Content uploaded by June J Pilcher
Author content
All content in this area was uploaded by June J Pilcher on Oct 07, 2017
Content may be subject to copyright.
Content uploaded by June J Pilcher
Author content
All content in this area was uploaded by June J Pilcher on Oct 07, 2017
Content may be subject to copyright.
ELSEVIER
Journal of Psychosomatic Research,
Vol. 42, No. 6, pp. 583 596. 1997
Copyright © 1997 Elsevier Science Inc.
All rights reserved.
0022-3999/97 $17.00 + .00
S0022-3999(97) 00004-4
SLEEP QUALITY VERSUS SLEEP QUANTITY:
RELATIONSHIPS BETWEEN SLEEP AND
MEASURES OF HEALTH, WELL-BEING AND
SLEEPINESS IN COLLEGE STUDENTS
JUNE J. PILCHER, DOUGLAS R. GINTER and BRIGITTE SADOWSKY
(Received
29
April
1996;
accepted
12
December
1996)
Abstract--Two studies assessed whether measures of health, well-being, and sleepiness are better re-
lated to sleep quality or sleep quantity. In both studies, subjects completed a 7-day sleep log followed
by a battery of surveys pertaining to health, well-being, and sleepiness. In subjects sleeping an average
of 7 hours a night, average sleep quality was better related to health, affect balance, satisfaction with life,
and feelings of tension, depression, anger, fatigue, and confusion than average sleep quantity. In addi-
tion, average sleep quality was better related to sleepiness than sleep quantity. These results indicate that
health care professionals should focus on sleep quality in addition to sleep quantity in their efforts to
understand the role of sleep in daily life. © 1997 Elsevier Science Inc.
Keywords:
Sleep habits; Health; Well-being; Sleepiness; College students.
INTRODUCTION
For many young people, college provides a degree of personal freedom not pre-
viously experienced. One of the life-style habits that young people frequently alter
upon entering college is sleep. Perhaps due to social and academic demands, many
college students choose an irregular sleeping pattern. This voluntarily irregular
sleeping pattern results in a degree of variability in sleep habits that is not often
available in nonclinical and non-shiftwork populations. Therefore, college students
provide a population in which variable sleep habits can be studied without the direct
influence of more clinical concerns, such as sleep apnea or shiftwork schedules. One
sleep-related area that has received little attention in college students and other
nonclinical populations is the relationship between sleep habits and subjective mea-
sures of health, well-being, and sleepiness.
One method of examining the relationships between sleep and measures of
health, well-being, and sleepiness is to classify sleep into two components, sleep
quantity and sleep quality. Although these components of sleep overlap to some ex-
tent, there is a qualitative difference between them. In addition to the more easy
quantifiable components of sleep such as number of awakenings at night, sleep la-
tency, and sleep duration, sleep quality includes largely subjective indices of sleep,
Department of Psychology, Bradley University, Peoria, Illinois, USA.
Address correspondence to: June J. Pilcher, Ph.D., Department of Psychology, Bradley University,
Peoria, IL 61625, USA. Tel: (309) 677-2590; Fax: (309) 677-2330; E-mail: pilcher@bradley.bradley.edu
583
584 J.J. PILCHER
et al.
such as depth of sleep, how well rested one feels upon awakening, and general satis-
faction with sleep [1, 2]. In an effort to better quantify the quality of sleep, research-
ers have developed subjective sleep quality indexes (e.g., [3, 4]). These question-
naires usually define sleep quality as a composite score of sleep quantity, length of
time to fall asleep, number of awakenings at night, length of time to fall back asleep
after awakening, a feeling of fatigue/restfulness upon awakening in the morning,
and general satisfaction with sleep.
The relationships between health and the two components of sleep, quantity and
quality, have been studied in some detail in clinical populations. For example, sleep
duration has been linked to cardiovascular disease [5-7] and gastrointestinal disor-
ders [8]. In addition, sleep itself has been shown to be a health risk in clinical popu-
lations, perhaps due to the physiological changes that occur during sleep [9]. Simi-
larly, poor sleep quality has been linked to increased health complaints in sleep
disorder (e.g., [10, 11]), clinical disorder (e.g., [12]), and shiftwork (e.g., [13, 14])
populations.
A few population survey studies have examined sleep habits in nonclinical popu-
lations using self-report data on sleep habits and health. Studies concentrating on
sleep quantity report that 7-8 hours of sleep at night is positively associated with
self-report health status and longevity [15, 16]. In contrast, a number of studies [17-
19] could not find significant differences in health between short and long sleepers.
Studies examining sleep quality have found a positive relationship between good
sleep quality and self-report health [20, 21]. However, no study to date has specifi-
cally examined the relationships between both sleep quality and sleep quantity and
self-report health in a nonclinical population.
Documenting physical health, however, may not present a total picture of general
health. For example, the World Health Organization has identified three major
components of health: mental, social, and physical. To better understand this more
general concept of health, one must examine facets of life in addition to physical
health, such as general well-being and mental health. Well-being is commonly iden-
tified by three major components: life satisfaction, negative affect, and positive af-
fect [22, 23]. The relationships between these components of general well-being and
sleep have not been thoroughly investigated in nonclinical populations. There is
some support for a relationship between measures of well-being and good sleep
quality [11, 24]. In contrast, studies focusing on sleep quantity and measures of well-
being have reported mixed results [25-27].
