Accounting for errors in dose estimates used in studies of workers exposed to external radiation.
Health Phys 1997 Dec; 74(1):22-29
Approaches for addressing errors in dose estimates used in dose response analyses of epidemiologic studies of health effects of workers exposed to ionizing radiation were discussed. Sources of errors in radiation dosimetry were considered. External dose estimates used in analyses of occupational data have generally been obtained from personal dosimeters worn by workers. Although the estimates are generally considered reliable, they are subject to a number of errors including the usual variations that occur in laboratory measurements, an inability to respond accurately to all radiation energies to which workers are exposed, and the fact that modern dosimeters are usually designed to estimate deep doses, that is, energy absorbed at a tissue depth of 1 centimeter, rather than the organ doses needed for epidemiologic studies. Overall biases and uncertainties in estimates of doses to lung tissue and bone marrow received by workers at the Hanford research facility (Hanford workers) were summarized. General methods for addressing errors in radiation dose estimates were discussed. Systematic errors can be addressed by calculating bias factors defined as the ratio of the recorded dose to the true dose. The true dose is the measured dose to the lungs or bone marrow. Bias factors determined from true doses to the lungs can be used in studies of the risk for all cancers except leukemia. For leukemia, bias factors are to be derived from true dose determined for bone marrow. Uncertainties are expressed as 95% confidence intervals (CIs) which can be determined from appropriate probability distribution functions. Random errors can be addressed by calculating regression coefficients that are based on expected values of the 'true' dose rather than on estimated doses. The expected doses can be obtained from an appropriate distribution function. These approaches were illustrated by applying them to data obtained in a study of the risk of any cancer excluding leukemia, and leukemia excluding chronic lymphocytic leukemia (CLL) in the Hanford workers. For all cancers excluding leukemia, application of the techniques slightly reduced the risk estimates, from 0.23 to 0.20 per sievert. For leukemia excluding CLL, the absolute value of the estimated risk was reduced by about 40% and the upper limits of the CI by about 1.3 to 1.5.
NIOSH-Publication; NIOSH-Grant; Grants-other; Ionizing-radiation; Dose-response; Occupational-exposure; Malignant-neoplasms; Risk-analysis; Radiation-measurement; Statistical-analysis; Epidemiology
Health Risk Assessment Dept Battelle Memorial Inst PO Box 999 Msin P7 82 Richland, WA 99352
Other Occupational Concerns; Grants-other
Battelle Pacific Northwest Laboratories, Richland, Washington