Chapter 20: Venous Thromboembolism

Venous thromboembolism (VTE), manifested as either deep venous thrombosis (DVT) or pulmonary embolism (PE), is a leading cause of morbidity and mortality in the critically ill patient. Standardizing for age and race, the incidence of VTE in the United States is approximately 70 to 120 cases per 100,000 individuals. The incidence of VTE increases exponentially as the population ages although no difference is apparent between genders. The high incidence of VTE is mainly dominated by PE diagnosed at autopsy and as a contributory factor in 4% to 11% of deaths. Despite appropriate therapy, 1% to 8% of patients with PE will not survive and others will experience long-term complications including postphlebitic syndrome (40%) and chronic thromboembolic pulmonary hypertension (4%). In the International Cooperative Pulmonary Embolism Registry, the death rate was 58% in patients who were hemodynamically unstable at the time of presentation and 15% among those who were hemodynamically stable.

VTE often eludes diagnosis due to its wide variety of presentations and diverse pathophysiological mechanisms. The diagnosis of PE is confirmed by objective testing in only about 20% of patients. Accurate diagnosis of VTE can minimize the risk of thromboembolic events and complications related to unnecessary anticoagulation and treatment. Consequently, the Agency for Healthcare Research and Quality ranks prevention of VTE as the first priority out of 79 preventive initiatives that can improve patient safety in healthcare settings.

Because DVT and PE have a spectrum of presentations, they have been approached by many clinicians in an algorithmic manner to address diagnosis, treatment, and prevention. The clinical presentation of VTE ranges from asymptomatic tachycardia to respiratory failure and hemodynamic instability. Additionally, VTE can occur spontaneously or from predisposing risk factors, catheter related, and may arise from genetic predisposition. In more than half of the cases, VTE is provoked by surgery, immobilization, advanced age, pregnancy, the use of oral contraceptives, or hormone-replacement therapy. The difficulty for clinicians lies in the early recognition and prompt treatment of VTE because of these widely diverse presentations. Prevention of DVT/PE in the ICU patient also poses many difficulties in choosing the appropriate regimen for prevention based on the risk-benefit ratios. The goal is to prevent recurrence because this can be fatal in approximately 5% of patients and prolonged therapy with anticoagulants in itself carries bleeding risks.

Risk factors for VTE include increased age (≥ 40 years), obesity, prior VTE, malignancy, prolonged immobilization, major surgery, trauma, cardiac failure, pregnancy, hormone replacement therapy, and oral contraceptive therapy. Critically ill patients are especially at high risk for VTE due to severe underlying disease, immobility, and central venous catheterization.

According to Virchow's triad, the pathophysiology of thrombosis lies in three mechanisms: vessel wall injury, stasis, and hypercoagulability. A failure of one of these components leads to clot formation and DVT/PE. Hypercoagulable states may be inherited or acquired. The inherited type should be expected in patients with recurrent or life-threatening VTE, family history of thromboembolism, those younger than 45 years old, no acquired risk factors, and females with multiple miscarriages. Inherited risk factors for VTE include factor V Leiden (activated protein C resistance), protein C and S deficiency, antithrombin III deficiency, prothrombin gene mutation, dysfibrinogenemia, and disorders of plasminogen. Acquired hypercoagulable states include antiphospholipid antibody syndrome (anticardiolipin antibodies, lupus anticoagulant) and hyperhomocysteinemia.

The location of the VTE is another factor in risk evaluation and treatment with anticoagulation. The most common site of DVT is the calf vein but only 15% to 20% will extend proximally into the deep proximal veins. The deep vein thromboses that are of significance are those that occur more proximally, which typically is in the posterior tibial vein and the common femoral vein. Those with proximal vein thromboses have a 90% likelihood of developing PE. Identifying high-risk thrombi in the proximal veins is imperative to prevent the progression to PE; superficial veins thromboses rarely are associated with PE. Hypercoagulable states can also lead to venous thromboses in the superior and inferior vena cava (IVC), renal veins, and hepatic veins.

Upper extremity DVT can occur and involve the brachial, axillary, and subclavian veins and can extend proximally to the brachiocephalic vein, internal jugular vein, or superior vena cava. They account for less than 10% of all DVTs and approximately 75% are provoked by the placement of pacemakers and central venous catheters either in the internal jugular or in subclavian vein. In addition, right atrial thrombi can occur due to venous catheters or cardiac disease, such as atrial fibrillation, cardiomyopathy, ventricular aneurysms, and malignancy. About 5% of patients with arm DVT will have a PE, 20% will have postthrombotic syndrome, and 8% will have recurrence if untreated.

