The Multiple Sleep Latency Test: Individual Variability and Time of Day Effect in Normal Young Adults

Transcription

1 Sleep 13(5): , Raven Press, Ltd., New York 1990 Association of Professional Sleep Societies The Multiple Sleep Latency Test: Individual Variability and Time of Day Effect in Normal Young Adults M. Clodore, O. Benoit, J. Foret, and G. Bouard INSERM U3 H6pital de la Salpetriere, Paris Cedex 13, France Summary: The influence of individual characteristics on diurnal physiological sleep tendency was investigated in young good sleepers. Fifty-five subjects underwent a Multiple Sleep Latency Test (MSLT) procedure. Among them 11 also participated in Repeated Test Sustained Wakefulness (RTSW) procedure. The MSL T results were analyzed as a function of both the number of sleep onsets per day and the time of day. Diurnal sleepiness seemed to be better appreciated by sleep onset (SO) frequency than by the traditional criteria of sleep latency. SO frequency, unlike latency, was influenced by factors such as usual sleep duration, morning/evening score, and RTSW procedure. Time of day effect was characterized by a decrease in sleep tendency at the beginning and at the end of the day (decrease in SO frequency and increase in SO latencies); between these two points a peak of sleepiness around 1400 was observed. The morning and evening periods of high alertness could represent important anchor points for the coupling of the sleep/wake and temperature rhythms. Key Words: MSLT-RTSW-Time of day-sleep schedules Morningness/eveningness. The Multiple Sleep Latency Test (MSLT) is a technique devised by Richardson et al. (1) to assess excessive daytime sleepiness. Carskadon and Dement (2) extended its validity to the measurement of the "physiological sleep tendency," i.e., "the difficulty maintaining arousal in a non-arousing environment." Since that time the MSL T has been widely used to measure sleep propensity after sleep deprivation, in sleep disorders (3-5), and with aging (6). In addition, an alternative method, the "repeated test of sustained wakefulness" (RTSW), was proposed by Hartse et al. (7) in order to better evaluate extreme diurnal sleepiness. The experimental conditions are the same as those for MSLT, except for the instruction to "try to stay awake." But few studies have been specifically devoted to the influence of the morningness/eveningness index on physiological sleep tendency in young adults with usual sleep/wake schedules. The present studies provide reference data for future studies on daytime alertness. Preliminary results have been published by Clodore et al. (8). Accepted for publication January Address correspondence and reprint requests to O. Benoit, Laboratoire d'etude du Sommeil, Batiment Pharmacie Laboratoires, 4eme Etage, 47 Bid de I'H6pital, Paris Cedex 13, France. 385

2 386 M. CLODORE ET AL. SUBJECTS Study 1 Fifty-five volunteers between the ages of 19 and 28 years underwent the MSLT protocol. They were free of any complaints of nocturnal sleep disturbance and of daytime sleepiness, and none had the habit of napping. Their habitual sleep lengths, verified by a 15-day sleep diary, ranged from 7 to 9 h. The subjects were grouped as a function of their Horne and Ostberg's score (9) as evening type (ET) (C), neither type (NT) (B), or morning type (MT) (A). Among ET and MT there were three "definitely evening type" and one "definitely morning type" subjects; the others were either of "moderate evening type" or "moderate morning type." Study 2 To study the influence of instruction on sleepiness, eleven of the "neither type" subjects underwent both the RTSW (D) and the MSLT procedures. The two tests were administered on nonconsecutive days in counterbalanced order. Table 1 shows the mean values of age, sex, and sleep schedules for the four groups of subjects. METHODS MSLT and RTSW were performed on days following a control night with sleep recordings. If not already awake, the subjects were awakened at 800. MSLT or RTSW were performed at 1000, 1200, 1400, 1600, 1800, and 2000 according to conventional protocol (2,10). After the noon test the subjects had a regular lunch without alcoholic beverage, coffee, or tea. For each test, subjects laid down on a bed in a private room. For the MSLT, they were asked to close their eyes and "try to fall asleep." For the RTSW, they were asked to close their eyes and "try to stay awake. " If no sleep occurred, the test was terminated after 20 min. If sleep did occur, the subject was awakened after 1.5 min of stage 1 in order to prevent cumulative sleep during the day. Test recordings were scored according to Rechtschaffen and Kales criteria (11). The sleep onset (SO) latency was defined as the interval (in minutes) between the beginning TABLE 1. Characteristics of the four groups of subjects Sleep schedules Average Average Tests Group Typology Sex Age bedtime waking time A 8MT 4F h 30 min 7 h 50 min 4M ± 2.04 ±.50 ±.54 MSLT B 27 NT 12F h 55 min 8 h 06 min 15M ± 3.8 ±1.1 ±.56 C 10 ET 4F h50min 6M ±1.4 ± 1.05 ± 1.0 MSLT D 11 NT 2F h 30 min 7 h 45 min RTSW 9M ± 2.2 ±.50 ±.54 Sleep schedules. The latter were calculated from the 2-wk sleep diaries. NT, Neither type; MT, morning type; ET, evening type; M, male; F, female. Sleep. Vol. 13. No

