An Automated Line-of-Therapy Algorithm for Adults With Metastatic Non–Small Cell Lung Cancer: Validation Study Using Blinded Manual Chart Review (Preprint)

Abstract

BACKGROUND Extraction of line of therapy (LOT) information from electronic health record (EHR) and claims data is essential for determining longitudinal changes in systemic anticancer therapy (SACT) in real-world clinical settings. OBJECTIVE The aim of this retrospective cohort analysis was to validate and refine our previously described open source LOT algorithm by comparing algorithm output with results obtained through blinded manual chart review. METHODS We used structured EHR data and clinical documents to identify 500 adult patients treated for metastatic non-small cell lung cancer with SACT from 2011 through mid-2018, assigning patients to training (n=350) and test cohorts (n=150), randomly divided proportional to the overall ratio of simple:complex cases (n=254:246). Simple cases were patients who received one LOT and no maintenance therapy; complex cases received more than one LOT and/or maintenance therapy. Algorithmic changes were performed using the training cohort data, after which the refined algorithm was evaluated against the test cohort. RESULTS For the simple cases, 16 instances of discordance between LOT algorithm and chart review pre-refinement were reduced to 8 instances post-refinement; in the test cohort there was no discordance between algorithm and chart review. For the complex cases, algorithm refinement reduced discordance from 68 to 62 instances, with 37 instances in the test cohort. Percentage agreement between LOT algorithm output and chart review for patients who received one LOT was 89% pre-refinement, 93% post-refinement, and 93% for the test cohort, while the likelihood of precise matching between algorithm output and chart review decreased with increasing number of unique regimens. Several areas of discordance that arose from differing definitions of LOTs and maintenance therapy could not be objectively resolved because of a lack of precise definitions in the medical literature. CONCLUSIONS Our findings identify common sources of discordance between an LOT algorithm and clinician documentation, providing for the possibility of targeted algorithm refinement.

Determining Line of Therapy from Real-World Data in Non-Small Cell Lung Cancer

“…There are several ways that our proposed rules better characterize LOTs in mNSCLC populations treated with targeted therapy compared to previously reported methods (Supplemental table 1). 2,[4][5][6][7] First, we distinguish between biologics and therapies targeting oncogenic drivers in mNSCLC. This distinction is important because biologics, such as angiogenesis inhibitors, may have little anti-tumor activity alone and are often added or removed from treatment regimens without a clear change in disease status, while targeted therapies have tremendous anti-tumor activity and are used in the setting of a change in disease status.…”

“…Third, we allow for a treatment pause and re-initiation within 60 days to be considered the same LOT, which may be more appropriate for targeted therapies typically taken daily by mouth that are paused and restarted for toxicity. [5][6][7] Retrospective determination of whether a treatment was given in the setting of metastatic disease versus early-stage disease is critical to determining real-world clinical endpoints. In this historical dataset, we used rules based on the approved therapies at the time of the dataset, but even while completing the analysis for this study the indications for several systemic therapies expanded to include early-stage disease, making the agent itself an unreliable indicator of early-stage treatment.…”