Five RCTs were included in the review (n=1,597 patients in text, 1,591 patients in table 1; range 59 to 500 patients). One trial scored 4 points on the Jadad scale, two RCTs scored 3, and two RCTs scored 2.
Primary endpoint: The pooled odds ratio for stenosis at six months was 0.50 (95% CI 0.38 to 0.66), favouring triple over dual therapy; heterogeneity was not significant (I2=0%).
Secondary endpoints: Coronary artery minimal luminal diameter favoured triple over dual therapy (mean difference 0.18mm, 95% CI 0.1 to 0.27). In-segment and in-stent late loss were reported and significantly lower in patients receiving triple therapy (three RCTs comprising 90% of the patients). Bleeding rates were not statistically significantly different between double and triple therapy groups (three RCTs).
Subgroup analyses: No significant difference in outcome was identified related to the type of stent used or the duration of thienopyridine therapy.
Sensitivity analyses: The results did not differ substantially when trials with Jadad scores of 2 or less were excluded, or for a random-effects rather than fixed-effect model.
The funnel plot was reported to be asymmetrical, raising the possibility of publication bias.