When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found
BMJ 2013; 347 doi: https://doi.org/10.1136/bmj.f3368 (Published 02 July 2013) Cite this as: BMJ 2013;347:f3368All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Wiener and colleagues convince that there is overdiagnosis of pulmonary embolism (PE) because of detection of sub-segmental emboli, aided by the technological advances in multidetector CT. They offer a multitude of plausible reasons why this may be so, including direct to consumer advertising.
They then disclaim: "we need to learn which small emboli need treatment" and "definitive evidence of overdiagnosis would, of course, be the finding that untreated patients never experienced harm from the pulmonary embolism during the rest of their lives".
This, I submit, is the crux of the problem. Physicians are unwilling to pronounce that treatment is not warranted, preferring instead to defer to the oft-used "more evidence is needed"; a truism that has an enviable lifespan.
If I were to deliberately ignore a sub-segmental PE, or my colleague refuse to treat it, to avoid overdiagnosis, and the patient later died from a fatal PE, I doubt we could present this article in our defense.
Until it is said in no uncertain terms that subsegmental PE should not be treated, overdiagnosis of PE will continue unabated, and the chief effect of articles such as this will be a reminder that medicine is not an exact science.
Competing interests: No competing interests
We appreciate all the interest in our analysis.
Dr. Quantrill's concern underscores the tension between doing too much versus too little. Based on substantial evidence, however, selectively imaging based on diagnostic algorithms is not only safe (1,2) but preferable because it reduces overtesting and overtreatment. Even in the study by Anderson et al, which only enrolled patients with a moderate to high pre-test probability of PE (by using a diagnostic algorithm to exclude low risk patients), there was no difference between CTPA and VQ scan in the primary outcome of VTE at 90 days or the secondary outcome of fatal PE or unexplained sudden death (the 0.6% difference in VTE rates cited by Dr. Quantrill was deemed both clinically and statistically insignificant). We thank Dr. Benger for pointing out the PERC criteria, which like the Wells and Geneva scores, provide another evidence-based approach for clinicians to minimize imaging of low risk patients.
We agree with Ripley and colleagues that excessive radiation exposure is an important harm of overuse of CT, and thank Dr. Benger and Dr. Roach for raising alternative first-line tests for VTE. While three-dimensional SPECT-VQ appears to have advantages over the planar scans used in the U.S., it will be important to confirm that this new technology results in improved outcomes and does not compound the problem of overdiagnosis of clinically insignificant PE because of its extremely high sensitivity.
Both Dr. Rogers and Haldar and colleagues highlight the role of the radiologist in PE overdiagnosis. Implementing systems of multiple readings may help reduce "over-calling" artifacts. Feeding back re-readings to radiologists may help calibrate them and improve inter-rater reliability. Regardless, radiologist reports like those described by Haldar and colleagues that directly help clinicians understand the meaning of ambiguous findings such as isolated subsegmental PEs, their clinical importance, and possible treatment actions (or inaction) are a welcome step forward.
1. van Belle A, Buller HR, Huisman MV, Huisman PM, Kaasjager K, Kamphuisen PW, et al. Effectiveness of managing suspected pulmonary embolism using an algorithm combining clinical probability, D-dimer testing, and computed tomography. JAMA 2006; Jan 11;295(2):172-9.
2. Perrier A, Roy PM, Sanchez O, Le Gal G, Meyer G, Gourdier AL, et al. Multidetector-row computed tomography in suspected pulmonary embolism. N Engl J Med 2005; Apr 28;352(17):1760-8.
Competing interests: No competing interests
While some small pulmonary emboli might not need treatment, the presence of emboli is not a physiological event and certainly should not be ignored. Presence of emboli is evidence of underlying pathology and should lead to screening for underlying malignancy and correction of provocative factors such as hormone replacement therapy.
Competing interests: No competing interests
We read with interest the recent paper by Renda Soylemez Wiener et al in the Journal (1). We share with these authors bewilderment at the excessive use of CT angiography (CTA) in the emergency department to rule out pulmonary embolism (PE). Many of these CTs are negative, patients are unnecessary exposed to radiation, and costly procedures and precious time are wasted.
A contributory factor to the exaggerated performance of this procedure is related to the liberal use of D Dimers test, an established method to rule out venous thromboembolism. As raised D dimer values are frequently observed, the next step by the attending physician is to perform CTA (Am J Med).
