You don’t need me to tell you that VTE is a huge problem. 10 million cases worldwide, a cost to the NHS of £466,000,000 and half a million European deaths every year spell it out. Every disease that has its own day (October 13th) must be pretty bad.
Whose problem is it? Well, some countries have decided that such an issue deserves its own dedicated experts. In the UK no single specialty owns the disease, but the burden of diagnostics is mostly left to emergency and acute medicine. Some centres may set themselves up to stream suspected disease to observational medicine clinics. Job done, no longer our problem. However, if you don’t think this is an EM problem let me share some of our regional data with you from Virchester. Imagine us as an ED seeing roughly 100,000 patients every year:
- Suspected VTE accounts for around 2-3% of our annual attendance
- We order around 1798 imaging investigations every year – a third being CTPA or V/Q scans
- We request just over 3500 D-dimer tests every year
There is also the issue of people returning with complex issues, including ongoing symptoms, bleeding, recurrence and end stage disease (such as PE with shock). These patients will always come to the out of hours provider. So like I said – a big problem. And at least partly our problem.
Firstly, let’s ask what we are already doing about it this problem. Well, we know from previous research that individual clinical signs and symptoms are unreliable for the diagnosis. However, this research has got slightly lost in translation to become the message that we should rely heavily on gold standard imaging. Put that alongside rising awareness of the condition in general, you get some serious over investigation. This is represented in the literature; pre-test probability for DVT in studies from the last 20 years has halved from 16.7% to 8.3%. For PE it is even worse, with recent estimates of 5% pre-test probability in some ED cohorts, down from 30% in the 90’s. For every 20 patients imaged for PE in many EDs, only 1 will test positive. That’s a lot of testing.
Is this approach doing any good? Well, that’s quite hard to prove. If you look at data from the turn of the century, you will see we are diagnosing far more pulmonary emboli but the attributable mortality rate remains static. What about reduction in CTEPHTN, post thrombotic syndrome and other VTE related morbidity I hear you cry? Hard to prove demonstrable benefit.
Is this approach doing any harm? Well, the answer to that is almost certainly yes. Firstly, we know anticoagulation has an attributable fatality rate of approx. 0.5-1% and a major bleeding rate of 2-3%. Without doubt we are anticoagulating more people as more VTE is diagnosed, primarily as imaging becomes more accessible and of higher quality. Secondly, our ‘gold standard’ tests are not flawless; patients we are placing on anticoagulation may not actually have a PE or DVT. Thirdly, CT comes with associated risks of radiation exposure and contrast induced nephropathy. Lastly, CT can often detect incidental nodules that necessitate further investigation and procedural assessment, with associated complex risk.
How can we redress the balance in suspected VTE? Should we be doing less for our patients? Could less be more?
Let’s think about this. How can we do less, remain safe and actually provide better overall care (do more) for our patients? To do so, we need to consider 3 things:
1 – You have a big brain and you should use it:
These are complex patients and not simple pathways. They require thorough evaluation to ascertain whether you think the risk of VTE is genuine and to develop an estimate of risk, via gestalt or clinical decision rule. The nuances of disease must also be considered – is this a potential recurrence, would it be provoked disease, is it at an unusual site? Then comorbidities; are they at risk of bleeding as well as clotting? All these factors need to be considered during your work up and treatment plan
2 – Understand and use the tools at your disposal:
There are many evidence based tools to help you work up a patient with suspected VTE and potentially avoid invasive imaging or misdiagnosis. To use them properly, you must understand their development, their compatibility with your local resources and how they perform within your local population.
3 – Share (or patient centre) your decision making and use predictive analytics where possible and reliable.
These are all cases where there are risks and benefits to be discussed. Firstly, tailor these risks to the individual in front of you. What is the likelihood of propagation/recurrence/embolization of this particular clot in this particular person? What is their bleeding risk if you were to find a clot? What of the available options would they prefer? It is up to you to translate the complexities of this into accessible information. Good job about that big brain of yours….
Let’s explore these 3 areas in a bit more detail.
USING YOUR BIG BRAIN
Whilst individual signs and symptoms may be of little use, an overall impression of risk or pre-test probability is essential to tailor the patient journey. There are at least 8 pre-test probability scoring systems, many of which have been externally validated to help you do this; these include the St Andre, Khan, Hamilton, Oudega, Revised Geneva, Constans and probably the most validated, the modified Wells score. These scores cannot be blindly applied – think about the area of symptoms, possible exclusion criteria and other relevant factors at the time of assessment, including provocation and recurrence.
These scores cannot exclude disease and are designed merely to develop a picture of risk. There is research suggesting you may be equally good at this – gestalt is not to be ignored and several papers would suggest the suspicion of a senior clinician to perform comparably to the best scoring systems.
Why is this important? Because for those patients with a low risk of disease, through gestalt or score, we now have a variety of tools at our disposal that may facilitate reassurance without the potential harms of invasive investigation or treatment.
USE AND UNDERSTAND THE TOOLS AT YOUR DISPOSAL
So, first out of the toolkit for low risk patients is the PERC rule – a set of criteria designed to facilitate exclusion of PE at the bedside without ancillary testing. Deemed of such use by ACEP that it has now made their top 5 ‘choosing wisely’ list for 2015. Certainly, the systematic review data would suggest a high level of performance, with pooled sensitivity of 97% across 13 cohorts and just under 15,000 patients. However, you need to remember the caveats with PERC – it is of primary benefit in populations with a low pre-test probability. With a NLR of 0.17, many suggest that the prevalence in your population needs to be at least <15% for application, maybe even as low as <7%. However, this assumes you are applying it blind. Recent studies with a higher disease prevalence (20%) have noted a high performance, providing an estimate of risk is taken via CDR prior to application, and only applying PERC to those deemed low risk.
Reworking the PERC data to look at those PERC negative patients who went on to eventually be diagnosed with PE, suggests pregnancy, post partum status and pleuritic chest pain to be the main confounders. Keep this in mind and use the rule with caution. But if your local pre-test probability is <7% or you have assessed the patient and suspect them to be low risk, PERC can help you avoid the diagnostic merry go round.
What about the over 50’s or those others who are PERC positive? Well, these patients will get a standard work up for VTE no doubt, with a validated assessment of risk followed by a D-dimer test for those deemed ‘unlikely’ to have the disease. Using the D-dimer appropriately unfortunately necessitates a bit of understanding. It is a continuous variable and as such, most studies have created a cut point for a positive/negative test based on the overall performance and diagnostic test characteristics. The standard is <500ng/ml in most large studies. However, there are a variety of assays available with little standardisation or harmonisation between them. Also, they are reported in variable units. For instance, the standard cut off of <500 is reported as Fibrinogen Equivalent Units in ng/ml or mcg/L. A D-dimer reported in standard (rather than FEQ units) is half this value, so your shop may use a cut point of <250 or around that area. Some report in mcg/ml and others in nmol/L. Confused much?
This is important to understand as the test may actually have a wider range of uses than originally envisaged. For instance, an adjustable age related cut off has been proposed since 2010, in attempt to improve the specificity of the test while retaining a reasonable sensitivity. This idea is gaining traction now (as a cut point of age * 10 in FEQ ng/ml – thus an 80 yo patient has a cut point of 800ng/ml rather than 500ng/ml), with large studies published in 2013, 2014 and just recently (epub ahead of print) all consistently supporting the merits of age adjustment.
We looked at this in Virchester recently. This is unpublished data, but if we had applied an age adjusted strategy for the first 6 months of this year to our suspected VTE population, we would have avoided 357 CUS scans, 57 CTPAs and 12 V/Q scans. More importantly, we would have avoided prescribing >2000 doses of therapeutic dose LMWH pending scan. There are caveats with this like all methods – we would have missed a single isolated distal deep vein thrombosis (which went untreated by the clinical team) and a PE on V/Q scan in an 85 yo patient who had presented with interscapular pain for 5 weeks. Length of symptoms and previous disease are confounders for the performance of all diagnostic pathways in VTE and this should be remembered when considering the application of any diagnostic test.
But, all in for a population of 985 referred for definitive investigation, this would have given us a sensitivity of >98%, NPV of >99% and NLR of 0.04. Not bad at all. This strategy of age adjustment is easy to look at within your local cohort – how do you think you would fair? What would you define as an acceptable miss rate?
Gestational adjustment has also been floated as a recent concept. This study suggests trimester cut points are worth considering and may pave the way to a reduction in unnecessary radiation exposure for this cohort as well.
The 3 minute walk test
So, those research savvy Canadians have been at it again. This time Ian Steill and co have examined the diagnostic test characteristics of a short 3 minute walk test in suspected PE, looking at elevation of the pulse rate by 10bpm and desaturation of >2% as markers for disease. When combined, the sensitivity was perhaps surprisingly high at 100%. Specificity was dreadful as you would expect, but this feasibility study raises interesting questions about going back to simple bedside assessment techniques during initial work up with the aim of preventing over-investigation. This is an easy study to replicate for assessment of external validity – how do you think your punters would fair?
How about if we could just wield the probe at time zero and reliably exclude VTE at the bedside prior to any of this radiation or anticoagulation nonsense? Well, that’s tricky in suspected PE but in DVT many have been training themselves up to perform compression ultrasound with interesting results. What does the literature say? Well, we seem to be fairly accurate following abbreviated training/educational programmes and certainly for proximal DVT our sensitivities are comparable to radiologists. However, the caveat of whole leg scanning and isolated femoral vein thrombus remains. In addition there is complex resource use here, and asking a senior physician to perform a dedicated lower limb CUS during a busy Monday morning in the ED is not without its challenges.
However, at the very least this information may add to your assessment of likelihood – can’t see a large clot? Maybe they do need that formal CUS in 3 days time but perhaps they could get away without heparin cover? Something to think about.
I want you to get the hell out
Say you go down a standard diagnostic pathway, your patient gets imaged and they have a VTE. Less is more doesn’t stop here. Ambulatory care and rapid ED discharge is becoming more routine for both DVT and PE as a direct result of its financial advantages, reduced in hospital risk and convenience. Dose reductions are worth considering as well – recent evidence suggests that prophylactic dose anticoagulation may enough for angry superficial thrombophlebitis and even in some distal disease. In addition, there are perhaps some patients with isolated subsegmental PE or isolated distal DVT that may not need any treatment at all – when risk of bleeding is high and disease burden is small, it is becoming increasingly easy with support of the evidence to argue for surveillance and follow up over the risks of anticoagulation. How is your department set up to discuss the options? Could you do better for these patients?
Share (or patient centre) your decision-making
As you can see, this is not always clear-cut. Where the evidence is not definitive, patients have a right to be informed of the options along with your suggestions. But this needs to be handled in a way they can understand and also in a way that is specifically tailored to them. Predictive analytics can help here and many have been working on individual computerised risk stratification tools, which can aid in tailored discussions of risk. At a personal level though, this discussion needs to be patient centred and open – what are the risks of testing and not testing. What are the risks of treatment versus conservative management? What are the options other than the current fairly narrow diagnostic corridor we offer most people? Other specialties can help here as well, such that a shared decision can be just that. Shared.
What to believe and when to change
Much of this work is relatively novel and untried at a local level. You will also find no mention of age adjusted d-dimer, 3 minute walk tests or EP ultrasound in the UK NICE guidelines for VTE. A British Thoracic Society guideline for ambulatory care of PE is coming in 2016; that is worth looking out for and the views of UK emergency physicians have hopefully been adequately represented.
There is always a need to question new work and examine its applicability at a local level. However, these options described above have the potential to save resource while improving patient outcomes. As such it is key that we at least seriously examine their merits. If you are not an ultra early adopter, then think about how you can gain more evidence locally prior to changing your pathways. Involve local stakeholders and see what they think. Ask your lab about age adjustment, ask your sonographers about training you up to scan. What to believe and when to change are always tricky concepts and a lecture in itself. But we can all start by thinking about change and what degree of evidence we would need to see before we alter our practice.
Putting it all together
If you do decide you like the sound of all these techniques and you have been bemoaning our over investigation rates for decades, then there are papers available to guide you. Jeff Kline has published his suggested algorithm recently, which incorporates most of the above and is free to access. The accompanying article is an excellent contemporary review of how we can do less, and how less might be more with regard to PE specifically.
My only caution here would be the complexity of the algorithm. A spider diagram like that can be very useful but can also be easy to ignore/complete incorrectly if you don’t have the accompanying education and understanding. Whilst senior sign off may be mandatory throughout North America, the UK is still playing catch up here and our departments remain primarily staffed by trainees and nurse practitioners. They do a great job, but it remains challenging.
Reducing the complexity of VTE to an algorithm also runs the risk of taking the thought out. I wonder if a more sensible approach is to raise awareness of the options available through the three points previously discussed. If you have a patient in front of you with suspected VTE tomorrow, consider the following:
- I have a big brain and I need to think about tailoring my decision making to the specifics of this complex case
- What ancillary diagnostic adjuncts are available and applicable to this patient in front of me, which may help reduce the risks associated with unnecessary diagnostic testing and/or therapy?
- What does my patient want, need and understand about the risks and benefits of this process?
Less is more. Make yourself a promise. Less TV more exercise. Less work more play. Less bakeoff more baking.
Less random testing; more focus on patient specific outcomes.
If you’re still reading, then thanks for sticking with me.
7 thoughts on “VTE masterclass with Dan Horner at #RCEM15”
Thanks for posting as I was in another room. Sounds like a great talk!
What about nasal capnography in ruling out PE: in this paper a cut-off of 4.3kPa had 100% sensitivity. Admittedly v different population (pts referred to medicine with suspected PE, 38% prevalence!) but worth an ED study? http://www.ncbi.nlm.nih.gov/pubmed/24715122
DOI: I talk about this paper in RCFN http://www.rcemfoamed.co.uk/portfolio/end-tidal-co2/
It’s an interesting concept. Kerstin is the person to ask about dead space and utility in PE diagnostics – she published on this in chest in 2005 and is around for the whole conference I think.
I think with an ICU hat on I would have queries about the reliability of nasal capnography and would need to see a decent prospective attempt at ED external validation, as you say.
All part of the puzzle though. And if your gestalt is low then additional tests such as these are all useful to further reduce your PTP in attempt to reduce unnecessary invasive diagnostic testing.
Cheers for posting
Yes, end tidal CO2 is a very interesting adjunct in the process of diagnosing PE. You might consider it in the unstable patient with possible high risk PE, as a pointer to the disease. As yet, no study has proven it has either high sensitivity or specificty. Part of the reason is that patients we work up for PE often have another lung pathology such as pneumonia and underlying COPD, which will affect the end tidal CO2. I looked at it in two prospective studies and did not find it discriminated well in patients with other co-morbidities. It seems to do better in young, healthy patients.
Great work. I love these summaries of talks.
Agree with Andy, particularly if you have not been able to get to the presentation
Pingback: Conference: VTE Dublin - Emergency Medicine Ireland
Pingback: Level Pegging? JC and the PEGeD study @StEmlyns • St Emlyn's