[The accuracy of mammography screening for breast cancer: a Meta-analysis]

Objective: To estimate the accuracy of mammography for population-based breast cancer screening. Methods: Studies of screening mammography performance were systematically reviewed in the Medline, Embase, Cochrane and China National Knowledge Infrastructure (CNKI) literature databases. The retrieval was performed by using related key words, including disease name, screening and intervention indicators and study type. A total of 1 167 records were retrieved by June 4, 2015 and those studies met the inclusion criteria were included for the current analysis. The numbers of true-positive, false-positive, false-negative and true-negative were originally extracted and calculated from each individual study. A QUADAS checklist was used to assess the quality of the studies. A curve of summary receiver operating characteristic (SROC) was applied as a comprehensive assessment of diagnosed accuracy. The pooled effect size of both sensitivity and specificity was quantitatively synthesized in both total population and subgroup with dense breast using the bivariate mixed-effects models. We conducted a sensitivity analysis in the studies with sample size over 100 000. Heterogeneity between studies was measured by the Q test and I statistic, and publication bias was evaluated by a funnel plot and the linear regression test. Results: A total of 48 publications with 8 551 873 individuals were identified for the final qualitative synthesis, most of them were from Europe area and American (38 studies), 8 were from Asia area, and 2 were from Oceania area. The studies were conducted during period of 1975-2013, and the age of study women ranged mostly from 40 to 75 years. The area under the SROC curve (AUC) was 0.95 (95%CI: 0.93-0.97). It was estimated that the pooled sensitivity and specificity were 0.81 (95%CI: 0.77-0.84) and 0.96 (95%CI: 0.94-0.96), respectively. The results were indicated stable and robust in sensitivity analysis. In the subgroup with dense breast, the sensitivity and specificity were 0.74 (95% CI: 0.61-0.83) and 0.93 (95% CI: 0.89-0.96), respectively. Funnel plot and test results showed there was no significant publication bias among the included studies. Conclusion: The accuracy of mammography is high for population-based breast cancer screening in women, although the sensitivity is lower in women with dense breast.