Objective. We sought to determine the contribution of withholding or withdrawing care to neonatal mortality in our hospital over a 10-year period from 1993 through 2002.
Background. Although not initiating or withholding intensive care for certain newborns has been practiced openly for the past 25 years, little information has been published concerning the results of these practices on neonatal mortality.
Design and Methods. All infants who were born in our hospital and remained in the hospital until their death were evaluated for the care they received at the time of birth and near the time of their death. The contribution of not initiating care or withdrawing care to the neonatal mortality rates in our hospital was determined. Information was obtained directly from the patients' charts as well as the neonatal database and monthly neonatal mortality and morbidity review. Other information, if needed, was obtained from the monthly ethics committee reviews of all nursery deaths. Information was collected relating to birth weight, gestational age, diagnosis, time of death, and year of death. Hospital and neonatal unit protocols were evaluated to determine how closely they were followed.
Results. During the 10-year period, 380 deaths (0.8%) of a total of 47820 live births occurred in our hospital. Care was not initiated or was withdrawn in close to 72% of those deaths; total care until death occurred in 28%. Total care for infants who died over the 10-year period decreased markedly as care not initiated or care withdrawn increased. Most of this increase in not initiating care and in withdrawal of care was in the smaller of the extremely low birth weight infants.
Conclusions. The majority of nursery deaths of infants born in our hospital occurred as the result of selected noninitiating of care or as a result of withdrawing care in infants not responding or considered to have a futile outcome. Only slightly more than one quarter of the infants received total care until the time of death
Infant and neonatal mortality rates (NMRs) are published annually by the Centers for Disease Control and Prevention in the United States and by the World Health Organization for countries that report these data, and each state and hospital usually also publish their own data. There has been remarkable improvement in the US NMRs in the past 40 years.1 These data show that neonatal mortality contributes the largest percentage to all childhood mortality. It is also clear that the smallest infants contribute the greatest percentage to neonatal mortality. These deaths are grouped into categories with congenital anomalies, prematurity, respiratory distress, and problems surrounding birth as the leading causes. Evidence from death certificates and national statistical data show that prematurity and congenital anomalies account for the largest percentage of neonatal deaths.1 Not often reported is the contribution to mortality of not initiating care or withdrawing care from these infants before death.
In this 10-year study in a large tertiary hospital with primarily high-risk patients, we determined the type of care received at the time of birth and death, the causes of death, and whether care was withdrawn before death.
DESIGN AND METHODS
All infants born at our hospital during the years 1993–2002 and who subsequently died before hospital discharge were evaluated for the cause of death and whether the infant did not have care initiated (CNI), had care withdrawn (CW), or received total care (TC) until the time of death. Outborn patients were not included, because we received <3 per year who expired during the time period.
CNI was the method of care determined in advance for infants with known lethal congenital anomalies or for infants who were considered to be so extremely immature as to have virtually no chance for survival or who would likely have a life with significant and unacceptable permanent handicap.
Care was withdrawn from infants who initially were resuscitated and given a chance to determine if they would respond but who did not improve. Infants who had lived for some time and were dying of complications of prematurity also had care withdrawn.
All infants with CNI or CW were given comfort care until they died. This care consisted of warmth, cleanliness, and oral feedings if applicable. Parents were encouraged to hold their infants when it was appropriate.
TC was designated for infants in whom no decline in care level occurred before death.
Type of care was compared according to birth weight (BW), gestational age (GA), age at death, broad categories of death, and year of death.
Causes of death were divided into 4 major groups: (1) extreme prematurity, which consisted of infants with a BW of <500 g and a few in the 500- to 600-g range for whom no other cause of death could be determined; (2) lethal congenital anomalies; (3) prematurity and all its complications including respiratory, infectious, intraventricular hemorrhage, and necrotizing enterocolitis; and (4) other, which represented infants with asphyxia, hydrops, and rare conditions that did not fit into the other categories.
For the first half of the time period that we evaluated, infants were routinely resuscitated if their BW was at least 450 g or GA was at least 23 weeks. During the last half of the time period, the policy was changed to not routinely resuscitate infants of <500 g. Between 500 and 599 g and at 23 weeks' gestation was considered to be a “gray zone” in which the parents' wishes were taken into consideration after counseling regarding our own and national outcomes. At 600 g and at 24 weeks' gestation, infants were resuscitated routinely.
