Pregnancy to Early Life Longitudinal (PELL) Data System: Research into Child Health Policy Child Health Services Research Interest Group AcademyHealth June 27, 2009 Chicago, Illinois Milton Kotelchuck, PhD, MPH Boston University School of Public Health Goals of Presentation • Discuss the importance of the creation of State population-based, longitudinal, linked MCH data systems for child health services research and policy • Describe the Massachusetts Pregnancy to Early Life Longitudinal (PELL) data system • Provide some examples of current policy-related usages of the PELL data system • Share lessons learned for the development and further enhancement of state population-based, longitudinal, linked data systems Value of Population-based Longitudinal Linked Data Systems • Provides information that free standing (silo) data sets can not • Can provide both numerators and denominators for analyses • Allows both individual and contextual level analyses • Supports life course models of health and development • Facilitates evaluation of programmatic and policy initiatives • Provides a child health services research database Massachusetts PELL Data System • Collaborative public-university partnership • Maternal and Child Health Department, Boston University School of Public Health • Massachusetts Department of Public Health (MDPH) • Centers for Disease Control and Prevention (DRH (MIHB, ASB); NCBDDD) PELL Overview • Population-based, relational, reproductive and child health data system • Funded to assess impact of prenatal environment on subsequent child (and maternal) health • Utilizes a broad range of public health data • Longitudinal data system – tracks mothers and children over time • Allows for multiple kinds of linkages and analysis • Breadth of data bases is expandable • Offers a conceptual and practical model for other states Nature of Pell Data System • Consists of a birth-related Core – birth/fetal deaths linked to hospital discharge delivery records • Other surveillance, hospital services and program data linked to birth and fetal death certificates • Data connected by unique random identifiers for births, children, and mothers • All confidential data reside on secure servers at the MDPH • Authorized by the Commissioner of Public Health under MA General Law Chapter 111, Sections 24A and 24B PELL and Child Health Services Policy Research • PELL started as a small research study; it hadn’t initially been envisioned as a larger data system • First study, however, was an evaluation of the MA Healthy Start program, with major policy and programmatic consequences • Now has expanded into an ongoing MDPH core data system for multiple program activities • PELL increasingly used for policy analyses, but these have just begun and its capacity is still under-utilized Core Data Sets in PELL Birth Certificate Hosp. Discharge Delivery (Mother) Fetal Death Record Fetal Death Certificate Hosp. discharge Birth record PELL Data System Program Participation Data Vital and Health Status Data WIC Early Intervention CORE Newborn Hearing Screening Birth Defects Registry Birth Certificate HD Delivery Mothers Child and Mother deaths Pregnancy-associated deaths Other MDPH programs (future) Linked birth-infant deaths Fetal Death HD Birth - Child Health Services Utilization Data Non-birth Hospital Discharge Contextual Data Observational Stays Geocoded birth data Emergency Department Census 2000 Data (2008) Other Contextual Data (future) Other Future Datasets: School, NICU, “ART”, Medicaid Conception Delivery Intrauterine (Fetal) Pregnancy (Mother) One Year Infancy/Mother Postpartum Five Years Early Childhood/ Interconceptional (Mother) Birth Certificate/ Fetal Death Hospital Discharge Observational Stay/ED WIC Early Intervention Infant/Child Death Maternal/Mother Death Examples of PELL Longitudinal Data Capacity Types Of Longitudinal Linkages • Non-Birth Hospital Related Services • Sequentially-Linked Births/Fetal Deaths • Program Use by Individuals and Families PELL Core Data: 1998 - 2006 Total Live Births Total Fetal Deaths Total Mothers Repeat Mothers 734,987 3,896 514,433 174,221 NOTE: