competition, quality and neonatal intensive care

8 downloads 0 Views 2MB Size Report
Oct 6, 2000 - Chapter 2. The Dynamics of Hospital Competition Across Economic ..... to direct patients to specialty centers for cardiac care at one period in time in the late 1970s. (55) One ...... 150722 Bakersfield Memorial Hospital. 3. 3. 3. 3.
COMPETITION, QUALITY AND NEONATAL INTENSIVE CARE IN CALIFORNIA, 1986-1997

By Ellen R. Shaffer

Dissertation/thesis

submitted to the Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy

Baltimore, Maryland 2001

Copyright Ellen R. Shaffer 2000 All rights reserved

ABSTRACT Problem: Regionalized systems of perinatal care have been shown to reduce neonatal mortality, particularly for low birth weight neonates and others at high risk. Regional systems were designed to direct high-risk births to high volume neonatal intensive care units (NICUs) at the tertiary level of care (Level III). During the 1980s and 90s, however, there was substantial growth in California of lower level, lower volume NICUs, with some evidence of higher mortality rates. Only 31.5% of low birth weight infants in California were delivered at Level III hospitals in 1990, far below the national goal of 90%. This study examines the relationship between particular competitive factors related to maternity care, the availability of lower level NICUs across distinct regions in California over time, and risk-appropriate admissions to tertiary level NICUs.

Methods: The study is a longitudinal, quantitative analysis that examined changes from 1986 to 1997 across 17 Maternal and Child Health Perinatal regions in California. It used data on hospital discharges and NICU levels compiled by the California Office of Statewide Health Planning and Development (OSHPD) and California Children’s Services, linked with birth and death certificates. The testing of the hypotheses involved the estimation of time series models, pooling data over 12 years and across 17 regions, and estimated competitive effects using random effects models. Dependent variables, for each region over time, are the number of lower level NICUs per region, the number of births per lower level NICU bed, and the percentage of all very low birth weight births at Level III and/or high 2

volume Level II+ hospitals. Explanatory competitive variables include licensed beds and occupancy rates for acute hospitals, perinatal units, and NICUs; number of births and low and very low birth weight births; Herfindahl indexes for hospital, perinatal and NICU discharges; and proximity of lower level NICUs to a Level III hospital. The shift in market power to the demand side is measured by the percentage of births covered by price-sensitive payors. Clinical need, which also represents demand, is measured by the number of births and very low birth weight births.

Results: Findings suggest that competitive factors including more competitive factors including an increasingly wide distribution of perinatal discharges, and declining hospital occupancy rates, were significantly associated with increasing numbers of lower level NICUs, and with greater availability of lower level NICU beds expressed as a smaller number of births per NICU bed. Measures of population need were not significantly related to expanded availability. Greater MediCal coverage was associated with fewer NICU units, but HMO coverage had no effect. Neither HMO nor MediCal coverage was associated with the number of NICU beds. Lower level NICUs were likely to have a higher percentage of births covered by an HMO. As expected, the percentage of very low birth weight births at Level III hospitals declined as the number of lower level NICUs increased. The percentage of very low birth weight babies at all risk-appropriate sites, including high-volume Level II+ hospitals, increased slightly between 1986 and 1997 from about 45% to about 49%, well below the Healthy People 2010 goal of 90%. A number of competitive factors were 3

associated with declines in risk-appropriate deliveries. A higher percentage of MediCal coverage was associated with a higher rate of risk-appropriate deliveries; the percentage of HMO coverage was not significant. There was significant variation among the 17 regions, including both competitive and population characteristics.

Conclusions: These findings suggest that the availability of lower level NICUs is not related to need, but is a response in part to competition among hospitals. Price competition through private sector HMOs failed to reduce availability or to improve risk-appropriate deliveries, although MediCal coverage was associated with improvements in both measures. The regional system for assuring the optimal distribution of high quality neonatal care is not performing up to potential. Availability of NICUs is consistent with the theory that hospitals in a price competitive environment may compete for market share by offering specialized services to differentiate themselves from competitors, and also to attract insured maternity patients.

Implications for Policy, Delivery or Practice: Over 500,000 women a year give birth in California, up to 24% of whom may be at risk for complications. Since competitive factors are related both to greater availability of lower level NICUs and to the rate of riskappropriate admissions, policy interventions to mitigate the effects of competition may be considered. These interventions could include developing and enforcing uniform standards for NICUs at each level, incorporating volume as well as staffing levels in the standards, 4

requiring that all NICUs meet the standards as a condition of operation and in order to receive reimbursement from any source, reinforcing the regional approach to configuring perinatal services in order to provide optimal neonatal survival, and strengthening Healthy People 2010 goals for risk-appropriate admissions.

5

Table of Contents Page Abstract

2

Table of Contents

5

List of Tables

8

List of Figures

10

List of Appendices

11

Chapter 1. Introduction and Background 1.1. Introduction

12

1.2 Background and Significance: Infant Mortality in the U.S.

14

1.3 Emergence of Regionalization of NICUs

15

1.4 Regionalization of Neonatal Intensive Care: Levels Established

18

1.5 NICUs: Volume, Level of Care and Outcomes

21

1.6 MCOs and Risk-Appropriate Care

22

1.7 California Perinatal Regions and Market Areas

22

1.8 Role of Kaiser Hospitals

24

1.9 MediCal Coverage and Reimbursement Rates

24

1.10 Growth of Level II NICUs in California

26

1.11 California regulation of NICUs

28

Chapter 2. The Dynamics of Hospital Competition Across Economic Environments in the U.S. 2.1 1945-1982: Medical Arms Race: Cost and Access at Odds 6

30

2.2 Price Competition

33

2.3 Summary of the Evidence

43

Chapter 3. Research Design and Methods 3.1 Conceptual Framework

47

3.2 Research Aims and Hypotheses

49

Chapter 4. Data Sources and Definition of Variables 4.1 Three Analysis Files

56

4.2 Major Variables

57

4.3 Approach to the Analysis

68

4.4 Data Analysis

74

Chapter 5. Results of Trend Analysis 5.1 Trends in Neonatal Care

79

Chapter 6. Results 6.1 Hypothesis IIa. Dependent variable #1: No. II/II+ NICU beds

91

6.2 Hypothesis IIa. Dependent variable #2: Ratio of births to NICU beds

97

6.3 Hypothesis IIb

106

6.4 Hypothesis III. Risk-appropriate delivery of VLBW births

107

6.5 Hypothesis III, Dependent variable #2: PVLB23

113

Chapter 7. Discussion

124

7.1 Limitations

128 7

7.2 Policy implications

131

7.3 Future studies

132

Appendices

134

Bibliography

182

Curriculum Vitae

187

8

LIST OF TABLES

Table

Title

Page

5.1

Change in number of hospitals by NICU level, VLBW births, and normal births, 1986-1997

79

5.2

Statewide neonatal care trends 1986-1997 Major dependent and independent variables

80

5.3

Regional trends in neonatal care: change in regional means, all regions

83

5.4

Percentage of mothers covered by HMO: with and without Kaiser regions; Regional means

85

5.5

Regional trends in neonatal care: change in regional means. Selected variables, non-Kaiser regions

85

5.6

Regional trends in risk-appropriate VLBW deliveries (High Volume II+/III). Regional means

87

5.7

Percentage VLBW at each level: statewide average

87

5.8

Market share: % of normal deliveries by hospital NICU level, 1986-1997

89

6.1

Hypothesis IIa. Dependent Variable #1: No. II/II+ NICUs (LV22PL)

6.2

Hypothesis IIa. Solution for random effects: significant regional variations

96

6.3

Trends in NICU beds

98

6.4

Trends in availability

99

6.5

Hypothesis IIa. Dependent Variable #2: Ratio of Births to II/II+ NICU beds (B22NB)

100

9

95

Table

Title

Page

6.6

Solution for random effects

102

6.7

Hypothesis IIb. Dependent variables: NICU levels

105

6.8

Hypothesis III. Percentage VLBW Births at Level III Hospital (VLB3P)

6.9

Percentage VLBW at Level III, without regions 3,5,10,11. Regional means

112

6.10

Solution for Random Effects

112

6.11

Hypothesis III. Dependent variable #2: percentage VLBW 117 births at high-volume Level II+ or Level III Hospital (PVLB23)

6.12

Solutions for random effects

119

6.13

Regions where PVLB23 increased

121

6.14

Regions where PVLB23 decreased

122

10

110

LIST OF FIGURES Figure

Title

Chapter 3 1

Page 46

Conceptual Framework

11

Appendices Appendix

Page

A

Infant Mortality Rates

134

B

Regionalization and Quality: Lower Risk-Adjusted Neonatal Mortality at Large II+ and Level III By Birthweight, 1992-93

136

C

California Perinatal Regions

137

D

Perinatal Region and County

138

E

California Hospital NICUs by Level and Year

142

F

California NICUs and Number NICU Beds by Region

149

G

Variable Definitions

159

I

High-Volume Level II+ Hospitals, 1997

168

J

High-Volume Level II+ Hospitals, 1986-1997

169

K

Market Share of Normal Deliveries by Hospital NICU Level, 1986-1997

174

L

Percentage VLBW Born at Level III, 1986 & 1997

176

M

Regional Analysis: % of Risk-Appropriate VLBW Births,1986-97, Volume + Level Used to Define Risk-Appropriateness

177

N

Percentage VLBW Births by Region and NICU Level: 1986 178

O

Percentage VLBW Births by Region and NICU Level: 1997

P

California Hospitals By NICU Levels, 1986-1997

Q

Results, Statistical Correlations

R

Results, Statistical Mixed Model Regressions 12

180

179

CHAPTER 1.1 Introduction Reducing neonatal mortality is one of the health objectives of the United States. Regionalized networks for perinatal care have been shown to be effective in improving neonatal survival. These networks facilitate access to hospitals that are organized and equipped to provide optimum care for neonates at high risk for morbidity and mortality, in part by identifying and designating levels for these hospitals. Virtually all low and very low birthweight infants require care at the highest level facilities, designated as Level III. National goals call for 90% of mothers and infants to receive risk-appropriate care, and California aims to ensure that 90% of very low birth weight births are delivered at Level III hospitals.(74) However only 31.5% of low birth weight infants in California were delivered at Level III hospitals in 1990 (75).

Neonatal intensive care units (NICUs) at the tertiary level of care (Level III), and highvolume Level II+ NICUs with an average daily census of at least 15, have been shown to reduce neonatal mortality for low birth weight neonates and others at high risk. (14, 62, 64, 71, 75, 78, 103, 104) In contrast, neonatal mortality rates at Level II and lower volume Level II+ NICUs are higher than rates at Level III and high-volume Level II+ NICUs, and are no better than rates at hospitals with no NICU. (14, 75, 95) Nevertheless, there was 13

substantial growth in California and elsewhere of lower level, lower volume NICUs during the 1980s. (32, 75)

In an earlier era, analysts attributed the duplication of expensive, specialized clinical facilities to a combination of nonprice competition among hospitals for physicians, and an open-ended reimbursement environment. (56) Since higher volume providers are associated with better outcomes for many conditions, wide availability of specialized services also was seen as a threat to quality, to the extent that it resulted in lower volume for each facility. Price-based competition was expected to dampen the "medical arms race," leading to reductions in costs, and with potential benefits in quality of care. Subsequent observers have found however that hospitals in the price-competitive environment in California have been able to reduce costs without concentrating specialized services in high-volume facilities. (107) California offers an excellent opportunity to examine the extent to which a market-driven, price sensitive reimbursement environment may affect the ability to achieve national health goals for low and very low birth weight infants.

Over 500,000 women a year give birth in California. Estimates of potentially high-risk births range from 6% to as many as 24%, (43) depending on the definition. Preterm delivery, which is strongly associated with low birth weight, and with infant morbidity and mortality, occurs in about 11% of cases.(75) If competition is linked to greater availability 14

of lower level NICUs and to worse outcomes, policy interventions may be sought by women, the public at large, providers, health plans, and elected officials.

These

interventions could include developing uniform standards for certification of new and existing NICUs at each level of care, including volume as well as staffing requirements for the highest levels of care, requiring certification by level for reimbursement by any payor, and strengthening the regional configuration of perinatal services required to provide optimal neonatal survival. (83, 100)

This study examined the relationship between hospital competition, price-sensitive purchasers, clinical need, and the availability of lower level NICUs across regions in California from 1986 through 1997. This period was dominated by price-based competition and increasing enrollment in managed care organizations. The study examined whether there is evidence that hospital competition and price-sensitive payors contributed to excess availability of NICUs, constrained it, or played no role. It also examined the effects of competitive forces and greater NICU availability on one indicator of the quality of neonatal care, specifically on risk-appropriate admissions to Level III and high-volume Level II+ NICUs.

The study is a longitudinal, quantitative analysis that examined changes from 1986 to 1997 across 17 perinatal regions in California. It used data from hospital discharge abstracts for infants, compiled by the California Office of Statewide Health Planning and Development, 15

linked with birth and death certificate data, when death occurred within a year of birth.

BACKGROUND AND SIGNIFICANCE 1.2 Infant mortality in the U.S. Infant mortality, defined as deaths per 1,000 live births during the first year of life, is higher in the U.S. than in 21 other industrialized nations. (67) The rate of preterm deliveries, about 11%, is about double that of many other developed nations. The mortality rate for AfricanAmerican infants, 14.1% in 1998, is more than double the 6% rate for white infants (60), a wider gap than in 1992.(43) The neonatal mortality rate (deaths per 1,000 from birth to day 28) was static at 4.8 in 1997 and 1998, with the 1998 rate for African American babies at 9.6 compared with 4.0 for all other races and ethnicities. Babies born at low birth weights, and particularly very low birth weights, either because they are born early or because intrauterine growth has been retarded, account for most newborn mortality and morbidity, followed by congenital anomalies. Low birth weight (LBW) is defined as a weight of 2,500 grams or less at birth, and very low birth weight (VLBW) as 1,500 grams, about 3 lb. 4 oz., or less. Infant mortality is a function of both low birth weight and birthweight-specific mortality (38).

Progress in identifying the complex causes of preterm births and associated low birth weight has advanced slowly in the U.S., and primary prevention strategies such as social support programs have not yet proven successful.(43) Progress toward the goal of 90% of 16

births at a risk-appropriate level of care could significantly improve neonatal survival. Progress toward achieving this goal may also serve as a measure of the organizational and financial effectiveness of the health care system.

1.3 Emergence and Regionalization of NICUs Neonatal Mortality: Advances in Public Health, Technology and Staffing Advances in public health, technology, and health care delivery have significantly reduced infant and neonatal mortality in the twentieth century. (See Appendix A.) Most sick newborns died within the first few hours of life until the earliest incubator was introduced in 1878.(11) After World War II, as sanitation, immunizations and antibiotics further reduced infant mortality past the first month of life, attention turned to the more persistent problems of low birthweight infants and death in the neonatal period (birth to 28 days). Nevertheless there was only gradual change in the rate of neonatal mortality (11) until the late 1960s. Social forces including improved education, expanded insurance coverage through Medicaid, subsidized nutrition programs, and the legalization of abortion helped reduce the infant and neonatal mortality rates after 1965.(19)

Neonatal intensive care units that combined trained staff and advanced technology have also been critical. New technologies developed in the late 1960s included effective mechanical ventilation (62) to treat respiratory distress syndrome, better monitoring to prevent blindness caused by oxygen therapy, and "microchemistry" lab tests using minute samples of blood. 17

(11) In 1995 a review of neonatal intensive care technologies enumerated 15 categories of commonly available procedures, from monitoring and diagnosis to respiratory support, pharmaceuticals, surgery, and psychosocial interventions. (45) The most important changes since 1989 are the more regular use of antenatal steroids, high frequency ventilation, and exogenous surfactant. (2)

In addition, NICU staffing is intensive and organized to respond to the special needs of patients.

Laboratory and X-ray technicians and respiratory therapists are assigned

exclusively to the NICU. Nurses at the highest level units typically care for one patient. Physician subspecialists such as pediatric anesthesiologists and surgeons are regularly on call and are experienced with NICU infants and their problems.

Special training for nurses and doctors who deliver neonatal intensive care began in the mid-1960s, leading to the development of neonatology as a board-certified pediatric subspecialty in 1975. (91) In 1995 there were 106 approved training programs, and about 2,500 board-certified neonatologists, as well as many pediatricians practicing as neonatologists who were not board certified.(45)

There was another another sharp drop in the neonatal mortality rate from 7.0 in 1985 to 5.8/1,000 in 1990 (67, 92) Based on data presented in the NCHS Chartbook 1997, the present study calculated that there was a 23% drop in mortality for infants in the 1,00018

1,499 gram group between 1989 and 1991, when other birth weight categories were changing only about 10%. (See Appendix A) The introduction of surfactant therapy to lower surface tension of the lungs was one factor contributing to the decrease in the neonatal mortality rate among VLBW babies between 1989 and 1990 (91), and to a 33% decrease in deaths due to respiratory distress syndrome between 1988 and 1990 (10). The birthweight threshold for viability as also fallen. Babies between 500 and 600 grams (1 pound 5 ounces), and between 22 and 24 weeks gestation, may now survive. Improvements in survival among very low birth weight infants have increased the demand for NICU services.

Overall, between 1970 and 1995, the neonatal mortality rate was reduced by two-thirds, from 15.1 to 4.9/1,000, while the postneonatal mortality rate (deaths from day 20 to one year) fell by less than 50%, from 4.9 to 2.7/1,000. (101)

1.4 Regionalization of Neonatal Intensive Care: Levels Established

Regionalization of high intensity specialized NICU services offers a rational method to distribute health care resources based on geography, while promoting quality and efficiency of care. (35) Regionalization depends on cooperation among hospitals to triage and transfer patients appropriately. When neonatal technologies emerged during the 1960s and 1970s, public policy was supportive of government programs to encourage 19

regionalization through planning authorities.

