Joint association between accelerometry-measured daily combination of time spent in physical activity, sedentary behaviour and sleep and all-cause mortality: a pooled analysis of six prospective cohorts using compositional analysis

Abstract

Objective: To examine the joint associations of daily time spent in different intensities of physical activity, sedentary behaviour and sleep with all-cause mortality.

Methods: Federated pooled analysis of six prospective cohorts with device-measured time spent in different intensities of physical activity, sedentary behaviour and sleep following a standardised compositional Cox regression analysis.

Participants: 130 239 people from general population samples of adults (average age 54 years) from the UK, USA and Sweden.

Main outcome: All-cause mortality (follow-up 4.3-14.5 years).

Results: Studies using wrist and hip accelerometer provided statistically different results (I²=92.2%, Q-test p<0.001). There was no association between duration of sleep and all-cause mortality, HR=0.96 (95% CI 0.67 to 1.12). The proportion of time spent in moderate to vigorous physical activity was significantly associated with lower risk of all-cause mortality (HR=0.63 (95% CI 0.55 to 0.71) wrist; HR=0.93 (95% CI 0.87 to 0.98) hip). A significant association for the ratio of time spent in light physical activity and sedentary time was only found in hip accelerometer-based studies (HR=0.5, 95% CI 0.42 to 0.62). In studies based on hip accelerometer, the association between moderate to vigorous physical activity and mortality was modified by the balance of time spent in light physical activity and sedentary time.

Conclusion: This federated analysis shows a joint dose-response association between the daily balance of time spent in physical activity of different intensities and sedentary behaviour with all-cause mortality, while sleep duration does not appear to be significant. The strongest association is with time spent in moderate to vigorous physical activity, but it is modified by the balance of time spent in light physical activity relative to sedentary behaviour.

Keywords: health; physical activity; sleep.

Figure 1

Structure of federated analysis employed. UK Biobank and National Health and Nutrition Examination Survey (NHANES) are open-source data sets.

Figure 2

Results of the sensitivity analysis showing Cox regression model coefficients with 95% CI bars for each of the isometric log ratio (ilr) coordinates ( $z_1$ , $z_2$ , $z_3$ ) in models including all studies, leave-one-out models and in models pooling studies using wrist accelerometers only and hip accelerometers only. For hip accelerometers, only $z_1$ was not calculated. NHANES, National Health and Nutrition Examination Survey; WHS, Women’s Health Study.

Figure 3

Results from pooled studies using wrist accelerometers. Dose–response relationship (with 95% CI—ribbons) between time spent in (A) sleep (relative to all other behaviours), (B) moderate to vigorous physical activity (MVPA) (relative to all other behaviours), (C) light physical activity (LIPA) (relative to all other behaviours), (D) sedentary behaviour (SB) (relative to all other behaviours), (E) joint association between time in MVPA, SB (presented as different levels) and LIPA (implied as it makes up the remaining time in the waking day time in LIPA=16 hours−time in MVPA−time in SB). HRs were computed with respect to the following reference composition defined as described in the Methods section: MVPA=20 min/day, LIPA=60 min/day, SB=14 hours and 40 min/day, sleep=8 hours/day (marked as a solid black dot). Compositions with different values in time in the primary behaviour were reported such that the remaining behaviours were in the same ratio as for the reference composition.

Figure 4

Results from pooled studies using hip accelerometers. Dose–response relationship (with 95% CI—ribbons) for waking day behaviours between time spent in (A) moderate to vigorous physical activity (MVPA) for different levels of sedentary time (with light physical activity (LIPA) making up the remaining time in the waking day), (B) LIPA for different levels of MVPA (with sedentary time making up the remaining time in the waking day), and (C) sedentary behaviour (SB) for different levels of MVPA (with LIPA making up the remaining time in the waking day). HRs were computed with respect to the following reference composition defined as described in the Methods section: MVPA=2 min/day, LIPA=229 min/day, SB=729 min/day, sleep=8 hours/day.

Figure 5

Results from pooled studies using hip accelerometers. HRs for different compositions of the waking day are presented as a heat map in (A). Moderate to vigorous physical activity (MVPA) and sedentary time are shown on the x-axis and y-axis, respectively. The remaining time in the waking day is made up of light physical activity (LIPA), black lines represent LIPA isotime lines. The dashed blue line represents composition with 30 min of MVPA. Heat maps of the lower and upper bounds of the 95% CIs are shown in (B) and (C). HRs were computed with respect to the following composition: MVPA=2 min/day, LIPA=229 min/day, SB=729 min/day, sleep=8 hours/day. SB, sedentary behaviour.