 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
PEDIATRICS Vol. 104 No. 5 November 1999, pp. 1064-1069
From the Department of Pediatrics, University of Iowa College of Medicine, Iowa City, Iowa.
 |
ABSTRACT |
|---|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
Objective. Our earlier studies about the changing pattern of neonatal mortality were based on data from 1978-1979 and 1982-1983. In this report, we provide a modern update of the causes of neonatal deaths to help focus prevention measures.
Study Design. We reviewed neonatal deaths for the years 1995 and 1996 and assigned a clinical cause of death. Deaths were reviewed and analyzed by place of birth and death, according to level of care. We provided an estimate of the change in causes of death between 1982-1983 and 1995-1996 and calculated an idealized neonatal mortality rate for each level of care and for the entire state.
Results. The major changes in causes of death occurred in the respiratory distress syndrome (RDS) and extremely low birth weight (ELBW) categories. RDS showed a substantial decline, but ELBW deaths increased by nearly the same amount that RDS declined. The idealized neonatal mortality rate calculations suggest that fewer potentially medically preventable deaths occur now than in 1982-1983. Currently, more neonatal deaths of level I births occur in level III facilities than in 1982-1983. However, at that time, Iowa had only one tertiary facility, and the level II regional hospitals were the site of more deaths.
Conclusion. We believe that assigning deaths to a clinical category rather than to a catastrophic final event helps focus on sources of preventable death in the medical care system. To reduce neonatal mortality, the incidence of lethal congenital malformations and ELBW infants must be impacted. Key words: neonatal mortality, infant mortality, causes of death, regionalized care, idealized mortality rate.
In 1986, we reported1 a comparison of causes
of neonatal deaths of infants born in Iowa hospitals in 1978-1979 with a similar population born in 1982-1983. We showed that lethal congenital malformations were the leading cause of death and that the
actual number of malformations increased from the first to the second
review period. There was a net reduction of 208 neonatal deaths in the
later review period, and 185 of these represented a reduction in deaths
occurring to infants born in Iowa level I hospitals.
We speculated, based on the calculation of an idealized neonatal
mortality rate, that additional reductions in Iowa's neonatal mortality rate would have to come from improved care in level II and
level III centers rather than from level I hospitals. The purpose of
this article is to provide an updated analysis of the causes of
neonatal deaths, according to level of care in Iowa. The most recent
statewide data will be compared with the 1982-1983 data from the
previous paper.
Review of neonatal deaths was conducted as described in two
earlier reports.1,2 Specifically, the dominant factor that
brought about the death was selected as the clinical cause of death.
For example, if an infant had Down syndrome but died of an infectious
process attributable to group B streptococci, the cause of death was
listed as bacterial sepsis. Similarly, if an infant with Down syndrome
died as a result of inoperable congenital heart disease, the cause of
death was listed as Down syndrome and was included in the category of
lethal congenital malformations.
An appropriately grown infant3 with a birth weight of
Another difference from our previous report is the number of hospitals
contained in the various levels of care. In 1982-1983, there were 131 hospitals with obstetric services. Of these hospitals, 11 served as
regional level II centers and only one, the University of Iowa
Hospitals and Clinics, served as a tertiary center. All other hospitals
were level I. In 1995-1996, there were 107 hospitals with obstetrics
departments. The number of regional level II centers dropped to
10, and 2 hospitals were added to the tertiary level. These hospitals
are in Des Moines, Iowa, and both have neonatologists and
maternal-fetal medicine specialists on staff. One hospital had been a
regional level II center in 1983, and the other hospital was a level I
facility.
During the 1982-1983 review of deaths, only 357 of the 490 deaths were
reviewed (72.9%). Currently all neonatal deaths are included in the
review. This is because the author (H.A.H.) has assumed a part-time
consulting position with the Iowa Department of Public Health that
requires review of all infant deaths in Iowa. Previously, the deaths
were included only if they were part of a hospital experience
encountered during visits by the Iowa Perinatal Team.2,9
Birth and mortality data from Iowa hospitals were provided by the Iowa
Department of Public Health. All neonatal deaths were matched to the
appropriate birth certificates.
Categories of Causes of Death
The causes of death were assigned to various categories.
Nonpreventable Causes
Congenital Malformations
All deaths caused by congenital malformations and/or syndromes
incompatible with life were assigned to this category regardless of
birth weight.