NICUs were initially distributed during the 1970s and 1980s on a regional basis, to assure access and high quality at a reasonable cost. (64) Driven by the initial scarcity of skilled personnel and the expensive equipment required to operate NICUs, regionalization was underway as early as a 1967 statewide program in Wisconsin, followed by programs in rural Arizona and urban Quebec and Montreal.

In the early 1970s, the March of Dimes convened medical specialty groups and consumers to recommend guidelines for a regional perinatal care system, covering both pregnancy and infant care. Their 1976 report, Toward Improving the Outcomes of Pregnancy (TIOP), defined central features of a "systematized, cohesive regional network," intended to be cost efficient, in which "the complexity of patient needs determines where, and by whom, the care should be provided": 1. Identification of the degree of complexity of care each hospital is capable of providing, by mutual agreements between hospitals and physicians, and based upon population needs; 2. Maximal use of highly trained staff; 3. A communication network to provide consultation and education for all staff; and 4. A transportation system to facilitate transfers for selected maternity cases and sick newborns. It was believed that timely assignment of high risk mothers to the highest level of care appropriate would minimize transfer of a low birthweight infant after birth, which is riskier. In return, community doctors and hospitals 20

would rely on tertiary care facilities to communicate with them regarding their patients and to return them when stable.(17)

It was anticipated that regions with 8,000 to 12,000 live births yearly would have an adequate population and economic base to support the large investment required to open one center to provide care for extremely high risk births. Most complications of pregnancy and newborn abnormalities, which are less severe, could be treated in units providing only moderately complex care, serving a population base of about 2,000 live births a year.

The report also recommended three levels of neonatal care reflecting the intensity of both technology and staffing, which remain the basis of NICU designations:

Level I: These hospitals provide services primarily for uncomplicated maternity and newborn patients. They can stabilize unexpected complications prior to transfer but offer no special equipment or staffing. The TIOP report recommended minimizing maternity services at this level where feasible. Where geographic remoteness or transportation difficulty require continuing Level I delivery services, hospitals are encouraged to develop channels for consultation, referral and transfer in the case of unexpected complications.

Level II: These hospitals can provide care for complications including respiratory 21

distress for a limited time period, offering some 24-hour services and more highly trained personnel than Level I. The TIOP report envisioned that these larger urban and suburban hospitals would care for the majority of complicated obstetrical problems and certain neonatal illnesses that are not life-threatening.

TIOP

anticipated that the level of care at these hospitals would vary, with some transferring out seriously ill newborns, and others upgrading to a Level III.

Level III: These hospitals provide the full range of services and resources, including advanced ongoing respiratory support and surgery, immediate 24-hour availability of subspecialists, and concentrated staffing by skilled nurses. They are designated to offer consultation services and continuing education for all community hospitals and staff, and transport services including equipment, staff, and coordination. They may also be engaged in clinical or basic research, and may or may not be located within academic medical centers.

Subsequently additional levels of care and definitions of those levels have emerged. While the three TIOP levels form the basis of the NICU classifications, there is variation among states and regions in how these levels are defined and designated. A Maryland committee convened by the state identified five levels of facilities in 1998, adding levels III+ and IV to the TIOP definitions. (73) A recent review of eleven states found that seven used the traditional TIOP designations, four states varied, and in almost all cases the designations 22

were established voluntarily. (99)

There are no uniform standards for designating NICU levels in California. NICUs must be licensed by OSHPD in order to operate in California. However, OSHPD does not designate different levels of care. NICUs may choose to be certified by California Children's Services (CCS), which does designate levels. Hospital NICUs must be certified by CCS to receive MediCal reimbursement.

Hospitals that choose to rely more heavily on private

reimbursement, for example, may decline to seek CCS certification. Both OSHPD and CCS are divisions of the state Department of Health.

The California Code of Regulations sets general parameters used by OSHPD for licensing Intensive Care Newborn Nurseries (ICNNs) in Title 22, Division 5, Chapter 1, article 6, Sections 70483 through 70489. The regulations describe general requirements, staffing and equipment to be provided by all ICNNs. For example, the requirements call for a pediatric cardiologist to be "available to the service" (Section 70485 (a)(3)), a practice certainly not common among lower level NICUs; this suggests that licensed NICUs may operate within a range of the services and staff listed, and have access to other services via transport to another facility.

The California Children's' Services Program (CCS), a program of the state Department of Health Services, issued standards for NICUs in 1988, which must be met to receive CCS 23

certification. NICUs seeking MediCal reimbursement must be CCS certified; hospitals that rely on private sector reimbursement may choose not to apply for certification. (49) CCS standards distinguish among levels, as follows: (12) Regional NICU (Tertiary): Provides a full range of medical and surgical care for severely ill neonates in a facility approved by CCS for tertiary care. Extracorporeal membrane oxygenation (ECMO) may be performed only in regional units also designated as ECMO centers.

Community NICU: Provides a full range of medical care services for severely ill neonates in a facility approved by CCS for "Standard" care. Only specifically approved hospitals at this level may perform surgery.

Intermediate NICU: Provides care for sick neonates who do not require intensive care, but require care at a higher level than provided in a general nursery. Infants requiring surgery or more than four hours of ventilatory support must be transferred out.

A 1997 report by the California Regional Perinatal Programs attempted to consolidate the terminology used by different sources, and identified Intermediate facilities as a Level II, but both Community and Regional NICUs as equivalent to Level III. (80). It also identified a designation of Primary that is equivalent to Level I. 24

Given that no one classification system is predominant, and for brevity of presentation, this study used the numerical classifications presented by Phibbs et al. (75): I, II, II+, III. Level I: In accordance with the TIOP report and the California Children's Services Program (CCS), this study regards Level I facilities as not being NICUs. Level II: Equivalent to TIOP Level II and CCS Intermediate level NICUs. Level II+: Equivalent to the CCS Community NICU; not defined by TIOP. Level III: The highest level of care, equivalent to TIOP Level III, and CCS Regional/Tertiary level.

1.5 NICUs: Volume, Level of Care and Outcomes

An 11-site demonstration program was initiated in 1975 by the Robert Wood Johnson Foundation to stimulate and evaluate perinatal regionalization. An evaluation of these programs in 1985 indicated that regionalization succeeded in shifting VLBW births to riskappropriate sites, and that this shift was associated with a decline in neonatal mortality. The percentage of VLBW births delivered at tertiary centers increased from about 30% in 1970 to almost 60% by 1978 in the selected demonstration regions, and 47% in the comparison regions. Regions with greater increases in the proportion of infants born in tertiary centers experienced greater declines in neonatal mortality rates, and changes in site of delivery were associated with regionalization. (64) 25

Additional research has shown that the level of the NICU significantly affects survival for high-risk neonates. A report of NICU care in New York City between 1976 and 1978 showed significantly lower neonatal mortality rates for low birth-weight infants (501-2250 g) at Level III NICUs, compared with both Level I and Level II units.(71) Paneth et al. (70) found that initial choice of hospital was key, as 42% of all very low birth weight deaths occurred in the first four hours, when transport was unlikely; within this time, mortality for Level II and Level III NICUs was similar, but it was much higher in Level I hospitals. Between 4 hours and 28 days, however, Level II mortality was much higher than in Level III; after 18 hours, Level II mortality was worse than Level I. (70)

Le Fevre et al. (51) found improved neonatal mortality at Level III hospitals in Missouri between 1980 and 1984 for both black and white infants born at 2250 grams or less, and for white infants weighing less than 1500 grams. There was no such improvement at lower level hospitals, including those most comparable to California's Level II+.

Some studies examined the volume of patients treated as well as the NICU level. Mayfield et al. (62) examined white singleton births in Washington state between 1980 and 1983, and found that the perinatal mortality rate for low birthweight babies under 2000 grams was lowest at Level III hospitals, with no benefit for these infants at either Level I or Level II facilities. The study also found that volume had an effect on perinatal mortality, but that 26

technology (represented by level) had a stronger effect. An important limitation, noted by this study, is the high degree of correlation between level and volume. All of the level II hospitals had over 1,001 deliveries a year, and all of the Level IIIs had over 2,000; and these were the highest two of five volume categories presented. Another limitation, which is not addressed, is that the volume categories report births at each level by the hundreds or thousands per year; within these categories, it is impossible to tell from the data whether or not some hospitals have a high daily census, or whether a particular volume level is more or less significant for mortality outcomes.

Horbar et al. (44) did not find a volume effect for neonatal mortality among VLBW births. However, this study examined a voluntary consortium of 62 hospitals throughout the US, and did not explore whether the effect of volume varied among hospitals at different levels.

A review of singleton births in California in 1990 (75) found that neonatal mortality rate for high-risk infants were lower in high-volume Level III NICUs with an average daily census of 15 or greater, compared with other NICUs. A more recent study of births at California NICUs in 1992 and 1993 shows that high volume may bring neonatal mortality rates at Level II+ NICUs down to Level III rates (14). Compared to hospitals without a NICU, hospitals with either a Level III NICU, or a Level II+ NICU with an average daily census of at least 15 patients a day, had significantly lower risk-adjusted neonatal mortality than other units. Risk-adjusted costs were not higher. In Level II NICUs, and lower volume Level II+ 27

NICUs, neonatal mortality outcomes were not different from hospitals without a NICU, and mortality was significantly higher than at the higher-level/higher volume units (see Appendix B).

Phibbs et al. (75) found that subsequent neonatal transfer to a Level III NICU did not decrease the disadvantage of birth at hospitals with no NICU, a Level II NICU, or a highvolume Level II+ NICU.

Cifuentes et al. (14) found that transfer decreased this

disadvantage only marginally.

Although some more limited studies have reached different conclusions, as cited above, (44, 47, 90) as comprehensive as the 1996 studies by Phibbs et al. and Cifuentes et al..

1.6 MCOs and Risk-Appropriate Care Higher quality is associated with higher volume for a number of conditions, including cardiac as well as neonatal care. In theory, managed care organizations (MCOs) could direct patients to providers that are both cost efficient and high quality, since health plans have access to some information about quality indicators, such as volume, and greater capability to analyze and use it than individual patients. The Kaiser system was documented to direct patients to specialty centers for cardiac care at one period in time in the late 1970s. (55) One study in southeast Florida in 1993 suggests however that this is not consistently the case. The market for cardiac care for Medicare beneficiaries was segmented between 28

commercial MCO patients, who went to low-volume hospitals disproportionately; and Medicare fee-for-service enrollees, who went to high-volume hospitals, including academic health centers. (23)

1.7 California perinatal regions and market areas

California is divided into 11 perinatal regions, each of which has at least one tertiary Level III NICU. The boundaries of the regions were first established by the state Department of Health Services in 1979, pursuant to state Senate Bill 775, which defined a perinatal region as "that service area determined on the basis of geographical, jurisdictional, and marketing factors to be an appropriate setting for the operation of a regional perinatal health system." (20) Two regions consist exclusively of Kaiser hospitals, one each in the north and the south. Los Angeles is further subdivided into seven subregions, which were created in the 1980s through a collaboration between the local health department and community providers, based on the factors stated in Senate Bill 775 (79); each is counted as a region for purposes of this study. This results in a total of 17 perinatal regions.

There is general agreement that geographic proximity is an appropriate measure for defining the market area for general acute hospital services, since patients prefer nearby hospitals. Geographic areas drawn to reflect the source of hospitals' admissions, e.g. by 29

zip code, are also considered superior to politically drawn districts such as counties. There is less agreement regarding specialty services, where the market area may be regional or national. (22) The 17 California perinatal regions combine a number of important features of a market area, including roughly equal population size, geographic proximity to a Level III hospital, and recognition of traditional referral patterns. City and county boundaries were crossed in drawing some regions, to reflect these considerations. Perinatal regions therefore provide a useful definition of the market areas for specialty neonatal services in California.

1.8 Role of Kaiser hospitals Kaiser is an independent system of HMOs, with at least one hospital in each of the 11 main perinatal regions. Salaried physicians and other health care professionals work exclusively for Kaiser, which also owns and operates clinics and hospitals. Kaiser members enroll and re-enroll annually, and once enrolled are covered only for services provided within the Kaiser system. On occasion Kaiser may contract with a non-Kaiser provider for specialty services, which could include NICU care. Thus Kaiser is in competition with other health plans, which offer access to a range of community hospitals, and also with other hospitals. Kaiser hospitals deliver about 12% of all births in the state. Kaiser ownership of a hospital is reported by OSHPD.

1.9 MediCal coverage and reimbursement rates, and DSH funding 30

The California Medicaid program, known as MediCal, has historically offered generous eligibility and benefits, but limited reimbursements to providers. (48, 105) MediCal covers 32 of 34 services considered optional by the federal program. In an attempt to lower the state's infant mortality rate of 7.9/1,000 births as of 1990, the state began in 1992 to waive asset requirements for pregnant women and children up to age one with incomes up to 185% of the federal poverty level. In 1994 the income limit was raised to 200% of the federal poverty level. In 1995 the program covered 6.8 million Californians at some time, at a cost of $17 billion. (105)

MediCal used a variety of methods to set rates for hospitals during the study period. In most areas it paid hospitals per diem rates negotiated with the state California Medical Assistance Commission (CMAC). In 1995 CMAC negotiated rates with about 260 of California's approximately 500 hospitals. Two counties, Santa Barbara and San Mateo, operated county organized health systems since the early 1980s, which accept full risk for providing MediCal services and function as a managed care plan for all MediCal enrollees in the county; three additional counties adopted this model in the 1990s. Beginning in 1991 the state began to offer additional options to enroll MediCal participants in managed care plans. The conversion to managed care was slower in MediCal than in the private sector, 31

and in 1997 only 35% of MediCal participants were enrolled in managed care. (105)

MediCal rates are generally lower than those in other states and than other plans in California. In 1995 the average MediCal expenditure per adult enrollee was $1,225, compared with the U.S. average for Medicaid programs of $1,728. (105) MediCal hospital rates are confidential, but a survey in 1999 found that MediCal reimbursement for the first visit of critical care was $107.26, compared with an average private PPO rate of about $217. (47)

In part as a result, California safety net hospitals relied heavily during the study period on federal Medicaid Disproportionate Share Hospital (DSH) reimbursements, which grants additional funds to both public and private hospitals that serve a high percent of Medicaid patients. MediCal DSH grew from an $11 million program in 1990 to a $2 billion program in 1992, an even faster rate of growth than the federal program. This trend slowed considerably after 1992. (105) This significant increase in the DSH program gave California hospitals a strong financial incentive to attract MediCal patients beginning in the early 1990s.

While MediCal reimbursed at rates below hospitals' costs throughout the study period, expansions in eligibility and DSH funding made MediCal patients more attractive to hospitals through the 1990s. 32

1.10 Growth of Level II NICUs in California

The number of NICUs has been increasing in California. According to Phibbs et al. (75), between 1980 and 1990, the number of hospitals providing Level II or Level II+ neonatal care more than doubled, from 53 to 116, while the number of Level III NICUs remained relatively constant at 25 (two level III units merged in 1990). Births during this period shifted from Level I hospitals to Level II and II+. Most new units were in urban areas and also low volume: 90% of level II births and half of Level II+ births occurred in units with an average daily census of less than 10. Over this period there was a moderate increase in the percentage of LBW births in Level III hospitals, from 24.4% in 1980 to 31.5% in 1990. (75) Since in the aggregate California's Level II and low-volume Level II+ units present no advantage in mortality over non-NICU hospitals, questions arise regarding why hospitals are adding new Level II and low-volume Level II+ NICUs.

Some have argued that improvements in technology offer expectant mothers the convenience of delivering at a nearby hospital with no threat to safety. (65) However, this argument was not supported by evidence regarding mortality outcomes.

Neonatology has become a more popular specialty, and some posit that there has been pressure from obstetricians and neonatologists to provide units in which they can practice, 33

particularly as cost pressures reduce other opportunities for specialists. The question remains, however, why the health care system is able to accommodate this influx of new providers. McCormick and Richardson (63) note that there may currently be a surplus of NICU beds and of neonatologists. The U.S. rate of about 3,000 active neonatologists, or 7.4 per 10,000 live births, is at least twice that of several other industrialized countries, and exceeds the upper bound calculated by the American Academy of Pediatrics in 1985 of 5 per 10,000 live births. (This factor is not explored in this study.) The reported number of about 3 NICU beds per 1,000 live births also exceeds the estimated need of 1/1,000. While the birth rate is down, the rate of very low weight births is increasing, and McCormick and Richardson have suggested that 1985 estimates may require revision.

A review and case study by the National Perinatal Information Center (NPIC) of challenges to regional arrangements (32) found that community-based physicians expressed frustration particularly with university-based tertiary care centers that excluded them from privileges and failed to re-route recovered infants despite overcrowding at Level III. Nonacademic tertiary centers, which stressed an open-door policy for obstetricians that allows them to follow high-risk patients throughout their pregnancies, seemed to avoid these problems.

Clinical uncertainty may play a role in the ongoing delivery of high-risk births at lower level hospitals, and the increased number of lower level units. (7) Birth abnormalities cannot always be predicted, and opinion may differ even after birth about which level of care is 34

required.

The departure from regionalized systems of care, sometimes referred to as deregionalization (4, 10, 33), may also be due to the shift in the broader institutional environment beginning around 1983, from one in which planning and inter-hospital cooperation were promoted, to a greater reliance on market mechanisms and price-based competition among providers intended to encourage cost control and efficiency.

