eLetters is an online forum for ongoing peer review. To submit an eLetter please go to the article you wish to respond to and click on the link that reads "eLetters: Submit a Response." Submission of eLetters are open to all health care professionals and experts in related fields.

eLetters to:

ARTICLES:
Bryan L. Burke, James M. Robbins, T. Mac Bird, Charlotte A. Hobbs, Clare Nesmith, and John Mick Tilford
Trends in Hospitalizations for Neonatal Jaundice and Kernicterus in the United States, 1988–2005
Pediatrics 2009; 123: 524-532 [Abstract] [Full text] [PDF]
*eLetters: Submit a response to this article

eLetters published:

[Read eLetters] Kernicterus Incidence Estimates Unreliable
Thomas B. Newman, M. Jeffrey Maisels, Willaim Beaumont Hospital, Royal Oak, Michigan   (25 February 2009)

Kernicterus Incidence Estimates Unreliable 25 February 2009
Top
Thomas B. Newman,
Professor of Epidemiology and Biostatistics and Pediatrics
University of California, San Francisco,
M. Jeffrey Maisels, Willaim Beaumont Hospital, Royal Oak, Michigan

Send letter to journal:
Re: Kernicterus Incidence Estimates Unreliable

newman{at}epi.ucsf.edu Thomas B. Newman, et al.

We applaud the effort of Burke et al. to provide US nationwide estimates for kernicterus incidence. However, the fact that about 70% of the infants with a diagnosis of kernicterus had no codes for either phototherapy or exchange transfusion undermines confidence in the results. Although the authors appropriately regarded these diagnoses as suspect, and excluded them from their incidence estimates, this high number of exclusions demonstrates the unreliability of the method.

The authors suggest some of these excluded subjects with kernicterus diagnoses may have been "rule out" diagnoses (e.g., if kernicterus was coded if the treating clinician had recorded "rule out kernicterus" as a discharge diagnosis). However, excluding infants with no phototherapy or exchange transfusion codes would not solve this problem, because the vast majority of infants in whom kernicterus is to be "ruled out" should at least receive phototherapy. Thus "rule-out" diagnoses (or other coding errors) could still be inflating the incidence estimates. On the other hand, our experience with legal cases suggests that a significant proportion of children ultimately diagnosed with kernicterus are not diagnosed in the newborn period. Thus, there is a potential for underestimation as well as overestimation of incidence, and the incidence of kernicterus in the US remains poorly quantified.

Luckily, unless these biases are changing significantly over time, the reassuring data on time trends may be more reliable than the absolute estimates of incidence.

Conflict of Interest:

None declared