OUP user menu

Investigation of the Effects of Alcohol on Sleep Using Actigraphy

Pierce Geoghegan, Mairead T. O'Donovan, Brian A. Lawlor
DOI: http://dx.doi.org/10.1093/alcalc/ags054 538-544 First published online: 17 May 2012

Abstract

Aims: To investigate the effect of alcohol consumption on the sleep and mood of healthy individuals in a college-based, mixed gender population. Methods: Forty-seven individuals participated in this study, of whom 33 consumed alcohol and were included in the analysis. Sleep quality was objectively recorded using actigraphy. Subjects completed a daily sleep diary and bipolar Profile of Mood States Questionnaire, recording the subjective perception of sleep quality and waking mood respectively. Results: Mean self-reported alcohol consumption among the drinkers was 84.6 ml ethanol/night. Mean total sleep time for those who consumed less than the mean reported intake was significantly reduced on alcohol. This reduction in sleep time was associated with increased wakefulness in the second half of the night, a truncated sleeping period and increased waking fatigue. This rebound wakefulness could not be demonstrated in those who consumed higher than the mean intake, though these individuals also reported increased waking fatigue. Conclusion: These results add weight to the clinical evidence that ethanol should not be used as a hypnotic due to its potential to affect both the quantity and quality of sleep. The finding that total sleep time is reduced on low doses of alcohol is novel and may arise from measuring sleep in an environment other than the sleep laboratory.

INTRODUCTION

Alcohol is recognized as probably the most frequently used sleeping aid in the general population (Johnson et al., 1998). Alcohol can have effects on sleep initiation via a sedative effect as blood alcohol concentration declines (Papineau et al., 1998), and this is likely the source of much of ethanol's popularity as an over-the-counter hypnotic.

As well as having effects on sleep initiation/induction, alcohol consumption is known to significantly affect the quality of sleep in normal young individuals, reducing wakefulness and suppressing rapid eye movement (REM) in the first half of the sleeping period with a corresponding increase in slow wave sleep (Yules et al., 1966; Prinz et al., 1980; Stone, 1980; Williams et al., 1983). These alcohol-induced changes have generally been interpreted as an overall increase in sleep quality for the first half of the night on alcohol.

However, these studies have also found evidence of disturbed sleep in the second half of the sleeping period on alcohol. There is a rebound increase in REM sleep, a reduction in slow wave sleep to below control levels and also a tendency for time spent awake and in light stage I sleep to be increased (Yules et al., 1966; Prinz et al., 1980; Stone, 1980; Williams et al., 1983). Given the peak blood alcohol concentrations achieved in these studies and given typical rates of ethanol metabolism, these observations are consistent with a metabolic rebound effect (Roehrs and Roth, 2001).

The observation that wakefulness and light stage I sleep are increased in the second half of the night on alcohol is interesting as it suggests that the mean arousal threshold is lower in the second half of the night on alcohol. It possible to speculate that, although total sleep time is generally reported to be unaltered in the sleep laboratory setting (Prinz et al., 1980), the sleep period may be truncated in the subjects' normal sleeping environment where arousing stimuli are under less control. To our knowledge, there has been no investigation of this point to date.

Actigraphy is an inexpensive and relatively accurate method of monitoring sleep (Sadeh and Acebo, 2002). This technique exploits differences in activity (movement) between sleep and waking states to infer sleep variables from a series of measurements of limb movement. Actigraphy offers the opportunity to objectively monitor the sleep of individuals in their normal sleeping environment at low cost and hence with relatively high power when compared with many polysomnographic studies. To our knowledge, no studies have yet attempted to use actigraphy to characterize the effects of alcohol consumption on sleep.

Aims

The current study aimed to investigate the changes induced by alcohol consumption in the sleep of healthy young individuals in their normal sleeping environment and the impact of these changes on the next day's measures of mood. The objectives of the current study were as follows:

  1. To analyse the effect of alcohol consumption on objectively and subjectively measured sleep quality in a college-based, mixed-gender population.

