ObjectiveTo systematically review the non-inferiority trials in the cardiovascular domain that utilize medical devices as interventions, and investigate its characteristics and threshold settings. MethodsThe PubMed, Embase, and CENTRAL databases were electronically searched to collect non-inferiority trials in the cardiovascular field involving medical devices from inception to July 26, 2023. Two reviewers independently screened literature and extracted data. The reported information included basic characteristics, features of non-inferiority trials, and threshold-setting features of the included studies. Data analysis was performed using Excel 2020 and R 4.2.1 software. ResultsA total of 214 studies were included, with 167 studies (78.0%) focusing on interventions related to coronary artery stents. The trials predominantly utilized a two-arm design (92.9%), with a prevalent use of non-inferiority absolute thresholds (96.7%) as the criteria for non-inferiority determination. In 150 studies (70.1%), non-inferiority thresholds were established based on estimated control group effect values, while 33 studies (15.4%) did not report the source of these values. The non-inferiority trial endpoint outcomes exhibited diversity, and there were substantial differences in threshold settings. The three most studied qualitative indicators were target lesion failure rates (2.1%-8.6%), target vessel failure rates (2.5%-19.6%), and major adverse cardiovascular events rates (2.1%-10.0%). Late lumen loss (0.1-0.4 mm) emerged as the most frequently studied quantitative indicator. After converting absolute non-inferiority thresholds for all indicators into relative thresholds, the range was 1.20-3.67. ConclusionSignificant variations in non-inferiority threshold settings are observed for identical endpoint outcomes across included studies, highlighting a lack of reporting on the rationale behind threshold settings.