Secondary Logo

Journal Logo

Original Research

Trends in Perinatal Regionalization and the Role of Managed Care

Dobrez, Deborah PhD1; Gerber, Susan MD, MPH2; Budetti, Peter MD, JD3

Author Information
doi: 10.1097/01.AOG.0000232557.84791.3e

Data Supplements

Click on the links below to access all the Data Supplements for this article.

Please note that Data Supplement files may launch a viewer application outside of your web browser.

The promotion of perinatal regionalization has been supported by the abundant literature that demonstrates decreased neonatal morbidity1–4 and mortality5–16 for high-risk neonates delivered at tertiary care facilities or perinatal regional centers. Consistent with these findings is goal #16–8 of the Healthy People 2010 initiative, which aims to “increase the proportion of very low birth weight (VLBW) infants born at level III hospitals or perinatal centers.”17 However, multiple studies have demonstrated some evidence of deterioration in regional networks through the early to mid 1990s in single states.14,18,19

The deregionalization of perinatal health care in the United States has been attributed in part to market forces such as the growth of health maintenance organizations (HMOs) and other managed care organizations.5,14,20 Concurrent with the rising concern of deregionalization is the growth of managed care participation. From 1989 to 1998, the number of enrollees in HMOs more than doubled, from approximately 30 million to more than 70 million Americans. In 2002, 178 million Americans were enrolled in managed care organizations, including 68.8 million in HMOs.21 Prior studies have demonstrated a role for managed care in influencing the medical marketplace22 and hospital length of stay.23 An earlier version of the present study, using data from a single state during a more limited time span, found no effect of managed care on perinatal regionalization in Washington.24 Deregionalization of perinatal care has many far-reaching implications. Of most immediate concern is the potential adverse effect on neonatal morbidity and mortality rates. In addition, the centralization of high technology resources for neonatal care may actually be cost-effective.25 For this reason, an analysis of perinatal regional networks raises issues not only of the best clinical interests of patients but also of the efficiency of health care providers and the entire health care system. The purpose of the present study was to examine trends in site of delivery for high-risk infants and to estimate the extent to which managed care has played a role in the reported deterioration of the perinatal networks by estimating the effect of HMO penetration on the site of delivery of low birth weight infants.


A sample of four states was chosen based on several factors: history of perinatal regionalization, geographic region, varying degrees of managed care penetration reported before the conduct of this study, and availability of data. The four selected states are Washington, Illinois, California, and North Carolina. Based upon growth in the percentage of physician revenue from managed care from 1985–1993, relatively small growth in managed care penetration was observed in North Carolina, moderate growth in Illinois, and relatively high growth in Washington and California.22 All nonmilitary hospitals with obstetric services and greater than 50 deliveries annually in these four states were included in the study. To capture changes across time in the degree of regionalized perinatal care, data were drawn across a 10-year period, from 1989 to 1998. Data were drawn from multiple sources, including state birth certificate databases, the American Hospital Association annual survey, and the Interstudy HMO Database.26

This study used three different classifications of birth weight, defined cumulatively. Low birth weight deliveries were liveborn deliveries weighing at least 500 g, and less than or equal to 2,500 g. VLBW and extremely low birth weight (ELBW) deliveries were liveborn deliveries weighing at least 500 g and less than or equal to 1,500 g and 1,000 g, respectively. We applied two outcome measures, the percentage of deliveries at a particular hospital that are low birth weight (LBW) and percentage of LBW deliveries occurring at level I, II, and III hospitals. The percentage of hospital deliveries that were low birth weight (low birth weight percentage; LBWP) was calculated as the number of LBW neonates divided by the total number of liveborn neonates at each hospital per year. This measure has meaning at a hospital level and can be used for comparisons across years; it is a relevant parameter for this study because hospitals have some degree of control over the mix of deliveries at their hospital and could modify that measure in response to health plan requirements. The percentage of all LBW deliveries occurring at level I, II, and III hospitals may be the most direct measure of regionalization. However, it cannot be used for hospital-level analyses, because it is measured across the state. Moreover, because the percentage of all LBW deliveries at a particular hospital would change across time with the number of hospitals, we used that as a summary measure, and regarded the percentage of deliveries at a particular hospital that are LBW as the primary outcome measure. Secondary analyses focused on VLBW and ELBW deliveries.

