Daytime Carryover of Triazolam and Flurazepam in Elderly Insomniacs

Transcription

1 Sleep, 5(4) Raven Press, New York Daytime Carryover of Triazolam and Flurazepam in Elderly Insomniacs Mary A. Carskadon, Wesley F. Seidel, *David J. Greenblatt, and William C. Dement Sleep Research Center, Stanford University School of Medicine, Stanford, California, and *Division of Clinical Pharmacology, New England Medical Center Hospital, Boston, Massachusetts Summary: The effects of triazolam (0.25 mg) or flurazepam (15 mg) were evaluated in 13 elderly (ages 64-79) insomniacs. Subjects were reasonably healthy, ambulatory, and complained of disturbed sleep. Sleep apnea syndromes were ruled out by nocturnal polysomnogram. Sleep, daytime sleepiness [Multiple Sleep Latency Test (MSL T) and Stanford sleepiness scale (SSS»), performance, and mood [Profile of Mood States (POMS)] were assessed on five consecutive days. Placebo was given on nights 1 and 2; active medications were given on nights 3-5. Sleep time was increased by approximately 1 h in both groups. MSLT showed increased sleepiness with flurazepam and decreased sleepiness with triazolam. MSLT scores were unrelated to nocturnal sleep parameters in the flurazepam group and showed a pattern of correlation, though nonsignificant, in the triazolam group. Vigilance was impaired with flurazepam and unchanged with triazolam. Other performance tests showed slight improvement or no change with drugs. Mood tended to be improved after flurazepam ingestion and unchanged after triazolam. These findings suggest that, although both compounds improve nocturnal sleep in elderly insomniacs, there is significant residual sedation with flurazepam and improved daytime alertness with triazolam. Neither compound had a significant effect on nocturnal respiration in these non-sleep-apneic elderly subjects. Key Words: Sleeping pills-elderly-insomnia-sleepiness. Sleep laboratory evaluations of hypnotic compounds have traditionally focused on the measurement ofnoctumal sleep stage changes in normal or insomniac adult subjects. As such, they have overlooked a large segment of the population who receive prescription sleeping pills and have also failed to assess the effects of these compounds on waking life. The 1979 Institute of Medicine study (1) of sleeping pills and insomnia noted that approximately one-third of sleeping pill prescriptions are written for persons aged 60 or older, who comprise only 15% of the popula- Accepted for publication August Address correspondence and reprint requests to Mary A. Carskadon, Ph.D., Sleep Research Center, Department of Psychiatry and Behavioral Science, Stanford University School of Medicine, Stanford, California

2 362 M. A. CARSKADON ET AL. tion. In many cases, these prescriptions include benzodiazepine hypnotics, which have received relatively little testing in older patients. The elderly represent a special class of insomnia patients: complaints of sleep disturbance are more prevalent than in younger individuals, and objective confirmation ofthe complaints is more often demonstrable in the sleep laboratory (2-4). As many as 35% of the elderly may have a sleep-related breathing abnormality that may be causally related to disturbed sleep (5,6). In addition, the elderly are particularly susceptible to toxic reactions or carry-over effects of hypnotics (7), and tend to show a longer plasma half-life of certain benzodiazepine hypnotics (8). In general, the complaint of disturbed sleep or insomnia is accompanied by or associated with a complaint of daytime malaise, fatigue, or drowsiness (9). If the disturbed sleep and daytime complaint are directly related, then an effective sleeping pill should improve not only nocturnal sleep, but also daytime wellbeing. Such improvement is difficult to demonstrate, and most hypnotic efficacy studies that have assessed waking behavior have been more concerned with the presence or absence of residual sedation than with a demonstrated improvement of daytime function. In the present study, we have examined sleep and waking behavior in a narrowly defined population of elderly insomniacs. Daytime sleepiness/alertness was evaluated using the multiple sleep latency test, which in noncomplaining young adults has shown improved alertness with