Abstract / Summary
Background Pancreaticoduodenectomy remains a complex operation with significant morbidity and mortality. Readmission remains an issue, contributing to health care costs, decreased quality of life and delay in additional therapies. Identification of actionable predictors may improve discharge planning and reduce preventable readmissions. We aim to develop and validate a clinically applicable risk score for 30-day readmission using national NSQIP data. Methods The NSQIP Targeted Pancreatectomy Participant Use Data File (2014–2023) was queried for adults undergoing pancreaticoduodenectomy. Patients were divided into training (n=34,556) and validation (n=8,144) cohorts. Univariable and multivariable logistic regression models were developed using preoperative variables alone and a combined perioperative model incorporating postoperative complications. Model discrimination was assessed using the area under the curve (AUC). Predictors from the final multivariable model were converted into an integer-based risk score. Results Among 34,556 patients, 6,006 (17.4%) experienced 30-day readmission. The median hospital length of stay for the overall cohort was 8 days. In the preoperative model, BMI ≥25, chronic obstructive lung disease (COPD), congestive heart failure (CHF), hypertension, bleeding disorders, absence of preoperative biliary stenting, absence of recent chemotherapy, and node-negative status were independently associated with readmission (all p<0.05); discrimination was limited (AUC 0.56). In the combined perioperative model, CHF lost significance, while postoperative complications emerged as dominant predictors; model performance improved substantially (AUC 0.68; validation AUC 0.67). Independent predictors included pancreatic fistula (OR 1.37, 95% CI 1.27–1.48), delayed gastric emptying (OR 1.70, 95% CI 1.58–1.83), and percutaneous drainage (OR 2.89, 95% CI 2.68–3.12). Using an optimal cutoff of 8.5 points, the risk score demonstrated 59% sensitivity and 72% specificity (p<0.001). Conclusions Thirty-day readmission following pancreaticoduodenectomy is driven predominantly by postoperative complications rather than baseline comorbidities. This validated risk score uses available perioperative variables to identify high-risk patients and may support risk-stratified discharge planning and early outpatient surveillance.