Background Physicians treating acute pulmonary embolism (PE) are faced with difficult management decisions while specific guidance from recent guidelines may be absent.
Methods Fourteen clinical dilemmas were identified by physicians and haematologists with specific interests in acute and chronic PE. Current evidence was reviewed and a practical approach suggested.
Results Management dilemmas discussed include: sub-massive PE, PE following recent stroke or surgery, thrombolysis dosing and use in cardiac arrest, surgical or catheter-based therapy, failure to respond to initial thrombolysis, PE in pregnancy, right atrial thrombus, role of caval filter insertion, incidental and sub-segmental PE, differentiating acute from chronic PE, early discharge and novel oral anticoagulants.
Conclusion The suggested approaches are based on a review of the available evidence and guidelines and on our clinical experience. Management in an individual patient requires clinical assessment of risks and benefits and also depends on local availability of therapeutic interventions.
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/3.0/
Statistics from Altmetric.com
Several guidelines on acute pulmonary embolism (PE) have been published.1–3 Guidance for various scenarios which challenge physicians in the management of acute PE are often not easily accessible in guidelines. Our institution runs an integrated PE service between respiratory and haematology physicians and a large tertiary pulmonary hypertension service. We are not infrequently referred complex acute PE cases from other centres. In this review we discuss the most clinically challenging scenarios.
Eight physicians with an interest in the management of acute and chronic pulmonary embolic disease compiled a list of 14 challenging clinical issues faced in their day-to-day practice. A PubMed search for each dilemma was performed, an initial review and suggested approach drafted followed by round-table discussion to achieve consensus regarding management. In many dilemmas, conclusions based on the available literature, were hampered by patient numbers and reporting bias. Suggested approaches were provided based on consensus.
Definition of PE severity
In the current paper we have adopted the American Heart Association (AHA) classification.3 Massive PE is defined as sustained hypotension (systolic blood pressure <90 mm Hg) for >15 min secondary to acute PE or a requirement of inotropes or signs of shock. Submassive PE is defined by evidence of right ventricular (RV) dysfunction and/or evidence of myocardial necrosis. Patients with none of these features are defined as low-risk.
Which patients with submassive PE should I thrombolyse?
The pro–con debate published in this issue of Thorax highlights the controversy regarding systemic thrombolytics in normotensive patients with PE.4 ,5 Clinical trials have demonstrated more rapid, immediate haemodynamic improvement and clot resolution following thrombolysis, but not clear mortality benefits.6 ,7 Recent data from a large unselected national registry demonstrated that thrombolysis in normotensive patients with acute PE was associated with increased mortality.8 Consideration for thrombolysis therefore requires risk stratification. Validated severity scoring systems, such as the PE Severity Index (PESI, table 1), can identify clinical features at the time of presentation associated with poorer outcome.9 European Society of Cardiology (ESC) guidelines suggest assessing for RV dysfunction (using echocardiography, CT or B-type natriuretic peptide) or ischaemia (troponin) to aid risk stratification.1 The presence of lower limb deep venous thrombosis (DVT) has also been associated with poorer survival.10 By combining these factors it is possible to identify a higher risk population with 30-day mortality >20% (table 2).11 A meta-analysis of randomised controlled trials (RCTs) of thrombolysis in massive and submassive PE published prior to 2004 reported a risk of major bleeding of 9.1% and intracranial haemorrhage (ICH) of 0.5% while a recent large RCT of tenecteplase in submassive PE (PEITHO) observed rates of major bleeding of 6.3% and ICH of 2% (compared with 1.5% and 0.2% respectively for heparin alone).12 Interestingly, bleeding risk was lower and mortality benefit higher in patients <75 years.
Suggested approach: In submassive PE we would not routinely administer thrombolysis. PESI score and the presence or absence of single or multiple poor prognostic factors should be balanced against factors associated with increased risk of bleeding (including age) in identifying suitable candidates for thrombolysis.
What is the risk of thrombolysis in a patient with recent surgery, previous stroke or intracranial space-occupying lesion?
