was stronger between sleep factors and subjective cognitive function than with objective .... also given questionnaire 1, which they were asked to complete.
J. Sleep Res. (2009) 18, 436–446
Sleep duration and cognitive function
doi: 10.1111/j.1365-2869.2009.00765.x
Self-reported sleep duration and cognitive functioning in the general population ERKKI KRONHOLM1, MIKAEL SALLINEN2,3, TIMO SUUTAMA4, R A I M O S U L K A V A 5 , P E R T T I E R A 6 and T I M O P A R T O N E N 7 1
Department of Chronic Disease Prevention, National Institute for Health and Welfare, Turku, 2Brain Work Research Center, Finnish Institute of Occupational Health, Helsinki, 3Agora Center and 4Department of Psychology, University of Jyva¨skyla¨, Jyva¨skyla¨, 5Department of Geriatrics, School of Public Health and Clinical Nutrition, University of Kuopio, Kuopio, 6Department of Health Sciences, Finnish Centre for Interdisciplinary Gerontology, University of Jyva¨skyla¨, Jyva¨skyla¨ and 7Department of Mental Health and Substance Abuse Services Unit, National Institute for Health and Welfare, Helsinki, Finland
Accepted in revised form 22 February 2009; received 26 November 2008
SUMMARY
This study investigated the relationship between self-reported sleep factors (sleep duration, insomnia, use of sleeping medicine, probable sleep apnoea and feelings of fatigue and tiredness) with cognitive functioning in 5177 people aged 30 years or older from a cross-sectional representative sample of the adult population in Finland (The Finnish Health 2000 Survey). Previous studies have indicated a U-shaped association between increased health risks and sleep duration; we hypothesized a U-shaped association between sleep duration and cognitive functioning. Objective cognitive functioning was assessed with tasks derived from the Consortium to Establish a Registry for AlzheimerÕs Disease test battery (verbal fluency, encoding and retaining verbal material). Subjective cognitive functioning and sleep-related factors were assessed with questionnaires. Health status was assessed during a health interview. Depressive and alcohol use disorders were assessed with the Composite International Diagnostic Interview. Medication was recorded during the health examination. Short and long sleep duration, tiredness and fatigue were found to be associated with both objectively assessed and self-reported decreased cognitive functioning. The association was stronger between sleep factors and subjective cognitive function than with objective cognitive tests. These data suggest that self-reported habitual short and long sleep duration reflect both realization of homeostatic sleep need and symptom formation in the context of the individualÕs health status. keywords
cognitive functioning, general population, sleep duration
INTRODUCTION A number of epidemiological studies have revealed an association of self-reported sleep duration with morbidity and mortality (Ayas et al., 2003; Hublin et al., 2007; Yaggi et al., 2006) [for reviews, see Grandner and Drummond (2007) and Youngstedt and Kripke (2004)]. The commonly reported finding is that both short and long habitual sleep durations Correspondence: Erkki Kronholm, Department of Chronic Disease Prevention, National Institute for Health and Welfare, Turku, Finland. Tel.: +358 2 331 6718; fax: +358 2 331 6720; e-mail: erkki.kronholm@thl.fi
436
are associated with impaired health and increased mortality. Another important but less extensively investigated question is whether habitual sleep duration is associated with variations in cognitive functioning. The impact of experimentally manipulated sleep duration on cognitive functions has been examined repeatedly. Numerous studies have demonstrated that both acute total and cumulative partial sleep loss lead to deteriorations in a wide range of cognitive functions such as sustained attention, executive functions and memory functions [for reviews, see Banks and Dinges (2007) and Jones and Harrison (2001)]. These studies suggest that short sleepers in the general population may show 2009 European Sleep Research Society
Sleep duration and cognitive functioning inferior cognitive performance when compared with 7–8 h sleepers. Experimental studies on sleep extension have shown primarily that only marginal improvements in daytime sleepiness and psychomotor performance can be seen following extended sleep: in some studies, detrimental effects have been reported [for a review, see Ferrara and De Gennaro (2001)]. However, substantial improvements have also been reported (Kamdar et al., 2004). In light of data from sleep restriction and extension experiments, long sleepers in the general population could be assumed to perform similarly to or even better than mid-range sleepers. However, because self-reported sleep duration is not an accurate measure of physiological sleep amount (Jean-Louis et al., 2000; Kripke, 2004) the results from experimental studies may not be applicable to habitual self-reported sleep duration in the general population. The few existing epidemiological studies have reported contradictory relationships between self-reported habitual sleep duration and cognitive functioning. In a sample of 1269 community-dwelling older people (age ‡60 years), selfreported short sleep (£6 h) and daytime napping (‡1 h) were associated with impaired self-reported cognitive functioning (Ohayon and Vecchierini, 2005). Long sleep (‡9.5 h), on the other hand, showed no relationship to the level of self-reported cognitive functioning. The most straightforward association was found between daytime sleepiness and self-reported cognitive