Improving diagnostics of pulmonary embolism with clinical prediction models in a clinic of pulmonary diseases

Abstract

Purpose of the Study: Clinical prediction models have been developed to assess the pre-test probability for pulmonary embolism (PE). The Wells model and the revised Geneva score are the two most well studied. Our purpose was to compare the two prediction models, and indentify the frequent clinical findings of PE in patients admitted to the University Clinic of Pulmonary and Allergic Diseases Golnik. Methods: In 100 random emergency department patients and hospital inpatients with suspected PE and performed pulmonary CT angiography (CTPA) as the gold standard, a retrospective assessment of the clinical probability of PE by the Wells rule and the revised Geneva score was made. ECG, D-dimer, NT-proBNP, arterial blood gas analysis, chest X-ray, CTPA and 13 other clinical findings were analysed as well. Results: Average age was 65 years (SD 14.5), 39 % were male. The overall prevalence of PE was 33 %. The rates of PE in low, moderate, and high PE risk groups as determined according to the Wells model and the revised Geneva score were 3.7, 53,1, 100, and 14.3, 32.1, 83.3 %, respectively. ROC analysis showed that the Wells model was statistically more accurate than the Geneva score with the area under the curve (AUC) in Wells model 0.85 (95 % CI 0.762–0.936) and in Geneva score 0.73 (95 % CI 0.612–0.838). Sudden dyspnea, active malignancy, venous thromboembolism (VTE) history, estrogen therapy, deep vein thrombosis (DVT) signs, ECG changes and lower PaCO2 were significantly more frequent in PE group. All patients with PE had an increased concentration of D-dimer, and no PE were diagnosed in the group of patients with normal D-dimer. CTPA was ordered in 17 % of patients with low pre-test probability of PE according to Wells criteria and normal D-dimer. Conclusions: The Wells model is more accurate than the Geneva scoring system for the diagnosis of PE in patients admitted to a pulmonary clinic. Additional findings, such as sudden dyspnea, estrogen therapy, ECG changes and lower PaCO2, should always be incorporated in clinical assessment of PE. Adding the Wells algorithm to the clinical pathway for PE management might slightly decrease the number of CTPA.