Statistics from Altmetric.com
The study by Keren and colleagues1 is a retrospective study, using infants in whom pre- and post-discharge TSB has been carried out, hence causing an inherent bias towards the same group.
The data for clinical risk have been collected from documents such as admission, intrapartum, and discharge forms. This retrospective collection can result in missing or ambiguous data, as has been accepted by the authors. Ideally, a study should be prospective using both methods on all neonates in a study group, and then the sensitivity and specificity (that is, false positives and false negatives) should be compared using actual data on follow up.
The clinical risk factor score includes factors that are interrelated such as vacuum and cephalhematoma. In cases where the cephalhematoma is caused by the use of vacuum the neonate gets a double rating. Obviously, authors have not found clinical risk factors more specific than pre-discharge TSB.
Contrary to this study, the AAP guidelines promote and support breast feeding and state that effective breast feeding can reduce substantially the risk for hyperbilirubinaemia.2 It is known that inadequate feeds increase the level of neonatal jaundice; hence the emphasis on “effective” breast feeds. The study subjects date from 1993–97 and the feeding habits (that is, breast/bottle/combination feeds) have been given a considerable amount of significance, which contradicts the AAP guidelines by the subcommittee on hyperbilirubinaemia.2
Newman et al state that, compared to early TSB levels (<48 hours of life), clinical risk factors combined with TSB significantly improve prediction of subsequent hyperbilirubinaemia.3
Suresh et al have studied the cost effectiveness of strategies to prevent kernicterus, and concluded that to prevent one case of kernicterus, the cost was $10 321 463 for universal follow up of early newborn discharge, $5 743 905 for routine predischarge TSB, and $9 191 352 routine predischarge transcutaneous bilirubin with selective follow up.4 They concluded that widespread implementation of these strategies would result in significantly increased healthcare costs with uncertain benefits.4
The AAP guidelines also focus on the rarity of kernicterus and aim to reduce the incidence of kernicterus, while minimising the risks of unintended harm such as maternal anxiety, decreased breast feeding, and unnecessary costs or treatments.2
They recommend a systematic clinical assessment before discharge and an early and focused follow up based on the risk assessment.3 Finally we must remember that we are all clinicians and we should use the lab report as an adjunct to our clinical knowledge.
Drs Kanjilal and Prasad make some important observations but are mistaken in several of their assertions. First, they suggest that because we limited our study sample to infants for whom pre- and post-discharge TSBs were performed, our results are affected by some form of selection bias. The bias they are referring to is verification bias, in which only patients with “positive” or more concerning test results have a follow up test to verify the original results. By decreasing the number of patients with “negative” test results, this bias has the effect of overestimating test sensitivity and underestimating specificity. However, as we point out in our manuscript, we studied infants enrolled in an early discharge follow up programme and minimised the potential for verification bias by restricting our sampling frame to months during which >75% of enrolled infants had post-discharge TSB measurements performed. In fact, for the majority of these months, >90% of enrolled infants had post-discharge TSBs measured.
The second point on which Drs Kanjilal and Prasad are mistaken concerns the inclusion of “interrelated” factors “like vacuum and cephalhaematoma” in our clinical risk factor scoring system. As summarised in table 2, vacuum extraction is included in the scoring system but cephalhaematoma is not. In fact, contrary to our expectation, cephalhaematoma was not associated with development of post-discharge TSB >95th centile. This simply may be a result of poor documentation of cephalhaematoma in the admission and discharge physical examinations (misclassification bias), but it raises concerns about the use of subjective factors in clinical risk factor scoring systems. Our results suggest that using more objective findings, such as use of vacuum extraction during delivery—a common cause of cephalhaematoma—may provide more accurate information about subsequent risk of hyperbilirubinaemia.
Finally, our finding that breast feeding increases the risk of hyperbilirubinaemia is not new and should not be interpreted as a recommendation against breast feeding. As paediatricians who routinely care for newborn infants, we recognise the benefits of breast feeding and strongly support its use. However, at the same time we are cognisant of the potential risks of dehydration and hyperbilirubinaemia posed by inadequate intake in breast fed infants. The results of our study should be interpreted as another reminder that healthcare systems and providers must work to ensure adequate lactation support for breast feeding mothers and early identification and treatment of breast feeding problems that may result in inadequate intake for infants.
As Drs Kanjilal and Prasad suggest, a prospective validation of alternative risk assessment strategies is needed to confirm the results of our study as well as other studies of alternative screening strategies. Additional studies are also needed to evaluate the incremental benefit of using clinical risk factors in addition to the pre-discharge TSB to predict which infants are at risk of developing severe hyperbilirubinaemia. And finally, more studies are needed to evaluate the cost effectiveness of alternative strategies for screening and tracking infants for their risk of developing severe hyperbilirubinaemia in order to prevent the occurrence of kernicterus, an uncommon but devastating, costly, and entirely preventable condition.