Abstract
Among the various clinical scoring methods used for the prediagnosis of pulmonary embolism (PE), Wells criteria is the most common. It relies on the findings and story of deep venous thrombosis (DVT), PE and malignancy. It is known that atrial fibrillation (AF) is a risk factor for PE like as DVT or malignancy. We aimed to evaluate the possibility of diagnosing more patients with PE by including AF in the Wells criteria. This prospective study included 250 patients admitted to the emergency department with PE findings. Wells scoring and Wells scoring with AF were performed for each patient. Out of 250 patients, 165 patients were diagnosed as PE. Wells score was >4 in 61.8% of patients with PE and 28.2% of patients without PE. Out of false negative 63 patients with PE, 21 of them had AF. According to Wells scoring with AF the score of 148 (89.7%) patients with PE diagnosis was ≥3, whereas the score of 45 (52.9%) patients without PE was ≥3. AF was detected in 15.8% of patients with PE. The sensitivity of Wells score with AF was significantly higher than that of the Wells score (p<0.001). As a result, when AF, which is one of an important PE cause such as DVT and malignancy, was added to the Wells criteria, an additional correct PE estimate was obtained in 46 patients. We recommend using Wells score with AF since prediagnosing more PE is more valuable than having some false negative PE predictions.
Significance of this study
What is already known about this subject?
Pulmonary embolism (PE) is an important cause of mortality and morbidity.
The rapid diagnosis and treatment of PE are very important.
Various clinical scoring methods have been developed.
One of these methods is Wells scoring. It consists six items: suspected deep venous thrombosis, alternative diagnosis less likely than PE, heart rate >100 beats/min, prior venous thromboembolism, immobilization within prior 4 weeks, active malignancy and hemoptysis.
Then, the patients are categorized according to their total points (likely or unlikely).
What are the new findings?
The patients with PE nearly have 10%–20% of atrial fibrillation (AF).
AF may be the main cause of PE in some cases as it is in stroke.
Even though AF may develop secondary to PE, it may as well cause PE.
Therefore, AF can also be treated as a risk factor for PE and this study investigated and showed the possibility of diagnosing more patients with PE by adding AF to the Wells criteria.
AF was found in 15.8% of patients with PE in the study.
When AF was added as a risk factor to Wells scoring, the sensitivity, overall accuracy and negative predictive value were found to be better than the ones in the Wells score.
How might these results change the focus of research or clinical practice?
Adding AF to Wells score can avoid underdiagnosis of patients with PE when the cause is AF.
Introduction
Pulmonary embolism (PE) is an important cause of mortality and morbidity. The overall incidence of PE was approximately 100–200/100 000 person-years and the rapid diagnosis and treatment of PE is a prognostic factor.1 In order to establish patients with high probability of PE rapidly, various clinical scoring methods have been developed. One of these methods is Wells scoring, the most used scoring system for the probability of PE since 1998.2 In 2000, the Wells score for PE was revised to reduce the number of risk categories to two: likely (more than four points) and unlikely (four points or less).3 It only determines the PE probability in the direction of the findings obtained by anamnesis and physical examination and mostly depends on the findings or story of deep venous thrombosis (DVT) and malignancy.4 But nearly 10%–20% of the patients with PE had atrial fibrillation (AF), and AF may be the main cause of these PE cases.1 5–8
One of the most common clinical findings in PE is AF and can be diagnosed by a simple physical examination. Even though AF may develop secondary to PE, it may as well cause PE. In cases with AF, thrombus development in the right atrium has been shown to be greater than anticipated in autopsies.5 Thrombus formation in the left atrium is a more common condition and right atrium thrombus is also shown in many studies.9 PE may develop either with the passage of the thrombus of the left atrium from the atrial septal defect or opening of foramen ovale due to high left atrial pressure or directly from the right atrium thrombi.
While suggesting AF is also a risk factor for PE like DVT and malignancy, this study was planned to investigate the possibility of diagnosing more patients with PE by adding AF to the Wells criteria.
Methods
The study was carried out between years 2015 and 2017 in Ankara Yıldırım Beyazıt University, School of Medicine and Ankara Atatürk Training and Research Hospital. All subjects gave their written consent to participate. This prospective study included 250 patients admitted to the emergency department with PE symptoms and findings such as chest pain, dyspnea, hemoptysis, cough, tachypnea, tachycardia and who are consulted to pulmonary diseases department. Both Wells score (table 1) and Wells score with AF (table 2) were performed for each patient considered as PE by a pulmonary physician.