The hypoxemia that is observed in patients with PE is caused by the ventilation-perfusion mismatch resulting from increased physiologic dead space and intrapulmonary shunting. This is commonly associated with increases in minute ventilation and airways resistance, decreased vital and diffusion capacities, and in patients with a potentially patent foramen ovale, progressive pulmonary hypertension may lead to intra-arterial right-to-left shunting and severe refractory hypoxemia. Hemodynamic instability may result from the obstruction of the pulmonary circulation and increase in pulmonary vascular resistance, which impedes right ventricular outflow leading to reduced left ventricular preload and ultimately, a decreased cardiac output. Patients with underlying cardiopulmonary disease are more likely not to tolerate increases in pulmonary artery pressures and thus are more susceptible to developing right heart failure.

The clinical presentation of VTE can extend to a wide spectrum of signs and symptoms which present a multitude of diagnostic challenges. Both DVT and PE can present with nonspecific signs and symptoms, such as shortness of breath, fever, tachycardia, pain, and even no symptoms at presentation. However, both can have specific presentations, such as patients with DVT can present with erythema, leg swelling, tenderness, or pain, whereas those with acute PE can present with chest pain, dyspnea, unexplained sustained hypotension, hemoptysis, syncope, increased P₂ on physical exam, right heart failure, respiratory failure, and death. It is important to note that many of the signs and symptoms of PE may frequently be seen in patients with concomitant cardiac and pulmonary disease, and that these manifestations may be a result of a coexisting disease or a superimposed acute PE.

Acute PE can be massive or massive. Massive PE is characterized by hypotension (defined as a systolic blood pressure < 90 mm Hg) or shock and accounts for 5% of all cases of PE. Submassive PE is variably defined on the basis of right ventricular (RV) enlargement, dysfunction, ischemia, or strain on echocardiography, computed tomography (CT), or with cardiac biomarkers (troponins or B-type natriuretic peptide, BNP) with no associated hemodynamic instability.

The diagnostic workup for VTE should be tailored to the severity of the clinical presentation and depends on whether the patient is hemodynamically stable or hemodynamically unstable. Clinical judgment and clinical prediction rules are useful in establishing a pretest probability of DVT and PE in which patients are typically classified into low-, moderate-, or high-risk categories. These clinical prediction rules should drive the diagnostic workup and facilitates the interpretation of diagnostic tests.

Acute DVT

Several imaging modalities are available to detect acute DVT, including venous ultrasonography, impedance plethysmography, computed tomography (CT) venography, magnetic resonance imaging (MRI), and contrast venography. Venous ultrasonography of the lower extremities is the preferred noninvasive test for the diagnosis of symptomatic proximal DVT. Failure to demonstrate collapsibility or flow indicates a DVT. Venous ultrasonography has a very high sensitivity and specificity of 95% and 98%, respectively. Additionally, both compression and Doppler ultrasonography can be used to diagnose upper extremity DVT.

Impedance plethysmography uses electrical current to estimate venous outflow obstruction when blood flows out of the leg venous system after release of a thigh pressure cuff. Failure to change impedance is presumptive evidence of a proximal DVT. However, other forms of venous obstruction and congestive heart failure can provide false-positive results.

MRI and CT venography can be highly accurate and have the advantage of evaluating both PE and DVT in a single study. However, MRI is expensive, time consuming, and restricted in patients with metallic devices and the additional irradiation associated with CT venography may not be desirable. Contrast venography is invasive and also requires radiocontrast material and should only be considered if noninvasive testing is nondiagnostic or impossible to obtain.

Pulmonary Embolism

Initial evaluation of a patient with suspected PE with nonspecific symptoms should include examination of the electrocardiogram for tachycardia, T-wave and ST-segment changes, and right- or left-axis deviation, arrhythmias, and less commonly for the classic pattern of S₁Q₃T₃, right ventricular strain, and new, incomplete right bundle branch block. Chest radiography is usually abnormal in more than 80% of patients with PE and helps to rule out and identify other conditions, such as pneumothorax, rib fracture, pneumonia, and pulmonary edema. Classic radiographic signs suggestive of PE include Westermark's sign (focal decrease in vascularity or oligemia distal to the pulmonary artery occlusion) and Hampton's hump (wedge-shaped opacity at the costophrenic angle indicating pulmonary infarction).

