Objective. Much has changed in neonatal intensive care unit (NICU) care over the past decade. High-frequency oscillation, inhaled nitric oxide, and antenatal corticosteroids are now widely available. We wondered how these medical advances had affected both the epidemiology and ethics of life and death for extremely low birth weight (ELBW) infants in the NICU.
Methods. We identified 1142 ELBW infants (birth weight [BW] < 1000 g) consecutively admitted to our NICU between 1991 and 2001. We abstracted BW, gestational age, survival or death, and length of stay in the NICU. Statistical analyses were performed by using linear regression and 2-way analysis of variance.
Results. Both increasing BW and later year were significantly associated with improved survival. However, for larger ELBW infants, survival was ∼90% for the entire decade, and large-scale improvement was hardly possible. For smaller infants, greater improvements were both possible and observed, at least early in the decade. From 1991 to 1997, overall ELBW survival increased steadily (∼4% per year). However, from 1997 to 2001, there was no significant improvement in survival for ELBW infants. There was no change in the distribution of deaths accounted for by BW subgroups within the ELBW population from 1991 to 2001. Median length of stay for infants who eventually expired before discharge rose from 2 days in 1991 to 10 days in 2001. As a consequence, during the past decade, the percentage of infants whose outcome was “undeclared” by day of life 4 rose from 10% to 20% for ELBW infants overall and to 33% for infants with BWs of 450 to 700 g. The percentage of ELBW NICU bed-days occupied by nonsurvivors remained very low (∼7%) from 1991 to 2001.
Conclusions. 1) Fewer infants in all ELBW subgroups are dying, compared with a decade ago, and the improvement has been most prominent for BWs of 450 to 700 g, at which mortality was and remains to be greatest. 2) This progress seems to have slowed, or even stopped, by the end of the decade. 3) Although most NICU nonsurvivors still expire early, doomed infants are lingering longer. 4) Nonsurvivors continue to occupy a constant (and extremely small) fraction of NICU bed-days.
Progress in neonatology is generally portrayed as inexorable: doing better and better with smaller and smaller. For approximately the first 20 years it was. A succession of manuscripts published between 1970 and 1990 bore witness to our success. Articles exploring whether 1500 g was too small of a birth weight were quickly followed by those asking the same about 1000, 800, and 500 g.
By 1990, virtually all neonatal intensive care units (NICUs) had survival rates of ≥90% for infants with birth weights (BWs) >1000 g.5–9 Consequently, for individual patients with a BW >1 kilogram, parental refusal of intervention was precluded in the absence of other, non–BW-related circumstances. At the other end of a relatively narrow BW spectrum, below ∼500 g, survival was dismal.5–9 At a minimum, parental requests for nonresuscitation of infants below this BW seemed supportable, under the broad rubric of futility. Thus, the entire ethical debate surrounding NICU care was played out along a dimension of ∼1 lb: from 500 to 1000 g.
Moreover, another dimension of BW-specific mortality contributed to the ethical discussion: time. Small, sick infants who expired tended to do so quickly, and the smallest and sickest expired the quickest.10 Two perhaps unanticipated consequences followed. First, if parents could “hold their breathes” for a few days, the outcome for their infants was much clearer, one way or the other. The likelihood of survival to discharge for infants who had survived to day of life (DOL) 4 was >70%, whether the BW was 600 or 900 g.10 Second, because so few infants lingered long before dying, the percentage of NICU bed-days devoted to doomed infants was surprisingly low, even for BW cohorts in which overall mortality was extremely high. Indeed, for every BW cohort in the NICU, >80% of NICU bed-days were devoted to survivors, independent of whether the percentage of infants who actually survived was 20% or 80%.11–14
These epidemiologic truths were recognized by the early 1990s. However, much has changed in NICU care in the past decade. High-frequency oscillatory ventilation and inhaled nitric oxide are widely available.15–17 Perhaps even more important, antenatal corticosteroids have now become standard therapy for women with threatened preterm delivery.18–20 We wondered how these medical advances had affected both the epidemiology and ethics of life and death for extremely low birth weight (ELBW) infants in the NICU during the past 10 years.
We identified 1142 infants with BWs <1000 g who were consecutively admitted to our NICU between 1991 and 2001. We abstracted from their medical records BW, gestational age (GA), race, gender, survival or death of the infant, and LOS in the NICU (equivalent to the day of death for those infants who expired). This study was approved by the institutional review board of the University of Chicago.
