Diagnosis of suspected deep vein thrombosis of the lower extremity


Deep vein thrombosis (DVT) and acute pulmonary embolism (PE) are two manifestations of the same disorder, venous thromboembolism (VTE). DVT of the lower extremity is subdivided into two categories:

●Distal (calf) vein thrombosis, in which thrombi remain confined to the deep calf veins or the muscular calf veins, and

●Proximal vein thrombosis, in which thrombosis involves the popliteal, femoral, or iliac veins.

This concept is supported by the fact that over 90 percent of cases of acute PE are due to emboli emanating from the proximal veins of the lower extremities. In addition, anticoagulation is highly effective therapy for both conditions. As a result, tests designed to diagnosis DVT are also of importance for the diagnosis of acute PE.

The tests that may be used to confirm the presence of suspected lower extremity DVT and an overview of the usefulness of noninvasive testing in diagnosing acute PE in adults will be reviewed here [3]. The evaluation of the patient with venous thrombosis, with emphasis on the indications for and methods of screening for a hypercoagulable state, the prevention of DVT, and issues related to DVT in children are discussed separately.  


When approaching the patient with suspected DVT of the lower extremity, it is important to appreciate that only a minority of patients (17 and 32 percent in two large series) actually have the disease. This low proportion illustrates the importance of noninvasive testing to minimize the likelihood of inappropriate anticoagulation.

Symptomatic patients without DVT have a variety of other disorders, including popliteal (Baker's) cysts, muscle pulls and tears, and venous insufficiency. This issue is discussed separately.  


Deep vein thrombosis of the lower extremity is subdivided into either distal (calf vein) or proximal (thigh) vein thrombosis . Proximal vein thrombosis is of greater importance clinically, since it is more commonly associated with serious, chronic disease (eg, active cancer, congestive failure, respiratory insufficiency, age >75), whereas distal thrombosis is more often associated with transient risk factors (eg, recent surgery, immobilization, travel) . As a further example, over 90 percent of cases of acute PE are due to emboli emanating from the proximal, rather than the distal, veins of the lower extremities, and the mortality rate of proximal DVT is higher than that of distal DVT .

A diagnosis of DVT raises other important clinical issues, such as the presence of an inherited or acquired hypercoagulable state as well as anatomic variations in venous return (eg, inferior vena caval anomalies, May-Thurner syndrome). An overview of these issues is presented separately.  

Proximal vein thrombosis — Proximal deep vein thrombosis is considered to be of more importance clinically, since it is more commonly associated with serious disease and possibly fatal outcomes. Proximal deep venous thrombosis can be identified by noninvasive testing in 20 to 50 percent of patients with acute pulmonary embolism when only a single study is performed  

In many cases of proximal DVT, silent PE has already occurred by the time that the patient is seen. In one study of 622 patients without symptoms of PE, for example, approximately 40 percent with proven DVT had a high probability lung scan. Some high-risk patients (eg, after major surgery) develop DVT without local signs and symptoms . Such patients may present with sudden and often fatal pulmonary embolism .

The American Thoracic Society (ATS) statement on the diagnostic approach to acute venous thromboembolism, as well as other ATS guidelines, can be accessed through the ATS website at www.thoracic.org/statements.

Calf vein thrombosis — Distal calf vein thrombosis is felt to be of lesser clinical importance than proximal vein thrombosis and its optimal diagnosis and treatment have yet to be defined  

One study, for example, evaluated 68 patients who were clinically suspected to have or were at high risk for DVT .

●Thrombosis was limited to the calf veins in 12; none had clinical or ventilation/perfusion scan evidence of pulmonary embolism

●Thrombosis involved both the proximal (thigh) and calf veins in 15; eight had scan evidence of pulmonary embolism

The accuracy of the Wells score and D-dimer test for the diagnosis of isolated distal DVT was assessed in a single center, cross-sectional study including 873 consecutive outpatients with suspected DVT, in whom pre-test clinical probability, D-dimer, and a complete compression ultrasonography of both lower limbs were performed. The prevalence of isolated distal DVT was 12.4 percent in this group. While the Wells test and D-dimer alone were of little diagnostic utility for isolated DVT, isolated distal DVT could be essentially ruled out by the combination of a low pre-test clinical probability and a normal D-dimer: the negative predictive value was 99 percent (95% CI 95-100) .

If anticoagulation is not administered for isolated asymptomatic distal venous thrombosis, serial noninvasive studies of the lower extremity should be performed over the next 10 to 14 days to assess for proximal extension of the thrombus, which has a higher incidence of pulmonary embolism.

Clinical examination — Given the high risk associated with proximal lower extremity DVT that is not treated and the potential risk of anticoagulating a patient who does not have a DVT, accurate diagnosis is essential. Classic symptoms of DVT include swelling, pain, and discoloration in the involved extremity. In addition, risk factors for DVT should be sought .

Physical examination may reveal a palpable cord (reflecting a thrombosed vein), ipsilateral edema, warmth, and/or superficial venous dilation . However, the clinical diagnosis of DVT is not sufficiently accurate since the symptoms and findings are often nonspecific. Thus, testing is generally indicated to confirm or exclude the diagnosis.

