Association between preoperative neutrophil percentage-to-albumin ratio and in-hospital mortality in patients undergoing cardiac surgery: a retrospective cohort study

Abstract Background Systemic inflammation and nutritional compromise both affect postoperative outcomes. The neutrophil percentage-to-albumin ratio (NPAR) captures both in one inexpensive test, but its association with in-hospital mortality after cardiac surgery remains unexamined in an East Asian perioperative registry. Methods We performed a retrospective cohort study using INSPIRE version 1.4, a public perioperative registry from a single South Korean academic medical center (2011–2020). Adults undergoing cardiac surgery with neutrophil percentage and albumin measured in the first blood sample drawn within 24 h of admission, timestamp-verified to precede surgery, were included. NPAR was analyzed continuously and in quartiles. The primary outcome was in-hospital mortality (time zero: end of surgery; discharge alive censored). Cox regression with three sequential models provided the main estimates; restricted cubic splines, subgroup analyses, model-adjusted receiver operating characteristic (ROC) analyses with DeLong comparisons, and an E-value analysis supplemented them. Sensitivity analyses used alternative NPAR timepoints, complete cases, added adjustment for surgery type, operative and cardiopulmonary bypass times, and 20 imputed datasets, among others. Results Among 3,145 patients, 113 (3.6%) died in hospital. After full adjustment, each one-unit NPAR increase was associated with a 6% higher hazard of in-hospital death (hazard ratio [HR] 1.06, 95% CI 1.02–1.10), and the highest quartile carried higher risk than the lowest (HR 2.33, 1.11–4.87; P for trend = 0.027). The association was approximately linear (P for non-linearity = 0.855), more pronounced in older patients (P for interaction = 0.001), and consistent across surgery types. NPAR discriminated in-hospital death better than neutrophil percentage alone (AUC 0.740 vs. 0.723; DeLong P = 0.044) and similarly to albumin (0.729). All sensitivity analyses supported these findings, with an E-value of 4.09 indicating robustness to unmeasured confounding. Conclusions In this single-center East Asian cohort, higher preoperative NPAR, an inexpensive inflammation–nutrition marker, was independently associated with in-hospital mortality, particularly in older patients. Given the exploratory nature of data-derived cutoffs and the modest incremental discrimination, these findings position NPAR as a simple adjunct meriting external, prospective validation for perioperative risk assessment rather than an established clinical threshold.