Contract No.: 240-93-0050 MPR Reference No.: 8166-111

EFFECT OF HEALTHY START ON MORTALITY AND BIRTH OUTCOMES

July 2000

Lorenzo Moreno Barbara Devaney Dexter Chu Melissa Seeley

Submitted to: Submitted by:

Health Resources and Services Administration Mathematica Policy Research, Inc. Parklawn Building, Room 14-36 P.O. Box 2393 5600 Fishers Lane Princeton, NJ 08543-2393 Rockville, Maryland 20857 (609) 799-3535

Project Officer: Project Director: Karen T. Raykovich Embry Howell ACKNOWLEDGMENTS

The authors thank the following federal and state entities for providing the vital records for this report:

C Alabama Department of Public Health

C City Department of Health

C California Department of Health Services

C District of Columbia State Center for Health Statistics

C Illinois Department of Public Health

C Indian Health Service (IHS), Public Health Service, U.S. Department of Health and Human Services

C Indiana State Department of Health

C Louisiana Department of Health and Hospitals

C Massachusetts Department of Health and Hospitals

C Michigan Department of Health

C New York City Department of Health

C Ohio Department of Health

C Pennsylvania Department of Health

C South Carolina Department of Health and Environmental Control

The authors also thank Patricia Ciaccio, Roy Grisham, and Walter Brower of Mathematica Policy Research, Inc. for editing the report; and Cindy McClure, Marjorie Mitchell, and Jane Nelson of Mathematica Policy Research for producing it.

iii CONTENTS

Chapter Page

EXECUTIVE SUMMARY ...... xi

I INTRODUCTION ...... 1

A. PROGRAM BACKGROUND ...... 2

B. THE NATIONAL HEALTHY START EVALUATION ...... 5

II STUDY METHODOLOGY ...... 7

A. SELECTION OF COMPARISON AREAS ...... 7

1. Selection Process ...... 9 2. Comparison Areas Selected ...... 11 3. Comparability of Project and Comparison Areas ...... 11

B. ANALYSIS VARIABLES ...... 14

C. ANALYSIS METHODS ...... 16

1. Data ...... 19 2. Model Specification and Statistical Analysis ...... 20

III EFFECTS OF HEALTHY START ...... 25

A. PRENATAL CARE ADEQUACY ...... 25

1. Any Prenatal Care ...... 26 2. Trimester When Prenatal Care Began ...... 26 3. Number of Prenatal Care Visits ...... 28 4. Kotelchuck Index of Adequacy of Receipt of Prenatal Care and Its Components ...... 28

B. BIRTH OUTCOMES AND INFANT MORTALITY ...... 32

1. Rate ...... 32 2. Low and Very Low Birthweight Rate ...... 34 3. Infant, Neonatal, and Postneonatal Mortality ...... 34

v CONTENTS (continued)

Chapter Page

C. SUMMARY ...... 41

IV SUMMARY AND CONCLUSIONS ...... 43

REFERENCES ...... 47

APPENDIX A: BIRTH OUTCOMES AND INFANT MORTALITY RATES: HEALTHY START PROJECT AREAS AND MATCHED COMPARISON AREAS, 1984 TO 1996 ...... A.1

APPENDIX B: ESTIMATES OF THE EFFECTS OF HEALTHY START ON PRENATAL CARE, BIRTH OUTCOMES, AND INFANT MORTALITY ...... B.1

vi TABLES

Table Page

I.1 CHARACTERISTICS OF THE HEALTHY START PROJECT AREAS ...... 4

II.1 HEALTHY START PROJECT AREAS AND COMPARISON AREAS ...... 12

II.2 DEFINITION OF KEY OUTCOME VARIABLES ...... 17

II.3 INDIVIDUAL CHARACTERISTICS THAT AFFECT INFANT MORTALITY AND RELATED BIRTH OUTCOMES ...... 18

II.4 AVAILABILITY OF BIRTH AND INFANT DEATH FILES ...... 21

III.1 EFFECTS OF HEALTHY START ON PRENATAL CARE INITIATION AND NUMBER OF VISITS ...... 27

III.2 EFFECTS OF HEALTHY START ON PRENATAL CARE ADEQUACY .....31

III.3 EFFECTS OF HEALTHY START ON LOW BIRTHWEIGHT RATE AND VERY LOW BIRTHWEIGHT RATE ...... 37

III.4 EFFECTS OF HEALTHY START ON NEONATAL AND POSTNEONATAL MORTALITY ...... 42

IV.1 SUMMARY OF OUTCOMES ANALYSIS RESULTS ...... 44

vii FIGURES

Figure Page

II.1 COMPARABILITY OF PROJECT AND COMPARISON AREAS: INFANT MORTALITY RATE ...... 13

II.2 COMPARABILITY OF PROJECT AND COMPARISON AREAS: RACE/ETHNIC COMPOSITION ...... 15

III.1 ADEQUACY OF PRENATAL CARE UTILIZATION ...... 30

III.2 PRETERM BIRTH RATE ...... 33

III.3 EFFECTS OF HEALTHY START ON PRETERM BIRTH RATE ...... 35

III.4 LOW BIRTHWEIGHT RATE AND VERY LOW BIRTHWEIGHT RATE .....36

III.5 INFANT MORTALITY RATE ...... 39

III.6 EFFECTS OF HEALTHY START ON INFANT MORTALITY ...... 40

ix EXECUTIVE SUMMARY

The Healthy Start program is a major initiative to reduce infant mortality and improve maternal and infant health status in communities with high infant death rates. In fall 1991, the Health Resources and Services Administration selected 13 urban areas and 2 rural areas as Healthy Start grantees to implement a five-year demonstration of community-based approaches to reducing infant mortality. The demonstration was extended through 1997, and, in fiscal year 1999, 94 projects in total received Healthy Start funding.

This report presents findings on the effects of Healthy Start on prenatal care utilization, key birth outcomes, and infant mortality in the 15 original project areas through 1996--the period of full implementation of the Healthy Start demonstration. The principal findings from the analysis of the effects of Healthy Start on birth outcomes and infant mortality are:

C Prenatal Care Utilization: Healthy Start is associated with significant improvements in many of the outcome measures for prenatal care utilization.

- In 8 of the 15 project areas, Healthy Start is associated with improved adequacy of prenatal care utilization: Baltimore, Birmingham, Chicago, New Orleans, New York City, Northern Plains, Oakland, and Philadelphia.

- Healthy Start is associated with increases in the adequacy of prenatal care initiation in 4 of the 15 project areas: Birmingham, New Orleans, New York City, and Philadelphia.

- Healthy Start is associated with improved adequacy of the number of prenatal care visits in 9 of the 15 project areas: Baltimore, Birmingham, , Chicago, New Orleans, New York City, Northern Plains, Oakland, and Philadelphia.

- Three project areas--New Orleans, New York City, and Philadelphia--show improvements in all measures of prenatal care as a result of Healthy Start.

C Preterm Rate: Four project areas have statistically significant declines in the percentage of born less than 37 weeks gestation: Birmingham, New Orleans, Oakland, and Philadelphia.

C Low Birthweight: Three project areas have statistically significant reductions in the percentage of infants with birthweight less than 2,500 grams: Birmingham, Detroit, and the District of Columbia.

xi C Very Low Birthweight: In Birmingham, Boston, and Pittsburgh, Healthy Start is associated with a statistically significant reduction in the percentage of infants with birthweight less than 1,500 grams.

C Infant Mortality Rate: In two project areas, New Orleans and Pittsburgh, Healthy Start is related to a statistically significant decline in the number of infant deaths per thousand live births. For Birmingham and Oakland, Healthy Start is associated with fairly large, but not quite statistically significant, declines in infant mortality (4.3 and 2.1 infant deaths per 1,000 live births, respectively).

In summary, the Healthy Start program demonstrates improvements in several birth outcomes across the 15 original project areas and reductions in infant mortality in 2 project areas. Healthy Start is particularly successful at linking women and their families to prenatal care, as shown by improvements in the adequacy of prenatal care utilization in 8 of the 15 original project areas.

xii I. INTRODUCTION

Infant death is a tragedy for both families and communities. Fortunately, most infants--more

than 99 percent of them--survive their first year of life. Nonetheless, with a rate of 7.1 infant deaths per 1,000 live births in 1997, the United States ranks 25th in the world, and significant disparities in infant mortality are found across communities and population subgroups (Ventura et al. 1998;

National Center for Health Statistics 1998). Efforts to reduce infant mortality are important not only

for reducing the number of deaths and eliminating these disparities, but also for improving the

quality of life for those infants who survive.

Healthy Start began as a major demonstration program sponsored by the Health Resources and

Services Administration (HRSA) to reduce infant mortality by 50 percent over five years and to

improve maternal and infant health in communities with high infant death rates. In fall 1991, HRSA

chose 13 urban areas and 2 rural areas in which to implement a five-year demonstration of community-based approaches to reduce infant mortality. Since 1991, HRSA has continued to

sponsor similar programs in other high-risk communities, reaching a total of 94 communities in

1999. However, the national evaluation, as well as this report, focuses exclusively on the 15 original

demonstration programs.

Healthy Start is a community-based and community-driven initiative in which local programs have designed and implemented interventions targeted at women, infants, their families, and their

communities. These projects developed approaches to reducing infant mortality, including outreach

methods; case management for pregnant women and infants; broad-based public information

campaigns; support services; individual and classroom-based health education; and enhanced clinical

services (Howell et al. 1997).

1 This report presents findings on the effects of Healthy Start on infant mortality and related birth

outcomes. This introductory chapter provides an overview of the Healthy Start program and project areas and summarizes the design of the impact analysis. Chapter II describes the outcomes measures, the data used, and key aspects of the analytical methodology. Chapter III presents findings from the analysis of the effect of Healthy Start on the receipt of prenatal care, key birth outcomes, and infant mortality, and the final chapter presents the study implications.

A. PROGRAM BACKGROUND

In fall 1991, 15 communities were selected as Healthy Start project areas and received first-year planning grants to design a comprehensive program for reducing infant mortality by 50 percent over five years, along with a plan for implementing the program. To be eligible, project areas had to have high infant mortality rates, as well as the capacity to organize a community-based effort to strengthen the maternal and infant health care system. Specifically, a project area had to have between 50 and

200 infant deaths per year and an infant mortality rate of at least 15.7 deaths per 1,000 live births (50 percent above the national average) for the five-year period 1984-1988.

Healthy Start has developed strong and continued congressional support since it was initially proposed. In fiscal year 1994, seven new “Special Projects” received approximately $1 million each for two years to implement components of the Healthy Start program. In the fiscal year 1997 budget,

Congress also appropriated $96 million to continue the existing Healthy Start projects for a sixth year, one year beyond the original five-year time frame. Congressional support for Healthy Start remains high, with $90 million appropriated for fiscal year 2000 to continue some components of existing programs and fund a total of 94 programs.

The original 15 Healthy Start project areas were Baltimore; Birmingham; Boston; Chicago;

Cleveland; Detroit; the District of Columbia; New Orleans; New York City; Northern Plains Indian

2 reservations; Northwest Indiana; Oakland; the Pee Dee region of South Carolina; Philadelphia; and

Pittsburgh. Twelve of these project areas were inner-city communities. Northwest Indiana was a cluster of four smaller cities within Lake County, Indiana (East Chicago, Gary, Hammond, and Lake

Station). Pee Dee included six rural counties--Chesterfield, Darlington, Dillon, Marion, Marlboro, and Williamsburg. The Northern Plains project area encompassed 19 American Indian tribal organizations in Iowa, Nebraska, North Dakota, and South Dakota.

