Reported Occurrence of Severe Epilation vs. Location among the Life Span Study Cohort of Atomic Bomb Survivors

There has long been a concern that clinical symptoms of acute radiation injury, reported by survivors in the questionnaires used by the Atomic Bomb Casualty Commission (ABCC) in the 1950s to gather information for creation of the Life Span Study cohort, were reported at low but non-zero proportions in relatively distal survivors. Because the data were self-reported, there were various potential sources of error that could have resulted in false-positive data. In the most definitive RERF studies, severe epilation (of scalp hair) has been considered to be the most specific and reliable of the reported symptoms potentially related to acute radiation injury. The data on severe epilation: 1,288 cases in Hiroshima and 384 in Nagasaki, have been subjected to a new, exploratory spatial analysis, to see if there are any patterns that might plausibly be associated with exposure to residual radiation. A method used by RERF investigators in 1983 to check for circular asymmetry about the hypocenters, by comparing compass octants in various directions, was repeated with DS02 dose estimates, and a parametric model for dose response including octant indicators was also fitted. In addition, the cities were divided into 200-yard-square geospatial cells and spatial plots were made of the proportions in the cells to allow visualization of any patterns that might exist. A result was that all of the cells with more than a few cases of severe epilation were within about 1.6 km of the hypocenters, consistent with the epilation expected for the corresponding DS02 doses. Among the more distal cells were a number of cells with one or two cases per cell, which are increasingly sparse at longer distances and otherwise appear to be randomly distributed in location — no pattern potentially associated with fallout is obvious on visual inspection.
In the future these data may be further analyzed with spatial methods appropriate for binomial random variables with possible spatial autocorrelation, which must also correct for expected proportions due to DS02 dose.