is shown
below, where XE is the mean of the experimental
group, Xc is the mean of the control group and Sw is Meta-analytic computations
the standard deviation pooled across both groups.
The actual computations for the meta-analyses were
(1) performed using a SAS (SAS Institute, 1990) PROC
MEANS program developed by Huffcutt et al. (52)
The formula for computing the pooled standard devi- that takes the d values from the various studies and
ation in the effect size formula is shown below, where combines them mathematically. The result is an esti-
NE and SE are the sample size and standard deviation, mate of the average effect size across the studies (i.e.
Sleep, Vol. 19, No.4, 1996
SLEEP DEPRIVATION 321

the average number of standard deviations the exper- Schmidt (2) noted, the more different the individual
imental distributions was offset from the control dis- means are from the overall mean, the more influence
tribution) and the variability observed around this av- that characteristic has on the strength of the experi-
erage. All computations are done weighting by sample mental effect.]
size, since studies based on a larger sample are more Similarly, we separated the studies into the three
stable than those based on a smaller sample (2,3). categories of dependent measure (cognitive task per-
It should be noted that Huffcutt et al.'s program formance, motor task performance and mood scales)
does not provide any tests of statistical significance. and conducted a separate meta-analysis for each cat-
Formal significance testing is typically not done in a egory. Finally, we conducted a meta-analysis in which
meta-analysis, as the meta-analytic procedures were we combined across the two prominent design char-

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


developed to avoid the problems and limitations as- acteristics to assess whether the effects of deprivation
sociated with significance testing (2,3). Moreover, for a particular dependent measure changed depending
since sampling errors tend to be averaged out when on the type of sleep deprivation (i.e. an interaction
combining across studies, results of a meta-analysis effect).
are thought to represent direct estimates of the strength A third goal was to do a supplemental assessment
of a relationship iQ the population. The average effect of whether performance on cognitive tasks changed
size represents the overall strength of a relationship, according to either the type of task (simple vs. com-
plex) and/or the task length (short vs. long). Thus, we
while the variability around the average reflects the
took the cognitive task studies which were already
degree to which other variables moderate the relation-
sorted according to the type of deprivation, further
ship. (Therefore, high variability does not imply a lack
sorted them into simple and complex task categories
of an effect but rather that the strength of the effect
and conducted a separate meta-analysis for each re-
depends strongly on other variables.)
sulting combination. Then, within each type of depri-
vation, we re-sorted the studies into short and long
Analyses duration categories and conducted separate meta-anal-
yses for each resulting combination. These procedures
Our first goal was to assess the overall effect of allowed us to assess whether, for a particular type of
sleep deprivation. More specifically, we attempted to deprivation, performance on a cognitive task depended
answer two questions. First, at a general level, how on the type and/or length of the task.
well do experimental subjects do, on average, relative Lastly, a fourth and similar goal was to assess
to control subjects in a sleep deprivatiori study? And whether performance on motor tasks varied according
second, how constant (i.e. stable) are the effects of to either the type of task (simple vs. complex) or the
sleep deprivation across different study designs? length of the task (short vs. long). As described in the
To answer these questions we conducted a meta- preceding paragraph, studies already separated by type
analysis of all 143 effect sizes collectively. The mean of deprivation were then further separated by type of
d score from this analysis indicated the average num- task and length of task.
ber of standard deviations that the sleep-deprived In closing, the above analyses were designed to fol-
group was different from the non-sleep-deprived Iowa "hierarchical" strategy, as is typically done in
group, collapsing across all the different study design a meta-analytic investigation (2). In particular, we
characteristics. The standard deviation in d scores was started at an overall summary level and then progres-
a reflection of the extent to which design characteris- sively made the analyses more and more specific. Two
tics affected the magnitude of the difference between comments should be noted in regard to this strategy.
deprived and nondeprived subjects. First, unlike cognitive and motor task performance, we
A second goal was to investigate specifically how did not break mood measures down by additional fea-
the effects of sleep deprivation vary according to the tures such as length and complexity. Conceptually,
two most prominent study design characteristics, the such features were not as meaningful with mood mea-
type of deprivation and the type of measure. Our at- sures as they were with cognitive and motor tasks. And
tempt here was to assess quantitatively how much dif- second, the number of studies being analyzed at any
ference each of these two characteristics makes in one time progressively decreased as the data set was
terms of the effects of sleep deprivation. To assess the split into more and more subcategories. Naturally, the
smaller the number of studies in a given category, the
influence of type of deprivation, we separated the stud-
more tentative the results become.
ies into the three main categories (short-term, long-
term and partial sleep deprivation) and conducted a
separate meta-analysis for each category. We then RESULTS
looked to see if there was a difference in the mean d Results of the overall analysis of all 143 coefficients
scores across the three categories. [As Hunter and combined are presented in the top portion of Table 2.
Sleep. Vol. 19. No.4. 1996
322 1. 1. PILCHER AND A. I. HUFFCUTT

