Published online June 2, 2008
PEDIATRICS Vol. 121 No. 6 June 2008, pp. 1288-1289 (doi:10.1542/peds.2008-0549)

This Article Extract Full Text (PDF) Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Request Permissions Citing Articles Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Barrington, K. J. Articles by Finer, N. N. Search for Related Content PubMed Articles by Barrington, K. J. Articles by Finer, N. N. Related Collections Premature & Newborn Social Bookmarking                         What's this?

LETTER TO THE EDITOR

Inhaled Nitric Oxide for Preterm Infants: In Reply

Keith J. Barrington, MBChB
Department of Pediatrics
McGill University
Royal Victoria Hospital
Montreal, Quebec, Canada H3A 1A1

Neil N. Finer, MD
Department of Pediatrics
University of California San Diego
La Jolla, CA 92093

We thank Truog et al for their interest in our review article. It was not clear to us while reading the original publication¹ that the main analysis of the primary outcome variable had used the multiple outputation technique. The authors' letter has confirmed that this indeed was the case.

We continue to have concerns about the use of this technique, however; it cannot be replicated by readers or reviewers of an article and is understood by very few. If, rather than using this new approach, data from all of the enrolled infants are included, the difference in the primary outcome does not reach conventional levels of significance (P = .09).² Thus, only by applying this novel technique, which does not seem to have been used for this purpose previously and treats the multiple infant siblings as a cluster, are the results statistically significant. The important fact that the decision to use this technique was made before the trial commenced is reassuring. The apparent intention was to counter the potential effect of multiple deliveries with a presumed greater correlation of results compared with randomly assigned singletons.

The problem of how to deal with multiple gestations in neonatal randomized, controlled trials is recurrent. Some investigators have chosen to randomly assign them individually and accept that there may be more refusals from parents if their infants cannot be guaranteed to be in the same group. Others have enrolled the second infant in a nonrandomized fashion into the same group as the first infant and then analyzed the data from all the infants by using standard techniques, sometimes with an increase in sample size to account for the presumed correlation. The multiple outputation technique systematically discounts the information obtained from the individual multiple siblings by running the analysis hundreds or thousands of times with the information from 1 multiple or another randomly deleted³; how the results of such an analysis should be reported is unclear. The technique also raises the ethical issue of whether it is appropriate to "undervalue" the data from the multiples compared with the singletons (the "n" for the analyses is based on the number of sets of multiples rather than the number of infants) and whether the parents of the multiple sets should be informed in the consent process that their infants' data will have less impact on the results than the data from the singleton infants enrolled.

Overall, it seems to us that this approach is not well understood and may undervalue the contribution of infants of multiple gestations, and the results of such analyses do not seem to be consistent with more standardized statistical approaches. These observations suggest that more study is required of the statistical techniques currently used to analyze neonatal prospective trials.

REFERENCES

1. Ballard RA, Truog WE, Cnaan A, et al. Inhaled nitric oxide in preterm infants undergoing mechanical ventilation [published correction appears in N Engl J Med. 2007;357(14):1444–1445]. N Engl J Med. 2006;355 (4):343 –353[Abstract/Free Full Text]

2. Barrington KJ, Finer NN. Inhaled nitric oxide for preterm infants: a systematic review [published correction appears in Pediatrics. 2008;121(2):451]. Pediatrics. 2007;120 (5):1088 –1099[Abstract/Free Full Text]

3. Follmann D, Proschan M, Leifer E. Multiple outputation: inference for complex clustered data by averaging analyses from independent data. Biometrics. 2003;59 (2):420 –429[CrossRef][Web of Science][Medline]

---

PEDIATRICS (ISSN 1098-4275). ©2008 by the American Academy of Pediatrics

CiteULike    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

This Article Extract Full Text (PDF) Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Request Permissions Citing Articles Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Barrington, K. J. Articles by Finer, N. N. Search for Related Content PubMed Articles by Barrington, K. J. Articles by Finer, N. N. Related Collections Premature & Newborn Social Bookmarking                         What's this?