Clinical Topic A-Z Clinical Speciality

Pulmonary embolism

Pulmonary embolism
D011655Pulmonary Embolism
CardiovascularHaematologyRespiratory
2013-06-01Last revised in June 2013

Pulmonary embolism - Summary

Pulmonary embolism is a condition in which one or more emboli, usually arising from a blood clot formed in the veins (or, rarely, in the right heart), are lodged in and obstruct the pulmonary arterial system.

This results in reduced gas exchange of the affected lung tissue, causing hypoxaemia and a reduction in cardiac output.

Large or multiple emboli may result in hypotension, syncope, shock, and sudden death.

The most common source of pulmonary emboli is deep vein thrombosis (DVT) in the lower limbs, which can be found in 70–80% of people with pulmonary embolism.

Only 15% of people with pulmonary embolism have signs of DVT.

Without anticoagulation, the risk of recurrent venous thromboembolism (DVT or pulmonary embolism) within 3 months of a pulmonary embolism is thought to be 50%.

Risk factors for venous thromboembolism include:

Recent surgery, hospitalization, immobilization, or lower limb trauma within the previous 12 weeks (and particularly the previous 6 weeks).

Previous DVT or pulmonary embolism.

Active cancer or cancer treatment.

Pregnancy and the puerperium (6 weeks postpartum).

Combined oral contraception and hormone replacement therapy.

Known thrombophilias.

Varicose veins, obesity, and increasing age (older than 60 years of age).

Suspect pulmonary embolism (PE) in a person with dyspnoea, tachypnoea, pleuritic chest pain, or features of deep vein thrombosis particularly if they also have a risk factor and an alternative diagnosis is unlikely. These features are present in 97% of people with PE.

Immediate admission to hospital should be arranged for pregnant women and people with significantly disturbed vital signs such as hypotension or an altered level of consciousness.

For all other people, assess the two-level PE Wells score to estimate the clinical probability of PE.

For people with a high Wells score indicating a likely PE, either:

Arrange hospital admission for an immediate computed tomography pulmonary angiogram (CTPA), or

If there will be a delay in the person receiving a CTPA, give immediate interim parenteral anticoagulant therapy and arrange hospital admission.

For people with a low Wells score (indicating that PE) is unlikely arrange a D-dimer test:

If the test is positive, either arrange admission to hospital for an immediate CTPA or, if a CTPA cannot be carried out immediately, give immediate interim parenteral anticoagulant therapy and arrange hospital admission.

If the test is negative, consider an alternative diagnosis.

If pulmonary embolism is confirmed in secondary care, follow up of the person in primary care should include:

Adequate monitoring of warfarin or other anticoagulation regimes.

Ensuring people with unprovoked pulmonary embolism (PE) are investigated for the possibility of an undiagnosed cancer if they are not already known to have cancer.

Thrombophilia testing in people with unprovoked PE.

Have I got the right topic?

216months3060monthsBoth

This CKS topic covers the detection of pulmonary embolism in primary healthcare, and the primary care management of people with suspected or confirmed pulmonary embolism. Secondary care management is discussed briefly.

This CKS topic does not cover the prophylaxis (primary prevention) of venous thromboembolism, or the management of pulmonary embolism due to non-thrombotic disorders (for example air, amniotic fluid, foreign bodies, or sepsis).

There are separate CKS topics on Anticoagulation - oral, Breathlessness, Deep vein thrombosis, DVT prevention for travellers, and Thrombophlebitis - superficial.

The target audience for this CKS topic is health care professionals working within the NHS in the UK, and providing first contact or primary health care.

How up-to-date is this topic?

How up-to-date is this topic?

Changes

Last revised in June 2013

February 2014 — minor update. Update to the text to state that if pulmonary embolism is suspected in a woman who has given birth within the past 6 weeks, she should be admitted immediately.

June 2013 — reviewed. A literature search was conducted in May 2013 to identify evidence-based guidelines, UK policy, systematic reviews, and key RCTs published since the last revision of this topic. The guideline has been updated to include recommendations on the diagnosis and management of pulmonary embolism made by the National Institute of Health and Clinical Excellence (NICE) guideline Venous thromboembolic diseases: the management of venous thromboembolic diseases and the role of thrombophilia testing [National Clinical Guideline Centre, 2012].

Previous changes

August 2012 — minor update. Minor typographical error corrected.

July 2011 — minor update. All references to the British Committee for Standards in Haematology (BCSH) guideline on oral anticoagulation with warfarin have been updated to reflect the latest guideline [Keeling et al, 2011]. Issued in September 2011

November 2010 to February 2011 — this is a new CKS topic. The evidence base has been reviewed in detail, and recommendations are clearly justified and transparently linked to the supporting evidence.

Update

New evidence

Evidence-based guidelines

No new evidence-based guidelines published since 1 May 2013.

HTAs (Health Technology Assessments)

No new HTAs since 1 May 2013.

Economic appraisals

No new economic appraisals relevant to England since 1 May 2013.

Systematic reviews and meta-analyses

Systematic reviews published since the last revision of this topic:

Sardar, P., Chatterjee, S., and Mukherjee, D. (2013) Efficacy and safety of new oral anticoagulants for extended treatment of venous thromboembolism: systematic review and meta-analyses of randomized controlled trials. Drugs 73(11), 1171-1182. [Abstract]

Primary evidence

Randomized controlled trials published since the last revision of this topic:

Agnelli, G., Buller, H.R., Cohen, A., et al. (2013) Oral apixaban for the treatment of acute venous thromboembolism. New England Journal of Medicine 369(9), 799-808. [Abstract]

Observational studies published since the last revision of this topic:

Kamel, H., Navi, B.B., Sriram, N., et al. (2014) Risk of a thrombotic event after the 6-week postpartum period. New England Journal of Medicine epub ahead of print. [Abstract] [Free Full-text]

New policies

No new national policies or guidelines since 1 May 2013.

New safety alerts

No new safety alerts since 1 May 2013.

Changes in product availability

No changes in product availability since 1 May 2013.

Goals and outcome measures

Goals

To support primary health care professionals:

To detect people with possible pulmonary embolism and manage them appropriately

To follow up people with confirmed pulmonary embolism, to check that they are receiving appropriate duration and monitoring of anticoagulation, thrombophilia screening if necessary, appropriate cancer screening, and appropriate management during pregnancy

Background information

Definition

What is it?

Pulmonary embolism is a condition in which one or more emboli, usually arising from a thrombus (blood clot) formed in the veins (or, rarely, in the right heart), are lodged in and obstruct the pulmonary arterial system [Camm and Bunce, 2005].

Provoked pulmonary embolism is a pulmonary embolism associated with a transient risk factor such as significant immobility, surgery, trauma, and pregnancy or puerperium. The combined contraceptive pill and hormone replacement therapy are also considered to be provoking risk factors. These risk factors can be removed, reducing the risk of recurrence.

Unprovoked pulmonary embolism is a pulmonary embolism occurring in the absence of a transient risk factor. The person may have no identifiable risk factor or a risk factor that is persistent and not easily correctable (such as active cancer or thrombophilia). Because these risk factors cannot be removed, the person is at an increased risk of recurrence.

Deep vein thrombosis is the term used to describe the formation of a thrombus in a deep vein, usually in one of the legs [Camm and Bunce, 2005].

Venous thromboembolism is a term used to encompass both pulmonary embolism and deep vein thrombosis.

[National Clinical Guideline Centre, 2012]

Pathophysiology

What is the pathophysiology?

In pulmonary embolism, lung tissue is ventilated but not perfused, producing an intra-pulmonary dead space and resulting in impaired gas exchange.

After several hours, alveolar collapse occurs, which worsens hypoxaemia.

This leads to a reduction in the cross-sectional area of the pulmonary arterial bed, which results in an elevation of pulmonary arterial pressure and a reduction in cardiac output.

The area of lung that is no longer perfused by the pulmonary artery may infarct, but often does not do so because oxygen continues to be supplied by the bronchial circulation and the airways.

Large or multiple emboli can abruptly increase pulmonary arterial pressure to a level of afterload which cannot be matched by the right ventricle.

Sudden death may occur, or the person may present with hypotension or syncope, which might progress to shock or death due to acute right ventricular failure.

[Camm and Bunce, 2005; Torbicki et al, 2008]

Sources of emboli

What are the sources of emboli?

The most common source of pulmonary emboli is deep vein thrombosis (DVT) in the lower limbs, which, if sensitive diagnostic methods are used, can be found in 70–80% of people with pulmonary embolism [Tapson, 2008; Torbicki et al, 2008].

For some people with pulmonary embolism, DVT in the lower limbs is not detected because the whole thrombus has already detached and embolized [Tapson, 2008].

Only 15% of people with pulmonary embolism have signs of DVT [Torbicki et al, 2008].

Some pulmonary emboli originate from thrombi in abdominal or axillary veins, or from the right ventricle [Camm and Bunce, 2005; Forgione, 2006].

Other sources of emboli [Camm and Bunce, 2005; Torbicki et al, 2008]

Tumours — most commonly prostate and breast cancers.

Fat — from long-bone fractures.

Amniotic fluid in pregnant women.

Sepsis — for example: tricuspid valve endocarditis in people who inject illicit drugs intravenously, or have infected indwelling catheters or pacemaker wires; people with peripheral septic thrombophlebitis; or recipients of transplanted organs.

Foreign bodies — during intravenous drug use, or from broken catheters, guide wires, vena cava filters, embolization coils, and endovascular stent components.

Air — admitted during surgery or from other communication between the environment and the venous system.

Incidence

How common is it?

The annual incidence of diagnosed pulmonary embolism in the UK has been reported as 3–4 per 10,000 people [Huerta et al, 2007]. This is likely to be an underestimate because it is based on clinical data only.

Studies in the US and Europe have reported the annual incidence of pulmonary embolism to be 4–21 per 10,000 inhabitants [Torbicki et al, 2008]. Variation between studies is due to differences in the use of clinical and postmortem data [Torbicki et al, 2008].

Clinical data may underestimate the true incidence of pulmonary embolism, whereas postmortem data may overestimate it (because clinically insignificant pulmonary emboli are detected).

Nevertheless, postmortem studies suggest that pulmonary embolism is clinically suspected in fewer than half of people who die as a result of pulmonary embolism [Meyer et al, 2010].

Risk factors

What are the risk factors?

Major risk factors for pulmonary embolism (PE) include:

Deep vein thrombosis (DVT).

Previous DVT or pulmonary embolism.

Active cancer.

