Ten randomised controlled trials (RCTs) were included in the review (n=2,724 patients). Two of the trials were rated as high-quality.
For a number of outcomes patient self-management yielded improved outcomes in comparison with conventional management.
Death, all causes: For all trials, the relative risk for death was 0.48 (95% confidence interval (CI): 0.29 to 0.79; p=0.004). There was no heterogeneity between trials.
Major complications: For all trials, the relative risk for major complications was 0.58 (95% CI: 0.42 to 0.81; p=0.001). For high quality studies only, the relative risk was 0.47 (95% CI: 0.26, 0.84; p=0.01). There was no heterogeneity between trials.
Minor complications: For all trials the relative risk for minor complications was 0.76 (95% CI: 0.63 to 0.91; p=0.003). Restricting analyses to high quality studies the relative risk was 0.41 (95% CI: 0.31 to 0.54; p<0.00001).
Time within therapeutic international normalised ratio target range: For all trials, the weighted mean difference for therapeutic international normalised ratio target range range was 4.36% (95% CI: 2.87 to 5.86; p<00001). There was significant heterogeneity between trials and application of a random-effects model still yielded a significant result, weighted mean difference 6.53% (95% CI: 2.24 to 10.82; p=0.003). When the analysis was restricted to high quality trials the difference between the interventions was not significant.
The funnel plot suggested the presence of publication bias for trials analysing minor complications and time within therapeutic international normalised ratio target range