Includes all births to MA residents (in and out of state) and in state births to non-MA residents Technical Issues: PELL Linkage Methodology • Linkage without common unique identifiers • Combination of Deterministic and Probabilistic matching • LinkPro 2.0 (inexpensive, easy to use, effective) • Set of linkage programs, with a linker file (similar to a relational database) • Unique identifier randomly generated to facilitate linkage and assure confidentiality • Dyadic linkage flexibility (via linker file) Data Base Structure Hosp Disch Mother data MOMUID Hosp Disch Child data HDMOMUID KIDUID HD MOM - Matched HDKIDUID HD MOM - Weighted HDKID - Matched All other vars Linker File ALLMOMUID Healthy Start Mother data MOMUID MOMUID KIDUID HSUID HDKID - Weighted All other vars EI data HS - Matched KIDUID HS - Weighted EIUID All other vars BC/FD data EI - Matched ALLMOMUID EI - Weighted MOMUID All other vars KIDUID All other vars PELL Core Linkage: Vitals to Hospital Discharge Data • No Unique Identifier • Core Linkage Variables: • Facility code • Medical Record Number (babies only) • Date of Birth/Date of Delivery (from hospital discharge file) • Sex (babies only) • Zip Code • Birth weight used to resolve duplicates PELL Core Linkage Rates 1998-2006 Type of birth Live births linked to HD^ birth record Linkage Rate* (# of Records) 99.2% (708,467) Fetal deaths linked to HD^ delivery record * Among occurrence births to Massachusetts residents ^ Hospital Discharge 98.7% (3,810) PELL Program Linkage Rates Program Early Intervention* records linked to live births Birth defect cases^ linked to live births and fetal deaths Linkage Rate (# of Records) 86.9% (46,857) 99.7% (5,299) * Among EI children enrolled between January 1998 – September 2003 and born between January 1998 – September 2000 ^ MA Birth Defects Monitoring Program, 1998-2002 Technical Issues Summary • Technical issues are no longer the principal barrier for linked data systems • Growing project experience and newer computer programs facilitates linking data • However, life would be much easier with a common and universal identifier Confidentiality/Access to Database Issues • Major issue in constructing linked databases • Linked databases involve/require confidential data • Access involves both political and professional concerns • Substantial inter-and intra-agency turf issues • More complex in Public/Private partnership • HIPAA further complicated access to health databases Confidentiality • Commissioner of Public Health 24AB confidentiality approval required for PELL development and each analytic study • M.G.L. Chapter 112 Section 24AB stronger confidentiality protection than HIPAA • All data are held and linked at MDPH on a secure server • Only “de-identified” or “limited use” data sets are created for approved analytic studies at MDPH or offsite • MDPH RaDAR (Research and Data Access Review) Committee provides an institutional mechanism to review research/confidentiality requests using state public health data bases PELL Current Analytic Studies 1. Impact of Multiple Births in Massachusetts Goal: To examine the effects of multiple pregnancies and multiple births on fetal/infant and maternal morbidity and mortality, and assess the use of acute health care resources 2. Interpregnancy Period Project Goal: To use longitudinal core PELL data to examine and refine measures of interpregnancy periods (IPP) 3. Pregnancy-Associated Injuries Goal: To assess the prevalence of pregnancy-associated injuries (PAI) and examine health disparities related to race/ethnicity, health insurance and age 4. Impact of Late Pre-term Delivery on Infant and Maternal Health Goal: Assess the impact of late preterm (34-36 weeks) births on infant and maternal health and health services usage and costs PELL Current Analytic Studies 5. Maternal and Infant Morbidity Associated with NoIndicated Risk Cesareans Goal: This series of studies will examine the risks for maternal morbidities associated with method (vaginal vs cesarean) of delivery, with a particular emphasis on cesareans associated with no identified medical risk. 6. Children with Craniofacial