Measurement of managed care penetration was limited to health maintenance organizations (HMOs), excluding other types of managed care such as preferred provider organizations. The number of HMO enrollees divided by the total county population was used as a measure of HMO penetration (collected from the Interstudy HMO Database26).

Two primary hospital characteristics were included in the model: hospital neonatal intensive care unit level (extracted from the American Hospital Association annual survey) and Medicaid penetration (available in birth certificate data for Washington and California only). The potential importance of Medicaid on hospital incentives and policy was measured as the proportion of deliveries for which prenatal care was paid for by Medicaid. Maternal characteristics such as race, age, history of prior preterm delivery, and smoking history were measured directly from data included in the birth certificate databases.

We began by describing the extent of regionalization in each of the four states, by calculating the percentage of all LBW deliveries at each hospital level for each year in the 10-year study period. Pair-wise statistical tests of equivalent means (two sample tests of proportion) were conducted between 1989 and 1998, to identify statistically significant change across the 10-year study period. A Bonferroni correction was applied to analyses for multiple comparisons (4 states ×3 hospital levels ×3 birth weight groups: LBW, VLBW, and ELBW; 36 comparisons), adjusting the level of significance, α=.05 to α=.001.

The focus of analyses was on changes at level III hospitals; specifically, evidence of a declining LBWP at level III hospitals would be suggestive of deregionalization. The effect of managed care penetration on the extent of perinatal regionalization was tested in a series of fixed effects panel data models for each state,27 with data aggregated to the hospital level for each year. The panel data model includes data for each year (up to 10 years) for each hospital. To control for within-hospital correlation in the covariates, dummy variables are included for each hospital. The use of panel data models with many counties per state (and a correspondingly large number of measures of HMO penetration per year) allows the identification of any HMO effect on perinatal regionalization, including an association with slowed regionalization, or deregionalization. A positive coefficient (eg, +1.0) for the measure of HMO penetration would indicate that an increase in HMO penetration by 1 percentage point is associated with an increase in the LBW percentage of 1 percentage point. All models included as covariates maternal race (white compared with nonwhite), trimester prenatal care was initiated (first compared with second or third, maternal age [aged younger than 20 years compared with 20 years or older]), and singleton birth (compared with other). Additional covariates were percentage of prenatal care paid for by Medicaid (California and Washington), percentage of married mothers and percentage of mothers who smoke (Illinois, NC, and Washington), and percentage of mothers with prior preterm delivery (Washington).

Models were conducted separately for each level, to allow the effect of managed care penetration and other factors to vary according to level of hospital. The coefficient standard errors were adjusted explicitly to control for the clustering effects introduced by county-level measurement of HMO penetration.28,29 The study was powered to detect differences in the percentage of LBW deliveries occurring at level III hospitals 5% or lower (depending on the number of LBW deliveries in each state) at the α=.0014 significance level (adjusted for multiple comparisons), requiring at least 1,424 LBW deliveries per state and hospital level. The study was further powered to detect an increase in the overall model R2 of 0.10 from the inclusion of HMO penetration in the regression models at the α=0.0014 significance level, requiring at least 71 hospitals over the ten year time period per state and hospital level. An exemption for this research was obtained from the Institutional Review Board of Northwestern University.


A total of 8,479,144 deliveries were reported at 615 separate facilities meeting our eligibility criteria (50+ annual deliveries, nonmilitary) in Washington, California, North Carolina, and Illinois across the 10-year study period. Facilities for which the level could not be verified for a given year were excluded from the study for that year, resulting in the exclusion of approximately 14% of the total deliveries (final n= 7,238,400 deliveries).

