Abstract
Introduction Poor sleep hygiene portends loss of physical and mental stamina. Therefore, maintaining a regular sleep/wake schedule on both weekdays and weekends is highly recommended. However, this advice runs contrary to the habits of university students, who may only experience recovery sleep if they “sleep in” on weekends. Pharmacy students at Duquesne sit for frequent examinations, typically commencing at 7:30 AM, and they complain about fatigue. Thus, we tested the hypothesis that longer sleep durations on both weekdays and weekends are linked to stronger academic performance.
Methods Students in their third year at Duquesne University were administered three surveys to collect daily data on sleep habits and factors that might influence sleep quality, such as having roommates, long commute times, and sleep interruptions. GPAs were collected from the Dean’s office, with permission from the students.
Results Longer weekend—but not weekday—sleep durations were significantly correlated with higher GPAs in men and not in women. Women achieved slightly higher cumulative GPAs than men. Students who fell asleep within 15 minutes of going to bed had higher GPAs than those who fell asleep after an hour or more.
Conclusion The present observations do not establish causal links, but, given the body of prior evidence on the salutary properties of sleep, men in this cohort may have reaped benefit from recovery sleep on weekends. Rather than recommending that students force themselves awake on weekends in an attempt to maintain a consistent sleep routine, the real-life habits of students should be considered.
Introduction
“Sleep … Balm of hurt minds … Chief nourisher in life’s feast.”
Macbeth (2.2.46-51) by William Shakespeare
During the sleep phase of the activity/rest rhythm, the glymphatic system of the mammalian brain performs its janitorial duties and clears the accumulated metabolites via the cerebrospinal fluid.1-3 Sleep deprivation studies further suggest that sleep loss-induced attentional deficits are preceded by electrophysiological lapses in neuronal function, and that the association between sleep loss and cognitive impairment is causal.4-10 Thus, sleep is linked to superior memory consolidation and academic performance,11-13 including in students enrolled in Pharmacy programs.14
In the Pharmacy program at Duquesne University, exams are administered at 7:30 AM in the morning (6:30 or 7:00 AM for special needs students), and classes commence at 8:00 AM. Despite the need for early-morning awakenings and awareness of the benefits of good sleep hygiene, anecdotal comments from Pharmacy students suggest that they often stay up late at night, cramming for the 7:30 AM examinations, and then “crash” on weekends by oversleeping. Thus, the central hypothesis was that longer sleep durations and consistent sleep habits would be associated with better academic performance in first-year pharmacy students.
Methods
Study Design
Ethics approval for three surveys was granted by the Institutional Review Board at Duquesne University. First, a homework assignment was administered in the Ability Based Laboratory Experience (ABLE) course at Duquesne University. Students register for this course in the second semester of Year 1 of the four-year professional phase, after completion of the two-year preprofessional phase. Out of 152 enrolled students, all completed the daily online Survey 1 (Appendix 1) to record bedtimes, sleep times, and awakenings for three consecutive weeks mid-semester. Survey 2 was voluntary. In Survey 2, demographic information and permission to publish the data from Survey 1 (on a separate page from demographic data) were collected from 125 students (Appendix 2).
A third, voluntary survey was deployed two months later to the same student body in their Continuous Professional Development course, to continue to assess additional lifestyle factors hypothesized to impact sleep quality and academic performance, such as participation on an athletic team, nap frequency and duration, hours spent working at a job, etc. (Appendix 3). One-hundred and twenty-five students participated in the latter survey. In Survey 3, permission was collected to acquire individual grade-point averages (GPAs) from the Dean’s office. Students could refuse to have their data analyzed and published at any time. Data were deidentified to protect the students’ anonymity.
Statistics
Data were analyzed in GraphPad Prism (Prism 8 for MacOS) and subjected to Prism’s default tests for heteroscedasticity (Bartlett’s, Brown-Forsythe, and Spearman’s test) and normality (Anderson-Darling, D’Agostino-Pearson omnibus, Shapiro-Wilk, and Kolmogorov-Smirnoff tests). When parametric assumptions were met, Pearson correlations, Student t tests, or ANOVAs were performed on data sets. Bonferroni post hoc tests were used for multiple comparisons after the appropriate ANOVA. For non-Gaussian data sets, the Kruskal-Wallis test was followed by the Dunn’s post hoc correction for multiple comparisons. Alpha was set at 0.05 (two-tailed).
Inclusion/Exclusion Criteria
Data were included in the analyses and graphs only if the student granted permission. Data were excluded only if the student did not grant permission, or failed to complete that specific part of the survey (i.e., some students did not answer every single question on each survey). Therefore, the number of students per group were added to every figure. No outliers were removed.
Results
Demographic data and a frequency histogram of GPAs are illustrated in Figure 1. The majority of participants were women, 21 years of age, not part of an athletic team, and had not transferred from another school to Duquesne University. More women commuted than men, but among the women, a larger percentage lived on campus.
