ﬁwpﬁ
. ; any

at
51“.:-
A .5.
I...

«n.3,... :. J .
%mm "#5

3...; ...
243...?“
u. ._..

we?

a. .
.41.:
u... km». 1a...
LA... L

:1?

in... ,
:1

iv}:

31 «.2! I

in... (tax!-

.\4h..5l 3 I

’51..

I!!! ‘3.“

 

.25.
s
ha:
.3;
.15-
2:5...
1;} ... “"5!
o. . . .3. ts: . I
nv‘r‘uﬁxwﬁ...»
. vi!‘ u‘pﬂ‘t“. ¢
. whithﬁxw .

1.: 2
Kn. .232 2
2 I... 1.5.!“

.a n).-. A Y, .... $5
1...! a: A:
:13 ..

. i...

um.w%§. 2%.? “W...

s .\.\ .v.

.51.. ‘

If.
,: . .( In.

4,5;~N,.w. &‘

 

 

LIBRARY
Michigan State
University

 

 

This is to certify that the
dissertation entitled

AN INVESTIGATION OF RACIAL DISPARITIES IN

INFANT MORTALITY ACROSS THE UNITED STATES:

THE ROLES OF
SOClO-DEMOGRAPHIC FACTORS,
BIRTH AND FETAL DEATH REGISTRATION
AND PERINATAL REGIONALIZATION

presented by

Crystal Pirtle Tyler

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Epidemiology

 

 

 

Major Professor’s Signature

5/’3/0‘f
' /

MSU is an Afﬁrmative Action/Equal Opportunity Employer

4 - -.-...-.---.------—-..-.--—..I—.-..-..-.--—.-.-.—----.-.-.—.-.-.—.—— -.—-.—.-.—.—o—.-

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K.IProj/Acc&Pres/CIRC/DateDue‘indd

AN INVESTIGATION OF RACIAL DISPARITIES IN INFANT MORTALITY
ACROSS THE UNITED STATES: THE ROLES OF
SOCIO-DEMOGRAPHIC FACTORS,

BIRTH AND FETAL DEATH REGISTRATION
AND PERINATAL REGIONALIZATION

By

Crystal Pirtle Tyler

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
EPIDEMIOLOGY

2009

ABSTRACT
AN INVESTIGATION OF RACIAL DISPARITIES IN INFANT MORTALITY
ACROSS THE UNITED STATES: THE ROLES OF SOCIO-DEMOGRAPHIC
FACTORS, BIRTH AND FETAL DEATH REGISTRATION
AND PERINATAL REGIONALIZATION
By
Crystal Pirtle Tyler

Objective: To determine the effect of racial inequalities in sociodemographic factors,
state fetal death registration requirements and reporting of non-viable births and perinatal
regionalization, deﬁned as birth hospital level and neonatal intensive care unit (N ICU)
transfer, on nationwide variation in racial disparities in infant mortality (IM).

Methods: National Center for Health Statistics (N CHS) live birth and linked infant
death records from 2000-2002, US. Bureau of the Census, 2000 Census of Population
and Housing, and Michigan Department of Community Health live birth and linked
infant death vital records from 1996-2006 were used to examine absolute
(black IM / white IM) and relative (black IM —— white IM) racial disparities in IM rates.

Results: Absolute and relative U.S. disparity measures were 6.99 per 1,000 live births
and 2.42, respectively. The absolute disparity measure was highly correlated with black
IM (r = 0.91) but not white IM (r = -0.03), while the relative measure was correlated with
both black IM (r = 0.57) and white IM (r = -O.51). Compared to racial inequalities in
other infant, maternal and state risk factors, inequalities in the proportion of very low
birthweight births were most correlated with disparities in IM. Mortality rates and racial
disparities were the highest among states with birthweight only fetal death reporting

criteria and among states with the highest proportion of non-viable births recorded in

birth certiﬁcates (RR=1.22; 95% CI=l .17-1 .37). The largest proportion of this difference

was accounted for by births S 22 weeks gestation (RR=1.71; 95% CI=1.43-2.04). The
case study in Michigan found that the majority of infants were born at a level 3 hospital.
The highest IM rates were seen among extremely preterm infants born at level 1 hospitals
compared to their level 3 counterparts (level 1 = 465.3 per 1,000 live births; level 3 =
363.9 per 1,000 live births) and among extremely preterm level 1 white births compared
to their black counterparts (white = 506.1 per 1,000 live births ; black = 383.3 per 1,000
live births). Extremely preterm black infants who were born at a level 1 hospital and
subsequently transferred to the NICU had a signiﬁcantly decreased risk of infant death,
compared to their white counterparts (RR=0.41; 95% CI=O.26-0.66).

Conclusion: Racial inequalities in the proportion of very low birthweight and very
preterm infant births along with state differences in reporting very low birthweight and
very preterm births were consistently associated with national variation in IM disparities.
Racial inequalities in perinatal regionalization did not account for higher infant or
neonatal mortality rates among black infants. A uniform deﬁnition of fetal death should
be adopted to reduce systematic differences in the reporting of live births and fetal
deaths, especially among deaths 5 22 weeks gestation. Efforts should be made to reduce
rates of extremely preterm and extremely low birthweight births where mortality rates
and racial disparities in risk of adjusted mortality were the highest and state disparities

were more correlated.

Copyright by
CRYSTAL PIRTLE TYLER
2009

DEDICATION

This work is dedicated to my husband, Marcus. Thank you so much for your patience

and understanding throughout this process. I love you very much.

AKNOWLEDGEMENTS

I would ﬁrst like to thank the chair of my dissertation committee, Dr. Nigel Paneth who
encouraged and supported me with invaluable advice and kind words throughout this
process. I would also like to thank my committee members: Dr. Sue Grady, Dr. Barbara
Luke, Dr. Violanda Grigorescu and Dr. David Todem for their patience with my ongoing
revisions. I would also like to thank Dr. Yona Cloonan for helping me conceptualize
when my brain was tired and Dr. Nicole Talge for accepting me as a roommate for the
ﬁnal months before my defense. Finally, I would like to thank my family and friends

who supported me throughout this process.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. xi
LIST OF FIGURES ......................................................................................................... xiv
LIST OF ABREVIATIONS ............................................................................................ xv
CHAPTER 1: INTRODUCTION ....................................................................................... l
1.1 Overview of the Burden and Epidemiology of Infant Mortality ........................... 1
1.1.1 Deﬁnition of Infant Mortality ....................................................................... 1
1.1.2 Causes of Infant Mortality ............................................................................ 2
1.1.3 Timing of Death ............................................................................................ 4
1.1.4 Maternal Behavioral and Medical Inﬂuences ............................................... 5
1.1.5 Importance of Understanding Racial Disparities in Infant Mortality ........... 7
1.2 Three Key Exposures in Relation to Infant Mortality ......................................... 10
1.2.1 Social Inﬂuences ......................................................................................... 11
1.2.2 Reporting Differences ................................................................................. 13
1.2.3 Perinatal Regionalization ............................................................................ 14
1.3 Signiﬁcance of Study .......................................................................................... 16
1.4 Speciﬁc Aims ....................................................................................................... 17
1.5 Study Population and Data Sources ..................................................................... 18
1.5.1 Aim 1 ........................................................................................................... 18
1.5.2 Aim 2 ........................................................................................................... 21
1.5.3 Aim 3 ........................................................................................................... 21
1.5.4 Strengths of Vital Statistics Data ................................................................ 23
1.5.5 Limitations of Vital Statistics Data ............................................................. 23

CHAPTER 2: THE EFFECT OF RACIAL INEQUALITIES IN
SOCIODEMOGRAPHIC FACTORS ON RACIAL DISPARITIES IN INFANT

MORTALITY .................................................................................................................. 25
2.1 Abstract ................................................................................................................ 25
2.2 Introduction .......................................................................................................... 26

2.2.1 Background on Infant Mortality .................................................................. 26
2.2.2 Racial Disparities in Infant Mortality .......................................................... 26
2.2.3 Disparity Measurement ................................................................................ 27
2.2.4 Nationwide Variation in Infant Mortality and Racial Disparities in Infant
Mortality ...................................................................................................... 28
2.2.5 Study Objective ........................................................................................... 29
2.3 Materials and Methods ......................................................................................... 29
2.3.1 Study Population ......................................................................................... 29
2.3.2 Data Source and Population ........................................................................ 30
2.3.3 Outcome Variables ...................................................................................... 31
2.3.4 Exposure Variables ...................................................................................... 32

vii

2.3.4.1 Vital Statistics Measures .............................................................. 32

2.3.4.2 Census Measures .......................................................................... 33
2.3.5 Data Analysis .............................................................................................. 33
2.3.6 Mapping ...................................................................................................... 34
2.4 Results .................................................................................................................. 34
2.4.1 Region and State Speciﬁc Number of Live Births and Infant Deaths ........ 34
2.4.2 Infant Mortality Rates, Disparity Ratios, Disparity Differences and Racial
Composition Scores .................................................................................... 38
2.4.3 Lowest and Highest Disparity Ratios and Disparity Differences ............... 43
2.4.4 Racial Inequalities Among Selected Infant Characteristics ........................ 46
2.4.5 Racial Inequalities Among Selected Maternal Characteristics ................... 52
2.4.6 Racial Inequalities Among Selected State Characteristics ......................... 57
2.4.7 Correlation Between Infant Mortality, Disparity Ratios and Disparity
Differences .................................................................................................. 62
2.4.8 Correlation Between Inequalities in Infant Antecedents and Disparities in
Infant Mortality .......................................................................................... 63
2.4.9 Correlation Between Inequalities in Maternal Characteristics and Disparities
in Infant Mortality ...................................................................................... 65
2.4.10 Correlation Between Inequalities in State Characteristics and Disparities in
Infant Mortality ......................................................................................... 67
2.5 Discussion ............................................................................................................ 68
2.5.1 Nationwide Variation in Infant Mortality Rates ......................................... 68
2.5.2 Racial Disparities in Infant Mortality ......................................................... 69
2.5.3 Risk Factor Inequalities and Racial Disparities in Infant Mortality ........... 69
2.5.4 Correlation Between Disparity Measures and Racial Disparities in Infant
Mortality ...................................................................................................... 71
2.5.5 Correlation Between Inequalities in Exposure and Disparities in
Outcome ...................................................................................................... 71
2.5.6 Other Reasons for National Variation in Infant Mortality .......................... 74
2.6 Conclusion ............................................................................................................ 75

CHAPTER 3: THE IMPACT OF FETAL DEATH REPORTING PRACTICES ON
RACIAL DISPARITIES IN INFANT, EARLY NEONATAL, LATE NEONATAL,

POSTNEONATAL, AND FETAL MORTALITY ......................................................... 77
3.1 Abstract ................................................................................................................ 77
3.2 Introduction ............... . .......................................................................................... 78

3.2.1 Nationwide Variation in Racial Disparities in Infant Mortality .................. 78
3.2.2 Fetal Death Reporting ................................................................................. 78
3.2.3 Study Objective ........................................................................................... 79
3.3 Materials and Methods ......................................................................................... 80
3.3.1 Study Population ......................................................................................... 80
3.3.2 Data Source and Preparation ....................................................................... 80
3.3.3 Outcome Variables ...................................................................................... 82
3.3.4 Exposure Variables ........................................................ - .............................. 8 3
3.3.5 Gestational Age Stratiﬁcation ..................................................................... 89
3.3.6 Data Analysis ............................................................................................... 89

viii

3.3.7 Mapping ...................................................................................................... 90
3.4 Results .................................................................................................................. 90

3.4.1 Mortality Rates, Disparity Ratios and Disparity Differences By Race ....... 90

3.4.2 Mortality Rates and Racial Disparities Among Infants _>_ 20 Weeks Gestation
and With Unknown Gestational Age By Fetal Death Registration Area ...94

3.4.3 Mortality Rates and Racial Disparities Among Infants 5 22 Weeks Gestation
and 23-28 Weeks Gestation By Fetal Death Registration Area ................. 97

3.4.4 Mortality Rates and Racial Disparities Among Infants 2 20 Weeks Gestation
and With Unknown Gestational Age By Proportion Non-Viable

Category ..................................................................................................... 99

3.4.5 Mortality Rates and Racial Disparities Among Infants 5 22 Weeks Gestation
and 23-28 Weeks Gestation By Proportion Non-Viable Category .......... 101

3.4.6 Relative Risk of Mortality ......................................................................... 103

3.5 Discussion ........................................................................................................... 107
3.5.1 Fetal Death Registration Area ................................................................... 107
3.5.2 Proportion of Non-Viable Births ............................................................... 109
3.5.3 Racial Differences in Fetal Deaths ............................................................ 110
3.5.4 Limitations of Study ................................................................................. 112
3.5.5 Strengths of Study ..................................................................................... 112
3.6 Conclusion .......................................................................................................... 113

CHAPTER 4: RACIAL DIFFERENCES IN THE EFFECT OF PERINATAL
REGIONALIZATION ON RACIAL DISPARITIES IN INFANT AND NEONATAL

MORTALITY, MICHIGAN 1996-2006 ....................................................................... 115
4.1 Abstract .............................................................................................................. 115
4.2 Introduction ........................................................................................................ l 16

4.2.1 Background on Infant and Neonatal Mortality .......................................... 116
4.2.2 Racial Disparities in Infant and Neonatal Mortality .................................. 117
4.2.3 Perinatal Regionalization ........................................................................... 117
4.2.4 Study Objective ......................................................................................... 119
4.3 Materials and Methods ....................................................................................... 120
4.3.1 Data Source ............................................................................................... 120
4.3.2 Outcome Variables .................................................................................... 120
4.3.3 Exposure Variables .................................................................................... 121
4.3.4 Exclusion Criteria ...................................................................................... 122
4.3.5 Statistical Analysis ..................................................................................... 122
4.4 Results ...................................................................................................... . .......... 123
4.4.1 Maternal and Infant Demographic Characteristics .................................... 123
4.4.2 Rates of Preterm Birth ............................................................................... 126
4.4.3 Perinatal Regionalization ........................................................................... 127
4.4.4 Neonatal Intensive Care Unit Transfer by Birth Hospital Level ............... 129
4.4.5 Infant and Neonatal Mortality Rates ......................................................... 132
4.4.6 Infant Mortality by Birth Hospital Level ................................................... 134
4.4.7 Racial Differences in the Adjusted Risk of Infant Mortality ..................... 141
4.5 Discussion ........................................................................................................... 144
4.5.1 Perinatal Regionalization ........................................................................... 144

ix

4.5.2 Mortality and Perinatal Regionalization .................................................... 145

4.5.3 Racial Disparities in Mortality .................................................................. 146

4.6 Conclusions ........................................................................................................ 148
CHAPTER 5: CONCLUSIONS .................................................................................... 150
APPENDICIES .............................................................................................................. 1 54
BIBLIOGRAPHY .......................................................................................................... 165

LIST OF TABLES

Table 1.1. Deaths and the percentage of total deaths for the 10 leading causes of infant

death: United States, 2004 .................................................................................................. 3
Table 1.2. Racial differences in cause speciﬁc infant mortality rates: United States, 2004
.............................................................................................................................................. 4
Table 2.1. Region and state speciﬁc number of live births and infant deaths by race:
United States, 2000-2002 .................................................................................................. 36
Table 2.2. State-speciﬁc number of live births and infant deaths among excluded states,
United States 2000-2002 ........................................................................................ ' ........... 3 7
Table 2.3. Infant mortality rates, disparity ratios, disparity differences and racial
composition scores by race: United States, 2000-2002 .................................................... 42
Table 2.4. Lowest and highest disparity ratios and disparity differences with
accompanying racial composition scores: United States, 2000-2002 ............................... 45
Table 2.5. State speciﬁc relative risks and risk differences of selected infant birth
certiﬁcate characteristics: United States, 2000-2002 .............. p .......................................... 51
Table 2.6. State speciﬁc relative risks and risk differences of selected maternal birth
certiﬁcate characteristics: United States, 2000-2002 ........................................................ 55
Table 2.7. State speciﬁc relative risks and risk differences of selected state census
characteristics: United States, 2000-2002 ......................................................................... 60
Table 2.8. Correlations between infant mortality, disparity ratios and disparity
differences: United States, 2000-2002 .............................................................................. 63

Table 2.9. Correlations between inequalities between proportion of infant characteristics,
disparity ratios and disparity differences: United States, 2000-2002 ............................... 64

Table 2.10. Correlations between inequalities between proportion of maternal
characteristics, disparity ratios and disparity differences: United States, 2000—2002 ...... 66

Table 2.11. Correlations between inequalities between proportion of state characteristics,
disparity ratios and disparity differences: United States, 2000-2002 ............................... 68

Table 3.1. Fetal death registration requirements, 1997 revision ....................................... 94

Table 3.2. Consolidated fetal death reporting requirements: United States 2000-2002 ...86

xi

Table 3.3. Borderline live births based on the proportion non-viable: United States 2000-
2002 ................................................................................................................................... 88

Table 3.4. Mortality rates, disparity ratios and disparity differences: United States 2000-
2002 ................................................................................................................................... 91

Table 3.5. Mortality rates and racial disparities among infants 2 20 weeks gestation and
with unknown gestational age by fetal death registration area: United States, 2000-2002
............................................................................................................................................ 96

Table 3.6. Mortality rates and racial disparities among infants S 22 weeks gestation and
with 23-28 weeks gestation by fetal death registration area: United States, 2000-2002
............................................................................................................................................ 98

Table 3.7. Mortality rates and racial disparities among infants 2 20 weeks gestation and
with unknown gestational age by proportion non-viable: United States, 2000-2002 ..... 100

Table 3.8. Mortality rates and racial disparities among infants 5 22 weeks gestation and
with 23-28 weeks gestation by proportion non-viable: United States, 2000-2002 ......... 102

Table 3.9. Fetal death registration area differences in risk of race speciﬁc early neonatal
mortality: United States 2000-2002 ................................................................................ 104

Table 3.10. Proportion non-viable category differences in risk of race speciﬁc early
neonatal mortality: United States 2000-2002 ................................................................. 106

Table 4.1. Maternal and demographic characteristics by race among preterm infants:
Michigan 1996-2006 ....................................................................................................... 125

Table 4.2. Hospital level at birth and neonatal intensive care unit transfer by race among
preterm infants: Michigan 1996-2002 ............................................................................ 129

Table 4.3. Neonatal intensive care unit transfer by hospital level at birth and race among
preterm infants: Michigan 1996-2006 ............................................................................ 131

Table 4.4. Infant and neonatal mortality rates by race among preterm infants: Michigan
1996-2006 ....................................................................................................................... 132

Table 4.5. Gestational week speciﬁc infant mortality rates by race: Michigan 1996-2006
......................................................................................................................................... 133

Table 4.6. Gestational week speciﬁc neonatal mortality rates by race: Michigan 1996-
2006 ................................................................................................................................ 134

xii

Table 4.7. Infant and neonatal mortality rates by hospital level at birth and race among
preterm infants: Michigan 1996-2006 .......................................................................... 136

Table 4.8. Gestational week speciﬁc infant mortality rates among level 1 births by race:
Michigan 1996-2006 ..................................................................................................... 138

Table 4.9. Gestational week speciﬁc infant mortality rates among level 3 births by race:
Michigan 1996-2006 ..................................................................................................... 139

Table 4.10 Gestational week speciﬁc neonatal mortality rates among level 1 births by
race: Michigan 1996-2006 ............................................................................................ 140

Table 4.11 Gestational week speciﬁc neonatal mortality rates among level 3 births by
race: Michigan 1996-2006 ............................................................................................. 141

Table 4.12. Multivariate analyses predicting the risk of infant death (white infants as the
referent) by hospital level at birth and neonatal intensive care unit transfer, Michigan

1996-2006 ...................................................................................................................... 143
Appendix A. Exposure variable speciﬁc adjustment for fully adjusted partial correlation
model 3 .......................................................................................................................... 155
Appendix B. Percentage of Borderline Birthweight Births and Deaths by Fetal Death
Classiﬁcation Area: United States, 2000-2002 .............................................................. 157
Appendix C. Percent of Borderline Infants Who Survive to Age 1 by Fetal Death
Classiﬁcation Area; United States, 2000-2002 .............................................................. 158
Appendix D. Percentage of Infant and Fetal Deaths with Unknown Birthweight and/or
Gestational Age by Fetal Death Classiﬁcation Area; United States, 2000-2002 ........... 159
Appendix E. Presence of Level 3 hospitals by year, Michigan 1996-2006 .................... 160

Appendix F. Age at death by hospital of birth level and hospital of death level among all
deaths, Michigan 1996-2006 ........................................................................................... 161

Appendix G. Concordance between hospital level at birth and hospital level at death
among extremely preterm (<28 weeks) neonatal deaths, Michigan 1996-2006 ............. 162

Appendix H. Concordance between hospital level at birth and hospital level at death
among extremely preterm (<28 weeks) white neonatal deaths, Michigan 1996-2006 ...163

Appendix I. Concordance between hospital level at birth and hospital level at death
among extremely preterm (<28 weeks) black neonatal deaths, Michigan 1996-2006 1 64

xiii

LIST OF FIGURES

Figure 1.1. Trends in infant mortality rates: United States, 1975-2000 ............................. 2

Figure 2.1. Infant mortality rates: United States, 2000-2002 ........................................... 39
Figure 2.2. Disparity ratios in infant mortality: United States, 2000-2002 ...................... 40
Figure 2.3. Disparity differences in infant mortality: United States, 2000-2002 ............. 41

Figure 2.4. Relative risk of low birthweight: United States, 2000-2002 .......................... 47
Figure 2.5. Risk difference of low birthweight: United States, 2002-2002 ...................... 48
Figure 2.6. Relative risk of very low birthweight: United States, 2000-2002 .................. 49
Figure 2.7. Risk difference of very low birthweight: Untied States, 2000-2002 .............. 50
Figure 2.8. Relative risk of teen pregnancy: United States, 2000-2002 ........................... 53
Figure 2.9. Risk difference of teen pregnancy: United States, 2000-2002 ....................... 54
Figure 2.10. Relative risk of poverty: United States, 2000-2002 ..................................... 58
Figure 2.11. Risk difference of poverty: United States, 2000-2002 ................................. 59
Figure 3.1. Fetal death registration area: United States, 1997 revision ............................ 85
Figure 3.2. Proportion of non-viable births: United States, 2000-2002 ............................ 87
Figure 3.3. Total infant mortality rates: United States, 2000-2002 .................................. 92
Figure 3.4. Disparity ratios in infant mortality: United States, 2000-2002 ...................... 93
Figure 3.5. Risk differences in infant mortality: United States, 2000-2002 ..................... 94
Figure 4.1. Rates of preterm birth: Michigan, 1996-2006 .............................................. 127
Figure 4.2. Hospital levels: Michigan, 1996-2006 ......................................................... 128

xiv

LIST OF APPREVIATIONS

Conﬁdence Interval ................................................................................. CI
Disparity Difference ............................................................................... DD
Disparity Ratio ..................................................................................... DR
Extremely Preterm Birth ....................................................................... EPTB
Infant Mortality ..................................................................................... IM
Low Birthweight ................................................................................. LBW
Michigan Department of Community Health .............................................. MDCH
National Center for Health Statistics ......................................................... NCHS
Neonatal Intensive Care Unit .................................................................. NICU
Neonatal Mortality ................................................................................. NM
New York City ................................................................................... NYC
Odds Ratio ........................................................................................... OR
Preterm Birth ...................................................................................... PTB
Relative Risk ......................................................................................... RR
Risk Difference ..................................................................................... RD
Sudden Infant Death Syndrome ............................................................... SIDS
Very Low Birthweight ........................................................................ VLBW
Very Preterm Birth ............................................................................. VPTB
United States ........................................................................................ US

XV

CHAPTER 1: INTRODUCTION

Improving our understanding of the burden and epidemiology of racial differences in
infant mortality (IM) is the primary focus of this research. A special emphasis will be
placed on statewide variation in racial differences in 1M rates at the national level by ﬁrst
focusing on differences between state level population characteristics and socioeconomic
factors and second to differences in fetal death reporting policies and practices. A case
study of perinatal regionalization in the State of Michigan and its impact on racial
differences in IM rates will also be presented.

1.1 Overview of the Burden and Epidemiology of Infant Mortality
1.1.1 Definition of Infant Mortality

The National Center for Health Statistics deﬁnes IM as any death which occurs
between birth and 364 days of age. The IM rate therefore is the number of deaths among
infants less that 1 year of age divided by the total number of live births and is expressed
per 1,000 live births. The US. IM rate in 2004 was 6.8 infant deaths per 1000 live births
[1, 2].

As with many diseases and conditions in the United States, there are racial differences
in IM rates. These racial differences are commonly referred to as disparities because the
term embodies both inequality (ie. differences in the health status between two groups)
and inequity (ie. systematic differences between two groups with different social
advantage). Although IM rates have declined for all races during the 20th century, there
are substantial disparities among racial groups [3]. Slower IM declines among blacks
compared to whites have led to a 25% increases in disparities in the US. [1, 2]. For

example, in 1980, the IM rate among whites in the United States was 8.7 per 1,000 live

births compared to an IM rate of 16.5 per 1,000 live births among blacks [4], meaning
blacks were 1.90 times as likely as white infants to die during the ﬁrst year of life. On the
other hand in 2005, the white IM rate was 5.7 per 1,000 live births compared to the black
IM rate of 13.6 per 1,000 live births, meaning blacks were 2.39 times as likely as whites

to die during the ﬁrst year of life (Figure 1) [5].

Figure 1.1. Trends in infant mortality rates (per 1,000 live births), United States
1975-2000.

30-..

 

8

 

_n
(I

 

_A
O

 

RupermeveBlrthl

 

 

 

1.1.2 Causes of Infant Mortality

The leading causes of IM among United States infants in 2004 are listed in Table 1.
In descending order, causes include: congenital malformations, conditions related to short
gestation, sudden infant death syndrome, maternal complications of pregnancy, accidents,
maternal complications of the placenta, respiratory distress syndrome, bacterial sepsis,

neonatal hemorrhage, and diseases of the circulatory system [6]. These 10 leading causes

of death accounted for 69% of all infant deaths in 2004. Studies have shown that vital

statistics data may underreport or incorrectly report variables such as cause of death [7]

and recommend cautious interpretation [8]. Common problems with cause of death data

include the reversal of underlying vs. proximate causes or if the cause of death is

unknown, sudden infant death syndrome (SIDS) is frequently used. For this reason,

Krous et a1 (2004) describe SIDS as a “term that has been used to describe unexpected

deaths of infants or young children when subsequent interventions fail to demonstrate a

deﬁnite cause of death” [9].

Table 1.1 Deaths and percentage of total deaths for the 10 leading causes of infant

death: United States, 2004.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cause of Death?“ Rank Deaths Percent
All causes] -- 27,936 100.0
Congenital malformations, deformation and chromosomal 1 5,622 20.1
abnormalities

Disorders related to short gestation and low birthweight, not 2 4,642 16.6
elsewhere classiﬁed

Sudden infant death syndrome 3 2,246 8.0
Newborn affected by maternal complications of pregnancy 4 1,715 6.1
Accidents (unintentional injuries) 5 1,052 3.8
Newborn affected by maternal complications of the placenta, 6 1,042 3.7
chord and membranes

Respiratory distress of newborn 7 875 3.1
Bacterial sepsis of newborn 8 827 3.0
Neonatal hemorrhage 9 6 1 6 2.2
Diseases of the circulatory system 10 593 2.1

 

*Based on the International Classiﬁcation of Diseases, Tenth Revision, 1992

'I'Data from the NCHS [6]

Differences in cause of death rates differ by race with black infants more likely to die

from disorders related to short gestation and SIDS and white infants more likely to die

from congenital malformations (Table 2).

 

Table 1.2. Racial differences in cause speciﬁc infant mortality rates: United States,
2004.

 

 

 

 

Mortality RateT

a:

cause 0f Death Total White Black RRT
All causes 677.5 566.1 1,359.6 2.4
Congenital malformations, deformation and 137.1 129.3 167.4 1.3

chromosomal abnormalities

 

Disorders related to short gestation and low birthweight, 112.1 77.1 297.2 3.9
not elsewhere classiﬁed

 

 

 

 

 

 

 

 

Sudden infant death syndrome 54.6 54.0 110.9 2.1

Newborn affected by maternal complications of 41.5 32.2 103.1 3.2
_pggnancy

Accidents 25.6 25.6 46.8 1.8

 

 

* Cause of death information is based on the International Classiﬁcation of Diseases,
Tenth Revision, 1992. Data are from the National Vital Statistics Reports [6].
1‘ Mortality rate per 1,000 live births
I RR=reIative risk (Black to White).
1.1.3 Timing of Death
Depending on the time at which IM occurs, it can be categorized as neonatal
mortality or postneonatal mortality. Neonatal mortality is deﬁned as infant death
between 1 and 27 days and is primarily related to maternal exposures and conditions
arising in pregnancy. As a result, low birthweight (<2500 grams) and very low
birthweight births (<1500 grams) account for the majority of deaths during this time
period [10]. Since birthweight is a representation of gestational age and grth in-utero,
prematurity and intrauterine grth restriction are two factors which contribute
signiﬁcantly to neonatal mortality [11]. At times, neonatal mortality is further dissected
into early neonatal mortality, deﬁned as deaths between 1 and 6 days of life.
Postneonatal mortality is deﬁned as infant death between 28 and 364 days of life.
Postneonatal mortality in developed countries primarily results from sudden infant death

syndrome, infections and homicide [12], but can also result from the late effects of

congenital malformations and preterm delivery. Although maternal exposures and

4

conditions may contribute to postneonatal mortality rates, these rates primarily reﬂect
infant exposures and experiences after birth. Postneonatal mortality may also include
effects of postponing deaths which would have occurred during the neonatal period but
were delayed because of medical intervention.

1.1.4 Maternal Behavioral Influences

There are a number of maternal factors that are associated with an increased risk of
IM, including little or no prenatal care, low socioeconomic position, alcohol use and
tobacco use during pregnancy [6]. These risk factors act independently and interactively
to inﬂuence antecedents to IM such as reduced fetal growth and premature birth, the
latter of which can lead to a twofold risk in IM [10].

The adequacy of prenatal care received by the mother, based on number and timing of
prenatal care visits relative to infant gestation, is associated with IM in a number of ways.
The American College of Obstetricians and Gynecologists recommend that all pregnant
women initiate prenatal care in the ﬁrst trimester and continue care at speciﬁed intervals
throughout pregnancy [13]. Research into the role of prenatal care in IM risk primarily
centers around whether prenatal care itself plays a part in reducing 1M risk or if reduction
in IM risk is due to characteristics of mothers who initiate prenatal care early [14]. For
example, Poma (1999) found that infants whose mothers did not receive prenatal care
were signiﬁcantly more likely to die than infants who received adequate prenatal care
(OR=4.07; 95% CI=3.74-4.43). Poma et al also found that mothers who did not receive
prenatal care were more likely to have other risk factors such as smoking and medical
complications [15]. Racial patterns prenatal care closely mirror those of IM rates with

blacks more likely than whites to have inadequate or late prenatal care. Furthermore,

when IM is examined by trimester of prenatal care initiation, blacks have the highest
rates of IM, regardless of timing of prenatal care [6].

Risky maternal behaviors such as smoking and alcohol use also put the infant at
increased 1M risk. Infants born to women who smoke during pregnancy have a 40%
higher IM risk than infants whose mother didn’t smoke [16]. Smoking can contribute to
IM risk through SIDS, fetal growth restriction and premature rupture of membranes. The
risk of each of these factors increases with the number of cigarettes smoked [16]. The
race speciﬁc effect of smoking is difﬁcult to understand because it differentially
inﬂuences IM risk across different racial groups. For example, among children of white
women, smoking during pregnancy is associated with an 85% increase in IM risk, while
it is only associated with a 53% risk among children of black women who smoke [17].
This racial difference in risk could be due to racial differences in the causal pathway from
smoking to IM.

Maternal moderate and heavy alcohol use are associated with poor pregnancy
outcomes such as fetal alcohol syndrome fetal growth restriction and preterm delivery
[18]. Self-reported alcohol use during pregnancy indicate that while overall rates are low
(0.8% in 2002) and similar between blacks and whites (0.9% and 0.8%, respectively)
[19]. One of the best protections that a mother can offer against poor pregnancy
outcomes such as low birthweight, preterm birth and IM is to actively plan for pregnancy

and enter into pregnancy with as few medical and behavioral risk factors as possible.

1.1.5 Importance of Understanding Racial Disparities in Infant Mortality: Racial
Composition

A health disparity is the quantity that separates a group from a speciﬁed reference
point on a particular measure of health that is expressed in terms of a rate, percentage,
mean or some other quantitative measure [20]. The term health disparity is almost
exclusively used in the United States, while the terms ‘health inequity’ and ‘health
inequality’ are more commonly used outside of the United States [21]. Reasons for this
difference in terminology primarily center on whether a judgment of what is avoidable
and unfair is included, and how these judgments are made.

