Have you ever heard this sort of conversation in your ED?
Junior ED doc: Can I run my patient by you, please?
Senior ED doc: Sure
Junior: I have a 40 year old man with chest pain. I don’t think it’s anything to worry about. It only lasted 20 minutes and the patient says it was just a “mild discomfort” rather than an actual pain. He’s got no past history and he only came because his wife got worried. I want to send him home. Is that OK?
Senior ED doc: Hmmm…
It’s tempting to think that, if a pain is very mild in severity, it’s less likely to be caused by something serious like a heart attack. But what’s the truth in that? We asked just that question – is severe chest pain more likely to be a heart attack? Our findings have just been published in the European Journal of Emergency Medicine. You can find the paper at this link.
[DDET So what did we do?]
You might have already read about how we derived and then externally validated the MACS rule. For this analysis we used some of the data collected during that external validation study. So, this was a substudy of a prospective diagnostic cohort study. We included 455 patients who came to the ED with a primary complaint of chest pain, which the doctor treating the patient suspected might be cardiac.
We asked the patients to tell us how severe their pain had been at its peak from 0 to 10 (the numeric rating scale). Then, we followed the patients up. All patients had to undergo troponin testing at least 12 hours after symptom onset. Because the pain score was recorded on admission, the doctors were essentially blinded to the troponin results. Using the troponin levels and other relevant clinical information, we adjudicated which patients had an acute myocardial infarction (AMI). Then, we worked out how pain score affects the probability that a patient has AMI.
[DDET What did we find?]
We found that patients with AMI did have more severe pain, on average, than patients who didn’t have AMI. But – while this did reach statistical significance (p=0.03) – the actual difference was small! Patients with AMI had a median pain score of 8; patients without AMI had a median pain score of 7. What’s more, as a diagnostic test the pain score is pretty useless. The area under the ROC curve was 0.58, which is only just better than flipping a coin to decide.
You can see from the chart below (which is similar to a chart in the main paper) that patients with STEMI reported significantly more pain than patients with non-cardiac chest pain, but patients with NSTEMI, unstable angina, other cardiac diagnoses and non-cardiac diagnoses all reported similar pain severity.
[DDET Did it matter if the patient was old or young, male or female, or diabetic?]
No. It didn’t really matter. There was no difference in pain severity based on age or diabetes. Women did report slightly more severe pain than men – in fact, female gender was an independent predictor of higher pain score! (Ed – does that mean that the old myth about men being wimps is untrue?!) However, here’s some extra data not published in the paper – #FOAMed – just for you! Before us guys get our hopes up that we’ve at least got some evidence for our macho status, it’s worth pointing out that – even though women reported more severe pain – the guys tended to get more morphine. Hmmm… What started out as a simple research question may have opened a can of worms about possible gender inequality! Worth exploring some more in future work, perhaps…
In fact, the predictors of receiving morphine on multivariate analysis were the severity of pain reported and the clinician’s estimate of the probability of ACS. Presumably, doctors are more likely to consider that men have ACS. Statistically, that’s true. But does that mean that we under-recognise ACS in women? There’s now some fairly strong evidence that this might be true.
Lastly, it’s also worth mentioning that we took a look at whether troponin could be used as a sort of ‘pain biomarker’ for patients with AMI. Troponin levels correlate with infarct size. So did the 12h troponin level correlate with pain score? Perhaps more myocardial injury causes greater pain.
In fact, we found absolutely no truth in that suggestion. There was no correlation between pain score and peak troponin level. The r score was, in fact, -.0.001 – you can’t really get a lot closer to zero.
[DDET So what’s the bottom line for our practice?]
The key take home message from this paper is that, if your patient with unexplained chest pain reports that it’s only very mild (or “not even a pain”), don’t immediately discount it based on the severity. It could still be AMI. Just like cardiac risk factors, pain score really can’t help to rule out AMI in the Emergency Department. If the patient’s symptoms are otherwise compatible with ACS and you haven’t found an alternative cause, use a proper, evidence-based rule out strategy.
All the best,
Huge thanks to Beth Haves, who was an author on this paper and a huge help with our study. Beth presented some of the initial analyses from this paper at the College of Emergency Medicine Annual Scientific Conference in 2012.
[DDET Check out this incredible poster she created!]