ELBW
In the 1982-1983 sample, all births <750 g were assigned to
this category, and a birthweight of 650 g was used to calculate the irreducible neonatal mortality rate (vide infra). For the 1995-1996 population, the ELBW upper limit was changed to 700 g
and also was used to calculate the irreducible neonatal mortality rate.
Potentially Preventable Causes
The primary objective of assigning a death to one of the
following categories was to determine the principal condition
contributing to the death.
RDS
All infants in this category had roentgenograms that, in the
opinion of a roentgenologist or neonatologist, were consistent with the
diagnosis of RDS (ground glass appearance, air bronchograms, hypoventilated). The onset of respiratory distress was within an hour
of birth, the course was consistent with RDS, the infant required
maintenance with a ventilator, and cultures of blood and cerebrospinal
fluid did not yield pathogens. RDS was listed as a principal cause of
death when the disease was severe, even when another major pathologic
diagnosis such as intraventricular hemorrhage may have been present.
When RDS was mild or moderate and recovery from the disease was
occurring or had occurred, another diagnosis such as intraventricular
hemorrhage was listed as the cause of death.
Bacterial Sepsis and/or Necrotizing Enterocolitis
Bacterial sepsis and/or necrotizing enterocolitis was listed as
the cause of death when the clinical course was consistent with the
diagnosis of bacterial sepsis. In a few instances, the diagnosis was
based entirely on clinical observations, but in most cases, cultures of
blood or cerebrospinal fluid were obtained and a pathogen isolated,
either before or after death. Patients with proven necrotizing
enterocolitis and a positive culture result were included in this
group.
Birth Asphyxia
Birth asphyxia was listed as the cause of death when 1- and
5-minute Apgar scores were Other Category
Infrequent causes of neonatal death were included in this
category, including viral diseases.
Estimation of Magnitude of Changes in Causes of Death (Table
3)
We have provided information that can be used to quantify the
magnitude of change in major sources of neonatal death between the two
study periods. Data are presented for the entire population of Iowa
hospital births and for each of the three levels of care. Because there
were >15 000 more births in the 1982-1983 period, it was necessary
to standardize births for the two periods. Accordingly, the difference
in total number of births in thousand was multiplied by the neonatal
mortality rate for the 1982-1983 period to compute the difference in
deaths that would have resulted if the number of births were the same
for both periods. These deaths were then subtracted from the 1982-1983
total. Then, the proportion of deaths that occurred in each level of
care was modified appropriately and that number was multiplied by the
proportion of deaths in each category. For the 1995-1996 period, the
actual number of deaths was used. For example, in 1982-1983, the total
number of births reported was 88 810. For 1995-1996, the total was
73 712. Thus, 15 098 more births were recorded for the earlier
period. We multiplied 15 098 by the combined neonatal mortality rate
for 1982-1983 (5.52/1000 live births) to determine the number of
deaths that the difference in live births would make (15.098 × 5.52 = 83.3). Then, we subtracted the 83 deaths from the 490 reported and used 407 deaths as the basis for 1982-1983. We computed
the percentage of births that occurred at each level of care from the
data provided in the 1986 article1 to determine the number
of deaths at each level of care. In level I, there were 188 births, and
188 divided by 490 is 38.37%. This contribution multiplied by 407 (0.3837 × 407) = 158.65, and this is the standardized number
of deaths that would have occurred in level I hospitals, if the number
of births had been the same in 1982-1983 as in 1995-1996.
TABLE 3
METHODS
Top
Abstract
Methods
Results
Discussion
References
700 g and without a lethal congenital malformation was included in
the extremely low birth weight (ELBW) category. To avoid dual or
multiple causes of death, other causes such as sepsis or respiratory distress syndrome (RDS) were not considered for this weight category. It is important to note that the ELBW category has changed slightly from the previous report of 1986. In that report, we used 750 g to
define the upper limit for the ELBW category, but used a 650 g
upper limit to calculate an irreducible mortality rate (vide infra).
The reduction to 650 g for 1983 was made to reflect the improved
outcomes of neonates >700 g.4 Because we incorporate the
combination of lethal congenital malformations and ELBW deaths as the
basis for determining potentially medically preventable deaths, we
wanted to use a birth weight appropriate for this
purpose.4,5 In retrospect, we believe that we were too
stringent in the 1986 paper and should have compromised at 700 g.
