The Role of Ultrasonography in Thromboembolic Disease Management in the Orthopaedic Patient

Abstract

Deep venous thrombosis (DVT) is a well-recognized contributor to increased morbidity and mortality following trauma and elective musculoskeletal procedures. Ultrasound has become a popular noninvasive modality for use in the detection of symptomatic DVT. However, its use as a screening tool in asymptomatic or postoperative patients has been questioned. The reliability of ultrasound rests mainly in the ability of the technicians performing the exam. Ultrasound has been shown to be less reliable in identifying asymptomatic calf thrombi; in institutions where ultrasound DVT surveillance is not performed routinely, the technique suffers from inadequate sensitivity to be utilized for routine screening purposes. Recognition of patients at high risk for DVT, along with an understanding of the limitations of ultrasound, will allow for appropriate clinical application of this modality.

Deep venous thrombosis (DVT) is a well-recognized complication following major orthopaedic procedures, both emergent and elective. Due to the inaccuracy of the bedside clinical diagnosis of DVT^4,13 various invasive and noninvasive surveillance modalities have been developed. Ultrasound has recently gained popularity for use in the diagnosis of symptomatic DVT due to its low morbidity and noninvasive nature. However, various studies have questioned its use as a screening tool to identify the presence and location of DVT in asymptomatic high-risk patients^{6,11,15,23,45}. An understanding of the benefits as well as the limitations of ultrasound will allow for a more appropriate application in the clinical arena.

While ascending venography remains the gold standard for the diagnosis of DVT, various noninvasive surveillance modalities have been introduced. These include [¹²⁵ I] labeled fibrinogen, impedance plethysmography, magnetic resonance venography, and ultrasonography. Due to the fear of transmission of infectious agents by transfusion of blood products [¹²⁵ I] labeled fibrinogen has been removed from the market and, therefore, is no longer an option for use⁴⁴.

Impedance plethysmography is based upon the electrical impedance measured between two electrodes placed on the calf. In patients with DVT proximal to the electrodes there is decreased electrical impedance as the leg becomes swollen with venous blood, thereby increasing its electrical conductivity. When using a tourniquet technique, an expected increase in electrical impedance with the release of the cuff occurs due to the decompression of the venous system. Failure of the impedance to change an appreciable amount after tourniquet release implies the presence of DVT^22,46. While plethysmography appears to be best suited to diagnose occlusive proximal thrombosis in symptomatic individuals, a lack of sensitivity exists in its ability to identify sizable thrombi if they remain nonocclusive³⁴.

Magnetic resonance venography (MRV), although still investigational, appears to be an effective modality in the detection of DVT. Recent studies have shown that when compared to ascending venography, MRV displays a sensitivity between 97 percent to 100 percent for proximal vein thrombosis^10,39. Further, MRV can identify thrombi located in the deep pelvic veins,³¹ an area difficult to visualize even with venography. Magnetic resonance venography is noninvasive, requires no potentially hazardous contrast media, and images the proximal deep venous system of both extremities simultaneously. With further study and greater affordability MRV may become a promising choice for surveillance of proximal thrombotic disease in the lower extremity and pelvis.

BASICS OF ULTRASONOGRAPHY

Ultrasound utilizes the reflections of sound waves to evaluate soft tissue structures⁴⁰. Real time imaging has improved the use of ultrasound with dynamic images that are viewed instantaneously. Two modes of ultrasound are used in the evaluation of DVT: B-mode and duplex ultrasound. B-mode (brightness modulation) imaging utilizes real time imaging combined with a linear array of transmission beams to produce a two dimensional image. This enables the visualization of venous anatomy without the use of contrast media.