Chapter 2 discusses the basic tension in the U.S. health care system between cost and access; the shift from an environment based predominantly on nonprice competition that accommodated some elements of planning, to one based on greater price competition to manage the tension between cost and access, and the implications of this shift for the regionalization of NICUs and other specialty services. Chapter 3 reviews the research design and methods of the study. Chapter 4 identifies data sources and the definition of the major variables. Chapter 5 summarizes the results of the review of trends in the major variables. Chapter 6 presents results of the statistical analysis. Chapter 7 presents a discussion of the findings, and strengths and limitations of the study.

35

CHAPTER 2. THE DYNAMICS OF HOSPITAL COMPETITION ACROSS ECONOMIC ENVIRONMENTS IN THE U.S.

2.1 1945-1982: Medical Arms Race: Cost and Access at Odds

The economic incentives of the U.S. health care system have long been at odds with efforts to control costs. After World War II, other industrialized countries balanced the public's interest in wide access to a broad range of health care services with an equal interest in cost control by aligning reimbursement incentives and regulatory mechanisms through political institutions. The U.S., however, left coverage largely to private insurance. Both private insurance, and later public coverage through Medicare and Medicaid, emphasized and reimbursed highly for expensive acute care and specialty services. Due to the configuration of financing and services, expanded access has been available only at the cost of financial inflation.

A well-defined market consists of a set of suppliers and demanders whose trading establishes the prices of products and services, and possibly also their quantity and quality. (21) Suppliers of products or services seek to maximize profits, and compete for demanders who will purchase their products or services. (21, 48)

Suppliers such as hospitals have market power to the extent that they can raise prices 36

without losing business. Demanders express market power by seeking alternative suppliers when prices increase, or deciding not to make a purchase. (48) Unlike most economic sectors, however, prior to 1982 health care providers (suppliers) also essentially controlled demand, as doctors ordered the tests and services for which they and other health care providers were subsequently reimbursed. Neither hospitals, physicians, nor insurance companies stood to gain from minimizing prices, costs or revenues.

Before the advent of managed care, patients generally selected a physician, and relied on the physician's recommendations when hospitalization was required. Hospitals thus competed directly for physicians, and less directly for patients. Hospitals competed for doctors both on the basis of quality and price. They signaled quality by offering high tech equipment and services that permitted doctors to practice at a complex level and also maximize their income. (107)

The resulting growth of technology and service capacity, beyond levels needed to maintain the health of the population, has frequently been termed the "medical arms race." The excess capacity of providers was seen as contributing to higher costs, higher prices, and excess utilization. (18). Competition among multiple suppliers for patients was thus inflationary. Hospitals in comparatively competitive markets, that is those with many nearby competitors, also were more likely to offer specialized clinical services than hospitals in less competitive markets. (66, 56) 37

Analysts found this consistent with the theory that "cost-based reimbursement made it fruitless for hospitals to engage in price competition. Instead they competed by being first to acquire the latest technology, attracting patients who associated the acquisition of technology with quality, and perhaps more importantly, physicians whose incomes could be augmented by the technology." (22, p. 187)

Financed by both employment-based private insurance and government programs, health care utilization and charges increased. Regulatory systems emerged to attempt to balance the effects of the financial system on access and equity.

In 1974, the National Health Planning and Resources Development Act was passed, to enable certificate of need (CON) programs to control the expansion of hospital capacity and duplicative technology. Hospitals generally succeeded in overcoming CON provisions through legal defenses and capture of regulatory bodies. One study found that CON did not reduce the total dollar volume of investment but shifted it from bed supplies to investment in new services and equipment. (88) CON was also viewed as a barrier to entry, benefiting established and well-endowed hospitals. Voluntary price controls during the late 1970s were similarly ineffective.

It was in this environment, however, that NICUs were developed and disseminated. While 38

cooperation among hospitals was not common during this era, it was possible. Costs were high and expertise rare. Doctors and hospitals embraced NICU regionalization both as a cost effective method to disseminate expensive technology, and as a way to impose barriers to entry. The staff required to operate NICUs included not just technicians to operate diagnostic and treatment equipment, but also nurses with advanced skills and physicians with advanced medical training who would take longer to produce. Efforts at regional cooperation were carried forward by medical leaders. By 1973 the American Medical Association had recommended limiting the number of NICUs and specialty practitioners, while promoting efforts to regionalize perinatal care.(91)

2.2 1982-1997: Price Competition

Throughout the 1980s, the financial reimbursement incentives changed to provide incentives for cost containment. Public policy at the federal level also changed to deemphasize health planning. Beginning in the early 1980s both the Medicare payment system and the state of California established policies intended to shift market power from suppliers to demanders, and from patients to payors. (85) In 1982, California authorized both private insurers and the state's Medicaid program, MediCal, to contract selectively with providers who could be chosen based on price. (66, 86) This meant providers could also be excluded for reasons including price, without violating antitrust regulations. Antitrust law, which has included hospitals since 1976, seeks to maintain a 39

sufficient number of competitors in a market to control prices. Collusion on price-setting is prohibited.

In 1983 Medicare instituted a fixed-rate prospective payment system. Medicare also moved to reimburse outpatient surgery at 100%, and to impose criteria for appropriateness of admission to hospitals, causing doctors and hospitals to shift care to outpatient settings, and contributing to declining hospital inpatient admissions.

By 1986 over a third of Californians belonged to prepaid health plans. In some cases, hospitals or other providers were subject to capitation arrangements, placing them at financial risk for expenses. Prepayment at fixed rates reversed earlier financial incentives to provide the maximum number of services at the highest cost, replaced by incentives to limit utilization and charges. Even so, many providers continued to receive fee-for-service payment, but at reduced levels.

During this period there was substantial growth of for-profit hospital systems, whose primary goal was specifically to maximize profits. The number of for-profit health plans and service providers such as hospitals and home health agencies increased, and ownership became increasingly consolidated through mergers and acquisitions.(15, 37) They competed for patients not only through attracting admitting doctors, but also by building sufficient market share and breadth of services that they could not be excluded by health plans. 40

Methods to build market share included opening ancillary services and satellite clinics designed to appeal directly to patients, marketing to patients likely to be profitable by offering services of interest to upscale clients, locating services in high-income neighborhoods, and turning away uninsured patients.

The combination of competition from prepaid health systems and declining reimbursements from Medicare and ultimately from other payors shifted hospitals' priorities and behaviors to reflect a price-competitive rather than a quality-competitive industry. Survival required greater market power, and hospitals consolidated into systems during the decade.

Hospitals face imperatives to satisfy purchaser and health plan price demands, unless they can position themselves to be indispensable to a provider network, for example as the sole provider of necessary services. When included in a health plan, they would be expected to offer a sufficient range of doctors and breadth of services, and a reputation for quality, to attract individual plan enrollees. (18, 21)

Health plans seek to attract business from purchasers, primarily employers and government programs, and secondarily individual enrollees, who may experience conflicting concerns regarding quality and price. Further they must be sufficiently selective to enforce market discipline among providers. Payors may desire that some excess capacity continue to exist to permit room to bargain. They thus may encourage the growth of NICUs. 41

Competition for paying patients in particular has become intensified. The percentage of uninsured people grew, but more importantly sources of cost-shifting to pay for uncompensated care shrunk during the 1980s, threatening revenues, profits, and survival. Beginning in 1989, maternity patients increasingly had a source of payment through Medicaid and MediCal expansions.

Managed care organizations have the potential to reduce unnecessary procedures driven by fee-for-service reimbursement. Historically prepaid group practice plans, the precursors to managed care, contributed to quality by emphasizing collaboration among providers, organized systems of care, and primary and preventive care services.

However, questions about access and quality of care are receiving renewed attention for several reasons, (101, 80) including concerns that capitation provides incentives to underserve, and the fact that insurance premiums are again beginning to rise. Some analysts claim that market-driven health care systems may pose risks of both underservice and overservice, depending on the population and the service, and that market incentives alone lack mechanisms for accountability that would necessarily result in high quality and affordable care.(8)

Price competition was expected to control costs and discourage excess capacity or 42

unnecessary specialized services and technology. While quality was not an explicit goal, some expected that reducing excess capacity would benefit the quality of care. (93)

In the case of NICUs, however, it would appear that price competition has not limited greater availability. Although market competition beginning in the 1980s encouraged a concentration of facilities, the competitive behavior required of those facilities was antithetical to the concentration of specialized services and procedures. Zwanziger and Melnick (107) found that hospitals respond to increased price sensitivity through "changes that either decrease costs with little reduction in demand, or increase demand while costs remain relatively constant." Lower level NICUs could well represent the second type of response. In addition, they found that even in a period of price competition, California hospitals in more competitive areas, measured by the Herfindahl index of concentration, tended to differentiate themselves from their peers by adding or closing services at the margin with relatively small patient volumes (as opposed to "universally available" services such as surgery).

This changing environment caused the March of Dimes to reconvene the Committee on Perinatal Care. It issued a 1993 report calling for renewed focus on the social factors associated with infant mortality and public health interventions, as well as on improvements in health education, prenatal care, system organization, access to inpatient and specialty services, documentation and evaluation, and adequate financing of prenatal care.(17) 43

The shift to price competition has several implications for specialized services such as NICUs. It might be expected to reduce such services for a number of reasons.

1. Since the presence of many competitors had the effect of inflating prices under costbased reimbursement, price competition would reduce the number of suppliers, and concentrate admissions in fewer hospitals. Financial incentives have resulted in downsizing of acute care hospitals, particularly among public, community-based, and faithbased hospitals, thus reducing hospital capacity in some geographic areas.(3)

2. Alternatively, price competition may not reduce the number of suppliers, but could lead to lower prices.

3. Price competition may reduce revenues available to hospitals, including revenues required to finance new purchases and expansions. (5)

4. To the extent that MediCal and private MCOs pay rates below cost, hospitals have little incentive to attract patients covered by these programs (22).

Several elements of price competition might induce hospitals to open or expand a NICU.

44

1. Reductions in hospital occupancy and revenues due to price competition may make hospital survival a key issue. Diversification of services has been found to be associated with hospital survival in the era following Medicare prospective payment. (53)

2. Patients in managed care plans often have to make prospective choices of providers. The most common form of managed care plan has been a form of preferred provider organization (PPO), which gives enrollees financial incentives to select providers within the PPO. People prefer a plan that includes their providers, and can select a panel of providers based on health conditions they believe they may experience in the future. (21) Thus hospitals with NICUs may be more likely to be selected by enrollees who anticipate a pregnancy, and may also be more attractive to managed care plans.

3. Patients generally prefer to choose a hospital close to home, but will travel further for specialized services (48). Proximity is the most important factor for prospective mothers, but they also have time in advance to make a selection, and are likely to consider quality in selecting a hospital for delivery. Mothers known to be at higher-risk are more likely to consider quality indicators. (76) NICUs may be perceived to add quality.

4. With market power shifted to health plans, hospitals had incentives to compete for health plan contracts, and inclusion in health plan networks, as well as directly for physicians and patients. (66) It is uncertain whether or not having a NICU is a 45

competitive factor, but it does increase the scope of services a hospital has available. For example, health plans may prefer admitting OB patients to the lowest cost hospital that can safely care for the widest range of cases, without incurring the costs associated with tertiary care centers. A Level II NICU provides some respiratory support for distressed newborns, a service that a non-NICU hospital does not provide. A hospital with a Level II NICU could promise some degree of support for high-risk pregnancies at a lower price than a tertiary care center.

5. Some reimbursement arrangements make it more remunerative for physicians and hospitals to retain distressed neonates rather than transferring them, which may be clinically more justifiable if a NICU is present. In California in 1990, 24% of pregnant women received high-risk prenatal care in 1990 (75). Routine referral of all OB cases with elevated risk would have represented a significant loss for most OB practices.

6. The rate of uninsurance distinguishes California from other states, which could motivate some competition for paying patients among already lean hospitals. In 1992 six million Californians, or 23% of the population under age 65, were uninsured, compared with a national average of 18%. (9) This probably makes competition for insured patients more intensive. In contrast, 97% of expectant mothers were insured by 1997 (see Table 5.2).

Obstetric patients are relatively healthy and thus use fewer resources. Women patients are 46

also viewed as the family decision makers about health care, and may consider that securing their loyalty during delivery can mean repeat business.1

On the other hand, high-risk MediCal mothers and severely distressed MediCal infants, may be less desirable. Phibbs, Mark, Luft et al. (76) found that high-risk women covered by Medicaid were less likely than those with private insurance to deliver in hospitals with NICUs. A 1991 report from Chicago pointed out that public hospitals, as providers of last resort, may be victims of "perinatal dumping." Fifty-four community hospitals that were not in the Cook County Hospital network nevertheless transferred pregnant women there; this total included 9 of the 11 other perinatal centers. The transferred women were more likely to be black or high risk, and to deliver pre-term births. 73% were receiving Medicaid; 22% were uninsured. (39)

Other reports suggest, however, that low-risk Medicaid obstetric patients became widely sought after by hospitals in the mid-1990s,(40) where maternity reimbursement by private plans had fallen to the Medicaid level or below, and DSH payments offered additional compensation.

7. Price competition undermines alternative arrangements for disseminating technology such as regionalization. Mergers among specialty hospitals likely to attract OB patients and shifting network alliances may also have disrupted cooperative arrangements among 47

hospitals in a regional system.(32, 83) MCOs may require a hospital or a hospital network to include a NICU as a condition of contracting.

8. The price of new technologies tend to fall over time (22, 89). As noted, multiple procedures and techniques are now available to improve the status of VLBW babies prior to delivery and to prolong their lives afterward, some of which are less expensive than those first available in the 1970s. (1)

In sum, the reduced emphasis on quality, the increased emphasis on costs and price by both providers and purchasers, the absence of a cooperative environment among hospitals with shifting alliances, and the absence of a countervailing presence establishing and enforcing industry-wide quality guidelines, enable if they do not fully explain the greater availability of Level II NICUs. Although these competitive effects have been widely discussed, there has not been an effort to examine the individual and combined effects of competitive factors across multiple regionalized networks.

2.3 Summary of the research evidence

The combination of technology and skill offered by NICUs has made an important contribution to lowering neonatal mortality, and extending the bounds of survival for very low birthweight infants. NICUs were initially distributed on a regional basis, due to their 48

high costs, limited availability of providers, and determined efforts at cooperation among providers. This distribution led to high volumes of admissions in NICUs, and there is evidence that volume has been a factor in the success of higher level NICUs at lowering neonatal mortality. (14, 64, 75)

Competitive pressures led to reports of deregionalization during the late 1980s, raising concerns about quality of care in lower level units (63, 83). Other evidence of decline in the system of cooperation was seen in patterns of transfer and referrals.

Recently low-risk Medicaid obstetric patients may have become widely sought after by hospitals, as maternity reimbursement by private plans has fallen to the MediCal level or below. High-risk MediCal mothers, however, and severely distressed MediCal infants, may be less desirable.

Healthy People 2000 calls for 90% of maternity admissions to be risk appropriate, and California standards seek 90% of VLBW births to occur in Level III hospitals. A number of factors to be examined by this study may contribute to or detract from attaining these goals. The 1985 RWJF study, which included San Diego County and two areas of Los Angeles as demonstration sites, found that 60% of high-risk infants were delivered at Level III hospitals in 1977-78 in demonstration sites, and 47% in non-demonstration sites. (64) By 1990 only 31.5% of low birth weight infants in California were delivered at a Level III. There was 49

substantial growth in California and elsewhere of lower level, lower volume NICUs during the 1980s (75).

Health policy shifted in the 1980s towards reliance on market forces to limit costs and excess capacity. As health plans and purchasers became the route to attracting patients during the 1980s, hospitals sought to differentiate themselves in ways that were responsive to new incentives and

requirements.

Health plans that limited access to providers

nevertheless had to provide the full range of service within their networks, in order to attract purchasers and consumers. Purchasers also wanted sufficient excess capacity to permit negotiations on the price of services. As hospitals consolidated into systems, those that could offer some level of high tech services, at a reasonable cost, were more likely to survive.

Obstetric patients are a likely target for specialized services, since they are

disproportionately insured, and are known to play a more active role in choosing a hospital than many other patients.

California state health agencies do not require that NICUs at various levels conform with state standards for staffing, equipment, or performance. Reliance on financial incentives may be inadequate to assure optimal neonatal services

This research examined the impact of these changes on the availability of NICUs in California, and on the risk-appropriate delivery of very low birth weight babies. 50

51

FIGURE 1. CONCEPTUAL FRAMEWORK REGULATORY ENVIRONMENT

MARKET COMPETITION ENVIRONMENT

COMPETITIVE Demand-Population Need VLBW Births per Region Total Births Per Region

% Uninsured (not modeled) Provider revenues

FACTORS: Supply and Demand Supply: Hospitals, NICUs Hospitals Maternity Hospitals No. III NICUs No. III NICU beds No. Level II/II+ NICUs No. Level II/II+ NICU beds

Balance of Supply and Demand Herfindahl Index: dispersion/concentration of discharges Occupancy: hospital, perinatal, NICU Proximity II/II+ to Level III Births as % of Hospital Discharges

Availability: Balance of Supply and Population Need Ratio births/ II/II+ NICU beds

REGIONALIZATION

OUTCOMES Risk-appropriate births % VLBW at Level III % VLBW at High-Volume Level II+ or Level III

Demand: Terms of Reimbursement HMO MediCal Other Uninsured

CHAPTER 3. RESEARCH DESIGN AND METHODS This chapter reviews the conceptual framework, hypotheses, data sources, and major variables for each hypothesis. In the study hypotheses, the relationship of competitive factors to the growth of Level II NICUs was examined and these trends were studied in relation to risk-appropriate perinatal care.

