Jacques Taillard, Clinique du Sommeil, CHU Pellegrin, 33076 Bordeaux cedex, France. Fax: +33 (0) 5 56 79 48 06; e-mail: email@example.com
The purpose of this study was to determine, in a large sample of adults of all ages (17–80 years), the effect of morningness/eveningness on sleep/wake schedules, sleep needs, sleep hygiene and subjective daytime somnolence. A total of 617 subjects (219 subjects per chronotype group) matched for age, sex and employment status, completed an abridged morningness/eveningness questionnaire, a questionnaire on sleep habits and the quality of sleep, and the Epworth Sleepiness Scale. Eveningness was associated with a greater need for sleep, less time in bed during the week compared to ideal sleep needs, more time in bed at the weekend, a later bedtime and waking-up time especially at the weekend, more irregular sleep/wake habits and greater caffeine consumption. These subjects built up a sleep debt during the week and extended their duration of sleep at the weekend. They did not, however, rate themselves more sleepy than other types, despite the fact that our results showed a clear link between subjectively evaluated daytime somnolence and sleep debt. Why they were less affected by sleep deprivation is not clear. This raises the question of individual susceptibility to the modification of sleep parameters.
In a 7/13 sleep/wake paradigm Lavie and Segal (1989) observed a phase difference of 2 h in the circadian alertness and sleepiness rhythm between morning type (M-types) and evening type subjects (E-types). Lack and Bailey (1994), applying an hourly sleep test under constant routine, came to similar conclusions. Kerkhof and van Dongen (1996), studying body temperature under constant routine, have likewise shown that morningness/eveningness results from a difference in the circadian phase (2.1 h) of the endogenous clock.
This phase difference can explain interindividual variations in behaviour, and particularly sleep/wake schedules. Carrier et al. (1997) have shown that morningness/eveningness can vary with age and has a marked influence on habitual sleep patterns and sleep parameters in the middle years of life.
E-types tend to vary considerably their bedtime, waking up time and length of sleep (Kerkhof 1985; Ishihara et al. 1988; Monk et al. 1994) and are capable in perfectly normal circumstances of sleeping over 10 h (the ‘sleep extensors’; Violani et al. 1997). This variability is essentially as a result of the difference between weekday and weekend sleep/wake schedules. On working days, when social determinants prevent them from choosing their preferential sleep schedules, E-type workers report less sleep than M-type workers (Ishihara et al. 1988). Another study has shown that within a framework of imposed sleep/wake schedules (better suited to the preferential sleep/wake schedules of M-types), young E-types are sleepier during the day, especially in the morning, than young M-types (Clodoréet al. 1990; Volk et al. 1994; Andrade and Menna-Barreto 1995). Complaints of sleepiness in the morning by E-types could be related to a sleep debt caused by their late bedtime or by the phase delay of their sleepiness rhythm.
The present study, based on subjective evaluation, seeks to clarify three questions.
•Are E-types sleep-deprived during the working day?
•Do they have less satisfactory sleep hygiene than M-types?
•Are E-types in consequence more frequently and severely sleepy than M-types?
The study was carried out on a sample population aged from 17 to 81 years.
2007 subjects (range; 17–81 years) took part in the study, which was integrated into a survey on sleep deprivation and automobile driving, and conducted at two toll booths near Bordeaux and Biarritz (France). As soon as an interviewer was available, drivers were systematically approached by the highway patrol and asked to participate.
Subjects first completed an abridged version of the Horne and Ostberg questionnaire in order to assess morningness/eveningness (Adan and Almirall 1991). This questionnaire establishes five behavioural categories (English version scoring): definitively morning types (score=22–25), moderately morning types (score=18–21), neither types (score=12–17), moderately evening types (score=8–11) and definitively evening types (score=4–7). For the purpose of this study we reduced the categories from five to three: morning type (score=18–25), neither type (score=12–17) and evening type (score=4–11).