It can be argued that the lower limit of viability is 22, 23, or 24 weeks, but few would expect all infants weighing 700 g to be
medically preventable, regardless of the intensity and appropriateness
of care offered.6-8
5, and the infant never achieved a stable
condition after birth. Another disease process may have been diagnosed
postmortem, but if the dominant process was believed to be asphyxia,
the death was so recorded.
Estimated* Change in Number of Neonatal Deaths by Category of Cause of
Death
Next, we multiplied the number of deaths by the percentage of deaths in a category, determined by our review process in 1982-1983. For example, for asphyxia in level I hospitals, the percentage was 18.4. Thus, 158.65 × 0.184 = 29.2. Then, we subtracted the actual number of deaths that occurred in that category for 1995-1996 (which was 17), and the difference was 12.2. The final numbers are shown in Table 3 and represent either a decrease for the 1983 numbers minus the 1996 numbers (which are shown as negative numbers) or an increase (which are reflected as positive numbers).
In the 1986 article, we provided a comparison of the results of a review compared with death certificate data. The contribution of ELBW, malformations, and RDS were similar in review and death certificate data, but asphyxia and infection were underestimated. Accordingly, we believe that applying the percentages obtained by the review process was a more accurate way to standardize the revised 1982-1983 data.
Idealized Neonatal Mortality Rate (Table 4)
Idealized neonatal mortality rates were calculated for each of the hospital categories using 1983 and 1996 data. As in the earlier report, the conditions considered to be beyond medical intervention were births of infants with lethal congenital malformations and ELBW infants. An idealized neonatal mortality rate was calculated as follows. The actual neonatal mortality rate was multiplied by the proportion that the two nonpreventable factors contributed to the overall death rate. For example, the neonatal mortality rate for level I hospitals in 1983 was 3.4/1000 live births, and 56.1% of deaths were attributable to the two causes considered to be nonpreventable. Thus, 3.4 × 0.561 = 1.9, the idealized neonatal mortality rate for 1983. Subtracting the idealized rate from the actual rate provides an excess death rate, and by multiplying this figure by the number of births in thousands, the potential number of medically preventable deaths can be estimated.
|
In each case in which it was necessary to estimate the number of deaths because of the standardization procedures, numbers were rounded to the nearest whole number for the sake of discussion.
| |
RESULTS |
|---|
|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
Table 1 shows the Iowa hospital of birth and death by level of care for neonatal deaths occurring during the 1982-1983 and 1995-1996 periods. Approximately the same percentage of infants were born and died in level I hospitals (14.0 vs 14.2). Also, nearly the same percentage of level I births were referred to higher levels of care (13.4 vs 13.7), but during the latter period, a greater percentage of these were level III referrals (5.3 vs 12.5).
|
We believe that the majority of high-risk pregnancies are referred to higher levels of care before birth, but currently we do not have a data system to document this. However, we did review the 1995-1996 data to examine the causes of death of those neonates who were born and died at level I hospitals.
Of the 48 reported cases, 20 died of extreme immaturity (birth weight <500 g) and another 5 were in the 500 to 700 g category and were considered very immature and could not be resuscitated and transferred. Another 5 infants were born severely depressed from such causes as placenta previa and placenta abruptio and were considered to have died from nonpreventable asphyxia. A total of 16 infants had lethal congenital malformations, and 2 others were considered to be potentially transportable after having been born severely depressed. One was a term infant and the other was preterm, but weighed >700 g. Thus, the vast majority of the nontransported level I births (46 of 48) died of causes that we believe were medically nonpreventable.
There was a decline of 11.2% in the percentage of neonatal deaths whose birth occurred in level II hospitals. The increase in percentage of neonatal deaths occurring in level III hospitals in 1995-1996 (10.7%) is very close to the decrease observed for the level II hospitals. As noted above, one level II hospital from the 1982-1983 period became a tertiary hospital, as did one level I hospital.
Table 2 compares causes of neonatal death
between the two periods according to levels of care. It must be
recognized that the deaths were ascribed to the level of care where the
birth occurred and not necessarily where the death occurred. In level I
hospitals, there was essentially no change in the percentage of
asphyxia deaths, but in the other two levels, asphyxia decreased during
the modern period. The upper limit for ELBW was reduced from 750 to
700 g, but the percentage of deaths from this cause still
increased at each level of care. The number and percentage of deaths in
the ELBW category of infants weighing 500 to 700 g are shown for
1995-1996. These data were added to illustrate the significance of
births with estimated gestational age 22 weeks (<500 g) to the
neonatal mortality rate in 1995-1996. The 1982-1983 data were not
available for comparison.
|
RDS decreased noticeably at each level of care. The percentage of deaths attributable to lethal congenital malformations changed very little from 1982-1983 to the modern period, but level II hospitals did demonstrate a 12.4% decrease. However, the level III hospitals increased by 14.1%. Overall, malformations accounted for a 0.7% increase statewide. Deaths attributable to bacterial sepsis decreased at each level of care. The percentage of other and unknown causes was essentially the same for both periods.