The project areas differed greatly from one another in terms of their geographic, cultural, and political environment (see Table I.1). Although the 1984 to 1988 period was used as the baseline period for calculating a project area’s infant mortality rate, Table I.1 focuses on the three-year period just prior to Healthy Start (1989 to 1991). During this three-year period, the infant mortality rate for

1989-1991 in Healthy Start communities ranged from 10.7 (Boston) to 24.9 (Detroit). The national rate was 9.3 per 1,000 live births in the same period. With the exception of Northern Plains, the rates of low birthweight were considerably higher than the U.S. average (7 percent).

All project areas had a high proportion of the population in poverty and a high proportion of minority residents. Except for Northern Plains, all had a relatively large population of African

Americans (ranging from 50.9 percent in Northwest Indiana to 97 percent in Baltimore). Five project areas (Boston, Chicago, New York City, Northwest Indiana, and Oakland) had significant proportions that were Hispanic.

Although not shown in the table, the Healthy Start project areas experienced a variety of problems such as high unemployment rates, community and domestic violence, substance abuse, substandard housing, homelessness, and public health problems such as HIV, tuberculosis, and many others. Infant mortality was only one part of a large and complex constellation of social and health problems.

3 TABLE I.1

CHARACTERISTICS OF THE HEALTHY START PROJECT AREAS

Births 1989-1991 Percentage of Infant Percentage Percentage Population Mortality Low African Percentage in Poverty, Project Area Total Rate Birthweight American Hispanic 1989

Baltimore 3,684 15.9 18.1 96.9 0.3 41.6

Birmingham 10,295 19.6 12.8 88.3 0.1 30.2

Boston 18,652 10.7 10.0 52.5 20.1 22.9

Chicago 19,093 19.8 14.0 62.5 26.0 46.4

Cleveland 16,460 17.2 15.0 90.6 1.8 34.4

Detroit 32,430 24.9 16.7 94.6 NA 39.8

District of Columbia 7,633 22.5 16.9 94.5 0.8 24.0

New Orleans 12,272 17.4 15.1 95.0 0.8 48.7

New York 33,670 18.5 14.5 68.7 26.1 36.0

Northern Plains 8,222 17.5 5.7 0.0a 0.0 49.6

Northwest Indiana 13,289 12.0 10.0 50.9 14.6 21.4

Oakland 13,863 12.4 10.7 52.6 23.4 26.3

Pee Dee, SC 11,535 15.8 10.8 57.3 0.5 24.4

Philadelphia 16,960 15.2 14.4 82.1 1.0 24.0

Pittsburgh 11,950 17.4 12.5 62.8 0.5 26.9

United States 12,310,077 9.3 7.0 16.0 14.5 13.1

SOURCE: State linked birth and infant death files. Data for Baltimore, Boston, and the District of Columbia are for 1990 and 1991 only; percentage in poverty is from the 1990 population census. a Northern Plains is 100 percent American Indian.

NA = not available.

4 B. THE NATIONAL HEALTHY START EVALUATION

The national evaluation of Healthy Start includes a detailed process and outcomes analysis.1

The process analysis documents the community context in which the projects operate, planning

processes and implementation, service delivery, and barriers to implementation. Findings from the

process analysis of Healthy Start are presented in a separate project report (Howell et al. 1997).

This report presents findings from the outcomes analysis. The objective of the outcomes analysis is to assess the effects of Healthy Start on adequacy of prenatal care, key birth outcomes, and infant mortality. These effects are obtained by estimating what the outcomes in the Healthy Start project areas would have been in the absence of the Healthy Start program. The analysis is based on a matched comparison-site design in which each Healthy Start project area is carefully matched with two comparison communities to determine whether changes observed in the Healthy Start project areas are attributable to the Healthy Start program or to other factors.

This report on the effects of Healthy Start on infant mortality and birth outcomes is one in a series of evaluation reports on the Healthy Start program. Other completed reports from the national evaluation include:

C “Implementing a Community-Based Initiative: The Early Years of Healthy Start” (Howell et al. 1994)

C “The Implementation of Healthy Start: Lessons for the Future” (Howell et al. 1997)

C “Evaluation of the Fetal and Infant Mortality Review Programs in the Healthy Start Program” (Baltay, McCormick, and Wise 1997)

C “The National Healthy Start Program: Report From a Survey of Post-Partum Women” (McCormick and Deal 1998)

1For additional detail on the design of the national outcome and process evaluation, see Devaney and McCormick (1993) and Raykovich et al. (1996).

5 C “Using Health Education to Combat Infant Mortality: The Healthy Start Experience” (Harrington et al. 1998)

C “Infant Mortality Prevention in an American Indian Community: Northern Plains Healthy Start” (Howell et al. 1999)

C “Case Management in Healthy Start” (Devaney, Foot, and Chu 1999)

C “Reducing Infant Mortality: Lessons Learned from Healthy Start” (Devaney et al. 2000)

In addition, the Health Resources and Services Administration has completed two key reports on community outreach in Healthy Start that informed the analysis in this report (McCann et al.

1996; Simon and Raykovich 1995).

6 II. STUDY METHODOLOGY

Based on a matched comparison area design, the outcomes analysis compares key outcomes in

the Healthy Start project areas with those in carefully matched comparison areas. In this type of design, comparison areas are selected through the use of criteria that ensure that the areas can serve as a proxy for what would have happened in the project areas in the absence of Healthy Start.

This chapter summarizes the process of selecting comparison areas for the 15 Healthy Start

project areas, presents the findings from a baseline analysis of the comparability of the Healthy Start

project areas and their matched comparison areas, and describes the data sources and analysis methods used to estimate the impacts of Healthy Start.

A. SELECTION OF COMPARISON AREAS

The challenge for the Healthy Start outcomes analysis is to separate movements in outcomes due to the program from movements resulting from other factors. Typically, this is accomplished by comparing outcomes for Healthy Start participants with outcomes for some other comparable group that does not receive the program services. The latter group would be similar to the Healthy

Start participants, except that this group would receive no Healthy Start services. The ideal method for conducting such comparisons is an experimental or random assignment procedure. With a random assignment evaluation design, individuals or communities would be selected randomly to receive program services; those not selected would serve as a control group.

In the context of the Healthy Start program, however, a random assignment evaluation was not feasible; neither project areas nor program participants were selected randomly. As discussed above, the Healthy Start project areas were selected on the basis of an application process, and the selection

7 of project areas predated the national evaluation. Because Healthy Start is a community-based

initiative, the methodological approach used in the outcome analysis was a comparison area design.

The matched comparison areas design assumes that observed differences in outcomes between

project areas and comparison areas can be attributed to Healthy Start. For this assumption to be true,

it is critical that the matched comparison areas be as similar as possible to the Healthy Start project

areas prior to implementation of Healthy Start. Ideally, comparison areas would resemble the project

areas in all ways except availability of the Healthy Start program. In particular, because the objective

of Healthy Start is to reduce infant mortality, project areas and their comparison areas should have

similar baseline infant mortality rates.

In addition, the face validity of the matched comparison areas is important. The comparison

areas should match the project areas as closely as possible along demographic characteristics such as number of births, population, degree of urbanization, and racial and ethnic distribution. While

the means and trends for infant mortality during the baseline period might be similar, the results of

the outcomes analysis may be suspect if, according to the other variables, the comparison areas do

not resemble the project area. The selection process relied on Healthy Start staff and local evaluators

to determine the face validity of the match, given their firsthand knowledge of the characteristics of

the Healthy Start project areas and their surrounding areas.

Ideally, the basic structures of health care delivery should be similar in Healthy Start project

areas and comparison areas, yet the comparison areas should have no interventions similar to the

program (such as outreach and coordinated case management). Although several interventions similar to the Healthy Start program were in operation during the baseline period in several potential comparison areas, none of them was of similar magnitude to Healthy Start. Again, collaboration

8 with local Healthy Start staff was fundamental to determine which areas should be discarded because

of the existence of programs similar to Healthy Start.

1. Selection Process

In all but one instance, each Healthy Start project area had two comparison areas, thus

generating a sample of 15 project areas matched to 29 comparison areas.1 The selection strategy

relied primarily on information available on state-linked birth and infant death certificates. This

strategy consisted of constructing a measure of similarity between the Healthy Start project area and

each geographic unit in a city or state considered as a potential comparison area.2 After considerable

exploratory work, the similarity measure included the following variables: (1) race and ethnic

composition, (2) infant mortality rate, and (3) the trend in infant mortality rate during the period

before Healthy Start. The trend coefficient provided the annual change in infant mortality rate

relative to a baseline year (usually 1984). The geographic units varied across the Healthy Start

project areas; depending on data availability, the geographic units were census tracts, zip codes,

counties, census tracts mapped to zip codes, health districts, or planning areas.

In general, the aim of the selection process was to identify comparison areas within the same

city as a Healthy Start project area. In several instances, however, geographic units within the same

city--those having a similar number of births and similar preprogram infant mortality rates and race

and ethnicity distributions--could not be identified. In these cases, the search included other cities

in the same state where the Healthy Start project area was located. If no suitable comparison areas

were found within the state, the search was expanded to include other states.

1The exception is Boston Healthy Start, where only one comparison area from the pool of available areas closely matched the racial and ethnic composition of the project area, as well as the level of infant mortality, in the baseline period.

2Geographic units where the average number of births is less than 10 per year were excluded.

9 For each potential comparison area, the process to select the comparison areas included the following seven steps:

1. Calculate the following percentages and rates for the years 1989-1992:

- Percentage of births to African American women, Hispanic women, and White women - Infant mortality rate

2. If sufficient years of data were available, calculate the trend in the infant mortality rate for the period 1984-1992.

3. For each variable in steps 1 and 2, calculate the absolute value of the difference between each geographic unit and the Healthy Start project area.

4. For each variable--race/ethnicity, infant mortality rate, and infant mortality trend-- compute the rank of the absolute difference between each geographic unit and the Healthy Start area. This rank indicates the relative similarity of each geographic unit to the Healthy Start area. For example, a rank of 3 for infant mortality showed that a geographic unit had the third-lowest difference between its infant mortality rate and the rate for the Healthy Start project area.

5. Compute the weighted sum of the ranks for the three variables--race/ethnicity, infant mortality rate, and infant mortality trend. The weights used in the sum were .60 for race and ethnicity, .20 for the infant mortality rate, and .20 for the infant mortality trend. For Healthy Start project areas where an infant mortality trend was not estimated, the weights used in the sum were .60 for race and ethnicity and .40 for the infant mortality rate.

6. Rank the geographic units by the weighted sum of the individual-variable ranks, thereby ensuring that the geographic units with the lowest rank are most similar to the Healthy Start area.

7. Select two sets of geographic units in order of their overall rank to serve as comparison areas for the Healthy Start area. Geographic units were combined so that the total number of births from 1989-1992, or the applicable time period, was similar to the number of births in the Healthy Start area.

Ideally, selected comparison areas would not be adjacent to the project area, so the validity of the comparison areas would not be threatened by potential spillover effects from the Healthy Start project. Comparison areas, however, were not limited to nonadjacent geographic units, unless local

10 evaluators and project staff indicated a problem of potential spillover effects. In addition,

comparison areas also were not formed by combining contiguous geographic units. Thus,

comparison areas could be constructed by combining geographic units from different parts of the city

or state, or even from out of state. Combining nonadjacent geographic units made it less likely that

the comparison areas would be contaminated by the influence of other maternal and child health

programs in operation. Finally, in some instances, local evaluators or project staff excluded some

geographic areas because of interventions similar to the Healthy Start program area in the

preprogram period or because they lacked face validity.

2. Comparison Areas Selected

Table II.1 reports the final comparison areas for the 15 Healthy Start project areas. In most

instances, the selected comparison areas are all in the same city or state where a project area is located. However, in the case of Baltimore, Birmingham, Boston, the District of Columbia, New

Orleans, and Pittsburgh, either one or both of the comparison areas include geographic areas in neighboring states. Similarly, for Cleveland, Northwest Indiana, and Pittsburgh, the comparison areas also include selected geographic units in Chicago. Geographic units in Chicago serve as components of the comparison areas because, among all the urban areas explored as potential

comparison areas for Cleveland, Northwest Indiana, and Pittsburgh, the geographic units in Chicago

have (1) similar predemonstration infant mortality rates, and (2) a similar racial and ethnic

composition.

