post

Clinical Case 090: Unstuffing the Turkey

You are at work in a busy little ED.  It’s just you and a bunch of junior MOs on the floor.  You are seeing all the patients that you can and handing out advice as you go.  Just after lunch one of your JMOs catches your eye…  “Can I run this patient past you boss?”

Mrs Clot is a 49 y.o. tourist in Broome.  She flew in from Sydney via Perth (4 hr flight, 1 hour lay over, then a 2.5 hr flight) about 3 days ago.

She has been doing all the usual things tourists do – laying on the beach, riding camels at sunset, drinking cocktails and eating too much.

She has also met the local mosquitoes [as is the norm] and has nasty, itchy, red looking ankles – the left maybe a bit more oedematous than the right.

This morning she experienced an episode of chest pain, it was moderately-severe, retrosternal, sharp and eased when she took some paracetamol and relaxed.

Her PMHx:  hypertension [untreated], hysterectomy 3 years ago for menorrhagia, no current medications

So – sounds pretty standard.  You review the ECG which is plum normal.  A CXR is also “unhelpful” – that is, shows no obvious cause for the pain.

The plan is to check a highly sensitive Troponin (6 hours post-pain) and then discuss the disposition if it is normal.

“I am a bit worried about the plane flight…” he says “Doesn’t that make a PE more likely?”

Well there is surprisingly not a lot of really high quality evidence. Lots of trials comparing TEDS to no TEDS, various anticoagulants etc. But the baseline risk is extremely low – so really need a massive trial to detect any real difference in PE or PE-mortality rates.. A few studies have shown modest relative risks around the 1.12 – 1.6 risk ratio level – but this is still a rare event – so from a handful per million flights to 2 handfuls in a million travellers? It does appear that risk factors are cumulative – e.g.. being obese, on hormones, having cancer, previous VTE etc will all add up to make the risk significant. Most of the trials look at “long-haul” – ie > 6 hours in a plane. And some would suggest a dose-risk curve – longer flight more risk. So hard to say what this means to domestic flights even in big countries like Australia

You notice that your JMO is looking a little worried as he checks the lab results….

You ask – “was the troponin up?”

“No.  The trop was negative. But…..   I added a D-dimer, just to be sure.. and…  it was 0.49”    [our normal range is < 0.36]

“I was just going to ignore it if it was positive – ‘cos it isn’t really a useful test then is it…?”

You decide to take a closer look at the patient.  She is as advertised – normal aside from a few midgy bites on her legs and maybe some early cellulitis as a result.

You check her PERC score – she has 0/8 criteria

Her Well’s score is firmly in the “low-risk” group.

The original Well’s paper would put it at ~ 1.3 % The PERC rule should get her down to 1.8 – 2 % The Geneva score was less discriminatory – it just states low as < 8 % risk of PE

Well it depends on the manufacturer and the “cut off value you use.  BUt often used numbers are: Sensitivity = 98%,   Specificity  =  38 %,    + LR =  1.58 ;    – LR =  0.05  

So we need to work out our TEST THRESHOLD ( the post-test probability below which we would be doing more harm than good if we were to pursue the diagnosis with further testing.

Test thresholds vary from disease to disease and can be tough to work out.  They are based upon the prevalence of the disease, the pretest probability, the predicted rate of false positives and the potential harm from treatment.  Pines et al (Acad Emerg Med. 2012 Sep;) calculated  the “mortality benefit threshold” for various risk-levels.  That is the pretest probability you would need to have in order to make embarking upon a PE work-up worthwhile […i.e. when is PE more likely to kill the patient than all the potential downsides of diagnostic interventions and therapy.]

So what does this mean for our traveller?

Pines and colleagues would say that our “mortality threshold” would need to be greater than 3.7 %.

So a bit of maths here – if her Well’s and PERC put her at (lets be generous) 2 % risk – worst case scenario.  We can choose to ignore the D-dimer… or we can use the known +LR of a D-dimer and plug it into the formula.

Pretest probability  X  LR+ = Post-test prob

So for our patient – roughly… 2 %  X  1.58  =  3.16 %

This is a beautiful thing.  We have not ignored the D-dimer.  We have accepted a positive and applied it to our pre-test risk and guess what – we are still below the “test threshold”.  She still has a risk which would make it more harmful to continue to a CTPA.

In fact – doing a CTPA on this patient would more likely result in a false positive result than a true positive if the radiologist reported it as showing a PE!

Turkey is unstuffed.

As Dr Anand Senthi stated in his wonderful ACEM WS lecture – always prognosticate before you investigate!

