Estimating lead time and sensitivity in a screening program without estimating the incidence in the screened group

Early indicators of the effectiveness of a screening test for chronic diseases such as breast cancer are the length of time the diagnosis is advanced by screening, the lead time, and the sensitivity of the screening test. This paper describes a model for simultaneously estimating the mean lead time and the sensitivity when only the number of cancers detected at the successive screenings and the number of cancers occurring in the time interval between the screening examinations are known. This model is particularly useful in assessing the effect of screening when the underlying cancer incidence in the screened group is unknown. The model is fitted to the data of 235 screen-detected breast cancer cases and 146 interval cancers diagnosed across 6 screening rounds of the program in Nijmegen. The maximum likelihood estimate for the mean lead time ranges from 1.3 years in the under age 50 group to 2.2 years in the age 50-65 group, both estimates having large confidence intervals. The corresponding sensitivity estimates are 0.92 and 1.00.