Statistisches Verfahren zur Begründung von Schwellenwerten

Summary

In this paper methods for estimating and testing threshold values are described. An important field of application is occupational medicine, where for many chemical compounds and other agents which are health hazards but not carcinogenic, so-called threshold limit values (= TLV, in FRG: MAK-Werte = maximum concentration at the workplace) are specified to reduce the risk for the workforce. A threshold in this situation means there is no risk of the agent below a certain concentration. The risk is increasing only at higher concentrations. A method for estimating this threshold is presented within the framework of the generalized linear models widely used in the analysis of the relationship between some explanatory variables and a dependent dichotomous outcome. In most programs available for using these models the concept of a threshold is disregarded. The threshold is estimated by maximum-likelihood techniques and tested with the log-likelihood-ratio statistics R. Also an approach for calculating a confidence-interval is presented. The method is finally applied on the data of two epidemiological studies analyzing the relationship between silica dust and silicosis and dust exposure in coal mines and chronic bronchitis.