European Journal of Internal Medicine 26 (2015) 18–22
Contents lists available at ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original Article
Knowledge of the diagnostic algorithm for pulmonary embolism in primary care B. Planquette a,b,c,⁎, D. Maurice a, J. Peron d,e,f, G. Mourin a,b, A. Ferre a,b, O. Sanchez a,b,c, G. Meyer a,b,c a
Service de pneumologie et de soins intensifs, Hôpital Européen Georges Pompidou, Assistance Publique Hôpitaux de Paris, Paris, France Université Paris Descartes, Sorbonne Paris Cité, France ISERM U970-PARCC, Paris, France d Hospices Civils de Lyon, Service de Biostatistique, F-69003 Lyon, France e CNRS UMR 5558, Laboratoire de Biométrie et Biologie Evolutive, Equipe Biostatistique-Santé, Villeurbanne, France f Université de Lyon, F-69622, Lyon, France b c
a r t i c l e
i n f o
Article history: Received 22 July 2014 Received in revised form 18 November 2014 Accepted 21 November 2014 Available online 11 December 2014 Keywords: Algorithms Differential diagnosis Disease management Pulmonary embolism General practice Primary care physician
a b s t r a c t Background: Diagnostic algorithms for pulmonary embolism (PE) have been validated in patients attending hospital emergency departments. However, general practitioners (GPs) are often the professionals of first resort for the majority of non-critical cases of PE. Aim: To evaluate the knowledge of the diagnostic algorithm for PE among GPs in France. Design and setting: Questionnaire-based survey of GPs with a private practice. Method: All GPs in the study area were sent a questionnaire including several questions on the diagnosis of PE and two clinical cases scenario with suspected PE. Factors associated with knowledge of the diagnostic algorithm were analysed by univariate and multivariate analyses. Results: Five-hundred and eight questionnaires were distributed and 155 (30.5%) were available for analysis. Only 55% of the GPs did know about clinical scores for the assessment of clinical probability of PE and 42% of the GPs were aware that clinical probability is needed to interpret the result of D-dimer testing. Forty GPs (26%) gave valid responses to both clinical cases, 54 GPs (35%) had one valid case out of the two and 61 (39%) gave invalid responses to both clinical cases. Participation in specific training on PE was significantly associated with valid responses to the two clinical cases in multivariate analysis (p b 0.017). Conclusion: The majority of GPs were unaware of the diagnostic algorithm for PE. Clinical probability was rarely assessed and knowledge about D-dimers was poor. Specific training on PE and greater awareness of clinical probability scores may promote knowledge of PE algorithm diagnosis. © 2014 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
1. Introduction Pulmonary embolism (PE) is a common and potentially fatal disease. Significant improvements have been achieved in the diagnostic approach to PE over the past 20 years and this ended-up with diagnostic algorithms endorsed by international guidelines [1]. These diagnostic algorithms have been developed and validated in outpatients referred to emergency departments, and are thus applicable in general medicine. A low clinical probability combined with a negative D-dimer test excludes the diagnosis of PE in primary care with a sensitivity of 97.3% [2]. About three-quarters of venous thromboembolic events take place in the community, thus, general practitioners (GPs) are often the health professionals of first recourse for clinically stable patients with suspected PE [3]. The opportunity for outpatient treatment is currently under evaluation in patients with low-risk PE [4,5]. This new perspective ⁎ Corresponding author at: Service de Pneumologie et de soins intensifs, Hôpital Européen Georges Pompidou, 20 rue Leblanc, F-75015 Paris, France. Tel.: +33 1 56 09 34 62; fax: +33 1 56 09 32 55. E-mail address: [email protected] (B. Planquette).
of ambulatory treatment may lead to GPs becoming major players in the initial management of PE. Since non-adherence to guidelines in the diagnosis of PE has been associated with a significant increase in the risk of adverse events, the knowledge of diagnostic guidelines by GPs is of great importance for the care of patients with suspected PE [6]. The aim of this study is to evaluate the state of knowledge of the diagnostic algorithm for PE among GPs in the south of Paris and its suburb. 2. Method 2.1. Study population This study was carried out among GPs with a private practice. The study area was defined choosing districts and towns around our hospital where GPs usually refer their patients in case of PE suspicion. The area included the 15th and 16th districts of Paris and two towns in the near suburb of Paris: Vanves and Issy-Les-Moulineaux. The addresses of GPs with a private practice in the study area were obtained via the website http://ameli-direct.ameli.fr. The data from this site were
http://dx.doi.org/10.1016/j.ejim.2014.11.005 0953-6205/© 2014 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22 Table 1 Questionnaire (general practitioners characteristics). 1) You are a. Female b. Male 2) How many years have you been in practice? a. Between 1 and 10 b. Between 11 and 20 c. More than 20 3) Do you have any hospital activity in addition to your private practice? (Emergency, outpatients clinics at hospital …) a. Yes b. No 4) Do you have continuing medical education (CME) activities? (Several answers possible) a. Subscription to French medical journals b. Subscription to international journals c. Affiliation to associations of doctors for CME 5) Have you already taken part in training concerning PE? a. Yes b. No 6) Have you already suspected a PE since setting up your practice? a. Yes b. No
crossed referenced with the directory of GPs available on the Conseil National de l'Ordre des Médecins (CNOM) website (http://www. conseil-national.medecin.fr). A questionnaire and two simulated clinical case scenarios were sent to the GPs by post (see supplementary appendix). A stamped-addressed return envelope was included with these documents. 2.2. Questionnaire The survey consisted of 11 questions (Tables 1, 3 and 4) covering different subjects: demographic characteristics and professional situation, continuing medical education sessions on PE (CME) knowledge about risk factors, clinical symptoms of PE and clinical probability scores, and use of D-dimer testing in case of suspected PE. 2.3. Simulated clinical cases scenario Two clinical cases describing patients with suspected PE were distributed to the GPs (Table 2). The clinical probability of PE was intermediate for case 1 and high for case 2. Each clinical case included two questions; the first question concerned the clinical probability of PE and the second was about the choice of a first-line diagnostic test.
