You are here

Article Text

Article menu

Download PDFPDF

Original article
Atrial fibrillation in acute pulmonary embolism: prognostic considerations
  1. Sérgio Nuno Craveiro Barra1,
  2. Luís Vilardouro Paiva1,
  3. Rui Providência1,2,
  4. Andreia Fernandes1,
  5. António Leitão Marques1
  1. 1Cardiology Department, Coimbra University and Hospital Centre, Coimbra, Portugal
  2. 2Coimbra's Medical School, University of Coimbra, Coimbra, Portugal
  1. Correspondence to Dr Sérgio Nuno Craveiro Barra, Centro Hospitalar e Universitário de Coimbra, Hospital Geral, Quinta dos Vales, 3041-801 S. Martinho do Bispo, Portugal; sergioncbarra{at}gmail.com

Abstract

Aims Although it is accepted that atrial fibrillation (AF) may be both the contributing factor and the consequence of pulmonary embolism (PE), data on the prognostic role of AF in patients with acute venous thromboembolism are scarce. Our aim was to study whether AF had a prognostic role in patients with acute PE.

Methods Retrospective cohort study involving 270 patients admitted for acute PE. Collected data: past medical history, analytic/gasometric parameters, admission ECG and echocardiogram, thoracic CT angiography. Patients followed for 6 months. An analysis was performed in order to clarify whether history of AF, irrespective of its timing, helps predict intrahospital, 1-month and 6-month all-cause mortality.

Results Patients with history of AF, irrespective of its timing (n=57, 21.4%), had higher intrahospital (22.8% vs 13.1%, p=0.052, OR 2.07, 95% CI 0.98 to 4.35), 1-month (35.1% vs 16.9%, p=0.001, OR 3.16, 95% CI 1.61 to 6.21) and 6-month (45.6% vs 17.4%, p<0.001, OR 4.67, 95% CI 2.37 to 9.21) death rates. The prognostic power of AF was independent of age, NT-proBNP values, renal function and admission blood pressure and heart rate and additive to mortality prediction ability of simplified PESI (AF: p=0.021, OR 2.31, CI 95% 1.13 to 4.69; simplified PESI: p=0.002, OR 1.47, CI 95% 1.15 to 1.89). The presence of AF at admission added prognostic value to previous history of AF in terms of 1-month and 6-month all-cause mortality prediction, although it did not increase risk for intrahospital mortality.

Conclusions The presence of AF, irrespective of its timing, may independently predict mortality in patients with acute PE. These data should be tested and validated in prospective studies using larger cohorts.

  • Pulmonary Embolism
  • Arrythmia
  • Cardiac Care, Treatment
  • Thrombo-Embolic Disease, Management
  • Acute Medicine-Other

https://doi.org/10.1136/emermed-2012-202089

Statistics from Altmetric.com

    Request Permissions

    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.

    Introduction

    To this date, the potential relationship between the most frequent atrial arrhythmia and acute venous thromboembolism is still controversial. Stein PD and colleagues did not find atrial fibrillation (AF) in 90 patients with documented submassive/massive pulmonary embolism (PE), suggesting it would appear more typically in patients with pre-existent cardiac disease,1 while other investigators reported AF occurred in 15–26% of patients with PE.2 The few large-scale studies analysing such potential relationship focused on diagnostic purposes and did not demonstrate any significant association between AF and PE.3 In fact, Gex G and colleagues concluded the presence of AF does not increase the probability of PE when this diagnosis is suspected.4 Nevertheless, it is widely accepted that PE may have its origin on right atrial thrombi5 resulting from AF enhancement of platelet aggregation and coagulation.6 ,7

    The reported prevalence of AF in acute PE varies widely between studies,1 ,2 and its diagnostic usefulness in this context remains controversial, but information regarding its role as a prognosticator is even scantier. The prognostic value of AF in patients admitted for acute PE has never been unequivocally demonstrated and, although it may be logical to expect AF to help predict outcomes in this condition, some studies have suggested the opposite.8 The prognostic power of the PESI9 or simplified PESI10 rules has been extensively validated,11–13 but these scores do not include history of AF or AF at admission, although the parameter ‘heart rate above 110 beats per minute’ may indeed include the majority of uncontrolled AF cases. The Low Risk PE Decision (LR-PED) rule, derived from a low-risk cohort of patients, was the first risk score for PE that included AF,14 highlighting the fact that rhythm may be additionally important (alongside rate) when attempting to prognosticate in acute PE.

