Journal of Adolescent Health 46 (2010) 124–132

Original article

Sleep Patterns and Predictors of Disturbed Sleep in a Large Population

of College Students
Hannah G. Lund, B.A.a, Brian D. Reider, B.A.b, Annie B. Whiting, R.N.c,
and J. Roxanne Prichard, Ph.D.b,*
a
Department of Psychology, Virginia Commonwealth University, Richmond, Virginia
b
Department of Psychology, University of St. Thomas, St. Paul, Minnesota
c
Massachusetts General Hospital, Boston, Massachusetts
Manuscript received March 26, 2009; manuscript accepted June 16, 2009

See Editorial p. 97

Abstract Purpose: To characterize sleep patterns and predictors of poor sleep quality in a large population of
college students. This study extends the 2006 National Sleep Foundation examination of sleep in early
adolescence by examining sleep in older adolescents.
Method: One thousand one hundred twenty-five students aged 17 to 24 years from an urban
Midwestern university completed a cross-sectional online survey about sleep habits that included
the Pittsburgh Sleep Quality Index (PSQI), the Epworth Sleepiness Scale, the Horne-Ostberg Morn-
ingness–Eveningness Scale, the Profile of Mood States, the Subjective Units of Distress Scale, and
questions about academic performance, physical health, and psychoactive drug use.
Results: Students reported disturbed sleep; over 60% were categorized as poor-quality sleepers by
the PSQI, bedtimes and risetimes were delayed during weekends, and students reported frequently
taking prescription, over the counter, and recreational psychoactive drugs to alter sleep/wakefulness.
Students classified as poor-quality sleepers reported significantly more problems with physical and
psychological health than did good-quality sleepers. Students overwhelmingly stated that emotional
and academic stress negatively impacted sleep. Multiple regression analyses revealed that tension
and stress accounted for 24% of the variance in the PSQI score, whereas exercise, alcohol and caffeine
consumption, and consistency of sleep schedule were not significant predictors of sleep quality.
Conclusions: These results demonstrate that insufficient sleep and irregular sleep–wake patterns,
which have been extensively documented in younger adolescents, are also present at alarming levels
in the college student population. Given the close relationships between sleep quality and physical and
mental health, intervention programs for sleep disturbance in this population should be considered.
Ó 2010 Society for Adolescent Medicine. All rights reserved.
Keywords: Sleep quality; Sleep disturbance; Adolescence; Stress; Mood; College students

Little doubt exists among health professionals about the with an increased risk of work absenteeism and accidents [1], as
fundamental importance of sufficient, restorative sleep in main- well as significant decrements in vitality, social functioning,
taining one’s physical and mental health. Troubled sleep is physical and mental health, and general quality of life [1–4].
considered both a predictive sign and symptom of many Sleep in younger adolescents (ages 12–17) has been exten-
illnesses, and is associated with substantial decrements in the sively documented. Because of a multitude of intrinsic and
quality of life. Briefly, chronic sleep disturbances are associated environmental factors, younger adolescents are particularly
vulnerable to disturbed sleep, and are one of the most sleep
*Address correspondence to: J. Roxanne Prichard, Ph.D., University of
deprived age groups in the country [5]. First, pubertal adoles-
St. Thomas, 2115 Summit Avenue, JRC LL56, St. Paul, MN 55105. cents experience a biologically based phase change in their
E-mail address: [email protected] circadian rhythm that delays sleep and wake onset, making
1054-139X/10/$ – see front matter Ó 2010 Society for Adolescent Medicine. All rights reserved.
doi:10.1016/j.jadohealth.2009.06.016
 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132 125

