Podcast – Diagnosis in Emergency Medicine Part 3 – The importance of prevalence

Hello and welcome to this an Emeline’s podcast. I’m still Iain Beardsell and I’m occasionally Simon Carley. We’re glad to have you back for episode three of this diagnosis type section of our podcast.

Simon, it may surprise some of our listeners to know that actually this podcast does go through some sort of quality control procedure before it gets released to the public. This happened with our last episode and a member of the team, the esteemed researcher and expert diagnostician Rick Body, wanted to make this comment Simon and post you the following question. So pin back your ears because it’s quite a tricky one.

Rick says, deep breath, one quick point. If the population has a prevalence of 10% and you use a test with 90% sensitivity and 70% specificity, that’s a high sensitivity troponin on admission, the post-test probability will be less than 2%. Would you discharge a patient on that basis or would you admit for a second sample? Now before you dive in there, there’s more.

To play devil’s advocate, I’m not so sure about the 2% figure. I see a patient with suspected cardiac chest pain probably more because people save them up for me. If I was full-time and worked four to five shifts per week that’s five patients per week, I’d therefore miss an acute myocardial infarction every 10 weeks which is five AMIs missed per year. The mortality of a treated AMI is around 5%, untreated we know it’s two to three times as high. Let’s be kind and set it at 10%. So it’ll take me two years for one patient to die of an acute myocardial infarction that I miss and it’ll probably take me five years to get over each one if the GMC lets me. Suddenly two percent seems a lot worse. So Simon I don’t know if you want to split that into parts and have a think about some of those things.

Okay but it brings up quite a few reasonable points there doesn’t it? I only think the first one we need to pick out is this issue of prevalence and risk and how many patients who are actually missing. So let’s go through his numbers again.

So in our last podcast we talked about identifying a low-risk group population so we streamed out for instance our cardiac patients with ECG changes and we ended up with a population who had a prevalence of disease of about 10% also the pre-test probability. Yeah happy with that? Yes we were there 10% prevalence.

10% so let’s take a thousand patients so a thousand patients who come through the door with chest pain who are going to investigate who have decided a low risk that means about a hundred of them will have the disease and those are the ones we’re looking for. Yeah we want to concentrate on a hundred people who might have it yeah. You’re chalking this down on your backboard? I’m thinking it through and writing it down.

And we said we’ve got a 98% sensitivity test yeah so that means we’re going to pick up 98 out of a hundred of those will miss two. But that’s two out of the hundred who have got the disease but also two out of the thousand patients that we investigated. So it’s only one in five hundred patients that we potentially miss a diagnosis. So we’re not actually missing the one in 50 we talked a little bit about in the last podcast we’re actually missing one in five hundred.

Yeah I mean we kind of we’re kind of doing both we’re missing one in 50 of people who got disease or one in five hundred of people who are investigating. And that’s really important because people use those two different phrases at the same time and it’s quite important that you know exactly which population you’re talking about when you have these discussions. So we’ve now got a population of a thousand patients of whom a hundred have the disease we’ve done our test we’re still left with the one in five hundred.

Well it’s interesting isn’t it because if you say one in five hundred oh again is that natural frequency thing isn’t it one in five hundred sounds like quite a lot that’s that size of our girl’s school you know it’s one per year of the same population that’s a worry but the question is what’s the consequence of a miss? So what happens if you miss an MI?

Well I guess it depends on how big the part of the myocardium is that’s affected. I guess there’s people wondering around all the time who’ve had missed MI’s who don’t have a percent hospital could be something bad or maybe something nothing at all. Yeah and that’s really important I think many people when they say a miss they think that it’s going to always happen negative consequence but let’s just explore that a little bit more.

You’re absolutely right nothing may happen. So for some conditions there may be no consequence whatsoever to a misdiagnosis.

What feelings are we talking about? Well you could even have something really quite bizarre so you’d have somebody who could have a very small subarachnoid bleed and then they’re never going to have one again in their entire life. You could argue that there’s no consequence from that. You could have somebody who you missed an appendicitis which are in its early stages which then resolved. There’s no consequence from that. In fact there’s even a benefit. In pulmonary embolus if you miss a diagnosis and the patient doesn’t have a consequence and they never had to go on to warfarin there’s even potentially a great benefit.

So you have to put all of these potential outcomes together in terms of consequence and then okay so that’s one group can put them to one side. What about those patients who actually are missed?

I think you said that it’s quite unlikely that they’ll be massive MIs. If we’re by the very nature of the fact that we’re potentially missing people they’re probably the group of patients who have the lowest burden of disease probably.

True. That is definitely true although we’re just we always using that probably word with the night whenever I hear the word probably I guess I could always substitute for it probably not.

