Overdiagnosis of Pulmonary Embolism by Pulmonary CT Angiography

Transcription

1 Cardiopulmonary Imaging Original Research Hutchinson et al. Overdiagnosis of Pulmonary Embolism by CTA Cardiopulmonary Imaging Original Research Barry Donald Hutchinson 1 Patrick Navin 1 Edith M. Marom 2 Mylene T. Truong 2 John F. Bruzzi 1 Hutchinson BD, Navin P, Marom EM, Truong MT, Bruzzi JF Keywords: artifact, false-positive, misdiagnosis, pulmonary CT angiography, pulmonary embolism DOI: /AJR Received October 10, 2014; accepted after revision January 23, Department of Radiology, University Hospital Galway, Newcastle Rd, Galway, Ireland. Address correspondence to B. D. Hutchinson (barryhutchinson82@gmail.com). 2 Department of Chest Radiology, M. D. Anderson Cancer Center, Houston, TX. AJR 2015; 205: X/15/ American Roentgen Ray Society Overdiagnosis of Pulmonary Embolism by Pulmonary CT Angiography OBJECTIVE. The purpose of this study is to evaluate the rate of overdiagnosis of pulmonary embolism (PE) by pulmonary CT angiography (CTA) in a tertiary-care university hospital. MATERIALS AND METHODS. This study is a retrospective review of all pulmonary CTA examinations performed in a tertiary-care university hospital over a 12-month period. Studies originally reported as positive for PE were retrospectively reinterpreted by three subspecialty chest radiologists with more than 10 years experience. A pulmonary CTA was considered negative for PE when all three chest radiologists were in agreement that the pulmonary CTA study was negative for PE. The location and potential causes for PE overdiagnosis were recorded. RESULTS. A total of 937 pulmonary CTA studies were performed over the study period. PE was diagnosed in the initial report in 174 of these cases (18.6%). There was discordance between the chest radiologists and the original radiologist in 45 of 174 (25.9%) cases. Discordance occurred more often where the original reported PE was solitary (46.2% of reported solitary PEs were considered negative on retrospective review) and located in a segmental or subsegmental pulmonary artery (26.8% of segmental and 59.4% of subsegmental PE diagnoses were considered negative on retrospective review). The most common cause of diagnostic difficulty was breathing motion artifact, followed by beam-hardening artifact. CONCLUSION. In routine clinical practice, PEs diagnosed by pulmonary CTA are frequently overdiagnosed, when compared with the consensus opinion of a panel of expert chest radiologists. Improvements in the quality of pulmonary CTA examination and increased familiarity with potential diagnostic pitfalls in pulmonary CTA are recommended to minimize misdiagnosis of PE. P ulmonary embolism (PE) is a common clinical diagnosis in patients presenting to the emergency department and in hospitalized patients. It is normally treated with long-term anticoagulation therapy to reduce the risks of death and the morbidity associated with chronic pulmonary venous thromboembolism. Because the clinical presentation is often nonspecific and can be mimicked by a range of other conditions, in routine practice, pulmonary CT angiography (CTA) is often used as the imaging method of choice for further investigation [1]. Pulmonary CTA has been shown to be highly sensitive and specific when pretest clinical diagnostic tools are used [2] but surprisingly inaccurate in patients with low pretest probability, with false-positive rates as high as 42% [3]. Unfortunately, adherence to referral guidelines for pulmonary CTA has repeatedly been shown to be low [4, 5]. Nonetheless, many clinicians will initiate anticoagulation therapy on the basis of a positive result, regardless of pretest probability [6], even in isolated subsegmental PE [7]. The risk of hemorrhage related to anticoagulation therapy is potentially significant. A large meta-analysis in 2003 [8] found a 7% annual risk of major bleeding and a 0.4% incidence of bleeding-related fatality in patients treated with oral anticoagulation therapy for venous thromboembolism for longer than 3 months. The practical implications of long-term anticoagulation therapy for the patient are also potentially significant, requiring frequent attendance to their medical practitioners for blood tests, consequent time off from work, potential adverse drug interactions with other medications, adjustments to travel and lifestyle, implications for future dental and medical procedures, and possible negative effects on life insurance status. With these considerations in mind, it is important to minimize the misdiagnosis AJR:205, August

