Derivation and evaluation of baseline creatinine equations for hospitalized children and adolescents: the AKI baseline creatinine equation

Academic Article


  • Background: Acute kidney injury (AKI) definitions incorporate baseline creatinine (Crb) values, but Crb are frequently unknown in pediatrics. Our primary aim was to derive and validate a novel AKI Baseline Creatinine (ABC) estimation equation and compare it to existing methods of estimating Crb values. Methods: We conducted a single-center retrospective analysis of pediatric patients (0–25 years) admitted from 2012 to 2019. Included patients required at least one outpatient Crb prior to hospitalization (gold standard). Novel equations were developed with demographic and initial creatinine data. Existing methods included back-calculating Crb based on Schwartz, Full Age Spectrum (FAS), and CKiD-under-25 (U25) equations. To determine an optimal equation, we compared novel and existing equations to the gold standard. Results: The optimal simplified equation (ABC) included only age and had R2 = 59.9% and 73.2% of values within 30% of true Crb. The precision increased significantly when the equation included age and minimum creatinine within initial 72 h (ABC-cr): R2 = 75.4% and 86.5% of values within 30% of true Crb. The best performing existing equation was the age-based FAS, which had R2 = 61.0% and 78.0% of values within 30% of true Crb. All other existing equations performed worse, some methods as low as 52.6% within 30% of true Crb. Conclusions: The newly derived ABC equation is simple, and the ABC-cr equation can more accurately estimate Crb by ≥ 25% compared to previous methods. The potential applicability of these equations is vast, including faster recognition of AKI on initial patient contact and improved standardization of pediatric AKI definitions, enhancing health services research. Graphical abstract: A higher resolution version of the Graphical abstract is available as Supplementary information. [Figure not available: see fulltext.]
  • Authors

    Published In

    Digital Object Identifier (doi)

    Author List

  • Braun C; Rahman AKMF; Macomb E; Askenazi D; Bjornstad EC
  • Start Page

  • 3223
  • End Page

  • 3233
  • Volume

  • 37
  • Issue

  • 12