In the US. public health ﬁeld, use of the term ‘health disparity’ generally takes on the
implication of injustice, but can still be separated from general inequalities in health. A
health disparity should be viewed as a chain of events characterized by differences in
environment, access to or quality of care or health status [21]. Whitehead deﬁnes seven
determinants of health disparities as a) natural, biological variation; b) health damaging
behavior that is freely chosen; c) the transient health advantage of one group over another
when one group is ﬁrst to adopt a health promoting behavior (as long as other groups
have the means to catch up fairly soon); d) health damaging behavior in which the degree
of choice of lifestyles is severely restricted; e) exposure to unhealthy, stressful living and
working conditions; i) inadequate access to essential health services and other basic
services; g) natural selection, or health related social mobility, involving the tendency for
sick people to move down the social scale [22]. The ﬁrst three categories in this

classiﬁcation are usually viewed as unavoidable and fair, while the last four categories

are viewed as avoidable and unfair. Racial inequalities in either of the categories (but
especially the last four) are the prime focus of health disparity research.

In addition to worldwide variation in terminology, concrete deﬁnitions and methods,
which are critical to the health disparity research ﬁeld, are the center of much debate. In
general health disparities are measured by comparing the health of one population
(disadvantaged) with the health of another population (referred to as the reference
population). Although a reference point can be arbitrarily chosen based on researcher
preference, choice of reference has important implications for intervention and policy.
Comparisons can either be made to groups (between- or within-group measures), relative
to a summary measure (population mean or standard deviation), or based on a standard
(ie. Healthy People 2010 goal) [20]. When between-group comparisons are made, the
reference can be the group with the largest number of people. This could be beneﬁcial
because the largest group would have the most stable rate. Between-group comparisons
could also be made the ‘best’ rate as the reference point (best could be deﬁned as the
highest or lowest rate depending on the outcome). Best rate comparisons may be
convenient because comparisons with all other groups would be in the same direction.
For within group comparisons, the reference is based on an internal scale of
measurement. This method could account for culturally relevant differences between
groups, but only provides an indirect measure of health disparity. In large, diverse
populations, the mean of the population could be used as a reference for a small group.
The mean would be less subject to variation over time, but it could be subject to outlying

population rates. Finally, a health standard or target could be used as a reference because

it doesn’t have sources of random variation associated with it. Despite the reference
choice, it has to be clearly speciﬁed in order for the disparity measure to have meaning.

While a number of disparity measures exist, each is based on its own assumptions and
therefore requires its own scientiﬁc interpretation [23-25]. A number of potential
disparity measures have been proposed but a clear consensus on the best type of
measurement (relative or absolute) is best has not been achieved [5, 20, 23-27]. Disparity
ratios (DR) are a commonly used relative measure for quantifying inequality between two
groups [20, 24-26, 28, 29]. In this research, the DR is the ratio of black IM to white IM
and its meaning can vary depending on the rates of those groups. For example, a high DR
can be due to a high black IM rate, a low white IM rate, or a combination of the two. In
contrast, a low DR can be due to a low black IM rate, a high white IM rate or a
combination of the two. In 1980, the black/white disparity ratio in IM was 1.90 and in
2005, the black/white disparity ratio was 2.39, but in order to gain meaning from these
DRs it is important to know the race-speciﬁc composition of the ratios and if it varies
over time.

The absolute disparity difference (DD) is the difference between the black 1M rate
and the white IM rate (ie. black IM — white IM). The DD reﬂects the actual (absolute)
size of the disparity by indicating how large a proportion of the disadvantaged group is
affected by the outcome. It can also be used to make comparisons across health
indicators, regardless-of unit of measurement [5, 20, 24]. From 1980 to 2005, the black-
white DD in IM was similar (ie. 7.8 excess deaths per 1,000 live births and 7.9 excess
deaths per 1,000 live births, respectively). When comparing the relative to the absolute

measure, the DR depends heavily on the baseline level of the measure of interest, while

the DD depends heavily on the overall rate of disease. Larger DDs are seen when overall
rates of disease are average and smaller differences are seen at the extremes [24, 26].
With the 1980 and 2005 IM rates, the same absolute decrease in white and black
mortality rates affects the DR but not the DD. Equally, the same proportion increase in
white and black mortality rates would be reﬂected in a change in the DD, but not in the
DR. In general, absolute and relative measures lead to similar results when applied to the
same measure at one point in time, but can lead to different conclusions when
comparisons are made at multiple points over time [21]. In an ideal situation, both the
difference and ratio would be measured simultaneously to enable the most meaningful
interpretation of health disparity data.

Currently, we don’t know if racial disparities in IM are composed of high black IM
rates, low white IM rates, or both. We also don’t know if there is geographic variation in
the racial composition of disparities in 1M across the US. In order to effectively reduce
racial disparities in IM, it will be important to determine the underlying race-speciﬁc
rates that comprise these disparities.

1.2 Three Key Exposures in Relation to Infant Mortality

State-level differences in IM have been apparent since the 19503 [30], with regional
and statewide variations within and between racial groups [4, 6, 29, 31-36]. While
studies have shown that southern states average the highest overall IM rates [29],
racial/ethnic disparities in IM are greatest among states in the north central region of the
US. [6]. In 2004, state-level DRs ranged from 1.42-3.48 with black infants nearly four
times as likely as their white counterparts to die during the ﬁrst year of life in some

states. Little is understood about the cause of these state-level differences in IM rates and

10

disparities. States vary in population risk characteristics, prevalence of LBW, and
geographic obstacles to delivery in care [4]; but these factors have not been well
examined in relation to disparities. The areas of sociodemographic factors, reporting of
fetal and infant deaths and perinatal regionalization have received little attention in
relation to state-level variation in 1M disparities; especially in relation to the racial

composition of disparities.

1.2.1 Social Influences

Socioeconomic factors such as education and income, differences in cultural norms
and practices, and institutionalized racism [3 7], are thought to be major contributors to
racial disparities in IM. Nearly parallel reductions in IM rates among all racial groups
coupled with increases in racial disparities, suggest that factors which inﬂuence racial
disparities in IM may differ from factors associated with IM rates over time. For
example, black infants are more likely to be born preterm or with a low birthweight than
white infants [3 8], which accounts for a large proportion of the black/white disparity in
infant mortality. But, while conditions related to short gestation account for a large
proportion of IM, black low birthweight infants have improved survival, compared to
their white counterparts [39]. In contrast, normal birthweight black infants are
considerably more likely to die before the age of one than their white counterparts [12].

Black/white differences in IM persist even after classiﬁcation on socioeconomic risk

factors. In 1992, Schoendorf et al found that black infants born to college educated

mothers have higher IM rates than white infants with similarly educated mothers [3 7].

ll

Furthermore, infants of black college educated women have higher 1M rates than do
infants of white women with less than a high school education [37].

Since the early 20th century, researchers have observed a link between standard of
living and morbidity and mortality [40]. The general theory states that social factors,
such as societal transitions in economic conditions, urbanization, improved sanitation,
and improvement in the status of women, have been primarily responsible for reducing
mortality rates [41]. A number of researchers have speculated that socioeconomic and
political empowerment may have a beneﬁcial impact on maternal health status, and that
the disparity in IM is a reﬂection of underlying political inequality [41, 42]. Based on
this hypothesis, if differences in mortality are reﬂections of powerlessness, where certain
groups (ie. minorities) have economic and political power, mortality rates should be
lower. Bird and Baurnan (1995) found that a substantial portion of the variance in state
level IM is accounted for by state structural characteristics such as the proportion black,

Proportion with greater than a high school degree and proportion with income below the
Poverty level [33]. Furthermore, this same study found that structural variables accounted
for more variance in state level IM rates than health services variables such as the number
0f physicians, proportion without health insurance and proportion receiving late or no
Prenatal care. States vary in structural sociodemographic characteristics; although racial
inequalities in these factors have not been well examined in relation to racial disparities
in infant mortality. In addition, comparisons of different disparity measures have not
been made in relation to state level disparities in IM. The interaction of socioeconomic

factors with race poses a challenge when separating the effect of these factors on racial

12

disparities in IM. Identifying populations at high risk of IM requires simultaneously
examining race and socioeconomic factors.

Critiques of state level, ecologic analyses of the effects of social factors on health
outcomes usually cite ecologic fallacies (ie. the conceptual model being tested
corresponds to the individual but the data used are in aggregate or group form), residual
confounding (ie. partial or incomplete control for variables of interest) or over controlling
(ie. controlling mediators in the causal pathway) as methodological issues. While these
critiques are relevant to a number of research questions, not all ecologic rsearch studies
attempt to make individual inferences based study results (ie. Bird et al, 1995) [33].
Furthermore, as with individual level analyses, ecologic analyses require methods to
ensure proper statistical control (ie. not adjusting for mediators on the causal pathway).
1 .2.2 Reporting Differences

Variation in the completeness and accuracy of reporting fetal and infant deaths can
influence racial disparities in both fetal and infant mortality rates [43]. Although all
States require the reporting of a live birth regardless of the length of gestation or weight,
there is considerable variation in reporting criteria for fetal deaths.

The 1992 revision of the Model State Vital Statistics Act and Regulations
rec(I’rnmended the following deﬁnition of fetal death: “. . .death prior to the complete
expulsion or extraction from its mother of a product of human conception. . .after such
expulsion or extraction the fetus does not breathe or show any evidence of life... ” [44].
IInprecision in recognizing or acknowledging very brief and faint signs of life may lead

to SYStematic variations in reporting a delivery as a live birth and subsequent infant death

Versus a fetal death [43]. Because there is only a recommended deﬁnition of fetal death,

13

systematic variation appears in state reported rates of very low birthweight fetal deaths.
Studies have examined the effect of reporting differences on proportion low birthweight
[45, 46], racial differences in perinatal mortality [43], and neonatal mortality [3 6].
Although there is conﬂicting evidence on the effect of reporting differences on state-level
differences in perinatal outcomes, Wingate (2006) found that there may be
underreporting of very low birthweight fetal deaths and recommended further analyses to
establish if black fetal death rates are underreported [43]. Due to the systematic
differences in state reporting of perinatal deaths, it is reasonable to suspect that
misclassiﬁcation of low birthweight infants as fetal deaths by certain areas (or vice versa)
, could lead to state differences in overall IM rates and/orracial disparities in IM.
1.2.3 Perinatal Regionalization

Extremely preterm infants born in hospitals with neonatal intensive care units (N ICU)
or transferred to such centers immediately after birth have lower mortality and morbidity
rates than comparable infants born in other settings [47]. The process of perinatal
regionalization involves a “regionally coordinated system focusing on levels of hospital-
based perinatal care” and has been shown to improves outcomes for both mothers and
newborns [48]. Perinatal regionalization incorporates the use of maternal and/or infant
transport services to ensure that low birthweight or at-risk infants are inborn or promptly
transferred to appropriate facilities (preferably with NICUs), regardless of where their
mother initially sought obstetrical care [49]. This system evolved to increase the number
of mothers and infants who had access to neonatologists, obstetricians and pediatricians.
Furthermore perinatal regionalization offered improved health care for mothers and

infants and has been adopted by many states and hospital systems [50].

14

In 1976, the March of Dimes Committee on Perinatal Health designated three levels
of perinatal care. The three basic levels as described in the latest American Academy of
Pediatrics (AAP)/American College of Obstetricians and Gynecologists (ACOG)
guidelines are as follows: level 1 hospitals are able to treat newborns without obstetric
complications and do not have a NICU; level 2 hospitals are able to treat moderately ill
newborns and they may or may not have a NICU; level 3 hospitals always have a NICU
and are able to treat the highest risk infants [48]. Recently, level 2 and level 3 hospitals
have been further categorized by factors such as the volume of extremely preterm infants
and/or the number of NICU beds [51]. Alternatively some states, such as Michigan,
recognize only two levels of care, bypassing the intermediate (level 2) category [52].

The effectiveness of perinatal regionalization can be determined by examining the
proportion of extremely preterm, low birthweight infants which are born at a level 3
hospital or by examining the proportion of extremely preterm, low birthweight infants
which are transferred to a NICU after birth [53-56].

Throughout the 19903, increases in the number of smaller, community NICUs with 1
or 2 neonatologists has led to breakdowns in the cooperative relationships between the
less specialized level land level 2 hospitals and the most specialized level 3 facilities
[57]. Furthermore, decreases in state funding of the regionalized transport of mothers and
infants as well as decreases in insurance funding of level 3 care (when less expensive
level 2 care is available) has led to de-regionalization in many states (ie. Michigan). This
de-regionalization is thought to have the largest effect on lower socioeconomic groups
with limited access to level 3 hospitals [3 9]. With the disproportionate representation of

racial/ethnic minorities in lower socioeconomic groups, perinatal de-regionalization

15

could impact on racial disparities in infant and neonatal mortality rates. In addition,
state-level differences in regionalization funding could account for state differences in
preterm or low birthweight IM rates. While some studies have examined the effect of
perinatal regionalization 1M rates, it is not clear whether racial differences in 1M can be
partially explained by differences in access to medical care for high risk neonates.
1.3 Signiﬁcance of Study

Understanding IM in general and racial disparities in IM in particular is necessary in
order to reduce these disparities. There is a lack of research examining the effect of
racial inequalities in social factors, statewide reporting and perinatal regionalization on
racial disparities in IM. Furthermore few examine these factors with respect to national
variation in IM disparities. This study focuses on the burden and epidemiology of
disparities in IM by examining on three important and timely exposures with the purpose
of inﬂuencing state and national policy. The distinction between this research and that of
others is the focal point of states as the unit of analysis. States were chosen as the primary
focus for three key reasons: ﬁrst, racial disparity in IM is a population level concept and a
critique of relative versus absolute disparities can not be done on an individual level.
Second, in order to examine reasons for national variation in IM rates and racial
disparities in IM, an examination of characteristics in aggregate allows for patterns,
which may not be apparent in individual level analyses, to be seen. Finally, the use of
states as the unit of analysis allows for a direct pathway between research and the
population level implementation of policy. In order to effectively reduce racial disparities
in 1M it is important to determine what is causing the state-level differences in the racial

composition of disparities in IM.

16

1.4 Speciﬁc Aims
In order to improve understanding of major contributions to national variation in
IM rates and racial disparities in IM, the aims of this dissertation research are as follows:

0 Speciﬁc Aim 1: Determine the inﬂuence of racial inequalities in infant,
maternal and state sociodemographic factors on racial disparities in IM. Both
relative and absolute disparity measures will be used to determine if results
are inﬂuenced by disparity measurement techniques.

0 Speciﬁc Aim 2: Determine the effect of fetal death registration requirements
and reporting of largely non-viable births on state level differences in infant,
early and late neonatal, postneonatal, and fetal mortality rates and racial
disparities in mortality rates.

0 Speciﬁc Aim 3: Examine the effect of perinatal regionalization (ie. hospital
level at birth and NICU access) on racial disparities in preterm infant and
neonatal mortality in Michigan.

1.5 Study Population and Data Sources

Vital statistics data has been used to track maternal and child health since the
early 19003 [58]. National vital statistics data ﬁles are compiled from data provided
through the Vital Statistics Cooperative Program and involves contracts between the
National Center for Health Statistics (NCHS), all 50 states, the District of Columbia and
New York City (which is separate from New York State for the purpose of death

registration.

17

1.5.1 Aim 1 Study Population and Data Source

Speciﬁc Aim 1: Determine the inﬂuence of racial inequalities in infant, maternal and
state sociodemographic characteristics on racial disparities in IM. Both relative and
absolute disparity measures will be used to determine if results are inﬂuenced by
disparity measurement techniques.

The source population for aim 1 included singleton live births and infant deaths from
2000-2002 among non-Hispanic, white (white) and non-Hispanic, black (black) infants
aggregated to each of the 50 states, Washington DC, and New York City (n=52
area3;10,999,362 live births; 66,566 infant deaths). The study population included live
births and infant deaths among non-Hispanic, white and non-Hispanic, black infants
aggregated to each of the states who had complete ascertainment of all variables of
interest and adequate race speciﬁc sample sizes within each state (n=4l areas; 10,586,415
live births; 64,502 infant deaths).

Numerator data on infant deaths were obtained from the NCHS U.S. Infant Death
Data Set for the years 2000-2002. Infant death data includes deaths to all infants in a
calendar year which can be linked to a birth certiﬁcate in the denominator ﬁle and

consists of information reported on death certiﬁcates, regardless of age. Fetal deaths are

not contained in mortality ﬁles. The death certiﬁcate includes demographic
Characteristics of the decadent, and underlying and contributing cause of death. The US.
death-registration system encompasses 50 States, Washington DC, and New York City
NYC (which is independent of New York State for the purpose of death registration). It is
believed that more than 99% of the births and deaths occurring in this country are

registered [59]. Coding is done by the NCHS and quality control for infant death data is

18

done by computer edit checks, code validations, and comparisons of tabulated data with
data for the previous year.

Denominator data on live births were obtained from the United States natality ﬁles,
which include data reported on the birth certiﬁcate shortly after the birth of a live-bom
infant. All live-bom infants in the United States are required to have a birth certiﬁcate,
regardless of viability or subsequent outcome (ie. infant death). Birth certiﬁcates are
ﬁlled out by the institution at which the birth occurred and are submitted to the state.
Data included on the birth certificate are maternal and paternal demographics, maternal
pregnancy and reproductive history, birthweight and gestational age at delivery, mode of
delivery and medical complications during delivery.

The Linked Birth/Infant Death Dataset is produced by the National Center for Health
Statistics (NCHS) [59]. Three years of cohort data were used in order to obtain sufﬁcient
case and live birth thresholds from which to produce stable IM rates (>19 deaths in the
numerator). Cohort data links infants born in one year to subsequent deaths regardless to
whether the death occurred during the birth year, or the following year [59]. Birth cohort
data ﬁles were preferred for detailed analyses because they follow a given cohort of

births for an entire year to ascertain mortality-speciﬁc information.

National birth and infant death records make use of state linked ﬁles for the
identiﬁcation of linked birth and infant death certiﬁcates and NCHS natality and
mortality computerized statistical ﬁles. When the birth and death of an infant occurs in
different states, copies of the records are exchanged by the state of death and the state of
birth in order for the record to be linked. In addition, if a third state is identiﬁed as the

state of residence at the time of birth or death, that state is also sent a copy of the

19

appropriate certiﬁcate by the state when the birth or death occurred. The NCHS natality
and mortality ﬁles, produced annually, include statistical data from birth and death
certiﬁcates that are provided to NCHS by states under the Vital Statistical Cooperative
Program. Data were coded according to uniform coding speciﬁcations, passed rigid
quality control standards, were edited and reviewed and are the basis for ofﬁcial US birth
and death statistics [59].

Additional data on sociodemographic characteristics of areas of residence were
obtained from the US. Bureau of the Census, 2000 Census of Population of Housing
(census). For each census dataset, summary ﬁle 3 (SF3) data were used to estimate state-
level (level 050) characteristics relating to natality, education, employment, and poverty.
Sampling was done by the housing unit and the sampling rate varied by census block to
account for differing population densities. For the US. census, the overall sampling rate
is l in every 6 housing units and approximately 95% of the total population was
enumerated. The overall ﬁnal response rate for the 2000 census was 67%. Imputation
methods were used by the census to account for non-response, although it is recognized
that black, Hispanic, and poor populations were undercounted.

1.5.2 Aim 2 Study Population and Data Source

Speciﬁc Aim 2: Determine the effect of fetal death registration requirements and
reporting of largely non-viable births on state level differences in infant, early and late
neonatal, postneonatal, and fetal mortality rates and racial disparities in mortality rates.

The source population for aim 2 included singleton live births, infant deaths, and fetal
deaths from 2000-2002 among non-Hispanic, white [22] and non-Hispanic, black (black)

infants (n=10,999,362 live births; 66,566 infant deaths; 154,119 fetal deaths). The study

20

population included live births and infant deaths among white and black infants who had
complete ascertainment of all variables of interest (n=10,980,512 live births; 66,569
infant deaths; 68,755 fetal deaths).

In addition to the United States linked birth/infant death dataset from 2000-2002, aim
2 also made use of US. Fetal Death Data. Fetal death statistics for every year are based
on all reports of fetal death received by the NCHS. Reporting requirements for fetal
deaths vary from state to state. Overall reporting is not as complete for fetal deaths as for
births and deaths, but is believed to be relatively complete for fetal deaths at 228 weeks
gestation or more [60]. National data on fetal deaths include fetal deaths occurring at a
stated or presumed gestation of 20 weeks or more. The fetal-death reporting system
encompasses the 50 States, Washington DC, and New York City. Coding is done by the
NCHS and quality control for fetal death data is done by computer edit checks, code
validations, and comparisons of tabulated data with data for the previous year.

1.5.3 Aim 3 Study Population and Data Source

Speciﬁc Aim 3: Examine the effect of perinatal regionalization (ie. hospital level at birth
and NICU access) on racial disparities in preterm infant and neonatal mortality in
Michigan.

The source population for aim 3 included singleton, preterm (gestational age < 37
weeks) live births and infant deaths from 1996-2006 among non-Hispanic, white [22] and
non-Hispanic, black (black) infants (n=111,671 live births). The study population
included preterm live births and infant deaths among white and black infants who had
complete ascertainment of all variables of interest (n=107,046 live births; 3,950 infant

deaths).

21

Live birth and infant death certiﬁcates were obtained from the Vital Records and
Health Data Development Section of the Michigan Department of Community Heath
(MDCH). The data were coded according to uniform coding speciﬁcations, passed rigid
quality control standards, were edited and reviewed and are the basis for ofﬁcial
Michigan birth and death statistics. Cohort mortality data for each of the years were used,
as opposed to period data, which links deaths of all infants born in a certain year,
regardless to whether the death occurred during the birth year, or the following year.
Birth cohort data ﬁles were preferred for these analyses because they follow a given
cohort of births for an entire year to ascertain mortality-speciﬁc information. Birth
certiﬁcates included information on hospital at birth, maternal demographic and
pregnancy characteristics and infant characteristics at birth.

The main exposure of perinatal regionalization was determined by two variables
obtained from the birth certiﬁcate: hospital level at birth (1 vs. 3) and infant transport to a
NICU (yes/no). Assignment of hospital level was based on the availability of a NICU at
any point during the 11 year study period. Hospitals which had a NICU during the study
period were designated level 3. Hospitals which did not have a NICU during the study
period were designated level 1. In rare instances where hospital level could not be
obtained from the MDCH, hospitals were contacted directly and asked if they had a
NICU at any point in time from 1996-2006. Three hospitals were contacted directly; one
of the three was designated level 3, white the other two were designated level 1. NICU
transfer was based on the birth certiﬁcate question which asked “Was the child
transferred to a neonatal intensive care unit”. Response options were yes, no and

unknown. Other secondary variables included maternal demographics (education, age,

22

smoking status, alcohol use, urbanicity and the Kessner prenatal care index) [61] and
infant demographics (sex and birthweight).
1.5.4 Strengths of Vital Statistics
The most obvious strength of vital statistics data is its comprehensiveness. Birth and
death registration are thought to be virtually complete, with records on >99% of all live
births and infant deaths in the United States [58]. Due to the completeness of vital
statistics data, they are less subject to the selection biases one could encounter in more
clinical study populations. Furthermore, since natality and mortality data have been
collected since the early 19003, it allows for multi-site comparisons over time. Detailed
information on demographic, geographic and medical factors also allows for population
stratiﬁcation with relative completeness of virtually all perinatal variables [7, 58].
Finally, vital statistics data are publicly available and relatively easy to use.
1.5.5 Limitations of Vital Statistics

Studies have shown that vital statistics data may underreport or incorrectly report
variables such as paternal education or cause of death. Variables with a low reliability or
a high proportion of underreporting (such as cause of death) were not used in our
analyses and therefore were not a source of bias for this study. With respect to this
research, other studies have provided evidence that obstetric procedures, complications of
labor and delivery and maternal and infant conditions have been underreported [7]. This
underreporting may not be random, especially for conditions associated with adverse
pregnancy outcomes. For example, maternal data may be less complete when a hi gh-risk
pregnant woman is transferred prior to birth, or an infant is transferred shortly after birth

[58]. Second, gestational age is based on the date of the last menstrual period. Potential

23'

problems with this measure include recall bias and misclassiﬁcation because of bleeding
post-conception [62]. With the 1989 birth certiﬁcate revision, a clinical estimate of
gestation improved gestational age misclassiﬁcation to some extent [58]. In response to
this limitation, aim 2 speciﬁcally examined state-level differences in the completeness
and accuracy of perinatal reporting among speciﬁc gestational age/birthweight groups.
Third, there could be clustering of maternal risk factors for twins and higher order
multiple births. Analyses excluded multiple births and only examined the effect of
sociodemographics, reporting, and perinatal regionalization among singleton births.
Finally, there are several limitations of the US. census data. Urban and minority
populations have been underrepresented in the census and the sampling scheme of the
SF-3 long form reﬂects similar issues. Nevertheless, the US. census data is recognized

as the most complete data source to enumerate population-based characteristics.

24

CHAPTER 2: THE EFFECT OF RACIAL INEQUALITIES IN
SOCIODEMOGRAPHIC FACTORS ON RACIAL DISPARITIES IN INFANT
MORTALITY, BY STATE

This chapter includes an examination of the effect of inequalities in sociodemographic
factors on state level differences in racial disparities in 1M and is study 1 for this three
paper dissertation option.

2.1 Abstract

Objective: To determine the effect of racial inequalities in infant, maternal and state
sociodemographic factors on nationwide variation in racial disparities in IM. We also
sought to determine if disparity/inequality measurement inﬂuenced ﬁndings.

Methods: National Center for Health Statistics, Division of Vital Statistics (NCHS)
live birth and linked infant death records among singleton, non-Hispanic white (white)
and non-Hispanic black (black) infants from 2000-2002 were used to calculate absolute
(black/white) and relative (black - white) disparities in infant mortality (IM). Racial
disparities in [M data were correlated, using the Pearson Correlation Coefﬁcient, with
racial inequalities in infant, maternal, and state sociodemographic factors. Inequalities in
infant and maternal factors were obtained from NCHS natality ﬁles, while inequalities in
state sociodemographic factors were obtained from the US. Bureau of the Census, 2000
Census of Population and Housing.

Results: Absolute and relative U.S. disparity measures were 6.99 per 1,000 live births
and 2.42, respectively. The absolute disparity measure was highly correlated with black
IM (r = 0.91) but not white IM (r = -0.03), while the relative measure was correlated with
both black 1M (r = 0.57) and white IM (r = -0.51). After adjustment for confounding

factors, relative inequalities in the proportion of very low birthweight births (VLBW)

25

exhibited the strongest correlation with relative disparities in 1M (r=0.75), while absolute
inequalities in low birthweight births (LBW) exhibited the strongest correlation with
absolute disparities in IM (r=0.67).

Conclusion: Absolute measures of disparity were strongly related to black IM rates,
while both black and white IM rates equally contributed to relative disparity measures. In
order to reduce state differences in racial disparities in IM, efforts should be made to
target women who are at high risk for a LBW or VLBW birth.

2.2 Introduction
2.2.1 Background on Infant Mortality

The infant mortality (IM) rate (number of infant deaths <1 year of age per 1,000 live
births) is commonly used to assess the health and well-being of populations [63]. In the
US, reducing the overall IM rate is consistent with the ﬁrst and second overarching
goals of Healthy People 2010. The ﬁrst goal is to increase years and quality of life [2,
64], and by reducing a person’s life by nearly 80 years, IM can be seen as the most
dramatic loss of years of life. From 1950 to 1991 the US. 1M rate declined an average of
3% per year [65]. However, despite this decline, the Healthy People 2010 target of 4.5
infant deaths per 1,000 live births has not yet been met.

2.2.2 Racial Disparities in Infant Mortality

The second goal of Healthy People 2010 is to eliminate health disparities [64].
Although IM rates have declined for all races during the 20th century, there are substantial
disparities among racial groups [3]. Slower IM declines among blacks compared to
whites have led to a 25% increases in disparities in the US. [1, 2]. For example, in 1980,

the IM rate among whites in the United States was 8.7 per 1,000 live births compared to

26

an IM rate of 16.5 per 1,000 live births among blacks [4], meaning blacks were 1.90
times as likely as white infants to die during the ﬁrst year of life. On the other hand in
2005, the white IM rate was 5.7 per 1,000 live births compared to the black IM rate of
13.6 per 1,000 live births among black infants, meaning blacks were 2.39 times as likely
as whites to die during the ﬁrst year of life.
2.2.3 Disparity Measurement

Related to the second healthy people 2010 goal is the question of how best to measure
racial disparities in health outcomes [20, 23, 25, 26]. While a number of disparity
measures exist, each is based on its own assumption and therefore requires its own
scientiﬁc interpretation [23-25]. A number of potential summary measures have been
proposed but a clear consensus on the best type of measurement (relative or absolute) has
not been achieved [5, 20, 23-27]. Disparity ratios (DR) are a commonly used relative
measure for quantifying inequality between two groups [20, 24-26, 28, 29]. In this
research, the DR is the ratio of black IM to white IM and its meaning can vary depending
on the rates of those groups. For example, a high DR can be due to a high black IM rate,
a low white IM rate, or a combination of the two. In contrast, a low DR can be due to a
low black IM rate, a high white IM rate or a combination of the two. In 1980, the
black/white disparity ratio in IM was 1.90 and in 2005, the black/white disparity ratio
was 2.39, in order to gain meaning from these DRs it is important to know the raCe-
speciﬁc composition of the ratios and if it varies over time.

The absolute disparity difference (DD) is the difference between the black IM rate
and the white IM rate (ie. black IM — white IM). The DD reﬂects the actual (absolute)

size of the disparity by indicating how large a proportion of the disadvantaged group is

27

affected by the outcome. The DD can also be used to make comparisons across health
indicators, regardless of unit of measurement [5, 20, 24]. In 1980 the DD in IM was 7.8
excess deaths per 1,000 live births, while the 2005 DD was 7.9 excess deaths per 1,000
live births. When comparing the relative to the absolute measure, the DR depends
heavily on the baseline level of the measure of interest, while the DD depends heavily on
the overall rate of disease. Larger DDs are seen while overall rates of disease are average
and smaller differences DDs are seen at the extremes [24, 26]. With the 1980 and 2005
IM rates, the same absolute decrease in white and black 1M rates affects the DR but not
the DD. Equally, the same proportion increase in white and black IM rates would be
' reﬂected in a change in the DD, but not in the DR.
2.2.4 Nationwide Variation in Infant Mortality and Racial Disparities in Infant
Mortality

State-level differences in IM and racial disparities in [M (relative and absolute
disparities) have been apparent since the 19503 [28], with regional and statewide
variations seen within and between racial groups [4, 6, 29, 31-36]. For example Allen
(1987) found that southern states had the highest total and white IM rates (12.1 and 9.8
per 1,000 live births, respectively) but the north central states had the highest black IM
rates [31]. Furthermore, Marks (1987) found that state-speciﬁc black/white DRs in IM
varied from 1.9 to 2.4 [4]. More recently, in their 2008 report, the Robert Wood Johnson
Foundation found that IM rates from 2000-2002 ranged from 4.6 per 1,000 live births in
Massachusetts to 11.0 per 1,000 live births in Washington DC [66]. While studies have
shown that southern states have the highest overall IM rates [67], little research has been

done on the highest disparity states. The factor(s) underlying state-level differences in 1M

28

rates are not well-understood and even less is known of state-level differences in racial
disparities. States vary in the prevalence of known risk factors for [M such as low
birthweight births [68] and adequacy of prenatal care [4]; although racial inequalities in
these risk factors have not been well examined in relation to nationwide variation in
racial disparities in infant IM. Furthermore, comparisons of different disparity measures
have not been made in relation to IM or risk factors for IM.
2.2.5 Study Objective

This study estimated the effect of racial inequalities in infant, maternal and state
sociodemographic factors on state level differences in racial disparities in IM. We also
determined if results were inﬂuenced by measurement techniques (relative vs. absolute)
used to describe inequalities in exposures (infant, maternal and state factors) and/or
disparities in outcome (IM).
2.3 Materials and Methods

We examined singleton live births and linked infant deaths to self-identiﬁed, non-
Hispanic white and non-Hispanic black (hereafter referred to as white and black) mothers
who resided in the US. at the time of their infant’s birth. Birth and death records missing
information on maternal race were omitted from the dataset and not used in this analysis.
2.3.1 Study Population

The study population included singleton live births and infant deaths from 2000-2002
among non-Hispanic, white [22] and non-Hispanic, black (black) infants aggregated to
each of the 50 states, Washington DC, and New York City (n=52 areas; 1 0,999,362 live
births; 66,566 infant deaths). The study population included live births and infant deaths

among non-Hispanic, white and non-Hispanic, black infants aggregated to each of the

29

states who had complete ascertainment of all variables of interest and adequate race
speciﬁc sample sizes within each state (n=41 areas; 10,5 86,415 live births; 64,502 infant
deaths).

2.3.2 Data Source and Preparation

Birth and linked infant death records were obtained from the 2000-2002 Linked
Birth/Infant Death Dataset, produced by the National Center for Health Statistics
Division of Vital Statistics (N CHS) [59]. Three years of cohort data were used to obtain
a sufﬁcient sample of infant deaths for the numerator and live births for the denominator
from which to produce stable IM rates. Cohort data links infants born in one year to
subsequent deaths regardless to whether the death occurred during the birth year, or the
following year [59]. Birth cohort data ﬁles were preferred over period data for this
analysis because they follow a given cohort of births for an entire year to ascertain
mortality.

