We appreciate the interest of Walsh et al. in our recent papers (1 ,2) and their careful analyses of the data from the first of these. However, we feel that suggestions in the final paragraph of their letter may lead to some confusion. It is true that failing to deselect some of the highest dose range, e.g. over 2 Sv, leads to small estimated upward curvature due to the leveling off of the dose response in the highest part of the dose range. Our remarks about the effect of increasing follow-up were in fact based on such a deselection.

In particular, what we intended to convey was the following. If analyses are restricted to survivors with dose estimates under 2 Sv, the apparent upward curvature increases with increasing follow-up time as shown below based on the data for solid cancer mortality. Here we...

You do not currently have access to this content.