Abstract

The United Kingdom Atomic Energy Authority mortality study was designed to investigate the relation between exposure to ionising radiation and mortality among the authority's employees. The present paper describes some of the problems encountered in assessing occupational exposure to low dose radiation and examines whether the study's conclusions about the relation between exposure and mortality could be affected by the methods used. The study covered the years 1946 to 1979 during which time the frequency with which personal film dosimeters were issued changed from weekly to monthly, and the threshold level below which measurements were not made decreased 20-fold. Exposure from "below threshold" readings made an important contribution to total exposure in the early years. Estimates, based on the remeasurement of a sample of old films, indicated that the average whole body exposure before 1961 may have been about double that which was measured. Furthermore, although records were kept of when dosimeters were lost or damaged, the associated exposures were unknown and could only be estimated. Workers whose dosimeter readings were missing for more than 5% of the time during which they were monitored had higher all cause mortality (p = 0.04) and higher mortality from accidents and violence (p = 0.05) than other radiation workers. The results of analyses of mortality in relation to whole body exposure were compared when the exposures included estimates of the below threshold and missing exposures and when these exposures were assumed to be zero. Some of the findings differed, but none changed sufficiently to alter the general conclusions. Although the trend in mortality from all cancers changed from one in which the increase with exposure was far from statistically significant (p = 0.3) when the below threshold and missing values were assumed to be zero to one that approached significance (p = 0.06) after they were estimated, calculations of the annual excess deaths from cancer per unit dose resulted in broadly similar estimates. Studies of workers exposed to ionising radiation usually focus on mortality in relation to whole body exposure. In the present paper its relation to neutron and surface exposure is also examined. Workers with measured neutron exposures had significantly lower all cause mortality than other workers with a radiation record (p = 0.03). Surface exposure was significantly related to mortality from all cancers (p = 0.02) and prostatic cancer (p less than 0.001). Some data on cancer registration are presented but these cannot be readily interpreted because cancer registration details were available only for ex-employees who may not be typical of the workforce as a whole.