2019 ESC Guidelines for the diagnosis and management of acute pulmonary embolism developed in collaboration with the European Respiratory Society (ERS)

2.1. Why do we need new Guidelines on the diagnosis and management of pulmonary embolism?

This document follows the previous ESC Guidelines focusing on the clinical management of pulmonary embolism (PE), published in 2000, 2008, and 2014. Many recommendations have been retained or their validity has been reinforced; however, new data have extended or modified our knowledge in respect of the optimal diagnosis, assessment, and treatment of patients with PE. These new aspects have been integrated into previous knowledge to suggest optimal and—whenever possible—objectively validated management strategies for patients with suspected or confirmed PE. To limit the length of the article text, additional information, tables, figures, and references are available as supplementary data.

These Guidelines focus on the diagnosis and management of acute PE in adult patients. For further details specifically related to the diagnosis and management of deep vein thrombosis (DVT), the reader is referred to the joint consensus document of the ESC Working Groups of Aorta and Peripheral Vascular Diseases, and Pulmonary Circulation and Right Ventricular Function [1].

2.2. What is new in the 2019 Guidelines?

2.2.2. Changes in recommendations between 2014 and 2019

2.2.3. Main new recommendations 2019

3.1. Epidemiology

Venous thromboembolism (VTE), clinically presenting as DVT or PE, is globally the third most frequent acute cardiovascular syndrome behind myocardial infarction and stroke [2]. In epidemiological studies, annual incidence rates for PE range from 39 to 115 per 100 000 population; for DVT, incidence rates range from 53 to 162 per 100 000 population [3, 4]. Cross-sectional data show that the incidence of VTE is almost eight times higher in individuals aged ≥80 years than in the fifth decade of life [3]. In parallel, longitudinal studies have revealed a rising tendency in annual PE incidence rates [4–7] over time. Together with the substantial hospital-associated, preventable, and indirect annual expenditures for VTE (an estimated total of up to EUR 8.5 billion in the European Union) [8], these data demonstrate the importance of PE and DVT in ageing populations in Europe and other areas of the world. They further suggest that VTE will increasingly pose a burden on health systems worldwide in the years to come.

PE may cause ≤300 000 deaths per year in the US, ranking high among the causes of cardiovascular mortality [3]. In six European countries with a total population of 454.4 million, more than 370 000 deaths were related to VTE in 2004, as estimated on the basis of an epidemiological model [9]. Of these patients, 34% died suddenly or within a few hours of the acute event, before therapy could be initiated or take effect. Of the other patients, death resulted from acute PE that was diagnosed after death in 59% and only 7% of patients who died early were correctly diagnosed with PE before death [9].

Time trend analyses in European, Asian, and North American populations suggest that case fatality rates of acute PE may be decreasing [4–7, 10, 11]. Increased use of more effective therapies and interventions, and possibly better adherence to guidelines [12, 13], has most likely exerted a significant positive effect on the prognosis of PE in recent years. However, there is also a tendency towards overdiagnosis of (subsegmental or even non-existent) PE in the modern era [14], and this might in turn lead to a false drop in case fatality rates by inflating the denominator, i.e. the total number of PE cases.

Figure 1 summarizes the existing data on global trends in PE, highlighting increasing incidence rates in parallel with decreasing case fatality rates over an ∼15 year period.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Trends in annual incidence rates (left panel) and case fatality rates (right panel) of pulmonary embolism around the world, based on data retrieved from various references [5, 6, 11, 14–17]. Reproduced with permission from [18]. PE: pulmonary embolism; US : United States. ^aPE listed as principal diagnosis. ^bAny listed code for PE was considered.

In children, studies have reported an annual incidence of VTE of between 53 and 57 per 100 000 among hospitalized patients [19, 20], and between 1.4 and 4.9 per 100 000 in the community overall [21, 22].

3.2. Predisposing factors

There is an extensive collection of predisposing environmental and genetic factors for VTE; a list of predisposing (risk) factors is shown in table 3. VTE is considered to be a consequence of the interaction between patient-related—usually permanent—risk factors and setting-related—usually temporary—risk factors. Since categorization of temporary and permanent risk factors for VTE is important for assessing the risk of recurrence, and consequently for decision-making on chronic anticoagulation, it is discussed in more detail in section 8 of these Guidelines.

Major trauma, surgery, lower-limb fractures and joint replacements, and spinal cord injury are strong provoking factors for VTE [23, 24]. Cancer is a well-recognized predisposing factor for VTE. The risk of VTE varies with different types of cancer [25, 26]; pancreatic cancer, haematological malignancies, lung cancer, gastric cancer, and brain cancer carry the highest risk [27, 28]. Moreover, cancer is a strong risk factor for all-cause mortality following an episode of VTE [29].

Oestrogen-containing oral contraceptive agents are associated with an elevated VTE risk, and contraceptive use is the most frequent VTE risk factor in women of reproductive age [30–32]. More specifically, combined oral contraceptives (containing both an oestrogen and a progestogen) are associated with an approximately two- to six-fold increase in VTE risk over baseline [32, 33]. In general, the absolute VTE risk remains low in the majority of the >100 million combined oral contraceptive users worldwide [34]; however, VTE risk factors, including severe inherited thrombophilia (discussed in section 8) [35], increase this risk. Third-generation combined oral contraceptives, containing progestogens such as desogestrel or gestodene, are associated with a higher VTE risk than the second-generation combined oral contraceptives, which contain progestogens such as levonorgestrel or norgestrel [36, 37]. On the other hand, hormone-releasing intrauterine devices and some progesterone-only pills (used at contraceptive doses) are not associated with a significant increase in VTE risk [33, 38]; consequently, and following counselling and full risk assessment, these options are often proposed to women with a personal or strong family history of VTE.

In post-menopausal women who receive hormone replacement therapy, the risk of VTE varies widely depending on the formulation used [39].

Infection is a common trigger for VTE [23, 40, 41]. Blood transfusion and erythropoiesis-stimulating agents are also associated with an increased risk of VTE [23, 42].

In children, PE is usually associated with DVT and is rarely unprovoked. Serious chronic medical conditions and central venous lines are considered likely triggers of PE [43].

VTE may be viewed as part of the cardiovascular disease continuum, and common risk factors—such as cigarette smoking, obesity, hypercholesterolaemia, hypertension, and diabetes mellitus [44–47]—are shared with arterial disease, notably atherosclerosis [48–51]. However, this may be an indirect association mediated, at least in part, by the complications of coronary artery disease and, in the case of smoking, cancer [52, 53]. Myocardial infarction and heart failure increase the risk of PE [54, 55]. Conversely, patients with VTE have an increased risk of subsequent myocardial infarction and stroke, or peripheral arterial embolization [56].

3.3. Pathophysiology and determinants of outcome

Acute PE interferes with both circulation and gas exchange. Right ventricular (RV) failure due to acute pressure overload is considered the primary cause of death in severe PE. Pulmonary artery pressure (PAP) increases if >30–50% of the total cross-sectional area of the pulmonary arterial bed is occluded by thromboemboli [57]. PE-induced vasoconstriction, mediated by the release of thromboxane A2 and serotonin, contributes to the initial increase in pulmonary vascular resistance (PVR) after PE [58]. Anatomical obstruction and hypoxic vasoconstriction in the affected lung area lead to an increase in PVR, and a proportional decrease in arterial compliance [59].

The abrupt increase in PVR results in RV dilation, which alters the contractile properties of the RV myocardium via the Frank–Starling mechanism. The increase in RV pressure and volume leads to an increase in wall tension and myocyte stretch. The contraction time of the RV is prolonged, while neurohumoral activation leads to inotropic and chronotropic stimulation. Together with systemic vasoconstriction, these compensatory mechanisms increase PAP, improving flow through the obstructed pulmonary vascular bed and thus temporarily stabilizing systemic blood pressure (BP). However, the extent of immediate adaptation is limited, as a non-preconditioned, thin-walled RV is unable to generate a mean PAP>40 mmHg.

Prolongation of RV contraction time into early diastole in the left ventricle (LV) leads to leftward bowing of the interventricular septum [60]. The desynchronization of the ventricles may be exacerbated by the development of right bundle branch block. As a result, LV filling is impeded in early diastole, and this may lead to a reduction in the cardiac output (CO), and contribute to systemic hypotension and haemodynamic instability [61].

As described above, excessive neurohumoral activation in PE can be the result of both abnormal RV wall tension and circulatory shock. The finding of massive infiltrates of inflammatory cells in the RV myocardia of patients who died within 48 h of acute PE may be explained by high levels of epinephrine released as a result of the PE-induced “myocarditis” [62]. This inflammatory response might explain the secondary haemodynamic destabilization that sometimes occurs 24–48 h after acute PE, although early recurrence of PE may be an alternative explanation in some of these cases.

Finally, the association between elevated circulating levels of biomarkers of myocardial injury and an adverse early outcome indicates that RV ischaemia is of pathophysiological significance in the acute phase of PE [63, 64]. Although RV infarction is uncommon after PE, it is likely that the imbalance between oxygen supply and demand can result in damage to cardiomyocytes, and further reduce contractile forces. Systemic hypotension is a critical element in this process, leading to impairment of the coronary driving pressure to the overloaded RV.

The detrimental effects of acute PE on the RV myocardium and the circulation are summarized in Figure 2.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2

Key factors contributing to haemodynamic collapse and death in acute pulmonary embolism (modified from Konstantinides et al. [65] with permission). A-V: arterio-venous; BP: blood pressure; CO: cardiac output; LV - left ventricular; O2: oxygen; RV: right ventricular; TV: tricuspid valve. ^aThe exact sequence of events following the increase in RV afterload is not fully understood.

Respiratory failure in PE is predominantly a consequence of haemodynamic disturbances [66]. Low CO results in desaturation of the mixed venous blood. Zones of reduced flow in obstructed pulmonary arteries, combined with zones of overflow in the capillary bed served by non-obstructed pulmonary vessels, result in ventilation/perfusion mismatch, which contributes to hypoxaemia. In about one-third of patients, right-to-left shunting through a patent foramen ovale can be detected by echocardiography; this is caused by an inverted pressure gradient between the right atrium (RA) and left atrium, and may lead to severe hypoxaemia, and an increased risk of paradoxical embolization and stroke [67]. Finally, even if they do not affect haemodynamics, small distal emboli may create areas of alveolar haemorrhage resulting in haemoptysis, pleuritis, and pleural effusion, which is usually mild. This clinical presentation is known as “pulmonary infarction”. Its effect on gas exchange is normally mild, except in patients with pre-existing cardiorespiratory disease.

