Use of risk projection models to estimate mortality and incidence from radiation-induced breast cancer in screening programs

The authors report on a method to calculate radiological risks, applicable to breast screening programs and other controlled medical exposures to ionizing radiation. In particular, it has been applied to make a risk assessment in the Valencian Breast Cancer Early Detection Program (VBCEDP) in Spain. This method is based on a parametric approach, through Markov processes, of hazard functions for radio-induced breast cancer incidence and mortality, with mean glandular breast dose, attained age and age-at-exposure as covariates. Excess relative risk functions of breast cancer mortality have been obtained from two different case-control studies exposed to ionizing radiation, with different follow-up time: the Canadian Fluoroscopy Cohort Study (1950--1987) and the Life Span Study (1950--1985 and 1950--1990), whereas relative risk functions for incidence have been obtained from the Life Span Study (1958--1993), the Massachusetts tuberculosis cohorts (1926--1985 and 1970--1985), the New York post-partum mastitis patients (1930--1981) and the Swedish benign breast disease cohort (1958--1987). Relative risks from these cohorts have been transported to the target population undergoing screening in the Valencian Community, a region in Spain with about four and a half million inhabitants. The SCREENRISK software has been developed to estimate radiological detriments in breast screening. Some hypotheses corresponding to different screening conditions have been considered in order to estimate the total risk associated with a woman who takes part in all screening rounds. In the case of the VBCEDP, the total radio-induced risk probability for fatal breast cancer is in a range between [5 x 10(-6), 6 x 10(-4)] versus the natural rate of dying from breast cancer in the Valencian Community which is 9.2 x 10(-3). The results show that these indicators could be included in quality control tests and could be adequate for making comparisons between several screening programs.