The actual physical signs of venous thrombosis can be quite unreliable. There may be pain and tenderness in the thigh along the course of the major veins ("painful deep vein syndrome"). Tenderness on deep palpation of the calf muscles is suggestive, but not diagnostic. Homans sign is also unreliable. The original description was based upon "an increased resistance to dorsiflexion of the foot" without associated discomfort or pain (despite the mistaken perpetuation of the latter symptom as part of Homans sign in textbooks). There is also the mistaken belief that eliciting Homans sign is dangerous; this concern is based upon a single case of pulmonary thromboembolism following calf pressure during a Doppler examination.

Other signs and symptoms have variable sensitivity, specificity, and accuracy for DVT (confirmed by specific testing). In one series, the respective values for the following local symptoms were:

●Edema – 97, 33, and 70 percent

●Pain – 86, 19, and 58 percent

●Warmth – 72, 48, and 62 percent

The differential diagnosis in patients with suspected DVT includes a variety of disorders, such as musculoskeletal injury and venous insufficiency  

In one study of 160 consecutive patients with suspected DVT who had negative venograms, the following causes of leg pain were identified :

●Muscle strain, tear, or twisting injury to the leg – 40 percent

●Leg swelling in a paralyzed limb – 9 percent

●Lymphangitis or lymph obstruction – 7 percent

●Venous insufficiency (reflux) – 7 percent

●Baker's cyst – 5 percent  

●Cellulitis – 3 percent

●Knee abnormality – 2 percent

●Unknown – 26 percent


Contrast venography has long been considered the reference test for the diagnosis of DVT  . The validity of using contrast venography for this purpose was established in a study that evaluated the course of 160 patients suspected clinically of having DVT in whom contrast venography was negative  . During a six-month follow-up, only 1.3 percent developed DVT (95% CI 0.4-5.4 percent).

However, venography is not recommended as an initial screening due to patient discomfort and difficulty in obtaining an adequate study. The frequency of this problem was illustrated in a report comparing impedance plethysmography and ultrasound to venography in 127 hospitalized patients with clinically suspected DVT of the lower extremity; venography could not be performed in 20 percent of cases because of contraindications or technical factors  .

Noninvasive tests with nearly equivalent diagnostic accuracy for DVT have drastically reduced the need for venography. Venography is currently reserved for situations in which ultrasound or impedance plethysmography is not feasible, when noninvasive studies are equivocal, or when noninvasive studies are discordant with a strong clinical impression . As an example, compression ultrasonography does not detect isolated thrombi in the iliac vein.


Although noninvasive tests of the lower extremities are more readily available and are associated with lower morbidity and cost, they may be less accurate than invasive techniques. Thus, accurate assessment of the role of noninvasive testing in the diagnosis of lower extremity thrombosis requires a direct comparison with the diagnostic results obtained from invasive testing. In addition, clinical outcomes need to be determined in prospective studies on consecutive patients and the results compared with the outcomes of patients whose therapy was based upon the results of standard invasive techniques.


Impedance plethysmography requires the patient to lie still while a thigh cuff is inflated. The change in blood volume at the calf is measured from the impedance of the calf as determined by electrodes wrapped around it. After rapid deflation of the cuff, the proportional change of impedance over the subsequent three seconds is used to measure venous outflow obstruction in a manner similar to spirometry. At present, however, many facilities have neither the equipment nor skilled personnel to perform impedance plethysmography, while the availability of ultrasonography is more widespread.

Limitations — Impedance plethysmography is associated with several shortcomings. Although the equipment is simple and inexpensive, the patient must lie still for at least two minutes and must be positioned correctly to avoid obstruction of venous outflow. As a result, the procedure can be difficult to perform in a paraplegic with involuntary muscle spasms and cannot be performed at all in a patient whose leg is in a plaster cast. Furthermore, false positive tests may result from poor filling due to severe arterial disease, or from venous outflow obstruction in patients with prior venous disease, raised venous pressure associated with heart failure, or raised intrathoracic pressure due to mechanical ventilation.

Accuracy — Despite these limitations, impedance plethysmography is useful for the diagnosis and clinical management of patients suspected of having deep vein thrombosis. Proximal vein thrombosis can be detected by this technique with a sensitivity and a specificity of 91 and 96 percent, respectively .

Two prospective outcome studies of patients suspected of having DVT found that no serious adverse events occurred if serial examinations were performed over two weeks and anticoagulants were withheld from patients with negative results:

●In a Canadian series, there were no fatal pulmonary emboli in 311 such patients, and only 1.9 percent developed DVT during a 12-month follow-up

●In a report from the Netherlands, no fatal pulmonary emboli were observed in 289 patients in whom treatment was withheld, and only 0.3 percent of patients developed DVT during a six-month follow-up .

Since detection of calf vein thrombosis is poor by impedance plethysmography, the clinical utility of essentially ignoring calf vein thrombosis was substantiated by these studies. However, impedance plethysmography only detects venous obstruction; as a result, it is prone to false positive results if this obstruction is due to causes other than venous thrombosis.

