USF Libraries
USF Digital Collections

Low documented risk cesarean sections and late-preterm births :

MISSING IMAGE

Material Information

Title:
Low documented risk cesarean sections and late-preterm births : the florida experience
Physical Description:
Book
Language:
English
Creator:
Clayton, Heather
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Preterm Birth
Cesarean
Infant Morbidity
Maternal Morbidity
Dissertations, Academic -- Community and Family Health -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: There are increasing concerns about the excessive use of cesarean delivery in the United States, as cesarean deliveries have been associated with adverse maternal and infant health outcomes. Currently, the cesarean section (C/S) rate for Florida is the second highest in the nation. Furthermore, preliminary reports from the Florida Department of Health (FDOH) have implicated the increasing rate of cesarean delivery to an increase in the rate of late preterm births (PTB) in Florida (births at 34 to 36 weeks gestational age). Information on the impact of late PTB associated with cesarean delivery on the rate of maternal and infant morbidity in Florida as well as corresponding utilization of health care services is scarce. Information on the validity of data sources used to investigate infant and maternal health outcomes in Florida is also scarce. Therefore, the objectives of this research project were: (1) to determine the validity of data sources used to investigate low documented risk C/S and late PTB, and (2) to assess the impact of low documented risk C/S on maternal and infant morbidity and subsequent healthcare utilization. To determine the accuracy of data elements reported on the Florida birth certificate and hospital discharge data, sensitivity, specificity, positive predictive value, negative predictive value, kappa statistics and likelihood ratios were calculated. To assess differences in morbidity by route of delivery, generalized estimating equations and vii survival analyses were employed. Markov Chain Monte Carlo methods were used to determine appropriate morbidities for inclusion in all analyses. Differences in accuracy of data by data source was observed, with linked birth certificate and hospital discharge data demonstrating improved accuracy compared to birth certificate and discharge data alone. Further, significant differences in the rate of maternal and infant morbidity by route of delivery were observed, with cesarean delivery increasing the risk of adverse health outcomes, and intensive use of healthcare services.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Heather Clayton.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0003409
usfldc handle - e14.3409
System ID:
SFS0027724:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003409
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Clayton, Heather.
0 245
Low documented risk cesarean sections and late-preterm births :
b the florida experience
h [electronic resource] /
by Heather Clayton.
260
[Tampa, Fla] :
University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: There are increasing concerns about the excessive use of cesarean delivery in the United States, as cesarean deliveries have been associated with adverse maternal and infant health outcomes. Currently, the cesarean section (C/S) rate for Florida is the second highest in the nation. Furthermore, preliminary reports from the Florida Department of Health (FDOH) have implicated the increasing rate of cesarean delivery to an increase in the rate of late preterm births (PTB) in Florida (births at 34 to 36 weeks gestational age). Information on the impact of late PTB associated with cesarean delivery on the rate of maternal and infant morbidity in Florida as well as corresponding utilization of health care services is scarce. Information on the validity of data sources used to investigate infant and maternal health outcomes in Florida is also scarce. Therefore, the objectives of this research project were: (1) to determine the validity of data sources used to investigate low documented risk C/S and late PTB, and (2) to assess the impact of low documented risk C/S on maternal and infant morbidity and subsequent healthcare utilization. To determine the accuracy of data elements reported on the Florida birth certificate and hospital discharge data, sensitivity, specificity, positive predictive value, negative predictive value, kappa statistics and likelihood ratios were calculated. To assess differences in morbidity by route of delivery, generalized estimating equations and vii survival analyses were employed. Markov Chain Monte Carlo methods were used to determine appropriate morbidities for inclusion in all analyses. Differences in accuracy of data by data source was observed, with linked birth certificate and hospital discharge data demonstrating improved accuracy compared to birth certificate and discharge data alone. Further, significant differences in the rate of maternal and infant morbidity by route of delivery were observed, with cesarean delivery increasing the risk of adverse health outcomes, and intensive use of healthcare services.
590
Advisor: Elizabeth Gulitz, Ph.D.
653
Preterm Birth
Cesarean
Infant Morbidity
Maternal Morbidity
690
Dissertations, Academic
z USF
x Community and Family Health
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.3409



PAGE 1

Low Documented Risk Cesarean Sections a nd Late Preterm Births: The Florida Experience by Heather Breeze Clayton A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy De partment of Community and Family Health College of Public Health University of South Florida Co Major Professor: Elizabeth Gulitz Ph.D. Co Major Professor: Hamisu M. Salihu, M.D. Ph.D. Donna J. Petersen Sc.D. Charles S. Mahan, M.D. William M. Sappenfield, M.D. Date of Approval: March 31, 2010 Keywords: P reterm B irth, C esarean, I nfant M orbidity, M aternal M orbi dity Copyright 2010 Heather Breeze Clayton

PAGE 2

Dedication I would like to dedicate this dissertation to my family my mother Sharon Clayton, my father Daniel Clayton, my other mother Sylvia Tucker, and my sister Tammy Clayton. Their love and support helped me to reach my goals, and I would not be who I am today without them. I love you all more than words can express.

PAGE 3

Acknowledgements While the dissertation is an independent exercise, it is not a solitary journey, but the result of guidance and support from many individuals. First, I would like to thank my dissertation committee members, Dr. Elizabeth Gulitz, Dr. Hamisu Salihu, Dr. William Sappenfield, Dr. Charles Mahan and Dr. Donna Petersen f or their guidance I would also like to thank Dr. Julie Baldw in, the Chair of the Department of Community and Family Health (US F), and Dr. Hamisu Salihu, the Research Director of the Lawton and Rhea Chiles Center (USF), for their support of the Florida Late Preterm Cesarean Investigation. I was very fortunate to h ave the opportunity to collaborate with the Florida Department of Health and with the state MCH Epidemiologist, Dr. William Sappenfield as I worked on my dissertation. Dr. Sappenfield as well as other faculty members, Dr. Salihu and Dr. Jeannine Corei l have taken me under their wing and provided me with a wealth of information that will serve me in my future. Several others provided support for this dissertation and I would like to take this opportunity to thank them Dr. Alfred Mbah (thanks for assisting with the Markov!), Kara Stanley Dave Hogeboom Dee Jeffers and Dr. Eugene Declercq. I was fortunate to receive the MCH Epidemiology Traineeship during my graduate studies which allowed me to further develop my skills as an MCH Epidemiologi st. Lastly, it is important to acknowledge the FDOH, the March of Dimes and the Florida Obstetric and Gynecologic Society for their funding of the Florida Late Preterm Cesarean Investigation, without which, much of this dissertation would not have been possible.

PAGE 4

! i Table of Contents List of Tables ................................ ................................ ................................ ... iv List of Figures ................................ ................................ ................................ .... v Abstract ................................ ................................ ................................ ............ vi Chapter One: Introduction ................................ ................................ ................. 1 Purpose of Study ................................ ................................ .................. 1 Dissertation Format ................................ ................................ .............. 1 Background ................................ ................................ .......................... 2 Trends in Cesarean Deliveries in the United States ............................. 3 Adverse Outcomes of Cesarean Deliveries ................................ ......... 4 Risk Factors for Cesarean Delivery ................................ ..................... 6 Preterm Birth ................................ ................................ ........................ 7 Risk Factors for Preterm Birth ................................ ........................... 10 Economic Consequences of C/S and Preterm Birth .......................... 11 Research Focusing on Both Late Preterm Birth and C/S ................... 14 Limitations of the Literature ................................ .............................. 15 Prelim inary Research ................................ ................................ ......... 16 Theoretical Framework ................................ ................................ ...... 16 The Social Ecological Model ................................ ............................. 19 Application of SEM to Research in Maternal and Child Health ....... 20 SEM as a Framework for Describing Low Documented Risk C/S ... 2 3 Individual Factors ................................ ................................ 24 Interpersonal Factors ................................ ............................. 24 Community/Institutional Factors ................................ .......... 25 Societal Factors ................................ ................................ ..... 25 Research Purpose ................................ ................................ ............... 26 Study Design ................................ ................................ ...................... 27 Specific Aims and Research Questions ................................ ............. 28 Research Questions ................................ ............................... 2 8 Data Sources for Proposed Research Questions ................................ 2 9 The FDOH abstraction investigation data file ...................... 2 9 Florida birth certificate data file ................................ ........... 31 Hospital discharge data file ................................ ................... 31 Case Definition Algorithm ................................ ................................ 32 Plan for the Dissemination of Study Findings ................................ ... 33 Manuscript Option Introduction ................................ ........................ 35

PAGE 5

! ii Chapter Two: Maternal and Child Health Journal Manuscript ..................... 36 Manuscript One Title ................................ ................................ ......... 36 Introduction ................................ ................................ ........................ 37 Materials and Methods ................................ ................................ ....... 38 Statistical analyses ................................ ................................ 41 Results ................................ ................................ ................................ 42 Discussion ................................ ................................ .......................... 46 Strengths and limitations ................................ ..................... 52 Conclusion ................................ ................................ ........... 54 Chapter Three: Obstetrics & Gynecology Manuscript ................................ ... 60 Manus cript Two Title ................................ ................................ ........ 60 Introduction ................................ ................................ ........................ 61 Materials and Methods ................................ ................................ ....... 63 Data source ................................ ................................ ........... 63 Exposure definition ................................ ............................... 64 Outcome measures ................................ ................................ 65 Statistical analyse s ................................ ................................ 68 Results ................................ ................................ ................................ 70 Discussion ................................ ................................ .......................... 76 Strengths and limitations ................................ ..................... 79 Conclusion ................................ ................................ ........... 80 Chapter Four: American Journal Obstetrics & Gynecology Manuscript...... ..91 Manuscript Three Titl e ................................ ................................ ...... 91 Introduction ................................ ................................ ........................ 92 Materials and Methods ................................ ................................ ....... 94 Data source ................................ ................................ ........... 94 Exposure definition ................................ ............................... 94 Outcome measures ................................ ................................ 9 6 Statistical anal yses ................................ ................................ 99 Results ................................ ................................ .............................. 101 Discussion ................................ ................................ ........................ 107 Strengths and limitations ................................ ................... 110 Conclusion ................................ ................................ ......... 112 Chapter Five: Discussion ................................ ................................ .............. 124 Theoretical Model ................................ ................................ ............ 125 Overview of Significant Findings ................................ .................... 126 Research question one ................................ ........................ 127 Research question two ................................ ........................ 128 Research question three ................................ ...................... 128 Research question four ................................ ........................ 129

PAGE 6

! iii Research question five ................................ ........................ 131 Infant morbidity ................................ ................... 131 Maternal morbidity ................................ .............. 131 Infant healthcare utilization ................................ 132 M aternal healthcare utilization ............................ 132 Research question six ................................ .......................... 133 Infan t health outcomes ................................ ......... 133 Maternal health outcomes ................................ .... 134 Consideration of Findings in Light of Existing Research ................ 135 Validity of birth certifica te, discharge, and linked data ...... 1 35 Infant morbid ity and healthcare utilization ......................... 137 Maternal m orbidity and rehospitalization ........................... 138 Public Health Implications and Policy Recommendations .............. 140 Individual level ................................ ................................ ... 140 Interpersonal level ................................ ............................... 141 Institutional level ................................ ................................ 144 Societal level ................................ ................................ ....... 145 Strengths and Limitations ................................ ................................ 146 Recom mendat ions for Future Research ................................ ........... 150 Conclusion ................................ ................................ ....................... 151 Literature Cited ................................ ................................ .............................. 153 Appendices ................................ ................................ ................................ ..... 173 Appendix A. Abstraction Instrument for Florida Late Preterm and Cesarean Delivery Investig ation ................................ ............... 174 Appendix B. Power Calculation Results ................................ ........ 192 Appendix C. Manuscript One Supplementary Tables .................... 195 Appendix D. Manuscript Two Supplementary Tables .................... 2 03 Appendix E. Manuscript Three Supplementary Tables ................... 2 06 Appendix F. University of South Florida Institutional Review Board Exempt Status Determination Letter ................................ ..... 2 09 About the Author ................................ ................................ ................. End Page

PAGE 7

! iv List of Tables Table 2.1 Characteristics of Study Sample by Data Source ..................... 55 Table 2.2 Accuracy of Data Elements Reported on the Florida ............... 56 Table 2.3 Comparison of Validity Indices in the Linked Birth ................ 57 Table 3.1 ICD9 Codes for Selected Infant Morbidities ............................ 82 Table 3.2 Selected Socio Demographic Characteristics of Low .............. 83 Table 3.3 Reported Infant Morbidity at Initial Hospitalization ................ 84 Table 3.4 Association Between Primary Cesarean Delivery and ............. 85 Table 3.5 Association Between Primary Cesarean Delivery and ............. 86 Table 4.1 ICD9 Codes for Selected Maternal Morbidities ..................... 114 Table 4.2 Demographic Characteristics of Low Classified Risk ............ 115 Table 4.3 Distribution of Maternal Morbidities at Birth ........................ 116 Table 4.4 Reported Maternal Morbidities at Initial ................................ 117 Table 4.5 Risk of Maternal Morbidity by Route of Delivery ................. 118 Table 4.6 Risk of Maternal Morbidity by Route of Delivery ................. 119

PAGE 8

! v List of Figures Figure 1.1 The Social Ecological Model ................................ .................. 20 Figure 1.2 The Social Ecological Model of Low Documented Risk ........ 26 Figure 1.3 Overview of Study Design ................................ ...................... 27 Figure 2.1 Description of Sample and Linkage Results ........................... 58 F igure 2.2 Low Risk Primary C/S Algorithm ................................ .......... 59 Figure 3.1 Markov Hierarchical Model for Infant Morbidity ................... 87 Figure 3.2 Markov Pooled Infant Morbidity Curve ................................ .. 88 Figure 3.3 Description of Study Sample After Application of Low ......... 8 9 Figure 3.4 Difference in Reported Numbers of Morbidity at Birth .......... 90 Figure 4.1 Markov Hierarchical Model for Maternal Morbidity ............ 120 Figure 4.2 Markov Pooled Maternal Morbidity Curve ........................... 121 Figure 4.3 Description of Study Sa mple After Application of Low ....... 122 Figure 4.4 Kaplan Meier Estimate of Time Till First ............................. 123 ! ! ! ! ! !

PAGE 9

! vi ! ! ! Low Documented Risk Cesarean Sections and Late Preterm Birth: The Florida Experience Heather Breeze Clayton ABSTRACT There are increasing concerns about the excessive use of cesarean delivery in the United States, as cesarean deliveries have been associated with adverse maternal and infant health outcomes. Currently, the cesarean section (C/S) rate for Florida is the s econd highest in the nation. Furthermore, preliminary reports from the Florida Department of Health (FDOH) have implicated the increasing rate of cesarean delivery to an increase in the rate of late preterm births (PTB) in Florida (births at 34 to 36 wee ks gestational age). Information on the impact of late PTB associated with cesarean delivery on the rate of maternal and infant morbidity in Florida as well as corresponding utilization of health care services is scarce. Information on the validity of data sources used to investigate infant and maternal health outcomes in Florida is also scarce. Therefore, the objectives of this research project were: (1) to determine the validity of data sources used to investigate low documented risk C/S and late PT B, and (2) to assess the impact of low documented risk C/S on maternal and infant morbidity and subsequent healthcare utilization. To determine the accuracy of data elements reported on the Florida birth certificate and hospital discharge data, sensitivit y, specificity, positive predictive value, negative predictive value, kappa statistics and likelihood ratios were calculated. To assess differences in morbidity by route of delivery, generalized estimating equations and

PAGE 10

! vii survival analyses were employed. Markov Chain Monte Carlo methods were used to determine appropriate morbidities for inclusion in all analyses. Differences in accuracy of data by data source was observed, with linked birth certificate and hospital discharge data demonstrating improved a ccuracy compared to birth certificate and discharge data alone. Further, significant differences in the rate of maternal and infant morbidity by route of delivery were observed, with cesarean delivery increasing the risk of adverse health outcomes, and i ntensive use of healthcare services.

PAGE 11

CHAPTER ONE PURPOSE OF STUDY !The purpose of this proposed research is to investigate the impact of low indicated medical risk primary cesarean section (C/S) on maternal and infant morbidity at late preterm delivery. The primary objectives of this proposed project are: (1) to determine the validity of data sources (e.g. Florida birth certicate data, hospital discharge data, linked birth certicate and hospital discharge data le) used to investigate low indicated risk C/S and late PTB, and (2) to assess the impact of low indicated risk C/S on maternal and late PTB infant morbidity and subsequent healthcare utilization. The long-term goal is to provide evidence-based data that can be used by healthcare providers, health ofcials, health organizations, public health policy makers, and health insurers to decrease unnecessary C/S procedures and to reduce maternal and infant morbidity. DISSERTATION FORMAT !This dissertation is in a manuscript-style format. This means that instead of the traditional format of an introduction, literature review, methods, results and discussion chapters, much of the methods and results take the form of three distinct manuscripts, with a total of 5 chapters: introduction, manuscript one, manuscript two, manuscript three, and synthesis of results with discussion and conclusions. The introduction section includes: purpose of research, review of the literature, preliminary research, theoretical model, specic aims and research questions, description 1

PAGE 12

of data sources and a plan for the dissemination of research. Each of the three manuscripts in this dissertation has the introduction, methodology, results and conclusion sections completed. While this dissertation is described in terms of three distinct manuscripts, the research forms a cohesive study, with each phase (manuscript) informing subsequent phases. In the fth (nal) section of the dissertation, results of all three phases are synthesized and discussed. BACKGROUND !For well over a century, pregnancy and birth in the United States has been viewed more as a disease state than a natural process (Kohler-Reissman, 1992). Birth was once considered the purview of women, with midwives providing care for women during pregnancy and delivery. However, starting as early as the 1800's, the medical profession rose to prominence in the United States. Midwives were gradually pushed out by physicians as a new specialty formed obstetrics. With the development of obstetrics came several changes in the culture of birth in the United States. Birth went from being viewed as a natural life process, centered on the family and the home, to an event fraught with the potential for danger. Thus births increasingly took place in hospitals, purportedly a more controlled, sterile environment. Over time, as the medical management of pregnancy and delivery increased, pregnancy became viewed as a medical condition requiring specialized supervision and/or intervention. !While obstetric practice changed the context of birth in the United States, several technologies were developed, or more commonly utilized to improve health outcomes for women and infants antibiotics, anesthesia, blood transfusion, forceps, oxytocin 2

PAGE 13

(induction of labor), episiotomy and cesarean delivery. These advances, in addition to public health efforts (e.g. improvement in sanitation, vaccinations for childhood diseases, improved nutrition, access to contraception, prenatal care) have improved populationlevel health outcomes such as maternal and infant morbidity as well as overall life expectancy (IOM, 2002). However, increased use of obstetric technologies such as cesarean deliveries (beyond those which are medically indicated), have been associated with an increase in the rates of several adverse events (Allen, O'Connell, Liston, & Baskett, 2003; Belizan, et al., 2007; Declercq, Cabral, Evans, Kotelchuck, Simon, Weiss, Heffner, 2007; Gilliam, 2006; Malloy, 2009; Miesnik & Reale, 2007). TRENDS IN CESAREAN DELIVERIES IN THE UNITED STATES !A cesarean section is dened as the delivery of an infant through an incision in the abdomen. In the case of an obstetric emergency, a C/S can be a life saving procedure. However, there are increasing concerns about the excessive use of C/S in the United States and elsewhere in the world. In 1965, the overall or crude C/S rate for the United States was just under 5% (Hamilton, et al., 2007). Currently, the C/S rate for the United States is 31.8%, and in the state of Florida, the C/S rate is 37.2% (Hamilton, Martin, Ventura, 2009). Approximately one out of every three infants in the United States is delivered via CS. According to recent literature, C/S rates between 5-10% have the best outcomes for mothers and infants, while C/S rates above 15% appear to result in more adverse outcomes (Althabe & Belizan, 2006; Althabe, Sosa, Belin, Gibbons, Jacquerioz & Bergel, 2006; Belizan, Althabe, & Cafferata, 2007). Currently, the Healthy People 3

PAGE 14

2010 goals focus on reducing C/S among low-risk (full term, singleton, vertex presentation) women to 15% (CDC, 2009). !There are several classications of C/S rates. The overall or total C/S rate includes all C/S deliveries. The repeat C/S rate refers to the rate of C/S deliveries among women who have experienced a prior C/S. The vaginal birth following C/S or VBAC rate refers to the rate of vaginal births among women with a previous C/S. The primary C/S rate is the rate of C/S among women without a history of C/S. The planned C/S rate describes C/S without labor (Declercq et al, 2007). The rate of C/S with labor is dened as an attempted vaginal delivery that resulted in a C/S. A low documented risk (low risk) C/S is the rate of C/S deliveries among women without documented medical indications for C/S delivery. For the purposes of this dissertation, the low documented risk C/S will be classied by the absence of 15 medical risk factors associated with C/S, as well as a non-intensive level of prenatal care usage (as determined by GINDEX) (Alexander & Cornely, 1987; Alexander & Kotelchuck, 1996). ADVERSE OUTCOMES OF CESAREAN DELIVERIES !The increase in the C/S rate is a major public health concern because a C/S delivery carries a higher risk for adverse maternal and infant health outcomes than a vaginal delivery (Allen, O'Connell, Liston, & Baskett, 2003; Belizan, et al., 2007; Declercq, Cabral, Evans, Kotelchuck, Simon, Weiss, Heffner, 2007; Gilliam, 2006; Miesnik & Reale, 2007). The medical consequences of C/S can be described as either maternal or infant related. Maternal consequences of C/S include: wound infection, hysterectomy, ureteral tract and vesical injury, abdominal pain, cardiac arrest, puerperal 4

PAGE 15

febrile morbidity, endometritis, venous thromboembolism, cystitis, rehospitalization, potential loss of reproductive ability, or maternal mortality (Allen, et al., 2003; Belizan, et al., 2007; Declercq, et al., 2007; Liu, et al., 2007; Miesnik et al, 2007; Wax, 2006). C/S consequences that impact infants include: injuries during delivery (possibly resulting in death), neonatal respiratory morbidity, gastrointestinal problems, skin conditions, increased length of hospital stay, as well as difculty with attachment and bonding (Alexander, Leveno, Landon, Thom, Spong, Varner, et al, 2006; Belizan, et al., 2007; Belizan, Cafferata, Althabe, & Buekens, 2006; Hansen, Wisborg, Uldbjerg, & Henriksen, 2007, 2008; Leung, Ho, Tin, Schooling, & Lam, 2007; Miesnik et al, 2007; van den Berg, van Elburg, van Geijn, & Fetter, 2001; Villar, et al., 2007). !Some of the most substantial adverse outcomes of C/S are manifest in subsequent pregnancies (Gilliam, 2006). These adverse outcomes include abnormalities of placentation, uterine scar dehiscence, increased risk of uterine rupture, and unexplained fetal death (Gilliam, 2006; Spong, et al., 2007). One outcome of C/S that has signicant public health importance is repeated CS. Cragin in 1916 coined an unfortunate, yet increasingly accurate phrase "Once a cesarean always a cesarean" (Cragin, 1916) Note that Cragin referred to classical cesarean (uterus is entered through a vertical fundal incision), which is now rarely performed. Despite a push by the CDC to promote vaginal birth following cesarean (VBAC), the rate of VBAC's has declined almost to extinction (CDC, 2005). As a result, the repeat C/S rate has increased, and is currently reported to be approximately 91% (Menacker, Declercq, & Macdorman, 2006). If a woman intends to have a large family, an initial C/S delivery can have a devastating impact on her long term reproductive goals as reproductive consequences of multiple 5

PAGE 16

cesarean sections can be quite substantial (e.g. infertility, high risk pregnancy, uterine rupture) (Pare, Quinones, & Macones, 2006; Silver, et al., 2006). RISK FACTORS FOR CESAREAN DELIVERY !Risk factors for C/S identied in the available literature can be describe as either clinical or non-clinical. Clinical factors can also be referred to as medical indications for C/S. Clinical factors are often the most important in a physician's decision making process with regard to cesarean sections, however they can vary by somewhat broad categories maternal morbidity, pregnancy complications (maternal or fetal), and complications with labor management (Korst, 2004). There are numerous clinical risk factors for cesarean delivery, and a majority are reported on the birth certicate: eclampsia, renal disease, uterine bleeding, fetal distress, diabetes, obesity, hypertension, incompetent cervix, placenta abruption, placenta previa, cord prolapse, prolonged labor, dystocia, lung disease, heart problems, malpresentation, multiple gestation, sexually transmitted infections, and anemia (Bailit, Dooley, & Peaceman, 1999; Ennen, et al., 2009; Gregory, Korst, Gornbein, & Platt, 2002; Joseph, Young, Dodds, O'Connell, Allen, Chandra, Allen, 2003; Kahn, 2009; Sheiner, 2004; Yasmeen, Romano, Schembri, Keyzer, & Gilbert, 2006). !Non-clinical risk factors include risk factors for C/S that are not related to medical necessity. Non-clinical risk factors that result in increased C/S discussed in the literature include maternal race/ethnicity, maternal age, health insurance/payer status, measures of socio-economic status (SES), type of hospital, scheduling concerns, reduced use of forceps, fatigue during pregnancy, lawsuit activity and defensive medicine practices 6

PAGE 17

(Aron, 2000; Brown, 2007; Grant, 2005; Gregory, Korst, Platt, 2001; Hsu, 2006; Joseph et al, 2003; Ko, 2003). Furthermore, there has been increasing attention to the change in practice styles of younger (newer) generations of obstetricians. Recent and incoming cohorts of residents and obstetricians conform their careers to their lifestyles and thus choose more dened shifts, fewer emergency and after-hour calls (ACOG, 2008). Furthermore, work hours for residents have been mandated capped at 80 hours per week. The possibility of maternal request C/S has also received a great deal of attention in the available literature ("ACOG Committee Opinion No. 394, December 2007. Cesarean delivery on maternal request," 2007; Coleman, Lawrence, & Schulkin, 2009; Lee & D'Alton, 2008a, 2008b). While C/S by maternal request is not a risk factor, it is a practice entity that needs to be recognized. Unfortunately, it is difcult to accurately measure as maternal request is not assessed by available public health data systems (e.g. birth certicate, hospital discharge data), though could be (NIH, 2006). PRETERM BIRTH !Overall, there has been a shift towards earlier delivery of infants regardless of gestational age (dened as the rst date of the last menstrual period to present date, measured in weeks) (Davidoff, 2006; IOM, 2007). Davidoff et al (2006) observed that spontaneous singleton live births at > 40 weeks decreased, while births between 34 to 39 weeks increased (p<0.001). Importantly, births with medical intervention followed a distribution similar to that of spontaneous births. Cesarean delivery when analyzed separately from induction, also demonstrated a trend towards earlier gestational age at birth (Davidoff et al, 2006). This shifting distribution of gestational age may partially 7

PAGE 18

be due to an increase in the use of obstetric interventions such as induction of labor and C/S (Ananth, 2005; Davidoff, 2006; IOM, 2007; MacDorman, 2002). !Currently, a term pregnancy is dened as "one that has completed 37 weeks of gestation and that delivers after the rst day of the 38th week of pregnancy" pg. 793 (Fuchs, Wapner, 2006). Similarly, premature birth (PTB) is dened by the American College of Obstetrics and Gynecology (ACOG) as the birth of an infant before 37 completed weeks of gestation (Fuchs et al, 2006). Engle (2006) noted that denitions for infants born preterm, term, and post-term have been well delineated, however the nomenclature for subgroups within these categories has continued to evolve. Previously, infants born from 34 completed weeks gestation through 36 6/7 completed weeks gestation were referred to as "near term". Recently, the classication for this subgroup of births has been changed to "late preterm" to reect the fact that these infants have a higher risk of morbidity than their term counterparts (Engle, 2006). !While improvements in the treatment of PTB infants in Neonatal Intensive Care Units (NICU), have dramatically increased survival there remains several signicant complications of PTB. The more immediate consequences for infants born preterm include: respiratory, gastrointestinal, central nervous system as well as, hearing and vision problems (IOM, 2007). In the long term, PTB infants may be at risk of cerebral palsy, mental retardation, learning difculties, behavior and social concerns, visual and hearing impairments, and overall poor health and growth (IOM, 2007; Morse et al, 2009; Petrini et al, 2009). While the risk of complications for late-preterm infants is less than those experienced by very preterm or moderately preterm infants, late-preterm infants have a higher risk of adverse outcomes than term infants (Engle, Tomashek, Wallman, 8

PAGE 19

and the Committee on the Fetus and Newborn, 2008; Wang, Dorer, Fleming, Catlin, 2004). Compared to term infants, late-preterm infants are more likely to have temperature instability, hypoglycemia, respiratory distress, jaundice, longer length of stay (LOS) at birth hospitalization and increased rehospitalization rates (Burgos, Schmitt, Stevenson, & Phibbs, 2008; McLaurin, Hall, Jackson, Owens, & Mahadevia, 2009; Wang et al, 2004). In addition to increased rates of morbidity, the mortality rate for late PTBs is higher than that for term infants. For example, Tomashek et al (2007) assessed the differences in mortality between late-preterm infants and term infants, and observed that mortality among late preterm infants was three times higher than that of term infants (Tomashek, Shapiro-Mendoza, Davidoff, Petrini, 2007). !Between 1990 and 2006, the overall rate of PTB in the United States increased from 10.6% to 12.8% a 21% increase in the rate of prematurity (Hamilton, et al, 2007). Prematurity is a concern of public health importance due to the substantial burden of mortality and morbidity among premature infants. As prematurity rates are directly related to infant mortality, preventing prematurity is an important goal when attempting to lower overall infant mortality rates (Mathews, 2007). The largest increase in prematurity occurred among late preterm infants, accounting for approximately 75% of all PTBs (Davidoff, 2006; Hamilton, et al, 2007). As a large proportion of premature births (~ 75%), late PTBs have a signicant impact on the healthcare system and the overall population health indices (Damus, 2008). Therefore, late PTBs are an important focus for public health prevention and research efforts. 9

PAGE 20

RISK FACTORS FOR PRETERM BIRTH !The risk factors for PTB are complex and multifactoral. There are several behavioral risk factors for PTB including substance abuse (e.g. tobacco, alcohol, illicit drugs), nutritional behavior (e.g. prepregnancy weight, gestational weight gain, dietary intake) and inadequate physical activity (Cnattingius, Hultman, Dahl, & Sparen, 1999; Goldenberg, Iams, Mercer, Meis, Moawad, Cooper et al, 1998; Hellerstedt, Himes, Story, Alton, & Edwards, 1997; Holzman & Paneth, 1994; IOM, 2007; Kesmodel, Olsen, & Secher, 2000; Lang, Lieberman, & Cohen, 1996; Lundsberg, Bracken, & Saftlas, 1997; Savitz, Dole, Terry, Zhou, & Thorp, 2001). Psychosocial factors such as maternal stress and anxiety have been suggested as risk factors for PTB, but with great variation in denitions and measurement of stress, the evidence has been somewhat mixed (IOM, 2007). !There are several maternal socio-demographic factors that have been associated with PTB. These factors include young maternal age, marital status (unmarried mothers have increased risk for PTB), race/ethnicity, and socioeconomic variables (Amini, Catalano, Dierker, & Mann, 1996; Branum & Schoendorf, 2005; Derrick, Luo, Bregman, Jilling, Ji, Fischer et al, 2004; Luo, Kierans, Wilkins, Liston, Mohamed & Kramer, 2004; Raatikainen, Heiskanen, & Heinonen, 2005; Zeitlin, Saurel-Cubizolles, & Ancel, 2002). Risk factors for preterm birth at the community level include: poverty, the social environment (e.g. crime, lack of cohesiveness), the physical environment (e.g. housing quality, public space, toxins) and the service environment (e.g. lack of services, goods, healthcare facilities) (IOM, 2007). Maternal medical risk factors for PTB are very similar to the risk factors for C/S delivery, including hypertension, systematic lupus 10

PAGE 21

erythematosus, hyperthyroidism, pre-pregnancy diabetes mellitus, maternal cardiac disease, restrictive lung disease, asthma, renal disorders, gestational diabetes, other hypertensive disorders of pregnancy, and multifetal gestation (IOM, 2007). ECONOMIC CONSEQUENCES OF C/S AND PRETERM BIRTH !A C/S can be considered a revenue maximizing procedure (Xirasagar & Lin, 2007). The charge for a C/S is often higher than a vaginal delivery, and with an average procedure length of less than an hour, a C/S may be more time efcient for the provider and healthcare institution (Doherty, et al., 2008). Furthermore, the maternal length of stay (LOS) is longer for C/S delivery than vaginal delivery, which translates into a further increase in the cost of C/S delivery (Declercq, et al 2007). This study compared planned primary C/S with planned vaginal births and reported that the average cost of a planned primary C/S was 76% higher than the average cost of a planned vaginal birth ($4,372 versus $2,487). Furthermore, the LOS for planned primary C/S was 77% longer (4.3 days versus 2.4 days). Importantly, the author also observed that there were more maternal rehospitalizations for planned primary CS. This was due to complications such as obstetrical surgical wounds, puerperal infection, genitourinary tract infections, inammatory diseases of the uterus, and delayed and secondary postpartum hemorrhage. An increase in postpartum medical care for elective C/S has also been reported by Liu et al (2008), although the authors conclude that this may not be clinically signicant as the difference in costs between C/S and vaginal birth postpartum care was $2.20 (Liu, Chen, & Lin, 2008). Furthermore, this study focused on postpartum outpatient visits, not rehospitalizations. They observed that women who requested C/S deliveries had a 42% 11

PAGE 22

greater likelihood of attending postpartum outpatient visits compared to vaginal births. It should be noted that this study was conducted in Taiwan, while Declercq used birth and discharge records from Massachusetts. There is great variation in costs and related health outcomes by healthcare systems. The healthcare system in the United States is the most expensive in the world but unfortunately our health outcomes do not correlate with the amount of funds expended in care (IOM, 2002). Given this, it is difcult to compare the costs associated with C/S procedures in the United States with those in countries with substantially different healthcare systems, cost structures and models of care. !There has been a substantial amount of research on the economic consequences of premature birth. Studies have focused on immediate costs, costs in the rst year of life and time points extending well into childhood (Petrou, Mehta, Hockley, Cook-Mozaffari, Henderson, Goldacre, 2003a; Petrou, 2005). Given the higher rates of mortality and morbidity among preterm infants, it is not surprising that PTBs have higher rates of utilization of healthcare services and increased costs compared to infants delivered at term (Kirkby, Greenspan, Kornhouser, Schneiderman, 2007; Petrou, 2003b; Russell, Green, Steiner, Meikle, Howse, Poschaman, Dias, Potetz, Davidoff, Damus, 2007; Underwood, Danielson & Gilbert, 2007). In an analysis of costs associated with preterm delivery in California from 1992 to 2000, it was reported that 15% of preterm infants required at least one rehospitalization during the rst year of life (Underwood, 2007). These rehospitalizations resulted in an average annual cost to the state of California of $41 million. Russell et al (2007) used the Healthcare Cost and Utilization (HCUP) data to examine the overall costs associated with prematurity in the United States in 2001, and 12

PAGE 23

reported that the 4.6 million infant hospitalizations in 2001 that included a diagnosis of PTB/low birth weight resulted in a total cost of $5.8 billion. The IOM report on preterm birth estimates the annual societal impact of preterm birth to be upwards of $26.2 billion (2007). Research on the long term consequences of prematurity have demonstrated that costs of prematurity extend well into childhood (Petrou, 2005; Petrou, et al, 2003). Petrou et al (2003) found that premature birth was the strongest predictor of excessive health costs in the rst 5 years of life. Furthermore, when following this same population for an additional ve years (total follow-up of 10 years), the cost differences between children born premature or term persisted (Petrou, 2005). Clearly, PTB is associated with substantial immediate and long term health consequences and corresponding increased costs for healthcare services. !The healthcare costs for late-preterm infants are not as substantial as infants born before 34 completed weeks of gestation, but they still have a considerable impact on the healthcare system given the large proportion of premature births that are late-preterm (~75%) (McIntire & Leveno, 2008). Researchers have documented that late PTBs have substantially higher morbidity rates than term infants and with that comes higher costs due to increased utilization of healthcare services. McIntire et al, (2008) reported that late-preterm infants required more intensive care and longer LOS, which directly translated into higher hospital charges for late-preterm infants. The mean hospital bill for late-preterm infants was approximately 2.5 times higher than the 39 week referent group ($3,098 versus $1,258). Researchers have consistently reported higher costs associated with late PTBs compared to term births (McIntire et al, 2008; ShapiroMendoza, Tomashek, Kotelchuck, Bareld, Weiss, & Evans, 2006; Tomashek, Shapiro13

PAGE 24

Mendoza, Weiss, Kotelchuck, Barield, & Evans, 2006; Wang, et al, 2004). Wang et al (2004) demonstrated that the average cost for late-preterm infants was 2.93 times higher than term infants, with an average cost difference of $2,630. On the subject of the economic consequences of prematurity, the research is rather clear premature births regardless of gestational age (GA) at birth are costly, not only in terms of health care dollars, but also increased morbidity and mortality (Clements, Bareld, Femi Ayadi, & Wilber, 2007; Kirkby et al, 2007; Russell et al, 2007; Underwood et al, 2007). RESEARCH FOCUSING ON BOTH LATE PRETERM BIRTH AND C/S !In response to the relationship between the increase in preterm rates and obstetric interventions prior to 39 weeks gestational age, researchers have begun to focus on late preterm C/S delivery and neonatal outcomes. Researchers have demonstrated an association between late preterm delivery and cesarean delivery (Fuchs & Wapner, 2008), however information on the impact of the observed association between late preterm birth and cesarean delivery has not been well explored. Only a few studies describe neonatal morbidity following C/S among late preterm infants. In a recent study by Melamed et al, cesarean delivery was demonstrated to be an independent risk factor for neonatal respiratory morbidity among a population of low risk spontaneous singleton late preterm deliveries (Melamed, Klinger, Tenenaum-Gavish, Herscovici, Linder, Hod et al, 2009). De Luca et al (2009) also examined the outcomes of late preterm infants and identied several adverse outcomes among infants delivered via elective C/S, namely higher rates of mortality, admission to specialty care, and respiratory morbidity (De Luca, Boulvain, Irion, Berner, Pster, et al, 2009). Yoder et al (2008) also described higher rates of 14

PAGE 25

respiratory morbidity following C/S at late preterm, although this observation did not reach statistical signicance, likely due to sample size constraints (Yoder, Gordon, and Barth, 2008). Malloy (2009) examined the impact of cesarean delivery for late preterm infants and observed an increased risk of mortality among infants delivered via C/S. At present, much of the research focused on late preterm infants have used term infants as a control group. This strategy is appropriate for studies focused on describing increased morbidity or rehospitalization of infants delivered late preterm compared to term, however it may not be ideal for determining the contribution of C/S at late preterm to infant morbidity, over and above that already experienced by late preterm infants. Using as a control group, late preterm infants delivered vaginally may aid in the understanding of the independent contribution of C/S delivery to adverse outcomes among a population of late preterm infants. LIMITATIONS OF THE LITERATURE !The literature has focused predominantly on (1) risk factors for C/S and subsequent adverse outcomes, (2) risk factors and adverse outcomes for PTB, (3) the economic impact of C/S delivery, and (4) the economic impact of preterm (and latepreterm) births. Researchers have only recently begun to explore the potential association between C/S delivery and late PTBs, and have not yet considered the contribution of C/S delivery to late PTB and subsequent health care utilization for both infants and their mothers. 15

PAGE 26

PRELIMINARY RESEARCH !MCH Epidemiologists at the Florida Department of Health (FDOH) have conducted several analyses to explore the connection between PTB and intrapartum intervention in Florida (Goodman, Sappeneld & Thompson, 2007). Using the Florida Final Birth Certicate File (and restricting analyses to singletons), they have examined trends in late PTBs and demonstrated that (1) the proportion of births that are preterm has increased by 18% from 1995 to 2006, (2) the proportion of births that are late preterm has also increased, and account for the majority of the increase in the overall proportion of preterm deliveries, and (3) this proportion is higher in Florida than for the nation as a whole. Further analyses have demonstrated that the observed increase in PTB in Florida is not attributable to increases (or shifts) in multiple gestations, maternal age, maternal race, maternal ethnicity, maternal education, parity or marital status. The risk of PTB for singletons was also examined by delivery route. Using vaginal delivery as a referent category, and adjusting for age, race, ethnicity, marital status and parity, the relative risk of PTB for primary cesareans was 1.47 (95% Condence Interval (CI) (1.45,1.48)). The observations provided by the FDOH have extreme importance for public health efforts in Florida as late preterm infants have higher risk for morbidity and mortality than term infants, which highlights a viable target for lowering the rate of infant mortality and morbidity in Florida. THEORETICAL FRAMEWORK !The available literature on health concerns such as C/S and preterm delivery have identied factors that operate on several levels the contribution of individual factors, 16

PAGE 27

interpersonal factors, community or institutional factors and societal factors. For this reason, it is important to utilize a theoretical model that can incorporate the multi-level etiology of low documented risk C/S singleton late-preterm live birth. One of the most widely used frameworks in public health, the Social Ecological Model (SEM), is frequently used to describe the multi-level determinants of health problems (Rimer & Glanz, 2005). !The SEM is a model, not a theory. This is an important distinction to make as the terms "theory" and "model" are often used interchangeably. The primary purpose of theory is to provide a systematic way to explain phenomena (Rimer et al, 2005; White, 2002). Theories therefore, are a systematic collection of concepts, denitions, propositions and relationships that work to generate new knowledge or change outcomes (Rimer, 2005; Torres, 1986). Scientic theories, in particular, contain empirically testable propositions (White & Klein, 2002). Conversely, models are used to describe phenomena, and are often a component of theory (Torres, 1986). According to Torres (1986), models provide a framework for meaning and understanding of theoretical concepts and their interrelations, primarily through visual (schematic) representations. Furthermore, models may use several theories to understand a particular situation or problem in a specied context or setting (Rimer et al, 2005). !The SEM is frequently used as a framework, or structure, that aids in the organization and understanding of the multiple levels of inuence for health outcomes. At each level of inuence, and between levels of inuence, several theoretical perspectives can be applied to either explain or change health outcomes. For example, at the individual level, the Transtheoretical Model may be used to prevent heart disease 17

