Diagnostic test comparisons in patients with deep venous thrombosis

New diagnostic modalities are often judged relative to accepted standard procedures. These comparisons are influenced by the accuracy of the standard test and the prevalence of disease in the study population. We evaluated the importance of these factors in the assessment of antifibrin scintigraphy when used to detect deep venous thrombosis.

Scintigraphy is compared to contrast venography in two populations of patients with different disease prevalence. We calculate the sensitivity and specificity by limb site (calf, knee, thigh) and the overall diagnosis for each modality. The sensitivity and specificity results obtained using venography as a gold standard are compared to those obtained using a maximum likelihood statistical procedure that does not require comparison to a standard test.

A significant variation in the apparent sensitivity, specificity and accuracy is found for antifibrin scintigraphy as related to limb site, disease prevalence and use of a gold standard. The value of antifibrin scintigraphy sensitivity (84.7%) and specificity (75.8%) predicted by the maximum likelihood analysis are substantially higher than those obtained from the estimates based on the use of venography as a gold standard for both high and low disease prevalence populations. The sensitivities and specificities of antifibrin scintigraphy (84.7% and 75.8%, respectively) and venography (71.7% and 80.7%, respectively) are comparable for the combined study group of 268 patients.

To obtain unbiased evaluations of a new diagnostic modality, it is essential to take into account the errors of the standard reference test and disease prevalence in the study population. The results of our analysis suggest that it may not be appropriate to use contrast venography as a gold standard in the assessment of new diagnostic imaging procedures for DVT.