CT angiography has essentially replaced ventilation/perfusion (V/Q) scan and pulmonary angiography as the diagnostic imaging modality of choice in patients with suspected PE. In the Prospective Investigation of Pulmonary Embolism Diagnosis II (PIOPED II) trial, CT angiography had sensitivity and specificity rates of 83% and 96%, respectively. Limitations of CT angiography include the risk of adverse reactions to contrast (nephrotoxicity or anaphylaxis) and lack of portability.

Ventilation-perfusion scans are generally reserved for patients with renal failure or allergy to contrast dye. A normal V/Q scan essentially rules out PE. However, V/Q scanning is diagnostic in only 30% to 50% of patients with suspected PE. In PIOPED II with the exclusion of patients with intermediate or low probability scans, the sensitivity of a high probability (PE present) scan finding was 77%, whereas the specificity of very low probability or normal (PE absent) scan finding was 98%. Contrast-enhanced magnetic resonance angiography may also be useful in patients with suspected PE in whom radiographic contrast material or ionizing radiation are relatively contraindicated (eg, renal failure and pregnancy).

Echocardiography (transthoracic or transesophageal) is particularly useful in patients suspected of acute PE who may not be stable for transport for CT angiography or V/Q scanning and may also help to guide management. Classic echocardiography signs of PE on echocardiography include dilatation and hypokinesis of the right ventricle, McConnell's sign (free wall RV hypokinesis that spares the apex), paradoxical motion of the interventricular septum, tricuspid regurgitation, and lack of collapse of the IVC during inspiration. Transesophageal echocardiography can confirm the diagnosis of PE by showing emboli in the main pulmonary arteries. Right ventricular dysfunction defined as RV dilation (apical 4-chamber RV diameter divided by LV diameter > 0.9) or RV systolic dysfunction is associated with increased mortality in patients with acute PE. Additionally, among hemodynamically stable patients with PE, the association between an increased troponin level and RV dysfunction on echocardiography identifies a subgroup of patients at particularly high risk for an adverse outcome.

Pulmonary angiography remains the gold standard for diagnosis of PE. However, it is invasive and associated with several complications including renal failure, respiratory failure and bleeding and thus is now rarely utilized for diagnosis of PE.

D-Dimer

D-dimer is a specific fibrin degradation product that has been widely studied in patients with acute DVT and PE. The rapid enzyme-linked immunosorbent assay is the most commonly used method for measurement of D-dimer. When used in conjunction with a low clinical pretest probability for VTE, D-dimer testing is very sensitive and has a high negative predictive value in excluding the presence of DVT. Several studies have shown that D-dimer testing can decrease the need for further testing such as repeat venous ultrasonography. However, the specificity rates of D-dimer testing are low particularly in patients with cancer, pregnant women, and hospitalized and elderly patients.

Arterial Blood Gas

Arterial blood gases (ABG) have a limited role in diagnosing acute PE. In those with PE associated with hemodynamic instability, the ABG typically shows hypoxemia with an increased alveolar-arterial (A-a) gradient. However, in the PIOPED study, 7% of patients with angiographically documented PE had completely normal ABG measurements. Due to the increase in minute ventilation, patients with PE can demonstrate normal or decreased pCO₂ and respiratory alkalosis.

Cardiac Biomarkers

Elevated levels of serum troponin I or T are noted in 30% to 50% of patients with a moderate or large PE and can indicate RV strain, ischemia, or impending myocardial infarction and are predictive of a poor outcome. One meta-analysis showed that elevated levels of troponin were associated with an increase in the short-term risk of death by a factor of 5.2 (95% confidence interval [CI], 3.3-8.4) and an increase in the risk of death from PE by a factor of 9.4 (95% CI, 4.1-21.5).

BNP (and its precursor, N-terminal pro-BNP) are released in response to increased cardiac filling pressure and are frequently elevated in patients with significant PE. Similar to patients with PE who have elevated troponins, patients with elevated levels of BNP and pro-BNP have an increased risk of an adverse in-hospital outcome as compared with patients with normal levels.