National birth and linked infant death records make use of state natality data for the
identiﬁcation of birth and infant death certiﬁcates used in the NCHS computerized
statistical ﬁles. When the birth and death of an infant occur in different states, copies of
the records are exchanged by the state of death and the state of birth in order for a linkage
to take place. In addition, if a third state is identiﬁed as the state of residence at the time
of birth or death, that state is also sent a copy of the appropriate certiﬁcate by the state
when the birth or death occurred. The annual NCHS natality and mortality ﬁles include
statistical data from birth and death certiﬁcates that are provided by states under the Vital

Statistical Cooperative Program. Data were coded according to uniform coding

30

speciﬁcations, passed rigid quality control standards, were edited and reviewed and are
the basis for ofﬁcial US. birth and death statistics [59].
2.3.3 Outcome Variables

Region- and state-speciﬁc IM rates were calculated by dividing the number of infant
deaths by the total number of live births. All IM rates were expressed per 1,000 live
births. Crude IM rates were calculated, in addition to rates stratiﬁed by race. Births and
deaths were assigned to the birth state, regardless of if the infant death occurred in
another state. Regions were deﬁned based on the Census of Population and Housing
categories and included the northeast, midwest, south and west. States were deﬁned as
each of the 50 states, the District of Columbia and New York City, which reports
separately from New York State in the Vital Statistics Cooperative Program.

We employed both a relative measure, the disparity ratio (DR) and an absolute
measure, the disparity difference (DD) of racial disparities in I’M. In the case of no
disparity, the value DR took was one and the value DD took was zero. All DDs were
expressed per 1,000 live births.

To characterize relative and absolute measures of racial disparities in 1M and
determine which measure most reﬂected race-speciﬁc IM rates, racial composition scores
were calculated. We deﬁned racial composition as the race-speciﬁc IM ranking of each
state, compared to the rest of the US. Racial composition was determined as follows:
ﬁrst, race-speciﬁc IM rates were calculated for each state; second, black IM rates were
broken into tertiles with the highest 1/3 of black IM rates receiving the racial composition
score of ‘high’, the middle 1/3 of black IM rates receiving the racial composition score of

‘medium’ and the lowest 1/3 of black IM rates receiving the racial composition score of

31

‘low’; third, the same criterion was used for white infants. This resulted in a black racial
composition score (high, medium, or low) and a white composition score (high, medium,
low) based on the relative rank of each state’s race-speciﬁc IM rate. For example, if a
state has a white IM rate in the highest tertile, and a black IM rate in the middle tertile,
their racial composition score would be as follows: white=high; black=medium. Racial
composition scores were also grouped into quartiles and quintiles but tertiles were found
to correspond best with our data. To prevent unstable IM rates by race, eleven states
(Alaska, Hawaii, Idaho, Maine, Montana, New Hampshire, North Dakota, South Dakota,
Utah, Vermont, and Wyoming) were excluded from analyses for having fewer than 20
black infant deaths [6]. Thus, all rates presented had at least 20 deaths in the numerator.
2.3.4 Exposure Variables

Racial inequalities in several sociodemographic risk factors for IM were examined in
relation to racial disparities in IM. Racial differences in exposure variables were referred
to as ‘inequalities’, while the term ‘disparities’ was used to describe racial differences in
the outcome.
2.3.4.1 Vital Statistics Measures

Infant and maternal sociodemographic factors were collected from birth certiﬁcate
records for each state. Infant factors included information on gestational age, speciﬁcally
preterm birth (PTB) or infants born less than 37 completed weeks gestation and very
preterm birth (V PTB) or infants born less than 32 completed weeks gestation and
birthweight, including low birthweight (LBW) or infants born less than 2,500 grams, and
very low birthweight (VLBW) or infants born less than 1,500 grams. Maternal

characteristics included information on teen pregnancy (maternal age less than 20 years),

32

education (less than a high school diploma), marital status (unmarried), tobacco use
during pregnancy (yes), alcohol use during pregnancy (yes) and inadequate prenatal care
(Kessner index) [61]. The presence of a father on the birth certiﬁcate (as reﬂected by the
presence or absence of paternal age) was also included [69] and will hereafter be referred
to as a maternal factor because of its reﬂection of the social circumstances surrounding
the pregnancy and time shortly after birth. Inequalities in the proportion (%) of each
variable were computed for the total study population and in race-speciﬁc analyses.
2.3.4.2 Census Measures

State sociodemographic factors were obtained from the long-form records of the US.
Bureau of the Census 2000 census of population and housing (census). Census variables
examined in this study included inequalities in the proportion of the following population
variables: foreign born, education (less than high school diploma, 2 high school graduate,
and 2 college graduate; education categories were not mutually exclusive),
unemployment, and poverty status. Proportions (%) of each state level variable were
collected from the census for white and black women (includes Hispanic ethnicity) and
for the entire state (includes all races and ethnicities).

As we did for IM disparities, we expressed racial inequalities in sociodemographic
risk factors using two inequality measures: relative risk (RR), ie. black proportion/white
proportion; and the risk difference (RD), ie. black proportion — white proportion. The
RD was expressed as a percentage.

2.3.5 Data Analysis
Pearson correlation coefﬁcients and partial correlations (r) were used to examine the

correlation between relative and absolute racial disparities in IM and relative and absolute

33

racial inequalities in infant, maternal and state sociodemographic factors. Statistical
control for potential confounding variables was undertaken in the following manner:
adjustments were made for inequalities in confounding variables that were signiﬁcantly
associated with both the exposure inequality of interest and disparities in IM in univariate
analyses, but were not on the causal pathway. Furthermore, state proportions of the
exposure variable of interest (ie. total percent low birthweight) along with the proportion
of the state population which was black were also held constant in partial correlation
models. In the case of multicollinearity between multiple confounding variables, the least
signiﬁcant predictor of racial disparities in IM, was omitted from the model (appendix
table 1). Statistical signiﬁcance was determined by using the cutoff of p<0.05.
2.3.6 Mapping

Maps were used to visualize national differences in [M rates and racial disparities
in IM rates. We also mapped racial disparities in selected infant, maternal and state
sociodemographic factors. Maps were created using ArcMap version 9.3 within ArcGIS
version 9. Continuous data were split into 4 categories based on quantile classiﬁcation
within ArcGIS. Darker colors indicate higher proportions/rates, while lighter colors

indicate lower proportions/rates.

2.4 Results

2.4.1 Region- and state-speciﬁc number of live births and infant deaths by race
From 2000-2002, there were 10,5 86,415 births to women in the United States (Table

2.1). Of these births, 64,502 (0.6%) resulted in deaths before the infant’s ﬁrst birthday.

There were 8,832,933 (83.4%) births to white women and 1,753,482 (17.6%) births to

black women. Of these births, 43,596 white and 20,906 black infants died before the ﬁrst

34

year of life. In regional analyses the south had the highest number of births
(total=4,158,949; 39.3%; white=3,l91,515; 36.1%; black=967,434; 55.1%) and deaths

(totalé28,323; 43.9%; white=l6,685; 38.3%; black=l 1,638; 55.7%).

35

Table 2.1. Region and state speciﬁc number of live births and infant deaths by race:
United States, 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

State 1' Live Births Infant Deaths
Total* White Black Total* White Black
United States 10,586,415 8,832,933 1,753,482 64,502 43,596 20,906
Regions (n=4 regions)
Northeast (7 states) 1,753,552 1,441,778 311,774 9,400 6,171 3,229
Midwest (10 states) 2,428,673 2,090,076 338,597 15,563 10,921 4,642
South (17 states) 4,158,949 3,191,515 967,434 28,323 16,685 11,638
West (7 states) 2,245,241 2,109,564 135,677 11,216 9,819 1,397
States (n=41 stateﬂ
Alabama 174,51 1 1 18,652 55,859 1,429 698 731
Arizona 228,807 220,804 8,003 1,31 1 1,212 99
Arkansas 106,819 84,761 22,058 799 559 240
California 1,347,631 1,249,979 97,652 6,441 5,470 971
Colorado 186,1 17 177,472 8,645 978 872 106
Connecticut 116,622 101,543 15,079 617 424 193
Delaware 30,477 22,727 7,750 232 133 99
District of Columbia 21,435 7,269 14,166 227 33 194
Florida ' 578,397 442,072 136,325 3,736 2,159 1,577
Georgia 374,581 248,465 126,116 2,915 1,388 1,527
Illinois 505,409 409,642 95,767 3,340 2,018 1,322
Indiana 246,807 219,355 27,452 1,615 1,291 324
Iowa 106,377 102,723 3,654 551 513 38
Kansas 109,882 101,642 8,240 700 590 1 10
Kentucky 157,389 142,979 14,410 93 8 803 135
Louisiana 187,736 108,294 79,442 1,572 646 926
Maryland 201,139 130,408 70,731 1,408 597 81 1
Massachusetts 216,989 193,279 23,710 833 643 190
Michigan 372,086 304,31 1 67,775 2,582 1,590 992
Minnesota 181,669 168,085 13,584 822 683 139
Mississippi 121,868 66,841 55,027 1,135 413 722
Missouri 214,413 182,038 32,375 1,430 990 440
Nebraska 69,546 65,504 4,042 404 352 52
Nevada 84,1 87 76,963 7,224 463 373 90
New Jersey 302,632 243,245 59,387 1,631 954 677
New Mexico 68,743 67,259 1,484 390 368 22
New York 373,985 331,401 42,584 1,917 1,413 504
New York City 303,1 13 194,577 108,536 1,638 741 897
North Carolina 331,036 248,622 82,414 2,406 1,315 1,091
Ohio 430,981 364,032 66,949 2,978 2,043 935
Oklahoma 127,901 1 14,229 13,672 907 739 168

 

 

*Total denotes both black and white infants
1' Inclusion criteria: Singleton, Black and White, US born, sufﬁcient infant deaths (>19)

36

Table 2.1 (continued). Region- and state-speciﬁc number of live births and infant
deaths? by race: United States, 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

State Live Births Infant Deaths

Total White Black Total White Black
Oregon 123,230 120,429 2,801 614 588 26
Pennsylvania 405,378 346,157 59,221 2,562 1,826 736
Rhode Island 34,833 31,576 3,257 202 170 32
South Carolina 158,295 103,256 55,039 1,287 531 756
Tennessee 223,831 175,669 48,162 1,81 1 1,078 733
Texas 1,032,807 913,226 119,581 5,380 4,223 1,157
Virginia 270,913 206,345 64,568 1,71 l 965 746
Washington 206,526 196,658 9,868 1,019 936 83
West Virginia 59,814 57,700 2,1 14 430 405 25
Wisconsin 191,503 172,744 18,759 1,141 851 290

 

 

 

 

 

The excluded states (due to insufﬁcient black deaths) are shown in table 2.2.

Table 2.2. State-speciﬁc number of live births and infant deaths among excluded

states, United States 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

State I Live Births Infant Deaths
Total* White Black Total“ White Black
Alaska 19,871 18,582 1,289 89 86 3
Hawaii 12,861 11,435 1,426 79 67 12
Idaho 58,359 58,105 254 358 357 1
Maine 38, 680 38,260 420 1 75 I 73 2
Montana 2 7, 703 2 7, 5 79 124 I 65 1 64 1
New Hampshire 40, 707 40, 126 581 1 6 7 164 3
North Dakota 19, 74 6 1 9, 4 78 268 121 1 18 3
South Dakota 25,060 24, 754 306 124 124 0
Utah 134,115 133,150 965 610 606 4
Vermont 18, 3 68 18, 263 I 05 76 76 0
W oming 17,477 17,307 170 100 100 0

 

 

 

 

 

 

*Total denotes both black and white infants
'1 States were excluded due to insufﬁcient cell speciﬁc infant deaths (<19)

37

 

 

2.4.2 Infant mortality rates, disparity ratios, disparity differences and racial
composition scores

Table 2.3 shows IM rates, DRs, DDs and racial composition scores among live born
infants in the US, ranked by DR from lowest to highest. In comparison to the national
IM rate of 6.09 per 1,000 live births, region-speciﬁc IM rates ranged from 5.26 per 1,000
live births in the West to 7.53 per 1,000 live births in the South. State-speciﬁc IM rates
ranged from 3.84 per 1,000 live births in Massachusetts to 10.59 per 1,000 live births in
the District of Columbia. Among white infants, the IM rate was 4.94 per 1,000 live births,
and state IM rates among whites varied from 3.33 per 1,000 live births in Massachusetts
to 7.02 per 1,000 live births in West Virginia. Among black infants, the national 1M rate
was 11.92 per 1,000 live births and ranged from 8.01 per 1,000 live births in
Massachusetts to 15.46 per 1,000 live births in Wisconsin. The national DR was 2.42
while the national DD was 6.98 per 1,000 live births. Regional differences in infant
mortality rates and disparity measures were of interest, especially when examining the
southern and midwest regions. While the south had the highest overall 1M rates, the

midwest had the highest black IM rates and subsequent disparities.

38

Figure 2.1 Infant Mortality Rates, United States 2000-2002

    

Rate per 1,000 live births

i' . 3.34.529
.1 :3
Q ,3 5.30-6.32

- 6.33 -7.19
- 7.20- 10.59

155 .110 620 Miles

0

39

Figure 2.2 Disparity Ratios in Infant Mortality, United States 2000-2002

   

Disparity Ratio

1 V ‘7 1.65-2.08

rm -‘ '

L__j 2.09-2.37

- 2.38- 2.52

- 2.53-3.14 o 150 300 500M115

40

Figure 2.3 Disparity Differences in Infant Mortality, United States 2000-2002

    

\,’\
Disparity Difference per 1.000 live births \\

, 3.65-5.57
Z f: 5.58-6.92
- 6.93 -7.95 o 150 300 600 Miles
- 7.96- 10.53

Each state was also characterized by a racial composition score. Eight states, all but
one on the east or west coasts were in the lowest tertiles for both black and white racial
composition scores: Massachusetts, New York City, Washington, Texas, California,
Minnesota, New Jersey and Maryland. Five states, all but one located in the south, were
in the highest tertiles for both black and high white racial composition scores: Alabama,

Mississippi, Kansas, Ohio and Tennessee.

41

Table 2.3. Infant mortality rates*, disparity ratios‘], disparity differences]: and
racial composition scores§ by race: United States”, 2000-2002

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

State Infant Mortali Rate Disparity Disparity Racial Composition
Total White Black Ratio Difference White Black

United States 6.09 4.94 11.92 2.42 6.99
Regions (n=4 re ions)
Northeast 5.31 4.31 10.65 2.50 6.34
Midwest 6.15 5.22 13.01 2.50 7.79
South 7.53 5.59 12.19 2.22 6.60
West 5.26 4.96 11.36 2.28 6.40
States (n=41 states)
Arkansas 7.48 6.60 10.88 1.65 4.29 H L
Kentucky 5.96 5.62 9.37 1.67 3.75 H L
West Virginia 7.19 7.02 11.83 1.68 4.81 H M
Washington 4.93 4.76 8.41 1.77 3.65 L L
Rhode Island 5.80 5.38 9.82 1.82 4.44 M L
Oklahoma 7.09 6.47 12.29 1.90 5.82 H M
Oregon 4.98 4.88 9.28 1.90 4.40 M L
Louisiana 8.37 5.97 11.66 1.95 5.69 H M
Indiana 6.54 5.89 11.80 2.01 5.92 H M
Iowa 5.18 4.99 10.40 2.08 5.41 M L
Texas 5.21 4.62 9.68 2.09 5.05 L L
Mississippi 9.31 6.18 13.12 2.12 6.94 H H
Georgia 7.78 5.59 12.11 2.17 6.52 H M
New York City 5.40 3.81 8.26 2.17 4.46 L L
Delaware 7.61 5.85 12.77 2.18 6.92 H M
Alabama 8.19 5.88 13.09 2.22 7.20 H H
Arizona 5.73 5.49 12.37 2.25 6.88 H M
California 4.78 4.38 9.94 2.27 5.57 L L
Kansas 6.37 5.80 13.35 2.30 7.54 H H
Pennsylvania 6.32 5.28 12.43 2.36 7.15 M M
Florida 6.46 4.88 11.57 2.37 6.68 M M
Nebraska 5.81 5.37 12.86 2.39 7.49 M H
Massachusetts 3.84 3.33 8.01 2.41 4.69 L L
Virginia 6.32 4.68 11.55 2.47 6.88 L M
Tennessee 8.09 6.14 15.22 2.48 9.08 H H
Ohio 6.91 5.61 13.97 2.49 8.35 H H
Colorado 5.25 4.91 12.26 2.50 7.35 M M
Missouri 6.67 5.44 13.59 2.50 8.15 M H

 

 

 

 

 

 

 

*Infant mortality rates and disparity differences expressed per 1,000 live births
TDisparity ratios expressed are (black IM) + (white IM)
IDisparity differences expressed are (black IM) — (white IM)
§ Racial composition scores are broken into tertiles (H=High; M=Medium; L=Low)

H Inclusion criteria: singleton, Black and White, US born, sufﬁcient infant deaths (>19)

42

 

Table 2.3 (continued). Infant mortality rates*, disparity ratiosi, disparity
differences]; and racial composition scores§ by race: United States”, 2000-2002

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

State Infant Mortality Rate Disparity Dgparity Corlrigglseiiion
Total White Black Ratio D1 erence White Black

North Carolina 7.27 5.29 13.24 2.50 7.95 M H
Maryland 7.00 4.58 11.47 2.50 6.89 L L
Minnesota 4.52 4.06 10.23 2.52 6.17 L L
Nevada 5.50 4.85 12.46 2.57 7.61 L M
South Carolina 8.13 5.14 13.74 2.67 8.59 M H
New Mexico 5.67 5.47 14.82 2.71 9.35 M H
New York 5.13 4.26 11.84 2.78 7.57 L M
Michigan 6.94 5.22 14.64 2.80 9.41 M H
Illinois 6.61 4.93 13.80 2.80 8.88 M H
New Jersey 5.39 3.92 11.40 2.91 7.48 L L
District of
Columbia 10.59 4.54 13.69 3.02 9.15 L H
Connecticut 5.29 4.18 12.80 3.07 8.62 L M
Wisconsin 5.96 4.93 15.46 3.14 10.53 M H

 

 

 

 

 

 

 

 

 

2.4.3 Lowest and highest disparity ratios and disparity differences

To determine if high and low disparity states differed with respect to disparity
measure used, table 2.4 includes the ten lowest and highest DR and DD states with
accompanying racial composition scores. The lowest 10 DRs ranged from 1.65 to 2.08
with the lowest DR found in Arkansas. The ten highest DR states ranged from 2.57 to
3.14 with the highest DR in Wisconsin. Geographically the highest DR states were
primarily located in the northeast and midwest regions. The lowest DDs ranged from 3.65
per 1,000 live births in Washington to 5.41 per 1,000 live births in Iowa. The highest
DDs range from 8.15 per 1,000 live births in Missouri to 10.53 per 1,000 live births in
Wisconsin. The two states with the lowest disparity ratios (Arkansas and Kentucky) were
characterized by low racial composition scores among blacks and high racial composition

scores among whites. Disparity, measures categorized most of the same states as high

43

disparity states and low disparity states, but reﬂected differences in composition scores
with more consistent black IM rates seen with the absolute disparity measure.

The combination of racial composition scores and disparity measures shed new night
on highest and lowest disparity states. The evaluation of both racial disparities and racial
composition scores leads to conclusion that Washington has the lowest relative disparities
in IM and Massachusetts has the lowest absolute racial disparities in IM since they have
low DRs and DDs in addition to low black and low white racial composition scores. On
the other hand, Tennessee and Ohio would be the highest disparity states because they
have high disparity measures, high black and high white racial composition scores.
Furthermore, a state such as New Jersey, which has low black and low white racial

composition scores, would not be listed as a high disparity state.

44

Table 2.4. Lowest and highest disparity ratios" and disparity differences’r with
accompanying racial composition scoresi; United States 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Disparity Racral. Disparity Racial.
State Ratio Composrtron State Difference Composrtron
White I Black White LBlack
Lowest Disparity Ratios Lowest Disparity Differences
Arkansas 1 .65 H L Washington 3 .65 L L
Kentucky 1 .67 H L Kentucky 3 .75 H L
West
Virginia 1.68 H M Arkansas 4'29 H L
Washington 1 .77 L L Oregon 4.40 M L
Rhode Island 1.82 M L Rhode Island 4.44 M L
New York
Oklahoma 190 H M City 4.46 L L
Oregon 1 .90 M L Massachusetts 4.69 L L
Louisiana 1.95 H M West Virginia 4.81 H M
Indiana 2.01 H M Texas 5.05 L L
Iowa 2.08 M L Iowa 5.41 M L
Highest Disparity Ratios Highest Disparity Differences
Nevada 2.57 L M Missouri 8.15 M H
5°“th. 2.67 M H Ohio 8.35 H H
Carolina
New Mexico 2.71 M H SOU‘h. 8.59 M H
Carolina
New York 2.78 L M Connecticut 8.62 L M
Michigan 2.80 M H Illinois 8.88 M H
Illinois 2.80 M H Tennessee 9.08 H H
New Jersey 2.91 L L DlsmCt 9f 9.15 L H
Columbia
31mm?” 3.02 L H New Mexico 9.35 M H
olumbra
@mecticut 53.07 L M Michigan 9.41 M H
Msconsin 3.14 M H Wisconsin 10.53 M H

 

 

 

 

 

"‘ Disparity ratios expressed are (black IM) -=- (white IM)
1 Disparity differences expressed are (black IM) — (white IM) per 1,000 live births
I Racial composition scores are broken into tertiles (H=High; M=Medium; L=Low)

45

 

 

 

 

2.4.4 Racial inequalities among selected infant characteristics

Table 2.5 shows RRs and RDs of infant characteristics which are thought to account
for a large proportion of the black excess in IM. States were ordered according to DRs in
IM from low to high. When examining the relative inequality measures, U.S. RRs were
highest among VLBW births (RR=2.9) with the highest RR seen among VLBW births in
Michigan (RR=3.5). The lowest RRs were seen among PTB with a US. R of 1.6.
Compared to the other states, the lowest overall RRs were seen in Oregon (PTB=1.4;
VPTB=1.8; VLBW=2.4). Among absolute inequality measures in the US, the highest
RDs were seen among preterm births, which had a national average of 6.1% and ranged
from 3.3% in Oregon to 8.8% in the District of Columbia. RDs in very preterm births

ranged from 1.1% in Oregon to 3.6% in Alabama.

46

Figure 2.4 Relative Risks of Low Birthweight, United States 2000-2002

 

Relative Risk

, 1.80-2.00
VH3 iii
[gig 2.01-2.10
- 211-220 W
_2.2r-2.oo o 150 300 600Milas

47

Figure 2.5 Risk Difference of Low Birthweight, United States 2000-2002

   

Risk Difference per 1,000 live births

[; 4.10-5.40
[:13 5.41-6.10

{—T'T—T—I—T—T—m
-°'”‘°‘5° o 150 300 600M119:

- 6.51 .730

48

Figure 2.6 Relative Risk of Very Low Birthweight, United States 2000-2002

     

Relative Risk
[8 H 2.40-2.60

Eff; 2.61-2.90
- 2'91'3'20 r—r—r—r—r—r—r-r‘l
- 3.21-3.50 o 155 310 620 Miles

49

Figure 2.7 Risk Difference of Very Low Birthweight, United States 2000-2002

   

Risk Difference per 1,000 live births
f

, HLm-mo
Egjlshlm
IIIIJhLW FTTTTTTTW
o 150 300 600 Miles
IIIISLIN

50

Table 2.5. State-speciﬁc relative risks" and risk differences? of selected infant birth
certiﬁcate characteristics: United Statesi, 2000-2002

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Low Very Low
(:3?) Preterm very Preterm Birthweight Birthweight
RR RD RR RD RR RD RR RD
United States 1.6 6.1 2.5 2.5 2.1 5.9 2.9 1.7
Arkansas 1.7 6.9 2.6 2.8 2.1 6.2 2.7 1.7
Kentucky 1.5 6.0 2.2 2.3 1.9 5.3 2.4 1.4
West Virginia 1.6 6.6 2.4 3.0 1.8 5.7 2.7 1.9
Washington 1.5 3.6 2.2 1.5 2.0 4.3 2.6 1.1
Rhode Island 1.8 6.8 2.1 1.8 1.9 4.7 2.6 1.6
Oklahoma 1.5 5.6 2.1 2.1 2.1 6.1 2.6 1.4
Oregon 1.4 3.3 1.8 1.1 2.0 4.1 2.4 1.0
Louisiana 1.8 8.1 3.1 3.6 2.2 6.8 3.2 2.0
Indiana 1.6 6.4 2.3 2.4 2.1 5.6 2.7 1.5
Iowa 1.8 7.1 2.6 2.8 2.2 5.7 2.9 1.5
Texas 1.5 5.2 2.1 2.2 2.0 5.6 2.7 1.5
Mississippi 1.6 7.2 2.6 3.2 2.0 6.4 2.7 1.7
Georgia 1.6 5.2 2.5 2.1 2.1 5.9 2.9 1.7
Nf‘” Y0“ 1.5 4.6 2.0 1.8 1.8 4.2 2.4 1.3
City
Delaware 1.6 6.0 2.6 2.7 2.2 6.8 2.9 1.9
Alabama 1.7 7.7 2.9 3.6 2.1 6.2 2.9 1.9
Arizona 1.4 4.1 1.9 1.6 2.0 5.2 2.5 1.2
California 1.5 4.1 2.3 1.8 2.2 5.2 2.9 1.5
Kansas 1.6 5.0 2.2 2.1 2.1 5.4 2.8 1.6
Pennsylvania 1.8 7.2 2.8 2.9 ' 2.3 6.7 3.1 1.9
Florida 1.7 6.5 2.5 2.6 2.0 5.5 2.8 1.6
Nebraska 1.7 6.1 2.5 2.3 2.4 6.4 3.1 1.7
Massachusetts 1.6 4.5 2.4 1.9 1.9 4.2 3.3 1.6
Virginia 1.7 6.4 2.7 2.6 2.3 6.0 3.0 1.8
Tennessee 1.6 6.4 2.6 2.9 2.0 6.2 2.8 1.8
Ohio 1.6 5.9 2.4 2.6 2.2 6.3 3.0 1.8
Colorado 1.5 5.1 2.3 2.3 1.9 5.9 2.8 1.6
Missouri 1.8 8.1 2.7 2.9 2.2 6.1 2.9 1.7
North. 1.7 7.0 2.6 3.1 2.2 6.5 3.0 2.0
Carolina

 

 

 

 

 

 

 

 

 

* Relative risks expressed are (black proportion) + (white proportion)

1 Risk differences expressed are (black proportion) - (white proportion), expressed as the
excess proportion (%) of black infants affected by the characteristic

I States ordered from high 1M disparity ratio to low IM disparity ratio

51

 

Table 2.5 (continued). State-speciﬁcrelative risks* and risk differences] of selected
infant birth certiﬁcate characteristics: United Statesi, 2000-2002

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Low Very Low
(:32?) P‘e‘e‘m very Preterm Birthweight Birthweight
RR RD RR RD RR RD RR RD

Maryland 1.7 6.4 2.8 2.8 2.3 6.4 3.1 1.9
Minnesota 1.4 2.8 2.4 1.8 2.1 4.5 3.1 1.5
Nevada 1.6 6.2 2.4 2.3 2.2 6.1 2.8 1.4
30m“. 1.7 6.7 2.8 3.0 2.2 6.8 3.1 1.9
Carolina
New. 1.5 5.0 2.0 2.3 1.8 5.3 2.8 1.8
Mexrco
New York 1.8 6.4 2.7 2.6 2.3 6.1 3.4 1.9
Michigan 1.9 7.9 3.1 3.2 2.5 7.3 3.5 2.0
Illinois 1.8 7.6 2.8 3.1 2.5 7.3 3.5 2.0
New Jersey 1.8 6.8 2.7 2.9 2.3 6.2 3.4 1.9
mm“ 9f 2.0 8.8 2.6 3.4 2.5 7.8 3.1 2.1
Columbia
Connecticut 1.6 4.4 2.9 2.7 2.2 5.7 3.3 2.1
Wisconsin 1.9 7.8 3.1 3.3 2.6 6.9 3.4 1.9 '

 

 

 

 

 

 

 

 

 

2.4.5 Racial inequalities among selected maternal characteristics

Table 2.6 shows RR and RD of selected maternal sociodemographic factors. States
were ordered according to DRs in 1M from low to high. Large state variations in
inequality measures were seen in each of the maternal characteristics, with the largest
nationwide variation in RRs and RDs were seen within marital status (United States
RR=2.5; range=1.7 in New York City and Arizona to 3.6 in Alabama; RD=40.6%; range

24.% in Arizona to 57.8% in Wisconsin).

52

 

Figure 2.8 Relative Risk of Teen Pregnancy, United States 2000-2002

 
  

Relative Risk

g M, 1.10-1.70

ETE 1.71 -2.00

- 2.01-2.40

- 2.41-3.50 o 150 300 600 Miles

53

Figure 2.9 Risk Difference of Teen Pregnancy, United States 2000-2002

  

Risk Difference

[ '7 2.00. 7.20

.77—51

a; 721-920

- 9-11- ”-00 1-1—r-r-l—rﬁ—l—l
_11.01.1s_30 0 150 300 600M1lcs

54

008 35%? E >5— 8 0:8 .35de S: nwE Bot 0838 808m H
$0508 x83 .00 £0 newsman $088 2: 8 003298 Aaomtomoa 833v - Anomtaoa x876 oh @038de moozouobmd xma .P
€898.35 Said 1 Acomtomoa 0628 can @082de 82c 3020M *

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0d- 5d 0d N.N dd- dd 0.3» 0N Wm m._ w.mN Wm odd d; 32an
0.N m4 dd d; dd d._ NdN d; 5.07 md N.Nﬁ d.N m6 #4 mmEomsz
5m d; 5d d; 0.m mg ddN \iA d.N-- dd VAN N.N 5.0 v; maomtaoa
Nd m; md M: 0.3- md mdv 0.m 0.0 m; d.mm We dd 5.“ manna—ace
0d 0N Nd N.“ N.T dd ddv mN 0.0 v4 mdm d.N 0.N_ vN ounBﬂoQ
#4- dd md dew h; 54 mdN N; v; H; v.2 N.N w.m m; EU 3.8.» 302
h; m; _d- dd N0- ed dgv N.N N. T d; d.vN dim ms 5; «E506
dd M: ed d.N d.:- md d.Nm m.m 5w v; HAM we d.: M: Edmmmmmmuz
0d 0; _d- dd wd- dd 1mm N.N d.NT Nd 0AN d.N N.m v; make-_-
NA 0; m4 N.N are m._ 0.0aa N.N v.3 mN wdm We v.2 vN ~32
d.m m.N 0d d; m0- 5d d.mv 1N d.N. v; mdN d.m Nd_ d._ wedded
Nd v; md d.N mm- ed :3 d.N wea— dd mdm dd 0.: d.N «53:54
dd d._ md NA dd v; m.mm _N md d; E: d.N m8 5; nowouo
0.N mg dd d; Yv- hd 0dv vN dd d; N.N-N d..m dd 04 «Boast—O
d.N N.N ed- bd md- d._ va d.N ﬁx 0; mAN d..m 0.0 n; 283 0022
Ed w; Nd- dd Nd- d; ﬁmN d._ _A- dd d.N-H d.N We 0; :3m:E£3
d.N «.4 0d m.N 0d d; d.mv TN N.N _A 0NN m.m d.N. 0A «Ewwh> #33
dd d._ 0d 0; N.N- dd 0.30 0.N We N; de m.m N.N. mg emu—03:3—
0.d- wd 0d M: ﬁdv d.N dd N._ mém 5m N-d_ h; mums-£2

825m 33::

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.NddN-dddN #nouﬁm
03:5 "noﬂﬁuuoﬁano 83523 5.3 ie.—8a:— 08923 we too-53.5.. it can 3.3—m... 95.29.. uﬁoonntouﬁw .0.N 035—.

55

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

d.m 0M: m.m £28835
dd dd m.N 3282500
md :2 Wm «58200 do 85me
Wm dd d.N Nome»; 302
0.0 0.: 0N £95:
m0 mdd N.N SwEomE
dd md~ 0.N xuo> 302
n; d.N : 02on 262
d m dd ad «5830 5:8
d.N dd n; 3652
d.m _dﬁ 0N «83:52
d m Nd d.N dang-82
dd dd M: 25030 552
Nd d.: d.N 05822
N. m _N 0. L 88200
md 54 _ N.N oEO
d.m dd 5; commence-_-
0.m Ndﬁ mN «Emma;
dd d0 d .N 6.38308me
. . 8718002
«ESE

swat/Raced

 

 

 

.nedu-ddcu #825 e825 3903.89.23»
835:3 5.5.— EEouaE 03923 no toes—ouch? it and ...an 9530.. 95925335 .2.2-£2.80 0.N 93.:-

 

56

2.4.6 Racial inequalities in selected state characteristics

Table 2.7 includes RS and RDs of selected state sociodemographic factors.
Relative inequality measures, demonstrate that blacks were more likely than whites to be
foreign born, less than high school educated, unemployed and in poverty
(RR = 12,12, 2.5, 2.7, respectively.) Absolute inequality measures demonstrate that
blacks were less likely than whites to have a high school degree, college degree, or be
employed (RD = -9.6%, -12.0%, -4.0%). Between states, there was little variation in
inequality measures for the proportion with greater than a high school degree and
employment status when measured with the RR; both categories ranged from 0.7 to 1.0.
In contrast, variation in RD among proportion with greater than a high school degree
ranged from -24.0% to 1.5%. Also of interest was the trend in education inequalities with
inequalities increasing as educational level increased. The highest racial inequalities

were in the proportion college educated (RR=O.5%; RD = 42.0%).

