This study demonstrates a novel data-driven method of summarising accelerometer data to profile physical activity in three diverse groups, compared with cut-point determined moderate-to-vigorous physical activity (MVPA). GGIR was used to generate average daily acceleration, intensity gradient, time in MVPA and MX metrics (acceleration above which the most active X-minutes accumulate) from wrist-worn accelerometer data from three datasets: office-workers (OW, N = 697), women with a history of post-gestational diabetes (PGD, N = 267) and adults with ≥1 chronic disease (CD, N = 1,325). Average acceleration and MVPA were lower in CD, but not PGD, relative to OW (−5.2 mg and −30.7 minutes, respectively, P < 0.001). Both PGD and CD had poorer intensity distributions than OW (P < 0.001). Application of a cut-point to the M30 showed 7%, 17% and 28%, of OW, PGD and CD, respectively, accumulated 30 minutes of brisk walking per day. Radar plots showed OW had higher overall activity than CD. The relatively poor intensity distribution of PGD, despite similar overall activity to OW, was due to accumulation of more light and less higher intensity activity. These data-driven methods identify aspects of activity that differ between groups, which may be missed by cut-point methods alone. Abbreviations: CD: Adults with ≥1 chronic disease; mg: Milli-gravitational unit; MVPA: Moderate-to-vigorous physical activity; OW: Office workers; PGD: Women with a history of post-gestational diabetes; VPA: Vigorous physical activity.
- MX metrics
- intensity gradient