Modelling the analysis of breast cancer screening programmes: sensitivity, lead time and predictive value in the Florence District Programme (1975-1986).

Statistical indicators of early efficacy were estimated in the Florence District Programme where breast cancer screening has been carried out since 1970. Analysis of screen-detected and interval cancer data by means of statistical modelling using GLIM allowed us to estimate the mean sojourn time (1.91 and 3.97 years in 40-49 and 50-69 year old women respectively), sensitivity (about 90% for those aged 50-69) and predictive value at the prevalence screening test (about 100%). Results were compared with estimates of the same parameters in the Health Insurance Plan and Swedish Two County studies.