Participants were then asked to reply to the Basic Nordic Sleep Questionnaire (BNSQ; Partinen and Gislason 1995), which includes more precise questions on usual sleep schedules, sleep need, sleep quality, sleep hygiene, sleep medication and breathing sleep disorders. Two of these questions concerned the frequency of daytime somnolence at work (have you suffered from an irresistible tendency to fall asleep while at work?) and during the day in general (do you feel excessively sleepy during daytime?). Replies were classed on a five-point scale of frequency (1: never or less than once per month; 5: daily or almost daily). Sleep hygiene, especially variation of bedtime and/or arising time, was evaluated by a supplementary question (do you vary your bedtime and/or arising time by more than 2 h?). The reply was classed on the same five-point scale as the BNSQ. The general level of daytime sleepiness was evaluated on the Epworth Sleepiness Scale (Johns 1991). Daily caffeine intake was also assessed. Usual time in bed (TIB) was defined as the duration between bedtime and waking-up time. Daily sleep debt or sleep gain during the working week was calculated by subtracting usual time in bed during the week from ideal sleep need (how many hours of sleep do you need per night? How many hours would you sleep if you had the possibility to sleep as long as you need to?). In this case sleep debt became a positive value. Sleep extension is defined as weekend TIB minus weekday TIB.
Intra- and intergroup comparison of sleep quantity variables was performed using the Wilcoxon and Kruskal–Wallis tests (KW) followed by the U Mann Whitney test (MW, for post hoc comparison). Variables were ranked using the Spearman correlation (rho). We used the Chi-square test to evaluate the relationship between sleep chronotypes and sleep quality, sleep hygiene and frequency of daytime somnolence. Results are expressed as mean± SD.
The sample interviewed consisted of 40.2% M-types (48±12 years), 48.9% N-types (40±13 years) and 10.8% E-types (35±11 years). The distribution of chronotypes was skewed towards morningness (chronotype score: 14.6±4.3). There was a significant correlation between morning/evening score and age (rho=0.38, P < 0.0005). E-types were much younger than M-types and N-types (KW, P < 0.0005).
In order to construct comparable groups, we matched the 219 E-types by age, sex, employment status (in, or not in active employment) with 219 M-types and 219 N-types.
The M-type group was composed of 56 females and 163 males (mean=36.8, SD=10.1, range=19–73), the N-type group of 53 females and 166 males (mean=35.4, SD=11.4, range=17–80), the E-type group of 55 females and 164 males (mean=35.4, SD=11.4, range=17–79). In the M-type group, 26 subjects were definitely morning type, 193 subjects were moderately morning type. In the E-type group, 192 subjects were moderately evening type and 27 subjects were definitely evening type.
There was no significant difference in nocturnal sleep breathing disorders between the three groups. E-types considered their sleep less refreshing than N-types (χ=17.3, P=0.02). M-types woke up more often during the night than E-types and N-types (χ=26.2, P=0.001). There was no significant difference in the consumption of hypnotics between the three groups.
Table 1. Sleep characteristics and Epworth score according to chronotype
As one would expect, during the week E-types went to bed later and woke up later than M-types and N-types. Although E-types subjectively expressed a greater need for sleep than other types (about 20 min by night, P < 0.03) they did not sleep more than the others on week nights. This group consequently presented a slight but significant daily sleep debt of 9±90 min (KW, P < 0.001), whereas the M- and N-groups showed a daily sleep excess.
As shown in Fig. 1, twice as many E-types as M-types presented a daily sleep debt, a daily sleep debt ≥1 h or a daily sleep debt ≥2 h. Ideal sleep duration was reduced by as much as 25% in 11.4% of E-types and only in 4.6% of M-types (χ=9.8, P=0.007). Three subjects restricted their weekday daily sleep to 50% of their ideal sleep duration (one N-type and two E-types).
During the weekend all subjects slept more, went to bed later and woke up later (Wilcoxon, P < 0.0005). E-types slept more than M-types (MW, P < 0.001), went to bed later (MW, P=0.03) and woke up later than other types (MW, P < 0.0005). M-types delayed their bedtime by 30 min and their arising time by 90 min. N-types and E-types delayed their bedtime by 60 min and their arising time by 2 h and 2.3 h, respectively. During the weekend 24.6% of M-types, 35.5% of N-types and 39.9% of E-types could be qualified as ‘sleep extensors’ (subjects capable of sleeping more than 10 h).
No correlation was found between weekend TIB and daily sleep debt, but there was a significant link between weekend TIB and ideal sleep duration (rho=0.237, P < 0.005). A positive correlation was also found (rho=0.349, P < 0.005) between sleep debt and weekend sleep extension (weekend TIB minus weekday TIB) and a negative correlation (ρ=−0.414, P < 0.005) between weekday TIB and weekend sleep extension.