Recent surgery, hospitalization, lower limb trauma, or other immobilization (including long-distance sedentary travel).

Pregnancy and, in particular, for 6 weeks' postpartum.

Other risk factors include:

Increasing age (older than 60 years of age).

Combined oral contraception and hormone replacement therapy.

Obesity (body mass index greater than 30 kg/m2).

One or more significant medical comorbidities (for example: heart disease; metabolic, endocrine, neurological disability, or respiratory pathologies; acute infectious disease; or inflammatory conditions).

Varicose veins.

Superficial venous thrombosis.

Known thrombophilias (thrombotic disorders).

Other: indwelling central vein catheter, nephrotic syndrome, chronic dialysis, myeloproliferative disorders, paroxysmal nocturnal haemoglobinuria, or Behçet's disease.

[BTS, 2003; Torbicki et al, 2008; Sweetland et al, 2009; National Clinical Guideline Centre, 2012]

Risk of death

What is the risk of death from pulmonary embolism?

Number of deaths

Each year between 2005 and 2008, pulmonary embolism was mentioned on the death certificates of 12,000–13,000 people in the UK [UK Parliament, 2010].

The number of deaths is thought to be considerably higher — up to 60,000 people may die as a result of pulmonary embolism each year in the UK [Cohen et al, 2007].

Risk of death

The risk of death from pulmonary embolism is dependent on the severity of the pulmonary embolism and whether the person receives treatment.

Untreated, the risk of death in people from pulmonary embolism is high.

Studies published before 1960 reported mortalities of 23–87% [Blondon et al, 2009].

When treated with heparin and anticoagulants, the risk of death is considerably lower.

In a systematic review of prospective studies (excluding people who received thrombolysis) [Douketis et al, 1998]:

During the 3-month period of anticoagulation following diagnosis, the risk of death from definite or probable pulmonary embolism was 2.3% (about 1 in 40 people). All deaths occurred in the first 2 weeks of anticoagulation.

There were no deaths during 265 patient-years of follow up after discontinuing anticoagulation.

In three subsequent prospective studies (including people who received thrombolysis), 3–6% (about 1 in 15–30) of people died of definite or probable pulmonary embolism during the 3-month period of anticoagulation following diagnosis [Conget et al, 2008; Donzé et al, 2008; Jimenez et al, 2010].

In two prospective studies of people who had received at least 3 months of anticoagulation following their first pulmonary embolism, the annual risk of death from definite or probable pulmonary embolism was 0.2% (1 in 500 people) during a mean of 4.5 years of follow up after stopping anticoagulation [Douketis et al, 2007].

In people with clinically massive pulmonary embolism (hypotension):

The risk of death within 90 days is about 50% [Kucher et al, 2006].

It is unclear whether thrombolytic treatment reduces the risk [Kucher et al, 2006; Torbicki et al, 2008; Dong et al, 2009].

Pulmonary embolism is the leading cause of maternal deaths in the UK, with a mortality rate of 1.56 per 100,000 pregnancies [CEMACH, 2007].

Other complications

What other complications are there?

Chronic thromboembolic pulmonary hypertension (CTEPH) occurs in 0.5–5% of people with treated pulmonary embolism.

In people with CTEPH, emboli are replaced over months or years by fibrous tissue. This can lead to chronic obstruction of the pulmonary arterial vasculature.

This is followed by progressive increases in pulmonary arterial pressure, leading to right heart failure.

[Torbicki et al, 2008]

Risk of recurrence

What is the risk of recurrence?

CKS identified no studies which accurately reported the risk of recurrence after pulmonary embolism as opposed to after any venous thromboembolism.

Without anticoagulation, the risk of recurrence of venous thromboembolism (deep vein thrombosis [DVT] or pulmonary embolism) within 3 months of a pulmonary embolism is thought to be 50% [Torbicki et al, 2008].

In a meta-analysis of cohort studies and randomized controlled trials, the incidence of recurrent venous thromboembolism after DVT or pulmonary embolism, following 3 months of anticoagulation, was 8% during the first year after stopping anticoagulation therapy [van Dongen et al, 2003].

An individual's risk of recurrence will depend on their risk factors. The risk of recurrence for someone with a provoked PE (a PE associated with a temporary risk factor that can be removed such as the contraceptive pill) is lower than for someone with an unprovoked PE who has a risk factor that cannot be removed (such as active cancer, obesity, or thrombophilia) [NICE, 2012].

When to suspect PE

When should I suspect pulmonary embolism?

When to suspect PE

Suspect pulmonary embolism (PE) in a person with dyspnoea, tachypnoea, pleuritic chest pain, or features of deep vein thrombosis. These features are present in 97% of people with PE.

Other features that may be present include:

Tachycardia (heart rate greater than 100 beats per minute).

Haemoptysis.

Syncope.

Hypotension (systolic blood pressure less than 90 mmHg).

Crepitations.

Cough or fever.

If PE is suspected:

Exclude other conditions that could explain symptoms including:

Respiratory conditions such as pneumothorax, pneumonia, acute exacerbation of chronic lung disease.

Cardiac causes such as acute coronary syndrome, acute congestive heart failure, dissecting or rupturing aortic aneurysm, pericarditis.

Musculoskeletal chest pain. Note that chest pain with chest wall palpation occurs in up to 20% of people with confirmed PE.

Gastro-oesophageal reflux disease.

Pregnancy.

Any cause for collapse such as vasovagal syncope, orthostatic (postural) hypotension, cardiac arrhythmias, seizures, cerebrovascular disorders.

Do not delay management of suspected PE for results of an ECG or chest X-ray. They have limited value in diagnosis because they are usually normal in someone with a PE. They may be done as part of investigations for breathlessness or chest pain when another diagnosis seems more likely.

ECG signs that may be present include: sinus tachycardia, non-specific ST-segment and T-wave abnormalities, right axis deviation, incomplete or complete right bundle-branch block, and, less commonly, T-wave inversion in leads V1–V3, P pulmonale, or the classical S1, Q3, T3 (S wave in lead 1, Q wave in lead 3, and T-wave inversion in lead 3).

Chest X-ray features that may be present include: atelectasis, pleural effusion, or elevation of a hemidiaphragm.

If PE is suspected, assess for risk factors for PE.

Major risk factors include:

Deep vein thrombosis (DVT). Suspect if there is unilateral leg pain, swelling, redness, increased temperature, or venous distension. However, only 15% of people with pulmonary embolism have signs of DVT. For further information on DVT see the CKS topic on Deep vein thrombosis).

Previous DVT or pulmonary embolism.

Active cancer.

Recent surgery, hospitalization, lower limb trauma, or other immobilization (including long-distance travel).

Pregnancy and, in particular, for 6 weeks' postpartum.

For other risk factors — see Risk factors.

If pulmonary embolism is likely following the above assessment — see Scenario: Managing suspected pulmonary embolism.

Basis for recommendation

Basis for recommendation

There is evidence that clinical features and risk factors, either alone or in combination, have low specificity and low positive predictive value for diagnosing pulmonary embolism. However, there is also evidence that, when used in combination, they have high sensitivity and reasonably high negative predictive values. Hence, the absence of clinical features or risk factors may be useful for ruling out pulmonary embolism, but their presence does not rule in the diagnosis.

Clinical features and risk factors

The statement that most people (97%) with pulmonary embolism have either dyspnoea, tachypnoea, pleuritic chest pain, or symptoms or signs of deep vein thrombosis is derived from a prospective study on the accuracy of clinical features and risk factors for diagnosing pulmonary embolism [Stein et al, 2007].

The list of clinical features and risk factors is derived from:

Evidence from a systematic review and meta-analysis of the value of individual clinical features in the diagnosis of pulmonary embolism [West et al, 2007].

The review found that individual clinical features only slightly raise or lower the probability of pulmonary embolism. However, the most useful features for ruling in pulmonary embolism (likelihood ratios significantly greater than 1) were syncope, shock, current deep vein thrombosis, leg swelling, sudden dyspnoea, active cancer, recent surgery, haemoptysis, leg pain, and possibly thrombophlebitis. The most useful features for ruling out pulmonary embolism (likelihood ratios less than 1) were absence of sudden dyspnoea, absence of any dyspnoea, and absence of tachypnoea.

Evidence from four other prospective and retrospective studies on the diagnostic accuracy of clinical features and risk factors.

Evidence from systematic reviews and individual studies of clinical prediction rules for ruling out pulmonary embolism.

There is evidence from five studies of the Wells clinical prediction rule that 1–29% of people assessed as having low clinical (pre-test) probability actually have pulmonary embolism. Given the variation in accuracy between studies, and the highest value, the Wells rule cannot safely exclude pulmonary embolism. However, individual variables within the clinical prediction rules have been incorporated into the recommendations on when to consider the possibility of pulmonary embolism.

Evidence from prospective studies, reviewed in guidelines on the diagnosis and management of acute pulmonary embolism by the European Society of Cardiology [Torbicki et al, 2008], on the prevalence of symptoms and signs in people with confirmed pulmonary embolism.

Evidence from retrospective cohort studies reported in a review article on pulmonary embolism in pregnancy [Bourjeily et al, 2010] and a report of the Confidential Enquiries into Maternal Deaths in the United Kingdom [CEMACH, 2007].

The information that only 15% of people with pulmonary embolism have signs of deep vein thrombosis is derived from data from two diagnostic studies in guidelines on the diagnosis and management of acute pulmonary embolism by the European Society of Cardiology [Torbicki et al, 2008].

The statement that chest pain reproducible on palpation does not exclude pulmonary embolism is based on evidence from a diagnostic accuracy study, which found that chest pain reproducible on palpation occurred in 20% of people with confirmed pulmonary embolism, had an insufficiently high negative predictive value (79%), and had a negative likelihood ratio close to 1 (0.83, 95% CI 0.60 to 1.14) [Le Gal et al, 2005].

Other conditions to consider when assessing someone with symptoms of possible pulmonary embolism

This information is based on:

Expert opinion in guidelines from the British Thoracic Society on the diagnosis and management of pulmonary embolism [BTS, 1997; BTS, 2003], and review articles [Winters and Katzen, 2006; Scarsbrook and Gleeson, 2007; Bourjeily et al, 2010; Brims et al, 2010; Meyer et al, 2010].

Evidence from:

A study reporting final diagnoses in people who had investigations to rule out pulmonary embolism [Akram et al, 2009].

A study reporting pathological findings on computed tomographic pulmonary angiograms (ordered to diagnose pulmonary embolism) that were thought to support alternative diagnoses [Hall et al, 2009].

