Moral Reflections on Neonatal Intensive Care

In this issue of Pediatrics, Itabashi et al1 report survival statistics for infants at 22 weeks' gestation born in Japan. The most striking finding is that their survival rate was much better than that reported anywhere else in the world for infants born at this gestation. Although remarkable, this continues a well-known trend. Reported gestational age–specific survival rates have been better in Japan than in other countries for at least a decade.What can we learn from these or any other gestational age–specific survival statistics? How do they help us understand and resolve the moral dilemmas associated with treatment decisions at the borderline of viability?The following are some thoughts.1. Survival statistics tell us where we are and, over time, where we have been and where we can expect to be going in the near future. They are a measure of progress and of our rate of progress. For the first 30 years of neonatology (roughly from 1965 to 1995), each passing year saw a gain in birth weight–specific neonatal survival rates; the chances of survival at each birth weight increased steadily. Leaders in the field wondered where this trend would end, writing articles with titles such as “1000 g: How Small Is Too Small?” However, as Horbar et al2 recently showed, and as others have since replicated,3 this trend seems to be over. There has been little improvement in either birth weight–or gestational age–specific survival rates in the United States for the past decade. Moreover, there seems to be nothing on the horizon that is likely to change this trend or to improve survival rates for quasi-viable infants in the foreseeable future. The Japanese data are notable for being both an interesting exception (they continue to make steady progress) and a best-case scenario. It is unlikely that anybody else well achieve survival rates any better than those reported here. For the first time, it seems that our field has found its limits, its asymptote, and, as it were, run out of new ideas.2. What are the implications of the end of progress in neonatology? The first is emotional. The field is entering a new era, one in which the heady, youthful optimism of its early days must give way to a more sober and mature realism. Second, we need to interpret outcome data differently than we did in the past. Previously, when neonatology was advancing so quickly, the inevitable 4- to 5-year lag between birth, NICU outcome, 2-year follow-up, data analysis, manuscript submission, revision, and publication left all of us with a built-in and valid way of ignoring or discounting the more disturbing results of follow-up studies. We could say that the data that were being reported as “new” in the literature were inevitably already out of date and, therefore, not relevant in predicting outcomes for the infants who we were currently treating. That is no longer the case. Now, what is published in 2004 about infants born in 2000 is pretty reliable about infants born in 2008.3. Outcome data do not tell us anything about cost-effectiveness. Data showing that many premature infants survive with neurocognitive deficits are often cited as evidence of the overall expensiveness of neonatology. Such claims are econometrically flawed and, at the very least, beg for context and comparison with other health expenditures. Neonatal intensive care is one of the most cost-effective tertiary care interventions in all of medicine. It is far more cost-effective than adult intensive care, coronary bypass surgery, solid organ transplantation, renal dialysis, or many other well-accepted interventions. In standard economic approaches, treatments are considered cost-effective if they provide a quality-adjusted life-year (QALY) for less than $50000.4 In neonatology, each QALY costs less than $10000 even for infants at the lowest birth weights.5Put another way, in the NICU, at least 90 cents of every dollar spent is devoted to an infant who will survive to go home. This is true for NICU populations as a whole and true even if the population of interest is restricted to ventilated infants with a birth weight of <1000 g.3 How can this be? It is very simple: (1) doomed infants die early; (2) the smallest and the sickest die the quickest; (3) survivors stay in the NICU a long, long time; and (4) survivors live a long time after discharge, so the cost per QALY is amortized over a lifetime. In each of these ways, the adult ICU is the opposite of the NICU. The healthiest adults get out of the ICU quickest, doomed adults linger in the ICU, and those who survive to discharge often die in the first year after going home.6 More than 50% of adult ICU dollars are spent on patients who will die without leaving the hospital.7 The disproportionate focus on the cost-effectiveness of NICUs, as opposed to adult ICUs, suggests a disturbing moral calculus rather than a rigorous economic one.4. Is there a prejudice against micropremies? Janvier et al8 have eloquently articulated this position, amassing large amounts of survey data suggesting that micropremies are more likely to be “left to die” than older children and adults with the same, or even worse, prognoses. These data seem intuitively believable, perhaps some evolutionary remnant of prehistoric infant mortality epidemiology that led some cultures not to name children for 30 days or not to allow grieving until the child reached 1 year.What should we do about this apparent prejudice in our more modern era? At a minimum, Janvier et al suggested that recognizing this prejudice is important and further suggested that august professional groups might want to reconsider policies that offer blanket statements that deal with classes of infants and ignore individual differences. Before we accept policy statements in neonatology that are based solely on gestational age, we should ask whether we would accept similar policies in geriatrics based solely on chronological age. Can you imagine a statement from the American Medical Association claiming that resuscitation of 85-year-olds is “optional?” At the very least, the outcome criteria or the cost-effectiveness thresholds applied for one age group should apply to others, even tiny premature