Purpose: To assess feasibility of implementing an automated method to identify patients who should have EGFR testing, and whether they have been tested, as a tool for quality improvement.
Background: Approximately 7% of veterans with metastatic, nsNSCLC have sensitizing mutation of EGFR, which predicts sensitivity to oral EGFR inhibitors. Prior studies have shown under testing for EGFR mutations in this population in VHA.
Methods: An endorsed quality measure (NQF and ASCO) for EGFR testing was utilized. Data to implement the measure were extracted from the cancer registry (ONC RAW), problem and encounter ICD codes, national oncology note template-generated health factors, laboratory test results, National Precision Oncology Program NGS testing, vital status, and pharmacy drug file to populate a SQL database. A dashboard in SharePoint allowed users to retrieve data based on national data access permissions. Descriptive statistics were used.
Results: The initial algorithm implementation was evaluated by comparison to manual review of patient records from one medical center. The second generation algorithm was then evaluated in the same manner at a second medical center (MC2). Among 117 cases identified during 2018, 68 (58%) were identified as having been tested and 49 (42%) not tested (31 living and 18 deceased patients). 48 of the non-tested samples were reviewed: 28 had not been tested, 14 had data documentation or coding problems (11 correctable by using the national note template), 1 correctable limitation of the national note template, and 5 limitations of the algorithm (all but 1 of which has been corrected). For stage 3 and stage VA-wide, there were 871 and 2832 cases, respectively, with documented testing rates of 26% and 36%, and a facility testing rate range of 0% to 100%.
Implications: The EGFR testing dashboard, in conjunction with appropriate structured documentation, has high accuracy of EGFR testing in patients with metastatic nsNSCLC. Current documented testing rates vary widely with a low system-wide rate, that can be improved through utilization of the dashboard.