Fourteen RCTs, with 7,818 participants, were included. All were randomised and reported on withdrawals and loss to follow-up, but only six trials reported allocation concealment.
Overall, freedom from angina at follow-up was significantly more likely in the percutaneous coronary intervention group than in the medical therapy group (OR 1.69, 95% CI 1.24 to 2.30). Significant heterogeneity was present.
Subgroup analysis indicated that the benefit of percutaneous coronary intervention was greatest in older trials (recruitment ended in 1990 to 1994; three RCTs) and was not significant in trials that completed recruitment in 2000 or later (five RCTs).
There was a statistically significant benefit of percutaneous coronary intervention at one year follow-up (eight RCTs) and one to five year follow-up (10 RCTs), but not in trials with follow-up longer than five years (five RCTs).
Meta-regression analysis revealed a statistically significant inverse relationship (p=0.021) between odds ratio for freedom from angina and number of evidence-based medications used in the trial. Results of other analyses were reported.
No evidence of publication bias was found.