3009 — Leveraging VINCI Data to Identify Patients at Risk for Delays in Follow-up of Abnormal Lung Imaging

Murphy DR, Houston HS&RD Center for Innovations in Quality, Effectivenss, & Safety; Meyer AN, Houston HS&RD Center for Innovations in Quality, Effectivenss, & Safety; Wei L, Houston HS&RD Center for Innovations in Quality, Effectivenss, & Safety; Bhise V, Houston HS&RD Center for Innovations in Quality, Effectivenss, & Safety; Wu L, Houston HS&RD Center for Innovations in Quality, Effectivenss, & Safety; Russo E, Houston HS&RD Center for Innovations in Quality, Effectivenss, & Safety; Sittig DF, University of Texas Health Science Center; Singh H, Houston HS&RD Center for Innovations in Quality, Effectivenss, & Safety;

Objectives:
Delays in lung cancer diagnosis are one of the most common reasons for VA malpractice claims. Often, they involve delays in diagnostic evaluation in response to abnormal imaging. Due to lack of suitable methods to identify patients at risk for delayed diagnostic evaluation, we tested the performance of an algorithm that used VINCI (VA's national clinical research database)-accessible data to identify (i.e. "trigger") patients with potential delays in follow-up.

Methods:
Using literature reviews and clinical expert input, we refined a set of previously developed criteria to identify patients that might be experiencing delays in diagnostic evaluation of a "suspicious for malignancy" chest imaging result. The trigger first identified the "red flag" ("suspicious for malignancy" chest imaging result), and then excluded patients who would not require further evaluation (e.g., terminal illness) or who had already received appropriate follow-up action (e.g., biopsy). The criteria were programmed into a structured query language (SQL) program and applied to VISN 12 data via VINCI. To confirm potential delays, clinicians reviewed trigger positive records and classified whether each record had a delay, no delay, or needed additional tracking to ensure future action (e.g., repeat imaging in 6 months). Trigger negative records (i.e. abnormal imaging result with no potential delay as determined by the trigger) were also reviewed. Reviewers were blinded to trigger status. We calculated PPV, NPV, sensitivity, and specificity of the trigger.

Results:
The trigger was applied to 208,633 patients seen throughout VISN 12 between January 1 and December 31, 2012, and identified 1847 chest imaging results suspicious for malignancy, of which 655 were trigger positive. Of these, we reviewed 400 randomly selected records and identified 158 (39.5%) with an actual delay, 84 (21%) that required additional tracking, and 158 (39.5%) without a delay (PPV = 60.5%; inclusive of patients requiring tracking). Of 84 requiring tracking, 62 (73.8%) received follow-up action within the planned timeframe. Of 100 "trigger negatives" reviewed, 92 were truly negative (NPV = 92%). Sensitivity and specificity were calculated as 96.8% and 36.8%, respectively.

Implications:
In a large VA network, a VINCI-based trigger algorithm reduced the number of records requiring review by 65% and improved the feasibility of detecting delays in diagnostic evaluation of chest imaging results that are suspicious for malignancy.

Impacts:
Strategies that involve transmission of triggered data from VINCI to the point of care could improve patient safety.