Energy Expenditure and Total Sleep Time: Effect of Physical Exercise

Transcription

1 Sleep, 5(2): Raven Press, New York Energy Expenditure and Total Sleep Time: Effect of Physical Exercise lain Montgomery, John Trinder, and Susan J, Paxton Department of Psychology, University of Tasmania, Hobart, Tasmania, Australia Summary: The energy conservation model proposes that the main function of sleep is to lower metabolic requirements periodically and thus to conserve energy. However, certain variations in energy expenditure, such as that produced by physical exercise, have not been found to be consistently related to sleep length. We hypothesized that, because sleep variables may adapt relatively slowly to metabolic changes, the effect of exercise on sleep time would be observed as a function of habitual exercise patterns, not of daily variations. The study consisted of a retrospective analysis of five experiments. Although the design of each experiment was idiosyncratic, all involved physically fit and/or unfit subjects whose sleep was assessed following daytime exercise and/or no exercise conditions. As predicted, fit subjects slept significantly longer than unfit subjects, and daytime exercise had no consistent effect on sleep duration. However, for several reasons, the relevance of the data to the energy conservation model is uncertain. Key Words: Physical fitness Exercise-Sleep-Total sleep time-energy balance. The hypothesis that sleep has an energy conservation function (1-7) predicts that total daily sleep quota will be related positively to levels of energy expenditure. Support for this relationship comes primarily from phylogenetic and ontogenetic data. Thus there is a substantial correlation across species between total sleep time (TST) and basal metabolic rate (BMR), and thus total energy expenditure (7). Also, both TST (8) and BMR decrease as a function of age (9). Further, it has been noted (2) that the appearance of homeotherrny coincides phylogenetically with electroencephalographically defined sleep and possibly ontogenetically with slow wave sleep (SWS; non rapid eye movement stages 3-4). However, apart from ontogenetic changes, variations in energy expenditure within species were not found to be consistently related to sleep length (7,10). A number of studies assessed sleep duration as a function of exercise. The mqjority (11-17) found no change, one (18) found an increase, and one (19) found a decrease. Thus the current literature on the effects of exercise on TST is clearly Accepted for publication January Address correspondence and reprint requests to John Trinder, Department of Psychology, University of Tasmania, Hobart, Tasmania, Australia. 159

2 160 I. MONTGOMERY ET AL. negative in that all but one study failed to find a facilitative effect. On the basis of this and other data, Eastman and Rechtschaffen (10) argued that while species may have adapted their sleep to different energy demands (ultimate causality hypothesis), sleep is not responsive to variations, within individuals, in energy expenditure (proximate causality hypothesis). There are two reasons why the conclusion with respect to the empirical relationship between exercise and TST may be premature. First, the majority of studies either held time in bed (TIB) constant, or restricted the recording period, and thus reduced the likelihood of observing differences between conditions. Second, with one exception (17), the negative studies assessed the effect of particular exercise sessions on the immediately-following night's sleep. However, it is possible that sleep patterns respond slowly to metabolic change, a view expressed by Dunleavy et al. (20). The data of Adey et al. (21) are consistent with this hypothesis in that TST was reported to average 5.5 h in a sample of quadraplegics. Also, we showed that sleep cycle length is shortened by long term elevations in energy expenditure produced by regular physical training, but not altered by day-to-day changes due to particular training sessions (22). In view of these considerations it is possible that, although variations in exercise may not affect the immediately-following night's sleep, TST may respond to habitual exercise patterns. Thus individuals who engage in regular physical exercise over a period of months or years, and who as a result are physically fit, will have more TST than individuals who are sedentary. Apart from studies in our laboratory, there has been one study evaluating the TST of regularly exercising subjects (fit athletes), as opposed to sedentary subjects (14). The results indicate that, although TIB was limited to 8 h, there was a trend (0.05 > P > 0.01) for fit athletes to have more sleep than nonathletes. We have now completed five experiments involving subjects who were exercising regularly and were physically fit and/or subjects who did not exercise and were unfit. Both groups were typically assessed following both exercise and no-exercise days. Individually, the five studies are equivocal with respect to the relationship between energy expenditure and TST. As a group, they clearly indicate that individuals who habitually exercise sleep longer than unfit sedentary individuals. The studies also confirm that daily variations in exercise levels are not related to sleep time. METHODS Subjects and design Experiment 1 (23). The sleep of eight fit subjects (four males and four females) and eight unfit subjects (four males and four females) was compared after both an afternoon of exercise and a day of no exercise. The mean ages of the fit and unfit subjects were 23.3 and 23.5 years, respectively. Experiment 2 (24). Twenty-four unfit subjects (14 male and 10 female) were assigned to four groups matched for age and baseline SWS. Each group was required to perform one offour levels of exercise graded from no exercise in group 1 to relatively exhausting exercise in group 4. The subjects' sleep was assessed on

