- Email: [email protected]
The Road to Overdiagnosis
Perspective
The Road to Overdiagnosis: The Case of Subsegmental Pulmonary Embolism Saurabh Jha, MD Key Words: Overdiagnosis; pulmonary embolism; subsegmental pulmonary embolism; PIOPED. ªAUR, 2015
P
ulmonary embolism (PE) is a dreaded condition which nearly every physician, in practice or training, is aware about. It is a leading and avoidable cause of sudden death (1). That PE is now overdiagnosed (2), merits reflection. PE’s journey from an acute emergency which physicians feared missing, to a frequently overdiagnosed entity leading to overtreatment, is instructive at multiple levels, showcasing the inevitability of overdiagnosis and exposing barriers to its reduction. Embolism to the pulmonary arteries, if large, can lead to right heart strain, cardiovascular collapse, and sudden death (3). PE is in the differential for pulseless electrical activity, a type of cardiac arrest. PE, more often, presents less dramatically and such presentations can be so nonspecific that a range of symptoms is attributable to PE. Emboli that do not cause cardiovascular collapse at their first presentation are not important for their own sake but because they herald a much larger clot burden in the future. This means that the significance of PE lies in its recurrence. That the risk of recurrence can be reduced by anticoagulation makes recurrent PE an avoidable harm. The suspicion for PE induces uncertainty exacerbated by the fact that death from recurrent PE is potentially avoidable. The imperative to diagnose PE is substantial, particularly in the young patient in the emergency department. With such a high degree of clinical uncertainty, responsibility for diagnosis is shared by imagers. The gold standard for the diagnosis of PE was once catheter angiogram (4), at which thrombus in the pulmonary arteries is straightforward to detect, as it appears as a filling defect. However, catheter angiography is invasive and operationally difficult to perform in everyone suspected of PE. The first test for the diagnosis of PE used to be ventilation-perfusion (VQ) scan, in which PE was inferred, rather than anatomically depicted. The efficacy of VQ scans in the diagnosis of PE was shown in the Prospective Investigation of Pulmonary Embolism Acad Radiol 2015; -:1–3 From the Department of Radiology, University of Pennsylvania, 3400 Spruce St, Philadelphia, PA 19104. Received May 24, 2015; accepted May 25, 2015. Address correspondence to: S.J. e-mail: [email protected]. edu, @RogueRad ªAUR, 2015 http://dx.doi.org/10.1016/j.acra.2015.05.001
Diagnosis (PIOPED) (5). Reporting of VQ scans is an exemplar of standardization. The imager grades the likelihood of PE as ‘‘high,’’ ‘‘intermediate,’’ and ‘‘low’’ probability based on objective criteria drawn from PIOPED. The imager then merges the findings on VQ with the prior probability of PE to give a context-dependent interpretation. Paradoxically, and despite reporting of VQ scans being evidence-based, VQ increases uncertainty. Why the uncertainty is increased is because the ‘‘intermediate’’ category includes a wide bandwidth of probabilities and is a frequent finding. In PIOPED, nearly 40% of patients with suspected PE had an ‘‘intermediate’’ probability VQ scan. Of those with intermediate probability VQ, only one-third had PE (5). Furthermore, by contextualizing the interpretation with pretest probability of PE, the imager explicitly involves the referring physician in the interpretation. For example, if the VQ scan is of low probability for PE, the chances of PE remain as high as 12% if the patient is of a high pretest probability for PE (5). When clinicians request a diagnostic test for PE they prefer clinical closure with a dichotomous answer; affirmative or negative–is there a pulmonary embolus or not, yes or no? VQ scans do not give a dichotomous answer, but multidetector computed tomography (MDCT) can. PIOPED 2 established MDCT’s role in the detection of PE (6). MDCT is the preferred test for PE (7) and the de facto gold standard for the diagnosis of PE (8). With improving spatial resolution of MDCT, imagers see smaller arteries, including subsegmental pulmonary arteries, and it is here that MDCT has expanded the definition of disease, increasing the reservoir for inevitable overdiagnosis. PE