Thirteen RCTs (n=7,352 patients) and 18 registry studies (n=26,521 patients) were included in the review. All the RCTs reported methods for allocation concealment; only one RCT was reported to be double-blind. Mean follow-up in RCTs ranged from seven to 24 months; the mean follow-up in registry studies ranged from six to 36 months.
RCTs: There was no difference between drug-eluting stents and bare-metal stents for mortality (RR 0.89, 95% CI 0.70 to 1.14; 13 RCTs), myocardial infarction (RR 0.82, 95% CI 0.64 to 1.05; 13 RCTs) or stent thrombosis (RR 0.97, 95% CI 0.73 to 1.28; 13 RCTs), but there was a significant effect in favour of drug-eluting stents for target vessel revascularisation (RR 0.44, 95% CI 0.35 to 0.55; 13 RCTs). The number needed to treat with a drug-eluting stent was 14 patients (95% CI 11 to 21) to prevent one target vessel revascularisation. There was no statistical heterogeneity between the trials for all outcomes. Results were consistent by trial size, and differing follow-up periods. Single-centre trials indicated greater benefit for drug-eluting stents than multicentre trials, but the confidence intervals widely overlapped.
Registry studies: There was a significant reduction in mortality with drug-eluting stents within one year of the index percutaneous coronary intervention (RR 0.68, 95% CI 0.54 to 0.88; 11 studies), but not within two years. There was no significant difference between drug-eluting stents and bare-metal stents for myocardial infarction within one and two years from the index percutaneous coronary intervention. However, target vessel revascularisation was significantly reduced with drug-eluting stents within one (RR 0.54, 95% CI 0.40 to 0.74; 11 studies) and two years (RR 0.71, 95% CI 0.61 to 0.83; 10 studies) of the index percutaneous coronary intervention. There was also a significant difference in favour of drug-eluting stents for stent thrombosis within one year of the index percutaneous coronary intervention (RR 0.52, 95% CI 0.31 to 0.88; nine studies), but not within two years.
There was no evidence of publication bias for the RCTs, but there was evidence of publication bias for the registry studies.