The Cumulative Perioperative Model: Predicting 30-Day Mortality in Abdominal Surgery Cancer Patients.

Area of Science:

• Oncology
• Surgical Outcomes
• Health Informatics

Background:

• Accurate prediction of mortality risk is crucial for optimizing patient care in major abdominal surgery.
• Existing models may not fully capture the dynamic nature of patient recovery post-surgery.

Purpose of the Study:

• To develop a cumulative perioperative model (CPM) for predicting 30 and 90-day mortality in abdominal surgery cancer patients.
• To evaluate the predictive performance of the CPM against ten established mortality prediction models.

Main Methods:

• A multivariate logistic regression model was constructed using data from 13,877 major abdominal surgical cases.
• The model incorporated variables such as race, admission source, Charlson Comorbidity Index, and surgical factors.
• Predictive ability was assessed using the C-statistic as model features were cumulatively added over time.

Main Results:

• The CPM demonstrated strong predictive performance, with C-statistics reaching 0.87 for 30-day mortality and 0.84 for 90-day mortality within six postoperative days.
• Initial C-statistics were 0.70 (30-day) and 0.71 (90-day), improving significantly with cumulative data.

Conclusions:

• The developed CPM achieves high discrimination for predicting mortality risk in abdominal surgery cancer patients.
• The CPM provides a continuous, cumulative assessment of mortality risk, serving as a valuable decision support tool.
• Implementation of the CPM may lead to improved patient outcomes, reduced healthcare costs, and more informed clinical decisions.