Behavioral Treatment and Sleep. The Effects of Regularizing Sleep-Wake Schedules on Daytime Sleepiness

Transcription

1 Sleep, 19(5): American Sleep Disorders Association and Sleep Research Society Behavioral Treatment and Sleep The Effects of Regularizing Sleep-Wake Schedules on Daytime Sleepiness *Rachel Manber, trichard R. Bootzin, :j:christine Acebo and :j:mary A. Carskadon *Department of Psychiatry, University of Arizona, Tucson, Arizona, U.S.A,; tdepartment of Psychology, University of Arizona, Tucson, Arizona, U.S.A.; and tdepartment of Psychiatry and Human Behavior, E. P. Bradley Hospital, Brown University Medical School, Providence, Rhode Island, U.S.A. Summary: The present study evaluated the differential effects of two manipulations of sleep-wake schedules on daily subjective ratings of daytime sleepiness of college undergraduate students. Two experimental conditions were compared: a sleep only group and a regularity group. Subjects in both conditions were given a lower limit for total sleep time (7.5 hours). Subjects in the regularity group received an additional instruction to keep a regular sleep schedule. The study was longitudinal and prospective. Following a baseline period (12 days), the experimental conditions were introduced. The experimental phase lasted 4 weeks and overall compliance was good. A follow-up phase (1 week) began 5 weeks past termination of the experimental phase. The findings indicated that when nocturnal sleep is not deprived, regularization of sleep-wake schedules is associated with reduced reported sleepiness. Subjects in the regular schedule condition reported greater and longer lasting improvements in alertness compared with subjects in the sleep only condition and reported improved sleep efficiency. Key Words: Sleep duration-daytime sleepiness-sleep-wake schedule. Although regular sleep schedules are commonly recommended as part of good sleep hygiene, very little is known about the relationship between the regularity of sleep-wake patterns and daytime sleepiness. The present study addressed this question by evaluating the effect of manipulating sleep-wake schedules on selfreported daytime sleepiness. The target population was college students, a population known for a high prevalence of daytime sleepiness, irregular sleep schedules, and short sleep durations (1-6). The sample included individuals with a high score on sleepiness and a high variability of sleep schedule as reported during the screening procedure. The relevance of total sleep duration to daytime sleepiness has face validity as well as scientific support. The sleep deprivation literature, particularly the literature on partial sleep deprivation, shows that insufficient sleep leads to sleepiness during the day (e.g. 7). There is also evidence that habitual Accepted for publication March Address correspondence and reprint requests to Rachel Manber, Ph.D., Department of Psychiatry, Arizona Health Science Center, 1501 North Campbell Avenue, Tucson, AZ 85724, U.S.A. 432 sleep durations of many adolescents produce chronic partial sleep deprivation (e.g. 8-10) and that this is likely to contribute to the prevalence of the sleepiness complaint in this age group. Much less is known about the relevance of the irregularity of sleep patterns to the experience of daytime sleepiness. In a series of studies, Taub and Berger (11) shifted the sleep of regular sleepers and examined the effects of this schedule manipulation on mood and performance. They demonstrated that both a delay and an advance of sleep by 2-4 hours led to increased levels of fatigue, decrements in performance, and deterioration in mood. On the other hand, Bonnet and Alter (12) found no improvement in performance when sleep schedules of irregular sleepers were regularized. Neither group of researchers measured the effect of their manipulation on daytime sleepiness. Survey studies of adolescents and college-aged students have found an association between irregular sleep patterns and elevated levels of sleepiness. For example, Billiard et al. (2) found that unintended daytime sleep episodes were more prevalent among subjects with irregular sleep-wake schedules (6.4%) than

2 SLEEP-WAKE SCHEDULES AND DAYTIME SLEEPINESS 433 TABLE 1. Sleepiness subscale L How often do you feel very tired and have to struggle to stay awake? a. Never b. Rarely c. Occasionally d. Often e. Very often o How often do you feel the need for a nap? a. Never b. Rarely c. Occasionally d. Often e. Very often o How often do you feel that your performance is affected by your sleepiness? a. Never b. Rarely c. Occasionally d. Often e. Very often o How great a problem do you have with sleepiness (feeling sleepy/struggling to stay awake) during the day? a. No problem b. Mild c. Moderate d. Pretty bad e. Severe I am excessively sleepy during the day. a. Strongly disagree b. Disagree o 0 c. Neutral o Scoring is indicated by the integer values placed under each scale item. d. Agree 3 e. Strongly agree 4 among subjects with regular sleep-wake schedules (2.2%). Strauch and Meier (6) found that individuals who expressed a need for more sleep practiced irregular sleep patterns more frequently than individuals who stated