 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Department of Pediatrics, Division of Neonatology, Children’s Hospital Medical Center and TriHealth Hospitals, Cincinnati, Ohio
Institute for Health Policy and Health Services Research, University of Cincinnati, Cincinnati, Ohio
Child Policy Research Center, Children’s Hospital Medical Center, Cincinnati, Ohio
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods. A population-based cohort study was conducted of all live births of 500 to 1499 g at the 19 hospitals in the greater Cincinnati region from September 1, 1995, through December 31, 1997 (N = 848). Primary outcome was the risk-adjusted, predischarge mortality or morbidity, including bronchopulmonary dysplasia, severe intracranial hemorrhage, severe retinopathy of prematurity, or necrotizing enterocolitis for VLBW infants who were born at subspecialty perinatal centers compared with those who were born at nonsubspecialty centers.
Results. The odds of death or major morbidity for VLBW infants who are born at nonsubspecialty perinatal centers is twice that of infants who are born at subspecialty centers despite controlling for demographic (odds ratio [OR]: 2.64; 95% confidence interval [CI]: 1.7–4.2) and practice characteristics (OR: 1.96; 95% CI: 1.2–3.2). The effect of birth hospital type on death or major morbidity was greater for infants of 1000 to 1499 g birth weight (OR: 3.42; 95% CI: 2.0–6.1) than for infants of 500 to 999 g birth weight (OR: 2.1; 95% CI: 1.0–4.8).
Conclusion. The current study lends strong support to existing Academy of Pediatrics and American College of Obstetricians and Gynecologists recommendations that deliveries at <32 weeks’ gestational age occur at subspecialty perinatal centers.
Key Words: very low birth weight • regionalization • neonatal intensive care • outcome • mortality
Abbreviations: VLBW, very low birth weight • SPC, subspecialty perinatal center • NICU, neonatal intensive care unit • BPD, bronchopulmonary dysplasia • ICH, intracranial hemorrhage • ROP, retinopathy of prematurity • NEC, necrotizing enterocolitis • SGA, small for gestational age • CRIB, Clinical Risk Index for Babies • RPD, Regional Perinatal Database • LBW, low birth weight • OR, odds ratio
The American Academy of Pediatrics and the American College of Obstetricians and Gynecologists guidelines for perinatal care recommend that hospital-based services be organized within geographic regions to provide optimal access to expertise and experience.1 According to these guidelines, very low birth weight (VLBW) infants (<1500 g) should be delivered at subspecialty perinatal centers (SPCs), defined by their ability to provide comprehensive care for pregnant women and neonates of all risk categories.2 Changes in medical economics and medical care have challenged existing systems of perinatal regionalization, resulting in proliferation of perinatal services.3 Increased competition for patients, managed care, and dissemination of perinatal technologies and expertise may discourage referrals of high-risk obstetrical patients. Earlier studies that tended to support the benefits of regionalized perinatal care were limited by lack of population-based data4–8 or lack of information on important confounders of perinatal outcomes. Short-term outcomes were limited to neonatal mortality with no information on the morbidities that may affect long-term outcomes.9–18 It is unclear whether primary management or interhospital transport of pregnant women who are at risk for preterm delivery to SPCs reduces neonatal mortality and morbidity. We used population-based data to test the hypothesis that the likelihood of death or major morbidity is reduced for VLBW infants who are born at hospitals with high-level, subspecialty perinatal and neonatal care compared with other available birth sites.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Definitions
A total of 19 hospitals have obstetric units in the Cincinnati region, which comprises 21 contiguous counties in southwest Ohio, southeast Indiana, and northern Kentucky. There were 2 SPCs, each with on-site, high-risk obstetric care and tertiary neonatal intensive care units (NICUs).2 All NICU admissions to the 2 Cincinnati SPCs were inborn. The Cincinnati region has 1 free-standing children’s hospital that provides tertiary neonatal care but no maternity services. Selected mothers with high-risk pregnancies and infants who required tertiary neonatal care were referred from the remaining 16 hospitals to 1 of these 3 facilities. The 2 SPCs and the free-standing children’s hospital were the only Cincinnati hospitals that provided 24-hour, on-site physician coverage for newborn care and the only hospitals that provided long-term, mechanical ventilation for neonates. Some of the non-SPC birth hospitals used continuous positive airway pressure, but none used surfactant or mechanical ventilation other than short-term stabilization for transport to the children’s hospital. For comparisons of outcomes, infants were assigned to their hospital of birth, not their location of care. Infants who were born at non-SPCs but subsequently were transported to the children’s hospital were assigned to the non-SPC of birth. Within the region, SPC use for delivery was not limited by insurance status.
