Despite its proven efficacy in reducing breast cancer mortality, little is known about the effectiveness of screening mammography in community practice. Of particular concern is its specificity outside academic medical settings and the implications of false positives on patient outcomes and costs of care. These are major issues for older women, since Medicare now covers biennial screening mammograms. Monitoring trends in the use of screening mammography, assessing the accuracy of mammography in community practice and studying the outcomes of routine screening in terms of follow-up procedures have been targeted by the Institute of Medicine as important topics for effectiveness research in breast cancer that might be initially addressed with Medicare claims data. However, it is not clear whether the limitations of claims data will allow for the generation of valid information on these topics. The study goals are two-fold: 1) to evaluate the validity and utility of the Medicare claims for screening mammography research and 2) to investigate the effectiveness of mammographic screening for community practice in Texas over the years 1992-1993. The validation issues pertaining to coding of screening mammograms, the accuracy of a claims based algorithm to determine mammography test results and the precision of a claims based estimate of specificity will be addressed through chart reviews at 14 facilities providing mammography services in southeast Texas. Samples of claims from the centers will be drawn from the Medicare data base and compared to information from patient records. The accuracy of an algorithm to ascertain a newly diagnosed cancer will be assessed with data from the SEER-Medicare linked data bases for 1991-1993. Based on this evaluation we will characterize the nature and extent of error in using the claims for mammography research. The performance of screening mammography in Texas will be assessed with claims based estimates of specificity, the ratio of benign biopsies to cancers and the biopsy rate. Findings will then be compared to those reported in the clinical trials. These estimates and the costs of follow- up procedures will also be generated for individual facilities. Statistical analyses will measure the variation in estimates among facilities and their association with facility characteristics.
Showing the most recent 10 out of 14 publications