The NICU at the University of Chicago is a level III regional center with ∼1000 admissions per year, 15 to 20 ventilated infants per day, and 50 inpatients per day for the past decade. Approximately 100 infants with BWs <1000 g are admitted each year. In our NICU, exogenous surfactant has been available for infants with presumed respiratory distress syndrome since 1989. The use of antenatal corticosteroids for women with threatened delivery of ELBW infants has increased from ∼15% in the early 1990s to ∼80% by 2001. High-frequency oscillatory ventilation became available in our NICU in 1994. Inhaled nitric oxide became available in 1996.
Statistical analyses as a function of study year were performed by using linear regression and 2-way analysis of variance for continuous variables and χ2 statistic for categorical variables. Statistical significance was accepted at the P < .05 level.
There were no significant changes in the risk characteristics of the ELBW population during the course of this study. Specifically, race, gender, GA, and BW distribution within the ELBW category did not vary as a function of study year.
Figure 1 displays BW-specific survival for 1142 ELBW infants admitted consecutively to the NICU at the University of Chicago from 1991 to 2001. Two points are apparent. Both increasing BW and later year were significantly associated with improved survival (both P < .001). However, there was a significant interaction effect; that is, the impact of time was not symmetrically felt as a function of BW. For larger ELBW infants, survival was ∼90% for the entire decade, and large-scale improvement was hardly possible. For smaller infants, greater improvements were both possible and observed, at least early in the decade. Later in the decade, improvement in BW-specific survival seems to have slowed or even ceased.
Figure 2 elaborates this latter point. Here, the overall survival for 1142 ELBW infants is presented as a function of birth year. The figure reveals 2 phenomena. From 1991 to 1997, ELBW survival increased steadily (∼4% per year; P < .01). However, from 1997 to 2001, there was no significant improvement in survival for the ELBW infants.
Figure 3 presents the distribution of deaths accounted for by BW subgroups within the ELBW population from 1991 to 2001. There was no change in relative proportion for any BW group. The smallest infants (450–600 g) consistently accounted for ∼40% of all ELBW deaths; the next smallest subgroup (601–700 g) accounted for ∼30% of all ELBW deaths; infants with BWs of 701 to 800 g accounted for ∼20% of all ELBW deaths; and the 2 largest BW subgroups accounted for <10% each of ELBW deaths.
Figure 4 displays one of the most important epidemiologic phenomena of the past 10 years from the perspective of personal ethics; that is, concerns for individual children and their parents. Figure 4 reveals that the median LOS for infants who eventually expired before discharge from the NICU rose steadily from ∼2 days in 1991 to 10 days in 2001 (r = .79; P < .01). In other words, in 1991, more than half of the doomed ELBW infants had expired by the end of DOL 2, but by 2001, it took 10 days to achieve the same effect. As Fig 4 also reveals, the average LOS for nonsurvivors paralleled the median for these same years, rising from 5 days in 1991 to 17 days in 2001 (r = .72; P < .02).
Figure 5 reveals another aspect of this phenomenon. In this figure, the percentage of doomed ELBW infants who have yet to expire are displayed as a function of DOL and year for the entire ELBW population (Fig 5A) and for the subgroup of infants with BWs of 450 to 700 g (Fig 5B). Two tendencies are apparent. First, ELBW infants tend to “declare themselves”; that is, with every passing DOL, the uncertainty in life-or-death outcome diminishes. However, this declaration seems to be growing increasingly fuzzy in recent years. As an example, between 1991 and 1992, only 10% of all NICU outcomes were undeclared by DOL 4. This percentage has risen steadily in the latter part of the decade and by 1999–2001 reached 20% for ELBW infants overall and 33% for infants with BWs of 450 to 700 g.
Although LOS for nonsurvivors rose between 1991 and 2001, the absolute number of nonsurvivors fell during this time. The consequences of these countervailing trends from the perspective of resource allocation for society at large are made clear in Fig 6. Figure 6 displays the ratio of ELBW NICU bed-days (combining NICU intensive care and NICU intermediate bed-days) occupied by infants who would not survive to discharge, compared with the total number of NICU bed-days occupied by all ELBW infants, survivors and nonsurvivors combined. The percentage of ELBW NICU bed-days occupied by nonsurvivors remained constant, and very low (∼7%), from 1991 to 2001. In no year did this ratio exceed 10%.