that they get enough sleep. Short habitual sleep duration, however, might have a greater impact on daytime sleepiness than the irregularity of sleep pattern. Even when the amount of sleep deprivation per night is small, its cumulative effects are substantial. In a recent study, Manber and Bootzin (13) found that sleepy individuals with irregular sleep schedules who regularized their sleep schedules but suffered sleep loss in the process experienced an increase in daytime sleepiness and a concomitant deterioration in daily mood and concentration ratings. The central question of the present study was whether irregular sleepers would manifest decreased sleepiness levels after regularizing their sleep-wake schedules under conditions that assured adequate sleep duration. Specifically, it was hypothesized that when irregular sleepers slept at least 7.5 hours at night and increased the regularity of their sleep schedule, they would report less sleepiness, and that the regularity of sleep schedules would contribute to decreased sleepiness above and beyond the benefits obtained by sleeping at least 7.5 hours at night. To test these hypotheses, two experimental conditions were compared. Both groups were assigned the same lower bound on nocturnal sleep duration. One of the groups received an additional instruction to keep a regular sleep schedule. Factors other than sleep schedules may also be associated with daytime sleepiness. Several of these factors, such as medication use, disorders of excessive daytime sleepiness, substance abuse, medicated depression, and frequent nightmares were controlled by exclusion criteria in the subject selection phase, whereas other factors, such as unmedicated depression and circadian rhythm disorders, remained beyond the scope of the present study and were controlled by random assignment to experimental conditions. Design METHODS The present study was longitudinal and prospective. It consisted of a baseline period (12 days), followed by an experimental phase in which the experimental conditions were introduced (4 weeks), and a follow-up phase (1 week) that began 5 weeks after the termination of the experimental phase. At the end of the baseline phase, subjects were randomly assigned to one of three experimental groups stratified for gender and for levels of daytime sleepiness as reflected in the first week of baseline diary records. The present report is based on a comparison of two of the three experimental groups, the sleep only group and the regularity group. A third group, the nap group, was excluded from this report because its most salient instruction (taking daily naps) was unrelated to schedule irregularity. To minimize the effects of external events such as midterm, school break, season, etc., all subjects entered the study together and proceeded through the experimental phases together. Subjects The subjects in the present study were recruited from an introductory psychology course at the University of Arizona and received course credit for participation. About 1,000 students were screened via a questionnaire, followed by a telephone interview of Sleep. Vol. 19. No

3 434 R. MANBER ET AL. those who met initial selection criteria. Selection criteria included a high score on a sleepiness subscale of the screening questionnaire (Table 1) and an irregular sleep-wake schedule based on self report in the screening questionnaire. Subjects were asked to indicate their bedtime and wake-up time on weekdays and weekend days. A subject was identified as having an irregular sleep schedule if she or he reported a difference of 2 hours or more in wake-up time between weekdays and weekends or a range of at least 2 hours in wake-up time on either weekdays or weekends. The criteria threshold of 2 hours was chosen parsimoniously to allow for sufficient sample size. It has also been used by others in manipulating regularity of schedules (11). The sleepiness subscale included five items (Table 1). The scores of the sleepiness subscale ranged between 0 and 17, and subjects with a score of 7 or greater were included if they met the other inclusion and exclusion criteria. Scores of the final sample (n = 39) ranged from 7 to 15, with an average of 9.6 and a standard deviation (SD) of 2.l. Exclusion criteria were based on responses to a symptom checklist in the screening questionnaire and followed by a brief telephone interview. Exclusion criteria were use of alcohol more than twice a week, use of drugs more than twice a month, use of medications that affect sleep and sleepiness induding antidepressants, symptoms that suggested sleep apnea or restless leg syndrome, and frequent nightmares (more than four times a week). Subjects who met study criteria were recruited by phone and asked to agree in advance to participate in the experiment regardless of the experimental condition to which they would be assigned. Three subjects dropped out during the initial part of the experimental phase. The values of the central sleep and sleepiness measures for these subjects lay within 1 SD from the mean values for their respective groups. Participation of one subject from each group was terminated during the active experimental phase because