Effect of Dose Uncertainty on the Assessment of Radiation Risks of Solid Cancer Incidence in a Cohort of Russian Participants in the Elimination of the Consequences of the Accident at the Chernobyl NPP
V.K. Ivanov,,S.Yu. Chekin,M.A. Maksioutov,A.I. Gorski,S.V. Karpenko,K.A. Tumanov,V.V. Kashcheev,A.M. Korelo,E.V. Kochergina,,,,,,,,
DOI: https://doi.org/10.33266/1024-6177-2022-67-4-36-41
2022-08-01
Medical Radiology and radiation safety
Abstract:Purpose: Investigation of the influence of the possible uncertainty of exposure doses of Russian participants in the liquidation of the consequences of the Chernobyl accident on the assessment of radiation risks of the incidence of solid cancer in this cohort. Material and methods: Epidemiological and dosimetric data on a cohort of Russian participants in the liquidation of the consequences of the accident at the Chernobyl NPP, registered in the National Radiation and Epidemiological Register (NRER), are used as initial data for assessing radiation risks. The assessment of radiation risks is carried out by the statistical method of maximum likelihood in the framework of a linear non-threshold model of excess relative risk. Uncertainties in the liquidator’s exposure dose in the adopted risk assessment method are considered in the form of two error models. Dose estimates based on data from individual dosimeters are characterized by a classical model of measurement errors. In the case of estimates of unknown individual doses from group dosimetry data or group route doses, the Berkson assignment error model is used. Results: A method for assessing radiation risks has been developed, accounting for the uncertainty in dose estimates, based on the observed likelihood function. When taking into account the uncertainty of estimates of individual doses in the cohort of Russian liquidators, the estimate of the coefficient of the excess relative rate per dose unit (ERR/Gy) for the incidence of solid malignancies decreases by 7%, compared with the estimate obtained directly from the doses registered in the NRER database. The ERR/Gy estimate derived from the doses recorded in the NRER database was 0.69 with a 95% confidence interval (CI) of 0.37–1.04. The estimate of ERR/Gy, obtained accounting for the uncertainty in estimates of individual doses of liquidators, was 0.64 at 95% CI (0.33–0.98). This estimate bias is not significant, since it is within 95% CI for both ERR/Gy estimates, the statistical range of which is of the order of magnitude of the estimates themselves. Conclusions: Considering the uncertainty of individual dose estimates in the cohort of Russian liquidators, the estimate of the excess relative rate per dose unit (ERR/Gy) for the incidence of solid cancer does not statistically significantly differ from the estimate obtained directly from the doses registered in the NRER database. The bias in the estimate of the radiation risk coefficient observed, due to the dose uncertainty introduced into the calculation, is due to the statistical properties of the traditional radiation risk models used for radiation epidemiology. The results obtained confirm the high stability and validity of the radiation risk assessments obtained earlier from the doses registered in the NRER for the Russian cohort of Chernobyl liquidators. Further research will allow generalization of the developed method for assessing radiation risks, accounting for the uncertainty of dose estimates, based on the observed likelihood function, to other types of radiation epidemiological risk studies, including case-control and case-cohort studies.