A Model for Predicting Significant Hyperbilirubinemia in Neonates From China.

OBJECTIVES To develop and validate a predischarge risk stratification model by using transcutaneous bilirubin (TcB) values and clinical factors to predict significant postdischarge hyperbilirubinemia in healthy term and late preterm Chinese neonates. METHODS In a prospective cohort study, 8215 healthy term and late preterm neonates in 8 hospitals in China underwent TcB measurement at <168 hours of age. TcB percentiles were calculated and used to develop an hour-specific nomogram, and 9 empirically weighted items were used to derive a prediction model. A risk stratification model was developed by combining the TcB nomogram with clinical risk scores to predict significant hyperbilirubinemia, defined as a postdischarge bilirubin level that exceeded the hour-specific recommended threshold value for phototherapy. Data from another 13,157 neonates were used to validate the model. RESULTS A TcB nomogram for every 12 hours of the studied interval was constructed from the development set. Gestational age, male gender, history of previous neonate who received phototherapy, bruising, feeding mode, weight loss, and early discharge were predictors of postdischarge significant hyperbilirubinemia. The combination of the TcB nomogram and clinical risk score provided the best prediction of significant hyperbilirubinemia with an area under the curve of 0.95 (95% confidence interval: 0.94-0.95) in the development data set and 0.94 (95% confidence interval: 0.93-0.94) in the validation data set. A risk stratification model with 6 distinct risk levels was developed and validated. CONCLUSIONS A risk classification model, combining discharge transcutaneous bilirubin values and clinical risk factors, separated term and late preterm Chinese neonates into 6 risk classes for the timely follow-up of postdischarge hyperbilirubinemia detection.