he did not comply with the experimental instructions. These subjects did not complete the experiment and were not included in the analysis. This report is based on a sample of 39 subjects (12 males and 27 females) who completed the experimental phase. It included 20 subjects in the sleep only group (7 males and 13 females) and 19 subjects in the regularity group (5 males and 14 females). Thirty-six subjects completed the follow-up phase (19 in the sleep only group and 17 in the regularity group). Subjects' ages ranged from 17 to 22 years, with an average of 18.8 (SD= 0.97) years. Procedures Throughout the study, subjects recorded information on a computer scannable sleep-wake diary twice each day, once at bedtime and once upon awakening in the morning. The diaries were collected weekly. Bedtime recording in the diary included ratings of daytime parameters; morning recording in the diary included information on the previous night's sleep. Diaries were reviewed as they were turned in, and subjects were called when problems in diary recording or compliance arose. Subjects called every morning and left a message on an answering machine with information on selected items from the diary (wake-up time, bedtime, sleepiness levels from the previous day, and time of call). These morning calls were designed to increase adherence with the experimental instructions and to increase the probability that subjects would indeed complete the diaries daily. Following the subject selection phase, all subjects attended an orientation meeting. They were given a brief description of the study and its procedures, they signed consent forms, and they were taught how to record information in the diaries. For the next 12 days (baseline), subjects completed the sleep-wake diaries and kept their normal schedule. No further instructions were given, and the baseline data were not used for further exclusion of subjects. During the recruitment and orientation meeting, subjects were told that the purpose of the study was to examine the relationship between sleep and daytime functioning in young adults. They were told that we invited subjects with varying levels of sleepiness and varying levels of schedule regularity to participate and that following baseline they would be randomly assigned to groups. They were told that they might be asked to maintain a regular sleep schedule and to sleep a minimum number of hours and that this would be determined based on random assignment. Subjects were randomly assigned to the experimental conditions at the end of the baseline phase and were given instructions according to their assigned condition. Assignment of subjects to the experimental conditions was stratified by gender and by the average level of sleepiness during the first week of baseline. Subjects in the sleep only group were instructed to sleep at least 7.5 hours each night and to avoid napping. The 7.S-hour target value for total sleep time was chosen based on the average total sleep time of the whole sample during the first week of baseline. This number was then rounded to the closest half hour to simplify the instructions. Subjects in the regularity group were given the same instructions, plus an additional instruction to maintain a regular schedule by going to bed and waking up during preassigned I-hour time-windows. The time windows for subjects in the regularity group were based on habitual sleep-wake schedules as reflected in the first week of diary entries and in consultation with each subject. All subjects, re-

4 SLEEP-WAKE SCHEDULES AND DAYTIME SLEEPINESS 435 gardless of group assignment, were asked to expose themselves to daylight upon awakening and to engage in stimulating activities immediately after getting out of bed. Instruction for light exposure was based on the documented effects of exposure to bright light on circadian rhythms (14-16) and on sleepiness (17), indicating that it might help individuals adjust to their new sleep-wake schedules. Finally, all subjects were asked to keep consumption of caffeine at baseline levels so as to avoid possible effects of caffeine on changes in sleepiness levels from baseline. Baseline levels of caffeine consumption were computed as the average caffeine consumption reported on diaries during the first week of baseline. Individual levels of caffeine consumption thus calculated were given to each subject at the beginning of the experimental phase. A I-week follow-up phase began 5 weeks after the termination of the experimental phase and consisted of daily recordings in the sleep-wake diaries. Measures All measures were based on daily self report. Four values were computed for each of these measures: a baseline value, two experimental values based on the first 2 weeks (Time 1) and the last 2 weeks (Time 2) of the 4-week experimental phase, and a follow-up value. Each of the four values was computed as the average of a given diary measure over the corresponding time period. Daytime sleepiness, the central outcome measure, was recorded daily in the bedtime portion of the diary. Subjects were asked to rate "how you felt overall today" on two five-point Likert sleepiness scales (and on several mood scales). One of the