Thermography and plethysmography in the diagnosis of deep venous thrombosis--a comparison with phlebography.

Ninety-two patients with suspected unilateral deep venous thrombosis (DVT) in the lower limb were examined by thermography, plethysmography and phlebography. ROC analysis (Receiver Operating Characteristics) was used to evaluate discrimination thresholds and to compare thermography and plethysmography (four variables) with phlebography. The sensitivity of thermography, 85% (94% for out-patients), was higher than that of plethysmography (58-79%) for the discrimination thresholds chosen. The specificity of thermography was low, 39% (42% for out-patients) or 55%, if obvious relevant clinical findings were included in the evaluation. The specificity of plethysmography was much higher (80-97%). Optimum combination of the four plethysmographic variables showed predictive values of 93-94%, while combination of thermography and plethysmographic variables showed higher predictive values (95-97%), mostly because of a higher sensitivity of thermography for distal DVT. A possible reduction of the number of phlebographic examinations by at least 50% and a cost reduction of 25% could have been obtained without any appreciable loss of diagnostic accuracy. A follow-up study of 112 consecutive patients, examined according to the recommended screening method, showed a reduction of phlebographic examinations by 62%.