Eleven studies, including 2,950 patients, were included.
The trial reports provided inadequate information on the method of randomisation. The outcome assessors were blinded for some or all outcomes in four trials. All the trials provided some information on withdrawals.
The meta-analysis showed no statistically significant difference between PCI and conservative therapy for death, cardiac death or MI, nonfatal MI, CABG or PCI during follow-up. There was generally no statistically significant heterogeneity between the studies; the exception was in the analysis of PCI during follow-up. The Bayesian analysis gave similar results.
In the sensitivity analysis, smaller trials with fewer events showed significantly less favourable effects with PCI compared to conservative therapy than larger trials with more events for the outcomes cardiac deaths and MI and nonfatal MI (p=0.024).
The subgroup analysis showed a statistically significant difference in the risk of death and subsequent PCI between trials that enrolled patients with a recent history of MI versus the other trials (p<0.05). In the subgroup of two trials that enrolled patients with relatively recent MI there was a statistically significant reduction in the risk of death (RR 0.40, 95% CI: 0.17, 0.95, p=0.037) and subsequent PCI (RR 0.42, 95% CI: 0.20, 0.91, p=0.029).