19
For a clinical case to be considered valid, it was necessary to have estimated the correct clinical probability and chosen an appropriate first-line diagnostic test. The questionnaire and clinical cases were tested on five GPs practising in the study area before the final versions were produced. The aim of this pre-test was to insure the clarity of the questions and to enable modifications to be made depending on the different responses. This pre-test was also performed to confirm that the test could be completed in a limited time period (mean time between 5 and 6 min) in order to increase the response rate. The questionnaires were sent by post between the 1st and the 15th of February 2013. The time allowed for the responses lasted two months. No reminder was sent, either by telephone or by email. 2.4. Statistical analysis The responses to the survey are described as number and percentage for qualitative variables. Two populations of GPs were defined: (i) GPs with valid responses to at least one clinical case and (ii) GPs with no valid answer to the two clinical cases. Factors associated with knowledge of the diagnostic algorithm were analysed by univariate analysis. p b 0.05 was considered statistically significant. A multivariate logistic regression was used to assess the factors associated with two valid responses to clinical cases. All factors of interest and potential confounding factors were included in the multivariate models. The added value of each covariate was evaluated using a likelihood ratio test. Statistical significance was set a priori at α ≤ 0.05 (2-tailed). All statistical analyses were performed using “R” statistical software (version 2.14.1; http://www.R-project.org/). 3. Results 3.1. Study population The study questionnaires were sent to 508 GPs. One-hundred and fifty-nine responses were received but four of these questionnaires were unusable because they were incomplete. One hundred and fiftyfive questionnaires were analysed (30.5%). Among the responders, 86 (55%) were males and 118 (76%) had been in practice for N 20 years. Only 25 (16%) had additional hospital activity and 18 (12%) subscribed to international medical journals. Among the responders, 105 (68%) underwent CME via doctors' associations or French medical journals and 45 (29%) had participated in
Table 2 Clinical cases scenario. Clinical case no. 1 A 66-year-old female complaining of unusual dyspnoea evolving over several days. She is a former hairdresser, recently retired with no medical history except for post-partum deep vein thrombosis. She is an active smoker. Her last biological examination b3 months ago was normal. Clinically: she has no fever, no thoracic pain, no cough, no sputum production. Normal cardio-pulmonary auscultation. Pulse 98/min, blood pressure 135/80, respiration rate 18/min. No oedema or pain in the lower limbs. Question 1. Expected answer: What is the clinical probability that the patient has a Low, or medium, or unlikely, or not strong, or intermediate; or expressed as a percentage: ≤40% PE? Question 2 Expected answer: Which first-line diagnostic examination would you D-dimer measurement propose? Clinical case no. 2 A 72-year-old hypertensive patient, who underwent radical prostatectomy 21 days ago for prostate cancer. His current treatment includes amlodipin and triptoreline. He complains of pleuritic chest pain. On examination: no fever, pulse 78/min, blood pressure 145/65, respiratory rate 19/min, normal chest sounds and he has a painful right calf, without oedema. His renal function is normal. Question 1. Expected answer: What is the clinical probability that the patient has a Strong, probable, high PE? Expressed as a percentage: ≥70% Question 2. Expected answer: Which first-line diagnostic examination would you Chest computed tomography angiography, or compression ultrasonography of the lower limbs, or propose? ventilation-perfusion lung-scan
20
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22
3.3. Responses to clinical cases
Table 3 Questionnaire and answers about PE risk factors. Questions
Expected answers
7) Among the following factors, which are the risk factors for PE? (Circle your choice or choices) a. Age N65 years b. Cancer c. History of DVT d. Heart failure e. Obesity f. Pregnancy and post-partum g. Oestro-progestative contraception h. Hormone replacement therapy
Answer 7
8) Among the following factors, which would make you suspect a PE? a. Recent chest pain b. Recent dyspnoea c. Malaise d. Haemoptysis e. Purulent sputum f. Tachycardia N 75/min
Yes Yes Yes Yes Yes Yes Yes Yes
Responses (n, %)
86 (55) 132 (85) 153 (99) 153 (99) 67 (43) 86 (55) 137 (88) 82 (53)
Answer 8 Yes Yes Yes Yes No Yes
142 (92) 153 (99) 105 (68) 85 (55) 82 (53) 82 (53)
DVT: Deep Venous Thrombosis.
specific training on PE. The majority of GPs (92%) have had suspected at least one PE since setting up practice. 3.2. Responses to the questionnaire The results regarding the question dealing with the risk factors for PE are shown in Table 3. Four risk factors were clearly identified by the majority of GPs: history of deep-vein thrombosis, oestroprogestative contraception, pregnancy and post-partum and cancer. Three clinical symptoms were considered to be highly suggestive of PE: recent dyspnoea, recent chest pain and unusual tachycardia N75/min. Despite its low prevalence in PE, haemoptysis was selected by 55% of the GPs (Table 3). Only 55% of the GPs were aware of clinical probability scores for PE, while 68% stated that they had used outpatient D-dimer testing. D-dimer testing appears misinterpreted in most of the cases. A large majority (81%) of GPs consider that D-dimer testing allows confirming PE diagnosis. More than one-third (35%) of the GPs were not aware of the predictive value of a negative D-dimer test. For 58% of GPs, there was no link between estimation of clinical probability and D-dimer measurement. Finally, in more than 80% of the cases, the validity of D-dimer tests in patients with comorbidities (cancer and pregnancy) was unknown (Table 4).