57

Figure 2.10 Relative Risk of Poverty, United States 2000-2002

 

  
  

Relative Risk

 

L 7 7: 1.60-2.40

5‘1 "*3 2.41 -2.80

- 2.81-3.30 W
- 3.31-4.90 0 ISO 300 600 Miles

58

Figure 2.1] Risk Difference of Poverty, United States 2000-2002

     

Risk Difference

'1 1 8.70- 12.30
E533 12.31-15.20
- 15.21-19.10
- 19.11-25.60 o 150 300 600M1'lcs

59

038 rooﬁng 2H 32 8 can“ mum-8&3 EH nwE 82m 3330 mood—w H
08% :23 E ocmtouogo 05 3 deuce-Cm 383 we Aeov condone.“ 308$
06 mm comm-“:98 .Acootomoa 8326 I Ego-Snood xoﬁa you 33298 mooaooobmv xmg .o
Eon-885 833v + Anomtomoa xoﬂa 3m 3&2me 86.: 03.20% .._

 

 

 

   

 

 

 

  

 

  

 

   

 

 

 

  

 

   
  

 

 
 

 

 
 

 

   
   

 

  
  

 

  

 

  

 

  

 

  

 

  

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2: 2 o- do 3 7 do do 2 2 d_ agaoéﬁ
d: oﬂ- do do- do do 3 do 2 moo-52
d2 do;- do do- o; do- do o2- do 2555
do- do do- do do 3 do- do 832

do- do 2 o- do do 2 do- do “can“?

doo- do oom- do do doo- do 03328

d_ _- do do- do do I- do dosed

do- do 2:. do do do- do 392%:

doo- do do- . do- do 83

do- do do- do 2. ~32

do- do do- do- do «Eds

o.o_- do 2- do «ﬁasco

as- d-o . :owEO

do- do oeoﬁzo

do: do 232 woe:

do- do Sowing?

do- do 2&5 a»?

do- do 393.3

do- do $503..

mound—m ﬂow—ED

 

 

 

.NceN
-..ch .Hnouaum coat—D "moratouoahaﬁ 25:3 33'. 633—3 .3 +895qu6 an... E... Low—3.. 9559.. oEuoantouﬁm .o..N 033—.

 

60

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m.mN ad on: dd- wd 53085»
5.2 e. m a6 5280:qu
m.N~ QM ad «5:330 mo auto-m5
o-N~ m.m o6 thumb—- 262
ﬁg Wm ado moons:
5.: . swig:
Nd“ zoo-w >5 Z
od 02on >62
”2 2:030 £50m
M: o ado-$02
adN 880552
dd wan—>52
de ammo-EU £52
QE EOmmmE
5w owﬁo~o0
mo: oEO
WE ammo-madca-
m.N_ «meow-“S
Nd M wtomsnommmmz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.Noonéooo #825
sea—D Serum-333:0 32.3 35» @3923. no toque-.056 an! and day—mm.— o>mu£ou oEuoo—nuouﬁm .GoanﬂncOV d.N cash.

61

2.4.7 Correlation between infant mortality, disparity ratios and disparity differences
To determine which measure of IM disparity was more correlated with race-speciﬁc
IM rates, table 2.8 shows Pearson correlation coefﬁcients between IM rates, DRs and
DDs. DRs were signiﬁcantly correlated with both white and black IM rates
(white IM r=-O.51; black IM r=0.57). While DDS were positively correlated with black
IM rates (r=0.91) but, were not at all correlated with white 1M rates (r=0.03). Relative
and absolute disparities consistently reﬂected black IM rates, while the relationship with
white IM rates was not as consistent. As expected, DRs and RDs were signiﬁcantly

correlated with each other (r=0.86).

62

Table 2.8. Correlations* between infant mortality‘i, disparity ratios: and disparity
differences§: United States, 2000-2002.

 

 

 

 

 

 

 

 

All States
Infant Mortality . . (n=41 states) . .
Disparity Disparity
Ratio Difference
Total 0.01 0.33
White -0.51 -0.03
Black 0.57 0.91
Disparity Ratio 1.00 0.86
Risk Difference 0.86 1.00

 

 

 

* Pearson correlation coefﬁcients (r)
‘1' Infant mortality rates expressed per 1,000 live births
I Disparity ratios expressed are (black IM) + (white IM)
§ Disparity difference is expressed as (black IM) — (white IM)
Bold indicates signiﬁcant correlation at 0.05 level

The next three tables determined which sociodemographic inequalities were most
correlated with racial disparities in IM. They also examined if Pearson correlation
coefficients between inequalities in exposures and disparities in outcomes differed by
disparity measure used. Model 1 shows the unadjusted Pearson correlation coefﬁcient,
model 2, shows adjustment for the total (white and black) state-speciﬁc proportion of the
exposure variable of interest and model 3 shows full adjustment for all confounding
variables, along with the percent of each state that is black.
2.4.8 Correlation between inequalities in infant antecedents and disparities in infant
mortality

Table 2.9 shows correlations between inequalities in infant sociodemographic factors
and disparities in IM. While both absolute and relative IM disparity measures were
positively correlated with a number of infant sociodemographics, inequalities in VLBW

were considerably more correlated with DRs and DDs in fully adjusted models than the

other infant factors (RR r=0.75 and 0.60, for DR and DD, respectively; and RD r=0.63

63

 

and 0.57 for DR and DD, respectively). Surprisingly, the LBW RD was not signiﬁcantly
associated with either IM disparity measure in fully adjusted models. Although both the
DR and DD were signiﬁcantly correlated with the same infant inequalities, the DR was
most correlated with inequalities in infant sociodemographic measures.

Table 2.9. Correlations. between proportion of infant antecedentsi', disparity ratios:
and disparity differences§: United States, 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

Relative Absolute
Infant Antecedent Disparity Ratio Disparity Difference
Inequality Model Model Model Model Model Model
1“ Zﬂ 3'" l 2 3
Preterm Ratio , 0.43 0.24 0.10 0.46 0.32 0.05
Preterm Difference 0.22 0.17 0.06 0.42 0.30 0.02
Veg Preterm Ratio 0.49 0.38 0.53 0.53 0.43 0.36
Very Preterm Difference 0.37 0.40 0.51 0.55 0.49 0.41
Low Birthweight Ratio 0.61 0.53 0.58 0.60 0.61 0.54
Low Birthweight Difference 0.46 0.52 0.65 0.67 0.67 0.67
Very Low Birthweight Ratio 0.75 0.70 0.75 0.62 0.63 0.60
V?” LOW B‘m‘we‘gh‘ 0.56 0.57 0.63 0.65 0.62 0.57
Difference .

 

 

 

 

 

* Pearson correlation coefﬁcients (r)

1 Infant antecedents are from all black and white infants, singleton infants born during
the study period and are aggregated to the state

I Disparity ratios expressed are (black IM) + (white IM)

§ Risk difference is expressed as (black IM) — (white IM)

Bold indicates signiﬁcant correlation at 0.05 level in fully adjusted models

|| Model 1: Unadjusted Model

1] Model 2: Adjusted for the proportion of each variable in the total (white and black)
population

** Model 3: Additionally adjusted for proportion black and confounding inequalities in

maternal and state sociodemographic risk factors

64

 

2.4.9 Correlation between inequalities in maternal characteristics and disparities in
infant mortality

Table 2.10 shows correlations between inequalities in maternal sociodemographic
factors and disparities in IM. Despite signiﬁcant correlation in unadjusted models,
inequalities in maternal factors were not signiﬁcantly correlated with disparities in 1M (as
measured by the DR or DD). While inequalities in teen pregnancy and the proportion
with unknown fathers were correlated with IM disparity measures in unadjusted models 1

and 2, they failed to reach statistical signiﬁcance in adjusted model 3.

65

Table 2.10. Correlations. between proportion of maternal birth certificate
characteristicsi', disparity ratiosi and risk differences§: United States, 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Relative Absolute
. Disparity Ratio Disparity Difference
Maternal Inequality Model Model Model Model Model Model
1” 21] 3** 2 1 2

Teen Pregnancy Ratio 0.62 0.55 0.27 0.48 0.56 0.31
T?“ Pregnancy 0.36 0.34 0.18 0.47 0.47 0.27
Difference
Unknown Father Ratio 0.33 0.36 0.32 0.31 0.36 0.20
Uf‘kmwn Father 0.21 0.13 -015 0.40 0.24 -0.08
Difference
< High School Ratio 0.31 0.21 0.06 0.23 0.27 0.00
< Hijh School Difference 0.23 0.09 -0.21 0.26 0.31 -0.04
Unmarried Ratio 0.27 0.32 0.10 0.38 0.40 0.05
Unmarried Difference 0.16 0.20 -0.14 0.33 0.35 -0.08
Tobacco Ratio 0.33 0.35 -0.13 0.21 0.19 0.00
Tobacco Difference 0.31 0.23 0.01 0.08 0.06 0.00
Alcohol Ratio 0.05 0.12 -0.02 -0.04 0.21 0.06
Alcohol Difference 0.1 1 -0.01 -0. 14 0.21 -0.06 -0.05
”3969”“ Prenatal ca“ 0.18 0.17 0.19 0.17 0.17 0.14
Ratio
Infdequa‘e Prenatal care 0.24 0.17 0.21 0.09 0.10 0.08
Difference

 

 

 

 

 

* Pearson correlation coefﬁcients (r)
T Maternal characteristics are from all black and white women who gave birth to a
singleton infant during the study period and are aggregated to the state
I Disparity ratios expressed are (black IM) + (white IM)
§ Risk difference is expressed as (black IM) — (white IM)
Bold indicates signiﬁcant correlation at 0.05 level in fully adjusted models

ll Model 1: Unadjusted Model

1] Model 2: Adjusted for the proportion of each variable in the total (white and black)

population

*"' Model 3: Additionally adjusted for proportion black and confounding inequalities in
maternal and state sociodemographic risk factors

66

 

2.4.10 Correlation between inequalities in state characteristics and disparities in
infant mortality

Table 2.11 demonstrates correlations between inequalities in state sociodemographic
factors and disparities in IM. Inequalities in foreign born status were negatively
correlated with both relative and absolute disparity measures in IM, with the largest
correlation seen between R in foreign born status and DD in IM (r =-0.48). This
negative correlation indicates that as racial differences in the proportiOn foreign born
increases, racial disparities in IM decrease. Inequalities in education, unemployment
status, and percent poverty were not signiﬁcantly correlated with racial disparities in IM

after adjustment for confounding variables.

67

Table 2.11. Correlations‘' between proportion of state census characteristicsi',
disparity ratiosl]: and disparity differences§: United States, 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

' Relative Absolute
. Disparity Ratio Disparity Difference
State Inequahty Model Model Model Model Model Model
H] 21] 3** l 2 3
Foreign Born Ratio -0.06 -0.05 -0.36 -0.29 -0.30 -0.48
Foreign Born Difference , -0.00 0.01 -0.26 —0.26 -0.26 -0.41
< High School Ratio 0.02 -0.04 -0.29 0.05 0.01 -0.31
< High School Difference 0.08 0.03 -0.19 0.13 0.10 -0.13
2 High School Ratio -0.18 -0.20 0.11 -0.20 -0.21 0.11
2 High School Difference -0.33 -0.36 -0.11 -0.34 -0.35 -0.10
2 College Ratio -0.39 -0.31 -0.32 -0.40 -0.40 -0.29
2 College Difference -0.18 0.10 0.14 -0.15 -0.13 -0.03
Unemployed Ratio 0.28 0.20 -0.26 0.32 0.31 -0.05
Unemployed Difference 0.12 0.13 -0.39 0.21 0.22 -0.17
Poverg Ratio 0.42 0.26 0.09 0.33 0.27 0.15
Poverty Difference -0.1 1 -0.04 -0.31 0.03 0.07 -0.10

 

 

 

 

 

 

* Pearson correlation coefﬁcients (r)

1 State characteristics are state-level census characteristics which represent all races, ages
and both sexes in a state

I Disparity ratios expressed are (black IM) -=- (white IM)

§ Risk difference is expressed as (black IM) — (white IM)

Bold indicates signiﬁcant correlation at 0.05 level in fully adjusted models

I] Model 1: Unadjusted Model
1] Model 2: Adjusted for the proportion of each variable in the total (white and black)

population

** Model 3: Additionally adjusted for proportion black and confounding inequalities in
maternal and state sociodemographic risk factors

2.5 Discussion

2.5.1 Nationwide variation in infant mortality rates

Considerable nationwide variation in IM was seen during the study period. The

highest IM rates were found in the southeastern part of the US. with the District of

Columbia and Mississippi having rates twice the national average. Our ﬁnding of

regional patterns in IM was similar to those of other studies which examined state level

differences in infant mortality [4, 32, 35, 36]. High southern IM rates such as those seen

68

in our results led the Southern Govemor’s Association to identify IM as a focus area for
increased effort [4].
2.5.2 Racial disparities in infant mortality

State level differences in IM were even more pronounced when examining race-
speciﬁc rates. Among white infants, southern states again had the highest IM rates, while
black IM rates were highest in the north and midwest. The highest disparities were
primarily found in the northern states with black infants in Wisconsin, Connecticut and
the District of Columbia three times as likely to die as their white counterparts. This
amounted to an excess of 10 deaths per 1,000 live births among black infants in the
highest disparity states. The DR as a measure of relative disparity and the DD as a
measure of absolute disparity had similar, but not identical, results when identifying the
highest and lowest disparity states. Among those identiﬁed as low disparity states, DDs
included more states with low IM rates (ie. New York City, Massachusetts and Texas).
On the other hand, categorizing disparity based on the DR allowed states with high IM
rates to be counted as low disparity states (ie. Louisiana and Oklahoma).

Racial composition scores were most predictive of absolute disparities and more
closely followed black IM rates than white IM rates. According to racial composition
scores for the lowest and highest disparity states, low disparities primarily reﬂected low
black IM rates (as opposed to high IM white rates), while high disparities primarily
reﬂected high black IM rates (as opposed to low white IM rates). To our knowledge, this
use of racial composition scores to characterize nationwide variation in disparity across
had not previously been examined.

2.5.3 Racial inequalities in sociodemographic factors

69

Similar to other studies of racial disparities in socioeconomic factors [26, 28, 37, 70-
72], our study found striking racial inequalities in socioeconomic factors at the infant,
maternal and state level. Similar to other studies, inequalities in gestational age and
birthweight categories were found to vary by state [29, 31, 68] with the highest rates
found in the south but the highest racial disparities found in the midwest. Racial
differences in infant factors also varied by inequality measure used [10]. Furthermore, we
also found that the prevalence of each infant factor inﬂuenced the magnitude of relative
and absolute inequalities in different ways. Relative inequalities were more pronounced
for small proportions (ie. VPTB and VLBW) while absolute inequalities were larger
when overall rates were larger (ie. PTB).

Among maternal risk factors, the largest inequalities were seen in the proportion of
teen pregnancies [73] and the proportion unmarried [10]. Although variation in maternal
inequalities were seen by state, comparable inequality patterns emerged between relative
and absolute measures. Other studies which examined maternal risk factors have found
racial inequalities in maternal age, education, income and prenatal care among other
factors [10, 28, 37] although these have not been well examined on a national level.

Similar to other studies, our study found racial inequalities in state level census
factors [28, 70-72]. Nationwide variation in inequalities were seen for each of the state
factors, with the largest ﬂuctuations seen among relative inequalities in the proportion
with greater than a college degree and the percent poverty. As with maternal
sociodemographic factors, studies have examined racial differences in these state-level
census variables, but state—speciﬁc racial differences have not been well examined on a

national level.

70

2.5.4 Correlation between disparity measures and infant mortality

Racial composition score ﬁndings were reinforced when examining correlations
between IM rates and disparity measures. While relative and absolute disparity measures
were signiﬁcantly correlated with each other, they exhibited different relationships with
IM rates. Absolute measures of disparity were strongly correlated with black IM rates,
while both black and white IM rates equally contributed to relative disparity measures.
Results indicated that the absolute disparity measure more closely reﬂected black IM
rates than did the relative disparity measure.
2.5.5 Correlation between inequalities in exposures and disparities in outcome

Racial inequalities in infant characteristics were strongly correlated to both relative
and absolute disparity measures in IM. Numerous studies have shown robust associations
between preterm/low birthweight birth and risk of infant death and between preterm/low
birthweight birth [1 l] and racial disparities in infant mortality [46, 52]. In addition, the
proportion of VLBW births has long been correlated with national differences in IM rates
[74]. While inequalities in both VPT and VLBW births were signiﬁcantly correlated with
disparities in IM, of interest were differences in the magnitude of prediction for each of
the infant factors. Inequalities in the proportion of VLBW births were the strongest
predictor of nationwide variation in racial disparities inIIM. Although other studies have
found strong relationships between the proportion of VLBW births and IM rates [74], it is
unclear why inequalities in VLBW births would be a stronger predictor than inequalities
in VPTB. A possible reason is misreporting of gestational age reporting versus the

reporting birthweight in vital statistics data [62].

71

The ~200 gram difference in birthweight [75] between blacks and whites could have
something to do with the higher correlation between inequalities in VLBW and
disparities in IM but there is also a ~6 day difference in gestational age [76] between the
races as well. Or, the effect of inequalities in the proportion of very low birthweight
births could just be the strongest predictor of racial disparities in IM. Despite reasons for
differences in the magnitude of the effect, VLBW births and VPTB are still the largest
risk factors for IM and should still be prevented.

Other than its association with preterm birth, the relationship between race, low
birthweight and risk of [M is not a straight forward one. Populations with a higher
percent of LBW births often have higher rates of IM. But, LBW infants from these same
populations often have lower mortality than LBW infants from populations with a more
favorable birthweight distribution (ie. black vs. white infants) [77]. Examined another
way, among LBW infants, risk of IM is relatively independent of the social and
demographic factors that affect the overall IM rate. It appears that those social factors
inﬂuence IM by altering the birthweight distribution (ie. increasing the proportion of low
birthweight births), whereas for a given birthweight, the inﬂuence of those factors is
relatively small, speciﬁcally among black infants [52]. This concept is referred to as the
LBW paradox. Despite this complex relationship between birthweight, race and IM, the
fact remains that the smallest infants are at the greatest risk of mortality [6]. Furthermore,
as this research shows, inequalities in the very low birthweight proportion are
consistently correlated with disparities in IM.

After adjustment, relative and absolute inequalities in maternal factors were not

signiﬁcantly correlated with racial disparities in IM. The strong positive correlation seen

72

between inequalities in teen pregnancy and disparities in IM were no longer signiﬁcantly
correlated after accounting for confounding variables such as education and marital
status. Another study examining 1995-1996 linked birth infant death data found that
racial disparities in IM risk varied by teen category, with infants born to 18-19 year old
blacks more likely than their white counterparts to die, while the risk of infant death was
lower for blacks whose mothers were less than 18 [78]. Overall, black women are more
likely to have a teen pregnancy, which increases IM risk. This excess in teen pregnancy
among black mothers could account for the signiﬁcant correlation between inequality in
teen pregnancy and disparities in IM. While there was a lack of correlation between
inequalities in maternal sociodemographic factors and disparities in IM in state level
analyses, these factors have still been found to inﬂuence the risk of individual infant
death.

With respect to correlations between inequalities in state level factors and disparities
in IM, inequalities in the proportion foreign born and in the percent poverty were
signiﬁcantly correlated with racial disparities in IM, even after accounting for other
confounding variables. Although numerous studies have evaluated the effect of foreign
born status on racial differences in infant outcomes [79, 80], to our knowledge a state-
level examination of inequalities in foreign born status has not yet been done.
Historically, the effect of poverty on racial disparities in IM has been examined by
numerous studies [81-83]. More recently, Sims (2007) examined the effect of urban area
poverty on racial disparities in IM and found that high poverty was signiﬁcantly
associated with black/white racial disparities in IM. But, unlike our results, the black

poverty coefﬁcient shrunk by 78% when other maternal factors were added to the model.

73

Other studies have found associations between in income inequality, education, medical
care, proportion black, and unemployment [33, 34, 84, 85], and population level IM
although many of these were modest associations. The most consistent state-level
predictor of IM in these studies was the proportion of the population which was black.
We also found that the proportion black was signiﬁcantly correlated with racial
disparities in 1M (results not shown). There are several theories as to why social
inequalities could be related to health [41]. Social capital theories assert that individual
and group level relationships inﬂuence population health either directly or indirectly
through proximal factors. Psychosocial theories hold that inequalities in social standing
create stress that can eventually damage a person’s health- Others have suggested that
areas with greater social inequalities may systematically under invest in health care and
housing which may lead to poor health status among disadvantaged groups [84].
Schoendorf (1992) found that racial disparities in IM persist even after classiﬁcation on
socioeconomic position. In contrast to black infants in the general population, black
infants born to college educated parents have higher IM rates than similar white infants
[37]. Cultural differences and institutionalized racism [5, 37 , 86, 87], are also thought to
be contributors to racial disparities in infant outcomes directly or through access to
medical care but further studies are needed to examine their true contributions. Although
inequalities in state factors were not as strongly correlated with IM disparities as were
inequalities in infant factors, this research is in line with suggestions that racial disparities
in 1M reﬂect inequalities among socioeconomic groups in state level characteristics [88].

2.5.6 Other reasons for national variation in disparities in infant mortality

74

This study examined the effect of inequalities in sociodemographic risk factors for IM
on racial disparities in IM. Another factor which could have accounted for nationwide
variation in IM disparities are state discrepancies in the reporting of live births and infant
or fetal deaths, especially at non—viable birthweights or gestations and state differences in
access to medical care. While it is unlikely that state to state variations in the quality of
vital records contributes to differences noted in our results, since approximately 99
percent of all births are reported in the US, state level variation in the reporting of fetal
and infant deaths could inﬂuence state IM rates and disparities differently [36, 43, 45]. In
addition, since black infants are more likely to be born at extremely preterm gestations [1,
2, 10], race may differentially affect the reporting of these non-viable births, therefore
leading to disparities in IM-

2.6 Conclusion

Our study found that relative and absolute measures of disparity provided similar, but
not identical results when examining inequality in infant, maternal and state
sociodemographic factors. They provided different results, however, when examining
state-level racial disparities in IM. The combination of racial composition scores and
disparity measures shed new night on highest and lowest disparity states. The evaluation
of both racial disparities and racial composition scores leads to conclusion that
Washington has the lowest relative disparities in IM and Massachusetts has the lowest
absolute racial disparities in IM since they have low DRs and DDs in addition to low
black and low white racial composition scores. On the other hand, Tennessee and Ohio
would be the highest disparity states because they have high disparity measures, high

black and high white racial composition scores. Furthermore, a state such as New Jersey,

75

which has low black and low white racial composition scores, would not be listed as a
high disparity state.

State differences in the proportion of VLBW births were the strongest predictor of
racial disparities in IM and efforts should be made to target women who are at high risk
for a VLBW birth. Furthermore inequalities in foreign born status and percent poverty
were also signiﬁcantly correlated with national variation in IM disparities. Inequalities in
maternal factors did not play a strong role in nationwide variation in IM disparities, but
are still important predictors of individual IM risk. Due to differences in IM disparity
results depending on the measure used, future studies examining reasons for state-level
differences in racial disparities in IM, should use both relative and absolute disparity
measures. Care should also be taken when examining nationwide variation in racial
disparities in IM and interpretation of results should include discussion of how results

differ, with the disparity measure used.

76

CHAPTER 3: THE IMPACT OF FETAL DEATH REPORTING PRACTICES ON
RACIAL DISPARITIES IN INFANT, EARLY AND LATE NEONATAL,
POSTNEONATAL, AND FETAL MORTALITY

This chapter includes an examination of the impact of state reporting practices on racial
disparities in infant, early and late neonatal, postneonatal and fetal mortality rates and is
study 2 for this three paper dissertation option.

3.1 Abstract

Objective: To determine the impact of state fetal death reporting requirements and
reporting of non-viable births on racial disparities in infant, early and late neonatal,
postneonatal and fetal mortality rates.

Methods: Birth and death certiﬁcate data from non-Hispanic white (white) and
non-Hispanic black (black) infants were obtained from the 2000-2002 Vital Statistics
Division of the National Center for Health Statistics linked birth/infant death and fetal
death dataset. Mortality rates were grouped by state fetal death reporting requirements
and by the proportion of non-viable births. Relative and absolute measures of racial
disparity were examined in relation to mortality. Logistic regression was used to examine
the effect of fetal death registration area and proportion of non-viable births on the risk of
race-speciﬁc mortality.

Results: Mortality rates and racial disparities were the highest among states with
birthweight only fetal death reporting criteria and among states with the highest
proportion of non-viable births recorded in birth certiﬁcates. The largest proportion of
this difference was accounted for by births S 22 weeks gestation. Racial disparities in

fetal deaths were highest among states which report all products of conception and

77

among states with the highest proportion of non-viable births. The absolute measure of
disparity was most sensitive to registration area differences in disparity.

Conclusions: Fetal death registration area and proportion non-viable differences
in mortality were most pronounced among black, extremely preterm infants. A uniform
deﬁnition of fetal death should be adopted to reduce systematic differences in the
reporting of live births and fetal deaths.

3.2 Introduction
3.2.1 Nationwide variation in racial disparities in infant mortality

Racial disparities in infant mortality (IM) rates (death prior to 1 year of age per 1,000
live births) have been a problem in the US. for decades with some states experiencing
higher disparities than others [1, 2, 4, 6, 29, 31—36, 63]. Little is understood of the cause
for state-level differences in IM rates and racial disparities in IM rates but states have
been found to vary in population risk characteristics, prevalence of low birthweight, and
geographic obstacles to delivery in care [4]. The reporting of fetal deaths and births at
the border of viability has received little attention in relation to state—level variation in IM
rates[36, 45] and even less in relation to racial disparities[43].
3.2.2 Fetal death reporting

Variation in the completeness and accuracy of reporting fetal and infant deaths
can inﬂuence both fetal and infant mortality rates [43]. Although all states require the
reporting of a live birth regardless of the length of gestation or weight, there is
considerable variation in fetal death reporting criteria. The 1992 revision of the Model
State Vital Statistics Act and Regulations recommend the following deﬁnition of fetal

death: “. . .death prior to the complete expulsion or extraction from its mother of a product

78

of human conception. . .after such expulsion or extraction the fetus does not breathe or
show any evidence of life such as beating of the heart, pulsation of the umbilical cord or
deﬁnite movement of voluntary muscles. . .each fetal death of 350 grams or more, or if
weight is unknown, of 20 completed weeks of gestation or more. . .shall be reported
within 5 days after delivery” [44]. Thirty eight of the US. registration areas use a
deﬁnition similar to this deﬁnition while twelve areas use a shortened or different
deﬁnition of fetal death-

Because there is only a recommended deﬁnition of fetal death, systematic
variation does appear in state reported rates of low birthweight fetal deaths. In addition,
imprecision in recognizing or acknowledging very brief and faint signs of life may lead to
systematic variations in reporting a delivery as a live birth or a fetal death [43]. Previous
studies have examined the effect of reporting differences on the proportion of low
birthweight births [45, 46], racial differences in perinatal mortality [43], and neonatal
mortality [36]. Although there is conﬂicting evidence on the effect of reporting
differences on state-level differences in perinatal outcomes, Wingate [43] found that there
may be underreporting of low birthweight fetal deaths and recommended further analyses
to establish if black fetal death rates are underreported. Systematic misclassiﬁcation of
very low birthweight infants as fetal deaths could lead either to an underestimation or an
overestimation of the overall IM rate and racial disparities in IM rates, depending on the
racial composition of the under- or over-reported deaths.

3.2.3 Study Objective
The objective of this study was to determine the impact of state differences in a) fetal

death registration requirements and b) the reporting of births at the border of viability on

79

absolute and relative racial disparities in reported infant, early and late neonatal,
postneonatal and fetal mortality rates. We hypothesized that reported infant and fetal
mortality rates and racial disparities in reported mortality rates would vary depending on
the a) restrictiveness of fetal death reporting requirements and b) proportion of live births
reported at the border of viability. We further hypothesized that racial disparities would
vary depending on the disparity measure (relative vs. absolute) used. The effects of
reporting requirements and the proportion of non-viable births were expected to vary by
gestational age at birth, with the strongest hypothesized effect on mortality rates and
disparities among extremely preterm infants.
3.3 Materials and Methods
3.3.1 Study Population

The source population included singleton, live births, infant deaths and fetal
deaths to non-Hispanic white and non-Hispanic black (hereafter referred to as white and
black) maternal residents of the US. (n=10,999,362 live births; 66,566 infant deaths;
154,712 fetal deaths). Due to documented, systematic variation in the reporting of infant
and fetal deaths less than 20 weeks gestation, these records were excluded (n=2,902 live
births; 2,331infant deaths; 80,783 fetal deaths). Usually, records missing gestational age
are excluded from analyses such as these, but due to the thought that missing gestational
age information could lead to state-level discrepancies in infant and fetal mortality,
records missing information on gestational age were examined as well (n=108,784 live
births; 2,174 infant deaths; 7,563 fetal deaths). The study population therefore included
10,996,460 live births, 64,235 infant deaths and 73,929 fetal deaths.

3.3.2 Data source and Preparation

80

Birth and linked infant death records and fetal death records were obtained from
the 2000-2002 Linked Birth/Infant Death Dataset, and the 2000-2002 US. Fetal Death
Dataset produced by the National Center for Health Statistics (N CHS) Division of Vital
Statistics [59]. Three years of cohort, as opposed to period, data were used in order to
obtain a sufﬁcient number of infant cases in the numerator and live births in the
denominator from which to produce stable mortality rates. Cohort data links infants born
in one year to subsequent deaths regardless to whether the death occurred during the birth
year, or the following year. Birth cohort data ﬁles were preferred for this analysis because
they followed a given cohort of births for an entire year to ascertain mortality-speciﬁc
information. Since fetal deaths do not have a corresponding birth certiﬁcate, this cohort
argument only applies to infant deaths.

National birth and linked infant death records make use of state linked ﬁles for the
identiﬁcation of linked birth and infant death certiﬁcates and NCHS natality and
mortality computerized statistical ﬁles. When the birth and death of an infant occur in
different states, copies of the records are exchanged by the state of death and the state of
birth in order for the record to be linked. In addition, if a third state is identiﬁed as the
state of residence at the time of birth or death, that state is also sent a copy of the
appropriate certiﬁcate by the state when the birth or death occurred. Fetal death statistics
for every year are based on all reports of fetal death received by the NCHS and include
fetal deaths occurring at a stated or presumed gestation of 20 weeks or more. Reporting
requirements for fetal deaths vary from state to state, therefore, reporting is not as

complete for fetal deaths as it is for live births and subsequent infant deaths. Fetal death

81

reporting is believed to be relatively complete for fetal deaths at 2 28 weeks gestation
[60].

The NCHS natality and mortality ﬁles, produced annually, include data from birth
and death certiﬁcates that are provided to NCHS by states under the Vital Statistical
Cooperative Program. Data were coded according to uniform coding speciﬁcations,
passed rigid quality control standards, were edited and reviewed and are the basis for
official US. birth and death statistics [59].

3.3.3 Outcome Variables

Reported mortality rates included IM (infant death between 1 and 364 days / live
births), early neonatal mortality (infant death during the ﬁrst 6 days of life / live births),
late neonatal mortality (infant death between 7 and 27 days of life / live births - early
neonatal deaths), postneonatal mortality (infant death between 28 days and 364 days of
life / live births — neonatal deaths) and fetal mortality (fetal death 220 weeks gestation /
live births + fetal deaths) and were expressed per 1,000 live births (table 2). Crude
mortality rates in addition to rates stratiﬁed by race were calculated.

Both relative, via the diSparity ratio (DR), and absolute, via the disparity difference
(DD) measures of racial disparity in mortality were employed to test which was most
sensitive when assessing nationwide variation in racial disparity. The DR, one of the
most commonly used measures of health disparity, was calculated by dividing the
reported mortality rate of the most disadvantaged group (black infants) by the reported
mortality rate of the most advantaged group (white infants). In the case of no disparity,
the value the DR took was one. The DD, a measure of the absolute disparity between

two groups, was calculated by subtracting the reported mortality rate among the most

82

advantaged group (white infants) from the reported mortality rate among the most
disadvantaged group (black infants). In the case of no disparity, the value the DD took
was zero. All DDs were expressed per 1,000 live births.

3.3.4 Exposure Variables

State level fetal death reporting requirements and the proportion of live births at
the border of viability were the main exposures of interest. For the purpose of this study,
state referred to birth state, as opposed to maternal state of residence. Although infant
‘exposure’ to state reporting requirements do not increase risk of IM or racial disparities
in IM per se, state reporting practices could inﬂuence whether a death is classiﬁed as a
live birth and subsequent infant death, or classiﬁed as a fetal death. The exposure-
outcome relationship was therefore the effect of reporting practices on the risk of being
classiﬁed as a infant death versus the risk of being classiﬁed as a fetal death.