More recently, duplex scanning has been introduced as another modality for DVT diagnosis. Duplex ultrasound combines B-mode imaging with pulsed-wave Doppler technology⁴². Doppler ultrasonography is based on the physical principles of Christian Doppler (1803 to 1853),⁸ who described changes in sound wave frequency when reflected from a moving object. When sound waves are reflected by a stationary object, they return at the same frequency at which they were transmitted. However, objects moving toward the transducer reflect waves of higher frequency, and objects moving away reflect waves at a lower frequency. This phenomenon is called the Doppler shift. Pulsed-wave Doppler can be utilized to evaluate the direction and pulsatile nature of blood flow within vessels. With the addition of color flow, Doppler ultrasound allows simplified evaluation of flow directionality with blue arbitrarily indicating flow toward the transducer and red indicating flow away from it. The advantages of color flow imaging are the improved identification of vessels by visualizing blood flow and the ability to establish the direction of flow to differentiate arteries from veins. However, its utility to enhance the identification of DVT in the clinical setting is still debated^24,27.

Venous augmentation is a term used to describe transducer and patient manipulations utilized to improve the sensitivity of surveillance for DVT. Criteria considered to show the presence of DVT include the direct visualization of thrombus, the absence of spontaneous flow by Doppler, absence of phasicity of flow with respiration, and the incompressibility of deep veins with probe pressure (Figure 1). In a study performed by Lensing et al.,²⁸ 220 consecutive outpatients with clinically suspected DVT underwent B-mode ultrasound evaluation using the single criterion of vein compressibility for diagnosis. All results were verified with concomitant venography. They concluded that ultrasound evaluation of vein compressibility is a highly accurate, simple and objective method for detecting symptomatic proximal DVT. These findings were confirmed by other studies^3,14. A more thorough overview of the sensitivities of commonly used augmentation techniques was described by Killewich et al²⁵. In their prospective double blind study, 47 patients, both symptomatic and asymptomatic, underwent both duplex and ascending venographic exams to assess ultrasound augmentation maneuvers that improve DVT detection. When analyzed individually, visualization of intramural thrombus and the absence of spontaneous flow both had low sensitivities (50 percent and 76 percent) but high specificity, respectively (92 percent and 100 percent). The compressibility of veins, in contrast with previously discussed studies, showed low values for both sensitivity (79 percent) and specificity (67 percent). Absence of flow phasicity with respiration was the single best diagnostic finding with a sensitivity and specificity of 92 percent. The most sensitive combination was that of visualization of thrombus and the evaluation of flow phasicity (sensitivity 95 percent). They concluded that isolated diagnostic criteria should not be used to diagnose DVT, but rather a combination of augmentation maneuvers be employed to optimize the reliability of ultrasonography. They further suggested that in equivocal cases of proximal DVT a contrast venogram should be performed to confirm the diagnosis.

Figure 1.

Intraluminal thrombus visualized with ultrasound (with permission)

PREVALENCE OF DVT

Following Trauma

It is widely believed that DVT and pulmonary embolism are major contributors to the morbidity and mortality following major trauma. The literature examining the prevalence and risk factors associated with DVT in this population is sparse. However, in a landmark study by Geerts et al., these questions were more thoroughly examined¹⁸. In a prospective trial of 349 patients involved in major trauma (injury severity score of at least nine), ascending venography was performed to determine the prevalence of DVT and to better define the characteristics of trauma that predisposes to thrombus formation. All patients in this study underwent bilateral ascending contrast venography 14 to 21 days following the traumatic event. These patients received no antithrombotic prophylaxis. Deep venous thrombi were discovered in the lower extremities of 201 of the 349 patients studied (58 percent). Proximal thrombosis was found in 18 percent, while calf DVT was diagnosed in 40 percent of patients. Further, in patients with lower extremity orthopaedic injuries, up to 80 percent developed DVT. Specifically, the incidence of DVT was 77 percent following tibia fractures, 80 percent after femur fractures, 61 percent after pelvic fractures, and 62 percent after spinal column injury. Of the 201 patients with DVT, only three had clinical signs or symptoms suggestive of the condition, suggesting the need for adequate screening techniques. It was concluded that safe antithrombotic prophylaxis is needed, as DVT is a common complication in patients with major multisystem trauma.