The number of hospitals in 1989 for California, Illinois, North Carolina, and Washington were 235, 151, 69, and 60, respectively. Although the distribution of hospital level was relatively stable across time for each state, substantial variation in the total number of hospitals was observed in some states. For example, the number of hospitals included in this study from California dropped from 235 in 1989 to 159 in 1998 (−32%), with 85% of the decline observed in 1997 and 1998. The number of hospitals in the remaining states included in this study changed by −3%, 9%, and 14% for Washington, North Carolina, and Illinois, respectively, across the 10-year study period, with some annual fluctuation. Some of the variations in number of hospitals are likely due to a real change in the number of separate facilities eligible for inclusion in the study, but others seem to be fluctuations in annual reporting to the American Hospital Association.

The facilities in this study varied substantially in multiple key characteristics. In 1998, the number of deliveries per hospital varied from as few as 54 at a level I hospital in Illinois to 6,742 at a level III hospital in California. On average, the largest hospitals were in California (1,672 deliveries) and the smallest hospitals were in Washington (1,071 deliveries). Across all four states, there were on average 631 deliveries at level I hospitals, 1,527 deliveries at level II hospitals, and 2,826 deliveries at level III hospitals.

The proportion of all LBW, VLBW, and ELBW deliveries at each level of hospital for 1989 and 1998 is given in Table 1. Substantial variation in regionalization was observed across the four states.

Table 1
Table 1:
Percentage of Low Birth Weight, Very Low Birth Weight, and Extremely Low Birth Weight Deliveries by Level of Hospital and State for 1989 and 1998

By state, Illinois and North Carolina show strong signs of increasing regionalization, because the percentage of births in level III went up in all low birth weight categories. Washington, which already had the highest level of regionalization in 1989, showed very little change.

California seems to have deregionalized VLBW births, with little change in ELBW and LBW deliveries. In 1998, the percentages of LBW, VLBW, and ELBW births in level III hospitals in California were by far the lowest of all four states.

Penetration of HMOs was measured on the county level for every county containing an eligible facility for this study. Mean penetration, measured by the percentage of the population enrolled in an HMO, and percentage increase in penetration, are reported in Table 2. Data from California for 1998 were excluded from the current study due to reporting error, so the California data are limited to the 1989–1997 period.

Table 2
Table 2:
Selected Facility Characteristics and Health Maintenance Organization Enrollment

A total of 36 regression models were conducted to test the effect of managed care penetration, measured by the percentage of each county’s population enrolled in an HMO, on the percentage of deliveries that are low birth weight (500–2,500 g), VLBW (500–1,500 g) and ELBW (500–1,000 g). The panel data models were conducted across the study period for each hospital level and for each state. These data are available online at Coefficients for HMO penetration for each state and level are presented in Table 3.

Table 3
Table 3:
Coefficients on Health Maintenance Organization Penetration From 36 Fixed Effects Regressions of Health Maintenance Organization Penetration on Percent of Deliveries that are Low Birth Weight, Very Low Birth Weight, or Extremely Low Birth Weight

Despite the substantial variation in the percentage of deliveries at each hospital that are LBW, and in the HMO penetration across counties, HMO penetration is not a significant predictor of the LBW or ELBW delivery rate at each hospital level, and is only a significant (positive) predictor of the VLBW for level III hospitals in North Carolina (a 1 percentage point increase in HMO penetration is associated with a 0.04 percentage point increase in the VLBW percentage at level III hospitals). Significant predictors included in the analysis but not reported in the table generally include the percentage of mothers who smoked during pregnancy, the percentage of mothers who are married, and the percentage of mothers who began prenatal care in the first trimester.


The goal of our study was to describe trends in regionalization of perinatal care and to identify factors that predict the extent of regionalization across the 10-year study period. This study was driven by evidence supporting the use of perinatal networks to reduce neonatal morbidity and mortality4–6,12 and by concerns about the reported deterioration of these networks. A strength of this study is the use of data from four very different states across an extended, 10-year time period.