Malformations in Massachusetts: Service Utilization and Associated Costs – Children with Down Syndrome in Massachusetts Goal: Assess acute hospital-related health care service utilization and associated costs, along with selected indicators of morbidity and mortality, for children born in Massachusetts from 1998 through 2002 with craniofacial malformations (i.e. oral clefts, craniosynostosis, microtia) –and Down syndrome 7. An Assessment of the Massachusetts Early Intervention Program Goal: Assess maternal and infant birth characteristics that predict EI referral from the entire Massachusetts population and compare infants who were referred to those infants who were not referred to EI on a variety of birth, infant, and maternal demographic characteristics PELL Current Analytic Studies 8. Gestational Diabetes, Sequential Pregnancies Goal: This series of studies will examine the risk factors for recurrence of gestational diabetes in sequential pregnancies and the extent of conversion to diabetes, as well as the infant and maternal health sequelae of GDM, with the aim of enhancing the MA Diabetes and Obesity reduction initiatives 9. Fetal Death Surveillance Goal: To improve the surveillance of Fetal Deaths, with a focus on improved classification of cause of death, enhanced measurement of risk, with the added information from hospital discharge records. 10. Autism Risk and Surveillance Goal: Assess the infant, maternal and programmatic characteristics associated rising temporal trends in early detection and referral to EI treatment of MA children (0-3) with Autism Spectrum Disorders. PELL Current Analytic Studies (Initiated) 11. Evaluation of new MA Obstetric Regulations Goal: To evaluate the new MA Obstetric Regulations, to see if they achieved their goals of improved maternal and infant transfers, volume performance levels, and decrease in inappropriate services at Level I and II hospitals. 12. Enhanced Assessment of Maternal Drug and Alcohol Treatment Needs in Massachusetts Goal: To improve the needs assessment capacity for the MA DPH Bureau of Substance and Alcohol Services, by better identifying need for substance treatment among women who have recently delivered infants but do not currently participant in MDPH treatment programs, through use of emergency room, hospital discharge and birth certificate data. 13. MOSART: MA Outcomes Study of Assisted Reproductive Technologies Goal: Assess the maternal, infant and child sequalae of specific ART treatments, through the linking of detailed clinical ART treatment data from SART with population-based data from PELL, considering both the underlying infertility and the impact of multiple-birth pregnancies; initial study on child health sequalae. Examples of PELL Child Health Policy Research Usage • Evaluation of the Early Intervention Program • Elective C-Sections • Autism Surveillance • Late Pre-term Births and Neonatal Morbidity Use of PELL Population-Based Data to Evaluate the MA Early Intervention Program Evaluation Question: Is EI serving the at risk of delay population? Data: Births and Linked PELL Core 1998-2000, Early Intervention Data through 2003 Clemens et al 2006, 2008; Barfield et al 2008 Background: EI in Massachusetts • Federally mandated - Individuals with Disability Education Act, part C • Integrated services to enhance development for children ages 0 - 3 with developmental delay, some states enroll at-risk children • Substantial state and 3rd party insurer investment • Multiple “service” levels Referral • MD, hospital, parents • Child Find Evaluation IFSP - Enrolled 4 Eligibility categories • developmental delay • medical diagnosis (established condition) • 4+ risk factors • clinical judgment Q3: Referral differs by risk Percentage of infants < 1200 g referred to EI within 12 months, by maternal race/ethnicity and insurance, MA Births 1998-2000 100 93 % referred 12 months 82 88 89 Hispanic Other 93 84 80 60 40 20 0 White Black Comm govt, none •After adjusting for additional birth risks, infants born to black mothers, mothers with no insurance and Boston region mothers were referred later and were more