Inclusion criteria are being older than 18 years, having symptoms of PE like chest pain, dyspnea, hemoptysis, cough and with the findings of tachypnea, tachycardia. The PE suspected patients who could not be evaluated by CT angiography because of reasons such as having renal failure or pregnancy were excluded from the study.
DVT contributes to Wells score with three points. DVT is a direct source of thrombus. Other criteria in Wells are the signs of PE or the preparing factors of thrombus formation. AF itself is also a source of thrombus. Therefore, AF contribution to updated Wells score was also defined as three points.
Subsequently, the necessary tests for PE diagnosis (ECG, postero-anterior chest X-ray, D-dimer, compression venous ultrasonography [CUS], thoracic CT angiography) were applied in the emergency department by the request of fellow in pulmonary diseases department. Especially thoracic CT angiography was applied for every patient for a definite diagnosis. CT pulmonary angiography was applied by 256 slices Siemens Somatom Definition Flash Single Slice 256. Patients who were not applied thorax CT due to renal insufficiency, pregnancy or allergy were excluded from the study. AF is evaluated by both physical examination and ECG.
The distribution of age was examined by the Shapiro-Wilk test. Since age did not distribute normally, the age of PE group and non-PE group were summarized by median (min-max) and compared by Mann-Whitney U test. Categorical variables were expressed as numbers or percentages and Χ2 test was used to compare these variables between the PE group and non-PE group. The diagnostic performance of Wells score with AF was examined by receiver operating characteristic (ROC) curve analysis. The area under the curve (AUC±SE), the cut-off point determined by the Youden’s index, sensitivity, specificity overall accuracy, positive and negative predictive values (PPV, NPV) for the cut-off point were calculated. Ninety-five per cent binomial CI for sensitivity, specificity, overall accuracy, PPV and NPV were determined. PPV and NPV values were compared as suggested by Moskowitz and Pepe.10 A two-tailed p value <0.05 was accepted as statistically significant. CI calculations and comparisons of diagnostic values were performed by using ‘acc.paired’, ‘sesp.mcnemar’, ‘pv.rpv’ function of DTComPair package of R V.3.2.2.11 All other analyses were performed via IBM SPSS Statistics V.21.0 (IBM, Armonk, New York, USA).
Results
Out of 250 patients, 165 patients (66.0%) were diagnosed as having PE and 85 patients (34.0%) were without PE. Characteristics and the symptoms of the patients are shown in table 3.
There was no difference between the groups with respect to age (p=0.106). The ratio of female/male was similar in groups. The evaluation of symptoms showed that dyspnea, cough, sputum and fever were more frequent in non-PE group (p<0.05), while chest pain, pleuritic pain, palpitation, syncope and hemoptysis were similar in both groups (p>0.05).
In this study, 40.6% of patients with PE (67 in 165) had DVT which was diagnosed by CUS. AF was seen in 41 (16.4%) of the patients and 26 of them were (15.8%) in the PE group. The incidence of cancer which is one of the Wells criteria found as 9.7%. Note that the incidence of cancer is lower than AF in the PE group (table 3).
In the PE group, patients have no history of AF and anticoagulation therapy. However, 9 out of 15 patients in the non-PE group had history of AF and 8 of them were having anticoagulation therapy. One patient was given acetylsalicylic acid 100 mg since her age was over 90 years.
Most frequently seen risk factors in patients with PE were immobilization (n: 61, 37%), AF (n: 26, 15.8%), major surgery (n: 22, 13.3%), travel (n: 18, 10.9%), malignancy (n: 16, 9.7%), congestive heart failure (n: 11, 6.7%) and obesity (n: 9, 5.5%) as shown in table 4 and figure 1. Pneumonia (44.7%), congestive heart failure (27.1%), acute lower respiratory tract infection (8.2%), acute coronary syndrome (4.7%) and exacerbation of chronic obstructive pulmonary disease (4.7%) were most frequently diagnosed diseases in non-PE groups.
Wells score was higher than 4 in 61.8% of patients with PE and 28.2% of patients without PE. When Wells scores for these patients were evaluated, 49.6% of the patients scored ≤4% and 50.4% of the patients scored >4 (table 5). Out of false negative 63 patients with PE, 21 of them had AF.
