Variation in the Treatment of Infants Born at the Borderline of Viability

Kaempf et al,1 in this month's Pediatrics, describe their experience with a formalized instrument for counseling pregnant women who are threatening to deliver between 22 and 26 weeks' gestation. They devised and pilot-tested their counseling instrument several years ago.2 Their report documents the satisfaction of women whose perinatal counseling at Providence St Vincent Medical Center (PSC) over 4 years was guided by the use of this counseling protocol.There were 320 births of infants between 22 and 26 weeks' gestation. Sixty of these deliveries were emergent and occurred without counseling. All of these infants were resuscitated at birth. Two hundred sixty births occurred after an opportunity for counseling using the guideline. Fifty (19%) women were interviewed extensively about their experience, and 25 (10%) were reinterviewed months later.The women were happy with the counseling experience. This seemed true whether the outcome was intact survival, survival with impairment, or death. The authors recommend that other centers should (or at least could) adopt a similar approach to counseling women in comparable circumstances.A few comments seem appropriate: one is methodologic and the rest are moral.Methodologically, the problem is in determining whether the good psychosocial outcomes were the result of the PSC protocol itself or were, perhaps, the result of the sort of NICU environment in which physicians would collaborate to develop such a protocol. It is impossible to tell from the data. There was no control group, not even a historical one. We do not know whether women who delivered at PSC before the protocol were generally dissatisfied, whether internal discord among the staff decreased, or even whether the types of decisions that followed protocol-driven counseling differed from those that occurred without the protocol. Perhaps the women were happy because the PSC doctors are really good doctors and sensitive communicators, with or without the protocol. Perhaps most parents who have a premature infant anywhere are generally satisfied with the skill and dedication of their medical team. To make claims about the specific effects of the counseling protocol, one would want either a historical control group that faced decisions without the benefit of a protocol or a concurrent control group that was counseled in a traditional, ad hoc fashion.Three moral issues are raised by this research: the goals of the protocol; the accuracy of the facts that are presented; and the actual process of decision-making that follows its use.At one level, the goals of the PSC protocol, like much counseling for critically ill patients, can be seen as internally inconsistent. One the one hand, the authors state, “We believe the choice of neonatal intensive care versus palliative comfort care in extremely premature infants rightfully belongs to medically informed parents.”1 On the other hand, they suggest a need to “collectively decide when the application of intensive care is or is not indicated” by using means to ensure that such decisions are “rational, compassionate, and responsible.” Thus, they want to both maximize parental autonomy and maximize the use of collective wisdom and responsibility. These 2 competing views might lead to tensions. Those tensions become clear from the data.The PSC guidelines recommend comfort care only at 22 to 23 weeks' gestation, parental choice at 24 weeks' gestation, and NICU treatment at ≥25 weeks' gestation. This roughly follows American Academy of Pediatrics recommendations and parallels the reported practice of many neonatologists in the United States.3 The large majority of women chose to follow the recommendations. That is not surprising either and is probably heartening for those who, like us, think that shared decision-making with a hefty dose of physician guidance is an appropriate way to help parents think through difficult decisions.Two groups of women in the study, however, deserve further attention. One group is of those who chose nonresuscitation for infants born at 25 and 26 weeks' gestation. Apparently, 6% of the parents chose palliative care at 26 weeks, and 9% chose it at 25 weeks. Survival rates for these infants at this hospital are 83% and 73%, respectively, at these gestations. Thus, PSC neonatologists would allow (although not encourage) nonresuscitation and nontreatment for infants with an 83% likelihood of survival. This is, presumably, in keeping with their focus on parental autonomy, but does it reflect collective wisdom and rationality? Are we willing to generalize this approach to other areas of pediatrics and let, say, trauma victims or patients with cancer with an 80% chance of survival to die if their parents opt for palliative care? And, if so, why curtail parental autonomy at 26 weeks? Why not 27 or 28 weeks? The implicit moral principle seems to be a probabilistic one—that an 83% survival rate with a 30% to 40% morbidity rate is insufficient to deem treatment obligatory but that 90% survival with 33% morbidity is sufficient?The second group of cases that deserves more attention and analysis are those in which the parents' “initial preferences” differed from the “final decisions.” These initial preferences were obtained after the use of the counseling protocol. Thus, we can assume