The Accuracy of Wells Clinical Model for Predicting Pretest Probability in Deep Vein Thrombosis (DVT) at Dr. Sardjito Hospital

Abstract

ABSTRACT
Background: The diagnosis of deep vein thrombosis (DVT) is an important issue in the medical field because a large number of cases are undiagnosed. The accurate diagnosis of the DVT is essential to prevent complications of acute pulmonary embolism and long-term complications. The complication such as pulmonary hypertension and post phlebitis syndrome. It also to avoid improper anticoagulant therapy associated with a high risk of bleeding. There are several models of clinical suspicion of DVT include Hamilton’s score and Wells’ score. The Wells clinical model for predicting pretest probability for DVT has been widely used and tested its validity in several countries in Europe. The accuracy of a Wells clinical model of pretest probability for DVT has not been done for the races of Asia especially in Indonesia.
Objective: This study aimed to test the accuracy and precision (sensitivity, specificity and predictive value) of Wells clinical model for predicting pretest probability for diagnose DVT for patients in the Dr. Sardjito Hospital.
Methods: This study was a diagnostic study with cross-sectional design. The subjects were patients with DVT symptoms who visited Tulip clinic, outpatient and inpatient in the department of Dr. Sardjito General Hospital during January 2011-May 2012. In this study, a total of 100 patients were participated the Wells clinical model of pretest probability for DVT examination, Doppler ultrasound and D-dimer examination. The negative results on the fi rst ultrasound examination will be examined again 1 week later.
Results: DVT was confi rmed in 56 patients (56%) from 100 patients. The baseline characteristics showed the number of women more than men and the mean age was 52.98 ±1.335 years. The 57% of subjects were women and the median age was ≥60 years. The most frequent risk factors of DVT were cancer 49 patients (49%), bed ridden 42 patients (42%) and elderly 33 patients (33%). The accuracy of the Wells clinical model of pretest probability for DVT was 81%, precision was 77.6%, sensitivity was 92.8%, specifi city was 65.9%, and negative predictive value was 87.9%.
Conclusion: The Wells clinical model for predicting pretest probability in DVT has a high accuracy, moderate precision, high sensitivity and poor specifi city to diagnose DVT at Dr. Sardjito Hospital.
Keywords: deep vein thrombosis/DVT, the Wells clinical model of pretest probability for DVT, accuracy test, precision