Abstract
Meta-analysis may increase the risk of random errors. Trial sequential analysis (TSA) has been developed to adjust for these random errors. We conducted TSA on the efficacy and safety of direct oral anticoagulants (DOACs) and vitamin K antagonists (VKAs) in left ventricular thrombus (LVT) patients in order to estimate how many additional patients should be required to draw definite conclusions. PubMed, Scopus, and Cochrane Library databases were searched for articles directly comparing DOACs and VKAs for LVT in LV thrombus resolution, stroke, any thromboembolism, major bleeding, any bleeding, and all-cause death. TSA was conducted with a cumulative Z-curve, monitoring boundaries, and required sample size. A simulated trial was run and TSA estimated the sample sizes of trials needed to draw definite conclusions. Of 4749 articles, 25 studies were used for the analysis. TSA revealed the current sample size already demonstrated superiority of DOACs in LV thrombus resolution and stroke, and futility in any thromboembolism and all-cause death. Two other outcomes did not achieve the required sample size. The sample size of new trials needed to demonstrate the superiority of DOACs over VKAs was estimated 400 for any bleeding. Corresponding trials needed to demonstrate no significant differences could be estimated for major bleeding and any bleeding (n = 200 and n = 2000, respectively). Current results show that the sample size required to draw definite conclusions was not reached for two outcomes, and there was a risk of random error. Further randomized controlled trials with sample sizes estimated by TSA will work effectively to obtain valid conclusions.