Another area associated with general well-being and health is a feeling of sleepi-
ness. For example, sleepiness increases the likelihood of a person getting into acci-
dents at work or during leisuretime [28]. Although few studies have examined
sleepiness in nonclinical populations, there is some support for a relationship be-
tween less sleep and increased sleepiness [29, 30] as well as poorer sleep quality and
increased sleepiness [20, 21].
Surprisingly few studies have attempted to investigate health, well-being, and
sleepiness simultaneously, and no study to date has looked at both sleep quality and
sleep quantity in relation to all three measures. The intent of the current investiga-
tion was to directly address this issue using self-report sleep logs and surveys on
health, well-being, and sleepiness. To better understand the relationships between
measures of sleep quantity and quality and measures of health, well-being, and
Sleep and health 585
sleepiness, we first examined the extent to which sleep quantity and sleep quality
overlap. Because length of sleep is one component of sleep quality, some correla-
tion would be expected between the two dimensions of sleep; however, the amount
of independence of sleep quality indices from pure sleep quantity measures is im-
portant to ascertain. Psychometrically, a high degree of overlap would make it dif-
ficult to isolate the individual effects of sleep quantity and quality. More practically,
it is important to know the extent to which quantity of sleep affects quality of sleep.
Second, we investigated whether the measures of health, well-being, and sleepiness
were differentially related to either aspect of sleep.
METHOD
Study 1
Subjects.
Volunteers were solicited from an upper division psychology class at a midwestern univer-
sity. The students were offered extra credit points as an incentive to participate. Of the 45 students en-
rolled in the class, 39 (28 females, 11 males) volunteered to take part in the study. Thirty of these volun-
teers (22 females, 8 males) completed the experiment. The mean age of the subjects was 20.9 years
(SD=0.98).
Procedures.
To assess sleep, health, and well-being during a stressful period, we administered the sur-
veys (described below) on the day preceding each subject's last final exam. The surveys were adminis-
tered between 6:00 and 9:00 V.M. in a classroom setting for all subjects and took between 30 and 60 min-
utes to complete. The order of survey presentation was counterbalanced to control for fatigue effects.
For the 7-day period prior to his or her survey date, each subject maintained a daily sleep log. To docu-
ment sleep habits under normal sleeping conditions in college students, we utilized self-report sleep logs.
Self-report estimates of sleep have been shown to be highly correlated with polygraphic measures of
sleep quantity and quality [31, 32]. In addition, self-report data have been shown to provide additional
information about the effects of psychological distress on sleep patterns not provided by quantitative
measures [33]. The subjects made entries in the sleep log each morning after awakening. The sleep log,
modeled after that used by Hawkins and Shaw [34], contained six questions: (1) total amount of time in
bed for longest sleep period of the day (in hours and minutes); (2) total amount of time asleep for major
sleep period (in hours and minutes); (3) rating of daily sleep quality (l=awful to 7=great); (4) time to
bed the previous night; (5) time out of bed in the morning; and (6) length in minutes of napping or dozing
during the day. The sleep log provided two estimates of sleep quantity (questions 1 and 2) and one esti-
mate of daily sleep quality (question 3).
Surveys.
Subjects completed all surveys on the day preceding their last final exam. As a general mea-
sure of sleep quality during the last month, each subject completed the Pittsburgh Sleep Quality Index
(PSQI) [3] as part of the survey battery. The PSQI has been shown to have strong internal validity (a
coefficient alpha of 0.83) and temporal stability (0.85 for an average of 28.2 days). The PSQI contains ten
different questions that relate to normal sleep habits (e.g., hours of actual sleep at night, trouble going to
sleep, overall sleep quality rating). Subjects were directed to make each response accurately reflect the
majority of days and nights during the last month.
Current level of sleepiness at the time of survey administration was measured using two self-report
scales. Self-report measures are frequently used to assess level of sleepiness both inside and outside of
laboratory settings and have been shown to relate to severity of sleep disturbance, previous time awake,
and performance [35-37]. The Stanford Sleepiness Scale (SSS) [35] required the subjects to rate their
current sleepiness on a seven-point scale (l=alert to 7=almost asleep). The Epworth Sleepiness Scale
(ESS) [36] instructed the subjects to rank their current chance of falling asleep when imagining them-
selves in eight different situations (e.g., watching TV, sitting and reading) on a four-point scale (0 = never
to 3=high chance).
The three major components of subjective well-being (positive affect balance, negative affect balance,
and general satisfaction with life) were assessed using three independent scales. The Satisfaction with
Life Scale (SWLS) [38] is commonly used as an indication of general well-being. The SWLS contains five
statements regarding subjective opinions of life (e.g., the conditions of my life are excellent). Using a
seven-point scale, the subjects had to agree (7) or disagree (1) with each statement. The SWLS has con-
sistently shown strong internal reliability (a coefficient alpha of 0.87) and moderate temporal stability
(0.54 for 4 years) [39]. To measure affect balance, we used the Bradburn Affect Balance Scale (ABS)
[40]. The ABS consists of ten yes/no questions, five which indicate a positive affect and five a negative
affect (e.g., Did you feel pleased about having accomplished something?). Subjects were told to consider
the past few weeks of their lives when responding to the questions. The ABS has been shown to have
586 J. J. PILCHER
et al.
