The Effects of Caffeine on Simulated Night-Shift Work and Subsequent Daytime Sleep

Transcription

1 Sleep. 18(1): American Sleep Disorders Association and Sleep Research Society The Effects of Caffeine on Simulated Night-Shift Work and Subsequent Daytime Sleep Mark J. Muehlbach and James K. Walsh Sleep Medicine and Research Center. St. Luke's Hospital. Chesterfield. Missouri. U.S.A. Summary: Thirty healthy volunteers were randomly assigned to either a caffeine or a placebo group to investigate the alerting effects of caffeine at night. Subjects adhered to a simulated night-shift schedule for 5 consecutive nights. On the first 3 nights, 2 mg/kg caffeine was added to decaffeinated coffee at 2220 and 0120 hours for the caffeine group. On nights 4 and 5 both groups received placebo. Each night, subjects completed five 60-minute sessions of a computerized simulated assembly line performance task (), a multiple sleep latency test () and questionnaires. Daytime sleep was recorded in the laboratory between 0900 and 1700 hours each day following nighttime testing. Caffeine decreased physiological sleep tendency on the night shift compared with placebo; however, the two groups performed at equivalent levels on the. On nights 4 and 5, when both groups received placebo, there were no differences between the groups on the MSL T, suggesting the absence of a discontinuation effect. There were no differences on daytime polysomnograms between the groups. Key Words: Caffeine-Sleepiness-Performance-Shift work. Recent evidence implicating sleepiness in major accidents (1,2) has intensified research of methods to improve alertness. The sleepiness/alertness continuum is influenced by such factors as quantity and quality of sleep, drugs and circadian rhythms (3). Typically, alertness is most compromised during nighttime hours, generally between 0100 and 0700 hours (4). Thus, people who work night-shift hours are very susceptible to inattention and falling asleep on the job. In fact, Akerstedt and Gilberg (5) reported that approximately 20% of night workers fall asleep on the job, and 75% report feeling sleepy during their work shift. The stimulating effects of caffeine have been recognized for centuries. It is estimated that, in the United States, the mean daily intake of caffeine per person is 3 mg/kg (6). This estimate is even higher for coffee drinkers, 4 mg/kg per day. Several studies have demonstrated that caffeine decreases physiological sleepiness and improves performance during daytime hours (7-10). The alerting effects of caffeine when administered prior to nighttime sleep have also been demonstrated (11,12). Accepted for publication July Address correspondence and reprint requests to Mark J. Muehlbach, Ph.D., Sleep Medicine and Research Center, St. Luke's Hospital, 232 Woods Mill Road, Chesterfield, MO 63017, U.S.A. 22 Few studies have investigated the effects of caffeine to promote alertness during nighttime work hours. Borland et al. (13) found that 300 mg caffeine, administered at 2300 hours, improved performance compared to placebo during extended work hours. Walsh et al. (14) reported that caffeine (4 mg/kg), administered prior to a single simulated night shift, significantly decreased physiological sleepiness, as measured by the multiple sleep latency test (MSL T), and improved ability to maintain wakefulness, as measured by the repeated test of sustained wakefulness (RTSW) (14). These effects were seen for both mild (two or fewer caffeinated beverages per day) and moderate (five to seven per day) caffeine users. In another study, caffeine (4 mg/kg) improved performance on a computerized simulated assembly line task () across the night compared to placebo (15). In both ofthe latter studies mentioned, a single dose of caffeine prior to the night shift failed to produce alerting effects at the end of the night shift. The present study was designed to investigate the effects of divided doses of caffeine on physiological sleep tendency and performance on three successive night shifts and on subsequent daytime sleep. Further, we examined sleep tendency and performance the 2 nights following discontinuation of caffeine.