The AUC for the Wells score with AF was calculated as 0.714±0.035 (figure 2). The Wells score with AF performed better than the chance-based classification (p<0.001). The optimal cut-off point was determined as ≥3 based on Youden’s index. The sensitivity and specificity for this cut-off point was calculated as 0.897 (95% CI 0.851 to 0.943) and 0.471 (95% CI 0.364 to 0.577), respectively (table 6).
The sensitivity and specificity of Wells score (>4) were 0.618 (95% CI 0.544 to 0.692) and 0.718 (95% CI 0.622 to 0.813), respectively. The sensitivity, overall accuracy and NPV of Wells score with AF were greater than the ones in Wells score (p<0.05). However, specificity of Wells score was greater than the one in Wells score with AF (p<0.001). The PPVs of the two scores were similar (p=0.066).
Discussion
There are many methods to determine the probability of PE diagnosis. Two of them are validated as the Wells score and the revised Geneva score (RGS). They have similar diagnostic performance characteristics and interobserver reliability. Higher numeric value means higher risk of PE.12
The Wells score which is used for finding a clinical probability of PE has been shown to be an important clinical assessment tool for acute PE (table 1). Clinical probability assessment of patients using the Wells criteria has been shown to classify patients’ risk of PE.4 Previously, according to the Wells score PE were trifurcated into low (<2 points), moderate (2–6 points) and high (>6 points) pretest probability with only low ones being considered safe for d-dimer testing. Later, van Belle et al divided investigative group according to the Wells score into two groups, ‘PE unlikely’ (Wells score <5) and ‘PE likely’ (Wells score >4). Significant low probability of PE with the combination of negative-dimer and a PE unlikely Wells score was found.13 Especially in patients with a high or intermediate probability (PE likely), diagnostic studies should be done to clarify PE. It is accepted that in low probability (PE unlikely) patients, when d-dimer test result is negative, mostly no additional studies are needed.14 However, in our study, 38.2% of patients in the Wells unlikely group were diagnosed as PE and five of them had d-dimer test under normal level (500 µg/L). This means these patients were misdiagnosed as non-PE.
PE is originated from DVT of lower extremities approximately in 70% of the patients.1 CUS has a sensitivity of >90% and a specificity of approximately 95% for symptomatic DVT. However, DVT can be found only in 30%–50% of patients with proven PE by CUS.15 PE may originate from causes other than DVT. Venous thrombi of upper extremities or pelvic veins or thrombi of cardiac origin with patients with AF may cause PE.5 6 16 In this study, 40.6% of patients with PE (67 in 165) had DVT diagnosed by CUS. AF was observed in 41 (16.4%) patients of which 26 (15.8%) were in the PE group. Note that the incidence of cancer, which is among the Wells criteria, is lower than that of the AF in the PE group.
AF is the most common chronic arrhythmia in the population encountered in clinical practice and is associated with high risk of morbidity and mortality. The prevalence of AF in the general population is reported to be 2%–3%. The frequency of AF increases with age, the presence of coronary artery disease, structural heart disease and elevated cardiac filling pressures. Stroke and heart failure are increased in patients with AF. AF is diagnosed by ECG.17 Most ischaemic cerebrovascular diseases derived from the heart because of AF as a result of stasis induced thrombi in the left atrium. Abnormal hemostasis, including increased fibrin turn-over and platelet activation is well described in patients with AF.18 Furthermore, AF may induce a hypercoagulability in the right atria. In some patients with PE and AF, echocardiography has shown thrombi in the right atrium. Moreover, autopsy examinations have demonstrated thrombi in the right atrium of patients with AF.9 These findings endorse the idea that PE may originate from intracardiac thrombi in patients with AF. Thus, AF may also cause PE by giving rise to thrombi formation. Additionally, PE may result in AF due to right atrial distention, stress and hypoxia. The relation between AF and PE has not been evaluated in detail, despite the fact that AF is the most common chronic arrhythmia and PE is the most frequently misdiagnosed life-threatening pulmonary vascular disease.5 6
