The Perinatal Periods of Risk Approach Sanil Thomas MS Biostatistics candidate April 27, 2010
Introduction Infant mortality rate (IMR) is a critical indicator of nation’s health IMR remains higher in United States than in other industrialized countries But IMR does not provide sufficient information to understand the factors that contribute to infant mortality
Introduction Traditional methods don’t include the fetal death counts for the analysis of mortality rates Fetal-infant mortality is a multi dimensional issue and a detailed analytical approach to fetal- infant mortality is needed to focus community initiatives for improving maternal and infant health.
Objectives To look at Feto-Infant mortality in a new way Apply PPOR framework for New York State To see the distribution of common risk factors by county level
Perinatal Periods of Risk (PPOR) Approach The Perinatal Periods of Risk Approach was developed by Dr. Brian McCarthy from the W.H.O. Perinatal Collaborative Center at CDC and other W.H.O. colleagues. Simple method that is based on a strong conceptual prevention The PPOR Data allow you to look at feto-infant mortality in new ways
PPOR : 6 Basic Steps Step 1: Assure Analytic and Community Readiness Step 2: Conduct Analytic Phases of PPOR Step 3: Develop Strategic Actions for Targeted Prevention Step 4: Strengthen Existing and/or Launch New Prevention Initiatives Step 5: Monitor and Evaluate Approach Step 6: Sustain Stakeholder Investment and Political Will
Analytic Phases of PPOR Phase 1: Identifies populations and periods of risk with the largest excess mortality. Phase 2: Explains why the excess deaths occurred.
PPOR Examines Deaths in TWO dimensions simultaneously: Age at death Weight at birth
Conception Birth 1 Year Fetal Infancy 20 wks28 wks 4 wks Spontaneous Abortion Early Fetal Late Fetal Neonatal Postneonatal Infant Feto-Infant Age at Death The First Dimension Of PPOR Analysis :
Second Dimension: Birthweight Very Low Birthweight (PPOR limit) = less than 1500 grams (3.3 pounds) Low Birthweight = less than 2500 grams (5.5 pounds) Normal Birthweight e.g., a 7.5-pound baby weighs 3,400 grams Birthweight
PPOR “Map” fetal & infant deaths Age at Death Birthweight g g Fetal (24+ wks) NeonatalPostneonatal
PPOR “Map” fetal & infant deaths g g Fetal Death Neonatal Post- neonatal Maternal Health/ Prematurity Maternal Care Newborn Care Infant Health Birthweight Age at Death
PPOR is about ACTION (each period of risk is associated with a set of possible areas for action) Maternal Health/ Prematurity Maternal Care Newborn Care Infant Health Preconception Health Health Behaviors Perinatal Care Prenatal Care High Risk Referral Obstetric Care Perinatal Management Neonatal Care Pediatric Surgery Sleep Position Breast Feeding Injury Prevention
PPOR: Phase 2 Poisson log linear modeling Covariates/Fixed effects a.Mother’s race b.Mother’s education c.Mother’s age d.Payor Random effect - County
PPOR: Phase 2 Fetal death was not used Predicted death counts were used to obtain smoothed death rates Modeling done for each county Relative risk calculated from the beta estimates
Data New York State Dept. of Health. a.Electronic records of births b.Linked birth-death cohort c.selected fetal deaths Phase 1 : Phase 2 :
Software used SAS Excel ArcGIS
Results
Phase 1 Results Obs County Total number of deaths Sum of Live births and Fetal deaths Death rates per 1000 live births and Fetal deaths (PPOR categories) Total Death Maternal health/prematuri ty Maternal careNewborn careInfant health 1ST LAWRENCE BROOME OSWEGO BRONX JEFFERSON SCHENECTADY KINGS ONEIDA ULSTER ALBANY RENSSELAER ERIE ONONDAGA QUEENS MONROE NEW YORK NIAGARA ORANGE RICHMOND ROCKLAND WESTCHESTER SUFFOLK NASSAU SARATOGA DUTCHESS
Phase 1 Results New York State Death Rates per 1000 live births and fetal deaths 2.58 Maternal Health/ Prematurity 1.63 Maternal Care 1.12 Newborn Care 1.21 Infant Health
Phase 2 Results Poisson Log Linear model estimates Solutions for Fixed Effects Effectmom_racemom_educnmom_agepayorEstimate Standard ErrorDFt ValuePr > |t| Intercept <.0001 mom_race asian mom_race black_alone <.0001 mom_race other_races mom_race z_white_alone0.... mom_educn HighSchoolorAssoc <.0001 mom_educn lessthan_HighSchool <.0001 mom_educn z_Bachelors&above0.... mom_age 35&above mom_age lessthan <.0001 mom_age z_20to payor medicaid payor other0....
Conclusions Infant mortality Higher risk ratio for black mothers relative to white mothers Higher risk ratio for mothers having education less than high school when compared to mothers having education more than bachelors Higher risk ratio for mothers of age less than 20 when compared to mothers of age between 20 and 34 Smoothed rates are higher in St.Lawrence, Erie, Schenectady, Oneida, Broome, Cortland etc Risk ratio for black mothers relative to white mothers are higher in the counties Orleans, Oswego, Chenango and Cortland
Limitation Missing data records out of records 246 deaths Inconsistent fetal data
Future study Detailed Phase 2 analysis including fetal deaths MHP and IH categories – Protocol for Phase 2 studies Cluster Analysis Spatial smoothing analysis
Reference Cai, J, Hoff GL, Dew PC et al. Perinatal periods of risk: analysis of fetalinfant mortality rates in Kansas City, Missouri. Matern Child Health J.2005;9: Cai J, Hoff GL, Archer R et al. Perinatal periods of risk analysis of infant mortality in Jackson County, Missouri. J Public Health Manage Pract. 2007;13:
Acknowledgments Dr. Glen D. Johnson, PhD, MS, MA Dr. Marilyn A. Kacica, M.D.,M.P.H
Questions???