2 ,, CAFFEINE EFFECTS ON NIGHT WORK 23 I., I I I TABLE 1. Mean screening night TST and daytime mean and mean percent correct responses Jor the caffeine group and the placebo Caffeine Placebo Total Sleep Time (minutes) Mean SD MSL T (minutes) Mean SD (%) Mean SD Subjects METHODS Thirty young adults were randomly assigned to one of two experimental groups (caffeine or placebo), which did not differ in age, gender, body weight or habitual caffeine use. The mean age of the caffeine group (eight males, seven females) was 24.7 years (range: years) with a mean body weight of 70.9 kg (range: kg), and the mean age of the placebo group (eight males, seven females) was 23.9 years (range: years) with a mean body weight of70.3 kg (range: S8.S kg). Subjects in the caffeine group reported routinely drinking an average of 1.1 (range: S) caffeinated beverages per day compared with an average of 1.3 (range: ) caffeinated beverages per day for the placebo group. Subjects signed an informed consent form and the experimental protocol was approved by the institutional review board. Subjects were paid for their participation in the study. Screening and training procedures Participants were recruited through media or other advertisements and underwent a medical history and a physical examination. One night of polysomnography (PSG) and a daytime were conducted to screen for sleep disorders and to adapt individuals to the laboratory and experimental procedures. The MSL T was conducted and scored according to established procedures (16). On the same day as the physical examination, subjects received a 3-S-hour training session on a performance measure and other experimental procedures. The is a computer-driven, continuous performance measure designed to be sensitive to variation in sleepiness/alertness and to have face validity with respect to "real world" job demands. The presents subjects with images of electronic circuit boards that pass across a video monitor as objects might travel on an assembly line conveyor belt. The participants perform quality control inspections on each circuit board and use a mouse interface to discard faulty "products" or repair certain types of defective boards. Four variables were identified for analyses: 1) correct responses, the percentage of discardable or repairable boards correctly handled; 2) correction time, the mean time from appearance of a faulty board until appropriate action was taken; 3) nonfaulty items handled, the percentage of nonfaulty boards that were incorrectly discarded or for which a repair was incorrectly attempted; and 4) empty items handled, the ratio of the number of times a repair or discard was attempted with the cursor between boards to the total number of boards which appeared. All subjects were required to demonstrate proficiency on the task as defined as a correct repair/rejection rate of 90% or more during the last two 60-minute blocks of performance training. Those subjects failing this criterion received additional training within a few days to determine if they would qualify for study participation. Subjects participated in four SO-minute sessions for further practice during the day of MSL T screening. Table 1 shows the mean PSG total sleep time (TST) for the screening night and the mean MSL T and the mean percent correct responses for the following day. There were no significant differences between the two groups on any of these variables. Experimental procedures All subjects were instructed to refrain from caffeinated beverages and alcohol the 7 days prior to nights scheduled in the laboratory and throughout the remainder of the study. No central nervous system active medications, including sedatives, painkillers and antihistamines, were allowed for at least 7 days prior to the start of the study and for the duration ofthe study. Following training, screening and adaptation, each subject participated in five consecutive night shifts and five daytime sleep periods in the laboratory. This schedule closely simulates a typical week of night shifts and is similar to those used in previous studies (17,18). Subjects received 300 ml of decaffeinated coffee between 2220 and 22S0 hours and an identical amount between 0120 and 0 ISO hours on all S nights of the study. On the first 3 nights, 2 mg/kg caffeine was added to each serving of decaffeinated coffee for the subjects in the caffeine group. Subjects were instructed to drink the liquid such that it was consumed evenly over the half-hour period. Subjects and laboratory personnel were blind to the conditions. No caffeinated beverages were consumed by either group other than those designated by the protocol. Sleep, Vol. 18, No. I, 1995