3. Comparability of Project and Comparison Areas

The Healthy Start project areas and their corresponding comparison areas are similar in their level of infant mortality before Healthy Start (see Figure II.1). In most instances, the 1989-1991

11 TABLE II.1

HEALTHY START PROJECT AREAS AND COMPARISON AREAS

Site Area Project Area Comparison Area 1 Comparison Area 2

Baltimore 15 census tracts, divided into two target areas Census tracts in Baltimore, MD Census tracts in Baltimore, and Washington, DC; and zip code MD; and zip code area in areas in Philadelphia, PA Philadelphia, PA

Birmingham 58 census tracts, divided into 12 neighborhoods/ Zip code areas in Huntsville, Zip code areas in Bessemer, service areas that encompass part of the City of Jefferson County, Mobile, Jefferson County, Mobile, Birmingham and Jefferson County Montgomery, and and Montgomery, AL Tuscaloosa, AL

Boston 79 census tracts, divided into nine Zip code areas in Springfield, MA naa neighborhoods/service areas and Queens, NY

Chicago Six contiguous neighborhoods, divided into two Community areas in Chicago Community areas in Chicago service areas (West Side and South Side)

Cleveland 16 neighborhoods, divided into nine service areas Geozip areas in Cincinnati, Geozip areas in Cincinnati that encompass part of the City of Cleveland and Columbus, and Toledo, OH; and and Cleveland, OH; and the City of Warrensville Heights community areas in Chicago community areas in Chicago

Detroit One-third of city of Detroit and all of Highland Zip code areas in Detroit Zip code areas in Detroit and Park, divided into three contiguous service areas Flint, MI

District of One service area for two contiguous Census tracts in the District of Zip code areas in Columbia neighborhoods (wards) Columbia Philadelphia, PA

New Orleans 59 contiguous census tracts, divided into 10 Census tracts in New Orleans; zip Census tracts in New service areas on the basis of geographic code areas in Baton Rouge, LA; Orleans; zip code areas in neighborhood boundaries and Mobile and Montgomery, AL Baton Rouge and Shreveport, LA; and Jefferson County and Montgomery, AL

New York City Three service areas, two of which are contiguous Health districts in New York City Health districts in New York City

Northern Plains 19 American Indian reservations in four states Alaska Indian Health Service Area Billings Indian Health (Iowa, Nebraska, North Dakota, and South Service Area Dakota)

Northwest Four service areas: East Chicago, Gary, Community areas in Chicago, IL Community areas in Indiana Hammond, and Lake Station, Indiana Chicago, IL

Oakland 52 census tracts (the “Oakland Flatland”), Zip code areas in Los Angeles, CA Zip code areas in Los divided into three contiguous service areas Angeles and Sacramento, CA

Pee Dee Six contiguous rural counties in the northeast Counties in South Carolina Counties in South Carolina corner of the state

Philadelphia One service area of 26 square miles in West and Zip code areas in Philadelphia, Zip code areas in Southwest Philadelphiab Chester, and Harrisburg, PA Philadelphia

Pittsburgh Six service areas in Pittsburgh and in Allegheny Geozip areas in Pittsburgh, PA; Geozip areas in Pittsburgh, County Cincinnati, Cleveland, and PA; Akron, Cincinnati, Columbus, OH; and community Cleveland, Columbus, and areas in Chicago Toledo, OH; and community areas in Chicago

SOURCES: Healthy Start Project continuation applications; linked birth and infant death certificates.

NOTE: A geozip area consists of a collection of census tracts mapped into a zip code area. aOnly one comparison area was identified for this project area. b An approximated definition of the project area was used because it was not possible to identify the geographic units used by the grantees from the linked birth/infant death files.

.

12 Figure II.1 Comparability of Project and Comparison Areas: Infant Mortality Rate

Infant Deaths per 1,000 Live Births, 1989-1991

Baltimore

Birmingham

Boston

Chicago

Cleveland

Detroit

District of Columbia

New Orleans

New York City

Northern Plains

NW Indiana

Oakland

Pee Dee

Philadelphia

Pittsburgh

0 5 10 15 20 25 30

Project Area Comparison Area 1 Comparison Area 2

Source: State vital statistics birth files, 1989-1991.

13 infant mortality rate for one or both of the site-specific comparison areas is within 10 percent of the

infant mortality rate for the Healthy Start project area. However, for Comparison Areas 1 and 2 in

Birmingham; and Comparison Area 2 in Detroit, New Orleans, Pee Dee, and Philadelphia, the infant

mortality rate differs by more than 10 percent from the rate for the Healthy Start project areas. When

applicable, the project areas and their comparison areas are also well matched on the trend in the

infant mortality rate during the baseline period (data not shown).

In addition, Healthy Start project areas and their matched comparison sites match closely in

terms of their racial and ethnic composition (see Figure II.2). For all Healthy Start project areas and

their comparison areas, the percentage of births to African American women is within 10 percentage

points, except for Comparison Area 2 in Birmingham, Cleveland, Detroit, and New York City. For

women of Hispanic origin, the match is also very good for project areas with large percentages of

Hispanic women--Chicago, New York City, Northwest Indiana, and Oakland.3

B. ANALYSIS VARIABLES

Analysis variables include outcome variables, individual control variables, time trends, and a

measure of the Healthy Start intervention. The outcome variables used in the analysis are those most

closely related to the central Healthy Start objectives of reducing infant mortality by 50 percent and

improving birth outcomes. Three main types of outcome variables are considered:

C Prenatal Care: Whether any prenatal care was received; whether prenatal care was initiated in the first trimester of pregnancy; average number of prenatal care visits; adequacy of prenatal care utilization and its components of initiation and receipt of services, as measured by the Kotelchuck index of adequacy of receipt of prenatal care (Kotelchuck 1994)

3Northern Plains Healthy Start is not shown in Figure II.2, because all residents in the project and comparison areas are classified as American Indian.

14 Figure II.2 Comparability of Project and Comparison Areas: Race/Ethnic Composition

Percentage of Women Giving Birth, 1989-1991

Baltimore

CA1

CA2

Birmingham CA1

CA2

Boston CA1

Chicago

CA1

CA2 Cleveland

CA1

CA2

Detroit CA1

CA2

DC

CA1 CA2

New Orleans

CA1 CA2

New York

CA1

CA2 NW Indiana

CA1

CA2

Oakland CA1

CA2

Pee Dee

CA1 CA2

Philadelphia

CA1 CA2

Pittsburgh

CA1

CA2

0 20 40 60 80 100 African American Hispanic

Source: State vital statistics birth and death files, 1989-1991 15 C Birth Outcomes: Rates of preterm delivery, low birthweight, and very low birthweight

C Infant Mortality: Infant death rates within one year after birth (infant mortality rate), before 28 days after birth (neonatal mortality rate), and between 28 days and one year after birth (postneonatal mortality rate)

Table II.2 defines each of these outcome variables.

Individual control variables measure individual characteristics that are expected to affect infant

mortality and related birth outcomes. These individual characteristics come from the birth files and

include two groups: (1) newborn characteristics, and (2) maternal characteristics (Table II.3).

Newborn characteristics are measured by the sex of the child and the plurality of the birth. Maternal

characteristics include age of the mother, race and ethnicity of the mother, marital status, education

of the mother, whether the mother has experienced previous deaths of live-born children, whether

the mother has experienced at least one previous live birth or pregnancy termination, and adequacy of prenatal care utilization (Kotelchuck 1994).1

C. ANALYSIS METHODS

The Healthy Start outcomes analysis has two parts: (1) a descriptive analysis that compares measures of prenatal care, birth outcomes, and infant mortality rates over time for the Healthy Start

project areas and their matched comparison areas; and (2) a multivariate analysis of the effects of the Healthy Start programs. The multivariate analysis controls for differences in maternal and infant

characteristics that might exist between the Healthy Start project areas and their matched comparison

areas.

1The Kotelchuck index is used as both an outcome variable and a control variable in the analysis. It is not used as a control variable in the analysis of prenatal care adequacy derived from the Kotelchuck indexes.

16 TABLE II.2

DEFINITION OF KEY OUTCOME VARIABLES

Variable Definition

Adequacy of Prenatal Care

Whether Received Any Prenatal Care The percentage of women who received any prenatal care

Whether Prenatal Care Was Initiated in the First The percentage of women who began receiving prenatal Trimester of Pregnancy care in the first trimester of pregnancy

Average Number of Prenatal Care Visits The average number of prenatal care visits reported in the birth certificate

Receipt of Adequate or Better Prenatal Care The percentage of women who received adequate or adequate plus prenatal care, as defined by the Kotelchuck index

Initiation of Adequate or Better Prenatal Care The percentage of women who initiated prenatal care by the fourth month of pregnancy

Receipt of Adequate or Better Prenatal Care Visits The percentage of women who received at least 80 percent of visits recommended by the American College of Obstetricians and Gynecologists from the time prenatal care began until delivery

Birth Outcomes

Preterm Birth Rate The percentage of live births born at less than 37 weeks gestation

Low Birthweight Rate The percentage of live births under 2,500 grams at birth

Very Low Birthweight Rate The percentage of live births under 1,500 grams at birth

Infant Mortality

Infant Mortality Rate The number of deaths of infants under one year of age per 1,000 live births

Neonatal Mortality Rate The number of deaths of infants under 28 days of age per 1,000 live births

Postneonatal Mortality Rate The number of deaths of infants between 28 days and one year of age per 1,000 live births

17 TABLE II.3

INDIVIDUAL CHARACTERISTICS THAT AFFECT INFANT MORTALITY AND RELATED BIRTH OUTCOMES

Characteristic Definition

Newborn Characteristics

Sex Male Female

Multiple Birth Singleton Twin or higher

Maternal Characteristics

Age at the Birth of the Child Less than 18 years of age 18-19 years of age 20-34 years of age 35 years and older

Race/Ethnicity White, not Hispanic African American, not Hispanic Asian/Pacific Islander Hispanic Other race/ethnicity

Marital Status Not married Married

Years of Education Less than 9 years 9-11 years 12 years More than 12 years Missing

Previous Death of Live Born Infants Yes No

Previous Pregnancies No previous pregnancy One or more previous pregnancies

Adequacy of Prenatal Care Inadequate Intermediate Adequate Adequate Plus Missing

18 1. Data

The basic source of data for the Healthy Start outcomes analysis is linked files of birth and

infant death records. Birth and death files are maintained by state vital registrars and are based on

national standard certificates of birth and death. These certificates represent the standard data set

for collecting and publishing comparable state and local vital statistics data; they facilitate uniformity

in the information on which local birth outcome and mortality statistics are based. In particular, the

birth certificate contains a consistent set of information on place of birth, maternal and paternal

characteristics, the timing and extent of prenatal care received, pregnancy history, basic demographic

characteristics, and birth outcomes.

States create the linked birth and infant death files by linking all deaths of infants under age one

to their birth records. This file forms the basis for determining cohort-based infant mortality rates

and infant mortality rates for population and birthweight subgroups. An alternative data source to

the linked birth and infant death files is unlinked birth and infant death files. Unlinked birth and

death files are the basis of period-based infant mortality rates, which are the number of deaths to

infants under age one in a given year per 1,000 live births in that same year.

Linked birth and infant death files are preferable to separate birth and infant death records, for

several reasons. First, linked birth and infant death records support an individual-level analysis of

infant mortality controlling for the characteristics of the mother or father of the infant under consideration. Second, they allow for comparison of infant mortality among birthweight groups.