Now I am reasonably ceertian that I hve made at least 1 or 2 errors in the logic here – and this is the beauty of #FOAMed.  I would love to have your input and corrections to allow me and the other readers to learn.

Let me know

 

Casey

Comments

  1. thanks Casey
    Did your JMO miss your orientation talk on PERC/PE workup?

    My one suggestion is that it seems a bit overcomplicated, this approach of calculation of post test prob and comparing to mortality threshold

    My view is if clinically low risk for PE , backed up by Wells score even and PERC rule out criteria met, then you are done. Ignore the D Dimer.
    this paper provides some comfort for this
    http://www.madisonemergencyphysicians.com/userfiles/clinical/2012-09article2.pdf

    doing the extra calculations may help you feel more confident but in my view are unnecessary given the evidence to support the approach of clinical rule outs.

    Also “we have not ignored the D -Dimer” ??? is that really true.
    The post test prob calculation can be done based on maths..you dont need to know the D Dimer result to do that( i.e dont need to order it before you can calculate your post test prob)
    ..so in fact you HAVE ignored the D DImer..regardless of what it showed!

    and another thing…even if you accidentally order a D Dimer and it comes back as highly positive..like 10 times the normal range upper limit….are you going to just ignore it? I wouldnt!
    bugger eh?
    i have had an actual case where low risk PE pt had a positive D Dimer and then confirmed bilateral PE on CTPA. One of those 0.5% that clinical rule out like PERC will miss! One of those lucky ones where a jnr MO did order a D Dimer!..and it wasnt ignored!

    Its important to share the cases where you get burned despite EBM practice!
    So you need to counsel your patients that you are going to send home ( with their possible PE), that no clever EBM calculations are perfect and that followup and review is needed because they still have a small risk that in fact, you got it wrong and they do have a PE!

    and in fact in this case, where the Dx is more likely to be ACS, despite neg trop at 6 hrs, you would still arrange f/up and review..and I think best practice would say you would counsel pt regarding the positive D Dimer and need to review at f/up for that as well, explaining the small risk still that you have not fully eliminated.
    another thing to add further confidence would be to Ultrasound her legs for DVT in the ED and document your findings if negative in a followup letter to her GP . She needs f/up for possible ACS by GP but now that you have positive D DImer ( albeit low risk), you need to advise of small risk of PE and arrange review of that as well!

    Cause how do you know she is not that 0.5%?

    • Minh
      A while since we had a good disagreement!

      “I saw a patient once who was put on warfarin for a tiny, not haemodynamically significant PE that was picked up on a scan looking for something else. She had an intracranial bleed and died”

      Anecdotes are fine, but the beauty of the test or mortality threshold is it allows us to quantify these events in both directions.
      What we need to know is the net harm: benefit of our interventions or investigations – anecdotes usually reveal the misses, rarely the remote harms of our decisions
      C

  2. Hi Casey,
    I thoroughly approve of you using likelihood ratios. Please be careful though, since you need to multiply pre-test ODDS by LR to get post-test ODDS.
    In the case you give this gives a post test probability of 3.12 which is still below your cutoff, and your advice was therefore right and safe, but for the wrong reason!
    Kind regards
    @Noo_wethead

    • Thanks Nick. Yes odds are not the same as risk / probability.
      My math isn’t that good!
      My understanding is that they are practically equivalent when your prevalence is low eg. less than 10%

      Thanks for your mathematical wisdom
      C

  3. Keeweedoc says:

    Hi Casey,
    Out of interest do you do a quick DVT scan in these patients?

    Had a D-Dimer negative PE here this week, Shouldnt have had a D-dimer in first place. The repeat test was positive (4 hours later added for academic interest)

    Keeweedoc

    • Hi Andre
      Of course I do a DVT scan – any opportunity to use my toys! Of course it is tough to stratify the relative risk reduction for these patients.
      I think the best scenario to use a CUS in a ? PE patient (one with chest symptoms) is when they have some leg signs – oedema, swelling, infection which is clouding the picture. If you can confirm cellulitis or exclude a proximal DVT then it must drop he risk. I am not sure if you can then put that into the Wells score as “no signs of DVT” – as this was a clinical call in the trials, not a sonographic call.
      It does help reduce the inherent bias in seeing a slightly dodgy leg in a chest pain work up
      C

  4. Thanks for the interesting case, Casey! What a beautiful, if slightly baffling, clinical decision process.

    Do your JMOs really call you “boss” though??

  5. nothing wrong with anecdote if its actually the truth!
    Andre just posted an anecdote of that 0.5% risk of DDimer neg PE!

    i would be interested to know what would have been the mortality threshold and pre/post test prob results for Andres case? please

    • Minh, I think we need to make the debate clear!