A prospective evaluation of people presenting to an emergency department with syncope [Sarasin et al, 2001].

A case report [Warburton et al, 2004].

ECG and chest X-ray findings associated with pulmonary embolism

This information is derived from guidelines on the diagnosis and management of acute pulmonary embolism by the European Society of Cardiology [Torbicki et al, 2008] and a textbook chapter [Stein and Firth, 2010].

Management

Management

Scenario: Managing suspected pulmonary embolism : covers the assessment of the clinical probability of pulmonary embolism and other primary care management of people with suspected pulmonary embolism, and briefly covers investigations that may be carried out in secondary care to confirm or exclude the diagnosis.

Scenario: Managing confirmed pulmonary embolism : covers the primary care follow up of people with confirmed pulmonary embolism, and briefly covers secondary care management.

Scenario: Managing suspected pulmonary embolism

Scenario: Managing suspected pulmonary embolism

216months3060monthsBoth

Primary care management

How should I manage a person with suspected pulmonary embolism?

Arrange immediate admission for people with suspected pulmonary embolism:

If they are severely ill with any of the following features:

Altered level of consciousness.

Systolic BP of less than 90 mmHg.

Heart rate of more than 130 beats per minute.

Respiratory rate of more than 25 breaths per minute.

Oxygen saturation of less than 91%.

Temperature of less than 35°C.

If they are pregnant, or have given birth within the past 6 weeks.

For all other people, assess the two-level PE Wells score to estimate the clinical probability of PE.

Clinical features of deep vein thrombosis (minimum of leg swelling and pain with palpation of the deep veins) — 3 points.

Heart rate greater than 100 beats per minute — 1.5 points.

Immobilization for more than 3 days or surgery in the previous 4 weeks — 1.5 points.

Previous deep vein thrombosis or pulmonary embolism — 1.5 points.

Haemoptysis — 1 point.

Cancer (receiving treatment, treated in the last 6 months, or palliative) — 1 point.

An alternative diagnosis is less likely than pulmonary embolism — 3 points. Alternative conditions to consider include:

Respiratory conditions such as pneumothorax, pneumonia, acute exacerbation of chronic lung disease.

Cardiac causes such as acute coronary syndrome, acute congestive heart failure, dissecting or rupturing aortic aneurysm, pericarditis.

Musculoskeletal chest pain. Note that chest pain with chest wall palpation occurs in up to 20% of people with confirmed PE.

Gastro-oesophageal reflux disease.

Any cause for collapse such as vasovagal syncope, orthostatic (postural) hypotension, cardiac arrhythmias, seizures, cerebrovascular disorders.

For people with a Wells score of more than 4 points (PE likely):

Either, arrange hospital admission for an immediate computed tomography pulmonary angiogram (CTPA).

Or, if there will be a delay in the person receiving a CTPA, give immediate interim low molecular weight heparin or fondaparinux and arrange hospital admission.

For people with a Wells score of 4 points or less (PE unlikely), arrange a D-dimer test:

If the test is positive, either arrange admission to hospital for an immediate CTPA or, if a CTPA cannot be carried out immediately, give immediate low molecular weight heparin or fondaparinux and arrange hospital admission. For more information on prescribing a low molecular weight heparin or fondaparinux, see Prescribing information.

If the test is negative consider an alternative diagnosis.

Basis for recommendation

Basis for recommendation

Immediate hospital admission for people with suspected pulmonary embolism (PE) who are severely ill

These recommendations are based on the National Early Warning Score assessment of acute-illness severity [Royal College of Physicians, 2012]. Recommendations are based on a review of the available evidence on the performance of early warning scoring systems and the expert opinion of the 19 members of the Development and Implementation group of the Royal College of Physicians.

Immediate hospital admission for people with suspected PE who are pregnant

These referral recommendations are based on expert opinion of the Royal College of Obstetricians and Gynaecologists [RCOG, 2010]. Referral for objective testing PE is required because it is not possible to accurately assess the risk of PE in primary care based on the usual methods of assessment because:

The two-level PE Wells score can not be used to assess PE risk in pregnant women.

The usefulness of D-dimer testing is limited by a high rate of false positive results in pregnancy.

Assessment of the clinical probability of PE using the two-level PE Wells score and D-dimer testing

These recommendations are consistent with recommendations made by the National institute of Health and Clinical Excellence (NICE) who concluded that pre-test probability scoring system, followed by a D-dimer test can safely rule out PE [National Clinical Guideline Centre, 2012].

This is supported by evidence that, in emergency department settings, the prevalence of pulmonary embolism (false-negative rate) is 0.2–2.7% in people with a low clinical (pre-test) probability using the Wells rule, combined with a negative D-dimer test. This compares favourably with a false negative rate for computed tomographic pulmonary angiography (otherwise the investigation of choice in people with suspected pulmonary embolism) of around 4% in the same population [Torbicki et al, 2008].

The Wells rule has not been evaluated in non-hospital primary care settings. Its generalizability to these settings may be questionable but can be justified.

In the initial derivation study for the Wells rule, chest X-ray, electrocardiography, and arterial blood gases were used to determine whether an alternative diagnosis was less likely [Wells et al, 1998]. Whilst these are not usually immediately available in primary care, their value is in excluding pulmonary embolism rather than in ruling it in. Their absence is likely to overestimate rather than underestimate the prevalence of pulmonary embolism, therefore making the rule no less safe for use in non-hospital primary health care.

The prevalence of pulmonary embolism will be lower in non-hospital primary care than in emergency departments, and so the predictive value of this diagnostic approach will be higher.

Interim low molecular weight heparin or fondaparinux

This recommendation is based on the expert opinion of the guideline development group (GDG) of NICE [National Clinical Guideline Centre, 2012].

The GDG considered the costs, potential adverse effects of treatment, and the risk of death from untreated PE. They concluded that if a patient has a “likely” probability of PE, treatment with either low molecular weight heparin, or fondaparinux should be started while waiting for confirmation, and stopped if the scan result is negative.

Low molecular weight heparin and fondaparinux are equally favoured based on considerations of evidence on efficacy and safety [National Clinical Guideline Centre, 2012].

Secondary care investigations

What investigations may be done in secondary care?

Secondary care investigations may include one or more of:

Computed tomographic pulmonary angiography — the investigation of choice for most people with high clinical probability of pulmonary embolism, or non-high clinical probability and a positive D-dimer test.

D-dimer testing — in people with a Wells score of 4 points or less when PE is thought to be unlikely.

Arterial blood gases — although up to 20% of people with pulmonary embolism have a normal arterial oxygen pressure.

Chest X-ray and electrocardiography — mainly to exclude alternative diagnoses.

Lower limb compression venous ultrasound — may be useful for pregnant women in whom irradiation from other imaging may be harmful.

Ventilation-perfusion or perfusion scintigraphy (isotope lung scanning) — may be done in certain circumstances (for example half-dose perfusion scintigraphy in pregnancy).

Echocardiography — for people with hypotension (clinically 'massive' pulmonary embolism), the absence of right heart failure excludes pulmonary embolism.

Basis for recommendation

Basis for recommendation

This information is derived from the National Institute of Health and Clinical Excellence guideline Venous thromboembolic diseases: the management of venous thromboembolic diseases and the role of thrombophilia testing [National Clinical Guideline Centre, 2012].

Scenario: Managing confirmed pulmonary embolism

Scenario: Managing confirmed pulmonary embolism

216months3060monthsBoth

Secondary care treatment

What treatment may be offered in secondary care?

If pulmonary embolism is confirmed:

For people who are haemodynamically stable, parenteral anticoagulation is started as soon as possible (low molecular weight heparin, or fondaparinux) and continued for at least 5 days or until the international normalized ratio (INR) is above 2 for at least 24 hours, whichever is longer. People who have renal impairment or are at increased risk of bleeding should be treated with unfractionated heparin.

For pregnant women, low molecular weight heparin is continued until the end of pregnancy.

For people with active cancer, low molecular weight heparin is continued until the cancer is considered cured or for at least 6 months.

For other people, warfarin (or occasionally rivaroxaban) is commenced within 24 hours of confirmation of the diagnosis.

Treatment is usually continued for 3 months in people with a provoked pulmonary embolism (people with a transient, correctable risk factor such as prolonged immobility following surgery, trauma, or pregnancy).

Treatment is continued for more than 3 months in people with an unprovoked pulmonary embolism (people with no identifiable risk factor or a risk factor that is persistent and unmodifiable such as obesity or active cancer).

For people who are haemodynamically unstable thrombolytic therapy or embolectomy may be offered.

Basis for recommendation

Basis for recommendation

Initial anticoagulation with parenteral anticoagulants

Parenteral anticoagulants are recommended in the initial phase of therapy based on expert opinion from the Guideline Development Group (GDG) of the National Institute for Health and Clinical Excellence (NICE) to ensure rapid anticoagulation to reduce the risk of harm from thrombosis propagation or further embolic events [National Clinical Guideline Centre, 2012].

Choice of parenteral anticoagulant

The NICE GDG identified trial evidence for people being treated for pulmonary embolism (PE) and deep vein thrombosis (DVT) with low molecular weight heparin (LMWH), unfractionated heparin (UH) and fondaparinux [National Clinical Guideline Centre, 2012]. Comparisons of their relative mortality reduction, rates of major complications such as major bleeding events, cost effectiveness, and ease of use were analyzed by the GDG. After consideration of this evidence the GDG recommended offering a choice of low molecular weight heparin (LMWH) or fondaparinux to patients with confirmed proximal DVT or PE, taking into account comorbidities, contraindications and drug costs, with the following exceptions:

For people with severe renal impairment or established renal failure (estimated glomerular filtration rate [eGFR] < 30 ml/min/1.73 m2), offer unfractionated heparin (UFH).

For people with an increased risk of bleeding, consider UFH because it has a shorter half-life and is more easily reversed if required.

For people with PE and haemodynamic instability, offer UFH.

Longer term anticoagulation for people with cancer

The NICE GDG reviewed the evidence on people with cancer with deep vein thrombosis or pulmonary embolism being treated with warfarin or parenteral anticoagulants and found that anticoagulation for 6 months with LMWH leads to better outcomes compared with switching to warfarin after initial LMWH treatment [National Clinical Guideline Centre, 2012]. The GDG also discussed the potential advantages of using LMWH in people with cancer. Based on their clinical experience:

It is difficult to maintain good INR control (which increases the risk of bleeding or more VTE events) while patients are on chemotherapy. This makes a strong case for the use of LMWH for patients with new VTE and active cancer, particularly if undergoing chemotherapy.