3.1 Conceptual Framework The literature establishes that the environment for financing and organizing health services shifted over time toward greater reliance on market forces. It is expected that this shift has had consequences for the supply and availability of NICUs, which in turn affected the outcomes of care. (Figure 1) While the supply of hospitals and NICUs was expected to have been responsive to population need, historically there has been an excess in supply. This may be attributable to competitive forces that seek to balance supply with demand, as reflected in measures of the number and concentration of nearby competitors, and a hospital's occupancy rate. Both regulatory forces, in the form of regionalization, and market interventions in the form of reimbursement policies and selective contracting, have been designed to effect a more appropriate distribution of NICUs. Supply, competitive forces, and population need contribute to population-based measures of availability, in this case the ratio of births to Level II and II+ NICU beds. Regionalization was expected to create a cooperative system in which providers agree to a distribution of supply, and also agree on a system for directing prospective high-risk births to appropriate sites. Payors may have different effects on supply, competition, availability and regionalization. 53

The presence of any type of insurance represents the ability to pay, which can lead to increases in demand and thus supply and availability. In a market environment, pricesensitive payors such as managed care organizations and MediCal attempt to gain greater market power by engaging in selective contracting based on price. Also, these payors may establish networks with a limited panel of hospital and physician suppliers. These panels may differ from those established by regional NICU systems, altering the predominance and effectiveness of regional systems. (Fatheree) Thus price-sensitive payors may limit supply, as regionalization attempts to do, in part by reducing prices and, therefore, revenues available for expansion.

Alternatively price-sensitive payors may stimulate increases in supply, for several reasons. First, hospitals could open or expand NICUs to attract MCO contracts. Second, price negotiations may intensify competition, leading hospitals to add specialized services such as NICUs.

Third, lower prices and reduced revenues could cause hospitals that choose to

specialize to do so by adding services that attract insured (e.g. maternity) as opposed to uninsured patients.

HMOs and MediCal are likely to have different effects on supply, competition and outcomes. MediCal historically paid lower per diem rates than private managed care plans. Regional dominance of MediCal as a payor could be expected to constrain supply of Level II and II+ NICUs. 54

In summary, the model examined the effects of changes in need, insurance coverage and managed care, and the balance of supply and demand on the availability of NICUs. In turn, the distribution and mix of NICUs were examined as factors in determining outcomes, as measured by the percentage of risk-appropriate births.

3.2 Research Aims and Hypotheses

The first aim was to examine trends in the numbers and levels of NICUs, numbers of VLBW births, and characteristics of hospital markets over the study period, 1987-1997. Trends were examined in the following areas:

1. Supply: number of hospitals

2. Supply of NICUs: number of NICUs, number of NICU beds by level, and number of lower-level NICUs in proximity to Level III units.

3. Demand a. Population need: number of births and very low birth weight births b. Demand-side market power through penetration by price-sensitive payors: HMO and MediCal 55

4. Balance of supply and population need: ratio of births to Level II/II+ NICU beds

5. Balance of supply and demand/competition: hospital and perinatal occupancy rates, and the Herfindahl Index, which measures the concentration or dispersion of hospital and perinatal discharges.

5. Outcomes: The percentage of risk-appropriate deliveries, defined as the percentage of VLBW babies delivered at a Level III and/or high-volume Level II+ hospital.

The study reviews and presents trends in market share of normal deliveries among hospitals with and without NICUs at various levels.

The second aim of the study was concerned with hospital and market characteristics associated with competition and the availability of lower level NICUs. Regionalization led to a rational distribution of NICUs early in their development, albeit based to some extent on the historical locations of academic health centers. But the growing availability of NICUs in the 1980s was more typical of the "medical arms race" associated with hospital competition. The central question is whether market forces, by imposing competition based on price, have constrained hospital competition and its effects, particularly the expanded availability of NICUs.

Alternatively, price competition could contribute to greater 56

availability, or could have no effect. The study hypothesized that increased demand-side market power, exemplified by HMO and MediCal payors, was associated with more intense competition and greater availability of NICUs, and further that NICU availability was not associated with clinical need.

The specific study hypotheses were:

Hypothesis IIa. Greater availability of lower level NICUs was associated with lower rates of perinatal and hospital occupancy, a higher percentage of deliveries as a percentage of all hospital discharges, and a more competitive distribution of perinatal discharges (Herfindahl index). Price-sensitive payors increased or did not constrain availability. Clinical need was not a significant factor.

Managed care is believed to have reduced hospital revenues by reducing the number of hospital admissions and length of stay. At the same time the percentage of uninsured in California has increased. While hospitals have always competed for patients, they are expected in this environment to compete most vigorously for insured and relatively healthy patients. By 1997, when 25% of Californians were uninsured, about 97% of mothers had insurance. In addition, as managed care ratcheted down reimbursement rates, MediCal reimbursements became increasingly attractive. Maternity cases were a substantial percentage of hospital discharges, ranging from 10 to 37% by region. It has been suggested 57

that hospitals offering NICU care intend to attract normal deliveries.

There is some evidence that expectant mothers play a more active role in choosing their hospital than patients with other conditions. The existence of a NICU may signal quality to these mothers.

Earlier research documents that the majority of lower level NICUs are located in close proximity to Level III NICUs. An association between proximity of competitors and growth in the number of competitors would be consistent with the theory that opening or expanding NICUs is intended in part to compete for patients.

HMOs and MediCal as price sensitive payors are expected to increase or fail to constrain NICU availability, for several reasons. Per Zwanziger and Melnick (107), hospitals are expected to attempt to differentiate themselves from their peers by offering marginal, low volume services such as NICUs. In addition, price competition may undermine regionalization. Regionalization was shown to constrain availability of NICUs and also to produce risk-appropriate admissions that reduced neonatal mortality. This arrangement likely depended on two basic factors, the high cost of NICU technology, and the willingness and ability of hospitals to cooperate. Advances in technology have reduced the cost of sustaining the life of increasingly smaller infants. Price competition has undermined some degree of the cooperation required for successful regionalization. Under some managed care 58

reimbursement policies obstetricians or hospitals may lose reimbursement as a result of transferring patients, providing a disincentive for transferring high-risk newborns from lower level hospitals to higher level ones.

Hypothesis IIb: Hospitals with lower level NICUs were associated with a higher degree of HMO penetration at the hospital and regional levels.

The basic tools of market competition may provide conflicting incentives.

Selective

contracting gives hospitals incentives to attract MCO contracts. MCO networks seek to include a full range of services among their providers, in part so that they can control costs, and in part in order to attract patients. NICUs may factor into hospital attempts to compete for contracts with managed care organizations. Hospitals with lower level NICUs offer the promise of caring for high-risk newborns at a lower charge than higher level NICUs, and also of avoiding the expense of a transfer.

A different way to explore the effect of price competition on availability is to test each NICU level for its relative association with competitive and payor variables. It was expected that lower level hospitals were associated with a higher degree of HMO penetration at the hospital level, and that higher level hospitals were not. This is consistent with the theory that hospitals opened lower level NICUs in part to seek or retain HMO contracts at their particular facility, as opposed to region-wide trends. Higher level hospitals were expected 59

to be competing for patients throughout the region, and will to be less influenced by their own particular rate of HMO coverage.

The third aim of the study was to determine whether there was an association between quality indicators and NICU availability, hospital competition, and price-sensitive payors. Quality was measured by the rate of risk-appropriate admission of VLBW infants.

The third aim was directed at understanding how VLBW admissions differ among more and less competitive regions and over time. It has already been established that survival outcomes for high risk infants are better at high volume tertiary hospitals. Standards of quality call for 90% of VLBW births to occur at risk-appropriate hospitals. Recent evidence suggests that high volume II+ NICUs can achieve comparable mortality reductions.

Hypothesis III. Declining rates of risk-appropriate admissions were associated with greater availability of lower level/lower volume NICUs, lower level/lower volume NICU beds, and competitive factors. Growth in price-sensitive payors did not constrain the decline.

Considerable research has addressed the relative roles of market competition and regulation with regard to the price of services, but there is little exploration regarding the effect of market competition on quality. It was hypothesized that lower level NICUs succeeded in attracting a certain percentage of high-risk mothers, and that the growing availability of 60

these units contributed to the decline in VLBW births occurring at higher level units.

The next chapter discusses the sources of data used in the study, and defines the major dependent and independent variables.

Following chapters will describe the research

methodology to test the hypotheses. Results and conclusions are presented in the final chapter.

61

Chapter 4. Data Sources and Definition of Variables.

This study examined changes from 1986 to 1997 across the 17 Maternal and Child Health Perinatal regions in California. It used Hospital Discharge and Birth/Death Cohort linked data files compiled by the California Office of Statewide Health Planning and Development (OSHPD). A list of NICUs and their levels was created by comparing lists from OSHPD, which licenses units; California Children’s Services (CCS), which certifies them; and primary research with key informants regarding the level at which the hospital was actually operating. 1986 was the first year for which both OSHPD and CCS classifications were available, and 1997, the most recent year that OSHPD data were available at the time of the analysis.

4.1 Two analysis files

Two analysis files were created to address the aims and hypotheses. First, a region-level data set was created which includes regional level characteristics of hospitals, the birth population, competition and markets for each study year. For each variable, for example hospital occupancy rate, this data set provides information about an entire perinatal region. The second data file, a hospital-level data set, includes the same set of variables for each hospital individually. This data set provides information on the hospital occupancy rate, or other variables, for a particular hospital. Each of these files was further divided into two 62

data sets, one in which "year" was a separate variable, so that each variable could either measure all years, or through programming could represent specific years; and a second in which each variable is also identified for each of the twelve years. The variable name for hospital occupancy rate would be HOCCRT in the former case; and HOCCRT86 for the latter, for the year 1986. A more complete description of each dataset, and the source of information for each variable, is found in Appendix G.

4.2 Major variables

The major variables can be divided into categories of supply, demand, balance of supply and demand, price-sensitive payors, and risk-appropriate admissions. Before defining these variables, the definition of perinatal region is provided.

1. Perinatal regions as units of analysis

The perinatal regions are established by the California Department of Health Services. Each region consists of a number of hospitals, and is anchored by one or more hospitals with a Level III NICU. While the regions are geographic designations, in some cases lines were drawn to recognize prevailing referral patterns. There were originally six regions, two in the north and four in the south. (77) The 1984 Report to the Legislature by the Department of Health Services lists the current 11 regions. These include two regions 63

consisting of all the Kaiser hospitals in northern California (region 10) and another consisting of all the Kaiser hospitals in southern California (region 11). Unlike the other perinatal regions, which are geographically defined, the Kaiser regions consist of all hospitals owned by the Kaiser system. Kaiser hospitals are physically located in other perinatal regions, but they are not counted as part of those regions. Region 6, which encompasses Los Angeles County, was further subdivided into 7 sub-regions. One of the L.A. subregions also includes Ventura and Santa Barbara Counties.

Each region has a Regional Perinatal Program, pursuant to 1979 California laws (SB 775 and SB 776, encoded in the California Health and Safety Code, Division 1, Part 1, Chapter 2, Article 2.6, Section 289 et seq.). The laws mandate that the State and the Programs facilitate and support the development of regional perinatal health systems, including coordinating services and education among the hospitals. As of 1996 there were four Perinatal Programs in Region 6 (Los Angeles). The Programs are financed primarily by federal Title V Maternal and Child Health block grants; Kaiser provides funding for the Programs in its two regions.

The overall and birth populations of the regions as of 1982 seem to reflect an attempt to distribute population among the regions as evenly as possible, balanced by the demands of geography. The regions had roughly equivalent populations in 1982, as noted in the 1984 report by the Department of Health Services, (20) ranging from 1.7 million to 2.5 million. 64

The birth populations varied more widely, from about 19,000 to 38,000 live births a year. The exception was the Los Angeles region, with 8.5 million population and about 141,000 live births in 1982; however, as noted, this region was divided into seven subregions, making the population of each subregion more comparable to the other regions. The two Kaiser regions had about 24,000 live births each in 1982; the study does not report separately the overall population of Kaiser enrollees.

A complete list of perinatal regions, and associated geographic areas and Level III hospitals, appears in Appendix D. The list of regions used for this study was compiled in 1994 by the Perinatal Regionalization Program of the State Department of Health Services, and reported in the Technical Report and User Manual for linkage of birth/death and maternal hospital discharge records (42). In identifying the NICU level of each hospital, as described below, questions about the assignment of hospitals to the L.A. subregions were resolved by personal communication with the perinatal regional coordinator for that area. The Manual lists an eighth sub-region for Los Angeles, set aside for hospitals that are not identified as located in any other regions. Since no hospitals are actually assigned to this region, it was not included in the study.

This study used 17 perinatal regions for purposes of descriptive and statistical analysis. Ten regions, excluding L.A., plus the 7 L.A. subregions, comprise a total of 17. Counting these L.A. subregions created some regions with relatively smaller numbers of hospitals than 65

other regions. However, the L.A. subregions more closely reflect a geographic market or catchment area for perinatal services than some of the larger multi-county regions.

2. Hospital supply characteristics

Several measures of supply were taken from OSHPD reports.

a. Number of general acute hospitals. This category excludes hospitals defined as acute psychiatric, chemical dependency and recovery hospital, and psychiatric health facility.

b. Maternity hospital. Using OSHPD data, a list of deliveries at each hospital was generated for each year under review. It was assumed that a hospital reporting 10 or more deliveries a year for any three years of the study would be considered to be offering a maternity service.

c. Number of NICUs by level. Hospitals with NICUs were designated by level to indicate the complexity and intensity of services offered. However, a central problem for this study as well as for prospective patients is that there is no standard definition of NICU levels in California, and little regular effort to apply the standards that exist. As described earlier, there are two sources that designate NICUs in the state. As reported by the Regional Perinatal Programs of California in 1997, (80) Intensive Care Newborn Nurseries (ICNNs) 66

are licensed by the Department of Health Services (DHS) Licensing and Certification Division under California Code of Regulations (CCR), Title 22, Division 5, Chapter 1, article 6, Section 70483-70489. Facilities may be licensed as either a general Perinatal Unit, which provides routine low-risk services for normal infants, equivalent to AAP/ACOG Level I; or an ICNN, providing either moderate-risk services or comprehensive, high-risk services, equivalent to Level II or Level III units. ICNN status is reported in the OSHPD Annual Report of Hospitals files.

California Children's Services (CCS), also an entity of the DHS, designates units as Intermediate, Community, and Regional, and does not designate primary levels of neonatal care. Intermediate NICUs provide moderate-risk services, equivalent to the AAP/ACOG designation for Level II. Community and Regional NICUs provide comprehensive, highrisk services, equivalent to Level III. CCS also records the facilities it has certified by level, including the year when an application was made, and the year when it was inspected and certified. Phibbs et al. (75) proposed referring to Community hospitals as II+, to indicate their somewhat lower level of services compared with Level III.

This study adopts the classifications used by Phibbs et al. (75) Primary/Level I hospitals are standard maternity services that care only for healthy newborns. Intermediate/Level II units care for moderately sick infants, but do not provide mechanical ventilation to assist with breathing for more than four hours. Community/Level II+ units provide longer term 67

ventilation support, but not all of the staffing and services provided by tertiary units. Tertiary/Level III units offer the full range of specialized neonatal care, including subspecialty consultants and surgery.

A process was undertaken to develop a complete and consistent list of hospital NICU designations. This involved comparing lists from OSHPD and CCS with surveys conducted by Phibbs in 1980 and 1990. Since CCS designations are more detailed, they were used where both OSHPD and CCS noted a NICU. Where only an OSHPD designation was available, the investigator consulted with key informants. The results were confirmed with a practicing neonatologist in northern California, several coordinators for the Perinatal Regional Programs, and Dr. Phibbs. A complete list of hospitals with NICUs, and their level designation, appears in Appendix E.

d. The number of NICU beds are reported annually by OSHPD, and were assigned the level designation associated with their hospital.

e. High-volume and low-volume Level II+ NICUs, and small level II/II+ NICUs.

In

May 2000 Cifuentes et al. released pre-publication data showing that high-volume level II+ NICUs, with an average daily census of 15 or greater, had a neonatal mortality rate similar to that of Level IIIs. Based on these data variables were created that broke down the NICU levels described above into high and low-volume level II+ NICUs. The trend analysis 68

presents normal deliveries and delivery of very low birthweight babies at these levels.

It is important to note however that Level II+ NICUs are not identified or formally designated according to volume by a state agency or any other entity. In total, 35 Level II+ hospitals had an average daily NICU census of 15 or greater in at least one of the years studied, while fewer met this criterion in any given year. For example, only 17 did so in 1997. This was not consistent from year to year, and hospitals fell on and off this list. For analytic purposes these hospitals are counted correctly by year and region, and are reflected accurately. To the extent that high-risk deliveries are finding their way to these hospitals, regardless of designation, this is worth noting. For most policy purposes, however, highvolume Level II+ hospitals are not well-defined entities.

For the above reasons, the study also reports on "small" lower level NICUs, those with fewer than 15 beds. Bed capacity is a more stable measure for evaluating availability than actual volume. Hospitals with fewer than 15 beds are generally by definition also low volume. (Though a few NICUs ran at over 100% of capacity for limited times, this was not usually the case in smaller NICUs for more than a year.)

For some analytical purposes, NICUs are aggregated to show the total of lower level units (all Level IIs plus low-volume Level II+) and higher level units (high-volume II+ plus all Level IIIs). 69

f. Number of lower level NICUs within 10 miles of a Level III NICU. This variable was adopted from Dr. Phibbs' linked files, and reflects the number of II or II+ NICUs within 10 miles of a Level III. Earlier research noted that the majority of lower level NICUs are in proximity to Level IIIs, particularly in urban centers. This provides a measure of potential competition among hospitals.

3. Availability: Balance of supply and population need Births and VLBW births per lower level NICU bed. Availability is an aspect of the fit between the patient's needs and the health care system (72). This variable measures NICU availability on a population basis. It is calculated by dividing total births, or VLBW births, per region by the total number of beds at level II and II+ NICUs. A higher ratio indicates less availability; a lower ratio indicates fewer births per bed, and thus greater availability.