3 MSLT: VARIABILITY WITH SUBJECT AND TIME OF DAY 387 of the test (lights out) and the first of three consecutive epochs of stage 1 sleep marked by the disappearance of alpha rhythm in over 50% of one 20-s epoch (7,12). Data analysis The mean latency scores were calculated both for the individuals and for the different groups. Tests with no sleep onset were not included in the analyses. An analysis of variance (ANOY A) completed by matched sample (-test was used for main effects of conditions and groups. The i test was used for the analysis of sleep frequency as a function of time of day. Individual data RESULTS Fig. 1 shows for all subjects (groups A, B + D, and C) the number of subjects as a function of the number of SO throughout the MSLT day session (median and mean between two and three sleep onsets per day) and the number of subjects as a function of the average individual sleep latency per day (mean latency of SO). The mean group latency was 13.1 ± 3.5 min. The correlation between the individual number of sleep onsets and their mean sleep latency was not significant; those who fell asleep the fastest were not necessarily those who fell asleep the most often during the day. Subjects were grouped as a function of the number of their sleep onsets (Table 2). Sex, age, and Horne and Ostberg scores did not differ among groups. In contrast, when the subgroup with one and two SO was compared with that with five and six SO per day, average bedtime (23 h 9 min versus 23 h 55 min) and sleep duration (8 h 50 min versus 8 h 5 min) differed significantly (one-way ANOYA, respectively, F = 4.1, P < 0.05 and F = 6.2, p < 0.02; df = 1,25). Subjects who usually had an earlier bed time and a longer sleep duration fell asleep less frequently during the MSLT. f' CII 18 u Q ::::I CII CI 10.Q E ::::I Z l Number of sleep onsets by day o Average sleep onset latency by day (mi nutes) FIG. 1. Distribution of individual sleep frequency and individual mean stage 1 latency of 55 subjects in MSLT. Each point is the mean of actual sleep latencies. Tests in which sleep onset did not occur were omitted from the calculation of the average. The six subjects with 6-min latencies should not be considered as pathological because, in fact, five of them fell asleep only once and one twice.

4 388 M. CLODORE ET AL. TABLE 2. Characteristics of subjects in function of the number of sleep onsets during the lvlslt Number of Number Habitual sleep schedules sleep onsets of Average Average Sleep Mean latency per day Typology subjects Sex Age bedtime waking time duration in MSLT 6 IMT 5 2M NT 9% 3F ± 2.1 ± 0.6 ±.55 ±.9 ± ET I3 9M NT 23.6% 4F ± 2.3 ± 1.1 ±1.l ± 0.61 ± and 3 2MT 6 4M NT 10.9% 2F ±1.7 ±.4 ±.6 ±.9 ± MT 7M NT 12F ± 2.06 ±.8 ±.7 ±.86 ±1.4 3ET 34.5% IMT 4M NT 6 2F ± 1.6 ±1.3 ±.28 ±.99 ± 3.5 2ET 10.9% 0 IMT 3M NT 6 3F ± 2.65 ±1.3 ±.76 ± 2.02 let 10.9% MT, Morning Type; NT, Neither Type; ET, Evening type; M, male; F, female. (according Horne and Ostberg questionnaire). Time of day effect Time of day effect was significant for both frequency for sleep onset (X 2 = 13.71, p < 0.01; df = 5) and stage 1 latency when the latter was calculated in a conventional manner, i.e., absence of SO was scored as a 20-min latency (F = 4.3, df = 5,330, p < 0.05). At 1400, the probability of falling asleep was the highest and the sleep latency the shortest. In contrast, when the ANOV A was performed only on the actual SO latencies (i.e., absence of SO was not counted), time of day effect was no longer significant (F = 1.9, df = 5,158). Fig. 2 shows the number of SO as a function of latency at different times of day. The median of stage 1 latencies varied as a function oftime (13.8 min at tooo, 11 min at 1400, and 15.5 min at 2000). At tooo and 2000 sleep onsets were clearly infrequent. Fig. 3A shows the relationship between the number of sleep onsets per individual and the time of day. Except for those subjects who either fell asleep at all six sessions or did not fall asleep at all, the probability of sleep onset was a function of time, with a peak at 1400 and a minimum at Fig. 3B displays, as a function of time of day, the mean latencies for all sleep onsets for all subjects, for those with five and six SO and for the six subjects with one SO (from the top to bottom of the figure). In all three, the SO latency was lowest at 1400 and highest at When compared among individuals (subjects with only one SO) or intra-individually (five and six SO), SO latency was longer at both tooo and For the subjects with two or three SO, SO latency did not show any significant variation with time. This might be explained by the fact that the subjects almost never fell asleep at tooo or Individual characteristics In Table 3, the number of SO and their mean latencies are presented for the three groups determined by the "morningness/eveningness" score (9). No difference was