      In simple terms – if you exhaustively test every patient with chest pain for a PE you will detect a lot of PEs, many of which would be harmless “lung lint” which would have no impact on the life of the person.
      However, if you set your “threshold for testing” too high then you will miss some clinically important PEs and your patient will be worse off.
      What we need to do is find the break-even point. The initial risk level below which further investigation is going to do more harm than good.
      This is a real thing – it can be calculated but is complex and there are a lot of variables – fortunately for PE there is a large body of data to make this possible.
      Andre’s patient might have been below this risk threshold – and should not have been investigated. The “PE” she was diagnosed with might have been inconsequential, but the warfarin, radiation etc is not.
      This is the nature of biostats – we are trying to predict who will benefit from investigation – one cannot retrospectively say – “Hey – you would have missed that clot!”
      None of us are so experienced that we have a good handle on the relative risks in such a complex situation – so it is best to fall back onto predictive models based on good data – rather than ” ANdre had a bad miss last week!” This is pure cognitive bias.
      This parallels the use of checklists for procedural work – allow us to cognitively unload and make rational decisions where there might be unavoidable bias…

      EG. I had a nasty resus last week and I think the Coroner will report a big PE as the COD. So now I ma going to be really paranoid and think PE is higher on the list than it is – suddenly every patients Wells is “moderate” and I am doing a whole bunch of D-dimers ahhhhh!
      Debate on!
      C

      • Minh Le Cong says:

        Casey, my point is that despite your biostats, you will get it wrong at some point. In fact you got it wrong in your math even in this case but luckily still made the right decision!
        My point of this is that you need to be honest and discuss this with your patient, that our tests are not perfect, our clinical judgement is not perfect and our risk calculations are not perfect and so …they need followup and review!
        Look at Andres case in the comments to demonstrate my point!

    • keeweedoc says:

      Hi guys.
      So case was an interesting one (losing a blog post here)
      Gut in late 50’s who had previous PE 3 years ago.
      presented classic pleuritic chest pain.
      “just like the last one”
      had been on warfarin for 2 years but stopped due to neuropsych side effects.
      then developed AF approx 6 months before presentation. rate controlled as didnt want to have a cardioversion. on dabigatran.

      Observations RR 16 Sats 94% Pulse 112.
      CXR clear
      USS no RV strain.
      no obvious leg source.

      so he’s a moderate wells (6%?) score but clearly I thought he had something.

      Ordered the CTPA and he was off getting it when the D-dimer came back negative.

      had a repeat on the ward 4 hours later which was positive but not massive.

      clexane for 1 month (refused warfarin despite counselling ect)

      Thoughts?

      • HI Andre
        I think his Well’s score is 6 – so that makes his PreTP either 16.2 or 28 % depending on which study you want to use.
        That is a pretty high number, and given the patient has another reasonable reason to anti coagulate – you could argue that he can just be advised to go onto anticoags [obviously he was of a different mindset!]
        Does he even need a CTPA? I know radical idea….
        After all – in small PE, without RV strain / troponin etc we don’t have any other really useful treatments. And if this is number 2 – then some would advocate life-long treatment

        Great example of why biostats are good, but we still need to use our brains and adjust all the above to individualise therapy with the patient involved.
        Cheers C

        • keeweedoc says:

          Cheers Casey,
          agree with your thoughts,
          the CTPA was done after discussion with the patient who was willing to consider warfarin for a period again but monitor for side effects (hopefully none and then continue life long) This changed the following day when he decided on clexane.

      • Minh Le Cong says:

        Andre thats a very interesting case and thanks for sharing!
        Its not a common one!
        lots of confounding issues in this one!

        He is on dabigatran already! so either he got PE despite adequate dabigatran..or he is noncompliant with dabigatran!
        The dabigatran surely would have affected the D Dimer test results!

        it would be important to confirm the PE Dx with CTPA as it means he is either getting PE despite dabigatran or noncompliance..either way it means he needs his anticoagulation Rx rationalised ( increase the dose or swap to proven agent like rivaroxaban, review compliance issues)

        on the issue of biostats being helpful here..hard to know because the calculations normally are for people NOT ON anticoagulants already!
        I see no harm in applying Well’s scoring as a guide but you see here how clinical gestalt was adequate in this case as clearly the CTPA was ordered despite D Dimer ( in fact CTPA was done before D DImer result known!)

        And this case is clearly not low risk given the presenting complaint, vital signs and past episode…so not sure if it applies to this discussion of low risk patients for PE and using something to make decision to send them home.

