Venous Thromboembolism Prophylaxis in Hospitalized Patients: A Clinical Practice Guideline From the American College of Physicians FREE

Venous thromboembolism (VTE), comprising pulmonary embolism (PE) and deep venous thrombosis (DVT), is a common clinical problem and is associated with substantial morbidity and mortality (1). Most hospitalized medical patients have at least 1 risk factor for VTE, and this risk persists for several weeks after discharge (1). Twenty-six percent of patients with undiagnosed and untreated PE will have a subsequent fatal embolic event, whereas another 26% will have a nonfatal recurrent embolic event (2). Studies show that between 5% and 10% of all in-hospital deaths are a direct result of PE (3 - 5). The incidence of PE in the United States is estimated to be 1 case per 1000 persons per year, and PE accounts for 200 000 to 300 000 hospitalizations per year (6 - 7).

The purpose of this guideline is to present clinical recommendations on prophylaxis of VTE in adult hospitalized medical patients and patients with acute stroke, based on the available evidence on the benefits and harms of prophylaxis of VTE in these patient populations. The target audience for this guideline is all clinicians, and the target patient population is all hospitalized nonsurgical patients who are at risk for VTE.

The guideline is based on a systematic evidence review that addressed the following questions:

Key question 1: What are the benefits and harms of subcutaneous low-dose heparin products for VTE prophylaxis in hospitalized medical patients?

Key question 2: What is the comparative effectiveness of different low-dose heparin products (low-molecular-weight heparin [LMWH], unfractionated heparin [UFH]) for VTE prophylaxis?

Key question 3: What is the effectiveness and comparative effectiveness of mechanical devices for VTE prophylaxis?

Key question 4: Do results vary by general medical patient populations: general medical inpatients and patients with acute stroke?

The systematic evidence review was conducted by the Minnesota Evidence-based Practice Center (8). The literature search included studies identified by using MEDLINE and the Cochrane Library for clinical trials of VTE prophylaxis. The authors reviewed titles and abstracts of identified references and used reference lists of pertinent randomized trials and systematic reviews to identify additional reports. The studies selected included English-language, randomized trials published between 1950 and April 2011. Included trials evaluated treatments that are commonly recommended and used to prevent VTE, including subcutaneous low-dose (<20 000 U/d) UFH or similar prophylactic doses of LMWH or related agents (such as fondaparinux) and graduated compression stockings or other mechanical measures (such as intermittent pneumatic compression). The primary outcome of interest was total mortality up to 120 days after randomization. Secondary outcomes included symptomatic DVT; all PEs; fatal PE; all bleeding events; and major bleeding events (variably defined by trials, but typically defined as a decrease in hemoglobin level >20 g/L, transfusion of ≥2 units of blood, or life-threatening bleeding at a critical site); and, for mechanical prophylaxis, effects on skin. Details regarding the review methods can be found in the accompanying evidence review (8). To guide our recommendations, we prioritized outcomes on the basis of clinical importance, starting with total mortality. In the absence of statistically significant effects on total mortality, we then weighted effects on all PEs versus effects on major bleeding events, followed by symptomatic DVT, all bleeding (including minor bleeding) events, and effects on skin.

This guideline rates the evidence and recommendations by using the guideline grading system of the American College of Physicians (ACP), which is based on the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) system (Table). Details of the ACP guideline development process can be found in the methods article (9).

Ten trials (10 - 19) (n = 20 717) evaluated medical patients without stroke. The results showed that compared with no heparin prophylaxis, heparin prophylaxis was not associated with a statistically significant reduced risk for mortality (risk ratio [RR], 0.94 [95% CI, 0.84 to 1.04]; I² = 0%; absolute reduction, 4 events per 1000 persons treated [CI, −11 to 3 events]) (moderate-quality evidence). However, heparin prophylaxis was associated with a reduced risk for PE (RR, 0.69 [CI, 0.52 to 0.90]; I² = 0%; absolute reduction, 4 events per 1000 persons treated [CI, −6 to −1 event]) (moderate-quality evidence) but an increased risk for all bleeding events (RR, 1.34 [CI, 1.08 to 1.66]; absolute increase, 9 events per 1000 persons treated [CI, 2 to 18 events]) (moderate-quality evidence). Although the risk for major bleeding events increased, the difference did not reach statistical significance (RR, 1.49 [CI, 0.91 to 2.43]; I² = 16%; absolute increase, 1 event per 1000 persons treated [CI, 0 to 4 events]) (low-quality evidence). Also, heparin prophylaxis resulted in an absolute reduction of 2 fewer symptomatic DVTs per 1000 patients treated (CI, −6 to 4 events), although the difference was not statistically significant (RR, 0.78 [CI, 0.45 to 1.35]) (low-quality evidence).