  2. To analyse patterns in the dose–response relationship.

  3. To characterize alterations in next-day measures of mood after alcohol.

METHODS

Study population

Forty-seven subjects (24 female, 23 male; mean age, 21.5 ± 1.05; range, 20–25) were recruited from a general college population. Individuals with medical, sleep or psychiatric disorders were excluded from participation in the study, as were individuals who regularly abuse illicit drugs or were identified as at risk for alcohol dependence when screened using the CAGE questionnaire. Ethical approval was received from the local research ethics committee. Subjects were fully informed as to the nature of the research and provided written consent for their involvement.

Alcohol consumption questionnaires

Data on alcohol consumption was collected by self-report. Subjects were provided with a separate alcohol consumption questionnaire for each day of the test period and instructed to fill the questionnaires out daily, immediately before going to bed. Subjects were advised that if they were too intoxicated to fill out the questionnaire before going to bed, the questionnaire should be filled out immediately upon waking the following morning to maximize the accuracy of the report. The questionnaire listed a series of commonly consumed beverages and subjects recorded the quantity of each drink (in volume) that they had consumed on that night. These values were converted to units of alcohol using the UK unit convention where 1 unit = 10 ml of ethanol. The subjects also recorded the duration of the drinking period, whether or not they had eaten before, during or after the drinking period and whether or not they had vomited before going to sleep as a result of their alcohol consumption.

Actigraphic sleep measurement

Actigraphy was used to objectively measure sleep quantity and quality. Participants were provided with wrist actigraphs (actiwatch, Cambridge Neurotechnology Ltd) which they were instructed to wear continuously throughout the study period. Short breaks were allowed where there was a danger of the actigraphs becoming damaged and subjects were provided with an activity diary to record any such removals. Each actigraph contains an accelerometer that monitors the intensity, amount and duration of movement in all directions. Activity levels were scored in 0.5 min epochs. Raw data were downloaded telemetrically and sleep and wakefulness were distinguished using a computerized sleep-wake scoring algorithm. This algorithm has been validated against polysomnography and has been peer reviewed (Kushida et al., 2001). The algorithm calculated the following variables:

  1. Total sleep time: assumed sleep time minus wake time, where assumed sleep time is the difference between sleep start and sleep end as determined by the algorithm.

  2. Sleep efficiency: the percentage of time spent asleep while in bed.

  3. Mean activity: the average value of activity counts per epoch over the assumed sleep period. Counts are a unit directly related to acceleration or g (Chen and Bassett, 2005).

  4. Sleep latency: the difference between bedtime and the algorithm-determined start of the sleep period.

The calculation of these variables involves manual input of the time that the subject turned off the lights (bedtime) and the time that subjects got up (get up time). These values were reported by subjects in the sleep diary.

Sleep diary

Actigraphic sleep measurements were supplemented by the use of a sleep diary. In the sleep diary subjects recorded the time they turned off the lights to go to sleep (bedtime) and the time subjects got up (get up time). They also recorded their subjective estimates of sleep latency (time taken to fall asleep) and number of night time awakenings. Additionally, the sleep diary contained a 4-point Likert scale which subjects used to rate the quality of each night's sleep (subjective sleep rating). Subjects also used the sleep diary to record whether or not they had napped on each day of the study period.

Measurement of waking mood

The bipolar form of the Profile of Mood States questionnaire (BI-POMS: Educational and Industrial Testing Service, San Diego, CA, USA) was used to evaluate the mood upon waking for each individual. This test uses six bipolar subjective mood scales: elated–depressed, composed–anxious, clearheaded–confused, agreeable–hostile, confident–unsure and energetic–tired. The questionnaire itself lists 72 items, all of which are adjectives/phrases describing mood. Subjects are asked to rate each item according to how they feel when they wake up (0: feel much unlike this, 1: feel slightly unlike this, 2: feel slightly like this, 3: feel much like this). The items are scored 0, 1, 2 or 3 according to the answer given. Mood on each scale is scored according to the subject's response to the 12 items (adjectives or phrases) specific to that mood scale, 6 of which represent the positive pole and 6 of which represent the negative pole of that particular bipolar scale. The total mood score (ST) on a given bipolar mood scale is defined as (ST = SP – SN + 18), where SP is the sum of the positive scores and SN is the sum of the negative scores for that scale. Therefore, total mood scores for each of the individual six scales may range from 0 to 36. Completed questionnaires were hand-scored using standard templates and double-checked by an independent researcher to ensure the accuracy of the original scores.

