Investigators have suggested that the effects of exposure to mercury vapor from dental amalgams be investigated by using data bases including various health outcomes measures and coronal carries DFS scores. In such data bases one is confronted with having to estimate amalgam exposure by an indirect method (for example, assume that all restorations on occlusal surfaces are amalgam). The effect of misclassification was investigated by using the NIDR Amalgam Study data base. In this study the coronal caries scores were obtained by the standard NIDR criteria used in epidemiological surveys. All restorative materials found in these study participants were also documented at the surface level. Thus, direct estimation of the total number of amalgam surfaces for each participant was possible as well. The effect of various indirect estimates of amalgam exposure were estimated. The result of assuming that all occlusal restorations involved amalgam material produced positively biased estimates of amalgam exposure which ranged from 30% to 70% of their true values. The magnitude of the bias was not consistent over the age range 40 - 70 years of age.