PAGE 28

among obese individuals by inuencing health behavior change (Prochaska, 2008; Prochaska & DiClemente, 1983; Sarkin, Johnson, Prochaska, & Prochaska, 2001). There are also theories that operate at the relationship level (e.g. Social Cognitive Theory) and the institutional/community level (e.g. Community-Based Participatory Research, Diffusion of Innovations Theory, Communication Theory) that work to explain or prevent health conditions (Israel, Paerker, Rowe, Salvatore, Minkler, Lopez, et al, 2005; Israel, Schulz, Parker, & Becker, 1998; Rimer et al, 2005). Finally, at the larger societal level, there are numerous theoretical perspectives to explain social norms, economic conditions and policy (e.g. Theory of Gender and Power) (Torres, 1986; Wingood & DiClemente, 2000). !While the term "ecology" was rst coined in 1873 by Ernst Haeckel (a German biologist), the historical roots and intellectual traditions of the SEM can be traced back to the work of Thomas Malthus (1766-1834) and Charles Darwin (1809-1882), who both conducted considerable study of humans within their environments (White et al, 2002). Thomas Malthus used ecology to describe the complex interplay between population growth, availability of food, and preventive checks to control population (e.g. moral restraint, wars, famine). In the publication The Descent of Man, and Selection in Relation to Sex, Charles Darwin described the process of evolution, natural selection and elimination. Charles Darwin argued that we "do our utmost to check the process of elimination; we build asylums for the imbecile, the maimed, and the sick; we institute poor laws; and our medical men exert their utmost skill to save the life of everyone to the last moment" (Darwin, 1871)( pp.168). 18

PAGE 29

This quote by Darwin illustrates an important perspective individuals, and in fact society, have the ability to interfere with, and impact health outcomes. Since the times of Thomas Malthus and Charles Darwin, the eld of ecology has undergone substantial development. In 1988 it was introduced for use in health promotion by McLeroy, Bibeau, Steckler & Glanz (1988). Since that time, it has been widely adopted in public health as a method by which to organize and understand the multi-level inuences of health outcomes. THE SOCIAL ECOLOGICAL MODEL !According to the SEM, health is inuenced on several levels individual, interpersonal, community/institutional, and societal (Rimer, 2005). The individual level (also referred to as the intrapersonal level) of inuence involves individual biologic characteristics and individual behaviors. At the individual level, individual characteristics impact behavior (personality, knowledge, attitudes, health beliefs) and subsequent risk for adverse health outcomes (Rimer et al, 2005). At the interpersonal level, relationships with primary groups such as family, friends, and peers impact social identity, support, and behavior, which can either ameliorate or exacerbate health risks. The community/institutional level involves norms, social networks, and standards that exist among and between individuals, groups, institutions and other organizations. The societal level includes policy, economics, media, and social inequities (e.g. sex, race/ ethnicity). !The most commonly portrayed levels of the SEM are the individual, interpersonal, community, institutional, and societal although it should be noted that 19

PAGE 30

many other intermediate levels could be applied to this diagram, and some levels are occasionally combined. For the purpose of this research, only the levels outlined in Figure 1.1 will be applied to the discussion of the SEM of low documented risk singleton late PTB. Figure 1.1 Social Ecological Model !"#$%$#&'( !")*+,*+-."'( /.00&"$)123 !"-4)&4."'( 5.6$*)'(3 APPLICATION OF SEM TO RESEARCH IN MATERNAL AND CHILD HEALTH (MCH) !Given the prominence (and importance) of the SEM in public health, it is not surprising that the SEM has also been widely utilized in MCH research and programs. For example, the SEM has been used to describe the complex etiology of a wide range of MCH health concerns, such as folic acid, adolescent pregnancy, low birth weight racial/ ethnic disparities, and childhood obesity (Jaffee & Perloff, 2003; Klein, Lytle, & Chen, 2008; Nitz, 1999; Quinn, Thompson, & Ott, 2005). Recently, researchers have utilized the SEM to investigate ecological factors that impact delivery mode and birth outcomes such as PTB. This is not surprising as the etiology of outcomes such as PTB are 20

PAGE 31

complex and include a variety of factors such as social, hormonal, environmental, and genetic (DeFranco, Lian, Muglia, & Schootman, 2008). !The use of the SEM is exible, meaning that researchers may utilize theories that address several levels of the SEM, or they may use the SEM to outline the complex etiology of a health issue, but only focus research on one or more levels of the SEM. For example, there has been signicant attention in the literature to investigating the contribution of individual and neighborhood factors for birth outcomes such as PTB, intrauterine growth restriction (IUGR), and low birth weight (LBW) (Ahern, Pickett, Selvin, & Abrams, 2003; DeFranco, et al., 2008; Farley, Mason, Rice, Habel, Scribner & Cohen et al, 2006; Kaufman, Alonso, & Pino, 2008; Masi, Hawkley, Piotrowski, & Pickett, 2007; Messer, Kaufman, Mendola, & Laraia, 2008; Nkansah-Amankra Luchok, Hussey, Watkins & Liu, 2009; O'Campo, Burke, Culhane, Elo, Eyster, Holzman et al, 2008; Urquia, Frank, Glazier, Moineddin, Matheson & Gagnon, 2009). Farley et al (2006) used the SEM to explore neighborhood factors such as tract-level median household income, neighborhood physical deterioration and neighborhood density of retail facilities and found that tract-level median income was associated with birth weight-for-gestational age and gestational age at birth. Masi et al (2007), used individual (demographics, socio-economic status) and census tract characteristics (economic disadvantage, violent crime, racial/ethnic group density) to model pregnancy outcomes, and found that group density was associated with PTB while neighborhood violent crime was associated with SGA. Nkansah-Amankra et al (2009) included in their investigation of neighborhood factors related to low birth weight and PTB, the role of four domains of maternal stress: nancial, emotional, spousal-related and traumatic 21

PAGE 32

stress. The authors reported that the effects of maternal stress for LBW and PTB outcomes vary by neighborhood context (Nkansah-Amankra, 2009). The interaction observed between maternal stress and neighborhood context, is understandable given the interconnectedness of levels of the SEM (reciprocal causation). Other modiers between the association of environmental and other neighborhood factors with PTB that have been explored include: individual smoking status, individual level socio-economic factors (e.g. insurance status, occupation, education), and social status (Ahern et al., 2003; Kaufman, et al., 2008). For example, Ahern, et al (2002) examined the contribution of smoking status and socioeconomic factors to PTB among African American (AA) and White women, and found that both smoking and neighborhood socioeconomic factors were associated with PTB among both AA and White women, however individual level insurance status modied the impact of neighborhood socioeconomic factors on PTB. From the information garnered from this study of PTB, Ahern et al (2002) suggest that behavioral/biological risk factors and socioeconomic factors be examined together in order to improve our current knowledge of the etiology of PTB. !The SEM has also been used to explore the etiology of C/S delivery. Brown (2007) used social ecologic factors to explore the association between defensive medicine and differences in delivery practice patterns. Brown included in the investigation, clinical indications for CS, teaching hospital status, insurance status (Medicaid), race/ ethnicity, maternal age, and the number of lawsuits per OB/GYN, hospital variation and hospital referral region variation (Brown, 2007). Brown (2007) reported that there was an association between defensive medicine practice patterns and C/S delivery, however controlling for hospital variation and hospital referral region variation resulted in more 22

PAGE 33

conservative (presumably less biased), yet still signicant association between defensive medicine and C/S delivery. !Other researchers have examined (and identied) institutional factors related to C/ S delivery. For example, Le Ray et al (2006) investigated institutional factors such as the level of perinatal care in the maternity unit, size of the maternity unit, and hospital status (e.g. teaching, prot status), as well as maternal characteristics (e.g. age, geographic origin) and obstetric practices among hospitals in France. Le Ray et al (2006) observed that high-risk maternity units had higher rates of C/S delivery for low risk nulliparas than maternity units that serve primarily low risk obstetric populations. Size and status of the institution did not impact C/S rates, but maternal characteristics such as advanced maternal age and non-French origin resulted in increased risk for C/S (Le Ray, Carayol, Zeitlin, Breart, & Gofnet, 2006). SEM AS A FRAMEWORK FOR DESCRIBING LOW DOCUMENTED RISK C/S !Undoubtedly, the etiology of low documented risk C/S singleton late PTB is complex. Several factors operating on several levels of the SEM contribute to low documented risk C/S delivery of singleton late PTBs. However, only several of these potential factors (and levels of the SEM) will be included in the SEM of low documented risk primary C/S singleton late PTB, due to limitations in data sources currently available to conduct investigations of C/S delivery and preterm birth. For example, factors such as peer, family, cultural and other inuences on mode of birth, are generally not available in population-based public health datasets. These factors are best addressed through qualitative research methodology. However, the factors investigated also depend largely 23

PAGE 34

on the purpose and objectives of the proposed research. As this dissertation is focused on (1) the accuracy of public health data sources (e.g. Florida birth certicate, hospital discharge data, linked birth certicate hospital discharge data le), and (2) the contribution of low documented risk C/S singleton PTB to maternal and infant morbidity and healthcare utilization, only those factors assessed in this study will be described within the framework of the SEM (Figure 1.2). Individual factors !There are several individual level factors of interest to this research. For example, the accuracy of reporting for maternal morbidity, obstetric history, obstetric complications and gestational age is of prime importance in determining an accurate estimate of the proportion of low documented risk C/S singleton late PTB's (Kahn, 2009). The maternal demographics such as race/ethnicity and insurance status will be used to explore important subgroup differences in maternal and infant morbidity outcomes among singleton late PTBs by mode of delivery. Interpersonal factors !Interpersonal factors have not been included in this dissertation research. This is due to the inability to capture (with available data) interpersonal level factors that contribute to low documented risk C/S or late PTB. The available literature has suggested several interpersonal factors such as peer relationships, family obligations, and patient/provider relationship (Cohen, 2005; Gamble, Creedy, McCourt, Weaver, & Beake, 2007; Mancuso, De Vivo, Fanara, Albiero, Priolo, Giacobbe, et al., 2008; Robson, Carey, Mishra, & Dear, 2008). These factors are best investigated via study designs that include 24

PAGE 35

the use of surveys or qualitative methodology such as focus groups or individual interviews. !Community/institutional factors !There are several community/institutional factors of interest in this research. Hospital factors such as accuracy of reporting to systems including vital records and hospital discharge data may play a role in either (1) the quality of data available for the investigation of maternal and infant morbidity and subsequent healthcare costs, or (2) the decision to perform C/S deliveries in low risk patients. Another important institutional factor is accurate reporting to vital records and administrative data systems. There may be variation in the accuracy of information reported by hospital C/S rate category (high C/S rate versus low C/S rate). This is not too surprising as each hospital is a small healthcare system with its own internal policies, procedures and culture. Researchers have demonstrated that lawsuit activity has lead to defensive medicine and a subsequent increase in C/S deliveries (Brown, 2007). Unfortunately, information on malpractice premiums and claims are not available in public health data sources such as the birth certicate or hospital discharge data, and therefore this factor cannot be included. !Societal Factors !Healthcare policy has the ability to inuence practice behaviors. Policy can be explored on several levels, but for the purpose of this study, only three policy venues will be considered: Florida state government, professional organizations (e.g. American College of Obstetricians and Gynecologists), and private health insurance organizations. Policy will not play an active role in the analyses to be conducted as part of this 25

PAGE 36

dissertation research, however policy considerations will be prominent in the discussion of research implications and prevention efforts. Figure 1.2 The Social Ecological Model of Low Documented Risk C/S RESEARCH PURPOSE !The purpose of this proposed research is (1) to investigate the validity of data sources used to investigate C/S singleton late-preterm live births, and (2) to explore the relationship between low documented risk primary C/S and singleton late PTB maternal and infant health outcomes and healthcare utilization. 26

PAGE 37

STUDY DESIGN !While the study design will be described as two separate and distinct components, it should be noted that this is a cohesive study of low documented risk primary C/S and singleton late PTB. This study utilizes a sequential, equivalent study design (Tashakkori, 1998) (Figure 1.3). The two components are separate, but equal in contribution to the research focus, and are undertaken in sequential order. Each study component informs the subsequent study component. Figure 1.3. Overview of Study Design PHASE ONE: QUANTITATIVE Validity of Florida Birth Certicate and Hospital Discharge Data Compared to Data Abstracted from Maternal Medical Hospital Charts Outcomes of Interest for Both the Birth Certicate and Hospital Discharge Data: 1. Sensitivity 2. Specicity 3. Positive Predictive Value 4. Negative Predictive Value PHASE TWO: QUANTITATIVE Analysis of the Contribution of Low Documented Risk Primary C/S Delivery to Singleton Late-Preterm Maternal and Infant Morbidity Outcomes of Interest 1. Infant Morbidity 2. Maternal Morbidity 3. Length of Stay (LOS) 4. Time Till Rehospitalization 5. Number of Rehospitalizations RESULTS 1. Accuracy of key public health data sources in Florida 2. Maternal and infant morbidity and healthcare utilization (immediate and within the rst year post partum). Implications of Research (by level of SEM) Public Health Policy Clinical Guidelines Provider and Patient Education 27

PAGE 38

The rst compon ent of the study is the validation of the Florida birth certicate and hospital discharge data. The results of the validation efforts will inform subsequent analyses of maternal and infant morbidity by providing information on the accuracy of key variables (e.g. gestational age, obstetric complications, maternal morbidity). In the second component of the study, maternal and infant morbidity and healthcare utilization are assessed by mode of delivery. The long term goal of this proposed research is to provide evidence based data that can be used by the Florida Obstetric and Gynecologic Society (FOGS), Florida Department of Health (FDOH), health organizations, policy makers, healthcare providers, and health insurers to reduce unnecessary C/S rate, maternal and infant morbidity and health care utilization. SPECIFIC AIMS AND RESEARCH QUESTIONS !Specic research questions have been posed to address the two specic aims (and phases) of this dissertation research: Specic Aim 1 : To determine the validity of data sources (e.g. Florida birth certicate, Florida hospital discharge data) used to investigate primary C/S delivery and late PTB outcomes using maternal medical charts as the gold standard. Research Questions: !1. What is the validity of the Florida birth certicate compared with maternal !medical charts? !2. What is the validity of the Florida Hospital discharge data compared with !maternal medical charts? 28

PAGE 39

!3. What is the validity of the linked Florida birth certicate and hospital !discharge data compared with maternal medical charts? !4. Is there a signicant difference between the validity of the linked Florida birth !certicate and hospital discharge data by hospital volume (high primary C/S !versus low primary C/S rate)? Specic Aim 2: Assess the impact of low documented risk C/S on maternal and late preterm infant morbidity. !5. What impact does low documented risk primary C/S have on maternal and !singleton late-preterm infant morbidity and healthcare utilization? !6. Is there variation by important subgroups (e.g. race/ethnicity, payer source)? DATA SOURCES FOR PROPOSED RESEARCH QUESTIONS !In order to address the proposed research questions, three data sources will be utilized: (1) FDOH hospital abstraction data le (from The Florida Late-Preterm and Cesarean Delivery Investigation), (2) Florida birth certicate data, and nally, (3) the Florida hospital discharge data le (AHCA data). !The FDOH abstraction investigation data le !The FDOH in partnership with the March of Dimes and FOGS conducted an investigation of singleton late PTBs attributed to primary C/S delivery. The objective of the investigation, known as The Florida Late-Preterm and Cesarean Delivery Investigation, was to compare a sample of medical charts from two hospital based cohorts: hospitals with high volume primary late PTB C/S deliveries and hospitals with low volume primary late PTB C/S deliveries. Initially, the FDOH selected 12 hospitals 29

PAGE 40

that each account for more than 2,000 births a year, with 6 hospitals reporting high C/S rates (39.35 to 58.33) and 6 hospitals reporting low C/S rates (11.85 to 25.07). Based on a power analysis conducted by FDOH epidemiologists, it was estimated that a total of 840 records were needed to be abstracted to achieve 90% power when alpha=0.05. Therefore, the FDOH randomly selected 70 late preterm live births delivered by primary cesarean delivery from each of the selected hospitals, using data years 2006 to 2007. Late preterm status was ascertained using dates based on last menstrual period from the birth certicate if and only if there was a two-week agreement with the reported clinical estimate. Primary cesarean delivery was ascertained based on birth certicate reporting of the procedure. Additional birth records (meeting same criteria) were randomly selected to replace live births with charts that could not be found for use in the investigation, or to replace records that did not meet inclusion criteria. !Trained abstractors used a structured abstraction tool (Appendix A) to abstract information from maternal medical charts. It should be noted that the FDOH did not perform re-abstraction of charts due to limited resources. For the abstraction, each trained OB-nurse abstractor was provided with a computer-generated listing of randomly selected live births with assigned study numbers and necessary identifying information for the abstraction. Abstractors then returned the list of births to the FDOH once the abstractions were complete. All forms were entered into a database and 10% of the records were veried by a reviewer to assure completion and accuracy. During the early phase of data collection, it was observed that many of the hospitals selected had records that were misclassied by the Florida Birth Certicate either the birth was vaginal or repeat CS. In order to assure that there were sufcient records that 30

PAGE 41

met the criteria for inclusion in the study, four additional hospitals were added. Currently, the abstraction study includes 16 hospitals, with a total of 1,249 abstracted records from maternal medical charts. The additional hospitals can also be classied as either high or low volume (two high volume and two low volume). !Hospitals were not informed of their volume status when selected for the investigation. Investigation results do not identify individual hospitals or doctors. IRB approval was not required for the FDOH investigation as the activity was classied as an investigation of a concern of public health signicance. As such, IRB approval from the FDOH or individual hospitals was not necessary. Once abstraction was completed, the abstraction data was linked to the Florida nal birth le for data years 2006 to 2007, as well as the Florida Hospital Discharge Data for 2006 to 2007. These linked and deidentied data les were used for this dissertation research. !Florida birth certicate data le !The Florida birth certicate data contains information on maternal demographics, risk factors, prenatal care utilization, obstetric procedures and birth outcomes. For this analysis, the Florida birth certicate data le includes births that occurred in Florida from 1998 to 2006. Hospital discharge data le The hospital discharge data le contained information on all hospital in-patient, ambulatory and emergency room charges for the time period 1998 to 2007. Charges for mothers and their infants as well as all subsequent rehospitalization episodes in the rst year postpartum were been linked. For the validation study, only the data years 2006 to 2007 were utilized to correspond to the Florida abstraction data le. However for 31

PAGE 42

analyses of maternal and infant morbidity and healthcare utilization, the data years 1998 to 2006 were used. It was necessary to include only those births from 1998 to 2006 so that a complete year of postpartum follow-up could be obtained. The hospital discharge data le contains ICD9 procedure and diagnosis codes. !All datasets used in this study were linked together using identiers such as names, social security numbers, birth dates and dates of service. The FDOH and its contractual partners performed this linkage and a de-identied le was created for all analyses. A benet of this data linkage is that mothers were matched with their infants, which allowed for an analysis of maternal demographics, risk factors, clinical procedures, and maternal and infant outcomes. CASE DEFINITION ALGORITHM !Low documented risk C/S was calculated through an algorithm established by the FDOH (in consultation with researchers and practitioners) (Goodman, Sappeneld, Mahan and Kogan, Submitted for Publication, 2010). This algorithm is a revised version of the Joint Commission Specications for Early Medically-Indicated Delivery (Joint Commission, 2010). According to this algorithm, all of the following will be excluded from the classication of low documented CS: !!Hypertension prepregnancy (Chronic) !!Hypertension gestational (PIH, Preeclampsia) !!Hypertension Eclampsia !!Diabetes prepregnancy (Diagnosis prior to the pregnancy) 32

PAGE 43

!!Gestational Diabetes (Diagnosis in this pregnancy) !!Prolonged Labor (> 20 hours) !!Moderate/Heavy meconium staining of the amniotic uid !!Fetal intolerance of labor !!Chorioamnionitis !!Non-vertex presentation !!Fetal presentation at birth other than cephalic !!A birth weight greater than 4,500 grams !!Any of these congenital anomalies: (Anencephaly, Congenital !! diaphragmatic hernia, Meningomyelocele/Spina bida, Omphalocele, !! Cyanotic congenital heart disease, Gastroschisia). !!Modied GINDEX indicates intensive prenatal care use (Alexander et al, !! 1987; Alexander & Kotelchuck, 1996). PLAN FOR THE DISSEMINATION OF STUDY FINDINGS !While this is one cohesive investigation of low documented risk C/S and singleton late PTBs in Florida, the results of this study can be grouped into three distinct manuscripts for publication: (1) validation of the Florida birth certicate and hospital discharge data, (2) the contribution of low documented risk C/S singleton late PTBs to infant morbidity, and (3) the contribution of low documented risk C/S singleton late PTBs to maternal morbidity. !The rst manuscript, provided in Chapter Two of this dissertation is titled "Accuracy of birth certicate and hospital discharge data by cesarean risk factors: The 33

PAGE 44

Florida Late-Preterm and Cesarean Delivery Investigation". There is very little known about the validity of public health data sources used to explore perinatal outcomes in Florida. Another important contribution of the validation study is to examine the validity of the birth certicate and hospital discharge data by hospital C/S rate category (high C/S rate or low C/S rate). The intended audiences for these results are healthcare providers such as obstetricians, public health policy makers and other key stakeholders. The Maternal and Child Health Journal has been consistently interested in issues surrounding C/S delivery, as well as the accuracy of data sources used by MCH researchers and policy makers. !The second manuscript, provided in Chapter Three is titled "The contribution of low indicated risk primary cesarean delivery to infant morbidity following singleton latepreterm birth". The primary publication target audience for this manuscript is obstetricians/gynecologists (OB/GYN's). As many of the results will directly pertain to clinical practice, OB/GYN's would be the most ideal audience. Therefore, this paper will be submitted to the journal Obstetrics & Gynecology. Obstetrics & Gynecology is one of the premier OB/GYN journals in the United States (if not internationally), and has an impact factor of 4.3 (LWW, 2009). !The third manuscript, provided in Chapter Four is titled "The contribution of low indicated risk primary cesarean delivery to maternal morbidity following singleton latepreterm birth." The target audience for this manuscript includes medical practitioners, public health professionals, the media, policy makers, and other key stakeholders (e.g. non-prot organizations such as the March of Dimes). For this reason, this manuscript 34

PAGE 45

will be submitted to the American Journal of Obstetrics & Gynecology, which has an impact factor of 2.9. MANUSCRIPT OPTION INTRODUCTION !This dissertation is presented in a manuscript format. The methodology and results of this dissertation will be described through three manuscripts. Each of the three manuscripts represent a separate, but related component of the dissertation, as each manuscript informs subsequent manuscripts. The validation study provided information on the accuracy of study variables used in subsequent manuscripts. The second and third manuscripts evaluated maternal and infant morbidity by calculating a distribution of reported morbidities (with Markov Chain Modeling), and used the median of the distribution as a cut off point for morbidities to be evaluated. These morbidities were then compared by mode of delivery. In order to keep the reference format consistent throughout the dissertation, journal-specic reference formatting will take place at the end of the dissertation. 35

PAGE 46

CHAPTER TWO MANUSCRIPT ONE TITLE !Accuracy of birth certicate and hospital discharge data by cesarean risk factors and late-preterm birth outcomes: The Florida Late-Preterm and Cesarean Delivery Investigation 36

PAGE 47

INTRODUCTION !The rate of cesarean section (C/S) delivery in the United States has increased for 11 consecutive years (Hamilton, Martin, Ventura, 2009). In 2007 alone, the C/S rate increased 2%, reaching a rate of 31.8 in the U.S. The C/S rate in Florida is higher than the national average (37.2% versus 31.8%) (Hamilton et al, 2009). Also during this time period, the United States has observed a shift in the epidemiology of gestational length among singleton births towards birth at an earlier gestational age (GA) (Davidoff et al, 2006). This shift in the distribution of GA at birth may be partially explained by concurrent increases in obstetric interventions such as induction of labor and C/S delivery (Ananth, 2005; Davidoff, 2006; IOM, 2007; MacDorman, 2002). Presumably, C/S deliveries at term should not have a large impact on the increasing rate of late-preterm birth in the US, however, given the inaccuracy of gestational dating during pregnancy (+ 2 weeks), deliveries occurring at "presumed term" may result in late-preterm infants (infants born between 34 to 36 completed weeks of gestation) (Engle, 2006; Fuchs & Wapner, 2006). Unfortunately, late-preterm infants have three times the mortality of term infants and substantially higher morbidity than infants born at term (Fuchs et al, 2006; McLaurin et al, 2009; Wang, 2004). The prevention of premature birth is a public health priority, and thus the potential contribution of primary C/S delivery to late-preterm birth is an important research focus. !While there has been substantial research on the accuracy of live birth certicates and some on hospital discharge data compared to the "gold standard" the medical record, little information is available on the accuracy of these data sources used to investigate C/S delivery and preterm birth in the state of Florida. To date, there has been 37

PAGE 48

no report of the accuracy of the Florida birth certicate data compared to maternal medical charts. Furthermore, it is not known whether hospitals with high or low C/S rates have variation in the accuracy of reporting to vital records and administrative databases. Systematic differences in the quality of reporting may exist, which could result in biased estimates of C/S rates overestimation or underestimation. This research seeks to address this decit in knowledge by: (1) investigating the accuracy of variables on the Florida birth certicate, the Florida hospital discharge data and the linked Florida birth certicate hospital discharge data le; and (2) to determine if there are differences in the accuracy of reporting maternal and infant hospitalization information and outcomes by C/S rate category of hospitals in Florida, high C/S rate versus low C/S rate. MATERIALS AND METHODS !The data used for this analysis was obtained from the Florida Investigation of Late Preterm and Cesarean delivery, a public health investigation conducted by the Florida Department of Health (FDOH) in response to concerns over Florida's increase in births at late preterm (gestational age of 34 to 36 completed weeks). This investigation used data from the 2006-2007 birth certicates to select a sample of singleton late preterm primary cesarean deliveries for abstraction from maternal medical charts. The FDOH used a two stage sampling method. In the rst stage, inclusion criteria required that hospitals had a minimum of 2000+ live births per year and at least 70 such deliveries. In the second stage, hospitals were ranked by primary cesarean rates and only the highest six (primary C/S rates 39.35 to 58.33) and lowest six (primary C/S rates of 11.85 38

PAGE 49

to 25.07) hospitals were selected for record abstraction. High and low C/S rate hospitals were selected to allow for a sufcient contrast to view potential differences in hospital practices, patient population or other factors that might explain the difference in C/S rates. !The FDOH conducted a power calculation to determine the necessary sample size for hospital comparisons. The calculation was based on several design assumptions: independent, retrospective, difference of two proportions and an uncorrected chi-square. Furthermore, the power calculation assumptions included an alpha of 0.05 and a beta of 0.10 (90% power), an assumption of a two-fold increase in the C/S delivery ratio between hospitals with high and low rates of primary C/S late-preterm delivery, and that the problem identied would explain at least half of the rate increase observed. Therefore, a total of 840 medical charts were needed to meet the FDOH investigation objective to compare high C/S and low C/S hospitals. During the initial maternal medical chart abstractions, a higher rate of misclassied repeat cesarean deliveries than anticipated was discovered, so additional records were sampled. The nal maternal medical chart abstraction le contained data from 1,249 births that occurred in 16 hospitals in Florida from 2006 to 2007, although this sample includes some misclassied data (e.g. repeat cesarean (N= 175, 14.0%), vaginal delivery (N=11, 0.9%), multiple gestations (N=4, 0.3%) and non-resident births (N=4, 0.3%)). The data from the Florida Late-Preterm and Cesarean Delivery Investigation was classied by Florida Statute as a public health investigation, and as such, was not subject to IRB approval according to the FDOH's IRB program. 39

PAGE 50

!As a result of some misclassication of data, out of the 1,249 maternal charts abstracted, 1,055 met the criteria for inclusion. Furthermore, after linking maternal medical chart and birth certicate data to hospital discharge data to create the linked data le (referred to as the linked data le' from this point forward), 944 of the 1,055 records (89.5%) were available for analyses; 138 records had missing or otherwise invalid data for SSN, and thus could not be matched to hospital discharge data (Figure 2.1). Three data sources were validated using data abstracted from maternal medical records in the Florida Investigation of Late Preterm and Cesarean Delivery: (1) live birth certicates, (2) hospital discharge data, and (3) the linked data le. The FDOH in partnership with the University of South Florida (USF), College of Public Health (COPH), linked the maternal medical chart data to these data sources. Once each linkage was complete, identiers were removed to assure condentiality of data. Only de-identied data was used in the analyses. !The validation includes items from several domains used to assess indications for primary cesarean delivery and risk for late-preterm birth: maternal medical history, obstetric history, complications in current pregnancy, complications and procedures in labor and delivery. Maternal medical records misclassied on mode of delivery were not fully abstracted. Thus validation efforts were conducted in two stages. In the initial stage, the mode of delivery was validated. The second stage focused on those correctly classied as primary C/S delivery. All available abstracted data elements--maternal medical history, obstetric history, pregnancy complications and labor and delivery elements--were included in this stage. 40

PAGE 51

!Statistical analyses !To assess the accuracy of maternal medical records compared to (1) birth certicates, and (2) discharge data, the sensitivity, specicity, positive predictive value (PPV), negative predictive value (NPV), and the likelihood ratio were calculated. Consistent with other validation studies of birth certicates and discharge data, the medical record was considered to be the "gold standard" with regards to comparisons of accuracy (Lydon-Rochelle, Holt, Cardenas, Nelson, Easterling, Gardella, Callaghan, 2005; Roberts, Bell, Ford & Morris, 2008). Measures of reliability were also performed for the two data sources validated using kappa statistic. This measure of reliability is a more accurate estimate when conditions or procedures are infrequent events and thus missing on a large proportion of records (Zollinger, Przybylski & Gamache, 2006). Kappa statistic values below 0.41 were considered to be "fair/poor agreement", values from 0.41 to 0.60 were considered "moderate", values between 0.61 and 0.80 were considered to represent "substantial agreement", while 0.81 to 1.00 represented "almost perfect agreement" (Landis & Koch, 1977). To assess whether the accuracy of the linked data varied by hospital C/S rate category, sensitivity, 95% condence intervals for sensitivity, PPV and Kappa statistics were calculated. !Analyses were unweighted, as tests (data not shown) indicated that the sample sizes of hospitals did not result in an overrepresentation of data by any one hospital which was a potential concern due to the larger number of records available for abstraction in high CS rate hospitals. All analyses were conducted in SAS version 9.2. 41

PAGE 52

RESULTS !Overall, 1,249 medical records were abstracted from 16 hospitals. Of those that were abstracted, 1,055 were correctly classied as singleton primary cesarean deliveries on birth certicates, while those that were misclassied included repeat cesarean deliveries (n=175, 14.0%), vaginal deliveries (n=11, 0.9%), multiple gestations (n=4, 0.3%) and non-resident births (n=4, 0.3%). Some variation in misclassication by hospital volume was found for both birth certicate and hospital inpatient discharge data (Figure 1). High volume CS hospitals were signicantly more likely to incorrectly classify a repeat CS delivery as a primary CS delivery in vital records (17.8% Repeat CS among high volume hospitals compared to 10.1% in low volume hospitals, p < 0.0001), and this nding was also similar for discharge data (17.5% repeat C/S among high volume hospitals, and 9.9% among low volume C/S hospitals, p<0.0001). Approximately 77.2% of late preterm singleton live births were correctly classied on birth certicates. As report of gestational age in discharge data is based upon underlying pathology (prematurity, small for dates, large for dates), gestational age originating from discharge data was not comparably collected as with birth certicates and maternal medical charts. Therefore, it was not assessed in this study. For birth certicate data, no signicant variation in the correct classication of late preterm birth was identied by hospital C/S rate classication. !Table 2.1 depicts selected sample characteristics such as demographic information, history of prior live births, number of prenatal visits in current pregnancy, birth weight and gestational age by proportion of the sample originating from hospitals classied as high rate C/S or low rate C/S and data source. Several of these 42

PAGE 53

characteristics are not commonly available in discharge data or comparably reported, and thus no results are presented for these variables. According to the maternal medical chart, a large proportion of the sample was Hispanic (36.30%), followed by NonHispanic White (36.21%), Non-Hispanic Black (22.46%) with about 5% of the sample classied as "other". This distribution was similarly reported by birth certicates, although the proportion of the sample reported as Hispanic was somewhat higher at roughly 43%. Discharge data was not consistent with maternal medical chart or birth certicate data indicating that the majority of the sample was Non-Hispanic White (43.22%), followed by NonHispanic Black (26.8%) and Hispanic (25.0%). Parity, infant birth weight and estimated gestational age (in weeks) were similarly reported in the maternal medical charts and in birth certicates. However, the number of prenatal visits reported in birth certicates were not consistent with maternal medical charts, with a larger proportion of visits numbering > 12, 42.7% of birth certicates versus 18.8% in medical chart data. !The validity of maternal medical conditions and risk factors as well as labor and delivery factors on birth certicates, discharge data and linked data sources compared to maternal medical chart data is provided in Table 2.2. Substantial variation in the sensitivity and PPV of maternal medical risk factors and labor/delivery factors was found for data elements in the birth certicate, suggesting that the ability of vital records to capture the presence of conditions listed in the maternal medical chart varied signicantly. The sensitivity ranged from 0 for renal disease to 0.76 for maternal obesity (data not shown). Data elements with at least 50% sensitivity and PPV rates include: chronic diabetes (Sens=0.54, PPV=0.69), hypertensive conditions of pregnancy --(Sens = 43

PAGE 54

0.55, PPV = 0.79), Trial of Labor (TOL) (Sens=0.56, PPV = 0.79), induction of labor (Sens = 0.52, PPV = 0.63), and nally, breech/malpresentation (Sens = 0.62, PPV = 0.95). Specicity and NPV did not vary as widely as sensitivity and PPV, with specicity ranging from 0.74 for TOL to 1.0 for several factors such as intrauterine growth restriction (IUGR), cerclage, HIV, placenta previa, and prolonged rupture of membranes (NPV data not shown). Kappa values and likelihood ratios (LR) (+/-) have indicated at least a moderate level of agreement for obesity, chronic diabetes, gestational diabetes, hypertensive conditions of pregnancy, induction of labor, and breech/ malpresentation. Only the Kappa values for obesity and breech/malpresentation indicate a better than moderate level of agreement (Kappa >0.60) (data not shown). !The validity of maternal medical conditions and risk factors as well as labor and delivery factors in discharge data compared to maternal medical chart data is also presented in Table 2.2. Overall, the validity indices for data elements were much higher when medical charts were compared to discharge data than vital records. Again, there was a great deal of variation in sensitivity and PPV for selected data elements, with sensitivity ranging from 0 for syphilis to 0.89 for hypertensive conditions of pregnancy. Data elements with sensitivity and PPV rates of at least 50% include IUGR, cerclage, chronic diabetes, chronic hypertension, gestational diabetes, hypertensive conditions of pregnancy, HIV infection, breech/malpresentation, placenta previa, placental abruption and prolonged labor. Also, rates of specicity and NPV were quite high, better than 90%, for the majority of data elements. The notable exception was TOL (Spec = 0.54, NPV = 0.47) (NPV data not shown). Kappa values and likelihood ratios (LR) (+/-) showed a moderate level of agreement for chronic diabetes, induction of labor and 44

PAGE 55

chorioamnionitis. Several data elements had levels of agreement that ranged from substantial to almost perfect agreement: IUGR, cerclage, chronic hypertension, gestational diabetes, hypertensive conditions of pregnancy, breech/malpresentation, placenta previa, placental abruption and prolonged labor. !Validity of selected data elements on the maternal medical charts was also compared to the linked data le (Table 2.2). In comparison to accuracy of the birth certicate and discharge data sources alone, rates of sensitivity and PPV were improved for many data elements with the linked data le. Sensitivity ranged from a low of 0.08 with renal disease to a high of 0.91 with hypertensive conditions of pregnancy and breech/malpresentation. About half of the data elements demonstrate a rate of sensitivity and PPV of at least 50%: IUGR, cerclage, chronic diabetes, chronic hypertension, gestational diabetes, hypertensive conditions of pregnancy, genital herpes, HIV infection, TOL, breech/malpresentation, induction of labor, placenta previa, placental abruption, and prolonged labor. Overall, the rates of specicity and NPV were higher than 90%. The exceptions were three data elements: hypertensive conditions of pregnancy (Spec = 0.89, NPV = 0.96), TOL (Spec = 0.39, NPV = 0.56) and fetal distress (Spec = 0.64, NPV = 0.71) (NPV data not shown). !Lastly, Table 2.3 compares the validity of selected data elements by hospital C/S rate classication for the linked data compared to data abstracted from the maternal medical chart. Large variation in sensitivity, PPV and Kappa was observed by data source and by hospital C/S rate classication. For the majority of data elements included in the analysis, rates of sensitivity were higher for hospitals classied as low C/S rate except for gonorrhea, genital herpes, fetal distress, breech/malpresentation, 45

PAGE 56

chorioamnionitis, and placental abruption. For 17 of the 26 data elements assessed, the PPV was higher for hospitals classied with a low C/S rate. The only elements with higher PPV among high C/S rate hospitals include chronic hypertension, heart problems, gestational diabetes, Gonorrhea, genital herpes, chorioamnionitis, placenta abruption, prolonged rupture of membranes, and meconium. Notably, for many of these factors, while the PPV was higher among the high C/S rate hospitals, sensitivity was higher among the low C/S rate hospitals: chronic hypertension, heart problems, gestational diabetes, prolonged rupture of membranes, and meconium. Each data element was tested for signicant differences in accuracy by hospital C/S rate classication. We observed signicant differences in accuracy for three data elements: anemia, attempted labor, and induction of labor. For each of these factors, hospitals with low C/S rate classication had higher rates of sensitivity compared to high C/S rate hospitals. DISCUSSION !After comparing data abstracted from maternal medical charts to birth certicate and discharge data, we observed some misclassication for cesarean delivery, with about 14% of birth certicate data and 13.9% of hospital discharge data classifying repeat cesarean deliveries as primary cesareans. Furthermore, we observed a very small proportion of deliveries incorrectly classied by both birth certicate and discharge data as singleton primary cesarean live births when they were actually vaginal deliveries or multiples based on maternal medical chart data. Based on the medical chart, late preterm births were correctly classied as such on the birth certicate 77% of the time. This 46

PAGE 57

suggests that there are some concerns about potential misclassication of mode of delivery, as well as gestational age at delivery a concern for researchers and policy makers who use public health data sources to investigate adverse outcomes of delivery mode and prematurity. !We found a very low rate of sensitivity and PPV for many data elements assessed in birth certicates. As there have been no reports on previous efforts to validate the Florida birth certicate, it is difcult to compare our results with other sources that were more representative of all births. However, our observations are consistent with many previously published reports of the validity of vital records data in other states (DiGiuseppe et al, 2002; Piper et al, 1993; Reichman et al, 2001; Roohan, Josberger, Acar, Dabir, Feder & Gagliano, 2003). Validation studies of the birth certicate data in New Jersey, Tennessee, New York and Ohio reported low rates of sensitivity for many medical risk factors, obstetric procedures and complications of labor and delivery (DiGiuseppe, Aron, Ranbom, Harper & Rosenthal, 2002; Piper, Mitchel, Snowden, Hall, Adams & Taylor, 1993; Reichman et al, 2001; Roohan et al, 2003). Interestingly, the sensitivity and PPV for many of the data elements we assessed in Florida vital records data was higher than the rates published by DiGuisueppe et al (2002), Reichman et al (2001) and Roohan et al (2003). Only Zollinger et al (2006), using 1996 birth data from Indiana consistently reported higher rates of sensitivity and PPV for the majority of data elements contained on the birth certicate. It should be noted though that many of the values for sensitivity and PPV reported by Zollinger were higher than 90%. Furthermore, our analysis is based on data reported on the 2006-2007 birth certicate, which Florida revised in 2004 in accordance with the new U.S. birth certicate format (Osterman et al, 47

PAGE 58

2009). Many of the published reports on the validity of birth certicate data are based on the earlier birth certicate format, which may not report information in a similar manner. !Consistent with previously published reports, we also observed improved accuracy with discharge data, more accurate than birth certicate data (Kahn et al, 2009; Lydon-Rochelle et al, 2005). The sensitivity (and Kappa values) of most data elements in Florida's discharge data ranged from moderate to high, and improved further once linked with birth certicates. As with our current study, Lydon-Rochelle et al (2005) compared the accuracy of birth certicate and linked birth certicate and discharge data to maternal medical charts and observed that the linked data le greatly improved sensitivity for many maternal conditions and pregnancy complications. There were some substantial differences in accuracy for many conditions in the linked data reported by Lydon-Rochelle et al compared to our results. For example, Lydon-Rochelle et al reported high rates of sensitivity for gestational diabetes (93.3 versus 70.0), diabetes mellitus (96.9 versus 83.0) and chronic hypertension (70.3 versus 64.0). However, for some conditions, the Florida linked data had higher sensitivity pregnancy-induced hypertension (91.0 versus 73.5) and placenta previa (82.0 versus 69.5). In our study, pregnancy induced hypertension included eclampsia while Lydon-Rochelle et al reported results for eclampsia and pregnancy induced hypertension separately. !Investigators have also focused on the accuracy of discharge data for assessing indications of cesarean delivery. Kahn et al (2009) evaluated the accuracy of birth certicate and discharge data and found that many indications for primary cesarean delivery were signicantly underreported in birth certicate data compared to discharge data. For example, according to risk algorithms using birth certicate data, 59.2% of 48

PAGE 59

primary cesareans had no indications for cesarean delivery, while using discharge data that proportion reduced to 3.9% (Kahn et al, 2009). Our present study supports this nding many indications and other risk factors for cesarean delivery (e.g. hypertensive conditions of pregnancy, hypertension, diabetes mellitus, gestational diabetes, IUGR, and breech/malpresentation) demonstrated higher rates of accuracy (particularly sensitivity) in discharge data compared to birth certicate data. According to our results, using linked birth certicate and discharge data together to classify cesarean risk groups would result in lower potential for underestimation of risk (more accurately classied risk groups). The results of Kahn's study, as well as the study by Korst et al (2004) provides compelling evidence for the need to utilize discharge data for studies that include pregnancy risk factors, obstetric procedures, and labor/delivery complications. This has also been supported by Lydon-Rochelle et al (2005) who concluded that "a strategy of using combined data sources was more accurate for the detection of maternal pre-existing medical conditions and the complications of pregnancy than single data source strategies" (pg. 133). !We observed signicant differences in accuracy of data elements related to pregnancy and labor/delivery (for vital records data, administrative data and the linked data le) by hospital C/S rate classication. Our nding of consistently higher rates of sensitivity and PPV for low C/S rate hospitals is novel. Hospitals with higher rates of C/ S delivery may be more likely to represent a high risk obstetric population. However, presumably, this higher risk population should translate into a higher rate of morbid conditions resulting in higher PPV for many conditions. While this was true for some data elements (e.g. heart problems, gestational diabetes, labor induction, prolonged 49