In view of the above pathophysiological considerations, acute RV failure, defined as a rapidly progressive syndrome with systemic congestion resulting from impaired RV filling and/or reduced RV flow output [68], is a critical determinant of clinical severity and outcome in acute PE. Accordingly, clinical symptoms, and signs of overt RV failure and haemodynamic instability, indicate a high risk of early (in-hospital or 30 day) mortality. High-risk PE is defined by haemodynamic instability and encompasses the forms of clinical presentation shown in table 4.

As an immediately life-threatening situation, high-risk PE requires an emergency diagnostic (upon suspicion) and therapeutic (upon confirmation or if the level of suspicion is sufficiently high) strategy, as outlined in section 7. However, the absence of haemodynamic instability does not exclude beginning (and possibly progressing) RV dysfunction, and thus an elevated PE-related early risk. In this large population, further assessment (outlined in sections 5 and 7) is necessary to determine the level of risk and adjust management decisions accordingly.

4.1. Clinical presentation

The clinical signs and symptoms of acute PE are non-specific. In most cases, PE is suspected in a patient with dyspnoea, chest pain, pre-syncope or syncope, or haemoptysis [73–75]. Haemodynamic instability is a rare but important form of clinical presentation, as it indicates central or extensive PE with severely reduced haemodynamic reserve. Syncope may occur, and is associated with a higher prevalence of haemodynamic instability and RV dysfunction [76]. Conversely, and according to the results of a recent study, acute PE may be a frequent finding in patients presenting with syncope (17%), even in the presence of an alternative explanation [77].

In some cases, PE may be asymptomatic or discovered incidentally during diagnostic workup for another disease.

Dyspnoea may be acute and severe in central PE; in small peripheral PE, it is often mild and may be transient. In patients with pre-existing heart failure or pulmonary disease, worsening dyspnoea may be the only symptom indicative of PE. Chest pain is a frequent symptom of PE and is usually caused by pleural irritation due to distal emboli causing pulmonary infarction [78]. In central PE, chest pain may have a typical angina character, possibly reflecting RV ischaemia, and requiring differential diagnosis from an acute coronary syndrome or aortic dissection.

In addition to symptoms, knowledge of the predisposing factors for VTE is important in determining the clinical probability of the disease, which increases with the number of predisposing factors present; however, in 40% of patients with PE, no predisposing factors are found [79]. Hypoxaemia is frequent, but ≤40% of patients have normal arterial oxygen saturation (S_aO2) and 20% have a normal alveolar–arterial oxygen gradient [80, 81]. Hypocapnia is also often present. A chest X-ray is frequently abnormal and, although its findings are usually non-specific in PE, it may be useful for excluding other causes of dyspnoea or chest pain [82]. Electrocardiographic changes indicative of RV strain—such as inversion of T waves in leads V1–V4, a QR pattern in V1, a S1Q3T3 pattern, and incomplete or complete right bundle branch block—are usually found in more severe cases of PE [83]; in milder cases, the only abnormality may be sinus tachycardia, present in 40% of patients. Finally, atrial arrhythmias, most frequently atrial fibrillation, may be associated with acute PE.

4.2. Assessment of clinical (pre-test) probability

The combination of symptoms and clinical findings with the presence of predisposing factors for VTE allows the classification of patients with suspected PE into distinct categories of clinical or pre-test probability, which correspond to an increasing actual prevalence of confirmed PE. This pre-test assessment can be done either by implicit (empirical) clinical judgement or by using prediction rules. As the post-test (i.e. after an imaging test) probability of PE depends not only on the characteristics of the diagnostic test itself but also on the pre-test probability, this is a key step in all diagnostic algorithms for PE.

The value of empirical clinical judgement has been confirmed in several large series [84, 85]. Clinical judgement usually includes commonplace tests such as chest X-rays and electrocardiograms for differential diagnosis. However, as clinical judgement lacks standardization, several explicit clinical prediction rules have been developed. Of these, the most frequently used prediction rules are the revised Geneva rule (table 5) and the Wells rule (supplementary table 1) [86]. Both prediction rules have been simplified in an attempt to increase their adoption into clinical practice [87, 88]; the simplified versions have been externally validated [89, 90].

Regardless of the score used, the proportion of patients with confirmed PE can be expected to be ∼10% in the low-probability category, 30% in the moderate-probability category, and 65% in the high-probability category [92]. When the two-level classification is used, the proportion of patients with confirmed PE is ∼12% in the PE-unlikely category and 30% in the PE-likely category [92]. A direct prospective comparison of these rules confirmed a similar diagnostic performance [89].

4.3. Avoiding overuse of diagnostic tests for pulmonary embolism

Searching for PE in every patient with dyspnoea or chest pain may lead to high costs and complications of unnecessary tests. The Pulmonary Embolism Rule-out Criteria (PERC) were developed for emergency department patients with the purpose of selecting, on clinical grounds, patients whose likelihood of having PE is so low that diagnostic workup should not even be initiated [93]. They comprise eight clinical variables significantly associated with an absence of PE: age <50 years; pulse <100 beats per minute; S_aO2 >94%; no unilateral leg swelling; no haemoptysis; no recent trauma or surgery; no history of VTE; and no oral hormone use. The results of a prospective validation study [94], and those of a randomized non-inferiority management study [95], suggested safe exclusion of PE in patients with low clinical probability who, in addition, met all criteria of the PERC rule. However, the low overall prevalence of PE in these studies [94, 95] does not support the generalizability of the results.

4.4. D-dimer testing

D-dimer levels are elevated in plasma in the presence of acute thrombosis because of simultaneous activation of coagulation and fibrinolysis. The negative predictive value of D-dimer testing is high, and a normal D-dimer level renders acute PE or DVT unlikely. On the other hand, the positive predictive value of elevated D-dimer levels is low and D-dimer testing is not useful for confirmation of PE. D-dimer is also more frequently elevated in patients with cancer [96, 97], in hospitalized patients [89, 98], in severe infection or inflammatory disease, and during pregnancy [99, 100]. Accordingly, the number of patients in whom D-dimer must be measured to exclude one PE (number needed to test) rises from 3 in the general population of an emergency department to ≥10 in the specific situations listed above.

As a number of D-dimer assays are available, clinicians should become aware of the diagnostic performance of the test used in their own hospital. The quantitative enzyme-linked immunosorbent assay (ELISA) or ELISA-derived assays have a diagnostic sensitivity of ≥95%, and can be used to exclude PE in patients with either low or intermediate pre-test probability. In the emergency department, a negative ELISA D-dimer can, in combination with clinical probability, exclude the disease without further testing in ∼30% of patients with suspected PE [101–103]. Outcome studies have shown that the 3 month thrombo-embolic risk was <1% in patients with low or intermediate clinical probability who were left untreated on the basis of a negative test result [104].

4.5. Computed tomographic pulmonary angiography

Multidetector CTPA is the method of choice for imaging the pulmonary vasculature in patients with suspected PE. It allows adequate visualization of the pulmonary arteries down to the subsegmental level [112–114]. The Prospective Investigation On Pulmonary Embolism Diagnosis (PIOPED) II study observed a sensitivity of 83% and a specificity of 96% for (mainly four-detector) CTPA in PE diagnosis [115]. PIOPED II also highlighted the influence of pre-test clinical probability on the predictive value of multidetector CTPA. In patients with a low or intermediate clinical probability of PE, a negative CTPA had a high negative predictive value for PE (96 and 89%, respectively), but its negative predictive value was only 60% if the pre-test probability was high. Conversely, the positive predictive value of a positive CTPA was high (92–96%) in patients with an intermediate or high clinical probability, but much lower (58%) in patients with a low pre-test likelihood of PE [115]. Therefore, clinicians should consider further testing in case of discordance between clinical judgement and the CTPA result.

Several studies have provided evidence in favour of CTPA as a stand-alone imaging test for excluding PE. Taken together, the available data suggest that a negative CTPA result is an adequate criterion for the exclusion of PE in patients with low or intermediate clinical probability of PE. On the other hand, it remains controversial whether patients with a negative CTPA and a high clinical probability should be further investigated.

Chronic thromboembolic pulmonary hypertension (CTEPH) is a potentially fatal late sequela of PE, but pre-existing CTEPH should not be missed in patients investigated for suspected acute PE. Signs of pre-existing CTEPH on CTPA are listed in supplementary table 2; the diagnosis and management of CTEPH is discussed in section 10.

The major strengths, weaknesses/limitations, and radiation issues related to the use of CTPA in the diagnosis of PE are summarized in table 6.

4.6. Lung scintigraphy

The planar ventilation/perfusion [V/Q (lung scintigraphy)] scan is an established diagnostic test for suspected PE. Perfusion scans are combined with ventilation studies, for which multiple tracers such as xenon-133 gas, krypton-81 gas, technetium-99m-labelled aerosols, or technetium-99m-labelled carbon microparticles (Technegas) can be used. The purpose of the ventilation scan is to increase specificity: in acute PE, ventilation is expected to be normal in hypoperfused segments (mismatched). Being a lower-radiation and contrast medium-sparing procedure, the V/Q scan may preferentially be applied in outpatients with a low clinical probability and a normal chest X-ray, in young (particularly female) patients, in pregnant women, in patients with history of contrast medium-induced anaphylaxis, and patients with severe renal failure [116].

Planar lung scan results are frequently classified according to the criteria established in the PIOPED study [117]. These criteria were the subject of debate and have been revised [118, 119]. To facilitate communication with clinicians, a three-tier classification is preferable: normal scan (excluding PE), high-probability scan (considered diagnostic of PE in most patients), and non-diagnostic scan [120–122]. Prospective clinical outcome studies suggested that it is safe to withhold anticoagulant therapy in patients with a normal perfusion scan. This was confirmed by a randomized trial comparing the V/Q scan with CTPA [122]. An analysis from the PIOPED II study suggested that a high-probability V/Q scan could confirm PE, although other sources suggest that the positive predictive value of a high-probability lung scan is not sufficient to confirm PE in patients with a low clinical probability [123, 124].

