An article I wrote, with that title, as an open peer commentary in the American Journal of Bioethics has just appeared on-line. The first 50 people to click on the link at the bottom of this post below can get a free full text copy, but I would ask you to check first if your institution gives you free full-text access, and let those who don’t have that privilege have the great pleasure of getting a free copy of this insightful analysis. It is, however, a response to 3 articles about the SUPPORT controversy (all available, with all the responses, if you have access on the page here). Some of the other responses are (also) great, such as the uncompromising brief article by Bill Meadow, which is less than a full page, so you can see it all on the first page preview even if you don’t have paid access, also articles by John Lantos, Thor Willy Ruud Hansen, Dominic Wilkinson and his colleagues from Oz, and, of course, Annie Janvier.
Some on the other hand are just annoying, a couple of bioethicists state in an article entitled ‘SUPPORT Asked the wrong question’ that, because the systematic review and meta-analysis of the older oxygen trials was reported as showing that restricting oxygen did not ‘unduly’ increase mortality, it should have been obvious that an increase in mortality was a likely outcome. And the fact that the authors of the Cochrane review used the word ‘unduly’ meant that they were worried there might be a difference.
The authors of that commentary clearly don’t realize that neonatology has changed since the 1950’s, and that those articles are just not informative for today’s medicine. When the studies that Lisa Askie and colleagues reviewed were performed there was no continuous oxygenation monitoring. At all. Of any kind. Most mildly preterm babies died. There were no ventilators. Just read a little about the details of the methodology of those studies and you will realize that they have no relevance for current practice. In contrast there were recently published observational studies including thousands of very preterm babies in modern NICUs which showed NO increase in mortality when saturation targets were lowered. They even repeat some of the ill-founded and uninformed opinions of other non-neonatologist critics that demonstrate that they too have no idea how neonatology works, or how we regulate oxygen therapy.
Neonatal Research 
you’re quoting us selectively and out of context. the cochrane review was most informative as a method for assessing whether the risk was foreseeable. we relied more on the 2003 review in Pediatrics by Cynthia Cole and her colleagues, wherein she acknowledges that more serious risks including that of mortality are possible, but resolving the issue would require more centers and more subjects. which is precisely our point. since you appear to like lantos, see his hastings center report piece with my CHOP colleague Chris Feudtner, http://onlinelibrary.wiley.com/doi/10.1002/hast.407/abstract, wherein they assert that SUPPORT was designed in concert with the international trials with an eye to using metaanalysis to determine whether mortality was increased (which is hardly consistent with the argument that mortality was a surprise and not at all foreseeable… ya can’t have it both ways); this all turns on what does it mean for a risk to be ‘foreseeable’; as a Bayesian, there was a signal about mortality and given the unequal values of mortality compared to ROP, there was an obligation to assure that mortality was not increased for any putative benefit in preserving eye function. see also my eletter to Pediatrics http://www.pediatricsdigest.mobi/content/134/4/651/reply#pediatrics_el_68802
cheers,
jon