A factor often ignored in analyses of neonatal mortality rate is the reporting practices of hospitals. This study examined hospital reporting practices by comparing hospitals where the reporting requirement of New York City Department of Health was followed and all births were reported regardless of gestational age and hospitals where gestational age was taken into account in the reporting.
In 2008, a survey was conducted among neonatologists in charge of neonatal intensive care units and obstetricians in charge of departments of obstetrics in New York City maternity hospitals. The survey collected information on hospitals’ definitions of live birth and fetal death, their resuscitation practices for extremely premature infants, and their reporting practices for live births and infant deaths.
Data from the 2008 survey and New York City’s linked live birth/neonatal death data sets for 2007-2009 were used to examine the impact of hospitals’ reporting practices on neonatal mortality rate. The neonatal mortality rate for hospitals where gestational age was taken into account was significantly lower than the hospitals where all live births were reported regardless of gestational age (2.68 versus 3.54, p=0.0033). Removal of infants born at less than 23 weeks gestation resulted in almost equal neonatal morality rates (2.53 versus 2.77, p=0.41).When the data was separated by race and insurance coverage, the difference in reporting practices resulted in significantly lower neonatal morality for births to white mothers (p=0.0032) and women covered by Medicaid (p=0.0005).
Neonatal Mortality Rate (NMR) is defined as the number of deaths in a year among infants who are less than 28 days old per 1,000 live births in the same year. In 2008, neonatal mortality rate in New York City (NYC) was 3. 6 [1], after decreasing 38% between 1997 and 2007 [2]. However, over the same period (1997-2007), prematurity-related infant deaths increased 32% in NYC.
Neonatal mortality rate is a function of gestational age. Based on the 2007-2010 linked live birth/infant death data set for the United States (US), the neonatal mortality rate for infants born at less than 24 weeks gestation was 16 times that of infants born at ≥ 24 weeks gestation [3]. Premature infants that survive have high morbidity rates, which include significant risk of brain injury [4]. For example, a longitudinal study of 4,004 infants born at 20-25 weeks gestation in the United Kingdom and Ireland from March to December 2005 found that only 811 infants survived to be admitted to the Neonatal Intensive Care Unit (NICU) [5]. Only 314 of the 811 infants survived to be discharged (7.8% of the original cohort). Of the discharged infants, a follow up at 30 months found disabilities of about 50% among those still alive. Of the 236 children followed to age six, 63 children classified at 30 months as having severe disabilities had moderate or severe disabilities at age six [6].
The morbidities associated with premature births have financial implications; the annual economic burden associated with premature births in the US in 2005 was at least $26.6 billion [7]. This represents a 357% increase from 2001, when a study conducted using hospital discharge data from the 2001 Nationwide Inpatient Sample estimated preterm birth/low birth weight cost at $5.8 billion [8]. Concerns about the high social and economic costs of premature births led the US Congress to pass the Prematurity Research Expansion and Education for Mothers who deliver Infants Early (PREEMIE) Act in 2006 [9]. In 2013 Congress reauthorized the PREEMIE Act through 2017 with revisions to enhance research and surveillance activities [10]. The Act authorizes the Centers for Disease Control and Prevention to conduct epidemiological studies on preterm births, conduct activities to improve national data to aid with surveillance, and bolster efforts aimed at preventing preterm births.
The gestational age at which aggressive costly therapies should be instituted for premature infants has been an ongoing ethical debate. The debate relates to the ethics of resuscitation, and the high costs associated with the care of extremely premature infants that are resuscitated. A 2006 report from the Nuffield Council on Bioethics, an independent United Kingdom organization, recommended that infants born at less than 22 weeks gestation should not be given intensive care and those born at 22 weeks should only be resuscitated after frank discussions with parents about survivability [11]. The European Resuscitation Council recommended that infants born at 25 or more weeks gestation should be resuscitated unless there are contra indications, and infants born at less than 23 weeks gestation and/or a birth weight of less than 400 grams should not be resuscitated [12]. The American Heart Association recommends that infants born at less than 23 weeks gestation and/or a birth weight of less than 400 grams should not be resuscitated and those born at 25 or more weeks gestation should almost always be resuscitated [13].
