Published online April 1, 2008
PEDIATRICS Vol. 121 No. 4 April 2008, pp. 732-740 (doi:10.1542/peds.2006-2797)
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow E-mail this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My File Cabinet
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Meadow, W.
Right arrow Articles by Schreiber, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Meadow, W.
Right arrow Articles by Schreiber, M.
Related Collections
Right arrow Premature & Newborn
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Facebook   Add to Reddit   Add to Technorati   Add to Twitter  
What's this?

ARTICLE

Just, in Time: Ethical Implications of Serial Predictions of Death and Morbidity for Ventilated Premature Infants

William Meadow, MD, PhDa,b, Joanne Lagatta, MDa, Bree Andrews, MD, MPHa, Leslie Caldarelli, MDa, Amaris Keiser, BAa, Johanna Laporte, BA, RNa, Susan Plesha-Troyke, OTRa, Madhu Subramanian, BAa, Sam Wong, BAa, Jon Hron, BAa, Nima Golchin, BAa and Michael Schreiber, MDa

a Department of Pediatrics
b MacLean Center for Clinical Medical Ethics, University of Chicago, Chicago, Illinois


   ABSTRACT
TOP
 ABSTRACT
METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
OBJECTIVES. For a cohort of extremely premature, ventilated, newborn infants, we determined the power of either serial caretaker intuitions of "die before discharge" or serial illness severity scores to predict the outcomes of death in the NICU or neurologic performance at corrected age of 2 years.

METHODS. We identified 268 premature infants who were admitted to our NICU in 1999–2004 and required mechanical ventilation. For each infant on each day of mechanical ventilation, we asked nurses, residents, fellows, and attending physicians the following question: "Do you think this child is going to live to go home or die before hospital discharge?" In addition, we calculated illness severity scores until either death or extubation.

RESULTS. A total of 17066 intuition profiles were obtained on 5609 days of mechanical ventilation in the NICU. One hundred (37%) of 268 profiled infants had ≥1 intuition of die before discharge. Only 33 infants (33%) with an intuition of die actually died in the NICU. Of 48 infants with even 1 day of corroborated intuition of die in the NICU, only 7 (14%) were alive with both Mental Developmental Index and Psychomotor Developmental Index scores of >69, and only 2 (4%) were alive with both Mental Developmental Index and Psychomotor Developmental Index Scores of >79 at corrected age of 2 years. On day of life 1, the Score for Neonatal Acute Physiology II value for nonsurvivors (38.2 ± 18.1) was significantly higher than that for survivors (26.3 ± 12.7). However, this difference decreased steadily over time as scores improved for both groups.

CONCLUSIONS. Illness severity scores become progressively less helpful over time in distinguishing infants who will either die in the NICU or survive with low Mental Developmental Index/Psychomotor Developmental Index scores. Serial caretaker intuitions of die before discharge also fail to identify prospective nonsurviving infants. However, corroborated intuitions of die before discharge identify a subset of infants whose likelihood of surviving to 2 years with both MDI and PDI >80 is approximately 4%.

Key Words: neonatal epidemiology • informed consent • neonatal ethics • illness severity scores • prognostication • Score for Neonatal Acute Physiology II

Abbreviations: DOL—day of life • SNAP-II—Score for Neonatal Acute Physiology II • SNAPPE-II—Score for Neonatal Acute Physiology II Perinatal Extension • PPV—positive predictive value • NPV—negative predictive value • BW—birth weight • MDI—Mental Developmental Index • PDI—Psychomotor Developmental Index

The most widely recognized epidemiological observation in neonatal intensive care is birth weight (BW)-specific mortality rates, that is, the observation that small, low-gestation infants are less likely to survive than larger, more-mature infants. Above ~1.5 pounds and ~25 weeks of gestation, virtually all NICUs currently have, and have had for ≥1 decade, survival rates of ≥80%.1–6 This level of success precludes discretionary parental refusal of intervention in the absence of other, non–BW-related circumstances. At the other end of a relatively narrow spectrum, below ~1 pound and ~23 weeks, survival rates for appropriate-for-gestational age infants are dismal.1–6 In a recent survey, a large majority of US neonatologists stated that they routinely resuscitate all infants with BWs of >700 g and gestational ages of >25 weeks, whereas they do not attempt delivery room resuscitation of infants of <450 g and <23 weeks.7 For infants with intermediate BW/gestational age, it generally is represented that the most appropriate decision-making process is one of negotiation between physician caretakers and family surrogates, acting in the best interests of the infant patient.8–14

However widely accepted it may be, this formulation is obviously incomplete. What is left out is time. It has been recognized for ~1 decade that, for NICU patient populations as a whole, BW-specific mortality rates change radically over just a few days.15 Because time of death for the sickest doomed infants is so short, infants who had a very low likelihood of survival at birth have a much brighter outlook for discharge to home after only a few days of NICU care.

Prognostications of nonsurvival, whether derived from algorithms or from clinical intuitions, have been studied in ICU settings from elderly patients to newborn patients.16–28 The vast majority of these studies obtained assessments at a single time point (admission to the ICU), to calibrate the severity of illness of the overall ICU population. These efforts have proved remarkably useful for addressing questions of public policy and quality control across ICUs.