Table I.--Summary of scales used in survey battery
Scales Minimum score Maximum score and meaning
Pittsburgh Sleep Quality Index 0
Stanford Sleepiness Scale 1
Epworth Sleepiness Scale 0
Cornell Medical Index (both subscales) (I
Bradburn Affect Balance Scale 0
Satisfaction with Life Scale 5
Profile of Mood States
Tension/anxiety 0
Depression/dejection 0
Anger/hostility 0
Vigor 0
Fatigue 0
Confusion/bewilderment 0
21 Severe sleep disturbance
7 Extremely sleepy
24 Extremely sleepy
100% More health complaints
9 Greater positive affect
35 Greater satisfaction
36 Greater tension
60 Greater depression
48 Greater anger
32 Greater vigor
28 Greater fatigue
28 Greater confusion
strong internal validity (a coefficient alpha of 0.76) and moderate temporal stability (0.50 for four tests
at least 2 months apart) [40]. For the purposes of this study, we used the five positive affect questions
and four of the negative affect questions. 1 In addition to using the ABS as a measure of positive and neg-
ative affect, we used the Profile of Mood States (POMS; Educational and Industrial Testing Service, San
Diego, CA) to measure general mood as a component of well-being. The POMS supplies a list of words
related to six mood states: tension/anxiety, depression/dejection, anger/hostility, vigor, fatigue, and con-
fusion/bewilderment. The subjects ranked each word on a five-point scale from "not at all" (0) to "ex-
tremely" (4) according to how they had felt during the past week.
To assess subjective feelings of health, we included The Cornell Medical Index (CMI) [41] in our bat-
tery of surveys. The CMI has been widely used as an aid in medical-history taking and significant correla-
tions (0.52 for women and 0.57 for men) have been obtained between physicians' rating and self-rating
of psychological and physical health [42]. The CMI poses yes/no questions regarding physical and psy-
chological health. We simplified the CMI for administration to college students by eliminating questions
that pertain to only one gender and by deleting questions that apply specifically to aging? The questions
in the CMI are grouped into clusters A through R with each cluster representing a different type of
health complaint. Clusters A through K represent physical health complaints, such as complaints about
eyes and ears and digestive processes. Clusters L through R represent health concerns related to psycho-
logical well-being, such as excessive nervousness and loneliness. All scales administered as part of the
survey battery are summarized in Table I.
Data analyses.
To consolidate the data from the sleep logs, we averaged each subject's response
across the 7 days for each of the six questions. The remaining surveys were scored according to the direc-
tions given for each scale. We calculated one PSQI score, one SSS score, one ESS score, one SWLS
score, one ABS score (total affect), and six POMS scores (tension/anxiety, depression/dejection, anger/
hostility, vigor, fatigue, and confusion/bewilderment). In addition, we created two health variables as
recommended in the scoring instructions for the CMI. We calculated the percentage of yes responses to
questions in clusters L through R to determine a single score for psychological health complaints. To
determine a total physical health complaints score, we computed the percentage of yes responses to
questions in clusters A through K. Higher numbers on all scales represent a greater frequency of the re-
lated occurrence (see Table I).
All statistical analyses were completed on SAS (SAS Institute Inc., CarT, NC). First, we determined
the degree of overlap between sleep quantity and sleep quality by conducting a correlational analysis
between these variables, resulting in four correlation coefficients. Second, we completed a correlational
analysis of the measures of sleep quantity and sleep quality with the measures of health, well-being, and
sleepiness. The correlational analyses between the measures of sleep and measures of health, well-being,
and sleepiness resulted in a total of 48 correlation coefficients. Third, to control for the effects of the
covariation between sleep quality and sleep quantity on the relationships between sleep quality and mea-
sures of health, well-being, and sleepiness, we completed a partial correlational analysis of sleep quality
after removing the variance due to both estimates of sleep quantity (time in bed and time asleep).
We chose to eliminate question 8, "depressed or very unhappy," from the ABS, due to a typographi-
cal error which resulted in responses that could represent either positive or negative affect.
ZThe exact questions used in our modified version of the CMI are available upon request.
Sleep and health
Table II.--Average sleep habits
587
Study 1
Sleep log question M SD
Study 2
M SD
1. Time in bed 7 hr 17 min 1.01
2. Time asleep 6 hr 41 min 1.07
3. Daily quality rating a 4.87 0.88
4. Time into bed 2:05 A.M. 1.57
5. Time out of bed 9:23 A.M. 1.50
6. Time napping (min) 27.27 23.55
7 hr 42 min 1.02
7 hr 4 min 1.11
4.99 0.97
1:45 A.M. 1.18
9:30 A.M. 1.02
17.71 21.61
" Daily sleep quality rating scale: 1 = awful to 7 = great.
Study 2
To better document the relationships between sleep quality and measures of health, well-being, and
sleepiness, we replicated study 1 with a larger group of subjects at a less stressful time of the semester.