PAGE 60

rupture of membranes, and meconium staining), however, for the majority of data elements, sensitivity and PPV was higher for low C/S rate hospitals, suggesting differences in reporting quality by hospital C/S rate classication. This observation is a cause of concern for studies of cesarean delivery as there are certainly strengths and limitations of each data source (birth certicate and discharge data), but also a substantial difference in quality by institutional rate of cesarean delivery. We applied a low medical risk algorithm for primary C/S (gure 2.2) to the medical abstraction data, the birth certicate data, the hospital discharge data and the linked birth certicate discharge data to determine the impact of variation in data accuracy in classication of low risk primary C/S. Using the classication system, 86.6% of women in the medical chart were considered high risk for primary C/S, compared to 63.6% in the birth certicate, 81.5% in hospital discharge data, and 83.1% in the linked data le. There were signicant differences in the classication of high risk for primary C/S by hospital C/S rate classication as well. High C/S rate hospitals had a consistently lower proportion of deliveries classied as "high risk primary C/S" compared to low C/S rate hospitals, for all data sources except the birth certicate. !Why hospitals with high cesarean section rates have lower levels of accuracy for data reported on the birth certicate and hospital discharge requires further study. A better understanding of how data is collected and reported by hospitals will need to be developed. From an ecological perspective, the issue of data quality is complex, and solutions for improvement must consider various levels of inuence points of error and opportunities for improvement. For example, quality of data depends on provider documentation, patient recall, medical coding clerk entries, institutional level support for 50

PAGE 61

data quality monitoring, and more macro level factors such as policy and other supports for data improvement. On a macro level, this issue may be aided by the proposed changes to the U.S. healthcare system. At present, the overall performance of the U.S. healthcare system is poor the United States is ranked 37th in the world, yet spends more money on healthcare than any other nation (Murray and Frank, 2010). Our healthcare system has been described as a cottage industry (Swensen, Meyer, Nelson, Hunt, Pryor, Weissberg et al, 2010) a system that has poor integration, unmeasured performance, customized care for individual patients, and very little effort for standardization. !According to Swensen et al (2010), the U.S. healthcare system pays for volume instead of value resulting in increased tests, exams, procedures and surgeries. The present concern over cesarean delivery is an ideal case study for the problems resulting from the current structure of the U.S. healthcare system. Rates of cesarean delivery continue to increase, despite evidence that C/S results in poorer infant and maternal outcomes. !One of the most pertinent proposed changes to the U.S. Healthcare system is the transition to standardized electronic medical records. Since 2009, the U.S. Government has been drafting policy to move the healthcare system towards "a nationwide, interoperable, private, and secure electronic health information system" (Blumenthal, 2010, pg. 382). Accurate information is essential for modern medicine. Without accurate information, physicians and healthcare institutions cannot perform optimally for their patients (Blumenthal, 2010). The results of our present investigation demonstrate substantial differences in accuracy, and suggest that medical charts (in their present form) may not be the appropriate "gold standard". With the adoption of electronic medical 51

PAGE 62

records, information can be collected in a standardized, centralized format resulting in not only more accurate data, but also improvement in patient care, and subsequent healthcare outcomes. !Strengths and Limitations !One of the primary strengths of this study is the use of three data sources to provide information about the quality of data used to investigate late preterm singleton primary cesarean deliveries in Florida. This is an important activity as the use of maternal medical charts is unrealistic for many public health investigations and other research or policy endeavors. Further, there have been no published reports on the accuracy of data elements related to cesarean delivery by hospital C/S rate classication. We have demonstrated that for many data elements related to maternal medical conditions, risk factors and complications during labor and delivery, low C/S rate hospitals have higher sensitivity and PPV than high C/S hospitals. However we caution that these ndings are based on a non-representative sample of singleton late preterm primary cesarean deliveries, such that caution should be used in generalizing these ndings to all live births. !There are several important limitations of this study that should be noted. One of the most important limitations is the non-representative sample. As the design of this study was not intended to validate the Florida birth certicate nor hospital discharge data, results must be interpreted in the context of the population studied; this context is worthy given the current national interest in cesarean and late preterm delivery. Our results are not representative of accuracy indices for all births in Florida. Secondly, this investigation did not estimate the reverse error on singleton late preterm primary 52

PAGE 63

cesareans; some primary cesarean deliveries may have been reported as repeat cesarean on birth certicates. Because we have no information on the accuracy of these births that were recorded on the birth certicate, the rate of misclassication is captured in only one direction. !It is also important to consider the purpose of each data source used in this investigation, as that may impact the thoroughness of reporting. The purpose of birth certicate data is to record medical and obstetric history, pregnancy and labor/delivery complications as well as some maternal and infant health outcomes. Hospital discharge data provides information on conditions upon which reimbursement is required and is also used for policy decisions. Some common pregnancy risk factors such as obesity, use of alcohol, tobacco, and so forth may not be pertinent for treatment nor reimbursement at the current admission event, and therefore may be more likely to be underreported in hospital discharge data. We observed this nding. Finally, the results of this analysis are based upon comparison to the maternal medical chart which is often reported in the literature as the gold standard'. The hospital medical record is not only a source of information for hospital related care, but it also holds an important status in terms of the medical-legal context, therefore we must assume for this analysis that information contained in the medical chart is accurate (DiGiuseppe et al, 2002). However, the accuracy and completeness of the medical chart related to delivery is not fully known. Some variables such as prenatal care utilization may be more accurate in outpatient records (DiGiuseppe et al, 2002). 53

PAGE 64

!Conclusion !To our knowledge, this is the rst study to investigate the accuracy of the Florida birth certicate, hospital inpatient discharge data and the linked data le using maternal medical charts as the gold standard. This investigation provides important information for the accuracy of public health data sources used to investigate late preterm singleton primary cesarean delivery an important public health issue. In accordance with previously published research, we found that discharge data and the linked data le were more accurate than birth certicate data alone (Kahn et al, 2009; Lydon-Rochelle et al, 2005). We also observed that for many maternal medical conditions and risk factors, as well as labor and delivery factors, hospitals classied as low C/S rate had more accurately reported data, even when using the linked birth certicate and hospital inpatient data le. Our results suggest using birth certicate data alone may result in an overestimate of the number of women who have no indicated medical risk factors for primary cesarean delivery or premature birth. Therefore linked birth certicate and discharge data should be used when possible. !Population-based data sources are important for policy makers, government ofcials and researchers who are working to prevent preterm birth (Herrchen, 1997; Roohan, 2003). It is essential to know the accuracy of the information recorded in each data source (Zollinger, 2006) as research ndings inform clinical practice as well as healthcare policy. Given the results of this study, we strongly advocate using both birth certicate and discharge data when conducting research focused on maternal and child health (MCH) populations. 54

PAGE 65

Table 2.1. Characteristics of study sample by data source (medical record, birth certicate, hospital discharge data) Table 2.1. Characteristics of study sample by data source (medical record, birth certicate, hospital discharge data) Table 2.1. Characteristics of study sample by data source (medical record, birth certicate, hospital discharge data) Table 2.1. Characteristics of study sample by data source (medical record, birth certicate, hospital discharge data) Characteristics Medical records (%) Birth Certicates(%) ACHA (%) a Race/Ethnicity b Hispanic 383 (36.30) 449 (42.56) 236 (25.35) Non-Hispanic White 382 (36.21) 350 (33.18) 408 (43.82) Non-Hispanic Black 237 (22.46) 225 (21.33) 240 (25.78) Non-Hispanic Other 53 (5.02) 31 (2.94) 47 (5.05) Parity (n) c 0 714 (70.00) 719 (70.15) 1 191 (18.73) 196 (19.12) 2 86 (8.43) 81 (7.90) # 3 29 (2.84) 29 (2.83) Prenatal Visits (n) 0-5 370 (35.07) 86 (8.15) 6-11 487 (46.16) 519 (49.19) # 12 198 (18.77) 450 (42.65) Birth Weight <2500 476 (45.12) 444 (42.09) 2500-3499 493 (46.73) 522 (49.48) 3500-3999 69 (6.54) 71 (6.73) # 4000 17 (1.61) 18 (1.71) Estimated gestational Age (wk) <34 58 (5.50) 0 (0.00) 34-36 815 (77.25) 1055 (100.00) # 37 182 (17.25) 0 (0.00) Footnotes: a N=944; b AHCA missing=13; c MR missing =35; BC missing =30 Footnotes: a N=944; b AHCA missing=13; c MR missing =35; BC missing =30 Footnotes: a N=944; b AHCA missing=13; c MR missing =35; BC missing =30 Footnotes: a N=944; b AHCA missing=13; c MR missing =35; BC missing =30 55

PAGE 66

Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts Table 2.2 Accuracy of Data Elements Reported on the Florida Birth Certicate, Hospital Discharge Data, and Linked Data File Compared to Maternal Medical Charts # of Cases in Medical Charts a Birth Certicate a Birth Certicate a Birth Certicate a Birth Certicate a # of Cases in Medical Charts b Hospital Discharge b Hospital Discharge b Hospital Discharge b Hospital Discharge b Linked File b Linked File b Linked File b Linked File b # of Cases in Medical Charts a # of Cases Sen Spe PPV # of Cases in Medical Charts b # Cases Sen Spe PPV # Cases Sen Spe PPV Maternal Medical Conditions & Risk Factors " " " " " " " Prior Preterm 70 12 0.13 1.00 0.75 64 4 0.05 1.00 0.75 13 0.14 1.00 0.69 IUGR 135 11 0.08 1.00 1.00 131 115 0.74 0.98 0.84 117 0.76 0.98 0.85 Cerclage 18 6 0.28 1.00 0.83 17 17 0.65 0.99 0.64 19 0.71 0.99 0.63 Anemia 141 8 0.04 1.00 0.63 130 110 0.40 0.93 0.47 118 0.41 0.92 0.45 Chronic Diabetes 41 32 0.54 0.99 0.69 41 38 0.76 0.99 0.82 49 0.83 0.98 0.69 Chronic Hypertension 100 33 0.25 0.99 0.76 95 65 0.60 0.99 0.88 76 0.64 0.98 0.80 Renal Disease 66 1 0.00 1.00 0.00 64 12 0.08 0.99 0.42 13 0.08 0.99 0.38 Heart Conditions 58 3 0.03 1.00 0.67 58 20 0.26 0.99 0.75 24 0.29 0.99 0.71 Asthma 101 9 0.08 1.00 0.89 94 26 0.27 1.00 0.96 31 0.31 1.00 0.94 Gestational Diabetes 122 63 0.41 0.99 0.79 110 82 0.65 0.99 0.88 99 0.70 0.97 0.78 Hypertensive Conditions of Preg 285 197 0.55 0.95 0.79 266 292 0.89 0.92 0.81 314 0.91 0.89 0.77 Gonorrhea 14 10 0.43 1.00 0.60 12 1 0.08 1.00 1.00 9 0.42 1.00 0.56 Genital Herpes 75 9 0.08 1.00 0.67 72 36 0.49 1.00 0.97 43 0.54 1.00 0.91 Syphilis 7 1 0.14 1.00 1.00 7 1 0.00 1.00 0.00 2 0.14 1.00 0.50 HIV 14 5 0.36 1.00 1.00 14 11 0.79 1.00 1.00 11 0.79 1.00 1.00 Labor/Delivery Factors " " " " " " " Trial of Labor 700 493 0.56 0.71 0.79 619 571 0.68 0.54 0.74 714 0.84 0.39 0.72 Augmentation 111 120 0.48 0.93 0.44 * * * * Induction 207 171 0.52 0.93 0.63 192 120 0.44 0.95 0.71 213 0.71 0.90 0.64 Chorioamnionitis 21 10 0.19 0.99 0.40 20 14 0.40 0.99 0.57 21 0.50 0.99 0.48 Placenta Previa 58 7 0.12 1.00 1.00 50 46 0.80 0.99 0.87 47 0.82 0.99 0.87 Placenta Abruption 68 9 0.12 1.00 0.89 61 50 0.69 0.99 0.84 55 0.74 0.99 0.82 Prolonged ROM 121 1 0.01 1.00 1.00 109 20 0.17 1.00 0.95 21 0.18 1.00 0.95 Prolonged Labor 195 9 0.03 1.00 0.67 179 133 0.64 0.98 0.86 136 0.64 0.97 0.84 Fetal Distress 324 182 0.18 0.83 0.32 297 246 0.29 0.75 0.35 355 0.42 0.64 0.35 Breech/ Malpresentation 267 174 0.62 0.99 0.95 236 254 0.89 0.94 0.83 262 0.91 0.93 0.82 Meconium 59 36 0.27 0.98 0.44 53 15 0.13 0.99 0.47 46 0.36 0.97 0.41 Assisted Delivery (Forceps/Vacuum) 20 7 0.05 0.99 0.14 18 12 0.28 0.99 0.42 12 0.28 0.99 0.42 Infant Outcomes " " " " " " " Infant Sex   * 0.96 0.97 * * * * Foot Notes *Not Applicable; a N=1055; b N=944 *Not Applicable; a N=1055; b N=944 *Not Applicable; a N=1055; b N=944 *Not Applicable; a N=1055; b N=944 " " " " " 56

PAGE 67

Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. Table 2.3 Comparison of validity indices in the linked birth discharge data les by hospital cesarean section rate classication. # of Cases in Medical Records # of Cases in Linked Sen Sen PPV PPV Kappa Kappa # of Cases in Medical Records # of Cases in Linked High CS Low CS High CS Low CS High CS Low CS Maternal Medical Conditions & Risk Factors " " " " Previous Preterm 64 13 0.00 0.20 0.00 0.90 -0.01 0.30 Intrauterine Growth Restriction 131 117 0.73 0.78 0.81 0.88 0.74 0.79 Cerclage 17 19 0.56 0.88 0.45 0.88 0.49 0.87 Anemia 130 118 0.19 0.55 0.32 0.49 0.17 0.42 Chronic Diabetes 41 49 0.71 0.89 0.62 0.73 0.66 0.79 Chronic Hypertension 95 76 0.55 0.69 0.82 0.80 0.63 0.71 Renal Disease 64 13 0.04 0.11 0.33 0.40 0.06 0.14 Heart Problems 58 24 0.24 0.34 0.78 0.67 0.35 0.43 Asthma 94 31 0.17 0.45 0.89 0.95 0.26 0.58 Gestational Diabetes 110 99 0.59 0.80 0.83 0.75 0.66 0.74 Hypertensive Conditions of Pregnancy 266 314 0.90 0.92 0.77 0.77 0.77 0.75 Gonorrhea 12 9 1.00 0.30 0.67 0.50 0.80 0.37 Genital Herpes 72 43 0.58 0.50 0.96 0.85 0.70 0.61 Syphilis in current pregnancy 7 2 0.00 0.25 0.00 0.50 0.00 0.33 HIV 14 11 0.40 1.00 1.00 1.00 0.57 1.00 Labor/Delivery Factors " " " " Attempted Labor 619 714 0.78 0.88 0.70 0.75 0.22 0.26 Fetal distress 297 355 0.45 0.39 0.29 0.43 0.00 0.14 Breech/Malpresentation 236 262 0.91 0.90 0.74 0.87 0.77 0.84 Induction 192 213 0.59 0.81 0.61 0.66 0.51 0.64 Chorioamnionitis 20 21 0.86 0.31 0.50 0.44 0.62 0.35 Placenta Previa 50 47 0.76 0.86 0.84 0.89 0.79 0.87 Placenta Abruption 61 55 0.82 0.67 0.88 0.76 0.84 0.69 Prolonged rupture of membranes 109 21 0.11 0.24 1.00 0.94 0.18 0.34 Prolonged labor 179 136 0.62 0.66 0.80 0.87 0.64 0.70 Assisted Delivery (Forceps/Vacuum) 18 12 0.17 0.33 0.20 0.57 0.17 0.41 Meconium 53 46 0.33 0.37 0.50 0.37 0.38 0.33 Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence Sen is Statistically Different at 95% Condence 57

PAGE 68

Figure 2.1. Description of sample and linkage results 58

PAGE 69

Figure 2.2 Low Risk Primary C/S algorithm Low Risk Primary C/S Algorithm : !! !!Hypertension prepregnancy (Chronic) !!Hypertension gestational (PIH, Preeclampsia) !!Hypertension Eclampsia !!Diabetes prepregnancy (Diagnosis prior to the pregnancy) !!Gestational Diabetes (Diagnosis in this pregnancy) !!Prolonged Labor (> 20 hours) !!Moderate/Heavy meconium staining of the amniotic uid !!Fetal intolerance of labor !!Clinical chorioamnionitis !!Non-vertex presentation !!Fetal presentation at birth other than cephalic !!A birth weight greater than 4,500 grams !!Any of these congenital anomalies: (Anencephaly, Congenital !! diaphragmatic hernia, Meningomyelocele/Spina bida, Omphalocele, !! Cyanotic congenital heart disease, Gastroschisia). !!Modied GINDEX indicates intensive prenatal care use (Alexander et al, !! 1987; Alexander & Kotelchuck, 1996). 59

PAGE 70

CHAPTER THREE MANUSCRIPT TWO TITLE !The contribution of primary Cesarean section among low risk mothers to infant morbidity in late-preterm birth. 60

PAGE 71

INTRODUCTION !The United States has experienced a shift in the epidemiology of gestational duration among singleton live births. According to Davidoff et al (2006), the distribution of live births by gestational age has shifted to the left meaning that births, on average, are occurring earlier. Subsequently, the preterm birth rate has increased by 21.0% (from 10.6% in 1990 to 12.8% in 2006) (Hamilton et al, 2007). The proportion of births that are considered late-preterm (34 to 36 completed weeks of gestation) contributed greatly to this increase, and currently comprise about 75% of all births classied as preterm (Davidoff et al, 2006; Hamilton et al, 2007; McLaurin et al, 2009). During this increase in preterm birth, there has also been a simultaneous rise in the rate of Cesarean delivery (C/S). During an 11 year period, the C/S rate in the United States increased by over 50%, and is currently at 31.8% (Hamilton, 2009). The C/S rate for Florida in 2007 was 37.2%, which was not only higher than the national average, but also the second highest C/S rate in the United States, surpassed only by New Jersey (Hamilton et al, 2009). !Some of the shift in the distribution of gestational age in the United States has been attributed to concurrent increases in obstetric interventions (Ananth, 2005; Davidoff, 2006; IOM, 2007; MacDorman, 2002). With the margin of error for gestational age dating (+ 2 weeks), it is possible that deliveries at "presumed term" could result in late-preterm infants (Engle et al, 2006; Fuchs & Wapner, 2006). At present, the American College of Obstetricians and Gynecologists (ACOG) recommend that elective C/S delivery not be performed until at least 39 weeks GA, so that with the +/two week margin of error in dating, preterm infants are less likely to result (ACOG Committee 61

PAGE 72

Opinion, 2007). Delivery of infants at presumed early term (37 to 38 weeks GA) may result in late preterm infants. This is a problem as late-preterm births have higher rates of morbidity compared to term births, and a mortality rate that is three times higher (Engle & Kominiarek, 2008; Escobar, Clark, & Greene, 2006; Escobar, Green, Hulae, Kincannon, Bischoff, Gardner, et al., 2005; Kramer, Demissie, Yang, Platt, Sauve & Liston, 2000; McLaurin, et al., 2009; Raju, Higgins, Stark, & Leveno, 2006; ShapiroMendoza, et al., 2006; Tomashek et al 2007; Tomashek, Shapiro-Mendoza, Weiss, Kotelchuck, Bareld, Evans et al, 2006; Wang, et al, 2004). Furthermore, C/S delivery alone has been demonstrated to elevate the risk for adverse maternal and infant outcomes and increased utilization of healthcare services (Allen, et al., 2003; Belizan, et al., 2007; Declercq, et al., 2007; Gilliam, 2006; Miesnik, 2007; Ophir, Strulov, Solt, Michlin, Buryanov, Bornstein, 2008). !Malloy (2009) reported that among late preterm births, low risk primary C/S delivery signicantly increased the risk for infant morbidity and mortality. This is an important nding, as C/S deliveries performed without medical indication represent an important opportunity for the medical community to modify infant birth outcomes. While there has been a signicant amount of research on the adverse infant health outcomes following C/S delivery, no study has yet assessed the impact of low risk C/S delivery (without medical indication) on morbidity associated with singleton late-preterm births over a one year period of follow-up. !Therefore, the purpose of this study is to determine if there is a difference in the burden of infant morbidity by mode of delivery among singleton late-preterm infants born in Florida from 1998 to 2006. The two primary research objectives are: (1) to 62

PAGE 73

investigate the potential impact of low risk C/S on singleton late-preterm infant in terms of morbidity and rehospitalization, and (2) to determine if there is variation across important subgroups (e.g. primary cesarean without indications of labor, race/ethnicity, and payer source). MATERIALS AND METHODS !Data source !The Florida linked birth certicate and hospital discharge le was utilized to compare singleton late-preterm infant morbidity and rehospitalizations by route of delivery. This linked birth certicate and discharge data le resulted from the Florida Birth Certicate (data collected and maintained by the Florida Department of Health (FDOH)) and the Florida In-patient hospital discharge data (collected and maintained by the Florida Agency for Healthcare Administration (AHCA)). These data sources were linked with deterministic-dominant and probabilistic methodologies and a match rate of 97.4% was achieved. The resultant linked data le contains information on Florida singleton live births from the period of 1998 to 2007. The linkage methodology was validated by comparing rates of maternal and infant health outcomes to published rates for those complications (submitted for publication, 2010). This unique database allows for longitudinal study of birth outcomes for both mothers and their infants. For this analysis we focused on births from the time period 1998 to 2006 only, to allow for a oneyear period of follow-up for all infants in the study. 63

PAGE 74

! Exposure denition !The exposure of interest in this analysis is low risk primary C/S, compared to vaginal birth without a prior C/S (referent group). In order to classify women as low risk for C/S, we applied an algorithm developed by the FDOH and the Florida Obstetric and Gynecology Society (FOGS) and other key stakeholders. This algorithm restricts the study population to women without complications that have been considered potential risk factors or medica l indications for a C/S delivery. This algorithm has been used previously by the FDOH in analyses of low risk primary C/S and late preterm birth (Goodman et al, Submitted for Publication, 2010) and is a revised version of the Joint Commission Specications for Early Medically-Indicated Delivery (Joint Commission, 2010). !Low Risk C/S Algorithm: !!Hypertension prepregnancy (Chronic) !!Hypertension gestational (PIH, Preeclampsia) !!Hypertension Eclampsia !!Diabetes prepregnancy (Diagnosis prior to the pregnancy) !!Gestational Diabetes (Diagnosis in this pregnancy) !!Prolonged Labor !!Moderate/Heavy meconium staining of the amniotic uid !!Fetal intolerance of labor !!Clinical chorioamnionitis !!Non-vertex presentation 64

PAGE 75

!!Fetal presentation at birth other than cephalic !!A birth weight greater than 4,500 grams !!Any of these congenital anomalies: (Anencephaly, Congenital !! diaphragmatic hernia, Meningomyelocele/Spina bida, Omphalocele, !! Cyanotic congenital heart disease, Gastroschisia). !!Modied GINDEX intensive prenatal care use category (Alexander et al, !! 1987; Kogan, et al, 1998). !While these criteria have been specically designed to identify low risk primary C/S, it has been applied to the entire study population (primary C/S and vaginal deliveries) to ensure comparability between the two groups resulting in a population of women who are similar for all factors except the expos ure of interest route of delivery. Variables from both the birth certicate and in-patient hospital discharge data were used to classify deliveries by risk status. All singleton late-preterm live births (births occurring between 34 to 36 completed weeks of gestation) that remained following the application of the low risk cesarean algorithm were classied as either a primary cesarean delivery or a vaginal delivery. Subgroup analyses were also performed by restricting the vaginal group further to include only unassisted vaginal deliveries (deliveries in which there was no report of forceps or vacuum extraction) and primary cesarean deliveries in which no indications of labor were present. !Outcome measures !In order to include infant morbidities that resulted in rehospitalizations or longer initial hospitalization, an epidemiologic method was established to determine the most frequent morbidities reported among infants. While inclusion of the most common 65

PAGE 76

morbid conditions is relatively clear-cut, there are numerous conditions with a lower rate of incidence that may warrant inclusion, but objective criteria for their selection must be established. At present, there are three strategies for morbidity selection frequently utilized by researchers: (1) select morbidities frequently reported in the literature, (2) select morbidities of interest to the researcher, or (3) select the most common morbidities that appear in the data to be analyzed based oftentimes on an arbitrary cut-off point. Depending on the purpose of the research, these methods may be appropriate. However, when there are a large number of possible morbidities for inclusion, it may be necessary to establish an objective methodology for assessment of morbidities to be included in analyses. !One methodology that has promise for selection of morbidity is the Markov Chain Monte Carlo (MCMC) method. The advantage of the MCMC methodology is that it adjusts for several sources of error in point estimation. Errors originate from the uctuation of morbidity rates from one population to another, as well as the uctuation in morbidity over time, which can adversely impacts generalizability of ndings. We used this objective epidemiologic method to determine the most frequent morbidities reported among infants following delivery. This methodology has been used in genomics research to select biomarkers for inclusion in analyses of disease etiology, where there are a large number of potentially relevant biomarkers, and a need to select models that are best supported by the data used to investigate disease processes (Zhao, Foulkes, & George, 2005). !This method uses a 3 stage hierarchical model to establish a cut-off point for selecting morbidities for inclusion in statistical analyses (Figure 3.1). To enhance 66

PAGE 77

computational efciency, we arranged morbidity data frequency by year of birth. In the rst stage of the hierarchical model, we obtained the frequency of each morbidity for each year. Since there were 27 morbidities and nine different years (1998-2006), this yielded 27 times 9 data points in the rst stage (9 data points, with one for each year of the study). The second stage involved modeling each morbidity separately (i.e. we derived the overall morbidity frequency for each morbidity by taking the average of individual yearly frequencies). This yields one frequency value per morbidity for the entire period resulting in 27 morbidity frequencies. Finally, in the third stage, an overall distribution of morbidity was created by pooling the data from stage two which yielded a single overall morbidity frequency as a summary estimate. The MCMC method simulated direct draws from hypothetical distributions generated from the three-level hierarchical model of infant morbidity. 10,000 simulations were generated, after a burnin period of 1,000 simulations. The median of the draws of the pooled proportion of morbidities generated by the 10,000 simulations was used to establish a cut-off point for morbidity selection (Figure 3.2). !Of the 27 infant morbidities included in our Markov modeling technique, 6 morbidities were selected for further epidemiologic analyses based on the median of the pooled morbidity curve (Median 29.1 per 1,000): feeding difculties, respiratory distress, perinatal infections, jaundice, hypoglycemia and transient tachypnea (Table 3.1). The WinBUGS framework (version 1.4) was used for the Markov modeling. !In addition to the morbidity outcomes established by this distribution-based method, utilization of healthcare services was also investigated by route of delivery: (1) length of stay (LOS) at delivery, the rst rehospitalization and the second 67

PAGE 78

rehospitalization (2) time till rst rehospitalization, and (3) number of rehospitalizations. The length of stay (LOS) at delivery was dened as the number of days from delivery to discharge, and for rehospitalization episodes, LOS was dened as the number of days from admission to discharge. Time till rst rehospitalization was dened as the time, in days, from an infants' birth date till date of admission at rst rehospitalization within the period of infancy. Number of rehospitalizations was dened as the number of rehospitalization episodes that occurred within the rst year of infancy following discharge from the birth hospitalization. !Statistical analyses !Prior to conducting analyses of infant health outcomes, a power calculation based on published rates of infant morbidity (respiratory distress) among late preterms was performed to assure that the sample size of this study was sufcient. In order to detect a difference with a small effect size (0.03) at 80% power, 780 births would be needed for this analysis. However, with a sample of over 60,000 late preterm births among low risk women during the study period 1998 to 2006, the sample size was sufcient for this analysis (Refer to Appendix B). !Stratied analyses were employed to assess potential confounding variables by mode of delivery. Potential confounders were also identied by a review of the available literature (e.g. maternal age, race/ethnicity, infant sex) (Kuklina, Meikle, Jamieson, Whiteman, Barled, Hillis et al 2009; Liu, et al., 2005; Shapiro-Mendoza, et al., 2006). To compare differences in morbidity outcomes between primary cesarean and vaginal deliveries for infant morbidity at initial birth hospitalization, hospitalization within the neonatal period (rst 28 days), and hospitalizations within the rst year of life, 68

PAGE 79

Generalized Estimating Equations (GEE) were performed, and adjusted relative risks and corresponding 95% condence intervals were calculated. It was necessary to use the GEE methodology (Hanley, Negassa, Edwardes, & Forrester, 2003; Zeger & Liang, 1986) as the unit of analysis in this study is the individual and not the rehospitalization event. Since some infants had more than one re-hospitalization, these events would be strongly correlated for a specic infant. As repeated events experienced by an individual are not independent, analyses using repeated readmission events were adjusted for intraclass correlation. Another source of intraclass correlation is siblings many women gave birth more than once during the nine-year study period, and the resulting infants would share similar intra-uterine and environmental experience that would result in common outcomes. This source of correlation was also corrected for through the application of GEE modeling techniques. !Means and standard deviations were calculated for Length of Stay (LOS) at the birth hospitalization, rst rehospitalization and second rehospitalization, and compared by route of delivery by the Wilcoxon Rank Sum Test. Hospital readmission is another measure of infant morbidity (Glazener, Abdalla, Stroud, Naji, Templeton & Russell, 1995; Liu, et al., 2005; Lydon-Rochelle, Holt, Martin, & Easterling, 2000). Differences in the number of rehospitalizations within the infant's rst year of life was assessed with Poisson Regression. Poisson Regression was employed because the outcome variable, number of hospitalizations, is a non-parametric count variable. Furthermore, the number of hospitalization events for each infant are not independent observations. This statistical test is designed for non-parametric count variables, and Poisson regression allows for adjustment of correlated observations, resulting in a measure of the relative 69

PAGE 80

risk (Pedan, 2001). The difference in time till rehospitalization (days from delivery to readmission), was compared with the Kaplan-Meier estimate. Cox proportional hazards models could not be used to derive an adjusted hazard ratio for rehospitalization by route of delivery because a violation of the proportionality assumption was detected by plotting the log-negative-log of the Kaplan-Meier estimate of survival function versus the log of time. The resulting curves were not parallel. Therefore, differences in the risk of rehospitalization were examined with the log-binomial distribution of GEE, resulting in an adjusted relative risk, and corresponding 95% condence interval. All analyses were conducted with SAS version 9.2. This study was approved by the Institutional Review Board of the University of South Florida. RESULTS !During the 1998 to 2006 study period, 127,364 singleton late preterm live births occurred in Florida (Figure 3.3). The low risk C/S algorithm excluded 54,460 births (42.8%) from the population of late preterm deliveries, resulting in a sample of 72,904 low risk late preterm deliveries (C/S n = 14,264 or 19.6% and vaginal n=58,640 or 80.4%). From this sample, all deliveries with a record of prior cesarean (e.g. vaginal births after cesarean (VBAC) and repeat C/S deliveries) were removed, resulting in a nal population of 61,724 deliveries, of which, 5,012 were primary cesarean (8.1%) and 56,712 were vaginal (91.9%). !Table 3.2 displays the demographic characteristics of the study population, overall and by delivery route. Approximately one-third (31.0%) of mothers were Black, while 18.9% were Hispanic, 47.1% were White, and 3.0% were classied as Other. The rate 70

PAGE 81

of cesarean delivery was signicantly higher among White non-Hispanic mothers compared to non-Hispanic Blacks, Hispanics and Others. The majority of mothers were 25 years of age or older at the time of delivery (54.5%). Mothers in this age group also had higher rates of cesarean delivery (67.1% of primary cesarean deliveries). We also observed higher rates of C/S delivery among women with a high level of educational attainment (43.4% of primary C/S mothers had a college level of education compared to 39.1% of mothers with a vaginal delivery). Cesarean deliveries were also signicantly higher among women with no prior live births (56.6% in the C/S group compared to 39.3% in the vaginal group), and women who were obese (16.2% compared to 13.7%). Planned payer source differed signicantly by delivery route, with public and charity/ self-pay reported more commonly for vaginal deliveries than C/S deliveries. Infants of C/S delivery were more likely to be male (56.2% compared to 43.8%), and to be of low birth weight (LBW <2,500 grams). !Bivariate and multivariable associations between mode of delivery and infant morbidity at the birth hospitalization (excludes subsequent rehospitalizations) are provided in Table 3.3. The most common morbidity reported regardless of the mode of delivery was jaundice (27.7% of study population), followed by respiratory distress (15.2%). All morbidities included in the analysis (e.g. feeding difculties, respiratory distress, perinatal infections, jaundice, hypoglycemia and transient tachypnea) were more common among infants delivered by C/S. A composite measure of morbidity was constructed that pooled the six morbidity diagnoses into one variable, to examine the overall burden of infant morbidity. With this classication (composite morbidity), 56.5% of primary cesarean infants experienced at least one of the six selected morbidities 71

PAGE 82

during the birth hospitalization, compared to 35.8% among infants delivered vaginally (ARR = 1.29, 95% CI = 1.26,1.33). The burden of morbidity at the birth hospitalization is also portrayed in Figure 3.4. Cesarean deliveries were more likely to report a greater number of morbidity diagnoses compared to vaginal deliveries. With each additional morbid diagnosis, the rate among infants delivered via C/S was higher than the rate among infants delivered vaginally (p value for trend <0.0001). !Table 3.4 presents the risk of morbidity by mode of delivery at the birth hospitalization, within the neonatal period, and within the rst year of infancy, adjusting for covariates. At each of the three time periods presented, C/S deliveries had signicantly higher rates of morbidity. The risk of morbidity attributable to C/S was highest for transient tachypnea (ARR=1.93, 95% CI (1.77,2.10)) at birth hospitalization, during the neonatal period (ARR=1.86 95% CI (1.71,2.03)), and in the rst year of infancy (ARR=1.83 95%CI (1.68,1.99)) and lowest for jaundice at birth hospitalization (ARR=1.17, 95% CI (1.13,1.22)), during the neonatal period (ARR=1.13 95% CI (1.09,1.17)), and in the rst year of infancy (ARR=1.09 95%CI (1.04,1.13)). Overall, infants delivered via cesarean had approximately a 29% higher risk of morbidity (composite measure) during the birth hospitalization than infants born vaginally, 26% increased risk during the neonatal period and 24% increased risk in the rst year of infancy. Results summarized in Table 3.4 do not include adjustment for maternal BMI status (maternal BMI was added to the Florida birth certicate in 2004, and thus is not available for the entire study period). !To determine if this may have impacted our results, we conducted a sensitivity analysis restricting our analyses of morbidity to the time period 2004-2006, and 72

PAGE 83

included BMI classication (normal weight, overweight, and obese) in our multivariable analyses. Overall, the results did not differ substantially from those outlined in Table 3.4 (data not shown), however there were some notable exceptions. The risk of some morbidities following C/S delivery increased with the inclusion of obesity as a covariate in multivariable models: transient tachypnea (ARR=2.29, 95% CI (2.01,2.62) at birth hospitalization), perinatal infections (ARR=2.32, 95% CI (2.01, 2.67) at birth hospitalization). !Some notable subgroup differences were observed in risk for morbidity. Specically, for two of the morbid conditions included in this analysis (composite morbidity and jaundice), cesarean delivery among non-Hispanic blacks was protective for morbidity. For example, the risk of composite morbidity following C/S delivery was low for Blacks at the birth hospitalization (ARR = 0.80, 95% CI (0.75,0.85), and low for jaundice (ARR=0.70, 95% CI:(0.64,0.77)) at the birth hospitalization. There was some variation in risk of morbidity following C/S by payer status at the birth hospitalization as well for transient tachypnea, jaundice and hypoglycemia. For example, the risk of transient tachypnea at birth hospitalization was highest for C/S with commercial payer (ARR 1.96, 95% CI:1.75,2.19) and lower for public payer source (ARR=1.70, 95% CI: 1.48-1.96). Similar patterns in risk were observed for hypoglycemia (C/S with commercial payer source (ARR=1.66,95% CI:1.43,1.92), C/S with public payer source (ARR=1.34, 95% CI:1.13,1.59). At birth hospitalization, charity/self-pay was associated with a signicantly increased risk for jaundice regardless of route of delivery (ARR=1.10, 95% CI:1.04,1.16). 73

PAGE 84

!Signicant interactions with regard to risk for some morbidities were observed. For risk of transient tachypnea, respiratory distress, and feeding difculties among C/S deliveries, there was a signicant interaction between race/ethnicity and low birth weight (LBW). For respiratory distress, transient tachypnea and feeding difculties, the risk for morbidity was consistently higher for LBW non-Hispanic White infants delivered via C/S than LBW Non-Hispanic Black infants delivered via C/S. For non-Hispanic Black infants with LBW, the risk of respiratory distress was ARR=2.87 (95% CI:2.56,3.22), the risk of transient tachypnea was ARR:2.34 (95% CI:1.86,2.94), and for feeding difculties, the risk was ARR=3.72 (95% CI:2.80,4.94). Among LBW non-Hispanic White infants delivered C/S, the risk for respiratory distress was ARR=3.86 (95% CI: 3.56,4.17), transient tachypnea ARR=2.40 (95% CI:1.98,2.90), and for feeding difculties ARR=4.55 (95% CI:3.66,5.66). For normal weight infants delivered via C/S and low birth weight infants delivered vaginally, the risk for morbidity for respiratory distress, transient tachypnea, and feeding difculties was consistently highest among nonHispanic White infants. !The association between primary C/S and healthcare utilization was also investigated. About 14% of the overall study population required rehospitalization. The rehospitalization rate was signicantly higher for infants delivered C/S (148.6 per 1,000) compared to vaginal delivery (135.6 per 1,000) (p=0.01). At the bivariate level, the number of rehospitalizations within the rst year was higher for C/S delivered infants (1.59 (mean) +1.2 (standard deviation) versus 1.31 + 0.8, p < 0.0001). Furthermore, there were signicant differences in LOS by route of delivery, with C/S infants experiencing a longer average LOS than vaginally delivered infants at the birth 74

PAGE 85

hospitalization (5.4 days + 7.8 for C/S versus 3.0 + 3.6 for vaginal, p <0.0001), rst rehospitalization (5.3 + 5.3 for C/S versus 3.6 + 4.0 for vaginal, p <0.0001), and second rehospitalization (6.8 + 6.5 for C/S versus 4.1 + 4.4 for vaginal, p <0.0001). Infants delivered via C/S were also rehospitalized earlier than vaginally delivered infants, assessed by days from delivery till rst rehospitalization (C/S 81.3 days + 91.0 versus vaginal 84.6 days + 94.8, p = 0.02). After adjusting for covariates (smoking, infant sex, race/ethnicity, maternal age, education, birth weight, payer status), the number of rehospitalization events by delivery route remained signicantly different infants delivered C/S had a 22% greater number of rehospitalization episodes in their rst year (Adjusted Rate Ratio (ARR) = 1.22, 95% CI:1.11-1.34). After adjustment for covariates, infants delivered via C/S had a 10% greater risk for rehospitalization than infants delivered vaginally (Adjusted Risk Ratio (ARR)=1.10, 95% CI:1.03-1.18). !We further restricted our study sample to vaginal deliveries that were unassisted (no report of vacuum extraction or forceps) and to primary C/S deliveries in which no indications of labor were reported (e.g. induction, augmentation, vacuum extraction, forceps). Following this restriction, our population encompassed 53,460 (96.6%) infants delivered vaginally, and 1,814 (3.4%) infants delivered via primary C/S. We repeated our analyses of morbidity and observed that the risks associated with C/S for the majority of morbidities were somewhat reduced, but still signicantly higher than infants delivered vaginally (Table 3.5). The risk of transient tachypnea increased, while the association between C/S and feeding difculties and jaundice was no longer signicant. We also repeated our analysis of healthcare utilization. The LOS at birth hospitalization, rst rehospitalization and second rehospitalization remained signicantly higher for C/S 75

PAGE 86

delivered infants. Further, C/S deliveries without labor still had a greater number of rehospitalizations in the rst year of life (ARR=1.17, 95% CI:1.06-1.28), while the risk of rehospitalization was no longer signicant (ARR=1.11, 95% CI:0.99-1.24). DISCUSSION !Late preterm births have become an important focus of public health research as their rate is increasing (Davidoff et al, 2006; Engle et al, 2007; Malloy, 2009; McIntire et al, 2008; Raju 2008; Wang et al, 2004), and they experience a greater burden of morbidity and mortality compared to births at term (Malloy, 2009). Researchers have suggested that the increase in the late preterm birth rate is associated with increased obstetric interventions such as cesarean delivery (Barros & Velez, 2006; Bettegowda Dias, Davidoff, Damus, Callaghan & Pertrini, 2008; Melamed, Klinger, TenenbaumGavish, Herscovici, Linder, Hod et al, 2009). From a policy perspective, it is important to understand the impact of the association between C/S and late preterm birth in terms of morbidity and utilization of healthcare services, as C/S deliveries performed without medical indication represent an important opportunity to modify infant birth outcomes. Thus, we have focused our research on a low risk population to determine the contribution of C/S to infant morbidity among infants delivered late preterm. !A large body of evidence has been established for the increased morbidity experienced by late preterm infants compared to term infants (Burgos et al, 2008; McIntire et al, 2008; McLaurin et al, 2009; Wang et al, 2004). However the impact of C/S on late preterm infant morbidity has not been widely investigated. In our analysis, we applied a low risk algorithm to both C/S and vaginal late preterm deliveries and used 76

PAGE 87

the resulting population of vaginally delivered late preterm infants for all comparisons with late preterm infants delivered via C/S. Therefore, the groups were similar with one exception route of delivery. !The most common morbidity reported was jaundice (27.7% of study population), followed by respiratory distress (15.2%). The rate and types of morbidity for late preterm infants in our study population were within the range of those reported by other studies, a somewhat surprising nding given the strict low risk classication used for our analyses (McLaurin et al, 2009; Melamed et al, 2009; Wang et al, 2004). After controlling for several important covariates (e.g. maternal age, parity, infant sex, birth weight, payer status, education, race/ethnicity) we observed a signicant increase in the overall risk of morbidity during the birth hospitalization, neonatal period, and the rst year postpartum for low risk late preterm infants delivered via C/S compared to those that were delivered vaginally. Other researchers have also conrmed an independent association between late preterm infant morbidity and cesarean delivery (Malloy, 2009). !The rate of rehospitalization among late preterm infants (14%) was similar to that reported by other studies, which ranged from 4.8% to 15.2% (McLaurin et al, 2007; Shapiro-Mendoza et al, 2006; Underwood et al, 2007). Furthermore, we observed that infants delivered via C/S had a signicantly higher number of rehospitalization events during their rst year of life (1.59 versus 1.31, or 22% more hospitalizations), and the time till rehospitalization differed signicantly, with C/S resulting in earlier rst hospitalization. Infants delivered via C/S also had a 10% greater risk for rehospitalization, although this risk was no longer signicant once the sample of C/S was restricted to C/S without labor. Researchers have examined the impact of late preterm 77

