Predictors of Low Birth Weight And Geographic Access to Prenatal Care in Kansas, 1999 National Association for Public Health Statistics and Information Systems Annual Meeting June 7, 2006 San Diego, CA V. James Guillory, DO, MPH, FACPM Glynda Sharp, MHA Greg Crawford, BS
Practical Applications of Research on MCH at State Level Identifies specific areas and sources of concern where birth outcomes and maternal health can be improved Provides important information so that resources can best be allocated to achieve the greatest improvement Informs policymakers Drives policy formation.
Traditional Predictors of Infant Mortality in the United States Low birth weight (<2500 grams)* Preterm births (<37 weeks gestation) Prenatal care Early vs late initiation Adequacy *ACCESS to Prenatal care Maternal factors health/morbidities (anemia, DM, HTN) behavioral (tobacco, ETOH, drugs, nutrition) demographic (race, teen birth rate, education, marital status) * The US ranks last of 13 developed nations in LBW
What We Know: Barriers to Adequate Prenatal Care Mothers in rural areas: Lack available local prenatal and obstetrical care Less access to health insurance less prenatal care visits start prenatal care later Increased morbidity Greater distance and travel time to providers Transportation problems Child-care difficulties for larger families
Importance of LBW The rates of low birth weight are relatively unchanged over the past 35 years LBW is a major predictor of infant morbidity and mortality
National Objectives Related to Infant Health Healthy People 2010 Reduce infant morbidity and mortality Decrease disparities in health
Comparison of Kansas Birth Outcomes to the United States and HP2010 Objectives, 1999 % Kansas% U.S.% HP 2010 Target Low birth weight (<2500 grams) Very low birth weight (<1500 grams) Total Preterm (<37 weeks)** Live births weeks7.05*6.4 Live births <32 weeks * n = 38,194 (> 350 grams and > 20 weeks gestation) **Rate has risen steadily during the 1990's by about 11 percent Low birth weight (<2500 g) Very low birth weight (<1500g) Preterm (<37 weeks)** Live births weeks Live births <32 weeks % Kansas% U.S.% HP 2010 Target
How Kansas Compares to the Nation: LBW 2002 Kansas: 46 out of 51 (51 is Best) Best: Alaska, 5.8%; Worst: D.C., 11.6% Mississippi, 11.2%
Research Objectives Assess the quality of birth certificate data for use in multivariate statistical analysis Assess relationship between factors and LBW Maternal characteristics Prenatal care Insurance status Geographic location Determine the predictors of low birth weight using regression analysis
Methods Kansas Birth Certificate data linked to Medicaid Claims data, 1999 Data file prepared by the Office of Vital Statistics, Kansas Department of Health and Environment Obtained IRB approval
Inclusion criteria Single live births Birthweight > 350 grams Gestation > 20 weeks Exclusion Criteria Missing data for dependent or independent variables of concern Methods continued...
Methods (continued) Primary outcomes Proportion of infants with low birth weight Predictors of importance Geographic distribution Insurance status Other maternal characteristics
Geographic Distribution based on 2000 Census Bureau County Designations Based on commuting patterns and their resulting economic resource flow 3 Designations Metropolitan Micropolitan Non-metropolitan Ratcliff MR. Creating Metropolitan and Micropolitan Statistical Areas. Office of Management and Budget Alternative Approaches to Defining Metropolitan and Non-metropolitan Areas. Federal Register, 63:244:
Statistical Methodology Univariate analysis Chi square for tests of significance of categorical variables Student’s t-test for analysis of continuous variables Statistical significance at alpha < 0.05 Multiple regression analysis Logistic regression Stepwise Only those variables with a p-value < were entered into the model
RESULTS
Kansas Births, 1999* 38,748 birthsNumber % Single Births37, Twin Births 1, Triplet Births Race White34, Black/African American 2, Native American Asian/HI/PI/Other Gender Male20, Female18, * All births
Association Between Maternal Demographic Characteristics and Low Birth Weight: Single Births (n=37081) Infants with Low Birth Weight (< 2500 grams) Maternal Race n % White * African American Native American Asian/HI/PI/Other Maternal Marital Status Married * Not Married Maternal Education <HS Grad * HS Grad Any College Maternal Age <20 years old * 20+ years old *p<0.0001
