Cancer Deaths among Radiation Workers 913 The numberof years (V) required to showa positive contribution of radiation exposure to cancer death rates for a specific population, risk level and dose rate may be found by scaling the appropriate n in table 5 using eqn (5) (which includes the smn value so that the effect of known variations in natural death rates on WV may be estimated). Oo-8 L 105 2x SMR N ~ RENEE“ Di* P= RDP, * 10-3 years. (5) 3.2. Lusk coefficients detectable in a given time Rearranging eqn (5) we sec that 2 SMR x LOT\4 R= (= 7] . (i DP, (9)6 Thus, using the values of m from table 5 and putting V years as the observation time, the lowest detectable risk coefficients may be found and some examples of these are shownin table 6. There is a one to one correspondence between the true number of cxcess cancer deaths and the radiation risk cocfiicient. However, since the observed excess cancers are the difference of two statistically Huctuating variables the corresponding risk coefficient can only be established to lie within a range of values. ‘lo give some indication of the magnitude involvedin trying to establish risk coefficients we have shown, in the four right hand columns of table 6, the 95% confidence interval of the risk cocfficicuts which corresponds to observed radiation cancer deaths equal to those predicted in table 3. We should cemphasize that colunins 2-5 in table 6 are independent of our calculated number of radiation-induced cancer deaths while columns 6-9 are based on our predicted values. 3.3. Other influences ‘The risk values detectable and the times required for a survey to yicld a positive identification of radiation risk as presented in this paper are subject to variations not covered by our statistical analysis. Systematic differences betaveen the control and the exposed group or from one year to another can be incorporatedin the sar(see section 2.2) but it is unlikely that these are known, ‘Tho effect is an additional spread on the natural cancer deaths and hence an increase in the required observation time. Another factor is the classification of a cancer death as in-service or ex-service (for example, if the cancer has’ infiuencedretirement). This suggests the desirability of not discriminating and supports the argument for full follow-up studies. ‘The effects of exposure to radiation for medical diagnostics or therapy must be omitted from the study because of the practical dificultics involved. Dose measurements of exposure to radiation for merical purposes are not made routinely and if they were they would be treated as sensitive confidential information. Jnven if a satisfactory assessment of doses were available, because