Comparison methods of the clinical probability scores in the diagnosis of the Pulmonary Thromboembolism

Introduction: Pulmonary embolism is a preventable disease that is sometime difficult to diagnosed, can be recurrent and has a high morbidity and mortality.Aim: In this study, we aimed to detect and compare the values of various clinical probability scoring methods in diagnosis of pulmonary thromboembolism, both in inpatients and the patients who appealed to emergency department.Materials and methods: In this study, registration forms of thromboembolism suspected 207 patients who were outpatients of Dicle University Medical Faculty Emergency Service and Chest Diseases Clinic and inpatients of various clinics were retrospectively analyzed between February 2007 - January 2010. Patients were separated as inpatients and outpatients into groups. Patients were scored by Wells, Geneva and Modified Geneva clinical probability scoring methods, according to their patient file datas. These clinical probability scoring methods were compared to each other. To find the differential threshold points of scoring systems, the Receiver Operating Characteristic (ROC) curves were used. Statistical analysis was performed with SPSS 15.0 program.Results: PTE (+) of 145 patients, 62 (43%) women, 83 (57%) male; PTE (-) of 62 patients 27 (43%) female and 35 (56%) were male. The most frequent symptom in PTE (+) and PTE (-) patients were shortness of breath and chest pain. Shortness of breath in PTE (-) patients and the chest pain in PTE (+) patients showed a high rate (p <0.001). Homans symptoms and signs of DVT were determined significantly higher in PTE (+) patients than PTE (-) group (p <0.001). DVT history were statistically determined significantly higher in PTE (+) patients than PTE (-) patient group (p = 0.02). When the PTE (+) and PTE (-) patient groups were compared in terms of percentage of inpatients and outpatients, no statistically difference was found (p = 0.14). To determine and compare the differential (threshold) values of scores towards diagnosis, Receiver Operating Characteristic (ROC) curves were performed and the under curve (AUC, Area under the Curve) areas were assessed. In the ROC curve of inpatients and outpatients, the threshold values of all three scoring methods were determined as 5. When the ROC curves of clinical probability scoring methods were compared in outpatients, it was determined that, the undercurve areas of Wells and Modified Geneva scoring methods were significantly larger when statistically compared with Geneva scoring method (p <0.001) (AUC values; Wells: 0.78, Geneva: 0.59, modified Geneva : 0.76). When the ROC curves of clinical probability scoring methods were compared in outpatients group, it was determined that, the undercurve areas of Modified Geneva scoring method was significantly higher when statistically compared with Wells and Geneva scoring methods (p<0.005) (AUC values; Wells: 0.70, Geneva: 0.54, Modified Geneva: 0.75).Conclusion: In the ROC curves of all three clinical probability scoring methods, it was seen that, the Modified Geneva scoring method has the largest undercurve area and secondly the Wells. So these two methods are determined more useful than Genova to reach the diagnosis.