2 Hutchinson et al. of PE. Common artifacts that can lead to a false-positive diagnosis of PE have been well described in the published literature [9 11]. Despite this, however, reported interobserver agreement varies widely, especially in the diagnosis of subsegmental PE [12]. Wide variations in concordance between general and subspecialist radiologists have been reported (89 100%) [13, 14], as well as between residents, fellows, and attending radiologists (87 93%) [15 17]. Although pulmonary CTA examinations are frequently interpreted by general radiologists in most centers, limited data exist on the interobserver agreement between general and subspecialist chest radiologists. A small number of studies [13, 18] have directly compared pulmonary CTA interpretation by general radiologists with that of a single subspecialist chest radiologist. There is very limited analysis of these discrepant cases in terms of PE location within the pulmonary arterial system and potential causes of misdiagnosis. In addition, the absence of a practical reference standard examination makes it difficult to draw conclusions regarding the accuracy of pulmonary CTA in routine clinical practice. The purpose of this study was to evaluate the rate of overdiagnosis of PE by pulmonary CTA in a tertiary-care university hospital by assessing the degree of discordance between the original reporting radiologists and an expert panel of subspecialty chest radiologists and to attempt to establish patterns of misdiagnosis to try to understand the causes underlying pulmonary CTA misinterpretation. Materials and Methods This retrospective study was conducted at University College Hospital Galway, which is a specialist oncology center and a university-affiliated tertiary-care medical center in Galway, Ireland. Approximately 130,000 imaging studies are performed annually in the University College Hospital Galway Radiology Department, which is staffed by 15 attending and nine resident radiologists. The Galway University institutional ethical review board approved this retrospective study and waived the requirement for written informed consent. An electronic search was performed of the approved finalized reports of all consecutive pulmonary CTA examinations performed over a 12-month period between August 1, 2012, and July 31, Data were collected by both electronic query and manual review of the electronic medical record. All scans were acquired on a 64-MDCT scanner (Somatom Sensation 64, Siemens Healthcare) Fig year-old man with dyspnea. Axial pulmonary CT angiogram (mediastinal window) shows density measurement in pulmonary trunk and main pulmonary arteries to calculate quality of contrast enhancement. Circles denote ROIs. Avg = average, Dev = deviation. in the craniocaudal direction with a collimation of 0.6 mm and gantry rotation of 500 milliseconds. Automated dose control software was used with 120 kvp and 200 ma maximum; ml of low-osmolar contrast medium (350 Omnipaque, GE Healthcare) was injected through an 18-gauge cannula sited in the antecubital fossa, at a rate of 4 5 ml/s, followed by a 20-mL saline bolus chaser injected at 4 ml/s. Optimal scan acquisition time was determined using a bolus-tracking technique with an ROI placed over the pulmonary trunk. Images were reconstructed with a matrix and a smooth kernel, with 1-mm axial and 1.5-mm coronal slice thickness and 0.8-mm slice overlap. Images were reviewed using IMPAX (version 6.5, AGFA Healthcare). All studies in which a definite diagnosis of PE was reported were selected for further analysis. Studies reported as nondiagnostic or negative for the presence of PE were excluded (because the purpose of our study was to evaluate the potential rate of overdiagnosed PEs, rather than the overall diagnostic accuracy of pulmonary CTA). All studies were anonymized for independent interpretation on stand-alone workstations by a panel of three subspecialist chest radiologists, each with at least 10 years experience in pulmonary CTA interpretation. One radiologist was among the 15 on-site attending radiologists. The other two panel members were reviewers from another tertiary referral center. The final consensus opinion of these three chest radiologists was used as a surrogate reference standard for the diagnosis of PE. Each examiner was blinded to the index report, PE location, clinical history, and other diagnostic test results. An initial interpretation was performed by each of the three chest radiologists independently, in which they recorded the presence or absence of PE, the most proximal level of PE, the lobar location of PE, and the overall quality of the examination (satisfactory or unsatisfactory for diagnosis). After this initial interpretation, a second analysis was then made of those studies in which there was any disagreement among the three chest radiologists (to minimize the risk of overlooking PEs because of interpretation fatigue after reading a large number of pulmonary CTA studies in succession). Where there was any persistent discordance among the three chest radiologists after this second review, the original report was accepted as being correct (i.e., positive