Assessing an Emerging Nationwide Population-based Mammography Screening Program in Taiwan Huay-Ben Pan1,2,3 Giu-CHenG Hsu4 Huei-LunG LianG1,2 CHen-Pin CHou1,2 yen-CHi WanG1,2 san-Kan Lee5 yi-HonG CHou2,6 Kam-Fai WonG7

Breast cancer is the most common cancer and ranks fourth among cancer deaths in women in Taiwan. In 2004, Taiwan’s Bureau of Health Promotion, Department of Health and Radiology Society of Republic of China collaborated to implement a program to monitor the quality and trained the people involved in a nationwide biennially mammography screening program for women between 50 and 69 years of age, which was started from a small population trial in 2002. In this study, we assessed the effects of this program by comparing evaluations of the women’s mammographic images obtained from Taiwan’s Bureau of Health Promotion in Taiwan with whether or not they were listed on Taiwan’s National Cancer Registry from 2004 to 2007, to which all pathology confirmed tumors are reported. A total of 311,193 consecutive mammograms were performed over the four-year period. Although prevalence of confirmed breast cancer was found to 0.63% in the mammograms evaluated the first year (2004), our later findings suggested that actual prevalence to be about 0.48%. By 2007, screening was improved as evidenced by an increased positive predictive value of an abnormal mammography (4.4%), a decreased recall rate for further work-ups (9.3%) and improved sensitivity (84.9%). Sensitivity was unusually high the first year of screening (86.1% in 2004) because of lack of mammography specialists training and the confounding of screening mammography results with diagnostic mammography results. Recall rate was high in the second year of screening (11.9% in 2005). We concluded that the mammography screening program was improved by our quality assurance and education program and would improvement would continue as these efforts continue and mammography-audit recommendations are adjusted based on the performance indicators we studied. Our results may help focus our future training programs and serve as valuable reference for other regions that are implementing population-based mammography screening programs. Correspondence Author to: Yen-Chi Wang Department of Radiology, Kaohsiung Veterans General Hospital, Kaohsiung, Taiwan No. 386, Da-Chung 1st Road, Kaohsiung 813, Taiwan J Radiol Sci 2011; 36: 1-7 Mammography screening in Taiwan 2 J Radiol Sci March 2011 Vol.36 No.1 increased by 14.4% from 918 in 1995 to 1,552 in 2007 [1]. It has been found in regions that do not have such programs in place, the mortality rate due to breast cancer is higher [2]. Therefore, these increases suggest efficient and accurate early screening is becoming more important. Mammography is the only tool that has been found to detect early asymptomatic breast cancer and has been found to increase survival rate by as much as 30% [3-6]. However, the accuracy of mammograms are affected by how they are used to make a diagnosis by the experience of the radiologists in recognizing and handling low positive rate of mammographic findings suggesting possible cancer. For example, in one study of a country’s screening program, it was found that the sensitivity and specificity of that program could be improved by providing special training to radiologists to improve their assessment of mammographic images. The proportion of cancers with definite preoperative diagnosis increased from 33.5% to 91.0% with structured radiological training from 2000 to 2005 [7]. Therefore, effective of screening not only depends on whether or not one receives a mammogram with good resolution but also on how it is accessed, i.e., quality of screening and assessment. Since 2002, all women in Taiwan between the ages of 50 and 69 years old can receive a free mammographic examination every two years. In 2004, to improve the effectiveness of this program, Taiwan implemented a program to ensure the quality of equipment and experience and training of the radiologists and mammographers participating in the program. With such a large nationwide program, it is important to assess the effects of such a program and judge its effectiveness and identify areas that may need improvement. To do this, using the database of the Bureau of Health Promotion in Taiwan, we compared the screenee’s mammographic reports with whether or not they were listed on Taiwan’s National Cancer Registry from 2004 to 2007, to which all pathology confirmed tumors are reported. The findings of such a study may provide valuable information for future policy decisions, and quality assurance and training programs. MATeRiAlS And MeThodS data collection The study was approved by the institutional review board at Veterans General Hospital Kaohsiung and was performed according to the Declaration of Helsinki principles. We collected data from the results of a free nationwide population-based mammography screening program funded and coordinated by Taiwan’s Bureau of Health Promotion. Our study period was from July 2004 to December 2007. It was in July 2004 that the Bureau of Health Promotion and the Radiology Society of Republic of China (RSROC) implemented a quality control program to a mammography screening program that was begun 2002. In that program, women 50 to 69 years old were self-referred or referred from nationwide community health centers to receive mammography screening every two years. The information obtained from the quality promotion program would be used as feedback for radiologic technologists and radiologists to improve their methods performing and interpreting of mammograms Strategy of screening mammography In this nationwide screening program, asymptomatic women 50 to 69 years old received mammographic examinations at 138 participating hospitals. These screening women had to be checked at least 24 months after any other mammogram examinations to fit the criteria of screening mammography. Each mammographic examination included mammograms taken from both a mediolateral oblique (MLO) view and a craniocaudal (CC) view. The assessments of the mammograms were performed by 293 board-certified radiologists with specific training in screening mammography. Mammography assessments were based on the Breast Imaging Reporting and Data Systems (BI-RADS) categorization system established by the American College of Radiology (ACR). In that system, category 0 represents need for further study, and 1 to 5, represents increasing risk of malignancy. Women with more than one lesion received only one BI-RADS assessment, the highest of the BI-RADS category was recorded. The examination reports included breast density, assessment category, and specific recommendations for follow-up. Women with BI-RADS category 0 were recommended or recalled to receive further diagnostic procedures, e.g., spot compression magnification or ultrasound examinations. definitions Negative mammography: a mammogram classified as BI-RADS category 1, 2, or 3. Positive mammography: a mammogram classified as BI-RADS category 0, 4 and 5. Recall rate: the percentage of those receiving BI-RADS category 0, 4 and 5, who were recalled to the hospital for further study. True-positive (TP) and false-positive (FN): defined by whether or not a participant with BI-RADS category 0, 4, of 5 was listed on Taiwan’s National Cancer Registry within 12 months of the mammography. True-negative (TP) and false-negative (FN): defined by whether or not a participant with BI-RADS category 1, 2, or 3 were not listed on the National Cancer Registry within the same period. Positive predictive value of recall (PPV1): the ratio of number of women who received BI-RADS category 0, 4, or 5 who were actually listed on the National Cancer Registry. Breast cancer prevalence: rate of women who received a first screening who were actually listed on the National Cancer Registry, regardless of their BI-RADS scores. Mammography screening in Taiwan 3 J Radiol Sci March 2011 Vol.36 No.1 Breast cancer incidence: rate of women who had a previous negative screening mammography who were actually listed on the National Cancer Registry, also regardless of their BI-RADS scores. Statistical Analysis The PPV1 and rates for false negative, recall, true positives, true positives plus false negatives, sensitivity, and specificity were calculated for the 2004 to 2006. Because the National Cancer Registry has not been published for 2007 as of the end of 2010, the 2007 values for false negative, breast cancer incidence, sensitivity of screening mammography, and specificity were estimated based on Baye’s rule and conditional probability. Cancer incidence for that year was estimated based on the previous year’s figures. Due to size of the population we study, we had a power of 99% at the 1% type I error rate for detect small differences, and thus there was little need to tests of statistical significance.