Eight RCTs (n=4,909) were included in the review.
Three of the 8 trials were concerned with secondary prevention and the remaining 5 (3 with high-risk and 2 with moderate-risk patients) were primary prevention. The authors reported that randomisation and allocation concealment were adequate in all trials; blinding was not used in any of the trials.
ICDs were effective in preventing sudden cardiac death (RR 0.43, 95% confidence interval, CI: 0.35, 0.53). This was the case for both primary prevention (RR 0.37, 95% CI: 0.27, 0.50) and secondary prevention (RR 0.50, 95% CI: 0.38, 0.66). There was no significant heterogeneity amongst these trials, nor any difference observed in summary effect estimates for the different types of ICD.
There was an overall beneficial effect of ICDs on all-cause mortality (RR 0.74, 95% CI: 0.67, 0.82). This was the case for both primary prevention (RR 0.72, 95% CI: 0.63, 0.84) and secondary prevention (RR 0.76, 95% CI: 0.65, 0.89). There was significant heterogeneity between primary prevention trials of high-risk and moderate-risk patients for total mortality results (P<0.001). No survival benefit with ICD therapy was observed in moderate-risk patients.
In the 5 trials that reported on total cardiac mortality, the RR was 0.81 (95% CI: 0.69, 0.96). No differences were noted for non-cardiac mortality between patients receiving ICDs and medical treatment alone in the 3 trials that reported on this outcome (RR 0.91, 95% CI: 0.60, 1.38).
Complication rates were higher for transthoracic ICD than for newer ICD models. Further details on adverse events were given in the paper.
Funnel plots did not suggest the presence of publication bias.