Algorithms for DVT and PE

Validated algorithms are important to evaluate patients with suspected DVT along with objective testing to confirm the diagnosis. Risk factors for DVT should be identified including malignancy, recent surgery, history of prolonged bed rest or immobilization, obesity, lower extremity trauma, pregnancy, and use of oral contraceptives or hormone replacement therapy. The 2012 American College of Chest Physicians (ACCP) consensus guidelines recommend the following goals for nonpregnant patients with a suspected first DVT of the lower extremity: reduce overall false negatives to 2% or less as defined by symptomatic DVT or PE within 3 to 6 months after a negative test; reduce the risk of fatal PE after testing less than 0.1%, and reduce the risk of fatal hemorrhage due to anticoagulation to less than 0.1%. The guidelines endorse using the Wells score for risk stratification of patients for likelihood of DVT into low-, moderate-, and high-risk categories. The Wells criteria for DVT include risk factors, such as active cancer, bedridden recently more than 3 days or major surgery within 4 weeks, calf swelling more than 3 cm compared to the other leg, collateral superficial veins present, entire leg swollen leg, localized tenderness along the deep venous system, pitting edema greater in the symptomatic leg, previously documented DVT and alternative diagnosis to DVT as likely or more likely. Wells score of 4 or less is consistent with a low (5%) pretest probability for DVT; 4.5 to 6 is consistent with a moderate (17%) pretest probability, and more 6 with a high (53%) pretest probability of DVT (Table 20–1).

In patients with a low pretest probability for DVT in the leg, the guidelines recommend checking either a moderately or highly sensitive D-dimer and compression ultrasound of the proximal leg veins rather than whole leg ultrasound. D-dimer testing is preferred over compression ultrasound of the proximal veins as the initial test. If D-dimer is negative, no further testing is necessary; if D-dimer is positive, compression ultrasound of the proximal leg veins should be performed, and if this is positive, then the patient should be treated without further testing.

In patients with a moderate pretest probability for DVT in the leg, the guidelines recommend checking either a highly sensitive D-dimer (unless the patient has a comorbid condition that would likely elevate the D-dimer level, such as cancer or recent surgery or trauma), compression ultrasound of the proximal leg veins, or whole leg ultrasound. If the highly sensitive D-dimer is negative, then no further testing is needed. If the highly sensitive D-dimer is positive, then compression ultrasound of the proximal leg veins or the whole leg is recommended. A negative ultrasound should lead to a repeat compression ultrasound in 1 week and checking a moderate/high sensitivity D-dimer. If the compression ultrasound is negative, no further testing is warranted. If the D-dimer test is negative, no further testing is needed. If whole leg ultrasound is chosen and is negative, no further testing is also recommended. If compression ultrasound of the proximal leg veins or whole leg ultrasound is positive, the patient should be treated without further testing. If whole leg ultrasound is only positive for isolated distal (calf vein) DVT, serial ultrasounds are recommended to ensure the DVT does not propagate proximally rather than treating with anticoagulation. Approximately 15% to 20% of calf vein thrombi can extend into the proximal veins especially within the first 7 days; thus the need for serial ultrasound examinations.

In patients with a high pretest probability for DVT in the leg, the guidelines recommend either compression ultrasound of proximal leg veins or whole leg ultrasound. If either ultrasound examination is positive, the patient should be treated without further testing. If the whole leg ultrasound is negative for DVT, no further testing is recommended. If the compression ultrasound of the proximal leg veins is negative, one can either repeat the ultrasound in 1 week and if negative no further testing is recommended, or check a high sensitivity D-dimer and if negative, no further testing is needed. If D-dimer is positive, repeat compression ultrasound in one week is recommended and if negative, no further testing. D-dimer should not be used as a stand-alone test to rule out DVT in patients with a high pretest probability for DVT.

A diagnostic algorithm (Figure 20–1) with a dichotomized version of the Wells clinical decision rule, D-dimer testing, and CT has been shown to be useful in guiding management decisions in almost 98% of patients with clinically suspected PE.

Anticoagulation

In patients with a high clinical suspicion of acute DVT, treatment with parenteral anticoagulants is recommended while awaiting the results of diagnostic tests. For those with documented DVT and PE, anticoagulation is the mainstay of treatment if no contraindications exist. The recently released 2016 ACCP guidelines suggest the use of direct oral anticoagulants (DOACs) such as dabigatran, rivaroxaban, apixaban, or edoxaban over vitamin K antagonist (VKA) in patients with DVT of the leg or PE and no cancer. For patients with DVT of the leg or PE and no cancer who are not treated with DOACs, the guidelines suggest the use of VKA therapy over low-molecular weight heparin (LMWH). Initial parenteral anticoagulation with unfractionated heparin or LMWH is given before dabigatran and edoxaban, is not given before rivaroxaban and apixaban, and is overlapped with VKA therapy. In patients with DVT of the leg or PE and cancer, LMWH is favored over VKA therapy and DOACs for the first 3 months. Routine use of compression stockings to prevent postthrombotic syndrome is not recommended in patients with acute DVT or PE who are treated with anticoagulants.

