Background: Hepatitis A virus (HAV) is the commonest cause of pediatric acute liver failure (PALF) in developing countries. Our objective was to develop and validate a HAV-etiology specific prognostic model in PALF.
Methods: All children with HAV induced PALF (IgM HAV reactive) were included. Outcome was defined at day 28. Only those with death or native liver survival were included. The model (Peds-HAV) was derived using the independent predictors of outcome and validated in a prospective independent cohort.
Results: Hepatitis A accounted for 131 (45.9%) of total 285 PALF. After excluding 11 children who underwent liver transplant, 120 children (74 survivors and 46 death) were included. The first 75 patients formed the derivation cohort and the next 45 patients formed the prospective validation cohort. In the derivation cohort, INR: OR 2.208, (95% CI 1.321-3.690), p = 0.003, grade of hepatic encephalopathy (HE): OR 3.078, (95% CI 1.017-9.312), p = 0.047 and jaundice-to-HE interval: OR 1.171, (95% CI 1.044-1.314), p = 0.007 were independent predictors of death. The final model comprised three criteria: (1) presence of grade 3-4 HE, (2) INR greater than 3.1, and (3) jaundice to HE interval more than 10 days. Presence of 2 or more of these criteria predicted death with 90% sensitivity, 81.4% specificity and 84.9% accuracy. Peds-HAV model was superior to existing prognostic models. In the validation cohort, Peds-HAV model predicted death with 83.3% sensitivity and 92.6% specificity.
Conclusion: Peds-HAV model is a simple, bedside, dynamic, etiology (HAV) specific prognostic model based on 3 objective parameters with optimum sensitivity and specificity, hence should be used as liver transplant listing criteria in HAV induced PALF.
Keywords: Acute liver failure; Etiology; Hepatic encephalopathy; Hepatitis A; International normalized ratio; Jaundice to encephalopathy interval; Liver transplant; Outcome; Pediatric; Prognosis.