4. Demand variables

a. Population Need: Number of births. This is the number of live births reported annually by hospitals to OSHPD. It excludes fetal deaths. (See Appendix G for the reporting form, with detailed descriptions.) According to Herrchen and Gould (42), fetal deaths accounted for .6% of deliveries in California in 1992.

70

The number of LBW and VLBW births are also reported by OSHPD, and defined as newborns weighing under 2500 grams and under 1500 grams, respectively.

b. Terms of Reimbursement: Price-Sensitive Payors

Price-sensitive payors are represented by HMOs and MediCal. This information is reported in the OSHPD Annual Report of Hospitals. HMOs include all categories of managed care plans, and do not distinguish among traditional HMOs and PPOs, for example. Births at Kaiser hospitals are recorded as HMO, and thus can only be distinguished from other HMO births if they are segregated by perinatal region. (Kaiser hospitals are listed as located in two separate perinatal regions, region 10 and region 11.) Kaiser is included as an HMO payor. MediCal is classified as MediCal and not HMO, although some MediCal beneficiaries belonged to managed care plans at times during the study period. MediCal patients delivering at Kaiser hospitals are listed as MediCal patients. Both HMOs and MediCal negotiate for services and contract selectively with hospitals based on price. These two payors thus well represent the presence of price competition in the regions.

5. Competitiveness/balance of supply and demand

Terms of competition

71

Competitiveness may be measured by at least two standard criteria: number of competitors, and concentration/dispersion of competitors, commonly measured by the Herfindahl Index. There is some evidence that number of competitors is a better measure for non-price competitive environments. (22) However, the Herfindahl Index is commonly used as an indicator of competition in price-competitive environments, (107) in part because market share has become a key concern.

a. Herfindahl Index. The Herfindahl Index measures the extent to which market share is divided among competitors evenly or not. If market share is divided equally among competitors, the index is zero; if all services of a specific category are concentrated in one hospital the index equals one (a monopoly). Market share is defined as the number of discharges for a hospital (from the OSHPD Annual Report of Hospitals) divided by the number of discharges for all hospitals in the region (derived from the OSHPD Annual Report of Hospitals). The market share is treated as a value between zero and 1. Each firm's market share is squared, and the squared results are added up. For example, if there is just one firm it has the entire market share. One squared is one, which is the Herfindahl index for that area. If there are two firms and one has a 90% share and the other has a 10% share, 0.9 squared is 0.81, 0.1 squared is 0.01. Summing those up the Herfindahl index is 0.82. A higher number indicates a greater degree of concentration, and less dispersion.

72

b. Perinatal and Hospital Occupancy Rates.

The perinatal and hospital occupancy rates are reported by OSHPD. Perinatal services are defined as maternity and newborn services for the provision of care during pregnancy, labor, delivery, postpartum, and neonatal periods with appropriate staff, space, equipment, and supplies, as defined in Title 22, Division 5, Sections 70547-70553, California Code of Regulations. OSHPD records maternity discharges as perinatal, and counts newborns separately. The perinatal occupancy rate is calculated as the number of perinatal patient days divided by the number of perinatal licensed bed days each year.

c. Proximity of Level II/II+ hospitals to Level III. Data supplied by Phibbs calculates distance from Level II/II+ NICUs in increments of 10, 15 and 25 miles. The number of competitors in proximity to a Level III is a measure of competition as well as of supply. A variable was created for each level of distance. The variables are strongly correlated. The closest measure, II/II+ NICUs within 10 miles of a Level III, was chosen as the most conservative measure of geographically-based competition.

6. Outcomes: Risk-Appropriate Deliveries Healthy People 2000 goals call for 90% of deliveries to occur at a risk-appropriate site. Virtually every VLBW infant would be expected to benefit from NICU care. The

73

percentage of VLBW births at each level of care was calculated by dividing the number of VLBW births at each level of care by the total number of VLBW births. Results are shown statewide for all years, and for 1986 and 1997; and as the average of regional averages. The risk-appropriate level is presented in two ways:

Percentage of VLBW births at a Level III hospital Percentage of VLBW births at a Level III or high-volume Level II+ hospital

Transfers into a NICU were not captured in this study.

4.3 Approach to the Analysis

The study was guided by two major hypotheses. First, it was expected that that there would be an association between greater availability of lower-level NICUs and both competitive forces and the growth of price-sensitive purchasers. Growth in lower level NICUs was not expected to be related to indicators of clinical need. This hypothesis assumed that hospitals continued to compete for normal maternity admissions in a price competitive environment by offering specialized services, and that the presence of a NICU was intended to attract managed care networks, physicians and patients. Second, it was also hypothesized that the risk-appropriate delivery of very low birth weight (VLBW) babies (i.e., VLBWs born at Level III or high-volume Level II+ hospitals) would be negatively 74

associated with greater availability of NICUs and with greater competition.

The trend analysis assessed the extent to which the major independent and dependent variables changed over time and varied across regions.

Hypothesis IIa: Greater availability of lower level NICUs and NICU beds was associated with lower rates of perinatal and hospital occupancy, higher percentage of deliveries as a percentage of all hospital discharges, and more competitive distribution of perinatal discharges. Price-sensitive payors decreased or did not constrain availability. Clinical need is not a factor.

Major independent variables for this hypothesis were selected based on the conceptual framework and previous research.

The availability of Level II and II+ NICUs and NICU beds was expected to be related to population demand/need, reimbursement incentives, Level II/II+ hospitals and beds and their proximity to Level III units, and the balance of supply and demand, including occupancy levels and distribution of births across hospitals (Herfindahl index). It was expected that the balance of supply and demand and HMO coverage was associated with increased availability, but that MediCal coverage was not. It was expected that the number of births were significant, reflecting hospitals interest in marketing to patients with normal 75

deliveries, but that the number of VLBW births, as a measure of population need for NICU services, were not significant.

The dependent variable is measured in two ways: Number of hospitals with a Level II or II+ NICU, and the ratio of annual births in a region to the number of Level II and II+ NICU beds. The former indicates increases in NICU availability hospitals and NICU capacity, while the latter is the more standard measure of availability relative to aggregate demand, i.e. the number of births in a region per NICU beds (Level II and II+) each year. A higher ratio would indicate more births per bed, or a more limited availability. A lower ratio would reflect fewer births per bed, or greater availability. This second measure is particularly useful in examining the extent to which the supply of NICU beds is associated with changes in need, which is also expressed as a population-based measure of births and VLBW births. The findings from these two models are expected to clarify the importance of competitive and need variables in the increase in hospitals with NICUs and in NICU beds.

Collinearity is likely to occur with some of the independent measures. Total births per region and total VLBW births are likely to covary, as might hospital occupancy and perinatal occupancy. The relationship of independent variables was examined and their contribution to explaining variation in the dependent measures assessed. The approach was to include the more predictive of collinear variables that were measuring the same dimension. 76

For each dependent variable, models were run both including and excluding the two Kaiser regions. The payor source for most deliveries at Kaiser hospitals is coded as HMO. To get a good picture of the influence of managed care organizations other than Kaiser, it was considered necessary to remove the two Kaiser regions from some models for sensitivity analysis. Region 10, northern California Kaiser, and region 11, southern California Kaiser, both consist entirely of Kaiser hospitals.

Hypothesis IIb: Greater availability of lower level NICUs were associated with higher per-hospital penetration of HMOs. Greater availability of higher level NICUs were associated with higher regional penetration of HMOs.

This hypothesis assumes that hospitals with smaller and lower level NICUs were more responsive to local as opposed to regional financial conditions, and also that offering a NICU was in part a method for attracting or maintaining managed care contracts. As a proxy for examining MCO contracts, this hypothesis tested whether hospitals with lower level NICUs were more likely to have relatively high rates of HMO-sponsored deliveries. To test this hypothesis, hospitals were divided by level and census, using the hospital-level data base. The hospital and region-level databases were merged, the hospitals at each NICU level were modeled with independent variables representing payors at the hospital and 77

regional levels. For this model the dependent variables are:

Level II NICU Low volume Level II+ High-volume Level II+ Level III

Independent variables are: Percentage of mothers with an HMO payor at the regional level Percentage of mothers with an HMO payor at the hospital level Percentage of mothers with a MediCal payor at the regional level Percentage of mothers with a MediCal payor at the hospital level

Hypothesis #III: Declining rates of risk-appropriate admissions were associated with greater availability of lower level/lower volume NICUs, lower level/lower volume NICU beds, and competitive factors. Payors did not constrain the decline.

Risk-appropriate deliveries were expected to be related to the availability of lower level and lower volume hospitals, the balance of supply and demand, and reimbursement incentives. As discussed above, Healthy People 2000 standards call for 90% of births at riskappropriate sites. Only one region met this standard in any year. This hypothesis therefore 78

tested how the variables affected relative degrees of risk-appropriate deliveries among the regions.

The dependent variable was measured in two ways. First, the percentage of VLBW births at a Level III is a useful measure because Level III NICUs are well-defined and accepted as appropriate for high-level births.

When using the percentage of VLBW births at a Level III, it was necessary to exclude regions 3 and 5, as no births occurred in the children's hospitals which are the only Level III hospitals in those regions.

Second, in light of evidence that risk-adjusted neonatal mortality outcomes are equally good in high-volume Level II+ hospitals, it was also useful to examine the distribution and determinants of births at this level as well. Therefore the percentage of VLBW births at either a Level III or high-volume II+ NICU was also used as a dependent variable. Highvolume Level II+ NICUs varied in their census during the period studies, and the implications of this for modeling this variable are discussed in the section on analysis.

For both dependent measures, models were run both including and excluding the Kaiser regions. These tests examined whether Kaiser regions skewed the variables measuring payor source and also NICU availability, since a far higher percentage of births in Kaiser 79

regions are reported as covered by an HMO than in other regions, and all but one Kaiser hospital had a NICU by 1997.

Ten independent and dependent variables were examined for collinearity, with a standard for a correlation coefficient of .60 or over to determine a strong correlation. (16) By this standard the two dependent measures are likely to be collinear, as are the two payor variables, the percentage of deliveries covered by an HMO or by MediCal.

Because volume is a consideration affecting risk-appropriateness, independent variables were created that identify hospitals by NICU level, and in the case of Level II+ units also by volume.

Level II NICUs Low Volume NICUs, with average daily census of less than 15 Total Low: Combination of Level II and Low Volume Level II+ High Volume Level II+ NICUs Level III Total High: Combination of High Volume Level II+ and Level III Other hospitals, with no NICU

4.4 Data Analysis 80

Statistical Methods

To accomplish Aim #1, descriptive data documented changes over time in the key variables: availability of NICUs and NICU beds at various levels; elements of hospital competition, including supply and demand variables; price-sensitive payors; and delivery of normal births, and risk-appropriate delivery of VLBW births, among the hospitals at each level.

Trends are reported both statewide and regionally. Statewide trends were established by simply adding up units, in the case of hospitals, or dividing the relevant statewide totals, in the case of the ratio of births to lower level NICU beds and the percentage of VLBW births at each NICU level.

Regional trends were needed to interpret the results of the statistical analysis, and were determined using the univariate procedure in SAS. The data files were based on regions, and the variables in the statistical models represent regional values. If the percentage of risk appropriate births, for example, is determined by summing up and then dividing the 17 regional averages, this number is likely to vary from one derived for the absolute statewide total, as described above. In order to quantify the effect of the independent variables, it was necessary to know the regional value for each variable at the beginning and end of the study 81

period.

To accomplish Aims #2 and #3, several hypotheses were tested. The testing of the hypotheses involved the estimation of time series models, pooling data over 12 years, from 1986 to 1997, and across 17 regions, yielding 12 times 17 observations, or n=204.

The number of independent variables was kept to a minimum due to limited degrees of freedom. All variables were examined to detect gaps, inconsistencies, and missing values. Preliminary analysis examined distributional characteristics of dependent and independent variables, and the bivariate relationships between independent and dependent variables. Correlations were used to guide the selection of variables and avoid collinearity. For Hypothesis IIa, 18 major independent and dependent variables were tested for correlations. Those with a correlation coefficient of .40 or over were considered strongly correlated. For Hypothesis III, with 10 variables, a coefficient over .60 was considered a strong correlation. (16) Sensitivity analysis accounted for the presence of Kaiser hospitals, one small region, and the two regions in which the sole tertiary level hospital is a children's hospital that does not perform deliveries.

The models estimated competitive effects using mixed effects models, which adjust for repeated measures, and for changes in both random and fixed effects over time (52). Mixed model regression varies from the general linear model, which structures the covariates 82

differently, and is better suited to testing repeated measures on an unchanging subject. (52) Mixed model regression is thus useful for testing the effects on the outcome variables of ongoing changes in the covariates between and within the regions over time. Specifically, the random statement makes it possible to model variation between experimental units, such as perinatal regions. A repeated statement modelled the covariance structure within experimental units to permit modeling variation. Random statements in the model allowed each region to have its own intercept and trajectory over time.

Where perinatal region and time are covariates and the response variable Yik is a random variable, a statistical model is: Yik= u + ai+ B1*X1ik + B2*X2ik+….+B5*X5ik + eijk Where: Yik = the dependent variable in region i at year k i = the perinatal region k = year u = the intercept ai = is the random intercept for area i B1 = the coefficient for X1; B2 is the coefficient for X2, etc. X1ik to X5ik = independent variables for region i at year k eik is random error associated with the independent variable in perinatal region i at time k The study tested whether covariates representing competition or clinical need explained the 83

increase in lower level NICUs within regions over the 12 year period.

To explore whether higher HMO maternity coverage at the individual hospital or regional level was associated with lower or higher level NICUs, models tested each level of hospital for the growth of market and competitive forces over time.

Finally, to examine whether a lower percentage of risk-appropriate deliveries could be predicted by greater availability of lower level NICUs, and by competitive variables, models tested dependent variables for risk appropriate deliveries for the growth of NICU availability, and competitive forces, over time.

The results of the trend analysis and of the mixed models are presented in Chapter 5.

84

CHAPTER 5. RESULTS OF TREND ANALYSIS

5.1. Trends in Neonatal Care: Statewide and Regional Averages, With and Without Outlier Regions

The following tables provide descriptive information on changes in the study variables between the beginning and the end of the study period. They summarize changes in NICU availability, in the distribution of normal births, and in risk appropriate deliveries for VLBW births. These trends provide the basis for the statistical analysis that follows.

Trends are presented both based on statewide totals, and using regional averages. The differences between these approaches is described below.

In addition, certain outlier regions are presented separately, first in the statewide analysis, and then for the regional data. These are the two Kaiser regions (Northern California Kaiser Region 10 and Southern California Kaiser Region 11), the two regions where the Level III hospital is a children's hospital that does not perform deliveries (East Bay Region 3 and Fresno Region 5), and a small region in Los Angeles (Region 67) that was assigned a Level III mid-way through the study period, and where all of the hospitals have a NICU. As discussed below, these could disproportionately affect results regarding the distribution of VLBW births. In addition the Kaiser regions are outliers regarding HMO enrollment.

Statewide trends were determined using all data for the state. Regional trends were also needed to interpret the results of the statistical analysis, and were determined by summing up and averaging the separate values for each variable in each region. The data files are based on regions, and the variables in the statistical models represent regional values. If the percentage of risk appropriate births, for example, is determined by summing up and then dividing the 17 regional averages, this number is likely to vary from one derived for the absolute statewide total. In order to quantify the effect of the independent variables, it is necessary to know the regional value for each variable at the beginning and end of the study period.

Licensed vs. certified NICUs The present study also counted the number of ICNNs licensed by OSHPD, and all NICUs certified by CCS, as of 1997. There were 169 ICNNs, of which 109 were certified NICUs; 60 ICNNs, or 35% of the total, were not certified.. The study did not cross reference back to determine whether all certified facilities are also licensed. All of the 16 NICUs identified by the present study as high-volume Level II+ as of 1997 were certified. Among the 56 low-volume Level II+ NICUs, 48 were certified, and 8 were not.

Statewide trends: Increases in lower level and low volume NICUs Table 5.1 presents the changes in key variables regarding NICU availability and the distribution of normal and VLBW births by NICU level. There were increases in the number of low level and low volume NICUs. Normal deliveries fell at hospitals with 86

no NICU and Level III hospitals, but increased at hospitals with low volume and low level NICUs, as well as at high volume Level II+ hospitals. Low volume II+ hospitals had the highest share of normal deliveries in 1997.

The percentage of VLBW deliveries also fell at hospitals with no NICU and at Level III hospitals, and increased somewhat at Level II+ hospitals. The two categories of high an low volume II+ hospitals combined had 43.6% of VLBW births in 1997, with 25.3% at the 54 low-volume II+ hospitals, and 18.4% at the 16 high-volume II+ hospitals. Although Level III hospitals lost market share of both normal and VLBW deliveries, the overall rate of risk-appropriate deliveries, defined as births at high volume II+ hospitals plus those at Level III, increased slightly, due to the increase at high-volume Level II+ hospitals,.

87

Table 5.1 Change in Number of Hospitals by NICU Level, VLBW Births, and Normal Births, 1986-1997 Statewide Totals for California Number of Hospitals, By NICU Level Year Level LowTotal High II Vol II+ Low Vol II+ Total 1986 43 26 69 2 Total 1997 75 54 129 16 # Change 32 28 60 15 % Change 74.4 111.5 87 750

Level III 26 25 1 3.8

Total High 28 41 17 60.7

No NICU 413 320 -93 -22.5

Percentage of All Normal Deliveries at Each Level Year Level LowTotal High II Vol II+ Low Vol II+ Total 1986 19.0 15.0 34.0 2.4 Total 1997 24.2 23.2 47.4 13.5 % Change 27.4 54.7 39.4 462.5

Level III 21.3 13.7 -35.7

Total High 23.7 27.3 15.2

No NICU 42.3 25.4 -40

Percentage of All VLBW Deliveries At Each Level Year Level LowTotal High II Vol II+ Low Vol II+ Total 1986 18.2 19.9 38.1 3.1 Total 1997 16.5 25.2 41.7 18.4 % Change -9.3 26.6 9.4 494.5

Level III 41.7 30.2 -27.6

Total High 44.8 48.6 8.5

No NICU 17.1 9.8 -42.7

Following in Table 5.2 is a more complete table of statewide trends in major dependent and independent variables.

