DefinePK

Abstract

Aim: Obtaining repeat bilirubin level after discontinuation of phototherapy is practiced in many neonatal units which is unnecessary. We compared the efficacy of two prediction rules in detecting rebound hyperbilirubinemia.
Methods: Subjects for this prospective cohort study were low risk neonates born ≥ 35 weeks gestation that received phototherapy and subjected for rebound bilirubin estimation. Infants with sepsis and hemolysis were excluded. We defined rebound hyperbilirubinemia as the return of total serum bilirubin (TSB) to phototherapy threshold within 72 hours of termination. The prediction scores by 2 variable model and 3 variable model were compared with the rebound bilirubin.
Results: Of the 467 neonates treated with phototherapy, 5.6% had rebound hyperbilirubinemia. Both models performed well with an area under the receiver operating characteristic curve of 0.974 (95% CI, 0.95–0.98) and 0.993 (95% CI, 0.98–0.99) for 2 variable and 3 variable models respectively. The sensitivity and specificity were 100% and 90.7% for 2 variable model and 100% and 98.2% for 3 variable model. Due to imbalanced data, precision-recall curve was put revealed 0.579 and 0.848 area under curve for 2 variable and 3 variable models respectively. Approximately 70% of infants had scores <20, which correspond to a <4% probability of rebound hyperbilirubinemia.
Conclusion: The risk of rebound hyperbilirubinemia can be quantified with 3 variable model. Clinical implementation of this score can avoid rebound bilirubin estimation