for PE). Next, a third and final analysis was performed of those studies for which there was a discrepancy between the consensus opinion of the three chest radiologists and the original report, guided by a partial unblinding of the original report to direct attention to the original reported PE. Where there was unanimous agreement among the three chest radiologists that a pulmonary CTA was negative, a final outcome of negative for PE was recorded. In addition, the following final data were recorded: patient demographics (age and sex); the most proximal PE location according to the modified Boyden classification [19] (pulmonary trunk, main pulmonary artery, lobar pulmonary artery, segmental pulmonary artery, or subsegmental pulmonary artery); number of PEs (solitary vs multiple); quality of contrast enhancement, assessed by calculating the average of the CT number measured in the pulmonary trunk and the right and left main pulmonary arteries with a circular ROI equal to the diameter of the vessel (Fig. 1); and interobserver agreement (modified kappa index) among the three chest radiologists and between the final consensus opinion of the three chest radiologists and the original reporting radiologists. Finally, the individual discordant cases (those that were considered to be negative for PE) were analyzed separately to attempt to establish a potential underlying cause for the misdiagnosis of PE on the original report, such as movement ar- 272 AJR:205, August 2015

3 Overdiagnosis of Pulmonary Embolism by CTA TABLE 1: Comparison of Contrast Enhancement, Patient Age, and Pulmonary Embolism (PE) Location in All and Discordant Pulmonary CT Angiography Examinations Reported as Positive for PE by the Original Radiologist Variable All Cases Discordant Cases Total 174 (100) 45 (25.9) Patient age (y), mean (range) 64 (17 99) 60 (23 91) Single-vessel PE 67 (38.5) 31 (46.2) Multiple-vessel PEs 107 (61.5) 14 (13.1) Mean quality of contrast enhancement (HU) a Pulmonary trunk PE 6 (3.4) 0 (0) Main pulmonary artery PE 28 (16.1) 1 (4.0) Lobar pulmonary artery PE 37 (21.3) 6 (16.2) Segmental pulmonary artery PE 71 (40.8) 19 (26.8) Subsegmental pulmonary artery PE 32 (18.4) 19 (59.4) Note Except where noted otherwise, data are number (%) of patients. a p = TABLE 2: Comparison of Solitary Pulmonary Embolism (PE) in All and Discordant Pulmonary CT Angiography Examinations Location of PE All Cases Discordant Cases Total 67 (38.5) 31 (46.2) Pulmonary trunk PE 1 (0.6) 0 (0.0) Main pulmonary artery PE 5 (2.9) 1 (20.0) Lobar pulmonary artery PE 6 (3.4) 4 (66.7) Segmental pulmonary artery PE 31 (17.8) 10 (32.3) Subsegmental pulmonary artery PE 24 (13.8) 16 (66.7) Note Except where noted otherwise, data are number (%) of patients. tifact from breathing or cardiac pulsation; poor contrast opacification of the pulmonary arteries due to Valsalva maneuver, cardiac insufficiency, or other cause of mixing of opacified and unopacified blood; beam-hardening attenuation artifact caused by adjacent high-density structures such as opacified veins, contrast material pooling in the inferior vena cava or right ventricle, or bony structures; and the presence of airspace disease obscuring the underlying pulmonary vasculature. Descriptive numeric values were used for patient and PE demographics (actual values, percentages, mean [± SD], and ranges). Comparisons between groups were performed using the paired t test and for ordinal categories using the chi-square test. A p value of 0.05 or less was considered statistically significant. All statistics were performed using SPSS (version 16, IBM). Results There were 937 pulmonary CTA examinations performed over the course of the 12-month study period. Of these, 174 studies (18.6%) were reported as positive for PE by the original radiologist (Table 1), and comprised 84 male and 90 female patients with a mean age of 64 years (range, years). PEs were reported as solitary in 67 cases (38.5%) and multiple in 107 cases (61.5%) (Table 2). PEs were more frequently reported in the peripheral segmental and subsegmental arteries (103 cases; Fig. 2 Bar chart depicting number of discordant cases of pulmonary embolism diagnosis per attending radiologist. Each letter anonymously represents each of 15 attending radiologists. Asterisk denotes member of panel of subspecialist chest radiologists. No. of Discordant Cases %) than in the more central and lobar arteries (71 cases; 40.8%) (Table 1). Twenty-four patients (13.8%) had a reported solitary subsegmental PE. The average quality of contrast enhancement was ± 88 HU (range, HU). On final analysis, the panel of three chest radiologists were of the consensus opinion that 45 (25.9%) of the original 174 index cases were negative for the presence of PE. Interobserver agreement between the panel members was almost perfect (weighted κ = 0.835). These discordant cases