Facility. Workers were initially categorized according to their facility of first employment---X-10, TEC, Y-12, or K-25. An individual who worked at more than one Oak Ridge facility was assigned to a fifth category ``MULT'', at the time of first entry into the second facility. In the dose-response analysis of white males ever employed at X-10 or Y-12 (X-10/Y-12), an individual could also enter follow-up as a multiple facility worker if his entry into X-10 or Y-12 was preceded by employment at K-25 or TEC. Entry dates were adjusted to the appropriate lag (two, ten, or twenty years).
Socioeconomic Status (SES). Each individual was categorized as either a nonmonthly paid worker or a monthly paid worker based on the variable Paycode in the demographic analysis file [2,6]. This variable is used as a surrogate for SES in the regression analyses with nonmonthly workers as the referent category so that estimated coefficients represent relative risks for monthly workers.
Length of Employment. Length of employment was calculated as the length of time between first hire and last termination at one of the Oak Ridge facilities. In all analyses, length of employment was categorized into two levels: level 1- one year or greater; or level 2- less than one year. The reference level used is one year or greater, so that estimated coefficients represent relative risks for ``short term'' workers relative to ``long term workers''.
Birth year. Workers were divided into five categories according to year of birth: before 1900, 1900-1909, 1910-1919, 1920-1929, and 1930 or after.
Age. Age at risk is divided into 15 groups (15-, 20-, ..., 80-, 85+) starting at 15-19 and continuing through 85+. Age was alternately treated as a factor (AG) and as a continuous variate A = (age -52.5)/100, where age is the interval mid-point.