Predicting Colorectal Surgery Readmission Risk: a Surgery-Specific Predictive Model.

Area of Science:

• Colorectal Surgery
• Health Services Research
• Predictive Modeling

Background:

• Current readmission risk models often lack specificity for surgical populations.
• Administrative data alone may not capture all factors influencing surgical patient readmissions.

Purpose of the Study:

• To develop and validate a predictive model for 30-day readmission risk specifically for colorectal surgery patients.
• To assess the performance of a comprehensive, surgery-specific model against existing indices.

Main Methods:

• Retrospective cohort study of 1549 colorectal surgery patients (2017-2019).
• Development of a multivariable logistic regression model using administrative, clinical, laboratory, and socioeconomic status (SES) data.
• Comparison of the new model's predictive accuracy (C-statistic) against two established readmission risk indices.

Main Results:

• The 30-day readmission rate was 19.62%.
• The developed surgery-specific model achieved a C-statistic of 0.70, outperforming two international indices (C=0.58 and C=0.60).
• Comprehensive datasets including SES and clinical data improved model performance.

Conclusions:

• Surgery-specific predictive models incorporating diverse data sources enhance 30-day readmission risk prediction in colorectal surgery.
• Comprehensive data, including socioeconomic and clinical factors, is crucial for accurate readmission risk assessment in surgical patients.