SEARCH

SEARCH BY CITATION

Keywords:

  • aging;
  • circadian rhythms;
  • human;
  • old;
  • sleep

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

The aim of this study was to determine whether naturally occurring inter-individual and intra-individual differences in bedtime selection in the elderly might be lawfully related to the amount of sleep that is obtained. A total of 128 seniors (63f, 65m) aged 70–92 years each provided a week of sleep diary data yielding a total of 896 subject-nights for analysis. From each subject-night the diary was used to derive measures of time in bed (TIB) and total sleep time (TST). These measures were used as dependent variables in mixed-effect linear models (nights nested within subjects) with the independent variable being bedtime for that subject-night, arbitrarily expressed as minutes since 19:00 hours. Although there were strong inter-individual and intra-individual differences, for both genders, bedtime had a statistically significant effect (P < 0.001) on both TIB and TST. We observed that later bedtimes were associated with less time in bed and less time asleep. On average between 7 and 8 min of less TIB and TST were associated with each 10-min delay in bedtime from 19:00 hours. These results are interpreted in terms of increases in sleep being derived from living in a better harmony with an earlier peaking circadian pacemaker characteristic of older age, although other possible mechanisms are also considered (e.g. age-dependent alterations in phase angle and homeostatic sleep need).


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

From both anecdotal observation and empirical evidence, it is clear that the sleep of older persons is quite different from that of younger adults. Compared to younger adults, even the most healthy of seniors spend about twice as much time during the night in unwanted wakefulness, and are more likely to wake up earlier in the morning than they would like to (Ohayon et al., 2004; Prinz, 2004). There are many possible reasons why this should be so, not least of which are the many troubles of ill health that advancing age is often heir to. Not only pain and discomfort, but also neurological disorders of late life such as Parkinson's disease and Alzheimer's disease are known to significantly impair sleep (Bliwise, 1993). Additionally, though it is plausible that at least some of this sleep dysfunction, especially in the healthy elderly, is a product of age-related changes in the phase (timing) of the human circadian system (reviewed by Monk, 2005; Prinz, 2004). In particular, the elderly often choose to go to bed and wake up several hours earlier than young adults (Buysse et al., 1992; Czeisler et al., 1992), have higher ‘morningness’ scores (more like a ‘morning lark’) on the Horne and Ostberg (1976) instrument (Monk et al., 1991), and have circadian phases (as indicated by temperature or melatonin rhythms) that are about 1 or 2 h earlier than those of the young (Duffy et al., 2002; Monk et al., 1995). This naturally raises the question of whether some of the sleep disruption of the elderly might be the result of bedtime choices at times that may be inappropriate for (i.e. at the wrong phase angle to) the timing of their endogenous circadian pacemaker (ECP). Thus, seniors might choose their habitual bedtime to be at a time that is at a different phase angle to their dim light melatonin onset (DLMO) or core body temperature minimum (Tmin), for example, than that chosen by younger adults (Carrier et al., 1999; Duffy et al., 2002). This hypothesis led to a body of work by which evening bright lights were used to delay the timing of the circadian pacemakers of those seniors experiencing unwanted early morning awakenings. Several authors have shown short-term beneficial effects on sleep of evening bright light treatment designed to accomplish such a phase delay (summarized in Campbell et al., 1995). Unfortunately, though, most older people do not like the light therapy, and a recent report has shown that maintenance light therapy for such patients is unlikely to work in the long term (Suhner et al., 2002). An alternative might be to leave the timing of an older person's circadian pacemaker alone, but to instead change the phase angle by changing the timing of bedtime. Before advocating such an approach, it would make sense to determine whether naturally occurring inter-individual and intra-individual differences in bedtime selection might be lawfully related to the amount of sleep that is obtained. That is the aim of the present study.

