Anticoagulation Therapy for Venous Thromboembolism

Abstract

Context

On the basis of theoretical rationale, heparoids and vitamin K antagonists are prescribed to prevent complications of venous thromboembolism (VTE, including pulmonary emboli [PE] and deep vein thrombosis [DVT]). They have been employed as the standard of care for treatment of VTE for over 40 years.

Objective

Critique the evidence supporting the efficacy of anticoagulants for the treatment of VTE in reducing morbidity and/or mortality.

Data Sources

This includes a search of reference lists and Medline.

Study Selection

This includes studies concerning the diagnosis and incidence of PE and DVT, efficacy of anticoagulants in preventing complications, risks of anticoagulant therapy, and the costs of diagnosis and the treatment of VTE.

Data Extraction

I analyzed references cited in reviews and meta-analyses of VTE, and from Medline searches concerning diagnosis and treatment. The data quality and validity of studies depended on the consistency of findings and statistical significance of the data.

Data Synthesis

No placebo-controlled trials of anticoagulants as treatment of PE with objective criteria for diagnosis have been published. Three randomized trials of anticoagulants vs no anticoagulants in DVT showed no benefit with heparin and vitamin K antagonists (combined all-cause mortality: anticoagulants = 6/66, un-anticoagulated controls = 1/60, P = .07). No placebo-controlled trials of low-molecular-weight heparins or thrombolytic drugs have been done; therefore, their efficacy in VTE depends entirely on randomized comparisons with unfractionated heparin. They have not been proven safer or more efficacious than unfractionated heparin. Thrombolysis causes more major and fatal bleeds than heparin and is no more effective in preventing PE. Diagnosing and treating VTE patients in the United States with anticoagulants costs $3.2 to $15.5 billion per year (1992 dollars). Bleeding and complications of angiography cause 1017-3525 deaths annually.

Conclusion

Anticoagulants have not been proven efficacious or safe in VTE. The bleeding risks and other complications of anticoagulation are unacceptably high. The use of anticoagulants for patients with VTE should be reconsidered.

Introduction

Risk factors for VTE include age > 65 years, prolonged bed rest, paralysis, congestive heart failure, myocardial infarction, cancer, stroke, trauma, and surgical procedures.[1-3] Anticoagulant therapy for VTE became established as the standard of care in the 1940s and 1950s before randomized trials were considered necessary to prove efficacy and safety. The risks and costs of anticoagulant therapy are significant; therefore, an overall assessment of the practice with evidence-based medicine principles is in order.

Diagnosis of Deep Vein Thrombosis

Venous thromboembolism (VTE) researchers regard venography as the "gold standard" test for lower extremity deep vein thrombosis (DVT).[4,5] However, venography is expensive, uncomfortable for the patient, difficult to repeat frequently, and itself causes DVT in 2% to 3% of examinations.[4,5] Of those suspected with DVT, about 60% to 75% do not have it diagnosed by venography.[6-11] No relationship between the duration of symptoms and the extent or the occlusiveness of venous thrombi has been demonstrated.[11]

Noninvasive tests (impedance plethysmography compression ultrasound with venous imaging, duplex ultrasonography, and color-flow Doppler ultrasonography) have largely replaced venography for diagnostic purposes in nonresearch settings.[4,12,13]

Incidence of DVT

Published figures of the annual incidence of diagnosed episodes of DVT per 100,000 people are 48,[1] 76,[14] 160,[15] 70, 160, and 206[16] (ie, 130,000-550,000 per year in the United States). The incidence of symptomatic DVT increases exponentially with age.[1]

Many medical and surgical conditions are associated with higher risks of VTE. The following incidences of asymptomatic and symptomatic DVT in general medical patients and after various surgical procedures have been estimated by Hyers and colleagues[17]:

• Hip fracture, 40% to 70%;

• Total hip replacement, 40% to 70%;

• Total knee replacement, 40% to 70%;

• Urologic surgery, 15% to 20%;

• General and gynecologic surgery, 15% to 20%;

• Neurosurgery, 15% to 20%; and

• Medical patients, < 15%.