Evidence from 8 trials (20 - 27) (n = 15 405) of patients with acute stroke showed that compared with no heparin prophylaxis, heparin prophylaxis was not associated with a statistically significant reduction in risk for mortality (RR, 0.91 [CI, 0.70 to 1.18]; I² = 32%; absolute reduction, 9 events per 1000 persons treated [CI, −29 to 18 events]) (low-quality evidence), PE (RR, 0.72 [CI, 0.50 to 1.04]; I² = 20%; absolute reduction, 3 events per 1000 persons treated [CI, −5 to 0 events]) (low-quality evidence), or symptomatic DVT (RR, 0.14 [CI, 0.00 to 7.09]; absolute reduction, 9 events per 1000 persons treated [CI, −10 to 57 events]) (low-quality evidence). Heparin prophylaxis was associated with an increased risk for major bleeding events (RR, 1.66 [CI, 1.20 to 2.28]; I² = 0%; absolute increase, 6 events per 1000 persons treated [CI, 2 to 12 events]) (moderate-quality evidence). The pooled trials were heterogeneous in their patient samples and treatment. The strongest evidence on the benefits and harms of VTE prophylaxis came from a single large randomized, controlled trial of patients with acute ischemic stroke (n = 14 578 [excluding patients randomly assigned to high-dose heparin]) (21). It found no statistically significant difference between low-dose heparin and no heparin in 14-day all-cause mortality (8.7% vs. 9.3%; RR, 0.94 [CI, 0.84 to 1.05]), fatal PE (0.51% vs. 0.40%; odds ratio [OR], 1.30 [CI, 0.77 to 2.18]), or all (fatal and nonfatal) PEs (0.68% vs. 0.83%; OR, 0.82 [CI, 0.55 to 1.21]). The study showed a statistically significant increase in 14-day hemorrhagic stroke or serious extracranial hemorrhage (1.3% vs. 0.80%; OR, 1.73 [CI, 1.22 to 2.46]) and a statistically significant decrease in 14-day recurrent ischemic stroke (2.6% vs. 4.0%; RR, 0.65 [CI, 0.54 to 0.80]).

For mortality, PE, and major bleeding events, pooled estimates for medical patients without stroke and patients with acute stroke were very similar. Although the point estimate for the effects on DVT risk was substantially stronger in patients with stroke than in medical patients without stroke, it was too imprecise to draw conclusions about differential risks. Combining evidence from the 2 populations (18 trials; 36 122 participants) showed that heparin was associated with a borderline statistically significant reduction in risk for mortality compared with no heparin prophylaxis (RR, 0.93 [CI, 0.86 to 1.00]; I² = 2%; absolute reduction, 6 events per 1000 persons treated [CI, −11 to 0 events]), a statistically significant reduction in risk for PE (RR, 0.70 [CI, 0.56 to 0.87]; I² = 0%; absolute reduction, 3 events per 1000 persons treated [CI, 5 to −1 events]), and no statistically significant decrease in symptomatic DVT (RR, 0.75 [CI, 0.43 to 1.30]; absolute reduction, 2 events per 1000 persons treated [CI, −6 to 3 events]). These trials also showed a statistically significant increased risk for all bleeding events (RR, 1.28 [CI, 1.05 to 1.56]; absolute increase, 9 events per 1000 persons treated [CI, 2 to 18 events]) and major bleeding events (RR, 1.61 [CI, 1.23 to 2.10]; I² = 0%; absolute increase, 4 events per 1000 persons treated [CI, 1 to 7 events]).

