www.elsevier.com/locate/sleep

Sleep Medicine 9 (2008) 184–187

Brief Communication

Do we really want more sleep? A population-based studyevaluating the strength of desire for more sleep q

Clare Anderson *, James A. Horne

Sleep Research Centre, Department of Human Sciences, Loughborough University, Leicestershire LE11 3TU, UK

Received 24 November 2006; received in revised form 14 February 2007; accepted 14 February 2007Available online 16 July 2007

Abstract

Objective: The sleep literature increasingly points to an apparent chronic sleep debt in the general population. We investigated thisby examining perceived shortfalls in daily sleep, using more indirect questioning methods.Methods: To determine perceived sleep deficits, 10,810 adults completed a simple questionnaire, which avoided leading questionsand provided information on sleep length, daytime sleepiness, desired sleep length, a choice of attractive daytime activities in a ‘‘freehour’’, and ‘‘stressful lifestyle’’. From this we assessed whether deficits were reflected by increased daytime sleepiness or opting formore sleep when given attractive waking alternatives. Respondents were divided according to age and sex.Results: Half of men and women seemed to desire more sleep, but this apparent sleep deficit was not correlated with daytime sleep-iness, for any age or sex group. Irrespective of deficit, few people opted for sleep when given waking alternatives. ‘‘Stressful lifestyle’’was independently related to this sleep deficit.Conclusion: Desiring more sleep may also be synonymous with a need for more ‘‘time out’’, as sleep deficit was unrelated to daytimesleepiness but rather related to ‘‘stressful lifestyle.’’ Extra sleep may not be the only anodyne for sleep deficit.� 2007 Elsevier B.V. All rights reserved.

Keywords: Sleep deficit; Daytime sleepiness; Sleep need; Stress

1. Introduction

Seemingly, society is becoming sleepless, with manyhealthy adults chronically having insufficient sleep [1–3] but being unaware of daytime sleepiness [4]. It maybe why many people can sleep beyond their usual dailynorm and yet seem to desire more sleep [5]. However,the average 7–7.5 h sleep has changed little, historically[6–11], and ‘‘hidden’’ sleepiness is mostly only detectableby super-sensitive measures [4]. While the individualdesire for more sleep may be indicative of an accruing

1389-9457/$ - see front matter � 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.sleep.2007.02.006

q Declaration of interest: The authors declare they have no conflictsof interest, financial or otherwise.

* Corresponding author. Tel.: +44 0 1509 223005; fax +44 0 1509228480.

E-mail address: c.anderson@lboro.ac.uk (C. Anderson).

sleep debt, for some this desire may simply reflect a needfor more ‘‘time out’’ [5]. We assessed the perceived needfor more sleep by avoiding leading questions, and deter-mined how real the need seemed in terms of increaseddaytime sleepiness and in respondents opting for moresleep when given attractive alternatives.

2. Method

As part of a national BBC TV programme, ‘‘How tosleep better’’, an interactive website was establishedjointly by the authors and the BBC, inviting partici-pants to anonymously complete a short, simple sleepquestionnaire limited to 10 questions and orientedtoward evaluating perceived sleep deficits. These ques-tions assessed actual versus desired amounts of sleep.Resulting differences were compared with (i) usual

C. Anderson, J.A. Horne / Sleep Medicine 9 (2008) 184–187 185

levels of daytime sleepiness, and (ii) a choice of attrac-tive daytime activities (including ‘‘more sleep’’) if givenan extra free hour. Data were collected before pro-gramme transmission.

The questions and possible responses included thefollowing:

1. Do you feel you have a problem with sleep? (yes,no, sometimes)

2. What is your age? (years)3. What is your sex?4. What time do you normally fall asleep on a normal

day? (hh:mm)5. What time do you wake up on a normal day?

