Eighteen RCTs were included in the review (n=12,446). Allocation concealment was considered to be adequate in five trials. The risk of bias was unclear in 13 trials.
Combination therapy was associated with a significantly decreased risk of moderate COPD exacerbations (RR 0.84, 95% CI 0.74 to 0.96, NNTB 31, 95% CI 20 to 93; 13 trials) and a reduced SGRQ score (WMD -1.88, 95% CI -2.44 to -1.33; eight trials). There was significant heterogeneity (range I2=29% to 50%) in these analyses.
FEV1 was significantly increased at pre-bronchodilator therapy (WMD 0.06, 95% CI 0.04 to 0.07; 12 trials) and post-bronchodilator therapy (WMD 0.04, 95% CI 0.02 to 0.05; eight trials). There was significant heterogeneity (pre-bronchodilator therapy I2=82% and post-bronchodilator therapy I2=64%). This outcome was associated with an increased risk of pneumonia (RR 1.63, 95% CI 1.35 to 1.98, NNTH 40, 95% CI 26 to 72; 12 trials; I2 = 20%).
Statistically significant lower scores were found for dyspnea and overall withdrawals and statistically higher rates were found for viral respiratory infections and oropharyngeal candidiasis in patients who received combination therapy. Combination therapy did not significantly decrease risk of severe COPD exacerbations, all-cause mortality, respiratory and cardiovascular mortality.
Sensitivity analysis did not materially alter the main findings. The authors stated that publication bias could be ruled out for all-cause, respiratory and cardiovascular mortality, but not for severe COPD exacerbations.