as little as 60 min change in total sleep time (10). We hypothesized that subjects receiving flurazepam, whose active metabolite (N -desalkylflurazepam) has a half-life of h (8), would show increasing sleepiness on this measure, regardless of changes in total sleep time. In subjects receiving triazolam, with an active half-life of 5-10 h (11), we predicted that sleepiness would not change or would improve if sleep at night were improved. METHODS Subjects Subjects were 13 noninstitutionalized, ambulatory volunteers who were diagnosed as insomniac based on history, physical examination, and nocturnal sleep recording. All received a diagnosis according to the Association of Sleep Disorders Classification of Sleep and Arousal Disorders (1979). The diagnoses included persistent psychophysiological DIMS in 10, DIMS associated with periodic leg movements (nocturnal myoclonus) in 2, and subjective DIMS complaint without objective findings in 1. Sleep apnea syndromes were ruled out in every case by screening with nocturnal sleep recordings. Volunteers were excluded if they reported previous adverse reactions to benzodiazepines. Physical examination and laboratory tests screened out subjects with serious medical illness. Subjects were randomly assigned to receive flurazepam, 15 mg (one male, five females; ages 67-72), or triazolam, 0.25 mg (three males, four females; ages 64-79). All procedures were approved by the Stanford University Committee for the Protection of Human Subjects, and each subject read and signed a statement of informed consent. A fee of $300 was paid to each subject upon completion of the study. Sleep. Vo/. 5. No

3 SLEEPING PILLS IN ELDERLY INSOMNIACS 363 Procedures Subjects lived at the Stanford summer sleep laboratory in groups of four or five from Sunday evening until Saturday morning. The three nighttime sleep conditions were placebo (nights 1 and 2), treatment (nights 3-5), and withdrawal (night 6). Daytime tests followed the placebo and treatment nights. The withdrawal night was included because we were uncertain of the extent of carryover and did not wish to send subjects home until approximately 36 h after the last dose of active medication. We made no attempt to assess withdrawal effects, since our focus was on daytime carryover. Matching capsules containing placebo or active medication were given with water on nights 1-5 at 10 min before lights out. The drug group was double-blind, although experimenters were aware on which nights the active compounds were given.. Individual rooms were used for sleep recordings and performance tests. Rooms were darkened for all sleep periods. Between tests, subjects were free to take walks, play games, or watch television. Each subject was continuously accompanied by a staff member. An orientation that included a demonstration and practice with all forms and tests was given Sunday evening. The experimental protocol included a fixed nightly bedtime from 22:30 to 08:00. During these hours, subjects were required to stay in bed (except for trips to the bathroom) and instructed to try to sleep. All meals were prepared at the sleep laboratory, and subjects were not permitted to consume alcoholic or caffeinated beverages. Nocturnal sleep recordings included EEG (referential central and occipital leads), EOG from right and left outer canthi, and chin EMG recorded in the standard manner (12). Additional variables recorded each night included nasal thermistor, abdominal or thoracic mercury-filled capillary strain gauges, electrocardiogram, and EMG from left and/or right anterior tibialis muscles. Sleep stage determinations were made in 30-s epochs according to standard criteria (12), including a 75-J.t V amplitude criterion for stages 3 and 4 sleep. Respiration during sleep was coded for apneas (respiration pauses > 10 s) and reduced amplitude breathing (amplitude of airflow and effort recordings reduced by > 50% for> los and followed by EEG arousal). These two measures were summed as sleeprelated respiration events. Periodic leg movements during sleep (13) were evaluated for those movements associated with EEG arousal and for total movements. Daytime measures included the Multiple Sleep Latency Test (MSLT), Stanford Sleepiness Scale (SSS) (14), morning and evening Profile of