442: Do Emergency Physicians Appropriately Select Patients for CT to Diagnose Pulmonary Embolus?

Abstract

Pulmonary embolus (PE) is a challenging disease that carries the potential for poor outcome and, if missed, is often the subject of medical malpractice litigation. Physicians in emergency departments (ED) may be overusing the computed tomography (CT) to diagnose pulmonary embolus in a defensive fashion. Clinical decision tools have been developed and validated regarding the utility of further diagnostic testing for PE. We sought to evaluate if the emergency physician decision to perform a CT conforms to the validated clinical decision algorithms for pulmonary embolus. This study is a retrospective chart review performed in a level 1, urban, tertiary care ED from October 24, 2008 to October 24, 2009. Inclusion criteria were patients aged 18 years or older who underwent chest CT imaging. We excluded subjects whose imaging was not specifically performed for PE or those with charts documenting the use of a validated PE diagnostic algorithm as part of the workup. Data collected included diagnostic predictors used in each of the rules: Pulmonary Embolism Rule-Out Criteria (PERC), Wells Criteria (Wells), and Revised Geneva Criteria (RGC), D-dimer, and the presence of PE on CT. Each score from the diagnostic algorithms was dichotomized into low/high probability for pulmonary embolism. High probability in this study indicates that the CT scan was indicated. The data were entered into the REDCap database and analyzed by SAS using Fisher, Chi-square, and logistic regression as appropriate. Of 565 ED visits that included imaging of the chest, there were 330 enrolled. The prevalence of PE in the study cohort was 11.5% (38/330) (95%CI: 0.08,0.15). The D-dimer assay was performed in 72.7% (240/330) of subjects with a mean of 2007 (95%CI 1639;2374); 64.6% of subjects had values >500. Of the patients who were ultimately diagnosed with PE on CT, 28.9% (11/38) did not have a D-dimer assay performed prior to CT imaging, and 2.6% (1/38) had a D-Dimer <500. The following odds ratios were observed in the positive PE cohort: admission status OR 7.58 (p500 and hypoxia were not significantly associated with PE (OR 1.47 p=0.37; OR 0.81 p=0.66). (Table 1) Utilizing the algorithms, according to PERC 95% (314/330), Wells 88% (291/330), and RGC 81% (269/330) of subjects met criteria for moderate or high pre-test probability for PE. The sensitivities of each test in predicting PE in this cohort were PERC 100% (95%CI 91;100), Wells 100% (95%CI 91;100), RGC 89% (95%CI 80;99). The specificities for each test were PERC 5% (95%CI 3;8), Wells 13% (95%CI 1;18), RGC 19% (95%CI 15;24).Tabled 1 In this retrospective study, the emergency physician's high clinical suspicion had a high predictive value for pulmonary embolus when compared to any single variable used in the 3 clinical scoring methods including D-Dimer >500. Most CT studies performed in the ED to diagnose pulmonary embolus are clinically indicated by validated prediction algorithms even when performed based on physician gestalt alone.