    This research aims primarily at studying the potential value of AF as a prognosticator in acute PE, namely its ability to predict all-cause mortality in the short and mid-term. In addition, we aim at analysing whether the presence of AF at admission adds prognostic value to past history of AF.

    Materials and methods

    Study design

    Retrospective cohort study included all patients with a diagnosis of acute PE between 1 January 2007 and 30 April 2011. Using collected baseline data at the time of PE diagnosis and outcome data from this cohort, we retrospectively assessed the intrahospital, 1-month and 6-month prognostic value of past history of AF, irrespective of its timing, duration and overall characteristics. We then assessed whether the presence of AF at admission would be additive to past history of AF in terms of outcome prediction.

    Patients and eligibility criteria

    Between 1 January 2007 and 30 April 2011, a total of 270 patients (age 70.1±15.8, from 28 to 97) were admitted at the emergency department of a tertiary referral hospital and university centre with a primary International Classification of Diseases diagnosis of PE (confirmed by thoracic CT angiography or transthoracic echocardiogram; in the latter case, through detection of inferior vena cava, intracardiac or pulmonary artery thrombus in the appropriate clinical context) or a secondary diagnosis of PE plus one primary diagnosis representing a complication of the PE itself (such as respiratory failure, cardiogenic shock or cardiac arrest). This sample included the presumably low-risk cohort of patients from which the LR-PED rule was derived.14 Incidental or perioperative PE cases were not included, but patients admitted for PE some days after being discharged following surgery would fulfil criteria for inclusion.

    Data collection

    Through extensive chart review done by two coinvestigators, the following data were collected: demographic features, cardiovascular and PE risk factors, history of AF, stroke, pulmonary conditions and heart failure, analytical study (including haemoglobin, glycaemia, NT-proBNP, C-reactive protein, creatinine, d-dimer, troponin I, prothrombin time), blood pressure, heart rate and gasometric parameters at admission, results of ECG and transthoracic echocardiography (when performed). History of AF was defined as previous electrocardiographic documentation of at least one episode of this arrhythmia (through 12-lead ECG, 24-h Holter, electrocardiographic or event monitoring), irrespective of its timing, duration (paroxysmal, persistent or permanent) and overall patient characteristics (lone AF, valvular or non-valvular AF) and symptomatology. In order to collect this information, a comprehensive review of available electrocardiographic recordings was made, along with a review of history notes from cardiology or internal medicine outpatient consultations or emergency department admissions. Table 1 describes the study sample. The simplified Pulmonary Embolism Severity Index (PESI) scores were retrospectively assessed. Data collection was blind, as the two coinvestigators assigned to this task were not aware of the purpose of the study.

    View this table:
    Table 1

    Sample characterisation

    Study endpoints

    The primary endpoint was all-cause mortality 1 month following diagnosis of acute PE. All-cause mortality is the most robust, unbiased and objective endpoint, as the specific cause of mortality in PE is often very hard to ascertain. Furthermore, 30-day mortality is the most common endpoint in studies evaluating potential risk scores for patients with PE.9 ,10 ,14 ,15 Secondary outcomes were intrahospital and 6-month all-cause mortality. Mortality was assessed using hospital chart review or proxy interviews.

    Statistical analysis

    Statistical analysis was performed using SPSS, V.17.0:

    • χ2 Test, Student t test and non-parametric equivalent tests when appropriate

    • Regression estimation to replace missing values (whenever the number of missing values was negligible, otherwise cases with missing values would be omitted)

    • Univariate analysis to (A) evaluate a potential association between history of AF and the study endpoints; (B) study whether AF at admission would improve outcome prediction

    • Multivariate analysis with logistic regression to: (A) assess whether AF independently predicted mortality in patients with PE; (B) study whether the presence of AF at admission or any time in the past would add prognostic value to the simplified PESI score

    • Kaplan–Meier curves to evaluate impact of AF on patient survival.