5 MSLT: VARIABILITY WITH SUBJECT AND TIME OF DAY 389 Median hliliu , 1400.!!l QI en c 0 Q. QI QI -en... 0 L QI.Q E ::::I Z ; j SO latency (minutes) FIG. 2. Distribution of stage I latency at different times of the day. Arrow represents median sleep latencies of actual SO. found, either for the number of sleep onsets or for the mean SO latencies. The only significant difference was found at 1000 and 1200; morning type subjects (group A) fell asleep less frequently than the two other groups (B + D and C) (X 2 = 8.94, df = 2, p < 0.01).

6 390 M. CLODORE ET AL. B 15 II... o L.. Q.I.0 E ::::J Z b+c 10 (/) " ::::J ::::J UI C i;"" UI - -- ::::J S] f...,.,......,.. '0 '2 '4' Time of day 10 I Time of dlly FIG. 3. (A) Distribution of sleep onsets as a function of number of sleep onsets (SO) per individual and as a function of time of day. Six subjects never fell asleep. (a) For all the subjects, (b) for subjects with 6 SO (N = 5), (c) for subjects with 5 SO (N = 13), (d) for subjects with 3 and 4 SO (N = 6), (e) for subjects with 2 SO (N = 19), (f) for subjects with 1 SO (N = 6). (B) Average stage 1 latency: (a) for all the subjects, (b + c) for subjects with 5 and 6 SO, (1) for subjects with I SO. 10 Comparison MSLT/RTSW The effect of instruction was studied by submitting 11 "neither type" subjects (group D) to the RTSW protocol with the instruction to "try to stay awake." The number of sleep onsets and the average SO latency are presented in Fig. 4. Compared with MSLT, RTSW resulted in a significant decrease in SO frequencies without change in sleep latency. Intra-individually, there was no relation between the two conditions (MSLT and RTSW) concerning either the number of SO or their latency. The daily distribution of both sleep onset frequency and sleep latencies obtained in RTSW confirmed the Sleep. Vol. ]3, No.5, 1990

7 MSLT: VARIABILITY WITH SUBJECT AND TIME OF DAY 391 l, n " ", TABLE 3. MSLT in three groups of subjects characterized by Horne and Ostberg questionnaire Number of sleep onsets (a) Number Number and mean latency to sleep onset (b) (min) of sleep of sleep as a function of time of day onsets/ onsets/ Mean Group subject trial latency Morning a I type A % N b = 8 ±1.4 (19/40) ± 3.3 ± 3.6 ± 2.9 ± 4.3 ± 2.0 ± 2.1 Neither a /18 type B+D % B b N = 37 ± 1.6 (112/198) ± 3.5 ± 3.9 ± 4.1 ± 3.9 ± 5.0 ± 4.5 ± 3.3 Evening a type C % b N = 10 ± 1.7 (28/60) ± 3.2 ± 4.6 ± 3.3 ± 5.0 ± 3.8 ± 3.3 ± 1.0 MSLT results: at 1400 sleep onsets were frequent and occurred with rather short latencies, while at 2000 all the subjects stayed awake (Table 4). DISCUSSION Sleep probability, measured in individuals by the number of actual sleep onsets during the MSLT session, shows a maximal dispersion (from 0 to 6 for the six tests). As far as we know few studies concerning the MSLT have taken into consideration this measure since tests without sleep were generally included as 20-min latencies in the calculation of the average sleep latency (13). The only exception was for narcoleptic patients in a study by Reynolds et al. (14) where the MSL T tests without sleep onsets were not taken into account. Therefore, we could not directly compare our results with those of other authors. However, individual and group mean sleep latencies were in the normal range (>6 min) in agreement with Dement and Carskadon (15), Hartse et al. (7), and Dement et al. '.- en t, QI,-.CI :;, en... CI GI.CI 2 E :;, z :: :: Number of sleep onsets Average sl eep onset latency by day by day (minutes) FIG. 4. Comparison of MSLT (_) and RTSW (1m) scores for 11 neither type subjects.