Figure 7 contrasts developments for ELBW infants with the cohort of larger BW infants. ELBW infants consistently accounted for approximately one-half of all bed-days occupied by nonsurvivors between 1991 and 2001. The average LOS for nonsurvivors with BWs >1000 g (12.8 ± 5.3 days) did not change significantly between 1991 and 2001.
Many technologic advances characterized NICU care in the 1990s. Consequently, and hearteningly, BW-specific survival for ELBW infants looks better than it did 10 years ago.21–24 Fewer infants in all ELBW subgroups are dying, compared with a decade ago, and the improvement has been most prominent in the BW groups between 450 and 700 g, in which mortality was and remains to be greatest. The good news is that we have been doing better and better with the infants with whom we need to be doing better.
Unfortunately, there seem to be some painful developments as well. Although most NICU nonsurvivors still expire early, they are not dying quite as early as they used to. We document here a steady rise in the median length of survival for doomed infants of approximately one-half day per year, from a median of DOL 2 to DOL 10. The average LOS for nonsurvivors paralleled this trend, rising from 5 to 17 days over the study period. It is clear that the timing of death for the subset of infants who do not survive is evolving.
These epidemiologic observations carry uncomfortable ethical consequences. A trial of therapy in the NICU takes longer than it used to. When the median day of NICU death was DOL 2, we were able to counsel NICU parents, even parents of infants born at 500 or 600 g, that if they could just “hold their breathes” for 2 or 3 days, we would have much different prognostic news to give them. Now we must ask them to hold their breathes for almost a week and a half to get the same news. Although the specific numbers almost certainly vary from NICU to NICU, the larger phenomenon of increased LOS for doomed infants seems to be widespread,25,26 a not-so-desirable side effect of improved survival rates for infants in this weight group.
The lengthening day of death for nonsurvivors forces reexamination of another of our previous observations: the vast majority of NICU bed-days are devoted to NICU survivors.11,12 When nonsurvivors expired quickly, it made intuitive sense (and turned out to be gratifyingly true) that these doomed infants consumed relatively few NICU bed-days. Consequently, the majority of bed-days associated with any BW cohort (even the smallest ones) was occupied by infants who remained alive to survive until discharge. However, if nonsurvivors are currently living longer than they used to, does it follow that more and more bed-days are devoted to the nonsurviving subgroup?
Perhaps not. During this same decade, overall survival has improved, particularly for the infants with the lowest BWs, who are at the greatest risk of dying. Because there are more and more survivors in each BW group, shouldn’t more and more NICU bed-days be devoted to survivors?
Which phenomenon dominates? Are more NICU bed-days now spent on nonsurvivors, because they are lingering longer, or are more days devoted to survivors, because there are relatively more of them? As Fig 6 reveals, the 2 trends have balanced each other out. Nonsurvivors occupied a constant (and extremely small) fraction of NICU bed-days (<10% in every study year and 7% for the decade overall). These observations continue to stand in sharp contrast with patients admitted to adult intensive care units (ICUs), in which nonsurvivors account for >50% of hospital bed-days.11,12
Interestingly, although NICU progress is generally portrayed as inexorable, this may no longer be quite accurate. There has been no recession in the lower limit of viability in our NICU over the past decade. Moreover, and perhaps more importantly, although BW-specific survival has improved during the past 11 years, this rate of improvement has slowed or even stopped. This phenomenon did not result from an increase in the risk characteristics of our ELBW population (there was no significant change in the proportion of white or male infants as a function of study year), nor was there a shift in the GA or BW distribution of our ELBW infants from larger to smaller. Moreover, to the extent that the risk characteristics of our ELBW population have changed at all, they have probably been reduced, reflecting the more frequent use of antenatal steroids as a function of study year.18–20
Nonetheless, our data and those of an even larger cohort reported by Horbar et al21 suggest that BW-specific survival may be reaching a plateau for ELBW infants. To reiterate, and in contrast to most public perception, the lower limit to successful NICU resuscitation has been relatively fixed over the past decade, and virtually all progress on the BW-specific mortality spectrum has been evidenced as a rising slope of survival for infants with BWs between 1 and 2 lb. Moreover, the rate of this rise is slowing or even stopped.