Nine trials (28 - 36) (n = 11 650) that compared LMWH with UFH in medical patients showed no statistically significant difference in mortality (RR, 0.91 [CI, 0.73 to 1.13]; I² = 25%; absolute reduction, 9 events per 1000 persons treated [CI, −28 to 13 events]) (moderate-quality evidence), PE (RR, 0.70 [CI, 0.44 to 1.11]; I² = 0%; absolute reduction, 2 events per 1000 persons treated ([CI, −4 to 1 event]) (low-quality evidence), or major bleeding events (RR, 0.89 [CI, 0.70 to 1.15]; I² = 0%; absolute reduction, 3 events per 1000 persons treated [CI, −7 to 3 events]) (low-quality evidence).

Five trials (37 - 41) (n = 2785) that compared LMWH with UFH in patients with acute stroke did not show statistically significant differences for mortality (RR, 1.00 [CI, 0.81 to 1.22]; I² = 1%; absolute reduction, 0 events per 1000 persons treated [CI, −23 to 26 events]) (moderate-quality evidence), symptomatic DVT (RR, 0.34 [CI, 0.11 to 1.00]; absolute reduction, 7 events per 1000 persons treated [CI, −9 to 0 events]) (low-quality evidence), PE (RR, 0.57 [CI, 0.25 to 1.34]; I² = 18%; absolute reduction, 4 events per 1000 persons treated [CI, −8 to −3 events]) (low-quality evidence), or major bleeding events (RR, 1.49 [CI, 0.73 to 3.06]; I² = 0%; absolute reduction, 4 events per 1000 persons treated [CI, −2 to 19 events]) (low-quality evidence).

Evidence from all trials (n = 14 435) comparing LMWH with UFH did not show a statistically significant difference between LMWH and UFH for mortality (RR, 0.94 [CI, 0.82 to 1.08]; I² = 16%) or major bleeding events (RR, 0.95 [CI, 0.75 to 1.20]; I² = 0%), although a nonsignificant difference favored LMWH for PE (OR, 0.67 [CI, 0.45 to 1.00]; P = 0.053; I² = 0%).

Evidence on clinical outcomes from randomized, controlled trials evaluating mechanical devices versus no mechanical devices is sparse. Three trials included in the systematic review compared mechanical devices with no mechanical devices, but 1 large trial (n = 2518) included 232 of the 247 deaths reported. It included patients with acute stroke and compared thigh-length graduated compression stockings with no stockings (42). The results showed no statistically significant difference in risk for mortality (RR, 1.11 [CI, 0.87 to 1.42]) (low-quality evidence), symptomatic DVT, or PE. However, risk for lower-extremity skin damage statistically significantly increased among patients treated with compression stockings (RR, 4.02 [CI, 2.34 to 6.91]; absolute increase, 39 events per 1000 patients treated [CI, 17 to 77 events]) (moderate-quality evidence). The 2 other studies showed no differences in rates of PE or mortality (43 - 44).

Four studies met inclusion criteria but could not be meaningfully combined with the studies described because they evaluated different clinical comparisons or interventions. One randomized, controlled trial (n = 300) that evaluated LMWH prescribed for different durations in medical patients (45) reported 2 deaths in patients who received LMWH only while immobilized (mean, 5.1 days) compared with no deaths in patients who received LMWH while immobilized and for 10 additional days (mean, 14.5 days) (RR, 0.00). A randomized trial (n = 90) of patients who were enrolled at least 7 days after stroke compared LMWH with intermittent pneumatic compression (46). Most patients received heparin prophylaxis before randomization. The study reported 2 symptomatic DVTs and 1 minor bleeding event in the LMWH group compared with no events in the compression group. Another study (n = 151) that included patients with acute cerebral hemorrhage compared compression stockings with or without pneumatic compression and found no statistically significant difference in risk for all-cause mortality through 3 months (22% vs. 31%; RR, 0.69 [CI, 0.4 to 1.20]) (47). One symptomatic DVT occurred in each group (RR, 1.04 [CI, 0.07 to 16]) (47). Another study found that proximal DVT occurred more in patients with stroke who wore below-knee stockings than in those who wore thigh-length stockings (48). One study (n = 6085) of hospitalized, immobile medical patients who had completed an initial 10-day course of open-label enoxaparin prophylaxis compared an additional 28 days of enoxaparin (40 mg) with placebo (49). After 90 days, the treatment group had a statistically significant reduction in risk for symptomatic VTE (from 28 [1.1%] to 8 [0.3%] events) and an increase in all bleeding events (from 116 [3.9%] to 186 [6.3%] events) and major bleeding events (from 10 [0.3%] to 25 [0.8%] events), with no difference in mortality risk (hazard ratio, 1.04).