In Table 3, we estimated the change in the total number of neonatal deaths by level of care according to the category of cause of death from period one to period two. As noted above, we standardized the 1982-1983 and 1995-1996 review periods by equalizing the total birth population and then calculating the incidence of the causes of death for the 1982-1983 population. The figures for 1995-1996 are actual deaths that occurred during that period. Deaths attributable to asphyxia were reduced at all three levels of care and particularly at the level I and II hospitals. ELBW deaths decreased at level I, but increased at both level II and level III hospitals. The number of deaths attributable to RDS decreased at all three levels of care. The number of deaths caused by malformations decreased at the level I and level II facilities but increased at level III hospitals. Bacterial sepsis decreased as a cause of death at all three levels of care. The last category of other and unknown causes showed little change.
As shown in Table 3, there was a reduction of 120 deaths for all categories except for ELBW, which showed an increase of 50 deaths. Thus, the net reduction was 70 deaths.
In Table 4, idealized neonatal mortality rates were calculated for the three levels of care for 1996, and 1983 results are provided from the 1986 paper for comparison. The actual neonatal mortality rates for each level of care has dropped since 1983, and the percentage of deaths attributable to lethal congenital malformations and ELBW have both increased since 1983. Because fewer deaths are considered now to be medically preventable, the excess neonatal mortality rates have dropped below 1.0 in level I and II hospitals and are down to l.8 in level III hospitals. The state excess rate is 0.9 per 1000 live births.
Given the assumption that lethal congenital malformations and ELBW births <700 g are currently medically nonpreventable, only 36 deaths were calculated to be potentially medically preventable in 1996, compared with 102 in 1983.
| |
DISCUSSION |
|---|
|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
In our previous article, we observed that in 1978-1979 and in 1982-1983, five major causes accounted for ~90% of all neonatal deaths. These include asphyxia, ELBW, RDS, lethal congenital malformations, and bacterial sepsis. Currently, these categories account for >95% of the deaths, and the combination of ELBW and lethal congenital malformations currently represents nearly 80% of the deaths.
In our earlier report, we found that the absolute number of lethal congenital malformations increased despite a statewide reduction of 208 neonatal deaths.1 Data from Table 3 reveal that the relative number of lethal congenital malformations decreased in the recent period, compared with standardized data from 1982-1983. In fact, only ELBW deaths showed an increase for 1995-1996, but one that was relatively large. Using data from Table 3, we estimated an increase in 50 deaths attributable to ELBW. Also, it must be noted that the upper limit for ELBW was decreased to 700 g in 1995-1996 from 750 g in 1982-1983. Thus, the relative increase in deaths of very small infants was larger than the estimate suggests.
In 1982-1983, the contributions of ELBW (28%) and malformations (31.4%) were comparable. The current data show that malformations continue to account for approximately one third of the deaths (32.1%), but the contribution of ELBW has risen to nearly one half of the total (47.5%). Reasons for the increase in deaths in the ELBW population are not apparent from our data. The definition of a live birth in Iowa states that the conceptus is live born if it shows any sign of life, regardless of birth weight or gestational maturity. A neonatal death is the death of a live born infant occurring within the first 27 days of life. Neither definition has changed over the period of this report (1982-1996). Thus, we believe that the increased incidence of ELBW deaths is not a statistical phenomenon. Rather, we speculate that improved obstetric care allows more of these small infants to be born alive.10,11 It is also possible that an increase in the number of multiple births has contributed to an increase in ELBW births.12,13 Unfortunately, our data do not permit an estimate of the number of births that result from some form of assisted reproduction. But, based on anecdotal information, we believe that assisted reproduction is an important factor leading to an increase in the number of ELBW births.14,15 An encouraging note referable to ELBW data is that level I hospitals report a net decrease in deaths from this cause, and the majority of the increase came from tertiary hospitals. The regional level II hospitals had an increase in ELBW deaths, but considerably less than the increase noted at the level III centers. The requirement for level III status in Iowa includes the presence of a maternal-fetal medicine specialist on staff. Of the 10 regional level II centers, 5 have neonatologists on staff and thus would be considered tertiary centers in some states. Accordingly, the birth of some ELBW neonates in regional level II centers does not necessarily constitute a breakdown in the regionalized system of perinatal health care. This is particularly true when time and distance factors are considered for emergent deliveries.
