Variations in the use of breast cancer screening, diagnostic and treatment services have long raised questions about the quality of breast cancer care and its impact on patient outcomes. Wide variations in mammographic interpretation and high rates of breast biopsies have raised particular concerns about the quality of screening mammography and how quality varies by different mammography providers and different population groups. ? ? In response to these concerns, the Institute of Medicine (IOM) has recommended that the use of screening mammography in community practice be carefully examined and that ways of monitoring performance should be developed. The IOM further recommends exploring the use of the Medicare claims database to describe trends in the use of screening mammography, examine geographic variations in screening incidence and study the outcomes of routine screening. Our research has shown that the Medicare data can be a valid source of information on screening, provided appropriate methods are used to define screening practices using information on the claims. We now propose to apply these methods in a population-based study of screening quality. ? ? The study will use SEER and Medicare claims data to examine screening performance measures among the SEER areas over the period 1998-1999. Our objectives are to: 1) investigate the relationship between screening performance (sensitivity and specificity of screening mammography) and the characteristics of radiologists and patients and 2) evaluate whether Medicare claims alone can provide valid measures of sensitivity and specificity compared to a linked SEER-Medicare database. ? ? The wide dissemination of screening mammography into community practice, particularly among older women, necessitates the ongoing assessment of its performance. Through this population-based study we will be able to compare screening performance by characteristics of patients and radiologists. Moreover, if we demonstrate that Medicare data alone provide valid measures of sensitivity and specificity, then our approach will allow for the generation of screening quality indicators (specificity) for virtually 100% of U.S. radiologists. ? ?
Showing the most recent 10 out of 14 publications
