Accuracy of mammography and clinical breast examination in the implementation of breast cancer screening programs in Colombia

Most evidence on breast cancer screening accuracy derives from high income countries. We evaluated screening accuracy and factors related to program implementation in Bogota, Colombia. Between 2008 and 2012 participants underwent clinical breast examination (CBE) and mammography. Positive results underwent histological verification. Adherence to screening protocols was analyzed. Sensitivity, specificity, and predictive values were estimated and adjusted by overdiagnosis. Impact of alternative screening algorithms on follow-up was explored, including combined screening tests and modified coding systems for mammography. In total, 7436 women aged 50-69 were enrolled; 400 discontinued and 1003 non-compliant with screening protocols. 23 cancer cases were diagnosed. Mammography sensitivity and specificity were 78.3% (95%CI 77.3-99.3) and 99.4% (95%CI 99.2-99.6). CBE sensitivity was 39.1% (95%CI 37.9-40.3) and specificity 83.4% (95%CI 82.6-84.3). Parallel mammography and CBE showed the highest sensitivity (95.6%) and combined as serial tests the lowest (positive CBE followed by mammography 13.0%). A simplified coding system for mammography (recall/no-recall) had 6.3% of positive results and a minor reduction in specificity compared with standard mammography, but reported the best balance between recall rates and screening protocol compliance. Call-backs had high rates of loss-to-follow-up; thus, alternative screening algorithms might help increase screening compliance and follow-up in low and middle income countries, particularly in populations with poor screening history and low access to health services.

Looking for trouble? Diagnostics expanding disease and producing patients

Novel tests give great opportunities for earlier and more precise diagnostics. At the same time, new tests expand disease, produce patients, and cause unnecessary harm in overdiagnosis and overtreatment. How can we evaluate diagnostics to obtain the benefits and avoid harm? One way is to pay close attention to the diagnostic process and its core concepts. Doing so reveals 3 errors that expand disease and increase overdiagnosis. The first error is to decouple diagnostics from harm, eg, by diagnosing insignificant conditions. The second error is to bypass proper validation of the relationship between test indicator and disease, eg, by introducing biomarkers for Alzheimer's disease before the tests are properly validated. The third error is to couple the name of disease to insignificant or indecisive indicators, eg, by lending the cancer name to preconditions, such as ductal carcinoma in situ. We need to avoid these errors to promote beneficial testing, bar harmful diagnostics, and evade unwarranted expansion of disease. Accordingly, we must stop identifying and testing for conditions that are only remotely associated with harm. We need more stringent verification of tests, and we must avoid naming indicators and indicative conditions after diseases. If not, we will end like ancient tragic heroes, succumbing because of our very best abilities.

Revised estimates of overdiagnosis from the Canadian National Breast Screening Study

We have re-estimated overdiagnosis of breast cancer from mammography screening by age group in the Canadian National Breast Screening Study (CNBSS), a randomized screening trial. In the CNBSS, participants were recruited in 15 centers. 89,835 women were randomized with informed consent, 50,430 age 40-49 and 39,405 age 50-59. Women aged 40-49 received annual mammography and physical examination (MA+PX) versus a single physical examination and usual care in the community (UC). Women aged 50-59 received (MA+PX) versus (PX-alone) annually. Individual randomization resulted in 44 almost identically distributed demographic and risk factors. Annual compliance over the five or four scheduled screens was 86-95%. The cumulative numbers of invasive and in situ breast cancers ascertained by year during screening and subsequent follow-up to 25 years post entry to the CNBSS in the mammography arm have been compared to those in the control arm. Estimates of overdiagnosis were derived using post-screening cessation cut-off points from 1 to 20years.⋯Overdiagnosis of invasive breast cancer at five years post cessation of screening for women aged 40-49 was estimated to be 32%, and 16% for women aged 50-59; 20years post cessation of screening 48% for women 40-49 and 5% for those 50-59. Including ductal carcinoma in situ with invasive cancer, estimates were 41%, 25%, 55% and 16%, respectively. We conclude that approximately 30% of invasive screen-detected breast cancers in women age 40-49 were overdiagnosed, and 20% of those screen-detected in women age 50-59. Including ductal carcinoma in situ, the estimates are 40% and 30%, respectively.