Abstract / Summary
The lactate-to-albumin ratio (LAR), integrating acute metabolic stress and chronic physiological reserve, has shown prognostic value in critical care, but evidence in surgical populations is limited. We investigated the association between preoperative LAR and postoperative all-cause mortality following non-cardiac surgery and evaluated its prognostic value. This retrospective cohort study included 12,606 adults undergoing non-cardiac surgery from the INSPIRE database. LAR was calculated as preoperative blood lactate (mmol/L) divided by serum albumin (g/dL), analyzed both continuously (natural log-transformed) and by quartiles. The primary outcome was 90-day all-cause mortality; secondary outcomes were 30-day and 180-day mortality. Multivariable Cox regression estimated adjusted hazard ratios (HRs), sequentially accounting for demographics, comorbidities, preoperative medications, laboratory findings, and surgical variables. Restricted cubic splines were used to assess non-linear associations. Incremental predictive value was evaluated using Harrell's C-statistics, net reclassification improvement (NRI), and integrated discrimination improvement (IDI), and bootstrap-based internal validation was also performed. The median preoperative LAR was 0.22 (IQR 0.16–0.34), and the overall 90-day mortality rate was 3.25%. Preoperative LAR showed a significant, non-linear dose–response association with postoperative mortality. After full adjustment, ln(LAR) was associated with higher 90-day mortality (HR, 1.09; 95% CI, 1.07–1.10; P < 0.001), and patients in the highest LAR quartile had a 3.82-fold higher risk compared with the lowest quartile (HR, 3.82; 95% CI, 2.56–5.71; P < 0.001). Consistent associations were observed for 30-day and 180-day mortality. Adding ln(LAR) modestly increased C-statistics but yielded significant improvements in NRI and IDI versus baseline models. Notably, ln(LAR) showed greater discriminative value than lactate and albumin entered as separate covariates in less-adjusted models, but this relative benefit was reduced when the model included the full set of clinical and intraoperative covariates. These findings remained stable on bootstrap-based internal validation. Among patients for whom preoperative lactate was clinically obtained, higher preoperative LAR was independently associated with increased mortality and modest improved risk reclassification beyond traditional predictors. Given the selective nature of preoperative lactate testing, these findings may not generalize to unselected, lower-acuity surgical populations, and external, multicenter validation is needed to confirm their broader applicability.