Subjects. Volunteers were solicited from two general introductory psychology courses. The students
were offered extra credit points as an incentive to participate. None of the subjects from study 1 were
permitted to participate in study 2. Of the 279 students enrolled in both courses, 99 (69 females, 30
males) volunteered to take part in the study. Of these volunteers, 87 (62 females, 25 males) completed
the study. The mean age of these subjects was 18.9 (so=l.1).
Procedures. The procedures used for the second study were similar to the first study, with the excep-
tion that the sleep logs were kept during the third week of the semester and the surveys were adminis-
tered on the fourth Wednesday of the semester. As in study 1, the subjects maintained the sleep log for
a 7-day period prior to the survey date. All subjects completed the surveys between 7:00 and 9:00 P.M.
in a similar classroom setting to that used in study 1. The order of the survey presentation was counter-
balanced to control for fatigue effects.
Measures. The sleep log and battery of surveys used in study 2 were identical to those used in study l.
Data analysis. For comparison purposes, the data from study 2 were initially analyzed exactly as the
data were analyzed in study 1. The same measures of sleep quantity, sleep quality, health, well-being,
and sleepiness were calculated and similar correlational analyses were conducted. In addition to the cor-
relation and partial correlation analyses completed in study 1, we tested whether the relationships be-
tween sleep quantity and measures of health, well-being, and sleepiness were significantly different from
the relationships between sleep quality and measures of health, well-being, and sleepiness. Using Fish-
er's r-to-z conversion [43], we first normalized all correlations and then conducted a chi-square analysis
to test for significant differences between the correlations for sleep quantity (time in bed, time asleep)
and sleep quality (average sleep quality, daily sleep quality) for each of the health, well-being, and sleepi-
ness variables.
In sum, study 2 was a replication of study 1 with three differences: (1) study 2 had more subjects; (2)
study 2 took place early in the semester, specifically avoiding the more stressful time around final exams:
and (3) the subjects in study 2 maintained their sleep logs for the same 7-day period.
RESULTS
Study 1
The sleep habits of the 30 subjects in study 1 are summarized in Table II. On av-
erage, the subjects were in bed for slightly more than 7 hours a night and reported
taking less than 30 minutes to go to sleep. Subjects usually had relatively late bed-
times and late rising times. In addition to their major sleep period of the day, sub-
jects reported an average of 30 minutes of napping each day.
In general, the correlation between the measures of sleep quantity and sleep
quality were relatively small. Amount of time in bed correlated -0.02 with daily
sleep quality rating and -0.17 with the PSQI. Estimated time asleep correlated 0.08
with daily sleep quality rating and -0.31 with the PSQI.
The results from the analyses of the relationships between measures of sleep
588 J.J. PILCHER
et al.
quantity and quality and measures of health, well-being, and sleepiness are summa-
rized in the first two columns of Table III. In general, health and well-being mea-
sures were better related to sleep quality than sleep quantity. Poor sleep quality, as
measured by the PSQI, was significantly correlated with increased physical health
complaints, and with increased feelings of tension, depression, anger, fatigue, and
confusion. Similarly, poor sleep quality, as measured by the daily ratings was sig-
nificantly correlated with increased physical health complaints and elevated tension,
depression, fatigue, and confusion. On the other hand, sleep quantity as measured
by average time in bed or average time asleep was not significantly correlated with
any measure of health or well-being. Increased sleepiness as measured by the SSS
was equally related to a decrease in sleep quality and quantity, whereas increased
sleepiness as measured by the ESS was not related to either sleep quality or quan-
tity. The third column of Table III shows the partial correlation of sleep quality ac-
counting for the covariance due to sleep quantity. The partial correlations for sleep
quality are virtually identical to the normal Pearson correlations for sleep quality
(middle column Table III) indicating that the relationships between sleep quality
and measures of health, well-being, and sleepiness are independent of any effect by
sleep quantity.
Study 2
As shown in Table II, the sleeping habits of the subjects in study 2 were very simi-
lar to those of study 1. Subjects in study 2 were in bed for approximately 7 hours
45 minutes each night and estimated sleeping for about 7 hours each night. Their
bedtimes were relatively late, as were their rising times, and they reported napping
an average of 18 minutes each day.
As in study 1, we calculated correlation coefficients between each sleep quantity
and quality measure as an estimate of overlap. Similar to study 1, the correlations
between sleep quantity as measured by time in bed and sleep quality were relatively
small. Amount of time in bed correlated 0.15 with daily sleep quality rating and
-0.08 with the PSQI. Estimated time asleep was moderately correlated with the
sleep quality measures (daily sleep quality rating: 0.34; PSQI: -0.27).
The results of the correlational analyses between measures of sleep quantity and
quality and measures of health, well-being, and sleepiness are summarized in the
first two columns of Table IV. As in study 1, health and well-being were clearly bet-
ter related to sleep quality than sleep quantity. Poor sleep quality as measured by
the PSQI was significantly correlated with increased psychological and physical
health complaints, and with many measures of well-being, including a more nega-
tive affect; less satisfaction with life; and increased feelings of tension, depression,
anger, fatigue, and confusion. Similarly, poor sleep quality as measured by the daily
ratings was significantly correlated with increased psychological and physical health
complaints and with many measures of well-being, including a more negative affect,
less satisfaction with life, and more tension, depression, anger, vigor, fatigue, and
confusion. On the other hand, sleep quantity as measured by either the average
time in bed or the average time asleep was not significantly correlated with either
measure of health and with only two measures of well-being. As estimated time in
bed and time asleep decreased, feelings of fatigue and confusion increased. In-
creased sleepiness as measured by the SSS was significantly correlated with a de-
Sleep and health 589
.&
.=,
,.c:
P~
c~
©
1..
