A simple clinical model for the diagnosis of deep-vein thrombosis combined with impedance plethysmography: potential for an improvement in the diagnostic process.

OBJECTIVES: We recently demonstrated the utility of a clinical model combined with ultrasonography to assist the diagnostic approach in patients with suspected deep-vein thrombosis (DVT). In this study we also sought to demonstrate that the model is useful with impedance plethysmography, a less accurate and less utilized diagnostic test. The original clinical model is slightly cumbersome to use; thus at the completion of the study we attempted to develop a simpler scoring system with a goal of maintaining accuracy.
DESIGN: An open, nonrandomized, multicentre trial.
SETTING: Three centres, two in Canada, and one in Italy.
SUBJECTS: Ambulatory patients with suspected deep-vein thrombosis.
INTERVENTIONS: All patients were assessed clinically to determine the probability for deep-vein thrombosis prior to performing impedance plethysmography and venography. We compared the accuracy of impedance plethysmography between the three pretest probability categories of high, moderate and low. All of the above were performed and interpreted by independent observers. When the study was completed, we revised the clinical model by first performing a simple regression analysis then a multiple logistic regression analysis; a scoring system was devised using the latter.
RESULTS: Impedance plethysmography is significantly more sensitive and less specific for all DVT in patients with high pretest probability for deep-vein thrombosis (P = 0.001). The post- test probability (positive predictive value) for deep-vein thrombosis with an abnormal impedance plethysmography result was significantly different (P = 0.0001) between the three pretest probability categories. Multiple regression analysis has provided a new model with only nine variables and a simple scoring system. The retrospective application of the revised clinical model, which is simpler to use, suggests it will provide similar results as the original clinical model when combined with impedance plethysmography. The combination of impedance plethysmography and the clinical model suggests patients are likely to have false positive results if they have a low or moderate pretest probability for deep-vein thrombosis and false negative results if the pretest probability is high. The combination of a low pretest probability and a normal impedance plethysmography result may exclude the need for serial testing, and represented more than 50% of our patient population.
CONCLUSIONS: The use of the clinical model in conjunction with impedance plethysmography would decrease the number of false positive and negative diagnoses and could markedly decrease the need for serial impedance plethysmography. Combining the clinical model with impedance plethysmography could overcome the fact that impedance plethysmography is clearly less accurate than venous ultrasound imaging. The use of the revised clinical model may increase acceptability and utility, but prospective testing is required before widespread use.