The incidence of DVT in pregnancy is unknown but suspected to be higher than in nonpregnant women.[18]

Havig[19] found DVTs in 161 of 261 (62%) unselected autopsies. Seventy-five percent of these had clear evidence of pulmonary emboli (PE) that had been clinically unsuspected. In an autopsy series of patients with PE, Lindblad and colleagues[20] reported that 87% (1448 of 1665) had DVT. In this series, the majority of patients with fatal PE had no clinical symptoms of a DVT before the onset of catastrophic embolism or any obvious heralding symptoms of PE. In another article by the same group in the same hospital, the study authors found that only 10 of 282 patients (3.5%) with symptomatic DVT confirmed by venography had clinical signs of PE, verified by lung scan in 4.[21] This apparent paradox provoked the question: "Could it be that the more symptomatic a DVT is, the less is the risk of PE? One possibility is that the phlebitic reaction giving rise to local symptoms also anchors the thrombus at the vessel wall.[20]"

Diagnosis of PE

The 1967 postmortem study of Linder and colleagues[22] illustrates the challenge of preventing deaths from PE by timely and accurate diagnosis. In this study before the widespread use of pulmonary angiograms and lung scans, the time interval between the onset of acute massive PE and death was less than 10 minutes in 76.2% of the 343 patients, about whom exact data were available from the charts.[22] Only 46 of 377 patients (12.2%) had clinical signs of DVT before the onset of PE and in only 43 (11.4%) did clinicians note symptoms of premonitory PE before the fatal episode.[22]

In people with adequate cardiovascular reserve, natural fibrinolytic mechanisms clear thrombi that migrate to the pulmonary artery within 2 weeks, with or without anticoagulants.[23]

With the advent of objective testing for PE over the past 30 years, we now know that about 75% of patients suspected of PE on clinical grounds do not have the diagnosis.[5,24,25] This means that conclusions from earlier studies using only clinical diagnoses of PE need to be reconsidered.

The prevailing view of the optimal PE diagnostic strategy comes from Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED), the largest study of the topic.[24, 26-28] Investigators of the PIOPED study concluded that pulmonary angiography is the gold standard test for PE. However, it is expensive and complications are significant. Ventilation perfusion scanning, if normal, rules out PE. A "high-probability" scan correlates with a positive pulmonary angiogram in 88% of cases and is considered sufficient evidence to treat with anticoagulants.[24] However, most scans are intermediate or low in probability for PE, requiring a confirmatory, noninvasive leg test (impedance plethysmography, compression ultrasound with venous imaging, duplex ultrasonography, and color-flow Doppler ultrasonography) or positive pulmonary angiography to warrant anticoagulant treatment.[24]

Despite the advances in diagnostic accuracy of PE, the correlation of antemortem diagnosis with postmortem confirmation remains poor. In 1969 before the widespread use of pulmonary angiography and lung scans, the range of antemortem diagnoses of PE being confirmed at autopsy was from 7% to 70%.[22,29,30] More recent published figures show rates of 7%,[31] 10%,[32] 15%,[33] 30%,[31] and 32%.[34] Karwinski and Svendsen[32] noted a significant drop in correct antemortem diagnoses of PE at the Gade Institute in Norway from 20% during 1960-1969 to 10% of the cases of PE seen at necropsy in 1980-1984.

Because of the dramatic fall in numbers of postmortem examinations in recent years, further clarification of the mortality of PE from autopsies is unlikely.

Incidence of PE

Studies show striking differences in incidence of PE depending on whether the reports concern symptomatic or asymptomatic patients, or postmortem findings. The difference between asymptomatic, symptomatic but nonlethal, and fatal PE depends primarily on the general health of the patient and cardiovascular reserve rather than the size of the emboli.[19,35,36] In normal individuals, the sudden occlusion of up to 50% of the pulmonary vasculature by a balloon catheter causes few if any symptoms or signs; therefore, even massive PE may be asymptomatic.[37] Conversely, small emboli in people with limited cardiovascular reserve may be lethal.

Pooling 3 studies looking at the incidence of abnormal perfusion scans in healthy volunteers reveals that 2.9% of healthy adults have pulmonary perfusion defects on lung scan with an intermediate or a high probability for PE.[35, 38-40] Even if only 40% of these lung scan abnormalities were confirmed by angiography, as is the case with intermediate probability scans, this indicates that over 3 million asymptomatic, healthy Americans have silent PE at the present moment. PE resolves angiographically within 2 weeks with or without anticoagulants,[23] so the yearly incidence in Americans with silent PE may be in the tens of millions.