The possibility of increasing the accuracy of impedance plethysmography by adding D-dimer measurements is discussed below  .

Recurrent DVT — If available, impedance plethysmography is the preferred test for the evaluation of suspected recurrent DVT. A distinct advantage of impedance plethysmography in this setting is that it normalizes at a rapid, predictable rate after a first DVT. In one series of 161 patients with proven DVT and an abnormal impedance plethysmography, a repeat test became normal in 67 percent by three months and 92 percent by nine months  . Normalization is less common with compression ultrasonography, occurring at a rate of 60 to 70 percent at one year . Magnetic resonance imaging may ultimately supersede impedance plethysmography in this setting because the concentration of methemoglobin within a thrombus increases over time; consequent alterations in signal characteristics permit an estimation of clot age .


A more direct approach to the diagnosis of DVT involves use of compression ultrasonography  . The chronicity of the thrombus may be inferred from the echogenicity of the clot because older clots appear more echodense. Nevertheless, there are no prospective clinical trials that validate this inference.

Procedure — The diagnosis of venous thrombosis using compression ultrasonography is made by the findings such as:

●Abnormal compressibility of the vein

●Abnormal Doppler color flow

●The presence of an echogenic band

●Abnormal change in diameter during the Valsalva maneuver

The American Thoracic Society (ATS) statement on the diagnostic approach to acute venous thromboembolism, as well as other ATS guidelines, can be accessed through the ATS website at www.thoracic.org/statements.

Prospective studies have demonstrated that lack of compressibility of a vein with the ultrasound probe is highly sensitive (>95 percent) and specific (>95 percent) for proximal vein thrombosis [39-41]. Color flow imaging, in addition to duplex Doppler ultrasound, is a less demanding study and is also highly accurate for the diagnosis of above the knee DVT . In comparison, the presence of an echogenic band, although sensitive for DVT, has a specificity of only about 50 percent. The variation of venous size with the Valsalva maneuver has a low sensitivity and specificity for the presence of DVT and is not performed in many centers.

Compression ultrasonography requires the patient to lie in both the supine and prone positions in order to permit examination of the popliteal veins. It is less demanding for the patient than impedance plethysmography in terms of the necessity of lying still.

Portable devices — The use of portable vascular ultrasonic machines for assessing the presence or absence of DVT in emergency departments (EDs) and critical care units (CCUs) has been the subject of a number of studies . As examples:

●In one report, ED physicians were given a 10-minute training session on the use of this apparatus at two specific points (common femoral and popliteal vessels). A total of 47 physicians performed 199 two-point compression ultrasonographic examinations in the ED. When compared with ultrasonographic reports from tests performed by the Department of Radiology, the sensitivity and specificity of the ED examinations were 100 and 99 percent, respectively . Additional studies will be required in order to develop a standard protocol for use of this modality in general ED practice  .

●A multicenter, retrospective review compared 128 bedside intensivist-performed compression ultrasonographic studies (IP-CUS) with a formal vascular study (FVS) performed by ultrasound technicians and interpreted by radiologists . All fellows and attending physicians performing IP-CUS had received formal training at a three-day critical care ultrasonography training course. When compared with the companion FVS study ordered immediately after completion of the IP-CUS study, IP-CUS studies had a sensitivity and specificity of 86 and 96 percent, respectively, with a median time delay between the ordering of the FVS and the FVS result of 13.8 hours. It was concluded that, before adopting an IP-CUS strategy, intensivists must be sure to achieve similar competence as the attending physicians and fellows involved in this study.

●A meta-analysis of 16 studies involving 2379 patients assessed the accuracy of emergency physician-performed ultrasound (EPPU) for the bedside diagnosis of DVT compared with either color-flow duplex ultrasound performed by a radiology department or vascular laboratory, or to angiography . Sensitivity and specificity of EPPU compared with the reference imaging test were 96 and 97 percent, respectively. Prospective studies are warranted to confirm the usefulness of EPPU in the management of patients with suspected DVT.

Limitations — Compression ultrasonography has several limitations.

●It does not detect isolated thrombi in the iliac vein or that portion of the femoral vein within the adductor canal.

●As with impedance plethysmography, the results are limited in patients with deformities or a plaster cast.

●Serial studies need to be performed when the initial test is negative; approximately 2 percent of patients with an initially negative ultrasound develop a positive study when retested seven days later. A single repeat study that is negative five to seven days after an initial negative study predicts a less than 1 percent likelihood of venous thromboembolism over months of follow-up .

●Patients with pelvic neoplasms or abscesses may demonstrate isolated noncompressibility of the femoral vein when thrombosis is absent .

Recurrent DVT — Compression ultrasonography is less useful than impedance plethysmography for recurrent DVT. As noted above, only 50 to 70 of studies return to normal at one year  compared with over 90 percent with impedance plethysmography .