In an earlier study of morningness–eveningness in the elderly (Monk et al., 1991) we have shown not only that on average, older adults (80–91 years) had higher (Horne and Ostberg, 1976) morningness scores than younger (21–30 years) adults, but also that within the older group there was a significantly positive correlation (ρ = 0.38, n = 34, P < 0.05) between morningness score and total sleep time from polysomnography (PSG) indicating that more sleep was obtained in those scoring higher on morningness. PSG sleep was recorded (second night) at the subject's own habitual bedtime and wake time as calculated from a 2-week sleep diary before the laboratory sleep study. Thus, within the elderly group, the more morning-type the orientation, the more sleep was obtained. Of course, a morning-type orientation suggests, among other behavioral indices, an earlier bedtime. Therefore, one plausible explanation for this finding is that those elderly people who made bedtime choices appropriate to their earlier phasing ECP thus obtained more sleep than did those who failed to accommodate the ECP changes and made the ‘wrong’ timing-of-bedtime choices. This also suggests that even if the precise phase of the ECP (e.g. from DLMO or Tmin) is not known, diary data might allow the question of the relation between bedtime choices and sleep duration to be examined.

The present report considers diary data collected using the Pittsburgh Sleep Diary [PghSD] (Monk et al., 1994) which allows the examination of not only the timing of sleep, but also allows the derivation of subjective measures of sleep latency (SL) and wake after sleep onset (WASO). Thus one can obtain (admittedly inexact and subjective) estimates of time in bed (TIB) and total sleep time (TST) [TST = TIB − SL − WASO] from the PghSD, and determine their relation to the clock time at which bedtime was selected on that night. By accessing our database of diary sleep measures collected over several years, we were able to assemble a sample of 128 healthy seniors (63f, 65m, 70–92 years) who had completed the PghSD for a full week, and to use those data to determine the relationship between clock time of bedtime to TIB and TST on a subject-night basis.

Method

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

The data to be reported were collected in a variety of protocols concerned with the sleep of healthy seniors (sometimes as a control group), in addition to the two present studies (R01 AG13396 and P01 AG020677) which have primarily supported this analysis. In all cases, subjects completed the PghSD for at least seven consecutive days (often 2 weeks or more) while living normally at home. This analysis focused on data from the first seven continuous days of the subject's first period of diary data collection. Subjects were carefully instructed in the completion of the diary which usually took place immediately before a laboratory PSG study (for which subject payment was given). The need for precision in subjects’ diary responses was emphasized, and the diary was printed in large font with instructions that it was to be kept with a pen on the bedside table. Informed consent was obtained.

Subjects included in this analysis were healthy seniors (70–92 years) who were not seeking help from sleep complaints, had no unstable medical condition and no personal history of mental illness. Stable medical conditions (e.g. hypothyroidism, hypertension) which were under treatment by medications not known to affect sleep or circadian rhythms were allowed. Subjects who were shift workers, recently returned from other time zones, or who habitually did not spend between 6 and 9 h in bed at night were not recruited for the original studies and therefore did not appear in the present sample. This analysis is from a total of 128 subjects (65m, 63f) who each gave seven nights of data, yielding a total of 896 subject-nights, 455 for men, 441 for women. The mean (SD) age of the 65 men was 78.4 years (4.9 years) and that of the 63 women was 79.4 years (5.3 years). The oldest man was 91 years of age, the oldest woman 92 years, the youngest in both genders was 70 years.

The data were analyzed using linear mixed-effect models. Two separate models were fit to the data, one with TST as the response variable and one with TIB as the response variable, with both measures expressed in minutes. In both models, responses were nested within individuals across nights. To account for correlation of nights within individuals, random intercepts were included in the models as well as first-order autoregressive distributions on the error terms. These models included bedtime (arbitrarily expressed as minutes since 19:00 hours) as a covariate with a random effect and gender as a covariate with a fixed effect.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

Table 1 gives the mean, median, SD and range values of bedtime, TIB and TST for the 128 subjects as a whole, and also divided by gender.

