Rethinking the sleep-health link

Sleep is important for the physical, social and mental well-being of both children and adults. In this paper, we discuss the need to consider sleep as a multidimensional construct and as a component of total 24-hour activity. First, we make a case for considering sleep as a multidimensional construct, whereby all characteristics of sleep (including duration, quality, timing, and variability) and their links with health are examined. Second, we argue that sleep should also be conceptualized as part of the daily spectrum of time-use, along with other types of activity. We propose novel statistical models, in particular compositional data analysis (CoDA), as appropriate analytical methods for a new sleep paradigm.