Thrombolysis after recent surgery
We identified 25 reports, including 64 patients, thrombolysed (the majority for PE) following major recent surgery13–37 (see online supplementary table S3). Major bleeding occurred in >50% of patients receiving thrombolysis within 1 week of surgery and in 20% of patients thrombolysed 1–2 weeks postoperatively. American College of Chest Physicians (ACCP) guidelines suggest that recent surgery (excluding recent brain or spinal surgery or trauma) is a relative contraindication and that the bleeding risk reduces significantly 2 weeks after surgery.
Thrombolysis in the presence of intracranial space-occupying lesions
A review of 12 patients with intracranial neoplasms thrombolysed for various indications identified ICH in a single patient (8.3%).38 Guillan et al39 identified five cases (five meningiomas, one cholesteatoma and one paranasal tumour) receiving systemic thrombolysis for stroke without complications. The risk of ICH is dependent on tumour type and localisation. A clinicopathological study showed the risk of microscopic and macroscopic spontaneous bleeding to be 50% in metastatic melanoma and ranging from 29.2% in oligodendroglioma to 2.8% in meningioma.40
Thrombolysis after recent ischaemic stroke
Previous ischaemic stroke within 3 or 6 months is a contraindication to thrombolysis in ACCP and ESC guidelines.1 ,2 A study involving 145 patients with a stroke within 3 months who received thrombolysis for a further stroke did not show an increase in ICH rate.41
Suggested approach: In patients with a massive PE within 1 week of surgery we would favour mechanical treatment if available. Within 1–2 weeks following surgery, thrombolysis may be an acceptable risk depending on the nature of the surgery. In our opinion previous ischaemic stroke is not an absolute contraindication to thrombolysis but there are no data to guide an acceptable timescale since the stroke. Selected intracranial space-occupying lesions, for example meningiomas, would not influence our decision to thrombolyse.
A patient with an acute cerebral infarct is found to have an acute PE: what should I do regards anticoagulation?
Patients rarely present with a stroke and PE simultaneously due to paradoxical embolisation across a patent foramen ovale (PFO).42 ,43 More frequently (1–10% of cases) patients may develop an acute PE following a stroke.44 PE is the most common cause of death 2–4 weeks post stroke.44 In the absence of anticoagulation, the majority of haemorrhagic transformation involves petechial bleeds with low risk of mass effect.45–47 However, low and intermediate dose heparin early after stroke presentation is associated with an increased rate of haemorrhagic transformation.48–50 Stroke guidelines advise delaying anticoagulation for 2 weeks post ischaemic stroke in patients with atrial fibrillation but give discordant advice regarding anticoagulation for coexisting PE. UK stroke guidance suggests anticoagulation for proximal DVT or PE while AHA guidelines do not recommend initial anticoagulation in patients with moderate to severe stroke.51 ,52
Suggested approach: The risk–benefit ratio for individual patients should be assessed; however, our general approach is to anticoagulate all patients with a cerebral infarct and PE. In patients with PE with a primary haemorrhagic stroke or recent significant haemorrhagic transformation we would consider inferior vena cava (IVC) filter insertion and delayed anticoagulation.
What is the optimal type and dose of thrombolytic agent and what should I do if a patient is already on low molecular weight heparin?
Thrombolytic agents for PE should be administered peripherally.2 Several thrombolytic agents have been studied: urokinase, streptokinase and recombinant tissue plasminogen activators (alteplase, reteplase, desmoteplase and tenecteplase).2 ,3 Alteplase is the most widely used thrombolytic agent for PE; recommended dosing in patients ≥65 kg is a loading bolus of 10 mg over 1–2 min followed by 90 mg infused over 2 h.53 In patients <65 kg the total dose administered is 1.5 mg/kg; for example a patient weighing 60 kg should receive a 10 mg loading bolus followed by 80 mg over 2 h. In patients already receiving intravenous heparin we stop the infusion prior to administration of alteplase, check activated partial thromboplastin time (APTT) 2 h following completion of administration and restart heparin when the APTT ratio is <2× the upper limit of normal. If there is good clinical response to thrombolysis we would convert to low molecular weight heparin (LMWH) 24 h following thrombolysis. If therapeutic LMWH had been administered prior to thrombolysis we would usually start heparin infusion as above but delay commencement to 18 h following the last dose of LMWH if once-daily dosing and 8–10 h if twice-daily dosing had been used. Two RCTs have investigated the efficacy and side effects of half-dose alteplase in predominantly submassive PE.54 ,55 Superior efficacy with no increase in bleeding risk was observed when compared with anticoagulation alone,54 and equal efficacy with less haemorrhage was seen when compared with standard-dose anticoagulation.55
Suggested approach: If thrombolysis is indicated for PE we would administer a 10 mg bolus of alteplase followed by a further 90 mg over 2 h (up to a maximum of 1.5 mg/kg). If thrombolysis is indicated but there is a high risk of haemorrhage we would consider using a half-dose regimen.