  6. Kirsty Challen (@KirstyChallen) says:

    Hi Casey and Minh

    I’m not sure you are answering the same question – Minh, you seem to be addressing the aim of finding all PEs; Casey, you are looking at finding enough PEs to benefit the patient without testing so many patients we harm them.

    I wonder if we might do better conceptualising some of our ED tests (like D-dimer and hsTn) as screening rather than diagnostic tests; they are done on large groups of patients, most of whom we don’t really think have the disease. In that case, we should look at criteria for screening tests (see, for example the UK NHS ones http://www.screening.nhs.uk/criteria). Some of them are particularly relevant:

    2. The epidemiology and natural history of the condition, including development from latent to declared disease, should be adequately understood – we don’t know what happens in untreated small PE (“lung lint”) – people probably do OK and have done for a long time before CTPAs.

    10. There should be an effective treatment or intervention for patients identified through early detection, with evidence of early treatment leading to better outcomes than late treatment – see previous!

    16. The opportunity cost of the screening programme (including testing, diagnosis and treatment, administration, training and quality assurance) should be economically balanced in relation to expenditure on medical care as a whole (ie. value for money) – we really don’t know this!

    20. Evidence-based information, explaining the consequences of testing, investigation and treatment, should be made available to potential participants to assist them in making an informed choice – now, unless we are David Newman or Jeff Kline, how many of us have enough of a handle to this to be able to explain it to the patient?

    So once again we conclude that we don’t know enough; at least in the case of PE there has been some calculation of a test threshold (see Pines above), but we lack this in many things – at least the upside is that there will be enough to keep me going through an academic career in EM 🙂

    • Seth Trueger says:

      Great thoughts but ddimer is NOT a screening test- it’s a diagnostic test. Screening tests are applied to asymptomatic populations of patients for secondary prevention; ddimer is used to diagnose VTE in symptomatic individuals. Using ddimer as a screen is how it INCREASES (when it should decrease) CT use.

  7. Anand Senthi says:

    Ok. Thanks for the twitter call up Casey. Good work on putting your approach out there. A few things:
    Firstly on the numbers – the meta-analysese I’ve reviewed suggest the following PTP’s:
    1. Clinical Gestalt Low Risk + PERC negative = 0.5%
    2. Wells <2 = 3% (initial studies suggested 1.5% but looking at more data I've pooled, it's about 2.8%)
    Widely accepted that 1 puts you below Test Threshold (TT). I argue 2 does as well especially if sPESI (simplified PESI score) = 0. i.e this is why I advocate the concept – "Prognosticate before you Investigate" … but that's a longer discussion. Writing a paper on this currently … hopefully if published we can chat about some more.
    On the current issues I completely agree with Casey. Anecdotes can cause harm because we focus on the anecdote in front of us but not on the deaths we don't see caused by the consequences of over investigation. This is bad medicine and harms our patients due to our psychological biases.
    As Casey pointed out, the key thing is we need to use the data we have to know:
    – the Test Thresholds (TT) for the tests we use
    – the sensitivity and specificity of the tests we use (or the likelihood ratios where this is rolled into one)
    – the Pre Test Probabilities (PTP's) we get from our various risk stratifying tools. The TT's apply to all tests but they are particularly relevant to PE as the tests are harmful and the false positive rates are extraordinarily large leading to frequently unnecessary treatment with a dangerous drug. Further PE in amubulatory ED patients with normal vitals (Casey's "Pulmonary Lint") has a very low mortality (likely <<5%) if untreated.
    So nothing we do will ever "exclude" PE. The best we can do is risk stratify patients to a place where we think based on the data we would do more harm by our investigating and therefore stop there. I agree of course patients need to understand this and be provided a safety net.
    Further if we do this, when the 1 in 100 or so patient comes back with a PE, Minh, using terminology like we were "WRONG" is entirely unhelpful and propagates our dangerous practice of over-investigation. If you followed a risk stratification methodology that takes into account the test threshold and the patient's PTP and make a patient centred and informed decision, then you can't be wrong.

