SEARCH

SEARCH BY CITATION

Keywords:

  • cognitive function;
  • elderly people;
  • sleep;
  • spain

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Participants and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The few studies that have examined the association between usual sleep duration and cognitive function have shown conflicting results. This cross-sectional study examined the association between sleep duration and cognitive function among 3212 people, representative of the non-institutionalized population aged 60 years and over in Spain. Sleep duration was self-reported, and cognitive function was measured with the Mini-Examen Cognoscitivo (MEC), a version of the Mini-Mental State Examination that has been validated in Spain. Linear regression, with adjustment for the main confounders, was used to obtain mean differences in the MEC between the categories of sleep duration (≤5, 6, 7, 8, 9, 10, ≥11 h day−1). The MEC score decreased progressively (became worse) across sleep categories from 7 to ≥11 h (P for linear trend <0.001). People who slept for ≥11 h had a significantly lower MEC score than those who slept for 7 h (mean difference −1.48; 95% confidence interval −2.12 to −0.85). This difference in the MEC was similar to that observed for a 10-year increase in age. The results did not vary significantly by sex (P for interaction >0.05). No association was observed between short sleep duration (<7 h) and cognitive function. We conclude that long sleep duration is associated with poorer cognitive function in older adults from the general population.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Participants and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Cognitive impairment in older people has been associated with a higher risk of dementia (Gauthier et al., 2006; Reisberg and Gauthier, 2008), disability (Blaum et al., 2002; Cigolle et al., 2007; Lee et al., 2005), nursing home admission (Aguero-Torres et al., 2001) and mortality (Bassuk et al., 2000; Frisoni et al., 1999). Thus, it is important to identify the determinants of cognitive impairment in the elderly.

Previous studies have observed that daytime sleepiness (Ohayon and Vecchierini, 2002) and sleep deprivation (Durmer and Dinges, 2005) are associated with poorer results on cognitive tests. However, the few studies that have analysed the association between usual sleep duration and cognitive function in older adults have yielded conflicting results. In a cross-sectional study of people aged 60 years and over in both sexes, Ohayon and Vecchierini reported that, in comparison with those who slept 7–8.5 h, those who slept ≤5 h more often had attention–concentration deficit, and those who slept 5–7 h more often had difficulties in orientation (Ohayon and Vecchierini, 2002). Moreover, in women aged 70–81 years, Tworoger et al. (2006) observed that sleeping ≤5 h was associated with poorer results on various cognitive tests; however, the association was no longer observed after 2 years prospective follow-up. On the other hand, Blackwell et al. (2006) did not find a cross-sectional association between sleep duration and cognitive impairment in older women. Finally, in a cross-sectional study in people aged 75–85 years in both sexes, Schmutte et al. (2007) observed that those who reported longer sleep duration performed significantly worse on a measure of verbal short-term memory. The specific reasons for the discrepant results between studies should still be elucidated, but the different approaches used to measure sleep duration (e.g. sleep within a 24-h period, night-time only, daytime sleep separately) might have played an important role. This cross-sectional study examined the association between usual sleep duration and cognitive function in older adults in Spain.

Participants and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Participants and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study design and participants

The study methods have been reported previously (Graciani et al., 2006; Lopez-Garcia et al., 2008). Briefly, in 2001, information was obtained on 4008 people representative of the non-institutionalized Spanish population aged 60 years and over. Subjects were selected using probabilistic sampling by multi-stage clusters. The clusters were stratified by region of residence and size of municipality. Census sections were then chosen randomly within each cluster, and the households in which information was finally obtained from the subjects were chosen within each section. Information was collected from a total of 420 census sections in Spain, and subjects were selected by age and sex strata. Subjects were replaced for interviews only after 10 failed visits by the interviewer, disability, death, institutionalization or refusal to participate. The study response rate was 71%. Reasons for non-response were ‘impossible to locate after several attempts’ (17%), ‘refused to be interviewed’ (6%) and remaining motives (6%).

Information was collected in the households by personal interview and physical examination, conducted by trained and certified staff. Subjects and an accompanying family member gave informed consent to participate in the investigation. The study was approved by the Clinical Research Ethics Committee of the La Paz University Hospital in Madrid.

Study variables

The dependent variable was cognitive function measured with the Mini-Examen Cognoscitivo (MEC), a version of the Mini-Mental State Examination (MMSE) (Folstein et al., 1975) that has been adapted and validated for use in elderly people in Spain (Lobo et al., 1999). The MEC is scored from 0 to 30 points; a higher score indicates better cognitive performance.