PAGE 88

delivery on LOS at the birth hospitalization. For example, McLaurin et al (2009) observed an increased risk of late discharge for late preterm infants, although the contribution of C/S delivery to LOS was not assessed. Tita et al (2009), demonstrated an increased LOS, following C/S. We observed that the LOS at birth hospitalization, and subsequent rehospitalizations was higher for infants delivered C/S. !The intention-to-treat model advocated by the National Institutes of Health (NIH) groups C/S delivery with indications of an attempt at labor with vaginal deliveries, resulting in a category of "planned vaginal delivery", which distinguishes between planned and unplanned C/S (de Luca et al, 2009; Declercq et al, 2007, NIH, 2006). To remain consistent with previous research conducted by the FDOH on late preterm infant morbidity and C/S, we used a low risk algorithm proposed by the FDOH and the Florida Obstetrics and Gynecologic Society (FOGS the state chapter of ACOG) instead of the Intention-to-Treat algorithm. For the purpose of assessing differences in morbidity and healthcare utilization of services, we were interested in comparing specic delivery routes. Grouping C/S into a planned vaginal group likely results in more conservative estimates in morbidity by route of delivery as signicant differences would be more difcult to detect with C/S deliveries present in both groups, although any observed differences would be noteworthy given this classication strategy (Macdorman, Declercq, Menacker & Malloy, 2008). With the low indicated risk late preterm status of our study population, presumably most deliveries were planned vaginal. To explore the impact of C/S delivery without indications of labor on late preterm infant morbidity, we further restricted our study population to primary C/S deliveries without labor (according to birth certicate and in-patient discharge records), and vaginal deliveries without forceps and/or 78

PAGE 89

vacuum extraction. Our revised rate of primary C/S deliveries without labor (3.4%) was similar to that reported among other studies (de Luca et al, 2008). For the majority of morbidities investigated in this study, we observed an increased risk with primary C/S with and without indications of labor. The risk for respiratory distress, perinatal infections, hypoglycemia and transient tachypnea remained higher among infants delivered via C/S regardless of reported labor attempt. Strengths and limitations !One of the major strengths of this study is our use of population-based data to investigate the association between low risk C/S and late preterm infant morbidity and healthcare utilization. We used a linked data le that contained the birth certicate and administrative data (in-patient hospital discharge) for 97.4% of the births that occurred in Florida from 1998 to 2007. This provided multi-year data, which allowed us to follow infants from birth through infancy. Further, using linked birth certicate and discharge data improved our ability to correctly identify maternal conditions used in our low risk algorithm, as linked data of this type has been previously reported to have improved accuracy (Kahn et al, 2009; Lydon-Rochelle et al, 2005; Shapiro-Mendoza et al, 2008). Another important strength of our study was the use of GEE for multivariate modeling to adjust for intraclass correlation. As infants are rehospitalized more than once (correlated events), and may have one or more siblings in our data (shared maternal inuence), it is important to reduce the impact of error in estimation introduced by these sources of correlation. GEE is a robust method for handling clustered data of this type (Hanley, Negassa, Edwardes, & Forrester, 2003; Zeger & Liang, 1986). 79

PAGE 90

!There are several limitations of this current study that warrant discussion. First, our analysis included only birth certicate and hospital in-patient discharge information. As such, we have no data on other healthcare encounters, such as outpatient visits, and as a result, our analyses are only focused on those conditions that resulted in the hospitalization of infants. While the use of population-based data sources is frequently an asset in terms of sample size and generalizability of results, there are potential weaknesses resulting from such data. With any population-based data, there is the possibility of misclassication, resulting in errors in case denition and risk estimation (Shapiro-Mendoza, Tomashek, Kotelchuck, Bareld, Nannini, Weiss et al, 2008). The birth certicate and in-patient hospital discharge data is the result of healthcare provider management of patient's medical chart, as well as the coding clerk's interpretation, prioritization, and submission of information from medical charts. Healthcare providers may not accurately record medical conditions and procedures, while different coding practices may result in data reporting errors (Northam, Polancich, and Restrepo, 2003; Shapiro-Mendoza et al, 2008). We previously conducted a study of the accuracy of cesarean delivery indications for late preterm births in Florida from 2006-2007, using maternal medical charts from 16 hospitals in Florida (n=1,249) and found greater accuracy for the linked birth certicate and hospital discharge data compared to either data source alone (Florida Late Preterm Cesarean Delivery Investigation manuscript in submission). !Conclusion !We chose to restrict our analyses to a low risk population for our investigation of the contribution of cesarean delivery to increased late preterm infant morbidity so that we 80

PAGE 91

could explore the implications of late preterm C/S under optimal conditions (healthy mothers with few pregnancy and/or labor complications) and with a comparison group that was similar to the case group with one exception route of delivery. With recent evidence of the association between cesarean delivery and the increase in late preterm births (Goodman, et al, Submitted for Publication, 2010), the implications for increased morbidity and healthcare expenditures must be explored. In our investigation, we observed that C/S increases the risk of several morbid conditions among infants born late preterm. Further, with an increased number of rehospitalization events, and longer LOS at rehospitalization, we caution that C/S likely contributes to greater healthcare expenditures. Our ndings contribute to the growing body of evidence regarding the consequences of cesarean delivery without medical indication. Further research regarding clinical decision-making for deliveries at late preterm is warranted in order to understand the connection between cesarean delivery and the increase in late preterm delivery. At present, the policy of the American College of Obstetricians and Gynecologists (ACOG) states that elective C/S (dened by ACOG as C/S by maternal request) should not be performed earlier than 39 weeks (ACOG Committee Opinion, 2007), which given a + 3 week error in gestational age dating based on the last menstrual period, may result in infants delivered late preterm (Fuchs et al, 2008). 81

PAGE 92

Table 3.1 ICD9 Codes for Selected Infant Morbidities Infant Morbidity Description ICD9 Code Feeding Difculties Disorder of stomach function and feeding problems of newborn and infant, symptoms involving the digestive system 536.2, 783.3, 779.3x, 787.0x Respiratory Conditions Any respiratory condition or distress (includes asphyxia, hypoxia, aspiration, and respiratory symptoms) 768.5,768.6, 768.7, 768.9, 769, 770.x, Perinatal Infections Infections specic to perinatal period 771.x Jaundice Perinatal jaundice (all types) 774.x Hypoglycemia Neonatal hypoglycemia (excludes mothers with diabetes mellitus (775.0)) 775.6 Transient Tachypnea Transitory Tachypnea of newborn 770.6 82

PAGE 93

Table 3.2 Selected Socio-Demographic Characteristics of Low Risk Late-Preterm Deliveries by Delivery Method, Florida 1998 to 2006. Table 3.2 Selected Socio-Demographic Characteristics of Low Risk Late-Preterm Deliveries by Delivery Method, Florida 1998 to 2006. Table 3.2 Selected Socio-Demographic Characteristics of Low Risk Late-Preterm Deliveries by Delivery Method, Florida 1998 to 2006. Table 3.2 Selected Socio-Demographic Characteristics of Low Risk Late-Preterm Deliveries by Delivery Method, Florida 1998 to 2006. Table 3.2 Selected Socio-Demographic Characteristics of Low Risk Late-Preterm Deliveries by Delivery Method, Florida 1998 to 2006. All Deliveries (N=61,724) Vaginal (no prior C/S) (N=56,712) All Primary C/S (N=5,012) P Value Characteristic N (%) N (%) N (%) Maternal Race/Ethnicity Non-Hispanic Black 19,152 (31.0) 17,824 (31.4) 1,328 (26.5) <0.0001 Non-Hispanic Other 1,856 (3.0) 1,736 (3.1) 129 (2.6) 0.0131 Hispanic 11,643 (18.9) 10,578 (18.7) 1,065 (21.3) 0.0611 Non-Hispanic White 29,064 (47.1) 26,574 (46.9) 2,490 (49.7) Ref Maternal Age < 15 Years 252 (0.4) 244 (0.4) 8 (0.2) 0.0504 15-19 Years 9,790 (15.9) 9,256 (16.3) 534 (10.2) 0.0193 20-24 Years 18,058 (29.3) 16,948 (29.9) 1,110 (22.2) Ref 25-34 Years 26,616 (43.1) 24,232 (42.7) 2,384 (47.6) <0.0001 35-44 Years 6,951 (11.3) 5,984 (10.6) 967 (19.3) <0.0001 45+ Years 57 (0.1) 48 (0.1) 9 (0.2) 0.0025 Educational Attainment a < High School 36,892 (60.1) 34,360 (60.9) 2,532 (50.1) <0.0001 College 21,134 (34.4) 19,097 (33.9) 2,037 (41.0) <0.0001 College for 5+ Years 3,337 (5.4) 2,934 (5.2) 403 (8.1) Ref Parity b 0 25,025 (40.7) 22,198 (39.3) 2,827 (56.6) Ref < 5 35,828 (57.3) 33,159 (58.6) 2,095 (42.0) <0.0001 5+ 1,261 (2.1) 1,189 (2.1) 72 (1.4) <0.0001 Body Mass Index c > 30 2,716 (14.0) 2,400 (13.7) 316 (16.2) 0.0023 25-29.9 4,178 (21.5) 3,755 (21.5) 423 (21.6) 0.4195 < 25 12,552 (64.5) 11,335 (64.8) 1,217 (62.2) Ref Smoking Status Yes 6,877 (11.1) 6,358 (11.2) 519 (10.4) 0.0649 No 54,847 (88.9) 50,354 (88.8) 4,493 (89.6) Ref Payer Source Public 29,954 (48.5) 27,864 (49.1) 2,090 (41.7) <0.0001 Commercial 26,875 (4356) 24,264 (42.8) 2,611 (52.1) Ref Charity/Self Pay 4,008 (6.5) 3,762 (6.6) 246 (4.9) <0.0001 Other 887 (1.4) 822 (1.5) 65 (1.3) 0.0178 Infant Sex Male 33,113 (53.7) 30,297 (53.4) 2,816 (56.2) 0.0002 Female 28,611 (46.4) 26,415 (46.6) 2,196 (43.8) Ref Infant Birth weight < 2,500 grams 13,684 (22.2) 11,969 (21.1) 1,715 (34.2) <0.0001 2,500-3,999 grams 46,987 (76.1) 43,817 (77.3) 3,170 (63.3) Ref 4,000-4,499 grams 1,051 (1.7) 924 (1.6) 127 (2.5) <0.0001 a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) 83

PAGE 94

Table 3.3 Reported Infant Morbidity at Initial Hospitalization (Delivery) among Low Risk Late Preterm Births by Mode of Delivery (Vaginal as Referent), 1998-2006, Florida Table 3.3 Reported Infant Morbidity at Initial Hospitalization (Delivery) among Low Risk Late Preterm Births by Mode of Delivery (Vaginal as Referent), 1998-2006, Florida Table 3.3 Reported Infant Morbidity at Initial Hospitalization (Delivery) among Low Risk Late Preterm Births by Mode of Delivery (Vaginal as Referent), 1998-2006, Florida Table 3.3 Reported Infant Morbidity at Initial Hospitalization (Delivery) among Low Risk Late Preterm Births by Mode of Delivery (Vaginal as Referent), 1998-2006, Florida Table 3.3 Reported Infant Morbidity at Initial Hospitalization (Delivery) among Low Risk Late Preterm Births by Mode of Delivery (Vaginal as Referent), 1998-2006, Florida Table 3.3 Reported Infant Morbidity at Initial Hospitalization (Delivery) among Low Risk Late Preterm Births by Mode of Delivery (Vaginal as Referent), 1998-2006, Florida Vaginal (N=56,712) Primary C/S (N=5,012) RR ARR* 95% CI* MORBIDITY N,% N,% " Composite Morbidity 20,320 (35.8) 2,831 (56.5) 1.58 1.29 (1.26,1.33) Jaundice 15,225 (26.9) 1,893 (37.8) 1.41 1.17 (1.13,1.22) Respiratory Distress 7,785 (13.7) 1,601 (31.9) 2.33 1.85 (1.76,1.94) Transient Tachypnea 3,124 (5.5) 610 (12.2) 2.21 1.93 (1.77,2.10) Perinatal Infections 2,511 (4.4) 516 (10.3) 2.33 1.87 (1.71,2.05) Feeding Difculties 1,850 (3.3) 302 (6.0) 1.85 1.41 (1.25,1.59) Hypoglycemia 2,314 (4.1) 353 (7.0) 1.73 1.51 (1.36,1.69) *Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight 84

PAGE 95

Table 3.4 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Low Risk for Cesarean Late Preterm Births (Vaginal Delivery as Referent), 1998-2006, Florida Table 3.4 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Low Risk for Cesarean Late Preterm Births (Vaginal Delivery as Referent), 1998-2006, Florida Table 3.4 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Low Risk for Cesarean Late Preterm Births (Vaginal Delivery as Referent), 1998-2006, Florida Table 3.4 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Low Risk for Cesarean Late Preterm Births (Vaginal Delivery as Referent), 1998-2006, Florida Table 3.4 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Low Risk for Cesarean Late Preterm Births (Vaginal Delivery as Referent), 1998-2006, Florida Table 3.4 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Low Risk for Cesarean Late Preterm Births (Vaginal Delivery as Referent), 1998-2006, Florida Table 3.4 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Low Risk for Cesarean Late Preterm Births (Vaginal Delivery as Referent), 1998-2006, Florida Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Birth Hospitalization Birth Hospitalization Hospitalizations in Neonatal Period Hospitalizations in Neonatal Period Hospitalizations in First Year of Infancy Hospitalizations in First Year of Infancy ARR* 95% CI ARR* 95% CI ARR* 95% CI MORBIDITY " " " Composite Morbidity 1.29 (1.26,1.33) 1.25 (1.22,1.29) 1.24 (1.20,1.28) Jaundice 1.17 (1.13,1.22) 1.13 (1.09,1.17) 1.09 (1.04,1.13) Respiratory Distress 1.83 (1.76,1.94) 1.83 (1.74,1.91) 1.78 (1.70,1.87) Transient Tachypnea 1.89 (1.77,2.10) 1.86 (1.71,2.03) 1.83 (1.68,1.99) Perinatal Infections 1.87 (1.71,2.05) 1.75 (1.61,1.91) 1.70 (1.56,1.86) Feeding Difculties 1.40 (1.25,1.59) 1.45 (1.29,1.62) 1.40 (1.26,1.56) Hypoglycemia 1.51 (1.36,1.69) 1.51 (1.35,1.68) 1.45 (1.30,1.62) *ARR=Adjusted Relative Risk. Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *ARR=Adjusted Relative Risk. Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *ARR=Adjusted Relative Risk. Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *ARR=Adjusted Relative Risk. Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *ARR=Adjusted Relative Risk. Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *ARR=Adjusted Relative Risk. Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight *ARR=Adjusted Relative Risk. Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight 85

PAGE 96

Table 3.5 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births Without Indications of Labor (Vaginal as Referent), 1998-2006, Florida Table 3.5 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births Without Indications of Labor (Vaginal as Referent), 1998-2006, Florida Table 3.5 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births Without Indications of Labor (Vaginal as Referent), 1998-2006, Florida Table 3.5 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births Without Indications of Labor (Vaginal as Referent), 1998-2006, Florida Table 3.5 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births Without Indications of Labor (Vaginal as Referent), 1998-2006, Florida Table 3.5 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births Without Indications of Labor (Vaginal as Referent), 1998-2006, Florida Table 3.5 Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births Without Indications of Labor (Vaginal as Referent), 1998-2006, Florida Primary C/S Delivery without Indications of Labor Primary C/S Delivery without Indications of Labor Primary C/S Delivery without Indications of Labor Primary C/S Delivery without Indications of Labor Primary C/S Delivery without Indications of Labor Primary C/S Delivery without Indications of Labor Primary C/S Delivery without Indications of Labor Birth Hospitalization Birth Hospitalization Hospitalizations in Neonatal Period Hospitalizations in Neonatal Period Hospitalizations in First Year of Infancy Hospitalizations in First Year of Infancy ARR* 95% CI ARR* 95% CI ARR* 95% CI MORBIDITY " " " Composite Morbidity 1.23 (1.17,1.29) 1.20 (1.14,1.26) 1.16 (1.10,1.22) Jaundice 1.07 (1.00,1.15) 1.04 (0.97,1.11) 0.99 (0.92,1.06) Respiratory Distress 1.72 (1.59,1.86) 1.70 (1.57,1.84) 1.64 (1.51,1.78) Transient Tachypnea 2.03 (1.73,2.31) 1.97 (1.73,2.24) 1.91 (1.67,2.17) Perinatal Infections 1.61 (1.37,1.90) 1.56 (1.34,1.82) 1.50 (1.28,1.74) Feeding Difculties 1.24 (1.00,1.54) 1.23 (1.00,1.51) 1.16 (0.95,1.41) Hypoglycemia 1.42 (1.18,1.71) 1.45 (1.21,1.75) 1.39 (1.15,1.67) Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight, Adjusted Relative Risk Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight, Adjusted Relative Risk Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight, Adjusted Relative Risk Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight, Adjusted Relative Risk Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight, Adjusted Relative Risk Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight, Adjusted Relative Risk Analyses adjusted for: smoking, infant sex, race/ethnicity, maternal age, parity, education, payer type, and birth weight, Adjusted Relative Risk 86

PAGE 97

Figure 3.1 Markov Hierarchical Model for Infant Morbidity Selection 87

PAGE 98

Figure 3.2 Markov Pooled Infant Morbidity Curve 88

PAGE 99

Figure 3.3 Description of Study Sample After Application of Low Risk C/S Algorithm Low Documented Risk C/S Algorithm (Applied to Cesarean and Vaginal Deliveries) Congenital Defects: Hernia (n=60,0.1%) Heart (n=4,332, 3.4%) Anencephaly (n=20, 0.0%) Spina Bifida (n=78, 0.1%) Gastroschisia/Omphalocele (n=302,0.2%) Diabetes Type I, II (n=5,127, 4.0%) Gestational Diabetes (n=8,279,6.5%) Chronic Hypertension (n=10,631,8.4%) Gestational Hypertension (n=20,995,16.5%) Eclampsia (n=1,309, 1.0%) Fetal Distress (n=4,170, 3.3%) Meconium Staining (n=4,079, 3.2%) Breech/Malpresentation (n=11,657, 9.2%) Chorioamnionitis (n=2,881, 2.3%) Prolonged Labor (n=6,220, 4.9%) Birthweight > 4,500 grams (n=454, 0.4%) Intensive Prenatal Care (n=6,723, 5.3%) Initial Study Sample Late Preterm Initial Study Sample Late Preterm Births N=127,364 Births N=127,364 Deliveries with Documented Risk N=54,460 (42.8%) All Cesarean Deliveries N=14,264 (19.6%) All Vaginal Deliveries N=58,640 (80.4%) Primary Cesarean N=5,012 (8.1%) Vaginal (no prior C/S) N=56,712 (91.9%) Primary C/S w/o Labor N=1,814 (3.4%) Unassisted Vaginal N=53,460 (96.6%) Deliveries with Low Documented Risk N=72,904 (57.2%) 89

PAGE 100

Figure 3.4 Difference in Reported Numbers of Morbidity at Birth Hospitalization Among Late Preterm Infants by Mode of Delivery, Florida 1998-2006. 7'8$"'(39*($%*+1 :+$0'+13/25 ; <=>? @?>; ?A>? =;>; ;3B.+C$#$)1 <3B.+C$#$)1 A3B.+C$#$)$*@D3B.+C$#$)$*90

PAGE 101

CHAPTER FOUR MANUSCRIPT THREE TITLE !The contribution of low indicated risk primary cesarean delivery to maternal morbidity following singleton late-preterm birth. 91

PAGE 102

INTRODUCTION !Over the past two decades, a dramatic shift in the epidemiology of gestational age at delivery for singleton infants in the United States has been observed. Davidoff et al (2006), report that the distribution of singleton live births by gestational age has shifted to the left by one week meaning that births are occurring a week earlier than in previous decades. During this same time period, the United States has also observed a 21% increase in the rate of prematurity (10.6% in 1990 to 12.8% in 2006) (Hamilton et al, 2007). The largest proportion of premature births (~ 75%), are those that occur between 34 to 36 completed weeks of gestation referred to as late preterm births (McLaurin et al, 2009). Any increase in the rate of late preterm births likely translates into an increase in prematurity overall, and thus late preterm births are an important focus of public health efforts to reduce infant morbidity and mortality (Davidoff et al, 2006; Hamilton et al, 2007; McLaurin et al, 2009). !According to Davidoff et al (2006) and others, the epidemiologic shift in gestational age at birth in the United States is likely related to the concurrent increase in obstetric interventions such as labor induction and C/S (Ananth, 2005; Davidoff, 2006; IOM, 2007; MacDorman, 2002). Over the past decade, the United States has experienced a dramatic increase in the rate of cesarean (C/S) delivery. Since 1996, the rate of overall C/S has increased over 40% (Declercq et al, 2007). Currently, 31.8% of births in the United States are via C/S (Hamilton, 2009). In Florida, the C/S rate is much higher (37.2%), and at present, is surpassed only by the state of New Jersey (Hamilton et al, 2009). 92

PAGE 103

!The increase in C/S delivery and late preterm births is a concern for infant health outcomes as C/S and prematurity have both been independently associated with adverse infant health outcomes (e.g. respiratory distress, jaundice, mortality). Another important piece of this puzzle, however, is the potential increase in maternal morbidity and rehospitalization following C/S delivery. Over the past decade, several researchers have associated C/S with an increased risk for maternal morbidity and rehospitalization (Declercq et al, 2007; Liu et al, 2005; Lydon-Rochelle et al, 2000; Ophir et al, 2008; Pallasmaa, Ekblad & Gissler, 2008). Specically, Declercq et al (2007) demonstrated that mothers with a planned primary C/S delivery were 2.3 times more likely to be rehospitalized in the rst 30 days postpartum (95% CI:1.74-2.9), while Liu et al (2005) and Ophir et al (2008) reported a similar association between C/S and maternal rehospitalization. !In order to understand the impact of late preterm birth C/S delivery on the healthcare system, it is important to investigate outcomes among mothers and their infants. In this study, we focused on maternal morbidity and rehospitalization. While previous studies have identied C/S as a risk factor for maternal rehospitalization and other specic morbidities, information on maternal outcomes by route of delivery following singleton late preterm birth is scarce. Therefore, the purpose of this study was to determine if there is a difference in the burden of maternal morbidity by mode of delivery for women who gave birth to singleton late-preterm births in Florida from 1998 to 2006. Our two primary research objectives were: (1) to investigate the potential impact of low indicated risk C/S on maternal morbidity and healthcare utilization, and (2) 93

PAGE 104

to determine if there is variation by important subgroups: primary C/S without indications of labor, race/ethnicity and payer source. MATERIALS AND METHODS Data source !This study utilized the Florida linked birth certicate and hospital discharge le to explore maternal morbidity and rehospitalization patterns by route of delivery. The Florida birth certicate (collected and maintained by the Florida Department of Health (FDOH)) and in-patient hospital discharge data (collected by the Florida Agency for Healthcare Administration (AHCA)) were linked with deterministic and probabilistic methodologies, achieving a 97.4% match rate. The linkage was validated by comparing maternal and infant morbidity rates to rates that have been previously published (methodology submitted for publication, 2010). The resulting linked database contains information on mothers who delivered a singleton live birth from 1998 to 2007. The primary advantage of this data linkage is the ability to examine longitudinal outcomes such as maternal rehospitalizations. In this study, we focused on deliveries that occurred from 1998 to 2006, to allow for a one-year period of follow-up. !Exposure denition !For this study, the exposure of interest was low indicated risk primary C/S delivery, which was compared to vaginal delivery (referent category). An algorithm was developed by the FDOH and the Florida Obstetric and Gynecologic Society (FOGS) for use in classifying women as low indicated (medical) risk for C/S delivery. The low risk C/S algorithm restricts the study population by removing potential risk factors or 94

PAGE 105

other medical indications for a C/S delivery. This algorithm has been used previously by the FDOH to investigate the association between low risk p rimary C/S and late preterm birth, and has also been used in a study of infant morbidity and rehospitalization by delivery route for singleton preterm deliveries (Refer to Manuscript Two; Goodman, Sappeneld, Mahan and Kogan, Submitted for Publication, 2010). !Low Risk C/S Algorithm: !!Hypertension prepregnancy (Chronic) !!Hypertension gestational (PIH, Preeclampsia) !!Hypertension Eclampsia!!! !!Diabetes prepregnancy (Diagnosis prior to the pregnancy) !!Gestational Diabetes (Diagnosis in this pregnancy) !!Prolonged Labor !!Moderate/Heavy meconium staining of the amniotic uid !!Fetal intolerance of labor !!Clinical chorioamnionitis !!Non-vertex presentation !!Fetal presentation at birth other than cephalic !!A birth weight greater than 4,500 grams !!Any of these congenital anomalies: (Anencephaly, Congenital !! diaphragmatic hernia, Meningomyelocele/Spina bida, !! Omphalocele, Cyanotic congenital heart disease, Gastroschisia). !!Modied GINDEX intensive prenatal care use category (Alexander et al, !! 1987; Alexander & Kotelchuck, 1996). 95

PAGE 106

!The purpose of this algorithm is to identify low risk primary C/S, however it has been applied to both primary C/S and vaginal deliveries in order to ensure comparability between the two modes of delivery. The universal application of the low risk C/S algorithm resulted in a population of women who were similar with one exception the route of delivery. Variables from both the birth certicate and in-patient hospital discharge data were used to classify deliveries by risk status. Two groups were available for analyses: (1) women who delivered via primary C/S, and (2) women who delivered vaginally, without a history of C/S. Further subgroup analyses were performed. The exposure variable was further restricted to include only those primary C/S deliveries without indications of labor, and as referent, unassisted vaginal deliveries (no report of vacuum or forceps delivery). Outcome measures !Numerous maternal morbidities have been reported in the available literature for both vaginal and cesarean delivery routes (Callaghan, MacKay & Berg, 2008). Further, maternal morbidities can be common for both cesarean and vaginal routes of delivery (e.g. hemorrhage), while some morbidities are more specic to C/S (e.g. abdominal wound infection), or vaginal deliveries (e.g. episiotomy). Depending on the focus of the research, the spectrum of maternal morbidities assessed can vary, ranging from common conditions to rare conditions, and from mild to severe. Inclusion of the most common morbidities is relatively clear-cut, however many conditions have lower rates of incidence, and distinguishing between those to include in analyses may be difcult without objective criteria for selection. For example, currently there are several strategies to assess maternal morbidity: (1) examine near miss maternal morbidity, (2) 96

PAGE 107

examine morbidities frequently reported in the literature, or (3) select the most common morbidities that appear in the data to be analyzed (oftentimes based on an arbitrary cutoff point). Depending on the purpose of the research, these methods may be appropriate. A major problem with maternal morbidity research is the low rate of morbidity experienced by women, as well as inconsistent denitions of morbidity outcomes (WHO, 2009). !With a large number of possible morbidities for inclusion in analyses, it is necessary to establish an objective methodology for assessment of maternal morbidity. A recent methodology that has promise for selection of morbidity is the Markov Chain Monte Carlo (MCMC) method. The MCMC takes important statistical applications into consideration. For example, morbidity rates uctuate from one population to another, and also uctuate over time, making it difcult to generalize ndings. The MCMC methodology is an objective, powerful and robust method that adjusts for uncertainties and sources of error in calculation of point estimates. In order to compare maternal morbidities by route of delivery, we used this objective epidemiologic method to determine the most frequent morbidities reported among mothers following delivery. The MCMC has been frequently applied in health research, particularly in genomics. Frequently in genomics studies, there are a large number of potentially relevant biomarkers, and a need to select models that are best supported by the data used to investigate disease processes (Zhao et al, 2005). Genomics researchers have used this methodology to select biomarkers for inclusion in analyses of disease etiology, such as HIV. 97

PAGE 108

!We developed a 3 stage hierarchical model using the Markov Chain Monte Carlo (MCMC) method to determine an appropriate cut-off point for selecting maternal morbidities for inclusion in statistical analyses (Figure 4.1). Computational efciency of the MCMC was enhanced by arranging maternal morbidity data frequency by year of delivery. The rst stage of the hierarchical model used the frequency of each maternal morbidity for each study year (1998-2006). Only 20 maternal morbidities considered to be "shared complications" of C/S and vaginal delivery were entered into the rst stage. For the second stage of the hierarchical model, each of the 20 morbidities were modeled separately (i.e. incidence rates for each maternal morbidity were obtained by averaging them yearly, resulting in one incidence per morbidity). The third stage resulted in an overall distribution of morbidity by pooling the incidence results from stage two (the result is an overall incidence representing a summary of all maternal morbidities included in the MCMC). 10,000 simulated direct draws from the hypothetical distributions were created by the three-level hierarchical MCMC model of maternal morbidity, following a burn-in period of 1,000 simulations. The median value of the pooled proportion of the 10,000 simulations of maternal morbidities created by these draws was used as the cut-off point for inclusion of maternal morbidities (Figure 4.2). Of the 20 maternal morbidities assessed with the Markov technique, 5 morbidities were selected for further epidemiologic analyses based on the median of the pooled morbidity curve (median = 25.5 per 10,000 women): postpartum hemorrhage, bladder repair, venous complications, unspecied febrile conditions, and puerperal infection (Table 4.1). The WinBUGS framework (version 1.4) was used for the Markov modeling. 98

PAGE 109

!Utilization of healthcare services such as rehospitalization, is another measure of maternal morbidity (Declercq et al, 2006 Liu et al, 2005; Lydon-Rochelle et al, 2000, and Ophir et al, 2008). To compare patterns of rehospitalization by route of delivery, three primary outcomes were assessed: (1) length of stay (LOS) (birth hospitalization, rst rehospitalization and second rehospitalization) (2) time till rst rehospitalization, and (3) number of rehospitalizations. LOS at the birth hospitalization was dened as the days from delivery to discharge, and LOS at subsequent rehospitalizations was dened as days from admission to discharge. The time till rst rehospitalization was dened as the time, in days, from the infant date of delivery to date of admission at the rst maternal rehospitalization event within the rst year postpartum. The number of rehospitalizations was dened as the number of maternal rehospitalization episodes during the rst year postpartum. Statistical analyses !Prior to conducting analyses of maternal health outcomes following late preterm delivery, a power calculation was performed based on published rates of maternal postpartum rehospitalization. In order to detect a signicant difference with a small effect size (0.03) at 80% power, 5,965 births would be required to compare maternal rehospitalization by delivery route. With a sample of over 60,000 low indicated risk late preterm births during the 1998 to 2006 study period, the sample size was sufcient for this analysis (full details in Appendix B). !A review of the literature and stratied analyses were performed to identify potential confounders for maternal morbidity by route of delivery. Generalized Estimating Equations (GEE) were used to compare differences in maternal morbidity by 99

PAGE 110

route of delivery at the initial birth hospitalization, hospitalization during the neonatal period (rst 28 days), and hospitalizations within the rst year postpartum. GEE was employed to adjust for intraclass correlation, and adjusted relative risks and corresponding 95% condence intervals were calculated. The GEE methodology was necessary, as the unit of analysis for this study was the individual, and not the rehospitalizaiton event (Hanley, Negassa, Edwardes, & Forrester, 2003; Zeger & Liang, 1986). Several mothers were rehospitalized more than once, and each rehospitalization event within the subject (mother), is correlated with the other rehospitalization events experienced by that subject. As repeated hospitalization events experienced by mothers were not independent, analyses using repeated readmission events were adjusted for this source of intraclass correlation. Another source of intraclass correlation is repeated delivery events within the nine-year period of this study. As many women gave birth more than once during the study, each delivery episode may share common characteristics such as pregnancy complications and environmental experiences that may result in similar outcomes. The application of GEE modeling techniques also allowed for this source of intraclass correlation to be controlled for in analyses. !The average LOS at the birth hospitalization and subsequent rehospitalizations by route of delivery was compared by the Wilcoxon Rank Sum Test. Differences in the number of maternal rehospitalization events in the year postpartum were compared with Poisson regression modeling. This statistical method was necessary as the outcome variable number of rehospitalizations is a non-parametric count variable. Further, the number of rehospitalizations for each individual are not independent observations. Poisson regression allows for an adjustment of correlated observations such as 100

PAGE 111

rehospitalization events (Pedan, 2001). Time till rst rehospitalization event was calculated as days from delivery of an infant to rst readmission episode. For the bivariate comparison in time till rst rehospitalization, the Kaplan-Meyer test was performed. The rate of rehospitalization was dened as a maternal hospitalization event in the rst year postpartum, per 1,000 deliveries. Cox proportional hazards regression models were constructed with the Robust Sandwich Estimator techniques to estimate adjusted hazard ratios (AHR) and corresponding 95% condence intervals for risk of rehospitalization by route of delivery. The Robust Sandwich Estimator technique was necessary to reduce potential estimation errors from intraclass correlation within the survival model. We tested the proportionality assumption of Cox proportional hazards regression by plotting the log-negative-log of the Kaplan-Meier estimates of survival function versus the log of time. The curves resulting from this test were parallel, and therefore we were not in violation of the proportionality assumption. !All analyses were conducted with SAS version 9.2. This study was approved by the Institutional Review Board of the University of South Florida. RESULTS !From 1998 to 2006, Florida reported 127,364 singleton late preterm live births (Figure 4.3). Application of the low indicated risk algorithm reduced the population of late preterm deliveries by 42.8% (n=54,460 deliveries). This resulted in a sample of 72,904 late preterm deliveries, of which, 14,264 (19.6%) were C/S and 58,640 (80.4%) were vaginal. Repeat C/S and vaginal births after cesarean (VBAC's) were also 101

PAGE 112

removed from the study population, resulting in a nal population of 61,724 deliveries, of which, 5,012 were primary cesarean (8.1%) and 56,712 were vaginal (91.9%). !The socio-demographic characteristics of the study population by route of delivery are provided in Table 4.2. Over half of the mothers were White (47.1%), followed by Black (31.0%), Hispanic (18.9%) and Other (3.0%). Rates of C/S delivery were signicantly higher for White mothers than for any other group. Most mothers were 25 years of age or older at the time of delivery (54.5%). The C/S rate among women with higher rates of educational attainment was also higher than that of women without a college education. Notably, the proportion of C/S deliveries among women with no prior live births was also higher (56.6% in the C/S group compared to 39.3% in the vaginal group). Planned payer source differed signicantly by delivery route, with public, charity/self-pay and other reported more commonly for vaginal deliveries than C/ S deliveries. Infants of C/S delivery were more likely to be low birth weight and male. !Table 4.3 compares differences in maternal morbidity by route of delivery. Maternal morbidities were classied as either a cesarean complication, a vaginal complication, or a shared complication (complications that may result from either route of delivery). Note that these classications are in reality, not completely exclusive to a specic delivery route, as it is possible that C/S deliveries that were attempted vaginal deliveries may report vaginal complications as well. For example, some cases of vaginal laceration and perineal lacerations were noted among C/S deliveries although the incidence was much lower for C/S deliveries compared to vaginal deliveries (e.g. perineal laceration reported by 2.1% of vaginal deliveries, and 0.2% of C/S deliveries). With the low risk classication of the study population, the overall rate of maternal morbidity was 102

PAGE 113

low, with no conditions reaching more than 3% incidence. Of the shared complications of C/S and vaginal delivery, the most commonly reported morbidity was postpartum hemorrhage (1.8% vaginal and 1.4% C/S, followed by bladder repair (1.0% vaginal and 0.1% C/S), and venous complications (0.06% vaginal and 0.5% C/S). !Table 4.4 presents the bivariate and multivariable associations between route of delivery and maternal morbidity at the birth hospitalization (excludes subsequent rehospitalizations). Three of the selected morbidities were more commonly reported among vaginal deliveries: postpartum hemorrhage, bladder repair and venous complications, while unspecied febrile conditions and puerperal infections were more commonly reported with C/S delivery. After controlling for potential confounders: smoking, race/ethnicity, maternal age, parity, education, and payer type, C/S delivery was protective for postpartum hemorrhage (Adjusted Relative Risk (ARR)=0.75, 95% CI 0.59-0.96). C/S delivery also appeared protective for venous complications (ARR=0.68, 95% CI 0.45,1.02), however the association did not reach statistical signicance. C/S delivery increased the risk for unspecied febrile conditions (ARR=3.84, 95% CI 2.89-5.10) and puerperal infection (ARR=8.09, 95% CI 5.33-12.27). !The risk of morbidity by route of delivery at the birth hospitalization, neonatal period and in the rst year postpartum, adjusted for potential confounders is presented in Table 4.5. At each of the three time periods, C/S delivery was protective for postpartum hemorrhage (ARR=0.75, 95% CI: 0.59-0.96) at birth hospitalization, during the neonatal period (ARR=0.63 95% CI: 0.48-0.82), and in the rst year postpartum (ARR=0.69 95%CI: 0.54-0.87). Also, for each of these periods of time, C/S appeared protective for venous complications, but the association did not reach statistical signicance. C/S 103

PAGE 114

remained a signicant risk factor for unspecied febrile conditions and puerperal infection for the initial birth hospitalization, the neonatal period and the rst year postpartum. The results presented in Table 4.5 do not include adjustment for maternal BMI status (maternal BMI was not present on the Florida Birth Certicate until 2004). !As obesity has been demonstrated to increase the risk for several adverse maternal outcomes such as preeclampsia, preterm birth, emergency C/S, C/S wound complications, and postpartum hemorrhage, its inclusion in analyses is warranted (Bhattacharya, Campbell, Liston, Bhattacharya, 2007; Satpathy, Fleming, Frey, Barsoom, Satpathy, and Khandalavala, 2008). To determine if inclusion of obesity measures may have impacted the results reported thus far, we conducted a sensitivity analysis restricting our study population to births from 2004-2006, and included the following categories based on BMI classication (normal weight (BMI < 25), overweight (BMI 25-29.9) and obese (BMI 30+)) in our multivariable analyses. Overall, the results did not differ substantially from those outlined in Table 4.5 (data not shown), with one exception. At the birth hospitalization, C/S was no longer signicantly protective for postpartum hemorrhage (ARR=0.79,95%CI:0.48,1.02), while C/S remained protective for postpartum hemorrhage during the neonatal time period (ARR=0.61,95% CI:0.40,0.92) as well as the rst year postpartum (ARR=0.66,95%CI:0.46,0.95). The association between C/S and venous complications remained insignicant, and the sample size for bladder repair was insufcient for multivariate analyses. For each of the three time periods, C/S was still signicantly associated with an increased risk for unspecied febrile conditions (e.g. ARR=3.48,95% CI:2.08-5.83 at birth hospitalization) and puerperal infection (e.g. ARR=5.47,95% CI:2.84,10.54 at birth hospitalization). 104

PAGE 115

!We observed some signicant subgroup differences in risk for morbidity. While C/S delivery was not signicantly related to venous complications, we observed that Black mothers (White mothers as referent) had lower risk for venous complications, at the birth hospitalization (ARR=0.42, 95% CI: 0.31-0.56), during the neonatal period (ARR=0.40, 95% CI: 0.30-0.53) and during the rst year postpartum (ARR=0.40, 95% CI: 0.30-0.52). The risk of puerperal infection however, was higher among Black mothers who delivered C/S (White mothers as referent) throughout the follow-up period, but was particularly high during the birth hospitalization (ARR=2.85, 95% CI: 1.86-4.37). The risk for puerperal infection was also consistently higher among women with a public payer type (ARR=1.66, 95% CI: 1.07-2.57 at birth hospitalization), while public payer type was protective for venous complications (ARR=0.64, 95% CI: 0.51-0.81). Due to limited samples, no signicant interactions were observed. !We investigated the association between primary C/S and utilization of healthcare services. 4.1% of the overall study population required rehospitalization (4.1% among women who delivered vaginally and 10.8% among women who delivered via C/S, p <0.0001). On average, mothers delivering via C/S have a signicantly longer LOS than mothers who delivered vaginally for the birth hospitalization as well as subsequent rehsopitalizations: birth LOS (4.4 (mean days) + 5.9 (standard deviation), p<0.0001 for C/S) compared to (2.4 + 2.4, p<0.0001 for vaginal); rst rehospitalization (3.7 + 5.1, p<0.0001 for C/S versus 3.0 + 2.9, p<0.0001 for vaginal); second rehospitalization (5.1 + 3.9, p<0.0001 for C/S versus 4.0 + 4.2, p=0.0006 for vaginal). The average length till rst rehospitalization event for mothers who delivered via C/S was 126.7 days (Std. Dev. + 117.1) versus 164.0 days (Std. Dev. + 118.0) for vaginal deliveries (p = < 0.0001). 105

PAGE 116

Differences in the time till rst rehospitalization by route of delivery were also assessed with the Kaplan-Meier estimates, with women delivering via C/S experiencing an earlier rate of rehospitalization compared to women who delivered vaginally (p<0.0001) (Figure 4.4). Also, at the bivariate level, the number of rehospitalization episodes within the rst year was higher for mothers who delivered C/S (2.0 + 2.3 versus 1.4 + 1.2, p < 0.0001). After adjusting for potential confounders (smoking, maternal race/ethnicity, maternal age, parity, education, and payer status) this observation remained, conrming that the number of rehospitalizations among women who delivered via C/S was about 36% greater than that of mothers who delivered vaginally (Adjusted Rate Ratio (ARR)=1.36, 95% CI:1.08,1.71). Also, after adjustment for potential confounders, the risk of rehospitalization was 57% higher among women delivering via C/S (Adjusted Hazard Ratio (AHR) = 1.57, 95% CI: 1.39,1.79). !To compare morbidity by route of delivery for C/S deliveries without labor, and unassisted vaginal delivery we further restricted our study population to vaginal deliveries that were unassisted (no report of vacuum extraction or forceps) and to primary C/S deliveries in which no indications of labor were reported (e.g. induction, augmentation, vacuum extraction, forceps). This further restriction resulted in a study population of 53,460 (96.6%) mothers who delivered vaginally, and 1,814 (3.4%) who delivered via primary C/S without indication of attempted labor (Table 4.6). We repeated our analyses of morbidity and observed that C/S was no longer protective for postpartum hemorrhage at the birth hospitalization (ARR=0.69, 95% CI: 0.46-1.04), but was still protective in the neonatal period (ARR=0.64, 95% CI:0.42-0.98) and the rst year postpartum (ARR=0.55, 95% CI:0.37-0.82). The risk for unspecied febrile 106

