Mammographic over-screening: evaluation based on probabilistic linkage of records databases from the Breast Cancer Information System (SISMAMA)

Abstract: The Brazilian Ministry of Health recommends biennial mammographic screening for women aged between 50 and 69 years. Since screening is opportunistic in the country, the actual periodicity varies. This study sought to test a methodology for estimating over-screening due to excessive periodicity, defined as a smaller than recommended interval between exams, and its association with socio-demographic characteristics. A cohort of women who underwent mammography in 2010, and whose result was normal, was assembled through probabilistic linkage SISMAMA records based on a set of personal identifiers. We used data from women living in the micro health region of Juiz de Fora/Lima Duarte/Bom Jardim, Minas Gerais State, Brazil, who were followed in the System until the end of 2012. The rate of over-screening was 150/1,000 women/year (95%CI: 144.9-155.9), affecting 21% of women. Over-screening increased by 24% during Pink October campaigns (adjusted HR = 1.24; 95%CI: 1.15-1.35). The shorter the time passed since the last mammogram, the greater the odds of over-screening. Compared with women who had never had a mammogram prior to 2010, women who had had one in the previous 2 years were two times more likely to be over-screened (adjusted HR = 2.01; 95%CI: 1.74-2.31) whilst those who had had a mammogram ≤ 1 year previously were three times more likely to be over-screened (adjusted HR = 3.27; 95%CI: 2.87-3.73). Over-screening was substantial in this population, excessively exposing women to the risks of screening with no additional benefits and overestimating mammogram coverage. The methodology proved to be successful and should be applied to representative populations in order to guide breast cancer control policies.