0
(D
I
©
II
II
II
~ o c5 c5
I I
III III
I
Iii III
II
oo~ ...~ . .
I II I Illlll
I II I
IIIII
o'~ ~-~
~'~,
=
~ ~.~_,~, .~
~ ~.
o'~ ~'&
~V
.,=.=,o
~.~V
590 J.J. PILCHER
et al.
=
!
~8
,9
0
84
£
II III IIII
II
I
6~W~ q~qq~qS
III I11 rill
I~
I
II III IIII
I 1 I I
IIII
~J~ rJ ~J
"-o "~ o c~
Cj C~ C~ ~.~ Q
~..~ ~
~ "~ "~ 0~
~. = ~
~.~c~
,...:
v
~V
'-~ "~ *
EE*
Sleep and health 591
Table V.--Study 2: Chi-square analysis between sleep quantity and sleep quality measures
Sleep quality measure
Time in bed Time asleep
PSQI Daily rating PSQI Daily rating
Cornell Medical Index
Psychological health complaints 9.96** 2.96 15.16"** 0.97
Physical health complaints 10.11"* 6.59* 15.73"** 3.18
Bradburn Affect Balance Sheet 1.35 8.57** 1.22 8.93**
Satisfaction with Life Scale 1.47 2.40 3.53 0.78
Profile of Mood States
Tension/anxiety 2.72 2.20 3.57 1.54
Depression/dej ection 4.77 * * 5.35 * 7.75 * * 3.07
Anger/hostility 8.86** 4.16" 16.16"** 0.99
Vigor 1.60 1.37 1.18 1.82
Fatigue 11.05"** 0.07 11.05"** 0.07
Confusion/bewilderment 20.03*** 0.37 25.89*** 0.00
Stanford Sleepiness Scale 4.01" 1.76 5.45* 1.04
Epworth Sleepiness Scale 9.88** 0.01 9.88** 0.0t
*p < 0.05; **p < 0.01; ***p < 0.001.
crease in both reports of sleep quality but neither measure of sleep quantity. In-
creased sleepiness as measured by the ESS was significantly correlated with poor
sleep as measured by the PSQI but not the daily sleep quality ratings or either mea-
sure of sleep quantity. The partial correlations for sleep quality accounting for the
covariance due to sleep quantity are given in the third column of Table IV. As in
the first study, the partial correlations for sleep quality are virtually identical with
the normal Pearson correlations for sleep quality (middle column Table IV). This
indicates that the relationships observed between sleep quality and measures of
health, well-being, and sleepiness are independent of any effect due to sleep
quantity.
The chi-square analysis indicated that a number of the correlations between sleep
quality and measures of health, well-being, and sleepiness were significantly differ-
ent from the correlations between sleep quantity and measures of health, well-
being, and sleepiness (Table V). More specifically, the correlations between the
PSQI and the measures, CMIX, CMIY, POMD, POMA, POMF, POMC, SSS, and
ESS, were significantly greater than the correlations for either of the sleep quantity
measures. Interestingly, the correlations between daily sleep quality ratings and
measures of health, well-being, and sleepiness were not as different from the corre-
lations pertaining to sleep quantity as those correlations with PSQI were. The corre-
lations between the daily sleep quality ratings and ABS were significantly greater
than the correlations for either of the sleep quantity measures. In addition, the cor-
relations between the daily sleep quality ratings and the measures, CMIY, POMD,
and POMA, were significantly greater than the correlations for sleep quantity as es-
timated by the time in bed.
In sum, the results from study 1 and study 2 were very similar. Both studies sug-
gested that the measures of sleep quality were only marginally related to measures
of sleep quantity. Furthermore, both studies clearly indicate that measures of
health, well-being, and sleepiness are better related to sleep quality than sleep
quantity. In addition, the partial correlations for sleep quality demonstrate that the
592 J.J. PILCHER
et al.
relationships between sleep quality and measures of health, well-being, and sleepi-
ness were independent of any covariation with sleep quantity. Finally, the chi-
square analysis in study 2 showed that the correlations associated with sleep quality,
especially the PSQI, were significantly greater than the correlations associated with
the sleep quantity measures.
DISCUSSION
The current results suggest that sleep quality is better related to measures of
health, well-being, and sleepiness than sleep quantity in a nonclinical population re-
porting an average of 7-8 hours of sleep at night. Specifically, in two separate groups
of college students, one during final exam week (study 1) and one during the first
third of the semester (study 2), poor sleep quality was correlated with increased
physical health complaints, as measured by the Cornell Medical Index, and to in-
creased feelings of anxiety, depression, anger, fatigue, and confusion, as measured
by the Profile of Mood States (POMS). Furthermore, poor sleepers in study 2 re-
ported a decrease in positive affect and a decrease in satisfaction with life. In addi-
tion, poor sleepers in both studies reported increased levels of sleepiness, especially
as measured by the Stanford Sleepiness Scale (SSS). Last, the results from the chi-
square analysis indicated that, of the two sleep quality measures, the Pittsburgh
Sleep Quality Index (PSQI) and daily sleep quality, the relationships between the
PSQI and measures of health, well-being, and sleepiness were more significantly dif-
ferent from the sleep quantity relationships.