However, in a series of 205 patients with suspected recurrent ipsilateral DVT, the incidence of confirmed VTE during a 6-month follow-up period was 1.3 percent in patients with stable or improved ultrasound findings . It has therefore been suggested that patients presenting with lack of compressibility of a previously fully compressible vein segment or those with an increased vein diameter >4 mm (as compared with earlier vein assessments) be considered to have recurrent DVT, while all others need further testing .

The safety of these diagnostic strategies to exclude recurrent VTE was explored in two studies:

●In 284 subjects in whom recurrent VTE had been excluded by the above criteria and anticoagulation was not given, eight subjects (2.8 percent) developed confirmed, non-fatal VTE over the subsequent three months. Of interest, six of these eight subjects had a superficial or distal vein thrombosis diagnosed at the time of the suspected recurrent VTE episode.

●A second report suggested that it may be safe to withhold anticoagulation from patients with suspected recurrent ipsilateral DVT who have stable or slightly increased (ie, up to 4 mm) vein diameter and negative D-dimer levels . Among 75 newly symptomatic patients who met these criteria and were not treated with anticoagulation, the incidence of recurrent VTE was zero at three months.

More data with longer follow-up are required before these approaches can be recommended.

Accuracy — Compression ultrasonography has been compared with ascending contrast venography for the detection of proximal vein thrombosis in clinically suspected  as well as asymptomatic DVT  . In one study of consecutive outpatients with clinically suspected DVT, compression ultrasonography had a sensitivity and specificity of 100 and 99 percent, respectively  . The high sensitivity of compression ultrasonography was confirmed in a cohort of 1702 patients in whom DVT was suspected clinically . The 1290 patients with negative studies on presentation and at one week had a cumulative rate of venous thromboembolic complications of 0.7 percent at six months, suggesting that very few DVTs were missed by compression ultrasonography.

A prospective outcome study supported both the clinical utility of compression ultrasonography and the view that it is superior to impedance plethysmography for the diagnosis of DVT [58]. In this trial, 985 consecutive outpatients with suspected DVT were randomly assigned to be tested with either impedance plethysmography or compression ultrasonography. Serial studies were performed on days one, two, and seven, and the following findings were noted:

●Although there were no fatal PEs in either group, compression ultrasonography produced fewer false positive studies.

●The incidence of venous thromboembolism over a six-month follow-up period in patients who had a negative test and were not treated with anticoagulants was not significantly different between impedance plethysmography and compression ultrasonography (2.6 versus 1.5 percent).

●The positive predictive value was 94 percent for compression ultrasonography and 83 percent for impedance plethysmography. As a result, 11 percent more patients were anticoagulated unnecessarily in the impedance plethysmography group.

The conclusion was that compression ultrasonography was superior to impedance plethysmography in guiding therapeutic strategy. Nevertheless, this study confirmed the utility of impedance plethysmography if compression ultrasonography is unavailable.

Extended (complete) lower extremity ultrasound — As noted above, routine compression ultrasonography has certain limitations, two of which are the lack of detection of iliac vein disease and the need for serial studies, with which the patient may not comply, if the first is test is negative and clinical suspicion is high. One approach that may overcome these limitations of venous ultrasound involves imaging the entire venous system, including distal (ie, calf) veins, at the time of initial presentation (ie, whole leg ultrasonography) .

Five similarly designed studies used complete lower extremity ultrasound as the sole diagnostic test in more than 3000 patients with suspected DVT . The incidence of DVT or symptomatic PE among patients with a negative initial study was between 0.24 and 1.9 percent during the first three months of follow-up. Two of the studies noted a 1 to 1.5 percent technical failure rate that prevented interpretation and necessitated additional testing .

A randomized trial compared the three-month incidence of objectively confirmed symptomatic VTE in 2098 consecutive symptomatic outpatients with a first episode of suspected DVT of the lower extremities who had an initially negative workup after being randomly assigned to one of the two following diagnostic protocols :

●Whole leg color-coded Doppler ultrasonography (US) performed once, or

●US of the common femoral vein at the groin and the popliteal vein down to its branching calf deep veins at the popliteal fossa plus D-dimer measurement. Subjects with an initially negative US study and a positive D-dimer underwent a second US study at one week, or earlier if clinically indicated (serial 2-point protocol) .

Subjects who tested negative were not anticoagulated. The three-month incidences of objectively confirmed symptomatic VTE for subjects testing negative via the whole leg and serial 2-point protocols were equivalent at 1.2 and 0.9 percent, respectively (observed difference 0.3%, 95% CI -1.4-0.8 percent).

A subsequent meta-analysis reviewed seven studies in which there was a negative whole-leg compression ultrasound study, anticoagulation was withheld, and at least 90 days of follow-up were obtained. In the 4731 subjects included in this study, the combined VTE event rate at 3 months was 0.57 percent (95% CI 0.25-0.89) .

However, the generalizability of these observations may be limited because the quality of complete lower extremity ultrasound studies is operator-dependent; thus, results may vary on the basis of operator experience and/or other technical factors. In addition, the majority of patients enrolled in these trials were outpatients, many had low pre-test probability of DVT  , and the sensitivity of complete lower extremity ultrasound among inpatients is less clear . At the present time, this approach should be reserved for patients enrolled in clinical trials at centers with significant experience in complete lower extremity ultrasound.