(hh:mm)6. What best describes your work hours? (normal

daytime, evening work, nightshift, rotating shifts,not applicable)

7. How much sleep do you feel you need each night?(h)

8. If you had an extra hour in the day, how wouldyou prefer to spend it? (playing sport/exercising;socialising; reading/relaxing; watching TV/film orlistening to radio; working; sleeping; other)

9. Epworth sleepiness scale10. Does your lifestyle often leave you feeling stressed

at the end of the day? (yes, no, sometimes)

Sleep period time (SPT) was calculated from ques-tions 4 and 5. The leading question, ‘‘How much moresleep do you need?’’ was avoided; information on sleepdeficit was derived from the difference between Q7 andSPT. This apparent deficit was categorised into Nil(60 h), Mild (0–1 h), Moderate (1–2 h) and Extreme(P2 h). Responses to Q8 were grouped as ‘‘inactive’’(relaxing, reading, watching TV), ‘‘active’’ (sport, exer-cise, socialising) and ‘‘sleep’’.

2.1. Participants

A total of 13,430 respondents aged 20–65 year com-pleted the questionnaire, which was manually screenedto exclude those providing incomplete/ambiguousanswers, shift workers, nightworkers, or those who wereunemployed. This left 10,810 (19.51% rejection, withmost [16.2%] being shift/nightworkers), who were subdi-vided by sex and four age categories (Table 1).

2.2. Daytime sleepiness

Daytime sleepiness was determined by the Epworthsleepiness scale (ESS [12]), assessing the likelihood offalling asleep under various circumstances. This scaleovercomes the problem of respondents being unawareof sleepiness, as it gauges actual falling asleep events,which are clearly indicative of sleepiness.

3. Results

All analyses of variance (ANOVAs) are two-way (sexand age). Significance levels for post-hoc Tukey testswere set conservatively at 0.001.

3.1. SPTs (Table 1)

There were main effects of age (F = 102.45,[3, 10,802], p < 0.0005) and sex (F = 89.439, [1,10,802],p < 0.0005), although the effect sizes were small (partialeta squared: age = 0.02; sex = 0.008); there was no sig-nificant interaction. Significant outcomes from theTukey tests showed the 20–29 year group sleeping longerthan all other groups, and the 30–40 year group sleepinglonger than older groups.

3.2. Perceived sleep need (Table 1)

Age (F = 90.77 [3;10802], p < 0.0001) and sex(F = 185.41, [1;10802], p < 0.0001) were again significantbut with no significant interaction. Again, the effect sizeswere small (partial eta squared: age = 0.2, sex = 0.17).Post-hoc tests showed younger groups to desire moresleep than both older groups.

3.3. Sleep deficit (Table 1)

Across all ages, 46.7% of men and 44% of womenreported wanting the same or less sleep than actuallyachieved. For men, the average deficit was 25.1 min(standard error (SE) = 1.7 min), and for women28.8 min (SE 1.7 min). There was a significant sex effect(F = 7.29, [1;10802], p < 0.007, partial eta squared =0.001) but no effect of age (p = 0.05) nor any interaction(p = 0.56).

3.4. Subjective daytime sleepiness – ESS (Table 1)

Across all ages, 19.8% of men and 19.9% of womenreported excessive daytime sleepiness (ESS P 10).There were significant age (F = 31.870 [3;10802],p < 0.0005) and sex (F = 43.26, [1; 10802], p < 0.001)effects, but with small effect sizes (<0.01). Tukey testsshowed the younger groups to have significantly lowerlevels of daytime sleepiness than the two older groups(except 20–29 year vs 30–39 year). However, withineach sex by age group there were no significant corre-lations between subjective sleepiness and extent of sleepdeficit (Table 1).

3.5. Choice of activity given extra hour (Q8)

Sleep deficits were compared with choices for theextra hour. Here, age bands were collapsed into two cat-egories (20–39 and 40–65 year), as the two younger and

Table 1Sleep and sleepiness characteristics of participants

Group (year) Sex n Mean age (y,m) Sleep period time a(h) Desired TST b(h) Sleep deficitc(h) ESSd Deficit · ESSe

20–29 Men 2370 24.11 (2.8) 7.41 (1.18) 7.79 (1.15) 0.39 (1.31) 7.09 (3.8) �0.07Women 2250 24.8 (2.8) 7.70 (1.14) 8.16 (1.14) 0.46 (1.28) 7.6 (3.7) �0.05

