Skip to main content

Risk estimates of lung cancer from the follow-up of uranium miners


The French and the Czech data sets were continuously improved and updated. The French data set now comprises 1785 miners followed up to 1985, a larger data set with about 6000 miners is in preparation. This data set has low dose rates in average. The Czech cohort S has 4320 miners with 708 cases, followed up to 1990. The newer cohorts L and N have 72 cases in the data set. The data sets are also included in an analysis of pooled data of 11 studies.
Both the French and the Czech data set show an increased rate of lung cancer in the Uranium miners. In the Czech cohort S, the relative risk is about 5; it is increased particularly in the period 4-14 years after exposure (ICRF). Also strongly increased in this cohort, but presumably not linked to Radon exposure are homicide and mental disorder.
Obtained coefficients for excess relative risk per Sievert reach from 0.35% for the French data set (CEN-FAR) to 0.71% for the Czech cohort S (NIPH) and up to 1.36% (0.52 to 3.44 at 95% confidence level) (ICRF) for the Czech cohort S when considering only miners 55+ years with exposures 5-14 years prior to observation. (The analysis of pooled data from 11 studies gives 0.49%.) The variation of the risk coefficients can be explained partly by statistical fluctuations, partly they are due to different composition of the subgroups with different risk. It is clear that a single number is a too crude measure for risk: it is necessary to consider risk as function of various confounders.
The risk function favored by the Czech data weighs cumulative exposures in different years since exposure (groups of 5-14, 15-24, and 25-34 years are considered) and has a weak dependence on the age at median exposure. The risk function has a strong dependence on time since exposure, and shows an inverse dose-rate effect. A detailed analysis of the data gave, that the inverse dose-rate effect is especially pronounced for high dose rates over 4 WL, which cause fewer lung tumor cases than extended exposures with the same cumulative dose. In the French data set, no effect of the period of exposure, and thus no inverse dose rate effect was found. This may be due to poor statistics, or due to the relatively low dose rates in this cohort.
An alternative model for analysing miners data in the clonal expansion model in its various forms. A complete set of parameters for such a model was derived by fitting it to the Colorado plateau miners (NRPB), for which individual smoking information is available.

GSF participated up to 1994 in the analysis of the Czech data. Later it was decided to move to mechanistic modelling. It was found, that the two existing analyses of an animal experiment of F.T. Cross et. al. with rats exposed to radon gave very different biological parameters; the same is true for different analyses of the Colorado plateau miners data, and the British doctors data for smoking only. The models in use can give comparable quality of fit to existing data by widely different parameter values. It was concluded, that a better understanding of the models is necessary.
Radon and its daughter products constitute the most important source of natural radiation exposure of the population. There are various animal data on the induction of lung cancer by radon and its daughter products, but neither these data nor epidemiological studies of domestic radon exposure can as yet provide quantitative risk estimates. Such estimates need to be based on sutdies of uranium miners exposed to higher levels of radon. Data for major cohorts were recently reviewed in a document of ICRP (ICRP 50), and reevaluated by a special Committee of the US-National Academy of Sciences (BEIR IV). When the analysis of BEIR IV were performed, it was not yet possible to include in the computations the extensive data from the follow-up of the Czechoslovakian uranium miners; these data could only be accounted for in summary form from publications based on simple computations.The evaluation of the original data with the same mathematical methods as used by the BEIR IV Committee remained therefore an important task. This has been started by the group involved in project 1 in collaboration with the Institute of Epidemiology in prague. The results of a preliminary analysis, although in essential agreement with those of the BEIR IV Committee, show important differences. It appears therefore necessary to complete the analysis by a comparison between models currently used for the Japanese data to those used for data from uranium miners. As a further important step the analysis would have to be extended to the very large study that has been initiated on the several hundred thousand miners of the Wismuth AG in East Germany.
Accordingly our project has a two-fold aim. One objective is the comparison of models to be used for the analysis of such studies, their application to the cohorts of Czechoslovakian uranium miners (project 1 and 4), to the cohort of French miners (project 2), and to the miners of the Colorado plateau (project 3). This includes the examination of confounding factors such as time after exposure or duration of mining.A special effort will be done to quantify the influence of smoking on the basis of the data from cohorts L and N of Czech miners (project 4). Second aim of the project will be the joint analysis of two cohorts of the French and the S-cohort of Czechoslovakian uranium miners as a preliminary step of a later effort to analyse the cohort of the Wismuth AG.

Funding Scheme

CSC - Cost-sharing contracts


Ingolstaedter Landstrasse 1
85764 Oberschleissheim

Participants (3)

Commissariat à l'Energie Atomique (CEA)
Centre D'etudes De Fontenay-aux-roses
92265 Fontenay-aux-roses
Imperial Cancer Research Fund
United Kingdom
The Radcliffe Infirmary, Gibson Building
OX2 6HE Oxford
United Kingdom

OX11 0RQ Didcot,harwell,chilton