Addition of pretest probability (Wells score) — As with all tests, ultrasonography for DVT is most useful when the results are combined with an assessment of pretest probability. While other pretest probability scoring systems are available, such as the Hamilton score , the Wells score for DVT (ie, Wells criteria for DVT) appears to be most commonly used.

One report of 593 patients with suspected DVT validated a measure of pretest probability in conjunction with an algorithm designed to minimize the use of venography or repeat ultrasonography . This measure of pretest probability is referred to as the Wells score or Wells criteria for DVT probability :

●Paralysis, paresis, or recent orthopedic casting of a lower extremity (1 point)

●Recently bedridden for longer than three days or major surgery within the past four weeks (1 point)

●Localized tenderness in the deep vein system (1 point)

●Swelling of an entire leg (1 point)

●Calf swelling 3 cm greater that the other leg, measured 10 cm below the tibial tuberosity (1 point)

●Pitting edema greater in the symptomatic leg (1 point)

●Collateral non-varicose superficial veins (1 point)

●Active cancer or cancer treated within six months (1 point)

●Alternative diagnosis more likely than DVT (eg, Baker's cyst, cellulitis, muscle damage, post phlebitic syndrome, inguinal lymphadenopathy, external venous) compression (-2 points)

A total of three to eight points indicates that there is a high probability of DVT, one to two points indicates that there is a moderate probability of DVT, and zero or fewer points indicates that there is a low probability of DVT.

DVT was documented in 3, 17, and 75 percent of patients with low, moderate, and high pretest probabilities, respectively. Serial ultrasonography was required in 28 percent and venography in 6 percent of patients; venous thromboembolism was diagnosed during a three month follow-up period in only 0.6 percent of patients thought not to have a DVT by this algorithm.

Modified Wells score — This measure has been modified (the modified Wells score or modified Wells criteria for DVT probability) to take one other clinical feature into account. Patients with a previously documented DVT are given an additional one point. According to the modified measure, DVT is likely among patients who have a score of two or greater and DVT is unlikely among patients who have a score of one or less .


Overview — The utility of measuring D-dimer, a degradation product of cross-linked fibrin, has been extensively studied for the diagnosis of both DVT and pulmonary embolus. D-dimers are detectable at levels greater than 500 ng/mL of fibrinogen equivalent units in nearly all patients with venous thromboembolism. In general, it is a sensitive test but lacks specificity for the diagnosis of DVT and is therefore, only useful when negative (ie, cutoff value <500 ng/mL).

The finding of elevated D-dimer concentrations alone is insufficient to establish the diagnosis of venous thromboembolism, because elevated D-dimer levels are not specific for VTE and are commonly present in hospitalized patients, particularly the elderly, those with malignancy, recent surgery, renal insufficiency, and a myriad of other conditions, including the second and third trimester of a normal pregnancy.

A meta-analysis of 11 studies pooling 1337 patients revealed a negative predictive value of D-dimer assays of 94 percent . When D-dimers were less than 500 ng/mL, the likelihood of venous thromboembolism remained less than 10 percent even when pretest probabilities were as high as 55 percent. This hypothesis was supported by a trial in 150 inpatients with suspected DVT in which a low pretest probability and a negative D-dimer by rapid RBC agglutination testing (SimpliRED) had a 96 percent negative predictive value, as well as by other similar studies in low risk populations.

Other studies have documented a reduced negative predictive value when the D-dimer assay was applied to a population with a high prevalence of DVT, such as those examined in an emergency department setting. While the negative predictive value of a normal D-dimer assay as a stand-alone test to rule out DVT may not be sufficiently high in cancer patients, the combination of a negative D-dimer assay and a low clinical pre-test probability appears to be more effective in such patients.  

D-Dimer values rise with age, further hampering its specificity in older patients . Adjusting values to improve its diagnostic utility in this population where DVT is prevalent may improve specificity  . One meta-analysis of 13 cohorts (12,497 patients), compared the sensitivity and specificity of conventional cutoff values for D-dimer (<500 ng/mL) to age-adjusted values (defined as age [years] x 10 ng/mL for patients aged over 50 years). Compared to a conventional cutoff value, higher specificities were reported for age-adjusted cutoff values (age 51 to 60 years: 63 versus 58 percent; 61 to 70 years: 50 versus 39 percent; 71 to 80 years: 44 versus 24 percent; >80 years: 35 versus 15 percent). Despite improved specificity, age-adjusted values for D-Dimer requires validation in clinical practice before they can be applied routinely in clinical practice, for the diagnosis of DVT.

Limitations — A negative D-dimer assay may be insufficient as a stand-alone test in patient populations with a high prevalence of venous thromboembolism  . The results of the above and other published studies concerning the usefulness of D-dimer measurement may be difficult to generalize, since available test methodologies and sensitivities vary . A systematic meta-analysis on this subject has concluded the following :

●D-dimer analysis using enzyme-linked immunosorbent assays (ELISA) yield the best sensitivities (approximately 95 percent) and negative likelihood ratios (approximately 10 percent) for excluding deep vein thrombosis and pulmonary embolism.