3 ENERGY EXPENDITURE AND TOTAL SLEEP TIME nights: a no-exercise baseline and 3 exercise nights. Four subjects, one from each group, were discarded from the present analysis, because the discarded subjects were substantially younger (average age, years) than the subjects in other experiments and because subjects in this study were all unfit with no agematched fit group. The average age of the remaining subjects was 22.7 years. In the statistical and graphical analyses, only the 10 subjects of groups 3 and 4, who were run in both exercise and no-exercise conditions, were included (the very light exercise of group 2 was considered to be of insufficient intensity). The average age of these subjects was Experiment 3 (25). Each of four groups contained six subjects. In one group, subjects were older (average age, 32 years) and fit, in another older and unfit, in another younger (average age, 22 years) and fit, and in the last younger and unfit. The sleep of the four groups was assessed on three occasions, the 1st, 3rd, and 5th days of a 5-day sequence. The fit subjects exercised on the afternoon of the 1 st day, but were not allowed to exercise for the remainder of the experiment. The unfit subjects were not permitted to exercise during the study. Experiment 4. Two groups participated in the experiment: eight male subjects who had long histories of participation in sports at a proficient level (athletes), and nine male subjects who had histories of nonparticipation in sports and exercise. Both groups were assessed on two different occasions separated by approximately three months. On the first occasion the athletes were physically unfit and on the second they were fit. At first the athletes were, on average, as unfit as the nonathletes because of a relatively prolonged period of abstinence from training, typically because of study commitments (minimal period required was 6 months). The nonathlete subjects were unfit on both occasions. On each occasion, the sleep of each subject was assessed on 4 nights, twice after both afternoon exercise and nonexercise conditions. For the athlete subjects, only the data collected when they were fit have been used. Experiment 5 (24). Each of 11 fit young males (average age, 19.5 years) completed four conditions: no-exercise baseline, a low-energy expenditure, and two high-energy expenditure conditions. In the latter two conditions the exercise was of either short (1 h) or long (6 h) duration. Each condition was performed on two separate occasions counterbalanced in a latin square design. The low-energyexpenditure condition was not included in the major analyses. The critical information, including mean V0 2max for the groups, is summarized in Table 1. As can be seen, the five experiments produced four samples of fit, and four of unfit, subjects. As described above, all subjects in the fit samples were measured on both exercise and nonexercise nights. This was also true oftwo unfit samples (experiments 1 and 4). Half the subjects in experiment 2 were assessed under both conditions; the remaining half did not exercise, or did so at a very low intensity. In the remaining study, experiment 3, the unfit subjects did not exercise. The subjects were recruited largely from the University of Tasmania community and were healthy, generally within the normal weight ranges for their physical builds, and not on medication. All subjects were given one night's adaptation to the laboratory prior to the experimental conditions. Sleep, Vol, 5, No, 2, 1982