Fetal death registration areas were classiﬁed by two sets of criteria. The ﬁrst was
the state adopted fetal death reporting requirements classiﬁcation system (table 1) and the
second set of criteria included four consolidated categories thought to capture most of the
reporting differences which may occur through birthweight or gestational age criteria
(table 2). Registration areas were consolidated to account for small sample sizes in a
number of the fetal death registration areas (ie. birthweight Z 350 grams: Kansas,
birthweight Z 400 grams or gestation of Z 20 weeks: Michigan, birthweight Z 500 grams
or gestation of Z 20 weeks: District of Columbia and gestation of Z 16 weeks:
Pennsylvania). The newly created registration areas included states which reported: 1. all
products of conception; 2. birthweight criteria or gestational age criteria; 3. birthweight

criteria, only; 4. gestational age criteria, only. Newly created fetal death registration

83

areas were then ranked from ‘most liberal’ which included states which report all
products of conception, regardless of the birthweight or gestational age to the ‘most
strict’ which included states which report based on report gestational age criteria, only.

Table 3.1. Fetal death registration requirements, 1997 revision.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

All products of human conception
Arkansas Hawaii
Colorado New York State Rh‘oglre iIrSiiznd
Georgia New York CitL g
Gestation of 220 weeks
Alabama Maine Oklahoma
Maryland
Alaska . Oregon
. . Minnesota
Calrforma Texas
. Nebraska
Connecticut Utah
. Nevada
Florida Vermont
. . New Jersey .
Illmors . Washington
Indiana North Carolina West Virginia
Iowa North Dakota W omin
Ohio y g
Birthweight of 2350 grams
Kansas
Birthweight of 2350 grams or Gestation of 220 weeks
Arizona Louisiana Montana
Delaware Massachusetts New Hampshire
Idaho Mississippi South Carolina
Kentucky Missouri Wisconsin
Birthweight of 2400 grams or Gestation of 220 weeks
Michigan
Birthweight of 2500 grams or Gestation of 220 weeks
District of Columbia
Birthweight of 2500 grams
New Mexico South Dakota Tennessee
Gestation of 21 6 weeks
Pennsylvania

 

84

Figure 3.1. Fetal Death Registration Areas, United States 1997 Revision

  

Area
[:] 1: All Products ofConcepuon
2: Birthweight or Gestational Age
- 3: Birthweight. only

- 4: Gestational Age. only 0 '50 300 600 Miles

85

Table 3.2. Consolidated fetal death reporting requirements, United States 2000-
2002.

 

 

 

 

 

 

 

 

 

 

 

 

tron Area 1 All products of human conception
Arkansas New York State
Colorado New York City
Georgia Rhode Island
Hawaii Virginia
3 ' tion Area-2: Bﬁamight or‘GESMOnm Ago1
Arizona Michigan
Delaware Mississippi
District of Columbia Missouri
Idaho Montana
Kentucky New Hampshire
Louisiana South Carolina
Massachusetts Wisconsin
"fRegistration ArEaSJ Birthwerj ' Tgli_t, Only "
Kansas South Dakota
New Mexico Tennessee
' Regamﬁon Ama‘4z?Gegﬁonal_-Age, Only-- - -' -
Alabama New Jersey
Alaska North Carolina
California North Dakota
Connecticut Ohio
Florida Oklahoma
Illinois Oregon
Indiana Pennsylvania
Iowa Texas
Maine Utah
Maryland Vermont
Minnesota Washington
Nebraska West Virginia
Nevada Wyoming

 

 

 

 

The second reporting exposure of interest was the proportion of non-viable live
births. This exposure was classiﬁed by the proportion of live births <500 grams and <23
weeks gestation within each state. The proportions for each state were ranked from
smallest to largest by the proportion <500 grams then by the proportion <23 weeks and

were subsequently were split into four categories (table 3).

86

Figure 3.2. Proportion of Non-Viable Births, United States 2000-2002

  

Caravan!

E 1: 0.7-0.11%

2012015,” 0 245 4110 00010101131»
- 3: 0 184119111.

- 4: 0 20m 0 102.5 325 050 was

87

Table 3.3. Borderline live births based on the proportion non-viable, United States

2000-2002.

 

Category 1
<500 grams: 0.03-0.09%
<23 weeks: 0.04-0.10%

 

 

 

Alas“ . South Dakota
New Hampshire Texas
Nevada Utah
(Egon Vermont
Montaiia “63811ng on
New York City yomlng
Category 2

<500 grams: 0.10-0.12%
<23 weeks: 0.10-0.12%

 

Arizona

 

 

California “bras“?
Colorado New Mexrco
Iowa North Dakota
Massachusetts Wgrthhiggiraiia
Minnesota
Category 3

<500 grams: 0.12-0.16%
<23 weeks: 0.13-0.16%

 

 

 

Arkansas Missouri
Florida New Jersey
Hawaii New York State
Indiana Virginia
Kansas . Ohio

Kentucky Pennsylvania
Maine Wisconsin

Category 4

<500 grams: 0.16-0.33%
<23 weeks: 0.17-0.37%

 

 

Alabama
Connecticut
Delaware
District of Columbia
Georgia
Illinois
Louisiana

 

Maryland
Michigan
Mississippi
North Carolina
South Carolina

Tennessee
Rhode Island

 

88

 

3.3.5 Gestational age stratification
Each state is required to report all fetal and infant deaths 220 weeks gestation so

reporting differences in mortality and disparity were stratiﬁed with this cut-point in mind.
The ﬁrst gestational stratiﬁcation included all live births, infant deaths and fetal deaths
220 weeks gestation and was selected as the ‘baseline’ estimator of reporting area
differences. Next, reporting differences among those missing information on gestational
age were examined. Since information on gestational age and/or birthweight was
required by most states for reporting purposes (table 3.1), mortality rates, relative and
absolute disparities among those missing information on birthweight or gestation age
were hypothesized to vary by fetal death registration area and proportion non-viable
category. Reporting differences were also hypothesized to ﬂuctuate around 23 weeks
gestation, since lungs are thought to be too immature to survive before this point. For
this reason, mortality rates and disparities among infants 322 weeks gestation were also
examined. Additionally, mortality rates and disparities between 23-28 weeks were
examined to see if any reporting differences among infants 5 22 weeks were still
apparent.
3.3.6 Data Analysis

Logistic regression was used to model the effect of fetal death classiﬁcation area and
the proportion of non-viable births on the race-speciﬁc risk of mortality. Gestational
week adjustments were made when appropriate to account for national variation in
gestational age at birth. Relative differences in gestational age speciﬁc mortality risk

between fetal death classiﬁcation areas and proportion non-viable categories were

89

quantiﬁed using the relative risk and statistical signiﬁcance was determined with a 95%
conﬁdence interval. All statistical analyses were conducted using SAS version 9.1.3.
3.3.7 Mapping

Maps were used to visualize national differences in IM rates and racial disparities
in IM rates. Continuous data were split into 4 categories based on quantile classiﬁcation
within ArcGIS. Darker colors indicate higher rates, while lighter colors indicate lower
proportions/rates. We also mapped state differences in reporting practices. Maps were
created using ArcMap version 9.3 within ArcGIS version 9.
3.4 Results
3.4.1 Mortality rates, disparity ratios and disparity differences by race

From 2000-2002, there were 10,996,460 (white = 9,238,461; black = 1,757,999)
reported live births and 64,235 (white = 44,397 and black =19,838) reported infant deaths
in our study population. This resulted in a reported IM rate of 5.8 per 1,000 live births
(white IM=4.8 per 1,000 live births; black IM =11.3 per 1,000 live births). The DR was
2.4 and DD was 6.5 excess deaths per 1,000 live births. There were 73,929
(white=52,181 and black=21,748) reported fetal deaths in our study population. This
resulted in a fetal mortality rate of 6.7 (white=5.6 per 1,000 live births; black=12.2 per
1,000 live births), a DR of 2.2 and a DD of 6.6 excess fetal deaths per 1,000 live births

(table 3.4).

90

Table 3.4. Mortality rates, disparity ratios and disparity differences, United States
2000-2002

 

 

 

 

 

 

 

. . Disparity Disparity
Mortality Rate“ Total White Black Ratio Di fference”
Infant 5.8 4.8 11.3 2.4 6.5
Early Neonatal 3.1 2.5 6.3 2.5 3.8
Late Neonatal 0.8 0.7 1.5 2.1 0.8
Postneonatal 2.2 1.8 4.0 2.2 2.2
Fetal 6.7 5.6 12.2 2.2 6.6

 

 

 

 

 

 

 

*expressed per 1,000 live births

91

Figure 3.3. Total Infant Mortality Rates, United States 2000-2002

   

Rate per 1,000 live births

]:] 3.84 - 5.21

5,22 -5.96

- 5.97 .700

- 7.01 -1o.59 o 155 310 1120 Miles

92

Figure 3.4. Disparity Ratios in Infant Mortality, United States 2000-2002

   

Disparity Ratio

]:] 0.00- 1.82

1.83-2.25

- 2'26 ’2'50 j—r-I—rTl—r—r—j
- 2.51 -3.14 o 150 300 600 Miles

93

Figure 3.5. Disparity Differences in Infant Mortality, United States 2000-2002

  

Disparity Difference per 1,000 live births

C] -S.78 -4.44
El] 4.45 -668
- 6.69-7.57 o 150 300 600 Miles
- 7.58 - 1053

3.4.2 Mortality rates and racial disparities among infants 2 20 weeks and with
unknown gestational ages by fetal death registration area

Table 3.5 shows mortality rates, relative disparities and absolute disparities
among infants with a reported gestational age of 2 20 weeks and infants with unknown
gestational ages by fetal death registration area. Among infants with a reported
gestational age of 2 20 weeks, the highest infant mortality rates were generally seen in

fetal death classiﬁcation area 3, with birthweight, only criteria. The lowest mortality

94

rates were generally seen in fetal death classiﬁcation area 1 which reported all products
of conception. In contrast to the mortality rates seen among infant deaths, area 1 had the
highest fetal mortality rates, while area 3 had the lowest fetal mortality rates.

Both absolute and relative disparity measures showed similar results, but the
largest variation between fetal death registration areas was seen with the absolute
measure. Nationally, DDs varied with differences ranging from 2.3 excess postneonatal
deaths per 1,000 live births to 6.8 excess infant deaths per 1,000 live births. DRs
demonstrated less national variation with black infants approximately 2.5 times as likely
as white infants to die during each period. The highest DRs and DDs for infant, early
neonatal, neonatal, and postneonatal mortality were found in registration area 3, while
registration area 1 had the highest fetal death DDs.

Infant, early neonatal and neonatal disparity measures among infants missing
information on gestational age were higher than disparity measures among infants with
complete gestational age information. DRs and DDs demonstrated similar disparity
ﬁndings, although the DD was more sensitive to large registration area differences in

disparity.

95

Table 3.5. Mortality among infants 220 weeks gestation and with unknown

gestational age by fetal death registration area, United States 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fetal Death Gestational Age 220 weeks Unknown Gestational Age
Classiﬁcation n=10,887,676 births n=108,784 births
Area Total White [Black 1 DR ] DD* Total ] White] Black | DR 1 DD*
Infant Mortality

United States 5.7 4.7 11.0 2.4 6.3 20.8 16.2 55.1 3.4 38.9
Area 1 5.8 4.6 10.0 2.2 5.4 65.1 41.9 151.5 3.6 109.6
Area 2 6.2 4.9 11.9 2.4 7.0 61.1 45.7 132.0 2.9 86.3
Area 3 6.7 5.6 13.3 2.4 7.6 74.6 51.5 135.7 2.6 84.3
Area 4 5.5 4.6 10.9 2.4 6.3 16.8 13.6 41.5 3.1 27.9

Early Neonatal Mortality

United States 2.7 2.2 5.4 2.5 2.5 15.8 11.8 45.0 3.8 33.2
Area 1 2.9 2.2 5.2 2.3 2.9 57.0 36.4 133.3 3.7 96.9
Area 2 3.0 2.3 5.9 2.5 3.6 55.1 40.0 125.3 3.1 85.4
Area 3 3.1 2.5 6.7 2.7 4.2 56.0 36.0 108.6 3.0 72.6
Area 4 2.6 2.2 5.2 2.4 3.1 12.2 9.6 32.4 3.4 22.8

Late Neonatal Mortality

United States 0.8 0.7 1.5 2.2 0.8 1.6 1.4 3.0 2.2 1.7
Area 1 0.8 0.7 1.4 2.1 0.7 - - - - -
Area 2 0.9 0.7 1.6 2.2 0.9 - -- - - -
Area 3 0.9 0.8 1.7 2.2 0.9 - -- - - -
Area 4 0.8 0.7 1.5 2.2 0.8 1.5 1.3 2.8 2.2 1.5

Postneonatal Mortality

United States 2.2 1.8 4.1 2.3 2.3 3.5 3.0 7.4 2.5 4.4
Area 1 2.1 1.7 3.4 2.0 1.7 - - - - -
Area 2 2.3 1.9 4.4 2.3 2.5 - - -- - -
Area 3 2.7 2.3 4.9 2.1 2.6 - - -- - -
Area 4 2.1 1.8 4.2 2.3 2.4 3.1 2.7 6.6 2.4 3.9

Fetal Mortality

United States 6.2 5.2 11.7 2.3 6.5 459.9 418.3 646.0 1.5 227.7
Area 1 7.6 5.9 13.4 2.2 7.4 971.8 970.1 976.9 1.0 6.8
Area 2 6.0 4.8 10.8 2.2 6.0 199.8 178.9 J 284.4 1.6 105.5
Area 3 4.1 3.5 7.7 2.1 4.1 88.4 96.1 67.5 0.7 -28.6
Area 4 5.8 4.9 11.0 2.2 6.1 65.5 58.3 116.9 2.0 58.6

 

 

 

 

 

 

 

 

 

 

 

*expressed per 1,000 live births

-- indicates inadequate cell speciﬁc count

Area 1= All products of conception

Area 2=Birthweight and gestational age criteria
Area 3= Birthweight criteria

Area 4= Gestational age criteria

96

 

3.4.3 Mortality rates and racial disparities among infants 522 weeks and 23-28
weeks gestation by fetal death registration area

Table 3.6 shows mortality rates and disparity measures among infants born at
reported gestations of 522 weeks and 23-28 weeks, by fetal death registration area.
Despite high mortality rates among infants 522 weeks reported gestation, black infants
were less likely than their white counterparts to die in most registration areas. DRs were
approximately 0.9 and DDs ranged from -77.0 to -16.0 for infant, early neonatal and
neonatal mortality. Any black survival advantage was not seen among infants in
registration area 3, as this was the only registration area which reported similar black and
white mortality rates (DR21.0). Despite this ﬁnding, relative and absolute disparity
measures demonstrated the largest black advantage among fetal death in registration area
3 (DR=0.4; DD=~159.9).

A black survival advantage was seen for most mortality periods among infants
between 23 and 28 weeks reported gestation, with DRs and DDS ranging from 0.8-1.0
and -48.0 to -13.0, respectively. Relative and absolute disparities were similar between
registration areas. Of note were large registration area differences in late neonatal
mortality rates and disparity measures. While black infants were less likely than their
white counterparts to die in areas 1 and 2, they were nearly three times as likely as their

white counterparts to die in registration area 4.

97

Table 3.6. Mortality rates among infants 522 weeks and 23-28 weeks gestation by
fetal death registration area, United States 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fetal Gestational Age :22 Weeks Gestational Age 23-28 Weeks
Death n=14,415 births n=50,909 births

Area Total ] White [ Black ] DR | DD* Total ] White | Black ] DR [ DD*

Infant Mortality

U.S. 787.3 801.1 770.2 1.0 -30.9 232.2 288.9 274.4 1.0 -l4.5
Area 1 837.2 877.8 801.2 0.9 -76.6 303.0 309.5 296.2 1.0 -l3.3
Area 2 784.2 820.9 748.9 0.9 -72.0 286.6 299.1 271.8 0.9 -27.3
Area 3 855.5 841.3 873.1 1.0 -31.8 289.0 296.3 275.1 0.9 -21.2
Area 4 770.3 776.6 761.1 1.0 -15.5 276.1 281.1 267.2 1.0 -13.9

Early Neonatal Mortality

U.S. 753.0 773.8 727.4 0.9 -46.4 187.8 199.9 167.2 0.8 -32.7
Area 1 808.9 855.3 767.8 0.9 -87.5 202.8 219.7 184.9 0.8 -34.8
Area 2 750.6 794.1 709.0 0.9 -85.1 190.0 207.5 169.2 0.8 -38.3
Area 3 822.2 802.4 847.0 1.1 44.6 193.0 207.1 166.2 0.8 -40.9
Area 4 734.1 748.8 712.5 1.0 -36.3 182.6 192.9 163.9 0.9 -29.0

Late Neonatal Mortality

US. 77.3 69.4 85.3 1.2 15.9 60.9 40.4 58.2 1.4 17.8
Area 1 97.2 - -- -- -- 64.9 82.7 65.0 0.8 -l7.7
Area 2 77.7 74.9 79.6 1.1 4.7 63.6 90.1 55.9 0.6 -34.2
Area 3 -- - -- -- -- 53.1 52.0 55.0 1.1 2.9
Area 4 69.8 58.1 85.0 1.5 26.9 59.4 21.2 56.9 2.7 35.7

Postneonatal Mortality

US. 66.7 55.5 78.3 1.4 22.8 60.2 49.4 76.0 1.5 26.6
Area 1 56.4 -- -- - -- 65.0 53.6 76.6 1.4 23.0
Area 2 61.4 -- 62.5 -- - 59.4 48.6 71.5 1.5 22.9
Area 3 -- -- - - - 69.6 63.9 80.0 1.3 16.1
Area 4 71.3 55.8 91.8 1.6 36.0 58.6 51.7 70.7 1.4 19.0

Fetal Mortality

U.S. 550.0 586.1 495.7 0.9 -90.4 226.8 245.9 195.6 0.8 -50.2
Area 1 653.9 689.3 614.9 0.9 -74.4 259.2 280.6 235.0 0.8 -45.6
Area 2 505.9 562.6 435.7 0.8 -127.0 210.4 235.4 178.3 0.8 -57.1
Area 3 190.9 254.5 94.6 0.4 -159.9 145.6 153.1 130.8 0.9 -22.4
Area 4 542.5 575.7 482.7 0.8 -93.0 227.5 246.3 191.0 0.8 -55.3

 

 

 

 

 

 

 

 

 

 

 

*expressed per 1,000 live births

-- indicates inadequate cell speciﬁc count
Area 1= All products of conception

Area 2=Birthweight and gestational age criteria
Area 3= Birthweight criteria
Area 4= Gestational age criteria

98

 

Next, we examined mortality rates, relative disparities and absolute disparity measures by
the proportion of non-viable births.
3.4.4 Mortality rates and racial disparities among infants 2 20 weeks and with
unknown gestational ages by non-viable category

Table 3.7 shows mortality rates and relative and absolute disparities among
infants with a reported gestational age 2 20 weeks and among infants missing gestational
age by the proportion of largely non-viable births. Among infants 2 20 weeks gestation,
the highest infant and fetal mortality rates and disparities were generally found in
category 4, which also had the highest proportion of non-viable births. Category 1, which
had the lowest proportion of non-viable births, had the lowest mortality rates and
disparities. The corresponding high infant and high fetal mortality rates seen when
examining proportion non-viable categories were contrary to the pattern of high infant
and low and low fetal mortality rates seen when examining fetal death registration areas-

Among births with unknown gestational ages, mortality rates and disparity
measures ﬂuctuated by proportion non-viable category with the highest rates in category

3.

99

Table 3.7. Mortality among infants 220 weeks gestation and with unknown
gestational age by Proportion Unviable, United States 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Non-viable Gestational Age 220 weeks Unknown Gestational Age
Category n=10,887,676 births 11:. 108,784 births:
Total ] White ] Black ] DR] DD* Total ] White ] Black ] DR ] DD*
Infant Mortality

United States 5.7 4.7 11.0 2.4 6.3 20.8 16.2 55.1 3.4 38.9
Category 1 4.9 4.4 8.5 1.9 4.1 32.6 29.8 45.8 1.5 16.1
Category 2 4.7 4.4 9.1 2.1 4.7 11.2 10.0 22.9 2.3 12.9
Category 3 5.9 4.9 11.1 2.3 6.3 95.6 68.5 177.6 2.6 109.0
Catﬁory 4 7.0 5.1 12.1 2.4 7.0 88.3 65.5 124.0 1.9 58.5

Early Neonatal Mortality

United States 2.7 2.2 5.4 2.5 2.5 15.8 11.8 45.0 3.8 33.2
Category 1 2.2 2.0 3.7 1.9 1.8 24.7 22.2 36.8 1.7 14.6
Category 2 2.3 2.1 4.3 2.1 2.2 7.3 6.6 13.9 2.1 7.3
Category 3 2.8 2.3 5.5 2.4 3.2 86.5 60.1 166.5 2.8 106.5
Category 4 2.3 2.4 6.1 2.5 3.7 78.7 57.6 111.8 1.9 54.1

Late Neonatal Mortality

United States 0.8 0.7 1.5 2.2 0.8 1.6 1.4 3.0 2.2 1.7
Category 1 0.7 0.7 1.2 1.9 0.6 3.1 2.9 - -- --
Category 2 0.7 0.6 1.3 2.1 0.7 1.2 1.1 2.6 2.3 1.4
Category 3 0.8 0.7 1.5 2.1 0.8 -- -- -- - --
Category 4 1.0 0.8 1.7 2.2 0.9 -- -- -- -- --

Postneonatal Mortality

United States 2.2 1.8 4.1 2.3 2.3 3.5 3.0 7.4 2.5 4.4
CategorLl 2.0 1.8 3.5 1.9 1.7 5.0 4.8 -- -- -
Category 2 1.8 1.6 3.4 2.1 1.8 2.8 2.4 6.6 2.8 4.2
Category 3 2.2 1.8 4.2 2.3 2.3 7.0 6.5 -- -- --
Category 4 2.6 1.9 4.4 2.3 2.5 7.7 -- -- - -

Fetal Mortality

United States 6.2 5.2 11.7 2.3 6.5 459.9 418.3 646.0 1.5 227.7
Category 1 6.0 5.1 12.4 2.4 7.3 168.7 149.4 249.4 1.7 100.1
Category 2 4.9 4.6 9.5 2.1 4.9 19.6 17.9 35.8 2.0 18.0
Category 3 6.3 5.3 11.3 2.11 6.0 431.6 418.6 467.7 1.1 49.1
Category 4 6.8 4.9 11.6 2.3 6.6 236.4 218.1 263.3 1.2 45.2

 

 

 

 

 

 

 

 

 

 

 

*expressed per 1,000 live births

-- indicates inadequate cell speciﬁc count

Category 1 = <500 grams: 0.03-0.09%; <23 weeks: 0.04-0.10%
Category 2 = <500 grams: 0.10-0.12%; <23 weeks: 0.10-0.12%
Category 3 = <500 grams: 0.12-0.16%; <23 weeks: 0.13-0.16%
Category 4 = <500 grams: 0.16-0.33%;<23 weeks: 0.17-0.37%

100

 

3.4.5 Mortality rates and racial disparities among infants 522 weeks and 23-28
weeks gestation by non-viable category

Table 3.8 shows mortality rates and disparities among infants with a reported
gestation of S 22 weeks and 23-28 weeks, by the proportion of largely non-viable births.
For both gestational age groups, the lowest mortality rates and disparities were generally
seen among category 1 births, while the highest mortality rates and disparities were seen
among category 4 births. Although the relative measure of disparity was similar between
each of the non-viable categories, the absolute measure showed more variation with DDs

ranging from -118.0 to -29.0.

101

Table 3.8. Mortality Rates Among Infants $22 Weeks and 23-28 Weeks Gestation by

Proportion Non-Viable, United States 2000-2002.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Non-Viable Gestational Age 322 Weeks Gestational Age 23-28 Weeks
Category n:=14,415 births n=50,909 births
Total | White ] Black 1 DR 1 DD* Total 1 White Black LDR 1 DD*
Infant Mortality
U.S. 787.3 801.1 770.2 1.0 -30.9 232.2 288.9 274.4 1.0 -14.5
Category 1 697.8 725.1 637.0 0.9 -88.1 260.8 263.8 253.7 1.0 -10.1
Catggory 2 773.1 783.7 731.9 0.9 -51.9 279.3 284.4 256.6 0.9 -27.8
Category 3 802.5 821.0 779.1 1.0 -41.9 281.3 288.0 270.7 0.9 -17.3
Category 4 813.5 843.1 795.4 0.9 -47.7 298.4 315.6 285.4 0.9 -30.3
Early Neonatal Mortality
U.S. 753.0 773.8 727.4 0.9 -46.4 187.8 199.9 167.2 0.8 -32.7
Category 1 659.5 695.5 579.2 0.8 -116.3 170.0 179.5 147.0 0.8 -32.4
Categog 2 744.5 757.2 695.6 0.9 -61.6 196.1 202.1 169.6 0.8 -32.4
Category 3 774.7 798.7 744.3 0.9 -54.4 187.6 198.2 171.0 0.9 -27.2
Category 4 772.0 809.4 749.1 0.9 -60.3 192.7 216.8 174.4 0.8 -42.4
Late Neonatal Mortaligy
US. 77.3 69.4 85.3 1.2 15.9 60.9 40.4 58.2 1.4 17.8
Category 1 48.0 - - - - 54.2 54.5 53.4 1.0 -1.1
Category 2 64.8 64.4 - - -- 58.2 60.3 49.0 0.8 -11.3
Category 3 64.5 58.2 70.8 1.2 12.6 61.8 66.0 55.5 0.8 -10.6
Category 4 112.3 118.0 109.7 0.9 -8.2 64.8 67.3 62.9 0.9 -4.4
Postneonatal Mortality
US. 66.7 55.5 78.3 1.4 22.8 60.2 49.4 76.0 1.5 26.6
Category 1 67.7 53.4 37.1 0.7 -16.3 58.5 51.0 64.0 1.3 13.0
Category 2 50.3 48.2 - -- - 48.1 45.6 58.6 1.3 13.0
Category 3 63.1 56.0 70.3 1.3 14.3 57.0 49.2 68.6 1.4 19.4
Category 4 78.6 59.0 74.8 1.3 15.8 70.8 58.9 71.5 1.2 12.7
Fetal Mortality
U.S. 550.0 586.1 495.7 0.9 -90.4 226.8 245.9 195.6 0.8 -50.2
Category 1 649.1 642.4 663.0 1.0 20.5 248.7 258.1 225.1 0.9 -33.1
Category 2 531.3 546.7 460.3 0.8 -86.4 237.2 245.4 197.5 0.8 -47.8
Category 3 549.9 592.8 480.5 0.8 -112.3 228.0 247.9 194.6 0.8 -53.3
Category 4 506.4 565.9 461.1 0.8 -104.7 208.0 232.9 187.9 0.8 —45.0

 

 

 

 

 

 

 

 

 

 

 

*expressed per 1,000 live births
-- indicates inadequate cell speciﬁc count
Category 1 = <500 grams: 0.03-0.09%; <23 weeks: 0.04-0.10%
Category 2 = <500 grams: 0.10-0.12%; <23 weeks: 0.10-0.12%
Category 3 = <500 grams: 0.12-0.16%; <23 weeks: 0.13-0.16%
Category 4 = <500 grams: 0.16-0.33%;<23 weeks: 0.17-0.37%

102

 

3.4.6 Relative risk of mortality

In order to determine if noted reporting differences in mortality were statistically
signiﬁcant, tables 3.9 and 3.10 show fetal registration area and proportion non-viable
differences in the race-speciﬁc risk of early neonatal mortality among infants 2 20 weeks
gestation, S 22 weeks gestation, and 23-28 weeks gestation. Early neonatal mortality was
chosen because it offers speciﬁcity in time of death. Furthermore, sample sizes for late
neonatal and postneonatal mortality were too small to calculate stable registration area
relative risks. Fetal death registration area 1, had the lowest mortality rates and was used
as the referent category.

Among infants 2 20 weeks gestation, signiﬁcant fetal death registration area
differences in early neonatal mortality risk were seen among black, but not white infants.
Black infants in fetal death registration areas 2 and 3 had signiﬁcantly higher mortality
rates than infants in registration area 1 (RR=1.15; 1.09-1.22, RR=1.33; 1.20-1.48,
respectively). This relationship became more apparent after sub-setting to black infants 5
22 weeks gestation in registration area 3 (RR=1.71; 1.14-2.57). In contrast, white infants
_<_ 22 weeks gestation, had a signiﬁcantly lower risk of death in registration areas 2 and 4
(RR=0.65; 0.52-0.82 and RR=0.55; 0.45-0.67, respectively). White infants within area 4
remained at decreased risk of mortality between 23 and 28 weeks gestation. Adjustment

for gestational week did not alter the ﬁndings.

103

:38 2:35ch b—ooumsﬁm a 332?: Sam

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ado-odooodo ado-odoomdo Aoo._-oo.ooodo oodoodoomdo Aodwoooovoa Goo-odooooo wooed.
ooo._-_dooodo Goa-odooodo 93-2.35 moon-dooooo a..._-o~.smd_ 25-332; 232
eon-odeodo aoaedooodo Aoo._-odooodo ado-odoomdo ANNA-8.9m: ANS-8.38; N82
oo._ oo._ oo._ oo._ oo._ oo._ :62
Room! 323 Roam 1 333 Roam 323 bﬁeoz
$363 wN-MN ﬂoors NN v 383 N 1 ow< Ecoanomow EncooZ

om< 38:830 ow< 353880 A .
.Nooo

ngN mauﬂum gum—u: obm—ﬂtéz —GHNEOOZ h—hﬂm vamOOA—mivuﬂm HG v—ma Em mﬂOﬁvhﬁth—H 60h< ﬂﬁmugummWQN— n—HQOA— —50'...— .@.M 0—n—NH.

104

Among infants 2 20 weeks gestation, a signiﬁcant proportion non-viable category
differences were seen for both blacks and whites, with mortality risk increasing with the
proportion of non-viable births. Category differences in early neonatal mortality became
more apparent when examining infants 5 22 weeks gestation. White infants born in
category 3 and 4 states were 70% more likely than white infants born in category 1 states
to die and black infants in categories 3 and 4 were twice as likely as their counterparts
born in category 1 states. Category differences were no longer signiﬁcant among infants F]
23-28 weeks gestation, with the exception of whites in category 4 states (RR=1.27; 1.16-

1.38). Adjustment for gestational week did not alter the ﬁndings.

 

105

doodo-odo 383 oovmxodo-odo use-a ooov u o booed
dado-2o ”£83 oov dado-So dream ooov l o bowed
dado-2o “goo; oov Mdado-2o ”mead ooov l N booed
aoodo-ooo ”3.83 RV Mdaooo-ooo ”Edam ooov u _ booed
:32 3.85:me bﬁozmuﬁm m 3822: Sam

 

 

 

 

 

 

 

 

 

 

Aodeod o2 33-3.52.— Aodoodcmoo Ago-no.3: sod-ode om; 93-2.32; viewed
ado-mono? Ammo-3.32; stood—rod God-o1: od— deodsoo; add-33:; omewaau
Aoo._-oo.:o_._ Goa-8:2; Goo-3.5%; Goa-odds..— advoiooﬁ AER-3:2; Nbooooeu
oo._ oo._ ooh oo._ oo._ oo._ _booaao
doom so; 5.85 1 323 doom 32% six
owﬂowwowmmamww owdwowmowc 333 ON N ow< 1289300 5:582

 

 

 

 

 

.NchrcceN 335 .3an Sum—8.5: 13.382 atom camoonmtooam he v—ma E moo-3.3.5: .o_aw_>rﬂez act-Sneak .c—.m 03.;

106

3.5 Discussion

This study was designed to determine the extent to which differences in fetal
death reporting requirements and the proportion of non-viable births inﬂuence absolute
and relative disparities in infant, early and late neonatal, postneonatal, and fetal mortality
rates. Fetal death registration area differences in the restrictiveness of reporting were
hypothesized to especially inﬂuence mortality rates among extremely preterm infants
with low survival rates [89]. In addition, since black infants are more likely to be born at
extremely preterm gestations, we explored the possibility that higher infant mortality
rates among black infants were due in part to systematic differences in the reporting of
extremely preterm live births and fetal deaths by registration area.
3.5.1 Fetal death registration area

Similar to other studies, our results demonstrated persistent racial disparities in
overall infant, early and late neonatal, postneonatal and fetal mortality rates [29, 36, 90-
92]. Fetal death registration area 3, with birthweight only criteria, had the highest
mortality rates, while fetal death registration area 1, which reported all products of
conception, had the lowest mortality rates and disparities. In contrast, the highest fetal
mortality rates were found among infants in registration area 1, while the lowest fetal
mortality rates seen in fetal death classiﬁcation area 3. The complementary high infant
mortality rates and low fetal mortality rates in area 3 suggest that fetal death registration
area differences in mortality rates were largely due to differences in reporting infants as a
live birth and subsequent infant death (ie. area 3) versus reporting as a fetal death (ie. area
1).

