Expert judgment and occupational hygiene: application to aerosol speciation in the nickel primary production industry.
In many situations characterized by sparse data, occupational hygienists have used subjective judgments that are claimed to be derived from their experience and knowledge. While this practice is widespread, there has been no systematic study of 'expert judgment' or the 'art' of occupational hygiene. Indeed, there is a need to address the question of whether there is such a thing as 'expert opinion' in occupational hygiene that is broadly shared by practicing professionals. This research, employing 11 experts who estimate an exposure parameter (the percentages of four nickel species) in 12 workplaces in a nickel primary production industry, provides a large dataset from which useful inferences can be drawn about the quality of expert judgments and the variability among the experts. A well-designed questionnaire that provided succinct information about the processes and baseline data served to calibrate the experts. The Bayesian framework has been used in this work to develop posterior means and standard deviations of the percentages of the four nickel species in the 12 workplaces of interest in the company. These estimates of the nickel speciation are at least as precise as--and most of the time more precise than--those provided by the sparse measurement data. There was a very high degree of agreement among the experts. A majority of the experts agreed among themselves 92% of the time, while almost two-thirds agreed 73% of the time. This, coupled with the fact that the experts came from varied backgrounds, seems to suggest that there is indeed some broad body of specialized knowledge that the experts are drawing on to reach similar judgments. It also seems that one type of expert is not necessarily any better than any other kind, and expertise does not necessarily require intimate familiarity with the workplace. In this example, the expert judgment exercise has indeed enhanced the quality of our knowledge of the exposure 'fingerprints' for the nickel industry workplaces studied and the combination of expert judgment and sparse data is better than the sparse data alone. For occupational hygiene exposure assessment, our experience suggests that such expert judgment methods can provide a cost-effective means to improve and refine information about workplace hazards. However, more study is warranted for situations where the domain of the quantity of interest has a much wider range of values, e.g. actual exposure values.