4 162 I. MONTGOMERY ET AL. TABLE i. Ages, sex distributions, and levels of jiines's for the fit and unfit samples in the five experiments Fit samples Sex V0 2max (L/min) Experiment n Age M F M F Unfit samples Procedures Sleep assessment. The laboratory procedures were essentially identical for all the experiments and the key personnel remained the same. The laboratory consisted of two sound-attenuated rooms with an adjacent equipment monitoring room. All subjects reported 1 h before their normal retiring times. The laboratory methods were according to standardized procedures (26). Subjects were allowed a degree of freedom in selecting the times they came in to the laboratory, were allowed to read in bed before lights-out if they wished, and, within certain limits, were free to select the times they were awakened. However, these parameters were determined on the basis of the subjects' usual sleep habits rather than on the necessities of individual nights. As a consequence of these procedures, it was more likely that sleep time could vary in response to the independent variables than is typical in many laboratory situations in which TIB is held constant. The sleep records were scored in 30-s epochs by trained scorers, with an interrater agreement of at least 90%, according to standardized procedures (26). In all studies except experiment 1, records were scored such that the subjects' identities and the experimental conditions were unknown. The major variables were defined as follows: TIB was the time from lights-out until morning arising, the latter generally at a time determined by the subject the previous night; sleep period time (SPT) was the time from the first stage 2 epoch until the final return to wakefulness or stage 1; TST was TIB minus all periods of wakefulness; and sleep onset latency (SOL) was the time from lights-out to the first stage 2 epoch. Physical fitness testing. Two types of criteria were used to assess the physical fitness of subjects, (a) aerobic fitness and (b) levels of regular exercise. In each experiment, each subject's aerobic fitness was assessed by a bicycle ergonometer submaximal exercise test (27,28). Average V0 2max values are presented in Table 1. In addition to meeting a predetermined V0 2max criterion, fit subjects were re- Sleep. Vol. 5, No, 2, 1982

5 ENERGY EXPENDITURE AND TOTAL SLEEP TIME 163 quired to have been actively engaged in sport and to train at least three times a week. Unfit subjects were required to have lower-than-average V0 2max scores and not to exercise or engage in sport regularly. Physical exercise Experiment 1. The exercise was a run of 4.5 miles over a hilly course. The unfit subjects were permitted to walk when necessary. Experiment 2. All exercise was conducted on a bicycle ergonometer. Group 2 exercised at 50% V0 2max for 45 min (three 15"min sessions with 10-min rests), group 3 exercised at 75% V0 2max for 60 min (four 15-min sessions with 10-min rests), and group 4 exercised as closely as possible to 100% V0 2max for 60 min (four IS-min sessions with 10-min rests). Subjects in group 4 usually required more than the stipulated number of rests. Experiment 3. Subjects engaged in their normal fitness activities and were required to exercise at normal intensity, producing moderate fatigue but not excessive stress. Experiment 4. Fit subjects engaged in their normal exercise activities to produce moderate fatigue but not exhaustion. Unfit subjects were required to run approximately 2 miles at a mild rate. Experiment 5. There were three exercise conditions in this experiment: a I-h walk of about 3 miles, a I-h run of about 8 miles, and a 6-h walk of approximately 18 miles total distance. Although the level and nature of the exericse varied across studies it was, with two minor exceptions, relatively demanding, particularly for the fit subjects. The exceptions, the use of a bicycle ergonometer at 50% V0 2max by unfit subjects in experiment 2 and a I-h walk for fit subjects in experiment 5, were not included in any of the major analyses. All exercise was conducted between 16:00 and 18:30 h with the exception of the 6-h walk in experiment 5, which nevertheless finished at 18:00 h. RESULTS Considered together, the results of the five experiments clearly indicate that TST is higher in fit than unfit individuals (See Fig. lc). In each of the three experiments in which both groups were run, fit subjects slept longer. Further, there was no overlap between the two groups over the five experiments, TST for each of the fit groups being higher than that for each of the unfit samples. In contrast, the effect of exercise was not consistent over studies (Fig. 2c). An analysis of the data treating groups as replications (2 x 2 analysis of variance, n = 4, with one missing value in the unfit exercise cell) showed a significant fitness effect [F = (1,5), p < 0.05], with both the exercise effect [F = 2.25 (1,5), p> 0.05] and the interaction [F = 2.84 (1,5), p > 0.05] nonsignificant. Further, as suggested by the analysis and illustrated in Table 2, there was no effect due to characteristic exercise patterns (exercise in the fit and nonexercise in the unfit). While the effect of physical fitness on TST was not statistically significant in any of the studies considered individually, the data, nevertheless, strongly support the