comprised 25 women (none of whom were pregnant) and 20 men (mean age, 60 years; range, years). Overall image quality was considered to be satisfactory for diagnosis in 170 examinations (98%) and inadequate for diagnosis in four examinations (2%). The average quality of contrast enhancement was ± 60.9 HU, versus a mean of ± 91.8 HU in the group with a concordant diagnosis (p = 0.002). There was discordance between the chest radiologists and the original radiologist in 31 of 67 (46.2%) cases of reported solitary PE, whereas discordance occurred in only 14 of 107 (13.1%) cases where multiple PEs were originally reported. Discordance was highest for cases of reported peripheral PEs (38/103 [36.9%] cases of segmental or subsegmental PEs), with the highest rate for reported solitary subsegmental PEs (16/24 [66.7%] of such cases). Discordance occurred most commonly in the lower lobes, with the most commonly involved vessels being the lateral basal and posterior basal segmental arteries of the left lower lobe; these vessels accounted for 35.5% of all discordant cases diagnosed at the segmental level. Interestingly, no discordant diagnoses were seen in the right middle lobe (where diagnostic difficulty might have been expected because of the more horizon- A B C D E F G H I J* K L M N O Attending Radiologist AJR:205, August

4 Hutchinson et al. No. of Discordant Cases Breathing (42.2%) Cardiac (11.1%) Mixing (11.1%) Valsalva (6.7%) Artifact Simulating PE tal course of the pulmonary arteries and their greater susceptibility to partial volume averaging effects). The distribution of instances of discordance between the chest radiologists and the original reporting radiologists was relatively even, varying from zero to six cases per radiologist (median, three cases), indicating that this was a generalized rather than an individual phenomenon (Fig. 2). Causes for the 45 cases of discordance included the following: 24 (53%) cases were due to motion artifact from breathing (19 [42.2%]) or cardiac pulsation (5 [11.1%]), eight (18%) cases were due to poor contrast opacification from Valsalva maneuver (3 [6.7%]) or contrast material mixing (5 [11.1%]), 10 cases (22.2%) were due to attenuation artifact secondary to beam hardening from adjacent high-density structures, and three cases (6.7%) were due to effects from adjacent airspace disease (Fig. 3). Discussion This study shows an unexpectedly high rate of overdiagnosis of PE by pulmonary CTA in a tertiary-care university hospital, with an overall rate of 25.9% of all positive pulmonary CTA examinations, increasing to as high as 66.7% of cases where a solitary subsegmental PE was originally reported. Discordance was greatest for solitary PEs, PEs located in segmental and subsegmental pulmonary arteries, and in the lower zones of the lungs. The positive predictive value of pulmonary CTA for the diagnosis of PE was only 74.1% in this study. The published overall rate of positive diagnosis of PE on pulmonary CTA varies from study to study (e.g., 15.4% [14], 16.4% [13], and 17.8% [6]) but usually ranges between 14% [20] and 22% [7]. Differing levels of Beam Hardening (22.2%) Airspace Disease (6.7%) Fig. 3 Bar chart showing relative frequencies of artifact thought to have been responsible for misdiagnosis in discordant cases. PE = pulmonary embolism. adherence to referral guidelines, which has been shown to significantly affect positivity rates, may explain some of this variation; for example, 30% of all pulmonary CTA examinations were positive for PE in the multicenter Christopher Study [2], which used strict adherence to a basic pretest risk stratification tool. The rate of index positive cases in our center was 18.6% for the 12-month period studied. After review by the panel of chest radiologists, this was revised downward to 13.8%. In our institution, there is no systematic use of pretest probability scoring (e.g., Well or Revised Geneva scores [21, 22]) and inconsistent use of d-dimer assays. Furthermore, many pulmonary CTA examinations in our institution are ordered by the emergency department before assessment by the admitting medical team. The combination of a lack of pretest probability assessment and either inconclusive or possibly erroneous pulmonary CTA results can cause difficulties in patient diagnosis and management, often leading to repeat imaging and unnecessary anticoagulation therapy. The risks and disadvantages of anticoagulation therapy include hemorrhage (occasionally devastating or fatal) [8], interactions with other medications, inconvenience in terms of attendance for repeated blood tests (which may require time off work), and cost (to both the patient and society) [23]. Furthermore, a diagnosis of PE carries with it implications for life insurance coverage, travel plans, and preparation for other medical or surgical procedures. The diagnosis of PE also