Evaluate algorithms for identifying treatment patterns for patients with metastatic colorectal cancer using administrative claims data.

299 Background: Administrative claims data are widely used to identify treatment patterns. However, as new drugs are brought to clinical practice, treatment regimens become more complicated for cancer patients receiving both chemotherapy and biologic agents. We evaluated the influence of using different treatment length algorithms on identification of treatment patterns for patients with metastatic colorectal cancer (mCRC). Methods: We identified patients aged 18 or older with at least one primary CRC diagnosis between 01/01/2008 and 03/31/2012 in the MarketScan databases. mCRC patients were included if claims indicated the presence of distant metastases (ICD-9-CM codes: 196.x, 197.0 to 197.4, 197.6 to 197.8, 198.x, and 199.0). All treatment claims were extracted and patients were followed for more than 6 months, until end of enrollment or study end date (03/01/2013). We identified all chemotherapy and/or biologic drugs given to a patient during 30, 60, or 90 days after initiation of treatment and compared differences in type and number of treatment lines (TLs). Results: A total of 2,046 mCRC patients received at least one TL. Using a 30-day cutoff period, 30% patients received 1 TL; 30% received 2 TLs; and 40% received ≥3 TLs. Results changed after using a 60-day cutoff period: 39% patients with 1 TL, 33% with 2 TLs, and 28% with ≥3 TLs. If using 90 days as a cutoff period, 44% received 1 TL; 35% received 2 TLs; and 22% received ≥3 TLs. For all cutoff periods, the most common first TL prescribed were FOLFOX +/- bevacizumab and 5-FU/LV. For patients receiving a second TL, FOLFOX was most frequently prescribed if using a 30-day or 60-day cutoff, while FOLFIRI/bevacizumab had a higher prescription rate with a 90-day cutoff. Regimens varied in third and fourth-line therapy using different cutoff periods and FOLFIRI +/- bevacizumab was more frequently used than other regimens. Conclusions: The cutoff period did not significantly influence treatment regimens detected for first and second TL. Using a 30-day cutoff period resulted in 30% patients with one TL and 40% with three or more TLs, while with a 90-day cutoff, 44% patients received one TL and only 22% received three or more TLs.