Deprived Nondeprived
TABLE 2. Meta-analyses of sleep deprivation overall and Group Group
by study design characteristic
\ \. ('
;'

Analysis aa SD(d)" N, TSS


Overall -1.37 2.08 143 1,932
Deprivation
Short-term -1.21 1.40 34 683
Long-term -1.27 2.10 79 799
Partial -2.04 2.55 30 450
Type of measure Scores on Dependent Measures
Motor -0.87 1.96 58 790
FIG. 1. Illustration of the overall difference between sleep-de-

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


Cognitive -1.55 1.75 65 949
Mood -3.16 2.53 20 193 prived and non-sleep-deprived subjects.

Abbreviations used: a, average effect size; SD(d), standard devi-


ation of effect sizes; N" number of study coefficients in the analysis;
TSS, total sample size from those coefficients. vation. In terms of type of measure, sleep deprivation
" Averages and standard deviations were computed using sample in general appeared to have the least effect on motor
size weighting.
tasks, a greater effect on cognitive tasks, and an even
greater effect on mood. [However, the average effect
As shown, the mean effect size collapsing across all size for motor tasks is still considered to be a large
study characteristics was -1.37, indicating that the experimental effect. For reference, an average effect
sleep-deprived subjects performed at a level 1.37 stan- size of 0.20 is considered to be a small experimental
dard deviations lower than the performance level of effect, 0.50 is considered medium, and 0.80 or greater
the non-sleep-deprived subjects. The difference of 1.37 is considered to be large (54).]
standard deviations between the two distributions is Results for type of dependent measure crossed with
graphically illustrated in Fig. 1. In more pragmatic type of deprivation are shown in Table 3. These results
terms, such a finding suggests that a person at the 50th in general suggest an interaction between these two
percentile in the deprived group (shown as the dark design characteristics. For motor performance tasks,
dot in Fig. 1) performs roughly equivalent to a person the means were fairly close across all three types of
at the 9th percentile in the nondeprived group. [This deprivation, suggesting that performance on motor
is based on the assumption that both the deprived and tasks is relatively unaffected by the type of depriva-
nondeprived groups roughly approximate a z distri- tion. For cognitive performance tasks, the means were
bution. Percentiles were obtained from Minium et al. dissimilar, with performance being considerably more
(53).] The relatively large standard deviation across decremented with partial sleep deprivation than either
the effect sizes (2.08) suggests that study design char- short-term or long-term deprivation. Similarly, mood
acteristics do make a considerable difference in terms appeared to be much more affected by partial depri-
of how deprived subjects perform relative to nonde- vation than by long-term deprivation. (There were no
prived subjects. studies of the effect of short-term deprivation on mood
Results of the meta-analyses for the two most prom- in the final data set.)
inent study design characteristics, the type of depri- Results of the supplemental analyses of cognitive
vation and the type of measure, are also presented in performance tasks are presented in Table 4. For short-
Table 2. As shown, the type of deprivation does appear term deprivation, performance on complex and long
to make a difference; partial sleep deprivation ap- tasks was considerably more decremented than on sim-
peared to have a considerably greater overall impact ple and short tasks, respectively. For long-term depri-
on subjects than either short-term or long-term depri- vation, opposite results were found, with performance

TABLE 3. Meta-analyses for type of measure and type of deprivation crossed

Type of sleep deprivation


Short-term Long-term Partial
Type of
measure aa SD(d)a N, TSS aa SD(d)a N, TSS aa SD(d)" N, TSS
Motor -0.77 1.31 10 173 -0.92 2.46 35 386 -0.85 1.33 13 231
Cognitive -1.36 1.39 24 510 -1.04 0.71 31 278 -3.01 2.88 10 161
Mood -2.75 2.25 13 135 -4.10 2.88 7 58
Abbreviations used: a, average effect size; SD(d), standard deviation of effect sizes; N" number of study coefficients in the analysis;
TSS, total sample size from those coefficients.
a Averages and standard deviations were computed using sample size weighting.