In a recent preliminary evaluation of this problem at our hospital, most of the CTA performed in the emergency department to rule out PE were negative. A small proportion showed minimal findings compatible with small infarctions and these patients were treated accordingly with anticoagulant therapy. Our findings further confirms the magnitude of the problem (1,2).
As Rubin ED remarked more than 40 years ago (Rubin ref), pulmonary embolism is overdiagnosed and overtreated (3). The current paper by Renda Soylemez Wiener, et al, shows us that not only is PE overdiagnosed but many diagnosed cases are subclinical and may not need therapy. This is a relative new issue that needs to be considered when confronting a patient with PE and minimal positive findings on imaging. Should we postpone therapy? Should we use less sensitive imaging modalities and come back to VQ scans? These are relevant questions considering the high morbidity and mortality secondary to anticoagulants administration. CTA is an important imaging tool and is effective for diagnosis of PE, but overdiagnosis is an undesirable effect. Regarding CTA for PE should we ask ourselves, once more if "..the emperor has no clothes.."?.
References
1. Wiener RS, Schwartz LM, Woloshin S. When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found. BMJ 2013;347 (f3368): 1-7.
2. Adams DM, Stevens SM, Woller SC, Evans RS, Lloyd JF, Snow GL, Adherence to PIOPED II Investigators. Recommendations for Computed Tomography Pulmonary Angiography. Am J Med. 2013:126:36-42.
3. Robin ED. Overdiagnosis and overtreatment of pulmonary embolism: the emperor may have no clothes. Ann Intern Med. 1977;87(6):775-81.
Competing interests: No competing interests
Weiner and colleagues(1) have opened this journal’s ‘too much medicine’ series with an excellent article that provokes thought and challenges current practice.
They have highlighted the uncertainty that exists with subsegmental pulmonary embolism that includes its diagnosis and management. They describe the harm that may come to patients from both the radiation exposure of CT pulmonary angiography as an investigation, and from anti-coagulation as treatment. The evidence they present, despite its limitations, supports the notion that this condition is overdiagnosed and overtreated.
However, we suggest that the advice on how to address this problem, would be difficult to follow for the majority of in-patients, and in particular, surgical patients.
In the UK, both the British Thoracic Society guidelines(2), and their successor, published by the National Institute for Clinical Excellence(3), recommend CT pulmonary angiogram (CTPA) as first line investigation for those rated as ‘likely’ on the Wells criteria. Those rated ‘unlikely’ are to be investigated with a D-dimer test, which, if positive, would then require CTPA.
According to NICE the VQ scan is to be offered to those with renal impairment, allergy to contrast media, or to those to whom the risk of irradiation is unacceptable.
Given these guidelines, it would seem difficult to reduce the number of medical and surgical inpatients suspected of having a pulmonary embolism, from receiving CTPA as investigation. Surgical patients will either be rated as ‘likely’, or will provide a D-dimer test that is considered difficult or impossible to interpret(4, 5)In addition, the reliability of clinical prediction rules for pulmonary embolism may decrease when applied to inpatients and trauma patients(6, 7). The proposed reduction in CTPA use therefore, would depend on the proportion of investigations that are provided on an outpatient or inpatient basis.
The alternative to treating all clinically stable patients with subsegmental pulmonary emboli with anti-coagulants is described as continued monitoring with serial ultrasonography. This seems a reasonable suggestion, however it would place an extra logistical burden on radiology departments, and require further compliance and attendance from patients.
Evidence guiding which pulmonary emboli can safely be left untreated would have significant value, both to the patient and the healthcare institution. The authors state those questions we need to answer with regard to the prognosis and outcomes of untreated subsegmental pulmonary emboli. We would also be interested to know if these outcomes differ for different patient groups, such as surgical, medical, inpatient or outpatient.
Whilst we congratulate the authors on highlighting this issue, we feel there is a great amount of work still to be done before a significant change in practice is seen. We look forward to seeing the results of the prospective cohort study they reference.
1. Wiener RS, Schwartz LM, Woloshin S. When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found. BMJ. 2013;347:f3368.
2. British Thoracic Society Standards of Care Committee Pulmonary Embolism Guideline Development Group. British Thoracic Society guidelines for the management of suspected acute pulmonary embolism. Thorax. 2003;58(6):470-83.
