Table 2.

Percent distribution of patients across instrument variable levels and percent distribution by method of detection (mammography vs. other methods) within each variable level, along with unadjusted ORs for detection by mammography

IVTotalMammography (N = 3,718)Other methods (N = 3,785)OR (mammography vs. other methods)P95% CI
 LA1,622 (21.6%)48.6%51.4%Ref.
 WI974 (13.0%)57.9%42.1%1.46<0.0011.24–1.71
 NC964 (12.8%)45.9%54.1%0.900.180.76–1.05
 CA1,499 (20.0%)46.8%53.2%0.930.310.81–1.07
 KY438 (5.8%)47.5%52.5%0.960.690.78–1.18
 GA2,006 (26.7%)50.6%49.4%–1.24
 Lobular427 (5.7%)39.6%60.4%Ref.
 Ductal or ductal/lobular5,462 (72.8%)48.3%51.7%1.43<0.0011.17–1.74
 Other1,614 (21.5%)56.5%43.5%1.98<0.0011.60–2.47
  • aDerived from an underlying, POC-BP generated 10-level categorization of histology codes, in which lobular breast cancers were distinguished from a combined category consisting of both ductal and ductal/lobular tumors and from a composite “other” category consisting of mucinous, tubular, comedocarcinoma, inflammatory, medullary, papillary, and all other tumors.

  • bA county-level, continuous variable defined as the estimated annual mammographic-capacity of the county (number of machines X an assumed capability of 6,000 mammographic exams/year) divided by the number of women aged over 40 in the county. Shown here are the means of mammography-capacity by method of detection. The underlying source of county-level data on mammographic-capacity was the federally supported Mammography Program Reporting and Information System (MPRIS), which provided data for a collaborative effort by the FDA and the CDC, initiated in 2008, to document facilities conducting mammography (and their machine capacity) at the US county level. Among analyses using the MPRIS was one conducted by the U.S. Government Accountability Office (U.S. GAO Mammography: Current nationwide capacity is adequate, but access problems may exist in certain locations. Washington, D.C: U.S. GAO, July 2006), in which it was assumed that one machine and one radiologic technologist could perform three mammograms per hour; from that was derived the assumption of 6,000 mammograms per machine per year. The MPRIS data for our estimates included the 2003–2005 period. As noted, the mammographic-capacity variable could not be created for patients residing in Minnesota, where state legislation and accompanying regulations continue to “prohibit the release of count-level data to outside entities” ( The Minnesota state cancer registry determined that the CDC constituted such an “outside entity,” notwithstanding the state's commitment to participate in the CDC-funded POC-BP study.