likely to be not referred to EI Q4: Services After Referral Evaluation: • Of referrals, 86% had an evaluation for eligibility • Children of teen mothers were 8% less likely to have an evaluation • Children born to mothers living in a high poverty town also were less likely to have an evaluation Enrollment: • Of eligible children, 90% enrolled • Children of teen mothers were less likely to enroll than children of older mothers Maternal characteristics of MA infants, referral, evaluation, and enrollment in Early Intervention (Births 1998-2000) Characteristic N (%) Births N (%) Referred N (%) Evaluated N (%) Enrolled 219,001 40,711 35,707 29,950 White 163,537 (74.9) 28,192 (69.4) 25,223 (70.7) 21,331 (71.4) Black 15,370 (7.0) 3,741 (9.2) 3,019 (8.5) 2,470 (8.3) 24,238 (11.1) 6,366 (15.7) 5,395 (15.1) 4,419 (14.8) Asian 11,243 (5.2) 1,459 (3.6) 1,268 (3.6) 1,069 (3.6) Other 4,074 (1.9) 887 (2.2) 743 (2.1) 610 (2.0) Overall Maternal Race Hispanic Barfield et al 2008 Policy Benefits to EI Program • Demonstrated child find success • Better identified populations who are not referred or disproportionately lost to followup after referral • Informed improved outreach strategies, service and program monitoring • Enhanced some hospital referral practices • Expanded evaluation capacity of EI program • Improved EI- PELL linkage data variables Maternal Outcomes with Medically Elective Caesareans Mass. 1998-2003 US Cesarean Rates, 1989-2005 32 30 28 % 26 %24 22 20 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 % Tot US 22.8 22.7 22.6 22.3 21.8 21.2 20.8 20.7 20.8 21.2 22.0 22.9 24.1 26.1 27.5 29.1 30.2 If in 2005 the cesarean rate was the same as in 1996, there would have been 339,015 fewer cesareans in the U.S. Source: National Center for Health Statistics Annual Birth Reports This group is key analytic difference from earlier analyses Medically Elective Primary Cesareans Source: Declercq et al. Obstet Gynecol. 2007.109: 669-677. Planned Primary Caesarean Rate (%), Mass., 1998-2004 3 2.63 2.17 1.62 1.73 2001 2002 % 1.33 1.03 0.43 0 1998 1999 2000 2003 Source: Declercq et al. Obstet Gynecol. 2007.109: 669-677. 2004 Rehospitalization of Mothers by Method of Delivery, Mass. 1998-2003 Planned Primary Caesarean (n = 3,334) Planned Vaginal Birth Rehosp. Rate/1,000 (n = 240,754) 1-30 days Postpartum 7.5 31-180 days Postpartum 9.8 181-365 days Postpartum 9.2 Rehosp. Rate/ 1,000 NIR Unadjusted Odds Ratio PCS (95% CI) Births 2.6 19.2 (2.01-3.33) 1.4 13.5 (1.02-1.86) 1.6 14.6 (1.19-2.13) * Adjusted for age, race/ethnicity, parity; Source: Declercq et al. Obstet Gynecol. 2007.109: 669-677. Adjusted* Odds Ratio (95% CI) 2.3 (1.74-2.90) 1.6 (1.08-1.98) 1.8 (1.32-2.38) Postpartum rehospitalization (1-30 days) rates by diagnosis for mother by MoD, Massachusetts, 1998-2003 Planned Vaginal (n= 240,754) Cause Rate/1,000 Major puerperal infection Planned Caesarean (n=3,334) Cause Rate/1,000 1.83 Complication of obstetrical wounds 6.60 Nonpurulent mastitis 0.73 Major puerperal infection 3.30 Delayed & secondary postpartum hemorrhage 0.62 Care and observation 2.40 0.54 Inflammatory diseases of uterus 1.50 0.46* Infections of the genitourinary tract 1.50 0.44 Delayed & secondary postpartum hemorrhage 1.20 Infections of the genitourinary tract Complication of obstetrical wounds Other complications * For unplanned cesareans alone rate was 3.97/1,000 Source: Declercq et al. Obstet Gynecol. 2007.109: 669-677. Average Hospital Length of Stay and Costs by Method of Delivery, Mass., 1998-2003 Planned Vaginal Planned Primary Cesareans Initial Hospital Stay Avg. Hospital Costs* (95% CI) $2,513 $4,373 ($2,507 -- $2,519) ($4,304 -- $4,441) Subsequent Rehospitalization in 12 months Postpartum Avg. Hospital Costs* (95% CI) $5,436 $6,100 ($5,221--$5,651) ($4,745 -- $7,455) * Charge data adjusted by cost to charge ratios obtained from the MA Div. of Healthcare Finance and Policy and adjusted for inflation to reflect 2003 dollars. Source: PELL Data System Source: Declercq et al. OBGYN 2007. 