two items was anchored at "sleepy" and "alert" and the other at "tired" and "energetic". The average of the two items defined the construct Sleepy. The construct has a high internal consistency (the average Cronbach alpha over the 12-day baseline period for Sleepy was 0.84, with a range of 0.75 to 0.93). In a previous study we found that similar daily ratings of sleepiness were sensitive to the effects of experimentally induced chronic sleep deprivation (13). Daily caffeine and alcohol consumption, as well as morning light exposure, were recorded daily in the diaries. The latter was included primarily as a check for compliance with instructions of morning light exposure. In addition, two nap variables included the number of planned naps and the total time spent napping (planned and unplanned). Sleep parameters were recorded every morning upon awakening. They included bedtime, wake-up time, time to sleep onset (minutes), time awake after sleep onset (minutes), number of awakenings after sleep onset, and sleep quality measured by a Likert scale (excellent to poor). Two measures of total sleep time were computed from the diary data-total nocturnal sleep time and total sleep time per 24-hour period. Total nocturnal sleep time was defined as the difference between time spent in bed and time spent awake while in bed (including time to sleep onset and time awake in the middle of the night). Daily sleep time was computed as the sum of total nocturnal sleep time and the time spent napping the following day. A measure of sleep efficiency was computed as the percent of time asleep at night relative to time spent in bed. Two measures of regularity of sleep were derived from the diary: regularity of bedtime and regularity of wake-up time. They were computed as the standard deviations of bedtime and wake-up time over each of the four time periods (baseline, Time 1, Time 2, and the follow-up week). Because the standard deviations during the follow-up phase were based on about half the number of days as in each of the other three phases, follow-up values were not entered into the analysis of variance (ANOV A) of the regularity measures. Analysis For the most part, there were no significant gender differences in the central variables at baseline or any other point in time (sleepiness, sleep variables, or mood variables), and there were no main effects for gender or interaction effects for group by gender or group by time by gender on these variables. As a result, males and females were pooled together in the analyses. One notable exception was the presence of a significant main gender effect on bedtime [F(l, 37) = 6.4, p < 0.05] and on wake-up time [F(1, 37) = 4.7, P < 0.05], whereby males went to bed and woke up about 45 minutes later than females, on the average. There were no group by gender or group by gender by time effects on these two measures. Adherence to instructions was tested by a series of 2 X 3 repeated measures ANOV As (2 groups by 3 points in time, excluding the follow-up phase). Significance levels for within-subject effects include Huynh-Feldt corrections for sphericity (14). Following Borgatta and Bohrnstedt (15), parametric statistics were used for all analyses, including the measures derived as averages of daily rating scales. The assessment of the differential effects of the two interventions was conducted via a series of 2 X 3 repeated measures analyses of covariance (ANCOV As; 2 groups, 3 points in time), with the value of a given dependent variable at baseline as a covariate. The presence of main effects for group guided further hypothesis testing. The AN COY A provides a more conservative measure of

5 436 R. MANBER ET AL '" = CIl = ~ =: r', / r bedtime -e- waketime L..---r I baseline time 1 time 2 Sleep Only Group '" 1.1 = CIl ~ =: ~ '1'" r , i baseline time 1 time 2 Regularity Group FIG. 1. Changes in regularity of bedtime and wake-up time during the study. Regularity is measured in hours as the standard deviation of bedtime and wake-up time over the corresponding period of time. Dispersion measure is the standard deviation. change than ANOV A of change scores because it corrects for error variability associated with the change scores (16). To assess the relationship between sleepiness and prior sleep time and rise time we performed a series of regression analyses with sleepiness as a dependent variable and total sleep time, rise time, and individuals as the dependent variables. These regression analyses were performed on daily diary entries, with individuals coded as dummy variables. The analyses were performed separately on the whole sample at baseline and within each group during the experimental phase. Baseline RESULTS The groups did not differ significantly at baseline on any of the variables related to the experimental manipulations, including wake time [t(37) = -0.75, P = 0.46], bedtime [t(37) = -0.36, P = 0.72], regularity of bedtime [t(37) = -0.02, P = 0.99], regularity of wake-up time [t(37) = -0.63, P = 0.53], total sleep time [t(37) = -0.83, P = 0.41], total nap time [t(37) = 1.73, P = 0.09], light exposure [t(37) = -1.19, P = 0.24], or caffeine consumption [t(37) = 0.82, P = 0.42]. The groups were also not