LAUSR

Cobalt is a commonly used element in metal industry. Exposure to workers occurs mainly by inhalation of cobalt-containing dust. For the evaluation of cobalt exposure, risk assessment and investigations on occupational diseases, measurements of cobalt in respirable dust are needed. Up to now, often only data for cobalt in inhalable dust are available, which is due to the earlier classification of the limit value in this fraction. Therefore, a possibility to convert cobalt concentrations mathematically from inhalable into respirable concentrations is desirable. In this study, 639 parallel measurements of cobalt concentrations in inhalable (cI(Co)) and respirable dust fractions (cR(Co)) were extracted from the non-public exposure database MEGA (Measurement data relating to workplace exposure to hazardous substances, maintained at the Institute for Occupational Safety and Health of the German Social Accident Insurance) and investigated by regression analysis. For the whole dataset regression shows high quality measures (correlation coefficient R = 0.888, adjusted coefficient of determination adj. R2 = 0.788 - R2 is adjusted to sample size). Further description of the data is achieved by splitting the dataset according to the type of sampling ('stationary' and 'personal') and three working activity groups, 'high temperature processing', 'filling/transport/storage', and 'machining/abrasive techniques' (0.845 ≤ R ≤ 0.876; 0.711 ≤ adj. R2 ≤ 0.762). As subgroups of 'high temperature processing' and 'machining/abrasive techniques' two further groups could be determined. These groups are called heuristic groups, since they have to be formed non-systematically by trial and error. These heuristic groups are 'welding' and 'grinding'. They are more selective on the included working activities with adj. R2 of 0.703 and 0.748 respectively. The resulting conversion functions of all groups are power functions with exponents between 0.704 and 0.794. For the estimation of cobalt in respirable dust in other studies, it is possible to use the conversion functions of the heuristic and working activity groups. Limitations of the possibility to use the conversion functions are discussed.