Physical Activity Measurement Accuracy in Individuals With Chronic Lung Disease: A Systematic Review With Meta-Analysis of Method Comparison Studies

Abstract

ObjectiveTo investigate the accuracy of physical activity measurement strategies in adults with chronic lung disease. Data SourcesMEDLINE, Embase, and CINAHL databases were searched from inception to September 30, 2014. Study SelectionStudies reporting validity data for devices measuring energy expenditure in comparison with indirect calorimetry or doubly labeled water measurements in chronic lung disease were included. Nine publications in chronic obstructive pulmonary disease (COPD) or cystic fibrosis (CF) from 2294 studies were identified. Data ExtractionTwo reviewers evaluated studies for quality using a modified version of the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) checklist and extracted data relating to population, setting, devices, activity protocols, and energy expenditure. Disagreements were resolved by consensus. Data SynthesisStudies were of high quality, with 8 studies scoring at least 9 out of 11 on the QUADAS checklist. In laboratory-based settings, the SenseWear multisensor accurately estimated energy expenditure during walking compared with indirect calorimetry (pooled mean difference, −0.7kcal/min; 95% confidence interval [CI], −2.5 to 1.1) in COPD, but overestimated it in CF. However, 2 studies in COPD and CF showed the SenseWear multisensor accurately estimated energy expenditure during lifestyle tasks compared with indirect calorimetry (pooled mean difference, .18kcal/min; 95% CI, −.13 to .49). The Digi-Walker pedometer underestimated energy expenditure compared with indirect calorimetry in COPD (mean difference walking, −2.4kcal/min; 95% CI −3.4 to −1.1; mean difference lifestyle tasks, −2.3kcal/min; 95% CI, −2.8 to −1.8). In free-living settings, the ActiReg multisensor accurately measured energy expenditure in COPD (mean difference, −21kcal/d; 95% CI, −133.9 to 91.9), whereas the Flex Heart Rate Method underestimated energy expenditure in CF (mean difference, −454.1kcal/d; 95% CI, −727 to −181.2). ConclusionsEnergy expenditure estimation was accurate from the SenseWear and ActiReg multisensors during laboratory-based and free-living testing. Future studies warrant investigation of activity measures in other lung diseases and in specific ranges of lung disease severity.