Evaluation of height-dependent and height-independent methods of estimating baseline serum creatinine in critically ill children

Baseline serum creatinine (bSCr) is required for diagnosing acute kidney injury (AKI). In children, bSCr is commonly defined as the lowest measurement within 3 months of admission. Measured values are often missing and estimating bSCr using height-based glomerular filtration rate (GFR) equations is problematic when height is unavailable. This is a retrospective cohort study including 538 children admitted to the intensive care unit (ICU) between 2003 and 2005 at two centers in Canada, with measured bSCr, height, and ICU-SCr values. We evaluated the bias, accuracy, and precision of back-calculating bSCr from height-dependent and height-independent GFR equations. Agreement of AKI defined using measured and estimated bSCr was calculated. Multivariate analyses were performed to assess the impact of bSCr estimation methods on the association between AKI and ICU mortality, length of stay, and duration of mechanical ventilation. Both methods underestimated bSCr by 1-3%, showed good accuracy (similar to 30% of patients with estimated bSCr within 10% of measured bSCr), but poor precision (wide 95% limits of agreement). The agreement between AKI defined by estimated versus measured bSCr was > 80% (kappa > 0.5). The height-independent method performed best in children > 13 years old; however, overall, both methods performed similarly across age subgroups. AKI was associated with longer stay, prolonged mechanical ventilation, and ICU mortality using measured and estimated bSCr. Height-dependent and height-independent bSCr estimation methods were comparable. This may have significant implications for performing pediatric AKI research using large databases, and in clinical care to define AKI when height is unknown.