Patients with cancer have a higher risk of major bleeding on anticoagulation compared to patients without cancer, which may relate to the underlying cancer and propensity for bleeding (e.g. ulcerated gastric cancer).

Longer term anticoagulation for pregnant women

The Royal College of Obstetricians and Gynaecologists recommends the use of LMWH based on evidence of its safety in pregnant women from large systematic reviews [RCOG, 2010]. Warfarin is teratogenic and contra-indicated in pregnancy [BNF 65, 2013].

Duration of treatment with anticoagulants

The recommendation to anticoagulate someone with a first episode of unprovoked pulmonary embolism for 3 months is based on the expert opinion of the NICE GDG, taking into account the current standards of practice [National Clinical Guideline Centre, 2012].

Evidence considered by the NICE GDG suggests that long-term anticoagulant treatment benefits patients with unprovoked PE based on considerations of reduced incidence of recurrent PE and cost effectiveness [National Clinical Guideline Centre, 2012]. Evidence suggests that anticoagulation only offers prevention while on treatment and therefore may have to be extended as long as the underlying risk factor(s) remain(s). Correctly identifying 'unprovoked' PE and where there may be permanent underlying risk factors as opposed to 'provoked' PE where the risk factors are likely to be temporary (for example, surgery, immobilisation) is the first step.

Follow up

How should I follow up a person with confirmed pulmonary embolism?

For people on warfarin ensure adequate monitoring — for more information, see the CKS topic on Anticoagulation - oral.

The target international normalized ratio (INR) is usually 2.5, keeping within the range of 2.0–3.0.

Specialists will usually make clinical decisions on the duration of treatment.

Ensure people with unprovoked pulmonary embolism (PE) are investigated for the possibility of an undiagnosed cancer if they are not already known to have cancer.

Initially undertake:

A full history and physical examination to look for evidence of malignancy.

A chest X-ray.

Blood tests including a full blood count, serum calcium, and liver function tests.

Urinalysis.

Consider referral for further investigations for cancer with an abdomino-pelvic CT scan (and mammogram in women) in all people over 40 years with a first unprovoked PE who do not have features of cancer based on the initial investigations above.

In people with an unprovoked PE, consider antiphospholipid testing (anti-cardiolipin or anti-beta glycoprotein I antibodies) before stopping anticoagulants.

In people with an unprovoked PE who have a first-degree relative who has had a DVT or PE, consider arranging hereditary thrombophilia testing (antithrombin, protein C, and protein S testing).

Basis for recommendation

Basis for recommendation

Investigations for undiagnosed cancer

These recommendations are based on the expert opinion of the guideline development group (GDG) of the National Institute of Health and Clinical Excellence (NICE) after consideration of the evidence [National Clinical Guideline Centre, 2012]:

Apparent unprovoked venous thromboembolism is associated with a significant increase in the risk of cancer within the first 1-2 years, with a standardized incidence ratio of 4.4 in a Swedish population-based study involving nearly 62,000 patients over 18 years. This is equivalent to approximately 11% of all patients.

Recent studies have shown there are differences in the optimum management of venous thromboembolism between patients with cancer and those without cancer. Thus it is important to identify these people, to not only manage their cancer at the earliest stage possible, but also to optimally manage their venous thromboembolism to reduce the risk of a potentially serious recurrence.

Thrombophilia testing

These recommendations are based on the expert opinion of the NICE GDG after consideration of evidence for thrombophilia prevalence, relative risk of venous thromboembolism recurrence for different types of thrombophilia, and effectiveness of treatment at preventing recurrences [National Clinical Guideline Centre, 2012]. No clinical trials comparing testing against no testing were found.

Important aspects of prescribing information relevant to primary healthcare are covered in this section specifically for the drugs recommended in this CKS topic. For further information on contraindications, cautions, drug interactions, and adverse effects, see the electronic Medicines Compendium (eMC) (http://medicines.org.uk/emc), or the British National Formulary (BNF) (www.bnf.org).

Prescribing issues - dalteparin

What should I be aware of when prescribing dalteparin?

Dalteparin should be administered subcutaneously.

The recommended dose of dalteparin for the treatment of pulmonary embolism is:

For the single-dose injections [ABPI Medicines Compendium, 2013b; BNF 65, 2013]:

Body weight less than 46 kg — 7,500 units once daily.

Body weight 46–56 kg — 10,000 units once daily.

Body weight 57–68 kg — 12,500 units once daily.

Body weight 69–82 kg — 15,000 units once daily.

Body weight 83 kg and over — 18,000 units once daily.

Dalteparin is contraindicated [ABPI Medicines Compendium, 2013b]:

In people with current (or history of) heparin-induced thrombocytopenia.

In people with acute gastroduodenal ulcer; cerebral haemorrhage; known haemorrhagic diathesis (a condition causing a predisposition to bleed) or other active haemorrhage; serious coagulation disorders; acute or sub-acute septic endocarditis; haemorrhagic pericardial effusion and haemorrhagic pleural effusion; and injuries to and operations on the central nervous system, eyes, and ears.

In people who have suffered a recent (within 3 months) stroke, unless due to systemic emboli.

More detailed information on prescribing dalteparin is available in the British National Formulary (BNF) (www.bnf.org) and the relevant Summary of Product Characteristics listed in the electronic Medicines Compendium (eMC) (www.medicines.org.uk) [ABPI Medicines Compendium, 2013b].

Prescribing issues - enoxaparin

What should I be aware of when prescribing enoxaparin?

Enoxaparin should be administered subcutaneously.

The recommended dose of enoxaparin for the treatment of pulmonary embolism is 1.5 mg/kg (150 units/kg) once daily [ABPI Medicines Compendium, 2013c; BNF 65, 2013].

Enoxaparin is contraindicated [ABPI Medicines Compendium, 2013c]:

In people with current (or history of) heparin-induced thrombocytopenia.

In people with acute bacterial endocarditis.

In people with active major bleeding, and conditions with a high risk of uncontrolled bleeding, including including recent haemorrhagic stroke and thrombocytopenia in people with a positive in-vitro aggregation test in the presence of enoxaparin.

In people with active gastric or duodenal ulceration.

In breastfeeding women.

More detailed information on prescribing enoxaparin is available in the British National Formulary (BNF) (www.bnf.org) and the relevant Summary of Product Characteristics listed in the electronic Medicines Compendium (eMC) (www.medicines.org.uk) [ABPI Medicines Compendium, 2013c].

Prescribing issues - tinzaparin

What should I be aware of when prescribing tinzaparin?

Tinzaparin should be administered subcutaneously.

The recommended dose of tinzaparin for the treatment of pulmonary embolism is 175 units/kg once daily [ABPI Medicines Compendium, 2011; BNF 65, 2013].

Tinzaparin is contraindicated [ABPI Medicines Compendium, 2011]:

In people with current (or history) of heparin-induced thrombocytopenia.

In people with generalised or local haemorrhagic tendency, including uncontrolled severe hypertension; severe liver insufficiency; active peptic ulcer; acute or subacute septic endocarditis; intracranial haemorrhage; or injuries and operations on the central nervous system, eyes and ears.

In women with an impending miscarriage.

In breastfeeding women.

More detailed information on prescribing tinzaparin is available in the British National Formulary (BNF) (www.bnf.org) and the relevant Summary of Product Characteristics listed in the electronic Medicines Compendium (eMC) (www.medicines.org.uk) [ABPI Medicines Compendium, 2011].

Prescribing issues - fondaparinux

What should I be aware of when prescribing fondaparinux?

Fondaparinux should be administered subcutaneously.

The recommended dose of fondaparinux for the treatment of pulmonary embolism is [ABPI Medicines Compendium, 2013a; BNF 65, 2013]:

Body weight less than 50 kg — 5 mg every 24 hours.

Body weight 50–100 kg — 7.5 mg every 24 hours.

Body weight over 100 kg — 10 mg every 24 hours.

Fondaparinux is contraindicated in people with [ABPI Medicines Compendium, 2013a]:

Active clinically significant bleeding.

Acute bacterial endocarditis.

Severe renal impairment (creatinine clearance less than 20 ml/minute).

More detailed information on prescribing fondaparinux is available in the British National Formulary (BNF) (www.bnf.org) and the relevant Summary of Product Characteristics listed in the electronic Medicines Compendium (eMC) (www.medicines.org.uk) [ABPI Medicines Compendium, 2013a].

Evidence

Evidence

Supporting evidence

This section summarizes the evidence that supports the recommendations about the detection, diagnosis, and exclusion of pulmonary embolism in primary healthcare.

Clinical features

Evidence on clinical features and risk factors for diagnosing pulmonary embolism

One systematic review found that individual clinical features only slightly raise or lower the probability of pulmonary embolism. Several features were found to be of more value than others for ruling in (shock, syncope, current deep vein thrombosis, and leg swelling) or ruling out (absence of sudden dyspnoea) pulmonary embolism. Other prospective studies (with a high risk of bias) suggest that clinical features and risk factors in combination have low specificity and positive predictive value, but high sensitivity and reasonably high negative predictive values. This means that the absence of some clinical features or risk factors may be useful for ruling out pulmonary embolism, but their presence does not rule in the diagnosis. In particular, the absence of dyspnoea, tachypnoea, chest pain, or deep vein thrombosis appeared to be useful in ruling out pulmonary embolism.

One systematic review of the diagnostic value of individual clinical features to determine the probability of pulmonary embolism identified 18 studies for inclusion (9597 people) [West et al, 2007].

Methods

It was appropriate that the review excluded case-control studies (people were selected on the basis of having, or not having, pulmonary embolism). It also excluded studies that measured the risk of developing pulmonary embolism rather than the probability of pulmonary embolism being present at the time of diagnosis (known elsewhere as 'clinical outcome studies'). This may not be justified, because follow up can be used as a type of reference standard; it is particularly helpful when no test has sufficient diagnostic accuracy to fulfil criteria as a reference standard. The review was otherwise methodologically sound. An assessment of study quality was done, although the authors did not use the validated QUADAS (Quality Assessment of Diagnostic Accuracy Studies) tool [Whiting et al, 2003]. The review did not consider combinations of clinical features.

Results

Individual clinical features only slightly raised or lowered the probability of pulmonary embolism, as illustrated by pooled likelihood ratios that were not considerably higher, or considerably lower, than 1.

