What proportion of breast cancers are detected by mammography in the United States?

Background: None of the recent national studies has ascertained how women's breast cancer was detected in the United States. To rectify this gap, questions were added to the leading national health survey. Mammography was of special interest because it is widely used in the US and has evidence of a mortality benefit.
Methods: We used the 2003 National Health Interview Survey (NHIS), to ascertain the self-reported modalities used to detect breast cancer in the non-institutionalized US population. The study included 345 women 40-84 years of age who, in 2003, reported a personal history of breast cancer. We examined the frequency of self-reported characteristics and tested for association with mammography-detected breast cancer using logistic regression.
Results: Among the survivors in 2003, the percentage of breast cancers detected by mammography was much higher after 2001 (59%) than before 1993 (29%). Breast cancer survivors with less than high school education or less than US$ 20,000 household income were less likely to report detection by mammography.
Conclusions: Women reported mammography-detected breast cancer at a slightly higher rate than published studies of mammography sensitivity and use would suggest. Lower rates of mammography-detected breast cancers among survivors with low income or low education raise the question whether mammography is underutilized as a diagnostic tool, especially for underserved women.