The main independent variable was usual sleep duration, obtained with the question: ‘How many hours do you usually sleep per day (including sleep at night and during the day)?’ Study participants reported the number of hours and minutes of sleep, which were then rounded to the nearest integer hour by the interviewer. The available information did not allow to distinguish between sleep duration in the night and during daytime (napping or siesta).

A structured questionnaire was used to obtain information on other variables of interest. Study participants were asked if a physician had ever diagnosed them with depression that required treatment. Lawton and Brody’s test was also used to assess limitations in the instrumental activities of daily living (IADL) (Lawton and Brody, 1969). ‘Probable dementia’ was defined according to those IADL that explained differences in MEC scores significantly in each sex after adjusting for age and educational level (Barberger-Gateau et al., 1992). Specifically, in women, ‘probable dementia’ was defined as the simultaneous presence of limitation for managing money, taking medication and shopping. In men, it was defined as limitation in managing money, taking medication and using public and private transportation (Graciani et al., 2006).

Information was also collected on variables that might confound the study relation, because in previous studies they have been associated with sleep duration, cognitive function or both. Specifically, we collected sex, age (in years) and leisure-time physical activity (sedentary, occasional activity, regular activity). Questions were also asked about tobacco use (never smoker, ex-smoker, current smoker) and consumption of alcoholic beverages (never drinker, ex-drinker, moderate drinker, heavy drinker). Heavy drinking was considered to be ingestion of >20 g day−1 of alcohol in women and >30 g day−1 in men, and moderate drinking was equal to or less than those amounts. Ex-drinkers were those reporting that they used to drink but do not drink now. Data were also obtained on coffee consumption (none, <1, 1–2, >2 cups day−1), educational level (no schooling, primary, secondary, university) and health-related quality of life, assessed by the SF-36 questionnaire. For this analysis, we specifically used the physical summary and the mental summary of the SF-36; where the higher the score the better the physical or the mental health (Lopez-Garcia et al., 2003). Subjects were also asked if they awoke while sleeping at night-time and if they took any medication for anxiety (including hypnotics) or depression. Data were collected similarly on social network, assessed as the number of participants’ social ties (marital status, cohabitation, frequent contact with friends and frequent contact with family) (Garcia et al., 2005), head of family’s work status (non-manual workers, land-owners and farm workers, skilled manual workers, unskilled manual workers), number of medical drugs consumed (0, 1–2, 3+) and on the number of the following chronic diseases diagnosed by a physician and reported by the subject: ischaemic heart disease, stroke, osteoarthritis, diabetes mellitus, chronic obstructive lung disease, Parkinson’s disease and cancer at any site.

Other confounders were obtained from physical examinations. In particular, weight and height were measured with standardized procedures (Gutierrez-Fisac et al., 2004); body mass index (BMI) was calculated as weight in kg divided by height in metres squared, and normal weight was defined as BMI 18.5–24.9 kg m−2, overweight as BMI 25–29.9 kg m−2, and obesity as BMI ≥30 kg m−2. Finally, blood pressure was measured six times in each subject by trained personnel using standardized methods, with calibrated mercury sphygmomanometers. Hypertension (HT) was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg or current treatment for high blood pressure (Banegas et al., 2002). HT was considered to be treated when subjects with known HT reported being in antihypertensive treatment.

Statistical analysis

Of the 4008 participants, 492 were excluded because they had been diagnosed with depression requiring treatment, 123 because of suffering dementia, 66 for reporting extreme values of sleep duration (≤3 or ≥16 h) and 27 for lack of data on confounding factors. In addition, 88 were excluded for not answering the complete MEC questionnaire. Thus, analyses were conducted with 3212 individuals (1521 men and 1691 women).

The association between sleep duration and cognitive impairment was summarized with odds ratios (OR) and their 95% confidence intervals (CI) obtained from logistic regression. Given the influence of age and educational level on cognitive function, and the high percentage of elderly Spaniards with low educational level, the recommended definition for cognitive impairment is a score of ≤22 on the MEC (sensitivity 89.8%, specificity 80.8%) (Lobo et al., 1999). Sleep duration was classified into the following categories: 4–5, 6, 7, 8, 9, 10 and 11–15 h day−1. To facilitate comparison with previous studies (Blackwell et al., 2006; Tworoger et al., 2006), we used 7 h sleep as the reference category. The analyses were adjusted for potential confounders using two models. The first model was adjusted only for sex and age. The second model was adjusted additionally for the remaining confounders. Both sleep duration and the confounders were modelled with dummy variables, with the exception of SF-36 summary scores and the number of social ties, which were entered as continuous variables. The possible dose–response relation was tested separately for short (<7 h) and long (>7 h) sleep duration, with the P for linear trend obtained by modelling sleep duration as a continuous variable.

