Self‐reported sleep duration – what does it mean?

Recent years have seen a dramatic increase in the number of papers that have examined the relationship between self-reported sleep duration and health-related effects. The annual number of papers bearing the words ‘sleep duration’ as part of the title increased from approximately three per year in the 1990s to more than 50 in 2008. These investigated the relationship between self-reported sleep duration and a large number of health-related conditions such as diabetes, cardiovascular morbidity, hypertension, metabolic syndrome and obesity, cancer, psychiatric disorders, neurological disorders and mortality (see a recent meta-analysis on the relationship between mortality and self-reported sleep duration and an accompanying editorial in Gallicchio and Kalesan, 2009; Grandner and Patel, 2009). Generally, both short sleep (<6 h) or long sleep (>8 h) were associated with detrimental health-related effects. Of note, most of these studies were not designed to investigate sleep-related effects. In many, self-reported sleep duration relied upon a single question and no information was provided on daytime sleep versus night-time sleep, nor on possible differences between weekdays and weekend sleep duration. These limitations raised the question of what is the true meaning of self-reported short or long sleep. Do they reflect actual physiological sleep amounts, or are they perhaps subjective indicators of functional states? The present issue of JSR includes two papers that investigated the relationship between self-reported sleep duration and cognitive functioning in large populations in Finland and Spain. One of them (Kronholm et al., 2009) provides interesting information on the meaning of subjective long and short sleep duration. They investigated the relationship between self-reported sleep factors such as sleep duration, insomnia, use of sleeping medications, probable sleep apnoea and fatigue and tiredness, with cognitive functioning in 5177 people older than 30 years from a cross-sectional representative sample of the adult population in Finland. Cognitive functioning was assessed objectively by verbal fluency, and encoding and retaining verbal material. In this study, short and long sleep duration and fatigue and tiredness were independent statistical predictors of low objectively measured cognitive functioning, even after controlling for health and socioeconomic factors. Sleep factors accounted for only a small proportion of the variation in the objective cognitive tests but for a much larger proportion of the variation in subjective cognitive function. Short sleep was associated with insomnia and depression, as well as with cardiovascular morbidities, while long sleep was associated with the use of neuroleptics, cerebrovascular and circulatory diseases and diabetes. Notably, a separate analysis of a subgroup of 2201 participants who did not report any illness and who also reported good, or relatively good, subjective health revealed that after adjustment for sociodemographic variables, the association between sleep factors and cognitive functioning was not found. Based on these results, Kronholm et al. concluded that ‘self-reported sleep duration can be interpreted partially as reflecting more strongly subjective symptom formation rather than physiological sleep amount per se’. Thus, both self-reported long and short sleep may be non-causal indicators of worsened functional state. In the second paper, Faubel et al. (2009) investigated the association between self-reported sleep duration and cognitive function among 3212 people representative of the non-institutionalized population aged >60 years in Spain. Cognitive function was measured by a version of the mini-Mental State Examination administered in a face-to-face interview. They reported that the frequency of cognitive impairment increased progressively across sleep duration categories from 7 to >11 h, while there was no association between short sleep and cognitive impairment. Adjustment for confounding variables attenuated the odds ratios of cognitive impairment by sleep duration to non-significant levels, but the linear trend remained significant, as was the difference in cognitive function between participants sleeping for 7 and >11 h. This difference was similar to that observed for a 10-year increase in age. Unlike Kronholm et al., Faubel et al. did not characterize their very long sleepers. However, they raised the possibility that cognitive function may alter individuals’ perception of sleep duration or that daytime sleepiness, which was not evaluated in their study, could mediate this association. They admitted that there is no known explanation of how sleep duration by itself may produce cognitive impairment. Kronholm et al. should be commended for providing information about the characteristics of self-reported long and short sleepers. In view of their results and the still-existing uncertainty regarding the meaning of self-reported short and long sleep, research strategies should progress beyond the ‘single question’ types of study in the form of ‘how many hours do you usually sleep?’ to strategies using objective documentation of habitual sleep duration, or studies aiming at understanding more clearly the true meaning of self-reported short and long sleep.