30–39 Men 1865 33.11 (2.10) 7.19 (1.13) 7.60 (1.02) 0.42 (1.23) 6.9 (3.8) �0.06Women 1294 33.10 (2.10) 7.55 (1.12) 8.00 (1.05) 0.45 (1.20) 7.2 (3.9) �0.05

40–49 Men 1125 44.1 (2.10) 6.96 (1.22) 7.46 (1.00) 0.50 (1.23) 7.4 (3.9) 0.04Women 711 44.2 (2.11) 7.26 (1.24) 7.8 (1.05) 0.54 (1.27) 5.5 (4.1) 0.17

50–65 Men 755 55.1 (4.1) 6.95 (1.36) 7.37 (1.08) 0.42 (1.37) 8.2 (4.2) 0.02Women 440 54.7 (3.8) 7.01 (1.51) 7.58 (1.11) 0.57 (1.47) 7.2 (4.1) 0.04

Total Men 6115 34.11 (10.8) 7.2 (1.21) 7.62 (1.08) 0.42 (1.28) 7.2 (3.9) �0.04Women 4695 32.11 (10.3) 7.52 (1.21) 8.01 (1.11) 0.48 (1.28) 7.1 (3.9) 0.00

a Sleep period time calculated from questions 1 (usual sleep onset time) and 2 (usual wake up time). Mean and (SD)�decimal hours.b Question 3 – How much sleep do you feel you need each night? – decimal hours.c ‘Desired’ minus ‘Actual’ � decimal hours.d Epworth sleepiness scale.e Correlation between sleep deficit and ESS score (r).

186 C. Anderson, J.A. Horne / Sleep Medicine 9 (2008) 184–187

two older groups within each sex were similar to eachother for deficits (see Table 1).

Fig. 1 shows how they chose to spend the extra time.Those responding ‘‘Other/Don’t know’’ to question 8(n = 354) were excluded. Irrespective of sleep group ordegree of sleep deficit, only a minority opted for ‘‘sleep’’(averaging 19% of men and 23.4% of women). There wasa significant difference between choice of activity forthose with a sleep deficit (F = 75.31, [3;10,802],p < 0.001); post-hoc tests showed those opting for sleephad significantly higher deficits (average, 49.8 min) than

Fig. 1. Percentage of men and women choosing how to spend an ‘‘extra houmoderate (1–2 h) or extreme (P2 h) sleep deficits.

those opting for inactive (22.8 min) or active (18 min)alternatives. However, the effect size was small (etasquared = 0.02), and there was no interaction betweenactivity choice and sex.

There was a significant effect of activity choice ondaytime sleepiness (F = 93.47, [3;10,802], p < 0.0001),with Tukey tests revealing those opting for sleep havinghigher ESS scores (8.2 ± 4.1) than those choosing inac-tive (7 ± 3.8) or active (7 ± 3.8) options. Again, thiseffect size was small (eta squared = 0.025), and therewas no interaction with sex.

r in the day’’, separated into those reporting nil (60 min), mild (0–1 h),

C. Anderson, J.A. Horne / Sleep Medicine 9 (2008) 184–187 187

3.6. Risk factors for sleep deficit

All data went into a logistic regression to identify sig-nificant (p < 0.01) risk factors for those reporting a sleepdeficit (i.e., >15 min). Data for perceived sleep deficitwere re-categorised as ‘‘yes’’ or ‘‘no’’ (P0.25 min vs60 min). Despite the low likelihood of opting for sleep,those opting for sleep had an increased risk (1.87, con-fidence interval (CI) 1.67–2.10) of having a sleep deficit.This is seen in Fig. 1, by comparing ‘‘nil deficit’’ col-umns with those indicating deficit. Additional signifi-cant risk factors for sleep deficit were ‘‘reportingproblem sleep’’ (1.82 increased risk, CI 1.67–1.99), and‘‘always stressed’’ (1.62, CI 1.42–1.85). Insignificant fac-tors included sex, age and daytime sleepiness(ESS P 10).

