Self-reported ‘sleep deficit’ is unrelated to daytime sleepiness

Physiology & Behavior 96 (2009) 513–517

Contents lists available at ScienceDirect

Physiology & Behavior j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p h b

Self-reported ‘sleep deficit’ is unrelated to daytime sleepiness Clare Anderson ⁎, Charlotte R. Platten, James A. Horne Sleep Research Centre, Dept. Human Sciences, Loughborough University, Loughborough, Leicestershire, LE11 3TU, United Kingdom

a r t i c l e

i n f o

Article history: Received 17 September 2008 Received in revised form 10 November 2008 Accepted 13 November 2008 Keywords: MSLT Sleep need Sleepiness Anxiety

a b s t r a c t Seemingly, many healthy adults have accrued a sleep debt, as determined by findings based on the multiple sleep latency test (MSLT). However, our recent, extensive survey found self-reported sleep deficit was not linked to daytime sleepiness determined by the Epworth sleepiness scale (ESS). Here, we report on the link between self-reported sleep deficit and gold standard measures of sleepiness: MSLT, Psychomotor vigilance test (PVT) and Karolinska Sleepiness Scale (KSS). Habitual sleep time in forty-three participants, from using a week long sleep diary and actiwatch data, compared with self-ratings of how much sleep they needed, provided estimates of apparent sleep deficit or otherwise. They were split into categories: ‘sleep deficit’ (Av. − 47 min), ‘sleep plus’ (Av. 47 min) or ‘neutral’ (Av. 0 ± 15 min), depicting perceived shortfall (or excess) sleep. Although the deficit group desired to sleep longer than the other groups, they actually obtained similar habitual nightly sleep as the neutral group, but less than the sleep plus group. ‘Survival curves’ based on those falling asleep during the MSLT showed no difference between the groups. Neither was there any difference between the groups for the PVT, KSS, or ESS. Here, factors other than sleepiness seem to influence self-perceived sleep deficits. © 2008 Elsevier Inc. All rights reserved.

1. Introduction Sleep debt is apparently becoming endemic in modern society [23,10,11,18], seemingly because of increasing waking demands. Laboratory evidence seems to further indicate this [4,3], with, for example, 8 h in bed considered inadequate due to observed neurobehavioural deficits [25]. Population studies over the last 40 years have consistently shown average daily sleep for UK healthy adults to be 7–7.5 h [24,17,13,2]. Nevertheless, this would seem to be inadequate [25], and as such society may be harbouring a hidden sleep debt. Whilst the individual desire for more sleep may be due to many healthy adults perceiving themselves to have insufficient sleep, our previous research based on 10,810 UK adults [2], suggested that this desire depends on the type of questions asked of respondents and, for a sizeable portion, that their perceived sleep deficit was synonymous with wanting more ‘time-out’, rather than more sleep per se. Moreover, self-reported sleep deficit was unrelated to daytime sleepiness (Epworth sleepiness scale — ESS) [15] but more closely linked with perceived ‘stress’. Interestingly, as a probe for the strength of desire for more sleep, and given the choice of extra daily sleep or relaxing waking activities, most of those with an apparent sleep deficit opted for the latter. That is, although they would like more sleep they would not forfeit ‘freewaking time’ to take this sleep. As one might expect, short sleepers (6–7 h/night) are more likely to have a genuine sleep debt due to voluntary curtailment of sleep [16], and ⁎ Corresponding author. Tel.: +44 1509 223005; fax: +44 1509 223940. E-mail address: [email protected] (C. Anderson). 0031-9384/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.physbeh.2008.11.009

are more likely to have MSLT scores of less than 5 min, deemed to be ‘pathologically’ sleepy (cf. [21,4]). Despite such a MSLT score being likely to be associated with performance degradation and unintentional sleep episodes (cf. [6,4,8], subjectively, sleepiness is often not reliably reported by participants [5,25]), who may be in denial about their sleepiness. Ascertaining actual sleep need is difficult outside of the laboratory and reported shortfall of sleep is usually based on introspection. Although our findings [2] may have, in part, been distorted by this confound, the ESS, helps overcome this problem by gauging actual incidents of falling asleep. Nevertheless, more objective measures of habitual sleep duration alongside gold-standard measures of daytime sleepiness may provide further insight into the link between desire for more (or less) sleep and daytime sleepiness. In a controlled laboratory environment we investigated whether a perceived desire for extra sleep, based on comparisons with actigraphically measured habitual sleep times, is associated with increased daytime sleepiness, as determined by objective measures of sleepiness including the ‘gold standard’ MSLT and psychomotor vigilance test (PVT) e.g. [8] or whether it might be linked to factors other than sleepiness (e.g. anxiety). 2. Methodology 2.1. Participants Forty-three healthy young (26.3 y ± 4.2 y) male (n = 19) and female (n = 24) participants were recruited after screening thus: via questionnaire to exclude those who were nappers (≥2/week); were not

