Nineteen controlled trials with 31,569 patients were included in the review.
Agreement on study inclusion, as measured by the kappa statistic, was 0.75. There was no evidence of publication bias.
Nine trials received the maximum quality score of 5. Allocation concealment was adequate in 8trials. With the exception of one trial, treatment was randomised. Blinding of the participants was achieved in all 19 trials, investigators in 17 trials and outcome assessors in 18. Intention-to-treat analysis was used in 15 trials.
ARBs did not lead to a statistically significant increase in the risk of MI when compared with placebo (OR 0.94, 95% CI: 0.75, 1.16), based on 11 trials (n=21,062). However statistically significant heterogeneity was noted. When compared with ACE inhibitors, the OR was 1.01 (95% CI: 0.87, 1.16), based on 9 trials (n=10,625). The OPTIMAAL trial of losartan versus captopril in patients with recent MI or ischaemic syndrome accounted for 86.8% of the weighted OR. Analyses using the fixed-effect model gave similar results.