How reliable is measurement of posture during sleep: real-world measurement of body posture and movement during sleep using accelerometers.

Objective . Understanding sleeping behaviours could improve prevention and treatment of sleep problems and associated health conditions. This study aimed to evaluate a method to assess body posture and movement during sleep using trunk-worn accelerometers for 28 nights. Approach . Participants (50 adults with low back pain (66% female); aged 32(±9) years) wore two activPAL-micro sensors (thigh, trunk) during their normal daily life for 28 consecutive days. Parameters related to body posture (e.g. time spent lying supine or prone) and movement (e.g. number of turns) during sleep were calculated for each night. Average values for each parameter were identified for different periods, the Spearman-Brown Prophecy Formula was used to estimate the minimum number of nights required to obtain a reliable estimate of each parameter, and repeatability of measures between different weeks was calculated. Main results . Participants spent 8.1(±0.8) h asleep and most time (44%) was spent in a supine posture. The minimum number of nights required for reliable estimates varied between sleep parameters, range 4-21 nights. The most stable parameters (i.e. requiring less than seven nights) were 'average activity', 'no. of turns', 'time spent prone', and 'posture changes in the first hour'. Some measures differed substantially between weeks. Significance . Most sleep parameters related to body posture and movement require a week or more of monitoring to provide reliable estimates of behaviour over one month. Notably, one week may not reflect behaviour in another week, and the time varying nature of sleep needs to be considered.