●A negative quantitative rapid ELISA result is as diagnostically useful as a normal lung scan or negative duplex ultrasonography finding for excluding VTE.

●The D-dimer test is unidirectional in that a negative test is useful in ruling out VTE but a positive test does not have a sufficiently high specificity or positive likelihood ratio to be helpful in increasing the certainty of a diagnosis of VTE.  

Addition to impedance plethysmography — The usefulness of impedance plethysmography or compression ultrasound may be improved when combined with an assay quantitating the amount of D-dimer. One study evaluated the clinical utility of D-dimer measurements and impedance plethysmography alone and in combination in 214 consecutive patients suspected of having DVT . All patients underwent contrast venography (as the gold standard), impedance plethysmography, and measurement of D-dimer levels. The individual sensitivity and specificity of impedance plethysmography and elevated levels of D-dimer for the diagnosis of DVT were 67 and 96 percent, and 93 and 77 percent, respectively. The combination of D-dimer assay and impedance plethysmography was highly accurate when the results were concordant:

●A negative D-dimer assay and normal impedance plethysmography had negative predictive values for all DVT and proximal DVT of 97 and 99 percent, respectively.

●A positive D-dimer assay and abnormal impedance plethysmography had positive predictive values for all DVT and proximal DVT of 93 and 90 percent, respectively.

The results of the two tests were discordant in 28 percent of patients; the tests could not be used reliably in combination in these patients.

A second prospective study of 398 patients with a first episode of suspected DVT found similar positive and negative predictive values for the detection of DVT [104]. It has therefore been suggested that the combination of D-dimer assay and impedance plethysmography, when concordant, may avoid the need for serial examinations.

Addition to compression ultrasonography — Three studies of patients undergoing compression ultrasonography found that a negative ultrasound, in conjunction with a normal D-dimer measurement, safely obviated the need for a follow-up ultrasound at seven days  , allowing treatment decisions to be made at the time of initial presentation in most patients.

Addition of Wells score — A D-dimer level less than 200 to 500 ng/mL by ELISA or a negative SimpliRED assay in conjunction with a low clinical probability of DVT appears to be useful and cost-effective in excluding DVT without the need for an ultrasound examination . This has been shown to hold in populations at high risk of having DVT, such as those with malignancy, as well as in populations in which the Wells score and the D-dimer may have reduced specificity, such as in the elderly.

●In two studies, one in 224 cancer patients and the other in 193 patients with colorectal cancer, the negative predictive values for a normal D-dimer test in conjunction with a low pre-test probability were 99 and 100 percent .

●In a combined analysis of three prospective studies, negative predictive values for a normal D-dimer test in conjunction with a low/unlikely pre-test probability were ≥99 percent for subgroups of patients <60, ≥60, ≥70, and ≥80 years of age  .

A trial of 1096 outpatients presenting with suspected DVT evaluated the patients as likely or unlikely to have DVT using a clinical model (Wells score for DVT probability)   and then randomly assigned them to undergo ultrasound imaging alone (control group), or to undergo D-dimer testing (D-dimer group) . All patients in the D-dimer group underwent ultrasound testing unless the D-dimer test (SimpliRED or IL-test) was negative and the patient was considered "unlikely" to have DVT. Results included:

●The overall prevalence of VTE during the three-month follow-up was 15.7 percent, and was 28 and 5.5 percent, respectively, in those categorized as "likely" or "unlikely" to have DVT in the pretest clinical model.

●For patients assigned to the D-dimer group who were judged "unlikely" to have DVT and had a negative D-dimer test (and therefore did not have ultrasonography), the incidence of subsequent VTE in the next three months was 0.9 percent. In this group, the negative and positive predictive values for the D-dimer test were 99 and 14 percent, respectively.

●For patients assigned to the control group who were judged "unlikely" to have DVT and had negative ultrasonography, the incidence of subsequent VTE was 1.5 percent.

Based on the above studies, a diagnostic strategy for investigation of suspected lower-extremity thrombosis combining pretest probability, D-dimer testing, and compression ultrasound imaging has been presented. The basics of this algorithm are as follows:

●The pretest probability of DVT is determined from the Wells score and a D-dimer test is performed.

●For those with a "low probability" score and a negative D-dimer, DVT is effectively ruled out. If the D-dimer test is positive, ultrasound is performed to rule out DVT.

●For those in whom DVT is likely (ie, Wells score ≥1) ultrasound is performed in all patients to rule out DVT. For the separate group in which the D-dimer test was positive and the initial ultrasound negative, a repeat ultrasound is performed one week later to rule out DVT.