PAGE 117

conditions and puerperal infection was still higher for C/S deliveries than vaginal deliveries regardless of period of follow-up. We also repeated our analysis of rehospitalization and observed that primary C/S deliveries without labor still had a higher number of rehospitalization episodes compared to vaginal delivery (ARR=1.50,1.14,1.98), and a signicantly longer LOS for birth and subsequent rehospitalizaitons. There was also a signicant association between route of delivery and risk for rehospitalization, with women delivering via C/S hospitalized 76% more than women who delivered vaginally (AHR = 1.76, 95% CI (1.45,2.12)). DISCUSSION !Our study used a population-based cohort of women who gave birth to singleton late preterm infants from 1998 to 2006. After application of a low risk algorithm, 61,724 mothers of singleton late preterm deliveries were available for an analysis of the impact of route of delivery on maternal morbidity and rehospitalization (5,012 primary C/ S and 56,712 vaginal). Overall, the rate of maternal morbidity among our study population was very low. This was not an unexpected observation as (1) we applied a conservative low risk algorithm, and (2) rates of maternal morbidity reported by other studies have also been relatively low as well (Liu et al, 2005). !While in our study, C/S delivery was protective for postpartum hemorrhage, evidence reported by other studies differs by method of vaginal delivery. For example, Liu et al (2005) and Ophir et al (2008) reported rates for postpartum hemorrhage were lower for C/S deliveries, while Lydon-Rochelle et al (2000) reported a rate of postpartum hemorrhage among C/S deliveries as higher than that of spontaneous vaginal deliveries, 107

PAGE 118

but not higher than assisted vaginal deliveries. We observed that C/S increased the risk for both unspecied febrile conditions and puerperal infection. Puerperal infection can result in fever, however unspecied febrile conditions as classied by ICD-9 code, specically exclude infection. This exclusion has been support by prior research, as researchers have reported that women who receive epidural analgesia often experience fever without an indication of an infection (Goetzl, Rivers, Evans, Citron, Richardson, Lieberman et al, 2004). Our nding that puerperal infection was more commonly reported following a C/S delivery was also described by other studies (Declercq et al, 2007; Liu et al, 2005). Liu et al (2005) suggests that puerperal infection is likely a condition that is exacerbated by surgical procedures and/or wound complications. !Researchers have frequently identied an increased risk of rehospitalization among mothers who delivered via C/S (Declercq et al, 2007; Liu et al, 2005; LydonRochelle et al, 2000, and Ophir et al, 2008). In our study population of mothers who delivered late preterm, we also observed an increased risk of rehospitalization for C/S. Furthermore, mothers who delivered via C/S were rehospitalized earlier than mothers who delivered vaginally (for rst rehospitalization), were hospitalized longer, and had a greater number of rehospitalization episodes in the rst year postpartum. It should be noted that our rates of rehospitalization were somewhat higher than those reported by other researchers. For the mothers in our study, the rate of rehospitalization in the rst year postpartum was 37 per 1,000 for vaginal deliveries and 90 per 1,000 for C/S deliveries. In the study by Liu et al (2005), the rate of maternal rehospitalization for spontaneous vaginal deliveries was 15.3 per 1,000 and 27.0 per 1,000 among C/S deliveries. Although not directly comparable due to different classication strategy, 108

PAGE 119

Declercq et al (2007) reported a maternal rehospitalization rate for "planned vaginal deliveries" of less than 10 per 1,000, and among C/S with no labor and no complications, less than 20 per 1,000. Our higher rates for maternal rehospitalization may be due to the late preterm delivery status of our study population. While our low risk algorithm removed vaginal and C/S deliveries with severe pregnancy and delivery complications, not all possible risk factors for maternal morbidity were included in the risk algorithm. !The National Institutes of Health outlined an Intention-to-Treat Model for research involving outcomes of C/S (NIH, 2006). This model classies C/S delivery with indications of labor (e.g. induction, augmentation, forceps, vacuum extraction) as a "planned vaginal delivery", which is then grouped with vaginal deliveries (de Luca et al, 2009; Declercq et al, 2007, NIH, 2006). We did not use this classication system for several reasons. First, our desire was to remain consistent with prior research conducted by the FDOH that reported an association between primary C/S and the increase in late preterm birth. That research used an extremely conservative low medical risk algorithm for C/S, which was developed by the FDOH and the Florida Obstetrics and Gynecologic Society (FOGS the state chapter of ACOG). Secondly, we were interested in assessing differences in maternal morbidity and rehospitalization by route of delivery. Undoubtedly, grouping C/S with indications of labor with vaginal deliveries would have resulted in a more conservative estimate of morbidity by delivery route, and observed differences would have been noteworthy (Macdorman, Declercq, Menacker & Malloy, 2008). However, with our strict low risk algorithm and the late preterm delivery status of our study population, presumably most C/S deliveries would have been planned vaginal (according to information contained in the birth certicate and in-patient 109

PAGE 120

discharge data, 63.8% of the low risk C/S populations had at least one reported indication of labor). To examine the impact of C/S without labor, we further restricted our study population to primary C/S deliveries without indications of labor, and vaginal deliveries without indications that delivery was assisted (i.e. forceps and/or vacuum extraction), as this category of vaginal deliveries has been demonstrated to have increased risk for morbidity as well (Lydon-Rochelle et al, 2000). We found that our revised rate of low indicated risk primary C/S without labor (3.4%) was similar to that reported by other studies (de Luca et al, 2008). Overall, our observations in the risk of morbidity or rehospitalization by route of delivery did not change appreciably with the restriction to primary C/S without labor indications. C/S remained protective for postpartum hemorrhage for all time periods except the birth hospitalization, while C/S remained a risk factor for both unspecied febrile conditions and puerperal infection regardless of period of follow-up. Rates of rehospitalization remained signicantly increased with C/ S deliveries compared to vaginal, although the rate of maternal rehospitalization for C/S was much higher following the restriction to C/S without labor indications (90 per 1,000 versus 196 per 1,000 (following restriction)). Strengths and limitations !A major strength of this study is the use of longitudinal population-based data to investigate maternal health outcomes and healthcare utilization following low indicated risk C/S. Due to the longitudinal nature of the data, we were able to follow mothers from delivery through the rst year postpartum. Another important strength of this study was the use of the Florida linked birth certicate and in-patient hospital discharge le, which contains data on 97.4% of the singleton live births that occurred in Florida 110

PAGE 121

from 1998 to 2007. Researchers have suggested that linked data of this type is ideal for perinatal research as accuracy is improved over that of either data source used alone, resulting in greater ability to correctly identify maternal conditions and infant health outcomes used in our low risk C/S algorithm (Kahn et al, 2009; Lydon-Rochelle et al, 2005; Shapiro-Mendoza et al, 2008). Methodologically, an important strength of our study was the use of GEE methodology for multivariate modeling of morbidity. As previously discussed, GEE adjusts for the intraclass correlation introduced by our longitudinal cohort study design specically mothers contributing more than one infant, and mothers repeated episodes of rehospitalization. GEE is a robust method for reducing the impact of intraclass correlation (Hanley, Negassa, Edwardes, & Forrester, 2003; Zeger & Liang, 1986). !There are several important limitations of this study that should be discussed. While linked birth certicate and in-patient hospital discharge data allows us to more accurately identify risk factors and outcomes, it does not provide data on healthcare encounters that occur outside the hospital (e.g. outpatient visits). As a result, our ndings are restricted to conditions that were severe enough to warrant hospitalization. Also, population-based data improves generalizability of results, but is not without some shortcomings. With population-based data, we cannot discount the possibility of misclassication, which could have resulted in errors in case classication and risk estimation (Shapiro-Mendoza et al, 2008). Previously, we conducted an evaluation of the accuracy of maternal medical charts (n=1,249) following late preterm primary C/S delivery among 16 hospitals in Florida (birth years 2006-2007) and observed greater accuracy for the linked birth certicate and hospital discharge data than the birth 111

PAGE 122

certicate or discharge data alone (Florida Late Preterm Cesarean Delivery Investigation -data not shown). !Conclusion !The association between C/S and increasing rates of late preterm delivery is disconcerting. Given the adverse impact of prematurity on infant morbidity and mortality, low indicated risk C/S late preterm deliveries are an important focus of research and preventive efforts. While there has been evidence of increased maternal morbidity following C/S, there has been very little information on the impact of C/S on mothers who delivered late preterm infants. In our investigation, we observed that C/S increases the risk of some maternal morbidities (puerperal infection, unspecied febrile conditions), and may be protective for other conditions (postpartum hemorrhage). Examining specic morbidities by route of delivery may not be ideal as there is variation in the types of morbidity experienced by each delivery route (e.g perineal laceration versus abdominal wound infection). Utilization of health care services such as LOS, rehospitalization rates, and timing of rehospitalization may be more ideal outcomes for research involving C/S as they are proxy measures of all morbidities that result in longer LOS or rehospitalization events. !While we have explored the outcomes of low risk C/S at late preterm, the underlying cause of low indicated risk C/S is not fully understood. Currently, American College of Obstetricians and Gynecologists (ACOG) policy states that elective C/S (dened by ACOG as C/S by maternal request) should not be performed earlier than 39 weeks (ACOG Committee Opinion, 2007), which given a + 2 week error in gestational 112

PAGE 123

age dating based on the last menstrual period, should presumably rarely result in infants delivered prior to 37 completed weeks of gestation. 113

PAGE 124

Table 4.1 ICD-9 Codes for Selected Maternal Morbidities. Maternal Morbidities ICD 9 Codes Postpartum Hemorrhage 666.x Bladder injury requiring repair 75.61 Venous complications 671.x (excludes venous complications prior to pregnancy) Unspecied febrile conditions 780.6 Puerperal infection 670.xx 114

PAGE 125

Table 4.2 Demographic Characteristics of Low Classied Risk Late-Preterm Delivery Overall and by Delivery Method, Florida 1998 to 2006. Table 4.2 Demographic Characteristics of Low Classied Risk Late-Preterm Delivery Overall and by Delivery Method, Florida 1998 to 2006. Table 4.2 Demographic Characteristics of Low Classied Risk Late-Preterm Delivery Overall and by Delivery Method, Florida 1998 to 2006. Table 4.2 Demographic Characteristics of Low Classied Risk Late-Preterm Delivery Overall and by Delivery Method, Florida 1998 to 2006. Table 4.2 Demographic Characteristics of Low Classied Risk Late-Preterm Delivery Overall and by Delivery Method, Florida 1998 to 2006. All Deliveries (N=61,724) Vaginal (no prior C/S) (N=56,712) All Primary C/S (N=5,012) P Value Characteristic N (%) N (%) N (%) Maternal Race/Ethnicity Non-Hispanic Black 19,152 (31.0) 17,824 (31.4) 1,328 (26.5) <0.0001 Non-Hispanic Other 1,856 (3.0) 1,736 (3.1) 129 (2.6) 0.0131 Hispanic 11,643 (18.9) 10,578 (18.7) 1,065 (21.3) 0.0611 Non-Hispanic White 29,064 (47.1) 26,574 (46.9) 2,490 (49.7) Ref Maternal Age < 15 Years 252 (0.4) 244 (0.4) 8 (0.2) 0.0504 15-19 Years 9,790 (15.9) 9,256 (16.3) 534 (10.2) 0.0193 20-24 Years 18,058 (29.3) 16,948 (29.9) 1,110 (22.2) Ref 25-34 Years 26,616 (43.1) 24,232 (42.7) 2,384 (47.6) <0.0001 35-44 Years 6,951 (11.3) 5,984 (10.6) 967 (19.3) <0.0001 45+ Years 57 (0.1) 48 (0.1) 9 (0.2) 0.0025 Educational Attainment a < High School 36,892 (60.1) 34,360 (60.9) 2,532 (50.1) <0.0001 College 21,134 (34.4) 19,097 (33.9) 2,037 (41.0) <0.0001 College for 5+ Years 3,337 (5.4) 2,934 (5.2) 403 (8.1) Ref Parity b 0 25,025 (40.7) 22,198 (39.3) 2,827 (56.6) Ref < 5 35,828 (57.3) 33,159 (58.6) 2,095 (42.0) <0.0001 5+ 1,261 (2.1) 1,189 (2.1) 72 (1.4) <0.0001 Body Mass Index c > 30 2,716 (14.0) 2,400 (13.7) 316 (16.2) 0.0023 25-29.9 4,178 (21.5) 3,755 (21.5) 423 (21.6) 0.4195 < 25 12,552 (64.5) 11,335 (64.8) 1,217 (62.2) Ref Smoking Status Yes 6,877 (11.1) 6,358 (11.2) 519 (10.4) 0.0649 No 54,847 (88.9) 50,354 (88.8) 4,493 (89.6) Ref Payer Source Public 30,308 (49.1) 28,194 (49.7) 2,114 (42.2) <0.0001 Commercial 27,535 (44.6) 24,855 (43.8) 2,680 (53.5) Ref Charity/Self Pay 2,952 (4.8) 2,801 (4.9) 151 (3.0) <0.0001 Other 929 (1.5) 862 (1.5) 67 (1.3) 0.0105 Infant Sex Male 33,113 (53.7) 30,297 (53.4) 2,816 (56.2) 0.0002 Female 28,611 (46.4) 26,415 (46.6) 2,196 (43.8) Ref Infant Birth weight < 2,500 grams 13,684 (22.2) 11,969 (21.1) 1,715 (34.2) <0.0001 2,500-3,999 grams 46,987 (76.1) 43,817 (77.3) 3,170 (63.3) Ref 4,000-4,499 grams 1,051 (1.7) 924 (1.6) 127 (2.5) <0.0001 a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) a Missing data N=361 (0.6%); b Missing data N=184 (0.3%); c Applies only to data years 2004-2006 (N=19,446) 115

PAGE 126

Table 4.3 Distribution of Maternal Morbidities at Birth Hospitalization of Late Preterm Infants by Delivery Route, 1998-2006, Florida Table 4.3 Distribution of Maternal Morbidities at Birth Hospitalization of Late Preterm Infants by Delivery Route, 1998-2006, Florida Table 4.3 Distribution of Maternal Morbidities at Birth Hospitalization of Late Preterm Infants by Delivery Route, 1998-2006, Florida Table 4.3 Distribution of Maternal Morbidities at Birth Hospitalization of Late Preterm Infants by Delivery Route, 1998-2006, Florida Table 4.3 Distribution of Maternal Morbidities at Birth Hospitalization of Late Preterm Infants by Delivery Route, 1998-2006, Florida Complications* Complications* Vaginal Delivery Vaginal Delivery Primary C/S Delivery " N(%) N(%) N(%) Cesarean Complications Cesarean Complications " Anesthesia Complications Anesthesia Complications 0 0 0 Abdominal Wound Complications Abdominal Wound Complications 36 (0.1) 36 (0.1) 44 (0.9) Pelvic Adhesions Pelvic Adhesions 4 (0.0) 4 (0.0) 8 (0.2) Vaginal Complications Vaginal Complications " Obstetric Trauma Obstetric Trauma 1625 (2.9) 1625 (2.9) 48 (1.0) Anal Conditions Anal Conditions 0 0 2 (0.0) Uterine Inversion Uterine Inversion 9 (0.0) 9 (0.0) 1 (0.0) Vaginal Laceration Vaginal Laceration 831 (1.5) 831 (1.5) 3 (0.1) Pelvic Prolapse Pelvic Prolapse 2 (0.0) 2 (0.0) 0 Perineal Laceration Perineal Laceration 1,172 (2.1) 1,172 (2.1) 9 (0.2) Uterine Rupture Uterine Rupture 3 (0.0) 3 (0.0) 11 (0.2) Retained Placenta Retained Placenta 339 (0.6) 339 (0.6) 9 (0.2) Shared Complications Shared Complications " Bladder Repair Bladder Repair 544 (1.0) 544 (1.0) 6 (0.1) Venous Complications Venous Complications 360 (0.6) 360 (0.6) 25 (0.5) DVT/Embolism DVT/Embolism 31 (0.1) 31 (0.1) 9 (0.2) Pneumonia Pneumonia 59 (0.1) 59 (0.1) 20 (0.4) OB Coagulation Disorder OB Coagulation Disorder 40 (0.1) 40 (0.1) 7 (0.1) Urinary Incontinence Urinary Incontinence 5 (0.0) 5 (0.0) 0 Hysterectomy Hysterectomy 19 (0.0) 19 (0.0) 26 (0.5) Lactation Disorder Lactation Disorder 1 (0.0) 1 (0.0) 0 Postpartum Hemorrhage Postpartum Hemorrhage 1,029 (1.8) 1,029 (1.8) 70 (1.4) Unspecied Febrile Unspecied Febrile 190 (0.3) 190 (0.3) 68 (1.4) Puerperal Infection Puerperal Infection 59 (0.1) 59 (0.1) 42 (0.8) Intestinal Obstruction Intestinal Obstruction 4 (0.0) 4 (0.0) 0 Cervical Laceration Cervical Laceration 107 (0.2) 107 (0.2) 5 (0.1) Cardiopulmonary Arrest Cardiopulmonary Arrest 1 (0.0) 1 (0.0) 3 (0.1) Pyelonephritis Pyelonephritis 15 (0.0) 15 (0.0) 4 (0.1) Breast Conditions Breast Conditions 36 (0.1) 36 (0.1) 19 (0.4) Cerebrovascular Disorder Cerebrovascular Disorder 8 (0.01) 8 (0.01) 5 (0.1) Cystitis Cystitis 14 (0.0) 14 (0.0) 2 (0.0) Hematoma Hematoma 31 (0.1) 31 (0.1) 2 (0.0) Cardiomyopathy Cardiomyopathy 8 (0.01) 8 (0.01) 4 (0.1) *International Classication of Diseases, 9th Revision, Primary or Secondary Diagnosis. *International Classication of Diseases, 9th Revision, Primary or Secondary Diagnosis. *International Classication of Diseases, 9th Revision, Primary or Secondary Diagnosis. *International Classication of Diseases, 9th Revision, Primary or Secondary Diagnosis. *International Classication of Diseases, 9th Revision, Primary or Secondary Diagnosis. 116

PAGE 127

Table 4.4 Reported Maternal Morbidities at Initial Hospitalization (Delivery) among Classied Low Risk for Cesarean Late Preterm Births by Mode of Delivery, 1998-2006, Florida a (Vaginal as Referent) Table 4.4 Reported Maternal Morbidities at Initial Hospitalization (Delivery) among Classied Low Risk for Cesarean Late Preterm Births by Mode of Delivery, 1998-2006, Florida a (Vaginal as Referent) Table 4.4 Reported Maternal Morbidities at Initial Hospitalization (Delivery) among Classied Low Risk for Cesarean Late Preterm Births by Mode of Delivery, 1998-2006, Florida a (Vaginal as Referent) Table 4.4 Reported Maternal Morbidities at Initial Hospitalization (Delivery) among Classied Low Risk for Cesarean Late Preterm Births by Mode of Delivery, 1998-2006, Florida a (Vaginal as Referent) Table 4.4 Reported Maternal Morbidities at Initial Hospitalization (Delivery) among Classied Low Risk for Cesarean Late Preterm Births by Mode of Delivery, 1998-2006, Florida a (Vaginal as Referent) Table 4.4 Reported Maternal Morbidities at Initial Hospitalization (Delivery) among Classied Low Risk for Cesarean Late Preterm Births by Mode of Delivery, 1998-2006, Florida a (Vaginal as Referent) Vaginal Delivery (N=56,712) Primary C/S Delivery (N=5,012) RR ARR a 95% CI a MORBIDITY " " Postpartum Hemorrhage 1,029 (1.8) 70 (1.4) 0.77 0.75 (0.59,0.96) Bladder Repair 544 (1.0) 6 (0.1) 0.12 *** *** Venous Complications 360 (0.6) 25 (0.5) 0.79 0.68 (0.45,1.02) Unspecied febrile conditions 190 (0.3) 68 (1.4) 4.05 3.84 (2.89,5.10) Puerperal Infection 59 (0.1) 42 (0.8) 8.05 8.09 (5.33,12.27) a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b RR(Relative Risk), ARR(Adjusted Relative Risk) ***Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b RR(Relative Risk), ARR(Adjusted Relative Risk) ***Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b RR(Relative Risk), ARR(Adjusted Relative Risk) ***Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b RR(Relative Risk), ARR(Adjusted Relative Risk) ***Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b RR(Relative Risk), ARR(Adjusted Relative Risk) ***Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b RR(Relative Risk), ARR(Adjusted Relative Risk) ***Insufcient sample size for multivariable methods. 117

PAGE 128

Table 4.5 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 1998-2006, Florida a Table 4.5 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 1998-2006, Florida a Table 4.5 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 1998-2006, Florida a Table 4.5 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 1998-2006, Florida a Table 4.5 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 1998-2006, Florida a Table 4.5 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 1998-2006, Florida a Table 4.5 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 1998-2006, Florida a Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Birth Hospitalization Birth Hospitalization Hospitalizations in Neonatal Period Hospitalizations in Neonatal Period Hospitalizations in First Year of Infancy Hospitalizations in First Year of Infancy ARR 95% CI ARR 95% CI ARR 95% CI MORBIDITY " " " Hemorrhage 0.75 (0.59,0.96) 0.63 (0.48,0.82) 0.67 (0.54,0.87) Bladder Injury ** ** ** ** ** ** Venous Complications 0.68 (0.45,1.02) 0.74 (0.50,1.11) 0.74 (0.50,1.08) Unspecied febrile conditions 3.84 (2.89,5.10) 4.05 (3.08,5.34) 3.87 (3.02,4.96) Puerperal Infection 8.09 (5.33,12.27) 6.28 (4.60,8.57) 5.70 (4.21,7.71) a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. 118

PAGE 129

Table 4.6 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women with Low Risk C/S without Labor and Women with Unassisted Vaginal Deliveries who delivered late preterm, 1998-2006, Florida a Table 4.6 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women with Low Risk C/S without Labor and Women with Unassisted Vaginal Deliveries who delivered late preterm, 1998-2006, Florida a Table 4.6 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women with Low Risk C/S without Labor and Women with Unassisted Vaginal Deliveries who delivered late preterm, 1998-2006, Florida a Table 4.6 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women with Low Risk C/S without Labor and Women with Unassisted Vaginal Deliveries who delivered late preterm, 1998-2006, Florida a Table 4.6 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women with Low Risk C/S without Labor and Women with Unassisted Vaginal Deliveries who delivered late preterm, 1998-2006, Florida a Table 4.6 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women with Low Risk C/S without Labor and Women with Unassisted Vaginal Deliveries who delivered late preterm, 1998-2006, Florida a Table 4.6 Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women with Low Risk C/S without Labor and Women with Unassisted Vaginal Deliveries who delivered late preterm, 1998-2006, Florida a Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Birth Hospitalization Birth Hospitalization Hospitalizations in Neonatal Period Hospitalizations in Neonatal Period Hospitalizations in First Year of Infancy Hospitalizations in First Year of Infancy ARR b 95% CI ARR b 95% CI ARR b 95% CI MORBIDITY " " " Hemorrhage 0.69 (0.46,1.04) 0.64 (0.42,0.98) 0.55 (0.37,0.82) Bladder Injury ** ** ** ** ** ** Venous complications 0.51 (0.24,1.07) 0.43 (0.19,0.97) 0.53 (0.26,1.10) Unspecied febrile conditions 3.98 (2.59,6.13) 4.76 (3.22,7.03) 4.85 (3.49,6.75) Puerperal Infection 9.68 (5.53,16.92) 9.58 (6.57,13.98) 8.02 (5.53,11.63) a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. 119

PAGE 130

Figure 4.1 Markov Hierarchical Model for Maternal Morbidity Selection 120

PAGE 131

Figure 4.2 Markov Pooled Maternal Morbidity Curve 121

PAGE 132

Figure 4.3 Description of Study Sample After Application of Low Risk C/S Algorithm Low Documented Risk C/S Algorithm (Applied to Cesarean and Vaginal Deliveries) Congenital Defects: Hernia (n=60,0.1%) Heart (n=4,332, 3.4%) Anencephaly (n=20, 0.0%) Spina Bifida (n=78, 0.1%) Gastroschisia/Omphalocele (n=302,0.2%) Diabetes Type I, II (n=5,127, 4.0%) Gestational Diabetes (n=8,279,6.5%) Chronic Hypertension (n=10,631,8.4%) Gestational Hypertension (n=20,995,16.5%) Eclampsia (n=1,309, 1.0%) Fetal Distress (n=4,170, 3.3%) Meconium Staining (n=4,079, 3.2%) Breech/Malpresentation (n=11,657, 9.2%) Chorioamnionitis (n=2,881, 2.3%) Prolonged Labor (n=6,220, 4.9%) Birthweight > 4,500 grams (n=454, 0.4%) Intensive Prenatal Care (n=6,723, 5.3%) Initial Study Sample Late Preterm Initial Study Sample Late Preterm Births N=127,364 Births N=127,364 Deliveries with Documented Risk N=54,460 (42.8%) All Cesarean Deliveries N=14,264 (19.6%) All Vaginal Deliveries N=58,640 (80.4%) Primary Cesarean N=5,012 (8.1%) Vaginal (no prior C/S) N=56,712 (91.9%) Primary C/S w/o Labor N=1,814 (3.4%) Unassisted Vaginal N=53,460 (96.6%) Deliveries with Low Documented Risk N=72,904 (57.2%) 122

PAGE 133

Figure 4.4 Kaplan-Meier Estimate of Time Till First Rehospitalization (Days) Among Women who Delivered Singleton Late Preterm Infants by Route of Delivery, Florida 1998 to 2006 Time Till First Rehospitalization (Days) Primary Cesarean Vaginal Log-Rank p < 0.0001 123

PAGE 134

CHAPTER FIVE !The increasing rate of cesarean deliveries in the United States has become a substantial public health concern. Cesarean deliveries have been demonstrated to increase the risk for adverse health outcomes for both mothers and their infants. Further, research demonstrating an association between Florida's increasing rate of deliveries at 34 to 36 weeks gestation and C/S delivery is important, as both premature births and C/S deliveries have been independently linked to an increased risk of morbidity. Information on the impact of C/S delivery at late preterm for both mothers and infants is scarce. Research on C/S delivery at late preterm is an important contribution as it provides an evidence base for programs and public health policy focused on preventing prematurity and reducing the C/S rate. !To address the dearth of information on C/S delivery at late preterm, this dissertation focused on three key areas: (1) an assessment of the accuracy of data used to investigate primary C/S late preterm birth, (2) an examination of the impact of C/S on late preterm infant morbidity, and (3) the impact of C/S at late preterm on maternal morbidity. Specic aims and corresponding research questions were developed to guide the dissertation research: Specic Aim 1 : To determine the validity of data sources (e.g. Florida birth certicate, Florida hospital discharge data) used to investigate primary C/S delivery and late PTB outcomes using maternal medical charts as the gold standard. 124

PAGE 135

! Research Questions !1. What is the validity of the Florida birth certicate compared with maternal !medical charts? !2. What is the validity of the Florida Hospital discharge data compared with !maternal medical charts? !3. What is the validity of the linked Florida birth certicate and hospital !discharge data compared with maternal medical charts? !4. Is there a signicant difference between the validity of the linked Florida birth !certicate and hospital discharge data by hospital volume (high primary C/S !versus low primary C/S rate)? Specic Aim 2 : Assess the impact of low documented risk C/S on maternal and late preterm infant morbidity. 5. What impact does low documented risk primary C/S have on maternal and !singleton late-preterm infant morbidity and healthcare utilization? !6. Is there variation by important subgroups (e.g. race/ethnicity, payer source)? THEORETICAL MODEL !The Social Ecological Model was used to guide this dissertation research because it has great utility in describing the complex etiology of increasing rates of cesarean delivery. Specically, the increase in C/S delivery has been attributed to several factors that operate on different ecological levels. For example, individual level factors such as increases in maternal age and morbid conditions as well as maternal request for cesarean delivery have been frequently discussed, as well as interpersonal factors such as patientprovider relationship and power dynamics. Institutional factors such as scheduling, labor 125

PAGE 136

policy, malpractice insurance and culture have also been discussed. Societal factors such as healthcare policy, and the healthcare system (to name a few) are of prime importance as they have contributed to the problem of increasing rates of cesarean delivery as well. The SEM aids understanding of the complexity of the increasing C/S rate issue, but also provides a foundation for the implications of research results as well as proposed avenues for prevention. !This dissertation was prepared in a manuscript format. Three manuscripts were drafted to describe each of the dissertation components. While each of these manuscripts are able to stand alone, they are also part of a cohesive body of research focused on the issue of maternal and infant morbidity following C/S at late preterm. To organize and integrate the discussion of the results and research implications, each of the research questions will be addressed and discussed according to the specic aims and research questions. Further, the strengths and limitations of the dissertation, as well as the public health implications and recommendations for further research will be discussed in the context of the overall dissertation research, and theoretical framework. OVERVIEW OF SIGNIFICANT FINDINGS !The signicant ndings of this dissertation research are organized and discussed in accordance the specic aims and related research questions. Specic Aim 1 : To determine the validity of data sources (e.g. Florida birth certicate, Florida hospital discharge data) used to investigate primary C/S delivery and late PTB 126

PAGE 137

outcomes using maternal medical charts as the gold standard. Obstetric nurse abstractors collected prenatal and delivery information from 1,249 maternal medical charts from 16 hospitals in Florida that were classied as either a high rate or low rate C/S hospital. Of the 1,249 records abstracted, approximately 15% were misclassied (e.g. repeat C/S, vaginal delivery, multiple gestation). Hospitals with high C/S rates had higher rates of misclassication for delivery route than hospitals with low C/S rates. About 25% of births were incorrectly classied as late preterm deliveries, although this rate of misclassication did not vary by hospital C/S rate category. Question One: What is the validity of the Florida birth certicate compared with maternal medical charts? !Substantial variation in the accuracy of maternal medical conditions, risk factors as well as labor and delivery complications was observed on Florida birth certicate data compared to maternal medical charts. The sensitivity ranged from a low of zero for conditions such as renal disease, to a high of 0.76 for maternal obesity. Several key conditions such as chronic diabetes, hypertensive conditions, trial of labor and labor induction had at least 50% sensitivity and PPV. Kappa values and likelihood ratios +/have indicated that most data elements on the Florida birth certicate had less than a moderate level of agreement, with the exception of chronic diabetes, gestational diabetes, hypertensive conditions of pregnancy, obesity, induction of labor, and breech/ malpresentation. 127

PAGE 138

Question Two: What is the validity of the Florida Hospital discharge data compared with maternal medical charts? !The validity of many data elements in the in-patient hospital discharge data from AHCA was higher than the indices observed on the birth certicate. Sensitivity and PPV varied by data element, ranging from 0 for syphilis to 0.89 for hypertensive conditions of pregnancy. Overall, the validity indices for data elements were much higher when medical charts were compared to discharge data than vital records. Many data elements had at least 50% sensitivity and PPV: chronic diabetes, gestational diabetes, chronic hypertension, hypertensive conditions of pregnancy, HIV, IUGR, cerclage, placenta previa, placenta abruption, breech/malpresentation, and prolonged labor. The specicity and NPV of most data elements was high (>90%), with one notable exception, TOL, which had a specicity of just over 50% and a NPV < 50%. The values for Kappa and likelihood ratio +/indicated that many variables had at least a moderate level of agreement, with several reaching near perfect agreement: cerclage, IUGR, gestational diabetes, chronic hypertension, hypertensive conditions of pregnancy, placenta previa, placenta abruption, breech/malpresentation, and prolonged labor. Question Three: What is the validity of the linked Florida birth certicate and hospital discharge data compared with maternal medical charts? !The linked Florida birth certicate and hospital discharge data le had greater accuracy rates than either the Florida birth certicate or discharge data alone. For example, sensitivity ranged from 0.08 (renal disease) to a high of 0.91 (hypertensive conditions of pregnancy as well as breech/malpresentation). About half of the variables evaluated had at least moderate validity (sensitivity and PPV of at least 50%): cerclage, chronic diabetes gestational diabetes, IUGR, hypertensive conditions of pregnancy, HIV 128

PAGE 139

infection, genital herpes, TOL, breech/malpresentation, induction of labor, placenta abruption, placental previa, and prolonged labor. For all but three data elements, the rates of specicity and NPV were greater than 90%. Only hypertensive conditions of pregnancy, TOL and fetal distress had lower than 90% specicity, with TOL as the lowest (Spec = 0.39, NPV = 0.56) Question Four: Is there a signicant difference between the validity of the linked Florida birth certicate and hospital discharge data by hospital volume (high primary C/S versus low primary C/S rate)? !The validity of selected data elements by hospital C/S rate classication for the linked Florida birth certicate discharge data le were compared to maternal medical chart information. Large variation across all indices of validity (sensitivity, PPV, and Kappa) was observed. For most of the data elements assessed by hospital C/S rate category, the rates of sensitivity were higher for hospitals classied as having a low C/S rate, expect for a few variables: placenta abruption, chorioamnionitis, breech/ malpresentation, genital herpes, gonorrhea, and fetal distress. Overall, 26 data elements were examined and for 17, the PPV was higher among hospitals with low C/S rates. For high C/S rate hospitals, the PPV was higher than low C/S rate hospitals for heart problems, gestational diabetes, chronic hypertension, Gonorrhea, genital herpes, placenta abruption, prolonged rupture of membranes, chorioaminonitis and meconium. Signicant differences in accuracy by hospital C/S rate category were observed. Only three variables signicantly differed in accuracy by hospital C/S rate classication: anemia, attempted labor, and induction of labor. Each of these three factors had higher sensitivity among hospitals with low C/S rates. 129

PAGE 140

Specic Aim 2 : Assess the impact of low documented risk C/S on maternal and late preterm infant morbidity. All late preterm singleton live births from 1998 to 2006 were included in analyses of the impact of C/S delivery at late preterm on maternal and infant health outcomes. During the study time period, there were 127,364 births that met inclusion criteria for this dissertation research. After applying the low risk algorithm outlined in both the introduction and manuscripts 2 & 3 of this dissertation, 54,460 births (42.8%) were excluded as they indicated higher risk for C/S delivery. This exclusion resulted in a sample of 72,904 births. Further restriction was necessary to include only those deliveries in which no prior C/S was reported. The nal study sample for both manuscripts (infant morbidity and maternal morbidity) was 61,724 deliveries (5,012 primary C/S and 56,712 vaginal deliveries). !With the exception of payer type, the socio-demographic characteristics were the same for both the infant and maternal morbidity analyses. The sample was predominantly White (47.1%) or Black (31.0%), and most women were 25 years or older. The rate of C/S was higher for women who were White, had higher levels of educational attainment, were obese or who reported commercial health insurance as the anticipated payer of services. 130

PAGE 141

Question Five: What impact does low documented risk primary C/S have on maternal and singleton late-preterm infant morbidity and healthcare utilization? Infant morbidity !Six infant morbidities were included in analyses, as determined by Markov: feeding difculties, perinatal infections, respiratory distress, jaundice, transient tachypnea and hypoglycemia, as well as a composite measure of morbidity (includes infants that experienced at least one of the six morbidities). The most commonly reported morbidity among infants was jaundice, followed by respiratory distress. Over half of the infants delivered via C/S reported at least one morbidity, whereas just over a third of vaginally delivered infants reported at least one morbidity. Overall, infants delivered C/S were more likely to report a greater number of morbidities experienced by individual infants. For each of the three time periods assessed (birth hospitalization, neonatal period, rst year of infancy), C/S delivery carried the greatest risk of infant morbidity. The highest risk of morbidity following C/S was transient tachypnea, while the risk was lowest for jaundice. Maternal morbidity !Five maternal morbidities were examined in this dissertation research, as determined by the Markov model: postpartum hemorrhage, bladder repair, venous complications, unspecied febrile conditions and puerperal infection. Of these morbidities, postpartum hemorrhage, bladder repair and venous complications were more commonly reported among vaginal deliveries, while puerperal infection and unspecied febrile conditions were more commonly reported with C/S delivery. After 131

PAGE 142

adjusting for socio-demographic covariates (e.g. smoking, race/ethnicity, parity, maternal age, education, and payer type), C/S delivery was protective for both postpartum hemorrhage and venous complications. Mothers who delivered via C/S had an increased risk for unspecied febrile conditions and puerperal infection. !Infant healthcare utilization Infant healthcare utilization was assessed by examining length of stay at the birth hospitalization, the number of rehospitalizations as well as the time till rst rehospitalization. Infant length of stay (LOS) at the birth hospitalization differed signicantly by route of delivery, with C/S delivered infants requiring longer LOS. About 14% of the study population required rehospitalization during the rst year of infancy. There was a higher rate of rehospitalization for infants delivered via C/S compared to infants delivered vaginally, although this difference did not remain after restricting to C/S without labor. Infants delivered via C/S had a higher number or rehospitalizations in the rst year of life, and were more likely to rehospitalized earlier than infants delivered vaginally. Maternal healthcare utilization The LOS at birth hospitalization and subsequent rehospitalizations was greater for mothers who delivered via C/S. The association between primary C/S and maternal rehospitalization was investigated. Overall, 4.7% of the study population required rehospitalization. The maternal rehospitalization rate for C/S deliveries was over twice that of vaginal deliveries (4.1% versus 10.4%). Mothers who delivered C/S had a greater number of rehospitalization events in the rst year post-partum. There was a 132

PAGE 143

signicant difference in the timing of rst rehospitalization by route of delivery, with mothers who delivered C/S rehospitalized earlier than mothers who delivered vaginally. The risk of rehospitalization was substantially higher among women who delivered via C/S. Question Six: Is there variation by important subgroups (e.g. race/ethnicity, payer source, cesarean without labor)? Infant health outcomes !Important sub-group differences were observed in overall risk for morbidity by delivery route. Infants of women who were identied as White had the highest risk for morbidity following C/S compared to vaginally delivered infants. A protective association between composite morbidity and jaundice regardless of route of delivery was observed among infants delivered by Black mothers. !Signicant interaction between race/ethnicity and low birth weight (LBW) status was observed for some of the infant morbidities assessed in this dissertation (respiratory distress, transient tachypnea, and feeding difculties). Specically, the LBW infants of Black mothers had higher rates of morbidity than infants with normal birth weight for transient tachypnea, respiratory distress and feeding difculty (average birthweight infants delivered by White mothers as referent group). The rates of morbidity for these conditions were substantially higher for LBW infants delivered by White mothers. !The study population was further restricted to C/S deliveries without labor and unassisted vaginal deliveries to assess differences in infant morbidity and healthcare. Following this restriction of the study population, the risk of transient tachypnea 133

PAGE 144

increased, and the association between primary C/S and jaundice and feeding difculties was no longer signicant. C/S deliveries without labor still had a longer LOS at the birth hospitalization and subsequent rehospitalizations, earlier readmission, and a higher number of rehospitalization episodes. !Maternal health outcomes !While no signicant interactions between maternal morbidity and sociodemographic covariates were observed, some important subgroup differences were noted. Specically, the association between Black mothers and venous complications was protective throughout the birth hospitalization, neonatal period and the rst year postpartum. Conversely, Black mothers had a higher risk for puerperal infection, regardless of postpartum length. Women with a public payer type also had a consistently higher risk of puerperal infection. As with the infant morbidity analysis, the primary C/S group was restricted to C/S without indications of labor and the vaginal delivery group was restricted to unassisted vaginal deliveries. After restricting primary C/S to C/S without labor, the protective association between C/S delivery and postpartum hemorrhage was no longer signicant at the birth hospitalization, but still present during the neonatal and rst year postpartum time periods. The risk of unspecied febrile conditions and puerperal infection observed among women who delivered C/S remained following the restriction to C/S without labor. The average number of rehospitalization events in the rst year postpartum increased for C/S deliveries, and the association between timing of rehospitalization and route of delivery remained with C/S deliveries requiring maternal rehospitalizations 134

PAGE 145

earlier than vaginal deliveries. Mothers delivering C/S without labor still had a longer LOS at the birth and subsequent rehospitalization events. The overall risk of rehospitalization for mothers delivering C/S increased. CONSIDERATION OF FINDINGS IN LIGHT OF EXISTING RESEARCH !Validity of birth certicate, discharge, and linked data !Overall, we observed that the birth certicate had variable accuracy compared to maternal medical charts ranging from low to moderate for most data elements. Unfortunately, there have been no previous studies of the accuracy of the Florida birth certicate (with the exception of one study examining prepregnancy weight and height), therefore comparisons of the dissertation results to other studies of the Florida birth certicated cannot be made (Park, Sappeneld, Bish, Bensyl, Goodman, Menges, 2009). However, there has been substantial research from other states demonstrating similar ndings for the accuracy of birth certicate data elements (DiGiuseppe et al, 2002; Piper et al, 1993; Reichman et al, 2001; Roohan et al, 2003). For example, the states of Tennessee, New Jersey, Ohio, and New York have all reported low rates of accuracy (sensitivity) for medical risk factor variables, as well as obstetric procedure and labor and delivery variables (DiGiuseppe et al, 2002; Piper et al, 1993; Reichman et al, 2001; Roohan et al, 2003). Many of the data elements on the Florida birth certicate had higher rates of accuracy than those reported by other states. 135

PAGE 146

!There have been several published reports demonstrating the improved accuracy with discharge data (as compared to the accuracy of birth certicate data) (Kahn et al, 2009; Lydon-Rochelle et al, 2005). Many data elements in the Florida hospital in-patient discharge data were more accurate that the birth certicate, when using maternal medical charts as the gold standard for comparison. Accuracy improved further once birth certicate and discharge data were linked. Lydon-Rochelle et al (2005) conducted a similar study of Washington State linked birth certicate and discharge data and also observed improved sensitivity for many pregnancy complications and maternal conditions. However, there was some variability in the accuracy of some conditions in the current results compared to the results of Lyndon-Rochelle (2006). Sensitivity for gestational diabetes, diabetes mellitus, and chronic hypertension was higher in LydonRochelle's investigation, while the sensitivity of placenta previa, and pregnancy-induced hypertension was higher for Florida. !Many of the data elements used to describe risk factors for C/S (e.g. hypertensive conditions of pregnancy, hypertension, diabetes mellitus, gestational diabetes, IUGR, and breech/malpresentation) had higher rates of accuracy in the discharge data le than the Florida birth certicate, and even higher rates of accuracy when using the linked birth certicate discharge data le, suggesting that using the linked data le would result in lower rates of misclassication an important nding for studies assessing the implications of low risk C/S delivery for maternal and infant health outcomes. LydonRochelle et al (2005) have supported this observation by stating that "a strategy of using combined data sources was more accurate for the detection of maternal pre-existing 136

