Considerations in using radiation dosimetry data in epidemiologic studies

From experience to date, it is concluded that cohorts can be divided into useable risk groups without need for extensive calculations of individual doses. Several possible risk group classification schemes are presented for use with standard methods of data analysis. The rationale, advantages, and disadvantages of each are discussed with respect to possible classification biases, and statistical and practical considerations. Data from a cohort of nuclear material workers are used to illustrate these dichotomous and ordinal schemes