3. Langford N, Stansby G, Avital L. The management of venous thromboembolic diseases and the role of thrombophilia testing: summary of NICE Guideline CG144. Acute medicine. 2012;11(3):138-42.
4. Dindo D, Breitenstein S, Hahnloser D, Seifert B, Yakarisik S, Asmis LM, et al. Kinetics of D-dimer after general surgery. Blood coagulation & fibrinolysis : an international journal in haemostasis and thrombosis. 2009;20(5):347-52.
5. Lippi G, Veraldi GF, Fraccaroli M, Manzato F, Cordiano C, Guidi G. Variation of plasma D-dimer following surgery: implications for prediction of postoperative venous thromboembolism. Clinical and experimental medicine. 2001;1(3):161-4.
6. Ollenberger GP, Worsley DF. Effect of patient location on the performance of clinical models to predict pulmonary embolism. Thrombosis research. 2006;118(6):685-90.
7. Young MD, Daniels AH, Evangelista PT, Reinert SE, Ritterman S, Christino MA, et al. Predicting pulmonary embolus in orthopedic trauma patients using the Wells score. Orthopedics. 2013;36(5):e642-7.
Competing interests: No competing interests
We read with interest the excellent and in-depth article by Weiner et al and wholeheartedly agree with the sentiments expressed regarding overdiagnosis of probably clinically insignificant pulmonary artery filling defects as well as overexposure to ionising radiation. However, the authors state that the average effective radiation dose is 10-15 millisieverts(mSv), which we believe is far too high for CT Pulmonary Angiogram protocols (CTPA). The effective radiation dose for CTPA protocol is generally between 2.2 and 7.5mSv [1].
In our institution, the average effective dose at present is 4.52 mSv, reduced from 5.4 mSv using primarily a reduced kVp protocol (from 120 to 80kVp).
1. Hunsaker AR, Lu MT, Goldhaber SZ, Rybicki FJ.Imaging in acute pulmonary embolism with special clinical scenarios.Circ Cardiovasc Imaging. 2010 Jul;3(4):491-500. doi: 10.1161/CIRCIMAGING.109.855981
Competing interests: No competing interests
Editor,
We welcome the recent article by Wiener & colleagues discussing overdiagnosis of pulmonary embolism (PE) and agree that there is evidence in the literature to suggest that widespread use of CT Pulmonary Angiogram (CTPA) is in part responsible for this(1).
We recently performed a study in our Acute Trust looking at our CTPA practice. We reviewed the scans of all patients having CTPA over a 12 month period (1st March 2011 – 28th February 2012). In particular, we reviewed all patients reported as having isolated subsegmental PE (ISSPE). All originally reported as ISSPE scans were reviewed by two Chest Radiologists, who rejected ISSPE if it was not visible on multi-planar reformats (i.e. was thought to be due to partial voluming). Finally, we reviewed the notes of patients originally reported as ISSPE to see if they had been anticoagulated.
We performed 1317/1435 (91.7%) technically adequate studies during this period. 6.9% were borderline (able to exclude large central clot) and 1.4% were non-diagnostic. Of the technically adequate studies, 21% had a positive study (not including ISSPE which we deemed to be a negative study). In addition, 44/1317 (3.3%) were reported as having ISSPE. Following review by two chest radiologists, only 16/44 (36%) were confirmed as ISSPE. 14/44 (32%) were thought to be negative. 14/44 (32%) were thought to have either multiple or higher order (e.g. lobar or segmental) emboli.
We reviewed the notes on 29/44 (66%) patients originally reported with ISSPE. 21/29 (72%) were anticoagulated; 9/29 (31%) were treated with tinzaparin and 12/29 (41%) were warfaranised. 8/29 (28%) patients were untreated with no outcome of VTE or PE.
We agree there is evidence that;
1) ISSPE is common in asymptomatic individuals
2) Patients with ISSPE who are not anticoagulated are not necessarily at risk of further Venous Thromboembolism (VTE)
In addition to the papers referenced by Wiener, Gandhi et al (2) prospectively performed CTPA on asymptomatic patients on the first post-operative day for primary hip (total hip arthroplasties [THA]) and total knee arthroplasties (TKA)2. 21 TKA and 27 THA patients were scanned. 11/21 (41%) TKA patients and 1/27 (5%) THA patients were found to have filling defects on their scan (not only restricted to ISSPE). All asymptomatic patients diagnosed with filling defects were discharged without anticoagulation and none had an additional diagnosis of PE or deep vein thrombosis at 3 month follow up.
