Chest Pain? HEART Score Attack
Gday, Bonjour.
I have just returned to Broome after a month or so eating and drinking my way around Europe. Heading back to work this week and I have been doing quite a bit of listening to the EM:RAP podcasts. Long flights and waiting in airports!
The one podcast that caught my eye (ears) was an episode of ERCast featuring Ammal Mattu (ED ECG guru and super nice guy). It was about the risk stratification of ED chest pain patients. This is a continual bug-bear in most city EDs, and the rural areas struggle even more with less access to specialists, highly-sensitive troponin assays and provocative testing. So when I heard this episode, my rural ED brain went into overdrive.
The scoring system that Dr Mattu was speaking of is the HEART score. Oh great! I hear you all moan – another Cardiology scoring system with a smart acronym… put it with the others eg. TIMI, GRACE… but wait – this one is a bit different. It actually can be used in small EDs, is simple, and now has been externally validated in trials including some centres in Australia. So it ticks all the boxes for my practice.
HEART was derived in the Netherlands by Drs Six and Backus – it has now been externally validated in other countries and settings. The HEART score also differs from previous scores in that it was derived in a general ED population (i.e. the folk we see, whereas TIMI and GRACE were done in CCU patients – ie. folk who already got admitted.
So what is the HEART score? It is in the image below. [ Or you can access an interactive version over at MDCalc here ]
Check out the shape of that curve – it is a beautiful sigma with a nice flat tail on the left! this is a discriminating tool with a good number of “low scoring” punters whom we can send home with confidence.
Other bonuses:
(1) It did not require a super high-sensitive troponin assay (not available in most rural places).
(2) All of the components of the score can be readily available within minutes – history, ECG, troponin. No serum rhubarb to send away to city labs!
(3) It did require a senior, experienced clinician to take the history, look at the patient and the ECG – this is something we DO have in spades in our EDs!
(4) You can use the HEART score with a single “admission” troponin and the low risk group has a 1.7% risk of badness over 6 weeks….
(5) Now – if you repeat the troponin in say 3 hours you increase the sensitivity (NPV) of the score to > 99%.. which is as good as it gets for these chest pain pathways. [Read Mahler et al in Circ Outcome from March 2015]
(6) And you can do all of this in 3.5 hours i.e. still discharge before the dreaded 4-hour target kicks in and you have to “admit” the low-risk chest pain for another troponin after midnight.
So I am sort of looking forward to my first shift back and secretly hoping to see a patient or two with low-risk sounding chest pain.
This score will not replace commonsense, or that sinking-feeling you get when you see the rotund man sweating bullets on the trolley… but it just might help us sort out the low-risk punters in a practical and EBM happy manner!
Let me hear your thoughts – are you going to use the in your practice?
Casey
Clinical Case 090: Unstuffing the Turkey
First10EM Journal Club: December 2018
Paediatric Constipation: with Dr Mel and Liz
About The Author
Casey Parker
I am a GP working in Broome, NW of Western Australia. I work as a hospital DMO (District Med Officer) doing Emergency, Anaesthestics, some Obstetrics and a lot of miscellaneous primary care. Also on the web as @broomedocs | + Casey Parker | Contact
Casey – I use the HEART score frequently to risk stratify patients and determine whether discharge home with outpatient follow up is appropriate. We are currently working on a protocol for our hospital so that others can use the HEART Pathway (Mahler et al) can be implemented.
we are just trying to create a pathway to include the HEART score. Have you done your protocol yet Swami???
Hi Casey,
The potential of HEART, or something like it, is exciting. EDs are screaming out for a way to work up chest pain more efficiently. I don’t want to rain on the parade, but when I considered the evidence not long ago, it just wasn’t convincing enough.
Firstly, I’m concerned about using it in rural practice. The validation studies (like Mahler’s) were done in urban areas with the safety net of outpatient follow-up and functional testing for patients who were discharged by the score (93.7% of Mahler’s patients had stress testing or cardiac imaging – I can’t imagine achieving that in our rural context).