3 24 M. J. MUEHLBACH ANDJ. K. WALSH The, and visual analog scales (VAS) for sleepiness/alertness followed the schedule outlined in Table 2 on each of the 5 study nights. The VAS for sleepiness/alertness is a IOO-mm line with one pole labeled "extremely sleepy" and the other pole labeled "extremely alert", on which subjects make a vertical line indicating level of sleepiness/alertness. Subjects had a minimum of a lo-minute break following each sleep latency test (SLT), just prior to the start of each. There was a 25-minute break following each session prior to the initiation of the subsequent SLT. During these break times, subjects were allowed to engage in sedentary activities, such as watching television or reading. Testing procedures for the caffeinated and decaffeinated coffee groups were conducted under normal room lighting conditions. Subjects were observed by technicians during the night to assure conformity to the study guidelines. Following each night shift, subjects in both conditions were exposed to 5,000-7,500 lux (measured by a Luna-Pro F light meter, Gossen, Germany) bright light for 30 minutes between 0830 and 0900 hours to simulate (and control) early morning light exposure, which typically occurs when night workers commute home. By artificially providing this light exposure, we could minimize the phase delay in circadian rhythms noted during prior studies that prevented morning exposure to sunlight (18). Breakfast was also provided between 0830 and 0900 hours. Daytime PSG began at 0900 and was terminated 8 hours later on each of the 5 days of the study for the two conditions. Sleep was scored according to standard criteria (19). All subjects slept in rooms with no windows to avoid exposure to sunlight. Subjects were allowed to leave the laboratory at approximately 1730 hours, following removal of electrodes and completion of postsleep questionnaires, and were asked to return by 2200 hours on nights scheduled for testing. Statistical analyses data underwent either reciprocal transformations (correction time) or arcsine transformations for proportions (20) prior to analysis. Treatment effects (nights/days 1-3) and discontinuation effects (nights/ days 4-5) were analyzed independently. All data were analyzed with multivariate analyses of variance (MANOV A) with factors for group, night (or day), time of night and their interactions, using the multivariate general linear hypothesis module of SYST A T (SYS TAT Inc., Evanston, IL, U.S.A.). Follow-up comparisons using analyses of variance (ANOVA) were made when statistically appropriate. In general, only significant effects are reported and discussed. Sleep, Vol. 18, No.1, 1995 TABLE hours 2330 hours 0100 hours 0130 hours 0300 hours 0330 hours 0500 hours 0530 hours 0700 hours 0730 hours 0830 hours 0900 hours Schedule for both groups on each of the 5 nights a Breakfast Polysomnography a VAS for sleepiness/alertness administered 5 minutes prior to each MSL T subtest. RESULTS The caffeine group received a mean of 142 mg (range: mg) of caffeine at each nightly administration (2220 and 0120 hours) on the first 3 nights. This is approximately equivalent to one 6-ounce cup of drip coffee at each administration. Mean MSL T sleep latencies for the two groups during caffeine administration are shown in Fig. 1. Analysis of the, collapsed across the first 3 nights, revealed the placebo group was significantly more sleepy compared with the caffeine group [F(l,28) = 5.7, p < 0.03]. Mean sleep latencies were 3.3 minutes, 2.1 minutes and 3.8 minutes longer on nights 1, 2 and 3, respectively, for the caffeine group compared with the placebo group. There was a significant main effect for night [F(2,27) = 5.7, p < 0.02]. From night 1 to night 3, the mean increased 3.0 minutes for the caffeine group and 2.5 minutes for the placebo group. A main effect for time of night was also detected [F(4,25) = 35.6, p < 0.001]. Physiological sleepiness increased as the night progressed regardless of group or night. However, a group by time of night interaction was found [F(4,25) = 3.61, p < 0.02] (see Fig. 2). Compared with the placebo group, the caffeine group took significantly longer to fall asleep at 2300 hours (p < 0.05) and 0300 hours (p < 0.05) and approached significance to take longer to fall asleep at 0500 hours (p < 0.06) and 0700 hours (p < 0.1). There were no differences between the caffeine and placebo groups on nights 4 and 5. Mean sleep latencies were 8.7 minutes and 9.7 minutes for the caffeine group and 10.0 minutes and 11.2 minutes for the placebo group on nights 4 and 5, respectively. Mean sleep latencies were longer on night 5 compared with night 4 regardless of group [F(2,27) = 3.9, p < 0.06]. Physiological sleepiness increased as the night progressed for both groups [F(4,25) = 3.9, p < 0.02]. I j