109:669-77. Policy Responses to C-Section Analyses • Two MA State panels examining rising C-section rates (MDPH, MMS) • New elective C-Section items on MA PRAMS surveys • Participation in two NICHD consensus conferences on elective C-sections and VBACs • Fostering national debate on elective C-sections • Renewed focus on late preterm deliveries, the majority of which are C-sections Surveillance for Early Diagnoses of Autism Spectrum Disorders in Massachusetts Susan E. Manning, MD, MPH LCDR U.S. Public Health Service CDC MCH Epidemiology Program Assignee Massachusetts Department of Public Health Autism Spectrum Disorders (ASD) • Autism, Asperger’s, pervasive developmental disorder, not otherwise specified (PDD-NOS) • Lifelong disorders with onset < age 3 years • Impaired communication/social functioning, unusual behaviors and interests • Prevalence: 66/10,000 (1 in 150) children aged 8 yrs (CDC, 2007) • Early diagnosis and treatment can improve developmental outcomes • MA Early Intervention Program (0-3) provides a very active and comprehensive specialty service program for children (0-3) diagnosed with ASD Methods • Population: In-state births to MA resident mothers, Jul 1999-Jun 2003 • Case-definition: – ICD-9 code for Autism, PDD-NOS, Asperger’s disorder in EI database or autism Specialty Services claim during Jul 1999June 2006 • Early ASD diagnosis: Diagnosis of ASD before age 36 months N = 1,935; 68% based on diagnoses, 32% on service claims • Data analysis – Trends in early ASD diagnoses – Comparison of rates by socio-demographic characteristics ASD/10,000 live births Trend in Early ASD Diagnoses, MA Births – Jul 1999-Jun 2003 100 90 80 70 60 50 40 30 20 10 0 69.9 75.9 56.0 52.2 54.1 * p<0.0001 1999 2000 2001 Birth Year Source: PELL Data System. * Cochran-Armitage trend test. 2002 2003 Trends in Early ASD Diagnoses, by Age at Diagnosis, MA Births – Jul 1999-Jun 2003 0-24 mos 24+ mos ASD/10,000 live births 75 60 45 52.9 35.7 37.7 * p<0.0001 58.4 41.6 30 15 13.0 15.1 13.6 16.6 0 1999 2000 2001 Birth Year Source: PELL Data System. * Cochran-Armitage trend test. 2002 17.2 * p=0.092 2003 Trends in Specific ASD Diagnoses, MA Births – Jul 1999-Jun 2003 Autism PDD-NOS ASD/10,000 live births 50 40 * p<0.0001 30 20 10 26.6 27.2 16.4 15.7 * p=0.003 22.9 17.5 10.5 19.0 12.1 13.3 0 1999 2000 2001 Birth Year Source: PELL Data System. * Cochran-Armitage trend test. 2002 2003 Trends in Early ASD, by Geographic Region, MA Births – Jul 1999-Jun 2003 Birth Year 1999 2000 2001 2002 2003 /10,000 /10,000 /10,000 /10,000 /10,000 Boston 43 54 50 65 88 0.003 Central 67 31 48 81 71 0.003 Metrowest 56 73 63 74 72 0.270 Northeast 59 60 60 76 78 0.025 Southeast 51 42 53 55 68 0.047 Western 27 49 54 70 88 <0.0001 Source: PELL Data System. * Cochran-Armitage trend test. p for trend* Early ASD Diagnoses by Maternal Age, MA Births – Jul 1999-Jun 2003 # ASD/10,000 births * 90 80 70 60 50 40 30 20 10 0 84.4 * 56.4 30.5 <20 20-34 Maternal age (yrs) * Statistically significantly different from age < 20 years (p<0.05). 35+ Early ASD Diagnoses by Race/ethnicity, MA Births – Jul 1999-Jun 2003 # ASD/10,000 births 70 66.4 58.0 60 * 50 * 46.3 39.2 40 30 20 10 0 White NH Black NH Hispanic Race/ethnicity * Statistically significantly different from White NH (p<0.05). Other Early ASD Diagnoses by Maternal Education, MA Births – Jul 1999-Jun 2003 # ASD/10,000 births 80 70 * * 65.0 70.8 Some college 4+ years 60 50 47.7 40 30 20 10 0 HS or less Maternal education * Statistically significantly different from HS or less (p<0.05). Early ASD Diagnoses by Insurance, MA Births – Jul 1999-Jun 2003 # ASD/10,000 births 80 70 67.5 * 60 47.4 50 40 30 20 10 0 Private Public Insurance * Statistically significantly different (p<0.05). ASD Conclusions • ASD diagnoses among MA children by age 3 appear higher than expected based on national estimate among 8 year-olds • Early diagnoses of ASD are increasing in MA • Socio-demographic disparities in early ASD diagnoses exist • Multiple explanations for increasing trend in early ASD diagnoses Public Health Implications • High cost of ASD services and parent demand driving intense focus on this population • PELL enhances monitoring trends and identifying disparities in early ASD diagnoses • Results used