Methods and impact of correcting heart rate data errors during exercise in older adults

Heart rate (HR) by time curves could be useful as a measure of treatment fidelity (TF). The purposes were to describe the frequency of common recording irregularities (e.g. errors) observed during exercise, validate a process to correct those errors, and determine whether there is a clinically meaningful benefit to data correction. In total, 1895 exercise sessions from 69 older adults with hip fracture were inspected for errors in the placement of event markers (duration of exercise) and signal artifacts. Marker errors occurred in 33% of the participants and corrected in 324 sessions. Signal artifacts occurred in 96% of the participants and eliminated in 702 sessions. Computer code was 85% accurate in detecting signal artifacts, compared to 97% for investigators. HR was significantly higher after correction, but the average change was only 0.69 + 1.20 beats per minute. An HR monitor showing HR by time curves (HRxTC) can be used to evaluate TF during exercise.