The association between sleep duration and cognitive function was estimated by beta coefficients and their 95% CIs obtained from linear regression. The beta coefficients estimate the mean differences in MEC across the sleep duration categories. The rest of the analytical strategy was the same as that described above for the logistic models.

As the study association could vary by sex, analyses were also performed separately in men and women. In the logistic models the interaction between sleep duration and sex was tested with likelihood ratio tests, which compared the model with the five interaction terms (the products of each sleep category by sex) with the model without these terms. In the linear regression models a test of variance was used to compare the models with and without interaction terms.

The statistical tests were two-tailed and statistical significance was set at < 0.05. The analyses were performed with sas version 9.1 for Windows (SAS Institute, Inc, 2004).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Participants and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The mean ± standard deviation (SD) age of participants was 71.6 ± 7.8 years (70.8 ± 8.0 in men and 72.3 ± 7.6 in women), and mean usual sleep duration was 8.0 ± 1.9 h (8.2 ± 2.0 in men and 7.9 ± 1.8 in women) (Fig. 1). A total of 639 (21.6%) individuals had cognitive impairment, 217 (34.0%) men and 422 (66.0%) women. Mean age among individuals with cognitive impairment was 75.9 ± 8.2 years (74.2 ± 8.4 in men and 76.3 ± 8.0 in women), while mean age among those without cognitive impairment was 70.6 ± 7.3 years (70.1 ± 7.8 in men and 70.7 ± 7.1 in women). Mean MEC score in the total study participants was 25.5 ± 4.5 (26.3 ± 4.3 in men and 24.9 ± 4.5 in women).

image

Figure 1.  Distribution of study sample by sleep duration, in each sex (1521 men and 1691 women).

Download figure to PowerPoint

Table 1 shows the characteristics of the study participants. Those with cognitive impairment were older and had a higher percentage of women, sedentary individuals, never-smokers, alcohol abstainers and people who did not drink coffee. They also had a poorer physical and mental score on the SF-36, and a higher number of diagnosed chronic diseases and of medical drugs consumed. Finally, individuals with cognitive impairment had a higher percentage of farm workers and of individuals with no schooling, who awoke during the night and who took anxiolytic medication.

Table 1.   Characteristics of study participants by cognitive impairment
 No cognitive impairment (n = 2573)Cognitive impairment (n = 639) P*
  1. SD: standard deviation.

  2. In men ≤30 g alcohol day−1; in women ≤20 g alcohol day−1.

  3. In men >30 g alcohol day−1; in women >20 g alcohol day−1.

  4. §Hypertension is systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg or current antihypertensive treatment.

  5. Antihypertensive treatment in patients with hypertension.

  6. *Obtained from analysis of variance (anova) for continuous variables and from chi-square test for categorical variables.

Age in years (mean ± SD)70.6 ± 7.375.9 ± 8.2<0.001
Sex (%)
 Men50.734.0<0.001
 Women49.366.0
Leisure-time physical activity (%)
 Sedentary37.255.2<0.001
 Occasional physical activity59.343.1
 Regular physical activity3.51.7
Tobacco use (%)
 Never-smoker61.274.0<0.001
 Ex-smoker27.618.1
 Current smoker11.27.9
Alcohol consumption (%)
 Never drinker45.059.1<0.001
 Ex-drinker10.612.6
 Moderate drinker32.522.7
 Heavy drinker11.95.6
Coffee consumption (%)
 None46.860.1<0.001
 <1 cup day−112.18.7
 1–2 cups 25.520.3
 >2 cups15.610.9
Educational level (%)
 No schooling42.780.4<0.001
 Primary education40.218.0
 Secondary11.81.5
 University5.30.1
SF-36 mental summary score (mean ± SD)51.5 ± 9.448.9 ± 11.2≤0.001
SF-36 Physical summary score (mean ± SD)45.6 ± 10.240.4 ± 11.3≤0.001
Wakes up during the night (%)13.620.7<0.001
Takes anxiolytic medication (%)10.313.60.02
Number of social ties (mean ± SD)2.9 ± 1.02.6 ± 1.1≤0.001
Head of family’s work status
 Non-manual workers14.95.9<0.001
 Land owners and farm workers17.737.1
 Skilled manual workers37.130.0
 Unskilled manual workers30.327.0
Medical drugs consumed
 015.814.00.03
 1–242.638.5
 3+41.647.5
Number of chronic diseases (mean ± SD)1.1 ± 1.01.4 ± 1.1≤0.001
Body mass index (%)
 Normal weight (18.5–24.9 kg m−2)17.120.40.09
 Overweight (25–29.9 kg m−2)48.043.9
 Obesity (≥30 kg m−2)34.935.7
Hypertension (%)§66.167.70.6
Antihypertensive treatment (%)53.457.00.2

