Another Friday another JC in Virchester. Lots of chin stroking and pontification on this one.
We looked at the ADAPT trial by Martin Than and colleagues. Essentially they assessed the diagnostic utility of an accelerated protocol using Timi risk score, 0 & 2 hour troponin I measurements for rapid exclusion of cardiac pathology in patients attending the ED with chest pain. I was rubbing my hands on this one. “think of the bed days we can save on the decision unit. Say goodbye to all the 12 hour boredom of the ‘rule out’ MI admission.”
So what did they actually do? Well, over 2 sites in Oz and NZ they assessed 1975 patients attending with chest pain of suspected cardiac origin. They performed baseline and 2 hour Troponin I measurements that were unavailable to clinicians. They performed usual diagnostic protocols and followed patients up for 3 months. Then an adjudication panel apportioned the diagnosis of Major Adverse Cardiac Events at the end of the study period. They subsequently looked back to ask 2 main questions:
Firstly, what was the sensitivity of their accelerated diagnostic protocol (did it catch all the true positives)?
Secondly, if we assume the ADP is highly sensitive and therefore ‘safe’ how many patients would have been potentially dischargeable directly from the ED within 4 hours?
The methodology was pretty sound. Pretest probability for MACE was a little on the low side (15.3%) compared to our prevalence (20%) or other studies (30%). But their use of MACE is in keeping with much of the literature on the topic. The adjudication committee was not independent but we had a long discussion about this and concluded that actually, if they were blind to the Trop I assay, they couldn’t introduce much bias here even if they wanted to. If they underestimated MACE based on some of the more subjective aspects for example, they could potentially heighten the sensitivity of the diagnostic protocol – but this would worsen specificity and consequent diagnostic utility. Thus they would be shooting themselves in the foot.
And what did they find?
An excellent sensitivity of 99.7% (95% CI 98.1 – 99.9)
A dischargeable proportion of 20% using the ADP, with 1 MACE event in this group only (0.25%).
These are very interesting findings from a large, methodologically robust diagnostic study from 2 countries with a distinguished author list. So, are we going to start applying the findings in practice? I’m not so sure. The findings must be taken in context with other concerns about the paper. We only had a few issues, but here they are:
- There was no mention of symptom duration at presentation in the cohort. Thus we don’t know how long these patients had had symptoms for prior to presentation. This is important: if most of the 1975 presented at 10 hours then actually, your 2 hour Troponin is predominately a 12 hour Troponin if you use time from symptom onset for your exclusion. We do Troponins 12 hours after symptom onset, but other systems (for example in many American papers) a measure of 12 hours after ED presentation is used.
- Their findings have not yet been externally validated. The sensitivity especially is likely to be slightly overestimated in a tightly controlled exploratory cohort. Something to be cautious of.
- They used Trop I rather than one of the new all reaching brand new spanking best in show highly sensitive troponin assays. This might mean that the findings are more immediately relevant if you’re not using a high sensitivity assay, but for us in Europe this is a concern. The proportion of patients who could be discharged is likely to be lower with a high sensitivity assay.
- The proportion of dischargeable patients using this rule was actually not huge at 20%. In the UK, this may be even less (only 7.4% of our patients had a TIMI risk score of 0/7 in previous work). It follows that the others all end up admitted to some sort of inpatient service. This is a fairly self fulfilling prophecy in terms of investigations and therapeutics, as noted by their figures of 74% ADP negative patients receiving further investigations and 14% recieving therapy of some form. There are risks with this. And remember, the actual prevalence is only 15.3% (increased to 18-19% if we discount the ADP positive patients). 60% are still getting a raw deal (full inpatient investigation and treatment with no clear diagnosis at the end) Are we potentially therefore doing more harm by using a strict diagnostic protocol such as this one that will no doubt encourage admission due to the limited specificity?
The accompanying editorial to this paper can be found here and is a really nice summary of how far we have come and how much work still needs to be done on this topic. It also highlights the recent work looking at undetectable HsTnT to exclude MI at the door by a certain Dr Body (who?) and Professor Carley (never heard of him). This kind of work interests me a little more if I am honest.
I want a test which I can use immediately on the young patient with low risk chest pain that will rapidly reassure me about an absence of cardiac pathology. I am happy that Troponin is not this test in the elderly and will often manage them as ACS irrespective of initial trop results.
But if you can find me a way of getting all those 30-40 year olds who wait 12 hours only to be told that they haven’t had a heart attack (no s***) but that we’re not really sure what’s wrong with them, off my decision ward, then I will applaud you.
Carry on the good work sirs.