Diagnosis of Acute Aortic Syndrome in the Emergency Department – The DAShED Study

Estimated reading time: 9 minutes

Acute Aortic Syndrome

Acute aortic syndrome is the absolute bane of the emergency physician’s existence. A rare, but incredibly dangerous diagnostic needle in a chest pain haystack. Currently, our only really reliable diagnostic test is a CT aorta angiogram, but just how many of these do we need to do to pick up one case of dissection?

In this post, we discuss a paper published in the Emergency Medicine Journal^​1​ looking at current practice in the UK and examining the use of clinical decision rules alongside clinical gestalt to diagnose AAS. Based on the very long list of collaborators at the end of the paper, I am sure many of you have not only heard of this trial but actively took part.

Introduction

Acute aortic syndrome (AAS) includes not just aortic dissection, but also intramural haematoma, and penetrating aortic ulcers. Chest pain is the most common presenting symptom (in 80%), which itself accounts for 7.6 million visits to Emergency Departments in the UK. There are about 4000 cases of AAS in the UK per year. Even the less able statisticians amongst us can see the numerical challenge this represents.

However, I can pretty much guarantee that every ED up and down the country has at least one tale of AAS being ‘missed’ and a patient coming to harm. It’s right up there with pulmonary embolism and subarachnoid haemorrhage as diagnoses we hate and sometimes (often) lose sleep over.

What is this study about?

In this study the authors’ primary objective was to look at the usefulness of previously described clinical decision tools – the ADD-RS^​2​, Canadian guideline^​3​, and AORTAs^​4​, with secondary objectives to establish rates of CT aortogram and patient characteristics.

What kind of paper is this?

This is a multicentre observational cohort study of patients across 27 UK Emergency Departments, where eligible patients were enrolled in a two-month period in autumn 2022. Some EDs recruited for the whole period, some recruited for much less. Three centres (Derby, Reading and Wexham Park) each recruited over ten percent of the total number. Recruitment could occur prospectively (37%), or retrospectively (48%) via daily searches of computer systems for patients presenting with symptoms that could relate to AAS or those who had aortic imaging performed.

Tell me about the patients.

Patients over the age of 16 with new-onset (within seven days) symptoms of possible AAS were eligible. These symptoms included chest, back or abdominal pain, syncope, or symptoms related to malperfusion. Remember that this was a pragmatic approach so it is likely that not all patients with these symptoms were included, and prospective inclusion would’ve very much had depended on the enthusiasm of clinicians or the fastidiousness of the research team.

What about the outcomes

For those prospectively recruited, clinicians were asked to fill in a six-page case report form. Data entry was via simple box-ticking for over 40 different questions. The base specialty of the clinicians and their level of seniority isn’t recorded. Although we would like to assume these unwell patients are being treated by senior emergency clinicians this may not be the case.

Crucially, clinicians were asked to rate their certainty of the diagnosis of AAS, whether it was the most likely diagnosis and if not what they thought the most likely diagnosis was. This was ideally done before any imaging had taken place. This was obviously not possible for the patients that were recruited retrospectively.

The study endpoints were the proportion of patients who had confirmed AAS and the ED clinician thought the diagnosis was possible, most likely or not possible; the performance against clinical decision rules; the rate of ordering of cross sectional imaging and the positivity rate (whether confirming AAs or another diagnosis); time based factors and 30 day mortality.

Where comparison was made with recognised decision rules a confirmed AAS diagnosis on CT or at surgery was taken as the reference standard.

What did they find?

The headline figures are perhaps some of the most interesting:

5548 patients presented in the study period with symptoms potentially attributable to AAS
Median age was 55, with broadly equal numbers across the sexes.
1082 patients had AAS considered as a possibility
407 patients had CT aortograms
47 patients had aortic pathology on CT with 14 patients having confirmed AAS (5 type A and 3 type B of the dissections) and 33 having other pathology like aneurysms (9 of which had ruptured)
Median time from attendance to confirmation of AAS was 6 hours (with a range from 2 hours to 11 days(!))

In terms of the various clinical decision rules then the analysis was compromised by missing data, a lack of a universally applied gold standard and small numbers. The authors have presented the data in the extensive supplementary material and whilst interesting and hypothesis generating it is good to see that the authors have not done much more with it than that. None of the scores, or combinations of score +/- clinical acumen or d-dimer reach sensitivities or specificities that we would expect to see with a high quality clinical decision tool.

What does this mean?

Diagnosing acute aortic syndrome is really hard. Like really tricky. We ‘overinvestigate’ to try to pick up a vanishingly small number of cases. Under 3% of all the CT aortograms done in this study looking for AAS were positive. Not wishing to state the obvious, but that means that 97% of these scans are negative. It’s clear that plenty of people are indeed ‘thinking aorta’. It is a diagnosis that seems to haunt us all. Of course, this study doesn’t tell us about ‘missed’ cases – those where the diagnosis was not considered.

Where the diagnosis was ‘obvious’ clinicians generally got this right (with a specificity of 97%), but with a disappointing sensitivity of only 45%. This does still give a likelihood ratio of 15, which we should rightly be proud of. In the more tricky cases, where AAS is a ‘possible’ the positive likelihood ratio is 5.9 – not quite as good, but still not bad.

The low prevalence of disease in the study population makes these calculations and the comparison to clinical decision rules tricky. After all in a population where the prevalence of disease is so low, if you just said none of them has AAS (or even any aortic pathology) you would be right over 95% of the time.

Conclusion

Despite the limitations of this study, which the authors acknowledge, this is an important piece of work. Being a clinician in an Emergency Department is an incredibly challenging role, and I think this study supports that. In amongst the mass of humanity attending we are expected to pick up these rare and life-threatening diagnoses, for which there are sometimes no really reliable clinical symptoms or signs, whilst simultaneously making sure waiting times are kept down and patients are ‘moved through’. Frustratingly then, when these diagnoses are ‘missed’ often with devastating consequences, not just for the patient and their loved ones, but for the whole clinical team, a critical retrospectoscope is applied to the individual case, without looking at the wider picture.

I once heard a well-renowned US Emergency Physician say that we should never make the diagnosis of aortic dissection, because in the quest to find the very rare case we would subject the vast majority to un-needed investigations, concern and worry, but of course, the world does not work like that (and neither do we).

The investigators’ final statement is that “The best decision aid to facilitate the decision to CT and to outperform clinician gestalt in AAS is not yet clear”.

I would suggest something more radical – often clinician gestalt is the best diagnostic test and we should be nurturing our emergency physicians as our most valuable diagnostic tools. However, clinician gestalt comes with experience and time, and it is unclear whether the opinions given in this paper were from trained emergency physicians/clinicians, or from other staff members who be more junior and on more general rotations. What is clear from this paper is that there is no easy solution to this complex diagnosis, and that for now we must support and develop our clinicians to seek it out. In reality that probably means education, supervision, support and review of cases^​5​. At St Emlyn’s we are big advocates of peer review ^​6​and this is yet another area of development where it may help.

CT scanners and troponin analysers get regular servicing and software updates to ensure they are working at their best, yet our most reliable and effective testing machines (us) are lucky to get a decent lunch break. In complex systems seeking complex conditions, it would be unlikely for us to find simple solutions, but there is always hope that we can find better screening and diagnostic processes in future.

References and further reading.