Eyer at al (2005) in their study aimed to show the clinican response and patient outcome associated with the radiologist’s report of ISSPE or inconclusive interpretation with CTPA(3). 192 patients were followed up. Patients with ISSPE more commonly received anticoagulation: 42 (43%) of 67 patients with ISSPE and 17 (14%) of 125 patients with inconclusive results. In the patients who did not receive anticoagulation, no recurrent PE was found on follow up.
In conclusion, we strongly believe that ISSPE on CTPA should usually be considered a negative result. We believe the finding is inconsistently reported and is present in asymptomatic populations. There is also evidence that if patients with ISSPE are not anticoagulated, they come to no significant harm (assuming there is no further clot in the legs).
We therefore add the following paragraph to CTPA reports which show ISSPE:
‘’There is no convincing evidence of pulmonary thromboembolism.The scan shows an ISSPE, which is of dubious clinical significance. A bilateral leg ultrasound should be considered. The decision to anti-coagulate should be based on the findings of this in conjunction with pre-test probability assessment, cardiopulmonary reserve assessment, and balanced against individual risk assessment of anticoagulation.”
This is supported by a consultant led VTE service which reviews all patients with a diagnosis of pulmonary embolism within 4 weeks of diagnosis.
Sananda Haldar, Specialty Registrar in Clinical Radiology
Akshay Garg, Foundation Year 1 doctor
Steve Barden, Consultant Acute Physician
Guy Burkill, Consultant Radiologist
Nigel Marchbank, Consultant Radiologist
1. Wiener RS, Schwartz LM, Woloshin S. When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found. BMJ 2013;347:f3368
2. Gandhi R, Salonen D, Geerts W.H, Khanna M, McSweeney S, Mahomed N.N.A Pilot Study of Computed Tomography-Detected Asymptomatic Pulmonary Filling Defects After Hip and Knee Arthroplasties Journal of Arthroplasty 2012; 27 (5) ,730-735
3. Eyer BA, Goodman LR, Washington L. Clinicians' response to radiologists' reports of isolated subsegmental pulmonary embolism or inconclusive interpretation of pulmonary embolism using MDCT. American Journal of Roentgenology 2005; 184:623-628
Competing interests: No competing interests
Concern must be raised in relation to the message sent out in stating that CTPA finds pulmonary emboli "that do not need to be found." (1) Great caution needs to be exercised if deciding not to treat confirmed PE on CTPA.
Anderson's study, (2) cited as providing the best evidence for overdiagnosis, was mainly out patient-based, very different from the UK experience of much higher risk in patients. This study actually showed a nearly 3 times increased rate of VTE (6/611 vs 2/561) in those randomised to V/Q scanning rather than CTPA, including one fatality.
In the authors' own work (3) mortality decreased by 3% after the introduction of CTPA scanning. Concerns regarding the adverse effects of anticoagulation appear to have been markedly exaggerated, increasing from around 3/100,000 prior to the introduction of CTPA scanning to about 5/100,000 afterwards, still remarkably low.
If CTPA is not performed, then PE will be missed and the opportunity to make other important diagnoses aswell, such as pneumonia or small lung cancers.
Failure to treat confirmed PE may result in recurrent PE, death, chronic pulmonary hypertension or post-thrombotic syndrome, and it is not currently possible to identify who will develop these complications. Although many small emboli may be of no clinical significance, given the low incidence of treatment side effects, the risk/benefit ratio appears to rest favourably with anticoagulation in most cases.
1 Wiener RS, Schwartz LM, Woloshin S. When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found. BMJ 2013;347:f3368.
2 Anderson DR, Khan SR, Rodger MA, Kovacs MJ, Morris T, Hirsch A, et al. Computer tomographic pulmonary angiography vs ventilation-perfusion lung scanning in patients with suspected pulmonary embolism: a randomized controlled trial. JAMA 2007;298:2743-53.
3 Wiener RS, Schwartz LM, Woloshin S. Time trends in pulmonary embolism in the United States:evidence of overdiagnosis. Arch Intern Med 2011;171:831-7.
Competing interests: No competing interests
While Wiener et al raise important issues regarding the overuse of CT pulmonary angiography, the relevance of incidental findings and whether subsegmental emboli should be treated, it is important to correct the assertion that V/Q scanning is inferior to CTPA.
