populations exposed to other radiations, such as from the atomic-bomb survivors in Hiroshima and Nagasaki who were exposed to primarily gamma rays and neutrons, to the doses estimated in the lung cells which are believed to become lung-cancers.
The biologically-motivated approach would use data from molecular, cellular, and animal studies to generate parameters which could be used to refine models of carcinogenesis. The miner data could be used in conjunction with the models to see if effects observed in nonhuman systems might be altered by responses in the intact human.
Finally, the empirical approach would use the state-of-the-art statistical methodologies to analyze epidemiologic data from miner and residential studies. In brief, of the three data sources (miner studies, domestic radon exposure studies, and populations exposed to γ rays), use of data from γ-ray exposed groups (particularly A-bomb survivors) was rejected due to the many assumptions needed to generate risk estimates for prolonged localized exposure to densely ionizing radiation, based on risks for acute whole-body exposures to γ rays. Use of data from studies of residential radon exposure was rejected for the primary risk estimation due to the very limited statistical power available in these studies. However, the residential data were used as validation and support for the low-dose risk estimates, in that the results were compared with, and shown to be consistent with, risk estimates extrapolated from analyses of the miner data.
Of the three modeling approaches possible for analyzing the epidemiologic data (dosimetric approach, biologically-motivated models, empirical approach), the dosimetric approach was rejected for the same reasons as was the use of A-bomb survivor data—the uncertainties involved in extrapolating from the effects of acute whole-body γ-ray exposures to prolonged localized exposure to densely ionizing radiation. The use of biologically-motivated models—such as the two-stage clonal expansion model—was rejected due to our limited knowledge of the complete mechanisms involved in radon-induced carcinogenesis. As described in chapters 2 and 3, however, mechanistic information was used to guide specific parts of the empirical modeling, whenever possible.
In summary, the committee chose to follow the overall methodology both of BEIR IV and the pooled analysis of Lubin and others (1994a), and to base risk estimates on an empirically-based analysis of miner data. This approach provided the committee with the invaluable starting point of existing data bases and methodologies, some well-characterized and others with deficiencies and limitations as described in chapter 3.
The principal place of exposure to radon is the home; the predominant contribution of exposure in the home reflects the amount of time spent there and the general pattern of radon concentrations in buildings (Harley 1991; NCRP 1991;