Estimated reading time: 8 minutes
There are many things that we do in Emergency Departments ‘just in case’. In this series we’ll be looking at some of these to see if we really do need to continue doing them in the current way, or perhaps there’s an opportunity for a change.
Some years ago, friend of St Emlyn’s, Cliff Reid coined the phrase ‘Dogmalysis’ which may well be appropriate with some of these discussions (depending on the outcome).
The first in our series will be one that happens everyday in the ED, many many times:
The ECG/EKG Thrust
Many patients as part of their initial assessment (as soon after arrival as possible and usually before seeing a clinician) will have a series of tests requested and whether you are new to Emergency Medicine, or been practicing for years, you will be very, very familiar with the phrase
“Could you just sign this ECG?”.
This is followed by a short period of examination of the ECG by the receiver of the request, the application of an indescifrable signature and usually the requested follow up “could you get an old one for me” or “let’s repeat it in 15/30/60* minutes. (*choose a number at random).
Now, let’s put aside the inherent bias in intial request (and its possible unintended consequences) and think about the value of an Electrocardiogram (ECG) for patients when they first arrive in the ED.
What is the purpose of these tests?
The reasoning behind performing these tests is twofold: firstly (and for the majority of patients) we are hoping to ‘increase efficiency’ so that the results are available when they are seen by a clinician and decisions can be made about their management, but more importantly we are also checking for diseases that need treatment as soon as possible. In the case of the ECG these include not only myocardial infarction, but also some significant electrolyte disturbances (such as disruptions in potassium levels).
Screening vs Focussed testing
Before we go any further lets pause to consider the difference between screening and focussed testing.
Screening is testing for disease before the manifestation of signs and symptoms. Now, we can’t really say that about our patients in the ED, most (?all) of them have symptoms of some description, although many who get an ECG have symptoms of diseases that will not have positive electocardiographic findings.
Before going any further we need to think about the patient population who are having these ECGs. Ideally this would be a patient in whom we suspect conditions that may manifest themselves in an abnormal ECG that requires immediate treatment for the best outcome, the highest risk one being the Occlusive Myocardial Infarction (cf STEMI).
But we all know this isn’t quite true: when I’ve asked why a patient has had an ECG I’ve received responses as wide ranging as: “a fast heart rate”; “chest pain”; “shortness of breath”; “collapse”; and my personal favourite “just in case”.
In each of these, we can only interpret the ECG (like any diagnostic test) when we know the prevalence of the disease in the population we are interested in.
In essence, most departments have very wide criteria for an ECG on arrival, where the prevalence of the disease we are really interested in from a time perspective (ie OMI) is pretty low. In one study1 there was a rate of about 5% for serious disease presenting to the ED with chest pain (and let’s not forget that OMI/STEMI was only a proportion of this). They are, in effect, performing screening.
What about the ECG in patients who actually have cardiac sounding chest pain?
In one study the ECG has a sensitivity of 56% and specificity of 94% for the diagnosis for STEMI (this translates into a positive Likelihood Ratio (LR+) or 2.14 and a negative LR of 0.11). These are figures for “expert interpretation” (often by two cardiologists), so you could argue that the accuracy in the general ED medical team may be worse even than this.
You all know that at St Emlyn’s we like to invoke Rev Bayes and his theorem at any opportunity, so let’s add in our prevalence and those diagnostic criteria to our group of patients to see what it gives us.
- Pretest probability = 0.05%
- LR+ = 2.14
- LR- = 0.11
- Post test probability = 10% of having an STEMI or 0.06% of not having one
In the group we are interested in, this means that the probability after the ECG of the patient having serious disease that may need urgent intervention is 1 in 10.
The ECG is not the only test
As we can see from the figures above, the ECG is actually a relatively poor test for acute myocardial infarction. In one (frankly seminal) study, lead by our own Rick Body, pain radiating to the right arm, vomiting, central chest pain and the presence of sweating all had better diagnostic characteristics than the ECG.
But what if I miss something?
This seems to be the phrase that is harming Emergency Medicine the most at the moment. We perform many tests indiscriminately for fear of ‘missing something’, without ever stopping to think of the consequences of this sweeping strategy.