514

C. Anderson et al. / Physiology & Behavior 96 (2009) 513–517

Table 1 Descriptive statistics for the groups Group

n

Descriptor Male: Female Ratio

Sleep Deficit

12 PSN N HST

1:1.2

Neutral 13 PSN = HST Sleep 18 PSN b HST Plus

1:0.86 1:2

Group – Average

1:1.26

–

Age

Mean Difference (HST–PSN)

25.8 −47 min (minimum −15 min)† 26.5 b1 min 26.2 +47 min (minimum 15 min)† 26.3 6.6 min

HST = habitual sleep time; PSN = perceived sleep need. 0.0005

ESS

Av. Bedtime (hh:mm)

Av. Risetime (hh:mm)

6.45 23:56

07.38

6.7 7

23:49 23:52

07.22 07:37

6

23:52

07:32

Table 2 Survival analysis using Kaplan–Meier curves to estimate average Sleep Onset Latency across all four naps for the three groups taking into account censored data (‘censored’ = still awake: SoL 20 mins) Group Sleep deficit

†

Neutral

denotes significant at the Sleep plus

extreme ‘morning’ or ‘evening’ types; consumed less than 150 mg caffeine daily and less than 20 units alcohol weekly; without any sleep or medical problems other than minor illnesses; were not on any medication causing daytime sleepiness. Random drug (urine) testing ensured all participants were free from recreational drugs. The study was approved by the University's Ethical Advisory Committee and participants were remunerated for their time. On an initial day, participants underwent a 20-minute practice session on the PVT to remove any practice effects and were familiarised with the laboratory environment prior to the main study day. 2.2. Procedure 2.2.1. Sleep times Participants wore actiwatches and kept sleep diaries for one week as a check on sleep habits and to obtain habitual sleep times (HST). They completed personality and state-trait anxiety questionnaires (see below), and a sleep questionnaire which included the (non-leading) question “how much sleep do you feel you need”, giving perceived sleep need (PSN). This was not used to select participants, but to group them into varying degrees of discrepancy between PSN and HST (i.e., reported sleep debt). Discrepancies between PSN and HST were categorised as described in Table 1. 27.9% of participants felt they required more sleep (at least 15 min: n = 12), 41.9% required less (at least 15 min: n = 18) and 30.2% were happy with their sleep length (n = 13). 2.2.2. Multiple sleep latency test (MSLT) On the main study day participants came to the laboratory at 09.00 h and were fitted with electrodes using the standard MSLT C3-

MSLT 10:00 12:00 14:00 16:00 Av. MSLT 10:00 12:00 14:00 16:00 Av. MSLT 10:00 12:00 14:00 16:00 Av. MSLT

Censored

Estimated SoL ± 95% CI

n

%

7 3 3 5 1 8 5 5 5 2 11 7 9 10 5

53.8 25 25 41.7 8.3 61.5 38.5 38.5 38.5 15.4 61.1 38.9 50 55.6 27.8

15.95 ± 1.57 11.76 ± 1.96 12.09 ± 1.73 13.25 ± 1.87 13.27 ± 1.50 17.53 ± 1.01 13.72 ± 1.54 16.45 ± 1.54 14.31 ± 1.42 14.89 ± 1.09 18.38 ± 0.73 15.19 ± 1.29 16.39 ± 1.15 15.66 ± 1.33 16.40 ± 0.96