88

Table 5.2. Statewide Neonatal Care Trends 1986-1997 Major Dependent and Independent Variables 1986 1997 % Change Supply No. Hospitals No. Maternity Hospitals (>10 births) No. II NICUs No. II+ NICUs No. III NICUs No. Small II/II+ ( |t| 0.0550 0.0092 0.0300 0.0386

Full Model 1b. Without Kaiser Regions 10 and 11 Solution for Random Effects Effect

Perinatal Region

Estimate

Pr > |t|

110

PNREGION PNREGION PNREGION PNREGION

4 61 63 66

211.15 185.80 -378.52 -218.43

0.0595 0.0573 0.0055 0.0260

Summary of findings, Number of II/II+ NICUs , and Number of Births per II/II+ NICU Bed: YEAR was an important factor in growing availability of both Level II/II+ NICUs and II/II+ NICU beds. The number of II/II+ NICUs within 10 miles of a Level III (L3L10M) also contributed to growing availability of lower level NICUs beds. (This variable was not modeled for number of lower level NICUs because of collinearity.) Declining hospital occupancy rates were associated with more Level II/II+ NICUs. Unlike the number of II/II+ NICUs, hospital occupancy rate did not play a significant role in bed availability. The perinatal Herfindahl Index was a significant factor for bed availability, but not the number of II/II+ NICUs. The percentage of MediCal coverage was significantly associated with constraining the number of NICUs, but neither HMO nor MediCal coverage had any effect on bed availability.

These findings suggest that NICU availability was driven to some extent by competition for normal deliveries, though the magnitude of the effects is small. The presence of price sensitive payors had no effect on constraining NICU availability.

In the case of population need, the number of births contributed to greater number of NICUs, but not to the number of NICU beds. Table 5.2 shows that both the number of II/II+ NICUs and the number of II/II+ NICU beds continued to increase when the number of births declined after 1991. As Table 6.7 shows, for most years the increase 111

in the number of beds parallels the increase in the number of NICUs until about 1994, when the number of beds slopes upward more sharply. Table 6.3 lists all existing NICUs and the years in which they added beds. It appears that the expansion of both beds and units in the early 1990s may have been responsive to the increased birth rate, and VLBW births, while later additions of beds were independent of both. Paradoxically, these later expansions may reflect to some extent the growth of highervolume II+ NICUs. The statistical results indicate that by region and across all years, bed expansions were responsive to competitive forces and payors rather than to population need.

Finally, the major contribution of the time variable, "year," suggests that there are important explanatory variables that are not in the model, or that when they appear they may not be predictable.

112

Table 6.10 Hypothesis IIb. Dependent Variables: NICU Levels Dependent Variables and Results II Low-vol. High-vol III NICU II+ NICU II+ NICU Independent Variable Estimate pr>t Estimate pr>t Estimate pr>t Estimate pr>t Intercept -0.6950 .0197 -0.9912 0.0008 -0.7156 100 (1987) 2 6 20-25% 14(1997) NICU Level

NICU beds

Butte NT Enloe Mem Sacramento ID#340947 Sacramento Mercy San Juan

2 0 3

Sacramento UC Davis

4

169

AvgOccRt

4 25 0 0 6 130 (1991) 12 100 (1992) 18 86 (1995) 40 30 80 (1992)

65 (1994) Sacramento Sutter Mem

4

Sacramento Kaiser-Sacto

3

8 (1994) NICU beds

29

County

Hospital

San Joaquin

Dameron

3

San Joaquin Gen

16 (1994) 40 (94-97) 2 6 187

San Joaquin

NICU Level

48 >100 (93&94) 55 (1994) 85 (1997)

AvgOccRt 6 >120 (91-93)

3 (1989)

San Joaquin Shasta

St. Joseph's Stockton

12(1987) 18(1988) 32 (1992) 3 10 (1995)

Mercy MC Redding

2

Solano

North Bay Med Ctr

Sutter Fremont Hosp Region 3: East Bay County Hospital

4 6 (1988) 3 3 11 (1992) 16 (1994) 2 (1993) 4 NICU Level

Alameda

Alta Bates

3

Alameda

Children's

4

Alameda Alameda Alameda

Eden Med Ctr Highland Summit

Alameda Alameda Contra Costa Contra Costa Contra Costa

Valley Care MC 2 (1991) Washington-Packard 2 (1997) Brookside 2 Merrithew

2 (1989) 2 2

2 (1993)

John Muir

2

170

171 100 58-40% 70 27 70-201% 100-200% 71 40 50-32%

NICU AvgOccRt beds 13 >100 82-47 (1995-7) 37 (1995) 40; 100-81 47(1988) 3 10 to 25 8 85-54(97) 8 >100 (9397) 2 15-50 2

23-50

6

26

5

44 to 106

Contra Costa Contra Costa

Mt. Diablo

3 (1990) 2 (1995)

San Ramon Regional MC

2 (1990)

County

Hospital

NICU Level

Region 4: Mid-Coast Monterey Natividad Sierra Vista Regional MC San Luis Obispo San Mateo San Mateo San Mateo Santa Clara Santa Clara

Santa Clara

Mills Memorial 2 Mills-Peninsula 2 (1997) Sequoia 2 Alexian 2 Community-LosGatos 2

Columbia/Good Samaritan

126 to 143 19 to 8

2

13 – 37

NICU AvgOccRt beds

2 (1990) 2

Santa Clara El Camino

9 5

2

2 3 (1990)

15 6

48 32 to 54

10 (1996) 128 5 0 5 0 6 0 30to67;101(97) 6 8 9 20,10,40,14 2 (1989) 4 102 to 120 16 56 to 63 (1993) 8 41 to 64 16 90 to >100 (1989) 3 17 to 36 10 57 (1997) 7 16;68;44 22 100 28 60 (1994) 25 (closed 100 to 171

Santa Clara O'Connor

2

Santa Clara San Jose MedCtr Santa Clara Sta Clara Valley MC

2 4

Santa Clara Stanford

4

Santa Clara Lucile Packard

4 47 (1991) 92 60 111;45 (1992) 66 (1995) 44 to32 25;92;63;84 2 6

Santa Cruz Dominican Sta Cruz Hosp – Soquel Santa Cruz Watsonville Comm Santa Cruz Dominican Sta Cruz

91)

2 10 2 3 (closed

171

26; 8; 30 40

90) Region 5: San Joaquin Valley County Hospital Fresno

Fresno Comm Hosp

County

Hospital

Fresno

Valley Children's Hosp

Fresno

Valley MC Fresno

Kern Kern

Bakersfield Mem Hosp

NICU AvgOccRt beds 2 6 22;15 19 (1997) 5

NICU Level

NICU AvgOccRt beds 41 62; 85 52 (1994) 82 62 63 (1996) 10 110 10 140 15 108; 75 (1986) 10 86; 40 10 77; 54 16 88 (1993) 28 90 (1994) 9 50 30 100 6 75; 40 6 62 8 77; 95 6 65

4

2 3 (1989) 3 3 3

Kern M C

Mercy Hosp Bakersfield Kern Stanislaus Doctors Med Ctr Stanislaus Emmanuel Med Ctr Stanislaus Memorial Hosp Modesto Stanislaus Modesto City Kaweah Delta District Hosp Tulare Region 7: Inland County Region County Hospital

Riverside

Desert Hosp

Riverside Riverside

Eisenhower MC Parkview Comm Hosp

Riverside Riverside

NICU Level

2 (1992) 3 (1990) 2 (1993) 2 (1995) 3 2 NICU Level

Riverside Comm Hosp MC

NICU AvgOccRt beds 3 4 150 30 (1993) 50 25 (1997) 75 2 8 35 3 5 125; 220

2 3

Riverside Gen/Univ MC

172

12 (1993) 13 13

>100 50; 40 110

15 (1992) 4 30

Loma Linda Univ MC San Bernardino

San Redlands Comm Bernardino County Hospital

58 (1989) 95 72 (1992) 75 3 8 64; 100; 90 NICU Level

San San Antonio Comm Bernardino

3

3 (1995) SB Comm Hosp of SB 2 SB St. Bernadine MC 3 SB St. Mary Regional MC 2 (1995) Region 8:Orange County County Hospital NICU Level Children's Hosp of Orange Cty Orange 2 Anaheim Mem Hosp

Orange

Hoag Mem/Presbyterian

Orange

Fountain Valley Rgnl MC-Euclid

NICU AvgOccRt beds 4 214

7 (1993) 20 (1997) 2 10

San San Bernardino Cty Bernardino

Orange Orange

80 125

15 5 20 8

160 53 140 65 100 65; 80 70

NICU AvgOccRt beds 8 70 12 (1992) 30 2 (1994) 8 20 3 8 165 14 (1989) 90 3 8 74; 94 12 (1993) 3 10

Orange

Mission Hosp Regnl MC UC Irvine MC

Orange

AMI Garden Grove MC 2 (1992)

Orange

Saddleback Mem MC

3

Orange Orange Orange

St. Joseph-Orange St. Jude MC Western Med Ctr

4 3 2

Orange

Martin Luther MC

2

4

173

30 8 12 (1994) 16 1 (1991) 14 6 8 16 (1996) 8

70 40; 65 115 ; 20 (1996) 129 50 43 50 85; 56 55; 105 110 62 100

Orange Irvine MC Children's Hosp @ Mission Orange Region 9:San Diego/Imperial County Region County Hospital Imperial Imperial San Diego San Diego San Diego

El Centro Regional MC Pioneers Mem Hosp Alvarado Hosp MC Scripps Mem-Chula Vista Children's Hosp San Diego

3 (1990) 8 3 22 (1994) 2 (1990) 16 3 (1993) 16

NICU Level 2 2 2 (1990) 2 4

10 35 90; 40 85 65 150 68 130 30 85 45 75 34;22 55

4 20 (1989) 4 40 2 (1992) 6 2 6

75 50 75 35 36; 10

Sharp Mem Hosp

3

San Diego

Grossmont

3

San Diego

Mercy Hosp & MC

2

San Diego San Diego San Diego

Palomar MC

San Diego

Tri-City MC

Scripps Mem-LaJolla

San Diego UC San Diego MC San Diego Sharp Chula Vista San Diego Pomerado Region 61:Los Angeles – Long Beach County Hospital LA

Downey Comm Torrance Mem MC Specialty Hosp of So. CA

NICU AvgOccRt beds 4 50; 78; 30 2 5 10 29 37 (1994) 14 61 (1992) 6 24 (1991) 12 19 (1992) 6 10 5

San Diego

Paradise Valley Hosp

110 50 20 45

3 (1992) 2 (1992) 2 (Closed 1994)

3

NICU Level 2 (1992) 2 3 (1990) 2(closed1992 )

Long Beach Comm Hosp

3 Long Beach Mem MC 4 Pioneer Hosp 2 (1990) St. Francis MC 2

174

NICU AvgOccRt beds 7 85 12 60 12 90 4 136; 25 20 71 6 10

65 60 290; 75 100

3 (1990) 10 3 20 (1996) LA County/MLK/Drew 4 10 43 (1993) Region 62:Los Angeles – South Bay County

Hospital

LA

St. Mary MC

NICU Level

Centinela Hosp MC Little Company of Mary

Mem Hosp of Gardena South Bay Hospital

NICU AvgOccRt beds 3 9 100 25 (1989) 89;60;35 3 10 50; 100 (1992 on) 2 5 95

9 (1989) 12 (1990) 2 (1990) 7

LA County/Harbor/UCLA

Region 63:Los Angeles/Ventura County Hospital NICU Level LA Antelope Valley MC

200 150 200 80

3

5

4

18

98 60 missing 150 til 93; 35 85; 50

NICU AvgOccRt beds 3 6 200 10 (1989) 120 27 (1994) 120; 0 Tarzana Encino Rgnl MC 3 21 60; 26 (97) Cedars Sinai 4 30 70 32 (1991) 80; 24 (97) No. Hollywood MC 3 (894 125; 70 93) Los Robles Rgnl MC 3 (1995) 10 60 Region 64:Los Angeles/Santa Barbara/Ventura County Hospital NICU NICU AvgOccRt Level beds Daniel Freeman Mem 3 8 100 (199396) 13 (1996) 95 Vencor Hosp LA 2 8 (til 60 1992) Providence Holy Cross MC 2 5 25 (89-95)

175

Northridge Hosp MC Sta.Monica UCLA MC

St. John's Hosp&HlthCtr

3 2 3 (1990)

UCLA MC Valley Presbyterian West Hills Rgnl MC

4 2 3 (1989) 2

W. Valley Hosp&HlthCtr

Sta. Barbara

LACty Olive View MC

2 3

Sta. Barbara Cottage Hosp

3

16 8 6

75; 100 100 20; 0

2 (1996) 20; 100 30 50 23 (1993) 60 25 85 25 90 4 80; 100; 68 11 (1993) 39; 29 2 40 (90-93) 24 60 8 150

22 (1996) 65 7 80 12 (1997) 82 Ventura 2 8 74 20 (1988) 25 30 (1991) 20 3 (1993) 30 75; 35 St. John's Regnl MC 2 (1993) 8 75 Region 65:Los Angeles County Hospital NICU NICU AvgOccRt Level beds Glendale Adventist MC-Wilson LA 3 8 >100 Ventura

Comm Mem HospSan Buenaventura Ventura County MC

3 (1996)

Tr Queen of Angels/Hollywood Presby

Good Samaritan Providence St. Joseph MC

2

6

45; 100 (91,92);80

3 (1989) 3

18 6

>100 >100 75

4

20 (1993) 48

LA County USC MC Region 66: Los Angeles County Hospital

NICU Level

Beverly Hospital Huntington Mem

3 4

176

85(86);41;96(94);7 5

NICU AvgOccRt beds 4 (1995) 127 10 100; 300

Hosp Citrus Valley MC,IC Campus Methodist Hosp of So. CA Pomona Valley Hosp MC

County

Hospital Presbyterian Intercommunity

51 (1994) 8

(91-93) 54 80

13 (1994) 3 (1989) 6

40; 60

3 (1991)

2 9 61; 145 3 (1989) 9 166; 265 3 33 (1992) 90 NICU NICU AvgOccRt Level beds 2 11 50; 106 (94)

CitrusValley/Queen ofValleyCampus

3

26 (1995) 42 16 75; 114 (94) 22 (1995)

Region 67: Los Angeles County Hospital LA

NICU Level

California Hospital MC

3

Garfield MC

3

White Memorial MC

4

Kaiser Oakland Hayward Walnut Creek

4 3 2

80

NICU AvgOccRt beds 14 110 26 (1993) 14

100 100; 125 (1997) 100 95; 71 72; 61 80 15; 77

20 29 (1995) Children's Hosp of LA 4 37 33 (19950 San Gabriel Valley 2 (1994) 6 MC Region 10: Kaiser North County Hospital NICU NICU AvgOccRt Level beds

Fresno Sacramento

18 >100 16 75; 60; 75 18 40 21 (1996) 45 2 (1995) 12 10; 24 4 9 162 (86);90 13 (1993) 60

177

San Francisco Redwood City Santa Clara Santa Teresa Vallejo Santa Rosa

Region 11: Kaiser South County Hospital LA

4 2 2 2 2 (1995) 2

NICU Level

Kaiser Sunset

4

Bellflower Harbor City Panorama City West Los Angeles

3 2 2 2

Woodland Hills

3

Anaheim

2

Riverside Fontana

3 3

San Diego

3

178

22 (1996) 6 4 19 12 16 4 11 (1995)

77;30 200 44 85; 24 80 ? 100 65; 30

NICU AvgOccRt beds 22 80;65 40 (1995) 40 38 60 3 40; 7 10 40; 30 4 40; 182(89);65 9 (1993) 54; 24 12 70 18 (1995) 45 8 70; 96 (1997) 15 22 8 65 16 (1993) 35 36 (1994) 14; 5 18 80; 100 24 (1992) 82 20 (1993) 65 33 (1997) 60

Appendix G: Variable Definitions File 1a. Region-level data set: all years In this data set, "year" is a variable. Each variable is reported for each region. Separate years can be identified by programming, and where "year" is included in a model, time series analysis can be conducted. The complete list of variables in this data set follows. Region-level data used in analysis, and definitions: Name B22NB BLUEBN BLUEBP BLUEMN BLUEMP CCS1_

Variable Ratio of births to # 2/2+ NICU beds Baby Payer: Blue C/S # Baby Payer: Blue C/S % Mom Payer: Blue C/S # Mom Payer: Blue C/S % CCS Level 1

CCS2_

CCS Level 2/2+

CCS3_

CCS Level 3

DCEN

Avg Daily Census, All NICUs

DCEN2A DCEN2S DCEN3A HD

Avg Daily Census, 2/2+ NICUs Avg Daily Census, Small 2/2+ NICUs Avg Daily Census, Lev 3 NICUs Hospital Discharges

HERFH HERFN

Herfindahl Index, Hosp Admits Herfindahl Index, NICU Admits

179

Source ERS

Definition No. births/No. 2/2+ NICU beds

OSHPD

Payor is non-HMO Blue Cross plan

CCS, Primary level: not a NICU OSHPD, key informants CCS CCS Intermediate (moderate risk) and Community (high-risk)level NICUs. CCS Tertiary/Regional NICU: comprehensive, high-risk services The number of NICU patient days (from the OSHPD Annual Report of Hospitals) divided by the number of days that the hospital was open (days open is equal to the number of licensed bed days divided by the number of licensed beds, both from the OSHPD Annual Report of Hospitals).