8 392 M. CLODORE ET AL. TABLE 4. MSLT and RTSW scores for 11 "neither type" subjects who completed the two procedures on separate days Number of sleep onsets (a) and mean latency to sleep onset (b, min) Number Number as a function of time of day of sleep of sleep Mean onsets/subject onsetsltrial latency / a MSLT ± \.9 59% ± b ± 2.6 ± 4.6 ± 4.6 ± 5.8 ± 4.6 ± 3.7 RTSW \.72 19/ a / ± \.6 34% ± b ±O ± 4.13 ± 1.8 ± 4.7 ± 6.3 (10). It is to be noted that no latency under 6 min was observed. In the present study subjects who had five or six SO per day slept on the average 45 min less than those who had only one or two SO. They were rather young (22 years of age), and although their habitual sleep duration was within the normal range (7-9 h), it can be hypothesized that their daily sleep need was greater than their actual sleep length. According to Carskadon and Dement (12,16) Dement and Carskadon (15), and Dement et al. (10) there is a negative correlation between night sleep reduction and daytime mean sleep latency. Other evidence in favor of such a hypothesis are the results obtained in subjects who had slept 10 h for two or three consecutive nights before the MSLT session (17): greater number oftests without sleep in young adults (70%) than in the present study (53%) and longer sleep latencies (16 versus 13.1 min) (18). In the same age range (18-22 years), Levine et al. (19) obtained a mean latency of 11.1 min with the inclusion of the 20-min tests. It is possible that in the present study, the higher sleep duration (>8 h on the average) accounted for the longer sleep latencies. Carskadon et al. (16), Richardson et al. 0), and Sugerman and Walsh (20), have shown that sleep latency varies consistently with the time of the night. Results concerning diurnal variation of sleep latency with time are more controversial. Scoring "no sleep tests" as being 20-min latencies, Carskadon and Dement (2,12,16) observed that sleep latencies were lowest in the early afternoon. Our results are in agreement with these authors, but the present study further analyzes the time of day effect on both SO frequency and sleep latency. On the one hand, for subjects falling asleep at each test, the daytime variation of sleep propensity is mainly reflected by SO latency changes, i.e., long latencies at 1000 and On the other hand, for the subjects who had only two or three SO, these SO occurred mainly at 1400 and 1600; with latencies equivalent to those of MitIer et al. (4). Therefore, all the subjects showed an increased sleep propensity at The RSTW protocol makes this pattern even more pronounced: compared with MSLT the total number of sleep onsets is decreased [Browman et al. (21), Hartse et al. (7)]. Our data show that this decrease is mainly due to the absence of SO at the beginning and at the end of the day. In contrast, the actual latencies were not modified by the change of instruction. The early afternoon increase in the "physiological sleep tendency" (2) has been observed in different conditions (22-24). Rather, it appears to be essentially determined by two periods of "enforced" alertness. The morning alertness zone appears to be weaker than the evening zone, as can be seen by the fact that more subjects had an SO with shorter latency at 1000 than at These results are in good agreement with data

