Economic conditions during pregnancy and preterm birth: A maternal fixed-effects analysis

Background Making causal inference regarding impacts of macrolevel economic conditions during pregnancy on pregnancy outcomes is hampered by the presence of unmeasured variables that may influence women's probability of giving birth under certain economic conditions (ie, exposure) as well as her pregnancy outcomes. Maternal fixed-effects (FE) analyses, in which the association between exposure and outcomes is estimated within mothers who had discordant outcomes, can control for such unmeasured variables when they are invariant across pregnancies. Methods We utilised a maternally linked data set of all singleton births in Michigan from 1990 to 2012 (n = 2 657 272 for full sample; n = 269 943 for FE analytic sample) to examine the relationship between state-level unemployment rates during pregnancy and preterm birth (PTB, <37 weeks' gestation). Measured maternal characteristics that change across pregnancies, for example, age, marital status, education, parity, and infant sex, were included as covariates in the model. Results Using an FE approach, we found that each one percentage point increase in state unemployment in the first trimester of pregnancy was associated with a modest 3% increase in odds of PTB. Our results were consistent with previously published results in a national sample and held across random- versus fixed-effect models, analytic samples, and outcome measures. Conclusions Our findings provide further evidence that economic downturn during early pregnancy may be associated with modest increases in PTB.