As an attempt to inform ethical decision-making for individual patients, however, these efforts have been less useful. Most patients, especially newborns, are admitted to an ICU for a trial of therapy, tacitly or explicitly agreed to by both the physician caretakers and patient surrogates. After a relatively brief period of intensive care, it is expected that the patients will "declare themselves." This declaration is a metaphor for the purported increase in accuracy of prognostic estimates as a function of length of ICU stay. In this view, the ethical appropriateness of continuing NICU support would be revisited periodically, illuminated ever more brightly by ongoing revision of increasingly accurate prognostications.

Unfortunately, few studies have observed individual patients longitudinally, assessing prospectively their subsequent likelihood of dying and the power of various purported prognostic indices over time.15,29,30 No study of which we are aware has examined serial predictions of subsequent morbidity in a group of extremely premature infants. We report both types of findings here.

We hypothesized that infants who were most likely to die or to survive with severe impairments would declare themselves, separating from their unscathed-to-be peers with every passing day. We tested this hypothesis by correlating both clinical intuitions obtained daily from attending physicians, fellows, residents, neonatal nurse practitioners, and registered nurses and serial assessments of 2 neonatal illness severity scores (Score for Neonatal Acute Physiology II [SNAP-II] and SNAP-II Perinatal Extension [SNAPPE-II]) with clinical outcomes at corrected age of 2 years for a cohort of extremely premature infants treated in our NICU.

Finally, we recognized that, from an ethical perspective, time is a bounded dimension and the only ethically relevant withdrawal in the NICU is ventilation. That is, except in rare circumstances, once an infant no longer needs the assistance of a mechanical ventilator, survival to NICU discharge is ensured. Consequently, we report data obtained for individual infants only during the time they required mechanical ventilation.


   METHODS
TOP
 ABSTRACT
 METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Patient Population
We identified 268 premature infants admitted to our NICU who required mechanical ventilation between 1999 and 2004 and were enrolled in 1 of 2 ongoing clinical studies, that is, the Nitric Oxide/Ventilatory Assistance study31,32 or Extremely Low Gestational Age Newborn study.33 We restricted our analysis to these patients because follow-up monitoring after NICU discharge would be facilitated by ongoing interactions between patients, families, and study coordinators. We designed our study to gather prognostic data (illness severity scores and caretaker intuitions) only during days in which the patients required mechanical ventilation. Infants who required nasal continuous positive airway pressure but not mechanical ventilation were excluded from this analysis. For infants who were reintubated after initial extubation, assessments were resumed on the day of reintubation. Each infant contributed only once to an outcome, no matter how many times the infant was placed on a ventilator.

The study population reported here comprised the large majority of extremely low gestation infants cared for in our NICU during this time period (rates of enrollment in both studies were >90%). The infants who were not enrolled generally were extremely small and sick and were not expected to survive. For the purposes of analyses of the impact of the passage of time on prognostic power, these infants are less relevant because they died so quickly.

Intuitions Protocol
For 268 patients, on each day of mechanical ventilation we asked nurses, neonatal nurse practitioners, residents, fellows, and attending physicians the following question: "Do you think this child is going to live to go home or die before hospital discharge?" Each respondent was allowed to answer "live," "die," or "uncertain" (if she or he could not answer comfortably on that day). Respondents were polled individually and privately, in an attempt to minimize the influence of other respondents on the opinions offered. The investigators spent several hours each day in the NICU, attempting to obtain, at a minimum, the opinions of the primary nurse and primary medical resident, as well as the attending physician and/or fellow, for each eligible infant.

SNAP-II Scoring
For 166 ventilated infants, we calculated modified SNAP-II and SNAPPE-II values on day of life (DOL) 0.5, 1, 3, 7, 14, 21, and 28 and weekly thereafter until either death or extubation.18,19,28 There were no significant differences in any subsequent analyses comparing SNAP-II and SNAPPE-II data. Consequently, SNAP-II data are reported in this article. SNAPPE-II analyses are available from the authors on request (Dr Meadow).

Follow-up Neurologic Assessment of Surviving Infants
Forty-two (16%) of 268 infants died before 2 years of age; 192 (85%) of the 226 surviving infants were assessed in our NICU follow-up program at corrected age of 24 months, by using the Bayley Scales of Infant Development Mental Developmental Index (MDI) and Psychomotor Developmental Index (PDI).34 The author responsible for assessing MDI and PDI scores was blinded with respect to both the SNAP-II values and clinical intuitions recorded during the NICU stay.

Statistical Analyses
Comparisons of SNAP-II and SNAPPE-II values over time, for both surviving and nonsurviving populations, were performed by using repeated-measures analysis of variance or linear regression analysis. Other comparisons between surviving and nonsurviving populations were performed with Student's t tests and {chi}2 tests. Positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity were calculated in the usual manner. For all evaluations, statistical significance was accepted at a value of P < .05. This study was approved by the institutional review board of the University of Chicago.


