Purpose: Although clinical examination has been reported to be unreliable in the diagnosis of deep vein thrombosis (DVT), this conclusion has often been derived from hospitalized patients (HP) and may not be applicable in an outpatient setting. This study was undertaken to define clinical parameters that might predict the diagnostic value of emergency venous duplex scanning (VDS).
Methods: Venous duplex scans performed over a 9-month period (interval I) in 154 outpatients (154 limbs) and 145 HP (145 limbs) with suspected DVT were reviewed.
Results: Eighteen percent of scans of outpatients and 31% of scans of HP were interpreted as positive for lower extremity DVT. With stepwise logistic regression analysis, criteria predictive of a negative result of outpatient VDS included (1) duration of symptoms greater than 7 days (p = 0.003), (2) thigh circumference difference relative to the uninvolved side of less than 3 cm (p = 0.001), and (3) no history of neoplasia (p = 0.03). This model, when applied prospectively to 68 outpatients (68 limbs) over the next 5 months (interval II), yielded a negative predictive value (NPV) of 96.7% (sensitivity 90.9%, specificity 50.9%, positive predictive value 26.3%). Of the 222 outpatients examined during intervals I and II, 98 (44%) met these three clinical criteria. Three of these 98 outpatients had DVT on VDS and thus would have been misclassified as having a negative result. With a similar logistic regression analysis for HP, clinical criteria achieved an NPV of only 75% (sensitivity 36%, specificity 90%, positive predictive value 62%).
Conclusion: Clinical assessment is unreliable in the diagnostic evaluation of HP with suspected DVT. In an outpatient population, however, clinical evaluation with the above criteria achieved an NPV of 96.7% in the diagnosis of DVT. These parameters may be useful as guidelines in determining the appropriateness of emergency outpatient VDS.