The prevalence of AF ranges from 15% to 24% in patients with acute PE. Sorenson et al found the prevalence of atrial fibrillation during 0–3 months prior to the index date higher in subjects with venous thromboembolism (VTE) than in controls.19 Kukla et al evaluated the prevalence of right side of the heart thrombi (RHT) and AF while assessing their impact on outcomes in patients with acute PE. AF was detected in 24% of the patients. RHT was observed in 5% of patients. The combination of AF and RHT was observed in 2% of the patients.6 CHA2DS2-VASc score (Congestive heart failure, Hypertension, Age≥75 [doubled], Diabetes, Stroke [doubled], Vascular disease, Age 65–74, and Sex category [female]) have been used for stroke prediction in patients with AF and serves to determine the need for anticoagulants. Most components of the CHA2DS2-VASc score also are potential risk factors for VTE. CHA2DS2-VASc and CHADS2 score have similar and modest predictive value for PE in patients with AF. Anticoagulant usage is associated with a decreased risk of PE in patients with AF.7 20
Enga et al investigated an association between AF and future risk of VTE in a population-based cohort. A total of 29 975 subjects were followed from 1994 up to 2010. VTE and AF during follow-up were recorded. One thousand six hundred four subjects were diagnosed with AF and 614 with incident VTE. The risk of VTE was increased during the first 6 months after diagnosis of AF (HR 8.44; 95% CI 5.61 to 12.69), and remained increased throughout the study period (HR 1.43; 95% CI 1.43 to 1.99) compared with those without AF. AF demonstrated higher risk estimates for PE (HR 11.84; 95% CI 6.80 to 20.63) than for DVT (HR 6.20; 95% CI 3.37 to 11.39) during the first 6 months.21 Wang et al evaluated whether AF was associated with an increased risk of PE; 11 458 patients newly diagnosed with AF and the comparison group comprising 45 637 patients without AF were compared. Both cohorts were followed up to evaluate the incidence of DVT and PE. The study showed that, after adjusting for age, sex and comorbidities, the risks of DVT and PE remained significantly elevated in the AF group compared with the non-AF group.22
Ögren et al carried out an autopsy study representing 84% of all in-hospital deaths during a 13-year period and intracardiac thrombus was found in 7% of all cases.23 Right-sided thrombosis was as common as left-sided thrombosis. Of patients dying from or with PE, 7% had right intracardiac thrombus as a potential cause, which was the only evident source for embolism in 4% of all cases.24 Their findings showed that the diagnosis of PE should be considered in patients with AF. Since 1969, AF and PE relationship is a known but an obscured fact.9 With all the results shown in these studies together with our clinical implications guided us into researching new PE prediction criteria. Since Wells criteria had been useful with some limitations, addition of AF to the Wells criteria was investigated. The possibility to detect left atrium thrombosis with echocardiography in the patients with AF and the outcome of large-scale intervention trials have led to an increased awareness of the associated risk of embolic stroke25–27 and guidelines about embolic stroke were designed and applied.28 29 Since the heart is symmetrical, it is expected that right cardiac thrombosis in the case of AF might have a similar role for the development of PE. Therefore, guidelines related to PE prevention should also take into account AF. Although 2014 ESC Guideline on the diagnosis and management of acute PE mentions AF as a risk factor for PE, it is still an obscured cause of PE.1 Our study results showed that Wells sensitivity, overall accuracy and NPV (in terms of predicting patients with PE) are increased when AF is added to the criteria (p<0.05). These results suggest that AF is an important cause of PE, just like malignancy and DVT. Although this has been observed in PE case series, due to the absence of population-based studies, the true prevalence is still not well known.
In conclusion, the relation between AF and PE is an important factor in diagnosis and therapy. The results of our study showed that WeIls scoring failed to estimate 46 patients since it does not consider AF as a cause of PE. We strongly recommend using the updated Wells score with AF since it includes AF as a scoring question which can facilitate diagnosing more patients with PE.
Footnotes
Contributors Planning: Canan Hasanoglu, Conduct: Canan Hasanoglu, Emine Argüder, Hatice Kılıç, Ebru Sengul Parlak, Ayşegül Karalezli, Reporting: Canan Hasanoglu, Emine Argüder, Revision: Canan Hasanoglu, Michael J McPhaul.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Ethics approval The regional committee of medical and health research ethics approved the study (no.: 2011-12-116).
Provenance and peer review Not commissioned; externally peer reviewed.
Patient consent for publication Obtained.