   RESULTS
TOP
 ABSTRACT
 METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Patient Demographic Features
Table 1 presents patient demographic data for the 268 ventilated NICU patients who constituted our study population. The median BW was 833 g, and the median gestational age was 26 weeks; 56% of our study patients were male, 67% were black, 18% were white, and 16% were Hispanic. A total of 226 (84%) of 268 infants survived, and 42 (16%) died (39 in the NICU and 3 after NICU discharge). Not surprisingly, NICU survivors had greater BWs and longer gestations than did nonsurvivors (median BW: 870 vs 684 g; median gestation: 26 vs 24 weeks; both P < .001). The average length of stay for NICU survivors was 91 days, which was significantly longer than the average length of stay for nonsurvivors (27 days; P < .001).


View this table:
[in this window]
[in a new window]

  		TABLE 1 Demographic Data for NICU Patients and Their Caretakers’ Intuitions

 
Serial Intuitions as Predictors of Death
Table 1 also reveals that 17066 intuition profiles were obtained for 268 ventilated infants on 5609 days of mechanical ventilation. The 39 NICU nonsurvivors received 1603 intuitions on 516 ventilator days. The 229 NICU survivors received 15463 intuitions on 5093 ventilator days. There was no significant difference in the number of daily intuitions obtained for nonsurvivors (3.1 intuitions per day) versus survivors (3.0 intuitions per day).

Intuition Profiles for Survivors
Intuition profiles for ventilated NICU survivors reflected 2 distinct hospital courses. A total of 162 (71%) of 229 survivor profiles were characterized by 100% accurate intuition of survival by every medical caretaker on every ventilator day. Alternatively, 67 NICU survivors (29%) had ≥1 hospital day characterized by an intuition of die before discharge. Indeed, 18 infants (8% of survivors) survived despite having ≥1 hospital day on which all respondents predicted death.

Intuition Profiles for Nonsurvivors
Intuition profiles for ventilated NICU nonsurvivors also reflected 2 distinct hospital courses. Seven (18%) of the 39 nonsurvivors had every day of their life characterized by an intuition of die before discharge. These 7 infants died early (average time of death: 8.7 days). In contrast, 32 (82%) of 39 profiled nonsurvivors had ≥1 day characterized by 100% intuition of survival (median number of 100% live days: 6 days). These infants died much later (average time of death: 21 days; P < .01).

Predictive Power of Intuitions of Die Before Discharge for Death in the NICU
A total of 100 of 268 profiled infants had ≥1 intuition of die before discharge. The predictive power of these caretaker intuitions for death was dismal; only 33 infants (33%) with ≥1 day of an intuition of die died actually died in the NICU. The power of increasingly stringent caretaker intuitions of die before discharge was not much better; 55 infants had ≥1 day of corroborated intuition of die (that is, >1 caretaker predicted die on ≥1 day of ventilation), and only 24 (44%) of those 55 actually died in the NICU. Forty-three infants had ≥1 day on which every caretaker predicted die before discharge, but only 24 (56%) of those 43 infants died.

Predictive Power of Intuitions of Die Before Discharge as Function of Length of NICU Stay
Additional insight into the lack of predictive power of caretaker intuitions of nonsurvival for subsequent death in the NICU can be derived by determining the PPV of a single day of 100% intuition of demise as a function of the DOL on which the caretaker intuitions were obtained. During the first 10 DOLs, 11 (55%) of 20 infants who had ≥1 day of unanimous intuition of die in the NICU died, which was not significantly different from the value for infants whose first day of unanimous intuition of die came after DOL 10 (13 of 23 infants; 56% died).

Serial Intuitions as Predictors of Morbidity
Forty-two of the 268 infants for whom caretaker intuitions were obtained in the NICU died before corrected age of 2 years (39 died in the NICU and 3 died after NICU discharge). Follow-up assessments of MDI and PDI scores at corrected age of 2 years were available for 192 (85%) of the 226 infants who survived. The 34 NICU survivors who were lost to follow-up monitoring were slightly heavier and more mature at birth (BW: 1078 vs 886 g; gestation: 27.2 vs 26.3 weeks; both P < .05), compared with the infants for whom follow-up data were obtained. The average MDI score of the 192 evaluated NICU survivors was 78 ± 20; the PDI score was 86 ± 22. Seventy-nine infants (41% of evaluated NICU survivors) had either MDI or PDI scores of <70 at corrected age of 2 years; 68 infants (35% of evaluated NICU survivors) had both MDI and PDI scores of >79.

Figure 1 displays the correlations between death or neurologic outcome (displayed on the y axis) and caretaker intuitions obtained while the infant was still on mechanical ventilation, (displayed on the x axis) as a function of increasingly stringent caretaker intuitions of die before NICU discharge. A total of 145 infants were never predicted to die on any day by any caretaker; 88 (61%) were alive with MDI and PDI scores of >69 at corrected age of 2 years. Eighty-nine evaluated infants had ≥1 day with ≥1 intuition of death, and only 25 (28%) of those infants were alive with MDI and PDI scores of >69 at 2 years (P < .01, compared with infants with no caretaker intuitions of die). Increasingly stringent criteria for intuitions of die were even more powerfully predictive. Of the 48 evaluated infants with even 1 day of corroborated intuition of die before discharge, only 7 (14%) had both MDI and PDI scores of >69, and only 2 (4%) had both MDI and PDI scores of >79.


