Seventeen RCTs (11,871 patients) were included in the review. The overall risk of bias was relatively low in most trials. Two trials were judged to be at high risk of bias for blinding. For allocation concealment and other sources of bias, the risk of bias was unclear for many trials.
Long-acting beta-agonists had a beneficial effect on chronic obstructive pulmonary disease (COPD) exacerbations compared with placebo (OR 0.81, 95% CI 0.75 to 0.88; Ι²=0%; 18 RCTs). The results for each drug (salmeterol, formoterol and indacaterol) compared with placebo were similar to the overall pooled result.
When results were stratified according to whether patients were treated with inhaled corticosteroids or not, salmeterol was equally effective in those treated and not treated compared with placebo. Formoterol was more beneficial than placebo in trials of patients treated with inhaled corticosteroids, but not better than placebo in trials of patients not treated with inhaled corticosteroids. There were no trials of patients treated with indacaterol but not treated with inhaled corticosteroids.
Long-acting beta-agonists had a beneficial effect on severe COPD exacerbations or withdrawals owing to exacerbations compared with placebo (OR 0.74, 95% CI 0.63 to 0.88; Ι²=4%; 16 RCTs). When stratified by drug, the effect was evident for salmeterol (OR 0.66, 95%CI 0.49 to 0.89; Ι²=0%; seven RCTs) and indacaterol (OR 0.42, 95% CI 0.21 to 0.83; Ι²=23%; two RCTs), but was not statistically significant for formoterol (OR 0.85, 95% CI 0.68 to 1.06; Ι²=0%; seven RCTs).
For all long-acting beta-agonists compared with placebo, the effect on severe COPD exacerbations or withdrawals owing to exacerbations was evident in patients who used inhaled corticosteroids (OR 0.68, 95% CI 0.52 to 0.89; Ι²=0%) and those who did not (OR 0.79, 95% CI 0.63 to 0.99; Ι²=25%). When these analyses were further stratified by drug, the only significant effects were in patients taking indacaterol and inhaled corticosteroids (OR 0.42, 95% CI 0.21 to 0.83; Ι²=23%) and in those taking salmeterol but not inhaled corticosteroids (OR 0.48, 95% CI 0.29 to 0.81; Ι²=0%).
Sensitivity analysis showed the meta-analyses was robust to excluding individual studies sequentially.
Funnel plots and Egger's test showed not evidence of publication bias for COPD exacerbations, but there was evidence of publication bias for severe exacerbations or withdrawals owing to exacerbations (Egger’s bias value -2.36, 95% CI -2.83 to -0.13).