Features that had the highest pooled positive likelihood ratios which were also statistically significant, making them the most useful features for ruling in pulmonary embolism if they are present, were: shock (positive likelihood ratio [LR] 4.07), syncope (2.38), current deep vein thrombosis (2.05), leg swelling (2.11), sudden dyspnoea (1.83), active cancer (1.74), recent surgery (1.63), haemoptysis (1.62), and leg pain (1.60). Thrombophlebitis also had a relatively high pooled positive LR of 2.20, but this was not statistically significant.

Features that had the lowest pooled negative LRs which were also statistically significant, making them the most useful features for ruling out pulmonary embolism if they are absent, were: sudden dyspnoea (negative LR 0.43), any dyspnoea (0.52), and tachypnoea (0.56).

All other clinical features had LRs near to 1.

CKS identified three other prospective studies that were not included in the systematic review [West et al, 2007]. This may be because several methodological quality criteria are not clear: whether the patients were consecutive, whether the reference standard was applied regardless of the findings of the clinical assessment, and whether the individuals who undertook the clinical assessment were blind to the results of the reference test(s).

One study described the clinical features of 215 non-pregnant adults without cardiac or respiratory disorders who had participated in trials of thrombolytic treatment [Stein et al, 1981].

The study did not assess the accuracy of the clinical features in the diagnosis of pulmonary embolism; all of the people included in the study had pulmonary embolism that had been confirmed radiologically. People with factors that made thrombolytic treatment a contraindication (for example recent surgery or haemorrhagic stroke) were excluded.

The most common symptoms were dyspnoea (in 84% of people), pleuritic pain (74%), apprehension (63%), and cough (50%). The most common signs were tachypnoea (respiratory rate of 20 breaths or more per minute; in 85% of people), tachycardia (heart rate of 100 beats or more per minute; 58%), accentuated pulmonary component of the second heart sound (57%), crepitations (56%), and fever (50%).

Of note, dyspnoea or tachypnoea occurred in 96% of people, and dyspnoea, tachypnoea, or deep vein thrombosis were present in 99%.

One prospective study of the accuracy of clinical features and risk factors for diagnosing pulmonary embolism used data from a large study on ventilation-perfusion scanning for diagnosing pulmonary embolism [Stein et al, 1991].

In people with suspected pulmonary embolism who had no previous cardiac or pulmonary disease, the presence of any predisposing factor (from a list of eight variables) had a sensitivity of 82% and a negative predictive value (NPV) of 80%, but a specificity of 35% and a positive predictive value (PPV) of 37%.

In people with suspected pulmonary embolism who had no previous cardiac or pulmonary disease, the presence of either dyspnoea, pleuritic chest pain, haemoptysis, or circulatory collapse had a sensitivity of 95% and an NPV of 82%, but a specificity of 11% and a PPV of 33%.

One subsequent prospective study reported similar results for the presence of either dyspnoea or chest pain: sensitivity 90%, NPV 78%; specificity 10%, and PPV 54% [Palla et al, 1995].

One prospective study of the accuracy of clinical features and risk factors for diagnosing pulmonary embolism used data from a large study on multi-detector computed tomography for diagnosing pulmonary embolism [Stein et al, 2007]. The study found that the presence of either dyspnoea, tachypnoea, pleuritic chest pain, or symptoms or signs of deep vein thrombosis (in people with suspected pulmonary embolism) had a sensitivity for diagnosing pulmonary embolism of 97% and an NPV of 77%, but a specificity of 13% and a PPV of 23%.

Clinical prediction rules

Evidence on clinical prediction rules for excluding pulmonary embolism

Clinical prediction rules use composite scores of individual clinical features and risk factors to determine the clinical (pre-test) probability of pulmonary embolism. CKS identified two systematic reviews and 13 original studies evaluating the utility of clinical prediction rules for excluding pulmonary embolism. The results were not pooled in either of the two systematic reviews. For this reason, CKS has summarized each study, focusing on those which use clinical prediction rules that are feasible for use in primary health care.

Several clinical prediction rules have been described, including the Wells rule, the revised Wells rule, the Geneva and revised Geneva rules, the pulmonary embolism rule-out criteria (PERC), and two other rules. The Wells rule has been the most thoroughly evaluated: the most valid data (from four studies) suggest that 1–12% of people with low clinical (pre-test) probability actually have pulmonary embolism. Although there are some questions about its generalizability to non-hospital primary care populations, the Wells rule is likely to be valid and effective when used in these populations, provided it is used in conjunction with near-patient D-dimer testing (see Clinical prediction rules plus D-dimer testing). Other rules have limitations. The revised Geneva rule may be less effective in excluding pulmonary embolism than the Wells rule. The Pulmonary Embolism Rule-out Criteria (PERC) seem to be effective (when combined with clinical gestalt — a method of global interpretation), but have been evaluated in only one study and have not been evaluated in combination with D-dimer testing. The (original) Geneva rule and two other rules are not feasible for use in non-hospital primary care populations.

One systematic review of techniques for diagnosis of deep vein thrombosis and pulmonary embolism identified eight studies that evaluated clinical prediction rules for the diagnosis of pulmonary embolism [Segal et al, 2007]. Results were not pooled because of qualitative heterogeneity.

One systematic review compared the accuracy of pre-test probability assessment for pulmonary embolism in studies using clinical 'gestalt' (general interpretation) with studies using clinical prediction rules [Chunilal et al, 2003]. Sixteen studies involving 8306 people were included. There were no direct comparisons between clinical gestalt and clinical prediction rules. The authors found similar accuracy between clinical gestalt and clinical prediction rules, but advocated the use of a clinical prediction rule because it can be used by less experienced clinicians.

Wells rule

Evidence on the Wells rule for excluding pulmonary embolism

The Wells clinical prediction rule is used to score a person as having a low, moderate, or high clinical (pre–test) probability of pulmonary embolism (see Primary care management in Scenario: Managing suspected pulmonary embolism). The clinical probability can then be combined with a D-dimer test to increase the utility of the rule (see Clinical prediction rules plus D-dimer testing).

Studies of the Wells rule have reported that 1–28% of non-pregnant adults who were assessed as being low clinical probability actually had pulmonary embolism. The value of 28% is an outlier, and may be due to study bias; the remaining studies reported values of 1–12%. Although the Wells rule has not been validated in non-hospital populations and has one subjective criterion ('alternative diagnosis less likely' variable) which could limit reliability, the Wells rule is likely to be valid and effectiveness when used in non-hospital primary care populations, provided it is used in conjunction with D-dimer testing (see Clinical prediction rules plus D-dimer testing).

In the initial study relating to the Wells rule, a relatively complex algorithmic clinical model (of 40 variables) was derived from published literature and consensus [Wells et al, 1998]. This model was piloted in 91 people with suspected pulmonary embolism, refined, and then applied to 1239 consecutive, non-pregnant, adult inpatients and outpatients with suspected pulmonary embolism.

All participants underwent ventilation-perfusion scanning and lower limb bilateral compression ultrasonography. The results were interpreted by physicians who were blinded to the findings of the clinical assessment. Subsequent investigations included contrast venography and pulmonary angiography, which were performed on selected people who had inconclusive results from previous investigations. Participants were also followed up for 3 months to determine if pulmonary embolism had been missed.

Out of 734 people found to have a low pre-test probability, 25 (3.4%) had pulmonary embolism.

One possible limitation is the lack of application of a reference standard to all people with suspected pulmonary embolism, but this is offset by the use of a follow-up period.

A subsequent study derived the Wells rule by retrospectively applying regression analysis to the results of a randomly selected 80% of the initial study population (derivation group), in order to identify key variables [Wells et al, 2000]. The rule was then validated in the other 20% of the original study population (validation group). The methods for this study are discussed in Clinical prediction rules plus D-dimer testing.

The proportion of people with low clinical probability who had a pulmonary embolism was 3.6% in the derivation group and 2.0% in the validation group.

A potential limitation of the Wells rule is its lack of generalizability to non-hospital primary health care populations. In particular, one of the variables is whether 'an alternative diagnosis is less likely'. The initial derivation study used chest X-ray, electrocardiography, and arterial blood gases to help determine this [Wells et al, 1998]. Whilst these investigations are not always immediately available, their value is in excluding pulmonary embolism rather than in ruling it in. Their absence is likely to lead to overestimates of the prevalence of pulmonary embolism rather than underestimates, therefore making the rule no less safe for use in non-hospital primary health care. Furthermore, the prevalence of pulmonary embolism will be lower in non-hospital primary care than in emergency departments, and so the predictive value of any diagnostic approach will be higher.

One subsequent prospective study evaluated the Wells rule in consecutive, non-pregnant, adult inpatients and outpatients (who had no indications for thrombolytic treatment) in several Dutch hospitals [Sanson et al, 2000].

The reference standard was ventilation-perfusion scanning and, if the result of ventilation-perfusion scanning was non-diagnostic, computer tomographic pulmonary angiography.

Out of 627 people who gave their informed consent, 110 (18%) were excluded from the analysis because the diagnosis remained uncertain. Out of the remaining 517 people, complete information for the Wells rule was only available in 413 (80%).

The proportion of people with low clinical probability who had a pulmonary embolism was 28% and the likelihood ratio for this test was 0.93 (95% CI 0.69 to 1.24), indicating a very low utility.

The study findings may have been biased by the high rates of people excluded from the final analysis (34%).

In one subsequent study, 1.3% of people with low clinical probability had a pulmonary embolism [Wells et al, 2001]. The methods for this study are discussed in Clinical prediction rules plus D-dimer testing.

In one study comparing the Wells rule with the Geneva rule, 19 out of 162 people with low clinical probability had pulmonary embolism after extensive investigation or during 3-month follow up (12%, 95% CI 7% to 17%) [Chagnon et al, 2002].

Revised Wells rule

Evidence on the revised Wells rule for excluding pulmonary embolism

The revised Wells clinical prediction rule has just two clinical probability categories: 'unlikely' (a score of 4 or less) or 'likely' (a score greater than 4) — see Primary care management in Scenario: Managing suspected pulmonary embolism for the scores of individual criteria. Whilst it seems to be effective at excluding pulmonary embolism, it is slightly less effective and has not been as thoroughly evaluated as the Wells rule itself.

In one study, which derived and first validated this dichotomous scoring system, the proportion of people designated as unlikely to have a pulmonary embolism who actually had a pulmonary embolism was 7.8% in the derivation group and 5.1% in the validation group [Wells et al, 2000]. The methods for this study are discussed in Clinical prediction rules plus D-dimer testing.

