Sleep pirates–are we really living through a sleep deprivation epidemic and what’s stealing our sleep?

There is a joke that is used to warn students about the dangers of ecological associations involving the (historical) ‘causal link’ between the declining numbers of pirates and increasing global warming. A key problem with this joke is that there are very few pirates left in the world: an assumption that is seldom questioned. Likewise in the field of sleep epidemiology, one of our key until recently untested assumptions is that a worldwide secular decline in sleep durations exists (the sleep deprivation epidemic). The problem is now that whilst the assumption is widely held numerous studies around the world using detailed time-use survey data have failed to confirm that adults in developed nations are sleeping less than they were a generation ago.1 Indeed the very recent good news from the USA is that average sleep durations have continued to subtlety rise through the last decade.2 Updated analyses of the time use survey data there indicate that between 2003 and 2016 average workday sleep durations have increased by about 1.4 min per year and weekend sleep by 0.8 min per year. Encouragingly for public health the authors suggest that people have been successfully convinced to go to sleep earlier. How is it that this assumption continues in the face of the evidence? So far the majority of studies that have found declines in sleep duration have done so using only simple single questions (i.e. ‘how much do you normally sleep within a 24-h period’) and not the more detailed time-use survey data. A problem with these simple questions is that humans seem fundamentally incapable of answering them reliably.3 These adults have also been exposed to a couple of decades of misrepresented sleep epidemiology reported in the media leading them to think that they are supposed to be sleep deprived. Would it be that surprising if they subsequently gently nudge down their self-estimates of sleep duration, giving the socially acceptable answer that they think health researchers want to hear, as is done for other health behaviours, such as healthy eating, for example? This could be a classic Hawthorne effect adversely operating across the population. In adolescents something different might be happening. We now have very recent data from two large scale repeated cross-sectional studies in the USA indicating that in late high school sleep durations indeed appear to have declined slightly.4 When insufficient sleep is defined as less than 7 h per night the proportions afflicted have risen from around 35–37% up to 41–43% between 2009 and 2015 having been relatively stable during the previous decade. The caveat here is that these studies have relied on simple self-reports of sleep duration and not on the potentially more trustworthy use of time-use survey data. Many of the traditional drivers of insufficient sleep in adolescents (work, homework, other extra-curricular activities and television watching) have actually declined over this period and so are probably not driving the increased prevalence. The investigators did find that the amount of time using portable internet connected devices such as smart phones is correlated with the decline (termed ‘new media screen-time’). As feared it could be that smart devices are plundering the sleep of adolescents since their introduction and almost complete market saturation since around 2010. But the causality could still be in the opposite direction- people who cannot sleep may be filling in that time with electronic media––(suggesting that) electronic devices are not the cause of the problem but rather a self-selected solution.5 Another possibility is that the 16-year-old children of today are sleeping less because they are simply more mature than the 16 year olds of 40 years ago. As such their sleep needs and habits could be more similar to those of early adulthood than childhood. The steady decline of the age of menarche over the past 100 years provides some evidence to support this suggestion. But this idea does not work for the previous decade (the noughties) where sleep durations remain relatively stable.4 The problem is that many of these are ecological associations. Whilst more biologically plausible than pirates affecting climate, they could indeed turn out to be spurious. Some of these combined social and biological effects may be too complex for risk factor epidemiology alone to untangle. For instance, in the USA through the time periods being examined the widely employed Gini co-efficient measure of income inequality has also generally been rising. The complex effects of this on household wellbeing could plausibly be something plundering teenagers sleep health. As with all of these well-meaning explanations we must still be sceptical of any ecological association, (including the usual suspects for ‘sleep stealers’) no matter how plausible they sound. Secular trends in sleep duration and health research at the University of Sydney have been supported by (National Health and Medical Research Council of Australia, NHMRC grants #571421 #1004528 & #1060992) (to or supporting N.S.M.). Tea Lallukka supported by the Academy of Finland (Grants #287488 and #294096). Conflicts of interest: None declared.