Mortality was defined as death of a liveborn infant before hospital discharge or age 120 days, whichever occurred first. Bronchopulmonary dysplasia (BPD) was defined as receipt of supplemental oxygen at 36 weeks’ postmenstrual age. Severe intracranial hemorrhage (ICH) was defined as grade III or grade IV ICH using the Papile classification.19 Severe retinopathy of prematurity (ROP) was defined as receipt of laser or cryotherapy for treatment of ROP. Severe necrotizing enterocolitis (NEC) was defined as Bell stage II or III.20 Gestational age was defined as the best obstetric estimate. Small for gestational age (SGA) was assigned when the infant’s birth weight was less than the 10th percentile for gestational age. Maternal hypertension or preeclampsia was defined as a maternal blood pressure >140 systolic or 90 diastolic, with or without edema and proteinuria, recorded in her medical record during the index pregnancy. Clinical Risk Index for Babies (CRIB) scores are based on information available within the first 12 hours of life, including birth weight, gestational age, congenital malformation, maximum base deficit, and minimum and maximum fraction of inspired oxygen.21 Higher scores reflect higher clinical risks. Antenatal glucocorticoid use was defined as receipt of betamethasone or dexamethasone before delivery to accelerate pulmonary maturity. Intrapartum antibiotics use was defined as administration of antibiotics to the mother during the admission that resulted in delivery.
Data Sources
Data for regional births were obtained retrospectively from 3 sources: the National Institute of Child Health and Human Development Neonatal Research Network registry, the Cincinnati Regional Perinatal Database (RPD), and hospital-based records. The Neonatal Research Network registry includes demographic and clinical data on all VLBW infants who either died in the delivery room or were admitted before 14 days of age to affiliated NICUs.22 In the Cincinnati region, both SPCs and the free-standing children’s hospital participate in the network. The RPD is a population-based, perinatal data collection system in which all regional birth hospitals provide aggregate monthly reports of all live births and predischarge deaths by race, birth weight, and presence or absence of lethal malformations. With Institutional Review Board approval, hospital-based data were obtained from medical records and maternity unit log books that record all live births and still births. When discrepancies between data sources could not be reconciled, hospital-based data, which could be verified by medical record review, were used. Nine VLBW infants reported in the RPD during the study period were not found as live births in hospital-based data sources and were not included in the analyses. Discrepancies were related to reporting the same infant twice, fetal deaths recorded as live births, or apparent errors in reporting birth weight in the RPD. Twenty-two VLBW infants included in the analyses were identified in hospital-based data sources but not found in the RPD. These discrepancies included infants who died in or were transferred from the delivery room to a different institution and inadvertently not included in regional reporting, infants with apparent errors in reporting birth weight, and infants not reported to the RPD for unidentified reasons.
Demographic and clinical data from VLBW infants who were born at 1 of the 2 SPCs or transferred to the children’s hospital were obtained from the network registry. Using the same data item definitions as the network registry, data on VLBW infants who were born and cared for at non-SPCs were obtained by review of mothers’ and infants’ medical records. Medical records of all VLBW infants who were born and/or cared for at non-SPCs were found and reviewed manually.
Outcomes and Confounding Variables
The primary study outcome was presence or absence of risk-adjusted, predischarge mortality or major morbidity, including BPD, severe ICH, severe ROP, or severe NEC. These morbidities were chosen because of demonstrated relationships to long-term neurodevelopmental outcomes of VLBW infants.22–24 Morbidities were counted among survivors and when possible also among infants who died. The composite outcome of death or major morbidity was chosen as the primary outcome to account for possible trade-offs between death and major morbidity. Secondary outcomes included death alone and death in combination with each major morbidity.
Information on factors previously shown to affect the selected study outcomes were included as independent variables in multivariable regression analyses. Possible confounders included both infant and maternal variables unrelated to the care provided by clinicians and practice variables related to clinician decision making. The infant and maternal variables were birth weight, gestational age, race (white or nonwhite), gender, multiple gestation pregnancy, SGA status, 5-minute Apgar score <5, and maternal hypertension or preeclampsia. Practice variables included CRIB score, use of antenatal glucocorticoids, and use of intrapartum antibiotic. We chose to evaluate separately practice variables to emphasize possible opportunities to modify care to improve outcome.
Cervical Dilation Data
Maternal transport to SPCs can potentially have an impact on VLBW outcomes only if mothers are candidates for transport. For estimating the number of mothers who were eligible for transport, information on cervical dilation at admission was obtained from mothers who delivered at non-SPCs. Cervical dilation, hospital admission, and delivery times and dates were obtained from the maternal assessment form completed on admission to the labor and delivery suite.