Table 1.   Mean, SD and range values of bedtime, time in bed (TIB) and total sleep time (TST) for the 128 subjects as a whole (all) and also divided by gender [63 women (W), 65 men (M)]
VariableMeanMedianSDRangeMin.Max.
Bedtime (all)23:2323:300 h56 m6 h50 m19:4502:35
TIB (all)7 h 29 m7 h 30 m1 h 16 m10 h 30 m2 h 20 m12 h 50 m
TST (all)6 h 53 m6 h 56 m1 h 23 m11 h 50 m45 min12 h 35 m
Bedtime (W)23:2523:300 h 57 m6 h 0 m20:3502:35
TIB (W)7 h 27 m7 h 25 m1 h 20 m10 h 10 m2 h 30 m12 h 40 m
TST(W)6 h 49 m6 h 55 m1 h 30 m11 h 40 m45 min12 h 25 m
Bedtime (M)23:2223:250 h 55 m6 h 20 m19:4502:05
TIB (M)7 h 31 m7 h 30 m1 h 12 m10 h 30 m2 h 20 m12 h 50 m
TST (M)6 h58 m7 h0 m1 h16 m11 h20 m1 h15 m12 h35 m

Figure 1 gives the scattergrams illustrating the relationship between TIB and bedtime (left panel) and TST and bedtime (right panel). Best fitting regression lines from the mixed-effect models are also plotted. As can be seen from these plots and from Table 1, there was a significant degree of variation in TIB and TST. The intraclass correlations for TST and TIB were moderate, with values of 0.51 for TST and 0.56 for TIB. Overall, there was a relationship whereby on average later bedtimes were associated with shorter TIB and with shorter TST. The mixed-effect model with TIB as dependent variable showed a highly significant relationship of TIB with bedtime (estimated coefficient = −0.75 min, SE = 0.06, P < 0.001) but not with gender (reference value male, estimated coefficient = −1.31 min, SE = 10.25, P = 0.90). Likewise, the mixed-effect model with TST as dependent variable showed a highly significant effect of bedtime on TST (estimated coefficient = −0.72 min, SE = 0.07, P < 0.001) but not with gender (reference value male, estimated coefficient = −7.4 min, SE = 10.66, P = 0.49). Thus every 10-min delay in bedtime from 19:00 hours was associated with an average decrease of 7.5 min of TIB and a decrease of 7.2 min of TST. Additionally, we also fit mixed-effect models for both TIB and TST with gender and bedtime interactions. The interaction terms for TIB (estimated coefficient = −0.14 min, SE = 0.11, P = 0.21) and for TST (estimated coefficient = −0.19 min, SE = 0.13 P = 0.16) both failed to achieve significance at the 0.05 level.

image

Figure 1.  Scattergram of diary time in bed (TIB) (left panel) and diary time spent asleep (TST) (right panel) plotted against bedtime for all 896 subject-nights. Best fitting regression lines from the mixed-effect models are also plotted. In both cases, the effect of bedtime was highly significant (P < 0.001).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

The circadian pacemakers of seniors tend to have a circadian phase that is several hours earlier than the circadian phase of younger adults (Czeisler et al., 1992; Duffy et al., 2002; Monk et al., 1995). Thus, one interpretation of the present results might simply be that in one sense ‘it pays’ to recognize this age-related phase advance and to adopt an earlier bedtime accordingly. While there were many inter-individual and inter-night differences, on average those seniors who did adopt an earlier bedtime on a given night appeared to spend longer in bed and obtained more sleep on that night than those who adopted the later bedtimes more typical of younger adults. One might refer to this as the ‘Living in harmony with the circadian pacemaker leads to better sleep’ explanation, and it is attractive in its parsimony.