But the fact that we’re not seeing huge numbers of people coming back in worse states would suggest that it’s probably okay. Probably. Agreed. Yeah probably probably. Agreed. Probably.

And then also we’re going to think about the natural history of the disease. So there are certain things where the adverse event can be very very sudden very precipitous and fatal. If you miss a myocardial disease and the person subsequently has a cardiac arrest then that’s obviously a tragedy. But that’s not the most common or most likely outcome. It’s much more likely that that person will have another set of symptoms or a further consequence of their disease which allows them sufficient time for them to come back seek medical attention and have it resolved at a later stage.

Now that’s obviously seen not perfect you do want to pick everything up first time around but we have to appreciate that not everything will pick up first time around. If we’ve got an opportunity to pick it up again then that’s a very valuable safety net for this group of patients.

I think there may be people listening to this who think well come on chaps live in the real world. You’re talking about a situation which it’s very theoretical it’s not the real world because the outcomes we’re interested in are not just whether the patient does poorly. We’re also interested in what do our colleagues think of us when we miss a patient. What does our hospital think of us? What does the patient think even if nothing bad’s happened? What does the patient and their family and perhaps even their lawyer think when they discover that we should have diagnosed them? Are we not just trying to treat ourselves when we’re looking at these things because we don’t consider all those patient outcomes you mentioned.

I know that if you have a serious incident in my trust and a missed AMI would be included in that. You’re then subject to a relatively supportive process but it’s still quite in depth anyway where they take you off and try and examine every aspect of why it was this was missed and then actions are put in place and you feel pretty bad about life and like Rick says it takes a while to get over even if there was no harm come to the patient. So do we not have to consider that as well as the true harm that’s been suffered by the person we’ve been looking after?

Okay there’s a number of aspects there which we can go into. The first is from an institution or from a legal perspective and I would say that if you adopt a set of guidelines and you adopt a principle of this is how we do things and it’s evidence based and we understand the diagnostic testing and we accept that to be good clinicians it doesn’t mean that we have to pick up everything first time every time and that is your departmental, college, institutional, national policy that’s not a problem that is your legal and institutional protection and something like a missed AMI is just a consequence of the way that we do things.

From a personal point of view and from a patient-centered point of view I think knowing this sort of thing changes the conversations that I have with patients. So that a number of years ago I would probably say at the end of a diagnostic process you do not have this. Dear Mrs. Miggins you do not have a diagnosis and I don’t do that so much anymore. What I do now is I say that we’ve investigated you and we can’t find anything particularly serious on this occasion very happy for you to go home. However if you do get any further symptoms or you’ve worried even if you get worried when you’re in the car park to quote Natalie May even if you’ve got the smallest concern come back and see us and there’s two reasons why I say that.

The first reason is because I think it’s the right thing to do. I think it’s a nice thing to do and it’s an honest thing to do. The second reason I say that is because if patients do have a consequence of a misdiagnosis or another diagnosis then they’re going to come back anyway and if the last thing they heard when they walked out the door was there’s nothing wrong with you they’ll come back angry. If the last thing they heard when they walked out the door is if you have any concerns come back and see us they won’t be angry and so I think managing that level of expectation and managing that level of risk in our conversations with patients is really important.

I think we’re lucky at the moment that we’re practicing in the UK and this is very much a UK based podcast but I’m aware that people around the world may dip into this and download it in other countries not least the United States of America where the approach may be different. I think we have to put a caveat on what we’re talking about that we both practice in the UK and these are ways in which we’re lucky enough to be able to think and to talk to our patients. I’m not sure talking to American colleagues perhaps that they’re able to practice in quite the same open way that we do.

I don’t know enough about it and we’ll have to drag somebody onto the podcast who can tell us more but what do you think around the world the approach to these things is where diagnoses are missed?

I had a conversation with a colleague who works in America who is embroiled in a situation which you’re describing really they’ve gone through a perfectly reasonable what I would imagine is a perfect reasonable diagnostic process a patient has fallen through the net and it’s been a missed diagnosis and there is a medical legal consequence from that. I still think that if you can demonstrate that you’re using the evidence appropriately that’s a good thing. I think the concern in some legal quarters and I don’t think it is just the US I think it does happen in the UK as well is that it’s an asymmetric assessment of a missed diagnosis they will only see the miss for that one individual patient that one individual patient that’s their experience. What we’re doing is we’re trying to keep a population safe and exposing to diagnostic interventions and subsequent therapies which are beneficial to the population as we talked in the previous podcast.