To better understand and interpret the relationships between sleep and measures
of health, well-being, and sleepiness, we investigated the degree to which the two
aspects of sleep, sleep quantity and sleep quality, overlap. Because sleep quantity
is one component of sleep quality, it was expected that the two would correlate to
some extent. The current data indicate that estimated time asleep, but not estimated
time in bed, was moderately correlated with the sleep quality measures, especially
the PSQI. However, this relatively small overlap did not directly affect the current
conclusions because the partial correlations (removing the variation due to both
measures of sleep quantity) between sleep quality and measures of health, well-
being, and sleepiness were virtually identical to the standard Pearson correlations on
sleep quality. Thus, the components of sleep quality other than simply sleep quan-
tity (e.g., number of awakenings at night, general satisfaction with sleep) appear to
be largely responsible for the relationships between sleep quality and measures of
health, well-being, and sleepiness. It must be noted, however, that these relation-
ships exist in nonclinical populations where the subjects report 7-8 hours of sleep
at night and that sleep duration outside of the 7-8-hour range may have a substan-
tial effect on health, well-being, and sleepiness not seen in the current study [44, 45].
The use of correlational data allow us to estimate the amount of variance in
health, well-being, and sleepiness accounted for by sleep quantity and sleep quality.
When examining the standard Pearson correlations in the two studies, it can be seen
that the proportion of variance explained ranges from 0.0 to 0.372. Whereas the
amount of variance explained is not large, it is consistent with other studies looking
at sleep, health, and well-being across a variety of populations (e.g., [5, 24, 27, 35,
36]). When the r 2 values are examined more closely, it can be seen that sleep quality
Sleep and health 593
accounts for more of the variance in health measures (study 1: M=0.20--0.11; study
2: M=0.14+0.02) and well-being measures (study 1: M=0.17--+0.10; study 2:
u=0.10+--0.05) than sleep quantity (health measures: study 1: U=0.01--0.01; study
2:U=0.01+0.02 and well-being measures: study 1: U=0.01+---0.01; study 2: M=
0.03-----0.04). However, sleep quality (study 1: M=0.07--0.07; study 2: M=0.05--0.02)
and sleep quantity (study 1: M=0.07--0.08; study 2: U=0.02+--0.01) account for ap-
proximately equal amounts of variance in sleepiness. While these data indicate that
sleep is only one of several life-style factors contributing to the overall health, well-
being, and sleepiness of the individual, sleep quality clearly accounts for more of the
variance in both health and well-being as measured by the surveys used in the cur-
rent study than sleep quantity.
One relevant point to consider is the external conditions present during study 1
and study 2. Study 1 was conducted in December, during final exam week, a pre-
sumably stressful time for college students, whereas study 2 was completed in Feb-
ruary, during the fifth week of the semester, a presumably less stressful time for col-
lege students. In spite of the difference in time of the semester for the sleep log and
survey administration, little difference in the pattern of relationships between mea-
sures of sleep and measures of health, well-being, and sleepiness was found. It is fea-
sible that the stability of the relationships found in the current set of studies may
be more dependent upon seasonal effects than on other external stimuli, such as
perceived stress due to final exams. However, the data could also indicate that the
relationships are relatively robust and are not easily altered by external conditions.
One notable finding was the lack of consistency in results between the Epworth
Sleepiness Scale (ESS) and the SSS. Our data indicated that the SSS was better re-
lated to sleep measures, particularly sleep quality, than the ESS. This difference
may be due to the manner in which the scales assess sleepiness. By asking the sub-
jects to project their sleepiness to a different setting, the ESS may be altering the
manner in which the subjects report sleepiness in comparison to the SSS. If both
scales were measuring sleepiness in the same manner, one would expect the two
measures to be highly correlated. However, the correlations between the ESS and
the SSS were relatively low. In study 1, the ESS correlated 0.15 with the SSS, and
in study 2 the ESS correlated 0.29 with the SSS. Therefore, it is likely that the two
sleepiness scales are measuring sleepiness in different manners and that the method
used by the SSS is more meaningful in relation to subjective sleep quality measures.
There are several experimental limitations to be considered when drawing con-
clusions from the current studies. Certainly, any self-report data must be cautiously
interpreted. However, there is substantial evidence that self-report estimates of
sleep are highly correlated with polygraphic measures of sleep quantity and quality
[31, 32] and as such provide a meaningful data set from which to draw conclusions.
