I was glad to see this article published as triple rule out CT for chest pain has been a topic that I’ve heard a little about, but personally it’s not something I use. This troubles me as in Virchester we have a long track record in the investigation of cardiac chest pain and this is an area that might interest us in the future, but we need to know more…..
I’m also interested from my role as a clinician. When patients present to the ED with chest pain we have a number of different diagnostic protocols designed to help colleagues perform the right tests in the right patients. We have a protocol for ‘pain compatible with cardiac disease’ and we have a protocol for pleuritic (respirophasic) chest pain. These are evidence based, well designed and defendable approaches to clinical practice, but there are problems. One such problem I see as a senior doc is when juniors push patients into a diagnostic protocol despite clinical presentations that are perhaps equivocal or contradictory for either chest pain pathway. Errors take place as a result of misapplying a good rule toi the wrong patient, a common failing amongst humans. Let’s take an example…
Mr Smith is 65 and presents to the ED with chest pain. He describes it as having started at rest following a session in the local gym where he is trying to get fit. He had done 20 mins on the cross trainer about an hour earlier and he described the pain as mostly in the centre of his chest with discomfort in his upper arm and upper chest. At the time he did feel a short of breath, sweaty and nauseous and the pain appeared to be worse when he took deep breaths, but to be honest the difference wasn’t that much. He mentions that he has only just got back to the gym following a period of rest as a result of an knee sprain 4 weeks ago (that left him pretty immobile for a week or so). He is a smoker and thinks that his maternal aunt died at the age of 38 of an embolus…, or was it a heart attack???
Now, I’m not intending to start another ‘shall we investigate for PE debate’ here. The fact of the matter is that this patient does not have symptoms that fit into any particular diagnostic box. It’s a mixed picture and as a result I see docs push patients into one diagnostic strategy or another. Picking the ‘wrong’ one might end in disaster……, so wouldn’t it be nice if there was a test that would tell us everything we might want to now about chest pain. Perhaps a triple scan that would look for everything that we need to know about…… coronary artery disease, aortic dissection and PE all in one rapid and convenient test. Now wouldn’t that be lovely 🙂
[DDET What’s a triple rule out scan anyway?] Basically, it’s a CT scan that images the coronary, aortic and pulmonary vasculature to look for coronary artery disease, pulmonary embolus or aortic dissection. This presents some technical difficulties in optimising the scan process to visualise all those areas whilst contrast is passing through different structures at different times. Whilst I understand this principle I would leave the techy details to our radiological chums. There’s a nice review article from Medscape here that explains in more detail how and what TRO CT is.
So a test looking for three rather important diagnoses at the same time has, at first look, some potential benefits. However, the technical elements of the scan, and other concerns means that we cannot just assume that this is fantastic. As EBM docs we should look for evidence that the test improves diagnosis, and that the improvement in diagnostic performance leads to an improvement in patient outcome.[/DDET]
[DDET What kind of paper is this?”] This is a systematic review and meta-analysis of a diagnostic test. It’s worth stopping at this point and asking yourself when you last saw one of those…….., you may struggle to remember the last as they are really not that common (Ed – do you not remember this one?). Diagnostic studies are often difficult to group together in a way that is meaningful during meta-analysis owing to the heterogenicity of the trials and the patients within them.
In a systematic review/MA the subjects are the papers and so we need to look carefully at how the authors selected the papers. This should include a selection policy based on whether the paper is relevant to the question asked, and also on the quality of the papers themselves.
Selection was based on finding papers that compared TRO scans against ‘another diagnostic modality’, which is fair enough, although I’m really interested to know about performance against a similar gold standard for the final diagnosis. I’m especially interested in understanding what it is we are actually looking for and what the gold standard is….., more of this later.