Performing only a perfusion scan might be acceptable in patients with a normal chest X-ray; any perfusion defect in this situation would be considered a mismatch. The high frequency of non-diagnostic scans is a limitation because they indicate the necessity for further diagnostic testing. Various strategies to overcome this problem have been proposed, notably the incorporation of clinical probability. Although the use of perfusion scanning and chest X-ray with the Prospective Investigative Study of Acute Pulmonary Embolism Diagnosis (PISAPED) criteria may be associated with a low rate of inconclusive results, the sensitivity appears too low to exclude PE and thus this approach may be less safe than CTPA [123, 125].

Several studies suggest that data acquisition in single-photon emission CT (SPECT) imaging, with or without low-dose CT, may decrease the proportion of non-diagnostic scans to as low as 0–5% [121, 126–128]. However, most studies reporting on the accuracy of SPECT are limited by their retrospective design [129, 130] or the inclusion of SPECT itself in the reference standard [127], and only one study used a validated diagnostic algorithm [131]. The diagnostic criteria for SPECT also varied; most studies defined PE as one or two subsegmental perfusion defects without ventilation defects, but these criteria are infrequently used in clinical practice. In addition, the optimal scanning technique (perfusion SPECT, V/Q SPECT, perfusion SPECT with non-enhanced CT, or V/Q SPECT with non-enhanced CT) remains to be defined. Finally, few outcome studies are available, and with incomplete follow-up [132]. Large-scale prospective studies are needed to validate SPECT techniques.

The major strengths, weaknesses/limitations, and radiation issues related to the use of V/Q scan and V/Q SPECT in the diagnosis of PE are summarized in table 6.

4.7. Pulmonary angiography

For several decades, pulmonary angiography was the “gold standard” for the diagnosis or exclusion of acute PE, but it is now rarely performed as less-invasive CTPA offers similar diagnostic accuracy [133]. The diagnosis of acute PE is based on direct evidence of a thrombus in two projections, either as a filling defect or as amputation of a pulmonary arterial branch [134]. Thrombi as small as 1–2 mm within the subsegmental arteries can be visualized by digital subtraction angiography, but there is substantial interobserver variability at this level [135, 136].

Pulmonary angiography is not free of risk. In a study of 1111 patients, procedure-related mortality was 0.5%, major non-fatal complications occurred in 1%, and minor complications in 5% [137]. The majority of deaths occurred in patients with haemodynamic compromise or respiratory failure. The amount of contrast agent should be reduced and non-selective injections avoided in patients with haemodynamic compromise [138].

The major strengths, weaknesses/limitations, and radiation issues related to the use of pulmonary angiography in the diagnosis of PE are summarized in table 6.

4.8. Magnetic resonance angiography

Magnetic resonance angiography (MRA) has been evaluated for several years regarding suspected PE. However, the results of large-scale studies [139, 140] show that this technique, although promising, is not yet ready for clinical practice due to its low sensitivity, the high proportion of inconclusive MRA scans, and its low availability in most emergency settings. The hypothesis that a negative MRA, combined with the absence of proximal DVT on compression ultrasonography (CUS), may safely rule out clinically significant PE is currently being investigated in an ongoing multicentre outcome study [Clinicaltrials.gov National Clinical Trial (NCT) number 02059551].

4.9. Echocardiography

Acute PE may lead to RV pressure overload and dysfunction, which can be detected by echocardiography. Given the peculiar geometry of the RV, there is no individual echocardiographic parameter that provides fast and reliable information on RV size or function. This is why echocardiographic criteria for the diagnosis of PE have differed between studies. Because of the reported negative predictive value of 40–50%, a negative result cannot exclude PE [124, 142, 143]. On the other hand, signs of RV overload or dysfunction may also be found in the absence of acute PE, and may be due to concomitant cardiac or respiratory disease [144].

Echocardiographic findings of RV overload and/or dysfunction are graphically presented in Figure 3. RV dilation is found in ≥25% of patients with PE on transthoracic echocardiography (TTE) and is useful for risk stratification of the disease [145]. More specific echocardiographic findings were reported to retain a high positive predictive value for PE even in the presence of pre-existing cardiorespiratory disease. Thus, the combination of a pulmonary ejection acceleration time (measured in the RV outflow tract) <60 ms with a peak systolic tricuspid valve gradient <60 mmHg (“60/60” sign), or with depressed contractility of the RV free wall compared to the “echocardiographic” RV apex (McConnell sign), is suggestive of PE [146]. However, these findings are present in only ∼12 and 20% of unselected PE patients, respectively [145]. Detection of echocardiographic signs of RV pressure overload helps to distinguish acute PE from RV free wall hypokinesia or akinesia due to RV infarction, which may mimic the McConnell sign [147]. It should be noted that in ∼10% of PE patients, echocardiography can show potentially misleading incidental findings such as significant LV systolic dysfunction or valvular heart disease [145]. Decreased tricuspid annular plane systolic excursion (TAPSE) may also be present in PE patients [148, 149]. Echocardiographic parameters of RV function derived from Doppler tissue imaging and wall strain assessment may also be affected by the presence of acute PE (figure 3). However, they probably have low sensitivity as stand-alone findings, as they were reported to be normal in haemodynamically stable patients despite the presence of PE [150, 151].

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3

Graphic representation of transthoracic echocardiographic parameters in the assessment of right ventricular pressure overload. A′: peak late diastolic (during atrial contraction) velocity of tricuspid annulus by tissue Doppler imaging; AcT: right ventricular outflow Doppler acceleration time; Ao: aorta; E′: peak early diastolic velocity of tricuspid annulus by tissue Doppler imaging; IVC: inferior vena cava; LA: left atrium; LV: left ventricle; RA: right atrium; RiHTh: right heart thrombus (or thrombi); RV: right ventricle/ventricular; S′: peak systolic velocity of tricuspid annulus by tissue Doppler imaging; TAPSE: tricuspid annular plane systolic excursion; TRPG: tricuspid valve peak systolic gradient.

Echocardiographic examination is not mandatory as part of the routine diagnostic workup in haemodynamically stable patients with suspected PE [124], although it may be useful in the differential diagnosis of acute dyspnoea. This is in contrast to suspected high-risk PE, in which the absence of echocardiographic signs of RV overload or dysfunction practically excludes PE as the cause of haemodynamic instability. In the latter case, echocardiography may be of further help in the differential diagnosis of the cause of shock, by detecting pericardial tamponade, acute valvular dysfunction, severe global or regional LV dysfunction, aortic dissection, or hypovolaemia [152]. Conversely, in a haemodynamically compromised patient with suspected PE, unequivocal signs of RV pressure overload, especially with more specific echocardiographic findings (60/60 sign, McConnell sign, or right-heart thrombi), justify emergency reperfusion treatment for PE if immediate CT angiography is not feasible in a patient with high clinical probability and no other obvious causes for RV pressure overload [152].

Mobile right-heart thrombi are detected by TTE or transoesophageal echocardiography (TOE), or by CT angiography, in <4% of unselected patients with PE [153–155]. Their prevalence may reach 18% among PE patients in the intensive care setting [156]. Mobile right-heart thrombi essentially confirm the diagnosis of PE and are associated with high early mortality, especially in patients with RV dysfunction [155, 157–159].

In some patients with suspected acute PE, echocardiography may detect increased RV wall thickness or tricuspid insufficiency jet velocity beyond values compatible with acute RV pressure overload (>3.8 m/s or a tricuspid valve peak systolic gradient >60 mmHg) [160]. In these cases, chronic thromboembolic (or other) pulmonary hypertension (PH) should be included in the differential diagnosis.

4.10. Compression ultrasonography

In the majority of cases, PE originates from DVT in a lower limb, and only rarely from upper-limb DVT (mostly following venous catheterization). In a study using venography, DVT was found in 70% of patients with proven PE [161]. Nowadays, lower-limb CUS has largely replaced venography for diagnosing DVT. CUS has a sensitivity >90% and a specificity of ∼95% for proximal symptomatic DVT [162, 163]. CUS shows a DVT in 30–50% of patients with PE [162–164], and finding a proximal DVT in patients suspected of having PE is considered sufficient to warrant anticoagulant treatment without further testing [165]. However, patients in whom PE is indirectly confirmed by the presence of a proximal DVT should undergo risk assessment for PE severity and the risk of early death.

In the setting of suspected PE, CUS can be limited to a simple four-point examination (bilateral groin and popliteal fossa). The only validated diagnostic criterion for DVT is incomplete compressibility of the vein, which indicates the presence of a clot, whereas flow measurements are unreliable. A positive proximal CUS result has a high positive predictive value for PE. The high diagnostic specificity (96%) along with a low sensitivity (41%) of CUS in this setting was shown by a recent meta-analysis [165, 166]. CUS is a useful procedure in the diagnostic strategy of patients with CT contraindications. The probability of a positive proximal CUS in suspected PE is higher in patients with signs and symptoms related to the leg veins than in asymptomatic patients [162, 163].

In patients admitted to the emergency department with haemodynamic instability and suspicion of PE, a combination of venous ultrasound with cardiac ultrasound may further increase specificity. Conversely, an echocardiogram without signs of RV dysfunction and a normal venous ultrasound excluded PE with a high (96%) negative predictive value in one study [167].

For further details on the diagnosis and management of DVT, the reader is referred to the joint consensus document of the ESC Working Groups of Aorta and Peripheral Vascular Diseases, and Pulmonary Circulation and Right Ventricular Function [1].

4.11. Recommendations for diagnosis

4.12. Computed tomography venography

When using CTPA, it is possible to image the deep veins of the legs during the same acquisition [115]. However, this approach has not been widely validated and the added value of venous imaging is limited [164]. Moreover, using CT venography is associated with increased radiation doses [168].

5.1. Clinical parameters of pulmonary embolism severity

Acute RV failure, defined as a rapidly progressive syndrome with systemic congestion resulting from impaired RV filling and/or reduced RV flow output [68], is a critical determinant of outcome in acute PE. Tachycardia, low systolic BP, respiratory insufficiency (tachypnoea and/or low S_aO2), and syncope, alone or in combination, have been associated with an unfavourable short-term prognosis in acute PE.

