OBJECTIVES: Coding, billing and reimbursement complexities limit use of claims data in assessing real-world biomarker testing, both single gene test (SGT) and next-generation sequencing (NGS). Using an electronic health record (EHR) database linked to claims, and EHR as the gold standard, we assessed the impact of different reimbursement coding algorithms on identifying biomarker tests in claims over time.

METHODS: Biomarker tests for patients with aNSCLC were identified from the nationwide Flatiron Health EHR –derived de-identified database (results date between 2013-2021) linked to Komodo Health’s Healthcare Map of de-identified patient-level claims data. For every documented NGS and SGT test in Flatiron Health, a ∓ 30 day period from the results date was employed to identify any biomarker testing claim (matching claim). Tests were identified using a combination of Proprietary Laboratory Analyses (PLA) and Current Procedural Terminology (CPT) codes. A hierarchical algorithm to identify tests was developed in which highly specific codes superseded less specific codes.

RESULTS: There were 16,770 SGT and 2,873 NGS tests identified in EHR among 5,390 patients. The matching rates of tests in EHR to claims were 71% and 77% for NGS and SGT, respectively, using the broadest algorithm. NGS declined to 46% when using a highly specific algorithm. Stacked coding using a combination of at least two CPT codes for molecular tests was the most commonly used method to bill for NGS (31.6%), followed by PLA (17.3%) and NGS specific codes (15.3%). The matching rates increased overtime with more PLA and NGS specific codes being used to bill for NGS.

CONCLUSIONS: Identifying NGS tests in claims remains challenging due to variations in coding and reimbursement, despite the increased use of NGS specific codes. This work could shed light on future efforts that aim to develop an algorithm to identify NGS tests in claims data.