Effects of random dosimetry errors and the use of data on acute symptoms for dosimetry evaluation

Two approaches are used to address questions regarding dose measurement errors. The first is to describe and compare the effects of random error for several dose treatments including the use of grouped and ungrouped data, and analyses with and without truncation at 600 rad. It is found that the ways in which measurement error is most likely to mislead are through downward bias in the estimated regression coefficients and through distortion of the shape of the dose response curve. The second approach makes use of data on acute symptons to identify survivors in particular shielding situations or locations whose dose estimates may be especially biased or subject to unusual amounts of random error. It is found that the dose-response curves for acute symptoms differ considerably by many of the factors studied, but it is not possible to separate differences resulting from varying degrees of random error from systematic bias. The analyses also suggest that doses of Hiroshima survivors are in general better estimated than doses of Nagasaki survivors, a situation which could easily bias city comparisons. 17 references