it physically harder to maintain earlier bedtimes [6,7]. and the remaining 3% identified as other or elected not to
Second, external factors like increasing caffeine consumption respond. However, there was a female responder bias;
and late night use of electronics further delay sleep onset although males and females were enrolled in equal propor-
[5,8,9]. Third, early start times for middle schools and high tions, females comprised 63% of the survey respondents.
schools demand earlier weekday risetimes [10–12]. Finally,
even with sufficient sleep times, adolescents have increased
Measures
daytime sleepiness and a greater physiological need for sleep
compared to prepubertal children, which may result from The online survey included five published scales related to
maturational changes in neuronal connectivity [13,14]. sleep, mood, and stress: (a) the Pittsburgh Sleep Quality Index
The consequences of this sleep deprivation are severe, (PSQI), (b) the Epworth Sleepiness Scale (ESS), (c) the Horne-
impacting adolescents’ physical and mental health, as well Ostberg Morningness Eveningness Scale (MES), (d) the
as daytime functioning. Population and clinic-based studies Subjective Units of Distress Scale (SUDS), and (e) the Profile
in younger adolescents (ages 11–17) have shown strong asso- of Mood States (POMS). The PSQI differentiates between
ciations between chronic sleep restriction and anxiety, ‘‘poor-’’ and ‘‘good-’’quality sleepers by measuring seven
depression, and somatic pain [9,15–17]. Younger adoles- areas: subjective sleep quality, sleep latency, sleep duration,
cents who report shorter sleep also show decrements in habitual sleep efficiency, sleep disturbances, use of sleep medi-
academic performance [5,18], and increased risk-taking cation, and daytime dysfunction over the past month [23].
behaviors including drug use and drowsy driving [9,19]. A Scoring is based on a 0–3 Likert scale, where a score of 3
12-month prospective study by Roberts et al. [20] demon- reflects the negative extreme. A global score greater than 5 is
strated that insomnia in younger adolescents significantly indicative of a poor-quality sleeper, whereas a score of 5 or
increased the risk for subsequent declines in social, psycho- less is indicative of a good-quality sleeper. For this study,
logical, physical, and mental health. global PSQI scores were split into three categories: optimal
By comparison, fewer studies have examined how sleep (5), borderline (6–7), and poor (8) sleep quality; these cate-
patterns change when older adolescents enter college, a time gories were created using the specified cutoff scores for the
of minimal adult supervision, erratic schedules, and easy purpose of achieving relatively even group sizes. The internal
access to over-the-counter (OTC), prescription, and recrea- consistency of the PSQI, estimated by Cronbach’s alpha, is .73.
tional drugs. Of these publications, most have focused on The Epworth Sleepiness Scale is a questionnaire used to
sleep patterns, fatigue, and academic performance identify excessive sleepiness associated with accumulated
[17,21,22]. Little is known about what factors contribute to sleep debt or clinical sleep disorders [24]. This eight-item scale
or exacerbate sleep difficulties in this population. The current assesses how sleepy one has felt in the past 6 months; partic-
study measures the extent of sleep deprivation and poor- ipants indicate the likelihood that they would fall asleep while
quality sleep in a large population of college students (ages doing certain activities (e.g., watching TV, sitting and talking
17–24), and extends the current literature on adolescent sleep to someone, or stopped at a traffic light), with responses from
by examining factors that are both precipitating and perpetu- (0 ¼ would never doze to 3 ¼ high chance of dozing. Scores
ating of poor sleep in this age group. Using a multibehavioral range from 0 to 24, with scores over 10 indicative of significant
analysis in a nonclinical population, we focused on three main levels of daytime sleepiness. Internal consistency for the ESS,
questions: (a) What are the sleep habits of college students? estimated by Cronbach’s alpha, is .75.
(b) What behavioral outcomes are associated with poor sleep The Horne-Ostberg Morningness Eveningness Scale is used
quality? (c) What physical, emotional, and psychosocial to distinguish between chronotypes (an endogenous character-
factors predict poor-quality sleep in college students? istic describing one’s preference for either morning or evening
patterns of activity) [25]. Scores range from 16 to 86, corre-
sponding to extreme eveningness (lower numbers) to extreme
Method morningness (higher numbers). Questions target individual
preferences for sleep and wake times, etc., such as: ‘‘Consid-
Participants
ering only your own ‘feeling best’ rhythm, at what time would
Participants were students between the ages of 17 and 24 you get up if you were entirely free to plan your own day?’’ In
(N ¼ 1,125; 420 male, 705 female) at a large private univer- response, participants are directed to select a time between the
sity in the Midwest. Of these participants, 27% were freshmen hours of 5:00 a.m. and 12:00 p.m. Internal consistency for the
(N ¼ 305), 27% were sophomores (N ¼ 312), 24% were MES, estimated by Cronbach’s alpha, is 86.
juniors (N ¼ 271), and 20% were seniors (N ¼ 232). The The Subjective Units of Distress Scale is used to assess
mean age was 20 (SD ¼ 1.3 years). Survey respondents an individual’s baseline level of stress [26]. Participants indi-
were representative of the age and ethnic diversity of the cate how stressed they feel on a typical day using a scale of 1
school; 86% of participants were Caucasian (N ¼ 978), 5% to 100, where 1 ¼ lowest possible stress and 100 ¼ highest
were Asian or Pacific Islander (N ¼ 58), 2% were African possible stress.
American (N ¼ 23), 2% were biracial (N ¼ 24), 1% were The POMS is used to assess how severely participants
American Indian or Alaskan Native, 0.4% were Hispanic, experience depression, tension, fatigue, confusion, vigor,
126 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132

