percent are from cancer (Campbell 1994). That appears unusually high, since national data show malignant neoplasms accounting for about 30 percent of male deaths in each 10-year age span covering deaths between ages 45 and 74. About 13 percent of years of potential life lost before age 65 is attributed to cancer deaths (NCHS 1992).
In 1976, following notice of a patient who associated his acute myelocytic leukemia with his presence at an atmospheric nuclear test, the Centers for Disease Control mounted an epidemiological study of military personnel who had attended that test—Shot SMOKY, a detonation of Operation PLUMBBOB—conducted at the Nevada test site in August 1957 (Caldwell et al. 1980, 1983). Findings of increased leukemias among participants generated concern that their health may have been adversely affected by participation in the atmospheric testing program. An extensive study of participants at five test series (chosen to represent a range of testing circumstances) was conducted by the Medical Follow-up Agency of the National Academy of Sciences (now within the Institute of Medicine) to pursue that hypothesis (Robinette et al. 1985). In 1989, the Defense Nuclear Agency informed MFUA that the data DNA had provided—and on which all MFUA analyses were based—incorrectly identified members of the participant cohort. DNA's initial estimate of the error was larger but, after detailed review, the congressional Office of Technology Assessment (OTA) estimated that approximately 15,000 names should have been but were not on the participant roster and another approximately 4,500 were wrongly included on the participant list (Gelband 1992). The total number of participants in that 1985 study was 49,148. MFUA (with support and concurrence from the OTA, the General Accounting Office, and congressional and Department of Defense staff) decided that the published study results (Robinette et al. 1985) should be withdrawn from discussion pending reexamination of the data and correction for possibly substantial errors in participant group identification. At the request of DNA, MFUA is redoing the Five Series Study with the more complete data. Results from the newer study are not expected before the end of 1997.
Other formal epidemiologic studies have not revealed distributions of rates that clearly confirm or refute radiation-caused mortality and long-term morbidity. Watanabe et al. (1995) compared military participants at Hardtack I, a 1958 U.S. test series in the Pacific, with a military comparison group. All-cause mortality (relative risk 1.10; 95% confidence interval 1.02–1.19) and digestive cancer mortality (RR 1.47; 1.06–2.04) mortality was higher among the Hardtack participants, but excess rates were not observed in deaths from all cancers, leukemia, or other hypothesized radiogenic cancers. The authors described the patterns of increased (and decreased) rates, but stopped short by neither concluding there were increased risks nor ruling them out.
Darby et al. (1993) studied mortality and cancer incidence in military participants of United Kingdom nuclear weapons tests and found no "detectable