Association of surgical risk prediction models with days alive out of hospital at 30 days in colorectal surgeries: a retrospective cohort study.

Abstract

Background: Surgical risk prediction models are widely used in perioperative settings to estimate postoperative risk, but their relationship with patient-centered outcomes remains unclear. Days alive out of hospital (DAOH) integrates postoperative complications, readmissions, and mortality into a single outcome. This study evaluated the associations between established surgical risk prediction models and DAOH following colorectal surgery.

Methods: This retrospective cohort study included adult patients undergoing colorectal surgery in 2023. Preoperative risk was assessed using the American College of Surgeons (ACS) Surgical Risk Calculator, POSSUM-morbidity, and P-POSSUM mortality scores. Patient characteristics, surgical variables, postoperative complications, readmissions, and mortality were recorded. DAOH at 30 days (DAOH30) and DAOH at 90 days (DAOH90) were calculated; low DAOH30 was defined as below the cohort median. The primary outcome was to evaluate the associations between risk estimates and DAOH30. Secondary outcome analyses included associations between risk estimates and DAOH90, multivariable logistic regression to identify factors associated with low DAOH30, and receiver operating characteristic analysis with area under the curve (AUC) to assess discriminative performance of risk models for low DAOH30.

Results: A total of 145 patients were analyzed, with a median DAOH30 of 23 (18-24) days. All risk models were negatively correlated with DAOH30 (ACS-any complication: rho = - 0.603; ACS-serious complication: rho = - 0.515; ACS-death: rho = - 0.447; ACS-readmission: rho = - 0.286; POSSUM-morbidity: rho = - 0.419; P-POSSUM-mortality: rho = - 0.396; all p < 0.001), with similar findings for DAOH90. In multivariable regression analysis, higher ACS-serious complication risk (OR: 1.177; 95% CI: 1.057-1.310; p = 0.003), higher ACS-any complication risk (OR: 1.172; 95% CI: 1.062-1.293; p = 0.002), and longer operative duration were independently associated with low DAOH30. ACS-serious and ACS-any complication estimates demonstrated good discriminative ability for low DAOH30 (AUC = 0.808 [95% CI: 0.735-0.880] and AUC = 0.828 [95% CI: 0.759-0.897], respectively).

Conclusions: Surgical risk prediction models were significantly associated with postoperative recovery measured by DAOH following colorectal surgery. The ACS Surgical Risk Calculator estimates for serious and any complication demonstrated good discriminative performance for reduced DAOH30, suggesting their potential utility in identifying patients at risk of impaired postoperative recovery.