- BIOETHICS FORUM ESSAY
OHRP and Public Citizen Are Wrong about Neonatal Research on Oxygen Therapy
On April 10, 2013, the advocacy group Public Citizen reacted to OHRP’s findings with a letter to Kathleen Sibelius, director of Health and Human Services. Public citizen criticized HHS for being too mild in its critique and too lenient in its proposed remedy. It claimed that the study was “highly unethical” because it “exposed 1,316 extremely premature infants to increased risks of either death or retinal damage.” It also claimed that egregious omissions in the informed consent process “caused parents to enroll their premature infants in this experiment under the false pretense that it was much safer for their infants than was known to be the case.” Public Citizen’s critiques go beyond informed consent. They suggest that the study should never have been done in the first place because enough was already known about oxygen therapy that “any study comparing the two experimental target levels of oxygen saturation would be both unethical and not compliant with requirements of HHS regulations.” Public Citizen demanded that the directors of both HHS and NIH personally apologize to all of the parents who enrolled their babies in the study.
OHRP and Public Citizen are deeply misinformed about the design of SUPPORT, about the well-being of the infants enrolled in it, and about oxygen therapy for neonates generally. Their criticisms of the study are not only off-target, but dangerous.
In the study, infants were randomized to doses of oxygen that would maintain their oxygen saturation levels at either 85-89% or 91-95%. At the time, standard care was to keep infants somewhere in the range between 85% and 95%, without any greater specificity. According to OHRP, the consent form accurately described this as the normal range for oxygen saturations and accurately stated that all infants who were in the study, as well as those who were not, would receive oxygen at concentrations that would keep oxygen saturations between 85%-95%.
At the time the study was designed and conducted, there was wide variation in standard practices of oxygen administration throughout the world. As one scholarly review of such therapy noted, “NICUs in the US today have a wide range of SpO2 [oxygen saturation] guidelines… A randomized, controlled trial is needed to establish a safe upper limit of SpO2 in the premature infant at risk for developing ROP [retinopathy of prematurity].”(1) Most clinicians and investigators at the time believed that lower oxygen levels were safer than higher ones in that they protected infants from retinopathy without increasing mortality. Very few people would have guessed that lower oxygen levels would lead to increased mortality, although many hoped that it would lead to lower levels of retinopathy. The SUPPORT study was designed to determine whether these beliefs and hopes were true. The consent form thus focused on ROP and stated, “the benefit of higher versus lower levels of oxygenation in infants, especially for premature infants, is not known.”
As it turned out, the study showed that the mortality rate was higher in the lower oxygen group than in the higher oxygen group. This surprising finding has now led OHRP to conclude retrospectively that the consent form was inadequate and for Public Citizen to conclude that the study was unethical and that the parents of babies who were enrolled deserve an apology. I believe that both OHRP and Public Citizen are learning the wrong lessons and drawing the wrong conclusions. To see why, I focus on three questions that bear on the ethics of this study and of neonatal research more generally: 1) Was the study design ethical and appropriate at the outset? 2) Were babies in the study harmed? and 3) Was the consent form adequate?
Was the SUPPORT study design ethical and appropriate at its outset?
Public Citizen says no. OHRP and the investigators (and their IRBs) say yes. OHRP and the investigators are right. At the time of the study design, there was significant uncertainty within the professional community about the optimum level of oxygen to provide to premature babies. In 2002, Tin reviewed the available data and concluded, “Oxygen must have been given to more infants than any other medicinal product in the last 60 years. Despite that, we still know very little about how much infants actually need, or how much it is wise to give. The depth of our ignorance is really quite embarrassing.”(2) Silverman echoed these sentiments, noting that “the optimum range of oxygenation… remains to this day, unknown.”(3) The recognition of the need for better data was not limited to the United States. An international consortium of researchers in neonatology recognized the need for – and conducted – similar trials. In fact, this is the only time in the history of neonatology when researchers around the world acknowledged the urgent need for a prospective randomized trial. There was clearly a state of “community equipoise” about oxygen dosing. Studies of the effects of oxygen on survival, retinopathy, and long-term neurological outcome were essential and ethically appropriate.
Were babies in the study at higher risk than babies who were not in the study?