Figure 1
View larger version (40K):
[in this window]
[in a new window]

  		FIGURE 1 Outcomes at corrected age of 2 years versus caretaker intuitions of "die before discharge" while the infant required mechanical ventilation in the NCIU for 234 patients. Values indicate MDI/PDI scores.

 
Table 2 elaborates on these observations by presenting the PPVs, NPVs, sensitivities, and specificities of 3 different intuition criteria in the NICU for 4 distinct outcomes at 2 years of age. Several points can be noted in Table 2. The PPVs of these criteria for the outcome of death were not good (60% at best), but they improved progressively as the outcome criteria were broadened to include infants who survived with higher MDI/PDI scores. Conversely, intuitions of die before discharge had relatively good NPVs for death in the NICU (attributable in part to the low prevalence of NICU death) but had poor NPVs for survival with low MDI/PDI scores. As revealed in Fig 1 and Table 2, 145 infants never had even 1 intuition of die before discharge, and only 88 (61%) of those 145 infants were alive with MDI and PDI scores of >69 at 2 years. Moreover, increasingly stringent caretaker intuitions of die had little effect on the NPVs. A total of 186 infants never had a corroborated intuition of die before discharge, and only 106 (57%) of those 186 infants were alive with MDI and PDI scores of >69 at 2 years.


View this table:
[in this window]
[in a new window]

  		TABLE 2 Predictive Power of Caretaker Intuitions of Die Before NICU Discharge

 
Serial SNAP-II Values as Predictors of Death
A total of 2145 SNAP-II values (and an equal number of SNAPPE-II values) were determined for 162 infants; 163 SNAP-II values were determined for 27 nonsurvivors (6.0 ± 4.6 values per nonsurvivor), and 1982 SNAP-II values were determined for 135 NICU survivors (14.7 ± 6.0 values per survivor; P < .001). Figure 2 presents the average SNAP-II values as a function of DOL for survivors and nonsurvivors during the first 56 DOLs. Two important points emerge from Fig 2. On DOL 1, the SNAP-II value for nonsurvivors (SNAP-II: 38.2 ± 18.1) was significantly higher than the SNAP-II value for survivors (SNAP-II: 26.3 ± 12.7; P < .001). However, the difference between SNAP-II values for populations of survivors and nonsurvivors diminished steadily and significantly over time as SNAP-II values improved for both groups (P < .005 for SNAP-II values versus time for both groups; P < .015 for difference in SNAP-II values between nonsurvivors and survivors). Table 3 presents the PPVs, NPVs, sensitivities, and specificities of SNAP-II values equal to the 50th percentile and the 90th percentile for NICU nonsurvivors for the outcome of survival, as a function of DOL. For both SNAP-II criteria, the PPVs for nonsurvival decreased over time (P < .05).


Figure 2
View larger version (16K):
[in this window]
[in a new window]

  		FIGURE 2 SNAP-II scores versus DOL for survivors and nonsurvivors. Bars represent SD.

 

View this table:
[in this window]
[in a new window]

  		TABLE 3 Predictive Power of Median and 90th Percentile SNAP-II Values for NICU Nonsurvivors

 
Serial SNAP-II Values as Predictors of Morbidity
Figure 3 presents the average SNAP-II values as a function of DOL for infants who survived with both MDI and PDI scores of >69 at 2 years, compared with infants who either died or survived with either MDI or PDI scores of <70. Two important points emerge from Fig 3. On DOL 1, the SNAP-II value for survivors with MDI and PDI scores of >69 (SNAP-II: 31 ± 16) was significantly higher than the SNAP-II value for infants who either died or had MDI or PDI scores of <70 (SNAP-II: 26 ± 12; P < .05). However, the difference between SNAP-II values for the 2 groups decreased steadily and significantly over time as SNAP-II values improved for both groups (P < .01 for SNAP-II values versus time for both groups; P < .001 for difference in SNAP-II values between groups). This phenomenon (narrowing of the difference in SNAP-II values over time) was replicated at every outcome threshold of MDI and PDI scores (death plus MDI or PDI scores of <49, <59, <69, or <79).


Figure 3
View larger version (15K):
[in this window]
[in a new window]

  		FIGURE 3 SNAP-II scores versus DOL for infants who survived with both MDI and PDI scores of >69 at 2 years, compared with infants who either died or survived with either MDI or PDI scores of <70. Bars represent SD.

 
Table 3 presents the PPVs, NPVs, sensitivities, and specificities of SNAP-II values equal to the median and 90th percentile for NICU nonsurvivors for the combined outcome of death or MDI or PDI scores of <70, as a function of DOL. For both SNAP-II criteria, PPVs decreased significantly over time (P < .05). This phenomenon was replicated at every outcome threshold of MDI and PDI scores (death plus MDI or PDI scores of <49, <59, <69, or <79). Moreover, it is apparent from Table 3 that the NPVs of SNAP-II values (either median or 90th percentile for nonsurvivors) correlated very poorly with the combined outcome of either death or MDI or PDI score of <70. This phenomenon was also replicated for every MDI/PDI threshold.