Several ethical consequences emerge from these observations. For infants with BWs >1000 g, BW-specific survival is so good that there are no ethically supportable claims for nonsupport as a function of either likelihood of death or excess cost based on BW alone. Over the past decade, BW-specific survival for infants with BWs of 800 to 1000 g has improved to the point at which the same ethical rubric probably applies. For the tiniest ELBW infants (BW 450 to 600 g) considered at the time of their birth, they remain relatively unlikely to survive. However, because at least half of these nonsurvivors will expire within 10 days, survival for the population of these infants who reach DOL 10 increases to at least 70%. Claims of futility are much less compelling when an infant is more likely to survive than not. Moreover, for almost all ELBW subpopulations, survival to hospital discharge is at least as good as survival to hospital discharge for any age group of intubated patients in an adult ICU.11 Finally, hospital days for ELBW infants admitted to the NICU are targeted to survivors much more precisely than hospital days for adults admitted to an adult ICU.11,12
Despite changes in mortality for ELBW infants during the 1990s, the essential ethics of neonatal intensive care have not changed. For most individual infants and their parents, uncertainty (at least from the perspective of life or death) remains acceptably low. Moreover, the likelihood of survival increases with every passing NICU day. Accurate and timely prediction of persistent residual morbidity for NICU survivors looms as the next ethical frontier.
- Received December 27, 2002.
- Accepted July 7, 2003.
- Address correspondence to William Meadow, MD, PhD, Department of Pediatrics, MC6060, 5825 S Maryland Ave, Chicago, IL 60637. E-mail:
- ↵O’Shea TM, Klinepeter KL, Goldstein DJ, Jackson BW, Dillard RG. Survival and developmental disability in infants with birth weights of 501 to 800 g, born between 1979 and 1994. Pediatrics. 1997;100 :982– 986
- Hack M, Friedman H, Fanaroff A. Outcomes of extremely low birth weight infants. Pediatrics. 1996;98 :931– 937
- ↵Hack M, Horbar JD, Malloy MH, Tyson JE, Wright E, Wright L. Very low birth weight outcomes of the National Institute of Child Health and Human Development Neonatal Network. Pediatrics. 1991;87 :587– 596
- ↵Meadow WL, Reimshisel T, Lantos J. Birth weight-specific mortality for extremely low birth weight infants vanishes by four days of life: epidemiology and ethics in the neonatal intensive care unit. Pediatrics. 1996;97 :636– 643
- ↵Meadow W, Frain L, Ren Y, Lee G, Soneji S, Lantos J. Serial assessment of mortality in the neonatal intensive care unit by algorithm and intuition: certainty, uncertainty, and informed consent. Pediatrics. 2002;109 :878– 886
- ↵Henderson-Smart DJ, Bhuta T, Cools F, Offringa M. Elective high frequency oscillatory ventilation versus conventional ventilation for acute pulmonary dysfunction in preterm infants. Cochrane Database Syst Rev. 2000;2 : CD000104
- ↵Wright LL, Verter J, Younes N, et al. Antenatal corticosteroid administration and neonatal outcome in very low birth weight infants: the NICHD Neonatal Research Network. Am J Obstet Gynecol. 1995;175 :269– 274
- ↵Horbar JD, Badger GJ, Carpenter JH, et al. Trends in mortality and morbidity for very low birth weight infants, 1991-1999. Pediatrics. 2002;110 :143– 151
- Piecuch RE, Leonard CH, Cooper BA, Sehring SA. Outcome of extremely low birth weight infants (500-999 grams) over a 12-year period. Pediatrics. 1997;100 :633– 639
- ↵Richardson DK, Gray JE, Gortmaker SL, Goldmann DA, Pursley DM, McCormick MC. Declining severity adjusted mortality: evidence of improving neonatal care. Pediatrics. 1995;102 :893– 899
- ↵Meadow WL, Lantos J, Frain L, Ren Y. Early deaths in Chicago and New England. Pediatrics. 1997;99 :754
- ↵Ellington M, Richardson DK, Gray JE et al. Early deaths in Chicago and New England. Pediatrics. 1997;99 :753– 754
- Copyright © 2004 by the American Academy of Pediatrics