Table 5 presents the main results for each clinical case. 3.3.1. Clinical case 1 In this case, the clinical probability was intermediate according to the Geneva revised score (9 points) and the Wells score (4.5 points) and according to guidelines, a D-dimer measurement should be used as first-line diagnostic test. Ninety-nine (64%) GPs gave a valid clinical probability of PE and 78 (50.4%) proposed D-dimers measurement as the first diagnostic test. Other first-line diagnostic tests proposed included a computed tomography pulmonary angiography (33 answers, 21%), lung scanning or compression ultrasonography of the lower limbs in 14 cases (9%), or another diagnostic test in 30 cases (20%). Overall, 68 of the 155 GPs (44%) gave a valid response to this clinical case. 3.3.2. Clinical case 2 In this case, the clinical probability was high according to the Geneva revised score (13 points) and the Wells score (8.5 points) and according to guidelines, CT scan, V/Q scan or compression ultrasonography should be used as first-line diagnostic test. Eighty-five GPs (55%) assessed the clinical probability of PE accurately and 88 (56%) selected chest computed tomography pulmonary angiography, compression ultrasound or lung scanning as the first-line diagnostic test. D-dimer measurement was selected by 32 GPs (21%) despite the high clinical probability. Other diagnostic tests not recommended by the guidelines were suggested by 35 GPs (23%) which include pulmonary MRI, chest X-ray, electrocardiogram, echocardiography, blood test such as complete blood count, and arterial oxygen saturation measurement. Overall, 68 of the 155 GPs (44%) gave a valid response to this clinical case. A total of 40 GPs (26%) gave valid responses to both clinical cases, 54 (35%) had a valid answer to one of the two clinical cases and 61 (39%) gave invalid responses to the two cases. 3.4. Factors associated with knowledge of the diagnostic algorithm In the univariate analysis (Table 6), two factors were significantly associated with two valid responses to clinical cases: participation in specific CME training sessions on PE (p = 0.0036) and knowledge of clinical probability scores for PE (p = 0.0034). In the multivariate analysis only one factor remained independently associated with two correct answers to clinical case: attending a specific CME course on the diagnosis of PE (p = 0.017).
Table 4 Questionnaire and answers concerning PE diagnosis strategy. Questions
Expected answers
Responses (n, %)
9) Do you know that there are validated scores to evaluate the clinical probability of PE? a. Yes b. No 10) In the case of suspicion of PE without signs of severity, would you always carry out outpatient D-tests? a. Yes b. No
Answer 9 Yes
85 (55)
Answer 10 No
49 (32)
11) Concerning D-dimer tests, are the following assertions correct? (Circle your response) * It allows confirming the diagnosis of PE YES NO DON'T KNOW * It is interpretable in patients with cancer YES NO DON'T KNOW * It is interpretable during pregnancy YES NO DON'T KNOW * It allow ruling-out, in some cases, the diagnosis of PE YES NO DON'T KNOW * It should be prescribed according to the clinical probability of PE according to a score YES NO DON'T KNOW
Answer 11
DK: Don't know.
Y
N
DK
No
126 (81)
17 (11)
12 (8)
Yes
20 (13)
85 (55)
50 (32)
Yes
29 (19)
65 (42)
61 (39)
Yes
100 (65)
32 (20)
23 (15)
Yes
65 (42)
14 (9)
76 (49)
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22
21
Table 5 Clinical cases: main results. Question CC1 What is the clinical probability that the patient has a PE? Which first-line diagnostic examination would you propose? CC2 What is the clinical probability that the patient has a PE? Which first-line diagnostic examination would you propose?
Expected answer:
Responses (n, %)
Low, or medium, or unlikely, or not strong, or intermediate; or expressed as a percentage: ≤40%
99 (64%)
D-dimer measurement
78 (50%)
Strong, probable, high Expressed as a percentage: ≥70% Chest computed tomography angiography, or compression ultrasonography of the lower limbs, or ventilation-perfusion lung-scan
85 (55%) 88 (56%)
CC: Clinical Case.
4. Discussion This study shows that the majority of GPs are unaware of the diagnostic algorithm for PE. Our study also identified the participation in specific training on PE and the knowledge of clinical probability scores as two factors associated with the use of a validated diagnostic algorithm. Most surveyed GPs knew the risk factors for PE included in clinical probability scores and most were good at recognising the symptoms of PE [7]. Conversely, the link between age and thrombosis was acknowledged by only 55% of the GPs and 55% cited malaise and haemoptysis as common symptoms of PE. More than one-half of the GPs were unaware of clinical probability scores for PE. However, some GPs may have evaluated clinical probability using their own gestalt [8–10]. A recent study highlights the importance of gut feeling in the GPs' capacity to suspect PE or to evaluate spontaneously clinical probability. In that sense, previous knowledge regarding the patient helps in detecting changes in their conditions, which seems very useful in suspecting PE [10]. GPs who knew about clinical probability scores were more likely to give valid responses to the clinical cases. The majority of GPs did not make the link between
Table 6 Variables associated with two valid clinical cases (univariate analysis). GPs with 2 valid clinical cases (%)
p-Value
Year of experience 1 to 10 20 11 to 20 17 N20 118
N
8 (40) 2 (12) 30 (25)
0.14
Parallel hospital activity N30 130 b30 25
33 (25) 7 (28)
0.79
Subscription to a french journal No 50 11 (22) Yes 105 29 (28)
0.45
Subscription to an international journal No 136 34 (25) Yes 19 6 (32)
0.55
Participation in a physician association No 50 10 (20) Yes 105 30 (29)
0.25
Training on pulmonary embolism No 110 21 (19) Yes 45 19 (42)
0.0036
Diagnosis of an embolism for the date of installation No 25 3 (25) Yes 143 37 (26)
0.95
Validated scores No Yes
0.0034
85 70
14 (17) 26 (37)
