Validating hip- and wrist-ActiGraph accelerometer cut-points for physical activity intensities in people living with coronary heart disease

Abstract

People with coronary heart disease (CHD) are encouraged to meet the public health moderate-to-vigorous aerobic physical activity (MVPA) guidelines for secondary prevention of cardiovascular disease. However, no accelerometer aerobic intensity cut-points are currently available to classify MVPA in this population. This study aimed to establish absolute and relative aerobic physical activity intensity accelerometer cut-points in people with CHD and compare the new with existing cut-points in an international cohort. Eighty-six participants with CHD performed a resting-metabolic-rate (RMR) assessment, activities-of-daily-living (ADLs) and a peak treadmill test with mixed-chamber gas analysis while wearing two ActiGraph GT3X accelerometers (hip and wrist). The average RMR was 2.8 ml.kg-1.min-1, 20% less than the commonly used 1 Metabolic Equivalent of Task (3.5 ml.kg-1.min-1). The study sample was randomly split into a training and independent validation set (2:1) allowing for cross validation. In the training set, there were significant positive correlations between accelerometer counts.min-1 (y-axis, vector-magnitude (VM)) and intensity (relative and absolute) across both accelerometer hip- and wrist-placements for all activities (p < 0.001). Using Generalized Estimating Equation modelling, there was a strong linear relationship between accelerometer counts and absolute intensity for hip-placement (R2 = 0.62–0.71), and weaker relationships for hip relative intensity (R2 = 0.40–0.47) and wrist-placement (R2 = 0.09–0.25). In the validation set, Bland-Altman plots found that the mean differences between predicted and actual absolute and relative intensity measures were negligible for all accelerometer counts.min-1 (y-axis, VM) and placements (hip, wrist), although the dispersion of the differences (95% limits of agreement) were wide. Hip VM counts.min-1 cut-points were found to best identify absolute and relative MVPA. In the international comparison (n = 176), participants completed significantly more MVPA using the new cut-points (p < 0.001). Thus, accelerometer cut-points developed in healthy individuals appear to underestimate physical activity intensity in this population and cut-points specific to people with CHD should be used.