Randomized trials of heparin versus no heparin therapy for medical patients did not show a statistically significant reduction in the risk for mortality or symptomatic DVT and showed an increased risk for bleeding events. However, PE was significantly reduced in medical patients. For patients with acute stroke, heparin increased the risk for major bleeding, with no effect on mortality, symptomatic DVT, or PE. Studies comparing LMWH with UFH did not show any differences in clinical outcomes. Mechanical prophylaxis was not associated with any improvement in clinical outcomes in patients with acute stroke and resulted in an increased risk for lower-extremity skin damage, although evidence on the effects of mechanical prophylaxis was sparse.

Recommendation 1: ACP recommends assessment of the risk for thromboembolism and bleeding in medical (including stroke) patients prior to initiation of prophylaxis of venous thromboembolism (Grade: strong recommendation, moderate-quality evidence).

The decision to initiate VTE prophylaxis in medical (including stroke) patients should be based on an individualized assessment of the risk for thromboembolism and bleeding, as well as an assessment of the potential harms against modest or even no benefit. Trials that evaluated the benefits and harms of heparin prophylaxis generally enrolled patients who were considered to be at higher risk for VTE. Risk factors for thromboembolism include presence of inherited conditions—such as factor V Leiden mutation, prothrombin gene mutation, protein S or C deficiency, and antithrombin deficiency—or acquired risk factors—such as surgery, cancer, immobilization, trauma, presence of a central venous catheter, pregnancy, drugs (for example, oral contraceptives, hormone replacement therapy, or tamoxifen), congestive heart failure, chronic renal disease, the antiphospholipid antibody syndrome, obesity, smoking, older age, and history of thromboembolism. Some evidence suggests that heparin is less beneficial in younger patients than in patients older than 75 years (5,49 - 50). Many risk assessment tools are available for estimating thromboembolism risk, but the current evidence is insufficient to recommend a validated tool. Although such instruments may be useful, decisions about heparin prophylaxis may also be based on general evidence regarding the risk factors for VTE and bleeding.

Heparin and related drugs are associated with an increased risk for bleeding. Risk factors for bleeding with anticoagulant therapy include older age; female sex; diabetes; hypertension; presence of cancer; acute or chronic alcoholism; liver disease; severe chronic kidney disease; peptic ulcer disease; anemia; poor treatment adherence; prior stroke or intracerebral hemorrhage; presence of bleeding lesions; bleeding disorder; and concomitant use of aspirin, nonsteroidal anti-inflammatory drugs, antiplatelet agents, antibiotics, statins, fibrates, and steroids.

Recommendation 2: ACP recommends pharmacologic prophylaxis with heparin or a related drug for venous thromboembolism in medical (including stroke) patients unless the assessed risk for bleeding outweighs the likely benefits (Grade: strong recommendation, moderate-quality evidence).

In hospitalized medical patients, prophylaxis with heparin is associated with a statistically significant reduction in PEs (absolute decrease, 4 events per 1000 persons treated) and increase in all bleeding events (absolute increase, 9 events per 1000 persons treated), a non–statistically significant increase in major bleeding events (absolute increase, 1 event per 1000 persons treated), and no effect on mortality or symptomatic DVT. In most patients, the clinical benefit of reduction of PEs outweighs the harm of increased risk for bleeding events.

