Accurate Preoperative Prediction of Discharge Destination Using 8 Predictor Variables: A NSQIP Analysis

With inpatient length of stay decreasing, discharge destination after surgery can serve as an important metric for quality of care. Additionally, patients desire information on possible discharge destination. Adequate planning requires a multidisciplinary approach, can reduce healthcare costs and ensure patient needs are met. The Surgical Risk Preoperative Assessment System (SURPAS) is a parsimonious risk assessment tool using 8 predictor variables developed from the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) dataset. SURPAS is applicable to more than 3,000 operations in adults in 9 surgical specialties, predicts important adverse outcomes, and is incorporated into our electronic health record. We sought to determine whether SURPAS can accurately predict discharge destination.

A "full model" for risk of postoperative "discharge not to home" was developed from 28 nonlaboratory preoperative variables from ACS NSQIP 2012-2017 dataset using logistic regression. This was compared with the 8-variable SURPAS model using the C index as a measure of discrimination, the Hosmer-Lemeshow observed-to-expected plots testing calibration, and the Brier score, a combined metric of discrimination and calibration.

Of 5,303,519 patients, 447,153 (8.67%) experienced a discharge not to home. The SURPAS model's C index, 0.914, was 99.24% of that of the full model's (0.921); the Hosmer-Lemeshow plots indicated good calibration and the Brier score was 0.0537 and 0.0514 for the SUPAS and full model, respectively.

The 8-variable SURPAS model preoperatively predicts risk of postoperative discharge to a destination other than home as accurately as the 28 nonlaboratory variable ACS NSQIP full model. Therefore, discharge destination can be integrated into the existing SURPAS tool, providing accurate outcomes to guide decision-making and help prepare patients for their postoperative recovery.