Sleep, Vol. 19, No.4, 1996


SLEEP DEPRIVATION 323

TABLE 4. Supplemental meta-analyses for cognitive measures

Type of sleep deprivation


Short-term Long-term Partial

da SD(d)a Nc TSS da SD(d)a Nc TSS da SD(d)a Nc TSS


Overall b -1.36 1.39 24 510 -1.04 0.71 31 278 -3.01 2.88 10 161
Type of task
Simple -0.37 0.41 4 76 -1.17 0.78 14 134 -3.48 3.01 7 123
Complex -1.53 1.43 20 434 -0.91 0.62 17 144 -1.49 1.69 3 38
Duration of task
Short -0.36 0.78 9 178 -1.64 0.85 3 32 1.18 0.00 1 23

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


Long -1.90 1.35 15 332 -0.96 0.66 28 246 -3.71 2.50 9 138
Abbreviations used: d, average effect size; SD(d), standard deviation of effect sizes; N c' number of study coefficients in the analysis;
TSS, total sample size from those coefficients.
a Averages and standard deviations were computed using sample size weighting.
b Data for "Overall" are from Table 3.

being considerably worse on short tasks than on long non-sleep-deprived subjects (i.e. a 1.37 standard de-
tasks and slightly worse on simple tasks than on com- viation difference between the distributions). Although
plex tasks. For partial deprivation, subjects did worse most of the sleep research community may concur
on tasks that were simple than those that were complex with these results, there are a surprising number of
and worse on tasks that were longer. However, the rel- scientists outside the sleep research field who have
atively small samples involved makes these findings concluded that sleep deprivation has no profound ef-
much more tentative. fect on performance and only a marginal effect on
Results of the supplemental analyses of motor per- mood. For example, many widely known professionals
formance tasks are presented in Table 5. As shown, outside the sleep research field writing introductory
there were no studies involving complex motor tasks texts in psychology and physiological psychology
in the final data set. For length of task, performance have stated that the effects of sleep deprivation on hu-
was worse on long tasks for all three types of depri- man functioning are minimal (55-59).
vation. Once again, the relatively small samples in- Another major finding of our investigation was that
volved makes these findings tentative. the effects of sleep deprivation vary according to two
key moderator variables. First, we found a substantial
DISCUSSION difference across the three dependent measures. Spe-
cifically, we found that cognitive performance was
Our results confirm that sleep deprivation has a sig- more affected by sleep deprivation than motor perfor-
nificant effect on human functioning. By quantitatively mance and that mood was much more affected than
combining across primary studies, we found that the either cognitive or motor performance. It is important
mean level of functioning of sleep-deprived subjects to note, however, that even on motor tasks the sleep-
was comparable to, that of only the 9th percentile of deprived subjects performed considerably worse than

TABLE S. Supplemental meta-analyses for motor measures

Type of sleep deprivation


Short-term Long-term Partial

da SD(d)a Nc TSS da SD(d)a Nc TSS da SD(d)a Nc TSS


Overall b -0.77 1.31 10 173 -0.92 2.46 35 386 -0.85 1.33 13 231
Type of task
Simple -0.77 1.31 10 173 -0.92 2.46 35 386 -0.85 1.33 13 231
Complex
Duration of task
Short -0.52 1.04 6 110 0.02 0.62 26 275 -0.27 1.18 10 162
Long -1.22 1.58 4 63 -3.26 3.51 9 III -2.20 0.24 3 69
Abbreviations used: d, average effect size; SD(d), standard deviation of effect sizes; N c' number of study coefficients in the analysis;
TSS, total sample size from those coefficients.
a Averages and standard deviations were computed using sample size weighting.
b Data for "Overall" are from Table 3.