5.2. Imaging of right ventricular size and function

5.3. Laboratory biomarkers

5.4. Combined parameters and scores for assessment of pulmonary embolism severity

In patients who present without haemodynamic instability, individual baseline findings may not suffice to determine and further classify PE severity and PE-related early risk when used as stand-alone parameters. As a result, various combinations of the clinical, imaging, and laboratory parameters described above have been used to build prognostic scores, which permit a (semi)quantitative assessment of early PE-related risk of death. Of these, the Bova [218–221] and the H-FABP (or high-sensitivity troponin T), Syncope, Tachycardia (FAST) scores [219, 222, 223] have been validated in cohort studies (supplementary table 4). However, their implications for patient management remain unclear. To date, only a combination of RV dysfunction on an echocardiogram (or CTPA) with a positive cardiac troponin test has directly been tested as a guide for early therapeutic decisions (anticoagulation plus reperfusion treatment versus anticoagulation alone) in a large randomized controlled trial (RCT) of PE patients presenting without haemodynamic instability [224].

5.5. Integration of aggravating conditions and comorbidity into risk assessment of acute pulmonary embolism

In addition to the clinical, imaging, and laboratory findings, which are directly linked to PE severity and PE-related early death, baseline parameters related to aggravating conditions and comorbidity are necessary to assess a patient's overall mortality risk and early outcome. Of the clinical scores integrating PE severity and comorbidity, the Pulmonary Embolism Severity Index (PESI) (table 7) is the one that has been most extensively validated to date [225–228]. The principal strength of the PESI lies in the reliable identification of patients at low risk for 30 day mortality (PESI classes I and II). One randomized trial employed a low PESI as the principal inclusion criterion for home treatment of acute PE [178].

In view of the complexity of the original PESI, which includes 11 differently weighed variables, a simplified version (sPESI; table 7) has been developed and validated [229–231]. As with the original version of the PESI, the strength of the sPESI lies in the reliable identification of patients at low risk for 30 day mortality. The prognostic performance of the sPESI has been confirmed in observational cohort studies [227, 228], although this index has not yet been prospectively used to guide therapeutic management of low-risk PE patients.

The diagnosis of concomitant DVT has been identified as an adverse prognostic factor, being independently associated with death within the first 3 months after acute PE [232]. In a meta-analysis investigating 8859 patients with PE, the presence of concomitant DVT was confirmed as a predictor of 30 day all-cause mortality (OR 1.9, 95% CI 1.5–2.4), although it did not predict PE-related adverse outcomes at 90 days [233]. Thus, concomitant DVT can be regarded as an indicator of significant comorbidity in acute PE.

5.6. Prognostic assessment strategy

The classification of PE severity and the risk of early (in-hospital or 30 day) death is summarized in table 8. Risk assessment of acute PE begins upon suspicion of the disease and initiation of the diagnostic workup. At this early stage, it is critical to identify patients with (suspected) high-risk PE. This clinical setting necessitates an emergency diagnostic algorithm (figure 4) and immediate referral for reperfusion treatment, as explained in section 7, and displayed in figure 6 and supplementary figure 1. Testing for laboratory biomarkers such as cardiac troponins or natriuretic peptides is not necessary for immediate therapeutic decisions in patients with high-risk PE.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4

Diagnostic algorithm for patients with suspected high-risk pulmonary embolism presenting with haemodynamic instability. CTPA: computed tomography pulmonary angiography; CUS: compression ultrasonography; DVT: deep vein thrombosis; LV: left ventricle; PE: pulmonary embolism; RV: right ventricle; TOE: transoesophageal echocardiography; TTE: transthoracic echocardiogram. ^aSee table 4 for definition of haemodynamic instability and high-risk PE. ^bAncillary bedside imaging tests may include TOE, which may detect emboli in the pulmonary artery and its main branches; and bilateral venous CUS, which may confirm DVT and thus VTE. ^cIn the emergency situation of suspected high-risk PE, this refers mainly to a RV/LV diameter ratio >1.0; the echocardiographic findings of RV dysfunction, and the corresponding cut-off levels, are graphically presented in figure 3, and their prognostic value summarized in supplementary table 3. ^dIncludes the cases in which the patient's condition is so critical that it only allows bedside diagnostic tests. In such cases, echocardiographic findings of RV dysfunction confirm high-risk PE and emergency reperfusion therapy is recommended.

In the absence of haemodynamic instability at presentation, further risk stratification of PE is recommended, as it has implications for early discharge versus hospitalization or monitoring of the patient (explained in section 7). Table 8 provides an overview of the clinical, imaging, and laboratory parameters used to distinguish intermediate- and low-risk PE. The PESI is—in its original or simplified form—the most extensively validated and most broadly used clinical score to date, as it integrates baseline indicators of the severity of the acute PE episode with aggravating conditions and the comorbidity of the patient. Overall, a PESI of class I–II or an sPESI of 0 is a reliable predictor of low-risk PE.

In addition to clinical parameters, patients in the intermediate-risk group who display evidence of both RV dysfunction (on echocardiography or CTPA) and elevated cardiac biomarker levels in the circulation (particularly a positive cardiac troponin test) are classified into the intermediate−high-risk category. As will be discussed in more detail in section 7, close monitoring is recommended in these cases to permit the early detection of haemodynamic decompensation or collapse, and consequently the need for rescue reperfusion therapy [179]. Patients in whom the RV appears normal on echocardiography or CTPA, and/or who have normal cardiac biomarker levels, belong to the intermediate−low-risk category. As an alternative approach, use of further prognostic scores combining clinical, imaging, and laboratory parameters may be considered to semi-quantitatively assess the severity of the PE episode, and distinguish intermediate−high-risk and intermediate−low-risk PE. Supplementary table 4 lists the scores most frequently investigated for this purpose in observational (cohort) studies; however, none of them has been used in RCTs to date.

A recent meta-analysis included 21 cohort studies with a total of 3295 patients with “low-risk” PE based on a PESI of I–II or an sPESI of 0 [234]. Overall, 34% (95% CI 30–39%) of them were reported to have signs of RV dysfunction on echocardiography or CTPA. Data on early mortality were provided in seven studies (1597 patients) and revealed an OR of 4.19 (95% CI 1.39–12.58) for death from any cause in the presence of RV dysfunction; elevated cardiac troponin levels were associated with a comparable magnitude of risk elevation [234]. Early all-cause mortality rates (1.8% for RV dysfunction and 3.8% for elevated troponin levels [234]) were in the lower range of those previously reported for patients with intermediate-risk PE [235]. Until the clinical implications of such discrepancies are clarified, patients with signs of RV dysfunction or elevated cardiac biomarkers, despite a low PESI or an sPESI of 0, should be classified into the intermediate–low-risk category.

5.7. Recommendations for prognostic assessment

6.1. Haemodynamic and respiratory support

6.2. Initial anticoagulation

6.3. Reperfusion treatment

6.4. Multidisciplinary pulmonary embolism teams

The concept of multidisciplinary rapid-response teams for the management of “severe” (high-risk and selected cases of intermediate-risk) PE emerged in the USA, with increasing acceptance by the medical community and implementation in hospitals in Europe and worldwide. Set-up of PE response teams (PERTs) is encouraged, as they address the needs of modern systems-based healthcare [301]. A PERT brings together a team of specialists from different disciplines including, for example, cardiology, pulmonology, haematology, vascular medicine, anaesthesiology/intensive care, cardiothoracic surgery, and (interventional) radiology. The team convenes in real time (face-to-face or via web conference) to enhance clinical decision-making. This allows the formulation of a treatment plan and facilitates its immediate implementation [301]. The exact composition and operating mode of a PERT are not fixed, depending on the resources and expertise available in each hospital for the management of acute PE.

6.5. Vena cava filters

The aim of vena cava interruption is to mechanically prevent venous clots from reaching the pulmonary circulation. Most devices in current use are inserted percutaneously and can be retrieved after several weeks or months, or left in place over the long-term, if needed. Potential indications include VTE and absolute contraindication to anticoagulant treatment, recurrent PE despite adequate anticoagulation, and primary prophylaxis in patients with a high risk of VTE. Other potential indications for filter placement, including free-floating thrombi, have not been confirmed in patients without contraindications to therapeutic anticoagulation.

Only two phase III randomized trials have compared anticoagulation with or without vena cava interruption in patients with proximal DVT, with or without associated PE [302–304]. In the Prevention of Recurrent Pulmonary Embolism by Vena Cava Interruption (PREPIC) study, insertion of a permanent vena cava filter was associated with a significant reduction in the risk of recurrent PE and a significant increase in the risk of DVT, without a significant difference in the risk of recurrent VTE or death [303, 304]. The PREPIC-2 trial randomized 399 patients with PE and venous thrombosis to receive anticoagulant treatment, with or without a retrievable vena cava filter. In this study, the rate of recurrent VTE was low in both groups and did not differ between groups [302]. A systematic review and meta-analysis of published reports on the efficacy and safety of vena cava filters included 11 studies, with a total of 2055 patients who received a filter versus 2149 controls [305]. Vena cava filter placement was associated with a 50% decrease in the incidence of PE and an ∼70% increase in the risk of DVT over time. Neither all-cause mortality nor PE-related mortality differed between patients with or without filter placement.

The broad indication for placement of a venous filter in patients with recent (<1 month) proximal DVT and an absolute contraindication to anticoagulant treatment is based mainly on the perceived high risk of recurrent PE in this setting, and the lack of other treatment options.

Complications associated with vena cava filters are common and can be serious. A systematic literature review revealed penetration of the venous wall in 1699 (19%) of 9002 procedures; of these cases, 19% showed adjacent organ involvement and ≥8% were symptomatic [306]. Lethal complications were rare (only two cases), but 5% of the patients required major interventions such as surgical removal of the filter, endovascular stent placement or embolization, endovascular retrieval of the permanent filter, or percutaneous nephrostomy or ureteral stent placement [306]. Further reported complications include filter fracture and/or embolization, and DVT occasionally extending up to the vena cava [303, 307, 308].

