Nineteen RCTs (8,987 participants: 4,574 with DES and 4,413 with BMS) were included.
Rates of restenosis were significantly lower in the DES groups compared with the BMS groups: 10.5% in the DES group (19 studies) versus 31.7% with BMS (OR 0.25, 95% CI: 0.22, 0.29, p<0.001), 7.6% in the sirolimus subgroup (9 studies) versus 36.8% with BMS (OR 0.14, 95% CI: 0.11, 0.17, p<0.01), and 12.4% in the paclitaxel subgroup (10 studies) versus 28.4% with BMS (OR 0.35, 95% CI: 0.30, 0.41, p<0.001).
Rates of MACE were significantly lower in the DES groups compared with the BMS groups: 19.9% in the DES group (19 studies) versus 10.1% with BMS (OR 0.46, 95% CI: 0.41, 0.52, p<0.001), 7.4% in the sirolimus subgroup (9 studies) versus 21.9% with BMS (OR 0.28, 95% CI: 0.22, 0.35, p<0.001), and 12% in the paclitaxel subgroup (10 studies) versus 18.3% with BMS (OR 0.62, 95% CI: 0.53, 0.73, p<0.001).
However, no significant differences were observed between DES groups and BMS groups with respect to the individual outcomes of mortality, Q-wave MI and non-Q-wave MI events.
No significant differences were observed between DES groups and BMS groups with regard to stent thrombosis.
In the sensitivity analyses, the removal of three of the largest studies did not alter the overall effect on MACE or binary restenosis.
In the effect model, the effectiveness of sirolimus stents improved with increasing baseline risk in the control groups; the effectiveness of the paclitaxel stents remained constant regardless of baseline risk.
Tests for publication bias suggested that 10 negative studies would be needed to affect the meta-analysis results for MACE or binary restenosis.