Comparison on the predictive value of four preoperative nutritional assessment tools for acute kidney injury after major surgery: a retrospective cohort study

BACKGROUND: Acute kidney injury (AKI) is a frequent yet severe complication following major surgery. Although malnutrition is an established risk factor for postoperative AKI, the comparative predictive value of different objective nutritional assessment tools remains poorly defined. MATERIALS AND METHODS: This retrospective cohort study consecutively included adult patients undergoing major surgery from two independent sources: a single academic hospital (derivation: 2018-2020; internal validation: 2021) and the MIMIC-IV database (external validation). Preoperative nutritional status was assessed using four objective tools, including the prognostic nutritional index (PNI), the geriatric nutritional risk index (GNRI), the controlling nutritional status (CONUT), and serum albumin. The primary outcome was postoperative AKI, defined as AKI occurring within 7 days after major surgery. RESULTS: The derivation, internal validation and external validation cohorts comprised 16 761 (1.9% AKI), 6243 (1.5% AKI), and 422 (36.3%AKI) patients, respectively. Malnutrition prevalence varied substantially by tool: PNI (2.0-19.9%), GNRI (18.5-28.1%), CONUT (49.3-69.4%), and serum albumin (4.9-28.4%). In the derivation cohort, all four nutritional assessment tools as continuous variables were independently associated with postoperative AKI after fully adjustment; when analyzed as categorial variables, only serum albumin showed independent associations for both mild and moderate-to-severe malnutrition. Serum albumin also demonstrated superior predictive performance (AUC: 0.697) and the highest incremental value for postoperative AKI, with this superiority maintained across both internal and external validation cohorts. Notably, serum albumin showed more pronounced inverse association with postoperative AKI in diabetics (aOR: 0.888, 95% CI: 0.832-0.948) than non-diabetics (aOR: 0.962, 95% CI: 0.923-1.003; P for interaction = 0.026). CONCLUSION: Malnutrition prevalence in surgical patients varies across nutritional assessment tools. Serum albumin demonstrated superior and consistent predictive value for postoperative AKI compared to nutritional indices (PNI, GNRI, and CONUT), with particularly enhanced performance in diabetic patients.