infants.5. The most common response to data suggesting prejudice against premature infants is the claim that post-NICU morbidity is more devastating than post–medical ICU morbidity—that the worst outcome of neonatology is not death in the NICU but the long-term survival of a neurologically devastated infant.9 Morbidity is clearly important to consider distinct from mortality. Here, 2 epidemiologic points are important, neither of which, in our view, has been sufficiently well described. First, morbidity does not tightly track mortality at the lowest birth weight. This is unprecedented in the history of neonatology, and its implications are underappreciated. In the early days of our subspecialty, a useful rule of thumb was that if an infant had a 10% chance of dying, then ∼10% of survivors could be expected to be neurologically impaired—50% survival then 50% morbidity. That is no longer true. Many outcome studies of infants born between 22 and 25 weeks' gestation have shown that, although survival rates differ at different gestational ages, the percentage of unimpaired survivors does not.10 The only real difference between an infant born at 23 week's gestation and one born at 26 weeks is the likelihood of getting out of the NICU; once infants have survived to discharge, the likelihood of “intact survival” is not much different between the 2 groups.Why should we explicitly describe outcomes for NICU survivors as opposed to all births? For many parents of micropremies, “trying and failing” is not the worst option—indeed, it may be the best option. We offered the infant our best shot and “kept the faith,” and the infant was just “too small.” People feel sad about the outcome and good about the process, and they move on. In this scenario, and for many parents, survival of a severely handicapped infant is the worst outcome. And, on this view, the likelihood of severe handicap as a percentage of infants who survive to discharge (not as a percentage of all births) is the most relevant statistic.6. Prognostication is obviously important. If we can distinguish in advance between infants who will survive and those who are doomed to die, we can offer the first group continued medical intervention and the second group quality comfort care. Prognostication for micropremies has traditionally been based on population statistics, and the most important characterization has been gestational age. That is the thrust of the Japanese article that prompted this commentary. More recently, Tyson et al11 incorporated other well-recognized features (race, gender, antenatal corticosteroids, small for gestational age) to refine population-based predictions for infants at the time of birth.This approach, although laudable, is more useful as an epidemiologic tool than it is as a guide to clinical decision-making for individual patients because it offers a snapshot of prognosis based only on data that are known at the moment of birth. However, irreversible decisions do not have to be made at the moment of birth. Infants can be admitted to the NICU for a “trial of therapy.” More information will become available over time that may help physicians to refine prognostications.127. What is our legal obligation to resuscitate infants born at 22 weeks' gestation? In the United States, legal obligations tend to be decided on a state-by-state and case-by-case basis. In Texas, Miller v HCA, Inc13 absolved doctors who overrode explicit parental requests for nontreatment. The courts articulated an “emergency exception,” wherein physicians who were confronted with a life-or-death decision in the delivery room were protected if they chose to intervene in an attempt to preserve life. The Texas court was silent on what the obligations are for a doctor who chooses not to intervene. In Wisconsin, the opinion handed down in Montalvo v Borkovec14 was different. Here, the court held that doctors were legally required to intervene if they thought there was a possibility of viability. Other countries have taken different approaches.158. What, in our view, is the most common source of moral distress for neonatologists? It is not a concern about money or resources, and it is not fear of malpractice suits. It is guilt over the long-term implications for families when a neurologically devastated infant survives. People often say we should take solace in the observation that although overall survival rates are much better now for extremely low birth weight infants than 20 or 30 years ago, the rates of severe cerebral palsy (CP) or other disabling neurologic morbidities have not changed. This epidemiologic truth does not mitigate the fact that improvements in neonatal intensive care have changed the locus of moral culpability. In the past, neurologic injury to newborns just happened. Today, the survival of neurologically impaired micropremies is a direct result of our decisions and our actions, and we feel culpable. It used to be that all of CP was God's fault; now roughly half of CP cases are our fault. That is hard to live with.Yet, perhaps the data from Saigal et al suggest a small source of solace.16 Most neurologically impaired NICU survivors judge their own quality of life to be acceptable. Doctors and nurses may be harder on themselves than they ought to be.9. What moral dilemmas are left in neonatology? Here are a few, and, sadly, updated reports of survival statistics are not likely to resolve them: (1) inevitable debate will persist over “how low is low enough?” (ie, at what prognostic threshold can invasive care be withheld or withdrawn?); (2) inevitable uncertainty will persist over the implications of confidence intervals for even the best prognostic indices; (3) inevitable disagreement will persist over which morbidities are “worth living with” and which are “worse than death”; and (4) disputes will arise over who should make these decisions and the process by which they should be made.17We close with an observation: the NICU follow-up clinic is our moral saving grace. We confront our “successes” and our “mistakes” visit by visit, week by week. In this way, we learn. We eagerly await the follow-up data from Japan to see whether their remarkable success in the NICU leads to better long-term outcomes.