    One final note we should acknowledge that none of the data is perfect and there is some uncertainty regarding all our estimates. We should acknowledge this uncertainty in our decision making and discussions with the patient.
    Therefore where the data is pretty clear (like gestalt low risk + PERC negative) it is an easy decision and patient discussion but where the numbers come out below but close to the TT I might have the following discussion with the patient:
    – "from the information we have I believe your risk of a blood clot in your lung is very low but not zero. Further even if you had a blood clot your risk of dying from it is low and probably less than 5%. The tests required to diagnose a blood clot carry some low but significant risks of harm. Based on the evidence we have, I believe the benefit performing investigating on you for a clot is likely less than or possibly about the same as the risk of harm caused by performing these investigations."
    I then ascertain the patient's concerns and beliefs and we come up with a decision together regarding whether to investigate and of course I document all of this.
    One thing that I find consistently in such discussions (and this has apparently has been displayed in studies) is that patients are far more risk tolerant than we are!
    I used to work in finance and the objective was managing investments and their risk profile based on the risk tolerance of the client NOT the adviser. Oddly in medicine we often focus on our own risk tolerance and that of our colleagues which is quite irrelevant.
    It is the risk tolerance of the patient that matters, not ours!

  8. Minh Le Cong says:

    thanks Anand! I agree with the majority of your comments

    My concerns are that this case seemingly portrays the use of additional biostats to increase the confidence of risk assessment to somehow allow better counselling of the patient.
    I remain unconvinced

    At the end of the day, you are stil going to have to explain to the patient what you think is going on, what you recommend to do about it and your followup advice
    Since the D Dimer was ordered and is abnormal the discussion should include this aspect, regrettably!
    If your gestalt and PERC calc = low risk 0.5% then I see no need to do anymore calculations just to allay your anxiety over the positive D Dimer.
    I think you need to simply explain the results of an abnormal D Dimer within the context of the low risk assessment and provide advice on followup and review.

    Now here is where it gets grey!
    If the patient then turns around and says they do NOT WANT to ACCEPT THE RISK AND insist on further investigation…what do you do?

    Anand you seem to indicate in this situation you would order further testing including the potentially harmful CTPA and flow on consequences.

    This is where the biostats and EBM does not help us much at all! In fact they may have contributed to a confused pt who sense the doctor’s indecision and decide to take matters into their own hands.

    I have had situations like this where I have told the patient I would not recommend further investigation and when they have insisted, I have recommended they seek a second opinion as I felt further testing was unwarranted!

    Because say for example every low risk patient decides that you as their doctor made an erroneous risk assessment and ALL insist on further testing..what are you going to say then? Order CTPA on them all.?

    This is the grey area and you either have conviction of what you believe in the EBM/biostats..or you dont.
    Hence my contention that the extra biostats calculations , in the end in this low risk pt group, are not helpful in the final clinical decision making for the individual.

    • Pik Mukherji says:

      Partial agreement Minh. We have to carefully peruse the stats to convince ourselves of the right course. But we should not give the pt. equal partnership in the decision. A recommendation and reassurance from the medical expert is what the pt. expects. When the pt. disagrees, I will venture that it is almost never about the math. So I agree that showing them a 3% value isn’t likely to be as fruitful as attentive listening, internal math, and reassurance. If they ask the logic, I’d share it with them, but if it gets to that point, there is likely a communication or trust issue that has already occurred.

      So: 3.7?

      And those “grey zone” pts. we keep talking about? I don’t think they’re folks who have a lower risk tolerance than their physician. I think they’re people who don’t trust your advice, often for reasons entirely separate from the medicine involved. And as far as the average person’s ability to consider risk – let’s just say that we suck at it, and that “blood clot” will occupy a larger representative area than the math says it should. Just as death by terrorist attack will always loom as a larger risk than death by fall in bathtub, even though the numbers don’t bear it out. (I include all medical doctors in this assessment- “WE suck at it.”) Getting off topic. It’s 3AM. G’night.

  9. Pik Mukherji says:

    Thanks for putting up the numbers- I admit I’ve never formally used the +LR on the D-dimer, but it makes total sense. And the “N of 1” case report in Annals of a PERC negative pt. with massive PE basically had the typical effect of scary anecdote- lots of folks in my shop became nervous about using PERC and started adding D-dimer to those cases, “just to be sure.” Which basically made it a worse testing algorithm than PERC alone.

    Only quibble is going to be accepting your 3.7% mortality threshold. I would be curious as to what the agreement from the esteemed panel is for that number. Putting aside things like the Hugli study with much higher baseline prevalences of PE, does the post-test prob. of 3.12 actually live in “you-can-ignore-it-land?”