9 MSLT: VARIABILITY WITH SUBJECT AND TIME OF DAY 393.'.: " obtained using different protocols (23,25,26). In these studies sleepiness exhibits a bimodal distribution with troughs at times similar to ours for externally synchronized subjects and at times related to the temperature curve for de synchronized subjects in time-free environment. The times of high alertness could correspond to the two' 'forbidden zones" of Lavie (23) or to the "wake maintenance zones" of Strogatz et al. (26). In between, the increase in sleep propensity observed in the early afternoon could correspond to a permissive zone for sleep. During this period sleep might occur according to situational facts such as environmental conditions and/or prior sleep deprivation and/or copious meal and alcohol. In the present data, the influence of the meal and the increased sleep propensity at 1400 cannot completely be excluded. However Stahl et al. (27) have shown that a standard meal has little influence if any on sleep onset latency around We found that the morningness/eveningness characteristic influences the probability of sleep at This difference in sleep tendency at a given time of the day may reflect the difference in the way both "wake maintenance zones" and temperature trough anchor the circadian system of humans (28). I, REFERENCES 1. Richardson G, Carskadon MA, Flagg W, Van Den HoedJ, Dement WC, Mitler MM. Excessive daytime sleepiness in man: multiple Sleep Latency measurement in narcoleptic and control subjects. Electroencephalogr Clin NeurophysioI1978;45:621-7, 2. Carskadon MA, Dement WC. The multiple sleep latency test: what does it measure? Sleep 1982;5:S Mitler MM. The multiple sleep latency test as an evaluation for excessive somnolence, In: Guilleminault C, ed. Sleeping and waking disorders: indications and techniques. Menlo Park, CA: Addison-Wesley, 1982: Mitler MM, Gujavarty KS, Browman Cp, Maintenance of wakefulness test. A polygraphic technique for evaluating treatment efficacy in patients with excessive somnolence, Electroencephalogr Clin Neurophysiol 1982;53: , Zorick F, Roehrs T, Koshorek G, Sicklesteel J, Hartse K, Willig R, Roth T, Patterns of sleepiness in various disorders of excessive daytime somnolence, Sleep 1982;5:S Richardson GS, Carskadon MA, Orav EJ, Dement WC. Circadian variation of sleep tendency in elderly and young adult subjects. Sleep 1982;5:S Hartse KM, Roth T, Zorick FJ. Daytime sleepiness and daytime wakefulness: the effect of instruction. Sleep 1982;5:SI Clodore M, Foret J, Benoit O. Diurnal variation in subjective and objective measures of sleepiness: the effects of sleep reduction and circadian types. Chronobiollnt 1986;3: , 9, Horne JA, Ostberg 0, A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms, lnt J Chronobiol 1976;4:97-110, 10. Dement WC, Seidel W, Carskadon MA. Daytime alertness, insomnia and benzodiazepines. Sleep 1982;5:S II. Rechtschaffen A, Kales A. Manual of standardised terminology, techniques and scoring system for sleep stages of human subjects. NIH Publications 204, Washington, D.C Carskadon MA, Dement WC. Sleep tendency during extension of nocturnal sleep. Sleep Res 1979;8:147, 13. Carskadon MA, Dement WC, Mitler MM, Roth T, Westbrook PR, Keenan S. Guidelines for the Multiple Sleep Latency Test (MSLT): a standard measure of sleepiness. Sleep 1986;9: Reynolds PA, Coble DJ, Kupfer DJ, Holzer BC. Application of the multiple sleep latency test in disorders of excessive sleepiness. Electroencephalogr Clin Neurophysiol 1982;53: Dement WC, Carskadon MA. An essay on sleepiness. In: Baldy-Moulinier M, ed. Actualites en medecine experimentale en hommage au Pr. P. Passouant. Montpellier: Euromed, 1981: Carskadon MA, Dement WC. Cumulative effects of sleep restriction of daytime sleepiness, Psychophysiology 1981;18: Miles LE, Dement WC. Sleep and aging, Sleep 1980;3: Carskadon MA. The second decade. In: Guilleminault C, ed, Sleeping and waking disorders: indications and techniques. Menlo Park, CA: Addison-Wesley, 1982: Levine B, Roehrs T, Zorick F, Roth T, Daytime sleepiness in young adults. Sleep 1988;11:S39-46.

10 394 M. CLODORE ET AL. 20. Sugarman JL, Walsh JK. Physiological sleep tendency and ability to maintain alertness at night. Sleep 1989;12: Browman CP, Gujavarty KS, Sampson MG, Mitler MM. REM episodes during the maintenance of wakefulness test in patients with sleep apnea syndrome and patients with narcolepsy. Sleep 1983 ;6: Carskadon MA, Dement WC. Midafternoon decline in MSLT scores on a constant routine. Sleep Res 1985;14: Lavie P. Ultrashort sleep-waking schedule III "gates" and "forbidden zones" for sleep. Electroencephalogr Clin Neurophysiol 1986;63: Zulley J, Campbell SS. Napping behavior during "spontaneous internal desynchronisation" sleep in synchrony with body temperature. Hum NeurobioI1985;4: Campbell SS, Zulley J. Ultradian components of human sleep/wake patterns during disentrainment. Exp Brain Res 1985;12: Strogatz SR, Kronauer RE, Czeisler CA. Circadian pacemaker interferes with sleep onset at specific times each day: role in insomnia. Am J PhysioI1987;253: Stahl ML, Orr WC, Bollinger C. Postprandial sleepiness: objective documentation via polysomnography. Sleep 1983;6:S Benoit 0, Foret J. Regulation circadienne des etats de veille et de sommeil. Neurophysiol Clin 1988;18: