Issue 3 / 4 February 2013

A SHARP rise in the number of referrals for pulmonary embolism-related imaging but no fall in mortality from the condition have called into question testing protocols, say WA researchers.

A retrospective audit of imaging trends in four WA teaching hospitals, published in the MJA, found a 34% increase in referrals for venous thromboembolism (VTE)-related imaging and a 64% increase in PE-related imaging during the 9-year study period. (1)

While WA hospital admissions for PE increased by 45% over the same period, the death rate from PE did not fall (P = 0.19), according to the audit. The increase in PE-related testing was driven entirely by increased referrals for computed tomography pulmonary angiography, which increased fivefold over the study period.

The research was prompted by widespread concern that a standard assessment used in WA teaching hospitals — the modified Wells score coupled with a D-dimer assay — was resulting in more, not fewer, referrals for imaging.

“There was no reduction in PE mortality in the state as a result [of increased testing], and the increase in hospital separations we have observed is likely to be explained by increased detection and overtreatment of clinically insignificant PE. These results call into question the wisdom of a ‘rule out’ strategy for PE, particularly among patients presenting to hospital emergency departments with non-specific symptoms, and indicate the need to re-evaluate the way in which D-dimer testing is used to screen patients for VTE in our hospitals”, the study authors concluded.

Dr Sally McCarthy, medical director of the Emergency Care Institute of NSW, said the study raised several critical issues, including the inability of protocols and guidelines to keep pace with the increasing sensitivity of testing.

“If new or more sensitive tests become available, we continue to manage the testing the same way, even though we’re in a dynamic environment”, she said.

Dr McCarthy said “shotgun testing” for PE was concerning. “It wastes not only the time of the person administering the test and the time of the laboratory in checking for it, but it … causes the patient to stay in emergency for longer”, she said. “That’s poor quality care and inefficient use of resources.”

Vascular surgeon Professor Ken Myers said the study indicated that a lot of taxpayer’s money was “going down the drain with no visible benefit”.

However, he said the trend was driven partly by doctors, who were anxious about missing a PE, and partly by patients, who were hesitant to rely only on a physician’s clinical judgement and demanded the latest technology to have the condition ruled in or out.

Professor Myers said this was understandable. “If I was the patient with the suspicion of a pulmonary embolism I would be down to have my pulmonary CT immediately.”

Dr McCarthy said patient demand was only a minor factor in the increased use of imaging.

“It’s driven much more by a misunderstanding of the complex assessment that has to occur on using a test appropriately.” She said guidelines and protocols were often developed by interest groups without appropriate consideration of the flow-on effects, such as the impact of unnecessary interventions or the opportunity cost to the system or other patients.

Dr McCarthy called for a more holistic assessment of the impact of testing protocols on both the patient and the system.

She said implementing “sensible test-ordering protocols” in emergency departments had been shown to improve efficiency and quality of care.

“There’s an over focus on pushing for diagnosis of everything and a lack of focus on the cost of poor quality testing.”

Professor Myers said the paper was an important reminder to clinicians to “think hard” before ordering expensive investigations.

However, he doubted it would have any impact on clinical practice.  “I don’t think that anyone is going to be game enough to put into practice what [this paper is] quite properly advocating, which is to cut back on the investigations and use your clinical judgement … to [better] select your patients”, he said.

“That’s what we were all taught in medical school, but in this technology-driven age, it’s taken a bit of a back seat.”

– Nicole MacKee

1. MJA 2013; 198: 100-103

Posted 4 February 2012

Photo courtesy of James Heilman MD through Wikipedia.

6 thoughts on “Huge jump in PE imaging

  1. Gabriel Blecher says:

    More and more pulmonary emboli are being diagnosed that are probably benign and do not warrant treatment. This is not the same disease PE was in the 1950s and 1960s when the only emboli diagnosed were those with haemodynamic significance, not the subsegmental ones we diagnose today. These are associated with minimal mortality.