significantly different on levels of reported daytime sleepiness, although there was a trend for lower sleepiness in the regularity group at baseline [t(37) = -1.90, P = 0.065]. The groups were also not significantly different on the unmanipulated sleep measures at baseline. Specifically, sleep efficiency [t(37) = -0.46, P = 0.65] and latency to sleep onset [t(37) = -0.54, P = 0.60] were not different for the two groups. Two sleep variables showed a trend toward significant differences at base- line. Subjects in the sleep only group reported lower sleep quality [t(37) = 2.37, P = 0.02] and a larger number of awakenings at night [t(37) = 2.81, P = 0.008]. (Because a total of 12 group comparisons were made at baseline, the threshold for significant p-values was set at ) Compliance Although there were individual differences in the degree to which subjects followed the instructions during the experimental phase, overall adherence was good. Total sleep time Compliance with the instruction to sleep at least 7.5 at night was good in both groups. Average total sleep time stayed above 7.5 hours during the experimental phase (7.4 ± 0.68 at baseline, 7.6 ± 0.63 in the first 2 weeks of the intervention, and 7.7 ± 0.59 in the last 2 weeks of the intervention). The percentage of subjects who stayed at or above an average of 7.5 hours per night during the experimental phase was 70% for the sleep only group and 58% for the regularity group. Naps Compliance with the instruction to avoid naps was good. Nap frequency was significantly reduced in each group [F(2, 38) = 22.2 for the sleep only group and F(2, 36) = 16.6 for the regularity group, p < 0.001], and there was a main effect for time [F(2, 37) = 37.9, p < 0.001], whereby the average number of naps per week decreased from 0.29 at baseline to 0.04 Sleep. Vol. 19. No

6 SLEEP-WAKE SCHEDULES AND DAYTIME SLEEPINESS '" '" 2.9 GI 2.7 = c. 2.5 GI GI 2.3 r'-l Sleep Only...- Regularity 1.5 baseline timel time 2 follow up FIG. 2. Changes in sleepiness during the study. Sleepiness is measured as the average of daily ratings on a scale of 1 to 5 across the corresponding time period. Greater numbers indicate greater sleepiness. Dispersion measure is the standard deviation. during the experimental phase. There was no significant group by time interaction. The percentage of subjects who napped less than twice a week on average during the experimental phase was 85% for the sleep only group and 100% for the regularity group. Average total time spent napping in the second half of the experimental phase was not significantly increased in the second half of the experimental phase in either of the groups. Regularity of schedules The instruction to adhere to a regular sleep schedule was given only to subjects in the regularity group. Figure 1 depicts the changes in the regularity measures across time for the two groups. Repeated measures ANOVA on the regularity (standard deviation) of bedtime and wake-up time revealed significant group by time interactions [F(2, 74) = 3.4, p < 0.05 for regularity of bedtime and F(2, 74) = 3.7, P < 0.05 for regularity of wake-up time]. Within the regularity group there was a significant main effect for time on regularity of wake-up time [F(2, 38) = 5.14, p::s 0.01] but not in the regularity of bedtime. In addition, planned comparisons revealed that subjects in the regularity group had significantly more regular bedtimes (t = 3.04, p < 0.01 at Time 1 and t = 2.25, P < 0.05 at Time 2) and wake-up times (t = 2.7, P < 0.05 at Time 1) than subjects in the sleep only group, whereas no such difference was detected at baseline (or followup). The percentage of subjects in the regularity group for whom the regularity of bedtime was 1 hour or less was 84% in the first half of the experimental phase and 74% in the second half. The percentage of subjects for whom the regularity of wake-up time was 1 hour or less was 74% in the first half of the experimental phase and 53% in the second half. Thus, compliance with the regularity instructions was generally good, although it decreased with time, and there was better compliance on regulating bedtime. Caffeine consumption Average reported caffeine consumption (in cups) was 1.56 ± 1.04 at baseline, 1.26 ± 0.71 at Time 1, and 1.35 ± 0.87 at Time 2, with a significant main time effect [F(2, 74) = 5.04, P < 0.05) and no significant group or group by time interaction effects. The percentage of subjects who stayed at or below their individual caffeine upper bound during the experimental phase was 85% for the sleep only group and 84% for the regularity group, and there were no significant changes in caffeine consumption during the two halves of the experimental phase. Morning light exposure Although subjects increased exposure to light gradually, the effect for time was not significant (F = 2.96, P = 0.08), with no significant group or interaction effects. Average morning light exposure (in hours) was 0.55 ± 0.38 at baseline, 0.61 ± 0.50 at Time 1, and 0.63 ± 0.18 at Time 2. There were differences in the degree to which individuals continued to follow experimental instructions at follow-up. Although, as indicated by group