Another encouraging datum was the reduction in deaths attributable to RDS. This reduction was shared in all three levels of care. Undoubtedly, the use of surfactant16-19 in the neonate and steroids prenatally20,21 contributed to the reduction. Currently, RDS accounts for only 2.7% of all neonatal deaths, but in 1982-1983, it was a much more significant factor, accounting for 15.4%.
Neonatal infection22-24 has been associated with the use of surfactant. Although our data do not provide evidence about the use of surfactant, it is used throughout the state. Accordingly, if the association of surfactant and sepsis were true in Iowa, one would anticipate the opposite of our results, because we found a decrease in deaths attributable to bacterial infection.
The category of asphyxia includes both medically preventable and nonpreventable causes. It is not possible by chart review to determine the truly preventable causes. However, during both review periods, there were cases in which care could have been better or more swiftly applied. Thus, it was disappointing to see that the relative incidence of deaths in these categories changed very little in the level I hospitals. However, there was a reduction at both level II and level III facilities.
In 1995-1996, level I hospitals reported reduced numbers of deaths in all categories. The reduction in lethal malformations was the most notable. Level II regional centers showed an increase in the number deaths attributable to ELBW, but decreases in the four other major categories of neonatal death. Level III hospitals reported an experience quite different from the other two in that they had a net increase of 49 deaths that came from an increase in ELBW infants and lethal congenital malformations. The overall reduction in the number of deaths statewide and the movement of deaths to the tertiary centers are consistent with the goals of a regionalized system of perinatal health care.
As noted in our previous article, the use of an idealized neonatal mortality rate to project potential for medically preventable deaths does not predict accurately the number of preventable deaths. However, we do believe that this approach does offer health researchers and educators an estimate of the magnitude and source of preventable deaths available overall and broken down at the various levels of care. The comparisons of the idealized rates for 1983 and 1996 are distorted somewhat by the use of a 650-g lower limit for 1983 and a 700-g lower limit for 1996. Unfortunately, we no longer had the raw data from 1983 to be able to recalculate the idealized rate using 700 g as the cutoff, and thus, used the data we had reported in 1986 for comparison. When we calculated the potentially medically preventable deaths, using 750 g as the upper limit for 1983, this would have reduced the total number to 89 versus 102, using 650 g. Regardless of the weight limit used for ELBW, the idealized rate will become, at best, an estimate. Many of the deaths that we currently consider in the preventable category are likely beyond medical intervention. However, we believe that only extremely immature neonates and those with lethal congenital malformations currently can be considered to be beyond the scope of the medical care system; therefore, they constitute the basis for the residual infant mortality or idealized rate.
Our statewide perinatal educational program has concentrated on reducing medically preventable perinatal deaths. We believe that our style of assigning deaths to a clinical category such as RDS rather than to a catastrophic final event, such intraventricular hemorrhage or an air embolism, has helped us to focus on the sources of preventable death within the medical care system.
Calculation of an idealized neonatal mortality rate also helps to underscore the large input that lethal congenital malformations and ELBW births have on the overall neonatal mortality rate, and thus, the infant mortality rate. Clearly, if rates in Iowa, and also very likely in the United States, are to be reduced significantly, fewer deaths must occur in both of these categories.
| |
FOOTNOTES |
|---|
Received for publication Jul 20, 1998; accepted Mar 2, 1999.
Reprint requests to (H.A.H.) the Department of Pediatrics, University of Iowa College of Medicine, University of Iowa Hospitals and Clinics, 200 Hawkins Dr, Iowa City, IA 52242. E-mail address: herman-hein{at}uiowa.edu
| |
ABBREVIATIONS |
|---|
ELBW, extremely low birth weight; RDS, respiratory distress syndrome.
| |
REFERENCES |
|---|
|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. Subhani, A. Combs, P. Weber, C. Gerontis, and J. D. DeCristofaro Screening Guidelines for Retinopathy of Prematurity: The Need for Revision in Extremely Low Birth Weight Infants Pediatrics, April 1, 2001; 107(4): 656 - 659. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 1999 American Academy of Pediatrics. All rights reserved. |
||
|
||
