Division of Cardiology (A.L., J.M.B., T.Y.W., S.V.), Duke University School of Medicine, Durham, NC.
Duke Clinical Research Institute, Durham, NC (A.L., J.M.B., T.Y.W., L.H.C., B.G.H., S.V.).
Circ Cardiovasc Interv. 2019 May;12(5):e007451. doi: 10.1161/CIRCINTERVENTIONS.118.007451.
Clinical event committees are commonly employed for event validation in clinical studies, but little is known about the comparative performance of administrative claims data versus clinician-triggered event adjudication for ascertainment of adverse events in structural heart disease studies.
Medicare claims were linked to 418 patients >65 years of age who underwent transcatheter mitral valve repair (MitraClip) for severe mitral regurgitation from 2007 to 2013 as part of the EVEREST II (Endovascular Valve Edge-to-Edge Repair Study II) High-Risk Registry or the REALISM (Real World Expanded Multicenter Study of the MitraClip System) Continued-Access Registry. Each registry adjudicated mortality, heart failure hospitalization, renal failure, ventilation, and bleeding/transfusion within 1 year. Concordance of claims-based outcomes with events was assessed in 3 ways: 1-year occurrence, cumulative incidence, and synchrony of first events. For event occurrence, positive predictive value (PPV) of claims versus adjudication was the highest for mortality (PPV=97%) and heart failure hospitalization (PPV=69%) but lower for bleeding (PPV=40%) and renal failure (PPV=19%). Whereas claims-based cumulative incidence for mortality, heart failure hospitalization, and renal failure were consistent with clinician-triggered adjudication, incidence curves for bleeding events and ventilation diverged, with claims identifying a greater number of events. When events were detected by both methods, however, over 75% of event dates matched exactly. Mitral valve reinterventions were identified through claims with perfect sensitivity and specificity relative to physician adjudication.
Ascertainment of mortality, heart failure hospitalization, and renal failure was highly concordant between physician adjudication and administrative claims. Further work is necessary to determine the role of administrative claims in event ascertainment in both prospective and retrospective studies of structural heart disease.
临床事件委员会常用于临床研究中的事件验证，但对于在结构性心脏病研究中确定不良事件时，行政索赔数据与临床医生触发的事件裁决相比的相对性能知之甚少。
将医疗保险索赔与 2007 年至 2013 年间因严重二尖瓣反流而行经导管二尖瓣修复术（MitraClip）的 418 名年龄>65 岁的患者相关联，这些患者是 EVEREST II（血管内瓣缘对缘修复研究 II）高危登记处或 REALISM（MitraClip 系统真实世界扩展多中心研究）持续访问登记处的一部分。每个登记处裁决了 1 年内的死亡率、心力衰竭住院、肾衰竭、通气和出血/输血事件。通过 3 种方式评估了基于索赔的结局与事件的一致性：1 年发生率、累积发生率和首次事件的同步性。对于事件发生，索赔与裁决的阳性预测值（PPV）对于死亡率（PPV=97%）和心力衰竭住院（PPV=69%）最高，但对于出血（PPV=40%）和肾衰竭（PPV=19%）较低。虽然基于索赔的死亡率、心力衰竭住院和肾衰竭的累积发生率与临床医生触发的裁决一致，但出血事件和通气的发生率曲线存在差异，索赔确定了更多的事件。然而，当两种方法都检测到事件时，超过 75%的事件日期完全匹配。通过索赔确定二尖瓣再干预，与医生裁决相比具有完美的敏感性和特异性。
医生裁决和行政索赔之间在死亡率、心力衰竭住院和肾衰竭的确定方面高度一致。需要进一步的工作来确定行政索赔在结构性心脏病前瞻性和回顾性研究中确定事件的作用。