NICU Resources Allocated as Function of Survival or MDI/PDI Scores
NICU resources were disproportionately devoted to survivors, as opposed to infants who died before discharge. NICU nonsurvivors accounted for 15% of study patients (39 of 268 patients) but only 5% of NICU bed-days (1046 of 21826 bed-days). However, the proportion of bed-days occupied by infants who would either die or survive with MDI or PDI scores of <70 was 10-fold greater than the proportion of bed-days occupied by nonsurvivors (9638 of 19089 bed-days; 50%).


   DISCUSSION
TOP
 ABSTRACT
 METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
It is often claimed that admission to a NICU acts as a trial of therapy for a sick newborn infant, whose prognosis will declare itself over time. At one level this is obviously true. The outcome for infants who have died is certain and does not change over time. Here we addressed the issue of the limits of our prognostic ability for infants who remain alive in the NICU. Moreover, in an attempt to restrict ourselves to ethically relevant questions, we limited our study population to infants who continued to require mechanical ventilation to remain alive in the NICU. For these infants, an alternative to continued NICU intervention exists, that is, withdrawal of mechanical ventilation. For some infants (ie, infants who are not dead yet but who eventually will die in the NICU and infants who will survive with severe neurologic disability), it is at least conceivable that prolonged, painful, medical interventions in the NICU are not in their best interests. We designed the current studies to investigate whether either serial intuitions or serial illness severity assessments would help us identify such infants during (as opposed to after) the time when ethical options could be exercised. The answer, in brief, is, "it depends on what you're asking."

We emphasize 4 points in detail. (1) SNAP-II values were progressively less helpful over time as predictors of death. (2) SNAP-II values were unhelpful at all times in distinguishing infants who would either die or survive with low MDI/PDI scores from those who would survive with better neurologic outcomes. (3) Intuitions of die before discharge were unhelpful in distinguishing infants who would die in the NICU from those who would survive until discharge. (4) Intuitions of die before discharge were very good, although not perfect, in identifying infants who would either die or have some neurologic impairment at 2 years. We take up each of these findings in turn.

First, as expected, SNAP-II values on DOL 1 were significantly higher for nonsurvivors than for survivors. This observation confirms previous research on illness severity scoring in newborn infants.18,19 However, there were always so many more survivors than nonsurvivors on every day of mechanical ventilation that the PPV of the median SNAP-II value for the nonsurviving population was never higher than 0.6, even on DOL 1. More importantly for the concept of a trial of therapy in the NICU, it was not clear to us (or in the literature) whether serial SNAP-II values would distinguish a population of nonsurvivors from a population of survivors with increasing clarity. It is clear now that they do not.

Second, we present what is, to our knowledge, the first description of serial illness severity scores as predictors of neurologic morbidity among ventilated NICU patients. Unfortunately, here too the message is clear; individual infants destined for either death in the NICU or survival with low Bayley Scale scores at corrected age of 2 years are not easily distinguished from infants who will survive with higher Bayley Scale scores, at least while they are undergoing ventilation in the NICU. The difference in SNAP-II values between these 2 groups narrowed, rather than widened, with the passage of time.

Third, intuitions of impending death, whether isolated, corroborated, or unanimous, whether offered by inexperienced residents or experienced neonatal nurse practitioners and attending physicians, were dismal predictors of death in the NICU. They were wrong approximately one half of the time.

Fourth, caretaker intuitions of die before discharge strongly indicated the prognosis for the combined outcome of death or survival with neurologic morbidity. For this study population overall, 47% of infants survived with MDI and PDI scores of >69; however, if even 1 ventilator day was characterized by a corroborated intuition of die in the NICU, then this number decreased to 15%. Even more striking, 31% of our study infants survived with both MDI and PDI scores of >79; however, if even 1 day was characterized by a corroborated intuition of die in the NICU, then this number decreased to 4%.

Many previous reports of neurologic conditions of NICU survivors at corrected age of 2 years used a MDI or PDI cutoff value of <70 (2 SDs below the mean) as a proxy for a permanent neurologic impairment or burden.31,32,35,36 However, reports by Hack et al37 and Patrianakos et al38 suggested that neurologic improvement of these premature infants after 2 years of age is not uncommon and the use of arbitrary MDI/PDI thresholds may label infants inappropriately and pejoratively. In response to these concerns, we reported our data here by using MDI/PDI values of <50, <60, <70, and <80, allowing for a richer interpretation of the admittedly incomplete neurologic assessment data available at corrected age of 2 years.

Several methodologic concerns in our study, and their implications, can be addressed directly. We recognize that 1 test of the generalizability of our observations involves the comparability of our NICU outcomes to those in other centers. We previously published extensive descriptions of our ventilation practices, our use of prenatal corticosteroid therapy, and the outcomes of ventilated premature infants in our NICU, demonstrating that rates of mechanical ventilation, as well as BW-specific mortality, intraventricular hemorrhage, necrotizing enterocolitis, bronchopulmonary dysplasia, and retinopathy of prematurity rates and rates of neurologic morbidity at 2 years of age, compared favorably with those reported elsewhere.31,32

Many ethicists have argued that withholding/withdrawing feeding is ethically comparable to withholding/withdrawing mechanical ventilation. However, most neonatologists and most parents disagree. In several studies, withholding/withdrawing mechanical ventilation was not-uncommonly practiced in most NICUs, whereas withholding/withdrawing feeding was rare and viewed as much more problematic.8,9,13,35,36,39,40 This observation spurred us to adopt a study design that centered on the possibility of withdrawal of ventilation, and not feeding, as a relevant parental option for certain NICU patients.

