DATA SOURCES AND PATTERNS OF CARE FOR BREAST CANCER

本申请是对10月1日提交的一份（CA/HS75545）的修订。

This application is a revision of one (CA/HS75545) submitted October 1, 1996 to the National Cancer Institute in response to their Program Announcement PA-95-069, Cancer Surveillance Using Health Claim-Based Data Systems. Researches have used administrative data sets, hospital discharge data, and/or cancer registry data to study geographic variation in patterns of care for breast cancer as well as to study the association of patient and hospital characteristics with patterns of care and mortality. Among their findings, they reported that rates of breast-conserving surgery varied across the country and were lower than expected. They also reported that some women (for example, older women and women who lived in rural areas) were less likely than others to receive standard treatments. Based on their findings, several authors suggested that the results of randomized clinical trials associated with breast cancer treatment and the dissemination of practice guidelines by national organizations such as the National Institutes of Health have not influenced breast cancer care in the United States. If their conclusions are not valid, however, any activities taken in response may be misguided. Resources would not only be wasted, but the results of these studies may cause unnecessary anxiety and confusion among women treated for breast cancer and among their providers. We have constructed two primary data sets comprised of data from medical records, patients, and physicians - one data set is from Massachusetts (N=1,497) and the other is from Minnesota (N-2,327). We propose to use these data to assess the accuracy of secondary data. The specific aims of this study are to 1) construct cohorts of patients with early-stage breast cancer using Medicare claims data, hospital discharge data, ad tumor registry data and compare the sample size, clinical characteristics, and demographic characteristics that result with those obtained from primary data; 2) determine the extent to which patterns of care reported from secondary data are consistent with patterns obtained from primary sources; 3) determine whether the degree of consistency between secondary data and primary data varies by the characteristics of patients or type of provider; and 4) develop an algorithm to adjust estimates derived from secondary data sets.