Finally, linked files allow for analysis of the experience of a birth cohort, an approach that implicitly

controls for the confounding that results when combining the experience of infants born in different

periods.

19 Beginning in summer 1994 and continuing through spring 1999, the evaluation team submitted

annual requests for linked birth and death records to vital registrars and/or state departments of

health for the years 1984 through 1996 (eight pre-Healthy Start years and five Healthy Start years).

The data request included both predemonstration and demonstration years, so that trends in infant

mortality and other key outcomes in the project and comparison areas could be compared for the

baseline period. Using statistical models, such differences can be netted out of the differences observed between the project and comparison areas during the program period.

For 12 of the 15 Healthy Start project areas, birth and infant death records are available for at least six predemonstration years and five program years; for three project areas (Birmingham,

Boston, and the District of Columbia), vital records files are available on a more limited basis (see

Table II.4). In addition, for Baltimore and Northern Plains, only unlinked birth and infant death

records for the period 1985-1996 and 1984-1996, respectively, are available.5

2. Model Specification and Statistical Analysis

The matched comparison site design for the outcomes analysis entails using vital records files

of births and infant deaths to estimate the effects of Healthy Start on infant mortality and related birth

outcomes. The model specification recognizes that both individual characteristics, such as the age and education of the mother, and Healthy Start may affect infant mortality and birth outcomes. The model also recognizes that the effect of Healthy Start may vary over time, reflecting the gradual implementation of Healthy Start. That is, 1992 is considered the first year of the Healthy Start

5Because linked birth and infant death files are not available for Baltimore and Northern Plains, the outcomes analysis of infant mortality is based only on aggregate counts of infant deaths and births. In addition, for Baltimore, the 1989 birth file does not have associated documentation and, therefore, could not be used in the outcomes analysis, and for Northwest Indiana, the 1993 linked birth infant death file was not available for this analysis.

20 TABLE II.4

AVAILABILITY OF BIRTH AND INFANT DEATH FILES

Project Area Type of File and Period Source

Baltimore 1985 to 1996 unlinked birth and death files Baltimore City Department of Health

Birmingham 1988 to 1996 linked birth and infant death files Alabama Department of Public Health

Boston 1988 to 1996 linked birth and infant death files Massachusetts Department of Health and Hospitals

Chicago 1984 to 1996 linked birth and infant death files Illinois Department of Public Health

Cleveland 1984 to 1996 linked birth and infant death files Ohio Department of Health

Detroit 1984 to 1996 linked birth and infant death files Michigan Department of Health

District of 1990 to 1996 linked birth and infant death files State Center for Health Statistics Columbia

New Orleans 1989 to 1996 linked birth and infant death files Louisiana Department of Health and Hospitals

New York 1985 to 1996 linked birth and infant death files New York City Department of Health

Northern Plains 1984 to 1996 unlinked birth and death files Indian Health Service (IHS)

Northwest Indiana 1985 to 1992 and 1994 to 1996 linked birth and Indiana State Department of Health death files

Oakland 1984 to 1996 linked birth and infant death files California Department of Health Services

Pee Dee 1984 to 1996 linked birth and infant death files South Carolina Department of Health and Environmental Control

Philadelphia 1986 to 1996 linked birth and infant death files Pennsylvania Department of Health

Pittsburgh 1984 to 1996 linked birth and infant death files Pennsylvania Department of Health

21 program, but implementation in all the Healthy Start project areas was gradual and not complete until

1994 or 1995.

To illustrate this model, let Yi denote an outcome for individual i, t denote a continuous variable that measures the time trend of the outcome relative to 1984 (t = 1 for 1984, 2 for 1985, and so

on); hsi denote a binary variable indicating that the woman resides in the project area; t1 denote a binary variable indicating the Healthy Start period (1992 or later) that takes values 1 for 1992, 2 for

1993, and so on; t1hsi denote the interaction between t1 and hsi; and Xi denote a vector of demographic and socioeconomic characteristics for individual i that are also assumed to influence infant mortality or other outcomes:

' % % % % % % (1) Yi á â1t â2hsi â3t1 â4t1hsi â5Xi åi .

In this specification, the Healthy Start intervention is measured by the variable t1hs, which measures the number of postimplementation years for women living in the Healthy Start project areas. This intervention variable equals 0 for all women prior to 1992; 1 for women living in the

Healthy Start project areas during 1992, 2 for Healthy Start residents in 1993; and 3, 4, and 5 for

Healthy Start residents in 1994, 1995, and 1996. This variable equals 0 for women living in

comparison areas, regardless of the year. Thus, the coefficient of t1hsi (that is, â4) is the annual change in the outcome variable attributed to Healthy Start. In addition, this specification controls for any preexisting or structural differences between the project and comparison areas by including

the dummy variable hsi. The variables Xi control for the observed demographic and socioeconomic characteristics of women and their infants, and the time trend (t) controls for exogenous trends in

the outcome variable that are not related to Healthy Start. Finally, åi is a random error term. The

22 model is fitted separately for each of the project areas. One-tailed tests of statistical significance are used to determine if Healthy Start reduces infant mortality and improves birth outcomes.

Because the analysis is based on individual-level records, the outcome variable, Y, is binary for most of the outcomes under consideration (for example, whether an infant died). Therefore, the model used is a multivariate logit, estimated by a maximum-likelihood estimation procedure appropriate for models for limited-dependent variables.

Estimates with Healthy Start and for the Counterfactual. As discussed above, the coefficients of the model specified in equation (1) are derived from multivariate logit models.

Because of the nonlinear multivariate logits, estimated coefficients from logits are not easily interpreted. The estimated coefficients, however, can be transformed to facilitate the interpretation of results. Specifically, the analysis results shown in the next chapter present regression-adjusted differences in predicted values of the outcome variables under two scenarios: (1) with the Healthy

Start program, and (2) without the Healthy Start program. The predicted values are based on the estimated coefficients from the multivariate logit or linear regression models. The statistical significance of the difference of the predicted outcomes corresponds to the statistical significance of the coefficient t1hs in equation (1). Although all years of data are used to estimate the effects of

Healthy Start on infant mortality and other outcomes, a specific year must be selected to generate the predicted values.6 Because of the overall goal to reduce infant mortality by the end of a five-year demonstration period, 1996 is the year used in generating the regression-adjusted predicted outcome values. In addition, given the model specification (see equation [1]), 1996 is the year that would be expected to show the effects of full Healthy Start implementation.

6That is, since equation (1) includes time trends t and t1, a specific value for these variables must be selected to generate predicted values. Nonetheless, as mentioned in the text, all years of data are used to estimate the model coefficients.

23 III. EFFECTS OF HEALTHY START

This chapter summarizes findings from an analysis of the effect of Healthy Start on adequacy

of prenatal care utilization, birth outcomes, and infant mortality. The basic strategy is to compare

infant mortality rates and other birth outcomes over time in the project areas with those in carefully

selected comparison areas. This comparison provides estimates of the effects of Healthy Start on

all women and infants residing in Healthy Start communities.

Section A discusses the findings on the adequacy of prenatal care, while Section B presents findings for infant mortality and related birth outcomes. For all Healthy Start project areas, the

analysis results are based on data from 1984 through 1996. To complement the analysis results presented below, Appendix A of this report provides descriptive tabulations of several key outcomes for each project area and its comparison areas for the four periods under consideration in this analysis: 1984-1988, the baseline years of the program; 1989-1991, the pre-Healthy Start years;

1992-1994, the period of early implementation; and 1995-1996, the years of full implementation of the program. Appendix B presents detailed estimates of the effects on the adequacy of prenatal care, birth outcomes, and infant mortality for each of the project areas.

A. PRENATAL CARE ADEQUACY

During pregnancy, it is expected that women will start their care early and receive a recommended number of visits. In addition, it is hoped that adequate prenatal care will translate into well-child care and age-appropriate . Since an important component of Healthy Start is to link low-income women and their families to needed services, especially during pregnancy, prenatal care utilization is an important outcome to examine.

25 The analysis of the relationship between Healthy Start and adequacy of prenatal care includes five measures of prenatal care utilization. All these measures reflect either the timing or the intensity

of prenatal care, or both. The Kotelchuck Index of Adequacy of Receipt of Prenatal Care

(Kotelchuck 1994) is the most comprehensive of the prenatal care measures used in the analysis.

The analysis also includes several of the components of this index: whether women received any prenatal care, whether prenatal care began in the first trimester of pregnancy, and the number of prenatal care visits.

1. Any Prenatal Care

The broadest measure of access to prenatal care is whether women received any prenatal care during pregnancy, regardless of its timing or quantity. Although the measure is crude, nevertheless it provides a summary of the extent to which prenatal care is utilized in a community.

As shown in Table III.1, the vast majority of pregnant women--more than 90 percent of them-- receive some prenatal care. In five project areas, Cleveland, New Orleans, New York City,

Philadelphia, and Pittsburgh, Healthy Start is related to higher percentages of women receiving any prenatal care. In New Orleans, the difference in the percentage of women receiving any care with and without the Healthy Start program is almost six percentage points, a large difference considering the high percentages of women receiving any care at all.

2. Trimester When Prenatal Care Began

Another common measure of utilization of prenatal care services is the percentage of women starting care in the first trimester of pregnancy. This measure indicates whether women began to receive care on a timely basis. Across the Healthy Start project areas, the percentage of women

26 TABLE III.1

EFFECTS OF HEALTHY START ON PRENATAL CARE INITIATION AND NUMBER OF VISITS

Percentage of Women Percentage of Women Receiving Any Prenatal Receiving Prenatal Care in Average Number of Care, 1996 the First Trimester, 1996 Prenatal Care Visits, 1996

With Without With Without With Without Healthy Healthy Healthy Healthy Healthy Healthy Project Area Start Start Start Start Start Start

Baltimore 95.6 95.2 70.1 71.8 10.0** 9.5

Birmingham 98.2 98.5 74.8 76.0 11.3 11.3

Boston 99.3 99.5 82.7 81.7 11.6 11.4

Chicago 95.4 95.1 71.0 70.7 9.9 9.9

Cleveland 96.6** 92.1 68.7 69.7 9.3 10.4

Detroit 95.4 95.3 68.0 69.8 10.6 10.9

District of Columbia 91.2 94.7 54.5 60.4 8.5 9.5

New Orleans 96.4** 90.8 70.7** 67.8 11.2** 10.3

New York City 96.2** 95.4 53.0** 50.0 9.0** 8.6

Northern Plains 97.4 97.2 67.8 67.7 8.9** 8.2

Northwest Indiana 97.0 97.6 71.1 72.8 11.0 11.3

Oakland 99.0 99.4 84.2 84.4 13.2 13.5

Pee Dee 98.0 98.5 70.3 73.3 10.7 11.2

Philadelphia 96.6** 95.4 66.9* 65.6 9.7** 9.2

Pittsburgh 97.4** 95.1 79.4 78.2 10.6 10.8

SOURCE: State vital statistics birth and death files, 1984-1986.

NA = not available.

*(**): Significantly different from estimate without Healthy Start at the .05 (0.01) level, one-tailed test.

27 receiving prenatal care in the first trimester in 1996 ranged from 54.5 percent in the District of

Columbia to 84.2 percent in Oakland (see Table III.1, middle of column). In three project areas,

Healthy Start is associated with significant increases in the percentage of women receiving first- trimester care: New Orleans, New York, and Philadelphia.

3. Number of Prenatal Care Visits

A third broad measure of prenatal care utilization is the reported number of prenatal care visits received. This measure, although readily available from the birth certificate, is difficult to compare across subgroups, because the number of visits received varies with the duration of pregnancy, as well as with the timing of when prenatal care services were first received and the health status of women.

For the Healthy Start project areas, the average number of prenatal care visits ranges from 8.5 in the District of Columbia to 13.2 in Oakland during 1996. In five project areas (Baltimore, New

Orleans, New York, Northern Plains, and Philadelphia), Healthy Start is associated with a statistically significant increase in the average number of prenatal care visits (Table III.1).