With regards to duration of therapy, at least 3 months of anticoagulation is recommended in patients with a proximal DVT of the leg or PE provoked by surgery or by a nonsurgical transient risk factor. After 3 months of treatment, patients with unprovoked DVT of the leg or PE should be evaluated for the risk-benefit ratio of extended therapy. Extended anticoagulant therapy is recommended in patients with a first VTE that is an unprovoked proximal DVT of the leg or PE and who have a low or moderate bleeding risk.

Administration of IV heparin is usually undertaken using the weight-based heparin dosing nomogram where patients receive a loading bolus dose of 80 units/kg followed by an 18 units/kg/h per hour infusion. The heparin dose is adjusted to maintain an activated partial thromboplastin time of 1.5 to 2.3 times control. IV heparin is favored in LMWH have greater bioavailability when given by SC injection, do not require strict laboratory monitoring, and have a lower risk of heparin-induced thrombocytopenia (HIT) compared with UFH. Enoxaparin, dalteparin, and tinzaparin are approved for use in the United States. Enoxaparin is typically given at 1 mg/kg SC twice daily or 1.5 mg/kg SC daily, while tinzaparin is given at 175 units/kg SC once daily. Monitoring antifactor Xa levels typically 4 hours after injection may be considered in patients who are morbidly obese, pregnant patients, and patients with renal insufficiency. Fondaparinux is given at a dose of 5 mg SC once daily for patients weighing less than 50 kg, 7.5 mg SC once daily for patients weighing 50 to 100 kg, and 10 mg SC once daily for patients weighing more than 100 kg. Caution should be observed in patients with renal impairment as both LMWH and fondaparinux are retained in these patients.

UFH is suggested for patients with renal impairment (creatinine clearance < 30 mL/min). Warfarin for the treatment of VTE is not ideal in ICU patients given the many drug and food interactions associated with its use and genetic variations in drug metabolism.

Complications of Anticoagulation

The major complications of anticoagulant therapy are hemorrhage and HIT. The frequency of major bleeding was reported at 1.9% and a fatal hemorrhage rate of 0.2% in a large meta-analysis involving over 15,000 patients treated with either UFH or LMWH. HIT is an immune-mediated adverse reaction to heparin that is associated with thrombocytopenia and can lead to venous and arterial thrombosis. Commonly, there is an otherwise unexplained fall in platelet count (absolute thrombocytopenia or > 50% decrease if the platelet nadir remains in the normal range) 5 to 10 days following exposure to heparin. In patients who develop HIT, the heparin should be immediately discontinued and direct thrombin inhibitors such as argatroban or lepirudin should be administered if anticoagulation continues to be required.

In patients receiving UFH or LMWH who develop clinically significant or devastating hemorrhage, protamine sulfate can be administered although the anticoagulant effect of LMWH is only partially reversed. Allergic reactions and bradycardia are side effects of protamine.

Thrombolytic Therapy

The guidelines recommend the use of systemically administered thrombolytic agents (alteplase, streptokinase, and alteplase) in patients with acute PE associated with hypotension (systolic BP < 90 mm Hg) who do not have a high bleeding risk. Thrombolysis may also be considered in selected patients with acute PE not associated with hypotension and with low bleeding risk whose initial clinical presentation or clinical course suggests a high risk of developing hypotension. In the Pulmonary Embolism Thrombolysis (PEITHO) trial published in 2014, the administration of tenecteplase plus heparin was associated a significant reduction in all-cause mortality or hemodynamic decompensation within 7 days, when compared to placebo plus heparin in patients with intermediate-risk PE (acute PE associated with RV dysfunction on echocardiography and myocardial injury as evidenced by a positive troponin I or T test). However, thrombolytic treatment with tenecteplase was associated with a 2% incidence of hemorrhagic stroke. Short-term infusion times (eg, 2-hour infusion of 100 mg of alteplase) through a peripheral vein are suggested over prolonged infusion times and over a pulmonary artery catheter, respectively. Urokinase and streptokinase are given as a loading dose (streptokinase 250,000 units over 30 minutes) followed by continuous infusion of 100,000 units/hour for 24 hours; urokinase 4400 units/kg over 10 minutes followed by a continuous infusion of 4400 units/kg/h for 12 to 24 hours. Heparin infusion is typically continued after thrombolytic therapy. Contraindications to thrombolysis include surgery in the past 10 days, recent puncture or invasion of noncompressible vessels, recent intracerebral hemorrhage or stroke, uncontrolled hypertension, recent trauma, pregnancy, hemorrhagic retinopathy, other sites of potential bleeding, and infective endocarditis. Thrombolytic therapy should be tailored to the complicated ICU patient and risks and benefits heavily weighed prior to administration.