PAGE 147

medical conditions and the complications of pregnancy than single data source strategies" (pg. 133). Infant morbidity and healthcare utilization Several researchers have demonstrated that late preterm infants have higher rates of morbidity than infants delivered at term (Burgos et al, 2008; McIntire et al, 2008; McLaurin et al, 2009; Wang et al, 2004). Researchers have also identied cesarean delivery as a risk factor for infant morbidity and rehospitalization (Tita et al, 2009). However, the consequences of C/S at late preterm have not been as widely investigated. After adjusting for several socio-demographic factors (e.g. maternal age, race/ethnicity, payer status), a signicant increased risk for morbidity was observed for infants delivered via C/S compared to infants delivered vaginally. Malloy (2009) also observed an increased risk of neonatal morbidity for preterm infants (32 to 36 weeks GA), as well as an increase in mortality among low risk infants delivered via C/S. !Researchers have described differences in infant LOS at the birth hospitalization by route of delivery, with C/S deliveries requiring longer LOS than vaginal deliveries (Tita, Landon, Spong, Lai, Leveno, Varner, Moawad et al, 2009). We also observed an increase in LOS among infants delivered via C/S. The rate of rehospitalization for late preterm infants (14%) in this research was similar previously published rates (4.8%-15.2%) (McLaurin et al, 2007; Shapiro-Mendoza et al, 2006; Underwood et al, 2007). Infants delivered via C/S had signicantly more rehospitalization episodes in their rst year of life, as well as earlier time to initial rehospitalization. 137

PAGE 148

!This dissertation did not use the intention-to-treat model advocated by the National Institutes of Health (NIH), which groups primary C/S deliveries with labor with vaginal deliveries, resulting in a "planned vaginal delivery" group. The purpose of the intention-to-treat algorithm is to distinguish between planned and unplanned C/S (de Luca et al, 2009; Declercq et al, 2007, NIH, 2006). As this research builds on research previously conducted by the FDOH (submitted for publication), it was important to remain consistent with the low risk algorithm used in Florida to examine C/S without medical indication. This algorithm was developed by the FDOH and Florida Obstetrics and Gynecologic Society (FOGS the state chapter of ACOG) to examine the association between C/S delivery and the increasing rate of late preterm deliveries. As these deliveries occurred at late preterm, presumably, most deliveries were planned vaginal (and planned term). The subgroup analysis of primary C/S without labor provides information on the group considered by the intention-to-treat algorithm to be "planned cesarean". The revised rate of primary C/S without indications of labor (3.4%) was similar to those reported by other researchers (de Luca et al, 2008). The results of the dissertation did not change substantially after restricting to C/S without labor indications. Maternal morbidity and rehospitalization The rate of maternal morbidity was low, which was not too surprising given the low risk classication of the study population. Further, rates of maternal morbidity reported by other studies have also been low (Liu et al, 2005). C/S delivery among mothers delivering late preterm was protective for postpartum hemorrhage. Other researchers have also identied the negative association between C/S and postpartum 138

PAGE 149

hemorrhage (Liu et al, 2005; Lydon-Rochelle et al, 2000; Ophir et al, 2008). A positive association for both puerperal infection and unspecied febrile morbidity and C/S was also identied. Researchers have also documented the increased risk of infection among mothers who delivered via C/S (Declercq et al, 2007; Liu et al, 2005). Maternal rehospitalization is perhaps a more ideal outcome for comparing the burden of morbidity by route of delivery, particularly as maternal morbidities vary greatly by route of delivery. For example, perineal lacerations (episiotomy) are most commonly observed among vaginal deliveries, while abdominal would infections are more likely with a C/S. Rehospitalization allows direct comparison of the burden of serious morbidity by route of delivery, without having to specify type of morbidity. An increased risk of rehospitalization was observed for mothers who delivered via C/S. This nding has been reported by several other researchers (Declercq et al, 2007; Liu et al, 2005; Lydon-Rochelle et al, 2000, and Ophir et al, 2008). In the present study, mothers who delivered C/S had a greater number of rehospitalization events in the rst year postpartum, and were rehospitalized earlier than women who delivered vaginally. Further, they had a longer LOS at the birth hospitalization as well as subsequent rehospitalizations. As with the analysis of infant morbidity, the Intention-to-Treat model suggested by the NIH was not used to classify low risk C/S. Again this research is an extension of research conducted by the FDOH, which developed a low risk C/S algorithm for use in Florida in collaboration with FOGS. To determine if restriction of primary C/S deliveries to those without indications of labor would impact study results, a subgroup 139

PAGE 150

analysis was performed. The results of maternal morbidity and rehospitalization did not change substantially following the application of C/S without labor restriction. PUBLIC HEALTH IMPLICATIONS AND POLICY RECOMMENDATIONS The discussion of the public health implications of the results as well as recommendations for public health policy changes is framed within the context of the Social Ecological Model (SEM), as it provides a framework for understanding the multilevel determinants of public health problems such as high C/S. The SEM also provides a framework for intervention as problems inuenced by several ecological levels require solutions that address each of these levels. Further, each of the levels of the SEM inuence each other, as each level is embedded in larger social or economic system. Individual level !There has been great debate in the C/S literature regarding maternal request for C/ S. The rate of this phenomenon is difcult to measure, however some studies have indicated that maternal request C/S is a relatively rare occurrence (Declercq et al, 2006, NIH, 2006). Maternal request was not examined, as the data sources did not contain this information. Other individual level factors such as socio-demographic characteristics were available for comparison by delivery route. C/S rates were higher among certain socio-demographic groups and behaviors. White, highly educated women with commercial insurance as an expected payer source had higher rates of C/S. This nding is likely more related to macro level inuences such as economics of the healthcare 140

PAGE 151

delivery system, and socio-economic status, and may be best addressed through policy at the level of the U.S. healthcare system. One intervention that would be suited for the individual level would be patient empowerment and education strategies. Education campaigns such as the March of Dimes Late Preterm Brain Development Card have been developed to prevent prematurity by stressing the importance of carrying a pregnancy to full term (March of Dimes, 2010). There has also been research on formalizing the content of the patient provider discussion about the risk and benets of C/S delivery so that mothers can make more informed decisions about their birth experiences (Milne, Gafni, Lu, Wood, Suave & Ross (2009). Interpersonal level !The interpersonal level was not directly assessed by this dissertation as data for interpersonal factors are rarely contained in population-based data sources. The interpersonal level however, is a vital component of any program to reduce C/S delivery rates as this level most often includes the agent of the C/S delivery the healthcare provider. As part of the informed decision process, the power imbalance between the physician and the expectant mother is substantial (Gamble et al, 2007). The physician is the gatekeeper of knowledge. Presentation of the risk and benets of C/S delivery is performed by the physician who may or may not perform this task sufciently (i.e. balanced) to allow for informed consent. Patient-provider interaction and decisionmaking processes for C/S is in need of further research. Although this is admittedly a controversial area, as healthcare decisions and provider-patient relationships are very protected, legally and culturally. 141

PAGE 152

!Institutional level The phase one validation component of this dissertation assessed differences in accuracy of data submitted by institutions with high C/S rates compared to institutions with low C/S rates. Some institutional level differences in accuracy were observed, with a tendency towards more accurate data among institutions with lower C/S rates. The reasons for this are unclear, and require further investigation. However, the validation activities have raised a valid concern about the use of the maternal medical record as the "gold standard" for comparisons of accuracy, as there are many sources of error that can occur in paper-based records with multiple handlers (nurses, physicians, coding clerks, etc). The institutional level differences observed demonstrated overall variation in accuracy. To reduce such error, a standardized electronic medical record is proposed. This idea is certainly not novel, as electronic medical records have been widely researched and advocated as a source of healthcare quality improvement in the form of care coordination and improvement in provider efciency and work ow (Kahn & Ranade, 2010; Lurio, Morrison, Pichardo, Berg, Buck, Wu et al, 2010). !The United States healthcare system was a major campaign issue in the 2008 Presidential Election, with Barack Obama elected on a platform of healthcare reform. Part of the reform proposed by President Obama includes a transition to electronic medical records. Since 2009, the President has drafted policy to move the U.S. healthcare system towards "a nationwide, interoperable, private, and secure electronic health information system" (Blumenthal, 2010, pg. 382). While accurate information is invaluable for epidemiologic assessment of maternal and infant health outcomes, it is also 142

PAGE 153

a vital component of ongoing quality patient care, and is vital for modern medicine (Blumenthal, 2010). With the adoption of electronic medical records, information can be collected in a standardized, centralized format resulting in not only more accurate data, but also improvement in patient care, and subsequent healthcare outcomes. !Despite a substantial evidence base for the adverse outcomes associated with low risk C/S delivery, and ACOG guidelines advocating that elective C/S be performed only when 39 weeks GA is achieved, increasing rates of C/S have contributed to increases in the rate of late preterm delivery. Researchers have identied several barriers to the adoption and implementation of clinical guidelines (Chaillet, Dube, Dugas, Gagnon, Poltras & Dumont, 2007). Perceived barriers include concerns over payment, time, cost, legal concerns and the patient-provider relationship. Chaillet et al (2007) conducted a qualitative study identifying barriers and facilitators for implementing guidelines aimed at reducing repeat C/S rates in Quebec, and observed that (1) providers felt a high level infrastructure was needed to assure safe VBAC, (2) providers felt that an anesthesiologist should be available around the clock, (3) providers had a fear of lawsuits should uterine rupture occur, and that (4) women preferred repeat C/S. Further research on the amelioration of perceived barriers for obstetrician's adoption of clinical guidelines is warranted, as the successes of provider and institutional level intervention for C/S deliveries are contingent upon provider participation and acceptance. Chaillet (2007) also identied provider reluctance to perform instrumental vaginal delivery. According to Bailey (2005), vacuum extraction and forceps have become a lost art. Due to insufcient training in instrumental vaginal delivery, many obstetricians prefer to perform a C/S when confronted by complicated labor. Medical school curricula, obstetric 143

PAGE 154

residency training, and provider continuing education opportunities should include instruction in instrumental vaginal delivery to increase the number of healthcare providers who are competent (and feel competent) to perform instrumental vaginal delivery. !Societal level The societal level was not directly assessed by this dissertation research, however it is a major component of the C/S increase. According to Cyr (2006), the C/S rate in the U.S. is an "indirect result of American public policy during the last century" (pg. 194). Only substantial changes in the provision of health and maternity care can reduce C/S rates. Among many academics, the poor performance of the U.S. healthcare system is undisputed. At present, the U.S. healthcare system is ranked 37 th in the world, yet the U.S. spends more money on healthcare than any other nation (Murray & Frenk, 2010). The U.S. healthcare system is a messy patchwork a poorly integrated system without adequate performance measures (Swensen et al, 2010). Care is often customized, and technology driven, with little effort for standardization. At present, the U.S. healthcare system pays for volume (i.e. number of visits, tests, procedures) rather than value. Unfortunately, the current issue with C/S delivery is an ideal case study for what is wrong with the U.S. healthcare system. The rates of C/S delivery continue to rise regardless of the substantial evidence base that has demonstrated that C/S without medical indication increases the risk for poor maternal and infant health outcomes and healthcare costs. !Several researchers have observed a correlation between the cost of malpractice insurance and the rate of C/S delivery. As malpractice rates increase, so does the rate of 144

PAGE 155

C/S (Murthy, Grobman, Lee, Holl, 2009; Yang, Mello, Subramanian, Studdert, 2009). Fear of liability may inuence provider choice of delivery method (Yang et al, 2009). A disconcerting study by Murthy, Grobman, Lee and Holl (2009) demonstrated an increase in late preterm induction for every annual $10,000 increase in obstetric malpractice premiums in Illinois. Tort reform is needed to reduce the medical liability associated with birth events, as rising costs of malpractice insurance exert substantial pressure on healthcare providers and impact healthcare decisions. There appears to be a disconnect in terms of legal perspectives of quality obstetric care, with the perception that cesarean delivery is a safer route of delivery. Insurers and the legal community need to become aware of the increased risk of infant and maternal morbidity following C/S delivery, when used without a medical indication. !At the beginning of the 20 th century, the U.S. healthcare system moved maternity care into the hospital. From that time, childbirth has become increasingly medicalized. Less than 1% of deliveries in the United States are performed in the home by a certied nurse midwife (CNMs) (Malloy, 2010). Many developed nations use midwives for uncomplicated low risk pregnancies (Malott, Davis, McDonald & Hutton, 2009). According to the Midwifery Model of Care, pregnancy and childbirth are normal, natural life processes that rarely require medical intervention. Included in this model of care is monitoring of a woman's physical, psychological and social wellbeing (Midwifery Task Force, 2008). Midwives provide prenatal care and education, and prepare women for the childbirth experience. Further, midwives use minimal technological interventions during childbirth. Not surprising, researchers have demonstrated that midwife attended births result in fewer C/S deliveries. In the hospital setting, births attended by nurse 145

PAGE 156

midwives have resulted in fewer obstetric interventions such as C/S, with no change in levels of neonatal morbidity (Cragin & Kennedy, 2006; Janssen, Ryan, Etches, Klein, & Reime, 2007). In addition to reduced rates of C/S delivery, the Midwifery Model of Care has been demonstrated to have higher "customer" satisfaction and may result in lower healthcare costs (Dahlen, Barclay & Homer, 2008; Parry, 2008; Stone, Zwanziger, Hinton & Buenting, 2000). At present, the Florida Medicaid program provides reimbursement for care provided by birth centers and licensed midwives (AHCA, 2010), although many women do not utilize these services and instead opt for delivery in the hospital setting. Increased awareness (and support) of midwifery services for prenatal care and childbirth is necessary to increase the number of women who opt for this model of care. STRENGTHS AND LIMITATIONS The most signicant strength of this dissertation research is the study design. This study utilized a sequential, equivalent study design, resulting in two separate yet interrelated phases a phase focused on the validation of data, the results of which informed the subsequent phase's investigation of maternal and infant morbidity. For example, the rst phase of the study demonstrated that many socio-demographic characteristics are more accurate (or more completely collected) by the Florida birth certicate. Therefore, in maternal and infant morbidity analyses, many of the sociodemographic variables such as race/ethnicity were only drawn from birth certicate data. However, the results of the rst phase of the analysis also demonstrated that many 146

PAGE 157

pregnancy complications and maternal risk factors were captured best by using information from both the birth certicate and discharge data the linked birth certicate discharge data le. So for conditions such as chronic hypertension, hypertensive disorders of pregnancy, gestational diabetes and prolonged labor, the data resulting from the linkage of both birth certicate and discharge data sources was used in all analyses of morbidity. Results from phase one of this dissertation provide evidence for the assertion that using the linked data le for many of the factors included in the C/S risk algorithm likely reduced the level of misclassication that would have resulted if analyses focused on birth certicate data alone (Kahn et al, 2009; Lydon-Rochelle et al, 2005; ShapiroMendoza et al, 2008). !In addition to the design of the study, another important and complementary strength was the data used for this dissertation. For the rst phase of the study, data from maternal medical charts were compared to the Florida birth certicate and in-patient discharge data sets (and the resultant linked birth certicate discharge data le) in order to assess validity of data elements contained in the birth certicate and discharge data. Access to data abstracted from maternal medical charts is rare, primarily because abstracting maternal medical charts is unrealistic for most public health and other research endeavors. Due to the expense and time needed to conduct a validation study of birth certicate and hospital discharge data using maternal medical records as a gold standard, only a few studies have been reported over the past decade. Further, data abstracted from maternal medical charts is stratied by C/S rate classication high C/S rate versus low C/S rate, which allowed for a comparison in accuracy of data elements for birth certicate and discharge data by this C/S rate classication. At present, there have 147

PAGE 158

been no reports on differences in the accuracy of data reported by hospital C/S rate classication. !A major strength of the second phase of this dissertation is the use of populationbased data to investigate maternal and infant morbidity and healthcare utilization patterns by route of delivery. The linked birth certicate and discharge data le contained information on 97.4% of births that occurred in Florida from 1998 to 2007. This longitudinal, multi-year data source provided an opportunity to follow mothers and infants from delivery to the rst year postpartum, by examining rehospitalization events following the initial birth hospitalization. Another important strength of both the maternal and infant morbidity analyses is the use of GEE to adjust for intraclass correlation in multivariable models. In the linked birth certicate discharge data le, infants and mothers experience repeated events rehospitalizations. Presumably, each rehospitalization event for an infant or mother is related to subsequent rehospitalizations, introducing a source of error in terms of correlation. Further, many mothers have more than one infant during the 1998 to 2007 study period. This results in clustered data (infants as well as delivery episodes from the same mother), another potential source of error in risk estimation. GEE provides a robust methodology for handling clustered data of this type (Hanley, Negassa, Edwardes, & Forrester, 2003; Zeger & Liang, 1986) Furthermore, the use of the Markov hierarchical model to create a threshold for inclusion of infant and maternal morbidities in analyses is novel, and provides an objective epidemiologic-based method to explore morbidity outcomes. 148

PAGE 159

!There are a few limitations in this dissertation that warrant discussion. The rst phase used data abstracted from maternal medical charts from the Florida Late Preterm Cesarean Investigation. This public health investigation only abstracted data from maternal medical charts that were believed to be singleton late preterm primary cesarean deliveries (according to the sample frame drawn from the Florida Birth Certicate for 2006-2007). Further, data was abstracted from 16 Florida hospitals (8 with high C/S rates and 8 with low C/S rates). As such, the results of the rst phase of the dissertation are not representative of accuracy indices for all births in Florida. Further, as only late preterm primary C/S deliveries were abstracted, it is not possible to estimate the reverse error meaning that some primary C/S deliveries may have been incorrectly classied as repeat C/S or vaginal deliveries on the Florida birth certicate. Since there is no information on the accuracy of these births, the rate of misclassication captured in phase one is uni-directional. The purpose of each data source must be considered in the validation process, as this can impact thoroughness of reporting. For example, the purpose of the birth certicate is to collect information vital to population records and public health surveillance and resource planning. Data elements such as medical and obstetric history, pregnancy and labor/delivery complications and selected maternal and infant health outcomes are reported on the birth certicate. The purpose of in-patient hospital discharge data is much different. Data elements reported in discharge data are primarily for the purpose of reimbursement and to inform state policy. Some important risk factors for adverse maternal and infant health outcomes such as obesity, alcohol use, tobacco and so forth, may not be widely reported if they are not pertinent justication for procedures and other treatment for which reimbursement is sought. 149

PAGE 160

!There were some important limitations in the second phase of the dissertation as well. With the use of birth certicate and in-patient hospital discharge data, it was only possible to evaluate conditions for which rehospitalization was required. The in-patient data le does not contain information on other healthcare encounters such as outpatient visits or emergency room visits that did not result in rehospitalization. The populationbased data used for this phase of the dissertation is a strength in terms of generalizability of results, but is also prone to an important limitation there is the possibility of misclassication. Misclassication within the study sample can adversely impact results, as errors in case denitions and outcome measures may result in erroneous estimation of risk (Shapiro-Mendoza et al, 2008). Population-based data sources such as the birth certicate and in-patient discharge data are the result of healthcare provider management of the medical record, as well as the medical coder's interpretation of data to be submitted from the medical chart. Providers and institutions may have differing recording practices, which could potentially result in data errors (Northam, Polancich, and Restrepo, 2003; Shapiro-Mendoza et al, 2008). Some of this error has been ameliorated by the use of results generated in the rst phase of this dissertation (e.g. selection of data sources for different variables, and using linked data when possible). RECOMMENDATIONS FOR FUTURE RESEARCH At present, there is a substantial body of evidence demonstrating that increasing rates of C/S delivery are adversely impacting the health of mothers and their infants. The present study contributes to this body of evidence by providing information on a 150

PAGE 161

population of C/S deliveries that were performed without indication, and at a gestational age that has been demonstrated to have high rates of infant morbidity regardless of delivery route. C/S at late preterm is a particular public health concern, as C/S further compounds the health issues experienced by infants delivered LPT, and is quite arguably, a preventable risk. Given the ndings presented here, three future avenues of research are advocated: (1) determine the economic impact of C/S at LPT, (2) design and evaluate clinical interventions to reduce provider and institutional C/S rates, and nally (3) conduct policy analyses to determine effective strategies for creating healthcare system changes that will not only benet overall patient care, but will also reduce the rate of C/S delivery. There has been a substantial amount of research focused on the outcomes of C/S delivery, but very little attention on interventions focused on the societal, institutional and intrapersonal levels of inuence. It is important to estimate the economic impact of C/S at LPT in Florida, as understanding both the burden of morbidity and the savings that could be realized from policy focused on reducing C/S rates could elevate C/S delivery on Florida's policy agenda. Additional research is needed to provide policy-makers and other healthcare professionals with the evidence needed to create effective programs of intervention, with the ultimate goal of reducing C/S rates throughout Florida. CONCLUSION The reported association between primary C/S delivery and increasing rates of late preterm birth is a major public health concern. There is a growing body of 151

PAGE 162

evidence that C/S increases the risk for adverse maternal and infant health outcomes. Coupled with the many known health disadvantages of prematurity (e.g. respiratory distress, feeding difculties, temperature instability), understanding the implications of C/ S at late preterm is paramount for public health researchers and key stakeholders such as policy makers, healthcare providers and advocacy organizations. This dissertation demonstrates that like C/S at term delivery, C/S at late preterm is also disadvantageous for mothers and their infants. Specically, C/S contributed to excess maternal and infant morbidity and resulted in more hospitalization episodes in the rst year postpartum than vaginal deliveries, likely translating into increased healthcare costs. Given the current economic landscape in the United States, as well as the State of Florida, reducing C/S without medical indication would be a viable target for improvement in the health of Floridians, but also may represent substantial savings in terms of healthcare dollars. 152

PAGE 163

LITERATURE CITED Ahern, J., Pickett, K. E., Selvin, S., & Abrams, B. (2003). Preterm birth among African American and white women: a multilevel analysis of socioeconomic characteristics and cigarette smoking. Journal of Epidemiology and Community Health, 57 (8), 606-611. Alexander, G. R., & Cornely, D. A. (1987). Prenatal care utilization: its measurement and relationship to pregnancy outcome. American Journal of Preventive Medicine, 3(5), 243-253. Alexander, G. R., & Kotelchuck, M. (1996). Quantifying the adequancy of prental care: a comparison of indices. Public Health Reports, 111, 408-418. Alexander, J. M., Leveno, K. J., Landon, M. B., Thom, E., Spong, C. Y., Varner, M.W., et al. (2006). Fetal injury associated with cesarean delivery. Obstetrics & Gynecology, 108(4), 885-90. Allen, V. M., O'Connell, C. M., Liston, R. M., & Baskett, T. F. (2003). Maternal morbidity associated with cesarean delivery without labor compared with spontaneous onset of labor at term. Obstetrics & Gynecology, 102 (3), 477-482. Althabe, F., & Belizan, J. M. (2006). Caesarean section: the paradox. Lancet, 368 (9546), 1472-1473. Althabe, F., Sosa, C., Belizan, J. M., Gibbons, L., Jacquerioz, F., & Bergel, E. (2006). Cesarean section rates and maternal and neonatal mortality in low-, medium-, and high-income countries: an ecological study. Birth, 33 (4), 270-277. American Academy of Pediatrics. Committee on Fetus and Newborn. (2010). Hospital stay for healthy term newborns. Pediatrics, 125(2), 405-9. American College of Obstetrics and Gynecology. (2007). Committee Opinion No. 394, December 2007. Cesarean delivery on maternal request. Obstetrics & Gynecology, 110 (6), 1501. 153

PAGE 164

American College of Obstetrics and Gynecology (2008). The Future of Ob-Gyn Practice. ACOG Today. Amini, S. B., Catalano, P. M., Dierker, L. J., & Mann, L. I. (1996). Births to teenagers: trends and obstetric outcomes. Obstet Gynecol, 87 (5 Pt 1), 668-674. Ananth, C. V., Joseph, K. S., Oyelese, Y., Demissie, K., & Vintzileos, A. M. (2005). Trends in preterm birth and perinatal mortality among singletons: United States, 1989 through 2000. Obstetrics & Gynecology, 105 (5 pt 1), 1084-1091. Aron, D. C., Gordon, H. S., DiGuiseppe, D. L., Harper, D. L., & Rosenthal, G.E. (2000). Variations in risk-adjusted cesarean delivery rates according to race and health insurance. Medical Care, 38 (1), 35-44. Bailey, P. E. (2005). The disappearing art of instrumental delivery: time to reverse the trend. International Journal of Gynaecology Obstetrics, 91(1) ,89-96. Bailit, J. L., Dooley, S. L., & Peaceman, A. N. (1999). Risk adjustment for interhospital comparison of primary cesarean rates. Obstetrics & Gynecology, 93 (6), 1025-1030. Barros, F. C., & Velez Mdel, P. (2006). Temporal trends of preterm birth subtypes and neonatal outcomes. Obstetrics & Gynecology, 107(5) ,1035-41. Belizan, J. M., Althabe, F., & Cafferata, M. L. (2007). Health consequences of the increasing caesarean section rates. Epidemiology, 18 (4), 485-486. Belizan, J. M., Cafferata, M. L., Althabe, F., & Buekens, P. (2006). Risks of patient choice cesarean. Birth, 33 (2), 167-169. Bettegowda, V. R., Dias, T., Davidoff, M. J., Damus, K., Callaghan, W. M., & Petrini, J. R. (2008). The relationship between cesarean delivery and gestational age among US singleton births. Clinics in Perinatology, 35 (2), 309-323, v-vi. Blumenthal, D. (2010). Launching HITECH. New England Journal of Medicine, 362, 382. 154

PAGE 165

Branum, A. M., & Schoendorf, K. C. (2005). The inuence of maternal age on very preterm birth of twins: differential effects by parity. Paediatrics Perinatal Epidemiology, 19 (5), 399-404. Brown, H. S. (2007). Lawsuit activity, defensive medicine, and small area variation: the case of Cesarean sections revisited. Health Economics Policy Law, 2 (Pt 3), 285-296. Burgos, A. E., Schmitt, S. K., Stevenson, D. K., & Phibbs, C. S. (2008). Readmission for neonatal jaundice in California, 1991-2000: trends and implications. Pediatrics, 121 (4), e864-869. Callaghan, W. M., MacKay, A. P., Berg, C. J. (2008). Identication of severe maternal morbidity during delivery hospitalizations, United States, 1991-2003. American Journal of Obstetrics and Gynecology, 199 (133), e1-133.e138. Centers for Disease Control and Prevention. (2005). QuickStats: Total and Primary Cesarean Rate and Vaginal Birth After Previous Cesarean (VBAC) Rate --United States, 1989--2003. Morbidity and Mortality Weekly Report, 54 (2), 46. Centers for Disease Control and Prevention. (2009). Healthy People 2010 Operational Denition Retrieved April 9, 2009, from http://ftp.cdc.gov/pub/health_statistics/ nchs/Datasets/DATA2010/Focusarea16/O1609a.pdf Chaillet, N., Dube, E., Dugas, M., Francoeur, D., Dube, J., Gagnon, S., et al. (2007). Identifying barriers and facilitators towards implementing guidelines to reduce caesarean section rates in Quebec. Bull World Health Organization, 85(10), 791-7. Clements, K. M., Bareld, W. D., Femi Ayadi, M., Wilber, N. (2007). Preterm birthassociated cost of early intervention services: an analysis by gestational age. Pediatrics, 119 e866-e874. Cnattingius, S., Hultman, C. M., Dahl, M., & Sparen, P. (1999). Very preterm birth, birth trauma, and the risk of anorexia nervosa among girls. Archives General Psychiatry, 56 (7), 634-638. 155

PAGE 166

Cohen, J. R. (2005). Patient satisfaction with the prenatal care provider and the risk of cesarean delivery. American Journal of Obstetrics Gynecology, 192 (6), 2029-2032; discussion 2032-2024. Coleman, V. H., Lawrence, H., & Schulkin, J. (2009). Rising cesarean delivery rates: the impact of cesarean delivery on maternal request. Obstetrics Gynecologic Survey, 64 (2), 115-119. Cragin, E. (1916). Conservatism in obstetrics. New York State Journal of Medicine, 104 (1). American Journal of Obstetrics Gynecology, 194(4), 932-6. Cragin L., & Kennedy, H. P. (2006). Linking obstetric and midwifery practice with optimal outcomes. Journal of Obstetrics Gynecology Neonatal Nursing. 35(6) 779-85. Cyr, R. M. (2006). Myth of the ideal cesarean section rate: commentary and historical perspective. American Journal of Obstetrics & Gynecology, 194(4), 932-936. Dahlen, H. G., Barclay, L. M., & Homer, C. (2008). Preparing for the rst birth: mothers' experiences at home and in hospital in australia. Journal of Perinatal Education, 17(4), 21-32. Damus, K. (2008). Prevention of preterm birth: a renewed national priority. Current Opinion Obstetrics & Gynecology, 20 (6), 590-596. Darwin, C. R. (1871). The Decent of Man and Selection in Relation to Sex (1st ed. Vol. 1). London: John Murray. Davidoff, M. J., Dias, T., Damus, K., Russell, R., Bettegowda, V. R., Dolan, S., & Schwarz, R. H. (2006). Changes in the gestational age distribution among U.S. singleton births: impact on rates of late preterm birth, 1992 to 2002. Seminars in Perinatology, 30 8-15. de Almeida, M. F., Guinsburg, R., da Costa, J. O., Anchieta, L. M., Freire, L. M., & Junior, D. C. (2007). Resuscitative procedures at birth in late preterm infants. Journal of Perinatology, 27 (12), 761-765. 156

PAGE 167

De Luca, R., Boulvain, M., Irion, O., Berner, M., & Pster, R. E. (2009). Incidence of early neonatal mortality and morbidity after late-preterm and term cesarean delivery. Pediatrics, 123 ,e1064-e1071. Declercq, E., Barger, M., Cabral, H. J., Evans, S. R., Kotelchuck, M., Simon, C., et al. (2007). Maternal outcomes associated with planned primary cesarean births compared with planned vaginal births. Obstetrics & Gynecology, 109 (3), 669-677. Declercq, E., Cunningham, D. K., Johnson, C., & Sakala, C. (2008). Mothers' reports of postpartum pain associated with vaginal and cesarean deliveries: results of a national survey. Birth, 35 (1), 16-24. DeFranco, E. A., Lian, M., Muglia, L. A., & Schootman, M. (2008). Area-level poverty and preterm birth risk: a population-based multilevel analysis. BMC Public Health, 8 316. Derrick, M., Luo, N. L., Bregman, J. C., Jilling, T., Ji, X., Fisher, K., et al. (2004). Preterm fetal hypoxia-ischemia causes hypertonia and motor decits in the neonatal rabbit: a model for human cerebral palsy? Journal of Neuroscience, 24 (1), 24-34. DiGiuseppe, D. L., Aron, D.C., Ranbom, L., Harper, D. L., & Rosenthal, G.E. (2002). Reliability of birth certicate data: a multi-hospital comparison to medical records information. Maternal and Child Health Journal, 6 (3), 169-179. DiMaio, H., Edwards, R. K., Euliano, T. Y., Treloar, R. W., & Cruz, A. C. (2002). Vaginal birth after cesarean delivery: an historic cohort cost analysis. American Journal of Obstetrics and Gynecology, 186 890-892. Doherty, D. A., Magann, E. F., Chauhan, S. P., O'Boyle, A. L., Busch, J. M., & Morrison, J. C. (2008). Factors affecting caesarean operative time and the effect of operative time on pregnancy outcomes. Australia New Zealand Journal of Obstetrics and Gynaecology, 48 (3), 286-291. Druzin, M. L., & El-Sayed, Y. Y. (2006). Cesarean delivery on maternal request: wise use of nite resources? A view from the trenches. Seminars in Perinatology, 30 305-308. 157

PAGE 168

Engle, W. A. (2006). A recommendation for the denition of "late preterm" (near-term) and the birth weight-gestational age classication system. Seminars in Perinatology, 30 (1), 2-7. Engle, W. A., & Kominiarek, M. A. (2008). Late preterm infants, early term infants, and timing of elective deliveries. Clinics in Perinatology, 35 (2), 325-341, vi. Engle, W. A., Tomashek, K. M., Wallman, C., and the Committee on the Fetus and Newborn (2008). "late-preterm" infants: a population at risk. Pediatrics, 120 1390-1401. Ennen, C. S., Boll, J. A., Magann, E. F., Bass, J. D., Chauhan, S. P., & Morrison, J. C. (2009). Risk factors for cesarean delivery in preterm, term and post-term patients undergoing induction of labor with an unfavorable cervix. Gynecology Obstetric Invesigations, 67 (2), 113-117. Escobar, G. J., Clark, R. H., & Greene, J. D. (2006). Short-term outcomes of infants born at 35 and 36 weeks gestation: we need to ask more questions. Seminars in Perinatology, 30 (1), 28-33. Escobar, G. J., Greene, J. D., Hulac, P., Kincannon, E., Bischoff, K., Gardner, M. N., et al. (2005). Rehospitalisation after birth hospitalisation: patterns among infants of all gestations. Archives of Disease in Childhood, 90 (2), 125-131. Farley, T. A., Mason, K., Rice, J., Habel, J. D., Scribner, R., & Cohen, D. A. (2006). The relationship between the neighbourhood environment and adverse birth outcomes. Paediatric Perinatal Epidemiology, 20 (3), 188-200. Fuchs, K., & Wapner, R. (2006). Elective cesarean section and induction and their impact on late preterm births. Clinics in Perinatology, 33 793-801. Gamble, J., Creedy, D. K., McCourt, C., Weaver, J., & Beake, S. (2007). A critique of the literature on women's request for cesarean section. Birth, 34 (4), 331-340. Geller, S. E., Rosenberg, D., Cox, S. M., & Kilpatrick, S. (2002). Dening a conceptual framework for near-miss maternal morbidity. Journal of the American Medical Womens Association, 57 (3), 135-139. 158

PAGE 169

Gilliam, M. (2006). Cesarean delivery on request: reproductive consequences. Seminars in Perinatology, 30 (5), 257-260. Glazener, C. M., Abdalla, M., Stroud, P., Naji, S., Templeton, A., & Russell, I. T. (1995). Postnatal maternal morbidity: extent, causes, prevention and treatment. British Journal of Obstetrics and Gynaecology, 102 (4), 282-287. Goetzl L., Rivers J., Evans T., Citron D. R., Richardson B. E., Lieberman E. et al. (2004). Prophylactic acetaminophen does not prevent epidural fever in nulliparous women: a double-blind placebo-controlled trial. Journal of Perinatology. 24(8), 471-5. Goldenberg, R. L., Iams, J. D., Mercer, B. M., Meis, P. J., Moawad, A. H., Copper, R. L., et al. (1998). The preterm prediction study: the value of new vs standard risk factors in predicting early and all spontaneous preterm births. NICHD MFMU Network. American Journal of Public Health, 88 (2), 233-238. Goodman, D, Sappeneld, W., & Thompson, D. (2007). Preterm Birth and Intrapartum Intervention in Florida: What is the Connection? Florida Department of Health. Goodman, D., Sappeneld, W. M., Mahan, C. S., & Kogan, M. D. (2010). Intrapartum !intervention and early delivery among women with low medical risk (Submitted !for Publication, 2010). Grant, D. (2005). Explaining source of payment differences in U.S. cesarean rates: why do privately insured mothers receive more cesareans than mothers who are not privately insured? Health Care Management Science, 8 5-17. Gregory, K. D., Korst, L. M., Gornbein, J. A., & Platt, L. D. (2002). Using administrative data to identify indications for elective primary cesarean delivery. Health Service Research, 37 (5), 1387-1401. Gregory, K. D., Korst, L. M., & Platt, L.D. (2001). Variation in elective primary cesarean delivery by patient and hospital factors. American Journal of Obstetrics and Gynecology, 184 1521-1534. Hamilton, B., Martin, J. A., & Ventura, S. J. (2009). Births: Preliminary Data for 2007. National Vital Statistics Reports, 57 (12), 1-23. 159

PAGE 170

Hamilton, B. E., Minino, A. M., Martin, J. A., Kochanek, K. D., Strobino, D. M., & Guyer, B. (2007). Annual summary of vital statistics: 2005. Pediatrics, 119 (2), 345-360. Hanley, J. A., Negassa, A., Edwardes, M. D., & Forrester, J. E. (2003). Statistical analysis of correlated data using generalized estimating equations: an orientation. American Journal of Epidemiology, 157 (4), 364-375. Hansen, A. K., Wisborg, K., Uldbjerg, N., & Henriksen, T. B. (2007). Elective caesarean section and respiratory morbidity in the term and near-term neonate. Acta Obstetrics and Gynecology Scandinavia, 86 (4), 389-394. Hansen, A. K., Wisborg, K., Uldbjerg, N., & Henriksen, T. B. (2008). Risk of respiratory morbidity in term infants delivered by elective caesarean section: cohort study. British Medical Journal, 336 (7635), 85-87. Hellerstedt, W. L., Himes, J. H., Story, M., Alton, I. R., & Edwards, L. E. (1997). The effects of cigarette smoking and gestational weight change on birth outcomes in obese and normal-weight women. American Journal of Public Health, 87 (4), 591-596. Herrchen, B., Gould, J. B., & Nesbitt, T. S. (1997). Vital statistics linked birth/infant death and hospital discharge record linkage for epidemiological studies. Computers and Biomedical Research, 30 290-305. Holzman, C., & Paneth, N. (1994). Maternal cocaine use during pregnancy and perinatal outcomes. Epidemiology Review, 16 (2), 315-334. Hsu, C., Shieh, G., Wu, C., Shen, H., & Tang, C. (2006). Risk adjustment for interhospital comparisons of caesarean section rates in Taipei municipal hospitals. European Journal of Obestetrics & Gynecology and Reproductive Biology, 127 190-197. Institute of Medicine (2002). The Future of the Public's Health in the 21st Century Washington, D.C.: National Academies. Institute of Medicine (2007). Preterm birth: causes, consequences and prevention Washington, D.C. : The National Academies Press. 160

PAGE 171

Israel, B. A., Parker, E. A., Rowe, Z., Salvatore, A., Minkler, M., Lopez, J., et al. (2005). Community-based participatory research: lessons learned from the Centers for Children's Environmental Health and Disease Prevention Research. Environmental Health Perspectives, 113 (10), 1463-1471. Israel, B. A., Schulz, A. J., Parker, E. A., & Becker, A. B. (1998). Review of communitybased research: assessing partnership approaches to improve public health. Annual Review Public Health, 19 173-202. Jaffee, K. D., & Perloff, J.D. (2003). An ecological analysis of racial differences in low birthweight: implications for maternal and child health social work. Health and Social Work, 28 9-22. Janssen, P. A., Ryan, E. M., Etches, D. J., Klein, M. C., & Reime, B. (2007). Outcomes of planned hospital birth attended by midwives compared with physicians in British Columbia. Birth, 34(2), 140-7. The Joint Commission. (2010). Specications for early medically indicated delivery. http://www.jointcommission.org /; accessed on March 22, 2010. Joseph, K. S., Young, D. C., Dodds, L., O'Connell, C. M., Allen, V. M., Chandra, S., et al. (2003). Changes in maternal characteristics and obstetric practice and recent increases in primary cesarean delivery. Obstetrics & Gynecology, 102 (4), 791-800. Kahn, E. B., Berg, C. J., & Callaghan, M. W. (2009). Cesarean delivery among women with low risk pregnancies. Obstetrics & Gynecology, 113 33-40. Kahn, M. G., & Ranade, D. (2010). The impact of electronic medical records data sources on an adverse drug event quality measure. Journal of the American Medical Informatics Association, 17(2) ,185-91. Kaufman, J. S., Alonso, F. T., & Pino, P. (2008). Multi-level modeling of social factors and preterm delivery in Santiago de Chile. British Medical Journal: Pregnancy Childbirth, 8 46. Kazandjian, V. A., Chaulk, C.P., Ogunbo, S., & Wicker, K. (2005). Does a cesarean section delivery always cost more than a vaginal delivery? Journal of Evaluation in Clinical Practice, 13 16-20. 161

PAGE 172

Kesmodel, U., Olsen, S. F., & Secher, N. J. (2000). Does alcohol increase the risk of preterm delivery? Epidemiology, 11 (5), 512-518. Kirkby, S., Greenspan, J. S., Kornhouser, M., & Schneiderman, R. (2007). Clinical outcomes and cost of the moderately preterm infant. Advances in Neonatal Care, 7 (2), 80-87. Klein, E. G., Lytle, L. A., & Chen, V. (2008). Social ecological predictors of the transition to overweight in youth: results from the Teens Eating for Energy and Nutrition at Schools (TEENS) study. Journal of the American Dietetic Association, 108 (7), 1163-1169. Klein, M. (2006). Not safer, not cheaper. Canadian Medical Association Journal, 175 (10), 1243. Ko, Y. L. (2003). Fatigue during pregnancy predicts caesarean deliveries. Journal of Advanced Nursing, 45 (5), 487-494. Kogan, M. D., Martin, J. A., Alexander, G. R., Kotelchuck, M., Ventura S. J., & Frigoletto, F. D. (1998). The changing pattern of prenatal care utilization in the United States, 1981-1995, using different prenatal care indices. Journal of the American Medical Association, 279(20),1623-8. Kohler Reissman, C. (1992). Women and medicalization: a new perspective. In G. Kirkup, Smith Keller, L. (Ed.), Inventing Women: Science, Technology and Gender (pp. 123-144.). Cambridge: Polity Press. Korst, L. M., Gregory, K.D., & Gornbein, J.A. (2004). Elective primary cesarean delivery: accuracy of administrative data. Paediatric and Perinatal Epidemiology, 18 112-119. Kramer, M. S., Demissie, K., Yang, H., Platt, R. W., Sauve, R., & Liston, R. (2000). The contribution of mild and moderate preterm birth to infant mortality. Fetal and Infant Health Study Group of the Canadian Perinatal Surveillance System. Journal of the American Medical Association, 284 (7), 843-849. Kuklina, E. V., Meikle, S. F., Jamieson, D. J., Whiteman, M. K., Bareld, W. D., Hillis, S. D., et al. (2009). Severe obstetric morbidity in the United States: 1998-2005. Obstetrics & Gynecology, 113 (2 Pt 1), 293-299. 162