Public Citizen and OHRP say yes. The investigators and IRBs say no. The investigators and IRBs are right. Every baby in the study received treatment that could have and perhaps would have been given to babies who were not enrolled in the trial. At the outste, nobody knew which treatment was safer. Babies who were randomized to one treatment or the other were thought to be at no higher risk of a bad outcome (or likely to have a good outcome) than babies whose treatment was decided based on clinical judgment. This was true at the beginning of the study and, we now know, it remains true now that the data have been analyzed.
How do we know? The fact that this study was conducted within the NICHD Neonatal Research Network actually allows the risk of the study to be quantified. That is because the network collects data on all babies born at these gestational ages, whether they are in a study or not. The data for babies in the study, and for all babies treated at these same hospitals, can be found in two papers, one in the New England Journal of Medicine and one in Pediatrics. The first, published May 10, 2010, reported survival rates and rates of severe eye disease for all babies in the SUPPORT study.(4) The other, published August 23, 2010, reported the same outcomes for all babies in the neonatal research network (some of whom were in the SUPPORT trial, most of whom were not).(5) Here is what they show. The babies in the “low oxygen” arm of the clinical trial had a mortality rate of 19.9%. The babies in the “high oxygen” arm of the study had a mortality rate of 16.2%. Babies in the network overall had a mortality rate of 24%. For severe retinopathy, the numbers are 8.6% (low oxygen group), 17.9% (high oxygen group) and 24.1% (overall group). In other words, babies in both arms of the study had higher survival rates and lower rates of retinopathy than babies who were not in the study. The fact that some babies in the study were in the overall group minimizes these differences. The true differences were probably larger. Babies enrolled in the study were protected, not endangered, by being in the study.
How can this be? There are a few possible explanations. It may have been a selection bias, as Rich and colleagues suggested.(6) Since consent for the study was sought prenatally, pregnant women who were receiving prenatal care were more likely to enroll their babies in the study than women who did not receive prenatal care. Another possible explanation is that damage from oxygen toxicity is not related to the absolute level but instead is associated with fluctuations in oxygen level, so that the babies in the trial, who received more stable levels of oxygen than babies not in the trial, may have been better off as a result. Or it may reflect the well-known but poorly understood phenomenon of “inclusion benefit,” by which subjects in either arm of a randomized trial have better outcomes than patients who were eligible for the study but not enrolled.(7) The fact is, we don’t know why babies in the trial did better. But we know that they did. Thus, to claim that babies were harmed by being in the study and that those harms could have been avoided by receiving conventional therapy is not just wrong, but dangerously wrong, as it will discourage research, discourage parents from enrolling their children in important research studies, and, by ignoring the actual data, suggest that even completed research ought to be ignored.
In spite of this data, OHRP suggested that the consent form should have stated that “the level of oxygen being provided to some infants, compared to the level they would have received had they not participated, could increase the risk of brain injury or death” (OHRP letter, page 10). Given the data, it is hard to know how to interpret OHRP’s reasoning. Their position is apparently that informed consent forms need to inform parents not only of known risks and of possible risks, but also of risks that the investigators did not think were possible – even after those risks have been shown not to exist. Essentially, there is no risk that does not fall into this category. By these criteria, consent forms should state something like, “ANY risk that you can imagine, and ALL risks that you cannot even imagine, and EVEN RISKS THAT HAVE BEEN SHOWN NOT TO EXIST, are possible as a result of participation in this study.”
Should the consent form have more accurately disclosed the goals of the study?
The goals of a study are different from the risks of the study. The SUPPORT study, according to its protocol as cited by OHRP, was designed to determine whether “a lower target range of oxygen saturation (85% to 89%), as compared with a higher target range (91% to 95%), would reduce the incidence of the composite outcome of severe retinopathy of prematurity or death among infants who were born between 24 weeks of gestation and 27 weeks 6 days of gestation.” The consent documents did not mention the goals of the study. They should have, and they should have explained that nobody knew if either arm of the study would do better or worse or whether babies in the study would do better or worse than those receiving “conventional” therapy outside the study. A model for such disclosure can be found in the consent form for a similar trial conducted in New Zealand, in which the consent form described the risks and benefits as follows:
“Too high oxygen in the blood for long periods may 1) contribute to abnormal development of the retina (a condition called retinopathy of prematurity – ROP) and affect vision – it is even possible for some babies with ROP to become blind; 2) contribute to changes in the lungs that mean the baby needs ongoing help with breathing for weeks or months (a condition called bronchopulmonary dysplasia – BPD); 3) be one cause of damage to brain cells and lead to developmental problems.
