Twelve RCTs were included in the review (7,680 patients); eight compared salmeterol/fluticasone with placebo (5,205 patients) and four compared budesonide/formoterol with placebo (2,475 patients). No trials were excluded from the analyses on the assessment of risk of bias. All trials were double-blinded. Discontinuation rates ranged from 5% to 34% in the active intervention groups and 5% to 44% in the placebo groups.
The results of the adjusted indirect comparison found that a significantly lower proportion of patients experienced a pneumonia averse event (OR 0.47, 95% CI 0.28 to 0.80) or a pneumonia serious adverse event (OR 0.41, 95% CI 0.19 to 0.86) with budesonide/formoterol compared with salmeterol/fluticasone. No significant difference between the treatment comparators was found for pneumonia-related mortality. No significant heterogeneity was found in the pairwise meta-analyses.
A sensitivity analysis, excluding one trial in which the salmeterol/fluticasone dose (50/250ug bid) differed from the other trials, was performed; the results did not substantially alter the findings of the primary analyses. Testing for small study effects was not possible.