Abstract: Objective To evaluated the prognostic value of prealbumin levels in patients with acute kidney injury (AKI)，which reason for causing AKI is acute tubular necrosis. Methods This was a prospective cohort study. Hospital- acquired AKI patients in the Ninth People’s Hospital from February 2012 to June 2014. laboratory measurements and clinical data were recorded. Cox proportional hazards models was uesd to estimate the risk of the ninety days mortality associated with serum prealbumin level at the beginning of the nephrology consultation. Results Three hundreds and forty-eight hospital-acquired AKI patients matched at least one of the Risk, Injury, Failure, Loss, End Stage (RIFLE) criteria for increased serum creatinine, and had requested a nephrology consultation. Clinicopathologic variables were compared between patients with a serum prealbumin level, ＜13.6 mg/dL (the exposed group; cutoff point, median) and patients with a serum prealbumin level≥13.6 mg/dL (the control group). In-hospital mortality rates associated with prealbumin levels were 48.3% for ＜13.6 mg/dL, and 21.4% for≥13.6mg/dL (χ2=13.622, P＜0.001). After adjusted analysis age, sex, hemoglobin, serum albumin, C-reactive protein, Liano score and serum total cholesterol, the presence of a serum prealbumin level, ＜13.6mg/dL was significantly associated with increased the ninety days mortality (HR 1.784，95% confidence interval 1.059～3.006, P = 0.029). In addition, each increment of 5 mg/dL of serum prealbumin was associated with an adjusted 24% decrease of the ninety days mortality (HR 0.754，95% confidence interval 0.606～0.938，P=0.011). Conclusions Low levels of serum prealbumin was an independent risk factor of death in hospital-acquired AKI patients

Key words: acute kidney injury, prealbumin, Prognosis