Impact of accelerometer epoch summary measure on associations between physical activity and all-cause mortality in Whitehall II and UK Biobank

Abstract

Abstract Accelerometer data are commonly reduced into epoch summary measures (ESMs) for analysis, e.g. ENMO (Euclidean Norm Minus One), MAD (Mean Amplitude Deviation), MIMS (Monitor Independent Movement Summary) or Counts. We compared associations with all-cause mortality of the volume and intensity of physical activity when derived from those four measures in the Whitehall II and UK Biobank cohorts. Volume (Average Acceleration, AvAcc) and intensity (Intensity Gradient, IG) were derived from each ESM. Associations with mortality were estimated using Cox models. 3733 (25.1% female, median age 68.3 years) and 89,848 (56.4% female, 63.5 years) participants were included from Whitehall II and UK Biobank, respectively. Median (IQR) follow-up was 11.0 (10.7, 11.3) and 8.0 (7.5, 8.5) years, with 563 (15.1%) and 3656 (4.1%) deaths. Associations with mortality were largely consistent between ESMs with the lowest mortality risk for those high (above the median) in both AvAcc and IG (Whitehall: HR = 0.59–0.68; Biobank: 0.55–0.61, reference: low/low), and IG associated with lower mortality risk, irrespective of AvAcc. AvAcc was associated with lower mortality irrespective of IG in Biobank only. In conclusion, associations of AvAcc and IG with mortality are broadly consistent across common ESMs, supporting comparability of activity-health findings across studies using different ESMs.