Which patients in an arrest or peri-arrest situation should I consider thrombolysing in the absence of definitive radiological evidence of PE?
If PE is suspected clinically in an acutely deteriorating patient who is too unwell for CT pulmonary angiogram (CTPA) then echocardiography may identify signs of acute right heart strain suggestive of acute PE.56 ,57 Thrombolysis can increase the return of spontaneous circulation and survival to discharge in patients with known or suspected PE who have cardiac arrest.58–60 British Thoracic Society guidelines suggest a bolus dose of 50 mg alteplase in the peri-arrest or arrest situation.61 Patients who have arrested and then regained circulation may also be suitable for emergency pulmonary embolectomy.62 Patients in whom the cause of arrest is unclear should not receive thrombolysis during cardiopulmonary resuscitation.2 A recent large RCT demonstrated that thrombolysis in out-of-hospital cardiac arrest when the cause of arrest is undifferentiated is not associated with significant mortality benefit.63
Suggested approach: Thrombolysis should be administered in the peri-arrest or arrest situation when PE is either known or suspected.
I feel it is too unsafe to perform thrombolysis: what are the surgical and non-surgical alternatives?
If thrombolytic therapy is contraindicated and a patient has significant accessible PE and persisting haemodynamic compromise then embolectomy, performed by an open surgical or catheter-based approach, should be considered.1 Older case series observed mortality rates for surgical embolectomy of >20%;3 however an intraoperative mortality of 6% was reported in 47 consecutive patients.64 Catheter-directed therapies include mechanical disruption of thrombi by catheter, ultrasound or pressurised saline injection.65 Suction may be used to perform thrombectomy or aspirate fragments of macerated emboli following other techniques. A recent meta-analysis observed 87% clinical success.66 Local intra-clot thrombolytic was used in 67% of cases and was associated with superior clinical success, postulated due to increased thrombus surface area exposed to thrombolysis after fragmentation. Major complications of catheter-directed therapy including pulmonary artery rupture and massive haemoptysis were seen in 2.4% of cases while haemodynamic deterioration due to fragmented emboli was unpredictable.66 Although the incidence of major bleeding in the meta-analysis was low (18 non-cerebral haemorrhages requiring transfusion and 1 intra-cerebral haemorrhage reported in 594 patients66) the absolute risk of bleeding related to intra-clot thrombolysis in an individual patient with a contraindication to systemic thrombolysis is not clear. There are few data comparing thrombolysis, surgical embolectomy and catheter-based intervention as primary treatment for massive PE. Current guidelines restrict surgical embolectomy to situations when thrombolysis has failed or is contraindicated.1 Although there is increasing interest in the expansion of surgical embolectomy to the initial management of massive PE, randomised trial data are required.62 ,64 ,67–70
Suggested approach: No comparative data exist to guide primary management of massive PE in the presence of a strong contraindication to systemic thrombolysis. Management is dependent on local availability of cardiothoracic surgery and catheter-based therapy.
A patient with a recent acute PE fails to respond to initial therapy: what should I do?