OSHPD

Pt. released, died, or transferred from acute to non-acute service

HERFP HLB

Herfindahl Index, Perinatal Admits Hospital Licensed Beds OSHPD

HLBD

Hospital Licensed Bed Days

HMOBN

Baby Payer: HMO/PHP #

HMOBP HMOMN HMOMP HOCCRT

Baby Payer: HMO/PHP % Mom Payer: HMO/PHP # Mom Payer: HMO/PHP % Hospital Occupancy Rate

HOSP HPD L22S L3L10M L3L15M L3L25M LV22PL LV22PP MATHSP MCALBN

Hospital Hospital Patient Days Lev 2/2+, 0-14 Beds Lev 2/2+ Within 10 Mi of Lev 3 Lev 2/2+ Within 15 Mi of Lev 3 Lev 3 w/in 25 miles Level 2/2+ ICN % Level 2/2+ ICN Mat hosp > 10 births Baby Payer: Medi-Cal #

MCALBP MCALMN MCALMP MCARBN MCARBP MCARMN MCARMP MH22N

Baby Payer: Medi-Cal % Mom Payer: Medi-Cal # Mom Payer: Medi-Cal % Baby Payer: Medicare # Baby Payer: Medicare % Mom Payer: Medicare # Mom Payer: Medicare % Mat Hosp > 10 Bth w Lev 2/2+

180

OSHPD

Maximum no. beds hospital is licensed to operate The number licensed beds multiplied by the number of days in the reporting period. No. babies for whom an HMO or Prepaid Health Plan is expected at admission to pay the greatest share of the bill, excluding MediCal and Medicare.

A measure of the utilization of beds over a reporting period. The measurement for the OSHPD Annual Report of Hospitals is derived by dividing number of the patient days by the licensed bed days in the reporting period.

MediCal is California's Medicaid program. MediCal patients covered by HMO are reported as MediCal.

MH22PP MHSPP N3VLBW NBTH ND NHBTH NLB NLB22P NLBD NLBW NOCCRT NOCHBN NOCHBP NOCHMN NOCHMP NPD NUPGD NVLBW OGOVBN

ICN % Mat Hosp > 10 Bth w Lev 2/2+ ICN % of All Hosps with 10+ Births # VLBW Births at a Level 3 # Births OSHPD NICU Discharges # Births NICU Licensed Beds # Lic Beds @ 2/2+ NICU NICU Licensed Bed Days # LBW Births NICU Occupancy Rate Baby Payer: No Charge # Baby Payer: No Charge % Mom Payer: No Charge # Mom Payer: No Charge % NICU Patient Days Opened or upgraded a NICU # VLBW Births Baby Payer: Other Govt #

OGOVBP OGOVMN OGOVMP OTHRBN OTHRBP OTHRMN OTHRMP OWNCH OWNKSR OWNNON OWNPRF OWNPUB P22NB PCTCH

Baby Payer: Other Govt % Mom Payer: Other Govt # Mom Payer: Other Govt % Baby Payer: Oth Non Gov # Baby Payer: Oth Non Gov % Mom Payer: Oth Non Gov # Mom Payer: Oth Non Gov % Owner: Church Owner: Kaiser Owner: Non-Profit Owner: For-Profit Owner: Public % of all NICU beds that are 2/2+ % Hosps: Church Owner

181

No. babies born alive

Free, special research, courtesy pts.

Payment by other American govt. agency not already listed, such as California Children's Services, ShortDoyle

PCTKSR PCTNON PCTPRF PCTPUB PD

% Hosps: Kaiser Owner % Hosps: Non-Profit Owner % Hosps: For-Profit Owner % Hosps: Public Owner Perinatal Discharges

PLB PLBD PLBW PNREGION POCCRT PPD

Perinatal Licensed Beds Perinatal Licensed Bed Days % LBW Perinatal Region Perinatal Occupancy Rate Perinatal Patient Days

PVLBW PVTBN PVTBP PVTMN PVTMP S17BN

% VLBW Baby Payer: Private # Baby Payer: Private % Mom Payer: Private # Mom Payer: Private % Baby Payer: Sect 17000 #

S17BP S17MN S17MP SELFBN SELFBP SELFMN SELFMP TITVBN

OSHPD

Discharge of mother from a maternity and newborn service for the provision of care during pregnancy, labor, delivery, postpartum, and neonatal periods with appropriate staff, space, equipment, and supplies (See Title 22, Division 5, Sections 70547-70553, California Code of Regulations, for details). Commonly called maternity or obstetrical. (Newborn counted separately.)

A unit of measurement denoting the services received by one perinatal inpatient in one 24-hour period. OSHPD

Private commercial carrier, non-HMO

OSPHD

Calif. Sec. 17000 identifies county as payor of last resort

Payer Payer: Sect 17000 % Mom Payer: Sect 17000 # Mom Payer: Sect 17000 % Baby Payer: Self #

OSHPD

Direct payment by the patient, not by any insurance or other third party

Baby Payer: Self % Mom Payer: Self # Mom Payer: Self % Baby Payer: Title V #

OSHPD

Title V Maternal & Child Health

182

program under Medicare TITVBP TITVMN TITVMP VLB2NB VLB3P WCBN WCBP WCMN WCMP YEAR YR86 YR87 YR88 YR89 YR90 YR91 YR92 YR93 YR94 YR95 YR96 YR97

Baby Payer: Title V % Mom Payer: Title V # Mom Payer: Title V % Ratio of VLBW births to # 2/2+ NICU beds % of all VLBW births at a level 3 Baby Payer: Workers Comp # OSHPD Baby Payer: Workers Comp % Mom Payer: Workers Comp # Mom Payer: Workers Comp % Year Year 1986 Year 1987 Year 1988 Year 1989 Year 1990 Year 1991 Year 1992 Year 1993 Year 1994 Year 1995 Year 1996 Year 1997

Workers' Compensation Insurance

Average Daily Census: This is the number of patient days (from the OSHPD Annual Report of Hospitals) divided by the number of days that the hospital was open (days open is equal to the number of licensed bed days divided by the number of licensed beds, both from the OSHPD Annual Report of Hospitals). The Herfindahl index consists of the market share of each firm in the market. Market share is defined as the number of admits for a hospital (from the OSHPD Annual Report of Hospitals) divided by the number of admits for all hospitals in the region (derived from the OSHPD Annual Report of Hospitals). The market share is treated as a value between zero and 1. Each firm's market share is squared, and the squared results are added up. For example, if there is just one firm it has the entire market share. One squared is one, which is the Herfindahl index for that area. If there are two firms and one has a 90% share and the other has a 10% share, 0.9 squared is 0.81, 0.1 squared is 0.01. Summing those up the Herfindahl index is 0.82. A higher number indicates a greater degree of concentration, and less dispersion. Lower

183

numbers are associated with greater dispersion, and hence with greater competition. A Herfindahl index of .82, for example, indicates a much more concentrated industry than 0.5.

Hospital Licensed Bed Days This is from the OSHPD Annual Report of Hospitals. Hospital Occupancy Rate This is equal to the number of patient days divided by the number of licensed bed days (both from the OSHPD Annual Report of Hospitals) NICU Discharges This is from the OSHPD Annual Report of Hospitals. No. LBW births This is from the OSHPD Annual Report of Hospitals. Perinatal Licensed Beds This is from the OSHPD Annual Report of Hospitals. Payor source information (HMO and Medi-Cal) came from linked birth-death files developed by Ciaran Phibbs, PhD, using OSHPD data. Following is a summary of most source variables. Others, like the ratio of births to the number of level 2/2+ beds, are derived from these variables: Description and Source NICU Level: OSHPD Annual Report of Hospitals, California Children's Services, and interviews with key informants. Hospital Average Length Of Stay: OSHPD Annual Report of Hospitals Hospital Discharges: OSHPD Annual Report of Hospitals Hospital Licensed Beds: OSHPD Annual Report of Hospitals Hospital Licensed Bed Days: OSHPD Annual Report of Hospitals Hospital Occupancy Rate: OSHPD Annual Report of Hospitals Hospital Patient Days: OSHPD Annual Report of Hospitals NICU ALOS: OSHPD Annual Report of Hospitals # Births: OSHPD Annual Report of Hospitals OSHPD reports the no. of live births from the majority of facilities. The California Health Information for Policy Project reported in 1996 on the results of matching OSHPD discharge data from 1992 with birth and death certificates.(Herrchen and Gould) By recording only live births, discharge reports exclude about .63% of deliveries that are fetal deaths (p.19). About .53% of births, or 3,204, occurred in facilities that do not report to OSHPD, such as birthing centers, 1.76% in military facilities, and .69% in other

184

uncoded facilities or out-of-state hospitals, for a total of 2.98% of overall births. There were no major changes in OSHPD data collection between 1986 and 1997, so the results for 1992 should be typical. Since the subject of this study is the distribution of births that occur in hospitals that do report to OSHPD, this minor discrepancy should have no bearing on the study. The OSHPD report only counts live births, as defined below. This excludes fetal deaths. (Confirmed by personal communication on 10/6/00.) The Herrchen report estimates that .6% of deliveries in 1992 resulted in fetal deaths. It is unlikely that these deliveries were candidates for NICU care or that delivery at a different level of care would make a difference, since NICUs are designed to prolong the life of distressed newborns who are delivered alive. However, to the extent that any of these fetal deaths might have been prevented by appropriate care at a higher level hospital, the present study does not reflect whether or not their site of delivery was appropriate. The OSHPD reporting form for births follows, from the Annual Report of Utilization Data: BIRTH AND ABORTION DATA A. Enter the number of the following events which occurred in your hospital during the calendar year. If a particular event did not occur in your hospital, enter a "0". Line No. EVENT TOTAL OCCURRING IN HOSPITAL 6 Total Live Births (Count multiple births separately)1 * 7 Live Births with Birth Weight Less Than 2500 grams (5lbs. 8 oz.) 2 8 Live Births with Birth Weight Less Than 1500 grams (3lbs. 5 oz.) 2 9 Induced Abortions Inpatient 3 10 Induced Abortions Outpatient (ambulatory)3

*The number of births shown on this line should be approximately the same as the number of discharges shown on Page 8, Line 2, Col. 3. Include LDR or LDRP births in table above. B. Enter the number 1 (yes) if the hospital had an alternative setting…….. 11____ (i.e. an approved birthing program) LDR 4 LDRP 4 If yes, your alternative setting was approved as (check correct alternative) 12_______ (Col. 1) (Col. 2) How many of the live births reported on line 6 occurred in your alternative setting?……..............…………………………………………………………… 13_______

185

Do not include C-Section deliveries. How many of the live births reported on line 6 were Cesarean Section deliveries?..14__ 1 LIVE

BIRTH The complete expulsion or extraction from its mother, in a hospital, of a product of conception, irrespective of the duration of pregnancy, which after such separation, breathes or shows any other evidence of life such as beating of 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; each product of such a birth is considered live born. When more than one live product of conception is expelled (multiple birth), each one constitutes a separate live birth. EXCLUDE live births occurring outside your hospital. 2 LIVE

BIRTHS UNDER 2500 GRAMS; UNDER 1500 GRAMS Of the total live births, how many weighed less that 2500 grams (5 lbs., 8 oz.); of the births weighing less than 2500 grams, how many weighed less than 1500 grams (3 lbs., 5 oz.)? 3 INDUCED

ABORTIONS Intentionally induced abortions (chemically or surgically), performed on an outpatient or inpatient basis, irrespective of gestational age. 4 LDR

(Labor, Delivery and Recovery) and LDRP (Labor, Delivery, Recovery and Post-Partum) LDR is a program for low-risk mothers with stays of less than 24 hours, including equipment and supplies or uncomplicated deliveries in a home-like setting and that has been approved by the Division of Licensing and Certification, Department of Health Services (L&C). LDR replaces ABC (Alternative Birthing Center). LDRP is a program similar to LDR but is not limited to low-risk deliveries and the stays are usually for more than one day. LDRP also is L&C approved.

C. Enter the number of newborn nursery days……………………………………15_____ NICU Discharges: OSHPD Annual Report of Hospitals NICU Licensed Beds: OSHPD Annual Report of Hospitals NICU Licensed Bed Days: OSHPD Annual Report of Hospitals # LBW Births: OSHPD Annual Report of Hospitals NICU Occupancy Rate: OSHPD Annual Report of Hospitals NICU Patient Days: OSHPD Annual Report of Hospitals # VLBW Births: OSHPD Annual Report of Hospitals Perinatal ALOS: OSHPD Annual Report of Hospitals Perinatal Discharges: OSHPD Annual Report of Hospitals Perinatal Licensed Beds: OSHPD Annual Report of Hospitals Perinatal Licensed Bed Days: OSHPD Annual Report of Hospitals Perinatal Region: Ciaran Phibbs linked B/D file Perinatal Occupancy Rate: OSHPD Annual Report of Hospitals Perinatal Patient Days: OSHPD Annual Report of Hospitals Lev 3 w/in 25 miles of Lev 2/2+: Derived from Ciaran Phibbs linked B/D file Lev 3 w/in 15 miles of Lev 2/2+: Derived from Ciaran Phibbs linked B/D file Lev 3 w/in 10 miles of Lev 2/2+: Derived from Ciaran Phibbs linked B/D file

186

License type: OSHPD Annual Report of Hospitals Ownership: Public OSHPD Annual Report of Hospitals Ownership: Non-Profit OSHPD Annual Report of Hospitals Ownership: For-Profit OSHPD Annual Report of Hospitals Ownership: Kaiser OSHPD Annual Report of Hospitals Ownership: Church OSHPD Annual Report of Hospitals Neonatal Mortality Rate: California Children's Services # Neonatal Deaths: OSHPD Annual Report of Hospitals Teaching Hospital (COTH): Linked B/D file Payor Source (All types) Linked B/D file Charges (baby and mom) Linked B/D file Maternal race Linked B/D file

187

Appendix H. High-Volume Level II+ Hospitals, 1997 PNREGION 2 3 4 5 7 7 9 11 11 61 64 65 65 66 67 67

OSHPDID HOSPITAL NAME 340950 010739 430779 500852 331164 361339 370694 190430 370730 190754 190812 190392 190758 190630 190125 190315

Mercy San Juan Hospital Alta Bates Medical Center Columbia Good Samaritan Hospital Doctors Medical Center Desert Hospital St Bernadine Medical Center Sharp Memorial Hospital Kaiser Fdn Hosp - Bellflower Kaiser Fdn Hosp - San Diego St Francis Medical Center Valley Presbyterian Hospital Good Samaritan Hospital Providence St Joseph Med Center Pomona Valley Hospital Med Ctr California Hospital Medical Ctr Garfield Medical Center

188

Appendix I. High volume Level II+ Hospitals, By Region, 1986-1997 OBS YEAR

PNREGION OSHPDID HOSPITAL NAME

95 96 97

2 2 2

340950 Mercy San Juan Hospital 340950 Mercy San Juan Hospital 340950 Mercy San Juan Hospital

1061 1062 1063 1064 1065 1067

89 90 91 92 93 94 96

2 2 2 2 2 2 2

391010 San Joaquin General Hospital 391010 San Joaquin General Hospital 391010 San Joaquin General Hospital 391010 San Joaquin General Hospital 391010 San Joaquin General Hospital 391010 San Joaquin General Hospital 391010 San Joaquin General Hospital

1399 1400 1401 1402 1403 1404

92 93 94 95 96 97

3 3 3 3 3 3

010739 Alta Bates Medical Center 010739 Alta Bates Medical Center 010739 Alta Bates Medical Center 010739 Alta Bates Medical Center 010739 Alta Bates Medical Center 010739 Alta Bates Medical Center

2070 2071 2072 2073 2074 2075 2076

91 92 93 94 95 96 97

4 4 4 4 4 4 4

430779 Columbia Good Samaritan Hospital 430779 Columbia Good Samaritan Hospital 430779 Columbia Good Samaritan Hospital 430779 Columbia Good Samaritan Hospital 430779 Columbia Good Samaritan Hospital 430779 Columbia Good Samaritan Hospital 430779 Columbia Good Samaritan Hospital

2399

96

5

100822 Valley Medical Center of Fresno

2481 2483

94 96

5 5

150736 Kern Medical Center 150736 Kern Medical Center

2765 2766 2767 2768 2769 2770

90 91 92 93 94 95

5 5 5 5 5 5

500852 Doctors Medical Center 500852 Doctors Medical Center 500852 Doctors Medical Center 500852 Doctors Medical Center 500852 Doctors Medical Center 500852 Doctors Medical Center

189

2771 2772

96 97

5 5

500852 Doctors Medical Center 500852 Doctors Medical Center

3080 3082 3083 3084

93 95 96 97

7 7 7 7

331164 Desert Hospital 331164 Desert Hospital 331164 Desert Hospital 331164 Desert Hospital

3164

93

7

331293 Parkview Community Hospital

3184

89

7

331313 Riverside Gen Hosp-Univ Med Ctr

3429 3432

94 97

7 7

361339 St Bernadine Medical Center 361339 St Bernadine Medical Center

3822

91

8

301317 Saddleback Memorial Medical Ctr

4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152

86 87 88 89 90 91 92 93 94 95 96 97

9 9 9 9 9 9 9 9 9 9 9 9

370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital 370694 Sharp Memorial Hospital

