Transcutaneous bilirubin nomogram for the first 144 hours in Thai neonates

Objectives: To develop an hour-specific transcutaneous bilirubin (TcB) nomogram for Thai neonates and to compare the ability of this nomogram with that of Bhutani's total serum bilirubin (TSB) nomogram for prediction of significant hyperbilirubinemia requiring phototherapy. Methods: Healthy Thai neonates, gestational age >= 35-week-gestation and birth weight >= 2000 grams were enrolled. Neonates who could not attend the postdischarge follow-up at our center were excluded. TcB measurements were routinely performed at 6 am and 6 pm using JM103 transcutaneous bilirubinometer until the neonates were discharged or received phototherapy. TcB levels were also measured at least once during 24-72 hours after discharge and thereafter depending on the pediatricians' decision. The nomogram was developed from the TcB data during age 12-144 hours of neonates who did not require phototherapy. The TcB values that obtained predischarge or before receiving phototherapy of all neonates were used to determine the predictive ability of this nomogram and Bhutani's TSB nomogram. Results: A total of 1071 neonates were included. Two hundred forty-one neonates (22.5%) required phototherapy. The nomogram was constructed using 4834 hour-specific TcB values. It provided a good prediction with the area under curve (AUC) of 0.89. The 75th percentile tract revealed sensitivity and negative predictive value (NPV) of 87.1 and 95.4% while that of the 40th percentile tract were 97.9 and 98.5% respectively. When Bhutani's nomogram was used, the AUC was 0.84. The sensitivity and NPV of the 75th percentile tract were 56.4 and 88.2%, and for the 40th percentile tract were 97.1 and 98.0% respectively. Conclusion: The newly developed TcB nomogram revealed slightly better predictive ability than Bhutani's TSB nomogram for term and late preterm Thai neonates who were the population with high prevalence of significant hyperbilirubinemia. The 40th percentile curve of both nomograms should be considered as an appropriate cut-off level for prediction.