eLetters to:

ARTICLES: Ahmed Cherif, Naima Khrouf, Sami Jabnoun, Chahnez Mokrani, Moez Ben Amara, Nedia Guellouze, and Samia Kacem Randomized Pilot Study Comparing Oral Ibuprofen With Intravenous Ibuprofen in Very Low Birth Weight Infants With Patent Ductus Arteriosus Pediatrics 2008; 122: e1256-e1261 [Abstract] [Full text] [PDF]

eLetters: Submit a response to this article

eLetters published:

More accurate measures of efficacy should be required for reporting clinical trials

Daniel Virella, Luis Pereira-da-Silva, Marta Oliveira and Micaela Serelha   (27 January 2009)

More accurate measures of efficacy should be required for reporting clinical trials 27 January 2009
Daniel Virella, Neonatologist NICU, Hospital Dona Estefânia, Lisbon Portugal, Luis Pereira-da-Silva, Marta Oliveira and Micaela Serelha Send letter to journal: Re: More accurate measures of efficacy should be required for reporting clinical trials danielvirella{at}oninetspeed.pt Daniel Virella, et al.	An interesting clinical trial comparing the efficacy of oral ibuprofen with intravenous ibuprofen for patent ductus arteriosus (PDA) closure in very low birth weight (VLBW) infants was recently published (1). Although this study is quite well designed, some methodological aspects should be addressed. The authors state that the differences found on the efficacy of both therapies are not statistically significant. Notwithstanding, neither the risk ratio nor the number needed to treat (NNT) or the number needed to harm (NNH) were estimated and the assumptions for the calculation of the sample size were not reached. Reviewing the published data we believe that they may be in fact significant. In the Results, the 95%CI of the efficacy risk ratio is presented without its estimate or the previous mention of its calculation. On the other hand, the risk ratio for adverse effects appears in the Discussion without being previously mentioned in the Results. In both cases, their interpretation lacks the insight due to a confidence interval that does not correspond to the confidence actually reached by the study itself. Dismissing the significance of clinical differences based on strict statistical analysis alone, has been a cause for delaying useful medical advances (2). Considering as non-significant 95% confidence intervals whose limits “tangentially touch” 1.0 may be clinically hazardous if not followed by proper discussion. Using the proper measures of efficacy (3) and the published data, it is possible to state that oral ibuprofen, compared with its intravenous form for treatment of PDA, has both a NNT and a NNH of almost 4.6 (95%CI 3.1-8.8). This means that for every 9 VLBW infants with significant PDA that are treated orally instead of intravenously, 2 more cases of success were achieved and occurred 2 cases less of adverse effects. If such measures of effect had been estimated, a more accurate insight on the clinical differences would have been reported and the impact of this study might have been greater. Nevertheless, we stress that the results of this important clinical trial deserve close attention from neonatologists. 1. Cherif A, Khrouf N, Jabnoun S et al. Randomized pilot study comparing oral ibuprofen with intravenous ibuprofen in very low birth weight infants with patent ductus arteriosus. Pediatrics 2008;122:e1256- 61. 2. Egger M, Smith GD. Meta-analysis: Potentials and promise. BMJ 1997;315:1371-4. 3. Hildebrandt M, Bender R, Gehrmann U et al. Calculating confidence intervals for impact numbers. BMC Medi Res Methodol 2006;6:32-9. Conflict of Interest: None declared