Residential mobility during pregnancy: patterns and correlates

Abstract

Information on patterns and correlates of residential mobility can be important in studies of environmental factors and birth outcomes. The objective of this study was to describe residential mobility patterns and possible sociodemographic correlates of residential mobility among pregnant women. We obtained information on 656 mothers of infants with birth defects (cases) and 335 mothers of infants without birth defects (controls) from the geocoded dataset of the Birth Defects Risk Factor Surveillance Study, a case-control study conducted in Atlanta, Georgia, from 1993 through 1997. Using geographic information techniques, we measured distances mothers moved between residential addresses, and evaluated the proportion of moves and movement patterns by trimester. We used multivariate logistic regression to evaluate possible correlates of residential mobility for case and control mothers, including race, age, education, occupation, socioeconomic status, smoking, parity, and pregnancy planning. About 22% of women moved during pregnancy and most of them moved during the second trimester (11.9%), with no variation by case-control status. Among mothers who moved 51% moved within the same county. Pregnant women were more likely to move if they were younger (20-24 years, adjusted odds ratio (aOR) 3.39, 95% confidence interval (CI) 2.12-5.42; > or = 30 years: reference), did not plan their pregnancy (aOR 1.66, 95% CI 1.18-2.34), and smoked (aOR 1.46, 95% CI 1.01-2.12). For these associations with mother's residential mobility, there were no appreciable confounding or effect modification effects by case-control status. In studies of pregnancy outcomes and potential environmental exposures based on residence at the time of delivery, residential mobility during pregnancy may not vary by case-control status, but it still needs to be considered as a possible source of exposure misclassification. Accounting for potential case-control differences in correlates of residential mobility could be useful in minimizing potential non-differential misclassification.