If you over investigate everybody then you get more false positives you have harm from some of the diagnostic processes and from a population point of view there is a position where you just have to say this is the point where we finished our diagnostic process for now we’ll leave it with an acceptance there are a couple of people who probably have missed diagnosis. Clearly the way to do that is this idea of shared decision making with the patient.

For people who don’t believe that Twitter is a valuable educational tool just yesterday, I was on Twitter and I noticed that there was a grand round happening in the US and a person I follow was at this grand round and Jerry Hoffman and David Newman both people who talk a lot about diagnoses were at this grand round and so I took the opportunity to say I could ask a couple of questions because this plays on my mind quite a lot this idea of over investigation to protect ourselves more than protecting a patient and so I tweeted and said it has the juggernaut of over diagnosis is this overtaken us is there anything we can do to stop it and I was really reassured that it appeared that both Jerry Hoffman and David Newman who I admire and respect hugely said no it’s not shared decision making maybe the way forward.

I think that’s what you’re describing we bring the patient on board we tell them the risks the benefits and perhaps we should call it the harms and the benefits rather than risks and we get them to become part of that decision making process we involve them again in a way that perhaps we haven’t in the past is that something you tend to do in your practice Simon?

Absolutely and I can give you a couple of examples where I think most people would understand it the best one I think is shared decision making is when we’re investigating pregnant ladies for PE who come in with critic chest pain because we have a process where we would go through in non-pregnant patients which would involve an early assessment early clinical risk early by chemical tests such as D dimers and risk stratification from there and then either moving forward into VQ testing and or CTPAs or increasingly CTPAs in pregnant ladies they’re un keen to go for the radiological investigation due to the exposure to ionizing radiation both to the fetus but also to the breasts and that is one of those few times where I think everybody understands that that’s a shared decision and so we have conversations with the patient to say well this will be our goal standard for diagnosis however do you want to go forward with this there are some alternative diagnostic interventions we can do for you such as pan leg ultrasounds follow up scans alterations to the way that we do VQ scans etc etc which we know are not going to be as good a goal standard as a CTPA but it’s a shared decision with the patient now if we can do it for them we can arguably do it for a much wider population of patients as well.

So this could be a way forward so that we could try and reassure Rick that maybe the numbers are smaller than he was thinking about that we’re actually going to and we keep using this word miss but I wish we could get away from it I wish we could think of another word instead of miss because it’s an inevitable part of a well thought out diagnostic process a lot of those patients aren’t going to come to harm and even if they do have a consequence that consequence will in all likelihood be small and we can be reassured by that but by sharing all of that with our patients it becomes a decision that we make together and then the way forward is one of harmony where we’re working together as a team to find the right answer rather than me doctor you patient you do as I tell you and I’m going to subject you to all sorts of stuff that you may or may not want.

Absolutely and I think going back to what we’re saying before I think it’s so important that those final few sentences when you finish up your consultation with your patient leave the door open that they’ve got the potential to come back in safety without feeling that they’re going to be judged and that’s good for them and it’s good for you.

So I mean I think that’s probably taken this long enough down this discussion for our third podcast for this series just shows what a single letter can get us talking about so please do write in comments on to our website or via Twitter we’d love to hear from you and we’d love to try and tackle anything you have about this topic so I think it’s one that we think about all the time normally Simon has some form of extra question that he’s now about to spring on me so I’m looking at him longingly no no no there’s always a question.

Oh I really thought that wasn’t going to be what’s today’s question so yeah I’ve got a question for you this week me and just for fun if there was one diagnostic test that you don’t have at the moment but you’d love to have in your department what would it be a diagnostic test that I do mean one that currently exists that I don’t have or a made-up test that I’d love to invent or let’s go for a made-up test now that’s even more difficult a made-up test I would quite like if it was available and I think it may be in the future either in the form of a delorean or a small glass shape ball structure I would like to be able to tell the future and I would like to be able to see 30 days in advance to see as a patient presents at the emergency department what their outcome may be and whether they’re going to be fine after 30 days or whether there might be something that I need to think about and then I can try and sort them out so that I can change their 30-day outcome how about you?

I think I still want a painometer I think I’d like a painometer yeah I’d like some method of being able to measure pain big not just to sort of judge people because I think that would be a bad thing I think that’s what people have thought about in the past about having a painometer so they can tell whether somebody’s genuine in pain not interested in that I’d like a painometer that would give you a minute by minute second by second level of pain so that I could adjust my therapies for analgesia and the ED because analgesia is so important to what we do it’s one of the most important things that we do.

I like that thought surely somebody’s listening to this who works for some big company and they can sort these out we’d like a crystal ball please and we’d like a painometer and hopefully by the time of the next podcast that would be great if that was all right so keep thinking about your diagnostic testing that really is it from us we will talk to you very soon take care everyone have fun.