The literature reports the incidence of clinically diagnosed symptomatic cases of PE per 100,000 people to be 10,[16] 15,[16] and 23[1] (ie, 27,000-57,000 per year in the United States). A study of fatal PE in previously normal healthy Americans aged 15-45 years showed an incidence of .27 cases per 100,000 per year[41] or about 240 per year in the United States.

Incidence of PE in High-Risk Groups

Postoperative orthopaedic patients screened for PE with ventilation-perfusion lung scans showed an incidence of 4%,[42] 6%,[43] 8%,[44] 14%,[45] 17%,[46] 20%,[47] and 31%.[48] With intraoperative transesophageal echocardiography to detect PE, Parmet and colleagues[49] found 29 of 29 total knee arthroplasty patients had showers of PE once the tourniquets on their legs were released. This study suggests that essentially everyone with major operations on the lower extremities have PE whether or not they can be detected by lung scanning.

Of people with symptomatic DVT, about 40% to 50% will have lung scan evidence of PE, mostly asymptomatic.[50-53]

The incidence of PE in healthy pregnant women is not published because of the potential risks of lung scans and angiograms to fetuses. From 1974 to 1979, 271 fatalities in pregnant or postpartum American women were due to PE. This gives an approximate incidence of maternal mortality due to PE of 1 in 74,000.[54]

Incidence of PE on Postmortem Examinations

Estimating the mortality of PE from death certificates has been shown to be grossly inaccurate with a sensitivity of less than 40% and the predictive value of a positive diagnosis of less than 50%.[55] This may be partly because studies in the literature report postmortem exams inconsistently (eg, classifications, such as "incidental," "contributing to death," or "fatal" PE). In addition, of those dying from PE, most have underlying terminal illnesses leading to prolonged bed rest or inactivity, venous stasis, and thrombosis associated with reduced cardiovascular reserve. Consequently, PE often do not significantly shorten the life expectancy in these fatal cases. Of those with postmortem diagnoses of PE, the proportion with underlying terminal illnesses has been reported as 95% (169/178)[33] and 96.5% (1867/1934).[32]

With these caveats regarding the significance of any data on the percentage of autopsies reporting PE as the cause of death, the published figures from postmortem series are 4.5%,[56] 6%,[30] 6%,[34] 9%,[32] 9% (medical patients only),[33] 9.4%,[20] and 15.7%[19] (ie, between 100,000 and 350,000 PE deaths in the United States per year). The most commonly quoted figures for annual mortality from PE in the United States are inexplicably below this range at 50,000[2] and 60,000,[3] perhaps because of the exclusion of patients with underlying terminal illnesses.

To estimate the actual mortality of PE in nonterminally ill people, we need to multiply the range of postmortem incidence of PE (ie, 100,000-350,000[19,20,30,32-34,56]) times the percentage of nonterminally ill people with postmortem diagnoses of fatal PE (ie, 3.5% to 5%[32,33]). This yields a range of clinically relevant PE deaths per year of only 3500-17,500 in the USA.

Efficacy of Anticoagulants in DVT

The only randomized, placebo-controlled trial used to justify heparin and vitamin K antagonists (eg, warfarin, phenprocoumon, and acenocoumarol) as treatment of DVT was reported in 1960 by Barritt and Jordan.[58] Ironically, the study population consisted of patients clinically diagnosed with PE, and the subset with DVT was not reported. This study noted that patients who had survived symptomatic PE and then received anticoagulants had a significantly lower mortality from PE (ie, 0/16 with anticoagulants vs 5/19 with placebo, P < .0007).[58] Anticoagulated patients received 3 days of intravenous heparin concurrently with oral nicoumalone (Sinthrome), a vitamin K antagonist, for a total of 14 days of anticoagulation. The clinical PE diagnoses of the investigators were not confirmed by pulmonary angiograms or lung scans.[58]

We now know that about 75% of those clinically diagnosed with PE do not have it.[5,24,25] Autopsy descriptions of the patients in Barritt and Jordan's[58] study show that in 4 of the 5 deaths, severe underlying diseases (eg, cerebral infarction and cavitary pneumonia with sepsis) caused the deaths, with PE only appearing as a contributing factor. This is consistent with reports that about 95% of people with fatal PE have severe, chronic underlying illness.[32,33] Egermayer[59] delineated a list of other problems with the trial:

• Doctors other than the investigators referred the patients into the trial, so the selection was not random;

• It was not double-blind;

• The investigators provided no information about the comparability of the anticoagulated and un-anticoagulated patients;

• Regarding nonfatal recurrences, no information was given other than 5 that occurred in the nontreated group; and

• Barritt and Jordan[58] emphatically stated that anticoagulation dissolves the existing clot, whereas researchers now claim that they only prophylax against further emboli.