Both absolute and relative disparity measures demonstrated similar patterns with

107

the highest infant disparities seen in fetal death registration area 3 and highest fetal
disparities seen in fetal death registration area 1. While both absolute and relative
disparity measures demonstrated similar ﬁndings by time at death (ie. early neonatal), the
absolute disparity difference was more sensitive to variation in registration area
disparities. Variation in disparity measure sensitivity was largely due to the size of
mortality rates, with the DD more sensitive to high overall mortality rates and the DR
more sensitive to lower overall mortality rates.

Although there were only a small number of live births with missing information
on gestational age, sizable registration area differences in mortality rates and disparity
measures were noted. The highest disparities were seen among infants born in
registration area 4 which required gestational age speciﬁc information. Wen et al also
found increases in racial disparities among infants missing birthweight and/or gestational
age information [93], although authors did not examine fetal death registration area
differences. These results suggest that either a) black infants missing information on
gestational age were more likely than their white counterparts to be reported as an infant
death than a fetal death in certain classiﬁcation areas or, more likely, b) black infants who
died were less likely than white infants to have gestational age or birthweight recorded.

Black infants, who have higher rates of extremely preterm birth, [91, 92, 94, 95],
also had a slight survival advantage at extremely preterm gestations. This survival
advantage, however, was not present in all fetal death registration areas. Black infants
:22 weeks gestation who were born in fetal death registration area 3 had early neonatal
mortality rates which were slightly higher than their white counterparts. This lack of a

black survival advantage among infants 522 weeks gestation in registration area 3 could

108

partially account for the higher absolute and relative disparities seen in this fetal death
registration area. Corresponding to this was the ﬁnding of much lower relative and
absolute fetal death disparities in fetal death registration area 3. In fact, both black and
white fetal mortality rates were lower in registration area 3 than in the other registration
areas. Seemingly, states within fetal death registration area 3 were more likely to report
births 522 weeks gestation as infant births and subsequent infant deaths than as fetal
deaths (as did the other fetal death registration areas). In order to determine if lower fetal
mortality rates in area 3 states which strictly required birthweight information could be
accounted for by a higher proportion of extremely low birthweight births [60, 67, 92], we
examined the percentage of extremely low birthweight births by fetal death registration
area (results not shown). No fetal death registration area differences were seen in the
percentage of births <500 grams or between 500 and 1000 grams.

In early neonatal predictive models, the effect of fetal death registration area on
mortality risk differed by gestational age at birth and race, with the largest differences
seen among infants S 22 weeks gestation. Black infants S 22 weeks gestation who were
born in fetal death registration area 3 were at signiﬁcantly increased risk of mortality,
while white infants in this same area were at decreased risk of mortality. In comparison
to blacks in other fetal death registration areas, mortality rates among black infants in
fetal death registration 3 were unusually high and accounted for much of the disparity
among states in fetal death registration area 3.

3.5.2 Proportion of non-viable births
With respect to the proportion of non-viable births, the highest mortality rates and

disparities were among infants born in category 4 states, which had the highest proportion

109

 

of non-viable births. Unlike the results seen by fetal death registration area, similar
results were seen among fetal deaths, with the highest fetal mortality rates also seen in
category 4 states. The parallel of high infant mortality rates and high fetal mortality rates
within category 4 leads to the conclusion that proportion non-viable category differences
in mortality rates are not caused solely by differences in reporting.

Relative and absolute disparity measures demonstrated similar results with the
highest disparities found among infant and fetal deaths in category 4. Other studies have
found similar results with the proportion of extremely low birthweight infants predicting
both IM rates [74], and racial disparities in IM [95]. When examining the race speciﬁc
risk of mortality by the proportion non-viable births, category differences among both
races were found to primarily reﬂect differences in mortality risk among infants 5 22
weeks gestation. Causes for race-speciﬁc proportion non-viable category differences in
mortality risk are unknown, especially after adjustment for gestational age (which did not
alter the ﬁndings). These category differences were most prominent among black infants,
which likely led to higher disparities in category 4.

3.5.3 Racial differences in fetal deaths

The question of whether lower black fetal mortality rates are real or a result of
racial differences in reporting has been the cause of some debate [36, 43, 60, 96].
Wingate et al found some evidence of underreporting among blacks with black infants in
their national study 20% less likely than white infants to be classiﬁed as a fetal death
versus an early neonatal or neonatal death. But, they conclude there is little evidence that
black-white differences in mortality are a function of racial differences in classiﬁcation of

fetal deaths and live births [43]. Cai et al found that fetal mortality rates among blacks <

110

28 weeks gestation and those < 1,000 grams were signiﬁcantly lower than their white
counterparts, but they attribute this to the formula used to calculate fetal mortality rates
[60]. Speciﬁcally, the denominator to calculate fetal mortality rates consists of the total
number of births at a speciﬁc gestational age + the number of fetal deaths. Since black
infants are more likely to be born at extremely preterm gestations, they have a larger
denominator, and a subsequently lower fetal mortality rate. This is the main argument
for the fetuses at risk approach, ﬁrst introduced by Yudkin [97]. On the other hand,
Allen et al found that black fetal mortality rates were 2.5 times the fetal mortality rate of
whites (95% CI = 2.2-2.9). This relationship did not remain after adjustment for
gestational age (RR = 1.2;0.9-1.4) [96]. We were unable to ﬁnd any studies which
examined fetal death reporting requirements in relation to mortality or racial disparities in
mortality.

Findings of lower black fetal mortality rates < 28 weeks suggest that there are two
sets of factors which lead to racial disparities in infant mortality which act at different
times (ie. before 28 weeks gestation and after 28 weeks gestation). For example, the ﬁrst
set puts black infants at greater risk for being born extremely preterm. Extreme
prematurity accounts for largest proportion of the black/white disparity in infant mortality
[91, 92, 94, 95]. However, for those fetuses who remain in utero 23-28 weeks gestation,
there did not appear to be an excess fetal mortality risk among blacks (tables 6 and 8).
This leads to the second factor which acts sometime after 28 weeks and puts blacks at
twice the mortality risk as whites both in utero and after birth (results not shown). This is
reinforced by the finding that among fetal deaths >28 weeks, there doesn’t appear to be

registration area differences in relative disparity.

lll

3.5.4 Limitations of Study

The ﬁndings of this study were subject to several limitations. First, by deﬁnition,
fetal deaths were truncated at 220 weeks gestation; with any deaths before 20 weeks
gestation classiﬁed as a miscarriage. Due to this left censoring, any disparities which
occurred before 20 weeks gestation were not measured in this study. Although some
miscarriages (<20 weeks gestation) were available for analysis, their inclusion could have
introduced bias because of possible confusion with late abortion, along with the limited
sample of mothers who present to hospitals after a miscarriage. Second, due to small
sample sizes, the original eight fetal death classiﬁcation areas were consolidated into four
which were though to reﬂect differences in restrictiveness between areas. The
consolidation of registration areas may have attenuated any differences which could have
been more visible between the original eight fetal death classiﬁcation areas. Third, due to
black/white differences in rates of preterm birth and subsequent denominators, racial
differences in fetal mortality rates should be interpreted with caution. This is especially
the case at extremely preterm gestations. Finally, excluding cases which were missing
information on race could have biased our results in light of the registration area
differences in mortality rates and disparity ratios by unknown gestational age and/or
birthweight categories. The basic hypothesis of this study was that fetal death
registration and proportion non-viable category differences would inﬂuence racial
differences in mortality. Since there were not ofﬁcial race requirements for each
registration area, excluding records with missing information on race was unlikely to
differentially inﬂuence mortality rates.

3.5.5 Strengths of Study

112

Despite these limitations, this study had several strengths. First, it made use of
national vital statistics data to determine if differences in fetal death reporting
requirements and the proportion of non-viable births inﬂuenced area differences in
mortality rates and disparity ratios. Second, it is the only study of its kind to examine the
effect of fetal death registration requirements on racial disparities in infant, early and late

neonatal, postneonatal and/or fetal mortality. Third, it made use of both absolute and

his.
relative disparity measures, which were found to vary in their sensitivity to registration
area differences. Finally, biologically relevant gestational age stratiﬁcation was used to
determine if any area or category differences were due to actual reporting or due to
differences in extremely preterm birth rates. r}

 

3.6 Conclusion

Overall, registration area differences and proportion non-viable differences in
mortality rates, absolute disparities and relative disparities were most pronounced among
infants S 22 weeks gestation. Although signiﬁcant area and classiﬁcation differences
were seen among both races, differences were most prominent among black infants.
Findings of proportion non-viable category similarities among infant and fetal deaths
indicated real category differences in not only the proportion of at-risk infants born (ie.
extremely preterm/low birthweight) but also in racial disparities between proportion non-
viable categories. Absolute disparities as measured by the DD were most sensitive to
area differences in disparities, especially when mortality rates were high.

State differences in fetal death reporting requirements lead to differences in the
reported number of live births, infant deaths and fetal deaths. These differences persist

even after limiting data to gestations which should be reported by all states. In order for

113

 

national vital statistics data to be meaningful, a uniform deﬁnition of fetal death must be
adopted. This would reduce systematic differences in the reporting of live births and
fetal deaths and give an accurate record of state mortality rates and disparities.

Future studies should examine proportion non-viable differences in relative and
absolute disparities. Studies should also examine registration area differences in the
continuum of miscarriages, fetal deaths and early neonatal deaths. Racial disparities in
infant mortality have persisted throughout the last century; determining what proportion
of the disparity is due to statistical artifact and versus an actual excess in deaths is critical

to solving the problem.

114

 

 

CHAPTER 4: RACIAL DIFFERENCES IN THE EFFECT OF PERINATAL
REGIONALIZAT ION ON RACIAL DISPARITIES IN INFANT AND
NEONATAL MORTALITY, MICHIGAN 1996-2006

This chapter includes an examination of racial patterns of birth hospital level and
neonatal intensive care unit (N ICU) transfer among preterm infants in Michigan. It also
explores the subsequent effect on racial disparities in infant and neonatal mortality and is
study 3 for this three paper dissertation option.
4.1 Abstract

Objective: We examined patterns of perinatal regionalization (birth hospital level and

neonatal intensive care unit (N ICU) transfer) among preterm infants in Michigan and the

 

g!
subsequent effect on racial disparities in infant and neonatal mortality.

Methods: Michigan Department of Community Health Vital Records on singleton,
preterm (<37 weeks) live births and infant deaths between 1996 and 2006 (n=107,046
preterm births and 3,950 infant deaths) were used to examine patterns of perinatal
regionalization and mortality (infant and neonatal). Perinatal regionalization measures
and mortality rates were examined by race (non-Hispanic white and non-Hispanic black)
and preterm gestational week. The adjusted relative risk of black infant death (with
whites as the reference category) was examined using logistic regression.

Results: The majority of infants were born at a level 3 hospital (preterm white 62.8%
vs. black 85.3%; very preterm white 83.1 vs. black 91.3% extremely preterm white
83.0% vs. black 89.9%). The highest infant mortality rates were seen among extremely
preterm infants born at level 1 hospitals (465.3 per 1,000 live births) compared to their
level 3 counterparts (336.9 per 1,000 live births) and among extremely preterm level 1

white births (506.1 per 1,000 live births) compared to their black counterparts (3 83.3 per

115

1,000 live births). Gestational week speciﬁc analyses revealed different race-speciﬁc
effects of hospital level at birth on mortality, with the largest racial differences seen
between 23-28 weeks gestation. After adjustment for demographic and hospital
characteristics, extremely preterm, level 3, black infants who were transferred to the
NICU had an increased risk of infant death compared to their white counterparts
(RR=1.25 95% CI=1.07-1.45), while extremely preterm, level 1, black infants who were
transferred to the NICU had a signiﬁcantly decreased risk of infant death, compared to
their white counterparts (RR=0.41 95% CI=O.26-0.66).

Conclusion: Racial disparities in preterm mortality differ by gestational week and
hospital level at birth. Efforts should be made to reduce rates of extremely preterm birth
where mortality rates and racial disparities in risk of adjusted mortality were the highest.
4.2 Introduction
4.2.1 Background on infant and neonatal mortality

The infant mortality rate (number of infant deaths <1 year of age per 1,000 live
births) is commonly used to assess the health and well-being of populations between and
within countries [63]. Depending on the time at which infant mortality occurs, it can be
categorized as neonatal mortality or postneonatal mortality. Neonatal mortality (infant
death between 1 and <28 days) accounts for two thirds of all infant mortality and is
primarily related to exposures and conditions related to pregnancy and soon after birth,
such as medical care. As a result, very preterm (<32 weeks gestation) and extremely
preterm births (<28 weeks gestation) account for the majority of deaths during this time
period [98]. Due to improvements in medical technology and neonatal care (ie.

mechanical ventilation) [99], and more recently the introduction and use of surfactant

116

therapy and antenatal steroids for extremely preterm infants [11]) the US. has seen
signiﬁcant declines in birthweight—speciﬁc infant and neonatal mortality rates over the
years. For example, from 1980 to 2004, the US. infant mortality rate declined 46% from
12.6 to 6.8 infant deaths per 1,000 live births [1, 2].
4.2.2 Racial disparities in infant and neonatal mortality

Despite this decline and improved care for extremely preterm infants, the infant
mortality gap between black and white infants in the US. has increased [1]. For example,
in 1980, the IM rate among whites in the United States was 8.7 per 1,000 live births
compared to an IM rate of 16.5 per 1,000 live births among blacks [4], meaning blacks
were 1.90 times as likely as white infants to die during the ﬁrst year of life. On the other
hand in 2005, the white 1M rate was 5.7 per 1,000 live births compared to the black IM
rate of 13.6 per 1,000 live births among black infants, meaning blacks were 2.39 times as
likely as whites to die during the ﬁrst year of life. Studies have shown that a large
portion of the excess mortality among black infants is due to higher rates of extremely
preterm birth and the subsequent neonatal mortality of black infants [46, 74].
4.2.3 Perinatal regionalization

Extremely preterm infants born in hospitals with neonatal intensive care units (N ICU)
or transferred to such centers immediately after birth have lower mortality and morbidity ’
rates than comparable infants born in other settings [47]. The process of perinatal
regionalization involves a “regionally coordinated system focusing on levels of hospital-
based perinatal care” and has been shown to improve outcomes for both mothers and
newborns [48]. Perinatal regionalization incorporates the use of maternal and/or infant

transport services to ensure that low birthweight or at-risk infants are inborn or promptly

117

 

transferred to appropriate facilities (preferably with NICUs), regardless of where their
mother initially sought obstetrical care [49]. This system evolved to increase the number
of mothers and infants who had access to neonatologists, obstetricians and pediatricians.
Furthermore perinatal regionalization offered improved health care for mothers and
infants and has been adopted by many states and hospital systems [50].

In 1976, the March of Dimes Committee on Perinatal Health designated three levels
of perinatal care. The three basic levels as described in the latest American Academy of
Pediatrics (AAP)/American College of Obstetricians and Gynecologists (ACOG)
guidelines are as follows: level 1 hospitals are able to treat newborns without obstetric
complications and do not have a NICU; level 2 hospitals are able to treat moderately ill
newborns and they may or may not have a NICU; level 3 hospitals always have a NICU
and are able to treat the highest risk infants [48]. Recently, level 2 and level 3 hospitals
have been further categorized by factors such as the volume of extremely preterm infants
and/or the number of NICU beds [51]. Alternatively some states, such as Michigan,
recognize only two levels of care, bypassing the intermediate (level 2) category [52].

The effectiveness of perinatal regionalization can be determined by examining the
proportion of extremely preterm, low birthweight infants which are born at a level 3
hospital or by examining the proportion of extremely preterm, low birthweight infants
which are transferred to a NICU after birth [53-56].

Throughout the 19903, increases in the number of smaller, community NICUs with l
or 2 neonatologists has led to breakdowns in the cooperative relationships between the
less specialized level land level 2 hospitals and the most specialized level 3 facilities

[57]. Furthermore, decreases in state funding of the regionalized transport of mothers and

118

 

 

infants as well as decreases in insurance funding of level 3 care (when less expensive
level 2 care is available) has led to de-regionalization in many states (ie. Michigan). This
de-regionalization is thought to have the largest effect on lower socioeconomic groups
with limited access to level 3 hospitals [3 9]. With the disproportionate representation of
racial/ethnic minorities in lower socioeconomic groups, perinatal de-regionalization
could impact racial disparities in infant and neonatal mortality rates. In addition, state-
level differences in regionalization funding could account for state differences in preterm
or low birthweight infant and neonatal mortality rates. While some studies have
examined the effect of perinatal regionalization on infant and neonatal mortality rates, it
is not clear whether racial differences in mortality can be partially explained by
differences in access to medical care for high risk neonates.
4.2.4 Study Objective

We examined patterns of race-speciﬁc perinatal regionalization among preterm births
in Michigan and the effect of perinatal regionalization on subsequent risk of infant and
neonatal mortality. Speciﬁcally, racial differences were estimated in the following:

1) preterm infant and neonatal mortality rates;

2) hospital level at birth and access to NICU treatment;

3) the hospital level distribution of access to NICU treatment; and

4) the effect of hospital level at birth and access to NICU treatment on infant and

neonatal mortality.
Michigan was chosen due to its high infant mortality rate (Michigan = 6.9 per 1,000 live
births vs. United States = 6.1 per 1,000 live births), high racial disparities in IM

(Michigan disparity ratio = 2.8 and risk differences = 9.4 per 1,000 live births vs. United

119

 

 

States disparity ratio = 2.4 and risk differences = 6.9 1,000 live births) and because of the
historical existence of a well-deﬁned perinatal regionalization system in the 19803 and
19903[56]
4.3 Materials and Methods

We examined Michigan live birth and linked infant death data from 1996-2006.

All preterm (gestational age <37 weeks) live births and infant deaths to non-Hispanic

 

white and non—Hispanic black (herein referred to as white and black), maternal residents
of Michigan were included. :

4.3.1 Data Source

 

Live birth and infant death certiﬁcates were obtained from the Vital Records and J

Health Data Development Section of the Michigan Department of Community Heath
(MDCH). Data were coded according to uniform coding speciﬁcations, passed rigid
quality control standards, were edited and reviewed and are the basis for ofﬁcial
Michigan birth and death statistics. Cohort mortality data for each of the years were used,
as opposed to period data, which links deaths of all infants born in a certain year,
regardless to whether the death occurred during the birth year, or the following year.
Birth cohort data ﬁles were preferred because they follow a given cohort of births for an
entire year to ascertain mortality-speciﬁc information. Birth certiﬁcates included
information on hospital at birth, maternal demographic and pregnancy characteristics and
infant characteristics at birth.
4.3.2 Outcome Variables

Crude infant mortality (death within the ﬁrst year) and neonatal mortality (death

within the ﬁrst month) rates were calculated, in addition to those stratiﬁed by race and

120

gestational week. Gestational age data were based on the clinical estimate of gestation
and were classiﬁed as preterm, deﬁned as <37 completed weeks gestation, very preterm,
deﬁned as <32 completed weeks gestation, and extremely preterm, deﬁned as <28 weeks
gestation.
4.3.3 Exposure Variables

Perinatal regionalization was determined by two variables obtained from the birth
certiﬁcate: hospital level at birth (level 1 vs. level 3) and infant transport to a NICU (yes
vs. no). We assigned of hospital level at birth based on the availability of a NICU, at any
point during the 11 year study period. Hospitals which had a NICU during the study
period were designated level 3, while hospitals which did not have a NICU during the
study period were designated level 1. 1n rare instances where hospital level could not be
obtained from the MDCH, hospital obstetrics departments were contacted by telephone
and asked if they had a NICU at any point in time from 1996-2006. Three hospitals were
contacted directly; one was designated level 3, white the other two were designated level
1. NICU transfer was based on the birth certiﬁcate question: “Was the child transferred to
a neonatal intensive care unit?”. Response options were yes, no and unknown.
Demographic third variables examined in addition the main exposures were obtained
from birth certiﬁcate records and included maternal factors: education (less than high
school, high school diploma, some college, college degree, graduate school or unknown),
age ( less than 20 years, 20—29 years, 30-39 years, 40-49 years, 50+ years, or unknown),
smoking during pregnancy (yes, no or unknown), alcohol use during pregnancy (yes, no
or unknown), adequacy of prenatal care (adequate, intermediate, inadequate or unknown),

and urbanicity (rural, micropolitan or metropolitan); and infant factors: sex (male, female

121

 

 

or unknown) and infant birthweight (500-999, 1000-1499, 1500-1999, 2000-2499, and
2500+).
4.3.4 Exclusion Criteria

To prevent possible bias in the registration of live births or fetal deaths by hospital,
the following exclusions were made in a sequential manner: birthweight < 500 grams
(n=2,5 82) and gestational age < 20 completed weeks (n=314). Furthermore, records
missing information on main exposures of interest: NICU transfer (n=703) and hospital
of birth (n=1,026) were also excluded. Before exclusions there were 111,671 singleton,
preterm births to black and white women in Michigan, the ﬁnal dataset contained
107,046 preterm births to black and white women.
4.3.5 Statistical Analysis

All statistical analyses were carried out using SAS version 9.1.3. We ﬁrst examined
the distribution of maternal and infant demographic characteristics, preterm birth rates,
and perinatal regionalization measures. Second, the proportion of NICU transfers by birth
hospital level were calculated to establish if variation in access to NICU treatment existed
by hospital level at birth. Third, in order to determine if mortality differed by hospital
level at birth or by access to NICU treatment, mortality rates were calculated by hospital
level at birth and by NICU transfer. Crude and race-speciﬁc analyses were carried out to
establish if regionalization patterns in Michigan differed by race. Statistically signiﬁcant
racial differences in exposures and outcomes of interest were quantiﬁed by the Mantel
Hanzel chi-square test. Differences were considered statistically signiﬁcant at the p-value
<0.05 level. Finally, logistic regression was used to determine the effect of hospital level

at birth and NICU transfer on mortality risk, after adjustment for demographic factors.

122

 

 

 

 

 

Relative differences in gestational age speciﬁc mortality risk between white and black
infants were quantiﬁed using the relative risk and statistical signiﬁcance was determined
with a 95% conﬁdence interval. White infants were used as the referent category. All
statistical analyses were conducted using SAS version 9.1.3.

4.4 Results

4.4.] Maternal and infant demographic characteristics among preterm infants

During the study period, there were 107,046 preterm births with 76,044 (71.0%)
to white women and 31,022 (29.0%) to black women. The majority of mothers had a high
school degree (33.9%), were between the ages of 20 and 29 (50.7%) and had adequate
prenatal care (66.7%). Smoking during pregnancy was prevalent in 20.0% of all preterm
births, while alcohol use during pregnancy was prevalent in 1.4% of all preterm births.
The majority of preterm infants were male (53.6%) and had a normal birthweight of
>2500 grams (50%).

Racial differences in the prevalence of numerous maternal characteristics were
apparent. Mothers of black preterm infants more likely than their white counterparts to
have less than a high school degree (31.4% vs. 17.5%), a teen pregnancy (19.2% vs.
10.5%), consume alcohol during the pregnancy (2.4% vs. 1.0%)and have inadequate
prenatal care (25% vs. 10.1%). Black infants were signiﬁcantly more likely than their
white counterparts to be born in a metropolitan county (99.7% vs. 87.7%), with only 12
black preterm births in rural counties over the 11 year period. With respect to sex, the
proportion of black preterm births were evenly split between males and females, while
among whites, 55% of all preterm births were male. More than half of white preterm

births were normal birthweight (54.8%) vs. 38% of black preterm births. Furthermore,

123

 

 

black infants were more than twice as likely as their white counterparts to weigh 500-999

grams (10.1% vs. 4.4%, respectively) (table 4.1).

124

.m'..: 2415.8." ._ - A. ' 4 . " ~
3 1'
- ' -'n -

 

 

Table 4.1. Maternal and infant demographic characteristics by race among preterm

infants, Michigan 19915-2006

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Maternal Total White Black
Characteristics n=107,046 n=76,044 (71.0 n=31,022 (29.0) p'value
Maternal Characteristics

Education <0.0001
< High School 23,018 21.5 13,283 17.5 9,735 31.4
High School Degree 36,307 33.9 25,070 33.0 11,237 36.3
Some College 24,306 22.7 17,762 23.4 6,544 21.1
College Degree 13,238 12.4 11,683 15.4 1,555 5.0
> College 8,026 7.5 7,118 9.4 908 2.9
Unknown 2,151 2.0 1,128 1.5 1,023 3.3
Maternal Age <0.0001
<20 13,912 13.0 7,972 10.5 5,940 19.2
20-29 54,308 50.7 38,429 50.5 15,879 51.2
30-39 35,934 33.6 27,545 36.2 8,389 27.1
40+ 2,889 2.7 2096 2.8 793 2.6
Unknown 3 0.0 2 0.0 1 0.0
Smoking During Pre ancy 0.4956
Yes 21,391 20.0 15,319 20.1 6,072 19.6
No 83,972 78.4 59,526 78.3 24,446 78.9
Unknown 1,683 1.6 1,199 1.6 484 1.6
Alcohol Use During Pregnancy 0.0030
Yes 1,468 1.4 731 1.0 737 2.4
No 103,777 97.0 74,021 97.3 29,756 96.0
Unknown 1,801 1.7 1,292 1.7 509 1.6
Kessner Prenatal Care Index <0.0001
Adequate 71,350 66.7 55,220 72.6 16,130 52.0
Intermediate 19,629 18.3 12,731 16.7 6,898 22.3
Inadequate 15,437 14.4 7,688 10.1 7,749 25.0
Unknown 630 0.6 405 0.5 225 35.7
Urbanicity (County of birth) <0.0001
Rural 2812 2.6 2797 4.7 12 0.1
Micropolitan 6624 6.2 6530 8.6 94 0.3
Metropolitan 97587 91.2 66694 87.7 30893 99.7

 

 

 

 

 

 

 

 

125

 

 

 

Table 4.1 (continued). Maternal and infant demographic characteristics by race
among preterm infants, Michigan 1996-2006

 

Infant Characteristics

 

Sex <0.0001

 

Male 57,411 53.6 41,663 54.8 15,748 50.8

 

Female 49,631 46.4 34,378 45.2 15,253 49.2

 

 

 

 

 

 

 

Unknown 4 0.0 3 0.0 l 0.0

 

Birthweight <0.0001

 

500-999 6,155 5.8 3,332 4.4 2,823 10.1

 

1000-1499 7,101 6.6 4,255 5.6 2,846 92

 

1500-1999 12,965 12.1 8,166 10.7 4,799 15.5

 

2000-2499 27,357 25.6 18,616 24.5 8,741 28.2

 

 

 

 

 

 

 

 

 

 

2500+ 53,466 50.0 41,673 54.8 11,793 38.0

 

4.4.2 Rates of preterm birth
Figure 1 shows racial differences in rates of preterm birth during the study period.
Signiﬁcant racial differences were seen in each preterm category with black infants 1.6
times as likely as whites to experience preterm birth (p-value <0.0001), 2.5 times as
likely as whites to experience very preterm birth (p-value <0.0001), and 3.5 times as

likely as white infants to experience extremely birth (p-value

126

 

 

<0.0001).

Figure 4.1. Rates of preterm birth by race, Michigan 1996-2006

 

 

_
O

 

00

 

 

 

 

 

 

 

Percent of Total Births

 

 

 

Preterm Very Preterm Extremely Preterm
Preterm Category

4.4.3 Perinatal regionalization

Rates of hospital level at birth and NICU transfer are shown in table 4.2. The
majority of preterm, very preterm and extremely infants were born at a level 3 hospital
(69.3%, 86.5%, and 86.2%, respectively). In race-speciﬁc analyses, black infants were
signiﬁcantly more likely than their white counterparts to be born at a level 3 hospital.

This relationship was seen across gestational categories.

127

The majority of extremely and very preterm infants were transferred to a NICU
after birth (79.2% and 78.1%, respectively). Despite this trend, signiﬁcant racial
differences in the proportion of infants transferred to the NICU were seen within each
preterm category. Among extremely infants, 83.2% of blacks were transferred to the
NICU compared to 75.8% of white infants (p-value <0.0001) and among very preterm
infants, 80.9% of blacks were transferred to the NICU compared to 76.1% of white
infants (p-value <0.0001). D

Figure 4.2 Hospital levels, Michigan 1996-2006.

 

 

 

 

Hospital Levels ; ,
+ Level 3 “3 "1

 

 

 

0 Level1 . , . i. .
3. -- ls— l . I. ”T. .l _1_J— "1"” ‘3 .1 I'l‘ -
l I : j l l' . :

 

128

Table 4.2. Hospital level at birth and NICU transfer by race among preterm infants,
Michigan 1996-2006

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Regionalization Total White Black

Measures n=137,702 (100%) n=100,680 (73.1%) n=37,022 (26.9%) p value

N L % N j % N l %
Hospital Level at Birth
Extremer Preterm Birth: <28 weeks <0.0001
1 864 13.8 577 17.0 287 10.1
3 5,382 86.2 2,820 83.0 2,562 89.9
Total 6,246 100.0 3,397 54.4 2,849 45.6
Very Preterm Birth: <32 weeks <0.0001
1 2,148 13.5 1,575 16.9 573 8.7
3 13,728 86.5 7,732 83.1 5,996 91.3
Total 15,876 100.0 9,307 58.6 6,569 41.4
Preterm Birth: <37 weeks <0.0001
1 32,873 30.7 28,318 37.2 4,555 14.7
3 74,173 69.3 47,726 62.8 26,447 85.3
Total 107,046 100.0 76,044 71.0 31,002 23 .0
‘ NICU Transfer

Extremely Preterm Birth: <28 weeks <0.0001
Yes 4,946 79.2 2,576 75.8 2,370 83.2
No 1,300 20.8 821 24.2 479 16.8
Total 6,246 100.0 3,397 54.4 ' 2,849 45.6
Very Preterm Birth: <32 weeks <0.0001
Yes 12,397 78.1 7,083 76.1 5,314 80.9
No 3,479 21.9 2,224 23.9 1,255 19.1
Total 15,876 100.0 9,307 58.6 6,569 41.4
Preterm Birth: <37 weeks <0.0001
Yes 33,792 31.6 21,981 28.9 11,811 38.1
No 73,254 68.4 54,063 71.1 19,191 61.9
Total 107,046 100.0 76,044 71.0 31,002 23.0

 

 

 

 

 

 

 

4.4.4 NICU transfer by birth hospital level

In order to determine if racial differences in the proportion preterm infants

transferred to a NICU were due to differences in hospital level at birth, table 4.3 includes

rates of NICU transfer by birth hospital level and race. The majority of very and

extremely preterm infants were transferred to the NICU, regardless of hospital level at

birth. Rates of NICU transfer were the highest among very and extremely preterm infants

129

 

 

 

born at a level 3 hospital (80.4% and 82.1%, respectively). In race-speciﬁc analyses,
black infants were signiﬁcantly more likely than their white counterparts to be transferred
to a NICU. This relationship was seen across preterm category and hospital level at birth,
with the largest racial differences seen among extremely preterm births in level 1

hospitals (55.6% vs. 71.8% p-value <0.0001).

 

130

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Em 3.8 mg on; 9.2: 3.: we $3. 3” 23a 92: Sag 3,;

3m 5.: Re Soda ms 34.? mi 83.” 2w 2.1% 3» SEN oz

:4 03.2 Em 8a.: 23 39% 2a m8 5: a? we 83 war
586v 38; av ”Em Ease 885v £83 av ”55 E25
5. can now EN 32 Man: SN mm mm 33 :2: $3 38%

2: 83 SN 82 we 33 SN 02 2:. 3 ﬁx a: 02

Ex 33. «a 9.3 vow E»: Mg 5. in as 08 83 new

886 £83 mmv ”gm Esau be, 88.? £025 va “Em €205 PS
9:. 83 2m 8% 0.2: S? N: 2a 4.3. E 92: new 38.

62 Mam 98 SW 0.: 8m 2a a is EN 9% 5 oz

new E3 98 $3 2” 23. w: SN 93 § 3.8 Rm s>
586v £83 ”NV “sea E25 inseam 885v £83 ”NV ”sea 8.95 zoaoexm

m .23 Beam sea _ .23 Eamon gm

Siﬂfx; 2 7x; 2 as; 2 3335:; z a; z _.x._ z Tamas
_. . seam H 333 H Boa e seam . as? H 33 _ BE

 

 

 

.wccﬂéaa— gamma—9:2 sodas—5 .539:— wneﬁa 3.2 an: 553 «a .95— ..Enme: .3 noun—:2. DUHZ Me 8.5— .m... «Bah.

131

 

4.4.5 Infant and neonatal mortality rates

Table 4.4 includes infant and neonatal mortality rates by gestational category and
race. Infant and neonatal mortality rates decreased as gestational ages moved closer to
term. Black infants were signiﬁcantly more likely than white infants to die during the
ﬁrst year of life at very preterm (white=169.5 per 1,000 live births; black=193.5 per
1,000 live births; p-value=0.0001) and preterm (white=30.8 per 1,000 live births;
black=51.9 per 1,000 live births; p-value=<0.0001) gestations. Although racial
differences in infant mortality were not statistically signiﬁcant among extremely preterm
infants, black infants had lower neonatal mortality rates (268.5 vs. 273.8 per 1,000 live
births), and higher infant mortality rates (364.7 per 1,000 live births vs. 346.2 per 1,000
live births) than white infants.