In patients with acute stroke, the pooled results from the evidence review showed no statistically significant benefit from heparin prophylaxis on mortality, PE, or symptomatic DVT. The pooled results also showed a statistically significant increase in risk for major bleeding events (absolute increase, 6 events per 1000 persons treated) that outweighed the potential reduction in PEs (absolute decrease, 3 events per 1000 persons treated). However, the pooled results showed wide CIs that also encompassed potential substantial net benefits. Seven of 8 studies that evaluated the effect of heparin on mortality were small (sample size range, 32 to 305 participants) and were published before 1996. Some did not describe use of CT to exclude intracranial hemorrhage before randomization. The strongest evidence in patients with stroke comes from the International Stroke Trial, a large study that randomly assigned 14 578 patients with suspected acute ischemic stroke to receive low-dose (5000 IU twice daily) heparin or no heparin (21). It found no statistically significant differences between low-dose heparin and no heparin in 14-day all-cause mortality, fatal PE, or all (fatal and nonfatal) PEs. Although the risk for hemorrhagic stroke or serious extracranial hemorrhage statistically significantly increased (absolute increase, 5 events per 1000 persons treated), this was offset by a statistically significant and larger decrease in risk for recurrent ischemic stroke (absolute decrease, 14 events per 1000 persons treated). Results of the International Stroke Trial and pooled estimates from patients with stroke were generally consistent with findings from pooled analyses of medical patients without stroke; thus, evidence was insufficient to conclude that risks and benefits of VTE prophylaxis differ between medical patients with stroke and those without stroke. Evidence on the risks and benefits in patients with stroke is relatively weaker than that in medical patients without stroke, although prevention of recurrent ischemic stroke may be an additional benefit in this population.

The optimal duration of heparin prophylaxis is uncertain. Almost all trials evaluated heparin therapy for patients during hospitalization. A recent study evaluated extended (posthospitalization) heparin therapy for high-risk (immobile) patients (49), but more research is needed to understand the effects of extended therapy on the balance of benefits and harms.

Clinical benefits and harms do not statistically significantly differ between LMWH and UFH. Fondaparinux has not been directly compared with heparin. All prophylactic heparins reviewed for this guideline are administered as subcutaneous injections. The dosage varies from 2 or 3 times daily for UFH to once daily for LMWH or fondaparinux. The average wholesale drug costs are about $10 per day for UFH, $35 per day for LMWH (generic enoxaparin is available), and $60 per day for fondaparinux. In 4 trials that compared UFH with LMWH and assessed heparin-induced thrombocytopenia, 7 cases of heparin-induced thrombocytopenia occurred out of about 1900 in patients receiving UFH and 1 case occurred out of about 1900 patients receiving LMWH (P = 0.07) (29,31,33,37). Hence, the choice of agent for prophylaxis of VTE should be based on ease of use, adverse effect profile, and cost of medication.

Recommendation 3: ACP recommends against the use of mechanical prophylaxis with graduated compression stockings for prevention of venous thromboembolism (Grade: strong recommendation, moderate-quality evidence).

Mechanical prophylaxis with graduated compression stockings was not effective in preventing VTE or reducing mortality and resulted in clinically important lower-extremity skin damage. Clinicians who initiate VTE prophylaxis should select heparin (or related drugs) rather than graduated compression stockings for patients in whom heparin can be used. In patients at high risk for bleeding events or in whom heparin is contraindicated for other reasons, intermittent pneumatic compression may be a reasonable option, because evidence suggests that it is beneficial in surgical patients. However, intermittent pneumatic compression has not been sufficiently evaluated as a standalone intervention in medical patients to reliably estimate benefits and harms.

See the Figure for a summary of the recommendations and clinical considerations.

ACP does not support the application of performance measures in medical (including stroke) patients that promotes universal venous thromboembolism prophylaxis regardless of risk.

In the United States, many organizations have developed performance measures intended to increase the appropriate use of VTE prophylaxis in hospitalized patients. However, in some clinical settings, performance measures have been based on rates of VTE prophylaxis in all patients, regardless of their underlying risk. The evidence reviewed for the clinical recommendations in this guideline does not support routine prophylaxis of VTE in all medical patients and emphasizes the tradeoff in harms and benefits. Clinicians caring for these patients must assess the risks and benefits before deciding whether to initiate prophylaxis. In some cases, not prescribing VTE prophylaxis may be justified because the estimated tradeoff between potential risks and benefits is small or unclear. Because no standard, accepted formula for risk assessment exists to identify which medical patients are likely to benefit from VTE prophylaxis, the decision is best left to physician judgment, and performance measures targeting all patients are inappropriate. Until we can better identify patients who truly benefit, performance measures that encourage VTE prophylaxis for all medical patients may encourage physicians to use prophylaxis in low-risk patients for whom the risks may exceed the benefit.