Hunter and Walker[60] criticized Barritt and Jordan[58] for conducting the trial because the benefits of anticoagulation had already been shown by Zilliacus[61] in 1946.

Published, randomized trials of DVT patients, including un-anticoagulated controls, include:

• An abstract-only report by Kakkar and colleagues[62] compared heparin, Malayan pit viper venom (Arvin), streptokinase, and placebo, resulting in 2 of 7 deaths in the heparin group and 0 of 6 in the placebo group.

• Ott and colleagues[63] published a placebo-controlled trial in which 2 of 11 patients died receiving heparin and warfarin, and 1 of the 12 placebo-treated patients died.

• Nielsen and colleagues[64,65] randomized 90 ambulatory patients with DVT into standard heparin and phenprocoumon vs phenylbutazone (ie, no anticoagulants). Two of 48 patients in the anticoagulated group died (one of PE), whereas 0 of 42 in the un-anticoagulated group died. About 50% of both groups had PE by lung ventilation-perfusion scanning, mostly asymptomatic.

Efficacy of Anticoagulants in PE

Without convincing evidence, a consensus has formed that diagnosis of more people with symptomatic and asymptomatic PE and treatment with anticoagulants may reduce deaths related to PE. We need to carefully look at the data underlying this assumption.

To assess the efficacy of anticoagulants in patients with previously diagnosed PE, we must know the risk of dying of subsequent PE without anticoagulants. The risk of dying of PE in patients who have survived documented PE and remain un-anticoagulated is poorly documented in the literature. Several investigators cite the mortality of untreated PE up to 30%, referencing the Barritt and Jordan[58] trial and uncontrolled studies mostly published before 1960.[61,66-68] Because all of these studies predated lung scans and pulmonary angiography, the PE diagnoses were clinical, making any conclusions very questionable. Another flaw of the older literature is that virtually all deaths were attributed to PE. We now understand that a large number of those deaths were due to an underlying terminal disease complicated by PE.

The trial of Barritt and Jordan,[58] the only randomized, placebo-controlled trial of anticoagulants in people surviving an episode of symptomatic PE, was discussed previously in reference to DVT. One subsequent study inadvertently provided data on patients who were documented to have PE on pulmonary angiography or lung scan and then not anticoagulated.

In the PIOPED study, 23 of 399 patients diagnosed with PE did not receive anticoagulant treatment in the hospital because the preliminary diagnosis was "no PE" by the local radiographer.[69] Four of the patients received anticoagulant treatment during the first 3 months. However, 19 patients received no anticoagulation. Of the 19 patients, 18 had the pulmonary angiogram interpretation changed from no PE to PE by the central panel of angiogram readers. One who had a positive lung scan and no angiogram died of underlying disease 4 days later. Thus, 0 of 19 patients with a confirmed diagnosis of PE but who received no anticoagulant treatment died of PE.

Not all of the older, uncontrolled studies of PE supported anticoagulation. Johnson and Charnley[70] retrospectively studied 7959 patients undergoing total hip replacement from 1962 to 1973 in Wrightington, England. The overall mortality from PE was 1.04% (80/7959). Using clinical evaluations only, they diagnosed nonfatal PE in 7.89% (628/7959) of whom 603 were available for study (see (Table 1)). The only death in the 603 patients with PE was in a patient anticoagulated with heparin. They described their major concern with heparin and warfarin: ". . .to achieve full anticoagulant levels quickly brings risks of wound hematoma during the second post operative week. The oozing of serum from an incompletely healed wound invites infection and the results of wound infection in total hip replacement are so distressing that the problem is made worse if the treatment was not absolutely necessary.[70]" This retrospective observational study suggests that heparin increases VTE recurrences in PE patients, and that warfarin does not improve outcomes as compared with no anticoagulation.

Table 1.