The extent of the impact of prematurity may not be evident in infant mortality data because the data are dependent on hospitals’ reporting practices, which may be influenced by definitions of live birth that differ from standard definitions used in health departments’ reporting requirements. If gestational age and/or viability are factors in the reporting of live births, this can significantly influence neonatal mortality data.
Almost all 57 US reporting areas have reporting requirements that include a definition of live birth, and all reporting areas require the reporting of live births regardless of gestational age [14]. The 2011 Revision of the Model State Vital Statistics Act and Regulations [15] defines live birth as follows:
The complete expulsion or extraction from the mother of a product of human conception, irrespective of the duration of pregnancy, which, after such expulsion or extraction, breathes or shows any other evidence of life, such as beating of the heart, pulsation of the umbilical cord, or definite movement of voluntary muscles, whether or not the umbilical cord has been cut or the placenta is attached. Heartbeats are to be distinguished from transient cardiac contractions; respirations are to be distinguished from fleeting respiratory efforts or gasps.
The Guidelines for Perinatal Care [16], which is published jointly by the American Academy of Pediatrics and the American College of Obstetricians and Gynecologists has the above definition with the underlined sentence omitted. The definition in the Guidelines for Perinatal Care is shared by most countries [17], and is in the reporting requirements of 41 reporting areas including NYC [14].
On the other hand, fetal death definitions and reporting requirements vary. Twelve reporting areas, including NYC, require the reporting of fetal death regardless of gestational age; 25 reporting areas require reporting at 20 or more weeks of gestation; and the other 20 areas have reporting requirements that includes gestational age and/or birth weight.
The aim of this study was to examine the statistical impact of differences in hospitals’ reporting practices on the neonatal mortality rate in NYC. This was done by combining the data from a survey conducted in 2008 with the NYC linked live birth/neonatal death data set for 2007-2009.
The study protocol was approved by the institutional review boards of both Binghamton University and New York City Department of Health and Mental Hygiene (NYCDOH). Data on hospital’s reporting practices were extracted from a survey conducted in 2008 [18]. The survey data was collected using a 15-minute telephone interview of hospital administrators. Administrators in obstetrics and neonatology were selected for the survey if they were responsible for hospital reporting policies, were involved in the training of new physicians regarding reporting policies and were involved in the care of extremely premature infants.
The survey collected detailed information regarding the hospitals’ resuscitation practices and policies. The survey also collected information on parental involvement in resuscitation decisions and the use of birth weight versus gestational age. The hospitals had similar resuscitation practices and policies, but there were differences in reporting practices that related to infants born at less than 23 weeks gestation. Based on the survey data, the hospitals were divided into two groups as outlined in table 1. Physicians who said gestational age was a component of their hospital’s criteria for live birth, but also indicated that they followed the NYCDOH requirement to report all live births regardless of gestational age were included in group 1.
Hospital Group |
Description |
Regional Perinatal Centers |
Level 3 Hospitals |
Level 2 Hospitals |
Group 1 (n=32)
|
Hospitals where the administrators provided hospital reporting practices that were in agreement with the reporting requirements of NYCDOH. These hospitals reported all live births regardless of gestational age. |
10 31.25%
|
19 59.38%
|
3 9.37%
|
Group 2 (n=6)
|
Hospitals where administrators reported gestational age as part of their hospital’s criteria for live birth reporting. The administrators in these hospitals reported that they did not follow the NYCDOH requirement to report all live births regardless of gestational age. |
1 16.67%
|
4 66.66%
|
1 116.67%
|
Table 1: Groups of hospitals surveyed.
Data from the linked live birth/neonatal death data sets for 2007-2009 were obtained from NYCDOH. The neonatal death file was used because NYCDOH would not include the post neonatal deaths due to concerns that these cases were potentially identifiable. The data sets included information on mother’s insurance coverage, mother’s race, and the infant’s gestational age at birth. The data sets from 2007-2009 were used to increase the number of data points for computing the neonatal mortality rates. Further, based on responses in the survey, 70% of the administrators interviewed had been in practice at their current hospital for at least six years.
Birthing locations that were not surveyed were removed from the data sets. This resulted in the exclusion of hospitals with no NICU, municipal hospitals, federal hospitals, birthing centers and births at home. The resulting data set contained 361,146 live births, which was 94% of all reported live births (93% of neonatal deaths) in NYC for the 3-year period.
