Department of Medical Sciences, Cardiology, Uppsala University, Uppsala, Sweden; Uppsala Clinical Research Center, Uppsala University, Uppsala, Sweden.
Green Lane Cardiovascular Service, Auckland City Hospital and University of Auckland, Auckland, New Zealand.
Am Heart J. 2019 Feb;208:65-73. doi: 10.1016/j.ahj.2018.10.010. Epub 2018 Nov 7.
Clinical Endpoint Classification (CEC) in clinical trials allows FOR standardized, systematic, blinded, and unbiased adjudication of investigator-reported events. We quantified the agreement rates in the STABILITY trial on 15,828 patients with stable coronary heart disease.
Investigators were instructed to report all potential events. Each reported event was reviewed independently by 2 reviewers according to prespecified processes and prespecified end point definitions. Concordance between reported and adjudicated cardiovascular (CV) events was evaluated, as well as event classification influence on final study results.
In total, CEC reviewed 7,096 events: 1,064 deaths (696 CV deaths), 958 myocardial infarctions (MI), 433 strokes, 182 transient ischemic attacks, 2,052 coronary revascularizations, 1,407 hospitalizations for unstable angina, and 967 hospitalizations for heart failure. In total, 71.8% events were confirmed by CEC. Concordance was high (>80%) for cause of death and nonfatal MI and lower for hospitalization for unstable angina (25%) and heart failure (50%). For the primary outcome (composite of CV death, MI, and stroke), investigators reported 2,086 events with 82.5% confirmed by CEC. The STABILITY trial treatment effect of darapladib versus placebo on the primary outcome was consistent using investigator-reported events (hazard ratio 0.96 [95% CI 0.87-1.06]) or adjudicated events (hazard ratio 0.94 [95% CI 0.85-1.03]).
The primary outcome results of the STABILITY trial were consistent whether using investigator-reported or CEC-adjudicated events. The proportion of investigator-reported events confirmed by CEC varied by type of event. These results should help improve event identification in clinical trials to optimize ascertainment and adjudication.
临床试验中的临床终点分类（CEC）可实现研究者报告事件的标准化、系统化、盲法和无偏倚判断。我们对稳定性试验的 15828 例稳定性冠心病患者进行了评估。
研究者被指示报告所有潜在事件。每个报告的事件均由 2 名审阅者根据预设的程序和预设的终点定义进行独立审阅。评估了报告和裁定的心血管（CV）事件之间的一致性，以及事件分类对最终研究结果的影响。
CEC 共审查了 7096 项事件：1064 例死亡（696 例 CV 死亡）、958 例心肌梗死（MI）、433 例卒、182 例短暂性脑缺血发作、2052 例冠状动脉血运重建术、1407 例不稳定型心绞痛住院和 967 例心力衰竭住院。CEC 共确认了 71.8%的事件。死因和非致命性 MI 的一致性较高（>80%），而不稳定型心绞痛（25%）和心力衰竭（50%）的一致性较低。对于主要结局（CV 死亡、MI 和卒的复合结局），研究者报告了 2086 项事件，其中 82.5%的事件被 CEC 确认。达拉普利与安慰剂治疗稳定性试验的主要结局的效果一致，无论使用研究者报告的事件（危险比 0.96 [95%CI 0.87-1.06]）还是裁定事件（危险比 0.94 [95%CI 0.85-1.03]）。
使用研究者报告或 CEC 裁定的事件，稳定性试验的主要结局结果一致。CEC 确认的研究者报告事件的比例因事件类型而异。这些结果应有助于提高临床试验中的事件识别能力，以优化确定和裁定。