A1, C4-A2 (C3–A2, C4–A1, EOG, EMG). They underwent standard MSLT testing at 10.00 h, 12.00 h, 14.00 h 16.00 h according to standard test criteria [5]. During each of these tests they were instructed to lie quietly with eyes closed and try to go to sleep. They were awoken when sleep onset criteria was met: the first three consecutive (30 s) epochs of stage 1 sleep (containing at least 50% sleep) or the 1st appearance of stage 2 sleep, whichever appeared first [20]. If the participant was still awake after 20 min, the test was terminated. EEG was recorded using Embla N7000©. EEG electrode impedances were maintained at b5 KΩ. High pass digital filtering (using finite impulse response digital filters) was set at 0.3 Hz and low pass filtering was at 40 Hz. EEG data were sampled at 100 Hz. 2.2.3. Psychomotor vigilance testing (PVT) An extended 30 min PVT [9] session was conducted at 16.30 h. Here, participants sat at a computer screen with their preferred finger on a response button (usually a thumb or index finger of the dominant hand) with which they responded as soon as a digital millisecond clock appeared on the screen. Interstimulus intervals averaged 7 s within a range 2–12 s. 2.2.4. Karolinska Sleepiness Scale (KSS) Subjective sleepiness (KSS) [1] was rated bi-hourly, commencing at 10.00 h. This is a 9 point scale: 1 = extremely alert, 2 = very alert, 3 = alert, 4 = rather alert, 5 = neither alert nor sleepy, 6 = some signs of sleepiness, 7 = sleepy, no effort to stay awake, 8 = sleepy, some effort to stay awake, 9 = very sleepy, great effort to keep awake, fighting sleep. 2.2.5. Psychometrics Participants completed the State-Trait Anxiety Inventory (STAI) [22] and Eysenck Personality Questionnaire (EPQ)[12]. The STAI is a 40-item questionnaire determining specific situation (state) and general trait anxiety. The EPQ provides personality characteristics, such as extroversion (i.e. sociable, dominant, impulsive, and active), neuroticism (i.e. anxious, depressed, moody, tense) and psychotocism (i.e. aggressive, dogmatic, and egocentric). 2.3. Statistical analyses

Fig. 1. Habitual Sleep Times (HST) and Perceived Sleep need (PSN) for those who selfreport too much (plus), too little (deficit) or adequate (neutral) nightly sleep. The deficit group took significantly less sleep then the plus group and desired more. The neutral group had similar HST as the deficit group but were satisfied with the amount of sleep taken.

One-way ANOVAs were used to compare difference between the three groups. Where data violates the assumption of homogeneity of variance, Brown-Forsythe corrections are reported. To overcome statistical problems caused by non-sleep episodes (20 min — censored data) on the MSLT, survival analysis with Kaplan–Meier survival curves are plotted. Survival curves show, for each time plotted

C. Anderson et al. / Physiology & Behavior 96 (2009) 513–517

515

3. Results Groups were split according to their HST and PSN as per Table 1. Paired t-tests for each group confirmed the sleep deficit group had a significantly negative discrepancy between HST and PSN (t = −7.930, df. 11, p b 0.0005) as they desired more sleep. The HST and PSN for the neutral group did not differ (p = 0.898) and the sleep plus group felt they needed significantly less sleep than they took (t = 5.976, df. 17, p b 0.0005). There were no significant differences between the groups for age (p = 0.992) or ESS (p = 0.87). Analyses confirmed differences between groups: for HST (F = 3.807, df. 2, 42, p b 0.03), with the sleep deficit group having lower HST than the sleep plus group (p = 0.025) — see Fig. 1 and for PSN (F = 14.933, df. 2, 32.128, p b 0.0005), with the sleep deficit group wanting more sleep than the neutral (p = 0.048) and sleep plus groups (p = 0.0005), and the sleep plus group desiring less than the neutral group (p = 0.03).

Fig. 2. Kaplan–Meier Survival curves for the three perceived sleep deficit groups. Cumulative survival equates to sleep onset latencies whereby 1.0 is 100% survival (awake). There were no between group difference in estimated latencies.

(minute), the portion of individuals “surviving” (i.e. still awake) at that time. A log rank test will be used to assess equality of survival functions (awake) across all time points.

3.1. MSLT The percentage of MSLT naps when sleep was achieved was 48.6% for those satisfied with their sleep (‘neutral’) compared with 62.5% and 55.8% for those categorised as ‘sleep deficit’ or ‘sleep plus’ respectively. Fishers exact test showed no between group differences, here. For all groups combined, no correlation was found between HST, PSN or amount of sleep deficit and MSLT (r b 0.29, p N 0.06), nor for any

Fig. 3. Kaplan–Meier Survival curves for each MSLT session (upper left: 10:00 h; upper right 12:00 h; lower left: 14:00 h; lower right: 16:00 h) for the three groups. Cumulative survival equates to sleep onset where by 1.0 is 100% survival (awake). There were no between group differences in estimated sleep onset latency, for any MSLT session.