Sleep, Vol. 19, No.4, 1996


324 J. J. PILCHER AND A. I. HUFFCUTT

the non-sleep-deprived subjects. This pattern of dif- ological concerns could account for all of the decre-
ferences among the three types of dependent measures ment found in cognitive tasks and mood.
is not surprising and is consistent with the viewpoints One clear direction for future research is to address
of many sleep researchers (1,11,19,24). why partial sleep deprivation may have such a pro-
That mood was more influenced than the objective nounced effect on mood and cognitive performance.
performance measures is not surprising. Since mood For example, partial sleep deprivation may alter cer-
is usually assessed using self-reporting methodology, tain circadian rhythm effects on performance and
it is possible that the subjects could be overestimating mood. While total sleep deprivation has been found to
the effect of sleep deprivation on their mood. How- interact with circadian rhythms (61,62), few studies
ever, it is important to note that on average, the sleep- have investigated the effects of partial sleep depriva-

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


deprived subjects reported mood ratings that were over tion on circadian rhythms. In addition, partial sleep
3 standard deviations worse than those of non-sleep- deprivation may be similar to fragmented sleep in that
deprived subjects. While part of these differences subjects in both cases obtain at least some sleep. Since
could be attributable to self-reporting error, it is likely sleep fragmentation has been shown to significantly
that sleep deprivation has a negative effect on mood. decrease performance and mood (63,64), it is possible
Second, we found a substantial difference across the that the effects of partial sleep deprivation more close-
three types of sleep deprivation. Unexpectedly, partial ly resemble those of sleep fragmentation than those of
sleep deprivation had a much stronger overall effect total sleep deprivation. Furthermore, partial sleep de-
on the dependent measures than either short-term or privation could have a unique effect on certain psy-
long-term sleep deprivation. On average, partially chological variables. Decreased interest and attention,
sleep-deprived subjects performed at a level 2 standard for example, are thought to be two prominent variables
deviations below that of the non-sleep-deprived sub- related to total sleep deprivation (65) and could be
jects, compared to about a 1 standard deviation differ- investigated with partial sleep deprivation. Similarly,
partial sleep deprivation could have certain physiolog-
ence for both long-term and short-term deprivation.
ical effects that are either different or more pro-
In addition, we found an interaction between the
nounced than those of total sleep deprivation. Al-
two key moderator variables, length of sleep depriva-
though numerous studies have been conducted on
tion and type of dependent mcasure. We found that
physiological changes following total sleep depriva-
detriments on motor task performance were relatively
tion (e.g. 66,67), few studies have specifically inves-
constant across the three types of sleep deprivation. In
tigated physiological changes following partial sleep
contrast, cognitive performance and mood were con-
deprivation. In sum, the effects of partial sleep depri-
siderably more decremented under partial sleep depri-
vation need to be more thoroughly investigated, par-
vation than under long-term or short-term deprivation.
ticularly since partial sleep loss is a relatively common
Narrative reviews clearly do not indicate such an
condition in our society.
overwhelming decrement in performance due to partial
There are several limitations that should be noted
sleep deprivation. For example, two reviews of the ef-
about our investigation. First, we could not use a num-
fects of sleep deprivation reported mixed findings from ber of the primary studies that we found because they
a variety of partial sleep deprivation studies (1,60). did not meet our established criteria. Although the
Similarly, more recent reviews concluded that the ef- meta-analytic technique does not require that all pos-
fects of partial sleep loss on medical residents' per- sible literature be utilized, it is important that coverage
formance were inconclusive (12,24). of the literature not be systematically biased. In our
It is possible that the difference in methodology be- case, there is no a priori reason to assume that the
tween the narrative reviews and our quantitative anal- articles we rejected were different in any systematic
ysis could account for the disagreement on the effects way from the articles that we used. Second, a general
of partial sleep deprivation. Alternatively, the disagree- concern about the meta-analytic technique is that it
ment could be attributable to differences in the studies combines across data that may be inherently positive.
reviewed. For example, four of the six partial sleep This same point, however, can be made concerning
deprivation studies in our meta-analysis used medical narrative reviews. In both cases, the reviewers are sim-
residents as subjects, and three of these studies spe- ply evaluating published data. Also, this concern may
cifically used medical-related tasks as dependent mea- not be as valid in this particular meta-analysis since a
su~es. It is possible that these tasks were more easily number of the studies that we used actually included
affected by sleep deprivation than more traditional nonsignificant data. Third, we were not able to draw
cognitive and motor tasks. However, given the mag- robust conclusions from our final level of analysis,
nitude of the differences between partially deprived which examined the influence of task length and com-
and control subjects, it is unlikely that these method- plexity on performance. Such analyses may become
Sleep, Vol. 19, No.4, 1996
SLEEP DEPRIVATION 325

