2012 HSR&D/QUERI National Conference Abstract
3084 — Using Electronic Triggers to Identify Patients at Risk for Diagnostic Delays in Lung Cancer
Laxmisan A, Murphy DR, and Reis BA, Houston VA HSR&D COE, Baylor College of Medicine; Thomas EJ, University of Texas Houston Medical School, Memorial Hermann Center for Healthcare Quality and Safety; Singh H, Houston VA HSR&D COE, Baylor College of Medicine;
Delays in follow-up of chest imaging studies are common and can potentially lead to poor outcomes and malpractice claims. We developed an electronic trigger to identify patients at high-risk for potential delays in lung cancer diagnosis.
Using literature review and expert input, we developed a set of rules to trigger patient records suggestive of missed or delayed follow-up of abnormal chest imaging. We then applied these rules to the data warehouse of a large, tertiary care VA facility to identify a training cohort of patients who met the following criteria: 1) age >30 years, 2) seen at the local facility between January 1 and December 31, 2008; 3) had an abnormal chest x-ray or abnormal chest computed tomography coded by the radiologist as “suspicious for malignancy;” and 4) had no evidence of follow-up care (repeat imaging, pulmonary consultation, or a lung biopsy) within 30 days of imaging test. We iteratively refined the trigger by conducting sequential chart reviews to identify clinical criteria warranting exclusion, such as patients receiving palliative care. The refined trigger was then applied to a validation cohort of all patients seen at the facility between January 1 and December 31, 2009. Chart reviews were performed using a standardized pre-tested data collection instrument and positive predictive value (PPV) of the trigger was calculated.
Of 89,132 patient records to which the trigger was applied, 131 (yield = 0.15%) were trigger positive. We then performed chart reviews on these patients and identified that 69 to be true trigger positives (PPV = 52.7 %); i.e. there was no documented follow-up action within 30 days . Of 62 patients that were false positive, the most common reason was presence of non-malignant lesions: 17/62 (27.4%).
EHR-based triggers can be useful in detecting patients at high risk for potential delays in follow-up of imaging suspicious for lung cancer. This method is far more efficient than non-selective record reviews to identify these patients.
We are now testing trigger-based “prospective” detection of delays. In future, these methods could potentially be useful for other VA facilities to identify missed follow-up of chest imaging studies suspicious for cancer.