Specifically, the committee reviewed the following NIOSH research:

1. Publications examining exposures at Hanford and ORNL in relation to cancer and non-cancer mortality—the committee also discusses the one single-site publication that used the LANL mortality data in a risk analysis;

2. The U.S.-based multisite studies that include Hanford and ORNL data (the Childhood Leukemia Case-Control Study and the Female Nuclear Workers Study);

3. The multisite case-control study of multiple myeloma and the ongoing multisite case-control study of leukemia;

4. Non-radiologic health studies focused primarily on beryllium-related issues; and

5. The contribution of DOE sites to international studies of cancer mortality among nuclear workers.

One of the most contentious issues in the fields of radiation protection and radiation epidemiology relating to cancer causation is that of the linearity of the dose response and the related question of dose-rate effects, i.e., whether the same dose delivered over an extended period has different effects than a dose delivered instantaneously. High-dose studies, particularly the A-bomb survivor study, show unequivocally that radiation can cause cancer, including most leukemias and most solid tumors. Interpolation of the high-dose studies down to the levels of exposure that are experienced in today’s occupational settings indicates that workers exposed near the maximum levels (2 rem/year) allowed by DOE will experience perhaps a ½ percent to 1 percent increase in the relative hazard of all tumors (for each year of such exposure). Current radiation protection standards and compensation programs are based upon the results of this extrapolation of high-dose effects down to low dose. The considerable uncertainty, however, in this high-dose extrapolation provides the fundamental rationale for conducting epidemiological studies of occupational radiation.

Direct studies of low-dose effects are desirable because there are scientific questions concerning the theoretical basis for extrapolation of human epidemiological data from acute radiation exposures in excess of about 200 mSv to lower doses delivered at the long-term chronic exposure rates experienced by most radiation workers. However, in direct studies of workers, the relatively low degree of excess risk poses enormous difficulties for epidemiological studies of current exposure levels, since even a perfect epidemiological study, where dose is known precisely and the chance for confounding is very limited, would require many decades of follow-up of hundreds of thousands of workers in order to accrue enough cases of cancer to have adequate statistical power to detect such relatively small increases. For example, to study the relationship between a spe-