Twenty-five trials (from 23 articles) were included in the review (82,868 patients); 11 trials specifically reported on cancer mortality (13, 568 patients in Table 1; 16,066 patients in text). All trials reported blinding status and intention-to treat. Twenty trials reported randomisation and sequence generation. Fifteen trials reported allocation concealment. Three trials had more than 20% losses to follow-up. The duration of follow-up ranged from one to 80 months.
Low-dose aspirin was associated with a statistically significantly lower risk on non-vascular mortality (RR 0.88, 95% CI 0.81 to 0.96; 24 trials; Ι²=0%) and cancer mortality (RR 0.77, 95% CI 0.63 to 0.95; 11 trials; Ι²=0%) compared with placebo or no treatment.
There was no significant difference in longer duration trials compared with shorter duration trials, or in trials with a higher dose compared with lower dose. Cumulative meta-regression indicated that significant effects were observed after an average of four years follow-up. Sequential analysis indicated that the optimal information size for nonvascular deaths was 33,586.