6.6. Recommendations for acute-phase treatment of high-risk pulmonary embolism^a

6.7. Recommendations for acute-phase treatment of intermediate- or low-risk pulmonary embolism

6.8. Recommendations for multidisciplinary pulmonary embolism teams

6.9. Recommendations for inferior vena cava filters

6.10. Recommendations for early discharge and home treatment

7.1. Diagnostic strategies

Various combinations of clinical assessments, plasma D-dimer measurements, and imaging tests have been proposed and validated for PE diagnosis. These strategies have been tested in patients presenting with suspected PE in the emergency department or during their hospital stay [101, 164, 171, 320], and more recently in the primary care setting [111]. Withholding of anticoagulation without adherence to evidence-based diagnostic strategies was associated with a significant increase in the number of VTE episodes and sudden cardiac death at 3 month follow-up [12]. The most straightforward diagnostic algorithms for suspected PE—with and without haemodynamic instability—are presented in figures 4 and 5, respectively. However, it is recognized that the diagnostic approach for suspected PE may vary, depending on the availability of, and expertise in, specific tests in various hospitals and clinical settings.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5

Diagnostic algorithm for patients with suspected pulmonary embolism without haemodynamic instability. CTPA: computed tomography pulmonary angiography/angiogram; PE: pulmonary embolism. ^aThe proposed diagnostic strategy for pregnant women with suspected acute PE is discussed in section 9. ^bTwo alternative classification schemes may be used for clinical probability assessment, i.e. a three-level scheme (clinical probability defined as low, intermediate, or high) or a two-level scheme (PE unlikely or PE likely). When using a moderately sensitive assay, D-dimer measurement should be restricted to patients with low clinical probability or a PE-unlikely classification, while highly sensitive assays may also be used in patients with intermediate clinical probability of PE due to a higher sensitivity and negative predictive value. Note that plasma D-dimer measurement is of limited use in suspected PE occurring in hospitalized patients. ^cTreatment refers to anticoagulation treatment for PE. ^dCTPA is considered diagnostic of PE if it shows PE at the segmental or more proximal level. ^eIn case of a negative CTPA in patients with high clinical probability, investigation by further imaging tests may be considered before withholding PE-specific treatment.

The diagnostic strategy for suspected acute PE in pregnancy is discussed in section 9.

7.2. Treatment strategies

7.2.1. Emergency treatment of high-risk pulmonary embolism

The algorithm for a risk-adjusted therapeutic approach to acute PE is shown in figure 6; an emergency management algorithm specifically for patients with suspected acute high-risk PE is proposed in supplementary figure 1. Primary reperfusion treatment, in most cases systemic thrombolysis, is the treatment of choice for patients with high-risk PE. Surgical pulmonary embolectomy or percutaneous catheter-directed treatment are alternative reperfusion options in patients with contraindications to thrombolysis, if expertise with either of these methods and the appropriate resources are available on-site.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6

Central Illustration. Risk-adjusted management strategy for acute pulmonary embolism. CTPA: computed tomography pulmonary angiography/angiogram; PE: pulmonary embolism; PESI: Pulmonary Embolism Severity Index; RV: right ventricular; sPESI: simplified Pulmonary Embolism Severity Index; TTE: transthoracic echocardiogram. ^aSee also emergency management algorithm shown in the supplementary data. ^bRefer to table 8 for definition of high, intermediate-high-, intermediate-low-, and low-risk PE. ^cCancer, heart failure and chronic lung disease are included in the PESI and sPESI (table 7). ^dSee supplementary table 12 for the Hestia criteria. ^ePrognostically relevant imaging (TTE or CTPA) findings in patients with acute PE, are graphically presented in Figure 3. ^fA cardiac troponin test may already have been performed during initial diagnostic work-up. ^gIncluded in the Hestia criteria.

Following reperfusion treatment and haemodynamic stabilization, patients recovering from high-risk PE can be switched from parenteral to oral anticoagulation. As patients belonging to this risk category were excluded from the phase III NOAC trials, the optimal time point for this transition has not been determined by existing evidence but should instead be based on clinical judgement. The specifications concerning the higher initial dose of apixaban or rivaroxaban (for 1 and 3 weeks after PE diagnosis, respectively), or the minimum overall period (5 days) of heparin anticoagulation before switching to dabigatran or edoxaban, must be followed (see supplementary table 8 for tested and approved regimens).

8.1. Assessment of venous thromboembolism recurrence risk

The risk for recurrent VTE after discontinuation of treatment is related to the features of the index PE (or, in the broader sense, VTE) event. A study, which followed patients after a first episode of acute PE, found that the recurrence rate after discontinuation of treatment was ∼2.5% per year after PE associated with transient risk factors, compared with 4.5% per year after PE occurring in the absence of known cancer, known thrombophilia, or any transient risk factor [331]. Similar observations were made in other prospective studies in patients with DVT [337]. Advancing the concept further, randomized anticoagulation trials over the past 15 years, which have focused on secondary VTE prevention, have classified patients into distinct groups based on their risk of VTE recurrence after discontinuation of anticoagulant treatment. In general, these groups are: (i) patients in whom a strong (major) transient or reversible risk factor, most commonly major surgery or trauma, can be identified as being responsible for the acute (index) episode; (ii) patients in whom the index episode might be partly explained by the presence of a weak (minor) transient or reversible risk factor, or if a non-malignant risk factor for thrombosis persists; (iii) patients in whom the index episode occurred in the absence of any identifiable risk factor (the present Guidelines avoid terms such as “unprovoked” or “idiopathic” VTE); (iv) patients with one or more previous episodes of VTE, and those with a major persistent pro-thrombotic condition such as antiphospholipid antibody syndrome; and (v) patients with active cancer [338].

Table 11 shows examples of transient/reversible and persistent risk factors for VTE, classified by the risk of long-term recurrence. As active cancer is a major risk factor for recurrence of VTE, but also for bleeding while on anticoagulant treatment [339], section 8.4 is specifically dedicated to the management of PE in patients with cancer.

Overall, assessment of the VTE recurrence risk after acute PE, in the absence of a major transient or reversible risk factor, is a complex issue. Beyond the examples listed in table 11, patients who are carriers of some forms of hereditary thrombophilia, notably those with confirmed deficiency of antithrombin, protein C, or protein S, and patients with homozygous factor V Leiden or homozygous prothrombin G20210A mutation, are often candidates for indefinite anticoagulant treatment after a first episode of PE occurring in the absence of a major reversible risk factor. In view of these possible implications, testing for thrombophilia (including antiphospholipid antibodies and lupus anticoagulant) [342] may be considered in patients in whom VTE occurs at a young age (e.g. aged <50 years) and in the absence of an otherwise identifiable risk factor, especially when this occurs against the background of a strong family history of VTE. In such cases, testing may help to tailor the regimen and dose of the anticoagulant agent over the long-term. On the other hand, no evidence of a clinical benefit of extended anticoagulant treatment is currently available for carriers of heterozygous factor V Leiden or prothrombin 20210A mutation.

A number of risk prediction models have been developed for the assessment of the risk of recurrence in an individual patient (supplementary table 13) [343, 344]. The clinical value and, in particular, the possible therapeutic implications of these models in the NOAC era are unclear.

8.2. Anticoagulant-related bleeding risk

Incidence estimates from cohort studies conducted more than 15 years ago reported an ∼3% annual incidence of major bleeding in patients treated with VKAs [345]. Meta-analyses of phase III studies focusing on the first 3–12 months of anticoagulant treatment showed an ∼40% reduction in the risk for major bleeding with NOACs compared with VKAs [346]. The risk of major bleeding is higher in the first month of anticoagulant treatment, and then declines and remains stable over time. Based on currently available evidence, risk factors include: (i) advanced age (particularly >75 years); (ii) previous bleeding (if not associated with a reversible or treatable cause) or anaemia; (iii) active cancer; (iv) previous stroke, either haemorrhagic or ischaemic; (v) chronic renal or hepatic disease; (vi) concomitant antiplatelet therapy or non-steroidal anti-inflammatory drugs (to be avoided, if possible); (vii) other serious acute or chronic illness; and (viii) poor anticoagulation control.

Existing bleeding risk scores and their current validation status are reviewed in supplementary table 14. The patient's bleeding risk should be assessed, either by implicit judgement after evaluating individual risk factors or by the use of a bleeding risk score, at the time of initiation of anticoagulant treatment. It should be reassessed periodically (e.g. once a year in patients at low risk, and every 3 or 6 months in patients at high risk for bleeding). Bleeding risk assessment should be used to identify and treat modifiable bleeding risk factors, and it may influence decision-making on the duration and regimen/dose of anticoagulant treatment after acute PE.

8.3. Regimens and treatment durations with non-vitamin K antagonist oral anticoagulants, and with other non-vitamin K antagonist antithrombotic drugs

All patients with PE should be treated with anticoagulants for ≥3 months [347]. Beyond this period, the balance between the risk of VTE recurrence and that of bleeding, which has been used to select candidates for extended anticoagulation after a first VTE event in the VKA era, is currently being revisited based on the lower bleeding rates with NOACs. However, despite the improved safety of these drugs compared with VKAs, treatment with NOACs is not without risk. Phase III clinical trials on the extended treatment of VTE have shown that the rate of major bleeding may be ∼1%, and that of clinically relevant non-major (CRNM) bleeding as high as 6%. Bleeding rates may be higher in everyday clinical practice [348, 349].

The NOAC trials that focused on extended VTE treatment are summarized in supplementary table 15. In all studies, patients with PE made up approximately one-third of the entire study population, while the remaining two-thirds were patients with proximal DVT but no clinically overt PE. Patients needed to have completed the initial and long-term anticoagulation phase to be included in the extended studies.

Dabigatran was compared with warfarin or placebo in two different studies (supplementary table 15) [350]. In these studies, dabigatran was non-inferior to warfarin for the prevention of confirmed recurrent symptomatic VTE or VTE-related death, and more effective than placebo for the prevention of symptomatic recurrent VTE or unexplained death [350]. The rate of major bleeding was 0.9% with dabigatran compared to 1.8% with warfarin (HR 0.52, 95% CI 0.27–1.02) [350].

Rivaroxaban was compared with placebo or aspirin in two different studies in patients who had completed 6–12 months of anticoagulation treatment for a first VTE event (supplementary table 15). Treatment with rivaroxaban [20 mg once a day (o.d.)] reduced recurrent VTE by ∼80%, with a 6.0% incidence of major or CRNM bleeding as compared to 1.2% with placebo [351]. Rivaroxaban given at a dose of 20 or 10 mg o.d. was compared with aspirin (100 mg o.d.) in 3365 patients [352]. Both doses of rivaroxaban reduced symptomatic recurrent fatal or non-fatal VTE by ∼70% in comparison with aspirin. No significant differences in the rates of major or CRNM bleeding were shown between either dose of rivaroxaban and aspirin [352].