New York State Department of Health classifies hospitals based on their level of neonatal care. Table 2, shows the types of maternity hospitals surveyed and the percent of physicians surveyed from each. More than 97% of the neonatologists surveyed were in charge of a NICU in either a regional perinatal center or a level three hospital; the ones most likely to care for extremely premature infants.
|
Neonatologists (%) (n=32) |
Obstetricians (%) (n=26) |
Response rate |
94 |
67 |
Type of hospital |
|
|
Public |
31 |
19 |
Voluntary |
69 |
81 |
Perinatal level of care of hospital |
|
|
Level 2 |
3 |
4 |
Level 3 |
66 |
65 |
Regional perinatal center |
31 |
31 |
Years in practice |
|
|
1-5 years |
0 |
8 |
6-10 years |
9 |
11 |
More than 10 years |
91 |
81 |
Years at current hospital |
|
|
1-5 years |
19 |
42 |
≥ 6 years |
81 |
58 |
Table 2: Characteristics of physicians surveyed.
The hospital of birth was the basis of the calculation of mortality rates. This was done to ensure that the mortality rates for the regional perinatal centers were not inflated by the inclusion of infants born at other facilities and later transferred.
For the hospitals where NYC reporting requirement was followed and all live births were reported regardless of gestational age (Group 1 hospitals) the NMR was 3.54 versus 2.68 for hospitals where gestational age was a factor in live birth reporting (Group 2 hospitals). This difference in NMR was statistically significant at 5% level of significance (p=0.0033). A two sample Z-test for proportions was used to illustrate the significant statistical difference. Based on the survey data, a gestational age of 23 weeks was the most consistent gestational age cut-off of the majority of physicians surveyed. For example, 97% of the neonatologists surveyed said they always resuscitate at gestational age of ≥ 23 weeks. Exclusion of infants born at less than 23 weeks gestation resulted in neonatal mortality rates of 2.77 and 2.53 for Group 1 and Group 2 hospitals respectively; a difference in NMR that was not statistically significant (Table 3). The response rate among the neonatologist was much higher than that of the obstetricians, with almost all neonatologists (94%) taking part in the survey. Use of only the responses from neonatologists resulted in an increase in the NMR among Group 1 hospitals (3.89 versus 2.68, p<0.0001).
|
Group 1 Hospitals (Births/Deaths) |
Group 2 Hospitals (Births/Deaths) |
Difference (95% Two-Sided Confidence Interval)
|
P Value |
Neonatal mortality rates when all neonatal deaths are considered |
3.55 (326,385/1,157)
|
2.68 (34,761/93)
|
0.87 (0.29, 1.45)
|
0.0033 |
Neonatal mortality rates when infants born at less than 23 weeks gestation are excluded |
2.77 (326,114/902)
|
2.53 (34,753/88)
|
0.24 (-0.32, 0.79)
|
0.41 |
Neonatal mortality rates when only the responses of neonatologists are used |
3.89 (273,690/1,064)
|
2.68 (34,761/93)
|
1.21 (0.62, 1.80)
|
<0.0001 |
Table 3: Differences in neonatal mortality rates between hospital groups.
Demographic factors such as race and income are related to both infant mortality and prematurity. In 2004, almost 50% of the deaths of infants of non-Hispanic black mothers were prematurity related [19]. The extent to which differences in reporting practices impact this high rate is not known. A study of infant mortality using data from New York City maternity hospitals estimated that if black mothers delivered in the same hospitals as white mothers there would be a 34.5% reduction in the black/white disparity in very low birth weight neonatal deaths [20].
In our study, the NMR among white mothers was 34% lower in Group 2 hospitals than in Group 1 hospitals (Table 4); a reduction that was statistically significant (p=0.0032). For infants born to black mothers the NMR was 19.4% lower in group 2 hospitals than in group 1 hospitals, but this difference was not statistically significant (p=0.1793). We used insurance coverage as an indicator of income level; with use of Medicaid serving as an indicator of poverty status. The NMR among mothers with Medicaid in Group 1 hospitals was 1.5 times that of those in Group 2 hospitals (p<0.001).