possible in the future as more primary studies become 17. Webb WB, Kaufmann DA, Devy CM. Sleep deprivation and
physical fitness in young and older subjects. J Sports Med Phys
available. Lastly, other moderator variables, such as Fitness 1981;21:198-202.
age or gender, may influence the interpretation of the 18. Martin BJ. Effect of sleep deprivation on tolerance of prolonged
effects of sleep deprivation. Similarly, these variables exercise. Eur J Appl Physio/ 1981 ;47:345-54.
19. VanHelder T, Radomski MW. Sleep deprivation and the effect
could be investigated as more studies become avail- on exercise performance. Sports Med 1989;7:235-47.
able. 20. Naitoh P. Sleep deprivation in human subjects: a reappraisal.
Nonetheless, these results allow us to draw two ma- Waking Sleeping 1976;1:53-60.
21. Babkoff H, Genser SG, Sing HC, Thorne DR, Hegge FW. The
jor conclusions. First, sleep deprivation has a substan- effects of progressive sleep loss on a lexical decision task: re-
tial effect on mood and motor and cognitive perfor- sponse lapses and response accuracy. Behav Res Methods In-
mance in humans. And, second, partial sleep depri- struments Computers 1985;17:614-22.
22. Cutler NR, Cohen HB. The effect of one night's sleep loss on