Procedure

Subjects were given an actigraph and instructed to wear it continuously for the duration of the test period which lasted for 7 days (from Monday to Monday). Subjects were also given a booklet containing the sleep diaries, POMS questionnaires, and alcohol consumption diaries. This booklet was divided into 7 sections corresponding to the 7 days of the testing period. For each day, the subjects were instructed to complete the morning section of the sleep diary and the entire POMS in the morning, immediately after getting up. They were instructed to complete the lifestyle questionnaire, evening section of the sleep diary and, if possible (see above), the alcohol consumption diary immediately before going to bed in the evening. Subjects were instructed to use the activity diary to record any removals of the actigraph during the day. Subjects were requested not to use alarm clocks in the morning for the duration of the study period to allow comparison of total sleep time on and off alcohol.

Subjects were allowed to consume alcohol on some but not all of the nights of the study period. Where alcohol was consumed on a weekday, comparison of sleep variables was with another weekday (where no alcohol had been consumed) and likewise where alcohol was consumed on a weekend, comparison of sleep variables was with another weekend night (where no alcohol had been consumed). This was done to try to control for potential confounders such as variation in environmental noise that might occur between the weekend and weekdays. It was emphasized that it was important to strictly adhere to the instructions on alcohol consumption. Alcohol consumption diaries were provided for each day of the test period. Subjects were instructed to fill out the diaries as soon as possible after each drinking period to maximize accuracy of the reports. If subjects consumed alcohol on days on which they were required to remain abstinent, they were instructed that they should record this consumption anyway and would not be penalized (subjects were paid a token amount for participation in the study).

Statistical analysis

Alcohol was consumed by 36 subjects. Three of these individuals were excluded as outliers for all comparisons of nights on and off alcohol. Thus, 33 subjects consumed alcohol and were included in the analysis. Mean alcohol consumption for these 33 individuals was 8.5 UK units per night. Subjects were divided into two groups based on reported quantity of alcohol consumed. Individuals who reported consuming less than the mean intake were assigned to the ‘low-dose’ group (n = 18). Individuals who reported consuming an amount of alcohol greater than the mean intake were assigned to the ‘high-dose’ group (n = 15). Comparisons of nights on and off alcohol were carried out using paired t-tests. Where dichotomous variables were recorded, comparisons were carried out using the McNemar test. Significance was reported for P < 0.05. Mean values for control and alcohol nights are reported ± SEM. Mean differences between control nights and alcohol nights are calculated as reported with corresponding P-values and 95% confidence intervals for (95% CI for difference) indicated. Note that a negative value for a mean difference indicates that the mean value of the variable in question was larger on nights where alcohol was consumed.

RESULTS

Alcohol consumption

Mean alcohol consumption per night for the 33 individuals included in the analysis was calculated as 8.5 ± 0.86 UK units. Individuals in the ‘low-dose’ group (n = 18) consumed an average of 4.9 ± 0.55 UK units of alcohol, while individuals in the ‘high-dose’ group (n = 15) consumed an average of 12.7 ± 0.95 UK units of alcohol. No individuals reported vomiting as a result of their alcohol consumption before going to sleep.

Demographic data

Demographic data for those who consumed alcohol, and for the ‘low-dose’ and ‘high-dose’ groups, are summarized in Table 1. There were no statistically significant differences in several key characteristics, although there was a non-significant trend for individuals in the high-dose group to have reported higher average alcohol consumption over the previous month (P = 0.07). Also, there was a greater proportion of females in the ‘low-dose’ group, though the difference was not statistically significant.

View this table:
300 Multiple Choices

This is a pdf-only article and there is no markup to show you.

