Scholarlink

Background <p>Accurate prediction of 30‑day mortality after transcatheter aortic valve replacement (TAVR) remains challenging. Existing models often overlook key preoperative laboratory variables. We aimed to develop and internally validate a preoperative risk prediction model using routinely available variables measured before the procedure.We state that predictors were preoperative‑only.</p> Methods <p>We retrospectively analyzed 1,673 consecutive patients who underwent TAVR at Fuwai Hospital between 2013 and 2023. Patients were randomly split into a training cohort (<i>n</i> = 1,171) and an internal hold out validation cohort (<i>n</i> = 502). Candidate predictors were prespecified as routinely available preoperative variables. All candidate predictors, including laboratory parameters (troponin, D‑dimer, HbA1c, uric acid, etc.), were collected during the preoperative evaluation, within 3 days prior to the TAVR procedure. Independent predictors of 30-day mortality were identified using multivariable logistic regression. Performance was assessed with receiver operating characteristic (ROC) analysis, calibration assessment (calibration plots and Hosmer Lemeshow test), and bootstrap optimism correction; a nomogram was constructed from the final model.</p> Results <p>30-day mortality was 3.4% in both the training (40/1,171) and validation (17/502) cohorts. The final preoperative model retained STS score, HbA1c, D-dimer, and uric acid as independent predictors. Discrimination was good (AUC=0.84 in the training cohort; AUC=0.78 in the internal validation cohort), with acceptable calibration (Hosmer-Lemeshow p = 0.211); optimism corrected performance closely matched the hold out out results. The resulting nomogram provides individualized risk estimates for 30-day mortality based on these preoperative predictors.</p> Conclusions <p>We developed and internally validated a parsimonious preoperative risk model and nomogram for 30-day mortality after TAVR using routinely available clinical and laboratory variables. This tool may support individualized risk stratification in elective/non salvage TAVR candidates. External validation particularly beyond East-Asian populations is essential before broader application.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Preoperative risk model and nomogram for 30‑Day mortality after TAVR: development and internal validation in a Chinese cohort

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月18日

查看全文
  • Haochao Li,
  • Chenyu Liu,
  • Pengfei Chen,
  • Shaoye Wang,
  • Xuanshu Li,
  • Xinjin Luo,
  • Yongquan Xie,
  • Xu Wang,
  • Liqing Wang

摘要

Background

Accurate prediction of 30‑day mortality after transcatheter aortic valve replacement (TAVR) remains challenging. Existing models often overlook key preoperative laboratory variables. We aimed to develop and internally validate a preoperative risk prediction model using routinely available variables measured before the procedure.We state that predictors were preoperative‑only.

Methods

We retrospectively analyzed 1,673 consecutive patients who underwent TAVR at Fuwai Hospital between 2013 and 2023. Patients were randomly split into a training cohort (n = 1,171) and an internal hold out validation cohort (n = 502). Candidate predictors were prespecified as routinely available preoperative variables. All candidate predictors, including laboratory parameters (troponin, D‑dimer, HbA1c, uric acid, etc.), were collected during the preoperative evaluation, within 3 days prior to the TAVR procedure. Independent predictors of 30-day mortality were identified using multivariable logistic regression. Performance was assessed with receiver operating characteristic (ROC) analysis, calibration assessment (calibration plots and Hosmer Lemeshow test), and bootstrap optimism correction; a nomogram was constructed from the final model.

Results

30-day mortality was 3.4% in both the training (40/1,171) and validation (17/502) cohorts. The final preoperative model retained STS score, HbA1c, D-dimer, and uric acid as independent predictors. Discrimination was good (AUC=0.84 in the training cohort; AUC=0.78 in the internal validation cohort), with acceptable calibration (Hosmer-Lemeshow p = 0.211); optimism corrected performance closely matched the hold out out results. The resulting nomogram provides individualized risk estimates for 30-day mortality based on these preoperative predictors.

Conclusions

We developed and internally validated a parsimonious preoperative risk model and nomogram for 30-day mortality after TAVR using routinely available clinical and laboratory variables. This tool may support individualized risk stratification in elective/non salvage TAVR candidates. External validation particularly beyond East-Asian populations is essential before broader application.