The authors quote results from planar V/Q scanning, a 2-dimensional technique still commonly performed in the United States due largely to the lack of a suitable ventilation agent being registered in that country. In many other parts of the world (including Europe, Canada and Australia), V/Q scanning is now routinely performed with 3-dimensional imaging using SPECT (Single Photon Emission Computed Tomography)[1]. Compared with planar V/Q scanning, SPECT is more sensitive[2, 3] and specific[3, 4], has better intraobserver reproducibility[1, 4] and a much lower inconclusive rate (typically <5%)[3, 5]. In studies directly comparing V/Q SPECT with CTPA, the sensitivity of SPECT has exceeded 95%, much higher than that reported for CTPA (68%-86%)[2, 6]. Compared with CTPA, V/Q SPECT has the added advantages of a lower radiation dose (particularly relevant in young women), less technically suboptimal studies and no contrast-related complications[7]. The statement by Wiener et al that CTPA is “much more sensitive” than V/ Q scanning is erroneous when compared with modern 3-D V/Q SPECT imaging.
Furthermore, the advent of hybrid SPECT/CT scanners can further improve accuracy by allowing V/Q SPECT (with its high sensitivity) to be combined with a low dose non-contrast CT scan which can identify structural pathologies such as pneumonia, abscess, pleural or pericardial effusions, malignancy and pulmonary infarction[8]. While the authors consider that “V/Q scans may make more sense” for investigation of PE in some patients, there are many compelling reasons (including higher sensitivity, lower radiation dose and absence of contrast related complications) why V/Q SPECT is superior to CTPA and why it should be the primary imaging test for most patients with suspected PE in clinical practice today.
1. Roach, P.J., D.L. Bailey, and B.E. Harris, Enhancing lung scintigraphy with single-photon emission computed tomography. Semin Nucl Med, 2008. 38(6): p. 441-9.
2. Reinartz, P., et al., Tomographic imaging in the diagnosis of pulmonary embolism: a comparison between V/Q lung scintigraphy in SPECT technique and multislice spiral CT. J Nucl Med, 2004. 45(9): p. 1501-8.
3. Bajc, M., et al., Diagnostic evaluation of planar and tomographic ventilation/perfusion lung images in patients with suspected pulmonary emboli. Clin Physiol Funct Imaging, 2004. 24(5): p. 249-56.
4. Collart, J.P., et al., Is a lung perfusion scan obtained by using single photon emission computed tomography able to improve the radionuclide diagnosis of pulmonary embolism? Nucl Med Commun, 2002. 23(11): p. 1107-13.
5. Leblanc, M., F. Leveillee, and E. Turcotte, Prospective evaluation of the negative predictive value of V/Q SPECT using 99mTc-Technegas. Nucl Med Commun, 2007. 28(8): p. 667-72.
6. Gutte, H., et al., Detection of pulmonary embolism with combined ventilation-perfusion SPECT and low-dose CT: head-to-head comparison with multidetector CT angiography. J Nucl Med, 2009. 50(12): p. 1987-92.
7. Roach, P.J. and M. Bajc, Acute pulmonary embolism. N Engl J Med, 2010. 363(20): p. 1972-3; author reply 1974-5.
8. Roach, P.J., et al., SPECT/CT in V/Q scanning. Semin Nucl Med, 2010. 40(6): p. 455-66.
Competing interests: No competing interests
Re: When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found
In their recent comprehensive analysis article Wiener and colleagues [1] make a compelling argument; the increased accessibility of CT angiography may have led to overdiagnosis and overtreatment of pulmonary embolism (PE) with resulting potential harms to patients as well as burdens to health systems.
We would agree with the views of the authors of this analysis regarding available evidence or lack thereof guiding practice for the management of patients with subsegmental PEs (SSPE). We would like though to comment that amongst these, cancer patients may be a distinct group which requires additional considerations.
Cancer patients are both more likely to develop thrombotic complications, as well as more likely to be diagnosed incidentally with PE since they are subjected routinely to repeat imaging often with multidetector row CT[2]. We perceive a cancer-related prothrombotic state due to factors relating to both the tumour as well as its treatment. Moreover, there is evidence that several groups of cancer patients may actually benefit from prophylactic anticoagulation even in the absence of relevant findings or history[3] .