Patients with VTE were randomized to receive two different doses of apixaban [2.5 or 5 mg twice a day (bis in die: b.i.d.)] or placebo after 6–12 months of initial anticoagulation (supplementary table 15) [353]. Both doses of apixaban reduced the incidence of symptomatic recurrent VTE or death from any cause compared with placebo, with no safety concerns [353].

Patients at high bleeding risk—based on the investigator's judgement, the patient's medical history, and the results of laboratory examinations—were excluded from the extension studies mentioned above; this was also the case for studies on extended anticoagulation with VKAs [330, 331]. This fact should be taken into account during triage of a patient for extended anticoagulation with one of the above regimens.

In a randomized, open-label study in high-risk patients with antiphospholipid syndrome (testing triple positive for lupus anticoagulant, anticardiolipin, and anti-β2-glycoprotein I), rivaroxaban was associated with an increased rate of thromboembolic and major bleeding events compared with warfarin (HR for the composite primary outcome 6.7; 95% CI 1.5–30.5) [354]. At present, NOACs are not an alternative to VKAs for patients with antiphospholipid syndrome.

In two trials with a total of 1224 patients, extended therapy with aspirin (after termination of standard oral anticoagulation) was associated with a 30–35% reduction in the risk of recurrence compared with placebo (supplementary table 15) [355, 356]. However, more recently, another trial demonstrated the superiority of anticoagulation with rivaroxaban, either 20 or 10 mg o.d., over aspirin for secondary prophylaxis of VTE recurrence [352].

A randomized, placebo controlled study evaluated sulodexide (2×250 lipasemic unit capsules b.i.d.) for the prevention of recurrence in 615 patients with a first VTE event without an identifiable risk factor, who had completed 3–12 months of oral anticoagulant treatment (supplementary table 15) [357]. Sulodexide reduced the risk of recurrence by ∼50% with no apparent increase in bleeding events. However, only 8% of patients in this study had PE as the index VTE event [357].

8.4. Recommendations for the regimen and duration of anticoagulation after pulmonary embolism in patients without cancer

8.5. Management of pulmonary embolism in patients with cancer

Five RCTs compared LMWH versus conventional VTE treatment (heparin followed by VKA) in the treatment of VTE in cancer-associated thrombosis [360–364]. In 2003, a significant reduction in VTE recurrence was reported with LMWH compared with conventional (VKA) treatment without an increase in bleeding complications [362]. In a more recent trial, long-term administration of tinzaparin failed to achieve a statistically significant reduction in overall VTE recurrence over conventional treatment (HR 0.65, 95% CI 0.41–1.03); however, the overall rate of recurrent VTE in the control arm was lower than that previously observed, probably as a result of the recruitment of patients with a lower cancer-specific thrombotic risk [360]. Overall, LMWHs were found to decrease the risk of recurrent VTE by 40% with a risk of major bleeding complications similar to that of VKAs [365]. Accordingly, LMWHs have become the standard of care. However, these agents are associated with a relevant cost and burden for patients. In addition, the absolute rate of recurrent VTE while on LMWH remains high (7–9%) compared with that observed in non-cancer patients with VTE on conventional treatment (1.5–3%) [365].

NOACs could make the treatment of VTE easier and more convenient in patients with cancer, due to their oral administration in fixed-dose regimens and their lower cost compared with LMWH. However, only 3–9% of patients included in phase III studies with NOACs for the treatment of VTE had concomitant cancer [260, 261, 312, 314, 351]. A randomized, open-label trial compared edoxaban with LMWH in the secondary prevention of VTE in 1050 patients with cancer-associated thrombosis (mostly symptomatic or asymptomatic PE) [366]. Edoxaban (60 mg o.d., reduced to 30 mg in subjects with moderate renal impairment, low body weight, or concomitant need for strong inhibitors of glycoprotein-P) was started after 5 days of LMWH and treatment was given for ≥6 months. Edoxaban was non-inferior to dalteparin in the prevention of VTE recurrence or major bleeding over 12 months after randomization (HR 0.97, 95% CI 0.70–1.36). Major bleeding occurred in 6.9% of the patients in the edoxaban arm and 4.0% in the dalteparin arm (difference in risk 2.9 percentage points, 95% CI 0.1–5.6). This difference appears to have been mainly accounted for by the high rate of bleeding in patients with gastrointestinal cancer allocated to the edoxaban group [366]. Similar results were reported by a randomized, open-label pilot trial comparing rivaroxaban with dalteparin in 406 patients with VTE and cancer, 58% of whom had metastases [367]. A significant decrease in the risk of recurrent VTE was observed with rivaroxaban (HR 0.43, 95% CI 0.19–0.99). The 6 month cumulative rate of major bleeding, which was mostly gastrointestinal, was 6% (95% CI 3–11%) for rivaroxaban and 4% (95% CI 2–8%) for dalteparin (HR 1.83, 95% CI 0.68–4.96). Corresponding rates of CRNM bleeds were 13% (95% CI 9–19%) and 4% (95% CI 2–9%), respectively (HR 3.76, 95% CI 1.63–8.69) [367].

Based on the currently available evidence, as described above, patients with acute PE and cancer, particularly those with gastrointestinal cancer, should be encouraged to continue LMWH for ≥3–6 months. This also applies to patients in whom oral treatment is unfeasible due to problems of intake or absorption, and to those with severe renal impairment. In all other cases, especially in patients with an anticipated low risk of bleeding and without gastrointestinal tumours, the choice between LMWH and edoxaban or rivaroxaban is left to the discretion of the physician, and the patient's preference.

Owing to the high risk for recurrence, patients with cancer should receive indefinite anticoagulation after a first episode of VTE. Although existing evidence is limited, it is conceivable that once cancer is cured the risk for recurrence decreases and anticoagulation can be stopped. However, the definition of cured cancer is not always clear. The risk of recurrence of PE in cancer was assessed in a cohort study of 543 patients and was validated in an independent set of 819 patients [368]. The proposed score to predict the risk of recurrence included breast cancer (minus 1 point), Tumour Node Metastasis stage I or II (minus 1 point), and female sex, lung cancer, and previous VTE (plus 1 point each). Patients with a score ≤0 were at low risk (≤4.5%) and those with a score ≥1 were at high (≥19%) risk of VTE recurrence over the first 6 months [368].

After the first 3–6 months, extended anticoagulation may consist of continuation of LMWH or transition to an oral anticoagulant. Two cohort studies have assessed the safety of extended treatment with LMWH (≤12 months) in cancer-associated thrombosis [369, 370]. In both studies, the incidence of bleeding complications was higher in the first months and then reached a plateau that remained unchanged after the sixth month. In the absence of conclusive evidence, the decision to continue with LMWH or to change to VKA or a NOAC should be made on a case-by-case basis after consideration of the success of anticancer therapy, the estimated risk of recurrence of VTE, the bleeding risk, and the preference of the patient. Periodic reassessment of the risk-to-benefit ratio of continued anticoagulant treatment is mandatory.

As mentioned in section 5, venous filters are principally indicated when anticoagulation is impossible due to active haemorrhage or an excessive bleeding risk. However, the risk of VTE recurrence in the absence of anticoagulation is particularly high in patients with cancer, and the insertion of a filter should not delay the initiation of anticoagulation as soon as it is safe to do so. There is no evidence to support the use of venous filters as an adjunct to anticoagulation treatment in patients with cancer.

A number of studies have reported that a proportion of patients presenting with PE in the absence of identifiable risk factors develop cancer within the first year after diagnosis [371]. Consequently, the optimal strategy to achieve early diagnosis of these occult cancers was investigated. Two large randomized trials failed to show that comprehensive CT of the abdomen or ¹⁸F deoxy-fluoro-glucose positron emission tomography were able to detect more cancers than limited screening in patients with an unprovoked VTE [372, 373]. Therefore, based on current evidence, the search for occult cancer after an episode of VTE may be restricted to careful history taking, physical examination, basic laboratory tests, and a chest X-ray (if no CTPA was performed to diagnose PE) [372, 374, 375].

In patients with cancer, incidental PE should be managed in the same manner as symptomatic PE, whether it involves segmental or more proximal branches, multiple subsegmental vessels, or a single subsegmental vessel in association with detectable DVT [376, 377].

8.6. Recommendations for the regimen and the duration of anticoagulation after pulmonary embolism in patients with active cancer

9.1. Epidemiology and risk factors for pulmonary embolism in pregnancy

Acute PE remains one of the leading causes of maternal death in high-income countries [379, 380]. For example, in the UK and Ireland, thrombosis and thromboembolism were the most common causes of direct maternal death (death resulting from the pregnancy rather than pre-existing conditions) in the triennium 2013–15, resulting in 1.13 mortalities per 100 000 maternities (www.npeu.ox.ac.uk/mbrrace-uk). VTE risk is higher in pregnant women compared with non-pregnant women of similar age; it increases during pregnancy and reaches a peak during the post-partum period [381]. The baseline pregnancy-related risk increases further in the presence of additional VTE risk factors, including in vitro fertilization: in a cross-sectional study derived from a Swedish registry, the HR for VTE following in vitro fertilization was 1.77 (95% CI 1.41–2.23) overall and 4.22 (95% CI 2.46–7.20) during the first trimester [382]. Other important and common risk factors include prior VTE, obesity, medical comorbidities, stillbirth, pre-eclampsia, post-partum haemorrhage, and caesarean section; documented risk assessment is therefore essential [383].

The recommendations provided in these Guidelines are in line with those included in the 2018 ESC Guidelines on the management of cardiovascular diseases during pregnancy [384].