Weekend sleep duration was significantly associated with cumulative GPAs collected from the Dean’s office (Figure 2A; one-way ANOVA; F(4, 107) = 2.621; p = 0.0389; passed heteroscedasticity and normality tests). Students who slept 10 or more hours per weekend night had significantly higher cumulative GPAs than students who slept 6 hours per weekend night. Weekday sleep was not significantly associated with GPAs. Women had slightly higher GPAs than men (Figure 2B; two-tailed t test; t = 2.418; df = 118; p = 0.0171; passed heteroscedasticity and normality tests). Thus, the impacts of gender and weekend sleep duration on GPAs were analyzed by two-way ANOVA (Figure 2C; passed heteroscedasticity and normality tests). A significant interaction between gender and hours of sleep on the weekend was observed (p = 0.0235, F(4, 102) = 2.954), as well as a significant effect of weekend sleep duration (p = 0.0059; F(4, 102) = 3.851). However, Bonferroni post hoc comparisons revealed that the potential impacts of longer weekend sleep durations were observed in men and not women (Figure 2C). Therefore, correlation analyses between weekend sleep and GPAs were plotted separately for men and women. These latter analyses confirmed a significant correlation between weekend sleep duration and GPA for men, but not women (Figure 2D-E; passed normality tests). In contrast, weekday sleep duration was not associated with GPA in men (Pearson r = 0.1468; two-tailed p = 0.3661) or women (Pearson r = 0.1772; two-tailed p = 0.1183).
The average standard deviation in sleep duration for each student across the survey period (adapted from Okano et al. as “inconsistency in sleep duration from day to day”11) did not differ between men and women (Figure 2F; passed heteroscedasticity but failed normality tests; Mann-Whitney U statistic 1588; two-tailed p = 0.7788) and was not correlated with GPA (not shown). Other notable measures were not statistically significantly related to GPAs, including the number and duration of naps, sleep interruptions, and the number of hours of job-related work per week. GPAs were also not significantly associated with commute duration. One exception was that the number of minutes to fall asleep after entering bed was significantly associated with GPAs from the professional phase, in a U-shaped pattern (Figure 2G; one-way ANOVA; F(4, 116) = 2.763; p = 0.0308; passed heteroscedasticity and normality tests). Subjects who fell asleep within 15 minutes, on average, had significantly higher professional GPAs than those who needed one or more hours. This advantage, however, was not observed in those who reported falling asleep immediately upon entering bed.
Discussion
The main finding of the present study is that weekend sleep duration explained a significant proportion of the variance in the GPAs of men, but not women. Our students diverge from other studies in that we failed to observe a correlation between academic performance and weekday sleep duration,11, 14 perhaps due to early exam schedules, combined with a high frequency of assignments and exams (four exams/semester for multiple courses). Given the lack of significant correlations between academic performance and weekday sleep durations, our central hypothesis was only partially supported. However, it should be noted that the other studies did not distinguish weekday from weekend sleep, and that the Zeek et al. study did not report the impact of gender.14
Given the collective findings, we speculate that men enrolled in our program may benefit from sleeping longer on the weekend, although it seems reasonable to recommend that both sexes catch up on lost sleep whenever weekday schedules are particularly hectic. It is known that women outperform men in academics, and they may enjoy a greater cognitive buffer against the negative sequelae of sleep loss.15 In contrast to previous studies,11 we did not observe higher sleep inconsistency in men compared to women. Thus, gender differences in academic performance in our student cohort are not readily explained by differences in sleep inconsistency.
The second main finding of the present study is the U-shaped graph of GPAs plotted as a function of the reported time to fall asleep. Taking one hour or more to fall asleep was associated with lower professional-phase GPAs than for those who required, on average, 15 minutes. Those who fell asleep as soon as their heads hit the pillow enjoyed no such advantage. These observations suggest that additional information on sleep-delaying factors, such as blue light exposure from electronic devices and anxiety-related insomnia, should be investigated in this cohort, particularly during the professional phase.
The major limitation of the current study was the reliance on self-reported survey data to assess sleep duration (due to financial constraints), rather than more expensive methods such as heartrate/activity-based sleep monitors (e.g., Fitbits) or electroencephalograms. On the other hand, sleep data were collected on a daily basis for three weeks, and are therefore independent of lapses in long-term memory recall, which can compromise survey data integrity.
Conclusion
Based on the current findings and a large body of sleep literature, we speculate that setting an early alarm on weekends in an effort to maintain the same sleep schedule as during the week may be counterproductive, especially in male students enrolled in academic programs with early-morning examinations or classes.
Acknowledgements
RKL conceived the study, wrote the paper, interpreted and analyzed data, and constructed figures. SLW entered and analyzed all the data, constructed figures, and contributed to experimental design, interpretation, and manuscript editing. MNC contributed to experimental design, collected survey data, and edited the manuscript. DCR contributed to experimental design and interpretation and edited the manuscript. We are indebted to the School of Pharmacy for their generous support of Dr. Leak’s lectures on the epidemiology and biological impact of sleep. We are also grateful to the Duquesne pharmacy students, for their kind participation. The authors have no conflicts to declare.
Appendix 1:
Survey One
(via Survey Monkey)
Question 1: What time did you go to bed?
Question 2: What time did you fall asleep?
Question 3: How many times did you wake up in the middle of the night? Why did you wake up?
Question 4: What time did you wake up in the morning?
Question 5: Did you feel refreshed within 30 minutes of waking up?