Mood States (POMS) (15), and two performance testing batteries each day. The MSLT was given at 2-h intervals from 09:30 to 19:30. Subjects were requested to lie quietly, close eyes, and try to fall asleep in their darkened bedrooms. Tests were ended after three consecutive epochs of sleep or after 20 min if the three-epoch criterion was not met. The score for each sleep latency test was the elapsed time from lights out until the first epoch of stage 1 sleep. The SSS was given at 30-min intervals every day, including immediately before each sleep latency test. The POMS was given at 10:00 and 21:30. Performance testing batteries were given at 10:05 and 14:00. Tests were administered in the same order in each battery. The first test in each battery was a modified and abridged version of the Williams Word Memory Test (16). For each Sleep, Vol. 5, No.4, 1982

4 364 M. A. CARSKADON ET AL. test, a list of 10 randomly selected four-letter words was chosen at random from 24 sets. Each word was typed on a card and shown to the subjec.t as the words were read and spelled aloud at lo-s intervals. Subjects wrote each word on a separate page of a small tablet and reviewed the list one time before it was removed. Subjects were then required to write, in any order, as many of the words as they could remember in 5 min. This test was scored for the number of words recalled and the number of false positive responses. The second test in each battery involved sorting 40 standard bridge playing cards (minus face cards) into piles according to four different sets of instructions (17). First, cards were sorted into four piles according to suit, and then into 10 piles according to number. The last two sortings were performed with cards facedown, into four piles and then 10 piles. Subjects were instructed to perform the sorting as quickly as possible, and each trial was scored for elapsed time. The third test was the digit-symbol substitution task (18). Subjects were given 90 s to complete as many problems as possible. Five alternate forms, with differing symbol keys, were used. The test was scored for the number of substitutions performed in 90 s. The final test of each battery was a modified and abridged form ofthe Wilkinson Auditory Vigilance test (19), performed by use of a program written for the Apple II computer (WFS). The test lasted 30 min and included 600 tones presented at 3-s intervals. The majority of tones were 0.5 s long; a I-s tone was presented at random within each block of 30 tones. Subjects were asked to respond to the longer tones by pressing the space-bar on the keyboard. A I-min practice test with a higher ratio of target tones was given immediately before each test. The auditory vigilance test was evaluated using two scores: the number of missed responses and the number of false positive responses. Blood samples were drawn by venipuncture at approximately 20:00 on days 2-5. Plasma was separated and frozen until the time of analysis. Both drugs were quantified in plasma samples by electron-capture gas-liquid chromatography using the technique of Greenblatt and colleagues (8,20). Analysis of variance was performed for each drug group across two conditions, placebo and treatment. Data from the withdrawal night were not included, because this condition confounded the effects of drug withdrawal with the effects of receiving no medication. Performance test results were analyzed as daily mean scores for each subject. The SSS was evaluated only for ratings given before each sleep latency test. RESULTS Plasma drug levels In all samples from subjects receiving triazolam, triazolam concentrations were undetectable (less than 0.2 ng/ml). Concentrations of N -desalkylflurazepam among flurazepam recipients showed an increase with successive days of treatment, with the following mean daily plasma levels: drug day 1 = 15.7 ng/ml; drug day 2 = 29.3 ng/ml; drug day 3 = 34.8 ng/m!. Sleep. Vol. 5, No.4, 1982

