Modified Sgarbossa criteria for myocardial infarction in left bundle branch block

Clinical background

Left bundle branch block and ventricular paced rhythm distort ventricular repolarisation to such a degree that the usual STEMI criteria cannot be applied. An acute coronary occlusion in a patient with left bundle branch block may be difficult or impossible to identify on the ECG without specific criteria, and delay to reperfusion therapy is associated with increased mortality. The modified Sgarbossa criteria serve this purpose: they identify patients who very probably have an acute coronary occlusion and who should therefore proceed to emergency reperfusion therapy.

The original criteria, published in 1996 from the GUSTO-1 database, used a weighted point scale in which the third criterion, absolute discordant ST elevation ≥5 mm, had low sensitivity and contributed to the underperformance of the whole rule in clinical use. The modified criteria replace this absolute measure with a proportional one: the magnitude of the ST elevation in relation to the QRS amplitude.

Applying the modified Sgarbossa criteria

The criteria are unweighted: at least one of three conditions must be met for the result to be positive.

1. Maximal concordant ST elevation (any lead, positive QRS) ≥1 mm, that is, ST elevation in a lead in which the QRS complex is predominantly positive.
2. Maximal concordant ST depression in leads V1–V3 ≥1 mm, that is, ST depression in leads in which the QRS is negative, where the depression runs in the same direction as the negative QRS vector.
3. Discordant ST elevation with an ST/S ratio ≤ −0.25 in the lead with the most pronounced discordant ST elevation, where the depth of the S wave in the same lead forms the denominator:

$$\text{ST/S ratio} = -\frac{\text{ST elevation (mm)}}{S\text{-wave depth (mm)}}$$

A ratio ≤ −0.25 means that the ST elevation constitutes at least 25% of the amplitude of the S wave. This replaces the original criterion of ≥5 mm discordant ST elevation irrespective of QRS size.

The derivation cohort consisted of 33 patients with angiographically verified acute coronary occlusion and left bundle branch block together with 129 controls presenting with chest pain or dyspnoea and left bundle branch block without occlusion, collected at three hospitals in the USA [1]. The reference standard was angiographically demonstrated coronary occlusion.

Interpretation in practice

A positive result multiplies the probability of acute coronary occlusion approximately ninefold (LR+ 9.0 in the derivation cohort) [1]. A negative result lowers the probability with an LR− of approximately 0.1, which is useful but not sufficient to exclude occlusion reliably where clinical suspicion is high.

Validation and performance

In the derivation study the modified criteria achieved a sensitivity of 91% (95% CI 76–98%) at a specificity of 90% (95% CI 83–95%) [1]. This was a marked improvement over the original criteria, which had a sensitivity of 52% (weighted) and 67% (unweighted) at comparable or lower specificity. The improvement was almost entirely due to the third criterion: discordant ST elevation with an ST/S ratio ≤ −0.25 was present in 79% of patients with occlusion versus 9% of controls, whereas the absolute criterion of ≥5 mm was present in only 30% versus 9%.

For ventricular paced rhythm, the criteria were validated in a multinational retrospective study across 16 centres (the PERFECT study), comprising 59 patients with occlusion myocardial infarction and 192 controls [2]. Sensitivity was 81% (95% CI 69–90%) and specificity 96% (95% CI 90–99%) against uncomplicated emergency department controls. Against a more demanding control group of patients who underwent angiography for suspected type 1 myocardial infarction but had no occlusion, specificity was 84% (95% CI 76–91%). Adding concordant ST depression in V4–V6 increased sensitivity to 86% (95% CI 75–94%).

A Swedish external validation at five emergency departments included 623 patients with left bundle branch block or ventricular paced rhythm and chest pain, of whom only 15 had occlusion myocardial infarction (prevalence 2.4%) [3]. Here sensitivity fell to 60% and specificity to 86%. The positive predictive value was low (9.4%) because of the low prevalence, whereas the negative predictive value exceeded 98%. In a larger Swedish study of 24,511 consecutive chest pain patients at five emergency departments, an expanded criteria set that included the modified Sgarbossa criteria for left bundle branch block and ventricular paced rhythm achieved a sensitivity of 41% at a specificity of 95% [4]. This reflects the fact that a substantial proportion of patients with occlusion are not captured by ECG criteria even when special rules for bundle branch block and paced rhythm are applied.

Limitations

The criteria apply only to patients with left bundle branch block or ventricular paced rhythm. With a normal QRS duration or right bundle branch block, the usual STEMI criteria should be used.

Sensitivity is not 100%. In the derivation cohort approximately 9% of occlusions were missed, and in external validations the miss rate was higher: 19% in the paced rhythm study [2] and 40% in the Swedish validation [3]. A negative result does not exclude occlusion and should not halt further investigation if clinical suspicion persists.

The positive predictive value depends strongly on prevalence. In an emergency department population with a low prevalence of occlusion myocardial infarction, as in the Swedish validation, most positive results may be false positives [3]. Conversely, in a patient with typical ischaemic chest pain and a high clinical probability, a positive result is considerably more reliable.

The ST/S ratio requires manual measurement of both the ST deviation and the S wave amplitude in the same beat. Sources of error include an incorrectly chosen J point, the influence of T waves, and repolarisation abnormalities not due to ischaemia. The relative measurement is nonetheless less sensitive to error than the original absolute criterion, particularly with deep S waves, where small ST elevations may be normal.

External validation in a low-prevalence emergency department population

The Swedish validation [3] shows that the modified Sgarbossa criteria perform less well in a Swedish emergency department population with a low prevalence of occlusion myocardial infarction than in the original American derivation cohort. A low positive predictive value at low prevalence means that in the Swedish setting a positive result should be considered together with the overall clinical picture and should not on its own trigger activation of the catheterisation laboratory. In a Swedish study of AI-based ECG interpretation, neither the standard STEMI criteria, the expanded criteria including the modified Sgarbossa criteria, nor computerised interpretation achieved more than 41% sensitivity for occlusion myocardial infarction in a consecutive emergency department population [4]. This underlines that ECG criteria alone are not sufficient and that clinical assessment, serial ECGs and troponin values remain central in Swedish practice.

References

1. Smith SW et al. Diagnosis of ST-elevation myocardial infarction in the presence of left bundle branch block with the ST-elevation to S-wave ratio in a modified Sgarbossa rule. Ann Emerg Med 2012. PMID: 22939607
2. Dodd KW et al. Electrocardiographic Diagnosis of Acute Coronary Occlusion Myocardial Infarction in Ventricular Paced Rhythm Using the Modified Sgarbossa Criteria. Ann Emerg Med 2021. PMID: 34172301
3. Lindow T et al. Comparison of diagnostic accuracy of current left bundle branch block and ventricular pacing ECG criteria for detection of occlusion myocardial infarction. Int J Cardiol 2024. PMID: 37931659
4. Lindow T et al. Improved Detection of Acute Coronary Occlusion Myocardial Infarction by an Artificial Intelligence Electrocardiogram Model in Swedish Emergency Departments. J Am Coll Emerg Physicians Open 2026. PMID: 42614578