9.2. Diagnosis of pulmonary embolism in pregnancy

9.2.2. Imaging tests

A proposed algorithm for the investigation of suspected PE in women who are pregnant, or ≤6 weeks post-partum, is shown in figure 7. Both maternal and fetal radiation exposure are low using modern imaging techniques (table 12) [385, 392–398]. For V/Q scans and CTPA, fetal radiation doses are well below the threshold associated with fetal radiation complications (which is 50–100 mSv) [399, 400]. In the past, CTPA has been reported to cause high radiation exposure to the breast [395, 401]; however, CT technology has evolved, and several techniques can now reduce radiation exposure without compromising image quality. These include reducing the anatomical coverage of the scan [393], reducing the kilovoltage, using iterative reconstructive techniques, and reducing the contrast-monitoring component of the CTPA [392, 393]. Modern CTPA imaging techniques may therefore expose the maternal breast to median doses as low as 3–4 mGy (table 12) [392]. The effect on maternal cancer risk with modern CTPA techniques is negligible (lifetime cancer risk is reportedly increased by a factor of 1.0003–1.0007); avoiding CTPA on the grounds of maternal cancer risk is therefore not justified [394].

View this table:

• View inline
• View popup

TABLE 12

Estimated amounts of radiation absorbed in procedures used to diagnose pulmonary embolism (based on various references [385, 392–398])

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7

Diagnostic workup and management of suspected pulmonary embolism during pregnancy, and up to 6 weeks post-partum. CTPA: computed tomography pulmonary angiography; CUS: compression ultrasonography; DVT: deep vein thrombosis; LMWH: low-molecular-weight heparin; PE: pulmonary embolism. ^aIf chest X-ray abnormal, consider also alternative cause of chest symptoms. ^bDVT in pelvic veins may not be ruled out by CUS. If the entire leg is swollen, or there is buttock pain or other symptoms suggestive of pelvic thrombosis, consider magnetic resonance venography to rule out DVT. ^cCTPA technique must ensure very low fetal radiation exposure (see table 12). ^dPerform full blood count (to measure haemoglobin and platelet count) and calculate creatinine clearance before administration. Assess bleeding risk and ensure absence of contra-indications. ^eSee table 8.

A normal perfusion scan and a negative CTPA appear equally safe for ruling out PE in pregnancy, as suggested by retrospective series [385, 386, 402–404]. Inconclusive results can be a problem (4–33% of investigations) [385, 386, 405], especially late in pregnancy [405]. A recent survey of 24 sites in the UK, representing a population of 15.5 million, revealed a similar rate of inadequate or indeterminate CTPA and scintigraphy scans, suggesting that the initial choice of imaging is best determined by local expertise and resources [406].

V/Q SPECT is associated with low fetal and maternal radiation exposure, and has promise in PE diagnosis in pregnancy [407]. However, further evaluation of this technique is required before its widespread incorporation into diagnostic algorithms. For MRA, the long-term effects of gadolinium contrast on the fetus are not known. In non-pregnant patients, technically inadequate images are frequently obtained and the rate of inconclusive scan results is high [140]. Therefore, use of this technique for diagnosing or ruling out PE during pregnancy cannot be recommended at present. Conventional pulmonary angiography involves significantly higher radiation exposure of the fetus (2.2–3.7 mSv) and should be avoided during pregnancy [400].

Overdiagnosis of PE is a potential pitfall that can have significant, lifelong implications for a pregnant woman, including the risk of bleeding at the time of delivery, the withholding of oestrogen contraception, and the requirement for thromboprophylaxis during future pregnancies. Consequently, avoiding PE overdiagnosis in pregnancy is as important as not missing a PE diagnosis.

9.3. Treatment of pulmonary embolism in pregnancy

LMWH is the treatment of choice for PE during pregnancy [384]. In contrast to VKAs and NOACs, LMWH does not cross the placenta, and consequently does not confer a risk of fetal haemorrhage or teratogenicity. Moreover, while UFH is also safe in pregnancy, LMWH has more predictable pharmacokinetics and a more favourable risk profile [408–411]. Although no RCT has evaluated the optimal dose of LMWH for the treatment of PE during pregnancy, currently published data favour similar dosing to non-pregnant patients, either with o.d. or b.i.d. regimens based on early pregnancy weight [408, 410]. For the majority of patients receiving LMWH treatment for PE during pregnancy [412, 413], it remains uncertain whether using serial measurements of plasma anti-activated coagulation factor X activity to guide dosing may be of clinical benefit. It is important to bear in mind that: (i) LMWH has a predictable pharmacokinetic profile, (ii) data on optimal anti-activated coagulation factor levels are lacking, and (iii) the assay itself has limitations [414]. In addition, there are no solid data on the clinical benefit versus harm of frequent, weight-based dose adjustments of LMWH during pregnancy. Thus, anti-activated coagulation factor X monitoring may be reserved for specific high-risk circumstances such as recurrent VTE, renal impairment, and extremes of body weight.

The use of UFH has been associated with heparin-induced thrombocytopenia and bone loss. It remains uncertain whether, and to what extent, the risk of bone loss is increased with LMWH use. In a recent observational cohort study, in which bone mineral density was measured by dual-energy X-ray absorptiometry 4–7 years after the last delivery in 152 women (92 of whom received prolonged LMWH during pregnancy), lumbar spine bone mineral density was similar in LWMH-treated women and controls following adjustment for potential confounders. No osteoporosis or osteoporotic fractures were reported [415].

Fondaparinux may be considered if there is an allergy or adverse response to LMWH, although solid data are lacking and minor transplacental passage has been demonstrated [416]. VKAs cross the placenta and are associated with a well-defined embryopathy during the first trimester. Administration of VKAs in the third trimester can result in fetal and neonatal haemorrhage, as well as placental abruption. Warfarin may be associated with central nervous system anomalies in the fetus throughout pregnancy. NOACs are contraindicated in pregnant patients [417].

The management of labour and delivery requires particular attention. In women receiving therapeutic LMWH, strong consideration should be given to planned delivery in collaboration with the multidisciplinary team to avoid the risk of spontaneous labour while fully anticoagulated. The incidence of spinal haematoma after regional anaesthesia is unknown in pregnant women under anticoagulation treatment. If regional analgesia is considered for a woman receiving therapeutic LMWH, ≥24 h should have elapsed since the last LMWH dose before insertion of a spinal or epidural needle (assuming normal renal function and including risk assessment at extremes of body weight).

In high-risk situations, for example in patients with recent PE, it is recommended that LMWH be converted to UFH ≥36 h prior to delivery. The UFH infusion should be stopped 4−6 h prior to anticipated delivery and the activated partial thromboplastin time should be normal (i.e. not prolonged) prior to regional anaesthesia [418].

Data are limited on the optimal timing of post-partum reinitiation of LMWH [419, 420]. Timing will depend upon the mode of delivery and an assessment of the thrombotic versus bleeding risk by a multidisciplinary team. LMWH should not be given for ≥4 h after removal of the epidural catheter; the decision on timing and dose should consider whether the epidural insertion was traumatic, and take into account the risk profile of the woman. For example, an interim dose of a prophylactic LMWH dose may be considered post-operatively (after caesarean section), once at least 4 h have elapsed since epidural catheter removal, and allowing for an interval of ≥8–12 h between the prophylactic and the next therapeutic dose. Close collaboration between the obstetrician, the anaesthesiologist, and the attending physician is recommended.

Anticoagulant treatment should be administered for ≥6 weeks after delivery and with a minimum overall treatment duration of 3 months. LMWH and warfarin can be given to breastfeeding mothers; the use of NOACs is not recommended [417].

High-risk, life-threatening PE during pregnancy is a rare, but potentially devastating, event. A recent systematic review included 127 cases of severe PE during pregnancy (and until 6 weeks post-partum) treated with thrombolysis, thrombectomy, and/or ECMO [421]. Both high- and intermediate-risk PE cases were included, and 23% of women experienced cardiac arrest. Reported survival rates were 94 and 86% following thrombolysis and surgical thrombectomy, respectively; however, these favourable rates may reflect reporting bias. Following thrombolysis, major bleeding occurred in 18 and 58% of cases during pregnancy and in the post-partum period, respectively [421]. Finally, fetal deaths occurred in 12 and 20% of the cases following thrombolysis and thrombectomy, respectively [421]. Thrombolytic treatment should not be used peri-partum, except in the setting of life-threatening PE. Typically, UFH is used in the acute treatment of high-risk PE.

Although the indications for vena cava filters are similar to those for non-pregnant patients (discussed in section 6), there is limited experience with their use in pregnancy and the risk associated with the procedure may be increased.

Suggestions for the anticoagulation management of PE in specific clinical situations (also) related to pregnancy, for which conclusive evidence is lacking, are presented in supplementary table 9.

9.4. Amniotic fluid embolism

Amniotic fluid embolism (AFE) is a rare condition that occurs during pregnancy or shortly after delivery. It remains one of the leading causes of direct maternal death (i.e. death resulting from the pregnancy rather than from pre-existing conditions) in high-income countries [422]. Diagnosis of AFE is challenging, being primarily a clinical diagnosis of exclusion. Awareness of AFE, prompt diagnosis, and aggressive life support are of critical importance. AFE is characterized by unexplained sudden cardiovascular or respiratory deterioration, often accompanied by disseminated intravascular coagulation [422], and occurring during pregnancy or after delivery [423, 424]. The reported incidence is approximately 2–7 per 100 000 maternities, with a mortality rate of 0.5–6 deaths per 100 000 deliveries [422, 425, 426]. Reported case fatality rates vary, reflecting the challenges in making the diagnosis and the rarity of AFE. In a retrospective Californian study including more than 3.5 million deliveries, a case fatality rate of 13% was reported, as in other US and Canadian studies [425]. Similarly, a case fatality rate of 19% was reported in a recent prospective UK population-based study with validated case criteria [422]. Recent literature have suggested that risk factors for AFE may include pre-existing cardiac, cerebrovascular, and renal disorders, placenta previa, polyhydramnios, stillbirth, chorioamnionitis, hypertensive disorders, instrumental delivery, and caesarean section [422, 425]. Management of AFE is supportive, and based on high-quality emergency care following the recognition and diagnosis of the condition, with prompt treatment of bleeding and coagulopathy [423]. Awareness of AFE should be integral to the education of involved physicians and to emergency algorithms.

9.5. Recommendations for pulmonary embolism in pregnancy

10.1. Persisting symptoms and functional limitation after pulmonary embolism

Cohort studies conducted over the past decade (summarized in Klok et al. [428]) have revealed that persisting or deteriorating dyspnoea, and poor physical performance, are frequently present 6 months to 3 years after an acute PE episode. The proportion of patients claiming that their health status is worse at 6 month follow-up than it was at the time of PE diagnosis varies widely, ranging between 20 and 75% [429–431]. The following baseline parameters and findings could be identified as predictors of exertional dyspnoea at long-term follow-up after PE: advanced age, cardiac or pulmonary comorbidity, higher body mass index, and history of smoking [429]; higher systolic PAP and RV dysfunction at diagnosis [430, 432, 433]; and residual pulmonary vascular obstruction at discharge [434].