and anger [27]. The shortened version includes a list of 30 weekend behaviors. Multivariate analysis of variance was
adjectives that relate to the six different mood states; partic- used to explore differences among optimal-, borderline-,
ipants are asked to rank on a five-point scale from ‘‘not at and poor-quality sleepers on a number of variables, including
all’’ to ‘‘extremely’’ how much they experience these mood mood (POMS), stress (SUDS), and caffeine and alcohol use.
descriptors on a typical day. Internal consistency for the Chi-squared analyses were used to assess differences
POMS, estimated by Cronbach’s alpha, is .79. between optimal-, borderline-, and poor-quality sleepers in
In addition to the five published scales and basic demo- ordinal and nominal variables. Multiple stepwise regression
graphic information, we also included questions relating to analyses were employed to determine predictors of sleep
academic performance, physical health, and psychoactive quality. Variables that have been shown in previous
drug use. To assess academic performance, we asked partic- studies to correlate with sleep quality (including individual
ipants to provide their grade-point average (on a 4.0 scale) components of the POMS, stress [SUDS], morningness/
and information about class attendance. To assess physical eveningness [MES], caffeine and alcohol use, frequency of
health, we asked students about regularity of exercise and exercise, and regularity of sleep scheduling [weekend over-
frequency of missing class because of illness. To assess sleep and bedtime delay]), but that are not themselves
psychoactive drug use, we assessed the average week and components of the PSQI score (e.g., total sleep time, sleep
weekend frequency and intake of caffeine, alcohol, nicotine, latency, pain during sleep), or direct measures of sleepiness
marijuana, and prescription and OTC stimulants and sleep (e.g., Epworth Sleepiness Score or the fatigue component
aids. Alcohol use was measured by total number of drinks of the POMS), were used as independent variables in the
(glass of wine, bottle/can of beer, shot of liquor, etc.) during regression.
the week (Sunday–Thursday) and during the weekend
(Friday–Saturday); caffeine was measured in drinks per
weekday or weekend day (8 oz. serving of coffee, espresso, Results
tea, soft drinks, hot chocolate, or 1.5 oz. of chocolate); nico-
Sleeping behavior: quantity and quality
tine was measured by the number of cigarettes per day, and
marijuana was measured by the number of uses per week, Overall, college students reported chronically restricted
as well as the number of inhalations per use. Students were sleep. Mean total sleep time (time spent actually sleeping,
asked to identify motivations for using particular drugs as opposed to being awake in bed) was 7.02 hours
(e.g., to increase wakefulness, to increase alertness, to be (SD ¼ 1.15). Twenty-five percent of students reported getting
social, to complement meals, to promote sleep, etc.). less than 6.5 hours of sleep a night, and only 29.4% of
students reported getting 8 or more hours of total sleep time
Procedure per night, the average amount required for young adults
Participants were recruited through an e-mail sent to all [28]. Sleep was particularly restricted on weeknights;
full-time undergraduate students (n ¼ 5,401). The first page mean weekday bedtime was 12:17 a.m. (SD ¼ 71 minutes)
of the survey informed participants of the purpose and nature and weekday rise time was 8:02 a.m. (SD ¼ 76 minutes).
of the study, assured them of their anonymity, and asked Sleep schedules were erratic. Mean bedtimes (1:44 a.m.,
participants to provide informed consent by clicking a state- SD ¼ 79 minutes) were delayed and mean rise times (10:08
ment before proceeding to the first data collection page of the a.m., SD ¼ 88 minutes) were extended on weekends. Addi-
survey. As incentives for participation, participants received tionally, 20% of students reported staying up all night at least
either class credit if they were enrolled in select psychology once in the last month, and 35% reported staying up until
courses (the type of credit depended on the class), or were 3 a.m. at least once a week.
entered into a raffle for a chance to win one of four monetary Figure 1 outlines the differences in bedtimes and risetimes
prizes ($25–$150 gift certificates). After the survey was by year, beginning in ninth grade and extending through the
completed, students were directed to a separate Web site to end of college (high school data are from the 2006 National
enter into the raffle. The survey was accessible online for Sleep Foundation Sleep in America Poll) [9]. Both weekday
4 weeks in the middle of the semester. The procedure was bedtimes and risetimes appear to be 75 minutes later in our
approved by the university’s institutional review board. cohort of first year college students when compared to a cohort
Approximately 21% of the University’s undergraduate of seniors in high school. Sleep schedule differed significantly
students completed the survey. Of these respondents, by year in school. First-year students had significantly later
students older than 24 were excluded from the study bedtimes and rise times than juniors and seniors during the
(<0.5% of respondents), as were students with incomplete weekends, F(3,994) ¼ 5.92, 7.06; h2 ¼ .018, .021, respec-
tively, p < .001, but not during the weekdays, leading to
surveys (<10% of respondents).
a more pronounced delay in weekend bedtime and weekend
oversleep among first year students and sophomores,
Analyses
compared to juniors and seniors, F(3,994) ¼ 4.57, 5.41;
t-Tests were used to test for gender differences and paired h2 ¼ .014, .026, respectively, p < .001. Sleep schedule also
t-tests were used to determine differences between week and differed significantly by sex. Males had significantly later
 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132 127

1:58 a week. Over one-third of students cited ‘‘other’’ reasons

1:44
1:44 for sleep disturbances at least once a week; of these responses,
1:27 the most common reasons provided for these disturbances
were stress (35%), excess noise (33%), and cosleeping
(sharing the bed with a partner; 7%).
Students also reported significant sleepiness-related
12:22
12:20 decrements in daytime performance (Table 2). Mean scores
12:45 12:14
Bed Time

12:11
on the Epworth Sleepiness Scale were 6.8 for weekdays
12:25 and 6.7 for weekends; 25% of students scored 10 or above
on the scale, indicating significant levels of daytime sleepi-
12:03
11:53 ness. Seventy-five percent of students reported feeling
‘‘dragged out, tired, or sleepy’’ once a week or more, and
15% reported falling asleep in class once a week or more.
11:02
10:51 week
weekend
10:32
Sleep quality, mood, and health
10:15
HS 9 HS 10 HS 11 HS 12 College College College Jr. College Sr. Poor sleep quality was associated with significantly higher
Fresh. Soph.
Year in School self-reported negative moods. Participants categorized as
having poor-quality sleep (PSQI scores 8) had significantly
10:26 greater negative mood subscale scores (anger, confusion,
10:06 10:09 10:08
9:54 9:52 9:51
depression, fatigue, and tension), compared to those with
9:49
good-quality sleep; for all cases, F(2,897) > 25, p <.001
(Table 3). Poor-quality sleepers also reported higher levels
of stress during the week and weekends, compared to
optimal-quality sleepers, F(2,916) ¼ 72.4; 37.7, respectively,
Rise Time

p < .001 (Table 3).