places the patient in a higher risk category for future events, which can influence investigations and management if the patient again seeks medical attention for similar symptoms. The significance of a falsepositive pulmonary CTA examination should be considered in this context. Previously published studies have shown differences between chest- and non-chesttrained radiologists in the diagnostic accuracy of pulmonary CTA interpretation. In a 2011 study of 70 isolated subsegmental PEs by Pena et al. [13], a reviewing thoracic radiologist reinterpreted 11% of these examinations as negative. In a separate abstract published by Miller et al. [18], a single thoracic radiologist found a false-positive or probable false-positive rate of 11% at all pulmonary artery levels in 508 cases. Compared with these previous studies, the current study is larger and uses a panel of three subspecialty chest radiologists as a more robust surro- A B Fig year-old woman with pulmonary embolism in left lower lobe pulmonary artery. A, Coronal pulmonary CT angiography image (mediastinal window) shows apparent filling defect (arrow) within left lower lobe pulmonary artery. B, Same image on lung window shows stair-step artifact (arrowhead) related to respiratory motion in right lower lobe artery and right major fissure (arrow) with rapid position movement. 274 AJR:205, August 2015

5 Overdiagnosis of Pulmonary Embolism by CTA A B Fig year-old woman presenting with pleuritic chest pain and deep calf tenderness. A, Axial image from pulmonary CT angiography shows apparent filling defect (arrow) in segmental pulmonary artery in right lower lobe, which is due to streak artifact secondary to beam hardening from high-density contrast material in right ventricle. B, Example of beam attenuation artifact from high-density contrast material in superior vena cava, creating pseudoembolus (arrow) in truncus anterior pulmonary artery. gate reference standard, rather than relying on one single radiologist s opinion. Although this was a single-center study, our department does not differ in any significant way from any other university hospital imaging center, with the same mix of inpatients, outpatients, emergency department patients, and pregnant patients as might be found in any equivalent tertiary referral center, and with a modern radiology department using conventional MDCT technology and a PACS for the performance and interpretation of pulmonary CTA examinations, staffed by a general mix of experienced subspecialty fellowship-trained radiologists. The high rates of discrepant pulmonary CTA interpretations found in this study raise concerns about the diagnostic accuracy of radiologists in the wider community. However, the generalizability of our results should be confirmed with a larger multicenter study. Causes of diagnostic difficulty in the interpretation of pulmonary CTA examinations are well recognized [9 11]. A full description of such interpretative pitfalls is beyond the scope of this discussion, but potential falsepositive findings are known to occur because of partial volume averaging effects secondary to motion (breathing and cardiogenic), poor contrast opacification from mixing of opacified and unopacified blood, beam-hardening attenuation artifact from high-density structures (e.g., contrast agent in the superior vena cava [SVC] and right atrium), and confusion with venous structures and mucus-filled bronchi [9, 10, 24]. In our study, the most common cause identified for the misdiagnosis of PE was motion artifact due to breathing, which accounted for 42.2% of cases. Breathing artifact has previously been shown to be the most common mimic of PE [18] as well as the most common cause of equivocal pulmonary CTA findings in up to 74% of cases [9]. Breathing artifact can most easily be identified on a lung window by the presence of the seagull artifact, the stair-step artifact, and rapid changes in position of vessels on contiguous image slices [10] (Fig. 4). Ways to reduce the level of breathing artifact include administering supplemental oxygen and scanning in the caudocranial direction [24]. The second-most-common confounding artifact was beam-hardening attenuation artifact (Fig. 5) from high-density structures, including pooled contrast agent in the SVC or other adjacent vessels, metallic structures such as pacemakers, or the patient s arms if they cannot be elevated above the chest. The use of a saline chaser helps clear pooled contrast agent from the SVC [10]. Apart from their proximity to a high-density structure, regions of low attenuation related to streak artifact have much higher densities (> 78 HU) than real thrombus and form indistinct borders with contrast agent in the vessels [10]. We also observed that beam-hardening attenuation artifact could often be tracked in a radial pattern from the source of the artifact and could