Our study demonstrated significant variation in the extent of regionalization of perinatal care across four states. We find in the current study that, although significant changes in concurrent years do exist for each measure in each state, trends are actually suggestive of some level of reregionalization in the later years of the study period for three of the four states, with increasing percentages of LBW deliveries occurring at level II or level III hospitals or both. When one considers all LBW deliveries, which includes many neonates who may be expected to have good outcomes even without level III care, increased use of level III hospitals was seen in Illinois and North Carolina, and increased use of level II hospitals was seen in Illinois and California. The VLBW and ELBW are the neonates that are most in need of high levels of medical care, and North Carolina and Illinois showed substantial movement of both to Level III hospitals as well. California showed a very different pattern strongly suggestive of deregionalization toward level II hospitals. Births in both of these categories remained in level III hospitals at very high frequencies in Washington.

Despite these changes in site of delivery, the percentages of high-risk deliveries (measured by deliveries of VLBW neonates) occurring at level III hospitals were substantially lower than the goal of 90% set by Healthy People 2010.17 In 1998, only 42% of VLBW deliveries occurred at level III hospitals in California, 74% in Illinois, 73% in North Carolina, and 80% in Washington.

Penetration of HMOs has increased substantially in all four states and presents financial incentives to restrict use of more expensive sites of care, as evidenced by a recent study of hospitals in Illinois that found that high HMO revenues were associated with nontransfer of infants less than 1,250 g born in nontertiary centers.30 However, HMO penetration was not found to play a significant role in the extent of regionalization across the study period, even after controlling for multiple maternal predictors of low birth weight. Health maintenance organization penetration was a significant predictor of the LBW, VLBW, or ELBW percentage in only one of 36 models, a finding that can be attributed to chance. In fact, it is possible that HMO penetration may have an effect opposite what was expected in this study. Prior work has demonstrated a link between HMO penetration and slower adoption of mid level neonatal intensive care units, which may actually promote the use of tertiary centers for the delivery of high-risk infants.31

Changes in the populations served by different hospitals are not entirely independent of the HMO effect under study, because HMO penetration may affect regionalization through HMO enrollment patterns (ie, lower-risk patients), as well as through restrictive HMO hospital referral policies or selective hospital contracting. Similarly, the types of shifts in patient populations referred to may reflect either “false” or “true” deregionalization, depending on, for example, whether shifts of low risk births to one level III hospital came from another level III hospital (“false” deregionalization for that hospital) or from a level I or II hospital (“true” deregionalization). To account for shifts in the populations served by hospitals, analyses controlled for maternal characteristics expected to affect high-risk delivery, including race, age, history of prior preterm delivery, and smoking history (aggregated to the facility level per year). Moreover, both the changes across time within given hospitals and across all hospitals were examined. Shifts of low-risk populations into level III facilities from lower level facilities would be detected by the across-hospital analysis to the extent that such lower-risk births are “coming from” lower-level hospitals.

The use of secondary data resulted in several limitations to the study. The four states were selected on the basis of a variety of factors, including expected HMO penetration, and do not therefore represent a generalizable sample. Hospital level data were not consistently available for the full study period for all eligible hospitals. In particular, substantial drop-off in hospital data were observed for California in the last 2 years of the study. The use of panel data models minimizes the potential effect of missing data by including cross-sectional and longitudinal data. Limited data were available from the birth certificate databases to control for predictors of birth weight. Omission of maternal predictors of birth weight (eg, clinical risk factors and detailed prior birth history) and non-HMO system trends (eg, measures of market competition) in the models reduced the predictive power of the analytic models. Another limitation with the use of birth certificate data is that predictors of birth weight available in the birth certificate databases are likely to be imperfectly reported.32 However, these omissions can be expected not to bias estimates of the HMO effect on regionalization provided that they are not correlated with facility-level measures of HMO penetration. In addition, level of hospital was not always available in electronic records and could not uniformly be confirmed by telephone. Missing observations were most likely not randomly distributed across hospital levels, possibly leading to bias in the study models. Patterns of regionalization, insurance mix, and other factors affecting the delivery of care are likely to be region-specific. The study focused on only one type of managed care, HMOs. However, HMOs are generally the most restrictive in their coverage, by limiting coverage to providers within a network and usually requiring referrals for specialty care. It is probably unlikely that a relationship between perinatal regionalization and other, less restrictive, forms of managed care exists, given that no significant relationship was found between HMOs and perinatal regionalization.

