Agreement of Medicare claims and tumor registry data for assessment of cancer-related treatment

BACKGROUND. Although health claims data are increasingly used in evaluating variations in patterns of cancer care and outcomes, little is known about the comparability of these data with tumor registry information. OBJECTIVES. To evaluate the agreement between Medicare claims and tumor reg istry data in measuring patterns of diagnostic and therapeutic procedures f or older cancer patients. RESEARCH DESIGN. Analysis of a database linking Surveillance, Epidemiology and End Results (SEER) registry data and Medicare claims in patients aged g reater than or equal to 65 years with cancer. SUBJECTS. 361,255 Medicare patients with invasive breast, colorectal, endom etrial, lung, pancreatic, and prostate cancer diagnosed between 1984 and 19 93. MEASURES. Concordance of SEER files with corresponding Medicare claims. RESULTS. Medicare claims generally identified patients who underwent resect ion and radical surgery according to SEER (ie, concordance greater than or equal to 85%-90%) but less likely biopsy or local excision (ie, concordance less than or equal to 50%). In some instances, claims also categorized pat ients as having more invasive surgery than was listed in SEER and also prov ided incremental information about the use of surgical treatment after 4 mo nths. SEER files and, to a lesser degree, Medicare claims identified radiat ion therapy not included in the other data source, and Medicare files also captured a significant number of patients with codes for chemotherapy. CONCLUSIONS. Medicare files may be appropriate for studies of patterns of u se of surgical treatment, but not for diagnostic procedures. The potential benefit of Medicare claims in identifying delayed surgical intervention and chemotherapy deserves further study.