This study followed the long-term developmental progress of children who were exposed to extreme neonatal hyperbilirubinaemia in Australia. The data is largely reassuring. The majority of infants for whom long-term follow-up results were available, did not suffer long-term neurodevelopmental sequelae. The estimated incidence of kernicterus ostensibly compares favourably with international estimates of 0.4 to 2.7 per 100 000 live births. 2–18 There are however significant methodological variations to consider, both within this study and its comparators. The majority of kernicterus incidence reports to date have been retrospective population-based record review studies of adverse neurodevelopment diagnoses following variably defined neonatal hyperbilirubinaemia. For example, Alkén and colleagues' recent Swedish large population cohort study of almost 1 million children estimated a kernicterus incidence of 1.3 per 100 000 based on medical record review up to 2 years of age of children who had TSB of ≥ 510µmol/L in the newborn period.2 Wu and colleagues also conducted an extensive medical record review study in a Californian population of 525,409 infants born between 1995 and 20113 with serum bilirubin levels at, or above the American Academy of Pediatrics (AAP) exchange transfusion threshold.25 A pediatric neurologist blinded to bilirubin levels, reviewed medical records for diagnostic evidence of CP secondary to kernicterus based on magnetic resonance image (MRI) findings and clinical evidence of dyskinesia. Ninety percent of the cohort had follow-up to 15 months of age. The incidence of CP incidence was 0.4% amongst ‘exposed’ infants compared to 0.1% in the ‘unexposed’. Three children in the ‘exposed’ group have cerebral palsy consistent with kernicterus. In contrast, Thomas Newman and colleagues' large, nested case-control study4 of Californian infants identified 140 cases from a population of 106 627 who had a peak TSB of ≥ 427µmol/L and compared their developmental outcomes with 419 controls. Follow-up data to 2 years of age was available for 94% of the cases and 89% of the controls. Formal evaluations were completed in 59% and 40% of cases and controls respectively to a mean age of 5.1 years of age ± 0.12 SD. The study did not find any cases of kernicterus in this cohort and there were no differences in neurological abnormalities, parental cognitive concern or reported behavioural problems. The Canadian Pediatric Surveillance Unit most recently reported a national kernicterus incidence of 2.7 per 100 000 based on a prospective study of paediatrician reports.10 Kernicterus was defined on the basis of neonatal bilirubin levels > 425µmol/L and either: two signs of adverse neurodevelopment consistent with kernicterus and/or enamel dysplasia of deciduous teeth; or abnormal MRI findings with bilateral lesions of the basal ganglia or midbrain. Ninety percent (18 out of 20) cases met criteria on the basis of MRI brain abnormalities. Outcome data for 14 of the 20 cases was gathered by way of a study-specific paediatrician-completed questionnaire at 12–18 months of age. Chronic bilirubin encephalopathy was evident in eleven. Three cases with abnormal neonatal MRIs had normal development and two cases with normal MRIs had abnormal neurodevelopment. The UK study11 estimated incidence of 0.9 per 100 000 live births was based on a definition of severe hyperbilirubinaemia as ≥ 510µmol/L and associated neonatal encephalopathy (impaired consciousness, hypotonia, opisthotonus and seizures). Five suspected cases of kernicterus were reported at 12 months of age by means of an unvalidated questionnaire completed by reporting clinicians. With regards to the risks associated with bilirubin encephalopathy, Ebbesen and colleagues’ population-based study of 502 766 infants utilising linked national data found bilirubin levels > 450µmol/L without associated intermediate or advanced ABE carried little risk of adverse neurodevelopmental sequelae.9 In this Australian study, of the children with known adverse outcomes, the most severely affected child presented with intermediate ABE in association with a bilirubin level of 630µmol/L secondary to haemolysis (case 1). With regards to the BSID-III findings of 25 cases in our study, all mean subscale results measured within the normal range of the standardised US population of this assessment tool. However, when considered in the local context, the performance of this cohort on BSID-III, might be considered lower than expected. Since its introduction in 2006, a tendency for the BSID-III to overestimate early childhood development has been documented by a number of authors.26,27 Peter Anderson and colleagues’ study26 of 202 healthy term Australian children at 2 years of age found all BSID-III mean scores were above the US normative means and thus rates of developmental delay much lower than expected. Chinta and colleagues27 also reported the 3 year BSID-III performance of 156 healthy term Australian infants. Of particular relevance to our study, there was no significant difference on the gross-motor scale performance amongst the Chinta cohort compared with the US norms.
One child (case 3) in this study has been diagnosed with global developmental delay and Autism Spectrum Disorder (ASD) in the context of a family history. ASD12,28 and other disorders of psychological and behavioural development, including attention-deficit disorder (ADHD)12 have been observed by some authors. Whilst further research is required to further explore this relationship, behavioural development of children affected by severe or extreme neonatal hyperbilirubinaemia should certainly be considered in their long-term surveillance and care.
Whilst the results of this prospective long-term outcomes study to 3 years of age using validated standardised assessment tools are reassuring, there are important methodological issues to consider. The voluntary reporting nature of enrolment, the indirect consent process and loss to follow-up led potentially to a significant underestimation of the incidence of impaired neurodevelopment. The study team were bound to the indirect consent process as the follow-up study was established as an adjunct to the original APSU surveillance study.19 In retrospect and in regards planning of future studies, enrolment in the follow-up study would have been offered at the time of the initial report. Similarly, nationwide follow-up centres and services would have been confirmed prospectively to ensure not only maximal cohort follow-up, but most importantly, all children and families received optimal early childhood surveillance and therapeutic intervention and support. Furthermore, minor neurological neurological dysfunction (MND)29 and abnormal general movements30 have been associated with neonatal hyperbilirubinaemia and we suggest future studies should not miss the opportunity to study General Movements in this patient population, in order to further proactively identify and treat early objective signs of abnormal neuromotor development.
The issues faced by the study team in setting up this ambitious study highlight complex barriers experienced by rare-disease researchers in establishing prospective national long-term outcomes studies. Long-term neurodevelopmental outcome data is key to determining the impacts of therapeutic interventions, not least in the field of perinatal research. In the context of an observational, non-interventional study such as this, the need to navigate complex local, state and national ethical approval processes seems unjustified. In fact, in order to progress our knowledge of long-term developmental outcomes in general, consideration of a central national ethics body in warranted.