clinical probability of PE and D-dimer testing, with 81% of the GPs considering the latter as a screening examination. The low rate of valid responses to the clinical cases highlights the absence of a structured diagnostic approach to guide diagnostic testing. Finally, there may be confusion between the specificity and sensitivity of D-dimer tests. Indeed, D-dimer level is rarely low during cancer [11] or pregnancy [12], resulting in a lower specificity in these contexts but its negative predictive value is preserved in these situations [1]. Our results show that 81% of GPs consider D-dimer test efficient to establish PE diagnosis, highlighting confusion between the negative and positive predictive values of such a test. D-dimer test is highly sensitive but nonspecific for the detection of PE. The majority of GPs subscribed to a medical journal(s) but this did not increase their knowledge about diagnostic algorithms, and even though there are many publications about PE in French journals of general medicine with more than 200 articles on PE published over the past 10 years in the two main journals intended for general practitioners in our country. In our survey, specific training on PE during CME was associated with a better knowledge of the diagnostic strategy than the subscription to one of these journals. However, the present survey is not able to evaluate how specific training on PE could have improved clinical practice for suspected PE but only addresses the knowledge of physicians about this issue. Our results are in line with previous studies suggesting that GPs are less likely to use guidelines than hospital clinicians [13]. Language could be an obstacle to the knowledge of guidelines as no guidelines about PE diagnosis have been published to date in French. Indeed, only 12% of the GPs stated that they were reading international medical journals. Our study has several limitations. The use of closed questions does not allow interpreting the nuances or different aspects of an opinion or knowledge. Thus, Q11 in our study on D-dimer test interpretation in pregnant women may be a source of bias. Furthermore, closed ended questions (e.g. Q8 and Q9) may prompt the GPs to give a response they would not have thought about otherwise and which does not reflect their knowledge. Conversely, open-ended questions, which may better assess the very complex process leading to the suspicion of PE in primary care, would have probably decreased the response rate and are more difficult to interpret. Therefore, the present survey focused the evaluation of knowledge. In addition, the databases used to define the study population did not include the addresses of locum GPs or GPs who were temporarily unemployed at the time of the survey. These GPs, who are often younger, may have given a different profile of responses [14]. Finally, the study area may not represent the whole French population of GPs, especially rural practitioners. Moreover, it is not unlikely that responders may have been more aware about PE management and guidelines than non-responders. This may represent the inevitable and common feature of survey and would falsely increase the level of knowledge. Although PE often presents in general practice, many GPs are unaware of the diagnostic algorithm for PE and do not follow guidelines. Estimation of clinical probability is anecdotal and knowledge about
22
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22
the use of D-dimer is limited. Specific training on PE and knowledge of clinical probability scores are significantly associated with a better knowledge of the diagnostic algorithm of PE. Learning points • General practitioners (GPs) are often the professionals of first recourse for non-critical cases of pulmonary embolism (PE). • The majority of GPs surveyed were unaware of the diagnostic algorithm for PE. • GPs should receive specific training on PE and clinical probability scores as part of their continuing medical education. Addendum B. Planquette, D. Maurice, O. Sanchez and G. Meyer contributed to the design of the study; D. Maurice collected the data; J. Peron was responsible for the statistical analysis. B. Planquette and D. Maurice analysed results and drafted the manuscript. O. Sanchez, G. Mourin, A. Ferre and G. Meyer reviewed the manuscript. All authors approved the final version of the manuscript. Funding Support for this project was received from Oxyvie, France in manuscript writing assistance. Conflict of interests The authors report no conflict of interest. References [1] Authors/Task Force Members, Konstantinides S, Torbicki A, Agnelli G, Danchin N, Fitzmaurice D, et al. 2014 ESC Guidelines on the Diagnosis and Management of Acute Pulmonary Embolism: The Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC) Endorsed by the European Respiratory Society (ERS). Eur Heart J Nov 14 2014;35(43): 3033–73. http://dx.doi.org/10.1093/eurheartj/ehu283 (Epub 2014 Aug 29). [2] Geersing G-J, Erkens PMG, Lucassen WAM, Büller HR, Cate HT, Hoes AW, et al. Safe exclusion of pulmonary embolism using the Wells rule and qualitative D-dimer testing in primary care: prospective cohort study. BMJ 2012;345:e6564.
[3] Spencer FA, Lessard D, Emery C, Reed G, Goldberg RJ. Venous thromboembolism in the outpatient setting. Arch Intern Med Jul 23 2007;167(14):1471–5. [4] Aujesky D, Roy P-M, Verschuren F, Righini M, Osterwalder J, Egloff M, et al. Outpatient versus inpatient treatment for patients with acute pulmonary embolism: an international, open-label, randomised, non-inferiority trial. Lancet Jul 2 2011; 378(9785):41–8. [5] Zondag W, Kooiman J, Klok FA, Dekkers OM, Huisman MV. Outpatient versus inpatient treatment in patients with pulmonary embolism: a meta-analysis. Eur Respir J Jul 2013;42(1):134–44. [6] Roy P-M, Meyer G, Vielle B, Le Gall C, Verschuren F, Carpentier F, et al. Appropriateness of diagnostic management and outcomes of suspected pulmonary embolism. Ann Intern Med Feb 7 2006;144(3):157–64. [7] Le Gal G, Righini M, Sanchez O, Roy P-M, Baba-Ahmed M, Perrier A, et al. A positive compression ultrasonography of the lower limb veins is highly predictive of pulmonary embolism on computed tomography in suspected patients. Thromb Haemost Jun 2006;95(6):963–6. [8] Chagnon I, Bounameaux H, Aujesky D, Roy P-M, Gourdier A-L, Cornuz J, et al. Comparison of two clinical prediction rules and implicit assessment among patients with suspected pulmonary embolism. Am J Med Sep 2002;113(4):269–75. [9] Chunilal SD, Eikelboom JW, Attia J, Miniati M, Panju AA, Simel DL, et al. Does this patient have pulmonary embolism? JAMA Dec 3 2003;290(21):2849–58. [10] Barais M, Morio N, Cuzon Breton A, Barraine P, Calvez A, Stolper E, et al. “I can't find anything wrong: it must be a pulmonary embolism”: diagnosing suspected pulmonary embolism in primary care, a qualitative study. PLoS One 2014;9(5): e98112. [11] Righini M, Le Gal G, De Lucia S, Roy P-M, Meyer G, Aujesky D, et al. Clinical usefulness of D-dimer testing in cancer patients with suspected pulmonary embolism. Thromb Haemost Apr 2006;95(4):715–9. [12] Boehlen F, Epiney M, Boulvain M, Irion O, de Moerloose P. Changes in D-dimer levels during pregnancy and the postpartum period: results of two studies. Rev Med Suisse Jan 26 2005;1(4):296–8. [13] Cabana MD, Rand CS, Powe NR, Wu AW, Wilson MH, Abboud PA, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA Oct 20 1999;282(15):1458–65. [14] Branchereau E, Branger B, Launay E, Verstraete M, Vrignaud B, Levieux K, et al. Management of bronchiolitis in general practice and determinants of treatment being discordant with guidelines of the HAS. Arch Pediatr Dec 2013;20(12): 1369–75.
