开发并验证免疫性血小板减少症患者颅内出血死亡预测模型:一项全国代表性多中心研究
基于中国27个中心2005至2020年数据,成人免疫性血小板减少症(ITP)患者颅内出血(ICH)死亡率为33.80%(48/142)。研究在107例训练队列中识别30天死亡危险因素并建立预后模型,内部验证ROC曲线下面积0.954(95%CI 0.910-0.998),外部验证(35例,11个其他中心)为0.942(95%CI 0.871-1.014),并建立在线应用用于预测30天死亡。
要点
- 成人ITP患者ICH死亡率为33.80%(48/142)。
- 30天死亡危险因素包括脑实质出血、ICH时血小板计数≤10×10⁹/L、合并严重感染、既往出血事件分级及入院格拉斯哥昏迷评分。
- 模型内部验证ROC曲线下面积0.954(95%CI 0.910-0.998),外部验证0.942(95%CI 0.871-1.014)。
核心结论
- 研究建立了成人ITP患者ICH后30天死亡预测模型,内部与外部验证均显示较好的区分度和校准度,并建立了在线应用供用户预测30天死亡。
研究设计与样本
- 全国多中心真实世界研究,纳入中国27个中心2005至2020年成人ITP合并ICH患者,训练队列107例,测试队列35例(来自11个其他中心)。
主要发现
- 成人ITP患者ICH死亡率为33.80%(48/142)。
- 30天死亡危险因素为脑实质出血、ICH时血小板计数≤10×10⁹/L、合并严重感染、既往出血事件分级及入院格拉斯哥昏迷评分。
- 内部验证ROC曲线下面积为0.954(95%CI 0.910-0.998),外部验证为0.942(95%CI 0.871-1.014)。
- 校准图显示估计风险与观察风险高度一致,决策曲线分析显示患者有相当的净获益。
关键数值与推荐
- 基于回归方程建立30天死亡预后模型,并通过bootstrap程序进行内部验证。
- 建立在线应用(47.94.162.105:8080/ich/)用于预测成人ITP患者发生ICH时的30天死亡。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
Intracranial hemorrhage (ICH) is a rare and life-threatening hemorrhagic event in patients with immune thrombocytopenia (ITP). However, its mortality and related risk factors remain unclear. Herein, we conducted a nationwide multicenter real-world study of ICH in adult ITP patients. According to data from 27 centers in China from 2005 to 2020, the mortality rate from ICH was 33.80% (48/142) in ITP adults. We identified risk factors by logistic univariate and multivariate logistic regression for 30-day mortality in a training cohort of 107 patients as follows: intraparenchymal hemorrhage (IPH), platelet count ≤10 × 109/L at ICH, a combination of serious infections, grade of preceding bleeding events, and Glasgow coma scale (GCS) level on admission. Accordingly, a prognostic model of 30-day mortality was developed based on the regression equation. Then, we evaluated the performance of the prognostic model through a bootstrap procedure for internal validation. Furthermore, an external validation with data from a test cohort with 35 patients from 11 other centers was conducted. The areas under the receiver operating characteristic (ROC) curves for the internal and external validation were 0.954 (95% confidence interval [CI], 0.910-0.998) and 0.942 (95% CI, 0.871-1.014), respectively. Both calibration plots illustrated a high degree of consistency in the estimated and observed risk. In addition, the decision curve analysis showed a considerable net benefit for patients. Thus, an application (47.94.162.105:8080/ich/) was established for users to predict 30-day mortality when ICH occurred in adult patients with ITP.
引用
- Chong S, Zhao P, Huang RB, et al. Developing and validating a mortality prediction model for ICH in ITP: a nationwide representative multicenter study. Blood Adv. 2022. PMID: 35679462. doi:10.1182/bloodadvances.2022007226
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。