Diagnostic yield of imaging for pulmonary embolism by presenting complaint in the emergency department: an observational study

ObjectivesOur primary objective was to determine whether the yield of pulmonary embolism imaging in the emergency department (ED) is different for patients presenting with “chest pain with cardiac features” than with other complaints. The yield of imaging was defined as the proportion of imaging tests that were positive for pulmonary embolism. Secondary objectives were to estimate the prevalence of pulmonary embolism, the use of imaging, and the yield of imaging for each presenting complaint category. Our hypothesis was that the presenting complaint influences the physician’s threshold for requesting imaging.MethodsWe performed an observational health records review study including all adult patient visits between 2018 and 2019 in three EDs in Hamilton (Ontario), Canada. The primary outcome was the diagnostic yield of imaging (computed tomography pulmonary angiogram or ventilation/perfusion scan). We performed a multivariable regression analysis using a generalized linear model, adjusting for confounders.ResultsDuring the study period, 518,787 patients were assessed and 6,700 received imaging for pulmonary embolism. Among the 29,834 triaged as having chest pain with cardiac features, 1,440 (4.8%) received imaging. Among the 488,953 patients with any other presenting complaint, 5,260 (1.1%) received imaging. The diagnostic yield of imaging was 4.2% for patients with chest pain with cardiac features, 8.6% for those with other presenting complaints, with an adjusted odds ratio of 0.62 (95% confidence interval: 0.45-0.87). The prevalence of pulmonary embolism at 30 days was 0.20% in patients with chest pain with cardiac features and 0.10% for all other presenting complaints.ConclusionsThe lower yield despite a higher prevalence means that the threshold for ordering imaging in people presenting to the ED with chest pain with cardiac features was lower than in other patients. Clinicians should keep in mind this possible bias when assessing their patients.