If a patient with acute PE fails to respond to initial anticoagulation, with worsening cardiovascular instability and/or respiratory failure, then thrombolysis should be considered. In the MAPPET-3 study of submassive PE, delayed thrombolysis was performed in 23% of patients treated initially with heparin, with no difference in mortality compared with patients receiving up-front thrombolysis.6 Although reperfusion is greater the earlier thrombolysis is given, benefit may be observed when administered up to 14 days from symptom onset.71 Failure to improve following thrombolysis may be related to persistent thrombus, complications such as lung infarction or infection or existence of chronic clot. Reassessment with additional imaging may therefore be required. In the presence of persistent clot, repeat thrombolysis or mechanical therapy may be considered. A single centre retrospective study of treatment in failed thrombolysis demonstrated that mortality in patients receiving repeat thrombolysis was 38% compared with 7% in patients undergoing embolectomy, although bias in management approach cannot be excluded.72 Supportive therapy for lung infarction may include ventilatory support, treatment of superadded infection and inotropic support. If underlying chronic thromboembolic disease is suspected, referral for pulmonary endartectomy and the use of bridging pulmonary vasodilator therapy should be considered. The role of pulmonary vasodilators in purely acute disease has also been assessed.73 Inhaled nitric oxide may improve gas exchange in acute PE.74 ,75 Limited data suggest possible benefit from nebulised iloprost 76 ,77 while a small RCT failed to demonstrate benefit from intravenous eposprostenol.78 There are limited reports of benefit from sildenafil in animal models and humans with acute PE.79
Suggested approach: Thrombolysis should be considered when a patient initially treated with anticoagulation alone develops worsening cardiovascular instability or respiratory failure. Failure to improve following thrombolysis should trigger reassessment for residual clot or complication of PE. Surgical embolectomy is preferable to re-thrombolysis for persistent obstructing acute PE.
How should I manage a pregnant patient with significant PE?
In non-massive PE, therapeutic LMWH has been shown to be safe and effective at preventing recurrent PE and does not cross the placenta.80 Warfarin administration is teratogenic in the first trimester but is also associated with neural abnormalities during any trimester and UK obstetric guidelines advise against its use during pregnancy.81 ,82 If PE is within a month of the expected date of delivery then a retrievable IVC filter should be inserted. A recent review identified 189 pregnant patients receiving thrombolysis for venous thromboembolism (VTE); major bleeding occurred in 2.6% with no maternal mortality.83 The peri-partum period poses a challenge with greater risk of haemorrhage associated with thrombolysis. The use of mechanical disruption, lower-dose catheter-directed thrombolysis and surgical embolectomy has been described and is dependent on local availability.30 ,84 ,85
Suggested approach: Therapeutic LMWH is the anticoagulant of choice in pregnancy. Systemic thrombolysis should be administered for massive PE in pregnancy; however if bleeding risk is high (eg, in the peripartum period) then surgical or mechanical methods are suggested, depending on local availability.
An echo demonstrates thrombus in the right atrium: what is the optimal management?
Right atrial thrombus occurs in 4–8% of patients with acute PE.86–90 Two main types of thrombus have been described: type A has high early mortality and consists of long, thin, worm-like mobile thrombi associated with clinically severe PE.91 Low cardiac output, higher pulmonary arterial pressure and more severe tricuspid regurgitation may slow transit of clot from peripheral veins to the pulmonary vasculature.90 Type B consists of immobile, non-specific thrombi with absence of associated PE in 60% of cases and low early mortality. A small proportion of thrombi are intermediate in character (type C), being mobile but not worm-like in shape, and have the potential to obstruct right atrial or ventricular outflow.91–93 CTPA is highly effective at identifying type A thrombi with a sensitivity of 100%, although false positives may be observed in patients with non-dilated right ventricles due to incomplete contrast filling.89 The optimal management of patients with right atrial thrombus is unclear. Two-week mortality in 42 patients treated with heparin, thrombolysis or surgical embolectomy was equally poor (20–25%).86 A systematic review of 177 cases observed lower mortality in patients receiving thrombolysis (11%) compared with anticoagulation (29%) and surgery (24%).94 In a series of 16 consecutively thrombolysed patients, right atrial thrombus disappeared in all patients within 24 h with 30-day survival of 100%.90 In a minority of patients thrombus may straddle a PFO leading to additional risk of systemic embolisation. A literature review of 88 such patients demonstrated similar mortality (14%) but higher incidence of stroke in patients treated with anticoagulation rather than surgical embolectomy.95 Patients treated with thrombolysis had a much higher mortality (36%), although they had more haemodynamic compromise. AHA guidelines therefore recommend surgical embolectomy as the optimal treatment in this group.