Contents lists available at ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original Article
Knowledge of the diagnostic algorithm for pulmonary embolism in primary care B. Planquette a,b,c,⁎, D. Maurice a, J. Peron d,e,f, G. Mourin a,b, A. Ferre a,b, O. Sanchez a,b,c, G. Meyer a,b,c a
Service de pneumologie et de soins intensifs, Hôpital Européen Georges Pompidou, Assistance Publique Hôpitaux de Paris, Paris, France Université Paris Descartes, Sorbonne Paris Cité, France ISERM U970-PARCC, Paris, France d Hospices Civils de Lyon, Service de Biostatistique, F-69003 Lyon, France e CNRS UMR 5558, Laboratoire de Biométrie et Biologie Evolutive, Equipe Biostatistique-Santé, Villeurbanne, France f Université de Lyon, F-69622, Lyon, France b c
a r t i c l e
i n f o
Article history: Received 22 July 2014 Received in revised form 18 November 2014 Accepted 21 November 2014 Available online 11 December 2014 Keywords: Algorithms Differential diagnosis Disease management Pulmonary embolism General practice Primary care physician
a b s t r a c t Background: Diagnostic algorithms for pulmonary embolism (PE) have been validated in patients attending hospital emergency departments. However, general practitioners (GPs) are often the professionals of first resort for the majority of non-critical cases of PE. Aim: To evaluate the knowledge of the diagnostic algorithm for PE among GPs in France. Design and setting: Questionnaire-based survey of GPs with a private practice. Method: All GPs in the study area were sent a questionnaire including several questions on the diagnosis of PE and two clinical cases scenario with suspected PE. Factors associated with knowledge of the diagnostic algorithm were analysed by univariate and multivariate analyses. Results: Five-hundred and eight questionnaires were distributed and 155 (30.5%) were available for analysis. Only 55% of the GPs did know about clinical scores for the assessment of clinical probability of PE and 42% of the GPs were aware that clinical probability is needed to interpret the result of D-dimer testing. Forty GPs (26%) gave valid responses to both clinical cases, 54 GPs (35%) had one valid case out of the two and 61 (39%) gave invalid responses to both clinical cases. Participation in specific training on PE was significantly associated with valid responses to the two clinical cases in multivariate analysis (p b 0.017). Conclusion: The majority of GPs were unaware of the diagnostic algorithm for PE. Clinical probability was rarely assessed and knowledge about D-dimers was poor. Specific training on PE and greater awareness of clinical probability scores may promote knowledge of PE algorithm diagnosis. © 2014 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
1. Introduction Pulmonary embolism (PE) is a common and potentially fatal disease. Significant improvements have been achieved in the diagnostic approach to PE over the past 20 years and this ended-up with diagnostic algorithms endorsed by international guidelines [1]. These diagnostic algorithms have been developed and validated in outpatients referred to emergency departments, and are thus applicable in general medicine. A low clinical probability combined with a negative D-dimer test excludes the diagnosis of PE in primary care with a sensitivity of 97.3% [2]. About three-quarters of venous thromboembolic events take place in the community, thus, general practitioners (GPs) are often the health professionals of first recourse for clinically stable patients with suspected PE [3]. The opportunity for outpatient treatment is currently under evaluation in patients with low-risk PE [4,5]. This new perspective ⁎ Corresponding author at: Service de Pneumologie et de soins intensifs, Hôpital Européen Georges Pompidou, 20 rue Leblanc, F-75015 Paris, France. Tel.: +33 1 56 09 34 62; fax: +33 1 56 09 32 55. E-mail address: [email protected] (B. Planquette).
of ambulatory treatment may lead to GPs becoming major players in the initial management of PE. Since non-adherence to guidelines in the diagnosis of PE has been associated with a significant increase in the risk of adverse events, the knowledge of diagnostic guidelines by GPs is of great importance for the care of patients with suspected PE [6]. The aim of this study is to evaluate the state of knowledge of the diagnostic algorithm for PE among GPs in the south of Paris and its suburb. 2. Method 2.1. Study population This study was carried out among GPs with a private practice. The study area was defined choosing districts and towns around our hospital where GPs usually refer their patients in case of PE suspicion. The area included the 15th and 16th districts of Paris and two towns in the near suburb of Paris: Vanves and Issy-Les-Moulineaux. The addresses of GPs with a private practice in the study area were obtained via the website http://ameli-direct.ameli.fr. The data from this site were
http://dx.doi.org/10.1016/j.ejim.2014.11.005 0953-6205/© 2014 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22 Table 1 Questionnaire (general practitioners characteristics). 1) You are a. Female b. Male 2) How many years have you been in practice? a. Between 1 and 10 b. Between 11 and 20 c. More than 20 3) Do you have any hospital activity in addition to your private practice? (Emergency, outpatients clinics at hospital …) a. Yes b. No 4) Do you have continuing medical education (CME) activities? (Several answers possible) a. Subscription to French medical journals b. Subscription to international journals c. Affiliation to associations of doctors for CME 5) Have you already taken part in training concerning PE? a. Yes b. No 6) Have you already suspected a PE since setting up your practice? a. Yes b. No
crossed referenced with the directory of GPs available on the Conseil National de l'Ordre des Médecins (CNOM) website (http://www. conseil-national.medecin.fr). A questionnaire and two simulated clinical case scenarios were sent to the GPs by post (see supplementary appendix). A stamped-addressed return envelope was included with these documents. 2.2. Questionnaire The survey consisted of 11 questions (Tables 1, 3 and 4) covering different subjects: demographic characteristics and professional situation, continuing medical education sessions on PE (CME) knowledge about risk factors, clinical symptoms of PE and clinical probability scores, and use of D-dimer testing in case of suspected PE. 2.3. Simulated clinical cases scenario Two clinical cases describing patients with suspected PE were distributed to the GPs (Table 2). The clinical probability of PE was intermediate for case 1 and high for case 2. Each clinical case included two questions; the first question concerned the clinical probability of PE and the second was about the choice of a first-line diagnostic test.
