Five RCTs (2,219 patients) were included in the review. Four trials had a Jadad score of 4; one trial scored 3 (out of a maximum of 7 points). Four trials had intention-to-treat analyses.
Self monitoring/management of anticoagulant therapy was associated with a reduction in thromboembolic events (OR 0.52, 95% CI 0.35 to 0.77; Ι²=0%; five RCTs) and all-cause mortality (OR 0.50, 95% CI 0.29 to 0.86; Ι²=0%; four RCTs). There was no evidence of a significant difference in major haemorrhage (OR 1.07, 95% CI 0.77 to 1.50; Ι²=0%; five RCTs) or minor haemorrhage (figures not reported, two RCTs).
Subgroup analysis, according to whether therapy was adjusted or not adjusted after self monitoring, mirrored the overall results for incidence of major haemorrhage and thromboembolic events. For all cause mortality, self monitoring with no therapy adjustment was associated with a significant reduction when compared with control groups (OR 0.53, 95% CI 0.30 to 0.93), but there was no significant reduction using self monitoring with self-adjusted therapy (OR 0.35, 95% CI 0.07 to 1.79).
For mean international normalised ratio (INR) within target range, all trials reported significant differences favouring the self monitoring/management groups when compared with control groups (range of mean INR within target range in intervention group was 43.2 to 78.3%; in control groups it was 22.3 to 65%). The self monitoring/management groups had more frequent tests compared with control groups; the ratio of test frequency between the groups ranged from 2.29 to 4.99.
Sensitivity analyses with the removal of one trial that did not use intention-to-treat analysis did not materially change the findings.
There was no evidence of publication bias.