5 SLEEPING PILLS IN ELDERLY INSOMNIACS 365 Nocturnal sleep Table 1 lists the nocturnal sleep measures for the two conditions of each group. Significant treatment effects in the flurazepam group included increased stage 2 sleep time and total sleep time, as well as a decrease in the number of arousals (of 15 s or longer). In addition, there was a nonsignificant tendency for wake time, wakefulness after the onset of sleep, and stage 2 latency to be reduced during the treatment condition. Nocturnal sleep in the triazolam group was similarly affected by treatment. Total wake time and stage 2 latency were significantly reduced, and there was a nonsignificant trend for the reduction of wakefulness after sleep onset. The trend for a reduction of the number of arousals with treatment was significant when calculated as arousals per hour of sleep. In addition, stage 2 sleep, slow-wave (stages 3 and 4) sleep, and total sleep time increased significantly with triazolam. The increase in slow-wave sleep time resulted from an increase of stage 3 sleep from a placebo mean of 20 min to 30 min with treatment. It should also be noted that in this highly selected group of subjects, neither compound had a significant effect on the number of sleep-related respiratory TABLE 1. Nocturnal sleep Flurazepam Triazolam Measure Placebo Treatment Placebo Treatment Total wake time 170 ± ± ± ± 85 a Wake after sleep onset 110 ± ± ± ± 75 Stage 1 latency 24 ± ± ± ± 14 Stage 2 latency 43 ± ± ± ± 18 a Stage 1, min 38 ± ± ± ± 17 Stage 2, min 237 ± ± 52" 169 ± ± 67 a Slow-wave, min 25 ± ± ± ± 31 a REM, min 88 ± ± ± ± 34 Total sleep time 387 ± ± 86" 317 ± ± 85 a REM latency 78 ± ± ± ± 48 Number of arousals 23 ± 8 18 ± 11 a 35 ± ± 15 Number of shifts to stage 1 11 ± 5 14 ± 6 13 ± 6 14 ± 8 Number of arousals per h of sleep 3.6 ± ± ±3.l 5.0 ± 3.0" Shifts to stage 1 per h of sleep 1.7 ± ± ± ± 1.4 Number of apneas 10.6 ± ± ± ± 14.4 Number of apneas per h of sleep 1.9 ± ± ± ± 2.5 Number of respiratory events/h of sleep 2.6 ± ± ± ± 2.7 Number of leg movements per h of sleep 24.3 ± ± ± ± 48.2 Number of leg movements w. arousallh sleep 8.6 ± ± ± ± 13.8 Values are means ± SD. a Significant difference between placebo and treatment (p <.05). Sleep, Vol. 5, No.4, 1982

6 366 M. A. CARSKADON ET AL. events. Arousals associated with nocturnal periodic leg movements were also not significanily affecied by flurazepam ami iriazoiam. Daytime sleepiness Table 2 summarizes the findings from the multiple sleep latency tests and SSS ratings. The mean daily MSLT score showed a significant difference for both groups with treatment; however, the change in MSLT score was in the opposite direction for the two treatments. Thus, subjects taking flurazepam became sleepier (lower MSLT score) on treatment, while subjects taking triazolam were less sleepy (higher MSLT score) on treatment days. As illustrated in Fig. 1, significant differences were apparent for each group at three times of day. In the flurazepam group, subjects were significantly more sleepy at 13:30, 17:30, and 19:30. Several subjects fell asleep virtually instantaneously on the 13:30 test on one or two occasions. Subjects in the triazolam group were significantly less sleepy on tests at 13:30, 15:30, and 19:30 following treatment. In both groups, a U-shaped MSLT profile was present during placebo and treatment, with greatest sleepiness in the afternoon regardless of condition. There was also a slight, nonsignificant, trend for MSL T scores to increase on consecutive drug days in the flurazepam group. Self-rated sleepiness on the SSS tended to decrease in both groups, although the reduction was significant only in the triazolam group. In the flurazepam group, SSS ratings were significantly lower than those on the placebo for ratings given at 15:30 on treatment days. For triazolam subjects, the SSS ratings at 13:30 and 15:30 were significantly lower during treatment than for placebo. To determine whether the MSLT changes might be associated with specific changes in nocturnal sleep, correlation coefficients were computed for each subject between daily mean MSLT scores and selected nocturnal sleep parameters. Mean correlation coefficients (averaged across subjects within groups) are presented in Table 3. Although none of the individual correlations achieved statistical significance (zero-mu t-test), there is a marked tendency in the triazolam group for high values to be associated with nocturnal measures that would, a priori, seem to TABLE 2. Daytime sleepiness Flurazepam Triazolam Measure Placebo Treatment Placebo Treatment Daily mean SL T 9.8 ± ± 4.5 a 13.3 ± ± 3.6 a Daily mean SSSb 3.1 ± ± ± ± 0.9 a 09:30 SSS 3.0 ± ± ± ± :30 SSS 3.2 ± ± ± ± :30 SSS 3.0 ± ± ± ± 1.0 a 15:30 SSS 3.5 ± ± 1.0 a 3.0 ± ± 0.9 a 17:30 SSS 2.9 ± ± ± ± :30 SSS 2.8 ± ± ± ± 1.2 Values are means ± SD. a Significant difference between placebo and treatment (p <.05). b SSS daily mean computed from ratings given immediately before each SLT. Sleep, Vol. 5, No.4, 1982