A retrospective analysis suggested that greater diagnostic utility might be obtained by altering the cutoff D-dimer level for those with decreased levels of pretest probability. From that analysis, a strategy was developed to rule out first time DVT that did not require ultrasonography in every patient suspected of first time DVT when the D-dimer was >500 ng/mL (uniform testing). In this prospective multicenter randomized trial, selective testing in 860 patients was used to rule out DVT without ultrasonography in those with a low pretest probability and D-dimer <1000 ng/mL, or in those with a moderate pretest probability and D-dimer <500 ng/mL, while ultrasonography without D-dimer measurement was employed in all inpatients and outpatients with a high pretest probability. The incidence of symptomatic venous thromboembolism during a three-month follow-up was not significantly different between the 860 subjects undergoing uniform testing or the 860 undergoing selective testing (0.5 percent in both groups) [120]. This selective testing approach may not apply to all assays of D-dimer or to patients with recurrent DVT.

Limitations of the Wells score — The Wells score was developed as an objective means for determining the pre-test probability of a patient having DVT, using patients from a secondary care outpatient setting . It may not perform as well in a primary care setting, due, in part, to differences in the population served, methodology, and interpretation of the individual criteria comprising the Wells score . In their original study, patients with a Wells score of ≤1 and a normal D-dimer had an incidence of DVT of 0.9 percent, while in this latter study, the percent of cases with missed DVT (false negatives) was much higher (2.9 percent) when the Wells score was combined with a negative D-dimer using sensitive quantitative techniques  .

A meta-analysis of 51 studies has shown that the overall assessment of clinical probability of DVT by use of the Wells score is more useful than any of the individual components comprising the score, with a negative likelihood ratio (0.25, 95% CI 0.21-0.29) similar to that of an empirical assessment. Nevertheless, when the population to be assessed contained a large percentage of elderly patients or those with a prior DVT or other comorbidities, as might occur in a primary care setting, the performance of the Wells criteria was reduced . Therefore, it seems prudent that the Wells criteria are used to complement rather than displace the clinician's empirical assessment.

Addition of Hamilton score — The Hamilton score is calculated as follows :

●Plaster immobilization of lower limb: 2 points

●Active malignancy (within six months or current): 2 points

●Strong clinical suspicion of DVT without other diagnostic possibilities: 2 points

●Bed rest (>3 days) or recent surgery (within four weeks): 1 point

●Male sex: 1 point

●Calf circumference >3 cm on affected side (measured 10 cm below tibial tuberosity): 1 point

●Erythema: 1 point

For this clinical probability score the unlikely versus likely cutoff is two or less points.

In a study of 542 consecutive ambulatory patients presenting to an emergency department for assessment of lower limb DVT, the combination of an unlikely probability Hamilton score and a negative Simplify D-dimer assay had a negative predictive value of 99 percent (95% CI 94.7-100) for ruling our DVT and obviating the need for ultrasound evaluation .

Clinical decision rule for primary care providers — A cost-effective clinical decision rule (AMUSE rule or primary care rule) has been developed specifically for the primary care setting, which uses seven objective clinical items and a rapid point-of-care D-dimer assay . This scoring system was employed in 1002 evaluable primary care patients suspected of having DVT primarily because of leg pain (87 percent) or leg swelling (78 percent). Patients with a total score ≤3 did not receive a referral for ultrasonography and did not receive anticoagulation, while those with a total score ≥4 were referred for ultrasonography. Results included the following:

●In the 500 evaluable subjects with a total score ≤3, the incidence of symptomatic VTE at three months was 1.4 percent (95% CI 0.6-2.9).

●In the 502 subjects with a score ≥4, the incidence of VTE on subsequent ultrasonography was 24 percent. The negative likelihood ratio is estimated to be 0.093 (95% CI 0.045-0.192).

These results are consistent with other studies that have indicated a low incidence of clinically evident VTE in patients with a low pretest probability and a negative D-dimer. They are also consistent with the current recommendations that all patients with a high pretest probability of DVT should undergo ultrasonography irrespective of the D-dimer test result.


Magnetic resonance venography is as accurate as contrast venography for the diagnosis of DVT  . This was illustrated in one study that evaluated 85 patients suspected of having DVT, all of whom underwent both magnetic resonance and contrast venography . Magnetic resonance venography had a sensitivity of 100 percent and a specificity of 96 percent. There were only three discrepancies out of 101 comparisons between magnetic resonance and contrast venography, and there was only one discrepancy out of 41 comparisons between magnetic resonance venography and compression ultrasonography.

Some investigators have claimed that magnetic resonance venography is superior to contrast venography in special circumstances such as acetabular fractures. In this setting, one report found that magnetic resonance venography yielded a 22 percent greater detection of thrombi than did contrast venography (10 of 45 comparisons) . However, the study design did not permit the determination of whether treatment of these additional cases altered outcome.

Although the diagnostic accuracy of magnetic resonance venography is comparable to that of contrast venography, outcome data are lacking. In addition, the present high cost of magnetic resonance venography makes it unlikely that it will gain prominence as a noninvasive test for DVT. However, magnetic resonance venography is a useful approach when contrast venography is required but precluded because of allergy to contrast material.

Direct thrombus imaging — Magnetic resonance direct thrombus imaging (MRDTI) may be more accurate than compression ultrasonography (CU) for the diagnosis of recurrent DVT. In one study of 43 consecutive patients with a first episode of VTE, MRDTI normalized over a period of 6 months in all patients, while CU remained abnormal in 31 percent.