19
For a clinical case to be considered valid, it was necessary to have estimated the correct clinical probability and chosen an appropriate first-line diagnostic test. The questionnaire and clinical cases were tested on five GPs practising in the study area before the final versions were produced. The aim of this pre-test was to insure the clarity of the questions and to enable modifications to be made depending on the different responses. This pre-test was also performed to confirm that the test could be completed in a limited time period (mean time between 5 and 6 min) in order to increase the response rate. The questionnaires were sent by post between the 1st and the 15th of February 2013. The time allowed for the responses lasted two months. No reminder was sent, either by telephone or by email. 2.4. Statistical analysis The responses to the survey are described as number and percentage for qualitative variables. Two populations of GPs were defined: (i) GPs with valid responses to at least one clinical case and (ii) GPs with no valid answer to the two clinical cases. Factors associated with knowledge of the diagnostic algorithm were analysed by univariate analysis. p b 0.05 was considered statistically significant. A multivariate logistic regression was used to assess the factors associated with two valid responses to clinical cases. All factors of interest and potential confounding factors were included in the multivariate models. The added value of each covariate was evaluated using a likelihood ratio test. Statistical significance was set a priori at α ≤ 0.05 (2-tailed). All statistical analyses were performed using “R” statistical software (version 2.14.1; http://www.R-project.org/). 3. Results 3.1. Study population The study questionnaires were sent to 508 GPs. One-hundred and fifty-nine responses were received but four of these questionnaires were unusable because they were incomplete. One hundred and fiftyfive questionnaires were analysed (30.5%). Among the responders, 86 (55%) were males and 118 (76%) had been in practice for N 20 years. Only 25 (16%) had additional hospital activity and 18 (12%) subscribed to international medical journals. Among the responders, 105 (68%) underwent CME via doctors' associations or French medical journals and 45 (29%) had participated in
Table 2 Clinical cases scenario. Clinical case no. 1 A 66-year-old female complaining of unusual dyspnoea evolving over several days. She is a former hairdresser, recently retired with no medical history except for post-partum deep vein thrombosis. She is an active smoker. Her last biological examination b3 months ago was normal. Clinically: she has no fever, no thoracic pain, no cough, no sputum production. Normal cardio-pulmonary auscultation. Pulse 98/min, blood pressure 135/80, respiration rate 18/min. No oedema or pain in the lower limbs. Question 1. Expected answer: What is the clinical probability that the patient has a Low, or medium, or unlikely, or not strong, or intermediate; or expressed as a percentage: ≤40% PE? Question 2 Expected answer: Which first-line diagnostic examination would you D-dimer measurement propose? Clinical case no. 2 A 72-year-old hypertensive patient, who underwent radical prostatectomy 21 days ago for prostate cancer. His current treatment includes amlodipin and triptoreline. He complains of pleuritic chest pain. On examination: no fever, pulse 78/min, blood pressure 145/65, respiratory rate 19/min, normal chest sounds and he has a painful right calf, without oedema. His renal function is normal. Question 1. Expected answer: What is the clinical probability that the patient has a Strong, probable, high PE? Expressed as a percentage: ≥70% Question 2. Expected answer: Which first-line diagnostic examination would you Chest computed tomography angiography, or compression ultrasonography of the lower limbs, or propose? ventilation-perfusion lung-scan
20
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22
3.3. Responses to clinical cases
Table 3 Questionnaire and answers about PE risk factors. Questions
Expected answers
7) Among the following factors, which are the risk factors for PE? (Circle your choice or choices) a. Age N65 years b. Cancer c. History of DVT d. Heart failure e. Obesity f. Pregnancy and post-partum g. Oestro-progestative contraception h. Hormone replacement therapy
Answer 7
8) Among the following factors, which would make you suspect a PE? a. Recent chest pain b. Recent dyspnoea c. Malaise d. Haemoptysis e. Purulent sputum f. Tachycardia N 75/min
Yes Yes Yes Yes Yes Yes Yes Yes
Responses (n, %)
86 (55) 132 (85) 153 (99) 153 (99) 67 (43) 86 (55) 137 (88) 82 (53)
Answer 8 Yes Yes Yes Yes No Yes
142 (92) 153 (99) 105 (68) 85 (55) 82 (53) 82 (53)
DVT: Deep Venous Thrombosis.
specific training on PE. The majority of GPs (92%) have had suspected at least one PE since setting up practice. 3.2. Responses to the questionnaire The results regarding the question dealing with the risk factors for PE are shown in Table 3. Four risk factors were clearly identified by the majority of GPs: history of deep-vein thrombosis, oestroprogestative contraception, pregnancy and post-partum and cancer. Three clinical symptoms were considered to be highly suggestive of PE: recent dyspnoea, recent chest pain and unusual tachycardia N75/min. Despite its low prevalence in PE, haemoptysis was selected by 55% of the GPs (Table 3). Only 55% of the GPs were aware of clinical probability scores for PE, while 68% stated that they had used outpatient D-dimer testing. D-dimer testing appears misinterpreted in most of the cases. A large majority (81%) of GPs consider that D-dimer testing allows confirming PE diagnosis. More than one-third (35%) of the GPs were not aware of the predictive value of a negative D-dimer test. For 58% of GPs, there was no link between estimation of clinical probability and D-dimer measurement. Finally, in more than 80% of the cases, the validity of D-dimer tests in patients with comorbidities (cancer and pregnancy) was unknown (Table 4).