Two factors not explored in this study may be factors in the regionalization of perinatal care. Hospitals may increase the number of subspecialists with the goal of enhancing their ability to care for high-risk neonates. In the years between 1978 and 1998 the American Board of Pediatrics certified more than 3,000 neonatologists, and 80% of practicing neonatologists today are younger than 50 years old.33,34 The increasing numbers of subspecialists may in particular be a factor in the increase in LBW deliveries at level II hospitals observed in California and Illinois. However, it is possible that the presence of neonatal intensive care units in pediatric hospitals without maternity units encourages the delivery of high-risk infants at community hospitals, with the intent to transfer for specialty care after delivery.

Our study demonstrates significant failures in meeting Healthy People 2010 goals for site of delivery of high-risk infants that cannot be blamed on the financial pressures introduced by managed care. Directed studies of market and institutional barriers to regionalization are needed to identify steps to improve states’ achievement of the Healthy People 2010 goals.


1.McCormick MC. The regionalization of perinatal care. Am J Public Health 1981;71:571–2.
2.Modanlou HD, Dorchester W, Freeman RK, Rommal C. Perinatal transport to a regional perinatal center in a metropolitan area: maternal versus neonatal transport. Am J Obstet Gynecol 1980;138:1157–64.
3.Obladen M, Luttkus A, Rey M, Metze B, Hopfenmuller W, Dudenhausen JW. Differences in morbidity and mortality according to type of referral of very low birthweight infants. J Perinat Med 1994;22:53–64.
4.Towers CV, Bonebrake R, Padilla G, Rumney P. The effect of transport on the rate of severe intraventricular hemorrhage in very low birth weight infants. Obstet Gynecol 2000;95:291–5.
5.Yeast JD, Poskin M, Stockbauer JW, Shaffer S. Changing patterns in regionalization of perinatal care and the impact on neonatal mortality. Am J Obstet Gynecol 1998;178:131–5.
6.Hernandez JA, Hall DM, Goldson EJ, Chase M, Garrett C. Impact of infants born at the threshold of viability on the neonatal mortality rate in Colorado. J Perinatol 2000;20:21–6.
7.Paneth N, Kiely JL, Wallenstein S, Marcus M, Pakter J, Susser M. Newborn intensive care and neonatal mortality in low-birth-weight infants: a population study. N Engl J Med 1982;307:149–55.
8.Paneth N, Kiely JL, Wallenstein S, Susser M. The choice of place of delivery: effect of hospital level on mortality in all singleton births in New York City. Am J Dis Child 1987;141:60–4.
9.Gortmaker S, Sobol A, Clark C, Walker DK, Geronimus A. The survival of very low-birth weight infants by level of hospital of birth: a population study of perinatal systems in four states. Am J Obstet Gynecol 1985;152:517–24.
10.McCormick MC, Shapiro S, Starfield BH. The regionalization of perinatal services: summary of the evaluation of a national demonstration program. JAMA 1985;253:799–804.
11.Harris TR, Isaman J, Giles HR. Improved neonatal survival through maternal transport. Obstet Gynecol 1978;52:294–300.
12.Phibbs CS, Bronstein JM, Buxton E, Phibbs RH. The effects of patient volume and level of care at the hospital of birth on neonatal mortality. JAMA 1996;276:1054–9.
13.LeFevre M, Sanner L, Anderson S, Tsutakawa R. The relationship between neonatal mortality and hospital level. J Fam Pract 1992;35:259–64.
14.Powell SL, Holt VL, Hickok DE, Easterling T, Connell FA. Recent changes in delivery site of low-birth-weight infants in Washington: impact on birth weight-specific mortality. Am J Obstet Gynecol 1995;173:1585–92.
