Cancer Deaths among Radiation Workers 5. 917 Conclusions The proportions of induced cancer deaths shown by tables 2 and 3 show clearly the need for follow-up studies of the causes of death of ex-radiation workers to supplement the records of actual workers. Moreover, we consider the proportions shown in these tables to be a lowerlimit as systematic factors “will tend to bias towards ex-workers deaths through a greater concentration of ill-health among those leaving. Since these factors are unquantifiable it seems wise to concentrate on the total (age-specitic) deaths in any analysis. We see also (particularly from table 5) that the exclusion of the over 65’s enhances the possibility of drawing conclusions. So we suggest that the analysis should concentrate on all radiation workers and cx-radiation workers between 16 and 65 althoughall included in the survey would be followed until death. This paper showsthat if a large survey (100000) on occupational exposure is made the first conclusions would not be expected for at least 20 years. However, if total exposures are muchless than 100000 manrad/yearorif the risk is less than 100 per 10° man rads—the time required to prove a positive effect of radiation on the incidence of deaths from cancer becomes very high and with little prospect of making statistically valid intermediate statements. Althoughthese prospects seem discouraging a survey has valuable contribu- tions to make. Vivstly, if the risk levels for low dose exposures are much higher than those anticipated, this will become evident at a much earlicr stage than suggested in table 5. For example, a factor of 3 increase in the risk reduces the time required by a factor of 9 (see eqn 4) so the effects of radiation would be detectable at the 5% significance level in under 10 years (and within 2 years at the 20% level). Althoughit is most unlikely that the actual risk is higher than the expectedrisk, the establishinent of a reliable base of data will provide the means to refute or ultimatcly to justify current estimations of levels of risk. secondly, a national survey may identify a rare form of cancer which can be radiation induced but which would be insignificant in data relating to small groups of radiation workers. While such cancers would account for a very small number of deaths, if they existed it would indicate environments where the working procedures should be reviewed. Analysis of any cancers which have low natural incidence would also provide an index against which thesignificance of the incidence of the cancerin particular industries may be assessed. Finally, any overall reduction in life expectancy for radiation workers may be investigated when sufficient data have beencollected. Resumis L'obsorvation ot Vanalyse des décés par tumours eancéreuses parmi los technicicns oxposés aux radiations Lapolitique énorgétiquo mondiale futuro dépend dans une cortaine mesuro do Voffot qu’a sur la mortulité cancéreuso lo dogré dirradiation nuquol les technicions sont exposés. Do Ik dépond aussi In décision priso sur los dépenses & prévoir pour réduiro les niveaux dirradiation éprouvées par leg techniciens. T.’oxposé discute certaines difficultés d’analyso do la situation ct il présente los résultats do calculs ostimant les mortalités par irradiation auxquolles on poub s’attendro pour chaquo groupe @ages particulier, do toutes les tumeurs cancérouses induites ainsi quo, séparémont,