Abstract
Background: Detailed analyses of physical activity (PA) measures in chronic obstructive pulmonary disease (COPD) have been insufficiently explored. We aimed to identify clusters of COPD patients based on objectively measured PA data, and to compare clinical characteristics, PA measures and PA hourly patterns between these clusters.
Methods: 1001 COPD patients (65% men; median age and FEV1: 67 yrs and 49%pred, respectively) from 10 countries were studied. PA measures and hourly patterns were analysed based on data from the multi-sensor Sensewear armband used for >=4 days. Principal component analysis was applied to PA data for dimensionality reduction, subsequently k-means cluster analysis was used to identify subgroups of COPD patients.
Results: 5 clusters were identified (Table 1).
Cluster 1 (very inactive) spent less time in moderate-to-vigorous intensity and more time in very light intensity, whilst cluster 5 (very active) presented an opposite behaviour. Cluster 1 also presented higher body mass index, lower FEV1 and worse dyspnoea compared to other clusters. PA hourly patterns revealed that in all clusters the peak of intensity occurred before midday, but also that more inactive clusters had a more similar pattern between week and weekend (Figure 1).
Conclusions: Five subgroups of COPD patients were identified with distinct PA measures and hourly patterns. These findings may serve as a basis for tailored interventions in COPD.
- © 2014 ERS