    • Assessment of the impact of adding AF to the simplified PESI score. A new risk score, entitled ‘Simplified PESI+AF’, was developed by adding one extra point in patients with history of AF, and compared with simplified PESI using receiver operating characteristic (ROC) curves in the prediction of primary and secondary endpoints. Area under the curve (AUC) comparisons performed using MedCalc for Windows V.9.2.0.1.

    Results

    Overall characteristics of study population

    Patients in our sample were older and had a higher prevalence of most comorbidities (in particular, heart failure, cerebrovascular disease and malignancy) compared with those in PESI-derivation samples.9

    Fifty-seven patients (21.1%) had history of AF. This information cannot be compared with that of PESI-derivation sample owing to the absence of information regarding prevalence of AF in the latter study. Seventeen (29.8%) patients had permanent AF, 27 (47.4%) were categorised as paroxysmal and/or persistent, and the remaining either as de novo AF or uncategorised AF (for lack of information). Thirty-two patients (11.9%) presented with AF at admission.

    AF as a prognosticator in acute PE

    Patients with history of AF were older (81.2±8.0 vs 67±16.1, p<0.001), had higher NT-proBNP (10 590±18 358 vs 3809±5700, p=0.002) and lower haemoglobin (12.1±1.8 vs 13±2.1, p=0.016) values, higher Geneva (36.7±26.1 vs 18.4±18.5, p<0.001) and simplified PESI (2.3±1.0 vs 1.4±1.3, p<0.001) scores. Heart failure was significantly more frequent in patients with previous AF (54.4% vs 15.4%, p<0.001), and there was a trend for higher prevalence of cerebrovascular disease (24.6% vs 14.8%, p=0.082). Hypotension and clinical and/or echocardiographic signs of right ventricular dysfunction were not more frequent in patients with AF, and fibrinolytic therapy was as likely to be given to these patients as it was for their counterparts.

    Primary endpoint: 1-month all-cause mortality

    History of AF was associated with a higher risk for the occurrence of the primary outcome in univariate analysis (table 2). Furthermore, the presence of AF at admission added prognostic value to previous history of AF (table 3).

    View this table:
    Table 2

    Risk of mortality according to the presence or absence of atrial fibrillation at admission

    View this table:
    Table 3

    Death rates according to risk stratification by the simplified PESI+AF and simplified PESI scores

    In multivariate analysis, a predictive model for the occurrence of the primary endpoint included variables ‘history of AF’ (p=0.002, OR 5.89, CI 95% 1.91 to 18.16) and ‘heart rate at admission’ (p=0.029, OR 1.018, CI 95% 1.002 to 1.033), excluding age, NT-proBNP, glycaemia, creatinine, C-reactive protein, troponin I and haemoglobin values, blood pressure, arterial oxygen tension and carbon dioxide tension at admission, and history of heart failure, diabetes mellitus or stroke. The Hosmer–Lemeshow test (p=0.950) indicated that the overall model fit was good.

    AF added prognostic value to simplified PESI. Using logistic regression, an overall model for 1-month all-cause mortality prediction included simplified PESI (p=0.002, OR 1.47, CI 95% 1.15 to 1.88) and history of AF (p=0.021, OR 2.30, CI 95% 1.13 to 4.69) (Hosmer–Lemeshow test: p=0.754).

    Secondary endpoints: intrahospital mortality and 6-month all-cause mortality

    History of AF associated with a higher risk for the occurrence of the secondary outcomes in univariate analysis (table 2). The presence of AF at admission added prognostic value to previous history of AF in terms of 6-month all-cause mortality prediction, although it did not help predict intrahospital mortality beyond previous history of AF (table 3).