Suggested approach: Thrombolysis is suggested for type A thrombus while type B thrombus may be treated with anticoagulation alone. Surgical embolectomy is suggested for thrombus straddling a PFO; if this is not available then anticoagulation alone is a reasonable approach unless thrombolysis is indicated due to the severity of the underlying PE. Surgical embolectomy is suggested for type C thrombus if the thrombus is extremely large and associated with risk of right atrial or ventricular outflow tract obstruction should it dislodge.
Which patients with acute PE may benefit from an IVC filter?
Retrievable IVC filter insertion in acute PE should be performed if anticoagulation is contraindicated or temporary cessation of anticoagulation within 1 month is envisaged. An RCT of IVC filter insertion involving 400 patients with proximal DVT receiving anticoagulation demonstrated a reduction in subsequent PE, counterbalanced by an increase in recurrent DVT with no effect on mortality.96 ,97 ACCP guidelines recommend against IVC filter insertion in patients with PE receiving anticoagulation, although they recognise that there is uncertainty regarding the risk and benefits in patients with hypotension.2 Retrospective analysis of data collected by the International Cooperative PE Registry found IVC filter insertion to be associated with a reduced 90-day mortality in the setting of massive PE, although only 10% of patients received IVC filters and two-thirds of patients did not receive thrombolysis.98 A large RCT of retrievable IVC filter insertion in patients with PE and associated DVT (PREPIC-2) has recently been presented in abstract form.99 No effect on recurrent PE, complications or mortality was observed.
Suggested approach: We generally limit IVC filter use in acute PTE to the small number of patients in whom anticoagulation is contraindicated. Routine placement of IVC filters in submassive PE and proximal DVT is not supported by current evidence. If possible we use retrievable filters, which should ideally be removed within the recommended time scale.
How can I differentiate between acute and chronic PE?
It is not infrequent to see patients with significant proximal chronic thromboembolic disease who have erroneously been thrombolysed. Several factors may suggest chronic rather than acute PE. Long duration of symptoms, a previous VTE, features of pulmonary hypertension on examination in the absence of systemic hypotension and or tachycardia and bilateral bruits due to stenoses (appreciated during breath hold on auscultation) favour a chronic process.100 Electrocardiographic and echocardiographic changes indicating longstanding increased RV afterload include a dominant R wave in V1 with absence of tachycardia and a systolic pulmonary artery pressure >60 mm Hg on echocardiography (the RV cannot acutely generate a higher pressure).101 McConnell's sign (RV free wall hypokinesis with preserved RV apical contraction) and the ‘60/60’ sign (pulmonary acceleration time below 60 ms with a tricuspid gradient of 30–60 mm Hg on echocardiography) suggest acute rather than chronic PE.56 ,57 An increased RV–LV ratio may be present in acute and chronic thromboembolic disease but the presence of RV hypertrophy and large bronchial arteries are suggestive of chronic disease.102 Within the pulmonary arteries, mural calcified thrombus forming an obtuse angle with the vessel wall, completely stenosed and narrowed segmental vessels and signs of recanalisation with contrast flowing either side of ‘webs’ of organised thrombi are suggestive of chronic disease.103 Within the parenchyma, peripheral wedge-shaped infarcts may be present in acute PE while a mosaic perfusion pattern with reduction in pulmonary arterial size in low attenuation areas of the lung suggests chronic disease.
Suggested approach: A chronic history, markedly elevated systolic pulmonary arterial pressures, RV and bronchial artery hypertrophy, thrombus calcification, webs and a mosaic perfusion pattern should raise suspicion of chronic rather than acute PE.
In which patients should I consider early discharge?
Clinical severity scoring systems have more clearly identified patients at low risk of complications from acute PE who may not require hospitalisation. PESI is the most widely validated scoring system and has been utilised in prospective randomised management studies to demonstrate the safety of such an approach, which may be appropriate in up to 40% of patients.104 ,105 ,106 Outpatient management of PE generally requires LMWH administration during oral anticoagulant initiation,105 although the introduction of oral factor Xa inhibitors provides the possibility of a more convenient ambulatory treatment for patients.107 ,108 Safe early discharge of patients is dependent on a robust multidisciplinary approach involving rapid imaging, accurate assessment and adequate support and follow-up mechanisms for the discharged patient. Admission of patients at very low or low risk may still be advisable due to patient concern or ongoing pain. Repeat PESI scoring after 48 h in patients initially assessed as unsuitable for discharge may reclassify them as appropriate for outpatient anticoagulation.109
Suggested approach: Patients with a very low or low PESI score may be offered early discharge and outpatient anticoagulation but a robust system of support and follow-up is mandatory.