Johnson and Charnley's[70] Retrospective Study of Clinically Diagnosed PE Patients

Treatment	Number	Recurrent Emboli	Later DVT	Impaired Wound Healing
Anticoagulant	295	38 (13.0%)	5 (1.7%)	18 (6%)
No anticoagulant	308	10 (3.2%)	3 (1.0%)	7 (2.3%)

PE = pulmonary emboli; DVT = deep vein thrombosis

Risk of Bleeding With Heparin

An analysis of clinical studies with heparin shows that bleeding risk is related to dose with a correlation coefficient of .56.[71] Data concerning the relationship between the efficacy of heparin in preventing recurrent VTE and its anticoagulant effect, as measured by the activated partial thromboplastin time, are conflicting.[72]

Patient factors that increase the risk of bleeding include recent surgery or trauma,[73] advanced age,[74] alcoholism,[75] aspirin use,[75] bed rest, immobility, obesity, congestive heart failure, cancer, hypercoagulable states (thrombophilia), childbirth, and pharmacologic doses of estrogens.[76]

The rate of major bleeding associated with anticoagulant use varies from 1% to 33%, largely because study criteria for defining the severity of bleeding vary significantly.[77] Independently observed studies demonstrate that the criteria for minor bleeding are also not reproducible because of observer variability.[78] According to recently published studies, the rate of major hemorrhage with unfractionated heparin ranges from 0% to 7%.[79] Fatal bleeding with a 5-14-day course of heparin ranges from 0% to 2%.[79] A meta-analysis of published studies shows that the average daily frequencies of fatal and major bleeding during heparin therapy were 0.05% and 0.8%, respectively.[80] Risk of bleeding from heparin in clinical settings with physicians who are not as experienced in anticoagulation may be higher.

With about 150,000-600,000 diagnosed cases of DVT and PE per year in the United States,[14-16, 81] between 450 and 1800 Americans treated for VTE die of heparin-related bleeding -- mostly intracranial -- each year. The incidence of nonfatal intracranial bleeding is not routinely separated from other major bleeds but is probably more than the number of fatal intracranial bleeds. This fatality rate due to heparin does not include deaths due to thrombosis from heparin-induced thrombocytopenia (HIT). The incidence of serologically proven HIT varies from 3.2% to 50% depending on the type of serologic assay, the patient population, and the heparin preparation.[82] HIT-related morbidity and mortality rates in large populations of VTE patients have not been reported.

Low-molecular-weight heparins have not been proven to cause less major or fatal bleeding than unfractionated heparin.[79,83-86]

Risk of Bleeding With Vitamin K Antagonists

In early studies, the risk of fatal bleeding with a course of vitamin K antagonists ranged from .2% to 2%.[87] The higher the ProTime INR, the higher is the risk of bleeding.[88] Comorbid conditions that raise the risk of bleeding include age greater than 65 years, a history of stroke, a history of gastrointestinal bleeding, severe anemia, renal insufficiency, recent myocardial infarction, aspirin or nonsteroidal anti-inflammatory medication use, and atrial fibrillation.[89]

A meta-analysis of prospective research studies by Levine and colleagues[90] showed that warfarin caused major bleeds in .8% to 4.1% patients per year (average, 1.7%) and fatal bleeding in .2% to 2.3% patients per year (average, 0.8%). In another meta-analysis by Landefeld and Beyth,[80] the average annual frequencies of major and fatal bleeding per year with warfarin therapy were 3.0% and .6%, respectively. During warfarin therapy, the risk for major bleeding during the first month of therapy is approximately 10 times the risk per month after the first year of therapy.[80] In nonresearch outpatient settings, the risk of fatal bleeding with vitamin K antagonists may be higher (eg, 2%/year,[91] 1.2%/year,[80] and 1%/year[92]).

With a conservative estimate, 450-1800 (0.3% per course x 150,000-600,000 patients treated for VTE) Americans die every year of bleeding from warfarin in VTE treatment.

Thrombolysis for VTE

Intracranial bleeding risk with thrombolysis for VTE is 2-4 times the risk with heparin.[93,94] Specifically, in an analysis of 5 published streptokinase studies, intracranial hemorrhage occurred in 6 of 312 PE patients, or 1.9% (95% confidence interval, .7% to 4.1%). Two of the 6 intracranial hemorrhages were fatal.[95] Thrombolytics do not reduce the incidence of postphlebitic syndrome,[96,97] and streptokinase gave no better protection against PE than heparin.[98]

In one study, only 7% (15 of 209) of VTE patients did not have a contraindication to thrombolytic therapy.[99] Thrombolytic therapy for VTE is not evidence-based.