8:08
7:59
Poor-quality sleepers also reported significantly more
8:03 7:59
physical illness than optimal- and borderline-quality sleepers,
c2(12, n ¼ 947) ¼ 39.9, p < .05. Twelve percent of poor-
quality sleepers reported missing class three times or more
in the last month because of illness, compared to less than
week
6:28 6:23 6:23
6:31
weekend 4% of borderline- or good-quality sleepers. Higher scores
on the PSQI were also associated with significantly increased
instances of falling asleep in class and skipping class for
HS 9 HS 10 HS 11 HS 12 College College College Jr. College Sr. reasons other than illness (Table 3).
Fresh. Soph.
Sleep quality was also related to the use of prescription,
Year in School
OTC, and recreational drugs to help regulate sleep and wake-
Figure 1. Bedtimes and rise times by year in school. Data from the high fulness (Table 3). Specifically, those with poor sleep quality
school bins are taken from the 2006 Sleep in America Poll (n ¼ 1,602) [9]. were more than twice as likely to report using OTC or
prescription stimulant medications at least once a month to
help keep them awake, compared to those with good sleep
bedtimes and risetimes during the week than did females, quality. However, the number of caffeinated drinks per day
t(998) ¼ 5.34, 2.18, respectively, p < .001, but not during did not significantly differ between PSQI groups. Likewise,
the weekends. more than 33% of students who scored equal to or greater
In addition to low sleep quantity, students also exhibited than 8 on the PSQI used prescription or OTC sleep aids at
poor sleep quality. Table 1 shows responses to individual least once a month, compared to less than 5% of optimal-
questions on the PSQI. Only 34.1% of students scored in quality sleepers. Finally, poor-quality sleepers reported
the ‘‘good’’ range of the PSQI (5), and 38% had PSQI scores drinking more alcohol per day than optimal-quality sleepers,
over 7, indicating poor-quality sleep. The primary contribu- and were twice as likely to report using alcohol to induce
tors to these high PSQI scores were restricted total sleep sleep compared to those with PSQI scores less than 8 (Table
time, low enthusiasm, long sleep latencies, and ‘‘other’’ 3). Furthermore, of poor-quality sleepers, those who said
factors regularly interfering with sleep (Question 5j). Specif- they used alcohol to induce sleep drank significantly more
ically, 52% of students reported lacking enthusiasm to get alcoholic beverages per week, compared to those drinkers
things done at least once a week, and 32% reported an who did not use alcohol to sleep: 12 versus 21,
inability to fall asleep within 30 minutes at least once t(282) ¼ 2.43, p ¼ .02.
128 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132

Table 1
Prevalence of sleep disturbances as measured by the PSQI
Pittsburgh Sleep Quality Index Bedtime Sleep latency Risetime Total sleep time
Mean, SD 12:21 a.m. 74 min 23.8 min, 19.2 min 8:05 a.m.,76 min 7.02 hours, 1.15 hours
How often have you had trouble sleeping Not during the past month Less than once a week Once or twice a week 3 or more times a week
because.
Cannot get to sleep within 30 minutes 26.2% 41.4% 21.5% 10.9%
Wake up in middle of night or early morning 44.2% 34.2% 16.0% 5.6%
Wake up to use the bathroom 56.1% 32.7% 7.4% 3.8%
Cough or snore loudly 87.5% 9.6% 2.1% 0.7%
Cannot breathe comfortably 80.6% 13.3% 4.7% 1.4%
Feel too cold 71.0% 23.4% 4.5% 1.1%
Feel too hot 33.5% 42.8% 19.3% 4.5%
Have bad dreams 70.2% 21.7% 6.5% 1.6%
Have pain 73.1% 19.2% 5.4% 2.3%
Other reasons 49.4% 17.3% 22.0% 11.0%
How often have you.
Taken medicine to aid in sleep? 82.1% 11.0% 4.0% 2.9%
Had trouble staying awake during social 75.7% 20.8% 3.3% 20.0%
activities?
Had a problem getting the enthusiasm to get 19.8% 30.1% 33.0% 17.2%
things done?
Rate overall sleep. Very good Fairly good Fairly bad Very bad
11.0% 55.0% 30.0% 3.9%
Global PSQI Optimal(1–5) Borderline(6–7) Poor(8)
34.1% 27.7% 38.2%
PSQI ¼ Pittsburgh Sleep Quality Index.

