PEDIATRICS Vol. 120 No. 5 November 2007, pp. 1225-1226 (doi:10.1542/peds.2007-2592)
LETTER TO THE EDITOR |
Meta-Analysis Combining 2 Previously Reported Trials on Respiratory Distress Syndrome in Neonates: In Reply
Fernando Moya, MDDepartment of Neonatology
Coastal Area Health Education Center
Wilmington, NC 28402
Sunil Sinha, MD, PhD, FRCP, FRCPCH
Department of Pediatrics and Neonatal Medicine
University of Durham and James Cook University Hospital
Middlesbrough TS4 3BN, United Kingdom
Ralph B. D'Agostino, Sr
Framingham Study: Statistics/Data Management
Department of Mathematics/Statistics and Public Health
Boston University
Boston, MA 02215
In our article,1 we reported the 1-year follow-up of preterm infants enrolled in 2 contemporaneous clinical trials that compared lucinactant, a peptide-containing synthetic surfactant, with colfosceril and animal-derived surfactants. The primary efficacy end points for these trials were described in their original publications.2,3 The primary conclusion from this 1-year follow-up was that lucinactant is at least as good as, if not superior to, animal-derived surfactants for prevention of respiratory distress syndrome. In response to this report, Lavin and Senn et al express concerns primarily related to our analytical approach to the data.
Any retrospective analysis, be it a pooled patient-level analysis or a meta-analysis, can be biased, and we addressed the limitations of previously published meta-analyses4,5 in our article. Although variation in birth weight, gestational age, prenatal steroid use, and Apgar score in previously published meta-analyses is both substantial and clinically relevant, in our report these differences were slight and clinically unimportant. These trivial differences should reduce concerns about the validity of our report.
We clearly described that the 1-year follow-up in both trials was planned before initiation and that all outcomes were prespecified. The study investigators remained blinded to 1 year corrected age. The aim of our article was to report those findings, the first such report among surfactant-comparison trials.
The analytical approach in our 1-year follow-up was also prespecified, and additional analyses performed beyond the prespecified plan were clearly delineated in the article to avoid misleading the reader. Indeed, our approach would satisfy the rigorous regulatory standard quoted by Senn et al. However, publication in top-quality peer-review journals is the clinician's gold standard for guiding treatment decisions. Given the relevance and impact factor of Pediatrics, which also published the lucinactant trials, our reporting of these 1-year outcomes has met all standards.
The time points chosen for the analyses were clinically relevant. The first 28 days after birth defines the newborn period, the saccular phase of normal lung development concludes at 36 weeks' gestational age, and infancy concludes at 1 year of age. Clinically, it is most helpful to present data to neonatologists in a way that is consistent with their considerations of the patient population for which they care. The 1-year corrected age time point provided a concrete and universally understood clinical milestone while also providing a reasonable period of follow-up when relative differences in the degree of lung health and overall well-being may affect survival. The fact that there were minimal (<2%) losses to follow-up at 1 year corrected age in both trials gives us substantial strength to affirm our conclusions.
Interestingly, after criticizing our analyses, both Lavin and Senn et al provide their own retrospective interpretation, contradicting their early admonitions of such analyses to draw conclusions. In addition, Lavin suggests a pooled analysis, which requires more assumptions of the data than a meta-analysis. As suggested by Lavin and Senn et al, a logistic regression may be suitable when clinical milestones, such as 28-day survival, create landmark analyses. With a longer-term observation period such as a 1-year follow-up, a more sensitive and powerful time-to-event analysis such as survival analysis is the analysis of choice, even by regulatory agencies. Given patient-population characteristics in our studies, we used analytical methods that avoid bias in either direction and are most sensitive for identifying significant effects. The majority of deaths occur within 28 days after birth. Thereafter, the number of deaths out to 1 year corrected age is modest. Therefore, the Wilcoxon test provided, as Senn et al assert, an analysis that is sensitive to short-term differences and avoids bias on the basis of the small changes after the first 28 days after birth.
Regarding the figures, we regret that there was an error in the original publication; however, a correction was subsequently published to correct this misprint.
Statistical analyses of data from clinical studies are a means to understanding such data. Clinicians need to understand the degree to which study results can be applied to their scope of practice, and statistics allow them to judge whether these results are beyond mere random observations. In our 1-year follow-up, a numerically superior survival rate was observed for infants who were treated with lucinactant compared with those treated with beractant, poractant alfa, or colfosceril palmitate, with no difference in morbidity despite the proportionally higher number of survivors in the groups treated with lucinactant. Discussion of alternative statistical methods will not change this observation. Clinicians at the bedside are now left to decide how the results will affect their practice.
REFERENCES
- Moya F, Sinha S, Gadzinowski J, et al. One-year follow-up of very preterm infants who received lucinactant for prevention of respiratory distress syndrome: results from 2 multicenter randomized, controlled trials [published correction appears in Pediatrics 2007;12:935]. Pediatrics. 2007;119(6) . Available at: www.pediatrics.org/cgi/content/full/119/6/e1361
- Moya FR, Gadzinowski J, Bancalari E, et al. A multicenter, randomized, masked, comparison trial of lucinactant, colfosceril palmitate, and beractant for the prevention of respiratory distress syndrome among very preterm infants.
Pediatrics. 2005;115
:1018
–1029
[Abstract/Free Full Text] - Sinha SK, Lacaze-Masmonteil T, Valls I, et al. A multicenter, randomized, controlled trial of lucinactant versus poractant alfa among very premature infants at high risk for respiratory distress syndrome.
Pediatrics. 2005;115
:1030
–1038
[Abstract/Free Full Text] - Soll RF, Blanco F. Natural surfactant extract versus synthetic surfactant for neonatal respiratory distress syndrome. Cochrane Database Syst Rev. 2001;(2) :CD000144[Medline]
- Halliday HL. Recent clinical trials of surfactant treatment for neonates. Biol Neonate. 2006;89 :323 –329[CrossRef][Web of Science][Medline]
PEDIATRICS (ISSN 1098-4275). ©2007 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  S Andersson and A Hellstrom Abnormal optic disc and retinal vessels in children with surgically treated hydrocephalus Br J Ophthalmol, April 1, 2009; 93(4): 526 - 530. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
