Accuracy of a wearable device for assessing sedentary behavior in Chronic Obstructive Pulmonary Disease

Abstract

Physical activity (PA) and sedentary time (ST) independently predict mortality in Chronic Obstructive Pulmonary Disease (COPD). Thus, accurate measurement of PA and ST is important. The accuracy of wearable fitness devices to measure PA has been assessed in health and chronic disease. Few studies have assessed the accuracy of ST measurement in consumer-available devices, and this was the aim of this study. <b>Methods:</b> People with COPD wore a FitBit Inspire™ and the validated ActivPAL4™ simultaneously for two weeks. Total ST (sitting + lying time) for valid data (24-hour wear time for ActivPAL™ and no evidence of FitBit™ removal confirmed visually on the web dashboard) was summarized. To assess FitBit™ accuracy, mean absolute percent error (MAPE), intra-class correlation (ICC), and Bland-Altman analysis was performed. <b>Results:</b> Across 15 participants (53% female; age: (mean±SD) 70.9±6.3; FEV₁ %predicted: 43.3±27.6), 130 valid days were identified. MAPE (%) was 7.2%. ICC was 0.5 (95% CI: 0.38-0.60). Bland-Altman analysis is shown in Fig 1. On average, FitBit™ underestimated daily ST by 16.3 min compared to ActivPAL™. <b>Discussion:</b> FitBit™ showed fair ST accuracy and a moderate level of agreement when compared to the validated ActivPAL4™. FitBit™ slightly underestimated ST on average and agreement between devices appears to decrease at lower ST values.&nbsp;In COPD, a FitBit Inspire™ may provide an accurate estimate of ST.