PAGE 173

Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33 (1), 159-174. Lang, J. M., Lieberman, E., & Cohen, A. (1996). A comparison of risk factors for preterm labor and term small-for-gestational-age birth. Epidemiology, 7 (4), 369-376. Le Ray, C., Carayol, M., Zeitlin, J., Breart, G., & Gofnet, F. (2006). Level of perinatal care of the maternity unit and rate of cesarean in low-risk nulliparas. Obstetrics & Gynecology, 107 (6), 1269-1277. Lee, Y. M., & D'Alton, M. E. (2008a). Cesarean delivery on maternal request: maternal and neonatal complications. Current Opinion Obstetrics and Gynecology, 20 (6), 597-601. Lee, Y. M., & D'Alton, M. E. (2008b). Cesarean delivery on maternal request: the impact on mother and newborn. Clinics in Perinatology, 35 (3), 505-518, x. Leung, G. M., Ho, L. M., Tin, K. Y., Schooling, C. M., & Lam, T. H. (2007). Health care consequences of cesarean birth during the rst 18 months of life. Epidemiology, 18 (4), 479-484. Levine, E. M., Ghai, V., Barton, J. J., & Strom, C. M. (2001). Mode of delivery and risk of respiratory diseases in newborns. Obstetrics & Gynecology, 97 (3), 439-442. Liu, S., Heaman, M., Joseph, K. S., Liston, R. M., Huang, L., Sauve, R., et al. (2005). Risk of maternal postpartum readmission associated with mode of delivery. Obstetrics & Gynecology, 105 (4), 836-842. Liu, S., Liston, R. M., Joseph, K. S., Heaman, M., Sauve, R., & Kramer, M. S. (2007). Maternal mortality and severe morbidity associated with low-risk planned cesarean delivery versus planned vaginal delivery at term. Canadian Medical Association Journal, 176 (4), 455-460. Liu, T. C., Chen, C. S., & Lin, H. C. (2008). Does elective caesarean section increase utilization of postpartum maternal medical care? Medical Care, 46 (4), 440-443. Lundsberg, L. S., Bracken, M. B., & Saftlas, A. F. (1997). Low-to-moderate gestational alcohol use and intrauterine growth retardation, low birthweight, and preterm delivery. Annals of Epidemiology, 7 (7), 498-508. 163

PAGE 174

Luo, Z. C., Kierans, W. J., Wilkins, R., Liston, R. M., Mohamed, J., & Kramer, M. S. (2004). Disparities in birth outcomes by neighborhood income: temporal trends in rural and urban areas, british columbia. Epidemiology, 15 (6), 679-686. Lurio, J., Morrison, F. P., Pichardo, M., Berg, R., Buck, M. D., Wu, M., Kitson, K., Mostashari, F., & Calman, N. (2010). Journal of the American Medical Informatics Association, 17(2), 217-9. Lydon-Rochelle, M., Holt, V. L., Martin, D. P., & Easterling, T. R. (2000). Association between method of delivery and maternal rehospitalization. Journal of the American Medical Association, 283 (18), 2411-2416. Lydon-Rochelle, M. T., Holt, V.L., Cardenas, V., Nelson, J.C., Easterling, T.R., Gardella, C., Callaghan, W.M. (2005). The reporting of pre-existing maternal medical conditions and complications of pregnancy on birth certicates and in hospital discharge data. American Journal of Obstetrics and Gynecology, 193 125-134. MacDorman, M. F., Declercq, E., Menacker, F., & Malloy, M.H. (2008). Neonatal mortality for primary cesarean and vaginal births to low-risk women: application of an "intention-to-treat" model. Birth, 35(1), 3-8. Malloy, M. H. (2009). Impact of cesarean section on intermediate and late preterm births: United States, 2000-2003. Birth, 36(1), 26-33. MacDorman, M. F., Mathews, T. J., Martin, J. A., & Malloy, M. H. (2002). Trends and characteristics of induced labour in the United States, 1989 -1998. Paediatric and Perinatal Epidemiology, 16 263-273. Malott, A. M., Davis, B. M., McDonald, H., & Hutton, E. (2009). Midwifery care in eight industrial countries: how does Canadian midwifery compare? Journal of Obstetrics and Gynaecology Canada, 31(10) ,974-9. Mancuso, A., De Vivo, A., Fanara, G., Albiero, A., Priolo, A. M., Giacobbe, A., et al. (2008). Caesarean section on request: are there loco-regional factors inuencing maternal choice? An Italian experience. Journal of Obstetrics & Gynaecology, 28 (4), 382-385. 164

PAGE 175

March of Dimes (2010). Late preterm brain development card. http:// www.marchofdimes.com/catalog/product.aspx?productcode=37-2229-07 Accessed on March, 19, 2010. Masi, C. M., Hawkley, L. C., Piotrowski, Z. H., & Pickett, K. E. (2007). Neighborhood economic disadvantage, violent crime, group density, and pregnancy outcomes in a diverse, urban population. Social Science & Medicine, 65 (12), 2440-2457. Mathews, T. J., & MacDorman, M. F. (2007). Infant mortality statistics from 2004 period linked birth/infant death data set. National Vital Statistics Reports, 55 (14), 1-33. Melamed, N., Klinger, G., Tenenbaum-Gavish, K., Herscovici, T., Linder, N., Hod, M. et al. (2009). Short-term neonatal outcomes in low-risk spontaneous, singleton, late preterm deliveries. Obstetrics & Gynecology, 114(2), 253-260. McIntire, D. D., & Leveno, K. J. (2008). Neonatal mortality and morbidity rates in late preterm births compared with births at term. Obstetrics & Gynecology, 111 (1), 35-41. McLaurin, K. K., Hall, C. B., Jackson, E. A., Owens, O. V., & Mahadevia, P. J. (2009). Persistence of morbidity and cost differences between late-preterm and term infants during the rst year of life. Pediatrics, 123 (2), 653-659. McLeroy, K. R., Bibeau, D., Steckler, A., & Glanz, K. (1988). An ecological perspective on health promotion programs. Health Education Quarterly, 15 (4), 351-377. Melamed N., Klinger G., Tenenbaum-Gavish K., Herscovici T., Linder N., Hod M., et al, (2009). Short-term neonatal outcome in low-risk, spontaneous, singleton, late preterm deliveries. Obstetrics & Gynecology, 114(2 pt 1.),253-60. Menacker, F., Declercq, E., & Macdorman, M. F. (2006). Cesarean delivery: background, trends, and epidemiology. Seminars in Perinatology, 30 (5), 235-241. Messer, L. C., Kaufman, J. S., Mendola, P., & Laraia, B. A. (2008). Black-white preterm birth disparity: a marker of inequality. Annals of Epidemiology, 18 (11), 851-858. Midwifery Task Force. (2008). The Midwives Model of Care. http://cfmidwifery.org/ mmoc/dene.aspx Accessed on March 19, 2010. 165

PAGE 176

Miesnik, S. R., & Reale, B. J. (2007). A review of issues surrounding medically elective cesarean delivery. Journal of Obstetrics and Gynecologic Neonatal Nursing, 36 (6), 605-615. Milne, J., Gafni, A., Lu, D., Wood, S., Sauve, R., & Ross, S. (2009). Developing and pre-testing a decision board to facilitate informed choice about delivery approach in uncomplicated pregnancy. British Medical Journal: Pregnancy Childbirth, 9,50. Morse, S.B., Zheng, H, Tang, Y, Roth, J. (2009). Early school-age outcomes of late preterm births. Pediatrics 123(4),e622-9. Murray, C. J. L., & Frenk J. (2010). Ranking 37th Measuring the Performance of the U.S. Healthcare System. New England Journal of Medicine, 362, 98-99. Murthy, K., Grobman, W. A., Lee, T. A., & Holl, J. L. (2009). Obstetricians' rising liability insurance premiums and inductions at late preterm gestations. Medical Care, 47(4), 425-30. National Institutes of Health (2006). Cesarean Delivery on Maternal Request NIH Stateof-the-Science Conference Statement on. NIH Consensus and State-of-theScience Statements, 23 (1). Nitz, K. (1999). Adolescent pregnancy prevention: a review of interventions and programs. Clinical Psychology Review, 19 (4), 457-471. Nkansah-Amankra, S., Luchok, K .J., Hussey, J. R., Watkins, K., & Liu, X. (2009). Effects of maternal stress on low birthweight and preterm birth outcomes across neighborhoods of South Carolina, 2000-2003. Maternal and Child Health Journal, Epub Ahead of Print Northham, S., Polancich, S, & Restrepo, E. (2003). Birth certicate methods in ve hospitals. Public Health Nursing, 20(4), 318-27. O'Campo, P., Burke J. G., Culhane J., Elo, I. T., Eyster J., Holzman C., et al. (2008). Neighborhood deprivation and preterm birth among non-Hispanic Black and White women in Eight geographic areas in the United States. American Journal of Epidemiology, 167(2),155-63. 166

PAGE 177

Ophir, E., Strulov, A., Solt, I., Michlin, R., Buryanov, I., & Bornstein, J. (2008). Delivery mode and maternal rehospitalization. Archives of Gynecology & Obstetrics, 277 (5), 401-404. Palencia, R., Gafni, A., Hannah, M. E., Ross, S., Willan, A. R., Hewson, S., & McKay, D. (2006). The costs of planned cesarean versus planned vaginal birth in the Term Breech Trial. Canadian Medical Association Journal, 174 (8), 1109-1113. Pallasmaa, N., Ekblad, U., & Gissler, M. (2008). Severe maternal morbidity and the mode of delivery. Acta Obstetrics and Gynecology Scandinavia, 87(6),662-8. Pare, E., Quinones, J. N., & Macones, G. A. (2006). Vaginal birth after caesarean section versus elective repeat caesarean section: assessment of maternal downstream health outcomes. British Journal of Obstetrics and Gynecology, 113 (1), 75-85. Park, S., Sappeneld, W. M., Bish, C., Bensyl, D. M., Goodman, D., & Menges, J. (2009). Reliability and validity of birth certicate prepregnancy weight and height among women enrolled in prenatal WIC program: Florida, 2005. Maternal and Child Health Journal, E-pub ahead of print. Parry, D. C. (2008). "We wanted a birth experience, not a medical experience": exploring Canadian women's use of midwifery. Health Care for Women International, 29(8), 784-806. Pedan, A. (2001). Analysis of count data using the SAS system. Statistics, Data Analysis and Data Mining, Tewksbury, M.A. Petrini, J. R., Dias, T., McCormick, M. C., Massolo, M. L., Green, N. S., & Escobar, G. J. (2009). Increased risk of adverse neurological development for late preterm infants. Journal of Pediatrics, 154(2),169-76. Petrou, S. (2003a). Economic consequences of preterm birth and low birthweight. British Journal of Obstetrics and Gynaecology, 110 Suppl 20 17-23. Petrou, S. (2003b). Economic consequences of preterm birth and low birthweight. International Journal of Obstetrics and Gynaecology, 110 (Supp 20), 17-23. 167

PAGE 178

Petrou, S., Mehta, Z., Hockley, C., Cook-Mozaffari, P., Henderson, J., & Goldacre, M. (2003). The impact of preterm birth on hospital inpatient admissions and costs during the rst 5 years of life. Pediatrics, 112 1290-1297. Petrou, S. (2005). The economic consequences of preterm birth during the rst 10 years of life. British Journal of Obstetrics and Gynaecology, 112 Suppl 1 10-15. Piper, J. M., Mitchel, E. F., Jr., Snowden, M., Hall, C., Adams, M., & Taylor, P. (1993). Validation of 1989 Tennessee birth certicates using maternal and newborn hospital records. American Journal of Epidemiology, 137 (7), 758-768. Prochaska, J. O. (2008). Decision making in the transtheoretical model of behavior change. Medical Decision Making, 28 (6), 845-849. Prochaska, J. O., & DiClemente, C. C. (1983). Stages and processes of self-change of smoking: toward an integrative model of change. Journal of Consulting and Clinical Psychology, 51 (3), 390-395. Quinn, L. A., Thompson, S. J., & Ott, M. K. (2005). Application of the social ecological model in folic acid public health initiatives. Journal of Obstetrics Gynecology and Neonatal Nursing, 34 (6), 672-681. Raatikainen, K., Heiskanen, N., & Heinonen, S. (2005). Marriage still protects pregnancy. British Journal of Obstetrics and Gynaecology, 112 (10), 1411-1416. Raju, T. N., Higgins, R. D., Stark, A. R., & Leveno, K. J. (2006). Optimizing care and outcome for late-preterm (near-term) infants: a summary of the workshop sponsored by the National Institute of Child Health and Human Development. Pediatrics, 118 (3), 1207-1214. Rimer, B., & Glanz, K. (2005). Theory at a Glance: A guide for health promotion practice. In N. C. Institute (Ed.), US Department of Health and Human Services Washington, D.C.: National Institutes of Health. Roberts, C. L., Bell, J. C., Ford, J. B., & Morris, J. M. (2008). Monitoring the quality of maternity care: how well ar labour and delivery events reported in population health data? Paediatric and Perinatal Epidemiology, 23 144-152. 168

PAGE 179

Robson, S., Carey, A., Mishra, R., & Dear, K. (2008). Elective caesarean delivery at maternal request: a preliminary study of motivations inuencing women's decision-making. Australia and New Zealand Journal of Obstetrics and Gynaecology, 48 (4), 415-420. Roohan, P. J., Josberger, R. E., Acar, J., Dabir, P., Feder, H. M., & Gagliano, P. J. (2003). Validation of birth certicate data in New York State. Journal of Community Health, 28 (5), 335-346. Russell, R. B., Green, N. S., Steiner, C. A., Meikle, S., Howse, J. L., Poschman, K., et al. (2007). Cost of hospitalization for preterm and low birth weight infants in the United States. Pediatrics, 120 (1), e1-9. Sarkin, J. A., Johnson, S. S., Prochaska, J. O., & Prochaska, J. M. (2001). Applying the transtheoretical model to regular moderate exercise in an overweight population: validation of a stages of change measure. Preventive Medicine, 33 (5), 462-469. Savitz, D. A., Dole, N., Terry, J. W., Jr., Zhou, H., & Thorp, J. M., Jr. (2001). Smoking and pregnancy outcome among African-American and white women in central North Carolina. Epidemiology, 12 (6), 636-642. Shapiro-Mendoza, C. K., Tomashek, K. M., Kotelchuck, M., Bareld, W., Weiss, J., & Evans, S. (2006). Risk factors for neonatal morbidity and mortality among "healthy," late preterm newborns. Seminars in Perinatology, 30 (2), 54-60. Shapiro-Mendoza C. K., Tomashek K. M., Kotelchuck M., Bareld W., Nannini A., !Weiss, J. et al. (2008). Effect of late-preterm birth and maternal medical !conditions on newborn morbidity risk. Pediatrics, 121(2),e223-32.. Sheiner, E., Levy, A., Tehillah, S., Silverberg, D., Katz, M., & Mazor, M. (2004). Maternal obesity as an independent risk factor for caesarean delivery. Paediatric and Perinatal Epidemiology, 18 (3), 196-201. Silver, R. M., Landon, M. B., Rouse, D. J., Leveno, K. J., Spong, C. Y., Thom, E. A., et al. (2006). Maternal morbidity associated with multiple repeat cesarean deliveries. Obstetrics & Gynecology, 107 (6), 1226-1232. Sousa, M. H., Cecatti, J. G., Hardy, E. E., Serruya, S. J. (2008). Severe maternal morbidity (near miss) as a sentinel event of maternal death. An attempt to use routine data for surveillance. Reproductive Health, 5,6. 169

PAGE 180

Spong, C. Y., Landon, M. B., Gilbert, S., Rouse, D. J., Leveno, K. J., Varner, M. W., et al. (2007). Risk of uterine rupture and adverse perinatal outcome at term after cesarean delivery. Obstetrics & Gynecology, 110 (4), 801-807. Stone, P. W., Zwanziger, J., Hinton, W., & Buenting, J. (2000). Economic analysis of two models of low-risk maternity care: a freestanding birth center compared to traditional care. Research Nursing Health, 23(4),279-89. Swensen, S. J., Meyer, G. S., Nelson, E. C., Hungt, G. C. Jr., Pryor, D. B., Weissberg et al, (2010). From Cottage Industry to Postindustrial Care The Revolution in Health Care Delivery. New England Journal of Medicine, 362, e12. Tashakkori, A. T. (1998). Mixed Methodology: Combining Qualitative and Quantitative Approaches Thousand Oaks, CA: Sage Publications, Inc. Tita, A. T., Landon, M. B., Spong, C. Y., Lai, Y., Leveno, K. J., Varner, M. W., et al. (2009). Timing of elective repeat cesarean delivery at term and neonatal outcomes. New England Journal of Medicine, 360(2), 111-20. Tomashek, K. M., Shapiro-Mendoza, C. K., Weiss, J., Kotelchuck, M., Bareld, W., Evans, S., et al. (2006). Early discharge among late preterm and term newborns and risk of neonatal morbidity. Seminars in Perinatology, 30 (2), 61-68. Tomashek, K. M., Shapiro-Mendoza, C. K., Davidoff, M. J., & Petrini, J. R. (2007). Differences in mortality between late-preterm and term singleton infants in the United States, 1995-2002. The Journal of Pediatrics, 151 450-460. Torres, G. (1986). Theoretical Foundations of Nursing Norwalk, Conn: Apple-CenturyCrofts. Underwood, M. A., Danielsen, B., Gilbert, W. M. (2007a). Cost, causes and rates of rehospitalization of preterm infants. Journal of Perinatology, 27 (10), 614-619. Urquia, M. L., Frank, J. W., Glazier, R. H., Moineddin, R., Matheson, F. I., & Gagnon, A. J. (2009). Neighborhood context and infant birthweight among recent immigrant mothers: a multilevel analysis. American Journal of Public Health, 99 (2), 285-293. 170

PAGE 181

van den Berg, A., van Elburg, R. M., van Geijn, H. P., & Fetter, W. P. (2001). Neonatal respiratory morbidity following elective caesarean section in term infants. A 5year retrospective study and a review of the literature. European Journal of Obstet rics and Gynecology and Reproductive Biology, 98 (1), 9-13. Villar, J., Carroli, G., Zavaleta, N., Donner, A., Wojdyla, D., Faundes, A., et al. (2007). Maternal and neonatal individual risks and benets associated with caesarean delivery: multicentre prospective study. British Medical Journal, 335 (7628), 1025. Wang, M. L., Dorer, D. J., Fleming, M. P., & Catlin, E. A. (2004). Clinical outcomes of near-term infants. Pediatrics, 114 (372-376). Wax, J. R. (2006). Maternal request cesarean versus planned spontaneous vaginal delivery: maternal morbidity and short term outcomes. Seminars in Perinatology, 30 (5), 247-252. Webb, D. A., & Robbins, J. M. (2003). Mode of delivery and risk of postpartum rehospitalization. Journal of the American Medical Association, 289 (1), 46-47. White, J. M., & Klein, D.M. (2002). Family Theories Thousand Oaks, CA: Sage Publications. Wingate, M. S., Alexander, G. R., Buekens, P., & Vahratian, A. (2007). Comparison of gestational age classications: date of last menstrual period vs. clinical estimate. Annals of Epidemiology, 17 (6), 425-430. Wingood, G. M., & DiClemente, R. J. (2000). Application of the theory of gender and power to examine HIV-related exposures, risk factors, and effective interventions for women. Health Education Behavior, 27 (5), 539-565. Xirasagar, S., & Lin, H. C. (2007). Maternal request CS--role of hospital teaching status and for-prot ownership. European Journal of Obstetrics and Gynecology and Reproductive Biology, 132 (1), 27-34. Yang, Y. T., Mello, M. M., Subramanian, S. V., Suddert, D. M. (2009). Relationship between malpractice litigation pressure and rates of cesarean section and vaginal birth after cesarean section. Medical Care 47(2), 234-42. 171

PAGE 182

Yasmeen, S., Romano, P. S., Schembri, M. E., Keyzer, J. M., & Gilbert, W. M. (2006). Accuracy of obstetric diagnoses and procedures in hospital discharge data. American Journal of Obstetrics and Gynecol, 194 (4), 992-1001. Yoder, B. A., Gordon, M. C., & Barth, W. H., Jr. (2008). Late-preterm birth: does the changing obstetric paradigm alter the epidemiology of respiratory complications? Obstetrics and Gynecology, 111 (4), 814-822. Zeger, S. L., & Liang, K. Y. (1986). Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 42 (1), 121-130. Zeitlin, J. A., Saurel-Cubizolles, M. J., & Ancel, P. Y. (2002). Marital status, cohabitation, and risk of preterm birth in Europe: where births outside marriage are common and uncommon. Paediatric Perinatal Epidemiology, 16 (2), 124-130. Zollinger, T. W., Przybylski, M.J., Gamache, R.E. (2006). Reliability of Indiana birth certicate data compared to medical records. Annals of Epidemiology, 16 1-10. 172

PAGE 183

Appendices 173

PAGE 184

Appendix A: Abstraction Instrument for the Florida Late-Preterm and Cesarean Delivery Investigation Florida Study of Late Preterm and Cesarean Delivery Abstraction Form ABSTRACTORS Abstracter initials and date ! ____ / ____ / ____ (mm/dd/yy) CASE IDENTIFICATION Birth Certicate !Birth Certicate Number!____________________ Hospital Medical Record !Hospital Name!______________________________________________________ !Date of delivery!!____ / ____ / ____ (mm/dd/yy) ! Mother's name (last/rst)_______________________________________________ !Mother's Date of Birth!____ / ____ / ____ (mm/dd/yy) !Hospital records number!_____________________ !Social Security Number!_______________________________________________ !Mother's residence (street address, City, Zip code, County)! !____________________________________________________________________ 174

PAGE 185

Appendix A: (Continued). NO INFORMATION COLLECTED ON THESE FORMS SHOULD COME FROM BIRTH CERTIFICATES OR BIRTH CERTIFICATE WORK SHEETS MATERNAL CHARACTERISTICS (Use the prenatal and delivery hospital record for this section) Mother's date of birth!!____ / ____ / ____ (mm/dd/yy) Married currently !!____ not stated!____ no !____ yes Race ____ black!____ white!____ otherspecify: ___________________ Ethnicity ____ not stated ____ Hispanic!____ otherspecify: ___________________ Place of Birth!!!____ U.S.!____ otherspecify: ___________________ Highest educational level ____ 8 th grade or less !____ 9 th -12 th grade, no diploma ____ HS diploma !____ Some college but no degree ____ Associate's degree !____ Bachelor's degree ____ Master's degree !____ Doctorate or professional degree Intended Hospital Payment !____ Medicaid!____ Self-pay !____ Private Insurance / HMO Source!____ otherspecify: __________________ PRENATAL CARE INFORMATION (Use the prenatal care record. If unavailable, use hospital record) Prenatal care received! !____ no !____ yes 175

PAGE 186

Appendix A: (Continued). Prenatal care records available!____ no !____ yes !____ partial !Available for last month of !pregnancy!!!____ no !____ yes !____ partial Date of rst prenatal visit: !____ / ____ / ____ (mm/dd/yy) !Gestational age at rst visit!________ weeks Date of last prenatal visit !____ / ____ / ____ (mm/dd/yy) (Review Doctor's note for a possible last visit) Number of prenatal visits! prenatal visits!!! Height !!!!____ feet!____ inches! Pre-pregnancy weight!!________ lbs !________ kg !____ not stated Last menstrual period!!____ / ____ / ____ EDC estimated by dates!!____ / ____ / ____ Final EDC on Prenatal Record!____ / ____ / ____ Ultrasounds: US date! Gestation! Est. Weight EDC ____ / ____ / ____ _____wks ________ gm ____ / ____ / ____ ____ / ____ / ____ _____wks ________ gm ____ / ____ / ____ ____ / ____ / ____ _____wks ________ gm ____ / ____ / ____ ____ / ____ / ____ _____wks ________ gm ____ / ____ / ____ ____ / ____ / ____ _____wks ________ gm ____ / ____ / ____ 176

PAGE 187

Appendix A: (Continued). ____ / ____ / ____ _____wks ________ gm ____ / ____ / ____ Amniocentesis ____ no! ____ yes results: _________________________! !Date of procedure!!____ / ____ / ____ (mm/dd/yy) !Estimated gestational age!____ not stated ________ weeks! !FLM Test!____ no ____ yes results: _______________________________ !L/S Ratio!____ no ____ yes results: _______________________________ PRIOR MEDICAL HISTORY (Use prenatal and hospital delivery records) Asthma!!!____ not stated! ____ no ____ yes Bleeding disorder!____ not stated! ____ no ____ yes !specify: ____________________ Cardiovascular disease____ not stated! ____ no ____ yes !specify: ____________________ Cancer ____ not stated! ____ no ____ yes specify: ____________________ Diabetes !Type 1 !!!____ not stated! ____ no ____ yes !Type 2! !!____ not stated! ____ no ____ yes Genetic / metabolic disorder ____ not stated! ____ no ____ yes !specify: ____________________ HypertensionChronic! ____ not stated! ____ no ____ yes HIV / AIDS!! ____ not stated! ____ no ____ yes !specify: ____________________ 177

PAGE 188

Appendix A: (Continued). Inammatory Bowel Disease! ____ not stated! ____ no ____ yes !specify: ____________________ Liver Disease / Hepatitis ____ not stated! ____ no ____ yes !specify: ____________________ Mental Illness / Depression!____ not stated! ____ no ____ yes !specify: ____________________ Obesity!!!____ not stated! ____ no ____ yes Renal disease !____ not stated ____ no ____ yes specify: ____________ Seizure disorder!!____ not stated! ____ no ____ yes !specify: ____________________ Systemic lupus erythematosus!____ not stated ____ no ____ yes Thyroid disorder !!____ not stated! ____ no ____ yes !specify: ____________________ Other medical conditions: !!!!!!! History of domestic violence !____ not stated! ____ no ____ yes !specify: ____________________ _____________________________________________________________________ _____________________________________________________________________ _____________________________________________________________________ Other ndings on prior medical history from prenatal and hospital records: _______________________________________________________________________ _______________________________________________________________________ 178

PAGE 189

Appendix A: (Continued). PAST OBSTETRICAL HISTORY (Use the prenatal care record. If unavailable, use hospital record) Gravida!____ pregnancies Para!____ deliveries History of previous C-section !____ not stated ____ no ____ yes !reason:_______________ Previous preterm birth!!____ not stated! ____ no ____ yes Previous fetal loss !____ not stated! ____ no ____ yes at which gestational ages: _______wks Uterine abnormality!____ not stated! ____ no ____ yes !specify:_____________________ Prior uterine surgery!____ not stated! ____ no ____ yes !specify:_____________________ Abnormal PAP smear(s)!____ not stated! ____ no ____ yes !specify:_____________________ Treatment(s) for abnormal PAP!____ not stated ____ no ____ yes specify:_____________________ Infertility problem !!____ not stated! ____ no ____ yes Infertility treatments!!____ not stated! ____ no ____ yes specify _____________ Last pregnancy outcome !Date of pregnancy ended ___ / ____ / ____ (mm/dd/yy)____ not stated !Outcome!____ live born ____ stillborn ____ spontAb ____ elective termination ____ not stated 179

PAGE 190

Appendix A: (Continued). Other history ndings: !!!!!!! OBSTETRICAL COMPLICATIONS DURING PREGNANCYNOT DELIVERY (Use prenatal & hospital records) Abruption prior to hospital admission!!! !Complete!!!____ no ____ yes !Partial!!!____ no ____ yes !Type not stated!!____ no ____ yes Anemia ____ not stated ____ no !____ yes !specify:_____________________ Anhydramnios!!____ not stated! ____ no ____ yes Birth defect!!____ not stated! ____ no !____ yes !specify:_____________________ Breech / malpresentation ____ not stated! ____ no ____ yes Chorioamnionitis ____ not stated ____ no ____ yes how long: __________________ Fetal growth restriction ____ not stated! ____ no ____ yes !specify:_____________________ Gestational diabetes!____ not stated! ____ no ____ yes Incompetent cervix!____ not stated! ____ no ____ yes Cerclage placed ____ not stated ____ no ___yes when: ______________________ Infections!!! !Gonorrhea! ____ not stated ____ no ____ yes when: ______________________ 180

PAGE 191

Appendix A: (Continued). !Herpes ____ not stated ____ no ____ yes when: ______________________ !HIV/ AIDS ____ not stated ____ no ____ yes when: ______________________ !HPV !Papillomavirus ____ not stated ____ no ____ yes when: ______________________ !Strep Group B ____ not stated ____ no ____ yes when: ______________________ !Syphilis ____ not stated ____ no ____ yes when: ______________________ !Other ____ not stated ____ no ____ yes! specify:_____________________ !Macrosomia ____ not stated ____ no !____ yes !Oligohydramnios ____ not stated ____ no ____ yes Preeclampsia / eclampsia / !Preg. induced hypertension!____ not stated! ____ no ____ yes Preterm labor!____ not stated ____ no ____ yes weeks gestation: _____________ Polyhydramnios !!____ not stated! ____ no !____ yes Placenta previa !!____ not stated! ____ no !____ yes Plurality ____ singleton! ____ twin!____ otherspecify: ________________ RH disease (isoimmunization)!____ not stated! ____ no !____ yes Substance abuse !Alcohol!!!____ not stated! ____ no !____ yes !Illicit drugs!!____ not stated! ____ no !____ yes !Tobacco!!!____ not stated! ____ no !____ yes 181

PAGE 192

Appendix A: (Continued). Urinary tract infection!!____ not stated! ____ no !____ yes Other complications!!______________________________________________ Prescription Medications:!!!! !__________________________________________________ !! !__________________________________________________ !__________________________________________________ Treatments for Complications:!! !__________________________________________________ !__________________________________________________ !__________________________________________________ Other ndings on current pregnancy including psychological and mental status: _______________________________________________________________________ _______________________________________________________________________ _______________________________________________________________________ _______________________________________________________________________ _______________________________________________________________________ _______________________________________________________________________ DELIVERY INFORMATION ( Use hospital delivery records) Date of hospital admission!____ / ____ / ____ (mm/dd/yy) !! Time of hospital admission!____ ____ : ____ ____ AM or PM Date of last menstrual period!____ / ____ / ____ (mm/dd/yy) 182

PAGE 193

Appendix A: (Continued). Date of estimated delivery-EDC!____ / ____ / ____ (mm/dd/yy) !! Initial delivery attendant! !Same as prenatal practice !____ not stated! ____ no ____ yes Certied Nurse Midwife !!____ not stated! ____ no ____ yes Transfer from another hospital!____ not stated! ____ no ____ yes Health status at the time of admission !Fever ____ not stated! ____ no ____ yesdegree: _______________ !Infection ____ not stated! ____ no ____ yesdescribe: ______________ !Hypertension ____ not stated! ____ no ____ yesdescribe: ______________ !Nausea! ____ not stated! ____ no ____ yesdescribe: ______________ !Diarrhea! ____ not stated! ____ no ____ yesdescribe: ______________ !Recent Trauma ____ not stated! ____ no ____ yesdescribe: ______________ !Rupture of membranes____ not stated ____ no ____ yesdescribe: _____________ !Vaginal bleeding !____ not stated! ____ no ____ yes !!!spotting light heavy n/s !Decreased fetal movements ____ not stated ____ no ____ yeshow long: ______ !Fetal heart tones!!____ not stated! ____ no ____ yeshow determined: ________ !Labor / contractions!____ not stated ____ no ____ yeshow long: _______ !Contraction frequency!____ not stated ____ minutes 183

PAGE 194

Appendix A: (Continued). !Contraction duration!____ not stated ____ minutes !Contraction strength!____ not stated! !!! ____ intermittent__ mild ___ moderate___ strong !Fetal station (circle one)!-4 -3!-2 !-1 !0 !1 !2 !3 !4 !Cervical length!!____ not stated! ____ cm! !!!Measured by ultrasound:__ yes no !Cervical dilatation!!____ not stated! ____ cm !Cervical effacement!!____ not stated! ____ % !Cervical consistency !describe: ____________________________ Weight at delivery!____ not stated! _____ pounds!_______ kilograms Weight gain during pregnancy ____ not stated _____ pounds!_______ kilograms Provider gestational age estimate !documented prior to delivery______ completed weeks Gestation / maturation conrmed during !this hospitalization or week prior Ultrasound ____ no!____ yes results: _____________________________ Amniocentesis ____ no ____ yes reason why: _________________________! FLM Test ____ no!____ yes results: _____________________________ L/S Ratio ____ no!____ yes results: _____________________________ Date of labor onset!!____ / ____ / ____ (mm/dd/yy) !____ no labor 184

PAGE 195

Appendix A: (Continued). Time of active labor onset!____ ____ : ____ ____ AM or PM Rupture of membranes! ____ not stated! ____ noor choose spontaneous or articial !Spontaneous rupture!!____ yes !Articial rupture!!____ yeswhy: ____________________________ Date of membrane rupture!____ / ____ / ____ (mm/dd/yy) !!! Time of membrane rupture!____ ____ : ____ ____ AM or PM Duration of membrane rupture! !prior to delivery!!____ hours!____ minutes Labor Induction!!!____ no !____ yes !Documented reason:! !__________________________________________________________ __________________________________________________________ __________________________________________________________ !If yes, check all that apply:! ____ Pitocin ____ Misoprostol or PGE1 [oral, vaginal, sublingual] ____ Prepidil [dinoprostone gel or prostaglandin E2] ____ Cervidil [dinoprostone vaginal insert] ____ Foley catheter [with or without extraamniotic saline infusion] ____ Other !Specify: ______________________ Date of labor induction!____ / ____ / ____ (mm/dd/yy) 185

PAGE 196

Appendix A: (Continued). Time of labor induction!____ ____ : ____ ____ AM or PM Labor augmentation!!____ no !____ yes !Documented reason:! !_________________________________________________________ !!!! !Date of augmentation!____ / ____ / ____ (mm/dd/yy) Time of augmentation!____ ____ : ____ ____ AM or PM Delivery method !Vaginal!!!____ no !____ yes !Cesarean!!!____ no !____ yes Forceps Used!!!____ no !____ yes Vacuum Extraction Used!!____ no !____ yes Anesthesia given for labor prior to deciding to do a !Cesarean delivery!!____ none!____ not stated !!!!!____ paracervical block !!!!!____ pudendal block !!!!!____ epidural !!!!!____ spinal!! Date of epidural!started!!____ / ____ / ____ (mm/dd/yy) !! Time of epidural! started!!____ ____ : ____ ____ AM or PM Length of stage 1 labor!!_________ hours!________ minutes 186

PAGE 197

Appendix A: (Continued). Length of stage 2 labor!!_________ hours!________ minutes Date of last cervical exam prior !to delivery!!!____ / ____ / ____ (mm/dd/yy) Time of last cervical exam!!____ ____ : ____ ____ AM or PM !Contraction strength!____ not stated! !!! ____ intermittent ____ mild ____ moderate ____ strong !Cervical length!____ not stated ____ cm!Description: _________________ !Cervical dilatation!!____ not stated! ____ cm !Cervical effacement!!____ not stated! ____ % Date of delivery!!!____ / ____ / ____ (mm/dd/yy) !! Time of delivery!!!____ : ____ AM or PM Live birth!!!_____ lbs ____ ozs !__________ grams Gender of fetus _____ male! ____ female!____ unknown ____ not stated Apgar score!!!_____ 1 min! ____ 5 min !____ 10 min Newborn resuscitation ____ not stated! ____ no ____ yesspecify: _________ Type of nursery referral!!____ not stated ____ NICU! ____ Intermediate !____ Normal nursery Provider gestational age estimate !documented after delivery!____ completed weeks !Estimation method: !____ not stated!Specify: ______________ 187

PAGE 198

Appendix A: (Continued). Nursery gestational age estimate !documented after delivery!____ completed weeks !Estimation method: !!____ not stated!Specify: ______________ OBSTETRICAL COMPLICATIONS AND PROBLEMS DURING LABOR AND DELIVERY ( Use hospital records ): Abruption Complete !____ no !____ yes Partial ____ no !____ yes Type not stated !____ no !____ yes Breech / malpresentation ____ not stated ____ no !____ yes !specify:_______ Chorioamnionitis ____ not stated ____ no ____ yes how long: __________________ Eclampsia!____ not stated! ____ no !____ yes Fetal distress / labor intolerance ____ not stated ____ no ____ yes specify:________ Fetal growth restriction ____ not stated ____ no ____ yes specify:_____________ Failed induction of labor ____ not stated ____ no ____ yes specify:_______________ Failure to progress / Prolonged stage of labor (1 or 2)!____ not stated ____ no ____ yes specify:__________ Hypertensive crisis!____ not stated ____ no ____ yes specify:_______________ Hemorrhage crisis ____ not stated ____ no ____ yes specify:_____________________ Infections!!! 188

PAGE 199

Appendix A: (Continued). !Gonorrhea ____ not stated ____ no ____ yes!when: ______________________ !Herpes!____ not stated ____ no ____ yes when: ______________________ !HIV / AIDS ____ not stated ____ no ____ yes when: ______________________ !Strep Group B ____ not stated ____ no ____ yes when: ______________________ !Syphilis!____ not stated ____ no ____ yes when: ______________________ !Other ____ not stated ____ no ____ yes specify:_____________________ Placenta previa !____ not stated! ____ no !____ yes Plurality ____ singleton! ____ twin!____ other! !specify: ____________________ Preterm labor!____ not stated ____ no ____ yes !weeks gestation: _____________ Prolonged rupture (> 12 hrs)!____ not stated! ____ no !____ yes duration: ____________________ Meconium staining!!____ not stated! ____ no !____ yes !thick: ____ yes!____ no Uterine rupture!!!____ not stated! ____ no !____ yes Other complications!!______________________________________________ DOCUMENTED REASONS FOR A CESAREAN (Use hospital delivery record and OB notes) If Cesarean, which of the following explanations are true Repeat Cesareanhad a previous Cesarean delivery!____ no !____ yes 189

PAGE 200

Appendix A: (Continued). Scheduled in advance to come to the hospital for a Cesarean delivery ____ no ____ yes Immediately referred to the hospital for a Cesarean in last 48 hours ____ no ____ yes Medical emergency Cesarean to save the life of mother or baby!! ____ no ____ yes Documented Reasons for Doing a Cesarean Delivery (Obtain from OB notes or delivery records; must be exact wording of the reasons for performing CS): 1) _____________________________________________________________________ 2) _____________________________________________________________________ 3) _____________________________________________________________________ 4) _____________________________________________________________________ If a Cesarean, any of the following reason? (Only check using the above documented reasons) !Birth defect / congenital anomally!____ yes !Breech / malpresentation!!!____ yes !Chorioamnionitis !!!!____ yes !Failed induction of labor!!!____ yes !Fetal distress / labor intolerance!!____ yes !Fetal growth!!!!____ yes !Force dystocia / failure to progress!____ yes 190

PAGE 201

Appendix A: (Continued). !Hemorrhage!!!!____ yes !Hypertension!!!!____ yes !Infection !!!!!____ yes !Macrosomia!!!!____ yes !Pelvic contraction or dystocia!!____ yes !Previous uterine surgery!!!____ yes !Other ____ yes Specify: ________________________________ Written comments about mother's thoughts, preferences, or requests regarding Cesarean delivery (Use prenatal and delivery hospital records, including nurses notes and please document the source of the information): _______________________________________________________________________ _______________________________________________________________________ _______________________________________________________________________ 191

PAGE 202

Appendix B: Power Calculation Results POWER CALCULATIONS Three power calculations were performed for this dissertation research: (1) the necessary sample size for comparisons of late-preterm infant morbidity by mode of delivery, (2) the necessary sample size for comparisons of maternal morbidity (classied as rehospitalization) by mode of delivery, and (3) the necessary sample size for comparisons of costs by mode of delivery. POWER CALCULATION ONE The rst power calculation focused on late-preterm infant morbidity by mode of delivery (vaginal versus cesarean). The majority of the studies on late-preterm birth outcomes by mode of delivery only report mortality. Of the four studies that assessed late-preterm birth morbidity following mode of delivery, all focused on respiratory distress (or indicators for respiratory distress) (de Almeida, et al., 2007; Levine, Ghai, Barton, & Strom, 2001; Malloy, 2009; Yoder, Gordon, & Barth, 2008) Thus, respiratory distress was used as the late-preterm infant morbidity outcome for the power calculation. While respiratory distress as reported by these studies occurred at the initial (delivery) hospitalization event, it serves as an indicator of infants that are likely to have further morbidity diagnoses in the rst year of life. Proportions of respiratory distress by mode of delivery were averaged across the four studies (0.179 CS, 0.093 vaginal). Study sample sizes were used to determine the proportion of late-preterm CS deliveries to vaginal deliveries (1:4.69). Those proportions were then entered into the PASS sample size calculation program, and 192

PAGE 203

Appendix B: (Continued). corresponding sample sizes were then calculated for several power assumptions (Table B-1) at an alpha level of 0.05. The effect size was small (0.27). Table B-1 Sample size determination for comparison of late-preterm respiratory morbidity by mode of delivery Table B-1 Sample size determination for comparison of late-preterm respiratory morbidity by mode of delivery Table B-1 Sample size determination for comparison of late-preterm respiratory morbidity by mode of delivery Table B-1 Sample size determination for comparison of late-preterm respiratory morbidity by mode of delivery Power Assumptions Cesarean (n) Vaginal (n) Total (N) 65% 91 427 518 70% 104 488 592 75% 119 559 678 80% 137 643 780 85% 159 746 905 90% 189 887 1,076 95% 238 1,117 1,355 99% 347 1,628 1,975 POWER CALCULATION TWO The second power calculation focused on maternal morbidity by mode of delivery (vaginal versus cesarean). Rate of rehospitalization was selected as an indicator for maternal morbidity. The rate of maternal morbidity varies by type of morbidity, however, as the hospital discharge data reports morbidities that resulted in rehospitalization, rates of rehospitalization by mode of delivery should sufciently capture maternal morbidity resulting in a hospitalization. Five studies reported maternal rehospitalization rates by mode of delivery (Declercq, et al., 2007; Liu, et al., 2005; Lydon-Rochelle, Holt, Martin, & Easterling, 2000; Ophir, et al., 2008; Webb & Robbins, 2003) The rehospitalization rate by mode of delivery was used for the sample size calculation, and was determined by averaging reported rehospitalization rates (2.19% CS, 1.10% vaginal). The ratio of CS to vaginal delivery across the study period (1998 to 2006) was determined by adding together all national CS rates (overall), and dividing 193

PAGE 204

Appendix B: (Continued). by the number of years to get an average CS rate (26.06). These proportions were then entered into the PASS sample size calculation program, and corresponding sample sizes were then calculated for several power assumptions (Table B-2) at an alpha level of 0.05. The effect size was small (0.03). Table B-2. Sample size determination for comparison of maternal rehospitalizations in the rst year postpartum by mode of delivery Table B-2. Sample size determination for comparison of maternal rehospitalizations in the rst year postpartum by mode of delivery Table B-2. Sample size determination for comparison of maternal rehospitalizations in the rst year postpartum by mode of delivery Table B-2. Sample size determination for comparison of maternal rehospitalizations in the rst year postpartum by mode of delivery Power Assumptions Cesarean (n) Vaginal (n) Total (N) 65% 818 3,139 3,957 70% 935 3,588 4,523 75% 1,071 4,110 5,181 80% 1,233 4,732 5,965 85% 1,437 5,514 6,951 90% 1,714 6,577 8,291 95% 2,171 8,331 10,502 99% 3,173 12,175 15,348 194