Many previous studies demonstrated that illness severity assessed late in an infant's NICU course could predict successfully a burdensome versus nonburdensome outcome. The presence of one or a combination of intraventricular hemorrhage, periventricular leukomalacia, retinopathy of prematurity, bronchopulmonary dysplasia, or spasticity at the time of NICU discharge was noted to be prognostically ominous.4,5,13,35,36 Unfortunately, from the perspective of parents, these caretaker intuitions are often too late. It does parents of children with these diagnoses little good to be concerned at the time of discharge; they are already quite likely to be concerned. Once NICU patients are successfully extubated, few ethically relevant options exist. We attempted to establish an algorithm to help parents while they still had the option of withdrawing ventilation.

Might our intuition data have reflected a self-fulfilling prophecy; that is, once nonsurvival was predicted, was the balance of NICU care biased to produce the demise of the infant? We consider this unlikely, for a number of reasons. First, clinical prognostication of nonsurvival was merely the first step in a sequence of actions that might have led to limitations of treatment. On occasion, these caretaker intuitions led to discussion with parents and other family members about decisions to withhold or to withdraw treatment. However, only rarely did those discussions lead to consensus to withdraw support. We noted previously that, in our NICU, very few infants who are in physiologically stable condition are extubated to die because of poor neurologic prognosis.39 These decisions reflect the expressed wishes of our parent population (primarily black and devoutly religious). Consequently, the deaths noted here reflect infants who were physiologically failing and not infants who were in physiologically stable condition with problematic neurologic prognoses.

We cannot rule out more-subtle effects of such self-fulfilling prophecies for our infants, analogous to the reduction in therapeutic interventions that has been reported for adults with do-not-resuscitate status.21,41 However, we note that any such effects would, by definition, have tended to make prognostication of nonsurvival more accurate, rather than less so; that is, the self-fulfilling prophesy hypothesis would have made patients who were predicted to die more likely to succumb after their caretaker intuitions had been made explicit. The possibility of any such effects biases against the observations noted here and serves to emphasize our conclusion that prognostication of nonsurvival is fraught with uncertainty.

Finally, we must note the implications of our observations of resource allocation in the NICU. We report here, as we and others noted previously,42–45 that ~95% of NICU bed-days (an excellent proxy for NICU resource expenditure) are directed toward infants who survive, as opposed to those who succumb in the NICU. This observation holds true even if the infants are small enough and sick enough to require mechanical ventilation. However, we also report here that approximately one half of all NICU resources are devoted to infants who either will die or will survive with MDI or PDI scores of <70. We stress that this observation is independent of our prognostic measures and seems to reflect realities in current NICU care.


   CONCLUSIONS
TOP
 ABSTRACT
 METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
Using either neonatal illness severity algorithms or clinical intuitions, we could do no better than ~50% in predicting whether sick, ventilated, NICU patients would die in the NICU or survive to be discharged. In contrast, if even 1 NICU ventilation day was characterized by a corroborated intuition of die in the NICU, then the likelihood that the infant would survive with both MDI and PDI scores of >79 decreased to 4%.

We draw 2 final conclusions from these observations. First, these data should inform the ongoing discussions we have with the parents of our infant patients while the infants require mechanical ventilation. Second, to serve our patients and their parents, we must continue to search for better prognostic tools.


   FOOTNOTES
 
Accepted Aug 31, 2007.

Address correspondence to William Meadow, MD, PhD, Department of Pediatrics, MC 6060, University of Chicago, 5825 South Maryland Ave, Chicago, IL 60637. E-mail: wlm1{at}uchicago.edu

The authors have indicated they have no financial relationships relevant to this article to disclose.


What's Known on This Subject

Few studies have evaluated individual patients longitudinally, assessing prospectively their subsequent likelihood of dying or surviving with neurologic disability.

 

What This Study Adds

No study of which we are aware has examined serial predictions of subsequent death and morbidity in a group of extremely premature infants. We report such findings here.

 


   REFERENCES
TOP
 ABSTRACT
 METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 