516

C. Anderson et al. / Physiology & Behavior 96 (2009) 513–517

Fig. 4. KSS over time for the three groups.

group individually (p N 0.152). One-way ANOVA showed no effect of group on average MSLT (p = 0.191). However, treating MSLT data with standard statistical techniques can be problematic as those participants who do not fall asleep pose problems with censored observations (i.e., all receiving the maximum score of 20 min). Most analyses (i.e., regression) treat these as continuous variables; thus distorting the outcome. Accordingly, analysis with Kaplan–Meier survival curves were undertaken (cf. [7,19]. Table 2 shows the survival analysis for sleep onset latency estimates, with 95% confidence intervals for all nap times and average MSLT sleep onset latencies. Log-rank (Mantel–Cox) tests showed no significant difference between groups (X2 = 2.65, df. 2, p = 0.265) for average sleep onset latency (see Fig. 2), nor for any individual MSLT session (p N 0.16) — see Fig. 3. 3.2. PVT performance For all groups combined, there was no overall correlation between HST and PVT performance (Mean RT: p = 0.154; PVT Lapse p = 0.777). Comparison between groups with a One-way ANOVA showed no effect of group on mean RT (p = 0.455) or number of lapses (p = 0.768). 3.3. Subjective sleepiness For all groups combined, there was no correlation between HST and ESS (p = 0.786). The One-Way ANOVA showed no significant effect of groups for ESS (p = 0.868), as scores were comparable for all three groups (See Table 1). Fig. 4 shows the change in KSS scores throughout the testing day for each group. A mixed two-way ANOVA (time X group) revealed no effects of time (p = 0.673), group (p = 0.925) nor interaction (p = 0.412).

Our previous research [2] addressed why individuals desired more sleep, and reported that, for many, this was not based on daytime sleepiness. In support of those findings we again report no link between perceived shortfall of sleep and daytime sleepiness from using the MSLT and PVT. Those desiring extra sleep were no more likely to fall asleep in the MSLT setting, nor fall asleep any faster than the other groups. Furthermore, there was no correlation between amount of deficit and sleep onset latency. Subjective sleepiness (ESS and KSS) also bore no resemblance to perceived shortfall of sleep. Contrary to [16] findings with their 17 participants, HST was not correlated with MSLT in our study. Whilst this may, in part, be due to our relatively short range of HSTs (6.5 h–8.5 h) versus 6.3 h–10.1 h reported by Klerman and Dijk, our participants were also less likely to fall asleep during the MSLT (our results are more similar to those of [19], and had a higher prevalence of maximum (20 min) MSLT scores in our dataset. Participants who do not fall asleep during the MSLT represent censored observations, which pose a problem, especially for correlations. Here, these censored observations are treated as continuous measurements of sleep latency, thereby reducing a bias which is apparent in many other findings which are uncensored. Our survival analysis (also utilised by e.g. [19,7] models censored observations. However, we still report no significant difference between groups for cumulative survival (i.e., no sleep onset). Those satisfied with their sleep had an estimated MSLT score of 14.43 min compared to 13.16 min for those who desired more sleep (neutral 14.53 min). The neutral and sleep deficit groups had similar cumulative survival, although, the greatest difference between these two groups was at the 14:00 h MSLT session, however, this was non significant. Moderate changes in sleep length are likely to impact sleepiness during the ‘post-lunch dip’ [14], and so, any difference between the groups would have been more apparent here. Whilst previous work [16] has commented that shorter habitual bedrest duration reflected sleep debt and not basal sleep need, our study focussed on perceived sleep need and, thus, was novel in this respect. Our sleep deficit and neutral groups had similar HSTs (12 min difference) and estimated MSLT scores (10 s difference), but differed by 35 min in perceived sleep need. Our previous work indicated that stress and anxiety may be factors [2]. However, here, we found no significant difference in trait anxiety between the three groups. This may be, in part, due to our sample all scoring relatively low for trait anxiety when compared with STAI scale norms (range 21–42) [22]. State anxiety also had no link with the desire for extra sleep. Although we find no link between perceived shortfall of sleep and the gold standard measures of sleepiness in our ‘normal sleepers’, for other people who consider themselves to be suffering from insomnia, selfdiagnosed inadequate sleep can lead to unwarranted use of hypnotic medication; a condition which may indeed be related to anxiety.

3.4. Anxiety/personality Acknowledgments Given that perceived sleep need was not linked to daytime sleepiness, we then compared the desire for more sleep with anxiety. Psychometric data was only collected from 36 participants. However, a one-tailed one-way ANOVA showed no effect of state (p = 0.293) or trait (p = 0.405) anxiety, nor any effect of any personality (p N 0.362) on the desire for more sleep. 4. Discussion Those who desire more sleep (apparent sleep deficit) had significantly shorter HSTs than those who perceived their sleep to be in excess of need (7.29 h vs. 7.58 h). However, neither group differed from the control group (7.41 h). Despite the sleep deficit group sleeping at the average UK 7.5 dh norm [2,13,24,17] their perceived sleep need was significantly higher (8.16 h) than the neutral (7.41 h) and sleep plus (7.08 h) groups.