is clot in the pulmonary arteries. However, a saddle embolus is different, in clinical severity, to an isolated subsegmental pulmonary embolus, although both fulfill the definition of PE. The significance of PE lies, however, not just in its occurrence but its potential for recurrence. This means that to declare that a subsegmental pulmonary embolus is not significant one must define what an acceptable risk of recurrence of PE is and show that subsegmental emboli do not meet that risk threshold. It is easier to show the significance of a condition than it is to prove its insignificance. This is an epistemological problem. Physicians are confronted with strong evidence of the 1
JHA
significance of large pulmonary emboli, which drives the testing (9), and in the process, they discover small pulmonary emboli. The imperative to not miss large PE is stronger than the imperative to not treat small PE. Although clinicians do not seek subsegmental pulmonary emboli, they are not confident enough to ignore them. The evidence for overdiagnosis of PE from MDCT is indirect, but compelling. The incidence of PE increased by 80% from 62.1 per 100,000 to 112.3 per 100,000 between 1998 and 2006, after the introduction of MDCT and despite prophylaxis for thromboembolic disease (10). The increased detection of PE has not reduced the age-adjusted mortality from PE, by a comparative degree, which reduced from 12.3 only to 11.9 per 100,000 during the same period. Stated differently, we are seeing more, treating more but not improving outcomes to the same extent. This points to redundancy and overdiagnosis. Of note, MDCT finds more subsegmental pulmonary emboli than VQ scans–1% of VQ scans (11) rated high probability show a subsegmental PE, whereas of the MDCT positive for PE, 15% have subsegmental PE (12). The clarity with which MDCT depicts small pulmonary arteries is not constant, as quality varies due to physiological and technical factors affecting contrast delivery. However, MDCT has established a floor of expectation that subsegmental pulmonary arteries will be visualized. This has led many imagers to hedge when they cannot see the subsegmental arteries by saying that isolated embolus to subsegmental pulmonary arteries cannot be excluded. The expansion of the disease to smaller arteries has pushed interpretation toward a higher sensitivity for the detection of subsegmental PE, at the expense of specificity, increasing false positives. Given that MDCT is the arbiter of PE, it is not possible to distinguish between a false positive diagnosis and an overdiagnosis of subsegmental PE. Such a distinction may not be important as the futility and risks of treatment apply both to the false positive and the overdiagnosed. The harm from overdiagnosis of PE is from anticoagulation, which has risks. In one series of 93 patients with subsegmental PE, the risk of bleeding was 5.3% and the risk of recurrence was 0.7% (13). Projecting these findings to a cohort of 1000 patients, 53 patients will have major bleeding to save seven patients from recurrent PE. Or nearly eight patients with subsegmental PE must be harmed to save one person from recurrent PE. Is this an acceptable trade-off? The overdiagnosis of PE is exacerbated by the use of the Ddimer test which, instead of gatekeeping MDCT, is a gateway to overdiagnosis (14). The threshold for positivity of a Ddimer is very low. This is deliberate to give D-dimer a high negative predictive value, at the expense of specificity, so that when the test is negative, the clinician is confident that the patient does not have a PE. However, the availability of the D-dimer has increased its use leading to a growing reservoir of declining pretest probability of PE. This phenomenon is known as indication drift, and it has been demonstrated in other settings (15). 2
Academic Radiology, Vol -, No -, - 2015