4. Kotelchuck Index of Adequacy of Receipt of Prenatal Care and Its Components

The Kotelchuck Index combines information on the start of prenatal care, the number of visits received, and the duration of pregnancy into a measure of the adequacy of prenatal care (Kotelchuck

1994). The index classifies prenatal care receipt into five classifications: no care, inadequate care, intermediate care, adequate care, and adequate plus care. The classification depends on the start of care (first trimester or later) and the number of recommended visits that were received, where the number of recommended visits depends on the duration of pregnancy, as specified by standards issued by the American Academy of Obstetricians and Gynecologists. Although the index does not

28 focus on the content of prenatal care, it is regarded as the most comprehensive measure of prenatal care receipt available from vital records.

Nationwide, the percentage of women who received adequate or better prenatal care increased from about 66 percent in 1984-1988 to 73 percent in 1995 (Kogan et al. 1998). Across all 15 project areas, the percentage of women who received adequate or better prenatal care increased from 51.6 percent in 1984-1988 to 59.8 percent in 1995-1996 (see Figure III.1).

In 8 of the 15 project areas, Healthy Start is associated with significant improvements in the adequacy of prenatal care (Table III.2). The estimated effects of Healthy Start are especially striking in Birmingham, Northern Plains, and Philadelphia, where the program is associated with an increase of 4.0 to 6.4 percentage points in the number of women receiving adequate or adequate plus prenatal care.

The Kotelchuck Index includes two components: (1) a subindex of initiation of prenatal care, and (2) a subindex of receipt of prenatal care visits. These two measures provide some indication of whether the improvements in adequacy of prenatal care discussed above result from better timing of receipt of care, from a larger quantity of services received, or from both. These two measures might also be linked to different components of the Healthy Start programs--outreach (prenatal care initiation) and case management (receipt of prenatal care services).

In general, Healthy Start is more strongly associated with the adequacy of receipt of prenatal care services than with the adequacy of prenatal care initiation. The following specific findings support this conclusion:

C In 9 of the 15 project areas, Healthy Start is associated with an increase in the percentage of women receiving an adequate or better number of prenatal care visits (Table III.2).

29 Figure III.1 Adequacy of Prenatal Care Utilization

Percentage of women receiving adequate or better prenatal care 80

1984-1988 1989-1991 1992-1994 1995-1996

59.8 58.1 60 56.3 57.6 51.6 52.5 51.9 49.8 51.6 49.2 48.0 46.9

40

20

0 All 15 Project Areas Comparison Areas 1 Comparison Areas 2

Source: State vital statistics birth files, 1984-1996.

30 TABLE III.2

EFFECTS OF HEALTHY START ON PRENATAL CARE ADEQUACY (Percentage of Women)

Adequate or Better Adequate or Better Prenatal Care Adequate Initiation of Number of Prenatal Care (Kotelchuck Index), 1996 Prenatal Care, 1996 Visits, 1996

With Without With Without With Without Healthy Healthy Healthy Healthy Healthy Healthy Project Area Start Start Start Start Start Start

Baltimore 60.4* 56.4 81.0 81.4 72.2* 68.9

Birmingham 74.2** 67.8 86.7* 85.4 83.7** 78.1

Boston 75.6 76.1 90.6 91.3 83.1* 81.2

Chicago 57.8** 56.2 81.6 81.3 66.8* 65.7

Cleveland 49.4 62.3 79.6 79.6 59.4 72.2

Detroit 62.1 67.6 79.3 80.2 75.4 80.8

District of Columbia 46.1 51.0 67.2 71.3 60.9 67.2

New Orleans 65.5** 62.3 82.3** 78.6 76.0** 72.8

New York City 47.2** 44.8 69.5** 67.8 65.5** 60.4

Northern Plains 45.8** 41.8 78.3 77.4 56.7** 51.7

Northwest Indiana 62.6 69.0 81.4 82.7 73.9 82.5

Oakland 83.0** 80.7 91.1 91.5 90.9** 88.6

Pee Dee 67.0 69.0 79.7 79.7 75.8 77.8

Philadelphia 57.8** 52.3 78.8** 75.8 70.7** 64.8

Pittsburgh 70.6 72.2 86.7 86.1 78.9 79.5

SOURCE: State vital statistics birth and death files, 1984-1986.

NA = not available.

*(**): Significantly different from estimate without Healthy Start at the .05 (0.01) level, one-tailed test.

31 C Improvements are particularly striking in Oakland Healthy Start, where more than 90 percent of women receive an adequate or better number of prenatal care visits.

C In 4 of the 15 project areas, Healthy Start is associated with an increase in the adequacy of prenatal care initiation.

In summary, the prenatal care analysis results document positive effects of Healthy Start on the adequacy of prenatal care utilization in more than half the project areas, which suggests that the program is successful at improving access to care.

B. BIRTH OUTCOMES AND INFANT MORTALITY

The focus of the Healthy Start program is reducing infant mortality and improving birth outcomes. Through its major interventions of case management and support services, Healthy Start attempts to link women to available services, with the ultimate goal of improving birth outcomes and reducing infant mortality. The analysis below focuses on the impacts on birth outcomes and infant mortality.

1. Preterm Birth Rate

Preterm birth is a major cause of infant mortality and morbidity. Infants born at gestational ages under 37 weeks are more likely to die in the neonatal period than term infants (Ventura et al. 1999).

Preterm newborns who survive are at greater risk of neurodevelopmental and respiratory disorders, as well as other problems (Berkowitz and Papiernik 1993).

For more than a decade, the rate of preterm birth has risen in the nation, from 9.9 percent in

1984-1988 to 11.0 percent in 1995-1996 (see Figure III.2). In the Healthy Start project areas, the preterm birth rate declined slightly across all the project areas between the periods 1984-1988 and

1995-1996, with slight increases in the 1989-1991 and 1992-1994 periods. Between 1984-1988 and

1995-1996, the preterm birth rate increased by about 0.7 percentage points in the comparison areas

32 Figure III.2 Preterm Birth Rate

Percentage of births less than 37 weeks gestation 25

1984-1988 1989-1991 1992-1994 1995-1996 20 18.0 17.4 17.4 16.6 16.7 16.6 16.5 16.4 16.4 16.3 15.6 15.8 15

10.7 10.9 11.0 10 9.9

5

0 United States All 15 Project Areas Comparison Areas 1 Comparison Areas 2

Source: State vital statistics birth files, 1984-1996.

33 The Healthy Start program is associated with statistically significant declines in the preterm

birth rate in 4 of the 15 project areas (see Figure III.3): Birmingham, New Orleans, Oakland, and

Philadelphia. In Birmingham and Philadelphia, the difference in the estimates with and without

Healthy Start is about three percentage points.

2. Low and Very Low Birthweight Rate

Low birthweight is a major determinant of infant mortality, especially in the neonatal period.

As a result, weight at birth has traditionally been used to measure the risk of infant mortality

attributable to causes ranging from simple prematurity to poor maternal health (Tompkins et al.

1985).

Nationwide, as shown in Figure III.4, low birthweight rates increased from 6.8 percent in 1984-

1988 to 7.4 percent in 1995-1996; the very low birthweight rate also increased from 1.2 percent in

1984-1988 to 1.4 percent in 1995-1996 (Ventura et al. 1999). These national trends in the rates of low and very low birthweight parallel the national trend in the preterm birth rate.

In three project areas (Birmingham, Detroit, and the District of Columbia), Healthy Start is associated with statistically significant lower rates of low birthweight (Table III.3). In the case of very low birthweight, three project areas (Birmingham, Boston, and Pittsburgh) show a statistically significant lower rate attributed to Healthy Start.

3. Infant, Neonatal, and Postneonatal Mortality

Infant mortality rates declined substantially between 1984-1988 and 1995-1996 in the Healthy

Start project areas. Across all project areas, infant mortality declined from 19.5 infant deaths per

1,000 live births in the baseline period of 1984-1988 to 15.8 in the early implementation period of

34 Figure III.3 Effects of Healthy Start on Preterm Birth Rate

Percentage of births less than 37 weeks gestation, 1996

22.4 Baltimore 22.5

17.6** Birmingham 20.5

19.6 Boston 19.4

Chicago 18.1 18.3

Cleveland 17.6 16.2

Detroit 18.4 18.6

20.9 DC 20.5

19.4* New Orleans 21.3

14.5 New York City 14.5

Northern Plains 12.2 12.6

NW Indiana 14.2 15.0

11.3** Oakland 12.6

14.5 Pee Dee 13.8

16.7** Philadelphia 18.1

15.5 Pittsburgh 14.2

With Healthy Start Without Healthy Start

Source: State vital statistics birth and death files, 1984-1996 Note: All rates are regression-adjusted. *(**): Significantly different from the estimate without Healthy Start at the .05(.01) level, one-tailed test. 35 Figure III.4 Low Birthweight Rate and Very Low Birthweight Rate

Percentage of births less than 2,500 grams 20

1984-1988 1989-1991 1992-1994 1995-1996

15 13.5 13.2 12.6 12.6 12.5 12.1 12.4 12.2 11.5 11.9 11.7 10.3 10

7.4 6.8 7.0 7.2

5

0 United States All 15 Project Areas Comparison Areas 1 Comparison Areas 2

Very Low Birthweight Rate

Percentage of births less than 1,500 grams 4

1984-1988 1989-1991 1992-1994 1995-1996

3.0 3 2.9 2.8 2.7 2.7 2.7 2.6 2.6 2.6 2.6 2.5 2.4

2

1.4 1.3 1.3 1.2 1

0 United States All 15 Project Areas Comparison Areas 1 Comparison Areas 2

Source: State vital statistics birth files, 1984-1996. 36 TABLE III.3

EFFECTS OF HEALTHY START ON LOW BIRTHWEIGHT RATE AND VERY LOW BIRTHWEIGHT RATE

Low Birthweight Rate, 1996 Very Low Birthweight Rate, 1996 (Percentage of Births) (Percentage of Births)

With Healthy Without Healthy With Healthy Without Healthy Project Area Start Start Start Start

Baltimore 17.8 17.6 3.5 3.8

Birmingham 12.3** 14.9 3.2** 4.6

Boston 11.8 12.5 2.9* 3.6

Chicago 13.0 13.1 2.4 2.5

Cleveland 14.5 12.3 3.4 2.1

Detroit 14.2** 15.8 3.0 3.4

District of Columbia 16.2* 17.9 4.4 4.6

New Orleans 14.2 15.1 3.1 3.4

New York City 11.6 11.6 2.5 2.6

Northern Plains 6.3 6.5 1.4 1.1

Northwest Indiana 10.4 9.5 2.1 1.9

Oakland 9.4 9.2 1.5 1.5

Pee Dee 11.7 11.2 2.4 2.2

Philadelphia 13.5 14.0 3.1 2.9

Pittsburgh 11.9 11.9 2.3* 2.8

SOURCE: State vital statistics birth and death files, 1984-1986.

NA = not available.

*(**): Significantly different from estimate without Healthy Start at the .05 (0.01) level, one-tailed test.

37 the program, and to 13.6 in the full implementation period of 1995-1996 (Figure III.5). Infant

mortality also declined in the comparison areas and nationwide. The magnitude of the decline in

infant mortality in the comparison areas was similar when comparing the pre-Healthy Start period

(1989-1991) with the full implementation period of 1995-1996, and the overall decline nationwide

was similar to the decline in the Healthy Start project areas.

