Purpose: To investigate the predictive value of lactate for prognosis in critically ill patients with AUD.
Methods: A retrospective cohort study was performed using data extracted from a freely accessible critical care database (MIMIC-III). We studied all patients with AUD from the database for whom lactate was available. The clinical outcomes were 30-day mortality. Analyses included LOWESS curve fitting, logistic multivariate regression model, receiver operating characteristic (ROC) analysis and subgroup analysis.
Results: A total of 1296 eligible critically ill patients with AUD were included and there were 223 non-survivors (17.2%). The non-survivors had a higher lactate than the survivors (p < 0.001). A nonlinear relationship between lactate and 30-day mortality was observed. Multivariate logistic regression indicated lactate could be an independent risk factors to predict the prognosis of critically ill patients with AUD. According to ROC curve analysis, the area under the curve predicted by lactate for 30-day mortality was 0.672 (95% CI, 0.634 to 0.711). Subgroup analysis did not find obvious interaction in most subgroups.
Conclusion: High lactate was associated with increased mortality in critically ill patients with AUD.
Keywords: alcohol use disorder, critically ill patients, lactate, mortality