    Whole-night actigraphic sleep measurements

    The results of paired t-test comparisons of actigraphic sleep quality on and off alcohol are shown in Table 2. A significant decrease (5 min) in mean sleep latency was observed after alcohol consumption [P = 0.043, 95% CI for difference (0, 10)]. There was also a significant decrease (35 min) in total sleep time with alcohol [P = 0.029, 95% CI for difference (3, 68)]. Time in bed was also significantly reduced so that there were no significant changes in mean sleep efficiency for the whole night when comparing nights on and off alcohol. Likewise, the mean activity score for the whole night did not appear to be altered significantly on alcohol.

    View this table:
    300 Multiple Choices

    This is a pdf-only article and there is no markup to show you.

      Dose–response patterns

      The results of paired t-test comparisons of nights on and off alcohol for the low-dose group are shown in Table 2. There was a significant decrease (47 min) in mean total sleep time on alcohol in the low-dose group [P = 0.04, 95% CI for difference (2, 92)]. There were no significant changes in mean sleep latency or mean sleep efficiency (for the whole night) with alcohol. Likewise, there did not appear to be any significant changes in mean activity scores following alcohol consumption in the low-dose group.

      The results of paired t-test comparisons of nights on and off alcohol for the high-dose group are also shown in Table 2. For the high-dose group, there was a decrease (8 min) in mean sleep latency on alcohol that approached statistical significance [P = 0.051, 95% CI for difference (0, –16)]. Although alcohol appeared to induce a decrease in mean total sleep time (22 min), this decrease was not statistically significant [P = 0.363, 95% CI for difference (−28, 73)]. As with the low-dose group, there were no significant differences in either the mean sleep efficiency or mean activity score on alcohol, though there was a non-significant trend for an increase (2.7%) in mean sleep efficiency [P = 0.158, 95% CI for difference (−6.7%, 1.2%)] in this high-dose group.

      Actigraphic sleep measurements in the first half of the sleeping period

      A summary of the results of paired t-test comparisons of actigraphic sleep quality in the first half of the sleeping period for all subjects on and off alcohol is found in Table 3. There was a non-significant trend for an increase (1.5%) in mean sleep efficiency in the first half of the night following alcohol consumption [P = 0.078, 95% CI for difference (−3.1, 0.2)]. There was also a significant decrease (2.1 counts) in the mean activity score on alcohol [P = 0.021, 95% CI of difference (0.3, 3.9)].

      View this table:
      Table 3.

      Summary of actigraphic measurements of sleep quality in the first half of the sleeping period on and off alcohol for all subjects who consumed alcohol and for the low- and high-dose groups

      All subjects (n = 33)Low-dose group (n= 18)High-dose group (n = 15)
      Control nightOn alcoholControl nightOn alcoholControl nightOn alcohol
      Mean SE (%)91.7 ± 0.7693.1 ± 0.5891.1 ± 1.2593.2 ± 0.7792.3 ± 0.7493.1 ± 0.91
      Mean AS (counts)6.81 ± 0.8404.71 ± 0.463*7.49 ± 1.4044.74 ± 0.6086.00 ± 0.7674.68 ± 0.734
      • SE, sleep efficiency; AS, activity score. n = 33 for all subjects comparisons. n = 18 for the low-dose group. n = 15 for the high-dose group. Mean values are given ± SEM.

      • *The mean value on alcohol differs significantly from the control value at the < 0.05 level.

      Dose–response patterns

      The results of paired t-test comparisons of the first half of the sleeping period on and off alcohol for the low-dose group are shown in Table 3. In the low-dose group, there was a tendency for alcohol consumption to result in an increase (1.8%) in sleep efficiency that was not statistically significant [P = 0.097, 95% CI for difference (−4.55, 0.41)]. There was a tendency for a parallel decrease (2.7 counts) in the mean activity score that approached statistical significance [P = 0.054, 95% CI for difference (0, 5.5)].

      The results of paired t-test comparisons of nights on and off alcohol for the high-dose group are also shown in Table 3. There was a tendency for an increase (0.7%) in sleep efficiency to occur in the high-dose group following alcohol consumption, though this result was not statistically significant [P = 0.508, 95% CI for difference (−3.0%, 1.6%)]. As in the low-dose group, there was a tendency for a parallel decrease (1.3 counts) in the mean activity score, though again this result was not statistically significant ([P = 0.227, 95% CI for difference (−0.9%, 3.6%)].

