Adverse birth outcomes contribute significantly to perinatal morbidity and mortality, yet few established risk factors exist. One important risk marker for all adverse birth events is black race. Most disparities research comes from two types of studies: geographically-defined population studies (using vital records) and clinic- or hospital-based studies. The women who self-select into clinical studies do not represent the population of reproductive women, but if these differences are associated with race and the degree to which the conclusions drawn from study samples are generalizable to population health has been inadequately studied. Risk factors besides race, like smoking, late initiation of prenatal care and inappropriate weight gain have been implicated in birth outcome disparities. The agreement between women's behavioral reporting in clinical research and reporting for birth records, and if reporting differs by race, has also not been systematically assessed. Racial disparity in adverse birth outcomes cannot be solely explained by individual-level factors, which has encouraged the examination of neighborhood-level influences, and the interaction of individual- and area-level effects. Modest but consistent effects of neighborhood-level factors on pregnancy outcomes, have been observed but this research has been limited to area deprivation. Non-socioeconomic area-level factors plausibly affect maternal and child health outcomes, including safety, physical activity facilitators, and racial residential segregation, which have been sparsely researched. To address these deficiencies, the study will match address data of women enrolled in the Pregnancy, Infection and Nutrition (PIN) study (n = 2006) to their birth records, year 2000 U.S. Census data and the directly observed neighborhood data collected on a subset of PIN women (n = 1500) in a five county NC study area - Alamance, Chatham, Durham, Orange and Wake. Directly observed neighborhood data were collected in 2001 and 2005 and most of the geocoded birth records (n = 120,000) for the five county area for the period corresponding to PIN deliveries (1999-2005) have been obtained. The research will improve our understanding of the racial disparity in birth outcome determinants by accomplishing three specific aims: Aim 1) evaluate the extent of selection bias for women who enroll in the longitudinal cohort studies, like the PIN study, and use sensitivity analysis to create selection probabilities. The selection probabilities generated from the population cohort will be compared with the PIN selection probabilities; a bias correction factor will be generated for use in future studies. Aim 2) evaluate the reliability of women's self-reported behavior (gestational weight gain, smoking, and alcohol intake data) in the clinical study and birth record setting using concordance values and kappa statistics to account for chance agreement. Aim 3) assess, using fixed slope random intercept multilevel logistic and linear modeling, neighborhood influences on health behaviors and birth outcomes using the PIN directly observed neighborhood and year 2000 census data to characterize maternal neighborhood. The area-level data will be spatially analyzed to identify spatial dependence among PTB and exposure variables of interest. We predict neighborhoods will differ by race, race will be associated with differential health behavior reporting, neighborhood differences will be associated with health behavior and birth outcomes, and that greater clustering will be observed among the clinic sample than among the population sample. The persistence of a pronounced racial disparity in adverse birth outcomes demands a critical look at current approaches to research and practice and the application of new methods for thinking about 'old' exposures. This research will contribute to understanding the factors underlying racial disparity in birth outcomes and ultimately improve population health.
Listed is descending order by year published.
Vinikoor LC, Messer LC, Laraia BA, Kaufman JS. Reliability of variables on the North Carolina birth certificate: a comparison with directly queried values from a cohort study. Paediatr Perinat Epidemiol. 2010 Jan;24(1):102-12.
Health Disparities, Neighborhood, Low Birthweight, Preterm