7 438 R. MANBER ET AL. TABLE 2. Sleep variables Sleep only group Baseline Time 1 Time 2 Follow-up TST (hours)*** Nap frequency** Bedtime Wakeup time Onset (hours)*** Efficiency (%)* 7.2:!: :!: :!: :!: :!: :!: :!: 0.69* 0.08 :!: 0.14*** 0.74 :!: :!: :!: :!: :!: 0.68** 0.06:!: 0.14*** 0.75 :!: :!: 1.4* 0.22 :!: :!: :!: :!: 0.18*** 0.93:!: :!: :!: :!: 3.2 TST indicates total sleep time. Onset indicates latency to sleep onset. Clock time is expressed in real numbers representing distance from midnight. Superscripts on group cell values indicate significance levels of differences relative to baseline within each group. Superscripts on variable names indicate main time effect. The F-ratios are based on repeated measures ANOV As, with baseline, Time I, Time 2, and follow-up values as the repeated measures. Significance levels are indicated: * p < 0.05; ** p < 0.01; *** P < The sleep only group had 20 subjects; the regularity group had 19 subjects. averages, all instructions were followed during the experimental phase, only two instructions were followed beyond the experimental phase. One was morning light exposure, which continued to stay above baseline levels at follow-up; the other was napping behavior, which continued to be below baseline level at followup. These are notable facts because subjects were only instructed to follow the experimental instruction during the experimental phase. Daytime sleepiness Changes in sleepiness level across time are depicted in Fig. 2. The ANCOV A with sleepiness at baseline as a covariate revealed a significant between subjects main group effect [F(l, 33) = 4.55, P = 0.04]. A planned comparison revealed that at the end of the experiment and at follow-up, sleepiness levels were significantly lower in the regularity group than in the sleep only group [t(37) = 2.4 for the end of the experiment and t(37) = 2.7 for follow-up; p < 0.05]. Separate repeated measures ANOV AS for each of the two groups revealed a significant main time effect on sleepiness in the regularity group [F(3, 48) = 4.753, P < 0.01] but no such effect in the sleep only group [F(3, 54) = 1.9, p = 0.15]. The results of the hierarchical regression analyses are summarized in Table 3. They suggest that prior TABLE 3. sleep time is a significant predictor for the next-day level of daytime sleepiness, both during the experimental phase and during baseline. Rise time was not found to be a significant predictor at any phase of the study. Four variables were tested for correlation with daytime sleepiness during baseline, yielding a Bonferroni corrected probability of as threshold for significance. These variables were prior total sleep time (r = -0.39, P = 0.013), sleep efficiency (r = 0.10, P = 0.56), regularity of wake-up time (r = -0.02, P = 0.92), and regularity of bedtime (r = 0.32, P = 0.05). In order to identify correlates of change in sleepiness, we adapted a conservative measure of change using a correction by simple regression for error in the baseline measures. To that end, we computed residual scores for sleepiness, regularity of bedtime, regularity of wake-up time, and total nocturnal sleep time at the end of the experiment by using baseline levels of each of these variables as covariates for the corresponding end of experiment values. Bonferroni correction yields a threshold for significance at p = Pearson correlation of these corrected scores indicated that in the regularity group, individuals who reported greater reductions in sleepiness at the end of the experiment relative to baseline had smaller reductions in nocturnal sleep time (r = -.48, P < 0.05). Hierarchical regression analyses Variables Step 1 Individuals Step 2 Rise time Total sleep time,.z = (39 subjects) Baseline phase n = 354 B = , P = 0.52 B = 0.219, P < ,.z = 0.318, F = 25.55, P < 0.01,.z = (20 subjects) Sleep only n = 508 B = , P = 0.61 B = 0.124, P < ,.z = 0.246, F = 11.2, P < 0.01 Experimental phase Regularity n = 496,.z = (19 subjects) B = 0.032, P = 0.49 B = 0.092, p = 0.048,.z = 0.258, F = 8.5, P < 0.05 The set of regression coefficients for each individual are of no theoretical interest and were thus omitted. F-values were computed for the incremental model from step 1 to step 2.

8 SLEEP-WAKE SCHEDULES AND DAYTIME SLEEPINESS 439 TABLE 2. Extended Regularity group Baseline Time 1 Time 2 Follow-up Group by time interaction 7.4 :±: :±: :±: :±: :±: ::': ::': 0.03*** 0.71::': ::': ::': 0.16* ::': ± 0.06*** 0.88 ± ::': ± 0.18*** ** 7.6::': ±0.19* 0.9 ::': ::': ::': ::': 2.3* F = 3.5, P = 0.04 F = 2.3, P = 0.14 F = 0.0, P = 0.96 F = 4.5, P = 0.01 F = 1.6, P = 0.21 F = 0.7, P = 0.49 Sleep parameters The results of the analyses of sleep parameters are summarized in Table 2. To better illustrate the effects of the experimental manipulations on sleep parameters, data are presented for all four time periods for each of the experimental groups separately. As instructed, subjects in the sleep only group slept, on the average, at least 7.5 hours per night. Their average sleep time increased from about 7.3 hours per night at baseline to an average of 7.7 hours per night [F(2, 38) = 7.1, P < 0.001], and they decreased their napping [F(2, 38) = 15.6, P = 0.000], so that total sleep per 24 hours (Fig. 3) remained