One study (originally of a more complex diagnostic strategy) retrospectively compared the revised Wells rule with the original Wells rule (which divided people into low, moderate, and high clinical probability of pulmonary embolism) [Righini et al, 2006].

The comparison was performed on a database of 965 consecutive outpatients with suspected pulmonary embolism, of whom data was available for both scores in 922 people. The reference standard for low or intermediate (or unlikely) probability was a negative D-dimer test, proximal lower limb ultrasonography, or helical computed tomography, or follow up lasting 3 months.

Pulmonary embolism occurred in 10% of people with low clinical probability (using the original Wells score) and in 13% of people designated as unlikely to have a pulmonary embolism (using the revised Wells rule).

It is unclear whether practitioners who were interpreting imaging results were blinded to the findings of clinical scoring. The study is also limited by being retrospective.

Geneva and revised Geneva rules

Evidence on the Geneva and revised Geneva rules for excluding pulmonary embolism

The Geneva and revised Geneva clinical prediction rules are used to score a person as low, intermediate, or high clinical (pre-test) probability of having a pulmonary embolism. Only the revised Geneva score is potentially feasible for use in non-hospital primary care settings. The proportion of people with low clinical probability by the revised Geneva rule who actually had pulmonary embolism was 8% in two studies.

Several studies [Perrier et al, 2004; Righini et al, 2004] have evaluated the Geneva rule since it was developed by means of a retrospective study [Wicki et al, 2001]. These studies are not described in detail here because some investigations needed to make up the Geneva score cannot be done in primary healthcare (including arterial blood gases).

One subsequent prospective study derived the revised Geneva rule, which relies on clinical variables only [Le Gal et al, 2006].

The study enrolled 965 people with suspected non-massive pulmonary embolism who consecutively attended three hospital emergency departments.

Several diagnostic investigations were used to confirm or refute the diagnosis, but a standardized diagnostic strategy was applied to all people. This included follow up for 3 months.

Logistic regression analysis was used to identify clinical variables statistically associated with pulmonary embolism. A revised Geneva score therefore includes:

Age older than 65 years (1 point).

Previous DVT or pulmonary embolism (3 points).

Surgery (under general anaesthetic) or fracture (of the lower limbs) within the past 1 month (2 points).

Active cancer (2 points).

Unilateral lower limb pain (3 points).

Haemoptysis (2 points).

Heart rate 75–94 beats per minute (3 points) or heart rate 95 beats/minute or greater (5 points).

Pain on lower limb deep venous palpation and unilateral oedema (4 points).

The clinical probability is low if the person scores 0–3 points, intermediate with 4–10 points, and high with 11 points or more.

The proportion of people with low clinical probability who had pulmonary embolism was 8%. Confidence intervals were not given.

One subsequent prospective study compared the revised Geneva rule with the Wells rule [Klok et al, 2008].

In 300 people with suspected pulmonary embolism attending a Dutch hospital consecutively, the clinical probability of pulmonary embolism was assessed prospectively using the Wells rule and retrospectively using the revised Geneva rule.

The proportion of people with low clinical probability who had pulmonary embolism was: 8.3% (95% CI 4.0 to 12.7) using the revised Geneva rule and 4.5% (95% CI 1.7 to 9.6) using the Wells rule. The area under the receiver-operating characteristic curve (another method of assessing diagnostic utility) was similar for the two methods (p = 0.1).

The study is limited by having a retrospective component.

Other clinical prediction rules

Evidence on other clinical prediction rules for excluding pulmonary embolism

CKS identified three other clinical prediction rules. One is not feasible in non-hospital primary care, and one was found to have a failure rate that is probably too high (13.3%). The Pulmonary Embolism Rule-out Criteria (PERC) have a low false-negative rate and seem to be easy to use in primary care, but CKS identified no studies of PERC combined with D-dimer testing — Clinical prediction rules plus D-dimer testing.

One prospective study derived and validated a prediction score [Stöllberger et al, 2000]. This prediction score requires investigations that are not always immediately available in non-hospital primary healthcare (including chest X-ray and electrocardiography), and so is not described here.

One prospective study developed a decision rule to identify people with clinically suspected pulmonary embolism who were high risk, and those who were low risk in whom it was safe to use D-dimer testing to exclude pulmonary embolism [Kline et al, 2002].

The main criterion for inclusion in the study was suspicion of pulmonary embolism sufficient for an emergency physician to request pulmonary vascular imaging. A total of 934 people participated in the study.

The reference standard involved a hierarchy of autopsy and various methods of imaging.

The assessors of the definitive diagnostic tests were 'unaware of the decision rule'.

After logistic regression, six variables were found to be significantly associated with the presence of pulmonary embolism. The decision rule determined a person 'unsafe' for exclusion of pulmonary embolism by D-dimer testing if the person had a ratio of heart rate (beats per minute) to systolic blood pressure (mmHg) of less than 1 or was older than 50 years of age, and had one of the following: unexplained oxygen saturation of less than 95% (non-smoker, no respiratory disease), unilateral leg swelling, recent surgery, or haemoptysis.

After exclusion of the 21% of the sample who were deemed 'unsafe' for exclusion by D-dimer testing, the pre-test probability of pulmonary embolism was 13.3% in the remaining sample (compared with 19.4% in the whole sample).

Study limitations were:

That the decision rule was derived from and validated in the same sample of people.

That enrolment was on a convenience basis rather than being consecutive patients.

One prospective study evaluated previously developed PERC, combined with clinical gestalt (a method of global interpretation) in 8138 people [Kline et al, 2008].

The sample of people were enrolled consecutively or randomly in some centres, and by convenience sampling in others.

The PERC required the person to have age less than 50 years, pulse rate lower than 100 beats/minute, oxygen saturation on air of 95% or greater, no haemoptysis, no exogenous oestrogen, no previous venous thromboembolism, no recent surgery or trauma (requiring hospitalization in the previous 4 weeks), and no unilateral leg swelling ('PERC negative').

Low suspicion also required a clinical gestalt pre-test probability of less than 15%.

Low clinical gestalt suspicion and PERC negative occurred in 1666 people, of whom 16 had a venous thromboembolism (false-negative rate 1%, sensitivity 97.4%, specificity 21.9%).

Clinical prediction rules plus D-dimer testing

Evidence on clinical prediction rules plus D-dimer testing for excluding pulmonary embolism

CKS identified five systematic reviews of diagnostic strategies combining low clinical (pre-test) probability with D-dimer testing to exclude pulmonary embolism. Although some of their inclusion criteria differed, resulting in differences in included studies, all found that low clinical probability combined with a negative D-dimer test were effective at ruling out pulmonary embolism in people who were not pregnant: the pooled prevalence of pulmonary embolism was 0.14–0.50%. This is likely to be better than with the most sensitive imaging (for example computed tomographic pulmonary angiography).

Two systematic reviews and two other studies of the Wells clinical prediction rule (combined with D-dimer testing) have been summarized here because this is the most thoroughly validated of the clinical prediction rules, although there are some issues about its generalizability to non-hospital primary healthcare (see Wells rule). Using the Wells rule, 0.2–2.7% of people designated low clinical probability were found to have pulmonary embolism. Urgent D-dimer tests may not be available outside of hospital settings, although the availability of near-patient D-dimer testing is increasing.

Wells clinical prediction rule plus D-dimer testing

CKS identified two systematic reviews. Two primary studies that did not meet the inclusion criteria of the most recent review (although one was included in the earlier systematic review) are also summarized here.

One systematic review of diagnostic strategies for excluding pulmonary embolism (in clinical outcome studies) identified five studies that used a combination of clinical probability and D-dimer testing [Kruip et al, 2003].

Amongst other criteria, to be included, studies had to:

Be prospective and involve consecutive patients.

Withhold anticoagulant treatment when pulmonary embolism was excluded.

Have a minimum follow up of 3 months (with fewer than 10% of people lost to follow up).

All of the identified studies evaluated the Wells rule combined with D-dimer testing. Four out of five studies evaluated low clinical probability with a normal (negative) D-dimer test in 894 people; one evaluated low to moderate clinical probability (with a normal D-dimer test).

The failure rate of normal D-dimer levels combined with low clinical probability was only 0.2% (upper 95% CI 0.8%), leading the authors to conclude that this strategy safely excludes pulmonary embolism.

One systematic review identified four prospective studies (including 1660 consecutive patients) in which anticoagulant treatment was withheld from people with clinically suspected pulmonary embolism which was 'unlikely' using the revised Wells clinical decision rule, who had normal D-dimer concentration, and in whom no further tests were performed [Pasha et al, 2010].

The quality of the included studies was judged to be high.

Follow up lasted 3 months in each study. The loss to follow up was very low. Different D-dimer tests, with varying accuracies, were used.

Meta-analysis was performed to determine the negative predictive value (NPV) of this strategy. A random-effects model was used. Moderate heterogeneity was suggested (I2 46%, 95% CI 0% to 81%).

The pooled incidence of venous thromboembolism after initial exclusion of acute pulmonary embolism based on a pulmonary embolism scoring of 'unlikely' and a negative D-dimer result was 0.34% (95% CI 0.04 to 0.96%), resulting in an NPV of 99.7% (95% CI 99.0% to 99.9%).

The authors concluded that pulmonary embolism can be safely excluded in patients with clinically suspected pulmonary embolism who have an 'unlikely' probability and a negative D-dimer test, and anticoagulant treatment can be withheld in these people.

There are methodological issues regarding the pooling of NPVs, owing to the effect of prevalence. This may explain the moderate heterogeneity.

One retrospective study sought to develop a scoring system (the Wells clinical prediction rule) that, when combined with D-dimer testing, would safely exclude pulmonary embolism in a large proportion of people in whom pulmonary embolism was suspected [Wells et al, 2000].

The scoring system was derived by retrospectively performing a logistic regression analysis on 40 clinical variables, that had formed part of a clinical prediction model in a previous study [Wells et al, 1998], in order to identify key variables. A randomly selected sample of 80% of the previous study population served as the derivation group, and the rule was then validated in the remaining 20% (validation group). The methods of the previous study are described in Wells rule.

The proportion of people with low clinical probability who had a pulmonary embolism was 3.6% in the derivation group and 2.0% in the validation group.

When combined with a negative D-dimer result, the proportion of people with low clinical probability who had a pulmonary embolism was 1.5% (4 out of 276) in the derivation group and 2.7% (2 out of 73) in the validation group.

When combined with a negative D-dimer result, the proportion of people designated as 'unlikely to have a pulmonary embolism' who in fact had a pulmonary embolism was 2.2% (10 out of 448) in the derivation group and 1.7% (2 out of 118) in the validation group.