impairment. Another study (Schmutte et al., 2007) found a relationship between self-reported sleep duration and objectively measured cognitive performance in non-demented community-dwelling older adults (n = 375; mean age, 79.6 years). The evidence was strongest for the relationship between long sleep (>9 h) and impaired performance on a verbal short-term memory test. Further evidence for the role of habitually insufficient sleep in cognitive functioning comes from a study among shift workers (n = 3237) (Rouch et al., 2005). In comparison with non-shift workers, male shift workers (but not females) showed slower cognitive processing. In addition, those who had stopped working shifts more than 4 years ago demonstrated no cognitive impairments. These results imply that habitually insufficient sleep due to a disrupted circadian rhythm leads to the deterioration of cognitive functioning, at least in men. For this study, our aim was to elucidate further the relationship between self-reported habitual sleep duration and cognitive functioning. In contrast to the above-mentioned epidemiological studies, we used a representative sample of the general adult population as our target group in order to obtain a general picture of the relationship in most adults. To examine the participantsÕ cognitive functioning, we employed both performance tests and self-assessment. In addition, we explored a variety of pertinent factors that are likely to influence the relationship between sleep duration and cognitive performance, such as insomnia symptoms (Kronholm et al., 2006; Schmutte et al., 2007), sleep-disordered breathing (Beebe and Gozal, 2002; Cohen-Zion et al., 2001, 2004; Engleman and Joffe, 1999; Telakivi et al., 1988), depression (Paradiso et al., 1997), alcohol consumption (Dao-Castellana et al., 1998), 2009 European Sleep Research Society, J. Sleep Res., 18, 436–446
437
drug use (Silva et al., 2003) and certain sociodemographic factors. METHODS This study was based on the Health 2000 Survey, a comprehensive nationwide interview and health examination survey carried out in Finland in 2000–2001 (Aromaa and Koskinen, 2004). The implementation, target population, sampling design, samples and methods of the survey are described in detail at http://www.ktl.fi/health2000 [see also Kronholm et al. (2006)]. In brief, from a sample comprised of 8028 people aged 30 years or older, 6986 people (87%) were met by the interviewers at the participantsÕ home or in an institution. During the interviews, the respondents were given an information leaflet and an informed consent form to sign and return. During the health interview, the participants were also given questionnaire 1, which they were asked to complete and bring along to the subsequent comprehensive health examination, which took place a few weeks afterwards in local health centres. If the invited participants did not attend, an abridged examination was conducted at home. The questionnaire was returned by 6460 people (80% of the original sample). During the health examination, the participants were given questionnaire 3 to be returned by mail. Questionnaire 3 was completed by 6269 individuals (78% of the sample); for all questionnaires, see http://www.ktl.fi/health2000. For the analyses of this study, 21 participants who were suspected to have dementia based on abbreviated Mini-Mental Sate Examination (not described here) performed at the health examination, and six participants who used medicine for AlzheimerÕs disease, were excluded. Thus, the final analysis sample (participants with complete information across cognitive performance variables) comprised 5171 participants (2333 men and 2838 women). In different statistical models missing information across variables caused slight variation in the number of participants. The study was approved by the Ethics Committee for Epidemiology and Public Health in the Hospital District of Helsinki and Uusimaa in Finland. MEASURES Cognitive functions Objective cognitive tests Verbal fluency (speech production) and encoding and retaining verbal material were assessed by means of selected tasks deriving from the Consortium to Establish a Registry for AlzheimerÕs Disease (CERAD) test battery which was developed originally for the purpose of assessing early phases of dementia and memory disturbances (Aromaa and Koskinen, 2004; Morris et al., 1989). The selected CERAD tests were carried out as part of the health examination of all attendees aged 30 years or over. To assess verbal fluency, participants were requested to list as many animals as possible within 1 min. Verbal fluency has been shown to be prefrontal
438
E. Kronholm et al.
cortex-focused and sensitive to sleep deprivation (Harrison and Horne, 1997, 1998; Herrmann et al., 2003). The outcome was reported in terms of words per minute (Jones and Harrison, 2001). Memory function was examined by means of three trials testing the learning of a list of 10 words by reading the list aloud three times. After each trial, the participant was asked to repeat the words he or she remembered. The outcome variable was the sum of scores across all three trials. (If a participant learned all 10 words successfully in the first or second trial, the following trial(s) was ⁄ were scored 10 points without an actual attempt). The outcome variable is herein referred to as the Ôlearning effectÕ. Delayed recall was assessed by asking the participants to repeat the list after 5 min. The recall rate (the percentage of remembered words from the number of originally learnt words) was used as an outcome variable in this test (Aromaa and Koskinen, 2004).
