Non-linear dose-response relationship between serum albumin and acute kidney injury in sepsis patients: a cohort study

Abstract Background The role of serum albumin on acute kidney injury (AKI) remains controversial. Additionally, research on this relationship in sepsis patients is sparse. Therefore, this research aimed to investigate the relationship between serum albumin level and probability of AKI in patients with sepsis. Methods This study was a retrospective cohort analysis of sepsis patients across the United States between 2014 and 2015 in the eICU Collaborative Research Database (eICU-CRD). To estimate the role of albumin on AKI by univariate, multivariate logistic regression and smooth curve fitting analysis. Results Among 5459 patients with sepsis, 32.94% (1798 patients) developed AKI. The results of the multivariate logistic regression analysis indicated that the albumin and AKI were negatively correlated (adjusted OR = 0.87, 95% CI 0.77–0.98, P = 0.0199). Moreover, a nonlinear relationship was observed between albumin level and probability of AKI with a inflection point at 2.1 g/dL. For albumin level < 2.1g/dL, each unit increase in serum albumin reduced the probability of AKI by 39% (adjusted OR = 0.61; 95% CI 0.44–0.85; P = 0.0032). However, for albumin levels above 2.1 g/dL, there was no significant association with the probability of AKI (adjusted OR = 0.99, 95% CI 0.83–1.16; P = 0.8691). Conclusion Serum albumin level below 2.1g/dL was defined as a risk factor for AKI in sepsis patients.