4 CAFFEINE EFFECTS ON NIGHT WORK a N 15 V> Q) -0- V> :; Caffeine 10 2 c --Placebo :J c 5 II) * -0- Caffeine --Placebo N a 23:00 1:00 3:00 5:00 7:00 nmecln FIG. 1. Mean (and SE) latencies for the caffeine group and FIG. 2. Mean (and SE) of each MSL T subtest for the caffeine group placebo group on nights 1-3. and placebo group collapsed across nights 1-3. * p < l. Subjective sleepiness/alertness Low values on the V AS represent relative subjective sleepiness and high values indicate relative subjective alertness. There was a significant group by night interaction detected for the first 3 nights [F(2,27) = 5.62, p < 0.01] (see Fig. 3). Follow-up analyses of the individual nights revealed that the caffeine group rated themselves significantly more alert on night 1 than the placebo group [F(I,28) = 4.99, p < 0.05]. There were no group differences on nights 2 and 3. There was a main effect for night [F(2,27) = 8.43, p < 0.001], indicating a general increase in subjective alertness from night 1 to night 3 for both groups. A significant timeof-night effect was also found [F(4,25) = 59.91, p < 0.001]. Both groups rated themselves progressively sleepier across the night. There were no group differences on nights 4 and 5 for subjective sleepiness/alertness. Ratings were 54.2 and 62.6 for the caffeine group and 56.5 and 66.6 for the placebo group on nights 4 and 5, respectively. Subjective alertness increased significantly for both groups from night 4 to night 5 [F(I,28) = 9.3, p < 0.005], yet alertness ratings decreased linearly from 2300 to 0700 hours for both groups on each night [F(4,25) = 9.7, p < 0.001]. Means and SD for the four variables on each of the 5 nights for both groups are shown in Table 3. Analyses of correct responses across the first 3 nights revealed a significant group by night interaction [F( 4,25) = 4.02, p < 0.05] (see Fig. 4). There was a nonsignificant trend for the caffeine group to make more correct responses on night 1 compared with the placebo group [F(1,28) = 3.17, p < 0.09]. There were no group differences on nights 2 and 3 for the number of correct responses. Significant main effects for night [F(4,25) = 3.69, p < 0.05] and time of night [F(4,25) = 15.32, p < 0.001] were found. In general, performance improved from nights 1 to 3, although performance continued to decrease across each night. There were no group differences for the number of correct responses on nights 4 and 5. However, on nights 4 and 5 there was a significant effect for night [F(1,28) = 10.22, p < 0.01], with better performance on night 5 compared with night 4, and a significant time-of-night effect [F(4,25) = 4.55, p < 0.01), with performance decreasing as the night progressed. There was no significant difference between the two groups for the length of time to make corrections on the. However, the caffeine group was able to make corrections nearly 1 second more quickly than the placebo group on night 1. There was a main effect for time of night [F(4,25) = 14.23, p < 0.001]; correction time generally increased across the night for both groups. There were no significant group differences found for nights 4 and 5. Similar to the results for correct responses, main effects for night [F(1,28) = 4.35, p < 0.05] and time of night [F(4,25) = 10.25, p < 0.001] were found for nights 4 and 5, regardless of group. Subjects made corrections more quickly on night 5 compared to night 4, yet correction time decreased as the night progressed. The caffeine group made significantly fewer attempts to make corrections on nonfaulty items across the first 3 nights compared with the placebo group [F(1,28) = 6.0, p < 0.03]. The number of attempts on nonfaulty items increased during the night for both groups [F( 4,25) = 5.0, p < 0.01] on all 3 nights. There were, however, relatively few of these errors on any night. There were no significant effects found for nights 4 and 5. There was a significant group effect for the number oftimes subjects responded with the cursor in between circuit boards during the first 3 nights [F(1,28) = 4.47, p < 0.05]. The placebo group made nearly twice the number of these errors on each of the first 3 nights compared with the caffeine group. However, the actual proportion of these errors was quite small. There were no significant effects for nights 4 and 5. Sleep. Vol. 18, No.1, 1995