    In multivariate analysis, a predictive model for intrahospital mortality prediction (Hosmer–Lemeshow test: p=0.10) included ‘age’ (p=0.021, OR 1.031, CI 95% 1.01 to 1.06) and ‘heart rate at admission’ (p=0.002, OR 1.028, CI 95% 1.01 to 1.05), excluding history of AF, NT-proBNP, glycaemia, creatinine, C-reactive protein, troponin I and haemoglobin values, blood pressure, arterial oxygen tension and carbon dioxide tension at admission, and history of heart failure, diabetes mellitus or stroke.

    A model for 6-month all-cause mortality prediction (Hosmer–Lemeshow test: p=0.953) included ‘history of AF’ (p=0.023, OR 2.49, CI 95% 1.14 to 5.44), ‘heart rate at admission’ (p<0.001, OR 1.039, CI 95% 1.02 to 1.06) and ‘age’ (p<0.001, OR 1.053, CI 95% 1.02 to 1.08), excluding all other variables mentioned before.

    AF added prognostic value to simplified PESI in terms of 6-month all-cause mortality prediction. The overall multivariate model included simplified PESI (p<0.001, OR 1.86, CI 95% 1.43 to 2.42) and history of AF (p=0.002, OR 3.09, CI 95% 1.49 to 6.38) (Hosmer–Lemeshow test: p=0.650).

    The impact of AF on patient survival and timing of events can be seen on a Kaplan–Meier curve (figure 1). It demonstrates the poorer prognosis of patients with history of AF, following their admission for acute PE (log-rank: p<0.001).

    Figure 1
    Figure 1

    Kaplan–Meier curves demonstrating the impact of atrial fibrillation on patient survival and timing of events.

    Simplified PESI+AF score versus simplified PESI

    Death rates, according to risk stratification by simplified PESI and simplified PESI+AF, are illustrated in table 3. ROC curves comparing both scores in the prediction of the primary endpoint are seen in figure 2. PESI+AF score demonstrated marginally superior AUC in the prediction of primary and secondary outcomes (AUC 0.682 vs 0.673, p=0.547; AUC 0.687 vs 0.670, p=0.205; AUC 0.772 vs 0.752, p=0.148 for intrahospital, 1-month and 6-month all-cause mortality, respectively), although differences were not statistically significant.

    Figure 2
    Figure 2

    Receiver operating characteristic curves comparing the simplified PESI and simplified PESI+atrial fibrillation scores in terms of 1-month mortality prediction.

    Discussion

    Our research suggests AF may be an independent prognosticator in symptomatic ambulatory patients with acute PE. Although patients with this arrhythmia were older and had higher prevalence of heart failure and/or cerebrovascular disease, the prognostic impact of AF 1 month and 6 months postadmission was independent of such variables, as shown by: (A) the inclusion of ‘history of AF’ along with ‘heart rate at admission’ in a multivariate predictive model for 1-month all-cause mortality that excluded age, NT-proBNP, blood pressure at admission and history of heart failure; (B) the inclusion of ‘history of AF’ along with ‘heart rate at admission’ and ‘age’ in a multivariate predictive model for 6-month all-cause mortality.

    The importance of AF as a mortality predictor was further suggested by its ability to add prognostic power to the simplified PESI rule, as shown by multivariate analysis. Also, the simplified PESI+AF score seemed slightly superior in its ability to discriminate patients at highest risk for 1-month all-cause mortality, showing higher sensitivity and negative predictive value for the detection of patients at risk of mortality. Patients categorised as low risk by this model showed lower mortality risk compared with their counterparts classified as low risk by the simplified PESI rule. It defined an apparently truly low-risk group.

    Furthermore, our study suggested a progressively increasing prognostic value of AF following discharge, supported by the higher impact on mortality as time passed. In fact, history of AF was not an independent predictor of intrahospital mortality, but its prognostic power was seen at the 1-month mark and even more pronounced at 6 months postdischarge.

    The presence of AF at admission showed additive prognostic impact. Patients with history of AF, and presenting with this arrhythmia at admission, showed higher 1-month and 6-month mortality risk, despite similar intrahospital death rates.

