Premature death among humans in the state of Missouri appears to have little relation to the composition of soils in the vicinity of their usual residency. Our conclusion contrasts with a considerable part of the literature, which implies that soil is a factor affecting human health. The difference may arise either because people living in a given environment may not expose themselves in such a way as to be affected by the soil or because other important factors may overshadow the effect of soils.

Other studies have suggested a correlation between the incidence of either human disease or human mortality and soil characteristics. Most correlations have been of the simple bivariate type. The opportunity for a multivariate study has arisen only with the availability of chemical data on soils and mortality rates of humans from a common area.

The distribution of elements in Missouri soils exhibits some regional pattern, but mortality distribution patterns are not regionally distinct. Therefore, stepwise multiple regression is used to define possible subtle relationships. This multivariate search tool identifies which of the 32 elements measured in 1,140 soil samples collected throughout Missouri best relate to human mortality risk. Differences in age, sex, and race are closely associated with mortality risk; adjustments for these factors have been made by direct methods. There is a possibility that other factors, presently not recognized, need to be measured and controlled before dismissing the idea of a soil influence on mortality. We are continuing the search for such factors.