Too low oxygen in the blood for long periods may 1) increase the risk the baby will not survive or contribute to poor growth; 2) raise blood pressure in the lungs and contribute to bronchopulmonary dysplasia; 3) damage the brain cells and lead to developmental problems.” (BOOST-NZ consent, July 2005, personal communication from Brian Darlow, principal investigator of BOOST-NZ)
The risks of reckless opinions
The most important ethical issues for the future of clinical research, both in neonatology and in other areas of medicine, are the issues of honesty, transparency, and safety. Trials should not be done if the risks are too high or if the study subjects aren’t informed of those risks. Interim analyses of data should allow studies to be stopped if one arm is clearly riskier or more beneficial than the other.
It is difficult to convey the need for randomized trials to any patient, but perhaps particularly to parents of critically ill children. Perhaps for this reason, there have been very few prospective randomized trials in neonatology. As a result, many therapies in neonatology have not been validated and may be dangerous. Neither parents nor babies are well served by a system that is exquisitely attentive to the risks of randomized trials but oblivious to the risks of unvalidated therapies. But that is exactly what OHRP and Public Citizen recommend. If their recommendations are followed, avoidable harm to babies will occur.
Their criticism of this important trial also discredits bioethics. Bioethics, at its best, advocates for the most vulnerable patients and criticizes doctors and researchers who put those patients at risk. In this instance, OHRP and Public Citizen have, in the name of bioethics and advocacy, criticized clinical investigators who were trying to prevent the use of harmful and unvalidated treatments and, instead, endorsed an anti-intellectual, unscientific approach to medical innovation for the most vulnerable patients.
The SUPPORT trial should be praised for its design. It was a crucially needed, rigorous evaluation of risks and benefits of supplemental oxygen for premature babies. Moreover, the study demonstrably protected babies. Study subjects were benefitted, not harmed, by being in the trial. The risks to babies in the trial were lower, not higher, than the risks to which babies not in the study were exposed. No apology is needed for that.
It is shocking that OHRP and Public Citizen did not see fit to understand the study or, apparently, to analyze the results before claiming that it was risky to babies. The real risk to babies comes from reckless and ill-informed opinion about highly ethical scientific studies. To minimize this risk, OHRP and Public Citizen should apologize to the investigators and to the parents of the babies in these studies for their sensationalistic misinterpretations of the support study. If they do not, and if fewer babies are enrolled in such studies, then more rather than fewer babies will die, more rather than fewer babies will go blind, and OHRP and Public Citizen will have been responsible for those harms to innocent babies.
1. C.G. Anderson, W.E. Benitz, A. Madan, “Retinopathy of Prematurity and Pulse Oximetry: A National Survey of Recent Practices,” Journal of Perinatology 24 (2004): 164-68.
2. W. Tin, “Oxygen Therapy: 50 Years of Uncertainty,” Pediatrics 110 (2002): 615-16.
3. W.A. Silverman, “A Cautionary Tale about Supplemental Oxygen: The Albatross of Neonatal Medicine,” Pediatrics 113 (2004): 394-96.
4. SUPPORT Study Group of the Eunice Kennedy Shriver NICHD Neonatal Research Network, “Target Ranges of Oxygen Saturation in Extremely Preterm Infants,” New England Journal of Medicine 362 (2010): 1959-69.
5. B.J. Stoll et al., “Neonatal Outcomes of Extremely Preterm Infants from the NICHD Neonatal Research Network,” Pediatrics 126 (2010): 443-56.
6. W. Rich et al.,“Enrollment of Extremely Low Birth Weight Infants in a Clinical Research Study May Not Be Representative,” Pediatrics 129 (2012): 480-84.
7. J.D. Lantos, “The ‘Inclusion Benefit’ in Clinical Trials,” Journal of Pediatrics 134 (1999): 130-31; W.A. Silverman, “Disclosing the ‘Inclusion Benefit,” Journal of Perinatology 22 (2002): 261-62; G.E. Vist et al., “Outcomes of Patients Who Participate in Randomized Controlled Trials Compared to Similar Patients Receiving Similar Interventions Who Do Not Participate,” Cochrane Database Systematic Review July 16, no. 3 (2008):MR000009.
Posted by Greg Kaebnick at 04/18/2013 05:03:48 PM |
Published on: April 18, 2013
Published in: Clinical Trials and Human Subjects Research