1 Department of Biostatistics, Harvard School of Public Health
2 Department of Pediatrics, Harvard Medical School, The Children's Hospital
Bartlett et al,1 in last month's issue of Pediatrics, report the results of an innovative clinical trial of extracorporeal membrane oxygenation (ECMO) in the treatment of infants with respiratory failure. The authors report that, prior to introducing ECMO, infants meeting their criteria for respiratory failure had an 80% or greater chance of early death despite optimal conventional care. Because continued randomization in a 50-50 fashion would have presented ethical problems if early results were very favorable to ECMO, the investigators employed an "adaptive allocation" scheme, that is, a design that modified the treatment assignment probabilities according to results for previously randomized patients. The design chosen was a "randomized play-the-winner" design.2
This article has been cited by other articles:
![]() |
J. Worrall Why There's No Cause to Randomize Brit J Philos Sci, September 1, 2007; 58(3): 451 - 488. [Abstract] [Full Text] [PDF] |
||||
![]() |
R. H. Bartlett Historical Perspectives: Extracorporeal Membrane Oxygenation (ECMO) NeoReviews, June 1, 2005; 6(6): e251 - e254. [Full Text] [PDF] |
||||
![]() |
A. Biswas Adaptive designs for binary treatment responses in phase III clinical trials: controversies and progress Statistical Methods in Medical Research, October 1, 2001; 10(5): 353 - 364. [Abstract] [PDF] |
||||