Low Body Mass Index Is Associated with a Long-Term Decline in Accelerometer-Measured Physical Activity in Patients with COPD: An Exploratory Analysis of Participants Co-Enrolled in Two Prospective Cohort Studies.

Background/Objectives: Reduced physical activity (PA) and increased sedentary behavior (SB) are associated with adverse outcomes in chronic obstructive pulmonary disease (COPD), but the factors underlying their long-term change are unclear; we investigated the baseline factors associated with these changes. Methods: Japanese outpatients with stable COPD co-enrolled in the SPACE (baseline) and E-PAC (follow-up) studies were analyzed retrospectively. Twenty-nine baseline clinical variables were assessed. PA and SB were measured with a waist-worn triaxial accelerometer and expressed in metabolic equivalents (METs). Annualized percentage changes (per year, relative to baseline) were analyzed by Spearman's rank correlation and multiple linear regression. Results: Thirty-two participants (all men; median age 71.0 years; median follow-up 5.0 years) were analyzed. Lower baseline body weight, BMI, upper arm circumference, FEV1.0, FEV1.0/FVC, and lowest percutaneous oxygen saturation (SpO2) during the 6 min walk test (6MWT) were correlated with greater annualized declines in PA. For SB, the annualized change was correlated only with the lowest SpO2 during the 6MWT, FEV1.0/FVC, and the hemoglobin concentration. In multivariable models adjusted for FEV1.0 %pred and the lowest SpO2 during the 6MWT, BMI remained associated with the declines in total PA and in the duration at ≥3.0 METs; the models for the step count and SB were not significant. Conclusions: In this exploratory analysis, nutrition-related anthropometric measures, poorer pulmonary function, and exercise-induced desaturation were associated with the long-term decline in PA, and a lower BMI remained associated after adjustment. For the change in SB, associations were found only in unadjusted correlation analysis and were not supported by the multivariable model. These hypothesis-generating findings require confirmation in a larger cohort.