      Actigraphic sleep measurements in the second half of the sleeping period

      A summary of the results of paired t-test comparisons of actigraphic sleep quality in the second half of the sleeping period on and off alcohol is found in Table 4. There was a non-significant trend for sleep efficiency to be decreased (1.8%) in the second half of the night following consumption of alcohol [P = 0.269, 95% CI for difference (−1.4, 4.97)]. This was accompanied by a parallel but non-significant trend for the mean activity score to be increased (2.85 counts) on alcohol [P = 0.219, 95% CI for difference (−7.488, 1.781)].

      View this table:
      Table 4.

      Summary of actigraphic measurements of sleep quality in the second half of the sleeping period on and off alcohol for all subjects who consumed alcohol and for the low- and high-dose groups

      All subjects (n = 33)Low-dose group (n = 18)High-dose group (n = 15)
      Control nightOn alcoholControl nightOn alcoholControl nightOn alcohol
      Mean SE (%)87.9 ± 1.0586.2 ± 1.4589.8 ± 1.0485.4 ± 2.26*85.7 ± 2.0687.1 ± 1.70
      Mean AS (counts)8.52 ± 1.21311.37 ± 1.9326.35 ± 0.68013.39 ± 3.315*11.11 ± 2.4168.94 ± 1.410
      • SE, sleep efficiency; AS, activity score. n = 33 for all subjects comparisons. n = 18 for low-dose group comparisons. n = 15 for high-dose group comparisons. Mean values are given ± SEM.

      • *The mean value on alcohol differs significantly from the control value at the < 0.05 level.

      Dose–response patterns

      The results of paired t-test comparisons of the second half of the sleeping period on and off alcohol for the low-dose group are shown in Table 4. In the low-dose group, there was a significant decrease (4.5%) in mean sleep efficiency in the second half of the sleeping period on alcohol [P = 0.017, 95% CI for difference (0.9%, 8%)]. A parallel and significant increase (7.0 counts) in the mean activity score during the second half of the sleeping period was also observed in the low-dose group [P = 0.046, 95% CI for difference (−13.9, −0.1)].

      The results of paired t-test comparisons of the second half of the sleeping period on and off alcohol for the high-dose group are shown in Table 4. Unlike the low-dose group, there was a non-significant increase (1.5%) in sleep efficiency in the second half of the night on alcohol in the high-dose group [P = 0.589, 95% CI for difference (−7.1, 4.2)]. Mean activity score was decreased (2.2 counts) on alcohol though again this result was not statistically significant [P = 0.43, 95% CI for difference (−3.6, 7.9)].

      Subjective sleep perception

      No significant changes in the subjective perception of sleep quality were observed when comparing nights on and off alcohol. Mean reported sleep latency was not significantly changed on alcohol and reported number of night-time awakenings was not significantly altered following alcohol consumption. Likewise, there did not appear to be any difference in how subjects rated their quality of sleep on and off alcohol. This pattern was true of both the low- and high-dose groups. In contrast, there was a non-significant trend towards an increased tendency for subjects to report that they had napped on days following a night's sleep with alcohol (P = 0.07).

      Waking mood

      Thirty-two subjects were included for comparison of waking mood on and off alcohol (three general outliers were excluded and a single individual was excluded due to missing data). Waking mood was investigated for each of the six bipolar scales in the POMS. The results of this investigation for the whole group are shown in Table 5. There were no significant changes in self-rated mood on the elated– depressed scale when comparing waking mood on and off alcohol. Likewise, there did not appear to be any significant changes in how subjects rated their waking mood on and off alcohol on either the composed– anxious scale or the agreeable–hostile scale. In contrast, there seemed to be a small but highly significant decrease in self-reported feelings of clear-headedness upon waking post-alcohol. This mean a decrease of 3.2 points for the score on the clearheaded-confused scale [P = 0.007, 95% CI for difference (0.938, 5.435)] was accompanied by a significant mean decrease of 1.97 points [P = 0.20, 95% CI for difference (0.329, 3.608)] for the self-reported waking mood state on the confident–unsure scale. However, the greatest difference between the subjective waking mood states on and off alcohol was seen in the energetic–tired scale where a highly significant decrease of 5.03 points was observed on alcohol [P = 0.008, 95% CI for difference (1.38, 8.67)]. This implies significantly higher levels of fatigue and sleepiness on mornings following alcohol consumption, possibly contributing to parallel decreases in self-reported confidence and clear-headedness. No differences were observed in this pattern when looking at either the low dose or high groups in isolation.