Experience is increasing with the use of computed tomography (CT) for establishing the presence of DVT. The protocols under most active investigation image the pulmonary arteries and the subdiaphragmatic deep veins (including the legs) at the same sitting, ideally with no additional contrast medium or venipunctures beyond what is required for a CT pulmonary angiogram [76,135-138]. In some reports, CT venography has performed in a manner comparable to ultrasonographic evaluation in the detection of femoropopliteal venous thrombosis. At present, the use of CT in this setting remains experimental, although the technique holds potential for simplifying the diagnosis of venous thromboembolism in the future.


This section will provide a brief overview of how noninvasive lower extremity studies contribute to the diagnosis and management of patients suspected of having acute pulmonary embolism. An extensive review of the diagnostic strategies for acute pulmonary embolism, including the use of invasive and noninvasive testing, is provided elsewhere.  

Ventilation-perfusion lung scanning remains the best validated noninvasive approach in the diagnostic evaluation of patients with suspected PE . The scan is most useful when it is negative or when it indicates a high probability for pulmonary embolism, particularly when the clinical (pretest) probability is also high:

●In 515 consecutive patients with a negative perfusion scan, withholding anticoagulants resulted in a 0.6 percent rate of non-fatal venous thromboembolic disease.

●The specificity of a high probability scan ranges between 90 and 92 percent, depending upon the criteria used . When present, a high probability scan is considered sufficient evidence to establish the diagnosis in a patient with a strong clinical likelihood of acute PE .

Unfortunately, the overall sensitivity of ventilation-perfusion scanning is low, ranging from 53 to 57 percent. As a result, approximately 71 percent of ventilation-perfusion scans in patients suspected of acute PE are nondiagnostic.

Since over 90 percent of PEs emanate from the proximal veins, some investigators have suggested that a reasonable approach is to withhold anticoagulants in patients in whom the ventilation-perfusion scans were nondiagnostic if serial impedance plethysmography revealed no evidence of proximal vein thrombosis . Using this strategy, venous thromboembolism occurred in only 3 percent of 414 patients. It is important to realize, however, that only patients without severe cardiopulmonary disease were enrolled in this protocol, because even a small embolus in such patients may have serious consequences. Severe cardiopulmonary disease was defined by the presence of pulmonary edema, right ventricular failure, systemic hypotension with blood pressure less than 90 mmHg systolic, syncope, acute tachyarrhythmias, abnormal spirometry with FEV1 less than 1 liter or vital capacity less than 1.5 liters, or respiratory failure (PaO2 less than 50 mmHg and/or PaCO2 greater than 45 mmHg while breathing room air).

More recently, other investigators have shown that the need for pulmonary angiography to diagnose pulmonary embolism could be reduced from 71 to 29 percent if this algorithm were to be implemented . Prospective confirmation of a similar algorithm was described in 1998.


For the approach to a diagnosis of deep vein thrombosis (DVT) of the lower extremity, we and others recommend initial evaluation using a validated clinical prediction rule in conjunction with a high sensitivity assay for D-dimer . Available prediction rules and D-dimer testing perform best when applied to a first episode of VTE, younger patients, those without significant comorbidities, and a short duration of symptoms.

A reasonable diagnostic approach is as follows, and is consistent with clinical practice guidelines from the American Academy of Family Physicians and the American College of Physicians:

●In selected patients with a low pretest probability of DVT according to the Wells score, the likelihood of DVT is low, and further testing (eg, ultrasonography) may not be needed unless the D-dimer is positive or unavailable.  

●Ultrasound evaluation is recommended for patients with intermediate to high pretest probability of DVT according to the Wells score. Repeat ultrasound or venography may be required for those with suspected calf vein DVT and a negative initial ultrasound investigation.

In most circumstances, compression ultrasonography is the noninvasive approach of choice for the diagnosis of patients with suspected DVT. If unavailable, impedance plethysmography with serial studies is an acceptable alternative. One exception noted above is that impedance plethysmography is preferred for possible recurrent DVT since it normalizes more quickly after a previous episode  than compression ultrasonography.

A positive noninvasive study in patients with a first episode of DVT usually establishes the diagnosis, with positive predictive values for compression ultrasonography and impedance plethysmography of 94 percent (95% CI 87-98 percent) and 83 percent (95% CI 75-90 percent), respectively. If the initial study is negative and the clinical suspicion of DVT is high, a repeat study should be obtained on day 5 to 7. Venography is currently used only when noninvasive testing is not clinically feasible or the results are equivocal.


Noninvasive testing of the lower extremities also has utility for the management of patients suspected of acute pulmonary embolism when the ventilation-perfusion scans are nondiagnostic and when the patients do not have a high cardiopulmonary risk. Because the morbidity of noninvasive studies is almost nonexistent, it may be reasonable to consider performing ultrasonography prior to exposing a patient to the risks of pulmonary angiography. Such applications must be undertaken tentatively until a diagnostic approach to suspected pulmonary embolism which incorporates noninvasive testing for DVT has been clinically validated.