In nine project areas, the estimated effect of Healthy Start is to lower infant mortality rates,

although these estimates are not statistically significant (see Figure III.6). In two project areas, New

Orleans and Pittsburgh, Healthy Start is related to statistically significant reductions in infant

mortality. In these two project areas, the difference in the infant mortality rate with and without the

Healthy Start program is between seven and eight infant deaths per thousand live births, a dramatic

difference considering the much smaller reductions in infant mortality nationwide during the same

period. Moreover, in Birmingham and Oakland, although the differences in the estimates with and

without Healthy Start are large, they are not quite statistically significant (p value = .06 and .07

respectively).

Neonatal and Postneonatal Mortality Rates. The overall infant mortality rate often is divided

into two categories based on time of death: neonatal deaths (deaths between 0 and 27 days after

birth) and postneonatal deaths (deaths between 28 days and one year after birth). This division of

the overall infant mortality rate corresponds largely to differences in causes of death. Neonatal

deaths generally result from events in pregnancy and delivery, such as inadequate fetal growth or altered fetal growth from congenital malformations. Postneonatal deaths are more often related to factors in the infant’s environment, such as pneumonia, gastroenteritis, injury, and the ill-defined

sudden infant death syndrome (Shapiro et al. 1968).

38 Figure III.5 Infant Mortality Rate

Infant deaths per 1,000 live births 25

1984-1988 1989-1991 1992-1994 1995-1996 20 19.5

17.8 17.5 17.0 16.9 16.5 15.8 15 14.2 13.6 13.9 13.5 12.5

10.4 10 9.3 8.3 7.5

5

0 United States All 15 Project Areas Comparison Areas 1 Comparison Areas 2

Source: State vital statistics birth files, 1984-1996.

39 Figure III.6 Effects of Healthy Start on Infant Mortality

Infant deaths per 1,000 live births, 1996

14.0 Baltimore 18.2

14.6 Birmingham 18.9

11.5 Boston 14.0

Chicago 13.3 12.7

Cleveland 17.0 11.7

Detroit 16.6 16.8

17.6 DC 18.8

11.3** New Orleans 18.3

10.0 New York City 11.1

Northern Plains 12.6 10.1

NW Indiana 14.0 8.9

7.2 Oakland 9.3

11.1 Pee Dee 11.6

11.4 Philadelphia 11.6

8.6** Pittsburgh 17.5

With Healthy Start Without Healthy Start

Source: State vital statistics birth and death files, 1984-1996 Note: All rates are regression-adjusted. *(**): Significantly different from the estimate without Healthy Start at the .05(.01) level, one-tailed test.

40 In two project areas (New Orleans and Pittsburgh), Healthy Start is associated with significant

reductions in neonatal mortality (Table III.4). The magnitude of the estimated reductionSabout 50 percentSis large and mirrors the effects on overall infant mortality. In Oakland and Pittsburgh,

Healthy Start also is associated with declines in postneonatal mortality.

C. SUMMARY

An overview of the findings presented in this chapter follows:

C Adequacy of Prenatal Care: In 8 of the 15 project areas, Healthy Start is associated with improvement in the adequacy of prenatal care utilization. Three project areas (New Orleans, New York, and Philadelphia) show improvements in all measures of prenatal care utilization.

C Preterm Rate: Four of the 15 project areas show a statistically significant decline in the percentage of infants born at less than 37 weeks gestation. The four are Birmingham, New Orleans, Oakland, and Philadelphia.

C Low Birthweight: In 3 of the 15 project areas, the difference in the percentage of infants with birthweight less than 2,500 grams with and without Healthy Start was statistically significant. The three are Birmingham, Detroit, and the District of Columbia.

C Very Low Birthweight: In three project areas (Birmingham, Boston, and Pittsburgh), Healthy Start is associated with a statistically significant reduction in the percentage of infants with birthweight less than 1,500 grams.

C Infant Mortality Rate: In two project areas (New Orleans and Pittsburgh), Healthy Start is associated with a significant reduction in infant mortality. For Birmingham and Oakland Healthy Start, the difference in the infant mortality rate with and without Healthy Start is almost significant.

41 TABLE III.4

EFFECTS OF HEALTHY START ON NEONATAL AND POSTNEONATAL MORTALITY

Neonatal Mortality Rate, 1996 Postneonatal Mortality Rate, 1996

With Healthy Without With Healthy Without Project Area Start Healthy Start Start Healthy Start

Baltimore 7.8 10.4 6.3 7.8

Birmingham 9.9 13.5 4.7 5.4

Boston 8.4 11.2 3.1 3.1

Chicago 7.7 7.2 5.6 5.6

Cleveland 11.6 5.8 5.4 6.3

Detroit 9.9 10.6 6.7 6.2

District of Columbia 12.4 12.4 5.5 6.8

New Orleans 5.4** 10.1 5.9 8.2

New York City 7.3 6.0 3.8 3.8

Northern Plains 6.7 5.5 6.0 4.6

Northwest Indiana 9.1 4.9 5.0 4.0

Oakland 4.1 4.1 3.1* 5.6

Pee Dee 7.3 7.5 3.8 3.9

Philadelphia 7.7 7.0 3.7 4.6

Pittsburgh 5.7** 12.6 2.9** 4.9

SOURCE: State vital statistics birth and death files, 1984-1986.

*(**): Significantly different from estimate without Healthy Start at the .05 (.01) level, one-tailed test.

42 IV. SUMMARY AND CONCLUSIONS

The outcomes analysis of Healthy Start examined a broad range of outcomes, including prenatal care adequacy, preterm birth rate, low and very low birthweight rate, and infant mortality. Table

IV.1 summarizes the analysis results for each of these outcomes. The principal results are the following:

C Prenatal Care Utilization. Healthy Start is associated with significant improvements in many of the measures of prenatal care utilization.

- In five project areas--Cleveland, New Orleans, New York City, Philadelphia, and Pittsburgh--Healthy Start is related to higher percentages of women receiving any prenatal care at all.

- Healthy Start is associated with increases in the percentage of women receiving first-trimester care and increases in the average number of prenatal care visits in three Healthy Start project areas--New Orleans, New York City, and Philadelphia. In Baltimore and Northern Plains, Healthy Start is associated with a higher average number of prenatal care visits.

- In 8 of the 15 project areas, Healthy Start is associated with improved adequacy of prenatal care utilization. These 8 project areas are: Baltimore, Birmingham, Chicago, New Orleans, New York City, Northern Plains, Oakland, and Philadelphia.

- Healthy Start is associated with increases in the adequacy of prenatal care initiation in 4 of the 15 project areas: Birmingham, New Orleans, New York City, and Philadelphia.

- Healthy Start is associated with improved adequacy of the number of prenatal care visits in 9 of the 15 project areas: Baltimore, Birmingham, Boston, Chicago, New Orleans, New York City, Northern Plains, Oakland, and Philadelphia.

- Three project areas--New Orleans, New York City, and Philadelphia--show improvements in all measures of prenatal care due to Healthy Start.

C Preterm Birth Rate. In 4 project areas, Healthy Start is associated with a lower preterm birth rate: Birmingham, New Orleans, Oakland, and Philadelphia.

43 X X Rate Infant Mortality X X X Rate Very Low Birthweight X X X Low Rate Birthweight X X X X Preterm Outcome Birth Rate X X X X X X X X X TABLE IV.1 TABLE Visits X X X X Initiation SUMMARY OF OUTCOMES ANALYSIS RESULTS ANALYSIS OF OUTCOMES SUMMARY Adequacy of Prenatal Care Adequacy of Prenatal X X X X X X X X Utilization Project Area District of Columbia New Orleans Detroit Oakland Philadelphia Baltimore Birmingham Boston Chicago Cleveland New York City Northern Plains Northwest Indiana Pee Dee Pittsburgh X denotes a statistically significant difference in outcomes with and without Healthy Start for a significance level of 5 percent or lower, one-tailed test. X denotes a statistically significant difference in outcomes with and without

44 C Low and Very Low Birthweight Rates. Three project areas--Birmingham, Detroit, and the District of Columbia--have significant reductions in the rate of low birthweight resulting from Healthy Start. In Birmingham, Boston, and Pittsburgh, Healthy Start is related to reductions in the rate of very low birthweight.

C Infant Mortality Rate. Infant mortality rates declined significantly in the Healthy Start project areas between the baseline period of 1984 through 1988 and 1996. Infant mortality rates declined by roughly the same magnitude in similar comparison areas and in the nation as a whole.

- In two project areas, New Orleans and Pittsburgh, Healthy Start is associated with significant reductions in infant mortality.

- In both New Orleans and Pittsburgh, Healthy Start is associated with reductions in neonatal mortality. In Oakland and Pittsburgh, Healthy Start is related to reductions in postneonatal mortality.

Three project areas--Birmingham, New Orleans, and Pittsburgh--have significant improvements in several birth outcomes and fairly large reductions in infant mortality attributed to Healthy Start.

Birmingham has the most consistent set of findings. Compared with its matched comparison areas,

Birmingham has statistically significant improvements in the adequacy of prenatal care utilization, statistically significant reductions in the preterm birth rate and rates of low and very low birthweight,

and a large (but not quite statistically) significant decline in the infant mortality rate. New Orleans

also shows statistically significant effects of Healthy Start: improvements in prenatal care adequacy,

including improvements in every available measure of prenatal care utilization and adequacy; a

decline in the preterm birth rate; and a reduction in the infant mortality rate. In Pittsburgh, Healthy

Start is associated with a significant reduction in both the very low birthweight rate and infant

mortality rate.

The analysis results also suggest some important improvements in birth outcomes for three

additional project areas--Baltimore, Oakland, and Philadelphia. In Baltimore, Healthy Start is related

to improvements in the adequacy of prenatal care utilization and a large, but not statistically

45 significant, decline in the infant mortality rate. The Healthy Start program in Oakland is related to improvements in the adequacy of prenatal care utilization and a reduction in the preterm birth rate.

In addition, infant mortality in the Oakland project area is very low and, by the end of the

demonstration period, is close to the national average. In Philadelphia, Healthy Start is associated with improvements in all measures of prenatal care utilization and a reduction in the preterm birth rate.

In summary, the Healthy Start program demonstrates improvements in several birth outcomes across the 15 original project areas and reductions in infant mortality in 2 project areas. Although a more in-depth synthesis of the process and outcomes analysis is the focus of the final synthesis report of the national evaluation (Devaney et al. 2000), Healthy Start is particularly successful at linking women and their families to care, as shown by both the focus on case management in the program interventions and the resulting improvements in the adequacy of prenatal care utilization in 8 of the 15 project areas.

46 REFERENCES

Baltay, Michelle, Marie McCormick, and Paul Wise. “Evaluation of the Fetal and Infant Mortality Review (FIMR) Programs in the Healthy Start Program.” Boston, MA: Harvard University School of Public Health, July 1997.

Berkowitz, G., and E. Papiernik. “Epidemiology of Preterm Birth.” Epidemiologic Review, vol. 15, 1993, pp. 414-443.

Devaney, Barbara, Barbara Foot, and Dexter Chu. “Case Management in Healthy Start.” Princeton, NJ: Mathematica Policy Research, Inc., March 1999.

Devaney, Barbara, Embry Howell, Marie McCormick, and Lorenzo Moreno. “Reducing Infant Mortality: Lessons Learned from Healthy Start.” Princeton, NJ. Mathematica Policy Research, 2000.

Devaney, Barbara, and Marie McCormick. “Evaluation Design: National Evaluation of Healthy Start.” Princeton, NJ: Mathematica Policy Research, Inc., December 1993.

Harrington, Mary, Barbara Foot, and Elizabeth Closter. “Using Health Education to Reduce Infant Mortality: The Healthy Start Experience.” Washington, DC: Mathematica Policy Research, Inc., September 2, 1998.

Howell, Embry, Barbara Devaney, Barbara Foot, Mary Harrington, Melissa Schettini, Marie McCormick, Ian Hill, Renee Schwalberg, and Beth Zimmerman. “The Implementation of Healthy Start: Lessons for the Future.” Washington, DC: Mathematica Policy Research, Inc., November 1997.

