Adjusted# association between physical activity variables and hospitalisation for chronic obstructive pulmonary disease (COPD) (multivariate Cox proportional hazards regression)
| HR (95% CI) | p-value | R2 | |
| Subjects n | 177 | ||
| Steps per day (change for each 1000 steps per day) | 0.79 (0.67–0.93) | 0.005 | 0.33 |
| High average intensity of physical activity (≥2.7 METs) | 2.71 (1.27–5.81) | 0.010 | |
| Interaction: steps×intensity | 1.28 (1.06–1.53) | 0.009 | |
| ≥1 COPD hospitalisation in the previous 12 months | 5.17 (2.95–9.06) | <0.001 | |
| FEV1 % pred | 0.96 (0.95–0.98) | <0.001 | |
| Physically active days per week (change for 1 day per week) | 0.79 (0.67–0.93) | 0.005 | 0.32 |
| High average intensity of physical activity (≥2.7 METs) | 2.12 (1.10–4.09) | 0.025 | |
| Interaction: days×intensity | 1.37 (1.10–1.70) | 0.005 | |
| ≥1 COPD hospitalisation in the previous 12 months | 5.19 (2.97–9.07) | <0.001 | |
| FEV1 % pred | 0.96 (0.94–0.98) | <0.001 | |
| Time in physical activity (change for 1 h per day) | 0.79 (0.63–0.99) | 0.039 | 0.31 |
| High average intensity of physical activity (≥2.7 METs) | 1.64 (0.95–2.85) | 0.078 | |
| Interaction: time×intensity | 1.36 (1.01–1.82) | 0.041 | |
| ≥1 COPD hospitalisation in the previous 12 months | 5.51 (3.16–9.62) | <0.001 | |
| FEV1 % pred | 0.96 (0.94–0.98) | <0.001 |
Data are presented as n, unless otherwise stated. An increase of 1000 steps per day at low average intensity of physical activity is related to reduced COPD hospitalisation risk (hazard ratio (HR) 0.79); high average intensity of physical activity without increasing the number of steps is related to increased COPD hospitalisation risk (HR 2.71); an increase of 1000 steps per day at high average intensity of physical activity is not related to COPD hospitalisation risk (HR 0.79×1.28=1.01). METs: metabolic equivalent tasks; FEV1: forced expiratory volume in 1 s. #: other potential confounders (sex, age, education, marital status, family members, working status, socioeconomic status, inhaled bronchodilators or corticosteroids, smoking status, smoking duration and intensity, Charlson index, cardiovascular comorbidities, modified Medical Research Council dyspnoea score, health-related quality of life, forced vital capacity, residual volume/total lung capacity, diffusing capacity of the lung for carbon monoxide, arterial oxygen tension, body mass index, fat-free mass index, 6-min walking distance, maximal oxygen uptake, lung density and structure, C-reactive protein, tumour necrosis factor-α, participation in a pulmonary rehabilitation programme and consumption of fruits, vegetables and cured meats) were not finally included in multivariate models because they did not relate to the outcome nor did they modify the coefficient estimate for the exposure >10%.