Predictors of sleep quality tional 3%, and morningness/eveningness (MES) accounting

for another 2% of the variance (Table 4). Alcohol per day,
Several lines of evidence point to stress as a major contrib- caffeine per day, exercise frequency, and daily hours of tele-
utor to poor sleep quality in college students. First, 20.1% of vision and video game exposure were not significant predic-
students reported stress interfering with sleep at least once tors of the PSQI score.
a week. Women were significantly more likely to report
stress-related sleep troubles than men; t(927) ¼ 5.49, p <.001.
Second, when asked to provide a written answer to PSQI ques- Discussion
tion 5j, ‘‘How often have you had trouble sleeping because of Overall, the results demonstrate that the epidemic of insuf-
other reason(s); please describe reasons,’’ the most common ficient sleep documented in high school students extends past
answers were related to stress. Answers such ‘‘stress about early and midadolescence to college students. Total sleep
school,’’ ‘‘racing thoughts,’’ or ‘‘worry about the future,’’ ac- time is similar between high school and college students,
counted for 35% of the responses, followed by excess noise but bedtimes and risetimes are shifted later by about
(33%), cosleeping (7%), and talking with friends (6%). Third, 90 minutes on both week and weekend days. The tendency
when asked ‘‘If your sleep is at all compromised, to what one to delay bedtimes and extend risetimes during weekends
factor do you most strongly attribute this?’’ in forced-choice also continues into young adulthood. In addition to short
question, the majority of students responded that academic sleep and irregular schedules, college students also experi-
(39%) or emotional (25%) stress most interfered with their ence low sleep quality, when assessed by standard measures.
sleep. By comparison, light or noise accounted for 17% of the Surprisingly, perceived stress (rather than sleep schedule
responses, illness or pain accounted for 8%, and cosleeping regularity, alcohol or drug use, exercise frequency, or elec-
accounted for an additional 4% of responses. Moreover, when tronics usage) provided the most explanatory power for
asked what factor most interferes with initiating sleep, 68% of poor sleep in this population.
students responded with stress, compared to 10% citing temper-
ature, and 8% responding with light or noise.
Limitations
To evaluate what factors predict sleep quality, we per-
formed a multiple stepwise regression using factors previ- The results of this study must be interpreted in light of the
ously shown to be related to sleep quality (e.g., mood, limitations inherent in its design. First, this sample consisted
caffeine and alcohol use, regularity of sleep schedule, and of college students from one geographic area who were
electronics exposure). The tension component of the Profile generally healthy and well-educated, and thus our findings
of Mood States predicted 21% of the variance in sleep quality are not necessarily generalizable to the United States young
(PSQI score), with stress (SUDS) accounting for an addi- adult population. However, basic sleep schedules (week
 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132 129

Table 2
Sample characteristics by class and gender
N Mean SD N Mean SD
ESS week Morningness/eveningness
Class 1 262 7.3 3.6 Class 1 273 49.3 9.0
2 270 7.1 3.9 2 279 50.4 9.3
Post hoc 3 244 6.7 3.7 3 255 49.4 9.9
1>4 4 205 6.2 3.7 4 213 50.3 10.1

*Gender M 359 6.2 3.6 Gender M 378 49.6 10.0

F 625 7.2 3.8 F 645 50.0 9.4
ESS weekend Caffeinated drinks/weekday
Class 1 262 6.7 3.5 *Class 1 305 0.7 1.0
2 270 6.8 3.7 2 312 0.7 1.1
3 244 6.6 3.8 3 271 0.9 1.1
4 205 6.4 3.5 1 < 3,4 2 < 4 4 232 1.0 1.2
*Gender M 359 6.0 3.6 Gender M 420 0.8 1.2
F 625 7.0 3.6 F 705 0.8 1.0
Total PSQI Caffeinated drinks/weekend day
Class 1 254 6.7 3.0 Class 1 305 0.9 1.0
2 254 6.9 3.1 2 312 1.0 1.3
3 236 7.3 3.3 3 271 1.2 1.6
4 202 7.1 3.1 1<3 4 232 1.1 1.6
Gender M 349 6.7 3.1 Gender M 420 1.0 1.3
F 600 7.2 3.2 F 705 1.1 1.5
SUDS week Alcoholic drinks/weekday
*Class 1 248 55.1 25.1 *Class 1 305 0.2 0.6
2 244 62.3 22.3 2 312 0.2 0.6
3 234 63.0 21.7 3 271 0.5 0.9
1 < 2,3,4 4 196 62.2 24.5 1,2 < 3,4 4 232 0.5 0.8
*Gender M 345 53.6 26.0 *Gender M 420 0.5 1.0
F 580 64.7 21.1 F 705 0.3 0.5
SUDS weekend Alcoholic drinks/weekend day
*Class 1 248 34.9 23.0 Class 1 305 2.8 4.2
2 243 40.2 23.1 2 312 2.5 3.7
3 234 40.9 22.7 3 271 3.0 4.0
1<3 4 195 39.1 24.1 4 232 3.2 4.0
*Gender M 345 31.8 23.4 *Gender M 420 3.9 5.1
F 578 43.0 22.3 F 705 2.2 3.0
Difference from total N reflects omissions in survey reporting. Asterisks indicate significant differences between class or gender. Bonferroni tested significant
differences (a ¼ .01) between classes are provided in the left side of the columns.
SUDS ¼ Subjective Units of Distress Scale; PSQI ¼ Pittsburgh Sleep Quality Index; ESS ¼ Epworth Sleepiness Scale.