Cost of Diagnosis and Treatment of VTE

Published estimates of the incidence of diagnosing DVT in the United States range from 130,000 to 550,000 cases per year. Assuming that 42% of suspected DVT patients actually have the diagnosis,[100] between 300,000 and 1,300,000 people undergo tests to diagnose DVT each year in the United States. Diagnosing and treating 1 DVT patient costs $13,055; ruling out DVT in 1 patient costs $9474 (1992 dollars).[100]

The cost of treatment of major bleeds must also be added (assume $10,000 for additional hospitalization per incidence). About 5% of DVT patients have major bleeding with heparin and about 6% with warfarin. Conservatively, the bleeding fatality rate with heparin averages .3% per course; vitamin K antagonists also average .3% bleeding mortality per 3-month course. Calculating the yearly cost of ruling in and out of DVT diagnosis and anticoagulating those with DVT goes as follows: 130,000-550,000 DVT patients x $13,055/diagnosis = $1.7 billion to $7.2 billion and 180,000-760,000 DVTs ruled out x $9474 = $1.7 billion to $7.2 billion. The cost of treatment of major bleeds follows: .11 x (130,000-550,000 patients) x $10,000 per episode = $140 million to $590 million and total cost of DVT diagnosis and treatment = $2.5 billion to $14.0 billion. Bleeding deaths from heparin (.3%/course) and warfarin (.3%/3-month course) would be estimated as follows: .006 x 130,000-550,000 DVT cases/year = 780-3300 bleeding deaths.

Estimates of the incidence of PE in the United States range from 27,000 to 57,000 US cases per year.[1,16] Hull and colleagues[101] found that only 26% (169 of 662) of suspected PE patients had scan or angiogram confirmation of the diagnosis. With these figures, 104,000-219,000 patients are investigated for PE each year, confirming 27,000-57,000 PE diagnoses and ruling out the 77,000-163,000 who do not have the diagnosis. With Hull and colleagues'[101] least expensive diagnostic scenario for ruling in or out of PE, the cost per year in the United States would be the following: $12,639 (cost per PE case) x 27,000-57,000 (PE cases/year) = $340 to $720 million and $4333 (cost per no PE) x 77,000-163,000 (no PE cases) = $330 to $700 million. The cost of treatment of major bleeds at $10,000/episode (major bleeding with heparin = 5% and warfarin = 6%) follows: .11 x (27,000-57,000) x $10,000/major bleed = $30 to $63 million, total cost of PE diagnosis and treatment = $.7 to 1.5 billion, and grand total cost of VTE diagnosis and treatment = $3.2 to $15.5 billion. Yearly bleeding deaths in PE patients in the United States from heparin (.3%) and warfarin (.5% per 6-month course) would be: .008 x (27,000-57,000) = 216-456 deaths. Yearly deaths from pulmonary angiographies (.5%[26] risk of death per case) would be: 104,000-220,000 x 29 angiograms/100 patients[26] x .005 = 151-319 deaths, total deaths per year from PE diagnosis and treatment = 367-775 deaths, and grand total yearly deaths from VTE diagnosis and therapy = 1017-3525 deaths.

Conclusion

DVT and PE occur frequently in healthy people and much more commonly in people with medical or surgical risk factors. In people with adequate cardiovascular reserve, natural fibrinolytic mechanisms clear thrombi that migrate to the pulmonary artery within 2 weeks.[23] Fatal PE occur primarily as an agonal event of terminally ill people.

Although heparoids and vitamin K antagonists have a theoretic rationale regarding prevention of death from PE, the medical scientific basis for the widespread use of anticoagulants in VTE is flawed. Because no placebo-controlled trials of low-molecular-weight heparins or thrombolytic drugs in VTE have been done, their efficacy and safety are not scientifically supported.

Diagnosing and treating 157,000-607,000 VTE patients with anticoagulants costs $3.2 to $15.5 billion per year (1992 dollars). Bleeding and complications of pulmonary angiography cost 1017-3525 lives annually.

Anticoagulants have not been shown to be efficacious in reducing morbidity and/or mortality or safe in VTE treatment.