Three RCTs were included (n=521). All were small studies (sample sizes were 38, 200 and 283).
Studies varied in quality, with Jadad scores of 1, 3 and 5. Methodological limitations included: unequal duration of follow-up in treatment groups; use of inadequate randomisation method; treatment groups not comparable at baseline; use of non-equivalent cointerventions; and lack of patient blinding, allocation concealment and intention-to-treat analysis. Two studies used previously validated dosing algorithms and adjusted for covariates; one study developed a new algorithm.
There was no significant difference between genotype-guided warfarin therapy and control in major bleeding (RR 0.69, 95% CI 0.16 to 2.9). There was no significant statistical heterogeneity for major bleeding, but studies differed with respect to intervention and control treatments, duration of follow-up and study quality. There was no evidence of publication bias.
There was no significant difference between genotype-guided warfarin therapy and control in time within the therapeutic International Normalised Ratio range, but there was significant heterogeneity (p=0.03, I2=72.5%) and a summary effect size was not calculated.