unchanged [F(2, 38) = 0.03, P = 0.96]. The increase in total nocturnal sleep time was achieved through an overall shift to later wake-up times [F(2, 38) = 3.9, P = 0.03]. These changes in sleep schedules were not accompanied by changes in reported sleep quality [F(2, 38) = 3.1, p = 0.08], latency to sleep onset [F(2, 38) = 0.87, P = 0.42], or time spent awake after sleep onset [F(2, 38) = 0.83, P = 0.43]. Subjects in the regularity group did not change their total amount of sleep at night [F(2, 36) = 0.44, P = 0.61] but reduced the time they spent napping [F(2, 36) = 16.58, P = ]. As a result, as can be seen in Fig. 3, they slept less per 24 hours during the intervention compared to baseline [F(2, 36) = 3.5, p =... '" :0 = SleepOnly Regularity , I r-----r------r----' baseline time 1 time 2 follow up FIG. 3. Changes in total sleep time per 24-hour period during the study. Dispersion measure is the standard deviation.., 0.05]. The changes in the duration and regularity of sleep schedules in the regularity group were accompanied by decreased latency to sleep onset [F(2, 36) = 7.8, P < 0.001] and increased sleep efficiency [F(2, 36) = 4.5, P = 0.01]. DISCUSSION The combination of adequate sleep duration and regular nocturnal sleep schedules was found in the present study to be superior to adequate sleep duration alone in decreasing reported sleepiness and improving sleep efficiency. Even though there was an overall significant reduction in reported daytime sleepiness for all subjects, subjects in the regularity group reported greater and longer-lasting improvements in daytime sleepiness. This improvement was observed despite the fact that during the 4 weeks of the experimental phase, subjects in the regularity group slept about half an hour less per 24-hour period compared to subjects in the sleep only group. Nevertheless, levels of daytime sleepiness continued to be significantly correlated with prior nocturnal sleep time within each group during the experimental phase, suggesting that, as expected, prior sleep time is central to next-day levels of daytime sleepiness. Factors other than the total amount of sleep and its regularity might have influenced the reduction in sleepiness. Examples of these factors include school schedule, exposure to light, time spent in study, etc. However, because all subjects were studied at the same time and because both groups received the same instructions, with the exception of regularity of schedules, it is not likely that these factors affected the groups differentially. One factor that might have affected the groups differentially is expectation. Although an effort was made to present the intervention in neutral terms, expectations were not measured, and thus their contribution to outcome cannot be assessed. In a previous study with a similar protocol, expectations were not correlated with outcome (13). Daytime sleepiness is a complex construct comprised of the subjective experience that people describe as "sleepy" as well as objective components including

9 440 R. MANBER ET AL. the propensity to sleep and its effects on the performance of vigilance tasks. The three aspects are complementary. The present study measured only the subjective aspect of sleepiness; the extent to which the results will generalize to the other two components remains an empirical question. Although there are no reports on the effects of schedule manipulation on the propensity to sleep as measured by multiple sleep latency tests (MSLT), two studies examined the effects of sleep schedules on performance. Taub and Berger (11,17,18) found that regular sleepers whose sleep schedules were altered experienced performance losses relative to baseline and that irregular sleepers had lower performance scores than regular sleepers. These results complement the findings of the present study. On the other hand, Bonnet and Altar (12) manipulated sleep schedules of irregular sleepers to produce very regular schedules and found only a trend towards improved performance. It is conceivable that the adjustment to the sharp change in regularity imposed in their study requires more time and that the trend toward improved performance might have eventually led to significant performance improvements. Future studies should examine the effects of regularizing sleep schedules on both subjective and objective components of sleepiness. The combination of regular and adequate nocturnal sleep length had a positive effect on sleep as well. The reported latency to sleep onset decreased and sleep efficiency increased during the experimental phase in the regularity group. Although these sleep findings were significant, the effect size is small, probably because baseline sleep efficiencies in our sample were already high. It is possible that in a clinical setting, where insomnia and irregular sleep habits often coexist, regularizing sleep schedules will yield greater effect sizes on sleep measures, thus supporting the efficacy of the common clinical recommendation to regularize sleep schedules. The noted improvements in sleep might have been a result of the reduction in the total amount of sleep per 24 hours. An alternative mediating factor might be the