The overdiagnosis of PE is a result of our intolerance of uncertainty, the want of dichotomous answers, and an emergence of technology. To mitigate overdiagnosis, specialty societies must state in no uncertain terms that the diagnosis, and treatment, of subsegmental pulmonary emboli is harmful and should be avoided. The management of PE is at the crossroads of multiple disciplines including emergency medicine, radiology, and hematology, and this begs the question which specialty will be tasked for issuing such a statement. Will emergency physicians, radiologists, and hematologists agree with each another? Will there be agreement between physicians of the same specialty? Can radiologists even agree that they must stop hedging that subsegmental pulmonary emboli cannot be excluded? The difficulty in ascertaining the insignificance of a condition is best illustrated in an article in the British Medical Journal about the overdiagnosis of PE (2). The authors, experts in overdiagnosis, argue convincingly that there is overdiagnosis of PE. However, they then add disclaimers which are quoted verbatim. ‘‘Definitive evidence of overdiagnosis would, of course, be the finding that untreated patients never experience harm from the PE during the rest of their lives and died from another cause, but no such studies exist’’ and ‘‘we need to learn which small emboli need treatment.’’ It is unlikely that these disclaimers will reassure physicians in the frontline that they can ignore subsegmental PE, even in patients who do not have malignancy or a prothrombotic state. As in other conditions, overdiagnosis of subsegmental PE is an information problem. To recap the triad of the information problem, many patients will not benefit from treatment of subsegmental PE, but some will, and we do not know who will or will not with certainty, ex ante. We have implicitly stated our preference, though, which is that it is better we over treat many patients with subsegmental pulmonary emboli than we under treat some. Until this calculus changes, overdiagnosis and overtreatment of subsegmental PE will continue. One way to reduce overdiagnosis of PE is to raise the bar for its suspicion through evidence and actionable thresholds. It is unlikely this can be achieved without a significant change in medical culture. REFERENCES 1. National Heart, Lung and Blood Institute. What is pulmonary embolism? http://www.nhlbi.nih.gov/health/health-topics/topics/pe (accessed May 24, 2015) 2. 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. 3. Handler JA, Feied CF. Acute pulmonary embolism. Aggressive therapy with anticoagulants and thrombolytics. Postgrad Med 1995; 97:61–62. 65-8, 71-2. 4. Sasahara AA, Stein M, Simon M, et al. Pulmonary angiography in the diagnosis of thromboembolic disease. NEJM 1964; 270:1075–1081.
Academic Radiology, Vol -, No -, - 2015
5. The Pioped Investigators. Value of the ventilation/perfusion scan in acute pulmonary embolus. The results of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED). JAMA 1990; 263(20):2753–2759. 6. Stein PD, Fowler SE, Goodman LR, et al. Multidetector computed tomography for acute pulmonary embolism. NEJM 2006; 354(22):2317–2327. 7. Weiss CR, Scatarige JC, Diette GB, et al. CT pulmonary angiography is the first-line imaging test for acute pulmonary embolism: a survey of US clinicians. Academic Radiology 2006; 13:434–446. 8. Russo V, Piva T, Lovato L, et al. Multidetector CT: a new gold standard in the diagnosis of pulmonary embolism? State of the art and diagnostic algorithms. Radiol Med 2005; 109(1-2):49–61. 9. Calder KK, Herbert M, Henderson SO. The mortality of untreated pulmonary embolism in the emergency department patients. Ann Emerg Med 2005; 45:302–310. 10. Wiener RS, Schwartz LM, Woloshin S. Time trends in pulmonary embolism in the United States: evidence of overdiagnosis. Arch Intern Med 2011; 171:831–837.
THE ROAD TO OVERDIAGNOSIS
11. Stein PD, Henry JW. Prevalence of acute pulmonary embolism in central and subsegmental pulmonary arteries and relation to probability interpretation of ventilation/perfusion lung scans. Chest 1997; 111: 1246–1248. 12. 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. J Thromb Haemost 2010; 8:1716–1722. 13. Donato AA, Khoche S, Santora J, et al. Clinical outcomes in patients with isolated subsegmental pulmonary emboli diagnosed by multidetector CT pulmonary angiography. Thromb Res 2010; 126:e266–e270. 14. Goldstein NM, Kollef MH, Ward S, et al. The impact of the introduction of a rapid D-dimer assay on the diagnostic evaluation of suspected pulmonary embolism. Arch Intern Med 2001; 161:567–571. 15. Escarce JJ, Chen W, Schwartz JS. Falling cholecystectomy thresholds since the introduction of laparoscopic cholecystectomy. JAMA 1995; 273(20):1581–1585.