More recently, a prospective cohort study enrolled 100 patients at five Canadian hospitals between 2010 and 2013, and followed them over 1 year [435]. As many as 47% of the patients had reduced maximal aerobic capacity, defined as peak oxygen consumption <80% of the predicted value on cardiopulmonary exercise testing (CPET). This functional outcome was associated with significantly worse generic health-related quality of life and dyspnoea scores, as well as with a significantly reduced 6 min walk distance [435]. Independent predictors of reduced functional exercise capacity and quality of life over time included female sex, higher body mass index, history of lung disease, higher pulmonary artery systolic pressures on the 10 day echocardiogram, and higher main pulmonary artery diameter on the baseline CTPA [436]. Of note, pulmonary function tests and echocardiographic results at follow-up were largely within normal limits, both in patients with and without reduced maximal aerobic capacity [435]. Lack of an association between exercise impairment, and persistent RV dilation or dysfunction, was also reported by a study of 20 survivors of massive or submassive PE [437].

Taken together, older and more recent cohort studies have suggested that muscle deconditioning, particularly in the presence of excess body weight and cardiopulmonary comorbidity, is largely responsible for the frequently reported dyspnoea and signs of exercise limitation after acute PE. This also means that, at least in the majority of cases, poor physical performance after PE does not appear to be attributable to “large” residual thrombi, or persisting/progressive PH and RV dysfunction. Ongoing prospective studies in large numbers of patients may help to better identify predictors of functional and/or haemodynamic impairment after acute PE, and their possible implications for shaping follow-up programmes [438].

As mentioned in section 6, it remains unclear whether early reperfusion treatment, notably thrombolysis, has an impact on clinical symptoms, functional limitation, or persistent (or new-onset) PH at long-term follow-up after PE. Consequently, prevention of long-term PE sequelae is, at present, no justification for thrombolytic treatment in the acute phase of PE.

10.2. Chronic thromboembolic pulmonary hypertension

10.3. Strategies for patient follow-up after pulmonary embolism

Figure 8 displays a proposed follow-up strategy for survivors of acute PE following discharge from hospital. Evaluation of the patients 3–6 months after the acute PE episode is recommended to assess the persistence (or new onset) and severity of dyspnoea or functional limitation, and to check for possible signs of VTE recurrence, cancer, or bleeding complications of anticoagulation. The severity of dyspnoea can be assessed using the Medical Research Council scale [160]; alternatively, the World Health Organization functional class can be determined (supplementary table 16) [289].

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8

Follow-up strategy and diagnostic workup for long-term sequelae of pulmonary embolism. CPET: cardiopulmonary exercise testing; CTEPH: chronic thromboembolic pulmonary hypertension; NT-proBNP: N-terminal pro B-type natriuretic peptide; PE: pulmonary embolism; PH: pulmonary hypertension; TTE: transthoracic echocardiography/echocardiogram; V/Q: ventilation/perfusion (lung scintigraphy). ^aAssess the persistence (or new onset) and severity of dyspnoea or functional limitation, and also check for possible signs of VTE recurrence, cancer, or bleeding complications of anticoagulation. ^bThe Medical Research Council scale can be used to standardize the evaluation of dyspnoea [160]; alternatively, the World Health Organization functional class can be determined (supplementary table 16) [289]. ^cAs defined by the ESC/ERS guidelines on the diagnosis and treatment of Pulmonary Hypertension (supplementary tables 17 and 18) [289]. ^dRisk factors and predisposing conditions for CTEPH are listed in table 13. ^eCardiopulmonary exercise testing, if appropriate expertise and resources are available on site; abnormal results include, among others, reduced maximal aerobic capacity (peak oxygen consumption), reduced ventilatory equivalent for carbon dioxide, and reduced end-tidal carbon dioxide pressure. ^fConsider CPET in the diagnostic work-up.

In patients complaining of persisting dyspnoea and poor physical performance, TTE should be considered as the next step to assess the probability of (chronic) PH and thus possible CTEPH. The criteria and levels of PH probability are defined by current ESC Guidelines [289], and are listed in supplementary tables 17 and 18. Patients with a high echocardiographic probability of PH, or those with intermediate probability combined with elevated NT-proBNP levels or risk factors/predisposing conditions for CTEPH, such as those listed in table 13, should be considered for a V/Q scan.

If mismatched perfusion defects are found on the V/Q scan, referral to a PH or CTEPH expert centre for further diagnostic workup is indicated. If, on the other hand, the V/Q scan is normal and the patient's symptoms remain unexplained, CPET may be performed. By providing evidence of reduced maximal aerobic capacity, CPET supports the need for further follow-up visits and helps to identify candidates for pulmonary rehabilitation, exercise, or weight-reduction programmes [435, 436]. CPET may also be helpful in patients with suspected CTEPH and coexisting left heart and/or respiratory disease; in such cases, it can help to establish the main limiting factor and thus set priorities for the treatment strategy [472].

For patients who report as free of dyspnoea or functional limitation at 3–6-month follow-up after acute PE but have risk factors/predisposing conditions for CTEPH (table 13), further follow-up visits may be scheduled and the patient must be advised to return if symptoms appear. Alternatively, TTE may be considered to assess the probability of PH (figure 8).

Apart from the recommended screening and diagnostic measures, an integrated model of patient care after PE should be provided, taking into consideration the infrastructure and possibilities offered by each country's health system. The model should include appropriately qualified nurses, interdisciplinary working with physicians in the care of both in-hospital and ambulatory PE patients, standardized treatment protocols adapted to the capacities of each hospital, and bidirectional referral pathways between general practice and the hospital. Such models ensure smooth transitions between hospital specialists and practitioners; provide continuity, and easy access to care along with information and education; and respect the patients’ preferences, and those of their families and social environment. In this context, nurse-led care models to deliver follow-up have been shown to be effective after acute coronary syndrome [473], in primary care-based management of chronic diseases [474], and in community based self-management initiatives [475]. A recently published study investigated the care of 42 patients followed at a pulmonary arterial hypertension (PAH)/CTEPH nurse-led outpatient clinic and showed positive results [476]. During patient follow-up visits, appropriately qualified nurses screen for signs and symptoms indicating VTE recurrence or complications of treatment, and assess adherence to medication. Nurses work collaboratively with patients using behavioural frameworks and motivational interviewing, to identify and modify associated risk factors (smoking cessation, diet, physical activity, and exercise). In addition, they promote self-management skills such as the use of compression stockings, safe increase in mobility, increased awareness of signs of recurrence, or complications.

10.4. Recommendations for follow-up after acute pulmonary embolism

13.1. Diagnosis

• The optimal method to adjust (based on the patient's age or in combination with clinical probability) the D-dimer threshold, permitting the exclusion of PE while reducing the number of unnecessary imaging tests to a minimum, remains to be determined.

• The diagnostic value and clinical significance of isolated subsegmental contrast-filling defects in the modern CTPA era remain controversial.

• No robust data exist to guide the decision on whether to treat incidental PE with anticoagulants compared with a strategy of watchful waiting.

• For patients presenting with non-traumatic chest pain, the benefits versus risks of “triple rule-out” (for coronary artery disease, PE, and aortic dissection) CT angiography need further evaluation before such an approach can be routinely recommended.

13.2. Assessment of pulmonary embolism severity and the risk of early death

• The optimal, clinically most relevant combination (and cut-off levels) of clinical and biochemical predictors of early PE-related death remain to be determined, particularly with regard to identifying possible candidates for reperfusion treatment among patients with intermediate-risk PE.

• The need for assessment of the RV status in addition to clinical parameters, to classify a patient with acute symptomatic PE as being at low versus intermediate risk, needs to be confirmed by further prospective management (cohort) studies.

13.3. Treatment in the acute phase

• The clinical benefits versus risks of reduced-dose thrombolysis and catheter-based reperfusion modalities in patients with intermediate–high-risk PE should be evaluated in prospective randomized trials.

• The place of ECMO in the management of acute high-risk PE awaits support by additional evidence from prospective management (cohort) studies.

• The optimal anticoagulant drug(s) and regimen in patients with renal insufficiency and CrCl <30 mL/min remain unclear.

• The criteria for selecting patients for early discharge and outpatient treatment of PE, and particularly the need for assessment of the RV status with imaging methods and/or laboratory markers in addition to calculating a clinical score, need to be further validated in prospective cohort studies.

13.4. Chronic treatment and prevention of recurrence

• The clinical value and the possible therapeutic implications of models or scores assessing the risk of VTE recurrence, and the risk of bleeding under anticoagulation, need to be revisited in the NOAC era.

• The effectiveness of extended treatment with a reduced dose, or apixaban or rivaroxaban, should be confirmed in patients with a high risk of recurrent PE.

• The evidence supporting the efficacy and safety of NOACs for the treatment of PE in patients with cancer needs to be extended by further studies.

• In patients with cancer, the anticoagulant regimen and dose after the first 6 months should be clarified and prospectively tested.

• The optimal time for discontinuing anticoagulant treatment after an episode of acute PE in patients with cancer is yet to be determined.

13.5. Pulmonary embolism and pregnancy

• Diagnostic algorithms for PE in pregnancy, using modern radiological imaging techniques and low radiation doses, need to be prospectively tested in adequately powered cohort studies.

• Controversy persists on the optimal LMWH dose and regimen for the treatment of PE during pregnancy.

• NOACs are not allowed in pregnancy. However, if exposure to these drugs occurs during pregnancy despite this warning, any possible effects on the fetus should be recorded to provide more precise information on the risks and complications of these drugs, and adapt the instructions to physicians in the future.

13.6. Long-term sequelae of pulmonary embolism

• The optimal follow-up strategy, including the spectrum of diagnostic tests that may be necessary, in patients with persisting symptoms and functional limitation after acute PE needs to be defined and prospectively validated.

• In the absence of persisting symptoms or functional limitation after acute PE, the criteria for identifying patients whose risk of developing CTEPH may be sufficiently high to justify further diagnostic workup require further elaboration and validation in prospective cohort studies.