Inferior Vena Cava Filter

The guidelines recommend the use of an IVC filter in patients with acute PE and contraindications to anticoagulation. No randomized clinical trial (RCT) has evaluated IVC filters as sole therapy in patients with DVT. IVC filters may increase the risk of recurrent DVT and do not reduce the risk of PE or alter mortality. If an IVC filter is indicated in a patient with acute DVT or PE because anticoagulant therapy is temporarily contraindicated (eg, active bleeding), a retrievable filter may be inserted and subsequently removed when it is safe to start anticoagulant therapy. Patients who have an IVC filter inserted should receive a conventional course of anticoagulation (eg, parenteral and long-term anticoagulation) if the contraindication to anticoagulation resolves and should be treated for the same length of time similar to patients who had not had an IVC filter placed. For upper extremity DVT, some institutions can place superior vena cava filters although these are associated with high complication rates than IVC filters; thus, their use should be confined to exceptional circumstances in specialized centers. The risk of recurrent pulmonary emboli after IVC filter placement is 2% to 3%. Complications of IVC filter placement include venous thrombosis at the site of filter insertion, procedural complications, filter malposition and migration, caval occlusion, and sepsis due to device infection. Permanent IVC filters may also predispose to postthrombotic syndrome.

Other Modalities

Surgical pulmonary embolectomy is recommended only in patients with acute PE who have contraindications to thrombolysis, have failed thrombolysis or catheter-assisted embolectomy, or shock that is likely to cause death before thrombolysis can take effect (eg, within hours), provided surgical expertise and resources are available. Mortality from emergency pulmonary embolectomy can be as high as 30% and should be performed only in highly specialized centers.

Other modalities that have investigated include interventional catheterization techniques for massive such as mechanical fragmentation of thrombus with a standard pulmonary artery catheter, clot pulverization with a rotating basket catheter, percutaneous rheolytic thrombectomy, or pigtail rotational catheter embolectomy. Pharmacologic thrombolysis and mechanical interventions are usually combined unless bleeding risk is high.

Due to the significant morbidity and mortality associated with VTE, prevention is pivotal particularly in high-risk patients, such as those admitted to the ICU. Prevention lies in risk stratification and focusing on identifying preventable or predisposing conditions, such as hypercoagulability and immobility. The Padua Risk Assessment Model (RAM) is the best available validated predictor for DVT/PE risk. Patients scored as low risk (RAM < 4) have a 0.3% risk of developing a DVT and do not require prophylaxis, while those scored as high-risk (RAM score ≥ 4) have a 2.2% risk of developing a DVT and require prophylaxis. The guidelines recommend against the use of pharmacologic prophylaxis or mechanical prophylaxis for acutely ill-hospitalized medical patients at low risk of thrombosis. However, the Padua prediction score should not be utilized in critically ill patients who already have an elevated risk for DVT/PE. Patients at moderate or high risk of developing DVT/PE by the Padua prediction score including all critically ill patients who are not bleeding or at high risk for bleeding should receive anticoagulant thromboprophylaxis with either LMWH, UFH, or fondaparinux. Screening for asymptomatic DVT with routine ultrasounds in critically ill patients is not recommended.

Intermittent pneumatic compression devices or graduated compression stockings should be used for patients who are bleeding or at high risk for bleeding. These patients should be switch to anticoagulant prophylaxis as soon as bleeding resolves or once the bleeding risk is considered to be low. Patients considered to be high risk for bleeding and warrant compression devices over anticoagulant prophylaxis include patients with active gastroduodenal ulcer, bleeding in 3 months prior to admission and platelets less than 50,000 mm³.