We read with interest the recent report on this subject(1). There are a few comments that we would like to make.

Firstly, the authors have not mentioned whether their center caters to inborn babies, or babies born elsewhere. From their statement in the methods "time of onset of jaundice (according to the observation of the family)" it appears that these were all outborn babies, because jaundice in inborn babies should have been detected earlier by the staff. Outborn babies are qualitatively different from inborn babies, because details of immediate post-natal events are often difficult to ascertain precisely, and it is also impossible to state whether or not the peak serum bilirubin level had already been crossed before the patient reached the destination hospital. This would make the peak bilirubin in hospital fallacious.

Secondly, the authors seem to have confused the terms "incompatibility" and "isoimmunization". A patient must have evi-dence of isoimmunization and/or hemolysis, before one can attribute the incompatibility as being causally related to the jaundice. How-ever, when the authors describe the etiology of hyperbilirubinemia, they use the term "incompatibility’, which is incorrect. They have wrongly implied that Rh incompatibility and ABO incompatibility are hemolytic causes, referring to the rest as "non-hemo-lytic". Hence, attempting to determine the prevalence of neurological abnormalities among those with just "incompatibility" is not justified.

Thirdly, the presence of signs of acute bilirubin encephalopathy was not a part of the exclusion criteria mentioned by the authors. Thus, it is unclear how many patients with acute encephalopathy during the episode of jaundice went on to develop neurological findings on follow up. Early bilirubin encephalopathy is an absolute indication for doing an exchange transfusion, whereas estab-lished encephalopathy is a contraindication for doing an exchange transfusion. Hence, the inclusion of these patients in the follow up cohort would vitiate the interpretation of the neurological findings vis-a-vis the bilirubin levels.

Fourthly, there is a measurement bias in the estimation of the duration of jaundice. The ascertainment of the onset of jaundice has been left to the clinical acumen of the family, which may be difficult to rely on. The measurement error may be in either direction. It is possible that the family members of those newborns, who had more intense jaundice at presentation, had noticed the onset of jaundice later than the family members who had sought medical attention for milder jaundice. Their delay in identifying the true onset may have itself resulted in the increased severity. On the other hand, in their anxiety about the seriousness of the disease, they may have over-estimated the duration of the jaundice.

Lastly, the authors have tried to relate the neurological outcome to duration of neonatal jaundice, irrespective of the level of jaundice. Previous studies have shown that it is the duration for which the indirect bilirubin remained in a high zone that correlated with outcome(2,3). Ozmert et al.(2) showed that the duration for which the indirect bilirubin, remained above 20 mg/dl correlated with neurological outcome, whereas Nilsen et al. found that in hemolytic jaundice, bilirubin levels above 15 mg/dl for more than 5 days were predictive of lower IQ scores.

Given these methodological drawbacks, we differ with the authors. We do not agree with their conclusions and key messages.