  1. Hack M, Fanaroff AA. Outcomes of children of extremely low birth weight and gestational age in the 1990s. Early Hum Dev. 1999;53 (3):193 –218[CrossRef][Web of Science][Medline]
  2. Hack M, Fanaroff A. Outcomes of extremely immature infants: a perinatal dilemma. N Engl J Med. 1993;329 (22):1649 –1650[Free Full Text]
  3. Hack M, Friedman H, Fanaroff A. Outcomes of extremely low birth weight infants. Pediatrics. 1996;98 (5):931 –937[Abstract/Free Full Text]
  4. O'Shea TM, Klinepeter KL, Goldstein DJ, et al. Survival and developmental disability in infants with birth weights of 501 to 800 g born between 1979 and 1994. Pediatrics. 1997;100 (6):982 –986[Abstract/Free Full Text]
  5. Horbar JD, Badger GJ, Carpenter JH, et al. Trends in mortality and morbidity for very low birth weight infants, 1991–1999. Pediatrics. 2002;110 (1):143 –151[Abstract/Free Full Text]
  6. Lorenz JM. Survival of the extremely preterm infant in North America in the 1990s. Clin Perinatol. 2000;27 (2):255 –262[CrossRef][Web of Science][Medline]
  7. Singh J, Fanaroff J, Andrews B, et al. Resuscitation in the "gray zone" of viability: determining physician preferences and predicting infant outcomes. Pediatrics. 2007;120 (3):519 –526[Abstract/Free Full Text]
  8. McHaffie HE, Cuttini M, Brolz-voit G, et al. Withholding/withdrawing treatment from neonates. J Med Ethics. 1999;25 (6):440 –446[Abstract/Free Full Text]
  9. Rebagliato M, Cuttini M, Broggin L, et al. Neonatal end-of-life decision-making with self-reported practices in 10 European countries. JAMA. 2000;284 (19):2451 –2459[Abstract/Free Full Text]
  10. American Academy of Pediatrics, Committee on Bioethics. Guidelines on forgoing life-sustaining medical treatment. Pediatrics. 1994;93 (3):532 –536[Abstract/Free Full Text]
  11. Cuttini M, Kaminski M, Saracci R, de Vonderweid V. The EURONIC project: a European concerted action on information to parents and ethical decision-making in neonatal intensive care. Pediatr Perinat Epidemiol. 1997;11 (4):461 –474[CrossRef][Web of Science][Medline]
  12. Silverman WA. Overtreatment of neonates? A personal retrospective. Pediatrics. 1992;90 (6):971 –976[Abstract/Free Full Text]
  13. Lantos JD, Tyson JE, Allen A, et al. Withholding and withdrawing life sustaining treatment in neonatal intensive care: issues for the 1990s. Arch Dis Child Fetal Neonatal Ed. 1994;71 (3):F218 –F223[Free Full Text]
  14. Lantos JD, Meadow WL. Neonatal Bioethics: The Moral Challenges of Medical Innovation. Baltimore, MD: Johns Hopkins University Press; 2006
  15. 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 NICU. Pediatrics. 1996;97 (5):636 –643[Abstract/Free Full Text]
  16. Stevens SM, Richardson DK, Gray JE, Goldmann DA, McCormick MC. Estimating neonatal mortality risk: an analysis of clinicians’ judgments. Pediatrics. 1994;93 (6):945 –950[Abstract/Free Full Text]
  17. Lemeshow S, Klar J, Teres D, et al. Mortality probability models for patients in the intensive care unit for 48 or 72 hours: a prospective multicenter study. Crit Care Med. 1994;22 (9):1351 –1358[Web of Science][Medline]
  18. Richardson DK, Corcoran JD, Escobar GJ, et al. SNAP-II and SNAPPE-II: simplified newborn illness severity and mortality risk scores. J Pediatr. 2001;138 (1):92 –100[CrossRef][Web of Science][Medline]
  19. Richardson DK, Gray JE, McCormick MC, Workman K, Goldmann DA. Score for Neonatal Acute Physiology: a physiologic severity index for neonatal intensive care. Pediatrics. 1993;91 (3):617 –623[Abstract/Free Full Text]
  20. Kruse JA, Thill-Baharozia MC, Carlson RW. Comparison of clinical assessment with APACHE II for predicting mortality risk in patients admitted to a medical intensive care unit. JAMA. 1988;260 (12):1739 –1742[Abstract/Free Full Text]
  21. Marcin JP, Pollack MM, Patel KM, Ruttimann UE. Decision support issues using a physiology based score. Intensive Care Med. 1998;24 (12):1299 –1304[CrossRef][Web of Science][Medline]
  22. International Neonatal Network. The CRIB (Clinical Risk Index for Babies) score: a tool for assessing initial neonatal risk and comparing performance of neonatal intensive care units. Lancet. 1993;342 (8865):193 –198[CrossRef][Web of Science][Medline]
  23. Cheung P, Prasertsom W, Finer NN, et al. Rescue high frequency oscillatory ventilation for preterm infants: neurodevelopmental outcome and its prediction. Biol Neonate. 1997;71 (5):282 –291[Web of Science][Medline]
  24. Lefebvre F, Gregoire MC, Dubois J, et al. Nursery Neurobiologic Risk Score and outcome at 18 months. Acta Paediatr. 1998;87 (7):751 –757[CrossRef][Web of Science][Medline]
  25. Schmidt B, Davis P, Moddemann D, et al. Long-term effects of indomethacin prophylaxis in extremely-low-birth weight infants. N Engl J Med. 2001;344 (26):1966 –1972[Abstract/Free Full Text]
  26. Brazy JE, Goldstein RF, Oehler JM, et al. Nursery Neurobiologic Risk Score: levels of risk and relationships with non-medical factors. J Dev Behav Pediatr. 1993;14 (6):375 –380[Web of Science][Medline]
