Each year, over 150,000 Americans are diagnosed with a pulmonary nodule. Evaluation of pulmonary nodules to identify the subset that are malignant creates a substantial burden on the healthcare system. Veterans are at particular risk for developing nodules because they are more likely to be current or former smokers.
Aim 1: To develop a VISTA-based algorithm to identify a cohort of Veterans with newly diagnosed pulmonary nodules at two VAMCs to determine frequency of nodules in these centers; Aim 2: to develop a structured coding form to track downstream resource utilization in these VAMCs attributable to the diagnosis of a pulmonary nodule.
We created algorithms to identify Veterans with pulmonary nodules at the White River Junction (WRJ) and Boston VAMCs by searching chest x-ray and CT reports for terms like "nodule" or "mass." Using a structured coding form, we abstracted data on related resource utilization from medical records of 300 Veterans with no prior history of cancer and a pulmonary nodule diagnosed between 2003-6. We ended data abstraction when any of the following occurred: nodule resolved; cancer was diagnosed; or 3 years passed since nodule diagnosis.
The algorithms identified 493 WRJ and 1875 Boston Veterans with a chest imaging report from 2006 suggestive of a pulmonary nodule. The positive predictive value was 78% at WRJ and 72% at Boston. Among the 300 Veterans whose charts were reviewed, nodule size was <1 cm in 71%. Substantial resource use, including 739 imaging studies (chest x-ray, CT, or PET scans), 70 pulmonary consults, 64 invasive procedures (biopsy or resection), and 18 hospitalizations directly related to the nodule occurred between the time of diagnosis and the end of resource tracking. Only 6.8% of Veterans with a pulmonary nodule turned out to have lung cancer. Management of pulmonary nodules did not closely adhere to professional guidelines; there were multiple instances at both sites of more aggressive and less aggressive management than recommended by guidelines.
This pilot work confirms that pulmonary nodules are common, and that computerized algorithms can identify these Veterans to facilitate research in a larger VA cohort. VA practice in managing pulmonary nodules is often not consistent with professional guideline recommendations; this is an area that can be targeted for improvement within the VA.
None at this time.
Health Systems, Cancer
Epidemiology, Diagnosis, Research Infrastructure, Prevention
Cancer, Guideline Development and Implementation, Organizational Planning, Outcomes, Predictive Modeling, Screening