Forty-three trials (n=31,020) were included in the review. Medications assessed were long-acting β2-agonists (23 trials), inhaled corticosteroids (13 trials), tiotropium (16 trials) and combination inhaled corticosteroid/long-acting β2-agonists (11 trials). Placebo treatment was the sole comparator in 29 trials. Forty-one studies scored 3 or more on the 5-item Jadad scale, which indicated good to high quality studies. Thirty-five trials used intention-to-treat analyses. Study durations ranged from four to 160 weeks.
Exacerbations (39 trials, n=28,232):
Traditional meta-analysis showed that compared to placebo the odds of having an exacerbation of COPD were significantly decreased: by 17% with use of long-acting β2-agonists (OR 0.83, 95% CI 0.76 to 0.90); by 28% with tiotropium (OR 0.72, 95% CI 0.65 to 0.80); and by 22% with combination regimens of inhaled corticosteroids and long-acting β2-agonists (OR 0.78, 95% CI 0.70 to 0.87).
Results from MTC analysis were similar, but additional decreases in the odds of an exacerbation were observed: by 15% when inhaled corticosteroids were compared to placebo (OR 0.85, 95% CrI 0.75 to 0.97); by 18% when tiotropium was compared to long-acting β2-agonists (OR 0.82, 95% CrI 0.72 to 0.93); and by 19% when compared to inhaled corticosteroids (OR 0.81, 95% CrI 0.69 to 0.94).
Mortality (28 trials, n=26,112):
In the traditional meta-analysis, mortality decreased by 21% with use of combination therapy of inhaled corticosteroids and long-acting β2-agonists compared to placebo (OR 0.79, 95% CI 0.65 to 0.96) and by 24% with use of combination regimens compared to use of inhaled corticosteroids alone (OR 0.76, 95% CI 0.63 to 0.93).
In the MTC analyses, statistically significant increases in mortality were found with tiotropium treatment compared to combination therapy of inhaled corticosteroids and long-acting β2-agonists (OR 1.84, 95% CI 1.07 to 3.17). Other comparisons of medications to placebo or to each other did not result in statistically significant reductions in mortality.
Withdrawals from trials (29 trials, n=22,487):
Patient withdrawals were significantly reduced with: all active treatments compared to placebo; tiotropium and combination therapies compared to monotherapy with long-acting β2-agonists; and combination therapy compared to inhaled corticosteroids. Similar trends were observed in the MTC analysis.
Statistical heterogeneity was reported for comparisons of exacerbations with inhaled corticosteroids with placebo (Q=15.9, p=0.04), combination therapy of inhaled corticosteroids and long-acting β2-agonists (Q=17.3, p=0.008) and patient withdrawals from trials when long-acting β2-agonists were compared to placebo treatments (p=0.04).
No significant variations from the results of the MTC meta-analyses were observed in: sensitivity or subgroup analyses conducted on the basis of follow-up periods; severity of COPD; studies that used intention-to-treat analyses; and studies that attained a Jadad quality score of 3 or more out of 5.