In addition, self-report measures on health, well-being, and sleepiness are fre-
quently used and provide meaningful data based on subjects' self-perceptions of
these variables (e.g., [46-48]). Another concern with the current design is the num-
ber of actual correlations computed. It is possible that some of the correlations
would be significant between measures of sleep and measures of health, well-being,
and sleepiness simply by chance. However, with the current data, this concern is less
valid. If the correlations reported were due mostly to chance, one would expect the
correlations to be more evenly distributed between the sleep quantity and sleep
594 J.J. PILCHER
et al.
quality categories. Because the vast majority of significant correlations occurred
with the sleep quality variables, this particular concern does not meaningfully affect
the conclusions from the current set of studies. In addition, the partial correlations
on sleep quality indicate that the relationships between sleep quality and measures
of health, well-being, and sleepiness are independent of any covariance with sleep
quantity, thus providing additional support for the conclusion that health, well-
being, and sleepiness are better related to sleep quality than sleep quantity.
In sum, our results lend further support to the importance of sleep and its rela-
tionships to health, well-being, and sleepiness in nonclinical populations sleeping an
average of 7-8 hours a night. Few studies prior to ours have used a comparably ex-
tensive set of measures of health, well-being, and sleepiness with nonclinical popu-
lations as those used here, and those studies that did [20, 21] yielded similar results.
Unlike these two earlier studies, our research specifically compared sleep quantity
to sleep quality. The current data indicate a predictable pattern of correlations be-
tween sleep quality and measures of health, well-being, and sleepiness. However, a
similar pattern was not found between these measures and sleep quantity. In addi-
tion, the relationships between measures of health, well-being, and sleepiness and
sleep quality were significantly greater than many of the relationships between mea-
sures of health, well-being, and sleepiness and sleep quantity. In addition, we found
that the relationships between sleep quality and measures of health, well-being, and
sleepiness are independent of the effect of sleep quantity on sleep quality. The cur-
rent results demonstrate that future research on sleep and preventative medicine in
nonclinical populations should focus on sleep quality in addition to sleep quantity
to better understand the role of sleep in daily health, well-being, and sleepiness.
Acknowledgments--The
authors thank Elizabeth S. Ott for her assistance in the initial data analysis in
both study 1 and study 2 and Eric E. Faulkner and Cheng Her for their assistance with the initial data
analysis in study 2. The authors also thank Allen I. Huffcutt for his helpful comments on this manuscript
during preparation. This research was supported by a Bradley University Research Excellence Commit-
tee Award to June J. Pilcher.
REFERENCES
1. Frankel BL, Coursey RD, Buchbinder R, Snyder F. Recorded and reported sleep in chronic primary
insomnia. Arch Gen Psychiatry 1976;33:615-623.
2. Parrott AC, Hindmarch I. Factor analysis of a sleep evaluation questionnaire. Psychol Med
1978;8:325-329.
3. Buysse D J, Reynolds CF III, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality In-
dex: a new instrument for psychiatric practice and research. Psychiatry Res 1989;28:193-213.
4. Urponen H, Partinen M, Vuori I. Hasan J. Sleep quality and health: description of the sleep quality
index. In: Peter JH, Penzel T, Podszus T, von Wichert P, eds. Sleep and health risk. Berlin:
Springer 1991:555-558.
5. Partinen M, Putkonen PTS, Kaprio J, Koskenvuo M, Hilakiv I. Sleep disorders in relation to coro-
nary heart disease. Acta Med Scand 1982;660(suppl):123-136.
6. Appels A, Mulder P. Imminent myocardial infarction. J Human Stress 1984;10:129-134.
7. Jenkins CD. The epidemiology of sudden cardiac death: incidence, clinical features, biomedical and
psychosocial risk factors. In: Beamish RE, Singal PK, Dhalla NS, eds. Stress and heart disease. Bos-
ton: Martinus Nijhoff 1985.
8. Elashoff JD, Van Deventer G, Reedy T J, Ippoliti A, Samloff IM, Kurata J, Billings M, Isenberg M.
Long term follow-up of duodenal ulcer patients. J Clin Gastroenterol 1983;5:50%515.
9. Fossey E, Fitzpatrick MF, Douglas NJ, Shapiro CM. Interaction and classification of sleep and health
risk. In Peter JH, Penzel T, Podszus T, von Wichert P, eds. Sleep and Health Risk. Berlin:
Springer 1991:174-179.
Sleep and health 595
10. Healey ES, Kales A, Monroe LJ, Bixler EO, Chamberlin K, Soldatos CR. Onset of insomnia: role
of life-stress events. Psychosom Med 1981;43:439-451.
11. Pailhous E, Benoit O, Goldenberg F, Bouard G, Payant C. Psychological profile and sleep organiza-
tion in young subjects with poor quality of sleep. Psychiatry Res 1988;26:327-336.
12. Hyyppa MT, Kronholm E. Quality of sleep and chronic illness. J Clin Epidemiol 1989;42:633-638.
13. Lavie P, Chillag N, Epstein R, Tzischinsky O, Givon R, Fuchs S, Shahal B. Sleep disturbances in
shift-workers: a marker for maladaptation syndrome. Work Stress 1989;3:33-40.
14. Phillips B, Magan L, Gerhardstein C, Cecil B. Shift work, sleep quality, and worker health: a study
of police officers. South Med J 1991;84:1176-1197.
15. Bellec NB, Breslow L. Relationship of physical health status and health practices. Prevent Med
1972;1:490-421.
