Calf sonography for detecting deep venous thrombosis in symptomatic patients: Experience and review of the literature

Purpose. We determined the accuracy of sonography in the detection of isola ted calf deep venous thrombosis (DVT) and the rate of indeterminate ultraso und examinations in patients with physical signs or symptoms suggestive of DVT. Methods. We reviewed the medical literature (MEDLINE) to determine the accu racy of sonography and the frequency of indeterminate studies in detecting isolated calf DVT in patients with physical signs or symptoms suggestive of DVT. A meta-analysis was used to derive summary measures of sensitivity, s pecificity, and accuracy from studies in which 5 or more isolated calf DVT were identified. Frequencies of indeterminate examinations were recorded fo r studies in which these data were provided, and we pooled these results wi th our own data for 196 patients. Results. The meta-analysis revealed that sonography correctly identified is olated calf DVT in 49 of 53 extremities (sensitivity, 92.5%; 95% confidence interval, 81.8-97.9%) and correctly identified the absence of calf DVT in 157 of 159 extremities (specificity, 98.7%; 95% confidence interval, 95.5-9 9.9%), yielding an accuracy of 97.2% (95% confidence interval, 93.9-99.0%) for ultrasound examinations considered diagnostic. However, when evaluating our patient population and the literature, we found a substantial number o f indeterminate studies (overall rate of 54.6% in 463 extremities), with a wide variation in the reported frequency of indeterminate studies (9.3-82.7 %). Conclusions. Sonography is highly accurate in detecting isolated calf DVT i n symptomatic patients, but indeterminate studies occur frequently, with a wide range of reported rates. Each ultrasound laboratory should evaluate it s own rate of indeterminate studies. (C) 1999 John Wiley & Sons, Inc.