are the most commonly used hypnotics in the treatment of insomnia in Finland (National Agency for Medicines and Social Insurance Institution, 2006). It has been reported that benzodiazepine-based hypnotic drugs affect the consolidation of memory during sleep (Silva et al., 2003). Therefore, use of hypnotics [Anatomical Therapeutic Chemical (ATC) classification: code N05C] was included in the sleep-related factors applied in the models predicting cognitive function. Probable sleep apnoea
During the home interviews self-reported memory function was assessed by means of the following question: ÔHow would you estimate your present memory? Is your memory: very good; good; satisfactory; poor; very poor?Õ (for subsequent analyses, the variables itemized were very good ⁄ good = ÔgoodÕ; satisfactory = ÔsatisfactoryÕ; and poor ⁄ very poor = ÔpoorÕ).
The following questions (all from questionnaire 3) and logic were used in order to determine the occurrence of probable sleep apnoea. If a participant answered ÔnoÕ to the question: ÔDo you snore when sleeping? (ask others, if you are not sure)Õ, sleep apnoea was considered to be unlikely. If the answer was ÔyesÕ, then the following additional questions were asked: (i) ÔHow often do you snore?Õ; (ii) ÔHow does your snoring sound?Õ (ask others, if needed); and (iii) ÔHave you noticed (or have others noticed) respiratory pauses when you sleep?Õ. Sleep apnoea was considered quite probable, if snoring was frequent (at least during 3–5 nights weekly) and either or both of the following items was positive: (i) snoring is loud and irregular with occasional respiratory pauses and ⁄ or stertorous breathing and (ii) respiratory pauses with a frequency of at least 1–2 nights weekly. In all other cases, sleep apnoea was considered unlikely.
Sleep-related measures
Health indicators
Sleep duration groups
During the health interview the participants were asked to report all illnesses diagnosed by a doctor. The reported diseases and symptoms were classified into 91 groups. From these, 21 most prevalent (range 11.3–0.9%) groups were considered as possibly modulating cognitive performance. A multivariable model was performed with all 21 disease groups as explanatory variables and word fluency as the dependent variable. The model revealed that 11 illness variables were independent statistical predictors of cognitive performance. These 11 variables were then selected to be used as health indicators possibly modulating cognitive performance. The considered illnesses and their prevalence in the study sample were hypertension (11.3%), musculoskeletal diseases (9.1%), arthrosis (7.1%), disease of the circulatory system (4.5%), coronary artery disease (3.8%), diabetes (3.6%), allergic reaction (2.5%), symptoms related to vascular diseases (1.9%), neurological diseases (1.8%), cerebrovascular disease (1.2%) and chronic respiratory system inflammation (0.9%). In addition, during the comprehensive health examination, the Composite International Diagnostic Interview (CIDI) to identify mental disorders was performed. The computerized version of the CIDI uses operationalized criteria for DSM-IV diagnoses and allows the estimation of DSM-IV diagnoses for major mental disorders [for details, see Pirkola et al. (2005)]. The 12-month prevalence rates of major depressive episodes and disorder as well as alcohol dependence were used in this
Self-assessed cognitive functioning
Self-reported 24-h sleep duration was used for the classification of the participants into four sleep duration groups. The participants were asked (questionnaire 3): ÔHow many hours do you sleep in 24 h?Õ (no time-frame was referenced for this question). The responses were recorded in whole numbers. For the purposes of this study, the participants were divided into four groups according to their sleep duration. The classification was based on previous reports of the associations between self-reported sleep duration and mortality and morbidity [for references, see Bliwise and Young (2007); Grandner and Drummond (2007) and Youngstedt and Kripke (2004). The groups were labelled as Ôshort sleepersÕ (6 h or less), Ô7-h sleepersÕ (7 h), Ô8-h sleepersÕ (8 h) and Ôlong sleepersÕ (9 h or more).
Insomnia and its daytime consequences To assess insomnia and its consequences, the following questions were asked: ÔHave you had any of the following usual symptoms and troubles within the last month (30 days): (i) sleeping disorders or insomnia and (ii) fatigue and tiredness? (questionnaire 1). Exceptional tiredness was assessed by the question: ÔAre you usually more tired during the daytime than other people of your age?Õ (questionnaire 3). Benzodiazepines
2009 European Sleep Research Society, J. Sleep Res., 18, 436–446
439
Sleep duration and cognitive functioning
Table 1 Sleep-related factors as explanatory variables (in general linear models) of objective cognitive tests in the adult general population Models explaining performance in word fluency test Model I
Model II
n = 5171; r2 = 0.0292
df = 7;
Model III
n = 5171; r2 = 0.1736
df = 11;
n = 5171; df = 24; r2 = 0.1813
Sleep factor
F
P
F
P
F
P
Fatigue and tiredness Exceptional tiredness Sleep duration
26.0 22.0 19.7