16. Frederick T, Frederichs RR, Clark VA. Personal health habits and symptoms of depression at the
community level. Prevent Med 1988;17:173-182.
17. Johns MW, Bruce DW, Masterton JP. Psychological correlates of sleep habits reported by healthy
young adults. Br J Med Psychol 1974;47:181-187.
18. Webb WB. Are short and long sleepers different? Psychol Rep 1979;44:259-264.
19. Spinweber CL, Johnson LC, Chin LA. Disqualified and qualified poor sleepers: subjective and objec-
tive variables. Health Psychol 1985;4:569-578.
20. Lugaresi E, Cirignotta F, Zucconi M, Mondini S, Lenzi PL, Coccagna G. Good and poor sleepers:
an epidemiological survey of the San Marino population. In: Guilleminault C, Lugaresi E, eds. Sleep/
wake disorders: natural history, epidemiology and long-term evolution. New York: Raven Press
1983:13-28.
21. Hyyppa MT, Kronholm E, Mattlar CE. Mental well-being of good sleepers in a random population
sample. Br J Med Psychol 1991;64:25-34.
22. Diener E. Subjective well-being. Psychol Bull 1984;95:542-575.
23. Chamberlain K. On the structure of subjective well-being. Soc Indicat Res 1988;20:581-6(14.
24. Weller L, Avinir O. Hassles, uplifts, and quality of sleep. Percept Motor Skills 1993;76:571-576.
25. Hartmann E, Baekeland F, Zwilling G. Psychological differences between long and short sleepers.
Arch Gen Psychiatry 1972;26:463-468.
26. Stuss D, Broughton R. Extreme short sleep: personality profiles and a case study of sleep require-
ments. Waking Sleeping 1978;2:101-105.
27. Wetzler HP, Ursano RJ. The positive association between physical health practices and psychologi-
cal well-being. J Nerv Ment Dis 1988;176:280-283.
28. Leger D. The cost of sleep-related accidents: a report for the national commission on sleep disorders
research. Sleep 1994;17:84-93.
29. Friedman RC, Bigger JT, Kornfeld DS. The intern and sleep loss. New Engl J Med 1977:285:
201-203.
30. Johnson LC, MacLeod WL. Sleep and awake behavior during gradual sleep reduction. Percept Mo-
tor Skills 1973;36:87-97.
31. Johnson LC, Spinweber CL. Quality of sleep and performance in the Navy: a longitudinal study of
good and poor sleepers. In: Guilleminault C, Lugaresi E, eds. Sleep/wake disorders: natural history,
epidemiology. New York: Raven Press 1983.
32. Jacobs EA, Reynolds CF, Kupfer DJ, Lovin PA, Ehrenpreis AB. The role of polysomnography in
the differential diagnosis of chronic insomnia. Am J Psychiatry 1988;145:346-349.
33. Shaver JLF, Giblin E, Paulsen V. Sleep quality subtypes in midlife women. Sleep 1991;14:18-23.
34. Hawkins J, Shaw P. Self-reported sleep quality in college students: A repeated measures approach.
Sleep 1992;15:545-549.
35. Hoddes E, Zarcone V, Smythe H, Phillips R. Dement WC. Quantification of sleepiness: a new ap-
proach. Psychophysiology 1973;10:431-436.
36. Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale.
Sleep 1991;14:540-545.
37. Gillberg M, Kecklund G, Akerstedt T. Relations between performance and subjective ratings of
sleepiness during a night awake. Sleep 1994;17:236-241.
38. Diener E, Emmons RA, Larsen RJ, Griffen S. The satisfaction with life scale. J Pers Assess
1985;49:71-75.
39. Pavot W. Diener E. Review of the satisfaction with life scale. Psychol Assess 1993;5:164-172.
40. Bradburn NM. The structure of psychological well-being. Chicago: Aldine 1969.
41. Brodman K, Erdmann AJ Jr, Wolff HG, Miskovitz PF, Cornell Medical Index health questionnaire
manual. New York: Cornell University Medical College, 1986.
42. Bowling A. Measuring health: a review of quality of life measurement scales. Philadelphia: Open
University Press 1991.
43. Cohen J, Cohen P. Applied multiple regression/correlation analysis for the behavioral sciences.
Hillsdale, New Jersey: Lawrence Erlbaum 1983.
596
J.J. PILCHER
et aL
44. Kripke DF, Simons RN, Garfinkel L, Hammond EC. Short and long sleep and sleeping pills: Is in-
creased mortality associated? Arch Gen Psychiatry 1979;36:103-116.
45. Wingard DL, Berkman LF. Mortality risk associated with sleeping patterns among adults. Sleep
1983;6:102-107.
46. Alfredsson L, Akerstedt T, Mattsson M, Wilborg B. Self-reported health and well-being amongst
night security guards: a comparison with the working population. Ergonomics 1991;34:525-530.
47. Gall R, Isaac L, Kryger M. Quality of life in mild obstructive sleep apnea. Sleep 1993;16(suppl):
$59-$61.
48. Hill DW, Welch JE, Godfrey JA III. Influence of locus of control on mood state disturbance after
short-term sleep deprivation. Sleep 1996;19:41-46.