3
The Road to Overdiagnosis: The Case of Subsegmental Pulmonary Embolism Saurabh Jha, MD Key Words: Overdiagnosis; pulmonary embolism; subsegmental pulmonary embolism; PIOPED. ªAUR, 2015
P
ulmonary embolism (PE) is a dreaded condition which nearly every physician, in practice or training, is aware about. It is a leading and avoidable cause of sudden death (1). That PE is now overdiagnosed (2), merits reflection. PE’s journey from an acute emergency which physicians feared missing, to a frequently overdiagnosed entity leading to overtreatment, is instructive at multiple levels, showcasing the inevitability of overdiagnosis and exposing barriers to its reduction. Embolism to the pulmonary arteries, if large, can lead to right heart strain, cardiovascular collapse, and sudden death (3). PE is in the differential for pulseless electrical activity, a type of cardiac arrest. PE, more often, presents less dramatically and such presentations can be so nonspecific that a range of symptoms is attributable to PE. Emboli that do not cause cardiovascular collapse at their first presentation are not important for their own sake but because they herald a much larger clot burden in the future. This means that the significance of PE lies in its recurrence. That the risk of recurrence can be reduced by anticoagulation makes recurrent PE an avoidable harm. The suspicion for PE induces uncertainty exacerbated by the fact that death from recurrent PE is potentially avoidable. The imperative to diagnose PE is substantial, particularly in the young patient in the emergency department. With such a high degree of clinical uncertainty, responsibility for diagnosis is shared by imagers. The gold standard for the diagnosis of PE was once catheter angiogram (4), at which thrombus in the pulmonary arteries is straightforward to detect, as it appears as a filling defect. However, catheter angiography is invasive and operationally difficult to perform in everyone suspected of PE. The first test for the diagnosis of PE used to be ventilation-perfusion (VQ) scan, in which PE was inferred, rather than anatomically depicted. The efficacy of VQ scans in the diagnosis of PE was shown in the Prospective Investigation of Pulmonary Embolism Acad Radiol 2015; -:1–3 From the Department of Radiology, University of Pennsylvania, 3400 Spruce St, Philadelphia, PA 19104. Received May 24, 2015; accepted May 25, 2015. Address correspondence to: S.J. e-mail: [email protected]. edu, @RogueRad ªAUR, 2015 http://dx.doi.org/10.1016/j.acra.2015.05.001
Diagnosis (PIOPED) (5). Reporting of VQ scans is an exemplar of standardization. The imager grades the likelihood of PE as ‘‘high,’’ ‘‘intermediate,’’ and ‘‘low’’ probability based on objective criteria drawn from PIOPED. The imager then merges the findings on VQ with the prior probability of PE to give a context-dependent interpretation. Paradoxically, and despite reporting of VQ scans being evidence-based, VQ increases uncertainty. Why the uncertainty is increased is because the ‘‘intermediate’’ category includes a wide bandwidth of probabilities and is a frequent finding. In PIOPED, nearly 40% of patients with suspected PE had an ‘‘intermediate’’ probability VQ scan. Of those with intermediate probability VQ, only one-third had PE (5). Furthermore, by contextualizing the interpretation with pretest probability of PE, the imager explicitly involves the referring physician in the interpretation. For example, if the VQ scan is of low probability for PE, the chances of PE remain as high as 12% if the patient is of a high pretest probability for PE (5). When clinicians request a diagnostic test for PE they prefer clinical closure with a dichotomous answer; affirmative or negative–is there a pulmonary embolus or not, yes or no? VQ scans do not give a dichotomous answer, but multidetector computed tomography (MDCT) can. PIOPED 2 established MDCT’s role in the detection of PE (6). MDCT is the preferred test for PE (7) and the de facto gold standard for the diagnosis of PE (8). With improving spatial resolution of MDCT, imagers see smaller arteries, including subsegmental pulmonary arteries, and it is here that MDCT has expanded the definition of disease, increasing the reservoir for inevitable overdiagnosis. PE is clot in the pulmonary arteries. However, a saddle embolus is different, in clinical severity, to an isolated subsegmental pulmonary embolus, although both fulfill the definition of PE. The significance of PE lies, however, not just in its occurrence but its potential for recurrence. This means that to declare that a subsegmental pulmonary embolus is not significant one must define what an acceptable risk of recurrence of PE is and show that subsegmental emboli do not meet that risk threshold. It is easier to show the significance of a condition than it is to prove its insignificance. This is an epistemological problem. Physicians are confronted with strong evidence of the 1