possible stabilizing effect of the intervention on the circadian rhythm. This alternative explanation awaits further research. Our results are consistent with recent findings by Akerstedt et al. (19), who found significant negative effects of a shift to irregular sleep schedule on sleep efficiency. In contrast, Bonnet and Alter (12) found no significant change in objective, as opposed to diary, measures of sleep latency or sleep efficiency following regularization of sleep schedules. Because correlations between electroencephalographic (EEG) and diary measures of sleep parameters are, in general, substantial (20), further explanation of the discrepancy between the two studies is needed. Bonnet and Alter (12) may have failed to find a significant relationship between regularity of schedule and objective sleep parameters because only 2 nights, one in each regularity condition, were compared. One night of sleep recording might not be sensitive enough to detect differences. It is not easy for college students who keep irregular sleep schedules to alter their sleep habits. Compliance with the regularity instructions in the present study was higher during the initial 2 weeks compared to the last 2 weeks of the experimental phase. Five weeks after the completion of the experimental phase, in the absence of instructions to continue to follow the experimental instructions, compliance with the regularity instructions, as well as with most other instructions, decreased to baseline levels. In a previous study, Manber and Bootzin (13) found that when college students were instructed to keep a more regular wake-up time and were advised (but not required) to go to bed earlier, they complied with the former but ignored the latter. As a result, the instruction to keep a regular wake-up time, so common in insomnia clinics, produced significant chronic sleep deprivation in the college sample and induced an increase in daytime sleepiness (13). Thus, in order to reap the full benefit of regular sleep patterns, instructions to college students need to address both bedtimes and wake-up times so as to avoid sleep deprivation. Continued support during the transition period might help individuals in this vulnerable age group adhere to more regular sleepwake schedules. Acknowledgments: This manuscript is based on the dissertation work of the first author. Special thanks are given to the dissertation committee members Alfred Kaszniak and Catherine Shisslak and the research assistants July AzimoY, Victoria Oswald, Roy Pardee, Katheryn Pinkart, Camilia Shaheed, and Martha Walton. REFERENCES 1. Billiard M, Alperovitch A. Epidemiology of excessive somnolence in draftees. In: Koella Wp, Obal F, SchulzH, Visser P, eds. Sleep '86. New York: Gustav Fischer Verlag, 1988: Billiard M, Alperovitch A, Perot C, Jammes A. Excessive daytime somnolence in young men, prevalence and contributing factors. Sleep 1987;10: Carskadon MA, Davis SS. Sleep-wake patterns in the highschool to college transition: preliminary data. Sleep Res 1989; 18: Carskadon MA, Dement WC. Daytime sleepiness: quantification of behavioral state. Neurosci Biobehav Rev 1987;11: Levine B, Roehrs T, Zorick F, Roth T. Daytime sleepiness in young adults. Sleep 1988;11: Strauch I, Meier B. Sleep need in adolescents: a longitudinal approach. Sleep 1988; 11 : Carskadon MA, Dement We. Cumulative effects of sleep restriction on daytime sleepiness. Psychophysiology 1981;18:

10 SLEEP-WAKE SCHEDULES AND DAYTIME SLEEPINESS Hicks RA, Mistry R, Lucero K, Lee L, Pellegrini, R. The sleep duration and sleep satisfaction of college students: striking changes over the last decade( ). Percept Mot Skills 1989;68: Carskadon MA. Adolescent sleepiness: increased risk in highrisk population. Alcohol Drugs Driving 1990;5/6: Carskadon MA. Patterns of sleep and sleepiness in adolescents. Pediatrician 1990;17: Taub J, Berger R. Perfonnance and mood following variations in the length and timing of sleep. Psychobiology 1973; 10: Bonnet M, Alter J. Effects of irregular versus regular sleep schedules on perfonnance, mood and body temperature. Bioi Psychol 1982;14: Manber R, Bootzin R. The effects of regular wake-up schedules on daytime sleepiness in college students. Sleep Res 1991 ;20: Huynh H, Feldt LS. Conditions under which mean square ratios in repeated measurements designs have exact F-distributions. J Am Stat Assoc 1970;65: Borgatta E, Bohrnstedt G. Level of measurement: once over again. In: Borgatta E, Bohrnstedt G, eds. Social measurement: current issues. Beverly Hills, CA: Sage Publications, Chronbach LJ, Furby L. How we should measure "change" or should we? Psychol Bull 1970;74: Taub J, Berger R. Acute shifts in the sleep-wakefulness cycle: effects on perfonnance and mood. Psychosom Med 1974;36: Taub J, Berger R. The effects of changing phase and duration of sleep. J Exp Psychol Hum Percept Perform 1976;2: A.kerstedt T, Hume K, Minors D, Waterhouse J. Regulation of sleep and naps on an irregular schedule. Sleep 1993;16: Carskadon MA, Dement WC, Mitler MM, Guilleminault C, Zarcon VP, Spiegel R. Self reports versus sleep laboratory findings in 122 drug-free subjects with complaints of chronic insomnia. Am J Psychiatry 1976; 133: