Noninvasive Diagnosis of Deep Vein Thrombosis

More than a million patients in the United States seek medical care each year for leg pain and swelling.^{1 - 2} Often, these symptoms raise the specter of deep vein thrombosis (DVT), which is diagnosed in roughly 20% of those persons evaluated.³ Proximal DVT—thrombosis of an iliac, deep femoral, or popliteal vein—is rightly feared. Untreated, proximal DVT often leads to symptomatic pulmonary embolism, which may be fatal, to progression or recurrence of the DVT, and to the postthrombotic syndrome of chronic pain, swelling, and difficulty walking. Anticoagulant therapy for proximal DVT prevents these complications in most patients⁴ ; however, this treatment is inconvenient, costly, and sometimes harmful, and must be avoided in patients whose symptoms are not attributable to DVT. The accurate diagnosis of proximal DVT is therefore critical.

Diagnosing DVT by physical examination is inaccurate, and until 25 years ago, radiocontrast venography was the only accurate diagnostic method. Because venography is uncomfortable and potentially harmful, noninvasive diagnostic tests have been developed, including impedance plethysmography, radio-labelled fibrinogen scanning, ultrasonography, tests for D-dimer, and clinical prediction rules. These tests have been rigorously evaluated, providing a model of the application of epidemiologic methods to build clinically useful knowledge about diagnostic strategies.⁵ For example, the pretest probability of DVT can be established by a clinical prediction rule,^{3 ,5} compression ultrasonography is highly sensitive (89%-96%) and specific (94%-99%) for symptomatic proximal DVT,⁵ and patients with either a low clinical probability of DVT or normal compression ultrasonography, and with a negative D-dimer test result, rarely have DVT and do not require further testing for DVT.^{5 - 7} Administering or withholding anticoagulant therapy on the basis of the results of serial compression ultrasonography is safe and effective.⁵

Even with these advances, important clinical questions remain in the diagnostic approach to DVT. Are there noninvasive alternatives to serial compression ultrasonography that would rapidly provide a definitive answer (and eliminate repeat testing 7-10 days later)? Would the management of DVT be improved by whole-leg color-coded Doppler ultrasonography, which may provide sufficient evidence at the time of presentation to initiate or withhold anticoagulant therapy without further testing?^{8 - 9}

In this issue of JAMA, Bernardi and colleagues¹⁰ report the results of a randomized trial that addresses these questions. The investigators evaluated 2098 outpatients presenting to 14 Italian hospitals with their first leg symptoms of DVT and compared 2 diagnostic strategies: (1) serial 2-point ultrasonography plus D-dimer testing, which consisted of compression ultrasonography of the deep femoral and popliteal veins in all patients on presentation, D-dimer testing in patients with normal ultrasonography, and repeat ultrasonography 1 week later only in patients with a positive D-dimer test; and (2) whole-leg color-coded Doppler ultrasonography in all patients on initial presentation.

The primary analysis was limited to patients with a negative diagnostic evaluation. In these patients, the 2 strategies had similar rates for the main outcome, which was symptomatic venous thromboembolism over 3 months (0.9% and 1.2% in the 2-point and whole-leg strategies, respectively; observed difference, 0.3%; 95% confidence interval [CI],–1.4% to 0.8%). Few patients were lost to follow-up, assessment and adjudication of the outcome were systematic and unbiased, and the 95% CI for the difference of 0.3% was within the prespecified threshold for equivalence. Thus, the study establishes that the 2 strategies led to comparable outcomes over 3 months. Whole-leg ultrasonography had the advantage that no patient required repeat testing. Serial 2-point ultrasonography plus D-dimer testing also had advantages—compression ultrasonography is simple, reliable, and widely available, and most patients (69%) with a negative initial ultrasound did not require repeat testing because their D-dimer test was negative.

