Evaluation of the accuracy of estimated baseline serum creatinine for acute kidney injury diagnosis.

Abstract

Modern epidemiologic studies of acute kidney injury (AKI) have been facilitated by the increasing availability of electronic medical records. However, pre-morbid reference serum creatinine (SCr) data are often unavailable in such records. Investigators substitute estimated baseline SCr with the eGFR 75 approach, instead of using actually measured baseline SCr. Here, we evaluated the accuracy of estimated baseline SCr for AKI diagnosis in the Japanese population. Inpatients and outpatients aged 18-80years were retrospectively enrolled. AKI was diagnosed according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria, using SCr levels. The non-AKI and AKI groups were selected using the following criteria: increase 1.5 times greater than baseline SCr ("baseline SCr") or increase 0.3mg/dL greater than baseline SCr in 48h ("increase in 48h"). AKI accuracy defined by the estimated reference SCr, the average SCr value of the non-AKI population (eb-GFR-A approach), or the back-calculated SCr from fixed eGFR=75mL/min/1.73m2 (eGFR 75 approach, or, eb-GFR-B approach in this study), was evaluated. We analyzed data from 131,358 Japanese patients. The number of patients with reference baseline SCr in the non-AKI and AKI patients were 29,834 and 8952, respectively. For AKI patients diagnosed using "baseline SCr", the AKI diagnostic accuracy rates as defined by eb-GFR-A and eb-GFR-B were 63.5 and 57.7%, respectively, while in AKI diagnosed using "increase in 48h", the AKI diagnostic accuracy rates as defined by eb-GFR-A and eb-GFR-B were 78.7 and 75.1%, respectively. In non-AKI patients, false-positive rates of AKI misdiagnosed via eb-GFR-A and eb-GFR-B were 7.4 and 6.8%, respectively. AKI diagnosis using the average SCr value of the general population may yield more accurate results than diagnosis using the eGFR 75 approach when the reference SCr is unavailable.