Table 5 presents the main results for each clinical case. 3.3.1. Clinical case 1 In this case, the clinical probability was intermediate according to the Geneva revised score (9 points) and the Wells score (4.5 points) and according to guidelines, a D-dimer measurement should be used as first-line diagnostic test. Ninety-nine (64%) GPs gave a valid clinical probability of PE and 78 (50.4%) proposed D-dimers measurement as the first diagnostic test. Other first-line diagnostic tests proposed included a computed tomography pulmonary angiography (33 answers, 21%), lung scanning or compression ultrasonography of the lower limbs in 14 cases (9%), or another diagnostic test in 30 cases (20%). Overall, 68 of the 155 GPs (44%) gave a valid response to this clinical case. 3.3.2. Clinical case 2 In this case, the clinical probability was high according to the Geneva revised score (13 points) and the Wells score (8.5 points) and according to guidelines, CT scan, V/Q scan or compression ultrasonography should be used as first-line diagnostic test. Eighty-five GPs (55%) assessed the clinical probability of PE accurately and 88 (56%) selected chest computed tomography pulmonary angiography, compression ultrasound or lung scanning as the first-line diagnostic test. D-dimer measurement was selected by 32 GPs (21%) despite the high clinical probability. Other diagnostic tests not recommended by the guidelines were suggested by 35 GPs (23%) which include pulmonary MRI, chest X-ray, electrocardiogram, echocardiography, blood test such as complete blood count, and arterial oxygen saturation measurement. Overall, 68 of the 155 GPs (44%) gave a valid response to this clinical case. A total of 40 GPs (26%) gave valid responses to both clinical cases, 54 (35%) had a valid answer to one of the two clinical cases and 61 (39%) gave invalid responses to the two cases. 3.4. Factors associated with knowledge of the diagnostic algorithm In the univariate analysis (Table 6), two factors were significantly associated with two valid responses to clinical cases: participation in specific CME training sessions on PE (p = 0.0036) and knowledge of clinical probability scores for PE (p = 0.0034). In the multivariate analysis only one factor remained independently associated with two correct answers to clinical case: attending a specific CME course on the diagnosis of PE (p = 0.017).
Table 4 Questionnaire and answers concerning PE diagnosis strategy. Questions
Expected answers
Responses (n, %)
9) Do you know that there are validated scores to evaluate the clinical probability of PE? a. Yes b. No 10) In the case of suspicion of PE without signs of severity, would you always carry out outpatient D-tests? a. Yes b. No
Answer 9 Yes
85 (55)
Answer 10 No
49 (32)
11) Concerning D-dimer tests, are the following assertions correct? (Circle your response) * It allows confirming the diagnosis of PE YES NO DON'T KNOW * It is interpretable in patients with cancer YES NO DON'T KNOW * It is interpretable during pregnancy YES NO DON'T KNOW * It allow ruling-out, in some cases, the diagnosis of PE YES NO DON'T KNOW * It should be prescribed according to the clinical probability of PE according to a score YES NO DON'T KNOW
Answer 11
DK: Don't know.
Y
N
DK
No
126 (81)
17 (11)
12 (8)
Yes
20 (13)
85 (55)
50 (32)
Yes
29 (19)
65 (42)
61 (39)
Yes
100 (65)
32 (20)
23 (15)
Yes
65 (42)
14 (9)
76 (49)
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22
21
Table 5 Clinical cases: main results. Question CC1 What is the clinical probability that the patient has a PE? Which first-line diagnostic examination would you propose? CC2 What is the clinical probability that the patient has a PE? Which first-line diagnostic examination would you propose?
Expected answer:
Responses (n, %)
Low, or medium, or unlikely, or not strong, or intermediate; or expressed as a percentage: ≤40%
99 (64%)
D-dimer measurement
78 (50%)
Strong, probable, high Expressed as a percentage: ≥70% Chest computed tomography angiography, or compression ultrasonography of the lower limbs, or ventilation-perfusion lung-scan
85 (55%) 88 (56%)
CC: Clinical Case.
4. Discussion This study shows that the majority of GPs are unaware of the diagnostic algorithm for PE. Our study also identified the participation in specific training on PE and the knowledge of clinical probability scores as two factors associated with the use of a validated diagnostic algorithm. Most surveyed GPs knew the risk factors for PE included in clinical probability scores and most were good at recognising the symptoms of PE [7]. Conversely, the link between age and thrombosis was acknowledged by only 55% of the GPs and 55% cited malaise and haemoptysis as common symptoms of PE. More than one-half of the GPs were unaware of clinical probability scores for PE. However, some GPs may have evaluated clinical probability using their own gestalt [8–10]. A recent study highlights the importance of gut feeling in the GPs' capacity to suspect PE or to evaluate spontaneously clinical probability. In that sense, previous knowledge regarding the patient helps in detecting changes in their conditions, which seems very useful in suspecting PE [10]. GPs who knew about clinical probability scores were more likely to give valid responses to the clinical cases. The majority of GPs did not make the link between
Table 6 Variables associated with two valid clinical cases (univariate analysis). GPs with 2 valid clinical cases (%)
p-Value
Year of experience 1 to 10 20 11 to 20 17 N20 118
N
8 (40) 2 (12) 30 (25)
0.14
Parallel hospital activity N30 130 b30 25
33 (25) 7 (28)
0.79
Subscription to a french journal No 50 11 (22) Yes 105 29 (28)
0.45
Subscription to an international journal No 136 34 (25) Yes 19 6 (32)
0.55
Participation in a physician association No 50 10 (20) Yes 105 30 (29)
0.25
Training on pulmonary embolism No 110 21 (19) Yes 45 19 (42)
0.0036
Diagnosis of an embolism for the date of installation No 25 3 (25) Yes 143 37 (26)
0.95
Validated scores No Yes
0.0034
85 70
14 (17) 26 (37)