Another important finding reported by Bernardi et al should help to inform management of suspected DVT. Deep vein thrombosis was diagnosed more often by whole-leg ultrasonography than by 2-point ultrasonography plus D-dimer testing (26% vs 22%, respectively; absolute difference, 4.3%; 95% CI, 0.5%-8.1%).¹⁰ This difference is attributable to the diagnosis of isolated calf DVT, which accounted for nearly a fourth of all DVT diagnosed by whole-leg ultrasonography but was not diagnosed by 2-point ultrasonography. Indeed, on the day of presentation, the 2 strategies detected proximal DVT with similar frequency (20.2% and 20.8%, respectively). Although whole-leg ultrasonography diagnosed isolated calf DVT, it did not reduce the 3-month rate of venous thromboembolism, suggesting that symptomatic but undiagnosed and untreated isolated calf DVT often has a benign course.

Earlier studies provide corroborative evidence. For example, in 1985, Hull et al¹¹ reported that serial impedance plethysmography alone failed to detect calf DVT but had long-term outcomes similar to serial impedance plethysmography plus fibrinogen leg scanning, which detected calf DVT twice as often as proximal DVT. In 2005, Kearon et al⁷ reported complementary findings—in symptomatic patients with a normal 2-point ultrasound of the leg, venography in patients with an abnormal D-dimer test diagnosed isolated calf DVT but did not decrease the rate of late venous thromboembolism compared with repeated 2-point ultrasonography in patients with a negative D-dimer test.

The increasing body of evidence that diagnostic strategies that rule out proximal DVT at presentation and 1 week later have similar outcomes to strategies that also rule out isolated calf DVT suggests that the search for isolated calf DVT is unnecessary as long as proximal extension is ruled out a week later.¹² This inference stands in contrast with the widely accepted American College of Chest Physicians guideline that isolated calf DVT should be treated with an anticoagulant for at least 6 weeks.¹³

This guideline, however, is based largely on 3 lines of evidence that is not definitive. First, in the only randomized trial restricted to patients with isolated calf DVT,¹⁴ 3 months of oral anticoagulant therapy reduced DVT recurrence at 1 year to 5% compared with 32% without oral anticoagulant therapy. This trial was quite small (n = 51), however, and it has not been reproduced. Second, in a randomized trial by Schulman et al,¹⁵ 6 months of anticoagulant therapy for venous thromboembolism was better than 6 weeks of therapy in reducing 2-year recurrence rates (9.5% and 18.1%, respectively; P < .001). The relative risk reduction was similar in patients with isolated calf DVT and in those with proximal DVT.¹⁵ However, the risk reduction in patients with isolated calf DVT was not statistically significant, and among patients with isolated calf DVT and a temporary risk factor (eg, recent surgery), DVT recurrence was both infrequent (3%) and unrelated to the length of anticoagulant therapy. Third, in a randomized trial¹⁶ that included 197 patients with isolated calf thrombosis, 6 weeks of anticoagulant therapy led to a low recurrence rate (1%) comparable with that with 12 weeks of anticoagulant therapy. Thus, the evidence that isolated calf DVT should be treated with an anticoagulant is not compelling.

How should clinicians approach patients with a possible first episode of DVT? Based on the available evidence, it would be reasonable to choose 2 tests initially—a clinical prediction rule and D-dimer test, a clinical prediction rule and 2-point ultrasonography, or 2-point ultrasonography and a D-dimer test. If both tests are negative, DVT is effectively ruled out and anticoagulation can be withheld safely.^{6 - 7 ,10 ,17} If DVT is not ruled out, 2-point ultrasonography should be performed if not already performed. If DVT has neither been ruled out nor diagnosed by ultrasound, a second ultrasound should be performed 1 week later; if that ultrasound is negative for DVT, no further testing is indicated. The results of the trial by Bernardi et al¹⁰ show that whole-leg ultrasonography has little advantage, unless a course of anticoagulant therapy for isolated calf DVT is preferable to repeating 2-point ultrasonography a week later.