    Data on this topic are surprisingly scarce. The fact that an association between AF and overall mortality has been previously demonstrated in several other conditions16–19 and even in otherwise healthy individuals20 may have swept away the focus on the importance and impact of AF in patients with acute PE. For seemingly good reasons, AF was expected to increase mortality risk in a cohort of patients with PE. It gives rise to thrombogenesis,6 ,21 ,22 and it has important haemodynamic implications. In fact, right atrial function plays an important role in the unfavourable haemodynamic milieu of PE, but AF removes it, decreasing right ventricle preload and, subsequently, left ventricle preload. Also, by compromising left atrial function and increasing heart rate, it further diminishes left ventricular filling, increases myocardial oxygen consumption and shortens diastole reducing myocardial oxygen supply. For these reasons, Koracevic et al believed it was logical to expect AF to help predict inhospital outcome of acute PE, but their research unexpectedly suggested AF was not related to inhospital mortality,23 contradicting an old study by Weber et al.24 Koracevic and colleagues analysed 140 patients with acute PE, of whom 15.7% had AF, and 13.6% died during hospitalisation, finding no association between the presence of AF and an increased risk for intrahospital mortality. Despite its merit, their study had some limitations, such as the lack of proper subgroup statistics (‘new onset AF’ vs ‘previously diagnosed AF’) and the absence of multivariate analysis regarding the potential influence of age, as mentioned by the authors themselves.23

    Our study included a larger cohort of patients with acute PE (approximately twice the patients studied by Koracevic et al23), a more versatile subgroup analysis (‘history of AF’ and ‘AF at admission’), multivariate analysis to demonstrate whether AF would be an independent predictor of mortality, and a 6-month follow-up. We corroborated the findings of Koracevic et al in regard to intrahospital mortality, as previously diagnosed AF was not an independent predictor of this secondary endpoint in our cohort. Nevertheless, history of AF predicted 1-month and 6-month all-cause deaths, and the presence of AF at admission was additive to history of AF in terms of 1-month and 6-month risk prediction.

    Importance of the study

    In this study, the role of AF as a prognosticator in acute PE has been evaluated in a reasonably sized cohort of patients. Its potential role as an independent predictor of 1-month and 6-month all-cause mortality has been conjectured after adjustment for some confounding variables and, to the best of our knowledge, this is the first study to date suggesting it. If validated in wider prospective studies, these findings may eventually influence the aggressiveness of treatment in patients with PE, and the duration of anticoagulant therapy. However, caution should be taken at the fact that more aggressive anticoagulation may even worsen outcome in a frail patient with multiple comorbidities. Furthermore, if it is known that AF caused a particular pulmonary embolus and associates with a poorer prognosis in such context, intervention might also be aimed at the arrhythmia itself, eventually with antiarrhythmic or, in selected cases with low comorbidity, ablation therapy, in addition to that directed at the clotting.

    Despite all these considerations, we highlight that, although AF may add independent prognostic value in this context, substantial therapeutic implications are not necessarily present, as AF may be a marker for unmeasured comorbidities or general frailty that will not be amenable to treatment. AF may also be a marker for general ill-health that is not measured by the variables included in the multivariate analysis. Furthermore, we could not show that the presence of this arrhythmia leads to unequivocal clinically worthwhile improvements in mortality prediction. Our findings thus require prospective validation. We can, nevertheless, conclude that the presence of an abnormal rhythm, such as AF, in addition to high ventricular rate, may be of value when assessing prognosis in acute PE and, therefore, as patients with PE and previously diagnosed AF share a poorer prognosis, they should be followed more closely/aggressively.

    Limitations of the study

    Although this is the largest study to date analysing the importance of AF in acute PE, our findings should be tested and prospectively validated in larger (preferably multicentre) cohorts. Our study sample is underpowered for the ROC curve analysis. A potential improvement in mortality prediction by the addition of AF to the simplified PESI score should be evaluated in samples comprising higher numbers of patients.

    The retrospective design of this study is not ideal to assess its primary aim, and potentially renders the findings susceptible to bias and confounding. The potential for hidden, unexplored, confounders should be underscored. We reinforce the importance of prospective validation of the main results of this research.

    Furthermore, a comparator group (eg, patients admitted for respiratory illnesses other than PE) could allow a more comprehensive evaluation and comparison of the prognostic role of AF in different respiratory conditions.