5 26 M. J. MUEHLBACH ANDJ. K. WALSH TABLE 3. Mean and SD for the four performance variables for the caffeine group and the placebo group on each of the 5 nights Caffeine group Placebo group Night I Night 2 Night 3 Night 4 Night 5 Night I Night 2 Night 3 Night 4 Night 5 Correct responses (%) Mean SD Correction time (seconds) Mean SD Nonfaulty items (%) Mean SD Empty items (%) Mean SD Daytime PSG Mean daytime PSG data for TST, sleep latency (SL), total wake time (TWT) and awakenings > 15 seconds are shown in Table 4. Analyses of the daytime PSGs revealed no significant differences between the caffeine group and the placebo group for TST, TWT, SL, the number of awakenings > 15 seconds, as well as stage 1 sleep, stage 2 sleep, stage 3-4 sleep and REM sleep, over the 3 days of caffeine administration. Even though there was no group difference for TST, the caffeine group slept a mean 62.6 minutes less on day 1 than the placebo group. However, 20 minutes of this mean difference was accounted for by a single subject. This subject slept a total of only 25.5 minutes on day 1. There were no group differences for any of the PSG variables on days 4 and 5. Subjects in both groups took significantly longer to fall asleep on day 5 compared to day 4 [F(1,28) = 7.44, p < 0.02J. No other significant effects were found on days 4 and 5. Postsleep Questionnaire (PSQ) Mean data from the postsleep questionnaire are shown in Table 5. Analyses revealed a significant group difference for subjective amount of sleep (SAS) [F(1,28) = 4.38, p < 0.05J collapsed across the first 3 days. The placebo group reported an average of over 34 minutes more sleep each of 3 days compared with the caffeine group. The largest difference in subjective amount of sleep was on the first sleep period when the caffeine group reported 67 minutes less sleep than the placebo group. However, 27 minutes were accounted for by the one subject who had only 25.5 minutes PSG TST. Thus, the mean difference in SAS for the two groups, excluding this one subject, was 40 minutes, which was still significant (p = 0.05). There were no significant effects for the postsleep questionnaire items on days 4 and 5. DISCUSSION These results indicate that 4 mg/kg of caffeine administered on 3 consecutive nights, in two equal doses (at 0220 and 0120 hours) each night, decreases physiological sleepiness during nighttime hours and, at least initially, enhances performance with no significant consequences on PSG measurement of daytime sleep. Discontinuation of caffeine did not result in any significant effects. These findings are consistent with other studies conducted in our laboratory, which show that caffeine administration improves performance and decreases physiological sleepiness during a single simulated night shift (13,14). During the first night shift, the and V AS for sleepiness/alertness were significantly improved for the caffeine group over the placebo group, and all other measures were in the same direction. Multiple sleep latency test data revealed that caffeine significantly decreased physiological sleepiness com- 80 E 70 E *...d 60. -o-caffeine Ol Placebo 7 c 0:: Night FIG. 3. Mean V AS for sleepiness/alertness for the caffeine group and placebo group on nights 1-3. * p < Sleep. Vol. 18. No. I. 1995

6 CAFFEINE EFFECTS ON NIGHT WORK 27 TABLE 4. Mean and SD daytime polysomnographic variables for the caffeine group and the placebo group on each of the 5 days Caffeine group Placebo group Day 1 Day 2 Day 3 Day 4 Day 5 Day 1 Day 2 Day 3 Day 4 Day 5 Total sleep time (minutes) Mean SD Total wake time (minutes) Mean SD Sleep latency (minutes) Mean SD Number of wakes (> 15 seconds) Mean SD ,. pared to placebo during caffeine administration. Mean sleep latencies were > 10 minutes on each of the first 3 nights for the caffeine group and < 10 minutes for the placebo group on those same nights. Furthermore, on night 1, scores for the caffeine group at 0500 and 0700 hours were nearly twice those ofthe placebo group. By night 3, mean latencies for the caffeine group were> 5.5 minutes and> 4.0 minutes longer at 0500 and 0700 hours, respectively. Consistent with the night 1 MSL T data, the caffeine group rated themselves 25% more alert than the placebo group on night 1. Subjects