also be identified in other nearby structures. Other artifacts responsible for misinterpretation included cardiac pulsatility (Figs. 6 and 7), which is most often seen in regions of the lung adjacent to the heart, such as the lingula and the paracardiac segments of the lower lobes; reduced mixing of contrast agent with unopacified blood, which can be due to excessive inflow of unopacified blood from the inferior vena cava or other veins, excessive breath-holding resulting in a Valsalva maneuver, or poor cardiac function and poor mixing of contrast agent; and obscuration of the pulmonary arteries by adjacent parenchymal disease. The latter is attributed to increased local vascular resistance, which leads to reduced flow and flow artifacts [10]. Interestingly, in our study, there were no discrepancies due to confusion between PEs and pulmonary veins or mucus-filled bronchi. This A B Fig year-old man with suspected pulmonary embolism. A, Axial image from pulmonary CT angiography at level of left lower lobe shows effects of cardiac pulsation artifact, which results in partial volume averaging effect between high density in left lower pulmonary arteries and low density of adjacent lung (arrows), which was confused with pulmonary emboli. B, On lung window, this artifact can be recognized by blurring of walls of affected arteries (arrows). AJR:205, August

6 Hutchinson et al. Fig year-old man with chest pain, hemoptysis, and elevated d-dimer assay. Axial image from pulmonary CT angiography shows cardiac pulsation artifact causing inhomogeneity in contrast material in left upper lobe pulmonary artery (arrow), which was confused with pulmonary embolus. Cardiac motion also causes movement of walls of ascending aorta and of main pulmonary outflow tract (arrowheads). might suggest that discrepancies in the diagnosis of PE arose not because of unfamiliarity with anatomy or to a lack of attention to detail when reading the scan but rather because of perceptual errors resulting from an underrecognition of the other causes of false-positive examinations, as summarized already. Our study also highlights the difficulty of performing audits of the accuracy of pulmonary CTA interpretation. The original decision by the treating physician to initiate anticoagulation therapy would be taken within the clinical context of the patient s presentation, history of thromboembolic disease, cardiac workup, d-dimer levels, and so forth, and is not based purely on the result of the pulmonary CTA scan. In our study, this clinical information was not included. In this regard, as a specific outcome from our study, it was considered to be neither clinically appropriate nor ethical to revisit the original clinical diagnosis several years later on the basis of the results of an academic study that had not been designed to reexamine all of the clinical information that was originally available. Rather, the purpose of our study was to examine the diagnostic difficulties in the use of pulmonary CTA as a diagnostic study in isolation. In the absence of a practical true reference standard, we opted to rely on the consensus opinion of three experienced chest radiologists, which would be difficult to reproduce on a routine basis because of the time involved in collating and reviewing the necessary examinations. Correlation of pulmonary CTA findings with clinical outcomes (e.g., recurrent thromboembolism or death) is a crude measure of accuracy and would also be difficult to achieve in routine clinical practice. The difficulty in performing regular audits of this very common imaging test highlights the risk of unrecognized diagnostic drift, where an established diagnostic test performs less well over time because of changes in practice and personnel and because of an absence of feedback or correlative reference standard test. Practical measures to reduce the risk of PE misdiagnosis could and should include any of the following: systematic use of pretest probability assessment (which would require buy-in from clinicians and incorporation into imaging protocols); radiology technologists being educated to optimize image quality, focusing on proper patient breathing technique and repeating examinations where appropriate; increased familiarization by radiologists with the range of potential diagnostic pitfalls; encouragement of the use of second opinions by interpreting radiologists, particularly for solitary subsegmental PEs; and regular review of positive pulmonary CTA cases (e.g., at monthly discrepancy or audit meetings). Some of these measures are easier to implement than others, but their importance is underscored by the implications of a false-positive diagnosis of PE. Conclusion When compared with the consensus opinion of a panel of three chest radiologists, we found a high rate of overdiagnosis of PE by pulmonary CTA in a department of mixed