JHA
significance of large pulmonary emboli, which drives the testing (9), and in the process, they discover small pulmonary emboli. The imperative to not miss large PE is stronger than the imperative to not treat small PE. Although clinicians do not seek subsegmental pulmonary emboli, they are not confident enough to ignore them. The evidence for overdiagnosis of PE from MDCT is indirect, but compelling. The incidence of PE increased by 80% from 62.1 per 100,000 to 112.3 per 100,000 between 1998 and 2006, after the introduction of MDCT and despite prophylaxis for thromboembolic disease (10). The increased detection of PE has not reduced the age-adjusted mortality from PE, by a comparative degree, which reduced from 12.3 only to 11.9 per 100,000 during the same period. Stated differently, we are seeing more, treating more but not improving outcomes to the same extent. This points to redundancy and overdiagnosis. Of note, MDCT finds more subsegmental pulmonary emboli than VQ scans–1% of VQ scans (11) rated high probability show a subsegmental PE, whereas of the MDCT positive for PE, 15% have subsegmental PE (12). The clarity with which MDCT depicts small pulmonary arteries is not constant, as quality varies due to physiological and technical factors affecting contrast delivery. However, MDCT has established a floor of expectation that subsegmental pulmonary arteries will be visualized. This has led many imagers to hedge when they cannot see the subsegmental arteries by saying that isolated embolus to subsegmental pulmonary arteries cannot be excluded. The expansion of the disease to smaller arteries has pushed interpretation toward a higher sensitivity for the detection of subsegmental PE, at the expense of specificity, increasing false positives. Given that MDCT is the arbiter of PE, it is not possible to distinguish between a false positive diagnosis and an overdiagnosis of subsegmental PE. Such a distinction may not be important as the futility and risks of treatment apply both to the false positive and the overdiagnosed. The harm from overdiagnosis of PE is from anticoagulation, which has risks. In one series of 93 patients with subsegmental PE, the risk of bleeding was 5.3% and the risk of recurrence was 0.7% (13). Projecting these findings to a cohort of 1000 patients, 53 patients will have major bleeding to save seven patients from recurrent PE. Or nearly eight patients with subsegmental PE must be harmed to save one person from recurrent PE. Is this an acceptable trade-off? The overdiagnosis of PE is exacerbated by the use of the Ddimer test which, instead of gatekeeping MDCT, is a gateway to overdiagnosis (14). The threshold for positivity of a Ddimer is very low. This is deliberate to give D-dimer a high negative predictive value, at the expense of specificity, so that when the test is negative, the clinician is confident that the patient does not have a PE. However, the availability of the D-dimer has increased its use leading to a growing reservoir of declining pretest probability of PE. This phenomenon is known as indication drift, and it has been demonstrated in other settings (15). 2
Academic Radiology, Vol -, No -, - 2015
