Hypoalbuminemia improves the ACS-NSQIP surgical risk calculator for gastrectomy

Background: The ACS-NSQIP Surgical Risk Calculator (SRC) is used to predict surgical outcomes, but its accuracy in gastrectomy has been questioned.^1,2 We investigated if adding hypoalbuminemia enhances its predictive ability in gastrectomy. Methods: We identified gastrectomy patients from the ACS-NSQIP database from 2005 to 2019. We constructed pairs of logistic regression models: one with the existing 21 preoperative risk factors from the SRC and another with the addition of hypoalbuminemia. We evaluated improvement using Likelihood Ratio Test (LRT), Brier scores, and c-statistics. Results: Of 18,070 gastrectomy patients, 34.5 ​% had hypoalbuminemia. Hypoalbuminemia patients had 2.34 higher odds of mortality and 1.79 higher odds of morbidity. Adding hypoalbuminemia to the RC model statistically improved predictions for mortality, cumulative morbidity, pulmonary, renal, and wound complications (LRT p ​< ​0.001). It did not improve predictions for cardiac complications (LRT p ​= ​0.11) Conclusion: Hypoalbuminemia should be considered as an additional variable to the ACS-NSQIP SRC for gastrectomy.