    -pik

  10. Anand Senthi says:

    Minh,
    Thanks for bringing the debate back here. Too much to say for twitter.
    For a start I don’t think you can ever truly simply “ignore” a d-dimer as you suggest with the PERC negative and low risk patient. You need to still consider what effect this has on their post test probability before you choose whether to ignore it or not. As a rough rule the positive d-dimer doubles your ( Pre Test Probability (PTP) – actually a bit less than that. For low PTP’s between 0 and 5% a positive d-dimer almost doubles your PTP while for higher PTP’s (eg above 10%) a positive d-dimer increases your PTP by about 1.5 times only.
    So for a PERC negative + low gestalt pt your PTP is 0.5%. With a positive d-dimer you can double that to 1% (actually 0.9%). So 1% is still below the TT so after considering the d-dimer you can safely ignore it.
    However it the patient’s PTP was higher eg a Wells <2 patient whose PTP is about 3% then whether you choose to consider and then ignore a inappropriately sent off d-dimer will depend partly on what you believe the true TT is (I believe it is over 5% and perhaps over 10% on the data I've reviewed but that's up for debate). It also depends on whether there is an explanation for the positive d-dimer – the stats I gave you are the average case. In a person with pneumonia, a positive d-dimer would not really change your PTP at all, (because it's specificity would drop from a low to horrendously low level) so you could ignore it.

    In regards to your general point re the issues regarding involving the patient in the decision and use of biostats:
    – biostats are very useful for you being able to understand the patient's risks and it is often better than gestalt alone (though gestalt is often part of the calcs eg Wells/PERC)
    – biostats help provide some concrete information to the patient about the risks and benefits that they can understand. eg Percentage risks and NNT's are eminently understandable for patients.
    – the amount you involve the patient in the decision process and be guided by their input will always depend on how clear cut the decision is. If a person presents with severe epigastric pain radiating to their back, I don't involve the patient in the decision to do a lipase … I just do it and I'm sure you are the same. However if a Well's Score <2 pt, they are in a grey area where the best evidence we have (which is not perfect) suggests they are likely below the test threshold. If they wanted to be investigated with the risk/benefits understood I would have no issue with that. In fact I think an RCT on this very group is both required and ethical.
    On the other hand a patient who has a clear h(x) of picking up something heavy causing immediate intercostal muscle strain and corresponding focal tenderness which precisely reproduces the pain, with normal vitals and they are low Gestalt + PERC negative then they clearly don't have a PE … no grey area here. In the very unlikely event that despite our discussion still "demand" a CTPA, then no, I probably would not accede to their request and I may follow your good example above and provide them with the right to a 2nd opinion.
    None of this stuff is black & white and it is all on a spectrum. I'm sure we all agree that the patient should be involved in the decision making process when there is some grey area and the decision is not clear cut. How much you involve them and how grey you feel the area is depends on the situation but we can empower ourselves and the patients if we have a firm quantitative understanding of what the real risks and benefits are and not accede to dogma and fears of litigation/criticism that promotes over investigation which loses more lives than it saves. As mentioned above taking into account the patient's risk tolerance, not our own, is critical.
    Also the way you deliver the information is important and you can do so without being seen as "weak or indecisive". Instead you can come across as knowledgable and sensitive to the patient's own fears , concerns and risk tolerance. I've been having these data based risk benefit discussions with patients on several topics, but particularly PE I(x), increasingly over the last 2 years, and I have not come across any issues. I certainly have had no experience with the hypothetical situation you suggest where every low risk patient wants to be investigated. People are unsurprisingly usually quite reasonable when presented with good information, in an appropriate manner at the appropriate time. Further I find patients really appreciate being involved in the decision where appropriate. Finally the data would suggest I'm managing the patient's risk as best as I can, taking into account the patient's risk tolerance, which at the end of the day is the most important thing.

    Pik,
    The mortality benefit threshold is just the Test Threshold flipped on its head. It is how much of a reduction in mortality you would you need to be able to produce by investigating the patient when their PTP is a certain number. i.e it changes with PTP.
    Casey was being very generous in his example with his mortality threshold or he felt the PTP was 15% … I'm unsure. Say you thought she was Wells <2 patient, her PTP would be 3% so the Pines study would suggest the mortality threshold is around 15% ish (somewhere between 10 and 24%). Here is the study
    http://www.ncbi.nlm.nih.gov/pubmed/22978741

    Given the data suggests that people like the lady in this case (ambulatory, ED pt with normal vitals) have an untreated PE mortality of likely less than 5%, it is really impossible that she would be above the mortality threshold cause the most you could reduce her mortality is only 5% (likely less) by investigating her and if positive for PE, treating her with anticoagulation (which has unproven benefit and proven harm – another discussion).
    So if you follow that line of thinking you would not investigate her.
    Casey's case though is complex. A lot would come down to your assessment of her legs and whether you really thought she had a DVT. If so, then her PTP would be a lot higher. I agree with Casey that I would not pay much attention to her short haul flight when even long haul flights have minimal increased risk attached to them.

