Probability sampling pregnant women in an urban core: A case study from the National Children's Study

BACKGROUND: Obtaining a probability sample of pregnant women presents challenges, particularly in an urban core. We summarize the challenges encountered during a pilot study of the National Children's Study (NCS), the methods used, and how representative the sample was of city births. The NCS will follow 100,000 prenatally recruited children until age 21 to examine the effects of environment on health. Current data stem from a recruitment pilot addressing one inner city Primary Sampling Unit (PSU) from a multistage national probability sample. Eligible women reside in randomly selected segments of the PSU and are pregnant or trying to conceive. Challenges include multiple native languages, poverty, low literacy levels, and mistrust of research.

METHODS: After listing all households (HH) in 15 segments within the PSU, we recruited door-to-door in 2011. Recruitment was enhanced through community engagement and media messaging. Analyses used descriptive statistics to compare characteristics (age, education, etc.) of eligible women and women consented to 2009 birth certificates. RESULTS: We listed 13,850 dwellings, contacted approximately 95% of HHs and enumerated 28.5%. By January 2012, we identified 200 eligible women and consented 150 (64.7%). The recruited women were 45% African-American and 21% did not graduate from high school. Recruited women were older, more educated, less likely to be African-American and more likely to be Hispanic than city births. We discuss possible explanations for these variances including segment formation and differential consent rates by race/ethnicity.

CONCLUSIONS: Results emphasize the challenges of probability sampling urban pregnant women in an inner city.