Statistical Analyses
Regional birth and death rates were calculated for all live births of 
500 g within the Cincinnati region during the study period. Low birth weight (LBW) rates were calculated for infants of 1500 to 2499 g, and VLBW rates were calculated for infants of 500 to 1499 g. Mortality rates are reported as the number of deaths per 1000 live births in each birth weight category. The risk-adjusted effects of birth hospital type (non-SPC vs SPC) on VLBW infant outcomes were evaluated using multivariable logistic regression performed in 2 stages. In the first stage, the effect of birth hospital type, controlling for nonpractice infant and maternal variables, was determined using a backward stepwise algorithm. In this algorithm, all variables were initially entered into the model. At each successive stage of the model-building process, the least significant predictor variable was removed until all variables that remained in the model met a significance level of .05. In the second analysis, a backward stepwise regression analysis was performed, including both the maternal and infant variables and the practice variables (CRIB score, antenatal glucocorticoid use, and intrapartum antibiotics use). In this model, practice variables were added to the model sequentially. Data were managed and analyzed using SAS statistical software (SAS Institute, Inc, Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
500 g, 66% of which (n = 127) occurred in infants 500 to 1499 g. Eighteen infants with lethal malformations were excluded from the regression analyses.
|
|
|
|
|
In separate models that included only infant and maternal covariables, birth in a non-SPC increased the odds of death or any major morbidity for both infants of 500 to 999 g and infants of 1000 to 1499 g (Table 6). At non-SPCs, the ORs for death or any major morbidity tended to be greater for infants of 1000 to 1499 g compared with those of 500 to 999 g (3.4 vs 2.1). With the exception of NEC, secondary outcomes followed a similar pattern. Overall, the ORs for death, BPD or death, ICH or death, or ROP or death tended to be greater for infants of 1000 to 1499 g compared with infants of 500 to 999 g.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In contrast to previous reports, this is the first population-based study to evaluate the effect of birth hospital characteristics on risk-adjusted morbidity as well as mortality. Earlier studies comparing outcomes of outborn and inborn infants4–7 were not population based. In these studies, the study population includes only those infants who were admitted to an SPC, resulting in potential selection biases.8 These studies omit healthier infants who remain at non-SPC, as well as the sickest infants who die at the non-SPC sites, resulting in potentially biased outcomes. Earlier studies that were population based used linked birth and death records as the primary data source. As a result, information on common VLBW infant morbidities that have an important impact on long-term outcome and possible confounding clinical variables were not available.9–18 In the absence of morbidity data, possible trade-offs between death and major morbidity cannot be evaluated.
As perinatal care becomes less well organized regionally, rigorous examination of outcomes for high-risk populations is specifically needed to evaluate the impact of changes in perinatal practices and access to care.25 Recent reports from California demonstrate these challenges. In California, a proliferation of community (level II+) NICUs resulted in increased numbers of VLBW births at these sites, with fewer births at both regional (level III) and intermediate (level II) NICUs. Mortality risk for VLBW infants was no different comparing community and regional NICUs but was significantly greater at hospitals with other levels of care.10 When California’s community NICUs were stratified further by average daily census, only community NICUs with an average daily census >15 had survival rates comparable to that of regional centers.9 In California, "deregionalization" resulting from increased numbers of community NICUs had no detrimental effect on mortality. However, if deregionalization results in multiple low birth volume sites, then mortality may be increased. In these 2 population-based studies of California births, the primary outcome of interest was death. Given the long-term impact of morbidities for this population, it would be important to know whether changes in care that maintained or even decreased mortality occurred at the expense of increased risk of long-term morbidity.
Our results are consistent with the California experience as well as other population-based reports. Mortality rates for VLBW infants who were born at non-SPCs were 1.5 to 3 times higher than for VLBW infants who were born at SPCs.9–18 Recently, results from the Canadian Neonatal Network compared outcomes of VLBW infants outborn to those inborn at SPCs.4 Outborn infants who required transport were at higher risk of death (OR: 1.7) as well as common morbidities. Outcome of infants who were cared for at hospitals that do not participate in the Canadian Neonatal Network were not included.
