Objective  The current work aims to investigate the effect of mammography-positive criteria on the cost-effectiveness of a breast cancer screening program, and provide a reference for the development of breast cancer screening programs in China.

  Methods  From July 2008 to September 2009, a cross-sectional breast cancer screening research project was developed in four cities: (a) Shenyang, located in northeast China; (b) Tianjin, situated in the coastal area of northern China; (c) Feicheng, an inland city in Shandong; and (d) Nanchang, located in southern China. A total of 21, 986 asymptomatic women, aged 45 to 69 years, volunteered to undergo breast mammography examination after signing an informed consent form. After a one-year follow-up, 65 new cases with breast cancer were diagnosed. The sensitivity and specificity of mammography were calculated on the basis of different criteria of positivity. A Markov model was used to simulate the entire screening process and calculate the cost-effectiveness ratios. The sensitivity analysis was performed using Tree Age pro 2011. The cost-effectiveness analysis was measured by the cost per life year saved (LYs), and the discounted rate was 3 %.

  Results  When the defined BIRADS 0 scale (used for further imaging scan or for comparison with the previous image data to draw conclusions) was classified as negative. A total of 55 new cases with breast cancer were detected. The sensitivity was 84.6 %, the specificity was 98.6 %, and the cost-effectiveness ratio was 45 632 ￥ / Lys. Moreover, when the defined BIRADS 0 was classified as positive, 56 new cases with breast cancer were identified, with a sensitivity of 86.2 %, a specificity of 93.9 %, and a cost-effectiveness ratio of 52 392 ￥ / LYs.

  Conclusion  This study suggests that different mammography-positive criteria have minimal effect on sensitivity, but a considerable effect on specificity and cost-effectiveness.