BACKGROUND:

National self-report surveys show minimal racial disparity in mammography, whereas analyses of administrative data show large disparity.

METHODS:

Using the 1998-2002 Medicare Current Beneficiary Surveys, which contain participants' self-report and claims data, we developed multivariable adjusted models examining factors associated with self-reported mammography and self-reported mammography verified by billing records.

RESULTS:

No racial/ethnic disparities were found in self-reported mammography. Verified mammography, however, revealed significant disparities for race, education, income, insurance, and health status.

CONCLUSIONS:

Race, education, income, insurance, and health status are associated with a lower likelihood of self-reported mammography verified by the existence of claims data. These data caution against exclusive reliance on self-report survey data to assess disparity in mammography.