E-types more often varied their bedtime and arising time by over 2 h than other groups (χ=37.7, P < 0.001). 50.2% of M-types, 41.1% of N-types and 27.5% of E-types never modified their sleep schedules. 5.9% of M-types, 9.5% of N-types and 20.1% of E-types modified their sleep schedules more than three times a week.
There were more smokers in the E-group than in the other groups (χ=16.6, P < 0.001). E-type smokers smoked more at bedtime than other smokers (χ=35.5, P < 0.001). E-types also consumed more coffee than other types (KW, P < 0.03, Table 1).
E-types did not rate themselves more sleepy than other types during the week (χ=7.8, NS) or at work (χ=8.8, NS).
No differences were observed in ESS scores between groups (Table 1); in the E-group, however, a correlation was observed between ESS score and the morningness/eveningness score (rho=−0.150, P=0.02). Morningness was associated with less daytime sleepiness.
A significant correlation was found for all groups between sleep debt and subjective somnolence but this correlation was more important in E-types (Table 2). It was also in E-types that we found the closest correlation between variation of sleep schedules and daytime sleepiness (rho=0.14, P=0.04).
Table 2. Spearman correlation between daily weekday sleep debt and daytime somnolence: in total sample and in M-types, N-types and E-types
E-types tended more than others to reduce their weekday time in bed. During the week 11.4% of this group slept only 75% of their ideal sleep duration against 4.6% of M-types. It has been shown that a repeated sleep restriction of as little as 2 h per night causes a cumulative increase in daytime sleepiness and waking neurobehavioural deficit (Carskadon and Dement 1981). In E-types, this sleep debt is not necessarily due to a reduction of sleep duration imposed by socio-economic schedules (Ishihara et al. 1988) but more to a greater need for sleep expressed by this chronotype. Another study, using regular sleep diaries and taking into account TIB during the week and the weekend, has also reported that morningness is associated with less TIB (Carrier et al. 1997). We found in the present study that during the weekend E-types slept more, went to bed and woke up later than M-types. They delayed their sleep wake schedules more than M-types, especially their arising time. This observation concords with previous reports on the sleep habits of E-types (Ishihara et al. 1988), the irregularity of their sleep/wake schedules (Monk et al. 1994) and their capacity to extend sleep duration (Violani et al. 1997).
Roehrs et al. (1983) suggested some time ago that an important difference between weekend and weekday sleep habits could suggest insufficient sleep. Our results confirm this hypothesis in that they show a link for all types between subjective daytime somnolence and the weekday sleep debt. There is also a link between the irregularity of sleep/wake schedules and subjective sleepiness, as previously shown (Manber et al. 1996). However, the E-types of our study, who cumulate repeated episodes of sleep restriction and irregular sleep/wake schedules and who logically should suffer most from daytime somnolence, do not consider themselves more somnolent during the day than N-types and M-types. Why they are so relatively unaffected by daytime sleepiness is not at all clear. Several explanations are possible.
•E-types may overestimate their ideal sleep needs; nonetheless they increase their sleep during weekend nights to reach these ideal sleep needs.
•E-types may underestimate their subjective sleepiness.
•E-types would show more plasticity with respect to sleep duration and sleep/wake schedules than other chronotypes (Ishihara et al. 1988). They may thus be less sensitive to sleep restriction than M-types, an observation that would concord with the results of a study carried out by Volk et al. (1994) on young subjects. This could explain why they adapt better to shift work (Akerstedt and Torsvall 1981).
•Significant daily caffeine intake could counteract the effects of repeated sleep restriction (Rosenthal et al. 1991).
There are some obvious limitations to the present study. It may be difficult for some individuals to estimate their need for sleep as well as their usual sleep schedules, especially when they often vary these schedules (e.g. E-types). Napping was not taken into account, although it has been shown that a short nap can have a positive effect on alertness (Gillberg et al. 1996). On a more general level, subjects’ perception of their sleep needs may be exaggerated. Once over the limit of essential sleep needs, variations may be of little importance and a 1.5/2 h weekday reduction in sleep may have little or no effect on the level of daytime somnolence (Horne and Wilkinson 1985).
The fact remains nonetheless that, in this survey, E-types would appear to react differently to sleep deprivation from other chronotypes. This question of individual susceptibility to the modification of sleep parameters needs to be further explored.
Accepted in revised form 24 August 1999; received 17 February 1999