7 SLEEPING PILLS IN ELDERLY INSOMNIACS (A) FLURAZEPAM GROUP (N=6) (B) TRIAZOLAM GROUP (N=7) > Uf- Z w 15 w'" f-z «0...J 0.. WW Cl w 10 ~v! wo >f «5 I I I I I --. PLACEBO TREATMENT ' '1930' '1530' ' TIME OF SLEEP LATENCY TEST FIG. 1. MUltiple Sleep Latency Test profiles are shown for baseline and treatment conditions for the flurazepam group (A) and the triazolam group (8). Treatment with flurazepam resulted in significantly lower sleep latencies at 13:30, 17:30, 19:30 (marked by asterisks), and for the overall daily mean. Sleep latencies with triazolam were lengthened at 13:30, 15:30, 19:30 (*), and overall. be related to daytime sleepiness. Thus, relatively high correlations with MSLT were found for total sleep time, total wake time, minutes and percentage of stage 2 sleep, and percentage of stage 1 sleep. In the flurazepam group, on the other hand, correlations tended to cluster around zero. [Note that correlation coefficients with a negative sign indicate that sleepiness is increased (MSLT decreased) with an increase in the nocturnal sleep measure.] Profile of Mood States The POMS data are summarized in Table 4. On the morning POMS, the flurazepam group showed significant improvement with treatment on all subscales, except vigor. No significant differences on the morning POMS were found in the triazolam group. On the evening POMS, the flurazepam group again showed TABLE 3. Average correlation of mean daily MSLT with selected nocturnal sleep parameters Parameter Total sleep time Total wake time Stage 1, min Stagc 1, percent Stage 2, min Stage 2, percent Slow-wave, min Slow-wave, percent REM, min REM, percent Number of body movements Leg movements per h of sleep Leg movements with arousallh sleep Flurazepam Triazolam Sleep, Vol. 5, No.4, 1982

8 368 M. A. CARSKADON ET AL. TABLE 4. Profile of mood states Flurazepam Triazolam Measure Placebo Treatment Placebo Treatment Morning (10:00) POMS Fatigue 8.2 ± ± 2.9 a 4.9 ± ± 4.6 Depression 9.6 ± ± 6.3 a 8.4 ± ± 11.5 Anger 5.3 ± ± 2.4 a 4.9 ± ± 6.7 Tension 5.3 ± ± 2.4 a 2.8 ± ± 6.4 Confusion 4.0 ± ± 2.7a 3.8 ± ± 5.2 Vigor 15.0 ± ± ± ± 7.8 Evening (21:30) POMS Fatigue 8.4 ± ± 3.3 a 4.9 ± ± 3.0 a Depression 7.8 ± ± 5.6 a 6.4 ± ± 8.0 Anger 2.3 ± ± 0.8 a 3.3 ± ± 4.0 Tension 4.9 ± ± 2.7a 2.5 ± ± 5.6 Confusion 3.1 ± ± ± ± 4.0 Vigor 14.0 ± ± ± ± 7.7 Values are means ± SD. a Significant difference between placebo and treatment (p <.05). significant improvement with treatment in four of the subscales. The triazolam group improved only on the fatigue sub scale of the evening POMS. Performance Table 5 summarizes the performance test results. No significant condition differences were present for either group on the word memory test. The flurazepam group improved significantly on the "by-suit"' card sort, and the triazolam group improved significantly on the "by-number" and "4-pile" card sorting with treatment. Both groups showed significant improvement on the digit-symbol substitution test with treatment. This finding may reflect a practice effect, however, because all subjects achieved better scores with each repetition of the test. The vigilance test showed significant increase in the number of missed responses with TABLE 5. Performance tests FJurazepam Triazolam Measure (daily means) Placebo Treatment Placebo Treatment Auditory Vigilance Test Number of misses 0.9 ± ± 2.3" 0.6 ± ± 1.3 False positive responses 0.4 ± ± ± ± 0.3 Card sort (s) By suit 38.4± ± 5.0 a 41.3 ± ± 8.4 By number 45.3 ± ± ± ± 9.9 a Into 4 piles 22.4 ± ± ± ± 5.5" Into 10 piles 24.1 ± ± ± ± 7.0 Word memory Words recalled 6.6 ± ± ± ± 1.2 Falsely recalled 1.0 ± ± ± ± 0.4 Digit Symbol Substitution 43.0 ± ± 5.5 a 44.6 ± ± 6.6 a Values are mean ± SD. a Significant difference between placebo and treatment (p <.05). Sleep. Vol. 5. No.4, 1982

