Accounting for pregnancy dependence in epidemiologic studies of reproductive outcomes.

The aim of this paper is to evaluate the contribution of hierarchical mixed models to the analysis of epidemiologic studies of environmental exposure and reproductive outcomes. We have re-analyzed, with a logistic-normal mixed model, four studies investigating the relation between the frequency of spontaneous abortions and paternal or maternal environmental exposures. The data include multiple pregnancies for some women. The fitted models allow for between-woman variation of the propensity for spontaneous abortion, by including a random intercept in the logistic model to adjust for within-woman correlations on pregnancy outcomes. We have discussed and implemented two estimation methods, maximum likelihood and Bayesian inference. We found similar values in the various epidemiologic studies of the between-woman variance of the intrinsic risk of spontaneous abortion. The size of this variance corresponds to a substantial variability in risk between women. Indeed, the risk of spontaneous abortion calculated for "nonexposed" pregnancies, that is, with mother's age, birth order, tobacco consumption, and maternal environmental exposure equal to the referent class, can vary, according to this model, from 2% to 17%.