This study is limited by the validation being retrospective, and by the limitations of the previous study. Around 50 people from the original sample of 1260 people did not have any D-dimer testing.

One prospective study assessed the safety of using the Wells clinical prediction rule combined with D-dimer testing to manage people presenting to the emergency department with suspected pulmonary embolism [Wells et al, 2001].

Participants were 930 consecutive, non-pregnant adults presenting to emergency departments in Canada in whom pulmonary embolism was suspected because of acute onset of new or worsening shortness of breath or chest pain.

The SimpliRED D-dimer test was used, which has a higher specificity and lower sensitivity than enzyme-linked immunoabsorbent assay D-dimer tests. This may affect the generalizability of the findings.

No imaging was done in people assigned a low clinical (pre-test) probability and a negative D-dimer result. These people, in whom pulmonary embolism was excluded, were followed up for 3 months to determine any false-negative results; follow up essentially served as the reference standard.

Including follow up, pulmonary embolism was diagnosed in 7 (1.3%) of 527 people with low clinical probability.

Of the 437 people with low clinical probability and a negative D-dimer result, only one person developed pulmonary embolism during follow up; the NPV of this strategy was stated as 99.5% (95% CI 99.1% to 100%), although the figures suggest an even higher NPV.

Any clinical prediction rule plus D-dimer testing

One systematic review of techniques for diagnosis of deep vein thrombosis (DVT) and pulmonary embolism identified four studies which combined clinical pre-test probability with D-dimer testing to exclude pulmonary embolism [Segal et al, 2007]. Two of the studies evaluated the Wells rule and two evaluated the Geneva score, both combined with D-dimer testing. Results were not pooled because of qualitative heterogeneity. The authors did not report the results of studies of pulmonary embolism separately from those of DVT. However, they concluded that 'when a D-dimer assay is negative and a clinical prediction rule suggests a low probability of DVT or pulmonary embolism, the NPV is high enough to justify foregoing imaging studies in many patients'.

One systematic review of the VIDAS® D-dimer test (a highly sensitive D-dimer test) in combination with non-high clinical pre-test probability to rule out pulmonary embolism included seven studies (six prospective management studies and one randomized controlled trial) [Carrier et al, 2009].

Studies were included which reported failure rates at 3 months.

A total of 5622 people had non-high (low, intermediate, or unlikely) pre-test probability; a D-dimer test was negative in 2248 (40%) people in this group.

The risk of a thromboembolic event within 3 months in people with non-high pre-test probability and negative D-dimer test was 0.14% (95% CI 0.05% to 0.41%).

The authors concluded that a non-high pre-test probability combined with a negative VIDAS® D-dimer result effectively and safely excludes pulmonary embolism.

Search strategy

Scope of search

A literature search was conducted for guidelines, systematic reviews and randomized controlled trials on the primary care management of pulmonary embolism (PE).

Search dates

October 2010 - May 2013

Key search terms

Various combinations of searches were carried out. The terms listed below are the core search terms that were used for Medline.

exp Pulmonary Embolism/, pulmonary embol$.tw

Table 1 . Key to search terms.
Search commands Explanation
/ indicates a MeSh subject heading with all subheadings selected
.tw indicates a search for a term in the title or abstract
exp indicates that the MeSH subject heading was exploded to include the narrower, more specific terms beneath it in the MeSH tree
$ indicates that the search term was truncated (e.g. wart$ searches for wart and warts)
Sources of guidelines

National Institute for Health and Care Excellence (NICE)

Scottish Intercollegiate Guidelines Network (SIGN)

Royal College of Physicians

Royal College of General Practitioners

Royal College of Nursing

NICE Evidence

Health Protection Agency

World Health Organization

National Guidelines Clearinghouse

Guidelines International Network

TRIP database

GAIN

NHS Scotland National Patient Pathways

New Zealand Guidelines Group

Agency for Healthcare Research and Quality

Institute for Clinical Systems Improvement

National Health and Medical Research Council (Australia)

Royal Australian College of General Practitioners

British Columbia Medical Association

Canadian Medical Association

Alberta Medical Association

University of Michigan Medical School

Michigan Quality Improvement Consortium

Singapore Ministry of Health

National Resource for Infection Control

Patient UK Guideline links

UK Ambulance Service Clinical Practice Guidelines

RefHELP NHS Lothian Referral Guidelines

Medline (with guideline filter)

Driver and Vehicle Licensing Agency

NHS Health at Work (occupational health practice)

Sources of systematic reviews and meta-analyses

The Cochrane Library :

Systematic reviews

Protocols

Database of Abstracts of Reviews of Effects

Medline (with systematic review filter)

EMBASE (with systematic review filter)

Sources of health technology assessments and economic appraisals

NIHR Health Technology Assessment programme

The Cochrane Library :

NHS Economic Evaluations

Health Technology Assessments

Canadian Agency for Drugs and Technologies in Health

International Network of Agencies for Health Technology Assessment

Sources of randomized controlled trials

The Cochrane Library :

Central Register of Controlled Trials

Medline (with randomized controlled trial filter)

EMBASE (with randomized controlled trial filter)

Sources of evidence based reviews and evidence summaries

Bandolier

Drug & Therapeutics Bulletin

TRIP database

Central Services Agency COMPASS Therapeutic Notes

Sources of national policy

Department of Health

Health Management Information Consortium (HMIC)

Patient experiences

Healthtalkonline

BMJ - Patient Journeys

Patient.co.uk - Patient Support Groups

Sources of medicines information

The following sources are used by CKS pharmacists and are not necessarily searched by CKS information specialists for all topics. Some of these resources are not freely available and require subscriptions to access content.

British National Formulary (BNF)

electronic Medicines Compendium (eMC)

European Medicines Agency (EMEA)

LactMed

Medicines and Healthcare products Regulatory Agency (MHRA)

REPROTOX

Scottish Medicines Consortium

Stockley's Drug Interactions

TERIS

TOXBASE

Micromedex

UK Medicines Information

References

ABPI Medicines Compendium (2011) Summary of product characteristics for Innohep 10, 000 IU/ml and Innohep Syringe 10, 000 IU/ml. Electronic Medicines CompendiumDatapharm Communications Ltd. www.medicines.org.uk [Free Full-text]

ABPI Medicines Compendium (2013a) Summary of product characteristics for Arixtra 2.5mg/0.5ml solution for injection, pre-filled syringe. Electronic Medicines CompendiumDatapharm Communications Ltd. www.medicines.org.uk [Free Full-text]

ABPI Medicines Compendium (2013b) Summary of product characteristics for Fragmin 10, 000 IU/0.4ml solution for injection. Electronic Medicines CompendiumDatapharm Communications Ltd. www.medicines.org.uk [Free Full-text]

ABPI Medicines Compendium (2013c) Summary of product characteristics for Clexane pre-filled syringes. Electronic Medicines CompendiumDatapharm Communications Ltd. www.medicines.org.uk [Free Full-text]

Akram, A.R., Cowell, G.W., Logan, L.J. et al. (2009) Clinically suspected acute pulmonary embolism: a comparison of presentation, radiological features and outcome in patients with and without PE. QJM 102(6), 407-414. [Abstract] [Free Full-text]

Blondon, M., Bounameaux, H. and Righini, M. (2009) Treatment strategies for acute pulmonary embolism. Expert Opinion on Pharmacotherapy 10(7), 1159-1171. [Abstract]

BNF 65 (2013) British National Formulary. 65th edn. London: British Medical Association and Royal Pharmaceutical Society of Great Britain.

Bourjeily, G., Paidas, M., Khalil, H. et al. (2010) Pulmonary embolism in pregnancy. Lancet 375(9713), 500-512. [Abstract]

Brims, F.J., Davies, H.E. and Lee, Y.C. (2010) Respiratory chest pain: diagnosis and treatment. Medical Clinics of North America 94(2), 217-232. [Abstract]

BTS (1997) Suspected acute pulmonary embolism: a practical approach. British Thoracic Society, Standards of Care Committee. Thorax 52(Suppl 4), . [Free Full-text]

BTS (2003) British Thoracic Society guidelines for the management of suspected acute pulmonary embolism. Thorax 58(6), 470-483. [Free Full-text]

Camm, A.J. and Bunce, N.H. (2005) Cardiovascular disease. In: Kumar, P. and Clark, M. (Eds.) Kumar & Clark Clinical Medicine. 6th edn. London: Elsevier Saunders. 725-871.

Carrier, M., Righini, M., Djurabi, R.K. et al. (2009) VIDAS D-dimer in combination with clinical pre-test probability to rule out pulmonary embolism. A systematic review of management outcome studies. Thrombosis and Haemostasis 101(5), 886-892. [Abstract]

CEMACH (2007) Saving mothers' lives: reviewing maternal deaths to make motherhood safer - 2003-2005. Confidential Enquiry into Maternal and Child Health.

Chagnon, I., Bounameaux, H., Aujesky, D. et al. (2002) Comparison of two clinical prediction rules and implicit assessment among patients with suspected pulmonary embolism. American Journal of Medicine 113(4), 269-275. [Abstract]

Chunilal, S.D., Eikelboom, J.W., Attia, J. et al. (2003) Does this patient have pulmonary embolism? Journal of The American Medical Association 290(21), 2849-2858. [Abstract]

Cohen, A.T., Agnelli, G., Anderson, F.A. et al. (2007) Venous thromboembolism (VTE) in Europe. The number of VTE events and associated morbidity and mortality. Thrombosis and Haemostasis 98(4), 756-764. [Abstract]

Conget, F., Otero, R., Jimenez, D. et al. (2008) Short-term clinical outcome after acute symptomatic pulmonary embolism. Thrombosis and Haemostasis 100(5), 937-942. [Abstract]

Dong, B.R., Hao, Q., Yue, J. et al. (2009) Thrombolytic therapy for pulmonary embolism (Cochrane Review). The Cochrane Library. Issue 3. John Wiley & Sons, Ltd. www.thecochranelibrary.com [Free Full-text]

Donzé, J., Le, G.al, G., Fine, M.J. et al. (2008) Prospective validation of the Pulmonary Embolism Severity Index. A clinical prognostic model for pulmonary embolism. Thrombosis and Haemostasis 100(5), 943-948. [Abstract]

Douketis, J.D., Kearon, C., Bates, S. et al. (1998) Risk of fatal pulmonary embolism in patients with treated venous thromboembolism. Journal of the American Medical Association 279(6), 458-462. [Abstract] [Free Full-text]

Douketis, J.D., Gu, C.S., Schulman, S. et al. (2007) The risk for fatal pulmonary embolism after discontinuing anticoagulant therapy for venous thromboembolism. Annals of Internal Medicine 147(11), 766-774. [Abstract]

Forgione, A.M. (2006) Managing patients with suspected pulmonary embolism. Journal of The American Academy of Physician Assistants 19(7), 22-28.