4505 4506

90 91

10 10

430805 Kaiser Fdn Hosp – Santa Clara 430805 Kaiser Fdn Hosp – Santa Clara

4573 4574 4575 4576 4577 4578 4579 4580 4581 4582

86 87 88 89 90 91 92 93 94 95

11 11 11 11 11 11 11 11 11 11

190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower

190

4583 4584

96 97

11 11

190430 Kaiser Fdn Hosp – Bellflower 190430 Kaiser Fdn Hosp – Bellflower

4684 4685 4686 4687 4692

89 90 91 92 97

11 11 11 11 11

370730 Kaiser Fdn Hosp – San Diego 370730 Kaiser Fdn Hosp – San Diego 370730 Kaiser Fdn Hosp – San Diego 370730 Kaiser Fdn Hosp – San Diego 370730 Kaiser Fdn Hosp – San Diego

4832

93

61

190475 Long Beach Community Hospital

4949 4950 4951 4952 4953 4954 4955 4956

90 91 92 93 94 95 96 97

61 61 61 61 61 61 61 61

190754 St Francis Medical Center 190754 St Francis Medical Center 190754 St Francis Medical Center 190754 St Francis Medical Center 190754 St Francis Medical Center 190754 St Francis Medical Center 190754 St Francis Medical Center 190754 St Francis Medical Center

5007 5008 5009 5010 5011 5012

88 89 90 91 92 93

62 62 62 62 62 62

190053 St Mary Medical Center 190053 St Mary Medical Center 190053 St Mary Medical Center 190053 St Mary Medical Center 190053 St Mary Medical Center 190053 St Mary Medical Center

5157

94

63

190034 Antelope Valley Hospital Med Ctr

5422

95

64

190230 Daniel Freeman Memorial Hospital

5504 5506

93 95

64 64

190568 Northridge Hospital Medical Ctr 190568 Northridge Hospital Medical Ctr

5596 5597 5598 5599 5600 5601 5602 5603 5604

89 90 91 92 93 94 95 96 97

64 64 64 64 64 64 64 64 64

190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital 190812 Valley Presbyterian Hospital

191

5657 5658 5660 5661

90 91 93 94

64 64 64 64

191231 Los Angeles Co Olive View Med Ctr 191231 Los Angeles Co Olive View Med Ctr 191231 Los Angeles Co Olive View Med Ctr 191231 Los Angeles Co Olive View Med Ctr

5721 5722

94 95

64 64

420514 Santa Barbara Cottage Hospital 420514 Santa Barbara Cottage Hospital

5792 5793 5794 5795

93 94 95 96

64 64 64 64

560481 Ventura County Medical Center 560481 Ventura County Medical Center 560481 Ventura County Medical Center 560481 Ventura County Medical Center

6005 6006 6007 6008 6009 6010 6011 6012

90 91 92 93 94 95 96 97

65 65 65 65 65 65 65 65

190392 Good Samaritan Hospital 190392 Good Samaritan Hospital 190392 Good Samaritan Hospital 190392 Good Samaritan Hospital 190392 Good Samaritan Hospital 190392 Good Samaritan Hospital 190392 Good Samaritan Hospital 190392 Good Samaritan Hospital

6104 6108

93 97

65 65

190758 Providence St Joseph Med Center 190758 Providence St Joseph Med Center

6378 6379 6380 6381 6382 6383 6384

91 92 93 94 95 96 97

66 66 66 66 66 66 66

190630 Pomona Valley Hospital Med Ctr 190630 Pomona Valley Hospital Med Ctr 190630 Pomona Valley Hospital Med Ctr 190630 Pomona Valley Hospital Med Ctr 190630 Pomona Valley Hospital Med Ctr 190630 Pomona Valley Hospital Med Ctr 190630 Pomona Valley Hospital Med Ctr

6399 6404 6405 6406 6407

88 93 94 95 96

66 66 66 66 66

190636 Citrus Valley Med Ctr-QV Campus 190636 Citrus Valley Med Ctr-QV Campus 190636 Citrus Valley Med Ctr-QV Campus 190636 Citrus Valley Med Ctr-QV Campus 190636 Citrus Valley Med Ctr-QV Campus

6495 6497 6498

88 90 91

67 67 67

190125 California Hospital Medical Ctr 190125 California Hospital Medical Ctr 190125 California Hospital Medical Ctr

192

6499 6500 6501 6502 6503 6504

92 93 94 95 96 97

67 67 67 67 67 67

190125 California Hospital Medical Ctr 190125 California Hospital Medical Ctr 190125 California Hospital Medical Ctr 190125 California Hospital Medical Ctr 190125 California Hospital Medical Ctr 190125 California Hospital Medical Ctr

6563 6564

96 97

67 67

190315 Garfield Medical Center 190315 Garfield Medical Center

193

Appendix K. Market Share of Normal Deliveries by Hospital NICU Level, 1986-1997 Perinata Year Level II Low Vol Total High Vol Level Total l Region Level II+ Low Level II+ III High Lev

No NICU

1 1

1986 1997

20 33

23.4 18.3

44 51.5

0 0

28 29.7

27.9 29.7

28 19

2 2

1986 1997

13 18

2.9 17.6

16 35.8

0 6.1

20.6 15

20.6 21

64 43

3 3

1986 1997

8 44

21 10.2

29 54.1

0 21.8

0 0

0 21.8

71 24

4 4

1986 1997

41 44

0 12.9

40.7 56.5

0 10.6

18.4 17.7

18.4 28.3

41 15

5 5

1986 1997

27 33

24 9.1

51 41.8

0 9.1

0 0

0 9.1

49 49

6.1 6.1

1986 1997

19 4

0 25.6

19.3 29.9

0 16.6

43.5 25.5

43.5 42.2

37 28

6.2 6.2

1986 1997

12 8

22.1 64.3

34 72

0 0

37.8 16

37.8 16

28 12

6.3 6.3

1986 1997

0 0

15.2 45

15.2 45

0 0

48.6 49.8

48.6 49.8

36 5

6.4 6.4

1986 1997

29 20

7.6 36.9

36.4 56.9

0 8.3

8.4 4.1

8.4 12.5

55 31

6.5 6.5

1986 1997

9 29

16.3 10.6

25.4 39.8

0 26.3

60.9 19.4

60.9 45.7

14 14

6.6 6.6

1986 1997

24 6

19.4 42.8

43.8 48.8

0 17.1

18.2 14.4

18.2 31.4

38 20

6.7

1986

0

0

0

0

26.3

26.3

41

194

Perinata l Region

Year

Level II

Low Vol Level II+

Total Low

High Vol Level II+

Level III

No NICU

16.4

Total High Lev 56.1

6.7

1997

0

0

0

39.8

7 7

1986 1997

5 17

8.8 30.9

13.9 47.5

0 13.5

6.6 6.9

6.6 16.1

80 36

8 8

1986 1997

17 18

17.7 32.1

35 50.5

0 0

11.2 14.7

11.2 14.7

54 35

9 9

1986 1997

13 40

19.9 24.3

33.2 64

19.8 21.4

10.7 5.4

30.5 26.7

36 9

10 10

1986 1997

24 31

11 24

35 72.3

0 0

38.3 27.7

38.3 27.7

27 0

11 11

1986 1997

35 24

32.4 15.9

67.3 40.1

16.8 44.2

16 8.5

32.7 52.7

0 7

Total Total

1986 1997

19 24.2

15 23.2

34 47.4

2.4 13.5

21.3 13.7

23.7 27.3

42.3 25.4

195

44

Appendix L. Percentage VLBW Born at Level III, 1986 & 1997 Regional Analysis: % VLBW Born at Level III Decline in 11 of 15 Regions (excludes Regions #3 And #5) Region

Sacramento Santa Clara

Riverside

Kaiser South Los Angeles Cty

1986 1997 change 1 2 3 4

61.2 67.5 62.7 38.3 0 0.0 66.2 42.2

5 7 8 9 10 11 61 62 63 64 65 66 67

0 34.7 20.9 16.7 64 18.2 74.2 43.3 59.3 15 78 41.7 0

196

0.0 17.6 44.8 27.8 34.7 15.0 63.1 35.3 57.4 6.2 44.1 16.2 25.5

9.4% -63.5% 0.0% -56.7% 0.0% -97.4% 53.3% 39.9% 2.1% -21.2% -17.6% -22.5% -3.3% -141.4% -77.1% -157.2% 100.0%

Appendix M. Regional Analysis: % of Risk-Appropriate VLBW Births,1986-97 Volume + Level Used to Define Risk-Appropriate Risk-Appropriate: % VLBW at Tertiary or Hi-Vol II+ Down in 7 Regions *

Not Risk-Appropriate: % VLBW at Intermediate, Low-Vol II+, or No NICU Up in Same 7 Regions * Hi-Vol II+

Region

1986

1997

1

61.2%

2*

1997

67.5%

32.5

-19.4%

62.7%

43.8%

-43.2%* 37.3

56.2

33.6%*



3

0

39.3%

100.0% 100

60.7

-64.7%



4*

66.2%

65.1%

-1.7%* 33.8

34.9

3.2%*



5

0

25.9%

100.0% 100

74.1

-35.0%



7*

34.7%

29.9%

-16.1%* 65.3

70.1

6.8%*



8

20.9%

44.8%

53.3% 79.1

55.2

-43.3%

9

50.6%

63.3%

20.1% 49.4

36.7

-34.6%

10 *

64.1%

34.7%

-84.7%* 35.9

65.3

45.0%*

11

37.4%

52.2%

28.4% 62.6

47.8

-31.0%



61

74.2%

78.7%

5.7% 25.8

21.3

-21.1%



62 *

43.3%

35.3%

-22.7%* 56.7

64.7

12.4%*

63 *

59.3%

57.4%

-3.3%* 40.7

42.6

4.5%*

64 *

15.1%

6.2%

-143.5%* 84.9

93.8

9.5%*



65

78.1%

81%

3.6% 21.9

19

-15.3%



66

41.7%

45.3%

7.9% 58.3

54.7

-6.6%



67

0

88.9%

100.0% 100

11.1

-800.9%



Total

44.8%

49.4%

9.3% 55.2

50.6

-9.1%

197

%Change

Present In 1997

% 1986 Change 9.3% 38.8



Appendix N. Percentage VLBW Births by Region and NICU Level: 1986

Region 1 2 3 4 5 7 8 9 10 11 61 62 63 64 65 66 67 Total

HOSPITAL NICU LEVEL Year Level II LowVol Total II+ Low 86 13.7 17.3 31.0 86 15.4 1.3 16.6 86 0.7 62.9 63.6 86 15.6 0 15.5 86 46.2 19.7 65.8 86 15.4 20.1 35.5 86 17.3 20.4 37.8 86 6.12 20.8 26.9 86 23.6 7.2 30.8 86 30.6 32.1 62.6 86 20.7 0 20.7 86 4.6 35.9 40.5 86 0 12.1 12.1 86 47.6 17.4 64.9 86 4.9 10.3 15.2 86 27.1 17.7 44.8 86 0 88 88 1986 18.1 18.9 37

198

High Vol Level III Total High NO NICU II+ 0 61.2 61.2 7.8 0 62.8 62.8 20.6 0 0 0 36.4 0 66.2 66.2 18.3 0 0 0 34.2 0 34.7 34.7 29.8 0 20.9 20.9 41.3 33.0 16.7 50.6 22.5 0 64.1 64.1 5.1 19.1 18.2 37.4 0 0 74.2 74.2 5.1 0 43.3 43.3 16.1 0 59.3 59.3 28.6 0 15.1 15.1 20 0 78.1 78.1 6.7 0 41.7 41.7 13.5 0 0 0 12 0 41.7 44.8 18.2

Appendix O. Percentage VLBW Births by Region and NICU Level: 1997 Region Year Level II LoVol Total Low Hi Vol II+ II+ 1 2 3 4 5 7 8 9 10 11 61 62 63 64 65 66 67 Total

97 97 97 97 97 97 97 97 97 97 97 97 97 97 97 97 97

8.68 19.2 9.15 20.9 42.75 12.6 17.81 11.2 47.27 3.8 12.11 33.6 17.95 24.5 15.19 15.2 31.17 34.1 15.78 27.0 2.23 17.0 7.33 56.0 0 42.0 25.99 59.6 11.25 5.5 3.3 44.1 0 0.0 16.5% 24.0%

27.9 30.1 55.3 29.0 51.1 45.7 42.4 30.4 65.3 42.7 19.3 63.3 42.0 85.6 16.7 47.4 0.0 40.5

199

0.0 5.5 39.3 22.9 25.9 12.3 0.0 35.5 0.0 37.2 15.6 0.0 0.0 0.0 37.0 29.1 63.5 18.4%

Level Total High No NICU III (Other) 67.6 67.6 4.5 38.3 43.8 26.1 0.0 39.3 5.3 42.2 65.1 5.9 0.0 25.9 23.0 17.6 29.9 24.4 44.8 44.8 12.8 27.8 63.3 6.3 34.7 34.7 0.0 15.0 52.2 5.1 63.1 78.7 2.0 35.3 35.3 1.3 57.4 57.4 0.6 6.2 6.2 8.2 44.1 81.0 2.3 16.2 45.4 7.2 25.5 88.9 11.1 31.1% 49.4% 10.1

Appendix P. California Hospitals By NICU Levels, 1986-1997 Region/Year Level II LowTotal Hi-Vol II+ Lev III Total Vol.II+ Low High 1 1986 4 2 6 0 3 3 1 1997 7 3 10 0 3 3

No NICU 40 36

2 2

1986 1997

3 4

1 4

4 8

0 1

2 2

2 3

60 55

3 3

1986 1997

1 8

1 1

2 9

0 1

1 1

1 2

31 23

4 4

1986 1997

8 12

0 2

8 14

0 1

2 3

2 4

30 22

5 5

1986 1997

3 5

3 3

6 8

0 1

1 1

1 2

52 49

7 7

1986 1997

1 4

1 5

2 9

0 2

1 1

1 3

41 32

8 8

1986 1997

3 5

2 6

5 11

0 0

2 3

2 3

34 27

9 9

1986 1997

2 8

2 3

4 11

1 1

2 2

3 3

27 20

10 10

1986 1997

2 6

3 2

3 8

0 0

3 3

3 3

6 1

11 11

1986 1997

4 4

0 3

7 7

1 2

1 1

2 3

3 2

61 61

1986 1997

3 3

2 2

3 5

0 1

2 2

2 3

21 18

62 62

1986 1997

1 1

1 4

3 5

0 0

1 1

1 1

8 6

200

Region/Year Level II 63 63

1986 1997

0 0

LowVol.II+ 2 4

64 64

1986 1997

5 6

2 7

65 65

1986 1997

1 1

66 66

1986 1997

67 67

1986 1997

Total 1986 Total 1997

Total Low 1 4

0 0

1 1

Total High 1 1

7 13

0 1

1 1

1 2

31 24

1 1

3 2

0 2

2 2

2 4

21 20

2 1

1 5

3 6

0 1

1 1

1 2

20 16

0 0

2 0

2 0

0 2

1 1

1 3

11 11

43 75

26 55

69 130

2 16

27 29

29 45

456 379

201

Hi-Vol II+

Lev III

No NICU 20 17

APPENDIX Q: Results of Statistical Correlations, Major Variables

19

Variable

Label

LV22PL bthp B22NB L3L10M HERFN HERFH HERFP HOSP MATHSP HOCCRT NOCCRT POCCRT HMOMP MCALMP NBTH NVLBW PVLBW NLBW PLBW

Level 2/2+ ICN Births as % of Hospital DC Ratio of births to # 2/2+ NICU beds Lev 2/2+ Within 10 Mi of Lev 3 Herfindahl Index, NICU Admits Herfindahl Index, Hosp Admits Herfindahl Index, Perinatal Admits Hospital Mat hosp > 10 births Hospital Occupancy Rate NICU Occupancy Rate Perinatal Occupancy Rate Mom Payer: HMO/PHP % Mom Payer: Medi-Cal % # Births # VLBW Births % VLBW # LBW Births % LBW

Variables: LV22PL bthp HERFP HOSP MCALMP NBTH

The CORR Procedure B22NB L3L10M HERFN HERFH MATHSP HOCCRT NOCCRT POCCRT HMOMP NVLBW PVLBW NLBW PLBW Simple Statistics

Variable LV22PL bthp B22NB L3L10M HERFN HERFH HERFP HOSP MATHSP HOCCRT NOCCRT POCCRT HMOMP MCALMP NBTH NVLBW

N

Mean

Std Dev

Sum

Minimum

Maximum

204 204 204 204 204 204 204 204 204 204 204 204 191 191 204 204

7.02451 0.18072 551.11484 3.52941 0.23086 0.08631 0.12007 29.31373 18.21078 53.30606 75.19771 56.97817 35.89266 37.42397 31619 357.97059

3.43617 0.04554 357.77503 1.82865 0.10881 0.04687 0.07931 14.35509 9.76117 6.67097 18.53828 12.85127 19.69376 17.17136 10231 147.74822

1433 36.86712 112427 720.00000 47.09450 17.60723 24.49440 5980 3715 10874 15340 11624 6855 7148 6450295 73026

1.00000 0.10238 189.34497 0 0.08682 0.03424 0.04654 10.00000 5.00000 38.49257 17.77616 29.76427 3.13837 0 11241 79.00000

15.00000 0.36591 3479 7.00000 0.66025 0.23206 0.41342 61.00000 45.00000 68.73193 156.66565 98.21751 100.00000 79.86642 51110 1236

Simple Statistics

Simple Statistics Variable PVLBW NLBW PLBW

N

Mean

Std Dev

Sum

Minimum

Maximum

204 204 204

1.14496 1691 5.43326

0.33292 593.64318 1.11758

233.57136 345008 1108

0.38968 611.00000 3.32390

3.20956 4992 11.97869

202

LV22PL LV22PL Level 2/2+ ICN

1.00000 204

bthp

B22NB

L3L10M

-0.13235 0.0592 204

-0.48755