Cancer Deaths among Radiation |) orkers 903 2.2. Natural death rates Table 1 shows the Registrar General’s figures (for 1972) on which we have based our non-radiation-induced deaths. We have considered ‘all-cancers’ (ICD 140-239 inclusive) and leukaemias (ICD 204-207 inclusive) in our comparisons of natural incidence with radiation-induced incidence. A specific group of workers would have a standardized mortality ratio (snr) to account for sclectiveness of cmployment (c.g. a requirement to be medicallyfit) and this may be accounted for as described in sections 2.3 and 3.1. The sur is simply the ratio of the death rate (from a given cause) in a specific group to that in the whole population, : ‘able 1. The Annual Death Statistics used in this work for comparison purposes (Registrar General 1972) Age group °%, deaths (all causes) % deaths (all cancers) (ICD 140-239) % deaths (leukaemias) (ICD 204-207) 16-25 26-35 36-£5 46-55 56-65 66-75 76-85 86-95 0-092 0-096 0-227 0-733 2-08 5-42 12-39 25-57 0-0096 0-0172 0-0502 90-1899 0-G094 1-3624. 21569 2-4963 0-0023 0-002 1 0-0030 0-G049 QO-O121 0-0250 0-0453 0-069] 2.3. Phe visk of death from radiation ‘Phree factors determine the radiation-induced death rate: (i) the annual exposure, (it) the total visk per unit exposure, and (i) the latent period of the risk. [é is convenient to work with an exposure to cach worker of 1 rad/ycar and to apply a scaling factor te find the effects duc to other average exposurelevels (see below). The choice of data for factors (ii) and (iii) has required judgement based on an assessment of other studies of populations exposed to radiation. ‘These sbudies indicate that the rate of radiation-induced cancer death varices con- siderably with time after exposure. ‘The largest group of people studied over a arene ert r meet Cat long period are the 23979 Japanesc survivors with exposures above 10 rad who are included in the life-span study of the Atomic Bomb Casualty Commission (Jablon and ato 1971). In this group the cxcess leukaemia rate has decreased slowly with time since the mid 1950’s and it may be predicted that all the radiation-induecd leukaemias will have occurred by the mid 1970's (Gross 1974). This leads to a risk coefficient of 30 per 10° manrad for radiationinduced leukacmia death. [xcess mortality from all other cancers (excluding lenkacmia) follows a different time pattern. After aw very low rate during the 5-year period 1955-60, the mortality rose in the next two 5-year periods. J’rom