However, even from a strictly circadian point of view, it may be over-simplistic to consider circadian pacemaker phase as the only determining factor. Although the evidence in the scientific literature for a simple age-related circadian phase advance is compelling, there is some evidence which tends to complicate the picture. First, there is the point that this phase advance in circadian pacemaker timing may not necessarily be the only factor influencing the early bedtimes and wake times characteristic of advanced age. In a series of constant routine studies, Duffy et al. (2002) found the usual age-related phase advance in the timing of the circadian plasma melatonin rhythm, but also showed that the timing of habitual sleep differed with age in its phase angle to the melatonin rhythm. Thus, compared to the young, older subjects were not only waking earlier relative to clock time, but also earlier relative to the phase of the melatonin rhythm (and thus presumably to the phase of their ECP). Clearly, a simple age-related change in the timing of the pacemaker was not the whole story, and another influence (changing the phase angle) has to be invoked. A second, and perhaps related, issue is that of age-related changes in the homeostatic build up in sleep need over the day. As pointed out by Dijk et al. (2000), both circadian and homeostatic mechanisms need to be considered when understanding age-related changes in sleep and its timing. Thus, an older person's earlier bedtime may result from an earlier phasing circadian pacemaker, a lowered threshold of sleep need at bedtime and/or a complex interaction of the two. Again, the story appears to be much more complicated than a simple age-related change in circadian phase. One plausible explanation for the present findings, for example, might hold that differences in an elderly person's selected bedtime on a given night may result from increased homeostatic pressure for sleep before that night, which itself might lead to longer sleep (as well as to an earlier bedtime). A further issue that merits discussion is that of whether extra sleep and, more particularly, extra time in bed are actually desirable in the elderly. It may well be that for some seniors, a reduction in TIB may be beneficial in consolidating sleep and reducing the amount of unwanted wakefulness experienced during the night (Hoch et al., 2001).

An equally fundamental question relates to the direction of causality. While the present explanation of the results has been phrased in terms of sleep benefits derived from ‘Living in harmony with the circadian pacemaker’, there are several quite different interpretations which can be considered which bear much less directly on issues of circadian phase. Thus, for example, it may simply be that those who constitutionally need (and thus tend to get) more sleep recognize this and go to bed earlier than their peers who constitutionally need less sleep. However, this presupposes that they do so in order that both groups can wake up at the same time in the morning – an explanation that makes sense for working individuals, but is less compelling for the predominantly retired 70+ year olds studied here. Then there arises the question of why elderly long sleepers do not simply choose to sleep-in late in the morning. When we studied the 896 wake times, we found a mean of 06:53 with a standard deviation of 1 h16 m around that figure. Thus it seems that many of these seniors habitually awoke before 07:00 hours and that comparatively few of them choose to sleep-in and get up late. This then leads back again to the issue of an early circadian phase perhaps making that strategy undesirable for seniors.

The present results suggest the utility of a possible strategy by which seniors choose their bedtimes to be more in line with their earlier peaking circadian pacemakers. This may form the basis of a behavior-based therapy by which the sleep of seniors might be improved without the need for costly (and sometimes inappropriately prescribed) hypnotics. We are currently performing a laboratory study whereby the timing of the sleep of seniors is manipulated in the laboratory. In a within-subject design involving three conditions bedtimes are (i) advanced by 2 h; (ii) delayed by 2 h; or (iii) left the same; and the subsequent sleep, circadian rhythms, mood, alertness and performance studied for 72 h. When available, these laboratory results will help further to address the present question of whether earlier bedtimes in the elderly are a good thing, and may also begin to address the question of the mechanism by which such advantages might accrue.

Conclusion

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

Although the direction of causality and underlying mechanisms remain to be articulated, it would appear that there is a correlation whereby seniors who retire to bed earlier in the evening spend longer in bed and obtain more sleep than those who retire to bed later. Results appear to be very similar for the two genders.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