Table 2 shows the association between sleep duration and cognitive impairment. In both, the model adjusted for age and sex and the model adjusted additionally for the rest of the confounders, the frequency of cognitive impairment increased progressively across sleep categories from 7 h to ≥11 h (P for linear trend <0.001 for age and sex model and 0.008 for total adjustment model). Furthermore, in the model adjusted for age and sex, the frequency of cognitive impairment was higher in those sleeping 10 h (OR 1.4; 95% CI 1.0–2.0) and ≥11 h (OR 2.2; 95% CI 1.5–3.2) than in those sleeping 7 h. The results did not vary significantly by sex (P for interaction >0.05 in both models). No association was observed between short sleep duration (<7 h) and cognitive impairment, either in the total sample or by sex.

Table 2.   Odds ratios (95% confidence interval) of cognitive impairment [Mini-Examen Cognoscitivo(MEC) score ≤22] by usual sleep duration
 P for linear trend (≤5–7 h)Sleep duration (per 24-h period)P for linear trend (7–≥11 h)
≤5678910≥11
  1. n, participants with cognitive impairment in each sleep duration category; N, participants in each sleep duration category.

  2. Model 1 adjusted for sex (except in stratified analyses) and age (60–69, 70–79, ≥80 years). Model 2 adjusted for sex (except in stratified analyses) and age (60–69, 70–79, ≥80 years), physical activity (sedentary, occasional, regular), tobacco use (never-smoker, ex-smoker, current smoker), alcohol consumption (never drinker, ex-drinker, moderate drinker, heavy drinker), coffee consumption (none, <1, 1–2, >2 cups day−1), educational level (no schooling, primary, secondary, university), SF-36 mental summary score, SF-36 physical summary score, night-time awakening (yes/no), intake of anxiolytic medication (yes/no), number of social ties, head of family’s work status (non-manual workers, land owners and farm workers, skilled manual workers, unskilled manual workers), medical drug consumption (0, 1–2, 3+), number of chronic diseases (0, 1, ≥2), body mass index (normal, overweight, obesity), hypertension (yes/no) and antihypertensive treatment (yes/no).

  3. The MEC is a version of the Mini-Mental State Examination that has been validated in Spain. The MEC is scored from 0 to 30 points, and higher score indicates better cognitive performance.

  4. *< 0.05; **< 0.001.

Total
 n/N (%) 62/292 (21.2)74/380 (19.5)79/494 (16.0)128/848 (15.1)107/542 (19.7)105/416 (25.2)83/238 (34.9) 
 Model 10.41.2 (0.8–1.7)1.1 (0.8–1.6)Ref.0.9 (0.7–1.2)1.1 (0.8–1.6)1.4 (1.0–2.0)*2.2 (1.5–3.2)**<0.001
 Model 21.01.0 (0.7–1.6)1.1 (0.7–1.6)Ref.0.8 (0.6–1.3)0.9 (0.6–1.3)1.1 (0.8–1.6)1.2 (0.8–1.9)0.008
Men
 n/N (%) 16/107 (14.9)27/161 6.8)28/229 (12.2)46/427 (10.8)31/260 (11.9)29/208 (13.9)40/130 (30.8) 
 Model 10.71.1 (0.6–2.2)1.2 (0.7–2.2)Ref.0.8 (0.5–1.3)0.8 (0.5–1.5)1.0 (0.6–1.7)2.5 (1.4–4.4)**<0.001
 Model 20.80.9 (0.4–1.9)1.2 (0.6–2.3)Ref.0.8 (0.5–1.5)0.7 (0.4–1.3)0.8 (0.4–1.5)1.5 (0.8–3.0)0.02
Women
 n/N (%) 47/185 (25.4)47/219 (21.5)51/265 (19.2)82/421 (19.5)75/282 (26.6)76/208 (36.5)43/108 (39.8) 
 Model 10.41.2 (0.8–2.0)1.1 (0.7–1.7)Ref.1.0 (0.7–1.5)1.4 (0.9–2.1)1.7 (1.1–2.7)*1.8 (1.1–3.0)*<0.001
 Model 20.71.3 (0.7–2.1)1.1 (0.7–1.9)Ref.1.0 (0.6–1.5)1.1 (0.7–1.7)1.4 (0.9–2.3)1.2 (0.6–2.1)0.1