Table 4.4. Infant and neonatal mortality rates (per 1,000 live births) by race among
preterm infants, Michigan 1996-2006

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. Total White Black

Mortalrty Rates N I Rate” N I Rate N 1 Rate p value
Extremely Preterm Birth: <28 weeks
Infant Mortality 2,215 354.6 1,176 346.2 1,039 364.7 0.1280
Neonatal Mortality 1,695 271.4 930 273.8 765 268.5
Live Births 6,246 3,397 2,849
Very Preterm Birth: <32 weeks
Infant Mortality 2,849 179.5 1,578 169.5 1,271 193.5 0.0001
Neonatal Mortality 2,055 129.4 1,183 127.1 872 132.7
Live Births 15,876 9,307 6,569
Preterm Birth: <37 weeks
Infant Mortality 3,950 36.9 2,342 30.8 1,608 51.9 <0.0001
Neonatal Mortality 2,646 24.7 1,627 21.4 1,019 32.9
Live Births 107,046 76,044 31,002

 

 

 

 

*Rate per 1,000 live births

 

In order to determine if race speciﬁc differences in infant and neonatal mortality
rates were due to an excess of births (and deaths) at earlier gestations, tables 4.5 and 4.6
include gestational week speciﬁc infant and neonatal mortality rates by race. In

132

 

gestational week speciﬁc analyses, blacks were at increased risk of infant mortality

between 20 and 21 weeks, 27-28 weeks and 34-36 weeks gestation (table 4.5).

Table 4.5. Gestational week specific infant mortality rates (per 1,000 live births) by

race among preterm infants, Michigan 1996-2006

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total White Black
Gestational Live Infant Live Infant Liv Inf Ratio
Week Births Deaths Rate“ Births Deaths Rate Birtlfs Deeftnhts Rate
Extremely Preterm: < 28 weeks
20 66 40 606.1 34 19 558.8 32 21 656.3 1.2
21 103 92 893.2 60 52 866.7 43 40 930.2 1.1
22 344 308 895.4 180 164 911.1 164 144 878.1 1.0
23 715 496 693.7 364 255 700.6 351 241 686.6 1.0
24 1,055 480 455.0 538 241 448.0 517 239 462.3 1.0
25 1,165 322 276.4 622 174 279.7 543 148 272.6 1.0
26 1,325 272 205.3 742 153 206.2 583 119 204.1 1.0
27 1,402 205 146.2 825 118 143.0 577 87 150.7 1.1
Nery Preterm: <32 weeks
28 1,805 179 99.2 1,055 102 96.7 750 77 102.7 1.1
29 1,849 144 77.9 1,142 97 84.9 707 47 66.5 0.8
30 2,635 163 61.9 1,579 99 62.7 1,056 64 60.6 1.0
31 3,266 148 45.3 2,101 104 49.5 1,165 44 37.8 0.8
Preterm: <37 weeks
32 4,991 143 28.7 3,247 100 30.8 1,744 43 24.7 0.8
33 6,847 153 22.4 4,602 107 23 .3 2,245 46 20.5 0.9
34 12,407 207 16.7 8,697 132 15.2 3,710 75 20.2 1.3
35 21,079 240 11.4 15,454 174 11.3 5,625 66 11.7 1.0
36 45,368 358 7.9 34,467 251 7.3 10,901 107 9.8 1.4

 

 

 

*Rate per 1,000 live births

Black infants were at lower risk of neonatal mortality at each gestational week

except among infants born at 20 weeks gestation and infants born at 36 weeks gestation

(table 4.6).

133

 

ttﬁ

 

Table 4.6. Gestational week speciﬁc neonatal mortality rates (per 1,000 live births)
by race among preterm infants, Michigan 1996-2006

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Gestation Total White Black .
Week Live Neonatal Rate“ Live Neonatal Rate Live Neonatal Rate R3110
Births Deaths Births Deaths Births Deaths

xtremely Preterm: < 28 weeks
20 66 36 545.5 34 18 529.4 32 18 562.5 1.1
21 103 82 796.1 60 48 800.0 43 34 790.7 1.0
22 344 275 799.4 180 146 811.1 164 129 786.6 1.0
23 715 406 567.8 364 214 587.9 351 192 547.0 0.9
24 1055 340 322.2 538 180 334.6 517 160 309.5 0.9
25 1165 216 185.4 622 123 197.8 543 93 171.3 0.9
26 1325 180 135.9 742 102 137.5 583 78 133.8 1.0
27 1402 129 92.0 825 83 100.6 577 46 79.7 0.8

ery Preterm: <32 weeks
28 1805 94 52.1 1055 60 56.9 750 34 45.3 0.8
29 1849 91 49.2 1142 68 59.5 707 23 32.5 0.6
30 2635 89 33.8 1579 62 39.3 1056 27 25.6 0.7
31 3266 82 25.1 2101 62 29.5 1165 20 17.2 0.6

reterm: <37 weeks
32 4991 87 17.4 3247 64 19.7 1744 23 13.2 0.7
33 6847 93 13.6 4602 70 15.2 2245 23 10.2 0.7
34 12407 122 9.8 8697 87 10.0 3710 35 9.4 0.9
35 21079 115 5.5 15454 96 6.2 5625 19 3.4 0.5
36 45368 167 3.7 34467 124 3.6 10901 43 3.9 1.1

 

 

*Rate per 1,000 live births

4.4.6 Infant mortality by birth hospital level

In order to determine if racial disparities in infant and neonatal mortality rates

were due to differences in perinatal regionalization, table 4.7 examines infant and

neonatal mortality rates by hospital level at birth. Hospital level differences in mortality

rates were seen across gestational categories, with extremely and very preterm level 1

infants experiencing the highest rates. For example, the very preterm infant mortality rate

was 234.4 per 1,000 live births among infants born in a level 1 hospital, compared to the

134

 

“ ,3!“ .
.-

 

infant mortality rate of 170.8 per 1,000 live births among infants born in a level 3
hospital. The most striking infant mortality rates were seen among extremely preterm,
white infants, who experienced a 50% infant mortality rate if they were born at a level 1
hospital, compared to a 31% infant mortality rate if they were born in a level 3 hospital.
Within race-speciﬁc analyses, black infants had signiﬁcantly higher infant mortality rates

than white infants of similar gestational ages. This relationship was seen across hospital

level at birth. P]

 

135

 

 

ﬁes 2,: 83 ea 3%...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. . . . . . 2E5
E. N eNN S N: E 5 a M: N N New Nm 3:
”.9 ma EN EN; N.NN NSN N.NN «N2 3: ONV m2 is 5282
88.9 N.Nm N} we... a: 9N4 N03 58.? we. 2: EN m8 3N 3N see
8303 nmv 5:5 5.535 8363 \rmv “:Em €8on
. . . . . 3:5
8a m NNN N NNN 2 E E _ a: N 35
52 SN pm: wow ES 82 E: z: Na: NaN 3: am 3%an
58.? 02: mm: 32 SN; 3: EN 533 N.NNN a: N.NNN Nam 0.va 3m SE
383 va “aim 8.535 bo> £83 mmv “aim E895 bo>
. . . $25
New N ONN N Nam m NaN 5 new 3:
SeN NS EN :6 NR a: 23 a 29. 2N Sam 02 3282
38¢ 9N8 Na 2: a; $3 :3 good 3% c: 3% NQN 22. N9. Has
$83 ”NV ”Sam €8on 325be 383 mV ”:Em €8on boEobxm
m 354 .358: Sam _ :53 .338: in
a: ersai z _aai z ﬁes: z, issuer z 782. z ...saE z _ 8am
an.» 33>

_ N85 _ 323 ﬁ 33. r M85 _ 333 309 _ buss:

 

 

33.33 5322 £55
.589:— uueﬁn 3a.. 15 5.3 3.321393: .3 35:: o>= c2: .55 8:: 3:33.: .328: can “:8:— SJ. 93:.

136

Gestational week speciﬁc mortality analyses by hospital level at birth were
undertaken to determine if hospital level infant and neonatal mortality rates differed by
gestational week. Tables 4.8 and 4.9 show race-speciﬁc infant mortality rates among
level 1 and level 3 births. Overall, race-speciﬁc mortality rates were comparable among
level 3 births, as opposed to the race-speciﬁc mortality rates among level 1 births. The
higher rates of extremely preterm mortality among white, level 1 infants in table 7 were
due to decreased white survival among births between 23-27 weeks gestation. On the
other hand, the higher rates of extremely preterm mortality among level 3 black births
were due decreased black survival during the same gestations. In fact, compared to black
level 1 births, birth at a level 3 hospital increased the risk of black infant mortality
between 23-27 weeks. On the other hand, white level 3 infants between 23-27 weeks had
lower mortality rates than their level 1 counterparts. Similar patterns were seen among

level 1 and level 3 neonatal mortality rates (tables 4.10 and 43.11, respectively).

137

 

 

 

Table 4.8. Gestational week specific infant mortality rates among level 1 births by
race, Michigan 1996-2006.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Gestational Total White Black .
Week Live Infant Rate” I..ive Infant Rate Dive Infant Rate Ratio
Bll’thS Deaths Blrths Deaths Blrths Deaths
Extremely Preterm: < 28 weeks
20 21 13 619.1 15 8 533.3 6 5 833.3 1.6
21 15 13 866.7 12 10 833.3 3 3 1000.0 1.2
22 68 63 926.5 53 48 905.7 15 15 1000.0 1.1
23 85 59 694.1 55 44 800.0 30 15 500.0 0.6 r}
24 134 75 559.7 85 55 647.1 49 20 408.2 0.6
25 137 56 408.8 90 39 433.3 47 17 361.7 0.8 :
26 160 45 281.3 116 39 336.2 44 6 136.4 0.4
27 173 38 219.7 119 30 252.1 54 8 148.2 0.6 i . .
Very Preterm: <32 weeks 3 I
28 206 23 111.7 152 14 92.1 54 9 166.7 1.8 E’- l
29 233 20 85.8 190 16 84.2 43 4 93.0 1.1 i
30 329 25 76.0 261 21 80.5 68 4 58.8 0.7
31 441 27 61.2 362 23 63.5 79 4 50.6 0.8 i
Preterm: <37 weeks I
32 824 30 36.4 657 21 32.0 167 9 53.9 1.7
33 1,196 30 25.1 998 26 26.1 198 4 20.2 0.8
34 3,040 45 14.8 2,586 37 14.3 454 8 17.6 1.2
35 7,173 60 8.4 6,247 52 8.3 926 8 8.6 1.0
36 18,014 105 5.8 15,985 92 5.8 2,029 13 6.4 1.1

 

 

 

 

 

 

 

 

 

 

 

 

*Rate per 1,000 live births

138

 

Table 4.9. Gestational week speciﬁc infant mortality rates among level 3 births by

race, Michigan 1996-2006.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Gestational Total White Black .
Week Dive Infant Rate" Dive Infant Rate hive Infant Rate R3110
Blrths Deaths Blrths Deaths Births Deaths

Extremely Preterm: < 28 weeks
20 45 27 600.0 19 11 579.0 26 16 615.4 1.1
21 88 77 875.0 48 41 854.2 40 36 900.0 1.1
22 276 239 866.0 127 111 874.0 149 128 859.1 1.0
23 360 353 980.6 309 209 676.4 321 222 691.6 1.0
24 921 393 426.7 453 179 395.1 468 214 457.3 1.2
25 1,028 260 252.9 532 134 251.9 496 126 254.0 1.0
26 1,165 221 189.7 626 112 178.9 539 109 202.2 1.1
27 1,229 165 134.3 706 87 123.2 523 78 149.1 1.2

Very Preterm: <32 weeks
28 1,599 154 96.3 903 88 97.5 696 66 94.8 1.0
29 1,616 122 75.5 952 80 84.0 664 42 63.3 0.8
30 2,306 136 59.0 1,318 77 58.4 988 59 59.7 1.0
31 2,825 120 42.5 1,739 81 46.6 1,086 39 35.9 0.8

reterm: <37 weeks

32 4,167 112 26.9 2,590 79 30.5 1,577 33 20.9 0.7
33 5,651 122 21.6 3,604 81 22.5 2,047 41 20.0 0.9
34 9,367 160 17.1 6,111 94 15.4 3,256 66 20.3 1.3
35 13,906 175 12.6 9,207 118 12.8 4,699 57 12.1 1.0
36 27,354 248 9.1 18,482 157 8.5 8,872 91 10.3 1.2

 

 

*Rate per 1,000 live births

139

 

 

Table 4.10. Gestational week speciﬁc neonatal mortality rates among level 1 births
by race, Michigan 1996-2006.

 

      

 

Extremely Preterm: < 28 weeks
20 21 13 619.1 15 8 533.3 6 5 833.3 1.6
21 15 11 733.3 12 9 750.0 3 2 666.7 0.9
22 68 61 897.1 53 46 867.9 15 15 1000.0 1.2
23 85 53 623.5 55 39 709.1 30 14 466.7 0.7

24 134 64 477.6 85 48 564.7 49 16 326.5 0.6
25 137 39 284.7 90 29 322.2 47 10 212.8 0.7
26 160 37 231.3 116 32 275.9 44 5 113.6 0.4
27 173 27 156.1 119 22 184.9 54 5 92.6 0.5
Very Preterm: <32 weeks

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

28 206 14 68.0 152 8 52.63 54 6 111.1 2.1
29 233 12 51.5 190 10 52.63 43 2 46.5 0 9
30 329 16 48.6 261 15 57.47 68 1 14.7 0 3
31 441 17 38.6 362 15 41.44 79 2 25.3 0 6
Preterm: <37 weeks
32 824 16 19.4 657 12 18.26 167 4 24.0 1 3
33 1,196 16 13.4 998 13 13.03 198 3 15.2 12
34 3,040 29 9.5 2,586 26 10.05 454 3 6 6 0.7
35 7,173 35 4.9 6,247 32 5.12 926 3 3.2 0.6
36 18,014 42 2.3 15,985 36 2.25 2,029 6 3.0 1 3

 

*Rate per 1,000 live births

140

 

Table 4.11. Gestational week specific neonatal mortality rates among level 3 births
by race, Michigan 1996-2006.

 

 

 

 

 

 

Gag/23‘ Live INeonatal Rate" Live eOnatal Rate Live Neonatal Rate . m9.
72951;. j Births Deaths ‘ Births Deaths Births Deaths _ '

 

 

 

 

 

 

Extremely Preterm: < 28 weeks
20 45 23 511.1 19 10 526.3 26 13 500.0 1.0
21 88 71 806.8 48 39 812.5 40 32 800.0 1.0
22 276 214 775.4 127 100 787.4 149 1 14 765.1 1.0
23 360 353 980.6 309 175 566.3 321 178 554.5 1.0
24 921 276 299.7 453 132 291.4 468 144 307.7 1.1
25 1,028 177 172.2 532 94 176.7 496 83 167.3 1.0
26 1,165 143 122.8 626 70 111.8 539 73 135.4 1.2
27 1,229 102 83.0 706 61 86.4 523 41 78.4 0.9
Very Preterm: <32 weeks ~. .. '
28 1,599 80 50.0 903 52 57.6 696 28 40.2 0.7 J
29 1,616 79 48.9 952 58 60.9 664 21 31.6 0.5
30 2,306 73 31.7 1,318 47 35.7 988 26 26.3 0.7
31 2,825 65 23.0 1,739 47 27.0 1,086 18 16.6 0.6
Preterm: <37 weeks
32 4,167 71 17.0 2,590 52 20.1 1,577 19 12.1 0.6
33 5,651 77 13.6 3,604 57 15.8 2,047 20 9.8 0.6
34 9,367 93 9.9 6,111 61 10.0 3,256 32 9.8 1.0
35 13,906 80 5.8 9,207 64 7.0 4,699 16 3.4 0.5

36 27,354 125 4.6 18,482 88 4.8 8,872 37 4.2 0.9
*Rate per 1,000 live births

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4.4.7 Racial differences in the adjusted risk of infant mortality

Table 4.12 examines racial disparities in the risk of infant death, using white
infants as the referent category. Due to effect modiﬁcation by hospital level and NICU
transfer, relative risks presented were stratiﬁed by the two exposures of interest (hospital
level and NICU transfer). Fully adjusted, statistically signiﬁcant results were bolded.
Aﬁer adjustment for confounding demographic and medical variables, the risk of infant

mortality among extremely preterm black infants who were delivered at a level 3 hospital

141

-_/_ ..

 

 

and subsequently transferred to a NICU was signiﬁcantly higher than the mortality risk
among their white counterparts (RR=1.25 95% CI=1.07-1.45). On the other hand,
extremely preterm black infants who were born in a level 1 hospital and transferred to a
NICU were signiﬁcantly less likely than their white counterparts to die during their ﬁrst
year of life (RR=O.41 95% CI=O.26-0.66). The risk of infant mortality among black, very

preterm infants was signiﬁcantly less than the risk for their white counterparts among

.3

level 1 NICU transferred infants (RR=0.58 95% CI=0.40-0.85) and among level 3 infants

who were not transferred to a NICU (RR=0.73 95% CI=0.57-O.94).

 

{ML

142

 

 

 

38 3855 we 388388 8.8 56232 $23:on
2:28 om: 3:028 £28:qu mam-28 wad—cam £028.60 .0? .03 EcouSmow “8328; ma§o=£ 05 88302 3808 88m: _.8<*

3235 352-200 $3 H

xma gum—om ..-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GN. 73. 8 3._ 93.8.8 N2. <2 <2 83 :82 +
ANN. 78. 8 o: 63.8.8 :3 $3.38 NS 8N. 788 N3 <8 +
5.788 c: 22.88 a; 83.88 83 ANS-3.8 who 3m +
28.73. 8 :1 8:788 88 ANN-.788 o? @788 o? twain-885
$872.8 8.2 $788 N: 82-5.8 a? AN_.N-wN.8 38 839.235
8.83 2v 53m Essa

8N. 7S. 8 42 8375.8 m2. <2 <2 88 :02 +
8N. 7s. 8 42 cad-$8 :8 33.3.8 and an. 72.8 3.0 <8 +
an. 73. 8 o: Ago-$8 Ed sad-2.8 3.0 2.758 who 26 +
Anna-4N8 a? God-$8 ”3 278.8 33 @738 N: 828855
AS. 7Nm.8 :1 8978.8 m3 @788 88 9878.8 :2 833.68:
383 NV "58m @322 be.»

5.78.8 m3 8 _ . 78.8 38 <2 <2 ...Bm 882 +
$4. 78. 8 22 5.78.8 £8 68.8.38 8.... 2:. 78.8 $8 <8 +
82-2.8 82 as. 78.8 :3 and-3.8 and cm. 738 m; 25 +
28.78. 8 3: A8. 72.8 :8 82.88 as $738 38 8288085
as. 73.. 8 8: §. 78.8 33 ANS-N88 ”to 28.728 :88 839.825

 

 

383 wmv ”58mm 8.5on beau-2m

143

 

 

Baa-.5 :02 _EWEH 82. 2,: £88... 82 _saase :02 oz_ bases.
. _233 _ . A_ :33. . .._ .35

 

 

 

25732 5222 .522. :02
8:8 :25 «a .95. .332:— .3 3.5.5.3.. ma 85E: 33.5 5.38 «ca-«E we saw.— oﬁ ”56:53 mourn—3 82.—«>552 .Ngé «Bah.

4.5 Discussion

Access to highly specialized care is critical among extremely preterm infants. Nearly
two thirds of the decrease in neonatal mortality over the past 20 years has been
attributable to the neonatal management and care of extremely preterm infants [100], as
well as improved capability of specialized centers [101].

This study investigated racial differences in rates of preterm birth and the effect of
hospital level at birth and access to NICU treatment on racial disparities in infant and
neonatal mortality rates. Signiﬁcant racial differences were found in rates of preterm
birth, which have been demonstrated in other studies [3 8, 55]. The largest racial
differences were seen among extremely preterm infants, who subsequently had the
highest risk of mortality.

4.5.1 Perinatal regionalization

The vast majority of very preterm infants in Michigan were delivered at level 3
hospitals, with only 14% delivered at a level 1 hospital. Other studies have found wide
variation in the proportion of very preterm/low birthweight infants born in a level 3
hospital: 82% of very low birthweight births [102], 60% very low birthweight births
[103], 82.2% of very low birthweight births [54], 81% of very low birthweight [53], 88%
less than 2000 grams [104], 81% of very low birthweight births [105], 68% of very
preterm births [106], 67% of very low birthweight births [107], 75% of very preterm
births [108], 78% of very low birthweight births [57], 79% of very preterm births [109],
and 73.3% of extremely preterm births [55]. While it is positive that the majority of very
preterm infants in each of these studies were born in a level 3 hospital, a problem with

interpreting the results is that many of them were based on research collaboratives in

144

 

 

 

large teaching hospitals. Differences in study population or hospital characteristics can
inﬂuence results with large area (ie. state) analyses more prone to ﬁnd lower rates of
level 3 births, while analyses from research collaboratives based around large teaching
hospitals (ie. Vermont Oxford Network) were more likely to ﬁnd higher rates of level 3
births.

Similar to other studies, black mothers in our study were more likely than their white
counterparts to deliver at a level 3 hospital, regardless of preterm gestational age category
[39, 110]. Contrary to our ﬁnding of higher rates of NICU transfer among black infants,
regardless of gestational age or birth hospital level, Bronstein et al found that race did not
affect the likelihood that infants born in level 1 hospitals would be transferred to a NICU
after delivery [39].

4.5.2 Mortality and perinatal regionalization

Mortality rates were highest among infants who were born at a level 1 hospital,
regardless of gestational age or NICU transfer status. These results support the 1976
Committee on Perinatal Health statement that survival improves for high risk infants that
are delivered at hospitals with a NICU, and supports other studies that found similar
results [53, 55, 57, 102, 104, 105, 109, 111, 112]. For example, Paneth et al found that
preterm and low birth weight infants were at a 24% higher risk of mortality if birth
occurred outside of a level 3 center, regardless of whether birth occurred at a level 1 or
level 2 hospital [113]. Historically prompt transfer from a level 1 to a level 3 hospital has
been found to greatly reduce mortality risk [114], but more recently, for infants who are

born in a level 1 hospital, subsequent transfer to a level 3 hospital has been found to only

145

 

 

marginally decrease the risk of mortality [104], further supporting the need for high risk
infants to be inborn at a level 3 hospital.
4.5.3 Racial disparities in mortality

Despite the ﬁnding that black infants were more likely than their white counterparts
to deliver at a level 3 hospital, mortality rates were elevated among black infants,
especially among level 3 births. The only exception was a signiﬁcantly decreased
mortality risk among black infants who were born at a level 3 hospital but not
subsequently transferred to a NICU. Since it is puzzling that any infant <32 weeks
gestation who was born at a level 3 hospital would not be transferred to a NICU, we
further examined the time of death for this small group of infants (565 white and 398
black) and found that 88% of white deaths and 85% of black deaths occurred before the
infants could be transferred (<1 hour). A perplexing elevation in risk among black
infants was seen among extremely preterm infants who were born at a level 3 hospital
and subsequently transferred to a NICU. This ﬁnding contradicts ﬁndings from four
previous perinatal regionalization studies which found lower mortality rates among very
preterm/low birthweight black infants compared to their white counterparts after
adjusting for hospital-speciﬁc characteristics [51, 102, 104, 112], but agree with those of
Bronstein et al who found that black low birthweight infants had a 35% elevated risk of
mortality after adjustment for socio-demographic characteristics [39]. In our analyses,
adjustment for elevated proportions of demographic risk factors such as inadequate
prenatal care, alcohol use and gestational week did little to reduce the black/white
disparity. Further analyses showed discrepancies in the birth hospital level and death

hospital level among extremely preterm black neonatal deaths (not shown). Speciﬁcally,

146

 

despite the ﬁnding that overall, black infants were more likely to be born at a level 3
hospital than white infants, among births that subsequently resulted in a death, extremely
preterm black infants were less likely than their white counterparts to be born at a level 3
hospital (black = 80% vs. white = 82%). Likewise, among all deaths, extremely preterm
black infants were less likely than their white counterparts to die at a level 3 hospital
(black = 77% vs. white = 78%). But, race-speciﬁc concordance analyses between hospital
level at birth and hospital level at death did not substantiate racial differences with 99%
of both white and black neonatal deaths among level 3 births occurring at a level 3
hospital as well (not shown).

Previous studies [111, 112, 115] have identiﬁed several hospital characteristics
associated with health outcomes among infants. Lower staff-to-infant ratios have been
associated with higher morality rates among very low birthweight infants [111], although
racial differences in mortality risk were not examined. Howell et al (2008) found that
black infants were more likely to be born at hospitals with higher risk-adjusted mortality
and this explained more than one third of the black/white disparity in extremely low
birthweight neonatal mortality [112]. Morales et al found that minority serving hospitals
had signiﬁcantly higher risk adjusted mortality rates for both white and black infants than
hospitals where less than 15% of infants were black (White OR=1.30, 95% CI=1.09-1.56;
Black OR= 1.29, 95% CI = 1.01-1.64) which were not explained by either hospital
characteristics or treatment variables [115]. It is possible that black infants in our sample
were more likely than white infants to be born at minority serving hospitals with higher
risk adjusted mortality rates, and this led to the excess risk among black infants across

preterm categories. An examination of hospital speciﬁc mortality rates demonstrated a

147

 

 

 

slight indication of higher mortality rates among minority serving institutions (>15%
black), but 79% of the level 3 hospitals in our study fell into this category, so it is
difficult to make strong inferences based on this information.

A growing body of evidence suggests that outcomes for surgical procedures and
medical diagnoses are better for patients treated in hospitals with higher volumes of
patients receiving similar procedures [57, 78, 116]. Upon examining 11 year volume
within our study, we found a trend in mortality rates with the highest rates seen in level 3
hospitals with >2000 preterm births (108.4 per 1,000 live births) and the lowest mortality
in hospitals with <1000 preterm births (49.4 per 1,000 live births). In contrast, Phibbs et
al found that compared with a high level of care and a high volume of very low
birthweight infants, lower levels of care and lower volumes were associated with
signiﬁcantly higher odds of death [57]. This is not always the case with high risk
neonatal and infant mortality. Some studies have not found a relation between hospital
volume and mortality. Horbar reported no effect of NICU patient volume on mortality
outcomes of very low birthweight infants [102]. The referral of very low birthweight
infants based on patient volume was minimally effective, accounting for only 1% of
hospital variation in mortality outcomes in a 2004 study [51]. Authors from the same
study found that the largest indicator of hospital quality was the mortality rate from the
previous year, which accounted for 34% of hospital level variation in mortality rates [51].
Reasons for delivery at certain hospitals and NCU transfer involve a number of factors
whose evaluation was not in the scope of this project such as place of residence [117],
distance to hospitals [118], managed care and HMOs [107], which could all vary by race.

4.6 Conclusions

148

 

Our study found signiﬁcant racial disparities in rates of preterm birth, hospital level at
birth and NICU transfer among various gestational categories. Despite higher rates of
level 3 births, racial disparities in preterm infant and neonatal mortality persist. Reasons
for this elevation in black mortality rates are unclear, and limitations in available vital
statistics data prevented the full exploration of hospital-speciﬁc characteristics.
Furthermore, national data on hospital level at birth and the prevalence of NICU transfers
are not available in the National Center for Health Statistics Linked Birth/Infant Death
Records, so it was not possible to see if this relationship differed by state. Future studies
should supplement state and national vital records with hospital speciﬁc characteristics
which inﬂuence infant mortality rates and racial disparities in infant mortality rates (ie.
infant to staff ratios). Although large teaching collaboratives, such as the Vermont
Oxford Network offer detailed hospital speciﬁc information, all infants are not born in
medical facilities with a NICU and therefore may be included in related studies. The
addition of data on all hospitals, especially level 1 hospitals with the highest mortality
rates, is essential for reducing overall infant mortality rates and racial disparities in infant

mortality.

149

 

 

CHAPTER 5: CONCLUSION

The primary focus of this research was to determine reasons for nationwide
variation in racial disparities in IM. Issues related to disparity measurement (relative vs.
absolute measures) and three key exposures (inequalities in sociodemographic factors,
infant and fetal reporting practices and perinatal regionalization) were examined in order
to gain a better understanding of the composition of racial disparities in IM. Study aims
were centered on state level factors which are amenable to intervention, with hopes of
inﬂuencing future research agendas along with state and national policy. These aims
allowed the untangling of a complex issue by simultaneously examining multiple factors
which could inﬂuence racial disparities in IM. Furthermore, this study evaluated two
policy-relevant disparity measures in relation to IM. Noteworthy results from this study
include the ﬁndings that racial inequalities in the proportion of very low birthweight and
very preterm infant births along with state differences in reporting very low birthweight
and very preterm births were consistently associated with national variation in IM
disparities. Racial inequalities in perinatal regionalization, however did not account for
higher infant or neonatal mortality rates among black infants in Michigan.

Our study found that relative and absolute measures of disparity provided similar, but
not identical results when examining inequality in infant, maternal and state
sociodemographic factors. They provided different results, however, when examining
state-level racial disparities in IM. State differences in the proportion of VLBW births
were the strongest predictor of racial disparities in IM. Furthermore, inequalities in
foreign born status and percent poverty were also signiﬁcantly correlated with national

variation in IM disparities. Inequalities in maternal factors did not play a strong role in

150

3',
1r-

- wimp __

 

Li."

 

nationwide variation in IM disparities in fully adjusted analyses, but relative and absolute
measures of inequality in teen pregnancy were signiﬁcantly correlated in partially
adjusted models. Due to different IM disparity results depending on the measure used,
future studies examining reasons for state-level differences in racial disparities in IM,
should use both relative and absolute disparity measures. Care should also be taken when
examining nationwide variation in racial disparities in IM interpretation of results should
include discussion of how results differ, depending on the disparity measure used. While
relative and absolute disparity measures were signiﬁcantly correlated with each other,
they exhibited different relationships with [M rates. Total IM was only correlated with
the absolute disparity measure, while white IM was only correlated with the relative
measure. Although both disparity measures were signiﬁcantly correlated with black IM,
this relationship was stronger when using the absolute measure of disparity as opposed to
the relative measure. This discovery reinforced the ﬁnding that the absolute disparity
measure more closely reﬂected black IM rates than did the relative disparity measure.
Due to differences in disparity measure associations with [M rates, future research on
racial disparities in IM should examine both relative and absolute disparity measures.
Furthermore, high correlation between inequalities in the proportion of very low
birthweight births and disparities in IM indicates that future research should work on
extracting more meaning from infant birthweight. Lower birthweight for gestational age
among black infants has yet to be fully explained.

Overall, registration area differences and viability differences in mortality rates,
absolute disparities and relative disparities were most pronounced among infants 5 22

weeks gestation. Simply excluding infants less than 20 weeks gestation and/or less than

151

 

500 grams did not erase registration area differences in IM rates or disparities. Although
signiﬁcant area and classiﬁcation differences were seen for both races, differences were
most prominent among black infants. That proportion non-viable categories were similar
among infant and fetal deaths indicates real category differences in not only the
pr0portion of at-risk infants born (ie. extremely preterm/low birthweight) but also
indicates real differences in racial disparities between these categories. Absolute
disparities as measured by the disparity differences were most sensitive to area
differences in disparities, especially when overall mortality rates were high.

State differences in fetal death reporting requirements lead to differences in the
reported number of live births, infant deaths and fetal deaths. These differences persist
even after limiting data to gestations which should be reported by all states. In order for
national vital statistics data to be meaningful, a uniform deﬁnition of fetal death must be
adopted. Based on the lowest IM rates and disparities seen in fetal death registration
areas which report all products of conception, we recommend reporting all products of
conception as fetal deaths. The reporting of all products of conception by each
registration area would allow the examination of racial differences in the continuum of
miscarriages, fetal deaths, early neonatal deaths and neonatal deaths with vital statistics
data. This would reduce systematic differences in the reporting of live births and fetal
deaths to give an accurate record of actual mortality rates and disparities. Future studies
should examine proportion non-viable differences in relative and absolute disparities.
Studies should also examine registration area differences in the continuum of

miscarriages, fetal deaths and early neonatal deaths.

152

 

Finally, with respect to perinatal regionalization, our study found signiﬁcant racial
disparities in rates of preterm birth, hospital level at birth and NICU transfer among
speciﬁc gestational categories in Michigan. Despite higher rates level 3 births among
black infants, racial disparities in preterm infant and neonatal mortality persist with black
infants twice as likely as their white counterparts to die during the ﬁrst year of life.
Reasons for this elevation in black mortality rates are unclear, and limitations in available
vital statistics data prevented the full exploration of hospital-speciﬁc characteristics.
Furthermore, national data on hospital level at birth and the prevalence of NICU transfers
are not available in the National Center for Health Statistics Linked Birth/Infant Death
Records, so it is not possible to determine if this relationship differs by state. Future
studies should supplement state and national vital records with hospital speciﬁc
characteristics found to inﬂuence infant mortality rates and racial disparities in infant
mortality rates (ie. infant to staff ratios). Although large teaching collaboratives, such as
the Vermont Oxford Network offer detailed hospital speciﬁc information, all infants are
not born in medical facilities with a NICU and therefore may be included in related
studies. The addition of data on all hospitals, especially level 1 hospitals with the highest
mortality rates, is essential for reducing overall infant mortality rates and racial
disparities in infant mortality.

