Receipt of Cancer Surgery: A Comparison of Cancer Registry Data and Medicare Claims

The Surveillance, Epidemiology, and End Results (SEER) Program collects data on cancer incidence and treatments given as the first course of therapy. Medicare claims are another source of cancer treatment-related information. We assessed the agreement between these data sources to identify cancer surgery.

We used the SEER-Medicare data to create a female breast and bladder cancer cohort that included patients diagnosed from 2004 to 2019 and continuously enrolled in Parts A/B fee-for-service from 4 months before at least 12 months following diagnosis, the window for surgery assessment. The presence/absence of Medicare surgery treatment claims was compared with SEER surgery data.

Of the 146,180 patients with breast cancer, 90.3% had surgery, and 6.4% did not have surgery per both SEER and Medicare. Surgery was identified in SEER only for 1.3% and in Medicare only for 1.6% (exact agreement=97%, κ=0.80). In the bladder cancer cohort (n=47,882), 90.9% had surgery and 1.8% did not, per both SEER and Medicare data. Surgery was recorded in SEER only for 2.2% and in Medicare only for 4.6% (exact agreement=93%, κ=0.31). For 0.1% of breast cancer patients and 0.3% of bladder cancer patients, surgery status was unknown per SEER, but Medicare surgery claims were present.

Overall, there was high agreement between SEER and Medicare claims in identifying breast and bladder cancer surgery, but there were some patients who received bladder cancer surgery per Medicare but not per SEER. It is possible that patients received surgeries in the community setting, which were not reported to SEER.