Table 3 shows the association between sleep duration and cognitive function. In both regression models, the MEC score decreased progressively (became worse) across sleep duration categories from 7 h to ≥11 h (P for linear trend <0.001 in both models). In the model adjusted for all the confounders, those who slept ≥11 h had a significantly lower MEC score than those who slept 7 h (mean difference in MEC −1.48; 95% CI −2.12 to −0.85). To put these results into context, a column labelled ‘clinical significance’ in Table 3 shows the mean difference in MEC associated with 1 year of ageing. According to these results, the difference in MEC associated with sleeping 11 h or more was similar to that observed for a 10-year increase in age. The results did not vary significantly by sex (P for interaction >0.05 in both models). No association was observed between short sleep duration (<7 h) and cognitive function.

Table 3.   Mean differences in the Mini-Examen Cognoscitivo (MEC) score (95% confidence interval) between categories of sleep duration
 P for linear trend (≤5–7 h)Sleep duration (per 24-h period)P for linear trend (7–≥11 h)Clinical significance
≤5678910≥11
  1. N, participants in each sleep duration category.

  2. Model 1 adjusted for sex (except in stratified analyses) and age (60–69, 70–79, ≥80 years). Model 2 adjusted for sex (except in stratified analyses) and age (60–69, 70–79, ≥80 years), physical activity (sedentary, occasional, regular), tobacco use (never-smoker, ex-smoker, current smoker), alcohol consumption (never drinker, ex-drinker, moderate drinker, heavy drinker), coffee consumption (none, <1, 1–2, >2 cups day−1), educational level (no schooling, primary, secondary, university), SF-36 mental summary score, SF-36 physical summary score, night-time awakening (yes/no), intake of anxiolytic medication (yes/no), number of social ties, head of family’s work status (non-manual workers, land-owners and farm workers, skilled manual workers, unskilled manual workers), medical drug consumption (0, 1–2, 3+), number of chronic diseases (0, 1, ≥2), body mass index (normal, overweight, obesity), hypertension (yes/no) and antihypertensive treatment (yes/no).

  3. The MEC is a version of the Mini-Mental State Examination that has been validated in Spain. The MEC is scored from 0 to 30 points, and higher score indicates better cognitive performance.

  4. To help to interpret the mean differences in the MEC, we provide the mean difference in MEC associated with 1 year of ageing.

  5. *< 0.05; **< 0.01; ***< 0.001.

Total
 N 292380494848542416238  
 Model 10.4−0.22 (−0.83 to 0.38)−0.09 (−0.65 to 0.47)Ref.0.11 (−0.35 to 0.57)−0.66 (−1.17 to −0.15)**−0.65 (−1.20 to −0.10)**−2.46 (−3.12 to −1.81)***<0.001−0.18
 Model 20.70.09 (−0.48 to 0.67)−0.06 (−0.58 to 0.47)Ref.0.17 (−0.27 to 0.60)−0.24 (−0.73 to 0.24)−0.18 (−0.70 to 0.34)−1.48 (−2.12 to −0.85)***<0.001−0.15
Men
 N 107161229427260208130  
 Model 10.5−0.15 (−1.09 to 0.79)−0.61 (−1.44 to 0.22)Ref.0.11 (−0.55 to 0.77)−0.49 (−1.22 to 0.24)−0.23 (−1.01 to 0.54)−2.80 (−3.69 to −1.91)***<0.001−0.14
 Model 20.90.31 (−0.57 to 1.19)−0.48 (−1.25 to 0.29)Ref.−0.02 (−0.63 to 0.59)−0.19 (−0.87 to 0.48)0.04 (−0.68 to 0.76)−1.97 (−2.82 to −1.11)***<0.001−0.11
Women
 N 185219265421282208108  
 Model 10.6−0.24 (−1.04 to 0.55)0.28 (−0.48 to 1.03)Ref.0.09 (−0.55 to 0.74)−0.84 (−1.55 to −0.13)*−1.04 (−1.81 to −0.26)**−2.08 (−3.03 to −1.12)***<0.001−0.20
 Model 20.9−0.21 (−0.97 to 0.56)0.13 (−0.59 to 0.84)Ref.0.10 (−0.52 to 0.72)−0.68 (−1.06 to 0.31)−0.48 (−1.23 to 0.26)−1.23 (−2.16 to −0.30)**<0.001−0.19

