Small is Beautiful Emergency Medicine
This study, and work that has come since, should be of real clinical value to us as senior EM clinicians – we currently are drilled to refer any ‘new’ LBBB and chest pain, but frequently cannot identify rapidly whether or not the LBBB is new or old. There is also evidence that new LBBB and pain alone being used to activate the cath lab results in ‘false activations’.
The European Society of Cardiology Guidelines say:
For us, activating the cath lab involves an interhospital transfer, so it’s important we make the right decision for our patients. Liaison with our cardiology colleagues is of course still recommended prior to going ahead with transfer, but I’d argue having read this paper that knowledge of the Sgarbossa criteria can help with our clinical confidence and decision making. It also certainly used to frequently come up in FRCEM exams! Here's my appraisal below. I suggest you read the paper and note your thoughts, then compare with mine. Please do comment (top right link) to add to the discussion!
This study is now 22 years old. Using a massive dataset of 26000 MI patients, only 131 had AMI with LBB and enzyme confirmation (CK-MB). The paper developed 3 ECG criteria that, in the context of LBBB, diagnosed acute myocardial infarction pre-enzyme results being available in these patients. The median age of patients was 68.5, 84% were male.
The study undertook blinded analysis of ECGs for these and other criteria (10 in total) – which had been predetermined. 10% of the ECG’s reviewed were seen by all 4 reviewers to look for interobserver variability – little was found. Reviewers were cardiologists, not emergency physicians, so not quite ‘the same’ as us as ECG reviewers.
Chi squared test with ROC curves (I think appropriate for univariate analysis) were used to correlate the enzyme +ve MI’s with the ECG variables. They used a predetermined ‘minimal acceptable specificity’ of 90% to then work out where on the ROC curve gave the best sensitivity. This is perhaps unusual but seems reasonable given need for high specificity of this test.
The study group then used likelihood ratios to identify the ‘best’ criteria from their list.
Then things get really complicated – multivariate analysis, which I’ve never got my head round, but that said the results of this are well and clearly presented in a flowchart.
Results were then validated in a very small (45) sample – but they obtained similar results to the derivation sample. Both derivation and validation studies were done in existing (GUSTO 1,and 2) registries. To further validate this a prospective study would be helpful, which the study group acknowledge.
Since then the Sgarbossa criteria have been modified. If the patient has LBBB, chest pain, and haemodynamic instability or heart failure, they should have coronary angiogram (the logic being the risk of cardiogenic shock is high, even if the LBBB is old). Point 3 has also been altered to being a discordant ST:S ratio of < or = 0.25 (see here). The best explanation I can find of this is from the Life in the Fast Lane link (below) “≥ 1 lead anywhere with ≥ 1 mm STE and proportionally excessive discordant STE, as defined by ≥ 25% of the depth of the preceding S-wave”. Still confusing, though.
I think the adjustment of the 3rd point makes it less easy to apply in clinical practice, though it has been prospectively validated (see here). Sgarbossa apparently supports the modified criteria’s use. What do you think?