A novel nomogram model to predict in-hospital mortality in patients with acute type A aortic dissection after surgery

Yifei Zhou, Rui Fan, Hongwei Jiang, Renjie Liu, Fuhua Huang, Xin ChenJournal of Cardiothoracic Surgery 2024 · DOI

TL;DR

Researchers have developed a new nomogram—a graphical predictive tool—to estimate the risk of in-hospital death for patients undergoing surgery for acute type A aortic dissection (ATAAD). By integrating preoperative coagulation markers (platelets and APTT) with clinical factors like age and surgical duration, the model achieved high predictive accuracy (AUC = 0.844), offering clinicians a data-driven way to identify high-risk patients.

What Problem Does This Paper Address?

Acute type A aortic dissection (ATAAD) is a cardiovascular catastrophe where the inner layer of the aorta tears, requiring immediate surgery. Even with modern surgical techniques, postoperative mortality remains stubbornly high, often cited around 20%.

The authors argue that existing mortality prediction models are outdated because they don't reflect recent surgical advances or the critical role of coagulation dysfunction. Bleeding is a leading cause of death in these patients, yet many previous models ignored preoperative blood clotting indicators. This paper seeks to fill that gap by identifying contemporary risk factors and building a user-friendly tool for bedside risk stratification.

Methods

The study followed a retrospective cohort of 341 patients who underwent "Sun’s procedure" (a complex aortic repair) between 2019 and 2023.

  1. Statistical Screening: The team used univariate and multivariate logistic regression to filter through dozens of variables—ranging from demographics to intraoperative data—to find "independent predictors" of death.
  2. Nomogram Construction: They built a nomogram based on six specific variables. In a nomogram, each variable is assigned points; the sum of these points corresponds to a specific probability of the outcome (in this case, death).
  3. Validation: The model’s performance was evaluated using:
  4. C-index/AUC: To measure how well the model discriminates between survivors and non-survivors.
  5. Calibration Curves: To check if the predicted probabilities match actual observed outcomes.
  6. Decision Curve Analysis (DCA): To determine the "clinical net benefit"—essentially asking if using this model leads to better clinical decisions than a "treat everyone" or "treat no one" approach.

Datasets

The study utilized a localized clinical dataset from Nanjing First Hospital.

  • Initial Enrollment: 445 patients with ATAAD.
  • Final Cohort: 341 patients (after excluding those with chronic dissections, previous heart surgeries, or incomplete data).
  • Outcome: 45 patients (13.2%) died during hospitalization.

The researchers used multiple imputation techniques to handle missing values, ensuring that the statistical power of the dataset wasn't compromised by minor gaps in medical records.

Key Insights and Ideas

  • The Coagulation Connection: The most significant "aha moment" is the confirmation that preoperative Platelet count and APTT (Activated Partial Thromboplastin Time) are independent predictors of mortality. ATAAD triggers a massive inflammatory response and consumes clotting factors; the model proves that checking the "fuel gauge" of the clotting system before surgery is vital.
  • Malperfusion Matters: Preoperative lower limb ischemia (loss of blood flow to the legs) was a major red flag. The authors suggest this isn't just about the legs; it indicates a systemic release of toxins from ischemic tissue that eventually leads to multi-organ failure.
  • The "Double Whammy" of AKI: Postoperative Acute Kidney Injury (AKI) had the highest Odds Ratio (4.62), meaning patients who developed AKI after surgery were over four times more likely to die.
  • Surgical Complexity: Cardiopulmonary bypass (CPB) time served as a proxy for surgical difficulty. Longer times correlate with higher systemic inflammation and more severe lung/kidney injury.

Key Results

The final model identified six independent risk factors: age, preoperative lower limb ischemia, APTT, platelet count, CPB time, and postoperative AKI.

  • Predictive Power: The model achieved an AUC of 0.844, which is considered very good for clinical prediction.
  • Stability: A sensitivity analysis (excluding all cases with any missing data) yielded a similar AUC of 0.852, suggesting the model is robust.
  • Clinical Utility: The Decision Curve Analysis (DCA) showed that the nomogram provides a higher net benefit across a wide range of threshold probabilities compared to standard clinical intuition.

Limitations and Future Work

The authors acknowledge several hurdles before this can be widely adopted:

  • Single-Center Bias: The data comes from one hospital in China. Surgical outcomes and patient demographics can vary significantly across different regions and healthcare systems.
  • Retrospective Nature: Because the data was collected after the fact, there is an inherent risk of selection bias.
  • Internal Validation Only: While the model performed well on the data it was built from (using bootstrapping), it has not yet been tested on an "external" dataset from a different hospital.
  • Future Directions: The authors suggest that future studies should look at the extent of the dissection (how far it spreads) to see if that correlates with the consumption of clotting factors. External, multi-center validation is the necessary next step.

Summary updated

Abstract

The new nomogram model has a good ability to predict the in-hospital mortality of patients with acute type A aortic dissection after surgery.
BibTeX
@article{zhou2024novel,
  author = {Yifei Zhou and Rui Fan and Hongwei Jiang and Renjie Liu and Fuhua Huang and Xin Chen},
  title = {{A novel nomogram model to predict in-hospital mortality in patients with acute type A aortic dissection after surgery}},
  journal = {Journal of Cardiothoracic Surgery},
  year = {2024},
  doi = {10.1186/s13019-024-02921-6},
}
Figure