Results in Tables 2 and 3 were similar when the analyses were repeated using 8-h sleep as the reference category.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Participants and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

In this population-based study of people aged 60 years and more in Spain, the frequency of cognitive impairment increased and cognitive function declined progressively across sleep categories from 7 h to ≥11 h. The reduction in cognitive function among those who slept ≥11 h was similar to that associated with a 10-year increase in age. In contrast, no association was observed between short sleep duration (<7 h) and cognitive function.

Given the cross-sectional design of the study, a temporal relation between sleep duration and cognitive function cannot be established. Thus, it is possible that cognitive function influences sleep duration. However, a recent study did not find that cognitive decline predicted total sleep duration in older women (Yaffe et al., 2007). This does not rule out the possibility that cognitive function may affect individuals’ perception of sleep duration. In fact, a study of postmenopausal women found that low self-reported quality of sleep, but not objectively measured sleep, was associated with difficulties in concentration and poorer results on cognitive tests (Regestein et al., 2004).

Another possible explanation is that some disorders, such as sleep apnoea, is responsible for both reduced cognitive function (Cohen-Zion et al., 2001; Spira et al., 2008) and altered sleep duration (Malhotra and White, 2002). However, this is unlikely to explain our findings, because the analyses were adjusted for important correlates of sleep apnoea, such as obesity, health-related quality of life (physical and mental summary of the SF-36) and awaking from sleep (Malhotra and White, 2002; Young et al., 2002). Daytime sleepiness has also been shown to be associated with poorer cognitive performance (Ohayon and Vecchierini, 2002) and might result in longer total sleep duration. Therefore, daytime sleepiness might have contributed to the association between sleep duration and cognitive function. Unfortunately, we did not have information on daytime sleepiness, but an attempt to take this variable into account was made by adjusting analyses for health-related quality of life. We also adjusted for chronic diseases, number of medical drugs taken and use of anxyolytics, which can lead to daytime sleepiness.

The last possible explanation is that sleep duration itself produces cognitive impairment. Some mechanisms by which short sleep duration could affect health are known; specifically, sleep restriction causes fatigue and daytime sleepiness, which translates to poorer cognitive performance (Ohayon and Vecchierini, 2002). However, the mechanisms by which long sleep duration could affect health are less well known. Long sleep duration could reflect both a greater physiological need for sleep and a mechanism to compensate for poor-quality sleep; in this regard, there are data supporting the association between disturbed sleep and cognitive impairment (Blackwell et al., 2006). Moreover, a relation between sleep breathing disorders, fairly frequent in elderly people, and cognitive impairment has also been observed. This relation could be mediated by daytime sleepiness (Cohen-Zion et al., 2001).

Our results differ from those found in previous studies. Ohayon and Vecchierini, (2002) conducted a study in elderly people in Paris, and found that those who slept ≤5 h had a significantly poorer score than those who slept 7–8.5 h on two of the six dimensions of the MacNair-R Cognitive Difficulties Scale; however, no association was found when cognitive function was evaluated using the MMSE. Similarly, the study by Tworoger et al. (2006) in older women in the United States found that those who slept ≤5 h had worse scores on various cognitive tests, including the MMSE. In contrast, Blackwell et al. (2006) did not find an association between sleep duration and the MMSE in older women in the United States, whereas Schmutte et al. (2007) found poorer verbal memory scores in American older men and women who slept for more than 9 h.