Howell, Embry, Barbara Devaney, Barbara Foot, Jane Griffin, Mary Harrington, Ian Hill, Marie McCormick, Renee Schwalberg, Amy Zambrowski, and Beth Zimmerman. “Implementing a Community-Based Initiative: The Early Years of Healthy Start.” Washington, DC: Mathematica Policy Research, Inc., November 1994.

Howell, Embry, Beth Zimmerman, and Elizabeth Closter. “Infant Mortality Prevention in American Indian Communities: Northern Plains Healthy Start.” Washington, DC: Mathematica Policy Research, Inc., January 1999.

Kogan, Michael D., Joyce A. Martin, Greg R. Alexander, Milton Kotelchuck, Stephanie J. Ventura, and Frederic D. Frigoletto. “The Changing Pattern of Prenatal Care Utilization in the United States, 1981-1995, Using Difference Prenatal Care Indices.” Journal of the American Medical Association, vol. 279, 1998, pp. 1623-1628.

47 Kotelchuck, Milton. “An Evaluation of the Kessner Adequacy of Prenatal Care Index and a Proposed Adequacy of Prenatal Care Utilization Index.” American Journal of Public Health, vol. 84, 1994, pp. 1414-1420.

McCann, Thurma, Bernice Young, Donna Hutten, Angela Hayes, and Beverly Wright. The Healthy Start Initiative: A Community-Driven Approach to Infant Mortality Reduction-Vol. IV. Community Outreach. Arlington, VA: National Center for Education in Maternal and Child Health, 1996.

McCormick, Marie, and Lisa Deal. “The National Evaluation of Healthy Start: Report from a Survey of Postpartum Women.” Washington, DC: Mathematica Policy Research, Inc., July 1998.

National Center for Health Statistics. Health United States, 1998 with Socioeconomic Status Chartbook. Hyattsville, MD: U.S.P.H.S., 1998.

Raykovich, Karen Thiel, Marie C. McCormick, Embry M. Howell, and Barbara L. Devaney. “Evaluating the Healthy Start Program: Design Development to Evaluative Assessment.” Evaluation and the Health Professions, vol. 19, no. 3, September 1996, pp. 342-362.

Shapiro, S., E. Schlesinger, and R. Nesbitt. Infant, Perinatal, Maternal and Childhood Mortality in the United States. Cambridge, MA: Harvard University Press, 1968.

Simon, Della, and Karen Thiel Raykovich. “The Role of Outreach Workers in the Healthy Start Program.” Rockville, MD: U.S. Department of Health and Human Services, Health Resources and Services Administration, November 1995.

Tompkins, Mark E., Greg R. Alexander, Kirby L. Jackson, Carlton A. Hornung, and Joan M. Altekruse. “The Risk of Low Birth Weight.” American Journal of Epidemiology, vol. 122, no. 6, 1985, pp. 1067-1079.

Ventura, Stephanie J., Robert N. Anderson, Joyce A. Martin, and Betty Smith. “Birth and Infant Deaths: Preliminary Data for 1997.” National Vital Statistics Reports, vol. 47, no. 4, October 7, 1998.

Ventura, Stephanie J., Joyce A. Martin, Sally C. Curtin, and T.J. Mathews. “Births: Final Data for 1997.” National Vital Statistics Reports, vol. 47, no. 18, April 29, 1999.

48 APPENDIX A

BIRTH OUTCOMES AND INFANT MORTALITY RATES: HEALTHY START PROJECT AREAS AND MATCHED COMPARISON AREAS, 1984 TO 1996 This appendix presents descriptive tabulations on key outcomes over time for each project area and its matched comparison areas. Baltimore Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 22.3 61.5 20.3 19.5 56.5 56.2 55.7 55.9 18.7 60 53.6 53.3 20 17.9 18.2 52.7 52.3 17.1 49.6 49.0 16.3 16.5 50 45.1 14.6 14.4 15 40 12.6 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation) 18.2 18.3 22.8 22.4 17.1 20.9 16.6 16.2 15.8 19.6 15.1 15.1 18.9 18.9 19.4 19.0 18.9 14.9 14.9 14.8 20 18.2 18.7 15 14.4 18.1

15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Baltimore unlinked birth and infant death certificates 1985-88 1990-91 1992-94 1995-96 Birmingham Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25.6 69.4 70.9 25 70 64.6 61.2 58.8 59.3 58.7 60.5 60.1 19.6 55.5 18.4 60 52.8 20 18.2 49.0 16.9 16.6 15.6 50 14.7 15.1 15 13.9 14.3 40 12.5 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

19.1 19.1 20 18.5 18.4 18.1 17.3 17.7 17.6 15 13.5 13.5 16.7 13.1 13.2 16.2 16.3 12.8 12.5 12.4 12.0 11.7 14.8 11.5 15 10.6 10.3 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Alabama linked birth and infant death certificates 1988 1989-91 1992-94 1995-96 Boston Project and Comparison Area

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30

80 74.6 72.7 69.0 25 70 67.2 60 57.4 20 48.8 49.6 50.2 50 15 13.4 13.4 40 11.6 10.7 10.6 9.4 10.0 30 10 8.2 20 5 10 0 0 Project Area Comparison Area Project Area Comparison Area

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20 15 16.0 16.3 14.7 15.2 11.2 14.0 10.7 15 13.5 10.4 10.2 10.5 10.4 12.8 9.9 12.1 10 9.4 10

5 5

0 0 Project Area Comparison Area Project Area Comparison Area Source: Massachusetts and New York City linked birth and infant death certificates 1988 1989-91 1992-94 1995-96 Chicago Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 61.2 19.8 60 56.2 57.4 20 18.8 19.0 18.6 54.1 52.5 17.6 18.0 51.0 51.2 49.8 49.9 48.6 15.9 50 46.0 14.8 15.5 42.5 15 14.1 13.7 40 12.5 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

19.6 19.6 19.0 18.8 20 18.5 18.2 18.3 14.0 14.1 17.7 17.7 17.9 15 13.5 17.2 17.4 12.9 13.0 12.6 12.2 12.4 12.0 12.3 12.4 11.2 15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Illinois linked birth and infant death certificates 1984-88 1989-91 1992-94 1995-96 Cleveland Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 62.6 20.9 21.2 20.0 55.8 57.4 60 20 17.9 18.4 17.9 18.1 18.0 52.2 52.6 17.2 17.1 50.3 48.3 50.4 50.2 48.9 16.6 50 46.2 46.9 15.9 15 40 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20.0 19.8 19.6 19.5 19.1 19.3 14.9 14.9 14.9 20 18.5 18.8 18.9 18.5 14.3 14.2 15 13.7 13.8 14.1 14.1 13.9 17.4 17.5 12.9 13.3 15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Ohio and Illinois linked birth and infant death certificates 1984-88 1989-91 1992-94 1995-96 District of Columbia Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30

80 24.5 25 70 22.5 22.0 20.0 60 20 18.2 49.1 16.7 16.5 16.6 47.5 46.4 50 43.4 44.6 43.3 41.4 41.6 15 14.1 40 35.3 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation) 18.4 22.7 17.5 21.7 17.0 17.1 21.4 21.1 16.4 16.6 16.7 20.7 20.6 16.1 16.1 20.1 19.9 20.0 20 15

15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2 Source: District of Columbia and Pennsylvania linked birth and infant death certificates 1990-91 1992-94 1995-96 Detroit Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30

80 74.4 74.6 25.0 24.9 72.2 73.7 71.6 70.1 71.8 70.9 25 23.6 70 67.3 66.0 21.9 62.6 63.0 21.2 19.9 19.2 19.6 19.4 60 20 17.6

50 14.7 15 13.9 40 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

16.7 20.3 16.0 19.9 19.5 15.3 20 18.9 18.9 14.8 14.4 14.4 18.4 18.3 18.1 15 13.9 13.9 13.9 17.3 17.3 13.5 13.2 16.2 15.8 12.1 15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Michigan linked birth and infant death certificates 1984-88 1989-91 1992-94 1995-96 New Orleans Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 63.8 21.8 58.6 60.5 55.4 57.4 60 53.1 20 17.8 52.1 50.5 17.4 17.0 47.4 16.0 50 15.2 14.6 15.2 15 40 11.1 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20.3 20.8 20.2 19.6 19.8 15.2 20 19.4 18.9 19.2 14.4 14.5 14.5 14.5 18.4 15 13.9 13.8 13.5 13.3

15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2 Source: Louisiana and Alabama linked birth and infant death certificates 1989-91 1992-94 1995-96 New York City Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 19.4 18.4 18.7 60 20 17.8 17.6 16.7 50 45.4 47.0 46.8 15.1 39.0 15 37.9 38.5 12.2 12.2 12.4 12.8 40 34.0 35.1 12.0 33.5 32.0 33.7 30 28.1 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20 14.2 14.5 18.0 15 13.6 12.9 16.6 16.6 15.6 15.7 15.5 16.0 16.0 11.9 12.3 12.1 12.2 11.7 11.5 14.8 14.4 14.5 11.3 11.0 15 14.0 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: New York City linked birth and infant death certificates 1985-88 1989-91 1992-94 1995-96 Northern Plains Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 62.8 64.4 60.1 60 20 18.9 52.3 17.5 49.9 16.1 45.6 15.3 15.0 50 44.3 14.2 40.8 40.6 15 13.9 37.5 38.0 12.7 40 33.8 10.5 9.8 30 10 8.5 8.9 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2 Low Birthweight Rate Preterm Birth Rate (Percentage of births less than 2,500 grams) (Percentage of births less than 37 weeks gestation) 20 25

20 15

15 12.5 11.8 11.8 11.5 10 11.0 11.4 10.9 10.9 10.4 10.3 10.5 9.7 10 6.2 6.3 6.2 5.7 5.9 6.1 5.2 5.3 5.2 5.2 5.6 5.6 5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Indiana Health Service (IHS) counts of births and infant deaths 1984-88 1989-91 1992-94 1995-96 Northwest Indiana Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 63.8 65.7 64.3 60.8 57.1 58.3 58.1 58.1 60 54.4 55.7 53.8 20 51.5 16.7 16.0 50 15.7 14.3 14.1 15 13.0 13.4 13.3 13.4 12.1 40 11.4 12.0 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20 15 17.4 16.9 16.4 16.0 15.8 16.0 15.8 15.1 14.6 15.0 14.9 11.3 14.0 10.8 10.9 10.7 10.8 10.8 10.7 15 9.9 10.1 10.0 10.2 10.1 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Indiana and Illinois linked birth and infant death certificates 1985-88 1989-91 1992, 1994 1995-96 Oakland Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 81.5 80 76.4 25 70 65.7 61.8 62.1 60 55.1 56.0 20 17.2 47.5 16.3 15.7 50 44.4 13.9 15 13.1 12.0 40 11.2 11.2 10.0 8.9 9.4 30 10 8.1 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20 15 15.6 15.1 14.5 11.5 11.3 14.1 13.8 13.8 10.7 10.7 10.6 15 13.1 9.9 10.1 12.5 12.9 12.8 9.5 11.7 12.1 10 8.9 9.2 8.8 8.2 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: California linked birth and infant death certificates 1984-88 1989-91 1992-94 1995-96 Pee Dee Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 71.5 71.5 25 66.5 70 63.6 58.3 58.1 59.8 60 55.6 54.1 55.5 20 50.5 17.2 48.1 16.0 50 15.8 15.5 15 14.0 14.1 12.5 12.4 12.5 40 11.6 11.5 30 10 6.6 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20 15