The search is extensive and well advised with expert librarian help (a good thing). They have also explicitly graded the papers using an objective scoring system QUADAS-2, again this is a good thing to do as objective scoring systems should be more reliable than authors impressions and opinions.[/DDET]
[DDET So what are we looking for in this study?] It’s essential that a diagnostic test has a clinically relevant outcome. In this study the authors have identified 4 areas that they consider important in the evaluation of TRO in the acute chest pain setting. These are
- Contrast volume
- Image quality
- Radiation exposure
- Diagnostic accuracy
Now, personally, I think one of those trumps the other three (DA) and arguably none of them are really patient outcomes. What I really want to know is whether use of this test is good for patients and whilst these measures are important they are to some extent proxies to true patient outcomes.[/DDET] [DDET What about the clever meta-analysis stuff….?] The stats of Meta-analysis are a little impenetrable to be honest. In this study they have used a random effects model which is appropriate as they are looking at different treatment effects. Having said that I find that the most important test as a reader is to look at the studies and ask myself whether grouping them together is appropriate, wise and sensible. The more varied the studies the less valid the analysis will be regardless of any statistical jiggery-pokery. [/DDET]
[DDET What papers did they find?] Of 733 papers scanned a final number of 11 were included. This included just one randomised controlled trial. A total of 3539 patients were included across all trials with 791 undergoing TRO scans.
Apart from the one RCT the authors found a selection of case control and cohort studies of variable quality, though there were no high quality papers according to QUADAS-2 scores.
Only 4 studies looked at diagnostic accuracy of TRO, using coronary angiography as a gold standard. The pooled diagnostic accuracy estimates in these studies were sensitivity of 94.3% (95% CI=89.1% to 97.5%) & a specificity of 97.4% (95% CI=96.1% to 98.4%). This sounds impressive doesn’t it? They also picked up 8 PEs and no aortic dissections. All the patients with PE were suspected as having a PE – which kind of makes you question how much benefit the ‘triple’ element of the scan really is.[/DDET]
[DDET Er, but are these ED patients in the EM Journal?] That’s a fair question. I’m reading the paper in Acad Emerg Med so I’ve lulled myself into the impression that this paper is about management of patients in the ED, but is it? In the text I’m struggling to find out whether the patients in these trials are actually patients like the ones I see in the ED. That’s crucial to my interpretation and understanding of what’s going on here.
So, back to the paper to find out more about the setting and patient types……, and there’s not that much detail really. I’m still a bit confused and so I’ve resorted to checking the originals out myself. Of the 4 papers looking at diagnostic accuracy (the big outcome for ED patients).
- Schuchlenz paper: We evaluated 72 consecutive chest pain patients from our emergency department/cardiac outpatient clinic with a low to intermediate likelihood of coronary. Interestingly I can only find this paper as a conference abstract online (this may be my searching prowess…..). Gold standard is unclear but appears to be coronary angiography.
- Johnson paper: All patients with acute chest pain were eligible for the study if referred by a colleague after initial diagnostic workup including physical examination, ECG and serum levels of creatinine and TSH. So not really an unselected group such as I see in the ED in Virchester. 58 patients in this study. Selective use of gold standard (angiography)
- Litmanovitch paper: 56 patients in this study with a normal or equivocal ECG and negative cardiac enzymes. Selective use of gold standard which was itself variable.
- Johnson paper: 55 patients with acute chest pain referred by cardiologists or emergency physicians with chest pain of uncertain origin. Patients were only referred if the diagnosis was in doubt. 20 cases had confirmatory angiography. Patients followed up for 5/12
[DDET The authors say the TRO can rule out coronary artery disease though….]I’m not so sure. The authors state that…‘..if a TRO CT is obtained in a patient with nontraumatic chest pain and no significant coronary stenoses are identified, ACS is effectively ruled out’
I am very reluctant to support this as in terms of diagnostic accuracy we have not seen any studies that meet criteria for a good diagnostic study.
Ed – what simple criteria are you using here?
An easy test for a diagnostic study is to ask if it has the following features…
- Was there an independent, blind comparison with a reference standard
- Was the diagnostic test evaluated in an appropriate spectrum of patients
- Was the reference standard applies regardless of the test result?
- Was the test validated in a second group of patients?
These are based on David Sackett’s work but they are features of many diagnostic checklists.
So, basically we can say….no. These studies do not meet criteria for a robust diagnostic study and I think the main conclusion of this study is ‘courageous’ if based on the data I can find in tracking back to the original papers.