Veterans have a high risk of lung cancer and VA is currently implementing CT screening for this population, making it imperative to improve our ability to deliver high quality care to this very large group. After 3 annual screening CT scans, 39% of individuals in a recent lung cancer screening trial had a pulmonary nodule detected that required further evaluation. Our ability to provide Veteran-centric lung cancer screening and pulmonary nodule care is hindered by several gaps in our understanding of the pulmonary nodule evaluation process.
Aim 1: Develop and validate a predictive model of lung cancer risk among a cohort of Veterans with incidentally-detected pulmonary nodules. Rationale: A prognostic model would enable clinicians and patients to improve counseling and access to knowledge by providing personalized, Veteran-centric care outside of specialty settings. Aim 2: Evaluate healthcare system resource utilization amongst a cohort of Veterans with incidentally-detected pulmonary nodules. Rationale: Understanding the resources utilized by Veterans and their clinicians will enable increased safety, efficiency, and standardization which in turn will allow increased access to quality care. Aim 3: Develop and validate an algorithm based on routinely collected clinical data to identify Veterans with a pulmonary nodule diagnosis.
We will conduct a comprehensive review of patients in pulmonary nodule registries at the VA Portland Health Care System and VISN 23 to address these knowledge gaps. These registries are linked to other routinely-collected administrative data and are currently clinically used to track patients with nodules.
We have requested and received access to our study data from the VA Center for Medicare/Medicaid Services, the VA Central Cancer Registry, and the Corporate Data Warehouse. Once our database was created we used the Portland VA Unsuspected Radiologic Findings and VISN 23 Lung Nodule Registries, found within the CDW, to develop our raw cohort of subjects. We used structured query language to apply the exclusion criteria as detailed in the protocol. After cleaning and validating our final cohort the additional databases were linked to the cohort. Our cohort contains 8893 Veterans from VISN 20 and 15132 Veterans from VISN 23 with an incidentally detected pulmonary nodule from 2000-2015. Approximately 405 (4.5%) of Veterans from VISN 20 were found to have developed lung cancer after the initial scan that identified a pulmonary nodule. Although there are data that we can obtain from administrative sources, there is still information that must be manually abstracted from patients' electronic medical records. These data include information on nodule characteristics, follow up recommendations, biopsy results, lung function tests, and smoking/asbestos history. Our VA Research Electronic Data Capture database, used for storing abstracted data, has been tested with over 100 practice patients and vetted by the study PI. Our statistician has worked to develop, refine, and test the analytic protocols. Importantly, this latter process will enable us to frequently test the utility of individual data elements in the predictive model.
There are likely opportunities to improve the utility and safety of the nodule evaluation process which will become crucial when CT screening is widely adopted. There is currently no mechanism for using administrative data to identify patients with pulmonary nodules because of a lack of a well-accepted diagnostic code. This information gap substantially hinders the ability of researchers, clinicians, and administrators to efficiently evaluate future screening interventions.
None at this time.