      View this table:
      Table 5.

      Summary of waking mood on and off alcohol for all subjects who consumed alcohol as measured by the bipolar Profile of Mood States Questionnaire

      Control morning (n = 33)Post-alcohol (n = 33)
      Elated–depressed25.1 ± 0.9024.1 ± 0.10
      Composed–anxious27.3 ± 0.9626.2 ± 1.01
      Clearheaded–confused26.1 ± 0.9622.9 ± 1.05**
      Agreeable–hostile26.6 ± 1.0325.5 ± 0.97
      Confident–unsure22.9 ± 0.9520.9 ± 0.92*
      Energetic–tired22.3 ± 1.1717.3 ± 1.54**
      Units of alcohol0 ± 08.35 ± 0.880
      • n = 33. Mean values are given ± SEM.

      • *The mean value observed on alcohol differs significantly from the control value at the P < 0.05 level.

      • **The mean value observed on alcohol differs significantly from the control value at the P < 0.01 level.

      DISCUSSION

      To our knowledge, this is the first study that has looked at the effect of alcohol on sleep using actigraphy. Laboratory studies utilizing polysomnography have generally shown that alcohol tends to improve sleep in the first half of the night and disrupt sleep in the second half of the night on alcohol as a consequence of metabolic rebound effects. This study offered the opportunity to assess these effects actigraphically and, furthermore, to determine whether any additional changes could be observed when the sleep took place in the subjects' normal sleeping environment.

      Mean total sleep time for the group as a whole was significantly reduced on alcohol. This reduction was observed in both the low- and high-dose groups, though the decrease was smaller in the high-dose group and was not statistically significant. The observation that total sleep time appears to be reduced following consumption of alcohol is novel. Sleep laboratory studies have not detected this effect (Yules et al., 1966; Prinz et al., 1980; Stone, 1980; Williams et al., 1983). However, these studies have reported that alcohol induced increased wakefulness and light stage I sleep. This change would lower the mean arousal threshold and in itself might be expected to truncate sleep outside the laboratory, where there are more environmental stimuli, as was observed here.

      On analysing sleep in the second half of the sleeping period on alcohol, it was observed that there was a trend for alcohol to induce a decrease in mean sleep efficiency and an increase in mean activity score for the whole group, implying increased levels of wakefulness and restless activity. This trend also appears in polysomnographic studies and is suggestive of a metabolic rebound effect (Stone, 1980; Williams et al., 1983; Roehrs et al., 1991). We hypothesized that, since increased wakefulness in the second half of the sleep period has been ascribed to a metabolic rebound effect, the effects would be less pronounced in the high-dose group, since higher levels of alcohol would take longer to be metabolized to non-physiologically active levels. In agreement with this hypothesis, mean sleep efficiency decreased and mean activity score increased significantly for the low-dose group in the second half of the sleeping period on alcohol, whereas there was a non-significant trend for mean sleep efficiency to be slightly improved in the high-dose group. This result provides further support for the hypothesis that rebound wakefulness is a metabolic rebound effect and the delay of onset when higher doses are consumed may explain the apparent difference between high- and low-dose groups in terms of effects of alcohol consumption on total sleep time.

      This is not the only possible explanation of the difference between the two groups. An alternative hypothesis would be that the difference represents a tolerance phenomenon. One could postulate that individuals drinking higher doses of alcohol on a regular basis could develop tolerance to the ability of alcohol to disrupt sleep in the second half of the night, without any tolerance to the sedative effect explaining why sleep latency is still reduced in this group without a trend to altered sleep efficiency. Consistent with this hypothesis, the ‘high-dose’ group reported higher average alcohol consumption over the previous month, though the difference did not reach statistical significance. The possibility that this is a tolerance phenomenon is potentially important as physiological tolerance is a risk factor for alcohol dependence and addiction. Alternatively, the difference between the groups could represent a sex difference in the response to alcohol, since the low-dose group contained proportionately more females (though this difference was not statistically significant).

