A Bayesian approach to the differential diagnosis of wide complex tachycardia

We use sensitivity, specificity, and likelihood ratios (LR) to interpret lab tests all the time. This is known as Bayesian analysis. It is an essential part of the clinical reasoning we do everyday. We  don't ordinarily apply Bayesian analysis to ECG interpretation but the concept is valid. A review published in 2000 outlines the Bayesian approach to the diagnosis of wide complex tachycardia, linked here:

The Bayesian approach improves the electrocardiographic diagnosis of broad complex tachycardia

Like any other uses of the Bayesian approach, one starts with the pretest probability which can be converted into prior odds. We have data on the likelihood ratios of various ECG findings in wide complex tachycardia. To apply a Bayesian analysis one must first determine the pretest probability (or prior odds). The authors posit prior odds of 4 (80% pretest probability) as appropriate for most cases, based on data from a study of unselected patients. They acknowledge that different prior odds may be applied based on clinical judgment and published data in selected patient groups. For example, in an adolescent with WCT a supraventricular mechanism is more likely with a probability of VT of around 40% in published studies, equated to prior odds of 0.67.

Table 2 from the paper, showing various  ECG findings and their likelihood ratios, is shown here.

Pretest odds are multiplied by the LRs of various findings sequentially. The final product is the post test odds. Post test odds of greater than or equal to 1 lean toward  VT. Odds of less than 1 point to a supraventricular mechanism. The further the number is from 1 the greater the strength of the diagnosis. 

When should a Bayesian approach be used as opposed to a conventional approach using algorithms and scoring systems? In WCT with bizarre QRS morphology measurements may be difficult causing problems in application to algorithms. A Bayesian approach may be advantageous in those situations. Moreover, for teaching and learning purposes, Bayesian  analysis provides a good exercise in clinical reasoning as applied to electrocardiography.