Association Between Prenatal Care, Medicaid and Low Birth Weight: Single Births Infants with Low Birth Weight (< 2500 grams) Trimester PN Care Began n % 1 st Trimester * None or Later Maternal Medicaid Status No Medicaid * Moms with Medicaid *p<0.0001
Under weight mother Chi square = 15.2 p = < Obesity >20% Chi square = 16.9 p = < Smoker Chi square = p = < Alcohol Use Chi square = 33.4 p = < Association Between Maternal Behavior and LBW
% Low Birth Weight < 2500 grams Uterine BleedingYes11.72 Uterine BleedingNo5.43 Hydramnios/Oligo HydramniosYes18.9 Hydramnios/Oligo HydramniosNo5.36 EclampsiaYes50 EclampsiaNo5.47 Pre-clampsiaYes17.46 Pre-clampsiaNo5.05 Association between Maternal Medical Factors Associated with Pregnancy and Infant Low Birth Weight % Low Birth Weight ( < 2500 grams) YesNoYesNoYesNoYesNo Uterine Bleeding Chi 2 = p = <.0001 Hydramnios / Oligo Hydramnios Chi 2 = p = <.0001 Eclampsia Chi 2 = p = <.0001 Pre-clampsia Chi 2 = p = <.0001
% Low Birth Weight < 2500 grams Other STDYes8.07 Other STDNo5.44 HemoglobinopathyYes28.57 HemoglobinopathyNo5.48 Cardiac DiseaseYes12.1 Cardiac DiseaseNo5.47 Chronic HypertensionYes17.32 Chronic HypertensionNo5.41 Lung DiseaseYes8.26 Lung DiseaseNo5.46 Association between Maternal Medical Factors Not Associated with Pregnancy and Infant Low Birthweight % Low Birth Weight (< 2500 grams) Yes NoYesNoYesNoYesNoYesNo Other STD Chi 2 = p = Hemoglobinopathy Chi 2 = p = Cardiac Disease Chi 2 = p = Chronic Hypertension Chi 2 = p = <.0001 Lung Disease Chi 2 = p =
Medicaid Non-Medicaid p-value % Hepatitis, B/HBsAg Genital Herpes AIDS or HIV Other STD <.0001 Anemia (HCT <30) <.0001 Hemoglobinopathy Cardiac Disease Diabetes Hypertension, Chronic Lung Disease <.0001 Renal Disease Association Between Medicaid and Conditions Not Associated With Pregnancy
Medicaid Non-Medicaid % % p-value Underwieght <10% <.0001 Obesity >20% <.0001 Smoker <.0001 Alcohol Use <.0001 Behavioral Factors and Medicaid
Maternal Comorbidity and Comorbidity Associated with Pregnancy: Comparison Between Medicaid and Non-Medicaid MedicaidNon-Medicaid p-value Any Comorbidity <.0001 Uterine Bleeding <.0001 Incompetent Cervix Isoimmunization Hydramnios/Oligohydramnios Eclampsia Pre-Eclampsia
1999 Kansas Births By Geographic Location Metropolitan23, % Micropolitan 8, % Non-metropolitan 4, %
Multiple Regression Analysis Covariates were assessed to determine the independent contribution to LBW deliveries in Kansas while adjusting for all known variables that were very highly statistically associated (p< 0.001) with LBW (adjusted p-value because of multiplicity)
Adjusted Odds Ratio Estimates—LBW* Point Predictor Estimate 95% CI p-value PRETERM <.0001 NONWHITE <.0001 EDUCATION PRIORBABY UNMARRIED UT. BLEEDING HYDRAMNIOS <.0001 ECLAMPSIA PRE-ECLAMPSIA <.0001 PREV. SGA/PT <.0001 CHRONIC HTN <.0001 OBESITY >20% <.0001 SMOKING <.0001 MEDICAID *Prenatal care drops out of the model
Conclusions Rate of preterm and low birth weight infants in Kansas and disparities in these outcomes was similar to the nation Women who did not have private insurance were less healthy and had worse birth outcomes Women with Medicaid had a 20% increase in the likelihood of having a LBW infant when adjusting for other factors.
Conclusions Geographic Location Does Matter Women in micropolitan counties Started prenatal care later based on average month of starting care Had fewer prenatal visits Were less likely to have private insurance Had multiple indicators that increase potential for worse birth outcomes less access to health care Poorer health
Conclusions Insurance matters Late or no prenatal care greatest for women without private insurance Birth outcomes worse for infants born to mothers on Medicaid Need to separate public insurance from no insurance Results suggest that birth outcomes may be improved through increased focus on maternal health rather than prenatal care
Conclusions (continued ) “Poor access indicators most often suggest the need to make health care more affordable.” (Pathman DE, Ricketts III TC, Konrad Thomas. How Adults’ Access to Outpatient Physician Services Relates to the Local Supply of Primary Care Physicians in the Rural Southeast. Health Services Research. 2006;41:79-102)
Thanks To: The Kansas Department of Health and Environment for funding a portion of this study and for providing data. The Kansas City University of Medicine and Biosciences for their generous support. Sue Min Lai, PhD, KUMC John Keighley, PhD, KUMC Pradeep Chandra, MSIV, KCUMB