15.2 15.3 15.1 15.1 11.5 11.4 10.8 15 14.0 13.9 10.3 10.3 10.7 13.2 13.3 13.3 9.8 10.0 9.9 10.0 10.1 12.6 12.8 12.8 10 9.5 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: South Carolina linked birth and infant death certificates 1984-88 1989-91 1992-94 1995-96 Philadelphia Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 25 70 20.3 19.7 60 56.4 55.2 20 17.5 17.3 49.9 51.2 16.6 49.1 49.3 47.8 16.4 50 44.8 44.2 15.2 14.5 42.4 41.9 13.8 39.5 15 40 11.5 10.7 10.9 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20.5 15.5 19.3 20 18.7 14.4 14.4 18.1 18.5 13.9 14.2 14.2 17.3 17.7 17.3 15 13.4 13.7 16.8 17.2 13.0 13.1 12.9 16.4 12.4 15.8 15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Source: Pennsylvania linked birth and infant death certificates 1986-88 1989-91 1992-94 1995-96 Pittsburgh Project and Comparison Areas

Receipt of Adequate or Better Prenatal Care Infant Mortality Rate (Kotelchuck Index) (Infant deaths per thousand live births) 90 30 80 69.6 70.1 25 67.5 70 65.6 65.3 64.5 61.5 60.9 60.8 62.5 62.2 59.6 19.3 60 20 17.4 16.6 16.5 15.5 15.6 15.6 50 14.8 15.3 14.7 14.7 15 40 11.4 30 10 20 5 10 0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2

Low Birthweight Rate Preterm Birth Rate

20 (Percentage of births less than 2,500 grams) 25 (Percentage of births less than 37 weeks gestation)

20 15 17.5 17.2 16.6 16.8 16.8 13.1 12.8 16.2 16.4 16.0 12.5 12.7 12.6 12.4 12.3 15.7 12.2 11.8 15.0 15.2 11.3 14.0 10.6 10.7 15 10 10

5 5

0 0 Project Area Comparison Area 1 Comparison Area 2 Project Area Comparison Area 1 Comparison Area 2 Source: Pennsylvania, Illinois, and Ohio linked birth and infant death certificates 1984-88 1989-91 1992-94 1995-96 APPENDIX B

ESTIMATES OF THE EFFECTS OF HEALTHY START ON PRENATAL CARE, BIRTH OUTCOMES, AND INFANT MORTALITY This appendix presents detailed analysis results from the outcomes analysis of the national evaluation of Healthy Start. This analysis uses multivariate logit models to estimate the effects of

Healthy Start on prenatal care, birth outcomes, and infant mortality. Estimated coefficients from logit models are not easily interpreted but can be transformed, however, to facilitate the interpretation of results. Specifically, the analysis results presented in the following figures present adjusted differences in the outcome variables under two scenarios: (1) with the Healthy Start program, and (2) without the Healthy Start program. The adjusted or predicted values are based on the estimated coefficients from the multivariate logit models. Although all years of data are used to estimate the effects of Healthy Start, a specific year must be used to generate the predicted values.

Because 1996 was the last year of the demonstration period and the latest year for which we have linked birth and death data, 1996 is the year used in generating the predicted outcome values. Impact of Baltimore Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 95.6 95.2 90 90 80 70.1 71.8 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Average Number of Prenatal Visits Percentage of Women Who Received Adequate or Better Prenatal Care (Kotelchuck Index) 12 80

10.0** 10 9.5 70 60.4* 60 56.4 8 50 6 40 30 4 20 2 10

0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 90 100 80 90 72.2* 81.0 81.4 70 68.9 80 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Baltimore Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 25 20 22.4 22.5 17.8 17.6 20 15

15 10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Births less than 1,500 grams per hundred births Infant deaths per thousand live births 4 20 3.8 18.2 3.5

15 3 14.0

2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Birmingham Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 98.2 98.5 90

90 80 74.8 76.0 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 11.3 11.3 80 74.2** 10 70 67.8 60 8 50 6 40

4 30 20 2 10 0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 90 100 83.7** 80 78.1 90 86.7* 85.4 70 80 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Birmingham Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 25 20

20.5 20 14.9 17.6** 15 12.3** 15 10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births Births less than 1,500 grams per hundred births 20 5 18.9 4.6

4 15 14.6 3.2** 3 10 2

5 1

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Boston Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 99.3 99.5 90 82.7 81.7 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Average Number of Prenatal Visits Percentage of Women Who Received Adequate or 11.6 11.4 Better Prenatal Care (Kotelchuck Index) 12 80 75.6 76.1 10 70 60 8 50 6 40 30 4 20 2 10 0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 90 83.1* 100 81.2 90.6 91.3 80 90 70 80 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Boston Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 19.6 19.4 20

15 15 12.5 11.8

10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Births less than 1,500 grams per hundred births Infant deaths per thousand live births 4 20 3.6

15 3 2.9 * 14.0

11.5 2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Chicago Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 95.4 95.1 90 90 80 71.0 70.7 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 70 9.9 9.9 57.8** 10 60 56.2

8 50 40 6 30 4 20

2 10

0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 80 100 70 66.8* 90 65.7 81.6 81.3 80 60 70 50 60 40 50 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Chicago Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20 18.1 18.3

15 15 13.0 13.1

10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20

3 15 13.3 12.7 2.4 2.5

2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Cleveland Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 96.6** 90 92.1 90 80 80 70 68.7 69.7 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 70 10.4 62.3 10 9.3 60

49.4 8 50 40 6 30 4 20

2 10

0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 80 100 72.2 70 90 80 79.6 79.6 60 59.4 70 50 60 40 50 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Cleveland Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20 17.6 16.2 15 15 14.5 12.3

10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Births less than 1,500 grams per hundred births Infant deaths per thousand live births 4 20

3.4 17.0

3 15

11.7 2.1 2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of District of Columbia Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 94.7 90 91.2 90 80 80 70 70 60.4 60 54.5 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Average Number of Prenatal Visits Percentage of Women Who Received Adequate or 12 Better Prenatal Care (Kotelchuck Index) 70

10 9.5 60 8.5 51.0 8 50 46.1 40 6 30 4 20

2 10

0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 80 100 70 67.2 90 60.9 80 60 71.3 70 67.2 50 60 40 50 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of District of Columbia Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 25 20 17.8 20.9 20.5 16.2* 20 15

15 10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births Births less than 1,500 grams per hundred births 20 5 18.8 4.6 4.4 17.6 4 15

3 10 2

5 1

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Detroit Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 95.4 95.3 90 90 80 80 70 68.0 69.8 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 10.6 10.9 80 70 67.6 10 62.1 60 8 50 6 40

4 30 20 2 10 0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 90 100 80.8 80 90 75.4 79.3 80.2 70 80 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Detroit Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20 18.4 18.6

15.8 15 15 14.2**

10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20

3.4 16.6 16.8 3.0 3 15

2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of New Orleans Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 96.4** 90 90.8 90 80 70.7** 80 70 67.8 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 11.2** 70 65.5** 10.3 62.3 10 60

8 50 40 6 30 4 20

2 10

0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 80 76.0** 100 72.8 70 90 82.3** 80 78.6 60 70 50 60 40 50 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of New Orleans Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 25 20

21.3 20 19.4* 15.1 15 14.2

15 10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20 18.3 3.4 3.1 3 15

11.3** 2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of New York City Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 96.2** 95.4 90 90 80 80 70 70 60 53.0** 60 50.0 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Average Number of Prenatal Visits Percentage of Women Who Received Adequate or Better Prenatal Care (Kotelchuck Index) 12 80

10 70 9.0** 8.6 60 8 50 47.2** 44.8 6 40 30 4 20 2 10 0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 90 100 80 90 70 80 65.5** 60.4 69.5** 70 67.8 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of New York City Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20

15 14.5 14.5 15

11.6 11.6

10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Births less than 1,500 grams per hundred births Infant deaths per thousand live births 4 20

3 15 2.5 2.6 11.1 10.0 2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Northern Plains Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 97.4 97.2 90 90 80 80 70 67.8 67.7 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 90 10 80 8.9** 8.2 70 8 60 50 45.8 ** 6 41.8 40 4 30 20 2 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Percentage of Women Who Received Adequate or Better Better Initiation of Prenatal Care Prenatal Care Services 100 100 90 90 80 78.3 77.4 80 70 70 60 60 56.7 ** 51.7 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Northern Plains Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20

15 15

12.2 12.6

10 10

6.3 6.5 5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20

3 15 12.6

10.1 2 10

1.4 1.1 1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of NW Indiana Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 97.0 97.6 90 90 80 71.1 72.8 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 11.0 11.3 80 69.0 10 70 62.6 60 8 50 6 40

4 30 20 2 10 0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 90 100 82.5 80 90 73.9 81.4 82.7 70 80 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of NW Indiana Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20

15.0 15 14.2 15

10.4 10 10 9.5

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20

3 15 14.0

2.1 2 1.9 10 8.9

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Oakland Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care

99.0 99.4 in the First Trimester 100 90 84.2 84.4 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits Better Prenatal Care (Kotelchuck Index) 16 90 83.0** 80.7 14 13.2 13.5 80 12 70 60 10 50 8 40 6 30 4 20 2 10 0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Percentage of Women Who Received Adequate or Better Prenatal Care Visits Better Initiation of Prenatal Care 100 90.9** 100 88.6 91.1 91.5 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Oakland Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20

15 15 12.6 11.3**

10 10 9.4 9.2

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births (results refer to 1995 data) 4 Births less than 1,500 grams per hundred births 20

3 15

2 10 9.3 1.5 1.5 7.2

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Pee Dee Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 98.0 98.5 90 90 80 73.3 70.3 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits Better Prenatal Care (Kotelchuck Index) 12 11.2 80 10.7 10 70 60.2 61.4 60 8 50 6 40 30 4 20 2 10

0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 90 100 80 75.8 77.8 90 79.7 79.7 70 80 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Pee Dee Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20

15 14.5 15 13.8 11.7 11.2 10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20

3 15

2.4 11.6 2.2 11.1 2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Philadelphia Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care in the First Trimester 100 96.6** 95.4 90 90 80 80 70 66.9* 65.6 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 70

9.7** 57.8** 10 9.2 60 52.3 8 50 40 6 30 4 20

2 10

0 0 With HS Without HS With HS Without HS

Percentage of Women Who Received Adequate or Percentage of Women Who Received Adequate or Better Better Initiation of Prenatal Care Prenatal Care Visits 100 80 70.7** 90 70 64.8 80 78.8** 75.8 60 70 60 50 50 40 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Philadelphia Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20 18.1 16.7** 15 15 13.5 14.0

10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20

3.1 3 2.9 15

11.4 11.6 2 10

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Pittsburgh Healthy Start on Prenatal Care Adequacy, 1996

Percentage of Women Who Received Any Prenatal Care Percentage of Women Who Began Receiving Prenatal Care 97.4** in the First Trimester 100 95.1 90 79.4 90 80 78.2 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10

0 With HS Without HS 0 With HS Without HS

Percentage of Women Who Received Adequate or Average Number of Prenatal Visits 12 Better Prenatal Care (Kotelchuck Index) 10.6 10.8 80 70.6 72.2 10 70 60 8 50 6 40 30 4 20 2 10

0 0 With HS Without HS With HS Without HS Percentage of Women Who Received Adequate or Better Percentage of Women Who Received Adequate or Prenatal Care Visits Better Initiation of Prenatal Care 80 78.9 79.5 100 70 90 86.7 86.1 80 60 70 50 60 40 50 40 30 30 20 20 10 10 0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test Impact of Pittsburgh Healthy Start on Key Outcomes, 1996

Preterm Birth Rate Low-Birthweight Rate Births less than 37 weeks gestation per hundred births Births less than 2,500 grams per hundred births 20 20

15.5 15 14.2 15

11.9 11.9

10 10

5 5

0 0 With HS Without HS With HS Without HS

Very-Low Birthweight Rate Infant Mortality Rate Infant deaths per thousand live births 4 Births less than 1,500 grams per hundred births 20 17.5

15 3 2.8

2.3*

2 10 8.6**

1 5

0 0 With HS Without HS With HS Without HS

*(**): p<.05(.01), one-tailed test