JK, ME, DA, PS, PM MS, RvH, KM, IM,PS, SN, SH, UH, and RM participated in data acquisition. JK, ME, MS, JP, and RBperformed analysis and interpretation moreover of the data. ME, DA, PS, MS, IA, DA, SN, RvH, JP,UH, JP, JV, and RB performed critical revision of the manuscript for importantintellectual content. AK and JK completed the statistical analysis. All of the authorsread and approved the final draft of the manuscript.Supplementary MaterialAdditional file 1:Table S1. Mortality (percentage, 95% CI) and number of patients forindividual cohorts, nondiabetes, and diabetes patients, for each of thethree domains of glycemic control. This file contains data detailing thenumber of patients from each of the nine centers, their mortalitypercentage, and the 95% CI of this percentage, stratified by diabeticstatus, for each of the “bands” of the three domains of glycemic controldescribed in the manuscript.

Click here for file(125K, DOC)
Blood NGAL and serum creatinine (sCr) were determined at ED presentation (T0), and at: 6 (T6), 12 (T12), 24 (T24) and 72 (T72) hours after hospitalization. A preliminary assessment of AKI by the treating ED physician occurred in 218 out of 665 patients (33%), while RIFLE AKI by expert nephrologists was confirmed in 49 out of 665 patients (7%). The ED physician’s initial judgement lacked sensitivity and specificity, overpredicting the diagnosis of AKI in 27% of the cohort, while missing 20% of those with AKI as a final diagnosis.The area under the receiver operating characteristic curve (AUC), obtained at T0, for blood NGAL alone in the AKI group was 0.

80. When NGAL at T0 was added to the ED physician’s initial clinical judgment the AUC was increased to 0.90, significantly greater when compared to the AUC of the T0 estimated glomerular filtration rate (eGFR) obtained either by modification of diet in renal disease (MDRD) equation (0.78) or Cockroft-Gault formula (0.78) (P = 0.022 and AV-951 P = 0.020 respectively). The model obtained by combining NGAL with the ED physician’s initial clinical judgement compared to the model combining sCr with the ED physician’s initial clinical judgement, resulted in a net reclassification index of 32.4 percentage points. Serial assessment of T0 and T6 hours NGAL provided a high negative predictive value (NPV) (98%) in ruling out the diagnosis of AKI within 6 hours of patients’ ED arrival. NGAL (T0) showed the strongest predictive value for in-hospital patient’s mortality at a cutoff of 400 ng/ml.ConclusionsOur study demonstrated that assessment of a patient’s initial blood NGAL when admitted to hospital from the ED improved the initial clinical diagnosis of AKI and predicted in-hospital mortality.