Save the Date!
ActiGraph Digital Data Summit 2021November 4 - 5 | Register for Event Updates Now
Inferred Time In Bed Independently Predicts Levels Of Daytime Activity And Sedentary Behavior
- Presented on June 17, 2013
Purpose Increased all-cause mortality has been consistently associated with longer (8-10 hours+) self-reported sleep duration. The possibility that longer sleep may impact survival through inactive lifestyles was proposed by Morgan (2007), and subsequently tested by Hartescu et al (2012) who concluded that, independent of health status, longer sleep duration, and the inevitably longer periods of time spent in bed, could represent inactivity and/or sedentary behavior. It may be possible to infer Time in Bed (TIB: i.e. the time spent in bed irrespective of time spent asleep) from the periods excluded from daytime data collection in instrumental surveys of physical activity. The present analyses explore this possibility and address the question: is inferred TIB predictive of daytime activity/sedentary behavior levels?
Methods Profiles of health and physical activity were obtained from a random community sample of 1917 adults aged 25+ assessed for the 2008 Health Survey for England. Sedentary time and medium-vigorous physical activity (both in minutes/day) were calculated from accelerometer data over a 7-day period. The 1 minute epoch Actigraph GT1M accelerometer data was analysed using KineSoft version 3.3.75. Only those who had valid accelerometry data for at least 1 day (i.e., at least 10 hours of wear) were included. Time in Bed (TIB) was inferred from accelerometer “non-wear” periods (devices were removed at bedtime, and replaced at waking time). To assess the strength of associations between sleep and daytime variables separate regression models were fitted with sedentary time (Model 1) and moderate-vigorous activity (Model 2) as dependent variables. In both models, TIB, age, sex, BMI, health status, and mental health were entered as covariates.
Results The modal inferred TIB duration was 600 minutes. In the adjusted regression models, longer inferred TIB duration was significantly associated with lower levels of medium-vigorous physical activity (r2 =0.17, F(6, 1910) = 66.66, p<0.01), but also with significantly lower levels of daytime sedentary behavior (r2 =0.13, F(6, 1910) = 49.08, p<0.01).
Conclusions While the explained variance is modest, longer TIB emerged as a significant predictor of both physical activity and sedentary behavior. The directionality of these relationships, however, indicates that lower PA, rather than higher levels of sedentary behavior may be contributing to TIB-mortality relationships, supporting the use of this proxy measure as an analogue of TIB. Confirmation of these findings will be required, but the consistency of the present findings supports the use of non-wear or “non-recording” time as a proxy for time spent in bed.