Localization of accessory pathways by the electrocardiogram: which is the degree of accordance of three algorithms in use?

We evaluated the extent of agreement among three algorithms used for the localization of accessory pathways in patients with overt preexcitation. By the use of one algorithm, three independent couples of observers localized the accessory pathway in 95 consecutive patients showing overt preexcitation in the 12-lead surface electrocardiogram. We defined the following regions: Left atrioventricular ring (LAVR), Right atrioventricular ring (RAVR), Left lateral/left anterolateral (LL/LAL), Left posterior/left posterolateral (LP/LPL), Left posteroseptal (LPS), Right midseptal (RMS), Right posteroseptal (RPS), Right posterior/right posterolateral (RP/RPL), Right lateral/right anterolateral (RL/RAL), and Right anterior/right anteroseptal (RA/RAS). The extent of agreement in each region was evaluated and compared with the expected one, as calculated from the reported. The extent of agreement was as expected: (1) high in the regions LAVR, RAVR, LL/LPS and (2) limited in the regions LPS, RPS, and (3) clearly lower than expected in the regions LP/LPL, RA/RAS, RMS, RL/RAL. In cases with total or partial disagreement, the number of electrocardiograms with duration of QRS complex smaller than 120 ms was greater than in cases with total agreement (30/46 vs 22/50, P < 0.05). The observed agreement among algorithms is clearly lower than the expected one. Minimal preexcitation, limited number of patients, and arbitrarily defined regions were possibly the reasons for some unexpected results.