6 164 I. MONTGOMERY ET AL. 480 El fit o unfit 101 Vi c ] <f! >- Vi c ] I- 0. <fl Vi c }: t- <fl r- --' a <fl II STUDIES v FIG. 1. Mean TIB (a), SPT (b), TST (c), and SOL (d) values for fit and unfit subjects in each experiment. Exercise and nonexercise conditions were averaged where applicable. view that subjects who habitually exercise have longer sleep quotas than more sedentary individuals. Unfortunately the proportion of male and female subjects varied across studies. However, it is unlikely that the variation in the proportion of female subjects would have produced artifactually the observed habitual exercise-fitness effect. In the three studies that included both fit and unfit groups, the proportions of male and female subjects were constant. The two groups not matched within a study (experiments 2 and 5) may be compared with groups from other studies with the same proportion of males and females. Such comparisons are consistent with the direction of the results. Further, as the statistical analysis used groups as replicates, the overall proportion of males and females is not relevant, though the

7 ENERGY EXPENDITURE AND TOTAL SLEEP TIME 165 LZl exercise FIG. 2. Mean TIB (a), SPT (b), TST (c), and SOL (d) values for exercise and nonexercise conditions in each experiment. Fit and unfit subjects were averaged where applicable. 480 ~ i 460 ID ;:: Vi 440 c ] l- f}; Vi 25 c i --' 0 (j) 15 ~1' "f ;: c- r- - '...'. II if- ;7-?, I J,', STUDIES r- non-exercise /r---.,! /.., [01 r- [bl Sf- 7r- [ ; " "....'., >[ III IV V imbalance of experiments 2 and 5 would be. Finally, the mean TST values for the 6 fit and 16 unfit females were 458 and 425 min, respectively. These values are quite consistent with the general results. There appeared to be two main sources of the increased TST in fit subjects: an increased TIB and a shorter SOL. Figure la shows that the fit subjects tended to spend a longer time in bed with the lights out than did the unfit subjects. The main effect of fitness was not statistically significant for TIB [F = 3.15 (1,5), p > 0.05] or SPT [F = 2.79 (1,5), p > 0.05). However, the graphs clearly indicate the contribution of these components to the TSTeffect. The contribution of SOL is more consistent (see figure Id), the main effect of fitness being significant [F = (1,5), p < 0.05]. {ci Idl

8 166 I. MONTGOMERY ET AL. TABLE 2. Average values for selecred sleep parameters as a function of physical fitness and exercise Fit Unfit Variable Ex Nex Ex Nex TlB SPT TST" SOL" SWS REM (% TST) W + MT + 1% SPT Values are expressed as minutes, unless noted otherwise. Abbreviations used: Ex, exercise; Nex, nonexercise; REM, rapid eye movement sleep; W, waking; MT, movement time. Other abbreviations are defined in text. " Fit and unfit values are significantly different, p < None of the other variables considered made a differential contribution to the sleep of the two groups (see Table 2). The situation with respect to SWS is of some interest, as this was the one significantly different variable between the two groups when the studies were considered independently (experiments 1, 3, and 4). The negative result [F = 2.43 (1,5), p > 0.05] was primarily because of a very low SWS level for fit subjects in experiment 5. With the exception of this study, the SWS data paralleled the TST data. Exercise conducted during the day had no consistent effect on any of the sleep variables, either as a main effect or as an interaction with fitness (see Table 2).1 The failure to observe an exercise effect on TST may have been due to the intensity of exercise used; the effect would seem more likely with more intense exercise (18). However, experiment 2, which varied intensity of exercise in unfit subjects, and experiment 5, which did so in fit subjects, offer no support for this hypothesis, the main effect of exercise level on TST being nonsignificant in each case [F = 0.36 (3,20), p > 0.05 for experiment 2; F = 2.59 (3,30), p > 0.05 for experiment 5]. Although the F value in experiment 5 approached significance, the ordering of the groups was not related to the intensity of the exercise. Thus, the failure to observe an exercise effect is unlikely to be due to an intensity of exercise component. This conclusion is reinforced by the nonsignificant interaction between exercise and fitness, as the fit subjects typically would have exercised at a higher absolute level than the unfit individuals. DISCUSSION The results demonstrate an empirical relationship between TST and habitual patterns of physical exercise. Thus they are consistent with the hypothesis that 1 The aim of the present paper was to consider the relationship between exercise and TST and thus to evaluate a particular theoretical issue. The effect of exercise on the distribution of sleep stages in each of the various designs has been presented in other papers in the context of the hypothesis that gave rise to the studies. Sleep. Vol. 5. No, 2, 1982