were able to recognize the alerting effects of caffeine during the first night, when placebo group alertness was most compromised. However, there were no subjective differences between the two groups on nights 2 and 3, even though mean MSL T latencies were longer for the caffeine group. Discrepancies between subjective and objective data are not surprising. Walsh et al. (14) reported that subjects did not feel more alert following caffeine administration compared to placebo, despite a marked effect on objective measures. Thus, subjects may not always appreciate the effects of treatment. The placebo group committed nearly twice as many (; -<>- Caffeine () 90 C -Placebo Q) f::' Q) a Night FIG. 4. Mean (and SE) correct responses for the caffeine group and the placebo group on nights 1-3. errors (correction attempts on empty items and nonfaulty boards) as the caffeine group during the first 3 nights on the performance measure. The improvement in correct responses on the was not significant, yet the caffeine group was able to perform approximately 4% better than the placebo group on night 1. Furthermore, despite a decrease in performance across the night for both groups, the caffeine group was able to make nearly 11 % more correct responses during the 0730-hour session on night 1, compared with the placebo group. There were virtually no differences between the two groups on nights 2 and 3. This may be due to a combination of two factors. First, all subjects had to perform at 90% or better to qualify for the study. Screening data showed that subjects in both groups were able to make over 95% correct responses prior to entry into the study, decreasing the potential for practice effects. However, this may have created a ceiling effect. Subjects simply could not perform above 100%. Second, nights 2 and 3 followed daytime sleep, thus physiological sleepiness was not as compromised as it was during night 1. In general, performance is influenced more by motivation and knowledge of task when physiological sleepiness is low. In the current study, task knowledge was very high. Assuming that motivational factors were equivalent between the two groups, the daytime sleep may have been enough for the placebo group to overcome significant performance decrements during the night. All nocturnal measures demonstrated a significant time-of-night effect regardless of group. This suggests that the titration of caffeine in the two doses was unsuccessful in preventing the drop in alertness and performance at the circadian trough. The failure of caffeine to maintain the measures across the night demonstrates the strength of the circadian rhythm. However, it may be possible that a larger dose of caffeine can Sleep. Vol. 18. No

7 28 M. J. MUEHLBACH ANDJ. K. WALSH TABLE S. Mean and SD postsleep questionnaire items for the caffeine group and the placebo group on each of the 5 days Caffeine group Placebo group Day I Day 2 Day 3 Day 4 Day 5 Day I Day 2 Day 3 Day 4 Day 5 Subjective amount of sleep (minutes) Mean SD Subjective sleep latency (minutes) Mean SD Il.l Subjective number of awakenings Mean SD overcome the circadian drop. Recent findings suggest that a single large dose of caffeine (400 mg) may be more effective than multiple administrations of smaller doses (150 mg) to improve performance and alertness during sleep deprivation (21). On the other hand, multiple administrations of small doses of caffeine ( mg) in combination with prophylactic naps, has been shown to be more effective for improving performance and alertness during sleep loss than a single large dose of caffeine (400 mg) alone (22). Polysomnographic sleep was not significantly affected by caffeine. This is not surprising because the dose used in this study should have been nearly completely metabolized by the time subjects went to bed. In a recent study, Landolt et al. (23) reported an increased sleep latency and decreased sleep efficiency at night approximately 16 hours after ingestion of 200 mg caffeine, even though caffeine levels were less than 3 Jlmol/l. In the current study, subjects in the caffeine group slept 62.2 minutes less on day 1 than the subjects in the placebo group. Even though this was not significant, the data are in the direction that suggests a detrimental effect of caffeine. However, 20 minutes of the mean difference on day 1 are accounted for by one subject in the caffeine group. This may be related to prior caffeine ingestion, the rotation of the sleep/wake schedule or possibly a combination of the