that these parents were well-informed. Eighty parents initially chose comfort care, but only 37 infants actually received only comfort care. The drop off from initial preferences occurred at every gestational age except 22 weeks and was largest at 23 weeks (two thirds of the parents, apparently, either changed their minds or their wishes were not followed). What happened in those cases? What does that tell us about the efficacy of the protocol? To the extent that parents changed their minds, why does it seem like they mostly opted for more treatment rather than less? Finally, should the efficacy of the protocol be judged by parental satisfaction, parental freedom, or the degree to which it shaped decisions to reflect the preferences of the experts?Much of the impulse for restrictive treatment policies in the NICU comes from concerns about the burdens of bad outcomes. Those burdens are both emotional and economic. Families can be ruined by either or both, family ruin that often seems like it could have, and would have, been avoided if the decision had been for nontreatment and comfortable death rather than ongoing treatment and impaired survival. Some studies, however, have suggested the opposite. In Sweden, some centers treat more aggressively than others. Perhaps surprisingly, impairment among surviving infants was higher in the centers that were less aggressive, not in the ones that were more aggressive.4 Thus, even if the goal of protocols for counseling is to help parents make the best decision on the basis of long-term prognosis, it is unclear what decision is right or whether this protocol will be better than one that is either more proactive or more palliative. Perhaps different centers will answer that question differently and develop counseling protocols that subtly lead parents in different directions.Practice variations are common in all domains of medicine.5 Most are frowned on,6,7 as they seem not to be based on explicit policies or philosophies but, instead, seem at best random and at worst careless or ignorant.8 By contrast, practice variations in delivery room resuscitation of 22- to 26-weeks'-gestation infants persist despite better outcome data than we have ever had before. Where uncertainty remains, it is well-quantified uncertainty. We know what facts we know, and what facts we do not know, about delivery room resuscitation for 22- to 26-weeks'-gestation infants, and as best as we can tell, we are not going to learn much more in the foreseeable future.Protocols reflect values as well as facts, but values create facts. A policy that limits treatment for infants born at 24 weeks' gestation will lead to low survival rates for those infants. Those low survival rates will seem to justify and validate the policy, even if the true causal relationship runs in the other direction. In the United Kingdom, the EPICure study showed that survival rates for 23- and 24-weeks'-gestation infants were poor. Bioethicists at the Nuffield Council cited this study in their recommendation that treatment be optional at these gestational ages.9 Optional treatment policies led to deteriorating outcomes for 23- and 24-weeks'-gestation infants,10 which seemed to justify the restrictive treatment policies.At the hospital level, support for this argument can again be found in the PSC data. At 26, 25, and 24 weeks' gestation, PSC and Vermont Oxford Network survival outcomes are comparable. But at 23 weeks, Vermont Oxford Network outcomes are 5 times better (27% vs 5%). It seems likely that the PSC policy of discouraging resuscitation led to nonresuscitation, which, in turn, led to low rates of survival and, circularly, justified its policy.So, are differences in intrainstitutional, interinstitutional, or international policies about NICU resuscitation for 22- to 26-weeks'-gestation infants a good thing or a bad thing? In moral gray zones, we think they are probably a good thing. The very notion of a moral gray zone suggests uncertainty about what choice might be right or wrong. However, gray zones are not static—they can change with time, technology, politics, economics, and morality. Treatment of infants with Down syndrome used to be considered a gray-zone issue. Today, in most industrialized countries, it is considered morally obligatory. Extracorporeal membrane oxygenation used to be “experimental”; today it is not. Liquid ventilation still does not work, nor does resuscitating 22-weeks'-gestation infants, except in Japan, where it may! This is how moral progress is made—slowly, elliptically, and tentatively, with many false starts.Good ethics starts with good facts, but good facts do not necessarily lead to moral consensus. In neonatology, doctors, professional societies, bioethicists, parents, and judges may study the same data and come to fundamentally different conclusions. The efficacy of counseling parents in gray zones must ultimately be evaluated not just by measuring parental satisfaction but also by examining the assumptions, facts, and communication techniques that go into the counseling protocols. Parents could be highly satisfied whether the counseling protocol that pleases them is itself morally problematic or praiseworthy. PSC doctors and moms seem to be happy with their guidelines and suggest we should all consider guidelines. We probably should, but we probably shouldn't all use the same ones.