Nine RCTs with 1,336 participants were included in the review but only 7 trials (6 warfarin and 1 heparin) were used in the meta- analysis. The statistical unit for the analyses was the dose, so the number of tests was 3,416 dosages carried out in 1,327 participants.
The pooled odds ratio (OR) for the effect of intervention on the quality of anticoagulation was 1.29 (95% CI: 1.12, 1.49) which was statistically significant but there was significant heterogeneity in this result (p < 0.02).
After excluding one small study with the largest effect, the OR was 1.25 (95% CI: 1.08, 1.45) and the homogeneity test was statistically not significant (P=0.12).
Results on the effect of intervention on clinical outcomes reported 14 major haemorrhages in 700 participants in the intervention group versus 25 among 636 in the control groups (2.0% versus 3.9%). Combining all major events (death, major haemorrhage and thrombo-embolic events) a total of 28 (4.4%) clinical events were reported in the intervention group and 39 (5.6%) in the control group.
In the two complementary additional analyses, the first additional analysis excluded the heparin study which did not change the prior heterogeneity results. The second additional analysis excluded the smallest trial which had the largest effect and this resulted in a finding of no heterogeneity between the studies.