OBJECTIVES

: The accurate identification of lines of therapy (LOTs), which are used to determine the sequences of care as disease advances, is critical to real-world evidence research. However, variables specifying the LOT are not available in most administrative claims and electronic medical records databases. The purpose of this research was to automate the identification of LOTs in the setting of NSCLC.

METHODS

: The Flatiron Electronic Health Records (EHR) were used to identify an adult cohort of patients with advanced NSCLC. Anti-cancer drugs were extracted and classified into chemotherapy, targeted therapy, or biologic therapy. The LOT rules are as follows: The initial regimen was the drugs administered during first 28 days of treatment; Maintenance therapy did not advance a line (in NSCLC, patients may receive single agent pemetrexed or bevacizumab after initiation of treatment with these agents in combination therapy); The LOT was advanced when a new chemotherapy drug was used, initiation of a biologic or targeted drug with discontinuation of all prior drugs, or a specific period without any anti-cancer therapy. A SAS macro was built to automate these rules from real-world data.

RESULTS

: A total of 37,505 NSCLC patients with advanced NSCLC were identified with 821,509 drug administration records. The SAS macro was applied to these data and identified that 51.43%, 26.5%, 11.65% and 10.42% of patients have one, two, three and ≥four lines of therapy, respectively. Paclitaxel plus platinum, Pemetrexed plus platinum, and Bevacizumab plus Pemetrexed plus platinum were the most common 1^st line regimens. Nivolumab, Erlotinib, and Pemetrexed were the most common 2^nd line regimens. Regimens and LOTs identified by the macro were consistent with national treatment guidelines.

CONCLUSIONS

: The SAS macro provides researchers a tool to quickly identify LOTs from real-world data while retaining flexibility to tailor the code to specific research questions.