The Diagnosis of Venous Thromboembolism (VTE) in the Emergency Room: Poor Compliance with Recommended Diagnostic Algorithms.

Background: The diagnosis of VTE remains problematic in the emergency room. There are a number of algorithms, for both pulmonary embolism (PE) and deep venous thrombosis (DVT), that have been validated for use. These incorporate clinical assessment of pretest probability, D-Dimer measurement, and non-invasive diagnostic imaging. The purpose of this study was to evaluate adherence to the VTE diagnostic algorithms (according to Wells, Anderson et al.) at Victoria Hospital, London Ontario (a tertiary centre).

Methods: The charts for all patients who presented with suspected DVT or PE during a two-month period were retrospectively reviewed. Patients were identified by those who had D-dimer testing or diagnostic imaging for VTE (D-dimers are only indicated for use in our hospital for the investigation of VTE). Patients were excluded if they were currently anticoagulated, pregnant, had upper limb DVT or were <18 years of age. All charts were reviewed by 3 persons to assign the Wells criteriae and to review the adherence to the recommended algorithms - determination was by consensus.

Results: There were 289 patients included, 157 women and 132 men with an average age of 57. Of 289 patients included, 242 were investigated for PE with 6 events occurring, for an event rate of 2.5%. Forty-seven patients were investigated for DVT, with 4 events, for an event rate of 8.5%. The algorithms were seemingly correctly applied at superficial analysis in 64.4% of all patients and incorrectly applied in 20.8% of all patients. In 14.9% of patients, the algorithms were not followed to completion due to admission to hospital or an alternative diagnosis developing. Consultants correctly managed in 68.37% of cases and Residents in 59.8%. Although 71% of patients with suspected PE were seemingly managed according to algorithm - the vast majority of these patients had a low pretest probability (217) and were investigated with D-dimer only, which was negative. There were 25 patients in the moderate to high probability group and 17 were incorrectly managed with either unnecessary D-dimer assessments or a lack of leg imaging after non-diagnostic V/Q scans or CT scans. Eight patients in the low probability group had imaging despite negative D-dimer tests. With an event rate of only 2.5%, a disproportionately large number of patients were investigated for PE. Only 30% of patients with suspected DVT were managed correctly according to algorithm, 41% for “unlikely” and 7% “likely”. Of the 32 patients in the “unlikely” group, 9 patients had U/S performed despite negative D-dimer and 5 patients had U/S without D-dimer assessment. For the 15 patients who were “likely”, 6 had D-dimers performed upfront and 6 did not have D-dimers done after initial negative imaging. Overall, of 46 patients who had chest imaging for PE, 13 V/Q scans and 1 CT scan were done that were not indicated and 7 patients who should have been imaged were not. In the DVT group, 39 U/S were done and 10 of these were not indicated.

Conclusions: The disproportionate amount of PE patients, and the overall low event rate for PE, suggest that D-dimers are being used indiscriminately as a screening test for the diagnosis of chest symptoms, an approach that has not been validated by studies. The algorithms for the diagnosis of DVT and PE are not being applied appropriately at our centre. The implications of this include the potential for missed diagnoses as well as the cost and potential clinical consequences of over investigation.