Identification of potential bias in Louisiana Pregnancy Risk Assessment Monitoring System 2011 Data

The Pregnancy Risk Assessment Monitoring System (PRAMS) is a population-based surveillance system designed to monitor and identify maternal behaviors before and during their pregnancy and during the early period of the infant’s life. The PRAMS survey is administered on an annual basis to a stratified random sample of postpartum women in approximately 40 U.S. states including New York City. During 2011, eligible postpartum women were mistakenly excluded from the 2011 Louisiana PRAMS sampling frame. In order to assess selection bias in the Louisiana PRAMS 2011 data, we compared the characteristics of those individuals who were versus were not included in the 2011 Louisiana PRAMS eligible sampling frame. Louisiana PRAMS (n=59,941) and Vital Records birth certificate (n=62,240) data from 2011 were used. Chi-squared and t tests were used to assess for potential selection bias. We also examined non-response bias by performing chi-square and multivariate logistic regression analyses on a linked birth certificate data and Louisiana PRAMS sample records (n=2,193) file. The results from these analyses showed that selection bias was minimum and would not significantly impact the results produced from the Louisiana PRAMS 2011 data. Further analyses showed that only 56% of the mothers who were selected to participate completed the survey and those mothers who did not complete the survey were more likely to be African American (OR=1.3, 95% CI=1.1-1.7), have a high school diploma (OR=1.4, 95% CI=1.1-1.8), single (OR=1.3, 95% CI = 1.1-1.7), and did not receive prenatal care (OR=2.3, 95% CI=1.1-5.1).