    The reason why all-cause mortality was chosen as the primary outcome has been stated before. However, this still represents a limitation of the present study. The cause of death is important when analysing the impact of AF on prognosis. AF is a major cause of cardioembolic stroke, which could be the cause of death. However, AF has already been shown to associate with higher mortality risk even when limiting analysis to subjects free of stroke.25 Thus, AF's impact on prognosis should not be thought of as limited to its association with cerebrovascular events.

    Although it has been previously shown to predict mortality in this context, history of malignancy was not included in the multivariate analysis for construction of the predictive models, as data regarding the status of the disease (active vs in remission/cured) was not available for a considerable percentage of patients. However, as cured cancer does not share the same prognosis as an active one, including malignancy irrespective of current status could add a further bias to the model.

    Few patients had been previously anticoagulated (3.7% of the sample). This should not be considered a true limitation though, as appropriately anticoagulated patients are at lower thromboembolic risk and, therefore, not expected to suffer a PE. We were not expecting a high percentage of previously anticoagulated individuals in a cohort of patients with PE.

    Data regarding anticoagulation levels during follow-up or time in therapeutic range are lacking. Nevertheless, patients were given low-molecular-weight heparin (enoxaparin) until an INR>2.0 was achieved. Thereafter, most patients performed serial INR measurements and warfarin daily doses were adjusted by accompanying physicians in the primary care setting according to current clinical practice.

    Conclusion

    Our data suggested a potential relationship between AF and short/mid-term prognosis in patients with acute pulmonary venous thromboembolism. The presence of an abnormal rhythm, in addition to high ventricular rate, may be of value when assessing prognosis in acute PE. While this may be linked to unknown/unexplored confounding factors, further prospective study is warranted. If an independent prognostic role is unequivocally demonstrated in future studies, it may alter clinical decision making regarding disposition, therapeutic and follow-up protocols.