On the other hand, cancer patients may also have an increased risk for haemorrhagic complications due to metastases, coexisting haematological toxicities from chemotherapy etc.
In our database of cancer patients with unsuspected PE identified prospectively and treated uniformly under our Centre’s dedicated service[4] one third (33%) had SSPE whilst we have also observed that even though for some patients the presence of a small PE may remain inconsequential for others it may be a progressive process[5].
To gain an understanding of current UK practice, we have recently conducted a cross sectional survey using hypothetical scenarios of cancer patients on treatment with chemotherapy and an unsuspected SSPE in staging imaging. The survey was conducted amongst 154 oncologists, chest and palliative care physicians between 9/9/2012 and 19/5/2013.
In an adjuvant setting (post-operative chemotherapy, no-active cancer), 75%, would immediately initiate anticoagulation [95% Confidence interval (CI) (67.5%,82.5%)] for a single–site SSPE, a percentage which would rise to 94 % for multiple-site SSPE [95% CI (89.6%,97.4%)]. For a patient with metastatic cancer the respective percentages were 84% [95%CI(78%,90.1%)] and 95% [95%CI(91.5%,98.6)].
For a patient with metastatic cancer and a single-site unsuspected SSPE , anticoagulation for life would be the recommendation of 26% of the surveyed physicians [95%CI(19.9%-34%)], a number that rise to 38%, 95%CI(29.8-46.1%) for multiple-site SSPE.
We also observed differences in the attitudes towards treatment between the different specialities. In the adjuvant setting, oncologists would be more likely to immediately anticoagulate for a single SSPE than palliative care physicians or chest physicians (84% vs 46% vs 56%, respectively, p=0.001). In the metastatic setting the differences were smaller (89% vs 69% vs 76%,respectively, p=0.057) but palliative care physicians remained less likely to immediately anticoagulate even in the case of multiple-site SSPE (85% vs 96%, p=0.014).
Despite the unknown clinical significance of SSPE[6], and the likelihood that even in cancer patients some of these SSPEs may have trivial effects on prognosis if left untreated, the majority of the physicians surveyed would opt for anticoagulation in patients with unsuspected SSPE regardless of its extent. In view of the increased thrombotic risk in cancer patients and evidence suggesting that unsuspected PE in cancer patients has similar outcomes as symptomatic PE[7], our opinion is that this practice should still remain the standard of care until better predictive factors are identified, optimally with an appropriately designed RCT.
References:
1. Wiener RS, Schwartz LM, Woloshin S. When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found. Bmj 2013;347:f3368 doi: 10.1136/bmj.f3368
2. Le Gal G, Righini M, Parent F, et al. Diagnosis and management of subsegmental pulmonary embolism. Journal of thrombosis and haemostasis : JTH 2006;4(4):724-31 doi: 10.1111/j.1538-7836.2006.01819.x|.
3. Agnelli G, George DJ, Kakkar AK, et al. Semuloparin for thromboprophylaxis in patients receiving chemotherapy for cancer. The New England journal of medicine 2012;366(7):601-9 doi: 10.1056/NEJMoa1108898|.
4. Palmer J, Bozas G, Stephens A, et al. Developing a complex intervention for the outpatient management of incidentally diagnosed pulmonary embolism in cancer patients. BMC health services research 2013;13(1):235 doi: 10.1186/1472-6963-13-235|.
5. Bozas G, Bradley R, Avery G, et al. Pre-existing pulmonary thrombi in cancer patientsdiagnosed with an unsuspected pulmonary embolism. Journal of Thrombosis and Haemostasis. 2013;11(issue Supplement s2):PB 3.60-3
6. Carrier M, Righini M, Wells PS, et al. Subsegmental pulmonary embolism diagnosed by computed tomography: incidence and clinical implications. A systematic review and meta-analysis of the management outcome studies. Journal of thrombosis and haemostasis : JTH 2010;8(8):1716-22 doi: 10.1111/j.1538-7836.2010.03938.x|.
7. den Exter PL, Hooijer J, Dekkers OM, et al. Risk of recurrent venous thromboembolism and mortality in patients with cancer incidentally diagnosed with pulmonary embolism: a comparison with symptomatic patients. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 2011;29(17):2405-9 doi: 10.1200/JCO.2010.34.0984|.
Competing interests: No competing interests