PAGE 205

Appendix C: Manuscript One Supplementary Tables Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1. Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Number of Cases Number of Cases " " " " " Birth Cert Medical record % Agree Sen Spe Sen 95% CI PPV NPV Kappa LR + LR Demographic Variable " " " " " Race/ethnicity * 0.86 * * * * Parity ++ * 0.96 * * * * Marital Status * 0.85 * * * * Maternal Medical Conditions & Risk Factors " " " " " LMP * 0.55 * * * # Prenatal Visits * 0.27 * * * Obesity from calculated BMI 226 227 0.90 0.76 0.94 (0.70, 0.82) 0.77 0.93 0.70 12.06 0.26 Previous preterm infant 12 70 0.94 0.13 1.00 (0.06, 0.23) 0.75 0.94 0.20 42.87 0.87 IUGR 11 135 0.88 0.08 1.00 (0.04, 0.14) 1.00 0.88 0.13 0.92 Cerclage (Incompetent cervix) 6 18 0.98 0.28 1.00 (0.10, 0.53) 0.83 0.99 0.41 277.8 0.72 Anemia 8 141 0.87 0.04 1.00 (0.01, 0.08) 0.63 0.87 0.05 10.61 0.97 Chronic Diabetes 32 41 0.97 0.54 0.99 (0.37, 0.69) 0.71 0.98 0.60 60.29 0.47 Chronic Hypertension 33 100 0.92 0.25 0.99 (0.17, 0.35) 0.76 0.93 0.35 29.76 0.76 Renal Disease 1 66 0.94 0.00 0.94 0.00 1.00 0.00 0.00 1.06 Heart Conditions 3 58 0.95 0.67 0.95 (0.00, 0.12) 0.03 1.00 0.06 12.33 0.35 Asthma 9 101 0.91 0.08 1.00 (0.03, 0.15) 0.89 0.91 0.13 80.00 0.92 Gestational Diabetes 63 122 0.92 0.41 0.99 (0.32, 0.50) 0.79 0.93 0.50 29.48 0.60 Gestational Hypertension, Preclapmsia, Eclampsia 197 285 0.84 0.55 0.95 (0.49, 0.61) 0.79 0.85 0.55 10.36 0.47 Gonorrhea 10 14 0.98 0.43 1.00 (0.18, 0.71) 0.60 0.99 0.49 112.79 0.57 Genital Herpes 9 75 0.93 0.08 1.00 (0.03, 0.17) 0.67 0.93 0.13 0.92 Syphilis 1 7 0.99 0.14 1.00 (0.00, 0.58) 1.00 0.99 0.25 0.86 HIV 5 14 0.99 0.36 1.00 (0.13, 0.65) 1.00 0.99 0.52 0.64 Alcohol 6 37 0.96 0.05 1.00 (0.01, 0.18) 0.33 0.97 0.08 13.87 0.95 Tobacco 52 108 0.93 0.39 0.99 (0.30, 0.49) 0.81 0.93 0.49 36.69 0.62 195

PAGE 206

Appendix C: (Continued). Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Table C-1 (Continued). Validity Indices of Florida Birth Certicate Compared to Maternal Medical Records, 2006-2007 Number of Cases Number of Cases " " " " " Birth Cert Medical record % Agree Sen Spe Sen 95% CI PPV NPV Kappa LR + LR Labor/Delivery Factors " " " " " Trial of Labor 493 700 0.61 0.56 0.71 (0.52, 0.59) 0.79 0.45 0.23 1.93 0.62 Augmentation 120 111 0.88 0.48 0.93 (0.38, 0.57) 0.44 0.94 0.39 6.73 0.56 Induction 171 207 0.75 0.52 0.93 (0.45, 0.59) 0.63 0.89 0.48 7.14 0.52 Chorioamnionitis 10 21 97.82 0.19 0.99 (0.05, 0.42) 0.40 0.98 0.25 19.00 0.82 Placenta Previa 7 58 95.16 0.12 1.00 (0.05, 0.23) 1.00 0.95 0.21 0.88 Placenta Abruption 9 68 94.22 0.12 1.00 (0.05, 0.22) 0.89 0.94 0.20 0.88 Uterine Rupture 1 2 99.72 0.00 1.00 (0.16, 1.00) 0.00 1.00 0.00 1.00 Prolonged rupture of membranes 1 121 0.89 0.01 1.00 (0.00, 0.05) 1.00 0.89 0.01 0.99 Prolonged labor 9 195 0.82 0.03 1.00 (0.01, 0.07) 0.67 0.82 0.04 7.50 0.97 Fetal distress 182 324 0.63 0.18 0.83 (0.14, 0.23) 0.32 0.70 0.01 1.06 0.99 Breech/Malpresentation 174 267 0.80 0.62 0.99 (0.56, 0.68) 0.95 0.89 0.69 52.08 0.38 Meconium 36 59 0.94 0.27 0.98 (0.16, 0.40) 0.44 0.96 0.31 13.49 0.74 Forceps 5 3 0.99 0.33 1.00 (0.01, 0.91) 0.20 1.00 0.25 87.71 0.67 Vacuum 2 18 0.98 0.00 1.00 (0.81, 1.00) 0.00 0.98 0.00 0.00 1.00 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 *Not Applicable " " " "   n=1028, missing 27 ++ n=1016, missing=39 196

PAGE 207

Appendix C: (Continued). Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 Table C-2. Validity Indices of Florida In-Patient Hospital Discharge Data Compared to Maternal Medical Records, 2006-2007 AHCA Record Medical Record % Agree Sen 95% CI Sen Spe PPV NPV Kappa LR + LR Demographic Variable " " " " " Race/ethnicity * 83.37 * * * * Maternal Medical Conditions & Risk Factors Maternal Medical Conditions & Risk Factors " " " " " Obesity from calculated BMI 51 213 79.70 0.17 (0.12, 0.23) 0.98 0.71 0.80 0.20 8.50 0.85 Previous preterm infant 4 64 93.43 0.05 (0.01, 0.13) 1.00 0.75 0.94 0.08 0.95 Intrauterine growth restriction 115 131 94.50 0.74 (0.66, 0.82) 0.98 0.84 0.96 0.78 37.00 0.27 Cerclage 17 17 98.73 0.65 (0.38, 0.86) 0.99 0.65 0.99 0.64 65.00 0.35 Anemia 110 130 85.59 0.40 (0.32, 0.49) 0.93 0.47 0.91 0.35 5.71 0.65 Chronic Diabetes 38 41 98.20 0.76 (0.60, 0.88) 0.99 0.82 0.99 0.53 76.00 0.24 Chronic Hypertension 65 95 95.13 0.60 (0.50, 0.70) 0.99 0.88 0.96 0.69 60.00 0.40 Renal Disease 12 64 93.01 0.08 (0.03, 0.17) 0.99 0.42 0.94 0.11 8.00 0.93 Heart Conditions 20 58 94.92 0.26 (0.15, 0.39) 0.99 0.75 0.95 0.36 26.00 0.75 Asthma 26 94 92.59 0.27 (0.18, 0.37) 1.00 0.96 0.92 0.39 0.73 Gestational Diabetes 82 110 94.92 0.65 (0.56, 0.74) 0.99 0.88 0.96 0.72 65.00 0.35 Gestational Hypertension, Preeclampsia, Eclampsia 292 266 91.11 0.89 (0.85, 0.93) 0.92 0.81 0.96 0.79 11.13 0.12 Gonorrhea 1 12 98.84 0.08 (0.00, 0.24) 1.00 1.00 0.99 0.15 0.92 Genital Herpes 36 72 95.98 0.49 (0.37, 0.60) 1.00 0.97 0.96 0.63 0.51 Syphilis in current pregnancy 1 7 99.15 0.00 (0.59, 1.00) 1.00 0.00 0.99 -0.002 1.00 HIV 11 14 99.69 0.79 (0.49, 0.95) 1.00 1.00 1.00 0.88 0.21 Alcohol 5 32 96.72 0.09 (0.02, 0.25) 1.00 0.60 0.97 0.15 0.91 Tobacco 35 100 91.42 0.27 (0.19, 0.36) 0.99 0.77 0.92 0.37 27.00 0.74 Labor/Delivery Factors " " " " " Trial of Labor 571 619 63.14 0.68 (0.64, 0.72) 0.54 0.74 0.47 0.21 1.48 0.59 Fetal distress 246 297 60.70 0.29 (0.24, 0.34) 0.75 0.35 0.70 0.04 1.16 0.95 Breech/Malpresentation 254 236 92.99 0.89 (0.85, 0.93) 0.94 0.83 0.96 0.81 14.83 0.12 Induction 120 192 84.95 0.44 (0.37, 0.52) 0.95 0.71 0.87 0.46 8.80 0.59 Chorioamnionitis 14 20 98.10 0.40 (0.19, 0.64) 0.99 0.57 0.99 0.46 40.00 0.61 Placenta Previa 46 50 98.31 0.80 (0.66, 0.90) 0.99 0.87 0.99 0.82 80.00 0.20 Prolonged rupture of membranes 20 109 90.36 0.17 (0.11, 0.26) 1.00 0.95 0.90 0.27 0.83 Prolonged labor 133 179 91.11 0.64 (0.56, 0.71) 0.98 0.86 0.92 0.68 32.00 0.37 Assisted Delivery (Forceps/ Vacuum) 12 18 97.88 0.28 (0.10, 0.53) 0.99 0.42 0.99 0.32 28.00 0.73 Meconium 15 53 94.28 0.13 (0.05, 0.25) 0.99 0.47 0.95 0.19 13.00 0.88 197

PAGE 208

Appendix C: (Continued). Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Table C-3. Validity Indices of Linked Florida Birth Certicate and Discharge Data File Compared to Maternal Medical Records, 2006-2007 Number of Cases Number of Cases " " " " Linked Med record Sen 95% CI Sen Spe PPV NPV Kappa LR + LR Demographic Variable " " " " " Race/ethnicity * * * * Maternal Medical Conditions & Risk Factors " " " " " Obesity from calculated BMI 223 213 0.78 (0.72, 0.83) 0.92 0.74 0.93 0.69 9.75 0.24 Previous preterm infant 13 64 0.14 (0.07, 0.25) 1.00 0.69 0.94 0.22 0.86 Intrauterine growth restriction 117 131 0.76 (0.67, 0.83) 0.98 0.85 0.96 0.77 38.00 0.24 Cerclage 19 17 0.71 (0.44, 0.90) 0.99 0.63 0.99 0.66 71.00 0.29 Anemia 118 130 0.41 (0.32, 0.50) 0.92 0.45 0.91 0.34 5.13 0.64 Chronic Diabetes 49 41 0.83 (0.68, 0.94) 0.98 0.69 0.99 0.74 41.50 0.17 Chronic Hypertension 76 95 0.64 (0.54, 0.74) 0.98 0.80 0.96 0.69 32.00 0.37 Renal Disease 13 64 0.08 (0.03, 0.17) 0.99 0.38 0.94 0.11 8.00 0.93 Heart Conditions 24 58 0.29 (0.18, 0.43) 0.99 0.71 0.96 0.39 29.00 0.72 Asthma 31 94 0.31 (0.22, 0.41) 1.00 0.94 0.93 0.44 0.69 Gestational Diabetes 99 110 0.70 (0.61, 0.78) 0.97 0.78 0.96 0.7 23.33 0.31 Gestational Hypertension, Pre-eclapmsia, Eclampsia 314 266 0.91 (0.87, 0.94) 0.89 0.77 0.96 0.76 8.27 0.10 Gonorrhea 9 12 0.42 (0.15, 0.72) 1.00 0.56 0.99 0.47 0.58 Genital Herpes 43 72 0.54 (0.42, 0.66) 1.00 0.91 0.96 0.66 0.46 Syphilis in current pregnancy 2 7 0.14 (0.00, 0.58) 1.00 0.50 0.99 0.22 0.86 HIV 11 14 0.79 (0.49, 0.95) 1.00 1.00 1.00 0.88 0.21 Alcohol 9 32 0.09 (0.02, 0.25) 0.99 0.33 0.97 0.13 9.00 0.92 Tobacco 62 100 0.49 (0.39, 0.59) 0.98 0.79 0.94 0.57 24.50 0.52 Labor/Delivery Factors " " " " " Trial of Labor 714 619 0.84 (0.80, 0.86) 0.39 0.72 0.56 0.25 1.38 0.41 Fetal distress 355 297 0.42 (0.36, 0.47) 0.64 0.35 0.71 0.06 1.17 0.91 Breech/Malpresentation 262 236 0.91 (0.86, 0.94) 0.93 0.82 0.97 0.81 13.00 0.10 Induction 213 192 0.71 (0.64, 0.78) 0.90 0.64 0.92 0.59 7.10 0.32 Chorioamnionitis 21 20 0.50 (0.27, 0.73) 0.99 0.48 0.99 0.48 50.00 0.51 Placenta Previa 47 50 0.82 (0.69, 0.91) 0.99 0.87 0.99 0.84 82.00 0.18 Prolonged rupture of membranes 21 109 0.18 (0.12, 0.27) 1.00 0.95 0.90 0.28 0.82 Prolonged labor 136 179 0.64 (0.56, 0.71) 0.97 0.84 0.92 0.67 21.33 0.37 Assisted Delivery (Forceps/ Vacuum) 12 18 0.28 (0.10, 0.53) 0.99 0.42 0.99 0.32 28.00 0.73 Meconium 46 53 0.36 (0.23, 0.50) 0.97 0.41 0.96 0.35 12.00 0.66 198

PAGE 209

Appendix C: (Continued). Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4. Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. % Agreement % Agreement Sen Sen Sen 95% CI Sen 95% CI PPV PPV Kappa Kappa High CS Low CS High CS Low CS High CS Low CS High CS Low CS High CS Low CS Demographic Variable " " " " " Race/ethnicity " " " " " Birth Certicate 78.09 88.82 * " * * Discharge File 74.48 92.20 * " * * Linked Birth & Discharge File # 81.49 90.30 * * * * Maternal Medical Conditions & Risk Factors " " " " " Obesity " " " " " Birth Certicate 88.87 90.89 0.72 0.78 (0.61, 0.82) (0.71, 0.85) 0.62 0.88 0.60 0.77 Discharge File 83.83 75.53 0.13 0.19 (0.07, 0.23) (0.13, 0.26) 0.48 0.87 0.15 0.23 Linked Birth & Discharge File 87.44 90.51 0.75 0.80 (0.63, 0.84) (0.72, 0.86) 0.58 0.87 0.58 0.76 Previous Preterm " " " " " Birth Certicate 95.73 92.05 0 0.18 (0.09, 0.31) 0 1.00 -0.001 0.25 Discharge File 0 90.93 0 0.07 (0.82, 1.00) (0.01, 0.18) 0 0.75 0 0.11 Linked Birth & Discharge File 95.32 92.20 0 0.20 (0.82, 1.00) (0.10, 0.35) 0 0.90 -0.011 0.30 Intrauterine Growth Restriction " " " " " Birth Certicate 89.61 86.83 0.07 0.09 (0.02, 0.16) (0.04, 0.18) 1.00 1.00 0.11 0.25 Discharge File 94.47 94.51 0.73 0.75 (0.60, 0.84) (0.63, 0.84) 0.81 0.87 0.74 0.77 Linked Birth & Discharge File 94.47 94.96 0.73 0.78 (0.60, 0.84) (0.66, 0.87) 0.81 0.88 0.74 0.79 Cerclage " " " " " Birth Certicate 98.52 98.84 0.30 0.25 (0.07, 0.65) (0.03, 0.65) 0.75 1.00 0.42 0.40 Discharge File 98.08 99.37 0.56 0.75 (0.21, 0.86) (0.35, 0.97) 0.50 0.86 0.52 0.80 Linked Birth & Discharge File 97.87 99.58 0.56 0.88 (0.21, 0.86) (0.47, 1.00) 0.45 0.88 0.49 0.87 Anemia " " " " " Birth Certicate 89.8 83.73 0.02 0.05 (0.00, 0.10) (0.01, 0.11) 0.50 0.67 0.03 0.07 Discharge File 87.24 83.97 0.19 0.54 (0.09, 0.33) (0.42, 0.65) 0.36 0.51 0.19 0.53 Linked Birth & Discharge File 86.6 83.33 0.19 0.55 (0.10, 0.33) (0.43, 0.66) 0.32 0.49 0.17 0.42 Chronic Diabetes " " " " " Birth Certicate 97.77 96.70 0.57 0.52 (0.29, 0.82) (0.32, 0.71) 0.57 0.78 0.56 0.61 Discharge File 98.51 97.89 0.64 0.81 (0.35, 0.87) (0.62, 0.94) 0.82 0.81 0.71 0.8 Linked Birth & Discharge File 97.87 97.47 0.71 0.89 (0.42, 0.95) (0.71, 0.98) 0.62 0.73 0.66 0.79 199

PAGE 210

Appendix C: (Continued). Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. % Agreement % Agreement Sen Sen Sen 95% CI Sen 95% CI PPV PPV Kappa Kappa High CS Low CS High CS Low CS High CS Low CS High CS Low CS High CS Low CS Chronic Hypertension " " " " " Birth Certicate 94.99 89.14 0.31 0.22 (0.17, 0.49) (0.12, 0.33) 0.79 0.74 0.43 0.29 Discharge File 96.17 94.09 0.52 0.65 (0.34, 0.69) (0.51, 0.76) 0.89 0.87 0.64 0.71 Linked Birth & Discharge File 95.96 93.67 0.55 0.69 (0.36, 0.72) (0.56, 0.80) 0.82 0.8 0.73 0.71 Renal Disease " " " " " Birth Certicate 95.18 92.05 0 0 * 0 0 0 -0.004 Discharge File 94.25 91.77 0.04 0.11 (0.00, 0.20) (0.03, 0.25) 0.33 0.44 0.06 0.14 Linked Birth & Discharge File 94.25 91.56 0.04 0.11 (0.00, 0.20) (0.03, 0.25) 0.33 0.40 0.06 0.14 Heart Problems " " " " " Birth Certicate 94.81 94.38 0.03 0.03 (0.00, 0.18) (0.00, 0.18) 1.00 0.50 0.06 0.06 Discharge File 94.89 93.88 0.24 0.28 (0.10, 0.44) (0.13, 0.47) 0.78 0.73 0.06 0.11 Linked Birth & Discharge File 94.89 94.94 0.24 0.34 (0.10, 044) (0.18, 0.54) 0.78 0.67 0.35 0.43 Asthma " " " " " Birth Certicate 90.54 91.66 0.04 0.12 (0.00, 0.13) (0.05, 0.25) 0.67 1.00 0.06 0.20 Discharge File 90.43 91.98 0.13 0.4 (0.05, 0.26) (0.26, 0.56) 1.00 0.95 0.10 0.30 Linked Birth & Discharge File 91.49 94.30 0.17 0.45 (0.08, 0.31) (0.30, 0.60) 0.89 0.95 0.26 0.58 Gestational Diabetes " " " " " Birth Certicate 91.84 92.06 0.31 0.51 (0.20, 0.44) (0.38, 0.64) 0.91 0.74 0.43 0.56 Discharge File 94.04 95.78 0.57 0.73 (0.42, 0.71) (0.60, 0.84) 0.83 0.91 0.64 0.79 Linked Birth & Discharge File 94.25 94.10 0.59 0.8 (0.44, 0.72) (0.67, 0.89) 0.83 0.75 0.66 0.74 Gestational Hypertension, Preclapmsia, Eclampsia " " " " " Birth Certicate 85.53 82.17 0.53 0.56 (0.44, 0.62) (0.49, 0.64) 0.78 0.80 0.55 0.54 Discharge File 91.06 91.14 0.89 0.89 (0.82, 0.94) (0.83, 0.94) 0.78 0.84 0.77 0.8 Linked Birth & Discharge File 90.85 88.81 0.9 0.92 (0.83, 0.95) (0.86, 0.96) 0.77 0.77 0.77 0.75 Gonorrhea " " " " " Birth Certicate 99.63 98.07 0.67 0.36 (0.09, 0.99) (0.11, 0.69) 0.67 0.57 0.66 0.44 Discharge File 0 98.10 0 0.10 (0.16, 1.00) (0.00, 0.45) 0 1.00 0 0.18 Linked Birth & Discharge File 99.79 97.89 1.00 0.30 (0.16, 1.00) (0.07, 0.65) 0.67 0.50 0.80 0.37 Genital Herpes " " " " " Birth Certicate 93.13 93.22 0.10 0.06 (0.03, 0.24) (0.01, 0.19) 0.80 0.50 0.16 0.09 Discharge File 95.95 96.00 0.50 0.47 (0.33, 0.67) (0.30, 0.65) 1.00 0.94 0.65 0.61 Linked Birth & Discharge File 96.38 95.78 0.58 0.5 (0.41, 0.74) (0.32, 0.68) 0.96 0.85 0.70 0.61 200

PAGE 211

Appendix C: (Continued). Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. % Agreement % Agreement Sen Sen Sen 95% CI Sen 95% CI PPV PPV Kappa Kappa High CS Low CS High CS Low CS High CS Low CS High CS Low CS High CS Low CS Syphilis in current pregnancy " " " " " Birth Certicate 99.44 99.41 0 0.25 (0.01, 0.81) 0 1.00 0 0.40 Discharge File 99.36 98.95 0 0 (0.29, 1.00) (0.40, 1.00) 0 0 0 -0.003 Linked Birth & Discharge File 99.36 99.16 0 0.25 (0.29, 1.00) (0.01, 0.81) 0 0.5 0 0.33 HIV " " " " " Birth Certicate 99.26 99.04 0.20 0.44 (0.01, 0.72) (0.14, 0.79) 1.00 1.00 0.33 0.61 Discharge File 99.37 100 0.40 1.00 (0.05, 0.85) (0.66, 1.00) 1.00 1.00 0.57 1.00 Linked Birth & Discharge File 99.37 100 0.40 1.00 (0.05, 0.85) (0.66, 1.00) 1.00 1.00 0.57 1.00 Alcohol " " " " " Birth Certicate 96.29 96.32 0 0.11 (0.01, 0.33) 0 0.50 -0.007 0.16 Discharge File 97.23 96.20 0.07 0.11 (0.00, 0.34) (0.01, 0.35) 1.00 0.50 0.13 0.17 Linked Birth & Discharge File 96.81 95.78 0.07 0.11 (0.00, 0.34) (0.01, 0.35) 0.33 0.33 0.11 0.15 Tobacco " " " " " Birth Certicate 92.77 92.83 0.12 0.57 (0.04, 0.25) (0.44, 0.69) 0.83 0.80 0.19 0.63 Discharge File 92.34 90.50 0.05 0.40 (0.01, 0.18) (0.28, 0.54) 1.00 0.76 0.09 0.48 Linked Birth & Discharge File 93.19 93.25 0.18 0.68 0.08, 0.34) (0.55, 0.79) 0.88 0.78 0.28 0.69 Labor/Delivery Factors " " " " " Attempted Labor " " " " " Birth Certicate 61.04 61.05 0.54 0.58 (0.49, 0.59) (0.52, 0.63) 0.78 0.80 0.25 0.22 Discharge File 58.72 67.51 0.61 0.74 (0.56, 0.67) (0.69, 0.79) 0.69 0.77 0.15 0.27 Linked Birth & Discharge File 65.11 71.52 0.78 0.88 (0.73, 0.83) 0.84, 0.92) 0.70 0.75 0.22 0.26 Fetal distress " " " " " Birth Certicate 57.14 69.19 0.26 0.10 (0.20, 0.34) (0.06, 0.16) 0.25 0.78 -0.044 0.12 Discharge File 60.85 60.55 0.25 0.32 (0.18, 0.33) (0.25, 0.40) 0.30 0.39 0.01 0.07 Linked Birth & Discharge File 51.91 62.45 0.45 0.39 (0.36, 0.53) (0.32, 0.47) 0.29 0.43 -0.004 0.14 Breech/Malpresentation " " " " " Birth Certicate 89.79 89.15 0.51 0.69 (0.42, 0.62) (0.62, 0.77) 0.96 0.94 0.61 0.73 Discharge File 92.56 93.04 0.91 0.88 (0.83, 0.96) (0.82, 0.93) 0.75 0.89 0.78 0.84 Linked Birth & Discharge File 92.13 93.04 0.91 0.90 (0.83, 0.96) (0.84, 0.95) 0.74 0.87 0.77 0.84 Induction " " " " " Birth Certicate 82.93 86.43 0.45 0.58 (0.35, 0.56) (0.48, 0.67) 0.51 0.75 0.38 0.57 Discharge File 86.17 83.76 0.31 0.55 (0.21, 0.42) (0.45, 0.65) 0.81 0.67 0.38 0.50 Linked Birth & Discharge File 85.75 86.49 0.59 0.81 (0.48, 0.70) (0.73, 0.88) 0.61 0.66 0.51 0.64 201

PAGE 212

Appendix C: (Continued). Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. Table C-4 (Continued). Comparison of validity indices in the Florida birth certicate and linked birth discharge data les by hospital cesarean section rate classication. % Agreement % Agreement Sen Sen Sen 95% CI Sen 95% CI PPV PPV Kappa Kappa High CS Low CS High CS Low CS High CS Low CS High CS Low CS High CS Low CS Chorioamnionitis " " " " " Birth Certicate 98.33 97.29 0.29 0.14 (0.04, 0.71) (0.02, 0.33) 0.33 0.50 0.30 0.21 Discharge File 98.94 97.25 0.67 0.31 (0.18, 0.90) (0.09, 0.61) 0.57 0.50 0.61 0.37 Linked Birth & Discharge File 98.51 97.04 0.86 0.31 (0.42, 1.00) (0.09, 0.61) 0.50 0.44 0.62 0.35 Placenta Previa " " " " " Birth Certicate 95.36 94.96 0.14 0.10 (0.04, 0.32) (0.02, 0.27) 1.00 1.00 0.23 0.18 Discharge File 98.08 98.52 0.71 0.86 (0.48, 0.89) (0.68, 0.96) 0.83 0.89 0.76 0.87 Linked Birth & Discharge File 98.29 98.52 0.76 0.86 (0.53,0.92) (0.68, 0.96) 0.84 0.89 0.79 0.87 Placenta Abruption " " " " " Birth Certicate 94.62 93.80 0.06 0.16 (0.01, 0.21) (0.06, 0.32) 1.00 0.86 0.11 0.26 Discharge File 98.30 95.99 0.79 0.61 * 0.92 0.77 0.84 0.66 Linked Birth & Discharge File 98.29 96.20 0.82 0.67 " 0.88 0.76 0.84 0.69 Prolonged rupture of membranes " " " " " Birth Certicate 90.54 86.62 0 0.01 (0.00, 0.08) 0 1.00 0 0.02 Discharge File 91.27 89.45 0.11 0.22 (0.04, 0.24) (0.13, 0.34) 1.00 0.93 0.18 0.32 Linked Birth & Discharge File 91.27 89.66 0.11 0.24 (0.04, 0.24) (0.14, 0.36) 1.00 0.94 0.18 0.34 Prolonged labor " " " " " Birth Certicate 82.93 80.62 0.02 0.04 (0.00, 0.07) (0.01, 0.10) 1.00 0.57 0.03 0.05 Discharge File 90.22 91.98 0.62 0.66 (0.51, 0.72) (0.55, 0.75) 0.8 0.91 0.64 0.72 Linked Birth & Discharge File 90.22 91.35 0.62 0.66 (0.51, 0.72) (0.55, 0.75) 0.8 0.87 0.64 0.70 Assisted Delivery (Forceps/ Vacuum) " " " " " Birth Certicate 98.33 96.90 0 0.08 (0.00, 0.36) 0 0.20 -0.006 0.10 Discharge File 98.08 97.68 0.17 0.33 (0.00, 0.64) (0.10, 0.65) 0.20 0.57 0.17 0.41 Linked Birth & Discharge File 98.08 97.68 0.17 0.33 (0.00, 0.64) (0.10, 0.65) 0.20 0.57 0.17 0.41 Meconium " " " " " Birth Certicate 96.10 91.86 0.26 0.28 (0.10, 0.48) (0.14, 0.45) 0.60 0.38 0.35 0.28 Discharge File 95.32 93.25 0.14 0.13 (0.03, 0.36) (0.04, 0.29) 0.43 0.50 0.20 0.18 Linked Birth & Discharge File 95.53 91.56 0.33 0.37 (0.15, 0.57) (0.21, 0.56) 0.50 0.37 0.38 0.33 202

PAGE 213

Appendix D: Manuscript Two Supplementary Tables Table D-1. Utilization of healthcare services by mode of delivery among late preterm infants who were delivered primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a Table D-1. Utilization of healthcare services by mode of delivery among late preterm infants who were delivered primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a Table D-1. Utilization of healthcare services by mode of delivery among late preterm infants who were delivered primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a Table D-1. Utilization of healthcare services by mode of delivery among late preterm infants who were delivered primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a ARR 95% CI P value # of Rehospitalizations b 1.17 (1.06,1.28) 0.0013 Rehospitalization c 1.11 (0.99,1.24) 0.0686 a Adjusted for maternal smoking, infant sex, race/ethnicity, maternal age, education, maternal obesity, parity, infant birth weight, payer status, b Adjusted Rate Ratio, c Adjusted Risk Ratio a Adjusted for maternal smoking, infant sex, race/ethnicity, maternal age, education, maternal obesity, parity, infant birth weight, payer status, b Adjusted Rate Ratio, c Adjusted Risk Ratio a Adjusted for maternal smoking, infant sex, race/ethnicity, maternal age, education, maternal obesity, parity, infant birth weight, payer status, b Adjusted Rate Ratio, c Adjusted Risk Ratio a Adjusted for maternal smoking, infant sex, race/ethnicity, maternal age, education, maternal obesity, parity, infant birth weight, payer status, b Adjusted Rate Ratio, c Adjusted Risk Ratio 203

PAGE 214

Appendix D (Continued). Table D-2. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among infants delivered in Florida, 1998-2006, unadjusted and adjusted analyses. Table D-2. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among infants delivered in Florida, 1998-2006, unadjusted and adjusted analyses. Table D-2. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among infants delivered in Florida, 1998-2006, unadjusted and adjusted analyses. Table D-2. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among infants delivered in Florida, 1998-2006, unadjusted and adjusted analyses. Table D-2. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among infants delivered in Florida, 1998-2006, unadjusted and adjusted analyses. Table D-2. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among infants delivered in Florida, 1998-2006, unadjusted and adjusted analyses. Table D-2. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among infants delivered in Florida, 1998-2006, unadjusted and adjusted analyses. Unadjusted Mean Unadjusted Std. Dev. P value a Adjusted Mean Adjusted Std. Dev. P value b Full Study Population " " " Birth Hospitalization " " " Primary C/S 5.4 7.9 <0.0001 3.5 1.0 <0.0001 Vaginal 3.0 3.6 2.3 1.0 First Rehospitalization " " " Primary C/S 5.3 5.3 <0.0001 3.6 1.0 <0.0001 Vaginal 3.6 4.0 2.7 1.0 Second Rehospitalization " " " Primary C/S 6.8 6.5 <0.0001 4.3 1.1 <0.0001 Vaginal 4.1 4.4 3.1 1.0 Cesarean w/o Labor " " " Birth Hospitalization " " " Primary C/S 4.5 5.8 <0.0001 3.3 1.0 <0.0001 Unassisted Vaginal 3.0 3.6 2.3 1.0 First Rehospitalization " " " Primary C/S 5.2 5.1 <0.0001 3.6 1.0 <0.0001 Unassisted Vaginal 3.6 4.0 2.7 1.0 Second Rehospitalization " " " Primary C/S 7.0 6.0 <0.0001 4.5 1.1 <0.0001 Unassisted Vaginal 4.0 4.2 3.0 1.0 a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, TukeyKramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, TukeyKramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, TukeyKramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, TukeyKramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, TukeyKramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, TukeyKramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, TukeyKramer with adjustment for multiple comparisons 204

PAGE 215

Appendix D: (Continued). Table D-3. (Obesity Sub-Analysis). Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births (Vaginal as Referent), 2004-2006, Florida a Table D-3. (Obesity Sub-Analysis). Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births (Vaginal as Referent), 2004-2006, Florida a Table D-3. (Obesity Sub-Analysis). Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births (Vaginal as Referent), 2004-2006, Florida a Table D-3. (Obesity Sub-Analysis). Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births (Vaginal as Referent), 2004-2006, Florida a Table D-3. (Obesity Sub-Analysis). Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births (Vaginal as Referent), 2004-2006, Florida a Table D-3. (Obesity Sub-Analysis). Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births (Vaginal as Referent), 2004-2006, Florida a Table D-3. (Obesity Sub-Analysis). Association Between Primary Cesarean Delivery and Infant Morbidity at Hospitalization During Delivery, the Neonatal Period, and in the First Year, among Classied Low Risk for Cesarean Late Preterm Births (Vaginal as Referent), 2004-2006, Florida a Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Birth Hospitalization Birth Hospitalization Hospitalizations in Neonatal Period Hospitalizations in Neonatal Period Hospitalizations in First Year of Infancy Hospitalizations in First Year of Infancy ARR 95% CI ARR 95% CI ARR 95% CI MORBIDITY " " " Composite Morbidity 1.25 (1.20,1.31) 1.22 (1.17,1.27) 1.22 (1.17,1.27) Feeding Difculties 1.65 (1.37,1.98) 1.41 (1.19,1.67) 1.39 (1.18,1.63) Respiratory Distress 2.33 (2.16,2.51) 1.93 (1.79,2.07) 1.89 (1.75,2.04) Perinatal Infections 2.32 (2.01,2.67) 2.06 (1.80,2.36) 2.01 (1.76,2.30) Jaundice 1.10 (1.04,1.16) 1.06 (1.00,1.12) 1.03 (0.97,1.09) Hypoglycemia 1.81 (1.52,2.16) 1.80 (1.51,2.14) 1.73 (1.45,2.06) Transient Tachypnea 2.29 (2.01,2.62) 2.24 (1.97,2.56) 2.17 (1.90,2.47) a Analyses adjusted for: smoking, infant sex, maternal obesity, race/ethnicity, maternal age, parity, education, payer type, and birth weight b ARR=Adjusted Relative Risk a Analyses adjusted for: smoking, infant sex, maternal obesity, race/ethnicity, maternal age, parity, education, payer type, and birth weight b ARR=Adjusted Relative Risk a Analyses adjusted for: smoking, infant sex, maternal obesity, race/ethnicity, maternal age, parity, education, payer type, and birth weight b ARR=Adjusted Relative Risk a Analyses adjusted for: smoking, infant sex, maternal obesity, race/ethnicity, maternal age, parity, education, payer type, and birth weight b ARR=Adjusted Relative Risk a Analyses adjusted for: smoking, infant sex, maternal obesity, race/ethnicity, maternal age, parity, education, payer type, and birth weight b ARR=Adjusted Relative Risk a Analyses adjusted for: smoking, infant sex, maternal obesity, race/ethnicity, maternal age, parity, education, payer type, and birth weight b ARR=Adjusted Relative Risk a Analyses adjusted for: smoking, infant sex, maternal obesity, race/ethnicity, maternal age, parity, education, payer type, and birth weight b ARR=Adjusted Relative Risk 205

PAGE 216

Appendix E: Manuscript Three Supplementary Tables Table E-1 (Obesity Sub-Analysis) Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 2004-2006, Florida a Table E-1 (Obesity Sub-Analysis) Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 2004-2006, Florida a Table E-1 (Obesity Sub-Analysis) Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 2004-2006, Florida a Table E-1 (Obesity Sub-Analysis) Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 2004-2006, Florida a Table E-1 (Obesity Sub-Analysis) Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 2004-2006, Florida a Table E-1 (Obesity Sub-Analysis) Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 2004-2006, Florida a Table E-1 (Obesity Sub-Analysis) Risk of Maternal Morbidity by Route of Delivery (Vaginal as Referent) at the Birth Hospitalization, the Neonatal Period, and in the First Year, among Women Classied As Low Risk for Cesarean, Late Preterm Births, 2004-2006, Florida a Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Primary C/S Delivery Birth Hospitalization Birth Hospitalization Hospitalizations in Neonatal Period Hospitalizations in Neonatal Period Hospitalizations in First Year of Infancy Hospitalizations in First Year of Infancy ARR b 95% CI ARR b 95% CI ARR b 95% CI MORBIDITY " " " Postpartum Hemorrhage 0.79 (0.48,1.02) 0.61 (0.40,0.92) 0.66 (0.46,0.95) Bladder Injury ** ** ** ** ** ** Venous complications 0.59 (0.29,1.20) 0.77 (0.40,1.47) 0.84 (0.46,1.54) Unspecied febrile conditions 3.48 (2.08,5.83) 3.88 (2.41,6.26) 3.42 (2.25,5.19) Puerperal Infection 5.47 (2.84,10.54) 5.90 (3.47,10.01) 5.07 (3.02,8.49) a Analyses adjusted for: smoking, race/ethnicity, maternal age, maternal obesity, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, maternal obesity, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, maternal obesity, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, maternal obesity, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, maternal obesity, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, maternal obesity, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. a Analyses adjusted for: smoking, race/ethnicity, maternal age, maternal obesity, parity, education, payer type b ARR=Adjusted Relative Risk **Insufcient sample size for multivariable methods. 206

PAGE 217

Appendix E: (Continued). Table E-2. Utilization of healthcare services by mode of delivery among women who delivered late preterm infants, primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a Table E-2. Utilization of healthcare services by mode of delivery among women who delivered late preterm infants, primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a Table E-2. Utilization of healthcare services by mode of delivery among women who delivered late preterm infants, primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a Table E-2. Utilization of healthcare services by mode of delivery among women who delivered late preterm infants, primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a Table E-2. Utilization of healthcare services by mode of delivery among women who delivered late preterm infants, primary cesarean without indications of labor compared to unassisted vaginal deliveries (referent) Florida 1998-2006 a ARR b AHR c 95% CI P value # of Rehospitalizations 1.50 (1.14,1.98) 0.0035 Rehospitalization 1.76 (1.45,2.12) <0.0001 a Adjusted for maternal smoking, race/ethnicity, maternal age, education, parity, payer status, b Adjusted Rate Ratio, c Adjusted Hazard Ratio a Adjusted for maternal smoking, race/ethnicity, maternal age, education, parity, payer status, b Adjusted Rate Ratio, c Adjusted Hazard Ratio a Adjusted for maternal smoking, race/ethnicity, maternal age, education, parity, payer status, b Adjusted Rate Ratio, c Adjusted Hazard Ratio a Adjusted for maternal smoking, race/ethnicity, maternal age, education, parity, payer status, b Adjusted Rate Ratio, c Adjusted Hazard Ratio a Adjusted for maternal smoking, race/ethnicity, maternal age, education, parity, payer status, b Adjusted Rate Ratio, c Adjusted Hazard Ratio 207

PAGE 218

Appendix E: (Continued). Table E-3. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among women who delivered a late preterm singleton infant, Florida, 1998-2006, unadjusted and adjusted analyses. Table E-3. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among women who delivered a late preterm singleton infant, Florida, 1998-2006, unadjusted and adjusted analyses. Table E-3. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among women who delivered a late preterm singleton infant, Florida, 1998-2006, unadjusted and adjusted analyses. Table E-3. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among women who delivered a late preterm singleton infant, Florida, 1998-2006, unadjusted and adjusted analyses. Table E-3. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among women who delivered a late preterm singleton infant, Florida, 1998-2006, unadjusted and adjusted analyses. Table E-3. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among women who delivered a late preterm singleton infant, Florida, 1998-2006, unadjusted and adjusted analyses. Table E-3. Differences in the mean LOS at birth hospitalization, rst rehospitalization and second rehospitalization by route of delivery among women who delivered a late preterm singleton infant, Florida, 1998-2006, unadjusted and adjusted analyses. Unadjusted Mean Unadjusted Std. Dev. P value a Adjusted Mean Adjusted Std. Dev. P value b Full Study Population " " " Birth Hospitalization " " " Primary C/S 4.4 5.9 <0.0001 3.4 1.0 <0.0001 Vaginal 2.4 2.4 2.2 1.0 First Rehospitalization " " " Primary C/S 3.7 5.1 <0.0001 2.7 1.0 <0.0001 Vaginal 3.0 2.9 2.3 1.0 Second Rehospitalization " " " Primary C/S 5.1 3.9 0.0006 3.7 1.1 0.0179 Vaginal 4.0 4.2 3.0 1.0 Cesarean w/o Labor " " " Birth Hospitalization " " " Primary C/S 4.1 5.0 <0.0001 3.1 1.0 <0.0001 Unassisted Vaginal 2.5 2.4 2.2 1.0 First Rehospitalization " " " Primary C/S 3.8 6.4 <0.0001 2.6 1.0 0.0002 Unassisted Vaginal 3.1 3.2 2.3 1.0 Second Rehospitalization " " " Primary C/S 5.4 4.3 0.0048 3.8 1.0 0.0136 Unassisted Vaginal 4.0 4.8 2.9 1.0 a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, Tukey-Kramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, Tukey-Kramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, Tukey-Kramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, Tukey-Kramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, Tukey-Kramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, Tukey-Kramer with adjustment for multiple comparisons a Unadjusted results, Wilcoxon Rank Sum Test; b Adjusted results (maternal age, race/ethnicity, education, parity, infant sex, payer type, smoking status), ANCOVA with log transformation, Tukey-Kramer with adjustment for multiple comparisons 208

PAGE 219

Appendix F: University of South Florida Institutional Review Board Exempt Status Determination Letter 209

PAGE 220

Appendix F: (Continued) 210

PAGE 221

About the Author !Heather Breeze Clayton received her Bachelor's Degree in Environmental Analysis and Design at the University of California at Irvine in 2000 and a Master of Public Health Degree from San Diego State University in Epidemiology in 2003. In 2003, Ms. Clayton was selected for the Association of Schools of Public Health/Centers for Disease Control and Prevention (ASPH/CDC) Fellowship in Reproductive Epidemiology. She worked as a ASPH/CDC fellow until 2005, when she enrolled in the Ph.D. program at the University of South Florida, College of Public Health. !Ms. Clayton was very active in teaching and research as a doctoral student, with over 17 publications during her ve year course of study. She received the Maternal and Child Health Epidemiology Traineeship, which provided additional opportunities for research and collaboration with the Florida Department of Health and the Lawton and Rhea Chiles Center for Healthy Mothers and Babies.