These results contrast with other reports comparing VLBW outcomes in different types of NICUs. Meadows et al,26 in a case-control study, found no difference in length of stay, duration of ventilation, or need for home oxygen between infants who were cared for at a regional SPC and those who were cared for at a community nursery with ventilation capabilities. Hobar et al,27 in a survey of 62 US NICUs, found no relationship between mortality and NICU volumes or presence of residency programs. The difference in results may be attributable to differences in study design and exclusion of nonparticipating community hospitals. Recently, Goodman et al28 found no relationship between availability of neonatal intensive care and neonatal mortality across a wide range of birth weights. In this study, availability of care was measured in terms of NICU beds and numbers of neonatologists rather than levels of care.
The effect of birth hospital type on the likelihood of death or major morbidity tends to be greater for infants of 1000 to 1499 g birth weight compared with infants of 500 to 999 g. Yeast et al12 and Powell et al15 reported a >2-fold increase in neonatal mortality for infants of 1000 to 1499 g when born at nontertiary hospitals. Menard et al11 found a similar increase in neonatal mortality for infants of 1000 to 1249 g born at nontertiary hospitals but no difference for infants of 1250 to 1499 g. Gould et al,10 using population-based data from California in the 1990s, found significantly increased mortality in VLBW infants who were <1000 g and born in nontertiary centers. No significant differences in mortality were found for infants of >1000 g. Some of these differences may be explained by use of mortality data alone and the inability to control for clinical variables that have an impact on outcome in this high-risk population. Although overall mortality or major morbidity rates in the 1000 to 1499 g birth weight group are lower than for infants of <1000 g, birth hospital type remains an important determinant of outcome in this group.
The composite of death or major morbidity was chosen as the primary outcome for this study because it accounts for possible trade-offs between mortality and major morbidity and because it avoids underestimation of possible adverse consequences associated with either type of birth hospital. Use of composite outcomes provides an estimate of the proportion of VLBW infants who are discharged from the hospital without selected morbidities identifiable during the NICU stay. Use of mortality rate alone as the outcome of interest would overestimate performance of sites where low mortality is achieved at the expense of high morbidity. Mortality data alone are particularly difficult to interpret at the lower end of the birth weight range, where survival is significantly altered by clinician and parent preferences regarding the aggressiveness of neonatal care. Use of composite outcomes is less likely to underestimate any negative impact of an intervention by accounting for both mortality and morbidity. The morbidities in this study were chosen because of their potential impact on long-term functional outcomes. Direct evaluation of long-term outcomes may better detect long-term disabilities not reflected in morbidities identified during the NICU hospitalization.
Composite outcomes were used in this study to evaluate the impact of site of delivery on specific morbidities. This approach may overestimate risk, as not all of the infants who died would have developed the morbidity. An alternative approach would estimate morbidity rates only in survivors; however, this approach underestimates risk by assuming that morbidity rates among infants who die are similar to that of survivors. When evaluating a possibly modifiable practice, such as location of preterm delivery, underestimating associated poor outcomes carries the possibility of adopting a practice that may be more harmful than reported. We therefore chose possibly to overestimate individual morbidities by using composite outcomes.
The effect of birth hospital type on VLBW outcome was assessed after adjustments for differences in risk associated with variations in important infant and maternal characteristics and clinical practice characteristics. Controlling for infant and maternal variables, the odds of death or major morbidity was 2.5 times higher for VLBW infants who were born at non-SPCs compared with SPCs. Although previous studies commonly made adjustments for differences in infant and maternal characteristics, risk adjustment has not previously been performed for practice variables. By evaluating the impact of practice style separately, we were able to illustrate the contributions of factors that result from clinical decisions separately from demographic characteristics. The CRIB score,21 based on information available in the first 12 hours of an infant’s life, may indicate inherent, nonmodifiable illness severity and/or obstetric and pediatric decision making.29–31 Both obstetric decisions and neonatal resuscitation capabilities may affect the neonate’s clinical condition. We chose to treat CRIB score as a practice variable to emphasize possible opportunities to modify care to improve outcome. To observe the influence of CRIB alone as an indirect measure of practice, we ran logistic models with CRIB and the nonmodifiable infant characteristics (Table 7). With CRIB as the only practice variable in the model, the OR for the effect of delivery site on death or selected major morbidities decreased from 2.64 to 1.96. Subsequent sequential addition of the antenatal antibiotics and antenatal glucocorticoids variables resulted in no change of the effect of site of delivery on the primary outcome. The lack of effect of antenatal practice variables may be attributable to interrelationships between variables not accounted for in the model, for example, between antenatal steroids and site of delivery. Controlling for practice variables, however, did not alter the conclusion that location of birth has an important effect on the likelihood of death or major morbidity. There seem to be additional factors that account for outcome differences between hospitals. Whether these factors are related to inherent patient risk or clinical practice remains to be determined. For individual secondary outcomes, no significant effect of birth hospital type was observed with the addition of practice variables to the models. This may have occurred because the event rates for individual morbidities are low, resulting in inadequate power to detect differences.
