Purpose: Accelerometers are used to assess physical activity intensity levels and durations across populations. This is done by dividing the device output into categories that correspond to light, moderate, and vigorous physical activity. Cut points provide where these dividing lines should be. However, there is not a consistent set of cut points for any given population. This makes inter-study comparison difficult and it is unknown how using different cut point sets affects outcomes. The aim of this study is to determine agreement between four different commonly used cut points. Procedure/Description: The NEXT Generation Health Study is longitudinal study funded by the NIH Intramural Research programs. NEXT Plus is a subset of the larger sample that wore accelerometers for one week intervals (n=150). The physical activity monitors used in this trial were the GT3X by ActiGraph. Data files were first converted to .agd files with a 10 second epoch using ActiLife software. Next, each cut point definition was used to give time spent in each intensity. The cut points used to evaluate the data were by Freedson, Romanzini (which has two sets), and Santos-Lozano. Physical activity guidelines from the CDC were applied to each cut point definition output. An agreement analysis was then calculated for each output. Statistical analyses were performed in SAS software version 9.4. P values < 0.05 were considered statistically significant. Results/Outcomes: There were significant differences in time spent in light, moderate, vigorous, and moderate and vigorous combined between each pair of cut point definitions (p<0.0001). Also, there was significant disagreement between each cut point definition regarding if individuals met the CDC guidelines (p<0.0001).
Implications/Future Direction: Cut point definition selection has a noteworthy effect on determining the duration of time spent in each intensity of physical activity. As this measure is often used as a main outcome of interest, past studies’ conclusions may be based on inaccurate data. These findings further complicate inter-study comparison when different cut point definitions are used. Future studies should determine if common cut point definitions used in other populations provide similar outcomes and perhaps rethink monolithic cut point definitions to express greater variability seen within groups.
