Emergency department variation in utilization and diagnostic yield of advanced radiography in diagnosis of pulmonary embolus

Background There is growing pressure to measure and reduce unnecessary imaging in the emergency department. Objective We study provider and hospital variation in utilization and diagnostic yield for advanced radiography in diagnosis of pulmonary embolism (PE) and to assess patient- and provider-level factors associated with diagnostic yield. Methods Retrospective chart review of all adult patients presenting to four hospitals from January 2006 through December 2009 who had a computed tomography or ventilation/perfusion scan to evaluate for PE. Demographic data on the providers ordering the scans were collected. Diagnostic yield (positive scans/total scans ordered) was calculated at the hospital and provider level. The study was not designed to assess appropriateness of imaging. Results There was significant variation in utilization and diagnostic yield at the hospital level (chi-squared, p < 0.05). Diagnostic yield ranged from 4.2% to 8.2%; after adjusting for patient- and provider-level factors; the two hospitals with an emergency medicine residency training program had higher diagnostic yields (odds ratio [OR] 2.0, 95% confidence interval [CI] 1.6-2.5 and OR 1.9, 95% CI 1.5-2.4). There was no significant variation in diagnostic yield among the 90 providers after adjusting for patient, hospital, and provider characteristics. Providers with < 10 years of experience had lower odds of diagnosing a PE than more experienced graduates (OR 0.8, 95% CI 0.6-0.9). Conclusions Although we found significant variation in utilization of advanced radiography for PE and diagnostic yield at the hospital level, there was no significant variation at the provider level after adjusting for patient-, hospital-, and provider-level factors.