Perinatal outcomes for the Cincinnati region are similar to the United States as a whole. The Cincinnati LBW rate for the study period was 72 per 1000 live births compared with a US rate of 75 per 1000 in 1997. The Cincinnati VLBW rate was 1.3% compared with 1.4% for the United States in 1997.32 Predischarge mortality rates were also similar to neonatal mortality rates reported by the National Center for Health Statistics.33 Availability of neonatal care resources within the Greater Cincinnati region would be considered high using criteria recently established by Goodman et al.28 For every 10 000 births within the Cincinnati region, there are 9 neonatologists and 51 NICU beds. Although Cincinnati outcomes are similar to the United States as a whole, findings from this study may not apply in other regions. There may be unique characteristics of the perinatal care delivery system in the Cincinnati region and unique risk profiles that limit application to other sites. In Cincinnati, for example, high-risk perinatal services tend to be aligned within hospital systems, whereas physician care for neonates is centralized in a single neonatology practice group. Twenty-five percent of VLBW infants within the study region were delivered at non-SPCs, which is similar to that reported from other regions during the late 1980s to early 1990s from 14% to 35%.10–12,15
Referral patterns determined by physician choice, instructions given to mothers regarding preterm labor, and the management of possible medical emergencies are important determinants of delivery site for VLBW infants. The extent to which each of these factors contributed to site of delivery in this study was not evaluated. Whether opportunities existed to shift delivery from non-SPCs to SPCs was evaluated, in part, by determining the degree of cervical dilation at the time of non-SPC admission. Of 198 non-SPC mothers whose medical records indicate cervical dilation at admission, 104 (52%) had cervical dilation <4 cm. Forty-six percent of these women (N = 48) delivered >12 hours after admission. For some women who eventually delivered VLBW infants, there seemed to be sufficient time to consider maternal transport.
VLBW infants represent roughly 1.3% of live births, yet these infants account for 60% of neonatal deaths and perhaps a larger share of life-long handicapping conditions that arise from perinatal events. Results from our study indicate that the outcome of VLBW infants is improved by delivery at an SPC. These results support the current American Academy of Pediatrics/American College of Obstetricians and Gynecologists recommendation that infants with estimated gestational ages <32 weeks continue to be delivered at SPCs whenever possible.2
|
ACKNOWLEDGMENTS |
|---|
We thank Richard Hornung, DrPH, at the Institute for Health Policy and Health Services Research, University of Cincinnati, for assistance in review and analysis of the manuscript.
|
FOOTNOTES |
|---|
Reprint requests to (B.W.) Division of Neonatology, Department of Pediatrics, 4th Floor, Children’s Hospital Research Foundation, Children’s Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229. E-mail: warnb0{at}chmcc.org
Dr Chenier’s current affiliation is Virginia Polytechnic Institute and State University, Office of Institutional Research and Planning, Blacksburg, Virginia.
Presented in part at the Society for Pediatric Research; Boston, MA; May 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  P. Van Reempts, L. Gortner, D. Milligan, M. Cuttini, S. Petrou, R. Agostino, D. Field, L. den Ouden, K. Borch, J. Mazela, et al. Characteristics of Neonatal Units That Care for Very Preterm Infants in Europe: Results From the MOSAIC Study Pediatrics, October 1, 2007; 120(4): e815 - e825. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Rautava, L. Lehtonen, M. Peltola, E. Korvenranta, H. Korvenranta, M. Linna, M. Hallman, S. Andersson, M. Gissler, J. Leipala, et al. The Effect of Birth in Secondary- or Tertiary-Level Hospitals in Finland on Mortality in Very Preterm Infants: A Birth-Register Study Pediatrics, January 1, 2007; 119(1): e257 - e263. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. A. Haberland, C. S. Phibbs, and L. C. Baker Effect of Opening Midlevel Neonatal Intensive Care Units on the Location of Low Birth Weight Births in California Pediatrics, December 1, 2006; 118(6): e1667 - e1679. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. B. Bartels, D. Wypij, P. Wenzlaff, O. Dammann, and C. F. Poets Hospital volume and neonatal mortality among very low birth weight infants. Pediatrics, June 1, 2006; 117(6): 2206 - 2214. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Ho and S. Saigal Current Survival and Early Outcomes of Infants of Borderline Viability NeoReviews, March 1, 2005; 6(3): e123 - e132. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2004 American Academy of Pediatrics. All rights reserved. |
||
|
||
