Biases in observational studies of the effect of prenatal treatment for congenital toxoplasmosis.

The paper reviews methodological difficulties that arise when using observational studies to evaluate the effect of prenatal screening and treatment. The principle of each difficulty is described and then illustrated by a clinical example of toxoplasmosis in pregnancy and its consequences. Methods to deal with these difficulties are described. Given the limitations of existing observational studies and lack of randomised controlled trials, a systematic review of cohort studies offers the best approach for exploring potential biases.