    References

    1. ↵
      1. Stein P,
      2. Dalen J,
      3. McIntyre K,
      4. et al
      . The electrocardiogram in acute pulmonary embolism. Prog Cardiovasc Dis 1974;17:247–57.
      OpenUrl
    2. ↵
      1. Martins L,
      2. Reiss I
      . A diagnosis of pulmonary embolism. J Emerg Med 2004;26:225–7.
      OpenUrlCrossRefPubMed
    3. ↵
      1. Rodger M,
      2. Makropoulos D,
      3. Turek M,
      4. et al
      . Diagnostic value of the electrocardiogram in suspected pulmonary embolism. Am J Cardiol 2000;86:807–9.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Gex G,
      2. Gerstel E,
      3. Righini M,
      4. et al
      . Is atrial fibrillation associated with pulmonary embolism? J Thromb Haemost 2012;10:347–51.
      OpenUrlCrossRefPubMed
    5. ↵
      1. Torbicki A,
      2. Galié N,
      3. Covezzoli A,
      4. et al
      . Right heart thrombi in pulmonary embolism. J Am Coll Cardiol 2003;41:2245–51.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Sohara H,
      2. Amitani S,
      3. Kurose M,
      4. et al
      . Atrial fibrillation activates platelets and coagulation in a time-dependent manner: a study in patients with paroxysmal atrial fibrillation. J Am Coll Cardiol 1997;29:106–12.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Aberg H
      . Atrial fibrillation. I: a study of atrial thrombosis and systemic embolism in a necropsy material. Acta Med Scand 1969;185:373–9.
      OpenUrlPubMedWeb of Science
    8. ↵
      1. Koracevic G,
      2. Atanaskovic V
      . Is atrial fibrillation a prognosticator in acute pulmonary thromboembolism? Med Princ Pract 2010;19:166.
      OpenUrlCrossRefPubMed
    9. ↵
      1. Aujesky D,
      2. Obrosky D,
      3. Stone R,
      4. et al
      . Derivation and validation of a prognostic model for pulmonary embolism. Am J Respir Crit Care Med 2005;172:1041–6.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Jiménez D,
      2. Aujesky D,
      3. Moores L,
      4. et al
      . Simplification of the pulmonary embolism severity index for prognostication in patients with acute symptomatic pulmonary embolism. Arch Intern Med 2010;170:1383–9.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Donzé J,
      2. Le Gal G,
      3. Fine M,
      4. et al
      . Prospective validation of the Pulmonary Embolism Severity Index. A clinical prognostic model for pulmonary embolism. Thromb Haemost 2008;100:943–8.
      OpenUrlPubMedWeb of Science
    12. ↵
      1. Moores L,
      2. Aujesky D,
      3. Jiménez D,
      4. et al
      . Pulmonary Embolism Severity Index and troponin testing for the selection of low-risk patients with acute symptomatic pulmonary embolism. J Thromb Haemost 2010;8:517–22.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Venetz C,
      2. Jiménez D,
      3. Mean M,
      4. et al
      . A comparison of the original and simplified Pulmonary Embolism Severity Index. Thromb Haemost 2011;106:423–8.
      OpenUrlCrossRefPubMed
    14. ↵
      1. Barra S,
      2. Paiva L,
      3. Providência R,
      4. et al
      . LR-PED rule: Low Risk Pulmonary Embolism Decision rule—a new decision score for low risk pulmonary embolism. Thromb Res 2012;130:327–33.
      OpenUrlCrossRefPubMed
    15. ↵
      1. Jiménez D,
      2. Yusen R,
      3. Otero R,
      4. et al
      . Prognostic models for selecting patients with acute pulmonary embolism for initial outpatient therapy. Chest 2007;132:24–30.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Jabre P,
      2. Jouven X,
      3. Adnet F,
      4. et al
      . Atrial fibrillation and death after myocardial infarction: a community study. Circulation 2011;123:2094–100.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Walkey A,
      2. Wiener R,
      3. Ghobrial J,
      4. et al
      . Incident stroke and mortality associated with new-onset atrial fibrillation in patients hospitalized with severe sepsis. JAMA 2011;306:2248–54.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Chamberlain A,
      2. Redfield M,
      3. Alonso A,
      4. et al
      . Atrial fibrillation and mortality in heart failure: a community study. Circ Heart Fail 2011;4:740–6.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Genovesi S,
      2. Vincenti A,
      3. Rossi E,
      4. et al
      . Atrial fibrillation and morbidity and mortality in a cohort of long-term hemodialysis patients. Am J Kidney Dis 2008;51:255–62.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Conen D,
      2. Chae C,
      3. Glynn R,
      4. et al
      . Risk of death and cardiovascular events in initially healthy women with new-onset atrial fibrillation. JAMA 2011;305:2080–7.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Ohara K,
      2. Inoue H,
      3. Nozawa T,
      4. et al
      . Accumulation of risk factors enhances the prothrombotic state in atrial fibrillation. Int J Cardiol 2008;126:316–21.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Kahn S,
      2. Solymoss S,
      3. Flegel K
      . Nonvalvular atrial fibrillation: evidence for a prothrombotic state. CMAJ 1997;157:673–81.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Koracevic G,
      2. Atanaskovic V,
      3. Salinger S,
      4. et al
      . Atrial fibrillation is not a predictor of in-hospital mortality in pulmonary embolism. Eur Respir J 2007;30(suppl 51):80.
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. Weber D,
      2. Phillips J
      . The prognostic value of supraventricular arrhythmias in acute pulmonary embolism. Vasc Dis 1966;3:393–7.
      OpenUrlPubMed
    25. ↵
      1. Benjamin E,
      2. Wolf P,
      3. D'Agostino R,
      4. et al
      . Impact of atrial fibrillation on the risk of death: the Framingham Heart Study. Circulation 1998;98:946–52.
      OpenUrlAbstract/FREE Full Text

    Footnotes

    • Contributors SNCB wrote the draft version of the paper. RP, LVP and AF provided further ideas for improving the manuscript and contributed to the final version. All authors collected data. ALM supervised the production of the article.

    • Competing interests None.

    • Ethics approval Institutional Board.

    • Provenance and peer review Not commissioned; externally peer reviewed.