In the most recent combined analyses of the data for workers in several countries exposed to low-level external radiation, Cardis et al. [36,25] present an excellent discussion of the importance of and difficulties encountered in studies of occupational effects of external radiation. The subgroup of workers from Oak Ridge that are in these combined studies are referred to as the ORNL subpopulation and correspond to X-10 only workers in this report. Their estimate of ERR per Sv for the ORNL subpopulation for all cancer excluding leukemia is 1.66 with 90% CI = (.04, 4.4) and for leukemia excluding CLL their estimate is -1.06 90% CI = (< 0, 4.8) ---see  Table V). These results are consistent with the results presented here for the larger X-10/Y-12 subcohort of Oak Ridge workers that were potentially exposed to external ionizing radiation. Cardis et al ---see  Table IV---give estimates of ERR per Sv for all cancer for their combined data of -0.02 90% CI: (-.34, .35); for leukemia the estimate is 1.55 with 90%CI: (-0.21, 4.7).
A recent high dose study by Pierce et al. presented risk estimates based on mortality in the Atomic Bomb Survivors through 1990 (see Table AI). Their excess relative risk estimates for all cancers [0.37 per Sv 90% CI= (0.31,0.44)] and lung cancer [0.42 per Sv 90% CI= (0.24,0.63)] are compatible with those derived here. Failure of this study to detect a significant dose-response for leukemia may be due to a lack of power. Another possibility is the presence of an unidentified toxic chemical that is only present at X-10 (e.g. some type of solvent) that is independent of, or negatively correlated with, external dose, and causes leukemia. This would also be a possible explanation of the increase in leukemia risk at X-10 relative to U.S rates and other Oak Ridge facilities.
The combined risk estimates derived from the present study are generally comparable to the positive dose response estimates in Pierce et al. for those categories where there were larger numbers of deaths. However, this study also suggests that there are differences in dose response between worker cohorts in the same location followed over about the same time period, and that these differences are associated with measurement issues. Specifically, dose response associations are stronger for X-10 workers, the population with more complete external radiation monitoring. Differences in dose response between X-10 and other workers are larger under longer latency assumptions that depend more on exposure classification based on radiation data from that historical period when monitoring differences between populations were greatest. This observation suggests that investigations of external radiation effects in combined facility studies should be sensitive to differences in other exposures and in radiation measurement characteristics between the cohorts.