9 SLEEPING PILLS IN ELDERLY INSOMNIACS 369 treatment in the flurazepam group. No significant changes were seen on the vigilance test for the triazolam group. DISCUSSION The nocturnal sleep data showed similar efficacy for both compounds. Total sleep time increased significantly with treatment in both groups. This increase is largely attributable to a significant increase in stage 2. The other sleep stages were unaffected by treatment, except for a small, but statistically significant, increase in stage 3 sleep with triazolam. With the exception of the stage 3 sleep increase in the triazolam group, these findings correspond to the results of similar studies on adult insomniacs. The paradoxical increase in stage 3 may reflect the very short half-life of triazolam. Although neither compound had a significant effect on nocturnal respiration, it should be recalled that the subjects had no preexisting sleep apnea syndrome and that low doses were administered. The effects of higher doses or the effects in patients with sleep apnea syndromes may be of greater significance. Daytime sleep recordings in healthy adult subjects by Ogura et al. (21) showed residual effects (increased sleep time) of flurazepam, 15 and 30 mg, in morning, afternoon, and evening recordings; residual effects oftriazolam were present only in the morning. The MSLT findings with flurazepam in the present study correspond somewhat with their results, showing increased sleepiness in the afternoon and evening. The MSLT findings with triazolam, on the other hand, showed reduced sleepiness in the daytime and not simply the maintenance of pretreatment levels. The difference in triazolam effects in the two studies may be related to the different subject groups that were used and to the greater latitude for improved sleep in the insomniac subjects of the present study. Thus, while the normal subjects of Ogura et al. increased their sleep by 5%, the insomniacs increased sleep time by nearly 20%. The trends present in the correlational analysis of MSLT and nocturnal sleep measures also suggest that improved alertness with triazolam was related to improved sleep at night. In contrast, the MSLT scores with flurazepam appeared to be unrelated to any drug-induced changes in nocturnal sleep. We suggest that the longer active half-life offlurazepam leads directly to daytime sedation and masks the beneficial effects of increased nocturnal sleep on daytime sleepiness/alertness. The shorter active half-life of triazolam eliminates the problem of daytime sedation, and the improved daytime alertness found with this compound results from the improvement in nocturnal sleep. In much the same way, the mood-altering effects found with flurazepam may reflect continued daytime sedation. The lack of marked mood effects with triazolam suggests that the improvement of sleep was not sufficient to produce changes in this measure of mood. The POMS findings with flurazepam do not correspond to findings in young adult "poor sleepers" (22) in whom morning and evening POMS scale scores did not change with 30 mg flurazepam. This difference may reflect an age-related difference in the way subjects completed the scale, or an age-related sensitivity to the drug. Performance test findings suggest that the subjects taking flurazepam were Sleep. Vol. 5. No

10 370 M. A. CARSKADON ET AL. unimpaired in cognitive performance, although vigilance was somewhat reduced. The reduced vigilance in the flurazepam group vias minima! on the test used in this study. Others have found cognitive and vigilance deficits with flurazepam (30 mg) in older adult poor sleepers (23) and in noncomplaining adults (24,25). Performance with triazolam tended to be unchanged (vigilance, memory) or to improve (card sorting; digit symbol substitution). In summary, flurazepam, 15 mg, and triazolam, 0.25 mg, tended to improve sleep in elderly patients with complaints of insomnia. Daytime testing, however, showed residual sedation with flurazepam and improvement of daytime alertness with triazolam. We suggest that the use offlurazepam as a sleep-inducing medication in elderly insomniacs should be limited to patients for whom daytime sedation is desirable, and for whom it would not constitute a hazard or an interference with mode of living. Triazolam appears to have equivalent effects on nocturnal sleep without inducing daytime sedation. On the