Background Large randomised controlled trials such as TORCH (towards a revolution in chronic obstructive pulmonary disease (COPD) health) report an increased risk of pneumonia associated with use of inhaled corticosteroids (ICS) in COPD but no corresponding increase in pneumonia-related mortality. However, these trials exclude patients who are elderly, comorbid, have co-existing lung conditions or use long-term oxygen therapy and may not be representative of ‘real-world’ practice. We hypothesised that ICS use in patients that are ineligible for TORCH would be associated with increased risk of pneumonia hospitalisations and mortality.
Methods We carried out an analysis of 2 independent cohorts. The EXODUS cohort included patients admitted with COPD exacerbation and considered outcomes over 1 year including pneumonia hospitalisations and pneumonia-related mortality. The Edinburgh pneumonia study included patients hospitalised with community-acquired pneumonia with the primary outcome of 30-day mortality. A secondary analysis of patients from this cohort with spirometry-confirmed COPD during clinical stability was conducted.
Results There were 977 patients included from the EXODUS cohort. 106 patients (10.8%) were hospitalised for pneumonia and 18 patients (1.8%) had pneumonia-related mortality within 12 months of initial admission. 497 patients (50.9%) would have been ineligible for the TORCH study. In a Cox proportional hazards model, adjusting for relevant confounders, patients who were ineligible for TORCH had an increased risk of pneumonia hospitalisation (HR 1.60; 95% CI 1.04–2.45) and an increased risk of pneumonia-related mortality (HR 6.1; 95% CI 1.7–22.0). Figure 1 shows a cox adjusted survival curve for pneumonia hospitalisations in patients eligible and ineligible for TORCH.
There were 376 patients with COPD included from the Edinburgh pneumonia study. The 30-day mortality rate was 12.0%. 186 patients (38.0%) would have been ineligible for TORCH. After adjustment for relevant confounders, ICS use in patients classified as ineligible for TORCH was associated with increased risk of 30-day mortality (HR 1.85 95% CI 1.00–2.41).