and weekend bedtimes and risetimes) in this report were oping depressive mood disorder [31,32]. Also, as in any
similar to ones at a Chinese university and a small liberal survey data, there is the potential for a recall bias.
arts college in New England [22,29]. Also, with respect to
differences between student and nonstudent sleep patterns,
Sleep and stress
Oginska and Pokorski [17] documented similar relationships
between shortened sleep and excessive drowsiness, poor In this age group, tension and stress seemed to be the most
mood, and tension in both university students and working important factors in predicting sleep quality, accounting for
young adults in a European population. 24% of the variance in PSQI score. Similarly, in a study of
Second, as this study consisted of a one-time survey, it is Chinese younger adolescents (N ¼ 1,629), perceived stress
impossible to determine directionality in the relationships was the most significant risk factor for poor sleep quality,
between poor sleep quality, mood, and stress, or to what accounting for 13.5% of the variance in sleep quality score
extent poor sleep is secondary to or predictive of stress [18]. Perceived stress can serve as predisposing, precipi-
and anxiety. However, much evidence suggests that this is a tating, and perpetuating factors for sleep difficulties in this
complex, bidirectional relationship. In a large population- population. First, the college lifestyle creates precipitating
based study (N ¼ 14,915), mood disorder diagnoses were factors that enhance stress-related sleeping difficulties (e.g.,
more often preceded by, rather than concurrent with, periods erratic schedules, high-stress periods like final exams).
of poor sleep [30]. Furthermore, in longitudinal studies, Second, students may be more susceptible to hyperarousal-
periods of disturbed sleep are significant predictors of devel- related sleep difficulties because of maturational changes in
130 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132

Table 3
Differences in behavior by PSQI group
Sleep schedule (df) F p 2
h Post hoc Optimal Borderline Poor
Total sleep time (h) 2,952 102.8 <.001 .01 O > B>P 7.61 7.08 6.47
Bedtime, weekday 2,948 9.5 <.001 .02 O,B < P 12:07 a.m. 12:13 a.m. 12:28 a.m.
Risetime, weekday 2,948 2.8 0.062 <.01 8:06 a.m. 7:54 a.m. 8:06 a.m.
Bedtime, weekend 2,946 5.5 0.004 .01 O<P 1:34 a.m. 1:42 a.m. 1:52 a.m.
Risetime, weekend 2,947 3.0 0.060 <.01 10:03 a.m. 10:03 a.m. 10:16 a.m.
Bedtime Delay (h) 2,946 0.8 0.448 <.01 1.46 1.48 1.38
Weekend Oversleep (h) 2,947 4.0 0.019 <.01 1.95 2.7 2.2
Morningness/Eveningness 2,949 33.1 <.001 .07 O > B>P 53 50 47
(df),N c2 p Optimal Borderline Poor
Stayed up to 3 a.m. >13/week 6,948 46.4 <.001 25% 32% 46%
All-nighter >13/month 6,947 30.8 <.001 12% 16% 28%
Stress and mood (df) F p 2
h Post hoc Optimal Borderline Poor
Anger 2,897 66.8 <.001 .13 O < B<P 7.48 9 10.61
Confusion 2,897 32.2 <.001 .07 O < B<P 8.6 9.56 10.31
Depression 2,897 71.2 <.001 .14 O < B<P 7.01 8.76 10.66
Fatigue 2,897 146.2 <.001 .25 O < B<P 9.44 12.09 14.92
Tension 2,897 81.1 <.001 .16 O < B<P 8.29 9.96 11.82
Vigor 2,897 28.4 <.001 .06 O > B>P 14.29 13.38 12.09
Weekday distress (SUDS) 2,916 72.4 <.001 .14 O < B<P 49.9 59.9 70.7
Weekend distress (SUDS) 2,916 37.7 <.001 .08 O < B<P 30.8 38 46.6
Psychoactive drug use (df) F p 2
h Post hoc Optimal Borderline Poor
Caffeinated drinks/day 2,952 0.53 0.59 <.01 1.0 0.990 1.08
Alcoholic drinks/day 2,952 3.42 0.03 <.01 1.07 1.24 1.35
(df),N c2 p Optimal Borderline Poor
Use OTC/Rx meds to wake >13/month 6,871 23.3 0.003 12% 22% 26%
Use OTC/Rx meds to sleep >13/month 6,949 118 <.001 4% 13% 33%
Use alcohol to get to sleep 2,681 14.0 <.001 5% 2% 10%
Sleepiness and performance (df) F p 2
h Post hoc Optimal Borderline Poor
Epworth SS weekday 2,917 42.2 <.001 .04 O < B<P 5.32 6.95 8.08
Epworth SS weekend 2,915 16.7 <.001 .10 O < B<P 5.58 6.83 7.34
(df),N c2 p Optimal Borderline Poor
Fall asleep in class 13/week 6,951 39.2 <.001 9% 12% 21%
Skip class >2/mo, due to illness 12,948 39 <.001 4% 3% 12%
Skip class >2/months, other reasons 10,950 29.3 <.001 16% 18% 22%
O ¼ optimal (PSQI >6), B ¼ borderline (PSQI ¼ 6–7), P ¼ poor (PSQI >7). a ¼ .01.
PSQI [ Pittsburgh Sleep Quality Index; OTC ¼ over the counter.

