Acute Kidney Injury Diagnostic Accuracy and Implications of Different Baseline Creatinine Equations

Abstract:Introduction Acute kidney injury (AKI) definitions rely on known baseline creatinine (Crb), unavailable in up to 75% of hospitalized children. New equations (ABC methods) for estimating Crb were derived from children without kidney disease. We aim to externally validate ABC methods in an international cohort and assess how different Crb equations alter AKI epidemiology. Methods AWARE was a prospective international study of critically ill children (age 0-25 years) from 32 PICUs. A subset of AWARE (n=2451) with known Crb (gold standard) is used to validate ABC methods using statistical measures of precision (R2) and accuracy (within 10% or 30% of gold standard). The entire cohort (n=4984) is used to determine how different Crb estimating equations (3 ABC equations and 4 published eGFR equations imputing Crb) impact AKI incidence and its association with key clinical outcomes, including 28-day mortality, using univariate and multivariate analysis. Results The ABC-Age equation (requiring only age) demonstrated similar accuracy and precision compared to existing Crb equations. The ABC-Creatinine equation (includes age and hospital creatinine value) outperformed existing Crb equations by up to 15% in precision (ABC-Creatinine R2=0.51 versus FAS R2=0.36) and 32% in accuracy (ABC-Creatinine 66% versus Original Schwartz 34%). For the entire cohort, AKI incidence varied from 7-12% depending on Crb definition. ABC equations are associated with clinical outcomes similarly to existing Crb equations. Conclusion ABC-Creatinine equation (with minimal variables) outperformed existing Crb equations for accuracy and precision as the optimal method for Crb estimation. Crb definition variability alters AKI incidence and epidemiology, necessitating standardization.