specialty radiologists, which appeared to be due to a lack of recognition of the imaging pitfalls that can be encountered in the interpretation of pulmonary CTA examinations. Increased education among radiography technologists, radiologists, and clinicians regarding these pitfalls should be encouraged. References 1. Rémy-Jardin M, Pistolesi M, Goodman LR, et al. Management of suspected acute pulmonary embolism in the era of CT angiography: a statement from the Fleischner Society. Radiology 2007; 245: van Belle A, Büller HR, Huisman MV, et al.; Christopher Study Investigators. Effectiveness of managing suspected pulmonary embolism using an algorithm combining clinical probability, d- dimer testing, and computed tomography. JAMA 2006; 295: Stein PD, Fowler SE, Goodman LR, et al. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med 2006; 354: Adams DM, Stevens SM, Woller SC, et al. Adherence to PIOPED II investigators recommendations for computed tomography pulmonary angiography. Am J Med 2013; 126: Corwin MT, Donohoo JH, Partridge R, Egglin TK, Mayo-Smith WW. Do emergency physicians use serum d-dimer effectively to determine the need for CT when evaluating patients for pulmonary embolism? Review of 5,344 consecutive patients. AJR 2009; 192: Ranji SR, Shojania KG, Trowbridge RL, Auerbach AD. Impact of reliance on CT pulmonary angiography on diagnosis of pulmonary embolism: a Bayesian analysis. J Hosp Med 2006; 1: Eyer BA, Goodman LR, Washington L. Clinicians response to radiologists reports of isolated subsegmental pulmonary embolism or inconclusive interpretation of pulmonary embolism using MDCT. AJR 2005; 184: Linkins LA, Choi PT, Douketis JD. Clinical impact of bleeding in patients taking oral anticoagulant therapy for venous thromboembolism: a meta-analysis. Ann Intern Med 2003; 139: Jones SE, Wittram C. The indeterminate CT pulmonary angiogram: imaging characteristics and patient clinical outcome. Radiology 2005; 237: Wittram C, Maher MM, Yoo AJ, Kalra MK, Shepard JAO, McLoud TC. CT angiography of pulmonary embolism: diagnostic criteria and causes of misdiagnosis. RadioGraphics 2004; 24: Bruzzi JF, Rémy-Jardin M, Kirsch J, et al. Sixteenslice multidetector computed tomography pulmonary angiography: evaluation of cardiogenic motion artifacts and influence of rotation time on image quality. J Comput Assist Tomogr 2005; 29: Ghanima W, Nielssen BE, Holmen LO, Witwit A, Al-Ashtari A, Sandset PM. Multidetector computed tomography (MDCT) in the diagnosis of pulmonary embolism: interobserver agreement among radiologists with varied levels of experience. Acta Radiol 2007; 48: Pena E, Kimpton M, Dennie C, Peterson R, Le Gal G, Carrier M. Difference in interpretation of computed tomography pulmonary angiography diagnosis of subsegmental thrombosis in patients with suspected pulmonary embolism. J Thromb Haemost 2012; 10: Costa AF, Basseri H, Sheikh A, Stiell I, Dennie C. The yield of CT pulmonary angiograms to exclude acute pulmonary embolism. Emerg Radiol 2014; 21: Yavas US, Calisir C, Ozkan IR. The interobserver agreement between residents and experienced radiologists for detecting pulmonary embolism and DVT with using CT pulmonary angiography and indirect 276 AJR:205, August 2015

7 Overdiagnosis of Pulmonary Embolism by CTA CT venography. Korean J Radiol 2008; 9: Shaham D, Heffez R, Bogot NR, Libson E, Brezis quency and causes of false-positive CTPA exams in community hospitals. Chest 2009; 136(4_ creasing the models utility with the SimpliRED d-dimer. Thromb Haemost 2000; 83: M. CT pulmonary angiography for the detection MeetingAbstracts):14S 22. Le Gal G, Righini M, Parent F, van Strijen M, of pulmonary embolism: interobserver agreement 19. Boyden EA. Segmental anatomy of the Couturaud F. Diagnosis and management of sub- between on-call radiology residents and special- lungs. New York, NY: McGraw-Hill, 1955 segmental pulmonary embolism. J Thromb Hae- ists (CTPA interobserver agreement). Clin Imaging 2006; 30: Ginsberg MS, King V, Panicek DM. Comparison of interpretations of CT angiograms in the evaluation of suspected pulmonary embolism by oncall radiology fellows and subsequently by radiology faculty. AJR 2004; 182: Miller WT Jr, Marinari LA, Mahne A. Fre- 20. Donato AA, Khoche S, Santora J, Wagner B. Clinical outcomes in patients with isolated subsegmental pulmonary emboli diagnosed by multidetector CT pulmonary angiography. Thromb Res 2010; 126:e266 e Wells PS, Anderson DR, Rodger M, et al. Derivation of a simple clinical model to categorize patients probability of pulmonary embolism: in- most 2006; 4: Lefebvre P, Laliberté F, Nutescu EA, et al. Allcause and potentially disease-related health care costs associated with venous thromboembolism in commercial, Medicare, and Medicaid beneficiaries. J Manag Care Pharm 2012; 18: Wittram C. How I do it: CT pulmonary angiography. AJR 2007; 188: AJR:205, August