Following Joint Replacement

Deep venous thrombosis is the most common complication threatening the life of a patient following total joint arthroplasty (Figure 2). The prevalence of DVT in patients undergoing joint replacement without prophylaxis is between 45 percent to 70 percent for total hips and 50 percent to 84 percent for total knees^16,41,21. With modern prophylactic techniques the DVT prevalence is reduced to 10 percent to 20 percent after hip and 22 percent to 50 percent following knee replacement. Pulmonary embolism (PE) from DVT is responsible for historical mortality rates ranging from 1.7 percent to 3.4 percent in unprotected patients^21,32,37,41.

Figure 2.

Venographic identification of a large femoral vein thrombus following total hip arthroplasty

It has been shown that the risk of PE is higher from proximal DVT than thrombi confined to the calf³². The clinical importance of calf thrombi is increasingly recognized; 17 percent to 23 percent of calf clots have been demonstrated to propagate to proximal deep veins and, subsequently, assume the same risk for embolism. Furthermore, calf thrombi historically account for 40 percent to 60 percent of all thrombi following hip replacement and as high as 95 percent following knee replacement^21,30,36. A more recent study has shown that with contemporary DVT prophylaxis up to 90 percent of thrombi after hip replacement were confined to the calf¹¹. Sensitive screening techniques are mandated as nearly all patients are asymptomatic for venous thromboembolic disease during the initial postoperative period encompassing the acute hospitalization.

ULTRASONOGRAPHY IN DVT DIAGNOSIS

Proximal DVT

The use of ultrasound as a surveillance modality to detect proximal thrombosis in asymptomatic patients has been previously examined^17,20,48. Woolson et al.⁴⁸ commented on the ability of ultrasound to detect proximal DVT following total hip replacement. In their study 150 ultrasound exams were compared with venographic results. Nineteen proximal thrombi were diagnosed with venography and 17 of these 19 were also identified by ultrasound (sensitivity 89 percent). Grady-Benson et al²⁰ studied ultrasound surveillance following total knee arthroplasty and detected seven of seven proximal thrombi (sensitivity 100 percent). Various reports have noted improvements in the ability of ultrasound to diagnose DVT as technicians become accustomed to study protocols^17,49. A recent study by Garino et al. compared ultrasonography with ascending venography following total joint surgery in an effort to determine factors that influence the accuracy of ultrasonography¹⁷. The study was performed on asymptomatic patients in two phases, and the results were compared to determine the effect of experience on the technician's ability to detect DVT with ultrasound. In the first phase, 121 patients who underwent total joint replacement were included and compared with 84 patients in a second phase. In phase one seven thrombi were diagnosed by venography in the proximal veins. Four of these thrombi were large measuring two to 15 centimeters; the remaining three were small, nonocclusive, and less than 1 centimeter in length. Ultrasound was unable to detect any of these thrombi (sensitivity 0 percent). By contrast, in phase two, after becoming accustomed to study protocols, seven proximal thrombi were diagnosed by ultrasound and five were confirmed by venography (two false positives; sensitivity 100 percent; positive predictive value 71 percent). When combined, these three widely cited papers only account for a total of 36 proximal thrombi with an overall ultrasound sensitivity of 81 percent (29 of 36). Yet, despite these limited data, ultrasound has prematurely become accepted as a viable surveillance option following total joint replacement. Initial use of concomitant venography should be practiced to provide an effective means of individual institutional evaluation of ultrasound screening accuracy in this challenging patient population.

