Twenty-four studies met the inclusion criteria (728,377 participants, range 51 to 717,396): five RCTs (862 participants), nine quasi-randomised studies and 10 cohort studies. Downs and Black scores ranged from 11 to 22 for observational studies and 21 to 29 for RCTs. The five RCTs all scored 3 on the Jadad scale.
Bleeding: Compared to usual care, pharmacist involvement significantly reduced the incidence of total bleeds (RR 0.51, 95% CI 0.28 to 0.94, I2 =0%; four RCTs and RR 0.71, 95% CI 0.52 to 0.96, I2=77%; 19 non-RCTs). There was no difference in incidence of major bleeding in RCTs, but a significant reduction was observed in 11 non-RCTs (RR 0.49, 95% CI 0.26 to 0.93, I2=46.7%).
Thromboembolic events: Compared to usual care, pharmacist involvement showed no difference in the incidence of thromboembolic events in RCTs, but a significant reduction was observed in 15 non-RCTs (RR 0.37, 95% CI 0.26 to 0.53, I2=3.7%).
Mortality: There was no significant difference between usual care and pharmacist involvement in RCTs and non-RCTs.
Results of sensitivity analyses were reported; effects of pharmacist involvement were reported as consistent across all subgroups. There was no evidence of publication bias for any analysis.