预测危重症患者静脉血栓栓塞症风险(PROVE-IT):模型开发与验证研究
本研究开发并外部验证了危重症患者住院期间静脉血栓栓塞症(VTE)的预后模型PROVE-IT。模型在26218例患者中开发,在1983例患者中外部验证。外部验证校准良好,但区分度仅为可接受(C统计量0.629),决策曲线分析提示临床获益未达统计学显著,作者认为目前尚不足以推荐临床实施。
要点
- PROVE-IT模型在26218例危重症患者中开发,用于预测ICU入住后住院期间VTE。
- 外部验证队列1983例,C统计量0.629,区分度仅为可接受。
- 外部验证校准良好(ICI=0.00664),但决策曲线分析临床获益未达统计学显著。
核心结论
- 开发并外部验证了危重症患者住院期间VTE的预后模型PROVE-IT,外部验证校准良好但区分度仅为可接受,决策曲线分析提示临床效用不明确,目前尚不足以推荐临床实施。
研究设计与样本
- 预后模型开发与外部验证研究,模型开发使用多因素logistic回归,纳入14个已确定的危重症患者VTE预后因素。
- 模型在26218例患者中开发,并在一个独立队列的1983例患者中进行外部验证。
- 结局为ICU入住后住院期间VTE。
主要发现
- 内部验证C统计量为0.681,外部验证C统计量为0.629。
- 内部校准:ICI=0.00231,E50=0.00184,E90=0.00313,Emax=0.452。
- 外部校准:ICI=0.00664,E50=0.00447,E90=0.00131,Emax=0.154。
- 决策曲线分析提示模型在1.1%至4.7%的风险阈值范围内可能有临床获益,但未达统计学显著。
关键数值与推荐
- 外部验证区分度C统计量0.629,内部验证C统计量0.681。
- 外部校准ICI=0.00664,E50=0.00447,E90=0.00131,Emax=0.154。
- 决策曲线分析风险阈值范围1.1%-4.7%,临床获益未达统计学显著。
临床怎么用
- 作者结论指出,模型临床效用不明确,目前尚不足以推荐临床实施。
局限性
- 外部验证区分度仅为可接受(C统计量0.629)。
- 决策曲线分析提示的临床效用不明确,临床获益未达统计学显著。
- 作者指出目前尚不足以推荐临床实施。
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
BACKGROUND: Venous thromboembolism (VTE) is a serious complication of critical illness. The persistently high rates of VTE may be the result of thromboprophylaxis failure in critically ill patients at particularly high risk of VTE. A prognostic model may facilitate patient-specific interventions, ultimately lowering the incidence of VTE in critically ill patients. OBJECTIVES: We aimed to develop and externally validate a prognostic model for VTE in critically ill patients: Predicting the Risk of Venous Thromboembolism in Critically Ill Patients model. METHODS: The outcome was in-hospital VTE after intensive care unit admission. Model development was performed using multiple logistic regression with 14 established prognostic factors for VTE in critically ill patients. External validation was conducted in an independent cohort. Model performance was evaluated in terms of discrimination (C-statistic), calibration (integrated calibration index [ICI], E50, E90, Emax), and clinical utility (decision curve analysis [DCA]). RESULTS: The Predicting the Risk of Venous Thromboembolism in Critically Ill Patients (PROVE-IT) model was developed in 26 218 patients and externally validated in 1983 patients. Discrimination was acceptable with a C-statistic of 0.681 upon internal validation and 0.629 upon external validation. Calibration was excellent internally (ICI = 0.00231; E50 = 0.00184; E90 = 0.00313; Emax = 0.452) and good externally (ICI = 0.00664; E50 = 0.00447; E90 = 0.00131; Emax = 0.154). DCA suggested that the model may have clinical benefit over a range of risk thresholds (1.1%-4.7%), although not statistically significant. CONCLUSION: We developed and externally validated a prognostic model for in-hospital VTE in critically ill patients. Our model was well-calibrated upon external validation, although the discrimination was merely acceptable. Additionally, the model's clinical utility, assessed by DCA, is unclear, which precludes its recommendation for clinical implementation at this stage.
引用
- Rivrud SCS, Heijkoop ÈRH, Hulshof L, et al. Predicting the risk of venous thromboembolism in critically ill patients (PROVE-IT): a model development and validation study. J Thromb Haemost. 2025. PMID: 40617504. doi:10.1016/j.jtha.2025.06.026
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。