A nationwide analysis of racial differences in hospital level at birth and NICU
transfers needs to be done. State differences in these factors could contribute
substantially to nationwide variation in IM rates and to variation in racial disparities in
IM. The addition of hospital level at birth and NICU transfer to the national, publicly

available linked birth, infant death dataset would provide an unbiased study population

153

 

publicly available for analysis.

154

 

 

APPENDICIES

155

 

Appendix A. Exposure variable speciﬁc adjustment for fully adjusted partial

correlation model 3.

APPENDIX

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Exposure Variable Model 3 Partial Correlation Ad'Lustments
Inequalities Maternal I State
Infant
Low Birthweight
Low Birthweight Very Low Birthweight Proportion Black
Unknown Father RR 2 High School RD
Very Low Birthweight Teen Pregnancy RD Poverty RR
Tobacco Use RR
Preterm
Preterm Very Preterm Proportion Black
Unknown Father RD 2 High School RD
Very Preterm Teen Pregnancy RD Poverty RR
Tobacco Use RR
Maternal
Teen Pregnancy Proportion Black
Teen Pregnancy Unknown Father RD 2 High School RD
Tobacco Use RR Poverty RR
Proportion Black
Unknown Father U611]: ﬂegaggr 2 High School RD
Poverty RR
. < High School Proportion Black
< ngh SChOOI Teen Pregnancy RD Unemployed RR
Unmarried Unmarried Proportion Black
Teen Pregnancy RD 2 ngh School RD
Tobacco Use Teeiol: 36:12:?“ Proportion Black
Alcohol Use
Alcohol Use Teen Pregnancy RD Proportion Black
< High School RR
Inadequate Prenatal Care Inadequate Prenatal Care Proportion Black
Tobacco Use RR
State
Foreign Born < High School RR Prloportron Black
orergn Born
Proportion Black
< High School “‘3‘ ”32‘?ng < High School
nmame Poverty RR
Proportion Black
_>_ High School T631 Pregnancy RD 2 High School
nmarrred DR
Poverty RR

 

 

 

156

 

 

Appendix A (continued). Exposure variable speciﬁc adjustment for fully adjusted
partial correlation model 3.

 

 

 

 

 

 

Exposure Variable Model 3 Partial Correlation Adjustments
Inequalities Maternal State
Teen Pregnancy RD Proportion Black
2 College Unmarried DR 2 College
Proportion Black
Unem 10 ed Unknown Father RR Unemployment
p y Teen Pregnancy RD 2 High School
Poverty RR
. Proportion Black
Poverty < High School RR Povergl

 

 

 

 

*Each of the exposure variables of interest are inequalities (ie. relative risk or risk
difference)

157

 

 

 

 

«route own 382830 nv 82

«E85 33035—5 Hm 83w
«IBM-6 own Raouﬂmow can Ewﬁaehm MN 374.
8088888 no 308888 =< "8 38¢.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

- . .. ..Hmwra 4.2-”WM”

 

v.8 mew 8.: 0: 8.0 _d _d v no.2.
md me 0.3 QM: md 28 #8 m 82¢
m6 ﬁmm 0.: NE v.8 ﬁo :8 N 82¢.
md EVN 8.2 <3 md 28 :8 ﬂ 8.2.

. ”E 3.0-_-

. 8......

.NacN

-eccu £33m 803:9 39.4 :etﬂuﬁmmﬂu .3qu ?«oh .3 .2325 8:“ 3t:— «again—ta «Etc—tom he swans—couch .m— 58:25<

 

158

Appendix C. Percent of Borderline Infants Who Survive to Age 1 by Fetal Death
Classiﬁcation Area; United States, 2000-2002

 

Are 0/ Birthweight <500 grams Birthweight 500-999 grams
3’ ° Total White Black Total White Black

 

 

Total 16.7 16.3 17.2 69.2 68.7 70.0

 

Area 1 10.5 8.9 12.0 68.1 66.9 69.3

 

Area 2 16.6 14.6 18.4 68.5 67.8 69.4

 

Area 3 10.4 11.4 9.1 68.4 67.8 69.6

 

 

 

 

 

 

 

 

 

Area 4 18.8 18.6 19.1 69.8 69.4 70.5

 

Area 1= All products of conception

Area 2=Birthweight and gestational age criteria
Area 3= Birthweight criteria

Area 4: Gestational age criteria

 

 

159

 

 

«tot-6 own ﬁaowﬁmow H8 «.65-

atoﬁo £303.35 Hm 82¢.

«tux-6 own Raouﬁmow 2.8 Emmozﬁﬁmmum 82
8038888 mo $2688 =< u— 8.2..

85 owe. Ecoﬂﬁmow 28 Ho: 20?
0: Wm wdﬁ .m . . H ﬂ m; . M: wd v «03.
”N m; v v.“ n. . v m. 82
5.5 m6 0.8 8.8 m8 . m4 m4 N 8.3.

.3 N8 m8 Nd 8.8 5.8 8.8 . . . . 2 o; 8 88¢.

    

85m 8. mNm 88 ad _A mm 5; ON. m4» . . . H m; H

«SN-88 .832 38.5 322. 8888830
.5qu Eom .3 «3. 1.53380 .825 «gamete-gm ESE—=3 .23 283A— 308 85 :5:— ac ousnouaom d 5.593

160

Ho» $5 How 325

o was .8888: 35 82.665 x

 

mmmmw

 

vﬁme

 

omomw

 

wwomw

 

mmomw

 

wwomw

 

vmomw

 

o_omw

 

wooﬁw

 

mooﬁw

 

voovw

 

«comb

 

VCOMB

 

m_om©

 

mﬁomo

 

oOOmo

 

scone

 

hocNm

 

oﬁoﬁv

 

mooﬁv

 

«comm

 

~80mm

 

Noemm

 

 

 

 

 

 

 

 

 

 

WASH-89‘: ET aim—SH ...mo

 

 

 

 

.a £5... 3.: m 325. we venowohm .m— 585.. .<

  

 

 

161

Appendix F. Age at death by hospital of birth level and hospital of death level
among all deaths, Michigan 1996-2006.

 

Age at Death Birth Level Death Level
1 3 1 3

 

 

Neonatal 601 (18.1) 2719 (81.2 390 (11.8) 2930 (88.3)

 

 

 

 

 

 

 

Postneonatal 216(17.9) 989(82.1) 353 (29.3) 852 (70.7)

 

162

 

Appendix G. Concordance between hospital level at birth and hospital level at death
among extremely preterm (<28 weeks) neonatal deaths, Michigan 1996-2006.

 

 

 

 

 

 

 

Birth Level Death Level

1 g 3
1 228 (60.5) 149 (38.5)
3 17 (0.9) 1821 (99.1)

 

 

*199 deaths <1 hour (87.3%)

163

 

 

Appendix H. Concordance between hospital level at birth and hospital level at death
among extremely preterm (<28 weeks) white neonatal deaths, Michigan 1996-2006.

 

 

 

 

 

 

 

Birth Level Death Level
1 3
l 182 (61.5) 114 (38.5)
12 (1.2) 998 (98.8)

 

 

164

 

Appendix I. Concordance between hospital level at birth and hospital level at death
among extremely preterm (<28 weeks) black neonatal deaths, Michigan 1996-2006.

 

 

 

 

 

Birth Level Death Level

1 3
l 46 (56.8) 35 (43.2)
3 5 (0.6) 823 (99.4)

 

 

 

 

165

'I1‘ :2 .1.’ y
9" *
f

 

 

 

BIBLIOGRAPHY

166

10.

BIBLIOGRAPHY

Iyasu, S., K. Tomashek, and W. Barﬁeld, Infant mortality and low birth weight
among black and white infants-- United States, 1980-2000. MMWR CDC Surveill
Summ, 2002. 51(27): p. 589-592.

Mathews, T.J. and KO. Keppel, Racial/Ethnic disparities in infant mortality--
United States, 1995-2002. MMWR CDC Surveill Summ, 2005. 54(22): p. 553-
556.

Alexander, G.R., M.S. Wingate, D. Bader, and MD. Kogan, The increasing
racial disparity in infant mortality rates: composition and contributors to recent
US trends. Am J Obstet Gynecol, 2008. 198(1): p. 51 e51-59.

Marks, J .S., J .W. Buehler, L.T. Strauss, C.J. Hogue, and J .C. Smith, Variation in
state-specific infant mortality risks. Public Health Rep, 1987. 102(2): p. 146-151.

Wise, P.H., The anatomy of a disparity in infant mortality. Annu Rev Public
Health, 2003. 24: p. 341-362.

Mathews, T.J. and M.F. MacDorman, Infant mortality statistics from the 2004
period linked birth/infant death data set. Natl Vital Stat Rep, 2007. 55(14): p. 1-
32.

DiGiuseppe, D.L., D.C. Aron, L. Ranbom, D.L. Harper, and GE. Rosenthal,
Reliability of birth certiﬁcate data: a multi-hospital comparison to medical
records information. Matem Child Health J, 2002. 6(3): p. 169-179.

Anderson, R.N., A.M. Minino, D.L. Hoyert, and H.M. Rosenberg, Comparability
of cause of death between [CD-9 and [CD-10: preliminary estimates. Natl Vital
Stat Rep, 2001. 49(2): p. 1-32.

Krous, H.F., J .B. Beckwith, R.W. Byard, T.O. Rognum, T. Bajanowski, T. Corey,
E. Cutz, R. Hanzlick, T.G. Keens, and EA. Mitchell, Sudden infant death
syndrome and unclassiﬁed sudden infant deaths: a deﬁnitional and diagnostic
approach. Pediatrics, 2004. 114(1): p. 234-238.

Ananth, C.V., K.S. Joseph, K. Demissie, and A.M. Vintzileos, Trends in twin
preterm birth subtypes in the United States, 1989 through 2000: impact on
perinatal mortality. Am J Obstet Gynecol, 2005. 193(3 Pt 2): p. 1076-1082.

167

 

 

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

Lee, K.S., N. Paneth, L.M. Gaitner, M.A. Pearlman, and L. Gruss, Neonatal
mortality: an analysis of the recent improvement in the United States. Am J
Public Health, 1980. 70(1): p. 15-21.

Scott, CL, S. Iyasu, D. Rowley, and H.K. Atrash, Postneonatal mortality
surveillance-- United States, 1980-1994. MMWR CDC Surveill Sum, 1998.
47(2): p. 15-30.

Beck, L.F., B. Morrow, L.E. Lipscomb, C.H. Johnson, M.E. Gafﬁeld, M. Rogers,
and BC. Gilbert, Prevalence of selected maternal behaviors and experiences, '
Pregnancy Risk Assessment Monitoring System (PRAMS), I 999. MMWR Surveill
Summ, 2002. 51(2): p. 1-27.

Gortmaker, S.L., The eﬂects of prenatal care upon the health of the newborn. Am
J Public Health, 1979. 69(7): p. 653-660.

Poma, P.A., Eﬂect of prenatal care on infant mortality rates according to birth-
death certiﬁcate ﬁles. J Natl Med Assoc, 1999. 91(9): p. 515-520.

Salihu, H.M., M.H. Aliyu, B.J. Pierre-Louis, and GR. Alexander, Levels of excess
infant deaths attributable to maternal smoking during pregnancy in the United
States. Matem Child Health J, 2003. 7(4): p. 219-227.

Mathews, T.J., F. Menacker, and M.F. MacDorman, Infant mortality statistics
from the 2002 period: linked birth/infant death data set. Natl Vital Stat Rep,
2004. 53(10): p. 1-29.

Lundsberg, L.S., M.B. Bracken, and AF. Saﬁlas, Low-to-moderate gestational
alcohol use and intrauterine growth retardation, low birthweight, and preterm
delivery. Ann Epidemiol, 1997. 7(7): p. 498-508.

Martin, J .A., B.E. Hamilton, P.D. Sutton, S.J. Ventura, F. Menacker, and ML.
Munson, Births: ﬁnal data for 2002. Natl Vital Stat Rep, 2003. 52(10): p. 1-113.

Keppel, K.G., Ten largest racial and ethnic health disparities in the United States
based on Healthy People 2010 Objectives. Am J Epidemiol, 2007. 166(1): p. 97-
103.

Carter-Pokras, O. and C. Baquet, What is a "health disparity"? Public Health Rep,
2002. 117(5): p. 426-434.

Whitehead, M., The concepts and principles of equity and health. Int J Health
Serv, 1992. 22(3): p. 429-445.

168

 

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

Harper, 8., J. Lynch, S.C. Meersman, N. Breen, W.W. Davis, and M.H.
Reichman, An overview of methods for monitoring social disparities in cancer
with an example using trends in lung cancer incidence by area-socioeconomic
position and race-ethnicity, 1992-2004. Am J Epidemiol, 2008. 167(8): p. 889-
899.

Harper, S., J. Lynch, S.C. Meersman, N. Breen, W.W. Davis, and ME.
Reichman, Harper et al respond to ”Measuring social disparities in health ". Am J
Epidemiol, 2008. 167: p. 905-907.

Messer, L.C., Invited commentary: measuring social disparities in health--what
was the question again? Am J Epidemiol, 2008. 167(8): p. 900-904; author reply
908-916.

Houweling, T.A., A.E. Kunst, M. Huisman, and J .P. Mackenbach, Using relative
and absolute measures for monitoring health inequalities: experiences from
cross-national analyses on maternal and child health. Int J Equity Health, 2007.
6: p. 15.

Pearcy, J .N. and K.G. Keppel, A summary measure of health disparity. Public
Health Rep, 2002. 117(3): p. 273-280.

Braveman, P., C. Cubbin, K. Marchi, S. Egerter, and G. Chavez, Measuring
socioeconomic status/position in studies of racial/ethnic disparities: maternal and
infant health. Public Health Rep, 2001. 116(5): p. 449-463.

Goldhagen, J ., R. Remo, T. Bryant, 3rd, P. Wludyka, A. Dailey, D. Wood, G.
Watts, and W. Livingood, The health status of southern children: a neglected
regional disparity. Pediatrics, 2005. 116(6): p. e746-753.

Stockwell, E.G., Infant mortality and socio-economic status: a changing
relationship. Milbank Mem Fund Q, 1962. 40: p. 101-111.

Allen, D.M., J .W. Buehler, C.J. Hogue, L.T. Strauss, and J .C. Smith, Regional
differences in birth weight-speciﬁc infant mortality, United States, 1 980. Public
Health Rep, 1987. 102(2): p. 138-145.

Baurnan, KB. and A.E. Anderson, Legal abortions and trends in fetal and infant
mortality rates in the United States. Am J Obstet Gynecol, 1980. 136(2): p. 194-
202.

169

 

-A\.

 

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

Bird, ST. and KB. Baurnan, The relationship between structural and health
services variables and state-level infant mortality in the United States. Am J

Public Health, 1995. 85(1): p. 26-29.

Brooks, CH, The changing relationship between socioeconomic status and infant
mortality: an analysis of state characteristics. J Health Soc Behav, 1975. 16(3): p.
291-303.

Kleinman, J .C., State trends in infant mortality, 1968-83. Am J Public Health,
1986. 76(6): p. 681-688.

Wilson, A.L., L.J. F enton, and DP. Munson, State reporting of live births of

newborns weighing less than 500 grams: impact on neonatal mortality rates.
Pediatrics, 1986. 78(5): p. 850-854.

Schoendorf, K.C., C.J. Hogue, J .C. Kleinman, and D. Rowley, Mortality among

infants of black as compared with white college-educated parents. N Engl J Med,
1992. 326(23): p. 1522-1526.

O'Campo, P., J .G. Burke, J. Culhane, I.T. Elo, J. Eyster, C. Holzman, L.C.
Messer, J .S. Kaufman, and BA. Laraia, Neighborhood deprivation and preterm

birth among non-Hispanic Black and White women in eight geographic areas in
the United States. Am J Epidemiol, 2008. 167(2): p. 155-163.

Bronstein, J .M., E. Capilouto, W.A. Carlo, J .L. Haywood, and R.L. Goldenberg,
Access to neonatal intensive care for low-birthweight infants: the role of maternal
characteristics. Am J Public Health, 1995. 85(3): p. 357-361.

Ogburn, W. and D. Thomas, The inﬂuence 'of business cycle on certain social
conditions. Journal of the American Statistical Association, 1922. 19: p. 324-340.

LaVeist, T., The political empowerment and health status of African Americans:
Mapping a new territory. American Journal of Sociology, 1992. 97(4): p. 1080-
1095.

McKnight, J .L., Health and empowerment. Can J Public Health, 1985. 76 Suppl
1: p. 37-38.

Wingate, MS. and GR. Alexander, Racial and ethnic diﬂerences in perinatal
mortality: the role of fetal death. Ann Epidemiol, 2006. 16(6): p. 485-491.

170

 

 

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

Kowaleski, J ., State deﬁnitions and reporting requirements for live births, fetal
deaths, and induced terminations of pregnancy (I 997 revision). N.C.f.H.
Statistics, Editor. 1997: Hyattsville, Maryland.

Phelan, S.T., R. Goldenberg, G. Alexander, and SP. Cliver, Perinatal mortality
and its relationship to the reporting of low-birthweight infants. Am J Public

Health, 1998. 88(8): p. 1236-1239.

Joseph, KS. and MS. Kramer, Recent trends in infant mortality rates and
proportions of low-birth-weight live births in Canada. CMAJ, 1997. 157(5): p.
535-541.

Campbell, M.K., Assessment of regionalized perinatal programs. J Dev Physiol,
1991. 15(2): p. 125-131.

Guidelines for prenatal care, A.A.o.P.A.C.o.O.a. Gynecology, Editor. 1997: Elk
Grove, IL and Washington DC p. 1.

Wise, P.H., L.R. First, G.A. Lamb, M. Kotelchuck, D.W. Chen, A. Ewing, H.
Hersee, and J. Rideout, Infant mortality increase despite high access to tertiary
care: an evolving relationship among infant mortality, health care, and
socioeconomic change. Pediatrics, 1988. 81(4): p. 542-548.

Van Mullem, C., A.E. Conway, K. Mounts, D. Weber, and CA. Browning,
Regionalization of perinatal care in Wisconsin: a changing health care
environment. WMJ, 2004. 103(5): p. 35-38.

Rogowski, J .A., J .D. Horbar, D.G. Staiger, M. Kenny, J. Carpenter, and J.
Geppert, Indirect vs direct hospital quality indicators for very low-birth-weight
infants. JAMA, 2004. 291(2): p. 202-209.

Paneth, N., The role of neonatal intensive care in lowering infant mortality, in A
pound of prevention: The case for university maternity care in the US, A. Press,
Editor. 1992. p. 49-63.

Bode, M.M., M. O'Shea T, K.R. Metzguer, and AD. Stiles, Perinatal
regionalization and neonatal mortality in North Carolina, 1968-1994. Am J
Obstet Gynecol, 2001. 184(6): p. 1302-1307.

Gerber, S.E., D.G. Dobrez, and RP. Budetti, Managed care and perinatal
regionalization in Washington State. Obstet Gynecol, 2001. 98(1): p. 139-143.

171

 

 

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

Kamath, B.D., T.L. Box, M. Simpson, and J .A. Hernandez, Infants born at the
threshold of viability in relation to neonatal mortality: Colorado, I 991 to 2003. J
Perinatol, 2008. 28(5): p. 354-360.

Paneth, NS. and MR. Rip, The uses of epidemiology in the evaluation of regional
perinatal services. 802 Praventivmed, 1994. 39(1): p. 3-10.

Phibbs, C.S., L.C. Baker, A.B. Caughey, B. Danielsen, S.K. Schmitt, and RH.
Phibbs, Level and volume of neonatal intensive care and mortality in very-low-
birth-weight infants. N Engl J Med, 2007. 356(21): p. 2165-2175.

 

SChOCDdOrf, KC. and A.M. Branum, The use of United States vital statistics in :
perinatal and obstetric research. Am J Obstet Gynecol, 2006. 194(4): p. 911-915. 1; '

Statistics, N.C.f.H., 2000-2002 linked birth/infant death dataset, C.f.D.C.a.P.
United States Department of Health and Human Services, Editor. 2000-2002.

 

Cai, J ., G.L. Hoff, F. Okah, P.C. Dew, G. Zaborac, X. Somoza, L. Jones, P. F
Livingston, M.J. Everhardt, and R. Archer, Fetal mortality: timing of racial
disparities. J Natl Med Assoc, 2007. 99(11): p. 1258-1261.

Kessner, D.M., J. Singer, C.E. Kalk, and ER. Schlesinger, Infant Death: An
analysis by maternal risk and health care, I.o.M.a.N.A.o. Sciences, Editor. 1973:
Washington DC.

Alexander, G.R., The accurate measurement of gestational age--a critical step
toward improving fetal death reporting and perinatal health. Am J Public Health,
1997. 87(8): p. 1278-1279.

Martin, J .A. and F. Menacker, Expanded health data ﬁom the new birth
certiﬁcate, 2004. Natl Vital Stat Rep, 2007. 55(12): p. 1-22.

Healthy People 2010: Understanding and improving health, U.S.D.o.H.a.H.
Services, Editor. 2000, Government Printing Office: Washington DC.

Singh, GK. and SM. Yu, Infant mortality in the United States: trends,
differentials, and projections, 1950 through 2010. Am J Public Health, 1995.
85(7): p. 957-964.

America 's health starts with healthy children: How do states compare? 2008,
Robert Wood Johnson Foundation: Commission to Build a Healthier America. p.
1-23.

172

 

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

Goldenberg, R.L., G.R. Cutter, K.G. Nelson, and J. Foster, Effects of very low
birth weights on fetal and neonatal mortality rates in Alabama. Public Health
Rep, 1989. 104(5): p. 488-492.

Thompson, L.A., D.C. Goodman, C.H. Chang, and TA. Stukel, Regional
variation in rates of low birth weight. Pediatrics, 2005. 116(5): p. 1114-1121.

Luke, B. and MB. Brown, Maternal risk factors for potential maltreatment
deaths among healthy singleton and twin infants. Twin Res Hum Genet, 2007.
10(5): p. 778-785.

Goza, F.W., E.G. Stockwell, and KS. Balistreri, Racial differences in the :1
relationship between infant mortality and socioeconomic status. J Biosoc Sci, 1
2007. 39(4): p. 517-529.

 

Polednak, A.P., Black-white differences in infant mortality in 38 standard
metropolitan statistical areas. Am J Public Health, 1991. 81(11): p. 1480-1482. '

Sohler, N.L., P.S. Arno, C.J. Chang, J. Fang, and C. Schechter, Income inequality
and infant mortality in New York City. J Urban Health, 2003. 80(4): p. 650-657.

Gage, T.B., F. Fang, E. O'Neill, and H. Stratton, Maternal age and infant
mortality: a test of the Wilcox-Russell hypothesis. Am J Epidemiol, 2009. 169(3):
p. 294-303.

Lee, K.S., N. Paneth, L.M. Gartner, and M. Pearlman, The very low-birth-weight
rate: Principal predictor of neonatal mortality in industrializedpopulations. J
Pediatr, 1980. 97(5): p. 759-764.

Kramer, M.S., C.V. Ananth, R.W. Platt, and KS. Joseph, US Black vs White
disparities in foetal growth: physiological or pathological? Int J Epidemiol, 2006.
35(5): p. 1187-1195.

Papiemik, E., G.R. Alexander, and N. Paneth, Racial differences in pregnancy
duration and its implications for perinatal care. Med Hypotheses, 1990. 33(3): p.
181-186.

Wilcox, A.J., On the importance--and the unimportance--0fbirthweight. Int J
Epidemiol, 2001. 30(6): p. 1233-1241.

Phipps, M.G., M. Sowers, and SM. DeMonner, The risk for infant mortality

among adolescent childbearing groups. J Womens Health (Larchmt), 2002.
11(10): p. 889-897.

173

 

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

David, R.J. and J .W. Collins, Jr., Diﬂering birth weight among infants of US.-
born blacks, African-born blacks, and (J.S.-born whites. N Engl J Med, 1997.
337(17): p. 1209-1214.

Rosenberg, K.D., R.A. Desai, and J. Kan, Why do foreign-born blacks have lower
infant mortality than native-born blacks? New directions in A frican-American
infant mortality research. J Natl Med Assoc, 2002. 94(9): p. 770-778.

LaVeist, T.A., Simulating the effects of poverty on the race disparity in
postneonatal mortality. J Public Health Policy, 1990. 11(4): p. 463-473.

Laveist, T.A., Segregation, poverty, and empowerment: health consequences for
Aﬁican Americans. Milbank Q, 1993. 71(1): p. 41-64.

Sims, M., T.L. Sims, and MA. Bruce, Urban poverty and infant mortality rate
disparities. J Natl Med Assoc, 2007. 99(4): p. 349-356.

Shi, L., L.H. Green, and S. Kazakova, Primary care experience and racial
disparities in self—reported health status. J Am Board Fam Pract, 2004. 17(6): p.
443-452.

Bird, S.T., Separate black and white infant mortality models: diﬁ’erences in the
importance of structural variables. Soc Sci Med, 1995. 41(11): p. 1507-1512.

Collins, J .W., Jr., R.J. David, R. Symons, A. Handler, S.N. Wall, and L. Dwyer,
Low-income African-American mothers’ perception of exposure to racial
discrimination and infant birth weight. Epidemiology, 2000. 11(3): p. 337-339.

Lu, MC. and N. Halfon, Racial and ethnic disparities in birth outcomes: a life-
course perspective. Matem Child Health J, 2003. 7(1): p. 13-30.

Singh, GK. and MD. Kogan, Persistent socioeconomic disparities in infant,
neonatal, andpostneonatal mortality rates in the United States, 1969-2001.
Pediatrics, 2007. 119(4): p. e928-939.

Paulson, J ., Unregistered deaths among extremely low birthweight infants--0hio,
2006. MMWR CDC Surveill Sum, 2007. 115(1): p. e44-e51.

Hessol, NA. and E. Fuentes-Afﬂick, Ethnic differences in neonatal and
postneonatal mortality. Pediatrics, .2005. 1 15(1): p. e44-51.

174

91.

92.

93.

94.

95.

96.

97.

98.

99.

100.

101.

Iyasu, S., J .E. Becerra, D.L. Rowley, and C.J. Hougue, Impact of very low
birthweight on the black-white infant mortality gap. Am J Prev Med, 1992. 8(5):
p. 271-277.

Luke, B. and MB. Brown, The changing risk of infant mortality by gestation,
plurality, and race: 1989-1991 versus 1999-2001. Pediatrics, 2006. 118(6): p.
2488-2497.

Wen, S.W., L.M. Chen, C.Y. Li, M.S. Kramer, and AC. Allen, The impact of
missing birth weight in deceased versus surviving fetuses and infants in the

comparison of birth weight-speciﬁcfeto-infant mortality. Chronic Dis Can, 2002.
23(4): p. 146-151.

Alexander, G.R., M. Kogan, D. Bader, W. Carlo, M. Allen, and J. Mor, US birth
weight/gestational age-specific neonatal mortality: 1995-1 997 rates for whites,
hispanics, and blacks. Pediatrics, 2003. 111(1): p. e61-66.

Kramer, MR. and CR. Hogue, Place matters: variation in the black/white very
preterm birth rate across US. metropolitan areas, 2002-2004. Public Health Rep,
2008. 123(5): p. 576-585.

Allen, C.L., T.M. Hulsey, and H. T.C., The inﬂuence of race on fetal outcome.
American Journal of Perinatology, 2005. 22(5): p. 245-248.

Yudkin, P.L., L. Wood, and CW. Redman, Risk of unexplained stillbirth at
different gestational ages. Lancet, 1987. 1(8543): p. 1192-1194.

Ananth, C.V., K.S. Joseph, Y. Oyelese, K. Demissie, and A.M. Vintzileos, Trends
in preterm birth and perinatal mortality among singletons: United States, 1989
through 2000. Obstet Gynecol, 2005. 105(5 Pt 1): p. 1084-1091.

Gregory, G.A., J .A. Kitterman, R.H. Phibbs, W.H. Tooley, and WK. Hamilton,
Treatment of the idiopathic respiratory-distress syndrome with continuous
positive airway pressure. N Engl J Med, 1971. 284(24): p. 1333-1340.

Richardson, D.K., J .E. Gray, S.L. Gortmaker, D.A. Goldmann, D.M. Pursley, and
MC. McCormick, Declining severity adjusted mortality: evidence of improving
neonatal intensive care. Pediatrics, 1998. 102(4 Pt 1): p. 893-899.

Pollack, L.D., An eﬂective model for reorganization of perinatal services in a

metropolitan area: a descriptive analysis and historical perspective. J Perinatol,
1996. 16(1): p. 3-8.

175

102.

103.

104.

105.

106.

107.

108.

109.

110.

111.

Horbar, J .D., G.J. Badger, E.M. Lewit, J. Rogowski, and PH. Shiono, Hospital
and patient characteristics associated with variation in 28-day mortality rates for
very low birth weight infants. Vermont Oxford Network. Pediatrics, 1997. 99(2):
p. 149-156.

Mehta, S., H.D. Atherton, P.J. Schoettker, R.W. Homung, P.H. Perlstein, and
U.R. Kotagal, Diﬂerential markers for regionalization. J Perinatol, 2000. 20(6): p.
366-372.

Cifuentes, J ., J. Bronstein, C.S. Phibbs, R.H. Phibbs, S.K. Schmitt, and WA.
Carlo, Mortality in low birth weight infants according to level of neonatal care at ﬂ_
hospital of birth. Pediatrics, 2002. 109(5): p. 745-751. a

Gould, J .B., A.R. Marks, and G. Chavez, Expansion of community-based
perinatal care in California. J Perinatol, 2002. 22(8): p. 630-640.

 

Vieux, R., J. Fresson, J .M. Hascoet, B. Blondel, P. Truffert, J .C. Roze, J. Matis, '

G. Thiriez, C. Amaud, L. Marpeau, and M. Kaminski, Improving perinatal
regionalization by predicting neonatal intensive care requirements of preterm
infants: an EPIPA GE-based cohort study. Pediatrics, 2006. 118(1): p. 84-90.

Dobrez, D., S. Gerber, and P. Budetti, Trends in perinatal regionalization and the
role of managed care. Obstet Gynecol, 2006. 108(4): p. 839-845.

Alidina, S., S. Ardal, P. Lee, L. Raskin, A. Sherman, and L.M. Young,
Regionalization reigns--but is care being delivered accordingly? An evaluation of

perinatal care delivery in a regionalized child health network. Healthc Manage
Forum, 2006. 19(2): p. 22-26.

Rautava, L., L. Lehtonen, M. Peltola, E. Korvenranta, H. Korvenranta, M. Linna,
M. Hallman, S. Andersson, M. Gissler, J. Leipala, O. Tammela, and U. Hakkinen,
The effect of birth in secondary- or tertiary-level hospitals in Finland on mortality
in very preterm infants: a birth-register study. Pediatrics, 2007. 119(1): p. e257-
263.

Gessner, 8D. and RT. Muth, Perinatal care regionalization and low birth weight
infant mortality rates in Alaska. Am J Obstet Gynecol, 2001. 185(3): p. 623-628.

Callaghan, L.A., D.W. Cartwright, P. O'Rourke, and M.W. Davies, Infant to staff
ratios and risk of mortality in very low birthweight infants. Arch Dis Child Fetal
Neonatal Ed, 2003. 88(2): p. F94-97.

176

 

112.

113.

114.

115.

116.

117.

118.

Howell, E.A., P. Hebert, S. Chatterjee, L.C. Kleinman, and MR. Chassin,
Black/white differences in very low birth weight neonatal mortality rates among
New York City hospitals. Pediatrics, 2008. 121(3): p. e407-415.

Paneth, N., J .L. Kiely, S. Wallenstein, and M. Susser, The choice of place of
delivery. Effect of hospital level on mortality in all singleton births in New York
City. Am J Dis Child, 1987. 141(1): p. 60-64. .

Paneth, N., J .L. Kiely, and M. Susser, Age at death used to assess the effect of
interhospital transfer of newborns. Pediatrics, 1984. 73(6): p. 854-861.

Morales, L.S., D. Staiger, J .D. Horbar, J. Carpenter, M. Kenny, J. Geppert, and J.
Rogowski, Mortality among very low-birthweight infants in hospitals serving
minority populations. Am J Public Health, 2005. 95(12): p. 2206-2212.

Luft, H.S., Regionalization of medical care. Am J Public Health, 1985. 75(2): p.
125-126.

Larson, E.H., L.G. Hart, and RA. Rosenblatt, Is non-metropolitan residence a
risk factor for poor birth outcome in the US. ? Soc Sci Med, 1997. 45(2): p. 171-
188.

Mori, R., M. Fujimura, J. Shiraishi, B. Evans, M. Corkett, H. Negishi, and P.
Doyle, Duration of inter-facility neonatal transport and neonatal mortality:
systematic review and cohort study. Pediatr Int, 2007. 49(4): p. 452-458.

177

 

  

 

3 1293 03062 6 66

”HIWIHHII‘HIHIWIIWWINN1M||W|HlllllI