3.7. Good vs problem sleepers

Comparison between those who perceived themselvesto be good versus (sometimes/always) problem sleepersrevealed a significant difference with SPT (mean: 7.32 hvs 6.55 h. F = 670.86, df 1;10,808, p < 0.0005), but nodifference between them in apparent sleep need (mean:7.48 h vs 7.45 h), nor with ESS (average, 7.2 vs 7.2).

4. Discussion

Inasmuch as we find that young adults sleep longerthan older adults, and women sleep longer than men,these findings reflect similar outcomes from other stud-ies [6–11], and indicate that our population is not unu-sual in these respects. Whereas women, here, reporteda significantly greater sleep need than did men (averag-ing 8.0 h vs 7.37 h), the difference in apparent sleep def-icit was small (men: 25.2 min, women: 28.8 min).Women were more likely to choose the extra hour sleep-ing (19% versus 23.4%), but according to the ESS, theywere no sleepier than men (scoring 7.2 vs 7.3). Overall,these sleep deficits did not alter with age, although moreof the younger groups opted to take the extra hoursleeping (men: 17.7% for younger vs 15.5% for older;women: 23.7% vs 18.7%).

Few people elected to take more sleep, even thosehaving the greatest sleep deficit. Furthermore, therewas no link between the desire for more sleep and day-time sleepiness. Interestingly, despite over 50% of ourparticipants indicating this desire, only 20% reportedexcessive daytime sleepiness and opted to take extrasleep.

The little difference between good and problem sleep-ers in perceived sleep need points to their different sleep

deficits being largely due to differences in estimatedSPTs, rather than to a distorted perception of sleepneeds. Despite the larger apparent sleep deficit for thosewith a sleep problem, their greater sleepiness was smalland non-significant (0.8 and 0.2 of an ESS score formen and women, respectively). None of the correlationsbetween ESS scores and sleep deficit was greater than0.23 (Table 1).

While ours was not a cross-sectional study, with apotential bias towards viewers with a sleep problem,the sample was very large and may provide a usefulinsight into the perceived shortfall of sleep. The mainoutcome was that few people elected for more sleepwhen given other alternatives, even those with a sleepproblem and having the greatest apparent sleep deficit.

Acknowledgement

We thank the British Broadcasting Corporation forfacilitating this study.

References

[1] Spiegel K, Leproult R, Van Cauter E. Impact of sleep debt onmetabolic and endocrine function. Lancet 1999;354:1435–9.

[2] Dinges D. Sleep debt and scientific evidence. Sleep2004;27(6):1050–2.

[3] Dement WC. Sleep extension: getting as much extra sleep aspossible. Clin Sports Med 2005;24:251–68.

[4] Van Dongen HP, Maislin G, Mullington JM, Dinges D. Thecumulative cost of additional wakefulness: Dose–response effectson neurobehavioral functions and sleep physiology from chronicsleep restriction and total sleep deprivation. Sleep2003;26(2):117–26.

[5] Hublin C, Kaprio J, Partinen M, Koskenvuo M. Insufficient sleep– A population-based study in adults. Sleep 2001;24(4):392–440.

[6] McGhie A, Russell SM. The subjective assessment of normalsleep patterns. J Ment Sci 1962;108:642–54.

[7] Tune GS. The influence of age and temperament on the adulthuman sleep–wakefulness pattern. Br J Psychol 1969;60:431–41.

[8] Reyner LA, Horne JA. Gender and age differences in sleep,determined by home recorded sleep logs and actimetry. Sleep1995;18(3):127–34.

[9] Groeger JA, Zijlstra FRH, Dijk DJ. Sleep quantity, sleepdifficulties and their perceived consequences in a representativesample of some 2000 British adults. J Sleep Res2004;13(4):359–71.

[10] Lichstein KL, Durrence HH, Riedel BW, Taylor DJ, Bush AJ.Epidemiology of Sleep. New Jersey: Lawrence Erlbaum Associ-ates; 2004.

[11] Ursin R, Bjorvatn B, Holsten F. Sleep duration, subjective sleepneed and sleep habits of 40 to 60 year olds in the Hordalandhealth study. Sleep 2005;28(10):1260–9.

[12] Johns MW. A new method for measuring daytime sleepiness: TheEpworth Sleepiness Scale. Sleep 1991;14(6):540–5.