Modeling epidemiologic studies of occupational cohorts for the quantitative assessment of carcinogenic hazards.
Stayner-L; Smith-R; Bailer-AJ; Luebeck-EG; Moolgavkar-SH
Am J Ind Med 1995 Feb; 27(2):155-170
Presently available methods for modeling epidemiological data for quantitative risk assessment were reviewed. Methods for use with retrospective cohort mortality studies of occupational groups for cancer risk estimations were highlighted. Analysis of a NIOSH study of work related lung cancer induced by cadmium (7440439) exposure was used to illustrate these methods. Statistical models based on standardized mortality ratios and internal comparisons, and biological models, including the multistage and two stage models of carcinogenesis, were considered. Risk prediction models were also discussed. Risk estimates for chronic exposure to variable levels of cadmium were calculated on the basis of selected models fitted to the NIOSH cadmium cohort mortality study. The authors conclude that the selection of a specific model may have a substantial effect on the resulting risk estimates. Although some models may be ruled out as inappropriate for risk assessment, there is probably no model that is absolutely correct. The best method may be to fit several models to the data to produce a range of risk estimates from the models which best fit the data.
NIOSH-Author; Mathematical-models; Statistical-analysis; Epidemiology; Occupational-exposure; Cancer-rates; Analytical-models; Risk-analysis; Lung-cancer; Mortality-data
American Journal of Industrial Medicine