Hall, W.B., Truitt, S.G., Scheunemann, L.P. et al. (2009) The prevalence of clinically relevant incidental findings on chest computed tomographic angiograms ordered to diagnose pulmonary embolism. Archives of Internal Medicine 169(21), 1961-1965. [Abstract] [Free Full-text]

Huerta, C., Johansson, S., Wallander, M.A. and Garcia Rodriguez, L.A. (2007) Risk factors and short-term mortality of venous thromboembolism diagnosed in the primary care setting in the United Kingdom. Archives of Internal Medicine 167(9), 935-943. [Abstract] [Free Full-text]

Jimenez, D., Aujesky, D., Diaz, G. et al. (2010) Prognostic significance of deep vein thrombosis in patients presenting with acute symptomatic pulmonary embolism. American Journal of Respiratory and Critical Care Medicine 181(9), 983-991. [Abstract]

Keeling, D., Baglin, T., Tait, C. et al. (2011) Guidelines on oral anticoagulation with warfarin - fourth edition. British Journal of Haematology 154(3), 311-324. [Free Full-text]

Kline, J.A., Nelson, R.D., Jackson, R.E. and Courtney, D.M. (2002) Criteria for the safe use of D-dimer testing in emergency department patients with suspected pulmonary embolism: a multicenter US study. Annals of Emergency Medicine 39(2), 144-152. [Abstract]

Kline, J.A., Courtney, D.M., Kabrhel, C. et al. (2008) Prospective multicenter evaluation of the pulmonary embolism rule-out criteria. Journal of Thrombosis and Haemostasis 6(5), 772-780. [Abstract]

Klok, F.A., Kruisman, E., Spaan, J. et al. (2008) Comparison of the revised Geneva score with the Wells rule for assessing clinical probability of pulmonary embolism. Journal of Thrombosis and Haemostasis 6(1), 40-44. [Abstract]

Kruip, M.J., Leclercq, M.G., van der Heul, C. et al. (2003) Diagnostic strategies for excluding pulmonary embolism in clinical outcome studies A systematic review. Annals of Internal Medicine 138(12), 941-951. [Abstract]

Kucher, N., Rossi, E., De Rosa, M. and Goldhaber, S.Z. (2006) Massive pulmonary embolism. Circulation 113(4), 577-582. [Abstract] [Free Full-text]

Le Gal, G., Testuz, A., Righini, M. et al. (2005) Reproduction of chest pain by palpation: diagnostic accuracy in suspected pulmonary embolism. BMJ 330(7489), 452-453. [Free Full-text]

Le Gal, G., Righini, M., Roy, P.M. et al. (2006) Prediction of pulmonary embolism in the emergency department: the revised Geneva score. Annals of Internal Medicine 144(3), 165-171. [Abstract]

Meyer, G., Roy, P.M., Gilberg, S. and Perrier, A. (2010) Easily missed? Pulmonary embolism. BMJ 340(), c1421.

National Clinical Guideline Centre (2012) Venous thromboembolic disease: the management of venous thromboembolic diseases and the role of thrombophilia testing (full NICE guideline). Clinical guidance 144National Centre for Health and Clinical Excellence. www.nice.org.uk [Free Full-text]

NICE (2012) Venous thromboembolic diseases: the management of venous thromboembolic diseases and the role of thrombophilia testing (NICE guideline). National Institute for Health and Care Excellence. www.nice.org.uk [Free Full-text]

Palla, A., Petruzzelli, S., Donnamaria, V. and Giuntini, C. (1995) The role of suspicion in the diagnosis of pulmonary embolism. Chest 107(1 Suppl), 21S-24S. [Abstract] [Free Full-text]

Pasha, S.M., Klok, F.A., Snoep, J.D. et al. (2010) Safety of excluding acute pulmonary embolism based on an unlikely clinical probability by the Wells rule and normal D-dimer concentration: a meta-analysis. Thrombosis Research 125(4), e123-e127. [Abstract]

Perrier, A., Roy, P.M., Aujesky, D. et al. (2004) Diagnosing pulmonary embolism in outpatients with clinical assessment, D-dimer measurement, venous ultrasound, and helical computed tomography: a multicenter management study. American Journal of Medicine 116(5), 291-299. [Abstract]

RCOG (2010) The acute management of thrombosis and embolism during pregnancy and the puerperium. . 2nd edn. Royal College of Obstetrics and Gynaecologists. www.rcog.org.uk [Free Full-text]

Righini, M., Aujesky, D., Roy, P.M. et al. (2004) Clinical usefulness of D-dimer depending on clinical probability and cutoff value in outpatients with suspected pulmonary embolism. Archives of Internal Medicine 164(22), 2483-2487. [Abstract] [Free Full-text]

Righini, M., LE, G.al, G., Perrier, A. and Bounameaux, H. (2006) Clinical probability assessment of pulmonary embolism by the Wells' score: is the easiest the best? Journal of Thrombosis and Haemostasis 4(3), 702-704.

Royal College of Physicians (2012) National Early Warning Score (NEWS). Standardising the assessment of acute-illness severity in the NHS. Royal College of Physicians. www.rcplondon.ac.uk [Free Full-text]

Sanson, B.J., Lijmer, J.G., Mac, G.illavry, M.R. et al. (2000) Comparison of a clinical probability estimate and two clinical models in patients with suspected pulmonary embolism. ANTELOPE-Study Group. Thrombosis and Haemostasis 83(2), 199-203. [Abstract]

Sarasin, F.P., Louis-Simonet, M., Carballo, D. et al. (2001) Prospective evaluation of patients with syncope: a population-based study. American Journal of Medicine 111(3), 177-184. [Abstract]

Scarsbrook, A.F. and Gleeson, F.V. (2007) Investigating suspected pulmonary embolism in pregnancy. BMJ 334(7590), 418-419. [Free Full-text]

Segal, J.B., Eng, J., Tamariz, L.J. and Bass, E.B. (2007) Review of the evidence on diagnosis of deep venous thrombosis and pulmonary embolism. Annals of Family Medicine 5(1), 63-73. [Abstract] [Free Full-text]

Stein, P.D. and Firth, J.D. (2010) Venous thromboembolism. In: Warrell, D.A., Cox, T.M. and Firth, J.D. (Eds.) Oxford textbook of medicine. 5th edn. Oxford: Oxford University Press. 16.16

Stein, P.D., Willis, P.W., III and DeMets, D.L. (1981) History and physical examination in acute pulmonary embolism in patients without preexisting cardiac or pulmonary disease. American Journal of Cardiology 47(2), 218-223. [Abstract]

Stein, P.D., Terrin, M.L., Hales, C.A. et al. (1991) Clinical, laboratory, roentgenographic, and electrocardiographic findings in patients with acute pulmonary embolism and no pre-existing cardiac or pulmonary disease. Chest 100(3), 598-603. [Abstract] [Free Full-text]

Stein, P.D., Beemath, A., Matta, F. et al. (2007) Clinical characteristics of patients with acute pulmonary embolism: data from PIOPED II. American Journal of Medicine 120(10), 871-879. [Abstract] [Free Full-text]

Stöllberger, C., Finsterer, J., Lutz, W. et al. (2000) Multivariate analysis-based prediction rule for pulmonary embolism. Thrombosis Research 97(5), 267-273. [Abstract]

Sweetland, S., Green, J., Liu, B. et al. (2009) Duration and magnitude of the postoperative risk of venous thromboembolism in middle aged women: prospective cohort study. BMJ 339(b4583), . [Abstract] [Free Full-text]

Tapson, V.F. (2008) Acute pulmonary embolism. New England Journal of Medicine 358(10), 1037-1052.

Torbicki, A., Perrier, A., Konstantinides, S. et al. (2008) Guidelines on the diagnosis and management of acute pulmonary embolism: the Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC). European Heart Journal 29(18), 2276-2315. [Abstract] [Free Full-text]

UK Parliament (2010) Deaths: thrombosis. Daily Hansard - Written AnswersUnited Kingdom Parliament. www.publications.parliament.uk [Free Full-text]

van Dongen, C.J., Vink, R., Hutten, B.A. et al. (2003) The incidence of recurrent venous thromboembolism after treatment with vitamin K antagonists in relation to time since first event: a meta-analysis. Archives of Internal Medicine 163(11), 1285-1293. [Abstract] [Free Full-text]

Warburton, M.S., Jackson, M.A., Norton, R. and Bhabra, M. (2004) Rare causes of haemoptysis in suspected pulmonary embolism. BMJ 329(7465), 557-558. [Free Full-text]

Wells, P.S., Ginsberg, J.S., Anderson, D.R. et al. (1998) Use of a clinical model for safe management of patients with suspected pulmonary embolism. Annals of Internal Medicine 129(12), 997-1005. [Abstract]

Wells, P.S., Anderson, D.R., Rodger, M. et al. (2000) Derivation of a simple clinical model to categorize patients probability of pulmonary embolism: increasing the models utility with the SimpliRED D-dimer. Thrombosis and Haemostasis 83(3), 416-420. [Abstract]

Wells, P.S., Anderson, D.R., Rodger, M. et al. (2001) Excluding pulmonary embolism at the bedside without diagnostic imaging: management of patients with suspected pulmonary embolism presenting to the emergency department by using a simple clinical model and d-dimer. Annals of Internal Medicine 135(2), 98-107. [Abstract]

West, J., Goodacre, S. and Sampson, F. (2007) The value of clinical features in the diagnosis of acute pulmonary embolism: systematic review and meta-analysis. QJM 100(12), 763-769. [Abstract] [Free Full-text]

Whiting, P., Rutjes, A.W., Reitsma, J.B. et al. (2003) The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Medical Research Methodology 3(), 25. [Abstract] [Free Full-text]

Wicki, J., Perneger, T.V., Junod, A.F. et al. (2001) Assessing clinical probability of pulmonary embolism in the emergency ward: a simple score. Archives of Internal Medicine 161(1), 92-97. [Abstract] [Free Full-text]

Winters, M.E. and Katzen, S.M. (2006) Identifying chest pain emergencies in the primary care setting. Primary Care 33(3), 625-642. [Abstract]