In this issue of The Journal, Caldwell and colleagues (p 1575) report a possible increase in leukemia among participants in military maneuvers at a nuclear bomb test site. We congratulate the authors on their thoroughness. One interpretation of the excess of leukemia suggests a greater effect per rem at low-dose levels. The present situation is remarkably similar to that surrounding the association between fallout from Nevada weapons tests and a reported increased leukemia incidence in Utah, also published with an accompanying editorial.1,2 In the present article and the previous one,1 more remarkable than the reported positive association is the extent to which the authors discuss difficulties in reaching conclusions from their data.
Additional reasons for their reluctance to draw conclusions can be adduced. Since large-scale studies of this nature are usually carried out with the expectation of demonstrating something not previously known, one asks why the study? What