15.Bode MM, O’Shea TM, Metzguer KR, Stiles AD. Perinatal regionalization and neonatal mortality in North Carolina, 1968-1994. Am J Obstet Gynecol 2001;184:1302–7.
16.Cifuentes J, Bronstein J, Phibbs CS, Phibbs RH, Schmitt SK, Carlo WA. Mortality in low birth weight infants according to level of neonatal care at hospital of birth. Pediatrics 2002;109:745–51.
17.U.S. Department of Health and Human Services. Healthy People 2010. Conference edition in two volumes. Washington (DC): Department of Health and Human Services; 2000.
18.Gould JB, Marks AR, Chavez G. Expansion of community-based perinatal care in California. J Perinatol 2002;22:630–40.
19.Mehta S, Atherton HD, Schoettker PJ, Hornung RW, Perlstein PH, Kotagal UR. Differential markers for regionalization. J Perinatol 2000;20:366–72.
20.Gagnon D, Allison-Cooke S, Schwartz RM. Perinatal care: the threat of deregionalization. Pediatr Ann 1988;17:447–52.
21.MCOL. Managed care fact sheets: managed care national statistics. 2006. Available at: Retrieved June 14, 2006.
22.Simon CJ, Dranove D, White WD. The impact of managed care on the physician marketplace. Public Health Rep 1997;112:222–30.
23.Wholey DR, Christianson JB, Engberg J, Bryce C. HMO market structure and performance: 1985–1995. Health Aff (Millwood) 1997;16:75–84.
24.Gerber SE, Dobrez DG, Budetti PP. Managed care and perinatal regionalization in Washington State. Obstet Gynecol 2001;98:139–43.
25.Rosenblatt RA, Macfarlane A, Dawson AJ, Cartlidge PH, Larson EH, Hart LG. The regionalization of perinatal care in Wales and Washington State. Am J Public Health 1996;86:1011–5.
26.Wholey D, Feldman R, Christianson JB. The effect of market structure on HMO premiums. J Health Econ 1995;14:81–105.
27.Greene WH. Econometric analysis. 3rd ed. Upper Saddle River (NJ): Prentice Hall; 1997. p. xliv, 1075.
28.Moulton BR. An illustration of a pitfall in estimating the effects of aggregate variables on micro units. Rev Econ Stat 1990;72:334–8.
29.Deaton A. The analysis of household surveys: a microeconometric approach to development policy. Baltimore (MD): Johns Hopkins University Press; 1997.
30.Wall SN, Handler AS, Park CG. Hospital factors and nontransfer of small babies: a marker of deregionalized perinatal care? J Perinatol 2004;24:351–9.
31.Baker LC, Phibbs CS. Managed care, technology adoption, and health care: the adoption of neonatal intensive care. Rand J Econ 2002;33:524–48.
32.Dietz PM, Adams MM, Kendrick JS, Mathis MP. Completeness of ascertainment of prenatal smoking using birth certificates and confidential questionnaires: variations by maternal attributes and infant birth weight. PRAMS Working Group. Pregnancy Risk Assessment Monitoring System. Am J Epidemiol 1998;148:1048–54.
33.Goodman DC, Fisher ES, Little GA, Stukel TA, Chang CH. Are neonatal intensive care resources located according to need? Regional variation in neonatologists, beds, and low birth weight newborns. Pediatrics 2001;108:426–31.
34.Pollack LD, Ratner IM, Lund GC. United States neonatology practice survey: personnel, practice, hospital, and neonatal intensive care unit characteristics. Pediatrics 1998;101:398–405.

Cited By

This article has been cited 1 time(s).

Pediatric Critical Care Medicine
Extracorporeal membrane oxygenation and term neonatal respiratory failure deaths in the United Kingdom compared with the United States: 1999 to 2005
Brown, KL; Sriram, S; Ridout, D; Cassidy, J; Pandya, H; Liddell, M; Davis, C; Goldman, A; Field, D; Karimova, A
Pediatric Critical Care Medicine, 11(1): 60-65.
PDF (278) | CrossRef
© 2006 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.