Acute kidney injury is an independent risk factor for pediatric intensive care unit mortality, longer length of stay and prolonged mechanical ventilation in critically ill children: a two-center retrospective cohort study

INTRODUCTION In adults, small (< 50%) serum creatinine (SCr) increases predict mortality. It is unclear whether different baseline serum creatinine (bSCr) estimation methods affect findings of acute kidney injury (AKI)-outcome associations. We characterized pediatric AKI, evaluated the effect of bSCr estimation approaches on AKI-outcome associations and evaluated the use of small SCr increases to predict AKI development. METHODS We conducted a retrospective cohort database study of children (excluding postoperative cardiac or renal transplant patients) admitted to two pediatric intensive care units (PICUs) for at least one night in Montreal, QC, Canada. The AKI definition was based on the Acute Kidney Injury Network staging system, excluding the requirement of SCr increase within 48 hours, which was impossible to evaluate on the basis of our data set. We estimated bSCr two ways: (1) the lowest SCr level in the three months before admission or the average age- and gender-based norms (the standard method) or (2) by using average norms in all patients. Outcomes were PICU mortality and length of stay as well as required mechanical ventilation. We used multiple logistic regression analysis to evaluate AKI risk factors and the association between AKI and mortality. We used multiple linear regression analysis to evaluate the effect of AKI on other outcomes. We calculated diagnostic characteristics for early SCr increase (< 50%) to predict AKI development. RESULTS Of 2,106 admissions (mean age ± SD = 5.0 ± 5.5 years; 47% female), 377 patients (17.9%) developed AKI (using the standard bSCr method) during PICU admission. Higher Pediatric Risk of Mortality score, required mechanical ventilation, documented infection and having a bSCr measurement were independent predictors of AKI development. AKI was associated with increased mortality (adjusted odds ratio (OR) = 3.7, 95% confidence interval (95% CI) = 2.1 to 6.4, using the standard bSCr method; OR = 4.5, 95% CI = 2.6 to 7.9, using normative bSCr values in all patients). AKI was independently associated with longer PICU stay and required mechanical ventilation. In children with no admission AKI, the initial percentage SCr increase predicted AKI development (area under the curve = 0.67, 95% CI = 0.60 to 0.74). CONCLUSIONS AKI is associated with increased mortality and morbidity in critically ill children, regardless of the bSCr used. Paying attention to small early SCr increases may contribute to early AKI diagnosis in conjunction with other new AKI biomarkers.