to inform EI program and anticipate future service demand and resource needs • Major EI program initiatives on disparities in early diagnoses, and early diagnosis criteria • Major initiative to link EI (0-3) ASD records to school special education records (3-21), to examine success of early ASD identification, continuity of diagnosis, and impact of EI services on subsequent school experiences • Multiple research initiatives Risk Factors for Neonatal Morbidity Among Healthy, Late Pre-term Newborns Carrie K. Shapiro-Mendoza Co-authors: KM Tomashek, M Kotelchuck, J Weiss, W Barfield, S Evans, A Nanini Background • Late pre-term births (34-36 weeks gestation) are increasing. • In 2002, 8.5% of U.S. births were late pre-term • Late pre-term newborns typically receive routine care in well-baby nurseries and are presumed low risk. • Limited knowledge about their neonatal outcomes. • Understanding which late pre-term infants are at greater risk is essential for preventing post discharge complications. PELL Late Pre-term Studies 1. Effects of late preterm and maternal medical conditions on newborn morbidity (Shapiro-Mendoza et al 2008) 2. Risk factors for neonatal morbidity and mortality among “healthy” late preterm newborns (Shapiro-Mendoza et al 2008) 3. Early discharge among late preterm and term newborns and risk for neonatal morbidity (Tomashek et al 2006) Study 1: Effects of late preterm and maternal medical conditions on newborn morbidity • Morbidity during the birth hospitalization: – Newborns with • Hospital stay < 5 nights and transfer to another medical facility • Hospital stay ≥ 5 nights and significant morbid diagnosis by ICD codes • Death prior to hospital discharge • Maternal medical conditions (assessed individually) Risk of morbidity during birth hospitalization by gestational age 37-41 3.0 Gestational age in weeks 34-36 22.2 41 2.8 40 2.5 39 2.6 38 3.3 37 5.9 36 12.1 35 25.6 34 51.7 0 10 20 30 Risk (%) 40 50 60 Risk of infant morbidity during birth hospitalization by GA and intrauterine exposure to selected maternal medical conditions, and effect measure modification, MA 1998-2003 Risk of Morbidity N (%) Maternal medical conditions & pregnancy complications No maternal medical conditions Term Late Preterm (n=386,688 (n=17,120) *Adjusted risk ratio for morbidity RR (95% CI) Term Late Preterm (n=386,688) (n=17,120) 5799 (3.0) 11321 (22.1) 1.0 7.0 (6.8, 7.2) Cardiac Disease 301 (4.4) 144 (25.4) 1.5 (1.4, 1.7) 8.6 (7.5, 10.0) Acute or chronic lung disease 778 (4.9) 479 (30.2) 1.6 (1.5, 1.7) 9.3 (8.6, 10.0) Diabetes mellitus 890 (5.0) 565 (28.4) 1.7 (1.6, 1.8) 9.3 (8.6, 10.0) 1345 (5.2) 1259 (32.5) 1.8 (1.7, 1.9) 10.8 (10.3, 11.4) 434 (7.1) 658 (37.2) 2.4 (2.2, 2.7 12.4 (11.6, 13.2) 1592 (7.2) 485 (30.0) 2.5 (2.4, 2.6) 9.8 (9.1, 10.6) Hypertensive disorders of pregnancy Antepartum hemorrhage Maternal infection Study 2 : Risk factors for neonatal morbidity among healthy late pre-term newborns • Isolated healthy late pre-term newborn population (2.6% of all MA singleton, vaginal births) • 6.1% of healthy late pre-terms had a subsequent rehospitalization (neonatal morbidity) • Risk factors for rehospitalization: breastfeeding at discharge, Asian race/ethnicity, primiparae, and L&D complications • First population-based study to identify risk factors for neonatal morbidity among healthy near-term newborns. Leading neonatal morbidity diagnoses Digestive disorders 3% Other 15% Respiratory problems 3% Feeding difficulties 4% Infection 14% Jaundice 61% Study 3: Early discharge among late pre-term and term newborns and risk for neonatal morbidity • Sample: All infants discharged less than two days post delivery • Being late pre-term was associated with a nearly two-fold risk of neonatal hospital readmission (controlling for multiple other variables) • First population-based study to identify risk factors for neonatal morbidity among preterm newborns discharged early. PELL Studies of Late Pre-term Infants: Policy Impacts • Knowledge base influencing practice and policy • Major changes in U.S. obstetrician perceptions of the health of late preterm infants and their rationale