Data were collected from the neonatal database and monthly mortality and morbidity reports that were kept by the neonatal department, as well as from patient medical charts if necessary. These data included notes of the status at the time of death and whether care was withdrawn. Autopsy reports, if available, were evaluated for final cause of death. Yearly neonatal and infant mortality rates were reviewed to determine if they were affected by the type of care.
All nursery deaths are routinely evaluated at a monthly mortality and morbidity conference attended by faculty and housestaff from both the newborn and obstetrical services. An ethics committee also reviewed these deaths at a monthly conference to monitor care and decision-making. All neonatal and infant patients born and cared for in our hospital until death during the study period were reviewed by the authors in a retrospective chart review. Prior decisions for most infants who did not undergo resuscitation at the time of birth because of known lethal congenital anomalies or extreme immaturity were made with the neonatal and obstetrical staff and the parents.
Parents were informed that if the situation was different than expected at the time of birth, the decision would be reevaluated. All parents, if time allowed before the birth, were given information regarding the possible complications and outcomes of infants of similar GA and BW in our own hospital. If the infant had known congenital anomalies, information was given to the parents regarding possible interventions and expected chances for survival and possibilities for interventional care. They then were given choices as to which type of care they wished their infant to receive, and in virtually all cases, their decision was honored. All of these decisions were recorded on the mother's chart. The information was also placed on an information board in the neonatal intensive care unit to alert the team to in-house imminent deliveries and the plan of care.
Complicated pregnancy conditions that might lead to a problem at delivery and cases of congenital anomaly were also discussed in a weekly conference between the obstetrical and neonatal teams if the delivery was not imminent. If there was no time for evaluation or consultation from the neonatal team before the time of birth, infants were resuscitated until the case was considered futile or it was obvious that it was a lethal congenital anomaly or a nonviable infant.
Permission to collect data for this project was approved by the Institutional Review board of the Los Angeles County-University of Southern California Medical Center.
During the 10-year period, there were 380 (0.8%) nursery deaths of 47820 live births. During this same period, yearly NMRs varied from a low of 4.7 to a high of 17.6 per 1000 live births (Table 1). When deaths of infants who were less <500 g in BW or had congenital defects were excluded, the NMR ranged from 0.9 to 4.6. Extreme prematurity (BW < 500 g) accounted for close to 32% of the total deaths, and congenital anomalies accounted for 36%, leaving all other causes such as complications of prematurity, sepsis, and asphyxia at just over 30%.
The type of care received for the entire period was CNI in 166 (44%) of the infants, CW in 106 (28%), and TC in 108 (28%). The type of care differed dramatically by BW and GA, with 98 (78%) of 125 deaths in those who had a BW of <500 g and 113 (71%) of 160 in those whose GA was <24 weeks not having care initiated at the time of birth. If withdrawal of care was added for the infants who had a BW of <500 g, 115 (92%) of 125 deaths in this BW category had either CNI or CW.
Figure 1 shows the percent of deaths according to type of care during 1993–2002. The year of death seemed to be the most significant of the various factors evaluated for the category of CNI and TC. During the first half of the time period, the number of infants who did not have care initiated was 70 (30%) of 234 compared with 96 (66%) of 146 during the last half of the time period, and the number who had TC until the time of death decreased from 88 (38%) of 234 in the first half of the time period to 20 (14%) of 146 in the last half. The combination of CNI and CW was 146 (62%) of 234 during the first half and 126 (86%) of 146 during the last half of the study.
Figures 2 and 3 show the type of care according to BW and GA. Of 125 infants with a BW of <500 g, 98 (78%) did not have care initiated compared with 26 (25%) of 105 of those with a BW of 500 to 999 g and 42 (28%) of 150 of those with a BW of ≥1000 g. One hundred thirteen (71%) of 160 infant deaths at <24 weeks' GA, 5 (14%) of 36 deaths at 24 to 25 weeks' GA, and 48 (26%) of 184 deaths at ≥26 weeks' GA did not have care initiated. Infants of <24 weeks' GA were not resuscitated because of extreme immaturity or condition at birth, and the cause of not resuscitating infants of ≥26 weeks' GA was usually a known lethal congenital anomaly. Figure 4 shows that 148 (78%) of 189 infants who died at <4 hours of age and 15 (26%) of 58 infants who died between 4 and 48 hours of age but only 3 (2%) of 133 of those who lived for >48 hours did not have care initiated.