The ability of ultrasound to detect proximal venous thrombi is dependent upon the clinical milieu in which it is applied. The sensitivity of ultrasound to detect proximal vein thrombosis in symptomatic patients has been shown to be 94 percent to 97 percent^2,3,25,28,47. In a meta-analysis performed by Wells et al., the accuracy of ultrasound to diagnose DVT in asymptomatic patients was assessed⁴⁵. Eleven studies with venogram controls were identified from the literature which met an established criteria for a low level of bias. In these studies the sensitivity of ultrasound to detect proximal thrombi ranged from 38 percent to 100 percent^15,44. More recently, two studies have evaluated the usefulness of the addition of color Doppler ultrasound in the identification of asymptomatic proximal thrombi^24,27. Both studies were performed as surveillance exams following total hip or knee replacement. Duplex ultrasound displayed sensitivities of 56 percent and 60 percent respectively when attempting to identify proximal thrombi. With the addition of color Doppler imaging one institution improved its sensitivity to 93 percent²⁴, while the second study showed no improvement²⁷. It has been suggested that these lower sensitivities are due to the presence of smaller and less occlusive thrombi than those present in symptomatic patients^23,29. In fact, one study found that ultrasound missed 60 percent of nonocclusive thrombi, less than five centimeters in length, in the thigh²³. Therefore, the reliability of ultrasound as a surveillance modality should be questioned even for proximal DVT when used in asymptomatic high-risk patients.

Distal DVT

The utility of ultrasound in the diagnosis of calf DVT has been shown to be poor^1,38 (Figure 3). A meta-analysis performed by Wells et al.⁴⁴ identified only two studies which evaluated the accuracy of ultrasonography with a defined low level of bias. B-mode ultrasonography was used in one study, and color flow Doppler was used in the other; both results were confirmed with venographic controls. Overall, 14 of 29 distal DVT were identified (sensitivity 48 percent). These results were confirmed by institutions that routinely utilize ultrasound for postoperative DVT surveillance. Grady-Benson et al.²⁰ were able to identify seven of eight distal thrombi, with two false positive and one false negative result (sensitivity 88 percent; specificity 98 percent; accuracy 98 percent; positive predictive value 78 percent; negative predictive value 50 percent). Further, in a study performed at our own institution data were combined from the University of Rochester and The Pennsylvania State University to determine ultrasound sensitivity¹¹. One hundred thirty-two postoperative total joint patients were managed at the University of Rochester. Sixteen distal thrombi were diagnosed by venography of which none were identified with ultrasound (sensitivity zero percent). At Penn State 36 distal DVT were identified by venography in 210 postoperative patients. Ultrasound identified only five distal DVT with one false positive (sensitivity 14 percent; specificity 99.4 percent; accuracy 85 percent; positive predictive value 83 percent; negative predictive value 85 percent). With the combined data of these two studies, only 12 of 60 distal thrombi were identified by ultrasound (sensitivity 20 percent). As these institutions routinely utilize ultrasound for DVT diagnosis we would expect the quality of the ultrasound exam to be better than the average. However, the sensitivities still do not meet the criteria for acceptance of a study employed for routine surveillance. There is a wide variation of color flow and duplex ultrasound sensitivities even among the more experienced institutions.

Figure 3.

Venographic visualization of calf vein thrombosis.

COMPLICATIONS OF PROPHYLAXIS

Sensitive surveillance modalities are desirable to limit extended anticoagulant prophylaxis to only those at risk for pulmonary embolism because bleeding risks associated with the use of anticoagulants have been well-documented. Coventry et al.¹² initially reported bleeding complications in 4.1 percent of patients following total hip replacement surgery when treated with full intensity warfarin. Following trauma, complications have been reported in up to 36 percent of patients⁷. More recently, based on a review of 25 studies, Landefeld et al.²⁶ noted a 3 percent annual frequency of major bleeding complications and further found the risk to be elevated ten-fold during the first month of anticoagulant therapy. In trauma the risk of DVT has been shown to increase with age^9,18. Further, anticoagulant related bleeding events have been noted to increase in the elderly due to the increased prevalence of comorbidities and polypharmacy, as well as vascular endothelial fragility⁵. These data underscore the need for the accurate surveillance of thrombosis as an integral component of a strategy to limit exposure of extended prophylaxis to only those who are at the greatest embolic risk.

