[Epidemiologic research on the environment and health: some methodologic aspects].

Research in environmental epidemiology deals with physical, chemical and biological agents whose presence--or relative absence--within the different media coming into contract with human beings (air, water, soil, food, etc...) may be harmful to human health. Some "major" environmental risk factors are well known. In a number of situations, however, environment-disease associations are "weak". This does not rule out the possibility that the exposures involved have a significant impact on human health, considering their prevalence which is frequently high. However, this complicates their study owing to the potential importance of biases as well as that of sampling fluctuations. Although increasing study size is of crucial importance, it is not sufficient to establish a clearcut distinction between "weak" associations and "diluted" ones. To improve our knowledge of health risks which are associated with environmental exposures, the basic methodological principles of epidemiological research--to define and adequately measure exposures, health outcomes, confounders and effect modifiers--may be very valuable to approach the study of "weak" associations: 1) identifying and quantifying the presence of the agents of interest in the environment, studying the distribution of environmental exposures among individuals and its determinants, taking into account the whole history of personal exposures and integrating adequately the short term time variability of exposures, giving special attention to the type and intensity of exposures may help in the definition and measurement of exposures; 2) carefully analyzing the interactions which may exist between the physical, chemical and biological agents of interest and the human body may greatly help in the elaboration, measurement and validation of relevant health outcomes (exposures to the target organs, early lesions and health impairments); 3) this same approach may also greatly contribute to the identification of constitutional or acquired individual characteristics which may interact with environmental agents in the development of diseases. While there is no guarantee that such approaches will successfully discriminate between "weak" and "diluted" associations, it is likely that inconclusive epidemiological evidence will be very difficult to avoid if such approaches are neglected by environmental epidemiologists.