What is the role of the newer oral agents in the management of acute PE?
At the time of writing, rivaroxaban, a direct Xa inhibitor, is the only new oral anticoagulant (NOAC) licensed and approved for the treatment and secondary prevention of DVT and PE in the UK.108 Other NOACs (dabigatran: a direct thrombin inhibitor; and apixaban and edoxaban: direct Xa inhibitors), however, have also been shown to be non-inferior to conventional anticoagulant therapy with favourable safety profiles in the treatment of patients with PE. In the dabigatran (RE-COVER)110 and edoxaban (HOKUSAI)111 studies, all patients initially received heparin for 10 and 7 days. The rivaroxaban (EINSTEIN-PE) and apixaban (AMPLIFY) studies excluded patients who had had more than 48 h of heparin, thus these therapies introduce the option of managing patients without parenteral anticoagulation.112 ,113 Patients with active malignancy were excluded from these studies (LMWH remains the standard of care for these patients), and NOACs should not be used in pregnant or lactating women, or in patients with significant renal impairment. Importantly, no regular monitoring is required.
Suggested approach: Other NOACs may also become licensed in the future but currently rivaroxaban can be considered an option for the acute management of haemodynamically stable patients with PE within its product license.
What should I do about an incidental or isolated subsegmental PE?
Demonstration of unsuspected PE occurs in up to 5% of thoracic CT scans performed for non-PE indications, the majority in the context of malignancy.114–118 In malignancy, most incidental PEs are lobar or segmental in distribution while VTE recurrence rate, mortality and complications are not significantly different between incidental and symptomatic PE.119 The ACCP guidelines suggest asymptomatic PE should be treated as symptomatic PE.2 Isolated subsegmental PEs are demonstrated in 1–5% of CTPAs performed for suspected PE and 10% of CTPAs with demonstrable PE.114 ,120–122 Optimal management of these patients is unclear. Pooled data from 105 patients with predominantly isolated subsegmental PE with no evidence of DVT on serial imaging who did not receive anticoagulation found no patients with recurrent PE after 3 months.120 The authors therefore suggested that the risk of haemorrhage may outweigh the benefit of anticoagulation in isolated PE, assuming negative serial compression ultrasound of lower limb veins.120 This approach has been challenged by a large prospective study which found similar risk factors and recurrence rates in patients with symptomatic subsegmental versus more proximal PE, although it is unclear how many patients had a single subsegmental PE.123
Suggested approach: Incidental and isolated subsegmental PE should generally be managed in the same manner as symptomatic and non-subsegmental PE.
The suggested approaches are based on a review of the available evidence and guidelines and on our clinical experience. For many of the dilemmas the evidence base is not substantial and is potentially hampered by reporting bias. Management in an individual patient will require clinical assessment of risks and benefits and will also depend on local availability of therapeutic interventions.
Review history and Supplementary material
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors RC and DGK conceived the article. All authors wrote sections and reviewed the whole of the manuscript.
Competing interests RC and DGK have received funding for conference attendance and honoraria for advisory boards and lecturing from Bayer and GSK, funding for conference attendance from Pfizer and their unit has received an unrestricted educational grant to part-fund a radiographer from Bayer. CAE has received funding for conference attendance and honoraria for lecturing from Bayer, and for conference attendance and honoraria for advisory boards and lecturing from GSK. JH has received funding for conference attendance and honoraria for advisory boards from Bayer. RJH has received honoraria for advisory boards and lecturing from Bayer, Boehringer Ingelheim and Sanofi Aventis. RMM has received funding to attend meetings from Daiichi-Sankyo, Bayer and Boehringer Ingelheim, and received honoraria for advisory board meetings and/or delivering educational sessions for Bayer, Daiichi-Sankyo and Pfizer. IS receives an annual unrestricted educational grant from GSK to support a clinical meeting. JVV has received funding to attend meetings from Boehringer Ingelheim and honoraria for advisory boards from Bayer.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.