Eleven trials (n=1,391) were included in the review. Overall study quality was good: six trials scored A, four trials scored B and one trial scored C.
Clinical worsening: Medication significantly reduced clinical worsening compared with placebo (OR 0.33, 95% CI 0.22 to 0.49; 11 trials). In subgroup analyses, reduction in clinical worsening remained significant for patients treated with bosentan (OR 0.26, 95% CI 0.13 to 0.50; five trials) and sildenafil (OR 0.34, 95% CI 0.18 to 0.64; three trials), but not for iloprost (three trials). There was no evidence of statistical heterogeneity.
Functional class: Medication significantly improved functional class compared with placebo (OR 2.81, 95% CI 1.95 to 4.03; eight trials). Improvement in functional class was significant for bosentan, sildenafil and iloprost.
Six-minute walk test: Medication significantly improved the six-minute walk test compared with placebo by an average of 31.13m in the bosentan group, 36.53m in the sildenafil group and 31.46m in the iloprost group (11 trials).
Haemodynamic parameters: Medication significantly improved all haemodynamic parameters compared to placebo. Pulmonary arterial systolic pressure was reduced by an average of 4.64mmHg (95% CI -6.02 to -3.26; five trials). Mean pulmonary arterial pressure was reduced by an average of 4.05mmHg (95% CI -4.54 to -3.56; four trials). Pulmonary vascular resistance was reduced by an average of 246.09dyn/s/cm-5 (95% CI -319.13 to -173.04; seven trials). The cardiac index was increased by an average of 0.40L/min-1/min-2 (95% CI 0.11 to 0.69; four trials). Cardiac output was increased by an average of 0.53L/min (95% CI 0.08 to 0.98; three trials). There was evidence of statistical heterogeneity for analyses of pulmonary vascular resistance, cardiac index and cardiac output.
Safety: There was no significant difference in the number of serious adverse events between participants who received one of the three study medications and those who received placebo (OR 1.09, 95% CI 0.69 to 1.71; 11 trials). The incidence of serious adverse events was lowest in patients who received bosentan and highest in patients who received iloprost (Χ2=57.134, p<0.0001). There was no evidence of statistical heterogeneity for this analysis.
An asymmetric funnel plot indicated possible publication bias.