Abstract
OBJECTIVE: This study aimed to investigate the association between the serum albumin-to-creatinine ratio (ACR) and the prognosis of sepsis. METHODS: Extracted clinical data of sepsis patients from the MIMIC-IV (v2.2) database. Based on the optimal ACR cutoff value, we divided the cohort into two groups and preformed propensity score matching to balance baseline characteristics. Explored the relationship between ACR and 28-day all-cause mortality using Cox proportional risk regression and Kaplan-Meier survival curves, and conducted subgroup analysis to evaluate the effect modifications across different patient populations. Applied the restricted cubic spline (RCS) curves to assess the nonlinear relationships, and the receiver operating characteristic (ROC) curve to assess the predictive performance. RESULTS: After screening and matching, a total of 1,418 sepsis patients were included. Cox regression and Kaplan-Meier analysis showed that a high ACR value might be associated with a low 28-day mortality risk. Subgroup analysis revealed a significant interaction between age and ACR, as well as renal disease and ACR. RCS analysis revealed a nonlinear association between ACR values and reduced mortality risk. When ACR was below 2,300, there was a negative association between ACR and mortality. However, no significant association was observed when ACR exceeded 2,300. ROC curve analysis indicated that combining ACR with age, sex, body mass index, SOFA score, white blood cell, hemoglobin, blood lactate improved the predictive performance for 28-day all-cause mortality (AUC = 0.730). CONCLUSION: A higher ACR value may associated with a lower 28-day all-cause mortality risk when ACR value was less than 2,300. Moreover, ACR had some predictive power for adverse outcomes in sepsis.