    Cheers

    Anand

  11. This is such a great discussion thread and thanks to all of you for sharing your insights without holding back.
    I’ll add two cents.
    1) I only trust PERC in low gestalt patient. “Twinge” of CP that is intermittent and no associated RFx, for example, would qualify. Pleuritic CP post air travel that is associated with dyspnea I don’t think would apply (although, depends). PERC will miss PEs, but most of the time you can find out that the patients wasn’t really low gestalt.
    2) One thing we need is to find out which PE is safe to not treat with anticoagulation. While doing the CT and the risks assoc with this is another discussion, it would be very helpful to know which “lung lint” for instance won’t require coumadin or rivaroxaban or dabigatran and therefore save the patient from bleeding risk.
    3) Agree with the approach that we need to be up front with patients about risk. I often say, “I can’t say for sure you don’t have a PE without a CT scan, but sometimes we see PE on scan and we wonder how significant it is in the first place – and perhaps you wouldn’t even need treatment or would even be harmed by it” Of course, to the lay person this is scary and often gets a response with a furrowed brow. If the patient cannot accept this, and asks for workup, then I oblige. Of course the conversation is not as simple as that. But for the purposes of this thread, I’ll leave it there. I cannot expect patients to feel completely comfortable with an approach that we as physicians cannot even come to consensus on. Lastly, as Minh stated, followup is keep for those patients where uncertainty remains.

    Casey, a wonderful case that clearly has sparked great discussion. Thanks for posting!

    -Elisha

    • Seth Trueger says:

      Re: “I only trust PERC in low gestalt patients”

      This is where people get in trouble with PERC. Entry into PERC requires low pretest prob; initially 15% but the recent metaanalysis shows PTP must be 8% or less. Not that we’re good at quantifying those numbers– but we’re great at gestalt “that patient is low risk.”

  12. Anand Senthi says:

    good points, though also worth bearing in mind that cohorts in Anglo Saxon countries you generally find that PERC works in most patients because almost our entire cohort that we investigate is low risk. We like to think we are not doing so but we are so PE paranoid that we end up finding few PE’s. Mainland Europe is the opposite, probably due to their civil law legal system which lacks tort law so getting sued is very difficult. When they generally investigate they really believe the patient has a PE and their PTP rates are far far higher. Also note while our gestalt is good we generally grossly over-rate the risk of PE due to our PE paranoia i.e in Kline PERC validation study, when docs through PTP was less than 15% it was actually 3%, when they thought it was 15-40% it was only 10% and when they thought it was >40% it was 31%.
    http://www.ncbi.nlm.nih.gov/pubmed/18318689
    So yeah agree PERC is only for the low risk but we should be mindful that when our gestalt suggest the patient is lowish risk eg <15% PTP, they are likely well below that.

    Absolutely 100% agree we need to figure out who really needs treatment. The biggest problem was created by the Barritt & Jordan study of 1960 that supposedly proved anticoagulation worked but proved nothing because it was a horrendous study in 38 patients with "clinically" diagnosed PE and a raft of methodological flaws. Consequently AC has been considered standard of care and no one's been game to do a placebo controlled trial. However the low risk patients we see in ED who are physiologically normal are the ones we should be studying the risks/benefits of investigating and treating.
    On that note, I've been most heartened by the sPESI score (simplified Pulmonary Embolism Severity Index). It uses 5 simple risk stratifying criteria, 3 physiological – PR, BP, sats – as well as 2 comorbidity criteria – cardiopulmonary disease (heart failure, COPD) and cancer to produce a simple 0 to 5 score. The few very recent truly prospective validations of this score I've located so far have a pooled all cause treated mortality of 0.1% and and PE mortality of 0% (I'm hoping to get this pooled data out there soon). A few things:
    – this focusses on clinical criteria rather than CTPA findings.. The latter if relevant to survival would probably affect the former. Importantly the former can be calculated before you even investigate and can therefore help you with that decision.

    – it is better than Echo and troponin at identifying a low prognostic risk group. Echo and troponin seem useful at further risk stratifying patients who are sPESI>0 into higher risk categories. However they are not useful for sPESI=0 patients.

    -this is treated mortality. However it certainly would risk stratify members of any group to a low mortality (within that group) eg Wells < 2 - this would be the very group that would be worth studying whether there really is any benefit to investigating and treating. I highly suspect there would not be. Anyway enough from me. Cheers all for the great discussion. Goodnight.

    • Amy Neilson says:

      Hi Casey,

      Thanks for the clinical question and link. This is an ongoing discussion in our ED as I imagine it is in many, and still different people do it differently… I and Neale, one of my colleagues, tend to err on the side of always looking to not CT… Others have a different threshold..