  2. Malvinder Parmar says:

    There are various issues resulting in the unnecessary use of imaging studies, not only for diagnosis of PE but other conditions as well, and some issues are as follows:
    1. Use of protocolized medicine – often removes whatever little thinking clinician need to use.
    2. Often D-dimer is requested by ER physicians, as a bundled test, for majority of patients and if the D-dimer is elevated, where there could be other reason for d-dimer elevation, the casulty officer orders CT to rule out PE, even when clinical likelihood is low.
    3. One of the clinician above says that “more sensitive tests” would be more useful. I should mention that it is incorrect, as sensitivity trumps specificity and highly sensitive tests often results in multiple other tests to rule out the disease. This is also seen in myocardial infarction as well, becasue of mild elevation of troponin, and often patients are told they had “heart attack” when in fact slight elevations are common with other multiple conditions and since the use of troponin, there is 30% or higher increase in diagnosis of myocardial infarctions when in reality that is not the case and often patients end up with further tests, including angiograms, to rule out disease.

    So we don’t need more sensitive tests but what we need are more specific tests.

  3. Sue says:

    Thanks for this important discussion. What we are doing, with increasing risk aversion and protocolisation, is ignoring Bayes theorem: if you apply a test to a low probability population, you will get lots of false positives. Since a false positive PE means at least six months of anticoagulation, this is no small thing. The most important test for PE is pre-test probability – based on history and examination. We also need to get away from the incident reporting that sees a small missed PE as a dangerous event – perhaps treating it is more dangerous.

  4. Etranger says:

    Is anyone really that surprised that we’re doing a lot more CTPA studies? They’ve become the next test done in ED after ECG/cardiac enzymes/D-dimers for virtually anyone presenting with chest pain, SOB or a whole host of non-specific complaints.

    What we don’t know (yet) is just what clinical significance all these small sub-segmental PE’s actually have. Do they really cause any long term morbidity or mortality or are they just a manifestation of a process that we’ve only recently become aware of with our enhanced ability to diagnosis small emboli?

    We also tend to forget the risks of anti-coagulation when we treat these cases. That said, it would be a game clinician who didn’t treat a known Pulmonary Embolism!

  5. Tim Bailey says:

    A very topical subject, thankyou for the opportunity to contribute. There is a recent article by Kline et al (within the past 12 months, NEJM I think) which reviews a large amount of the world literature on PTE, it’s diagnosis, treatment, mortality and also the mortality and morbidity associated with diagnosis (radiation and IV contrast) and treatment (anticoagulation).
    Without reproducing the article (which is beautifully written and analysed and I would recommend as important reading), in summary the findings were as follows.
    Taking the most conservative analysis of the data analyzed out of most of the current world literature on the subject. It would appear that with current practice, we are probably KILLING (either now or in future years) about FIVE TIMES MORE people than we are saving.
    What can we so to change this? 1. Establish clinical trials (multi-centre DBRCTs) to examine the problem. 2. Probably avoid any specific tests for PTE unless thee is some sort of significant physiological disturbance associated with the presentation (eg. Tachpnoea, hypoxia, associated marked tachycardia with elevation of central venous pressure).

  6. david mountain says:

    Talk about jumping to conclusions from a study with serious methodological flaws but which pushes many clinicians entrenched unhappiness with PE diagnosis.
    Even a brief look at the data shows that it does not support the underlying hypothesis (that D-Dimer testing has driven CTPA testing and that all the “new” PEs diagnosed are clinically insignificant).
    D-Dimer testing increased for 3 years to its maximum level from 2002-2005, 2005 was the first year any signficant increase in CTPAs/ PE testing occurred (probably driven more by Radiologists spruiking of the great new test that gives great pictures/detail etc). It is implausible to sustain a hypothesis that DD testing has a 3 year lag until it increases testing in a specific test CTPA when both VQ/CTPA were available before.

    The mortality rate did not change during the almost 10 years this study has data for. This statistical data rather than support that clinically unimportant PEs were being diagnosed suggests the opposite. PE is rarely put on a death certificate unless diagnosed pre-mortem. Therefore if all the new PEs were inconsequential you would expect the perceived mortality rate to have dropped very signficiantly – no one dies from inconsequential, unimportant PE. Another hypothesis is that D-Dimer testing increases improve the yield for follow up testing and that increased CTPA testing has picked up signficant numbers of additional and relevant PEs which when treated have a low mortality rate.

    This study is just a hypothesis generator and those suggesting it offers good evidence for major changes in practice are just using any old information to bolster already fixed prejudices and biases. It proves nothing except that if you look at data with a bias you see what you want to see!

Leave a Reply

Your email address will not be published.