  27. Ambalavanan N, Nelson KG, Alexander G, et al. Prediction of neurologic morbidity in extremely low birth weight infants. J Perinatol. 2000;20 (8):496 –503[CrossRef][Medline]
  28. Meadow WL, Frain L, Ren Y, Lee G, Soneji S, Lantos JD. Serial assessment of mortality in the neonatal intensive care unit by algorithm and intuition: certainty, uncertainty, and informed consent. Pediatrics. 2002;109 (5):878 –886[Abstract/Free Full Text]
  29. Zimmerman JE, Wagner DP, Draper EA. Improving intensive care unit decisions: supplementing physician judgment with caretaker intuitions of next day risk for life support. Crit Care Med. 1994;22 (9):1373 –1384[Web of Science][Medline]
  30. Wagner DP, Knaus WA, Harrell FE. Daily prognostic estimates for critically ill adults in intensive care units. Crit Care Med. 1994;22 (9):1359 –1372[Web of Science][Medline]
  31. Mestan KK, Marks JD, Hecox K, Huo D, Schreiber MD. Neurodevelopmental outcomes of premature infants treated with inhaled nitric oxide. N Engl J Med. 2005;353 (1):23 –32[Abstract/Free Full Text]
  32. Schreiber MD, Gin-Mestan K, Marks JD, Huo D, Lee G, Srisuparp P. Inhaled nitric oxide in premature infants with the respiratory distress syndrome. N Engl J Med. 2003;349 (22):2099 –2107[Abstract/Free Full Text]
  33. Laughon M, Bose C, Allred E, et al. Factors associated with treatment for hypotension in extremely low gestational age newborns during the first postnatal week. Pediatrics. 2007;119 (2):273 –280[Abstract/Free Full Text]
  34. Bayley N. Manual for the Bayley Scales of Infant Development. 2nd ed. San Antonio, TX: Psychological Corp; 1993
  35. Wood NS, Marlow N, Costeloe K, Gibson AT, Wilkinson AR. Neurologic and developmental disability after extremely preterm birth. N Engl J Med. 2000;343 (6):378 –390[Abstract/Free Full Text]
  36. Vohr BR, Wright LL, Dusick AM, et al. Neurodevelopmental and functional outcomes of extremely low birth weight infants in the National Institute of Child Health and Human Development Research Network 1993–1994. Pediatrics. 2000;105 (6):1216 –1226[Abstract/Free Full Text]
  37. Hack M, Taylor HG, Drotar D, et al. Poor predictive validity of the Bayley Scales of Infant Development for cognitive function of extremely low birth weight children at school age. Pediatrics. 2005;116 (2):333 –341[Abstract/Free Full Text]
  38. Patrianakos A, Lee K, Gin-Mestan K, Caldarelli L, Schreiber M. Correlation between neurologic performance at two years with school readiness. Presented at the annual meeting of the Pediatric Academic Society; Toronto, Ontario, Canada, May 8, 2007
  39. Singh J, Lantos JD, Meadow W. End-of-life after birth: death and dying in a NICU. Pediatrics. 2004;114 (6):1620 –1626[Abstract/Free Full Text]
  40. de Leeuw R, de Beaufort A, de Kleine MJ, van Harrewijn K, Kollee LA. Foregoing intensive care treatment in newborn infants with extremely poor prognoses. J Pediatr. 1996;129 (5):661 –666[CrossRef][Web of Science][Medline]
  41. Elstein AS, Christensen C, Cottrell JJ, Polson A, Ng M. Effects of prognosis, perceived benefit, and decision style on decision making in critical care. Crit Care Med. 1999;27 (1):58 –65[CrossRef][Web of Science][Medline]
  42. Lantos J, Mokalla M, Meadow W. Resource allocation in neonatal and medical ICUs: epidemiology and rationing at the extremes of life. Am J Respir Crit Care Med. 1997;156 (1):185 –189[Abstract/Free Full Text]
  43. Meadow WL, Lantos JD, Reimschisel T, Mokalla M. Distributive justice across generations: epidemiology of ICU care for the very young and the very old. Clin Perinatol. 1996;23 :597 –608[Web of Science][Medline]
  44. Tyson JE, Younes N, Verter J, et al. Viability, morbidity, and resource use among newborns of 501- to 800-g birth weight. JAMA. 1996;276 (20):1645 –1651[Abstract/Free Full Text]
  45. Stolz JW, McCormick MC. Restricting access to neonatal intensive care: effect on mortality and economic savings. Pediatrics. 1998;101 (3):344 –348[Abstract/Free Full Text]
PEDIATRICS (ISSN 1098-4275). ©2008 by the American Academy of Pediatrics

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us   Add to Digg Digg   Add to Facebook Facebook   Add to Reddit Reddit   Add to Technorati Technorati   Add to Twitter Twitter    What's this?


This article has been cited by other articles:


Home page
 PediatricsHome page
W. Meadow and J. Lantos
Moral Reflections on Neonatal Intensive Care
Pediatrics, February 1, 2009; 123(2): 595 - 597.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow E-mail this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My File Cabinet
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Meadow, W.
Right arrow Articles by Schreiber, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Meadow, W.
Right arrow Articles by Schreiber, M.
Related Collections
Right arrow Premature & Newborn
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Facebook   Add to Reddit   Add to Technorati   Add to Twitter  
What's this?