Technical Report No. 17-87
The LD50 associated with exposure to the atomic bombing of Hiroshima and Nagasaki: A review and reassessment
Fujita S, Kato H, Schull WJEditor’s note: Publications based on this report were published in J Radiat Res (Tokyo) 30:359-81, 1989, and J Radiat Res (Tokyo) 32S:154-61, 1991.
Summary
Data accumulated by RERF-ABCC have been used to estimate the LD50/60 associated with the A-bombing of Hiroshima. A range of values emerge, varying slightly with the method of estimation used. This range, derived from DS86 marrow doses, and based on a linear fit to equally weighted estimates of the probabilities of death at various doses, is 2.3-2.6 Gy. A linear estimate in which the probabilities of death at the various doses are weighted by the inverse of their variances is somewhat lower, 2.2 Gy. These estimates include deaths in the first day, and the severely injured (burns, trauma) who survived the first day but succumbed later to their injuries. If inclusion of the latter groups biases downwards the estimate 17.5% or so, as one study suggests, and the range of the LD50/60 is adjusted upwards by this amount, it would be 2.7-3.1 Gy. Given the nature of the data, this range would appear to be the “best” estimate.
These values are discussed in terms of possible ascertainment biases, nonlinear models, competing causes of mortality, and estimates of this parameter from other surveys of survivors in Hiroshima. It is shown that the LD50/60 is relatively insensitive to the curve of mortality fitted to the observations, but the choice of model can influence substantially estimation of the LD 95/60. It is also noted that the slope of the mortality curve is shallower than animal experimental evidence leads one to expect; presumably, this reflects the greater genetic heterogeneity in the human species than in most experimental animals, the difficulties inherent in separating deaths attributable to radiation from those stemming from other causes, notably burns and trauma, and inaccuracies in the dose estimates themselves. Finally, it is shown that there is a surprising concordance in the various estimates from Hiroshima given the different groups of survivors involved and the methods used to estimate the LD50/60.
Data accumulated by RERF-ABCC have been used to estimate the LD50/60 associated with the A-bombing of Hiroshima. A range of values emerge, varying slightly with the method of estimation used. This range, derived from DS86 marrow doses, and based on a linear fit to equally weighted estimates of the probabilities of death at various doses, is 2.3-2.6 Gy. A linear estimate in which the probabilities of death at the various doses are weighted by the inverse of their variances is somewhat lower, 2.2 Gy. These estimates include deaths in the first day, and the severely injured (burns, trauma) who survived the first day but succumbed later to their injuries. If inclusion of the latter groups biases downwards the estimate 17.5% or so, as one study suggests, and the range of the LD50/60 is adjusted upwards by this amount, it would be 2.7-3.1 Gy. Given the nature of the data, this range would appear to be the “best” estimate.
These values are discussed in terms of possible ascertainment biases, nonlinear models, competing causes of mortality, and estimates of this parameter from other surveys of survivors in Hiroshima. It is shown that the LD50/60 is relatively insensitive to the curve of mortality fitted to the observations, but the choice of model can influence substantially estimation of the LD 95/60. It is also noted that the slope of the mortality curve is shallower than animal experimental evidence leads one to expect; presumably, this reflects the greater genetic heterogeneity in the human species than in most experimental animals, the difficulties inherent in separating deaths attributable to radiation from those stemming from other causes, notably burns and trauma, and inaccuracies in the dose estimates themselves. Finally, it is shown that there is a surprising concordance in the various estimates from Hiroshima given the different groups of survivors involved and the methods used to estimate the LD50/60.