This study was funded by the Economic and Social Research Council, Ref: RES-000-23-0954. The authors would like to thank Kate Jordan for her help with data collection. References [1] Åkerstedt T, Gillberg M. Subjective and objective sleepiness in the active individual. Int J Neurosci 1990;52(1–2):29–37. [2] Anderson C, Horne JA. Do we really want more sleep? A population-based study evaluating the strength of desire for more sleep. Sleep Med 2008;9:184–7. [3] Belenky G, Wesensten NJ, Thorne DR, Thomas ML, Sing HC, Redmond DP, et al. Patters of performance degradation and restoration during sleep restriction and subsequent recovery: a sleep-dose response study. J Sleep Res 2003;12:1–12. [4] Carskadon MA, Dement WC. Cumulative effects of sleep restriction on daytime sleepiness. Psychophysiology 1981;18:107–13. [5] Carkadon MA, Dement WC, Mitler MM, Roth T, Westbrook TR, Keenan S. Guidelines for the Multiple Sleep Latency test (MSLT): a standard measure of sleepiness. Sleep 1986;9(4):519–24.

C. Anderson et al. / Physiology & Behavior 96 (2009) 513–517 [6] Carskadon MA, Harvey K, Dement WC. Sleep loss in young adults. Sleep 1981;4:299–312. [7] Carskadon MA, Harvey K, Duke P, Anders TF, Litt IF, Dement WC. Pubertal changes in daytime sleepiness. Sleep 1980;2:453–60. [8] Dinges DF, Pack F, Williams K, Gillen KA, Powell JW, Ott GE, et al. Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4–5 hours per night. Sleep 1997;20(4):267–77. [9] Dinges DF, Kribbs NM. Performing while sleepy: effects of experimentally-induced sleepiness. In: Monk TH, editor. Sleep Sleepiness & Performance. Chichester, John Wiley & Sons; 1991. p. 97–128. [10] Dinges D. Sleep debt and scientific evidence. Sleep 2004;27(6):1050–2. [11] Dement WC. Sleep extension: getting as much extra sleep as possible. Clin Sports Med 2005;24:251–68. [12] Eysenck HJ, Eysenck SBG. Manual of the Eysenck Personality Scales (EPS Adult). London: Hodder & Stoughton Educational; 1991. [13] Groeger JA, Zijlstra FRH, Dijk DJ. Sleep quantity, sleep difficulties and their perceived consequences in a representative sample of some two thousands British adults. J Sleep Res 2001;13:359–71. [14] Harrison Y, Horne JA. Long-term extension to sleep — are we really chronically sleep deprived? Psychophysiology 1996;33(1):22–30. [15] Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep 1991;14(6):540–5. [16] Klerman EB, Dijk DJ. Interindividual variation in sleep duration and its association with sleep debt in young adults. Sleep 2005;28(10):1253–9.

517

[17] McGhie A, Russell SM. The subjective assessment of normal sleep patterns. J Ment Sci 1962;108:642–54. [18] National Sleep Foundation. “Sleep in America” Poll; 2003. [19] Punjabi NM, Bandeen-Roche K, Young T. Predictors of objective sleep tendency in the general population. Sleep 2003;26(6):678–83. [20] Rechtschaffen A, Kales AA. Manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Washington D.C.: U.S. Government Printing Office; 1968. [21] Richardson GS, Carskadon MA, Flagg W, Van den Hoed J, Dement WC, Mitler MM. Excessive daytime sleepiness in man: multiple sleep latency measurement in narcoleptic and control subjects. Electroencephalogr Clin Neurophysiol 1978;45:621–7. [22] Spielberger CD, Gorsuch RL, Lushene RE. State Trait Anxiety Inventory (STAI). Consulting Psychologists Press; 1970. [23] Speigel K, Leproult R, Van Cauter E. Impact of sleep debt on metabolic and endocrine function. Lancet 1999;354:1435–9. [24] Tune GS. The influence of age and temperament on the adult human sleep-wakefulness pattern. Br J Psychol 1969;60:431–41. [25] Van Dongen HPA, Maislin G, Mullington JM, Dinges DF. The cumulative cost of additional wakefulness: Dose response effects on neurobehavioural functions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep 2003;26(2):117–26.