contrary, triazolam improved daytime alertness in these elderly subjects and therefore might be recommended for patients for whom daytime sedation is not wanted. We feel that the extremely brief sleep latencies in the MSLT in the flurazepam subjects may present a significant risk if sustained alertness is required, as in driving an automobile. In terms of nocturnal sleep parameters, the benefits of both compounds appear to be similar in magnitude. Acknowledgments: This research was supported in part by National Institute on Aging grant AG 02504, by National Institute of Mental Health award MH to William C. Dement, and by The Upjohn Company, who also provided the medications. We thank William Hebert, Duley Petta, and Ann Pursley for their help with sleep data analysis. REFERENCES I. Institute of Medicine. Report of a study: sleeping pills, insomnia, and medical practice. Washington: National Academy of Science, Miles LE, Dement WC. Sleep and aging. Sleep 1980;3: Reynolds C, Coble P, Black R, et al. Sleep disturbances in a series of elderly patients: polysomnographic findings. JAm Geriatr Soc 1979;28: Coleman R, Miles L, Guilleminault C, et al. Sleep-wake disorders in the elderly: a polysomnographic analysis. JAm Geriatr Soc 1981;29: Carskadon M, Dement W. Respiration during sleep in the aging human. J Gerontal 1981 ;36: Ancoli-Israel S, Kripke DF, Mason W, Messin S. Sleep apnea and nocturnal myoclonus in a senior population. Sleep 1981;4: Greenblatt D, Allen M, Shader R. Toxicity of high-dose flurazepam in the elderly. Clin Pharmacal Ther 1977;21: Greenblatt D, Divoll M, Harmatz J, et al. Kinetics and clinical effects of flurazepam in young and elderly insomniacs. Clin Pharmacal Ther 1981 ;30: Association of Sleep Disorders Centers. Diagnostic classification of sleep and arousal disorders, first edition, prepared by the Sleep Disorders Classification Committee, HP Roffwarg, Chairman. Sleep 1979;2: Carskadon M. Determinants of daytime sleepiness: adolescent development, extended and restricted nocturnal sleep. Doctoral dissertation, Stanford University, II. Breimer D. Pharmacokinetics and metabolism of various benzodiazepines used as hypnotics. Br J Clin Pharmacal 1979;8: Rechtschaffen A, Kales A, eds. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Brain Information ServicelBrain Research Institute, University of California at Los Angeles, Sleep. Vol. 5. No.4, 1982

11 SLEEPING PILLS IN ELDERLY INSOMNIACS Coleman R, Pollak C, Weitzman E. Periodic movements in sleep (nocturnal myoclonus): a case series analysis. Ann Neural 1980;8: Hoddes E, Zarcone V, Smyth H, et al. Quantification of sleepiness: a new approach. Psychophysiology 1973;10: McNair D, Lorr M, Droppleman L. Manualfor the POMS. San Diego: Educational and Industrial Testing Service, Williams H, Gieseking C, Lubin A. Some effects of sleep loss on memory. Percept Mot Skills 1966;23: Bond A, Lader M. Residual effects of hypnotics. Psychopharmacologica 1972;25: Kornetsky C, Vates TS, Kessler EK. A comparison of hypnotic and residual psychological effects of single doses of chlorpromazine and secobarbital in man. j Pharmacol Exp Ther 1959;127: Wilkinson RT. Methods for research on sleep deprivation and sleep function. In: Hartman E, ed, Sleep and dreaming, Boston: Little Brown, 1970: Greenblatt D, Divoll M, Moschitto L, et al. Electron-capture gas chromatographic analysis of the triazolobenzodiazepines aloprazolam and triazolam. J Chramatogr 1981 ;225: Ogura C, Nakazawa K, Majima K, et al. Residual effects of hypnotics: triazolam, flurazepam, and nitrazepam. Psychopharmacology 1980;68: Church M, Johnson L. Mood and performance of poor sleepers during repeated use offlurazepam. Psychopharmacology 1979;61: Oswald I, Adam K, Borrow S, et al. The effects of two hypnotics on sleep, subjective feelings and skilled performance. In: Passouant P, Oswald I, eds, Pharmacology of the states of alertness. Oxford: Pergamon Press, 1979: Roth T, Piccione P, Salis P, et al. Effects oftemazepam, tlurazepam and quinalbarbitone on sleep: psychomotor and cognitive function. Br J Clin Pharmacol 1979;8: Roth T, Hartse K, Zorick F, et al. The differential effects of short- and long-acting benzodiazepines upon nocturnal sleep and daytime performance. Drug Res 1980;30: Sleep, Vol. 5. No.4, 1982