clinical probability of PE and D-dimer testing, with 81% of the GPs considering the latter as a screening examination. The low rate of valid responses to the clinical cases highlights the absence of a structured diagnostic approach to guide diagnostic testing. Finally, there may be confusion between the specificity and sensitivity of D-dimer tests. Indeed, D-dimer level is rarely low during cancer [11] or pregnancy [12], resulting in a lower specificity in these contexts but its negative predictive value is preserved in these situations [1]. Our results show that 81% of GPs consider D-dimer test efficient to establish PE diagnosis, highlighting confusion between the negative and positive predictive values of such a test. D-dimer test is highly sensitive but nonspecific for the detection of PE. The majority of GPs subscribed to a medical journal(s) but this did not increase their knowledge about diagnostic algorithms, and even though there are many publications about PE in French journals of general medicine with more than 200 articles on PE published over the past 10 years in the two main journals intended for general practitioners in our country. In our survey, specific training on PE during CME was associated with a better knowledge of the diagnostic strategy than the subscription to one of these journals. However, the present survey is not able to evaluate how specific training on PE could have improved clinical practice for suspected PE but only addresses the knowledge of physicians about this issue. Our results are in line with previous studies suggesting that GPs are less likely to use guidelines than hospital clinicians [13]. Language could be an obstacle to the knowledge of guidelines as no guidelines about PE diagnosis have been published to date in French. Indeed, only 12% of the GPs stated that they were reading international medical journals. Our study has several limitations. The use of closed questions does not allow interpreting the nuances or different aspects of an opinion or knowledge. Thus, Q11 in our study on D-dimer test interpretation in pregnant women may be a source of bias. Furthermore, closed ended questions (e.g. Q8 and Q9) may prompt the GPs to give a response they would not have thought about otherwise and which does not reflect their knowledge. Conversely, open-ended questions, which may better assess the very complex process leading to the suspicion of PE in primary care, would have probably decreased the response rate and are more difficult to interpret. Therefore, the present survey focused the evaluation of knowledge. In addition, the databases used to define the study population did not include the addresses of locum GPs or GPs who were temporarily unemployed at the time of the survey. These GPs, who are often younger, may have given a different profile of responses [14]. Finally, the study area may not represent the whole French population of GPs, especially rural practitioners. Moreover, it is not unlikely that responders may have been more aware about PE management and guidelines than non-responders. This may represent the inevitable and common feature of survey and would falsely increase the level of knowledge. Although PE often presents in general practice, many GPs are unaware of the diagnostic algorithm for PE and do not follow guidelines. Estimation of clinical probability is anecdotal and knowledge about
22
B. Planquette et al. / European Journal of Internal Medicine 26 (2015) 18–22
the use of D-dimer is limited. Specific training on PE and knowledge of clinical probability scores are significantly associated with a better knowledge of the diagnostic algorithm of PE. Learning points • General practitioners (GPs) are often the professionals of first recourse for non-critical cases of pulmonary embolism (PE). • The majority of GPs surveyed were unaware of the diagnostic algorithm for PE. • GPs should receive specific training on PE and clinical probability scores as part of their continuing medical education. Addendum B. Planquette, D. Maurice, O. Sanchez and G. Meyer contributed to the design of the study; D. Maurice collected the data; J. Peron was responsible for the statistical analysis. B. Planquette and D. Maurice analysed results and drafted the manuscript. O. Sanchez, G. Mourin, A. Ferre and G. Meyer reviewed the manuscript. All authors approved the final version of the manuscript. Funding Support for this project was received from Oxyvie, France in manuscript writing assistance. Conflict of interests The authors report no conflict of interest. References [1] Authors/Task Force Members, Konstantinides S, Torbicki A, Agnelli G, Danchin N, Fitzmaurice D, et al. 2014 ESC Guidelines on the Diagnosis and Management of Acute Pulmonary Embolism: The Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC) Endorsed by the European Respiratory Society (ERS). Eur Heart J Nov 14 2014;35(43): 3033–73. http://dx.doi.org/10.1093/eurheartj/ehu283 (Epub 2014 Aug 29). [2] Geersing G-J, Erkens PMG, Lucassen WAM, Büller HR, Cate HT, Hoes AW, et al. Safe exclusion of pulmonary embolism using the Wells rule and qualitative D-dimer testing in primary care: prospective cohort study. BMJ 2012;345:e6564.
[3] Spencer FA, Lessard D, Emery C, Reed G, Goldberg RJ. Venous thromboembolism in the outpatient setting. Arch Intern Med Jul 23 2007;167(14):1471–5. [4] Aujesky D, Roy P-M, Verschuren F, Righini M, Osterwalder J, Egloff M, et al. Outpatient versus inpatient treatment for patients with acute pulmonary embolism: an international, open-label, randomised, non-inferiority trial. Lancet Jul 2 2011; 378(9785):41–8. [5] Zondag W, Kooiman J, Klok FA, Dekkers OM, Huisman MV. Outpatient versus inpatient treatment in patients with pulmonary embolism: a meta-analysis. Eur Respir J Jul 2013;42(1):134–44. [6] Roy P-M, Meyer G, Vielle B, Le Gall C, Verschuren F, Carpentier F, et al. Appropriateness of diagnostic management and outcomes of suspected pulmonary embolism. Ann Intern Med Feb 7 2006;144(3):157–64. [7] Le Gal G, Righini M, Sanchez O, Roy P-M, Baba-Ahmed M, Perrier A, et al. A positive compression ultrasonography of the lower limb veins is highly predictive of pulmonary embolism on computed tomography in suspected patients. Thromb Haemost Jun 2006;95(6):963–6. [8] Chagnon I, Bounameaux H, Aujesky D, Roy P-M, Gourdier A-L, Cornuz J, et al. Comparison of two clinical prediction rules and implicit assessment among patients with suspected pulmonary embolism. Am J Med Sep 2002;113(4):269–75. [9] Chunilal SD, Eikelboom JW, Attia J, Miniati M, Panju AA, Simel DL, et al. Does this patient have pulmonary embolism? JAMA Dec 3 2003;290(21):2849–58. [10] Barais M, Morio N, Cuzon Breton A, Barraine P, Calvez A, Stolper E, et al. “I can't find anything wrong: it must be a pulmonary embolism”: diagnosing suspected pulmonary embolism in primary care, a qualitative study. PLoS One 2014;9(5): e98112. [11] Righini M, Le Gal G, De Lucia S, Roy P-M, Meyer G, Aujesky D, et al. Clinical usefulness of D-dimer testing in cancer patients with suspected pulmonary embolism. Thromb Haemost Apr 2006;95(4):715–9. [12] Boehlen F, Epiney M, Boulvain M, Irion O, de Moerloose P. Changes in D-dimer levels during pregnancy and the postpartum period: results of two studies. Rev Med Suisse Jan 26 2005;1(4):296–8. [13] Cabana MD, Rand CS, Powe NR, Wu AW, Wilson MH, Abboud PA, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA Oct 20 1999;282(15):1458–65. [14] Branchereau E, Branger B, Launay E, Verstraete M, Vrignaud B, Levieux K, et al. Management of bronchiolitis in general practice and determinants of treatment being discordant with guidelines of the HAS. Arch Pediatr Dec 2013;20(12): 1369–75.