Of course, one of the ironies is that many patients with occlusive myocardial disease have a normal ECG, and so we may be falsely reassured at the very start of their attendance. And don’t get me started on the ‘”I’ve printed off an old ECG and it looks the same” which may be a topic for this series in the future.
Come on! It’s only an ECG…
Yes, it may only be an ECG, and indeed is a relatively cheap and painfree test, but there are often unappreciated knock on effects. The ECG machine may now not be available for a patient who really needs it; the nursing staff could be doing other things, that arguably matter more, and it is well documented how interruptions (in this case “The Thrust”) can affect clinical and cognitive performance.
Being the Thruster
I don’t think we should underestimate the stress on those asking for the ECG to be looked at. Often these are more junior members of the ED team, who are simply doing their job. It is not their decision who needs an ECG on arrival, yet all too often I see senior doctors sigh, tut, roll their eyes and even be blatantly rude. If we can be more discerning about who gets an ECG on arrival, and ensure all members of the team are supportive of this, it’s not hard to imagine that not only do the interruptions decrease, but the ECGs would get closer examination and reduce the stress on all concerned. In the interim we should all adopt the position that if we as senior clinicians approved the policy of taking and signing ECGs, then the very least we can do is to be polite and thank people for doing the job that we asked them to do! If you’re a senior doc and don’t like the system then change the system, never, ever blame the people doing the job you or your colleagues have asked them to do. Ever.
So you want me to stop doing ECGs?
NO! That’s not what I am saying. But what I would argue is that we can limit who gets them on arrival significantly (after clinical assessment there may be reasons that an ECG is entirely valid).
What about automated analysis?
This could be an inviting option. Although it does feel a bit of a cop out – we don’t target our testing more appropriately, we simply rely on a computer reading. I think the jury is out as to whether these machines are accurate enough to be totally trusted.
So, who should get an ECG on arrival?
For the immediate ECG prior to clinical assessment, I believe that the patient should have signs and symptoms for acute coronary ischaemia and be over the age of 40 (when disease prevalence starts to increase). You may want to be conservative when you set your parameters. The decision to perform an ECG (or not) should be backed up by a simple sets of questions.
- Does the patient have chest pain consistent with cardiac disease? AND/OR
- Does the patient have symptoms or signs consistent with significant cardiac disease or eletrolyte disturbance? AND
- Is the patient over 40 (if younger than 40 confirm need for ECG with senior member of team)
Let’s not forget that this is only a small part of the process: all patients should have a set of vital signs and ‘triage’ assessment within 15 minutes of arrival. This may cause them to be directed to a resuscitation area, or a clinician asked to see them as soon as possible and make the need for an ECG more apparent.
This doesn’t mean that patients won’t or can’t have an ECG further on during their assessment; it just means that it won’t be part of the ‘screening’ that happens on arrival.
If you limit the number of ECGs being done, the doctor assessing the ECG will know that the patient is already from a group where OMI is suspected and be of significant experience not only to interpret the ECG accurately, but also to action what may need to happen in the event of positive findings. The should them examine the ECG for evidence of myocardal ischaemia and infarction (ST elevation or elevation)
Oh, and when the Thrust happens, it should be accompanied by the phrase.
“Please could you assess this ECG, which is from a patient in whom I suspect is having an acute cardiac occlusion based on the age and these symptoms…”.
The Real Question
I think the science here makes sense and I would happily stop us performing at least 50% of the ECGs that are done on arrival to the ED, but the deeper and more difficult question is
“Are we brave enough?”
Do we keep doing tests ‘just because’, due to a notional fear of ‘missing something’ or actually start to redirect our limited resources to those who need them.
Hsia RY, Hale Z, Tabas JA. A National Study of the Prevalence of Life-Threatening Diagnoses in Patients With Chest Pain. JAMA Intern Med. 2016;176(7):1029–1032. doi:10.1001/jamainternmed.2016.2498
N. Herring, D.J. Paterson, ECG diagnosis of acute ischaemia and infarction: past, present and future, QJM: An International Journal of Medicine, Volume 99, Issue 4, April 2006, Pages 219–230
Body R, et al. The value of symptoms and signs in the emergent diagnosis of acute coronary syndromes. Resuscitation. Vol 81, Issue 3, p281-286, March 01, 2010