two. Furthermore, TST on days 2 and 3 for the caffeine group were much nearer to those of the placebo group. On the other hand, subjectively, subjects in the caffeine group reported significantly less sleep than the placebo group. Yet again, a large portion of the mean difference was accounted for by one subject. There were no differences between the caffeine and placebo groups on any of the dependent measures during the last 2 nights, when both groups received placebo, indicating no significant effects of caffeine discontinuation. That is, neither a conditioned response nor a "rebound" sedation developed as a result of prior caffeine ingestion. This study indicates that caffeine, at commonly used daily doses consumed at the appropriate time, reduces sleepiness and performance decrements during nighttime work hours. Further, sleep at times common for night workers was not markedly disturbed by nocturnal caffeine consumption in this study. Finally, abrupt discontinuation of nighttime use of caffeine did not produce untoward effects on subsequent nights. REFERENCES I. Mitler GS, Carskadon MA, Dement WC, Dinges DF, Graeber RC. Catastrophes, sleep and public policy: consensus report. Sleep 1988; 11: Wake Up America: A National Sleep Alert. Report of the National Commission on Sleep Disorders Research. Washington, DC: U.S. Government Printing Office, January Roth T, Roehrs T, Zorick F. Sleepiness: its measurement and determinants. Sleep 1982;5:S Richardson GS, Carskadon MA, Orav EJ, Dement WC. Circadian variation of sleep tendency in elderly and young subjects. Sleep 1982;5:S Akerstedt T, Gilberg M. The circadian variation of experimentally displaced sleep. Sleep 1981;4: Barone JJ, Roberts H. Human consumption of caffeine. In: Dews PB, ed. Caffeine: perspectives from recent research. Berlin: Springer-Verlag, 1984: Johnson LC, Spinweber CL, Gomez SA. Benzodiazepines and caffeine: effect on daytime sleepiness, performance and mood. Psychopharmacology 1990; I 0 I: Lieberman HR, Wurtman RJ, Emde GG, Roberts C, CovieJla ILG. The effects oflow doses of caffeine on human performance and mood. Psychopharmacology 1987;92: Lumley M, Roehrs T, Asker D, Zorick F, Roth T. Ethanol and caffeine effects on daytime sleepiness/alertness. Sleep 1987; 10: Zwyghuizen-DoorenbosA, Roehrs TA, Lipschutz L, Timms V, Roth T. Effects of caffeine on alertness. Psychopharmacology 1990; 100:36-9. II. Karacan I, Thornby JI, Anch AM, Booth GH, Williams RL, Salis PJ. Dose-related sleep disturbances induced by coffee and caffeine. Clin Pharmacol Ther 1976;20: Nicholson AN, Stone BM. Heterocyclic amphetamine derivatives and caffeine on sleep in man. Br J Clin Pharmacol1980; 9: Borland RG, Rogers AS, Nicholson AN, Pascoe PA, Spencer MB. Performance overnight in shiftworkers operating a daynight schedule. Aviat Space Environ Med 1986;57: Sleep, Vol. 18, No.1, 1995

8 CAFFEINE EFFECTS ON NIGHT WORK Walsh JK, Muehlbach MJ, Humm TM, Dickins QS, Sugerman JL, Schweitzer PK. Effect of caffeine on physiological sleep tendency and ability to sustain wakefulness at night. Psychopharmacology 1990;101: IS. Schweitzer PK, Muehlbach MJ, Walsh JK. Countermeasures for night work performance deficits: the effects of napping or caffeine on performance at night. Work and Stress 1992;6: Carskadon MA, Dement we, Mitler MM, Roth T, Westbrook PR, Keenan S. Guidelines for the multiple sleep latency test (): a standard measure of sleepiness. Sleep 1986;9: Walsh JK, Schweitzer PK, Anch AM, Muehlbach MJ, Jenkins NA, Dickins QS. Sleepiness/alertness on a simulated night shift following sleep at home with triazolam. Sleep 1991; 14: Walsh JK, Sugerman JL, Muehlbach MJ, Schweitzer PK. Physiological sleep tendency on a simulated night shift: adaptation and effects oftriazolam. Sleep 1988;11: Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring systems for sleep stages of human subjects. Los Angeles: Brain Information Service/Brain Research Institute, Winer BJ. Statistical principles in experimental design. New York: McGraw-Hill, Kelly T, Gomez S, Engelland S, Naitoh P. Repeated administrations of caffeine during sleep deprivation does not affect cognitive performance. Sleep Res 1993;22: Bonnet MH, Gomez S, Wirth 0, Arrand DL. Dose-response effects of prophylactic naps and caffeine on sleep latency and performance during sleep loss. Sleep Res 1994;23: Landolt HP, Werth E, Borbely AA, Dijk DJ. Effect of caffeine (200 mg) administered in the morning on sleep and EEG power spectra at night. Sleep Res 1994;23:68. Sleep, Vol. 18. No. I, 1995