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


vation has a greater negative effect on mood and cog- mood and memory in normal subjects. Compr Psychiatry
nitive performance than either short-term or long-term 1979;20:61-6.
sleep deprivation. 23. Reynolds CF, Kupfer DJ, Hoch CC, Stack JA, Houck PR, Ber-
man SR. Sleep deprivation in healthy elderly men and women:
Acknowledgements: We thank Cristin B. Dooley for effects on mood and on sleep during recovery. Sleep 1986;9:
492-501.
her assistance with the initial organization and analysis of 24. Leung L, Becker CEo Sleep deprivation and house staff perfor-
the data. mance. J Occup Med 1992;34: 1153-60.
25. Buck L. Sleep loss effects on movement time. Ergonomics
1975;18:415-25.
REFERENCES 26. Donnell JM. Performance decrement as a function of total sleep
loss and task duration. Percept Mot Skills 1969;29:711-14.
1. Krueger OP. Sustained work, fatigue, sleep loss and perfor- 27. Jacques CHM, Lynch JC, Samkoff JS. The effects of sleep loss
mance: a review of the issues. Work Stress 1989;3:129-41. on cognitive performance of resident physicians. Fam Pract
2. Hunter JE, Schmidt FL. Methods of meta-analysis: correcting 1990;30:223-29.
error and bias in research findings. Newbury Park, CA: Sage 28. Steyvers FJJM. The influence of sleep deprivation and knowl-
Publications, 1990. edge of results on perceptual encoding. Acta Psychol 1987;66:
3. Hunter JE, Schmidt FL, Jackson GB. Meta-analysis: cumulating 173-87.
research findings across studies. Beverly Hills, CA: Sage Pub- 29. Eilers K, Nachreiner F. Time of day effects in vigilance perfor-
lications, 1982. mance at simultaneous and successive discrimination tasks. In:
4. Huffcutt AI, Arthur W Jr. Hunter and Hunter (1984) revisited: Costa G, Cesana G, Kogi K, Wedderburn A, eds. Shiftwork:
interview validity for entry-level jobs. J Appl Psychol 1994;79: health, sleep and performance. Frankfurt: Verlag Peter Lang,
184-90. 1989:467-72.
5. Svartberg M, Stiles Te. Comparative effects of short-term psy- 30. Haslam DR. Sleep deprivation and naps. Behav Res Methods
chodynamic psychotherapy: a meta-analysis. J Consult Clin Instruments Computers 1985;17:46-54.
Psychol 1991;59:704-14. 31. Jazwinska EC, Adam K. Diurnal change in stature: effects of
6. Peers IS, Johnston M. Influence of learning context on the re- sleep deprivation in young men and middle-aged men. Exper-
lationship between A-level attainment and final degree perfor- ientia 1985;41:1533-35.
mance: a meta-analytic review. Br J Educ Psychol 1994;64:1- 32. Glenville M, Wilkinson RT. Portable devices for measuring per-
18. formance in the field: the effects of sleep deprivation and night
7. Benca RM, Obermeyer WH, Thisted RA, Gillin Je. Sleep and shift on the performance of computer operators. Ergonomics
psychiatric disorders: a meta-analysis. Arch Gen Psychiatry 1979;22:927-33.
1992;49:651-68. 33. Symons ID, VanHeider T, Myles WS. Physical performance and
8. Hudson 11, Pope HG, Sullivan LE, Waternaux CM, Keck PE, physiological responses following 60 hours of sleep deprivation.
Broughton RJ. Good sleep, bad sleep: a meta-analysis of poly-
Med Sci Sports Exerc 1988;20:374-80 .
. somnographic measures in insomnia, depression, and narcolep-
34. Akerstedt T, Froberg IE. Psychophysiological circadian rhythms
sy. BioI Psychiatry 1992;32:958-75.
in women during 72h of sleep deprivation. Waking Sleeping
9. Knowles JB, MacLean AW. Age-related changes in sleep in de-
1977;1:387-94.
pressed and healthy subjects. Neuropsychopharmacology
35. Angus RG, Heslegrave RJ, Myles WS. Effects of prolonged
1990;3:251-59.
10. Koslowsky M, Babkoff H. Meta-analysis of the relationship be- sleep deprivation, with and without chronic physical exercise,
tween total sleep deprivation and performance. Chronobiol Int on mood and performance. Psychophysiology 1985;22:276-82.
36. Babkoff H, Thorne DR, Sing HC, Genser SG, Taube SL, Hegge
1992;9: 132-36.
11. Johnson Le. Sleep deprivation and performance. In: Webb WB, FW. Dynamic changes in work/rest duty cycles in a study of
ed. Biological rhythms, sleep, and performance. Chichester, sleep deprivation. Behav Res Methods Instruments Computers
U.K.: Wiley, 1982:111-41. 1985;17:604-13.
12. Samkoff JS, Jacques CHM. A review of studies concerning ef- 37. Bond V, Balkissoon B, Franks BD, et al. Effects of sleep de-
fects of sleep deprivation and fatigue on residents' performance. privation on performance during submaximal and maximal ex-
Acad Med 1991;66:687-93. ercise. J Sports Med Phys Fitness 1985;26:169-74.
13. Martin BJ. Sleep deprivation and exercise. Exerc Sport Sci Rev 38. Brodan V, Kuhn E. Physical performance in man during sleep
1986;14:213-29. deprivation. J Sports Med Phys Fitness 1967;7:28-30.
14. Pickett OF, Morris AF. Effects of acute sleep and food depri- 39. Denisco RA, Drummond IN, Gravenstein JS. The effect of fa-
vation on total body response time and cardiovascular perfor- tigue on the performance of a simulated anesthetic monitoring
mance. J Sports Med Phys Fitness 1975; 15:49-56. task. J Clin Monit 1987;3:22-4.
15. Martin BJ, Gaddis GM. Exercise after sleep deprivation. Med 40. Friedman RC, Bigger JT, Kornfeld DS. The intern and sleep
Sci Sports Exerc 1981;13:220-23. loss. N Engl J Med 1971;285:201-3.
16. Reilly T, Deykin T. Effects of partial sleep loss on subjective 41. Friedman J, Globus G, Huntley A, Mullaney D, Naitoh P, John-
states, psychomotor and physical performance tests. J Hum son L. Performance and mood during and after gradual sleep
Movement Stud 1983;9:157-70. reduction. Psychophysiology 1977; 14:245-50.

Sleep, Vol. 19, No.4, 1996


326 1. 1. PILCHER AND A. I. HUFFCUIT

42. Holland GJ. Effects of limited sleep deprivation on performance 55. Benjamin LT Jr, Hopkins JR, Nation JR. Psychology, 3rd edi-
of selected motor tasks. Res Q 1968;39:285-94. tion. New York: Macmillan College Publishing Co., 1994.
43. Horne JA. Sleep loss and "divergent" thinking ability. Sleep 56. Pinel JPJ. Biopsychology, 2nd edition. Boston: Allyn & Bacon,
1988;11 :528-36. 1993.
44. Legg SJ, Patton IF. Effects of sustained manual work and partial 57. Carlson NR. Physiology of behavior, 5th edition. Boston: Allyn
sleep deprivation on muscular strength and endurance. Eur J & Bacon, 1994.
Appl Physio/ 1987;56:64-8. 58. Shaver KG, Tarpy RM. Psychology. New York: Macmillan Pub-
45. Linde L, Bergstrom M. The effect of one night without sleep lishing Co., 1993.
59. Weiten W. Psychology themes and variations, 3rd edition. New
on problem-solving and immediate recall. Psychol Res 1992;54:
York: Brooks/Cole Publishing Co., 1995.
127-36.
60. Naitoh P. Sleep loss and its effects on peiformance. Bureau of
46. Myles WS. Sleep deprivation, physical fatigue, and the percep- Medicine and Surgery, Dept. of the Navy, 1969; Rep. No. 68-3.
tion of exercise intensity. Med Sci Sports Exerc 1985;17:580-4. 61. Monk TH, Fookson JE, Kream R, Moline ML, Pollak CPo Weitz-
47. Quant JR. The effect of sleep deprivation and sustained military man MB. Circadian factors during sustained performance: back-

Downloaded from https://academic.oup.com/sleep/article-abstract/19/4/318/2749842 by guest on 28 August 2019


operations on near visual performance. Aviat Space Environ grounds and methodology. Behav Res Methods Instruments
Med 1992;63:172-6. Computers 1985; 17: 19-26.
48. Reznick RK, Folse IR. Effect of sleep deprivation on the per- 62. Naitoh P, Englund CE, Ryman DH. Circadian rhythms deter-
formance of surgical residents. Am J Surg 1987;154:520-5. mined by cosine curve fitting: analysis of continuous work and
49. Robbins J, Gottlieb F. Sleep deprivation and cognitive testing in sleep loss data. Behav Res Methods Instruments Computers
internal medicine house staff. West J Med 1990;152:82-6. 1985; 17 :630-41.
50. Symons JD, Bell DG, Pope J, VanHelder T, Myles WS. Elec- 63. Bonnet MH. Performance and sleepiness as a function of fre-
tromechanical response times and muscle strength after sleep quency and placement of sleep disruption. Psychophysiology
deprivation. Can J Sport Sci 1988;13:225-30. 1986;23:263-7l.
51. Webb WB. A further analysis of age and sleep deprivation ef- 64. Bonnet MH. Infrequent periodic sleep disruption: effects on
sleep, performance and mood. Physiol Behav 1989;45:1049-55.
fects. Psychophysiology 1985 ;22: 156-61.
65. Meddis R. Cognitive dysfunction following loss of sleep. In:
52. Huffcutt AI, Arthur W Jr, Bennett W. Conducting meta-analysis
Burton E, ed. The pathology and psychology of cognition. Lon-
using the 'PROC MEANS' procedure in SAS. Educ Psychol don: Methuen, 1982:225-52.
Measurement 1993;53: 119-31. 66. Fiorica V, Higgins EA, Iampietro F, Lategola MT, Davis AW.
53. Minium EW, King BM, Bear G. Statistical reasoning in psy- Physiological responses of men during sleep deprivation. J Appl
chology and education, 3rd edition. New York: John Wiley & Physiol 1968;24:167-76.
Sons, 1993. 67. Kant GJ, Genser SG, Thorne DR, Pfalser JL, Mougey EH. Ef-
54. Hedges LV, Olkin I. Statistical methods for meta-analysis. Or- fects of 72 hour sleep deprivation on urinary cortisol and indices
lando, FL: Academic Press, 1985. of metabolism. Sleep 1984;7:142-46.

Sleep, Vol. 19, No.4, 1996

Das könnte Ihnen auch gefallen