CONCLUSIONS

Management Strategy

In the postoperative setting, ultrasound is an unreliable modality when used for surveillance of asymptomatic DVT. It is therefore difficult to justify full intensity anticoagulant therapy for presumed DVT with its inherent risks based on the results of this test alone. The complications of intravenous heparin therapy following total joint arthroplasty have been studied by Patterson et al. who found an overall 30 percent incidence of bleeding complications which increased to 45 percent in patients within the first six days postoperatively³⁵. With the current trend of decreasing postoperative hospitalization following total joint procedures, most contemporary DVT surveillance occurs within the first seven days after surgery. The results of ultrasound screening in this setting are unreliable and should be confirmed prior to the commencement of anticoagulant treatment.

The treatment of calf vein thrombosis remains controversial. Lotke noted a 72 percent incidence of DVT following total knee arthroplasty, with most thrombi located in the calf³⁰. The 1.1 percent incidence of clinically important embolic events in this group of patients led to the recommendation that prophylaxis be deferred. However, recent data following joint replacement confirms propagation rates from the calf to the thigh in 17 percent to 23 percent of patients^21,33,37. In a study performed by Pellegrini et al., it was estimated that one in five fatal pulmonary emboli may arise from isolated calf vein thrombosis³⁶. Further, contemporary prophylaxis techniques, while limiting the number of proximal thrombi, have a limited effect on calf vein thrombosis. In trauma patients, Geerts et al noted a 40 percent prevalence of calf vein thrombosis in unprophylaxed patients¹⁸. However, when low-molecular-weight heparin was added as prophylaxis the prevalence only dropped to 32 percent¹⁹. The high prevalence of calf thrombi combined with the associated risks of propagation puts a large number of patients at risk for delayed pulmonary embolism and accentuates the need of a safe and effective strategy for surveillance and/or extended prophylaxis.

The high prevalence of deep venous thrombosis following orthopaedic procedures and musculoskeletal trauma necessitates an appropriate management strategy. Ultrasound, when used as a surveillance modality, displays a poor sensitivity in detecting asymptomatic calf and thigh thrombosis. Alternatively, an approach of extended prophylaxis for all, without surveillance, is an acceptable method to prevent pulmonary embolism. While this approach avoids the issue of surveillance reliability, it exposes the majority of patients who are not at risk for venous thromboembolic disease to the unnecessary risks of anticoagulant therapy. We prefer a surveillance modality be utilized to identify patients at risk for embolic events, as well as identify these patients not at risk, so prophylaxis may be limited to those at greatest risk for a thromboembolic event. At our institution we continue to rely on venography to screen for DVT in postoperative total joint patients; outpatient post-discharge anticoagulation is used only for patients with identified DVT on contrast venography.

The multiply injured patient presents unique challenges to the detection of DVT. Lower extremity edema makes intravenous access to the foot difficult. The presence of casts, external fixation devices, and implanted fixation hardware further hinders radiographic venous evaluation. Patients must also be hemodynamically stable when transported to the radiology department for venography. Thus, in the trauma patient, the inherent difficulties in obtaining venographic studies preclude its routine use. We have adopted a protocol in which all at risk trauma patients receive DVT prophylaxis preferably with low molecular weight heparin as soon as clinically safe. Each patient undergoes weekly duplex ultrasound with color flow to screen the proximal venous system. In those physiologically stable patients with equivocal ultrasound results, we have a low threshold to obtain a venogram to confirm the diagnosis. Since these high-risk patients are already receiving prophylaxis, these studies are used to detect failure of prophylaxis and the need for more aggressive treatment.

The limitations of ultrasound must be realized when this method is selected for DVT screening in the high-risk postoperative or post-injury orthopaedic patient. The ultrasound exam is highly dependent upon technician experience; its reliability should be determined in each institution by comparison with contrast venography before it is adopted as the sole surveillance method for thromboembolic disease.