Reeves Mathew J, Fonarow Gregg C, Xu Haolin, Matsouaka Roland A, Xian Ying, Saver Jeffrey, Schwamm Lee, Smith Eric E
From the Department of Epidemiology and Biostatistics, Michigan State University, East Lansing (M.J.R.); Division of Cardiology (G.C.F.), and Department of Neurology (J.S.), Geffen School of Medicine, University of California, Los Angeles; Duke Clinical Research Institute, Durham, NC (H.X., R.A.M., Y.X.); Department of Biostatistics and Bioinformatics, Duke University, Durham, NC (R.A.M.); Department of Neurology, Duke University Medical Center, Durham, NC (Y.X.); Massachusetts General Hospital, Boston, MA (L.S.); and Hotchkiss Brain Institute, University of Calgary, Canada (E.E.S.).
Circ Cardiovasc Qual Outcomes. 2017 Oct;10(10). doi: 10.1161/CIRCOUTCOMES.117.003748.
Hospital profiling is typically undertaken using risk-standardized 30-day mortality, but obtaining these data for hospitals can be difficult. We sought to determine whether risk-standardized in-hospital mortality could serve as an adequate proxy for risk-standardized 30-day mortality data for the purposes of identifying outlier hospitals.
Acute ischemic stroke cases entered into GWTG (Get With The Guidelines)-Stroke between 2003 and 2013 were linked to fee-for-service Medicare files to obtain 30-day mortality. Risk-standardized mortality rates (RSMR) for in-hospital and 30-day mortality were generated using previously developed risk score models, and the proportion of hospitals classified as statistical outliers compared. We also assessed the impact of using the combined outcome of in-hospital mortality or discharge to hospice. A total of 535 332 ischemic stroke patients from 1494 GWTG-Stroke hospitals were included; mean age was 80 years, 59% female, and 19% nonwhite. At the hospital level, mean in-hospital RSMRs and 30-day RSMRs were 6.0% and 14.6%, respectively, but the correlation between the 2 was modest (=0.53). Overall agreement in the designation of outlier hospitals between in-hospital and 30-day RSMRs was 78%, but chance-corrected agreement was only fair (κ=0.29). However, when using the combined outcome of in-hospital mortality or discharge to hospice (risk-standardized mean =11.8%), the correlation with 30-day RSMR was much stronger (= 0.83) and outlier agreement improved substantially (κ=0.60).
When used to identify outlier hospitals with high or low mortality, the agreement between risk-standardized in-hospital mortality and 30-day mortality was modest. However, the combined outcome of in-hospital mortality or discharge to hospice showed much better agreement with 30-day mortality. This composite outcome could serve as a proxy for 30-day mortality when used to identify low- or high-performing hospitals.
医院概况分析通常采用风险标准化的30天死亡率，但获取医院的这些数据可能很困难。我们试图确定风险标准化的院内死亡率能否作为风险标准化的30天死亡率数据的适当替代指标，用于识别异常医院。
将2003年至2013年纳入“遵循指南-卒中”（GWTG-Stroke）的急性缺血性卒中病例与按服务收费的医疗保险档案相链接，以获取30天死亡率。使用先前开发的风险评分模型生成院内和30天死亡率的风险标准化死亡率（RSMR），并比较被归类为统计异常值的医院比例。我们还评估了使用院内死亡率或转至临终关怀机构这一综合结局的影响。共纳入了来自1494家GWTG-卒中医院的535332例缺血性卒中患者；平均年龄为80岁，59%为女性，19%为非白人。在医院层面，院内RSMR的平均值和30天RSMR的平均值分别为6.0%和14.6%，但两者之间的相关性一般（r=0.53）。院内和30天RSMR在异常值医院认定方面的总体一致性为78%，但校正机遇后的一致性仅为中等（κ=0.29）。然而，当使用院内死亡率或转至临终关怀机构这一综合结局时（风险标准化平均值=11.8%），与30天RSMR的相关性更强（r=0.83），异常值一致性显著提高（κ=0.60）。
在用于识别死亡率高或低的异常医院时，风险标准化的院内死亡率与30天死亡率之间的一致性一般。然而，院内死亡率或转至临终关怀机构这一综合结局与30天死亡率的一致性要好得多。当用于识别表现不佳或出色的医院时，这一复合结局可作为30天死亡率的替代指标。
