A new statistical appraisal of baseline ECG features for localization of accessory pathways

In order to develop a better approach to accessory pathway localization, the authors analyzed 18 separate electrocardiographic characteristics in 93 consecutive patients. All patients had successful radiofrequency catheter ablation of a single accessory pathway causing preexcitation, prolonging QRS duration to > 100 ms, with 56 pathways ablated on the right side of the heart and 37 ablated on the left side. Eight individual locations were specified at the time of ablation, five right sided and three left sided. Initially, a simple statistical approach was used, with group mean statistical comparison for both delta-wave frontal axis and QRS horizontal transition lead being compared between adjacent locations by Student's t-test. Overlap between individual values was assessed by calculating the percentage of total pathway numbers in each of the two adjacent locations being compared that lay within an arc of 2 SDs in the direction of the location being compared. The statistical significance of all 18 variables was tested using univariate analysis and oneway analysis of variance for continuous variables and the chi-square test for categorical data. Post hoc tests performed were Scheffe's test for the analysis of variance and Bonferroni's test for the chi-square test. All electrocardiographic characteristics were then reanalyzed in a stepwise linear discriminant analysis using BMDP statistical software. A jack-knifed classification was made to assess the accuracy of classification. The statistical level for a characteristic to enter the model was set at P < .05 and the statistical level for a characteristic to be removed from the model was set at P > .10. Six linear discriminant analyses were performed: (1) right side versus left side, (2) right free wall versus right septum, (3) right anterior free wall versus right posterior free wall, (4) right anterior septum versus, right midseptum versus right posterior septum, (5) left anterior pathway versus left posterior pathways, and (6) left posteroseptal versus left posterior free walls.(ABSTRACT TRUNCATED AT 250 WORDS)