Sincere thanks to the faculty and staff of the various projects from which these diary data were drawn, to Margaret Rosenberg and Jean Miewald for data analysis, and to our wonderful subjects for their gracious co-operation. Primary support for this work was provided by National Institute on Aging grants R01 AG13396 and P01 AG020677. Other support was provided by NIH grants MH076981, MH52266, MH66227, MH61566, MH24652, PR054093, MH37869, MH71944, RR00056 and K25MH076981.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  • Bliwise, D. L. Sleep in normal aging and dementia. Sleep, 1993, 16: 4081.
  • Buysse, D. J., Browman, B. A., Monk, T. H., Reynolds, C. F., Fascizka, A. L. and Kupfer, D. J. Napping and 24-hour sleep/wake patterns in healthy elderly and young adults. J. Am. Geriatr. Soc., 1992, 40: 779786.
  • Campbell, S. S., Terman, M., Lewy, A. J., Dijk, D. J., Eastman, C. I. and Boulos, Z. Light treatment for sleep disorders: consensus report. V. Age-related disturbances [Review] [30 refs]. J. Biol. Rhythms, 1995, 10: 151154.
  • Carrier, J., Monk, T. H., Reynolds, C. F., Buysse, D. J. and Kupfer, D. J. Are age differences in sleep due to phase differences in the output of the circadian timing system? Chronobiol. Int., 1999, 16: 7991.
  • Czeisler, C. A., Dumont, M., Duffy, J. F., Steinberg, J. D., Richardson, G. S., Brown, E. N., Sanchez, R., Rios, C. D. and Ronda, J. M. Association of sleep-wake habits in older people with changes in output of circadian pacemaker. Lancet, 1992, 340: 933936.
  • Dijk, D. J., Duffy, J. F. and Czeisler, C. A. Contribution of circadian physiology and sleep homeostasis to age-related changes in human sleep. Chronobiol. Int., 2000, 17: 285311.
  • Duffy, J. F., Zeitzer, J. M., Rimmer, D. W., Klerman, E. B., Dijk, D. J. and Czeisler, C. A. Peak of circadian melatonin rhythm occurs later within the sleep of older subjects. Am. J. Physiol. Endocrinol. Metab, 2002, 282: E297E303
  • Hoch, C. C., Reynolds, C. F., Buysse, D. J., Monk, T. H., Nowell, P., Begley, A. E., Hall, F. and Dew, M. A. Protecting sleep quality in later life: a pilot study of bed restriction and sleep hygiene. J. Gerontol. B Psychol. Sci. Soc. Sci., 2001, 56: 5259.
  • Horne, J. A. and Ostberg, O. A self-assessment questionnaire to determine morningness–eveningness in human circadian rhythms. Int. J. Chronobiol., 1976, 4: 97110.
  • Monk, T. H. Aging human circadian rhythms: conventional wisdom may not always be right. J. Biol. Rhythms, 2005, 20: 366374.
  • Monk, T. H., Reynolds, C. F., Buysse, D. J., Hoch, C. C., Jarrett, D. B., Jennings, J. R. and Kupfer, D. J. Circadian characteristics of healthy 80  year olds and their relationship to objectively recorded sleep. J. Gerontol., 1991, 46: M171M175.
  • Monk, T. H., Reynolds, C. F., Kupfer, D. J., Buysse, D. J., Coble, P. A., Hayes, A. J., Machen, M. A., Petrie, S. R. and Ritenour, A. M. The Pittsburgh Sleep Diary (PghSD). J. Sleep Res., 1994, 3: 111120.
  • Monk, T. H., Buysse, D. J., Reynolds, C. F., Kupfer, D. J. and Houck, P. R. Circadian temperature rhythms of older people. Exp. Gerontol., 1995, 30: 455474.
  • Ohayon, M. M., Carskadon, M. A., Guilleminault, C. and Vitiello, M. V. Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan. Sleep, 2004, 27: 12551273.
  • Prinz, P. N. Age impairments in sleep, metabolic and immune functions. Exp. Gerontol., 2004, 39: 17391743.
  • Suhner, A. G., Murphy, P. J. and Campbell, S. S. Failure of timed bright light exposure to alleviate age-related sleep maintenance insomnia. J. Am. Geriatr. Soc., 2002, 50: 617623.