Statistical methods for the evaluation of health effects of prenatal mercury exposure

Environmental risk assessment based on epidemiological data puts stringent demands on the statistical procedures. First, convincing evidence has to be established that there is a risk at all. In practice this endeavor requires prudent use of the observational epidemiological information with delicate balancing between utilizing the information optimally but not over-interpreting it. If a case for an environmental risk has been made, the second challenge is to provide useful input that regulatory authorities can use to set standards. This article surveys some of these issues in the concrete case of neurobehavioral effects in Faroese children prenatally exposed to methylmercury. A selection of modem, appropriate methods has been applied in the analysis of this material that may be considered typical of environmental epidemiology today. In particular we emphasize the potential of structural equation models for improving standard multiple regression analysis of complex environmental epidemiology data. Copyright (C) 2003 John Wiley Sons, Ltd.