Twenty RCTs (68,680 participants; nine trials had sample sizes over 1,000). Randomisation was conducted in all trials, but clearly described in only 12. Loss to follow-up was described in 19 trials. Intention-to-treat analysis and double-blinding was used in 16 trials. Allocation concealment was carried out in 14 trials. The authors did not give an overall view on the quality of each trial but it seemed that seven trials were of unclear risk of bias and four were at high risk based on the table of results.
There were no statistically significant associations between omega-3 PUFA supplementation and all-cause mortality (RR 0.96, 95% CI 0.91 to 1.02; 17 RCTs, Ι²=12%), cardiac death (RR 0.91, 95% CI 0.85 to 0.98; 13 RCTs, Ι²=6%), sudden death (RR 0.87, 95% CI 0.75 to 1.01; seven RCTs, Ι²=8%), myocardial infarction (RR 0.89, 95% CI 0.76 to 1.04; 13 RCTs, Ι²=35%) and stroke (RR 1.05, 95% CI 0.93 to 1.18; nine RCTs, Ι²=14%). Results for absolute risk reductions were not statistically significant.
Subgroup analyses did not alter the main findings. Cumulative meta-analysis showed a weakened intervention effect over time, although a substantial increase in study size was reported at the point at which the effect levelled out. There was no evidence of publication bias.
Opposite directions of effect were reported for all-cause mortality and cardiac death in two RCTs that evaluated dietary counselling interventions. The differences could not be explained by study-specific characteristics.