Associations Between Common Physical Activity Accelerometer Metrics: NHANES 2011-2014

Researchers have a variety of metrics to choose from when summarizing raw data obtained during accelerometer-based physical activity assessments. Common metrics include activity counts, mean amplitude deviation (MAD), Euclidean norm minus one (ENMO), and monitor independent movement summary (MIMS) units. Associations between these metrics have yet to be fully explored in diverse and nationally representative samples. PURPOSE: To examine associations between commonly used accelerometer-based physical activity metrics. METHODS: We performed a secondary analysis of 13,485 US participants (age ≥ 3; 48.1% male) with wrist-worn accelerometer data collected during the 2011-2014 National Health and Nutrition Examination Survey (NHANES). Linear regression and correlations were used to quantify the nature and magnitude of associations between mean daily vector magnitude (VM) activity counts, MAD, ENMO, and MIMS units. Survey procedures were used for all analyses to account for the complex, multi-stage design of NHANES. RESULTS: Regression-based graphical depictions indicated linearity between all potential variable combinations (see Figure). All variables were significantly correlated (all p < 0.001). Association magnitudes for VM activity counts were highest with MIMS units (r = 0.995, 95% CI: 0.995-0.996) and lower for MAD (r = 0.874, 95% CI: 0.869-0.879) and ENMO (r = 0.770, 95% CI: 0.761-0.778). MAD demonstrated similar association magnitudes with ENMO (r = 0.904, 95% CI: 0.900-0.909) and MIMS units (r = 0.889, 95% CI: 0.885-0.894). The association for ENMO and MIMS units was among the lowest observed (r = 0.788, 95% CI: 0.779-0.795). CONCLUSIONS: All quantified accelerometer-based physical activity metrics were strongly and significantly associated with each other. Future research is needed to better understand the comparability of these accelerometer-based metrics as exposures in physical activity epidemiological research.