Understanding the epidemiology of bilirubin neurotoxicity is important for improving care. We were excited to read the study by Qattea et al., “Neonatal hyperbilirubinemia and bilirubin neurotoxicity in hospitalized neonates: analysis of the US Database.”1

This study evaluated bilirubin neurotoxicity based on International Classification of Diseases (ICD) codes for bilirubin neurotoxicity from the newborn period. One way to confirm the accuracy of an ICD diagnosis is to assess its association with other ICD codes that should accompany it. In the case of kernicterus, one would expect that if bilirubin levels are high enough to cause neurotoxicity, the infants would be treated with exchange transfusion. However, in this study, only 14% (205/1437) of infants with codes for bilirubin neurotoxicity received exchange transfusion (20% of the infants with neurotoxicity and isoimmunization and 13% of infants with neurotoxicity without isoimmunization). This indicates that the ICD codes may lack specificity as the vast majority of infants with “bilirubin neurotoxicity” did not get exchange transfusion.

Assuming that the codes were accurate for bilirubin neurotoxicity in newborns, this should not be conflated with kernicterus. Codes from the newborn period lack sensitivity because kernicterus is often not diagnosed until after 6 months of age.2 Many infants with acute bilirubin encephalopathy who receive timely treatment do not go on to develop kernicterus.3,4

Another important consideration when analyzing findings from large administrative dataset is the population and denominator under study. In this paper, the analysis is not clear as to the population studied. If the study focused on birth hospitalizations, then the study would miss readmissions or hospital transfers, both of which are common for bilirubin neurotoxicity. If the study included all hospitalizations for infants of specific ages, then the analysis would have to be careful not to include multiple hospitalizations per subject and would also need the method to define an appropriate population-level denominator. Using a denominator based on discharges instead of the population would be difficult to interpret.

The authors found an increased risk of bilirubin encephalopathy in Black infants, which they attributed to the increased prevalence of hemoglobinopathies. It should be highlighted that the drivers of bilirubin encephalopathy in Black neonates are more complex, including multiple biologic factors like G6PD deficiency or symptomatic ABO hemolytic disease and important non-biologic factors such as social determinants of health and structural racism.5 Disorders of globin synthesis are usually not clinically evident until after the first few months of life and are not a reported cause of kernicterus in Black or other neonates.

We hope that these comments are helpful for the authors of this study and for the many readers invested in improving long-term neurologic outcomes for patients with hyperbilirubinemia.