|
NMR Group 1 Hospitals (Births/Deaths) |
NMR Group 2 Hospitals (Births/Deaths) |
Ratio (Group 1 /Group 2) |
P Value |
Mother’s race |
Black
|
5.67 (91,053/516) |
4.57 (7,447/34) |
1.24 |
0.1793 |
White |
2.73 (182,276/497) |
1.80 (21,635/39) |
1.51 |
0.0032 |
Other |
2.71 (53,056/144) |
3.52 (5,679/20) |
0.77 |
0.3236 |
Mother’s insurance |
Medicaid |
3.81 (175,058/667) |
2.54 (22,796/58) |
1.5 |
0.0005 |
Non-Medicaid |
3.24 (151,327/490) |
2.93 (11,965/35) |
1.11 |
0.5498 |
Gestational age |
<23 weeks |
940.96 (271/255) |
625.00 (8/5) |
1.51 |
0.0658 |
23-27 weeks |
219.13 (2,154/472) |
242.42 (198/48) |
0.9 |
0.1914 |
28-36 weeks |
5.87 (29,491/173) |
7.21 (3,328/24) |
0.81 |
0.3799 |
>36 weeks |
0.65 (294,415/192) |
0.38 (31,220/12) |
1.7 |
0.026 |
Table 4: Impact of race, insurance coverage and gestational age on NMR.
The NMR for black mothers was significantly higher than white mothers within each group of hospitals (Table 5). At 9.61 deaths per 1,000 live births, the NMR of non-Medicaid black mothers in Group 2 hospitals was higher than any other group; this NMR was 6 times that of white mothers not covered by Medicaid (p<0.001) and 5 times that of white mothers covered by Medicaid.
|
Black NMR (Births/Deaths) |
White NMR (Births/Deaths) |
Ratio (Black NMR /White NMR) |
P Value |
Insurance |
Medicaid |
5.08 (67,526/343) |
2.83 (97934/277) |
1.8 |
0.0001 |
Non-Medicaid |
6.68 (30974/207) |
2.44 (105,977/259) |
2.73 |
0.0001 |
Hospitals |
Group 1 |
5.67 (91,053/516) |
2.73 (182,276/497) |
2.08 |
0.0001 |
Group 2 |
4.57 (7,447/34) |
1.80 (21,635/39) |
2.53 |
0.0009 |
Hospital and Insurance |
Group 1 |
|
|
|
|
Medicaid |
5.13 (61,848/317) |
2.98 (84,313/251) |
1.72 |
0.0001 |
Non-Medicaid |
6.51 (29,205/190) |
2.51 (97,963/246) |
2.59 |
0.0001 |
Group 2 |
|
|
|
|
Medicaid |
4.58 (5,678/26) |
1.91 (13,621/26) |
2.4 |
0.006 |
Non-Medicaid |
9.61 (1,769/17) |
1.62 (8,014/13) |
5.92 |
0.0007 |
Table 5: Within hospital differences in NMR based on race and insurance coverage.
Advances in NICU technology may be a factor impacting the data. Most (81%) of the maternity hospitals in NYC are either Regional Perinatal Centers (RPCs) or level 3 hospitals. Further, the majority (96%) of the hospitals included in this study were either RPCs or level 3 hospitals (Table 2). These high level hospitals have the technologies available to care for the most acutely ill infants. Therefore, it is likely that these hospitals will attempt to save more extremely premature infants. Despite the technological advances death rates remain high among extremely premature infants. Based on the 2007-2009 data set used, 92% of the infants born at less than 23 weeks gestation died within 24 hours of birth.
Fetal deaths at gestations of 18 weeks or more are reported to NYCDOH on a certificate of spontaneous termination. Examination of fetal deaths at ≥ 18 weeks gestation indicates the possibility that extremely premature infants that were not reported as live births by group two hospitals were reported as fetal deaths. This data was not analyzed for this study due to concerns regarding inconsistencies in the reporting of fetal deaths [21]. Based on the survey responses there was indication that some physicians who indicated lack of compliance with live birth reporting requirements also lacked compliance with fetal death reporting requirements.
The data sets used for the study did not contain information on post neonatal deaths. However, it is unlikely that the inclusion of post neonatal deaths would have had a significant impact on the findings since the majority of the infants born at less than 23 weeks gestation did not make it to the post neonatal period. In addition, the analysis of the data sets with infants born at less than 23 weeks gestation excluded resulted in an insignificant difference between the two groups of hospitals (Table 3).