for early deliveries • Supports March of Dimes Prematurity Prevention initiatives • U.S. increasing C-section rate slows in 2007, associated with decrease in late preterm Csection deliveries • (Irony) MA obstetricians telling PELL staff to examine late preterm deliveries • Late pre-term delivery a major new CDC research focus PELL and Child Health Policy Research: Lessons Learned • University (Private) – Public Partnership Critical • Ongoing Data System, Not One Time Research Data Linkage Project • Capacity for Continued Evolution and Enhancement • Science of Using Linked Data Systems • Align with MCH Life Course Models PELL and Child Health Policy Research: Lessons Learned • Individual Level Data, Creative Linkages • Multiple Outcome Measures (utilization, health status, costs) • Capacity for Child Health Services Research in Three Domains – Monitoring trends – Program evaluation – Policy assessment • Data Redundancy as a Source of Strength PELL and Child Health Policy Research: Lessons Learned • Long-term Commitment Required for Development of State Data Systems • Policy Makers – Child Health Policy Researchers an Uneasy and Often Changing Relationship • Current Limitations to Further Development – – – – – FERPA limitations on access to school health data Confidentiality issues in era of linked data systems Retrospective versus real time data systems Ambulatory data systems not well incorporated Linkage of Medicaid, HMO data remains problematic • PELL Data System Offers Model for Other States Conclusions • Population-based, longitudinal, linked State based data systems, like PELL, can influence public policy at the state and national level • Policy influenced by providing both new knowledge and new programmatic and policy evaluations • Advancing the science of linked data systems is critical for improving policy development • PELL still in early stages of influencing MA State public health policy and programs PELL Data System Advancing the MCH Sciences by Using Linked Data Systems Advancing the MCH Sciences of Using Linked Data Systems Gestational Diabetes • Impact on maternal health and on subsequent pregnancies • Impact on childhood obesity and diabetes Percent of deliveries with gestational or other pre-existing diabetes, Massachusetts, 1998-2004 Vitals^ PELL* Year Gest. Diabetes Only Gest. Diabetes Only Other preexisting diabetes All diabetes 1998 2.4 3.5 0.8 4.3 1999 2.6 3.7 0.9 4.6 2000 2.8 4.1 0.9 5.0 2001 3.0 4.3 0.9 5.2 2002 3.3 4.3 1.0 5.4 2003 3.4 4.5 1.0 5.5 2004 3.5 4.7 1.2 5.9 ^ Gestational diabetes reported on birth or fetal death certificate only * Diabetes reported on birth or fetal death certificate, or hospital discharge delivery record Gestational diabetes, other diabetes, or both, as reported on hospital discharge vs. birth/fetal death certificates, MA residents, CY 1998-2004 Birth/Fetal Death Cert No Gest. Diabetes Diabetes No Diabetes 521,738 Gest. Diabetes Hospital Discharge Data Other Diabetes Both (Gest and Others) Total 6,763 411 50 528,962 3,345 12,713 329 85 16,472 Other Diabetes 1,098 461 1441 15 3,015 Both (Gest. and other) 354 871 95 32 1,352 526,535 20,808 2,276 182 549,801 Total Gest Only: 22,821 Other: 5,249 Total: 28,070 Source: PELL GDM Research Policy Implications • Multiple MA State initiatives on GDM, diabetes and obesity reduction efforts directed at pregnant women, children and adults • Data strategy a key component of the MA State initiatives • Improvements in GDM measurement (PRAMS, birth certificates) • New PELL research on sequential births • Improved communication of incidence and risk MK involvement in linked data systems • MDPH – Evaluation of MA WIC Program, through linkage of WIC participation records and Vital Statistics data – ROSS Reproductive Outcome Surveillance System • UNC-CH – Evaluation of NC Medicaid Baby Love Program; Assessment of Maternal Morbidity – Linkage of Birth Certificates, Medicaid, WIC records • HRSA – Facilitate linkage of Medicaid, CDC/NCHS, and HRSA data – Support State data linkage efforts • BUSPH – Uterine Rupture/VBAC study – Linkage of birth certificates and hospital discharge records