The breakdown of type of care as related to the general categories of causes of death is shown in Fig. 5. Among the entire group, extreme prematurity (100 [60%] of 166) was the leading reason that care was not initiated. Congenital anomaly (45 [42%] of 106) was the main reason that care was withdrawn. The infants who had complications of prematurity and also those with congenital anomalies each represented 39 (36%) of 108 of those receiving TC until death. The number of congenital anomalies was almost equal in all 3 groups and represented 33% of those for whom care was not initiated, 42% of those for whom care was withdrawn, and 36% who received TC until death.
NMR is considered to be a marker for quality of care for countries, states, and hospitals. The World Health Organization and individual countries publish both neonatal and infant mortality rates that are often used as a marker for good maternal and infant care for that country. For the past several decades, the United States has improved its NMR each year through 2002 and met 1 of the goals in the objectives for the year 2000 in the area of decreasing infant mortality.1 NMR variability among hospitals and states in the United States has been attributed to various causes including level of intensive care, poor neonatal or maternal care, disparities of race and socioeconomic status, and lack of certain medical interventions or resuscitation skills. Hospitals that show an increase in their NMR could theoretically be considered by various regulatory groups to be failing in certain goals set by state or national organizations if numbers alone are used without evaluating the reasons for the difference. Our increase in NMR during the last 2 years of the time period, for example, partially reflects a marked decrease in total deliveries in our formerly very large delivery service while the number of high-risk deliveries remained unchanged. The total number of infants not being resuscitated increased because the percentage of infants with very low BW and very low GA increased and also because of the reinforcement of our policy to not resuscitate certain extremely low BW (ELBW) infants. The policy has been accepted and honored by all members of our team including obstetric and neonatal personnel and our bioethics committee.
Although not initiating care or withholding intensive care for certain newborns has been practiced for many years, little information has been published concerning the result of these practices on neonatal or infant mortality during the past 3 decades. The reason for this may include a lack of information on birth and death certificates, the reluctance of some hospitals and physicians to acknowledge their practices regarding resuscitation or withdrawal of care, or disagreement among various members of health care teams. There also may be a fear of governmental or other “watch-dog” groups that evaluate hospitals or neonatal units, such as the experience during the era of “Baby Doe” regulations. Although these regulations were struck down in 1986, the earlier controversies and the signs posted in each hospital advising any person to report newborns with handicaps who were not being treated engendered distrust of the system. This distrust may have lingered in some circles, although the outcry among physicians was generally responsible for abolishing those regulations and bringing life-and-death decisions back as the responsibility of the physicians and parents.
During the first half of our evaluation period, the decisions were made in relation to the improvement in neonatal intensive care, the improvement in ventilation techniques, the increased use of prenatal steroids for maturing the lungs and increased use of postnatal surfactant therapy, and from the literature, which showed a continuing decrease in NMR for ELBW infants. Our policy for resuscitation was to include infants down to 450 g in BW and 23 weeks' gestation. During the last half of our evaluation period, the decision to increase the weight limit for considering resuscitation to 500 to 600 g and the gestation to 23 weeks as a gray zone and to ≥24 weeks and/or 600 g for full resuscitation was based on the very poor outcome during the first half of the period of infants with a BW of <600 g at 2 years of age in our follow-up premature population.2 The change in policy resulting from the availability of this important information emphasizes the need for each unit to obtain follow-up data for its own infants. Decisions about viability then can be based on local performance and the actual odds faced by the infant. During both periods, parents were given information about survival statistics and common follow-up problems with premature infants. In the last half of the period, there was more emphasis on the complications and long-term follow-up based on our own data and those in the literature.