We do not know the reasons for the discrepancies between studies, but the following factors may have contributed to a certain degree. The first concerns the different characterizations of sleep duration, because in one study the longest sleep category was ≥8.5 h (Ohayon and Vecchierini, 2002), while in others it was ≥9 h (Schmutte et al., 2007; Tworoger et al., 2006), and in ours it was ≥11 h. Moreover, the larger percentage of individuals with long sleep duration in our study compared with others (Ohayon and Vecchierini, 2002; Tworoger et al., 2006) makes it easier to observe associations in this sleep range. Secondly, in two studies the data on cognitive function were collected by telephone (Ohayon and Vecchierini, 2002; Tworoger et al., 2006), whereas in the remaining studies (Blackwell et al., 2006; Schmutte et al., 2007) and our own study, face-to-face interviews were held. Thirdly, there are differences in the measurement of sleep duration across studies. Two studies assessed self-reported duration of night-time sleep (Ohayon and Vecchierini, 2002; Schmutte et al., 2007). Another investigation (Tworoger et al., 2006) and our study measured self-reported sleep duration for a 24-h period; Blackwell et al. (2006) were the only investigators to obtain objective measures of nocturnal sleep with actigraphy. Finally, it is possible that differences in exposure to light, lifestyles (e.g. diet, physical activity) and work and leisure-time hours between a Mediterranean country such as Spain, and Paris and the United States have contributed to the different results, as these variables may be associated both with sleep duration and cognitive function.

Some methodological comments are needed to interpret our results correctly. First, the higher rate of cognitive impairment in women than in men could be due to the higher age of women (mean age 72.3 years versus 70.8 years in men) and to their lower educational level. In fact, while 56.1% of women lacked schooling, the corresponding percentage for men was 43.3%. Secondly, sleep duration was self-reported; none the less, although the results differ with respect to actigraphy, the two measures have good correlation (Lockley et al., 1999). Furthermore, although mean sleep duration in our study (8 h) was longer than reported in older people in France (6–8 h) (Ohayon and Vecchierini, 2005) or the United States (7 h) (National Sleep Foundation), it was similar to that obtained in another representative sample of the Spanish population (National Statistics Institute), which supports the validity of our sleep estimate. Thirdly, although our analyses were adjusted for nocturnal awakenings and anxiolytics consumption, we did not have information on sleep disorders or sleep quality, which constitutes a limitation of the study; nor did we have data on the different phases of sleep. This might have been considered in the analyses, because the latency and density of the rapid eye movement (REM) phase have been associated with cognitive deficits in elderly people, which may be mediated through a cholinergic dysfunction (Spiegel et al., 1999). Fourthly, our information on sleep duration includes both night-time and daytime sleep. This makes our results comparable with those of other studies in the field (Tworoger et al., 2006) and permits better characterization of sleep in elderly people because they often take naps during the day (Cooke and Ancoli-Israel, 2006; Wolkove et al., 2007). Unfortunately, the question on sleep duration did not distinguish between sleep during the night and during daytime. As a result, individual participants might have interpreted the question quite differently, leading to a certain heterogeneity in the response; it also made it impossible to assess a differential association between each sleep time and cognitive function. Fifthly, as depression and cognitive decline may have a common pathogenesis (Comijs et al., 2004), and it is not easy to assess cognitive function in patients with depression, individuals with a diagnosis of depression or pharmacological treatment for depression were excluded from the analysis, as other authors have done (Blackwell et al., 2006). This improves the internal validity of our results, but limits extrapolation to people with depression, which is frequent in elderly people. Finally, we adjusted our analyses for a considerable number of covariables; moreover, little variation in the results was seen between the age- and sex-adjusted analyses and those adjusted additionally for other confounders. For all these reasons, the likelihood of residual confounding is small.

In conclusion, our results suggest that there is an association between sleeping ≥11 h and cognitive impairment. This association is clinically relevant, because its magnitude is similar to the cognitive decline associated with ageing 10 years. Future studies should use prospective designs and objective measures of sleep duration in addition to self-reports. They should also distinguish between night-time and daytime sleep, and take into account the influence of sleep breathing disorders and daytime sleepiness on the study relationship. Lastly, pathophysiological mechanisms that explain the relation between sleep duration and cognitive function should also be investigated.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Participants and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This study was funded by FIS grant 06/0366. Raquel Faubel received a fellowship from the Madrid City Council in the ‘Residencia de Estudiantes’. Esther López-García had a ‘Ramón y Cajal’ contract from the Ministry of Education. The funding bodies had no role in data extraction and analysis, writing of the manuscript or in the decision to submit the article for publication.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Participants and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References