The overdiagnosis of PE is a result of our intolerance of uncertainty, the want of dichotomous answers, and an emergence of technology. To mitigate overdiagnosis, specialty societies must state in no uncertain terms that the diagnosis, and treatment, of subsegmental pulmonary emboli is harmful and should be avoided. The management of PE is at the crossroads of multiple disciplines including emergency medicine, radiology, and hematology, and this begs the question which specialty will be tasked for issuing such a statement. Will emergency physicians, radiologists, and hematologists agree with each another? Will there be agreement between physicians of the same specialty? Can radiologists even agree that they must stop hedging that subsegmental pulmonary emboli cannot be excluded? The difficulty in ascertaining the insignificance of a condition is best illustrated in an article in the British Medical Journal about the overdiagnosis of PE (2). The authors, experts in overdiagnosis, argue convincingly that there is overdiagnosis of PE. However, they then add disclaimers which are quoted verbatim. ‘‘Definitive evidence of overdiagnosis would, of course, be the finding that untreated patients never experience harm from the PE during the rest of their lives and died from another cause, but no such studies exist’’ and ‘‘we need to learn which small emboli need treatment.’’ It is unlikely that these disclaimers will reassure physicians in the frontline that they can ignore subsegmental PE, even in patients who do not have malignancy or a prothrombotic state. As in other conditions, overdiagnosis of subsegmental PE is an information problem. To recap the triad of the information problem, many patients will not benefit from treatment of subsegmental PE, but some will, and we do not know who will or will not with certainty, ex ante. We have implicitly stated our preference, though, which is that it is better we over treat many patients with subsegmental pulmonary emboli than we under treat some. Until this calculus changes, overdiagnosis and overtreatment of subsegmental PE will continue. One way to reduce overdiagnosis of PE is to raise the bar for its suspicion through evidence and actionable thresholds. It is unlikely this can be achieved without a significant change in medical culture. REFERENCES 1. National Heart, Lung and Blood Institute. What is pulmonary embolism? http://www.nhlbi.nih.gov/health/health-topics/topics/pe (accessed May 24, 2015) 2. 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. 3. Handler JA, Feied CF. Acute pulmonary embolism. Aggressive therapy with anticoagulants and thrombolytics. Postgrad Med 1995; 97:61–62. 65-8, 71-2. 4. Sasahara AA, Stein M, Simon M, et al. Pulmonary angiography in the diagnosis of thromboembolic disease. NEJM 1964; 270:1075–1081.
Academic Radiology, Vol -, No -, - 2015
5. The Pioped Investigators. Value of the ventilation/perfusion scan in acute pulmonary embolus. The results of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED). JAMA 1990; 263(20):2753–2759. 6. Stein PD, Fowler SE, Goodman LR, et al. Multidetector computed tomography for acute pulmonary embolism. NEJM 2006; 354(22):2317–2327. 7. Weiss CR, Scatarige JC, Diette GB, et al. CT pulmonary angiography is the first-line imaging test for acute pulmonary embolism: a survey of US clinicians. Academic Radiology 2006; 13:434–446. 8. Russo V, Piva T, Lovato L, et al. Multidetector CT: a new gold standard in the diagnosis of pulmonary embolism? State of the art and diagnostic algorithms. Radiol Med 2005; 109(1-2):49–61. 9. Calder KK, Herbert M, Henderson SO. The mortality of untreated pulmonary embolism in the emergency department patients. Ann Emerg Med 2005; 45:302–310. 10. Wiener RS, Schwartz LM, Woloshin S. Time trends in pulmonary embolism in the United States: evidence of overdiagnosis. Arch Intern Med 2011; 171:831–837.
THE ROAD TO OVERDIAGNOSIS
11. Stein PD, Henry JW. Prevalence of acute pulmonary embolism in central and subsegmental pulmonary arteries and relation to probability interpretation of ventilation/perfusion lung scans. Chest 1997; 111: 1246–1248. 12. 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. J Thromb Haemost 2010; 8:1716–1722. 13. Donato AA, Khoche S, Santora J, et al. Clinical outcomes in patients with isolated subsegmental pulmonary emboli diagnosed by multidetector CT pulmonary angiography. Thromb Res 2010; 126:e266–e270. 14. Goldstein NM, Kollef MH, Ward S, et al. The impact of the introduction of a rapid D-dimer assay on the diagnostic evaluation of suspected pulmonary embolism. Arch Intern Med 2001; 161:567–571. 15. Escarce JJ, Chen W, Schwartz JS. Falling cholecystectomy thresholds since the introduction of laparoscopic cholecystectomy. JAMA 1995; 273(20):1581–1585.
3