      A potential confounder in analysing variations in total sleep time is sleep displacement. Subjects were observed to go to bed on average 1 h 19 min later on nights on which they consumed alcohol. Sleep displacement may itself disrupt sleep quality via circadian mechanisms (Lavie, 2001). However, this has generally been demonstrated for displacements many orders in magnitude greater than that observed here, e.g., in shift work (Lavie, 2001). In this context, it is worth noting that mean sleep displacement was actually larger in the high-dose group, whereas the significant difference in total sleep time was observed in the low-dose group.

      In parallel with changes in sleep, the waking mood was altered following alcohol consumption. The greatest difference was seen on the energetic-tired scale of the BI-POMS, where a large and highly significant decrease was seen on waking following a night's sleep with alcohol. This implies significantly higher levels of fatigue and sleepiness on mornings following alcohol consumption consistent with the existing literature (Yesavage and Leirer, 1986; Roehrs et al., 1991). In fact, the increase in physiological (objective) sleepiness levels is likely to be even greater than that indicated here as there has been consistent evidence that subjects under-report levels of sleepiness when compared with the actual physiological values obtained on the multiple sleep latency test. Other studies investigating the effects of alcohol-induced sleep disruption have demonstrated that this increase in fatigue persists throughout the day (Yesavage and Leirer, 1986; Roehrs et al., 1991), providing some support for our observation that there was a non-significant trend for increased tendency for subjects to nap following a night's sleep on alcohol. Increased levels of sleepiness are known to be associated with higher risk of social and occupational hazard, particularly in relation to automobile accidents, so that the importance of this effect should not be underestimated (Dinges, 1995). Worryingly, subjects themselves do not seem to regard disturbed sleep as the source of these increased feelings of fatigue following alcohol consumption as subjects did not rate the quality of their sleep any differently upon waking after alcohol consumption. This was true for both the high- and low-dose groups.

      Alcohol reduced actigraphically measured sleep latency. This is consistent with the sedative effect of ethanol reported previously (Yules et al., 1966). The effect size is probably an underestimate as sleep latencies are relatively short on the control night for many individuals, leading to a difficulty in detecting alcohol-induced sedation due to a ‘floor effect’. Additional error may have occurred due to inaccuracies of self-reported bedtime, circadian effects, later sleep times on nights where alcohol was consumed, as well as inaccuracies of actigraphy itself when analysing sleep-wake transitions (Sadeh and Acebo, 2002).

      There were a number of limitations in this study. Firstly, since the study was observational, not all subjects actually consumed alcohol, with the result that the power to detect small effect sizes would have been low. Secondly, though actigraphy offers an objective measure of sleep in the subjects' normal sleeping environment, it is less reliable than polysomnography, which is the accepted gold standard of sleep measurement. Finally, the dose of alcohol administered was self-selected so that differences in dose–response patterns may be due in part to trait differences between individuals that chose to consume high doses of alcohol and those that chose to consume low doses of alcohol. Also, reported and actual alcohol intakes are likely to differ, particularly at high doses where alcohol is known to interfere with recall for self-report. Additionally, the subjects were not blinded, i.e. they knew that they had consumed alcohol. This may have affected their subjective ratings of sleep and mood in particular.

      This study adds further weight to the clinical evidence that alcohol should not be used as a hypnotic, demonstrating the ability of alcohol to affect both the quality and quantity of sleep obtained. The novel findings regarding the effect of ethanol highlight the possibility that actigraphy may enable the investigation of phenomena which arise in the subjects’ normal sleeping environment, but not in the sleep laboratory. Future studies may wish to confirm this point by attempting to replicate the association between rebound wakefulness and premature truncation of the sleeping period with additional controls which are absent in this study (such as random assignment to dosing groups and equivalence of bedtime on and off alcohol).

      Funding

      This work was supported financially by Trinity College Dublin.

      References

      View Abstract