The bottom line for me is it breaks the rules of common sense and biological plausibility:
It is highly subjective with the History component and doesn’t account for the significance of the risk factors (e.g. is Diabetes on insulin equal to positive family history?). It stratifies age into three discrete groups arbitrarily (when we know the risk of MACE increases continuously with age). It allows you to discharge someone with an abnormal Troponin and weights all 5 components equally (is age>45 really the same significance as a positive Trop?).
Looking at the Mahler study. They had a low MACE rate 12/1070 (about 1%) to begin with. This is because Mahler only entered low risk patients into his study. It doesn’t take a much to achieve an event rate of <1% in discharged patients from that baseline using almost any algorithm. NPV is heavily affected by the underlying rate of MACE (pre-test probability) which is why, as a raw number, it looks so good.
Here's the kicker. Of those patients who had MACE (12 of them), 5 (almost half) were classified low risk on the HEART score and discharged early. Here's a quote from Mahler's own article:
"Given the potentially devastating outcome from missing a patient with MACE, high sensitivity is a critical characteristic of a chest pain decision aid. Unfortunately, the exclusive use of the HEART score lacked sensitivity (58.3%) and had only fair diagnostic accuracy (AUC= 0.72)."
One more decision rule that is developed in a “high testing” environment to avoid overtesting, that, if applied in a “low testing” environment would lead to increased testing!
Right now we don’t have a four hour rule in our rural ED in Canada.
Most of our population with chest pain is over 65, and have at least 3 risk factors. Many have had prior stenting.
We routinely send these people home after a negative 6 hour regular troponin and the absence of dynamic serial EKG changes, with a referral to a clinic where they get some further testing (within a few days to two weeks) under the supervision of a cardiologist if their history is moderately or highly suspicious.
Those are scores up to 7 (2 for age, 2 for risk factors, 2 for history, 1 for abnormal baseline EKG) that we are sending home from the Emergency Department: I don’t think the HEART score will help us. I don’t think I’ll be sending home a 44 year old with a moderately suspicious history and a markedly elevated troponin and a normal EKG (score of 3) any time soon either.
The paper you reference includes elective stenting within 30 days as a major adverse cardiac event (MACE). In our practice environment an elective stent after an outpatient work-up is a major rationale to have the clinic! In this study nobody died and only 6% of people had a troponin bump. The authors say no-one had an adverse outcome after the index visit, so the 30 day MACE after discharge was 0!
The paper presents the information poorly, but it appears that there were 7 of 7 NSTEMIs picked up by the serial troponins and 1 PCI during the index hospitalization that was not an NSTEMI that was missed by the serial troponins. If the person did not have an NSTEMI did they really need the PCI? Maybe there were dynamic EKG changes? In effect maybe all the benefit of the score comes from the EKG and the troponin and the other parts of the score are red herrings that lead to unnecessary testing!
If any of the EDs I work in will let me use this, I may well use it.
A comment: The troponin level remains a bit of a mystery to me in this sense – we really have no way of quantifying what the false positive rate is, especially among women. I suspect it is quite high and calling any patient with a high trop a non-stemi (often done) will never help us or them. I think there are probable many women, especially NIDDM sufferers, falsely diagnosed as having had a heart attack. And many of them go on to have at least one angiogram, and most will have some narrow vessels, and many of these have been stented, and most get repeatedly seen in ED with further, probably non-cardiac caused, pains. Men the same but far less so I guess.
I think the delta troponin is useful when you have a small bump and you suspect a non-ACS cause.
Casey
We’ve discussed the HEART score at our journal club, but didn’t think it added anything to our current practice. High risk, older patients are overdiagnosed and the question here is what you can do for them to improve outcomes other than optimize medical treatment. Do they really benefit from being admitted for a day? Young, low risk patients are underdiagnosed. I would not discharge a 44 year old diabetic with a highly suspicious history or a 44 year old without risk factor but an almost 3 fold rise in troponin.
It seems to me that the main utility of the Heart Score is to facilitate early discharge of low to no-risk chest pain by justifying a single or delta 3 hour troponin rather then a delta 6-8 hour troponin. This practice could drastically reduce length of stay, free up beds in the very low risk and save everybody time.