the neuroendocrine system. Developmental changes in the Of particular concern is the tendency for older adolescents
HPA axis during adolescence result in increased perisleep to self-medicate sleep-wakefulness. Self-administration of
onset cortisol secretion [33]. This neuroendocrine hyperac- OTC medication in an older adolescent population is associ-
tivity could contribute to both the hyperarousal observed in ated with psychological distress [36]. In our sample, poor-
delayed sleep onset [34] as well as increased feelings of quality sleepers reported higher alcohol consumption and
anxiety and depression. Third, college students may have more frequent use of alcohol and OTC drugs to help regulate
not yet developed sufficient coping strategies for handling their sleep/wake schedule. A potential consequence of such
stressful events, and subsequently experience more internal- behaviors is the stimulant–sedation loop (use of caffeine
izing, rumination, and worry [35]. Thus, biological factors and other stimulants to counteract daytime sleepiness, and
(e.g., hyperarousal of the autonomic nervous system and subsequent use of depressants to counteract the effects of
HPA axis overactivation) provide a predisposition for the stimulants). Students who get caught in this pattern
stress-induced sleep difficulties, stressful events common in may be at a higher risk for developing drug dependence
this population (e.g., midterm examinations, relationship [37]; approximately 90% of adolescents entering drug rehab
troubles) precipitate bouts of sleep difficulties, and rumina- programs report self-medicating with psychoactive drugs to
tion and worry can perpetuate the sleep difficulty. control sleep and combat fatigue [38].
 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132 131

Table 4
Stepwise multiple regression predicting PSQI scores
F(5,877) ¼ 882, p < .001 B Std. Error Beta t Sig.
(Constant) 7.449 3.453 2.158 .031
Predictors, R2 ¼ 29%
POMS-Tension .118 .041 .143 2.857 .004
SUDS .024 .005 .182 5.278 <.001
MES .060 .014 .187 4.290 <.001
POMS-Depression .112 .036 .149 3.088 .002
POMS-Anger .102 .038 .119 2.669 .008
Nonsignificant variables
Age .080 .071 .034 1.125 .261
Sex .152 .210 .024 .724 .469
Ethnicity .115 .288 .012 .398 .691
Caffeine/day .103 .083 .038 1.247 .213
Alcohol/day .105 .067 .052 1.561 .119
Bedtime delay .178 .104 .054 1.707 .088
Weekend oversleep .031 .082 .013 .377 .706
GPA .214 .250 .027 .858 .391
Hrs Exercise/week .065 .071 .028 .906 .365
Hrs TV/Video/week .018 .055 .010 .323 .747
POMS-Confusion .010 .046 .009 .222 .825
GPA ¼ grade-point average; PSQI [ Pittsburgh Sleep Quality Index; SUDS ¼ Subjective Units of Distress Scale; MES ¼ Morningness Eveningness Scale;
POMS ¼ Profile of Mood States.

Recommendations [5] Wolfson AR, Carskadon MA. Understanding adolescents’ sleep

patterns and school performance: a critical appraisal. Sleep Med Rev
These results highlight a growing need for professionals to 2003;7:491–506.
focus on the quality as well as the quantity of sleep when [6] Taylor DJ, Jenni OG, Acebo C, et al. Sleep tendency during extended
wakefulness: insights into adolescent sleep regulation and behavior. J
promoting mental and physical health in adolescents and
Sleep Res 2005;14:239–44.
young adults. College students who are consistently getting [7] Crowley SJ, Acebo C, Carskadon MA. Sleep, circadian rhythms, and
poor-quality sleep are at risk for problems far more serious delayed phase in adolescence. Sleep Med 2007;8:602–12.
than simply struggling to function in daily activities. As [8] Van den Bulck J. Television viewing, computer game playing, and
chronic insomnia is a risk factor for major mood [39] and Internet use and self-reported time to bed and time out of bed in
secondary-school children. Sleep 2004;27:101–4.
substance abuse disorders [40], physicians, college health-
[9] National Sleep Foundation: Sleep in America Poll, 2006 [Online].
care professionals, and residence life workers should be Available at: http://www.sleepfoundation.org. Accessed July 15, 2008.
more proactive in screening for sleep difficulties and in artic- [10] Carskadon MA, Wolfson AR, Acebo C, et al. Adolescent sleep
ulating the importance of sufficient, restorative sleep in patterns, circadian timing, and sleepiness at a transition to early school
college students’ well-being. days. Sleep 1998;21:871–81.
[11] Hansen M, Janssen I, Schiff A, et al. The impact of school daily
schedule on adolescent sleep. Pediatrics 2005;115:1555–61.
[12] Wolfson AR, Spaulding NL, Dandrow C, et al. Middle school start
times: the importance of a good night’s sleep for young adolescents.
Disclosure Statement Behav Sleep Med 2007;5:194–209.
[13] Campbell IG, Higgins LM, Trinidad JM, et al. The increase in longi-
The authors have indicated no financial conflicts of tudinally measured sleepiness across adolescence is related to the
interest. maturational decline in low-frequency EEG power. Sleep 2007;30:
1677–87.
[14] Carskadon MA, Harvey K, Duke P, et al. Pubertal changes in daytime
References sleepiness. Sleep 1980;2:453–60.
[15] Roberts RE, Roberts CR, Chen IG. Functioning of adolescents with
[1] Léger D, Massuel MA, Metlaine A, SISYPHE Study Group. Profes- symptoms of disturbed sleep. J Youth Adolesc 2001;30:1–18.
sional correlates of insomnia. Sleep 2006;29:171–8. [16] Fredricksen K, Rhodes J, Reddy R, et al. Sleepless in Chicago: tracking
[2] Léger D, Scheuermaier K, Phillip P, et al. SF-36: Evaluation of quality the effects of adolescent sleep loss during the middle school years.
of life in severe and mild insomniacs compared with good sleepers. Child Dev 2004;75:84–95.
Psychosom Med 2001;63:49–55. [17] Oginska H, Pokorski J. Fatigue and mood correlates of sleep length in
[3] Katz DA, McHorney CA. The relationship between insomnia and three age-social groups: school children, students, and employees.
health-related quality of life in patients with chronic illness. J Fam Pract Chronobiol Int 2006;23:1317–28.
2002;51:229–35. [18] Chung KF, Cheung MM. Sleep–wake patterns and sleep disturbance
[4] Roth T, Jaeger S, Jin R, et al. Sleep problems, comorbid mental disor- among Hong Kong Chinese adolescents. Sleep 2008;31:185–94.
ders, and role functioning in the national comorbidity survey replica- [19] O’Brien EM, Mindell JA. Sleep and risk-taking behavior in adoles-
tion. Biol Psychiatry 2006;60:1364–71. cents. Behav Sleep Med 2005;3: 133–33.
132 H.G. Lund et al. / Journal of Adolescent Health 46 (2010) 124–132

