Individual patient meta-analysis of self-monitoring of an oral anticoagulation protocol

Background and aim of the study: Oral anticoagulation with vitamin K antagonists is effective for the prevention and treatment of thromboembolic events. Recent systematic reviews have shown that self-monitoring improved the quality of oral anticoagulation therapy (OAT), with patients spending more time in the therapeutic range than traditionally monitored patients, and with a concomitant decrease in the incidence of adverse effects. However, methodological and reporting heterogeneity has limited the strength of the reviews' conclusions. Differences were noted in terms of the assessment of outcome measures and the analysis methods used. For instance, not all used an intention-to-treat analysis, which may have over-inflated the results. Interpretation was limited by missing data: for example, it was not possible to combine mean tests in range, mean time in range, or to determine the level of deviant values. Time-to-event data (e.g., death, thromboembolic events) were reported as numbers of events, which prevented adequate analysis. In order to overcome these limitations and allow further investigation of the data, the study aim is to undertake an Individual Patient Data (IPD) meta-analysis. Methods and study design: The IPD analysis will include data from randomized trials that have compared self-monitoring (self-testing or self-management) OAT versus a control group, and that measured adverse events defined as major hemorrhage, thromboembolism, and death. The data to be requested for each trial will include: outcomes, demographic and psychosocial (e.g., quality of life) data. The primary outcomes of interest will be time to major hemorrhage, thromboembolism, and death. The secondary outcomes will be minor hemorrhage, percentage time within range, percentage tests within range, and patient satisfaction. The primary analysis will be by intention to treat, and multilevel models with patients and trials as the two levels, will be explored to investigate treatment effects on various outcomes. Patient-level covariates will be incorporated into the models in an attempt to account for statistical heterogeneity, as well as to investigate interactions with treatment effect. Conclusion: Predictive models should lead to the identification of those most likely to benefit from self-monitoring of oral anticoagulation, and potentially also to a targeted and a more cost-effective use of the intervention.