9 .) ENERGY EXPENDITURE AND TOTAL SLEEP TIME 167 sleep length is adaptive to relatively sustained, but not to short-term, variations in the level of physical exercise within individuals. Further, the data are consistent with the view that sleep variables respond slowly to changes in metabolic factors (20). Although the results are consistent with the proposed empirical relationship, they cannot be interpreted as unequivocally supporting the energy conservation model. There were two general reasons for this conclusion. First, the magnitude of the demonstrated effect was insufficient to account for the assumed difference in energy expenditure between the two groups of subjects. Thus it is probable that changes in TST in response to variations in physical exercise are of only minor relevance to total energy expenditure, within individual members of a species. Nevertheless, the observation that TST responds at all to variations in physical exercise lends credence to the view that sleep in general has an energy conservation function. Second, none of the present studies included an assessment of energy balance. Thus food intake data were not available, leaving the results open to various interpretations in terms of dietary differences between the groups of subjects. Further, in the absence of complete energy expenditure information it cannot be stated with certainty that the total waking-period energy expenditure of the exercising subjects was greater than that of the sedentary groups, because adjustments could have been made during other times in the wake period. Finally, it is not known if metabolic rate during sleep varied between the two groups. Although any of these alternatives is possible, there is no direct evidence to support them and indeed there are data to the contrary. For example, neither short-term food deprivation (29-33) nor changes in dietary components (34) appear to affect TST. With respect to daily energy expenditure, there is considerable evidence indicating that individuals who exercise regularly have higher levels of total energy expenditure (27,35), though there are no data on metabolic rate during sleep as a function of physical exercise. In conclusion, though the hypothesized experimental relationship was demonstrated, there are methodological considerations that do not allow a clear interpretation of the data in terms of the energy conservation model. In particular the data do not give direct support to the proximate causality version of the theory. Acknowledgments: Research reported in this paper was supported by Australian Research Grants Council awards and a University of Tasmania Special Research Grant to John Trinder and lain Montgomery. REFERENCES I ~ Allison T, Van Twyver H~ The evolution of sleep. Natural History 1970; 79: Berger RJ. Bioenergetic functions of sleep and activity rhythms and their possible 'relevance to aging. Fed Proc 1975; 34: ~ Heller HC, Walker JM, Florant GL, Glotzbach SF, Berger RJ. Sleep hibernation: eiectrophysiological and thermoregulatory homologies. In: Wang L, Hudson JW, eds, Strategies in cold: on natural torpidity and thermogenesis. New York: Academic Press, 1978: Murray EJ. Sleep, arousal and dreams. New York: Appleton Century Crofts, Snyder F. Toward an evolutionary theory of dreaming. Am J Psychiatry 1966; 123: Webb WB. Sleep behaviour as a biorhythm. In: Coloquhoun WP, ed, Biological rhythms and human performance. London: Academic Press, 1971;

