Simultaneous validation of five activity monitors for use in adult populations

and five activity monitors: activPAL3 Micro, activPAL, ActiGraph GT1M, ActiGraph wGT3X-BT, and GENEActiv. Receiver operating characteristic curves and analysis were used to develop and validate cut-points for the vertical axis counts (all activity monitors) and sum of the vector magnitude (ActiGraph wGT3X-BT and GENEActiv) for 15 second (all devices) and 60 second (ActiGraph devices) epochs. A random coefficients statistical model was used to derive MET predictive equations for all activity monitors. Bland-Altman plots examined the variability in device error. No 1.5 MET cut-points were developed for the activPAL devices. All developed cut-points had high levels of sensitivity and specificity. When cross-validated in an independent group, high levels of sensitivity and specificity remained (≥77.4%, monitor and intensity dependent). The mean bias based on the Bland-Altman plots ranged from -0.03 METs to 0.35 METs (monitor dependent). This is the first study to develop and validate cut-points for five activity monitors simultaneously with high levels of sensitivity and specificity (≥77.4%). This is potentially a step toward cross-comparison/harmonization of data; however, further validation in a free-living environment is warranted.