      I agree with Minh that maybe this is a little over-complicated (I have a Maths degree and an MPH and got lost – so I take my hat off to you though! :-)), but then again everyone thinks differently. If the Well’s is genuinely low risk, I would head to PERC, and if PERC rule out all good, then I would stop then.

      Unlike Minh, Neale and I have agreed at times on ignoring the D-Dimer in that situation as it was not indicated. We tend to err on side of thinking likely picking up ‘lung lint’ as you say and thus subjecting people to unnecessary anticoagulation and often repeat unnecessary CTs. Always up for evidence to support this. Only aware of there being increased diagnoses of PEs but no increased mortality and going off this.

      Do have a question about this patient and the low risk Well’s – there is the option of giving her 3 points for PE being the leading diagnosis… I realise this is a sticky bit and tends to be used or not used depending on which bit of gestalt you are trying to err on the side of.. With her HTN I could certainly live with ACS as leading DDx.

      However, if you say there is no convincing alternative, and then she gets to intermediate Well’s. I would d-dimer all intermediate Well’s and if positive go on to consider looking for PE.

      High risk Well’s I tend to just look for PE as even -ve D-Dimer not particularly reassuring given error margin.

      Agree with Minh re: 2nd opinion is they are adamant for CTPA and they are genuinely low risk and PERC -ve. Trying to put myself in that position… I guess in reality I might be convinced to do a D-dimer in this situation even though low not intermediate and cross my fingers and toes it is negative….But probably that is unwise…

      The use of CTPA vs just treating got a brief mention above. So did USS and treat (can’t find that bit now but saw it before..maybe on Twitter?)…
      We had a patient the other day who was pretty much having an ‘end of the bed PE’. He was SOB, feeling unwell, CP, had Ortho surg 1 week ago, exquisitely tender, swollen, cool (L) lower leg on surgical side, previous bilateral PE following other Ortho surg….
      USS showed the expected DVT. Neale and I were erring of the side of stopping there and saying, “Yes, you have a PE. Crack on”. The thought was obviously the old thought about if you find a DVT does it really matter about the PE as you are treating it anyway? The only caveat I have with this conceptually is, is the PE large enough to affect function? In that way, I am wondering, for those in whom PE is clinically diagnosed by Sx in the presence of DVT, is it adequate to start anticoagulation, skip the radiation, and do a TTE? This is my preference if defensible.
      (Someone else took over the patient and did a CTPA, as it happened. And he did have a PE. And the plan did not change…)…

      Thanks for the link. Much appreciated. Really interesting to see how everyone thinks this though.

      Cheers

      Amy

  13. Anand Senthi says:

    hey Amy,
    quick response as on holiday and papers not at my finger tips but:
    – absolutely agree that we should be considering DVT U/S first and if positive together with s(x) of PE we should just assume PE and treat. No added value of CTPA and signif reduct in risk. There is a paper on this about it being a cost effective first step- I’ll try and get you the ref soon. It absolutely should precede CTPA (and replace if positive) in a pt with PE s(x) who has signs of DVT but there’s a strong argument to suggest perhaps it should be first test post positive d-dimer in all patients ?PE patients. Something related also to note that it is often quoted that pt’s with PE only have DVT on U/S in 30-50% of cases. However what hasn’t been factored in to those stats is that CTPA’s have a high false positive rate in low risk pt’s (50% of positives are false positives if Wells <2) so quite possible that 30-50% figure is an underestimate as many of those cases where CTPA positive and U/S negative may have been false positives. Food for thought.
    – Kline talks about high incidence of CTPA recidivism in pt's d(x) with PE. Real risk.
    – agree if choose to I(x) low or intermediate risk d-dime is right first option

    Re the case of 15.yo PE on twitter: be great to have full case details to discuss.

  14. Anand Senthi says:

    in addition – re whether you should consider Echo on patients that you have diagnosed with PE by diagnosing DVT first, I would suggest a slight amendment to that approach in accordances with my earlier comments above:
    I would calculate their sPESI score. If sPESI = 0, no added benefit of echo and troponin – extremely low mortality. If sPESI>0 then echo and troponin worthwhile. Also consider lactate – some good early evidence of raised lactate being a a really good prognostic marker in such patients. Exactly what to do with these higher risk prognostic patients is unclear but would treat as inpatient and may consider thrombolysis.
    http://www.ncbi.nlm.nih.gov/pubmed/23306454

    Also found the paper re cost effectiveness of a d-dimer then U/S approach in all patients presenting with PE s(x)
    http://www.ncbi.nlm.nih.gov/pubmed/20605261

Speak Your Mind

*