Sensitivity of Medicare claims to identify recurrence.

Abstract

6522 Background: Cancer registries do not collect recurrence data. As a result, researchers have used health claims to impute recurrence for cancer patients by identifying cancer therapy or metastatic disease after initial treatment. The validity of this approach has not been established. We used the linked SEER- Medicare data to assess the sensitivity of Medicare claims for cancer recurrence. Methods: Although the SEER-Medicare data do not explicitly capture recurrence, we applied criteria to identify patients likely to have had a recurrence, defined as patients diagnosed with Stage II/III colorectal (CRC) and female breast cancers from 1994-2003 who received initial cancer surgery, had at least a 90-day break in treatment following initial treatment, who then died from cancer in 1994-2008. For these patients, we reviewed all claims from the end of the treatment break to death for indicators of recurrence. Indicators included additional therapy (cancer surgery, chemotherapy, RT), diagnosis codes for metastasis, and end-of-life care. We used multivariate logistic regression analysis to evaluate patient factors associated with receipt of additional cancer therapy. Results: The first indicator of recurrence for CRC patients (n=6,910) and breast patients (n=3,826) was additional therapy (38.8% and 35.2%), metastasis diagnosis codes (36.0% and 45.7 %), or end-of-life care (16.7% and 12.8 %). Eight percent of patients had no indicator of recurrence before death; 40% of patients had no additional cancer therapy before death. Patients age 70 and older were less likely to have additional therapy (p<0.05), in adjusted analyses. Conclusions: Identifying recurrence through additional cancer therapy will miss about 40 percent of patients with recurrences; particularly older patients. Over 20% of patients with recurrence were first identified by hospice admission, late in the course of their illness, or had no indicator of recurrence in the claims. We conclude that Medicare data have limited potential to identify cancer recurrence. These findings may not apply to younger cancer patients, who are more likely to undergo additional cancer treatment for recurrence.