An NLP tool to identify molecular diagnostic testing in veterans with stage IV NSCLC.

Authors

null

Julie Ann Lynch

VA Health Care System, Salt Lake City, UT

Julie Ann Lynch, Michael J. Kelley, Kyung Min Lee, Anna Hung, Yanhong Li, Bradley J Hintze, Trudy Pendergraft, Melissa Pavilack, Olga Efimova, Scott L DuVall, Shelby D. Reed

Organizations

VA Health Care System, Salt Lake City, UT, Duke University Medical Center, Durham, NC, Department of Veteran Affairs Salt Lake City Health Care System, Salt Lake City, UT, Duke Clinical Research Institute, Durham, NC, Department of Veteran Affairs Medical Center, Durham, Durham, NC, AstraZeneca, Wilmington, DE, Health Economics and Outcomes Research, AstraZeneca US, Gaithersburg, MD, VA Medical Center Salt Lake City, Salt Lake City, UT, Department of Veteran Affairs Medical Center, Durham, NC

Research Funding

Pharmaceutical/Biotech Company
Astra Zeneca, Department of Veterans Affairs.

Background: ASCO Quality Oncology Practice Initiative (QOPI) encourages hospitals to determine if patients (pts) with stage IV NSCLC adenocarcinoma have activating EGFR mutations. Guidelines recommend tumor testing to identify ALK, BRAF, EGFR, ROS1 and other gene alterations before treatment. Studies have shown EGFR testing underutilization. We developed and tested a tool to identify EGFR tests in Veterans Affairs (VA) electronic health record (EHR). We examined whether Veterans with newly diagnosed NSCLC-IV underwent EGFR testing. We measured EGFR testing trends and analyzed differences by patient, VA Medical Center (VAMC) and region. Methods: VA EHR data identified Veterans with NSCLC-IV diagnosed 2013-2017, who survived 45+ days and had 2+ visits with a VA oncologist within 120 days of diagnosis. All NSCLC histologies were included. Demographics and VAMC were obtained from VA Corporate Data Warehouse. EGFR testing results performed outside VA were from commercial laboratory data. We deployed a natural language processing (NLP) tool to identify EGFR tests in VA EHR clinical notes. Testing rates and characteristics associated with testing were examined by descriptive analysis. Results: Of 3484 pts, 623 (18%) had evidence of EGFR testing. There was a 244% rise in testing. 54 (9%) pts diagnosed in 2013 were tested vs 186 (25%) diagnosed in 2017 (χ2 82.3, p-value 0.00). No statistically significant differences by sex, race, or urban/rural address existed. Testing decreased as pts aged (χ2 27, p-value 0.00). 35% of pts age ≤49 and 12% of pts age ≥80 were tested. Testing varied widely by VAMC and region (VAMC χ2 795.6, p-value 0.00; region χ2 90.3, p-value 0.00). 28% of pts in the Pacific were tested vs 10% of pts in the Southeast. The main contributor to regional variation was VAMC differences. VAMCs that conducted EGFR testing within their own laboratories (64% and 53%), or were co-located with the national precision oncology program (NPOP, 59%) tested the most pts. NPOP was temporally associated with increased testing nationally. 9% of pts diagnosed in 2017 were tested using NPOP. Conclusions:EGFR testing underutilization in the VA persists. QOPI and the NLP tool for this initiative will help system-wide quality improvement initiatives.

Disclaimer

This material on this page is ©2024 American Society of Clinical Oncology, all rights reserved. Licensing available upon request. For more information, please contact licensing@asco.org

Abstract Details

Meeting

2019 ASCO Quality Care Symposium

Session Type

Poster Session

Session Title

Poster Session B: Patient Experience; Safety; Technology and Innovation in Quality of Care

Track

Patient Experience,Technology and Innovation in Quality of Care,Safety

Sub Track

Use of IT/Analytics to Improve Quality

Citation

J Clin Oncol 37, 2019 (suppl 27; abstr 318)

DOI

10.1200/JCO.2019.37.27_suppl.318

Abstract #

318

Poster Bd #

P1

Abstract Disclosures