10 168 I. MONTGOMERY ET AL. 7. Zepelin H, Re~htschalleIi A. Nlanuualian sleep, longevity, and energy conservation. Brain Behav Evol 1974; 10: Ellingson RJ. Development of wakefulness-sleep cycles and associated EEG patterns in animals. In: Clemente CD, Purpura DP, Mayer FM, eds, Sleep and the maturing nervous system. New York: Academic Press, 1972: Altman PL, Dittmer DS, eds, Metabolism. Bethesda: Federation of American Society for Experimental Biology, Eastman CI, Rechtschaffen A. Effect of thyroxine on sleep in the rat. Sleep 1979; 2: Desjardins J, Healey T, Broughton R. Early evening exercise and sleep. Sleep Res 1974; 3: Hobson JA. Sleep after exercise. Science 1968; 162: Horne JA, Porter JM. Time of day effects with standardized exercise upon subsequent sleep. Electroencephalogr elin Neurophysiol 1976; 40: Walker JM, Floyd TC, Fein G, Cavness e, Lualhati R, Feinberg I. Effects of exercise on sleep. J Appl Physiol 1978; 44: Webb WB, Agnew HW, Jr. Sleep and waking in a time free environment. Aerospace Med 1974; 45: Webb WB, Friedman J. Attempts to modify sleep patterns in the rat. Physiol Behav 1971; 6: Zloty RB, Burdick JA, Adamson JD. Sleep of distance runners. Acta Nerv Super (Praha) 1973; 15: Shapiro CM, Griesel RD, Bartel PR, Jooste PL. Sleep patterns after graded exercise. J Appl Physiol 1975; 39: Van Twyver H, Webb WB. Modifiability of a sleep pattern in the rat. Psychophysiology 1968; 4: Dunleavy DLF, Oswald I, Brown P, Strong JA. Hyperthyroidism, sleep and growth hormone. Electroencephalogr elin Neurophysiol 1974; 36: Adey WR, Bors E, Porter RW. EEG sleep patterns after high cervical lesions in man. Arch Neurol 1968; 19: Trinder J, Stevenson J, Paxton SJ, Montgomery 1. Physical fitness, exercise and REM sleep cycle length. Psychophysiology 1982; 19: Griffin SJ, Trinder J. Physical fitness, exercise and human sleep. Psychophysiology 1978; 15: Paxton SJ, Montgomery I, Trinder J, Newman J, Bowling A. Sleep after exercise of variable intensity in fit and unfit subjects. Aust J Psychol (in press). 25. Trinder J, Bruck D, Paxton SJ, Montgomery I, Bowling A. Physical fitness, exercise, age and human sleep. Aust J Psychol (in press). 26. Rechtschaffen A, Kales A, (eds). A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Brain Information Service/Brain Research Institute, University of California at Los Angeles, Astrand I, Rodhal K. Textbook of work physiology. New York: McGraw-Hill, Astrand I. Aerobic work capacity in men and women with special reference to age. Acta Physiol Scand 1960; 49, Suppl Borbely AA. Sleep in the rat during food deprivation and subsequent restitution offood. Brain Res 1977; 124: Iwahara S, Yang K-M, Yamazaki S. Psychophysiological effects of prolonged food and water deprivation in the albino rat. Jpn Psychol Rev 1970; 13: Karacan I, Rosenbloom AL, Londono JH, Salis PJ, Thornby 11, Williams RL. The effect of acute fasting on sleep and the sleep growth hormone response. Psychosomatics 1973; 14: MacFadyen UN, Oswald I, Lewis SA. Starvation and human slow wave sleep. J Appl Physiol 1973; 35: Parker DC, Rossman LG, Vanderlaan EF. Persistence of rhythmic human growth hormone release during sleep in fasted and non isocalorically fed normal subjects. Metabolism 1972; 21: Phillips F, Chen CN, Crisp AH, et al. Isocaloric diet changes and electroencephalographic sleep. Lancet 1975; 2: Bjorntorp P, Sjostrom L, Sullivan L. The role of physical exercise in the management of obesity. In: Munroe JF, ed, The treatment of obesity. Great Britain: MTP Press, 1979: