Measuring Sedentary Behavior in People with Chronic Obstructive Pulmonary Disease (COPD)

Introduction: Sedentary behavior (SB) is a growing health problem, especially for older people with chronic diseases such as COPD; in fact people with COPD are among the least active. SB is defined as waking behavior less than or equal to 1.5 METS while sitting, reclining, or lying. Accelerometers, such as the ActiGraph (AG) are commonly used to measure SB, but the ActivPAL (AP) is considered the gold standard. There are several option for processing AG data, including use of filters and non-wear time algorithms, which will affect measures of sedentary time. People with COPD walk slowly, increasing the chances that some physical activity will be undetected with the AG and they are so sedentary that some sedentary time could be misclassified as non-wear time with the AG.

Purpose: To examine the effects of two filter and three non-wear algorithms when processing AG sedentary time data and to determine which methods have the strongest agreement with AP measured sedentary time.

Methods: Thirty-four older adults with COPD wore AG and AP monitors concurrently for 7 consecutive days. Each participant’s AG data was processed using six different methods, using all possible combinations of two filters (normal and low frequency extension) and three non-wear algorithms (60 min., 90 min., and 120 min.) that identify non-wear time based on the minimum duration of no activity. The Bland-Altman method was used to assess concordance in sedentary behavior time (min./day) between AP and each of the AG estimates.

Results: Mean sedentary time was 649.3 (SD 116) min./day as measured by the AP. When measured by AG mean sedentary time ranged from 639 (102) to 686 (104) min./day, depending on the combination of filter and non-wear algorithm used in data processing. Concordance correlation coefficients between AP-measured sedentary time and AG-measured sedentary time ranged from 0.388 to 0.511. The AG low frequency extension filter with the 60-minute non-wear algorithm resulted in the highest concordance correlation and a low mean difference between sedentary time measured by the two devices (4.4 (106) min./day).

Conclusion: For people with COPD the AG measures of sedentary time are reasonable accurate if the appropriate filter and non-wear algorithm are used. The combination of the AG low frequency extension filter with the 60-minute non-wear algorithm provides the most accurate method for processing AG sedentary data in older adults with COPD.