Although there are multiple studies and articles on the survival rates and outcomes of infants cared for in intensive care units, there is little information on how type of care has affected these rates. Duff and Campbell3 in 1973 were among the first to publish information about moral and ethical dilemmas in the special care nursery, including the withdrawal or withholding of care. In that early article, congenital anomalies were virtually the only basis for discontinuing care in 14% of infants who did not survive. In 1983, Dutch pediatricians began following all infants born in the country with a BW of <1500 g or GA of <32 weeks. In 2000, they published their results, which showed an overall low percentage (10%) of very poor outcomes by school age, but more than half had serious difficulties with behavior or learning disorders, especially in the very immature.4 This study led to the changed practice of not routinely resuscitating infants of <24 weeks' GA in a large neonatal unit in the Netherlands.
In a British Medical Journal comment in 2001 on the above-mentioned practice change in the Netherlands,5 it was reported that the Royal College of Paeditrics and Child Health in the United Kingdom established a policy in 1997 emphasizing that each case of withholding or withdrawing care had to be decided on its own merits. Lantos et al6 in 1994 presented the problems and economic and burdensome costs of sustaining treatment for life-threatening and potentially futile care. They concluded that the children's suffering was of primary importance and that costs and burden to family were also very important. They suggested that treatment decisions rest with parents and physicians. Partridge and Wall7 in 1997 reviewed their own experience at the University of California Hospital in San Francisco over a 3-year period from 1989 to 1992 regarding use of analgesics for infants when care was withheld or withdrawn. They found that 73% of deaths in their unit occurred as a result of selective nontreatment or withdrawal of care because of the perception of futility by the treating neonatologists. The reporting hospital is a university hospital in an academic center with a mixed public and private clientele in which the patients differ both racially and socioeconomically from our institution. Despite the differences, this rate was strikingly similar to our own (72%).
Evaluation of our own experience and this review of the limited published literature leads us to speculate that the basis for not resuscitating selected infants and the basis of withdrawing care for futile situations is the most common cause of nursery death and represents a large portion of the neonatal mortality experienced in the United States. It also leads us to believe that, in most situations, these practices are currently conducted with careful evaluation of each situation, collaboration with obstetrical colleagues and consultants in many subspecialty fields of pediatrics, with social worker and nursing staff input, and careful consideration of the parents' and caregivers' input and feelings. In none of the cases that we evaluated in our own unit was there a question regarding the validity of the decision; if there was disagreement, a mechanism was in place to resolve the differences. The final decision always favored extra time or care to give every benefit of the doubt to the infant in question.
It is important for each hospital or neonatal unit to develop policies for a plan of care for ELBW infants based on local and national standards and their own experience while keeping current on new methods that may be able to improve outcome in the future. Pediatric and neonatal physicians should consult with colleagues in obstetrics and with the parents before a delivery with risk factors for serious mortality or morbidity. The parents must have as much information as possible to participate in the decisions. To prevent any misuse or variance between type of care based on personal styles, there should also be frequent meetings and reviews including mortality and morbidity committees or an ethics review committee. Outcomes should be reviewed and policies adjusted based on new information.
With the advances in neonatal care, if newborns are mature enough and without lethal congenital defect, they most likely will survive. For there to be additional decreases in NMR in the United States, we will need to focus more on research for the prevention of prematurity, especially between 22 and 28 weeks of gestation, and additional research to prevent lethal congenital conditions, both of which contribute the greatest numbers to infant mortality rates.
- Accepted April 18, 2005.
- Address correspondence to Lorayne Barton, MD, MPH, Women's & Children's Hospital, Room L-919, 1240 N Mission Rd, Los Angeles, CA 90033. E-mail:
No conflict of interest declared.
- ↵Centers for Disease Control and Prevention, National Center for Health Statistics. Health, United States, 2004. Available at: www.cdc.gov/nchs/hus.htm. Accessed April 18, 2005
- ↵Sheldon T. Dutch doctors change policy on treating preterm babies. BMJ.2001;322 :1383
- ↵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 :F218– F223
- ↵Partridge JC, Wall SN. Analgesia for dying infants whose life support is withdrawn or withheld. Pediatrics.1997;99 :76– 79
- Copyright © 2005 by the American Academy of Pediatrics