[20] Roberts RE, Roberts CR, Duong HT. Chronic insomnia and its negative [31] Breslau N, Roth T, Rosenthal L, Andreski P. Sleep disturbances and
consequences for health and functioning of adolescents: a 12-month psychiatric isorders: a longitudinal epidemiological study of young
prospective study. J Adolesc Health 2008;42:294–302. adults. Biol Psychiatry 1996;39:411–8.
[21] Medeiros ALD, Mendes DBF, Lima PF, et al. The relationships [32] Chang PP, Ford DE, Mead LA, et al. Insomnia in young men and subse-
between sleep–wake cycle and academic performance in medical quent depression. The Johns Hopkins Precursors Study. Am J Epide-
students. Bio Rhythm Res 2001;32:263–70. miol 1997;146:105–14.
[22] Tsai LL, Li SP. Sleep patterns in college students: gender and grade [33] Forbes EE, Williamson DE, Ryan ND, et al. Peri-sleep-onset cortisol
differences. J Psychosom Res 2004;56:231–7. levels in children and adolescents with affective disorders. Biol Psychi-
[23] Buysse DJ, Reynolds CF 3rd, Monk TH, et al. The Pittsburgh Sleep atry 2006;59:24–30.
Quality Index: a new instrument for psychiatric practice and research. [34] Drake C, Richardson G, Roehrs T, et al. Vulnerability to stress-related
Psychiatry Res 1989;28:193–213. sleep disturbance and hyperarousal. Sleep 2004;27:285–91.
[24] Johns MW. Reliability and factor analysis of the Epworth Sleepiness [35] Jose PE, Ratcliffe V. Stressor frequency and perceived intensity as
Scale. Sleep 1992;15:376–81. predictors of internalizing symptoms: gender and age differences in
[25] Horne JA, Otsberg O. A self-assessment questionnaire to determine adolescence. N Z J Psychol 2004;33:145–54.
morningness–eveningness in human circadian rhythms. Int J Chrono- [36] Stasio MJ, Curry K, Sutton-Skinner KM, et al. Over-the-counter medi-
biol 1976;4:97–110. cation and herbal or dietary supplement use in college: dose frequency
[26] Wolpe J. The Practice of Behavioral Therapy, 3rd ed. New York: and relationship to self-reported distress. J Am Coll Health 2008;
Pergamon Press, 1982. 56:535–47.
[27] McNair DM, Lorr M, Droppleman LF. Profile of mood states. EdITS [37] Brower KJ, Aldrich MS, Robinson EA, et al. Insomnia, self-medica-
Manual for the Profile of Mood States. San Diego, CA: EdITS/Educa- tion, and relapse to alcoholism. Am J Psychiatry 2001;158:399–404.
tional and Industrial Testing Service, 1992. [38] Bootzin RR, Stevens SJ. Adolescents, substance abuse, and the treat-
[28] Roehrs T, Shore E, Papineau K, et al. A two-week sleep extension in ment of insomnia and daytime sleepiness. Clin Psychol Rev 2005;
sleepy normals. Sleep 1996;19:576–82. 25:629–44.
[29] Lund HG, Whiting AB, Prichard JR. The relationship between sleep [39] Neckelmann D, Mykletun A, Dahl AA. Chronic insomnia as a risk factor
quality, mood, and stress in college Women. Sleep Abstr 2006;29. for developing anxiety and depression. Sleep 2007;30:873–80.
[30] Ohayon MM, Roth T. Place of chronic insomnia in the course of [40] Roane BM, Taylor DJ. Adolescent insomnia as a risk factor for early
depressive and anxiety disorders. J Psychiatr Res 2003;37:9–15. adult depression and substance abuse. Sleep 2008;31:1351–6.