Adjustment of QT dispersion assessed from 12 lead electrocardiograms for different numbers of analysed electrocardiographic leads: comparison of stability of different methods.
OBJECTIVE--Normal electrocardiographic recordings were analysed to establish the influence of measurement of different numbers of electrocardiographic leads on the results of different formulas expressing QT dispersion and the effects of adjustment of QT dispersion obtained from a subset of an electrocardiogram to approximate to the true QT dispersion obtained from a complete electrocardiogram. SUBJECTS AND METHODS--Resting 12 lead electrocardiograms of 27 healthy people were investigated. In each lead, the QT interval was measured with a digitising board and QT dispersion was evaluated by three formulas: (A) the difference between the longest and the shortest QT interval among all leads; (B) the difference between the second longest and the second shortest QT interval; (C) SD of QT intervals in different leads. For each formula, the "true" dispersion was assessed from all measurable leads and then different combinations of leads were omitted. The mean relative differences between the QT dispersion with a given number of omitted leads and the "true" QT dispersion (mean relative errors) and the coefficients of variance of the results of QT dispersion obtained when omitting combinations of leads were compared for the different formulas. The procedure was repeated with an adjustment of each formula dividing its results by the square root of the number of measured leads. The same approach was used for the measurement of QT dispersion from the chest leads including a fourth formula (D) the SD of interlead differences weighted according to the distances between leads. For different formulas, the mean relative errors caused by omitting individual electrocardiographic leads were also assessed and the importance of individual leads for correct measurement of QT dispersion was investigated. RESULTS--The study found important differences between different formulas for assessment of QT dispersion with respect to compensation for missing measurements of QT interval. The standard max-min formula (A) performed poorly (mean relative errors of 6.1% to 18.5% for missing one to four leads) but was appropriately adjusted with the factor of 1/square root of n (n = number of measured leads). In a population of healthy people such an adjustment removed the systematic bias introduced by missing leads of the 12 lead electrocardiogram and significantly reduced the mean relative errors caused by the omission of several leads. The unadjusted SD was the optimum formula (C) for the analysis of 12 lead electrocardiograms, and the weighted standard deviation (D) was the optimum for the analysis of six lead chest electrocardiograms. The coefficients of variance of measurements of QT dispersion with different missing leads were very large (about 3 to 7 for one to four missing leads). Independently of the formula for measurement of QT dispersion, omission of different leads produced substantially different relative errors. In 12 lead electrocardiograms the largest relative errors (> 10%) were caused by omitting lead aVL or lead V1. CONCLUSIONS--Because of the large coefficients of variance, the concept of adjusting the QT dispersion for different numbers of electrocardiographic leads used in its assessment is difficult if not impossible to fulfil. Thus it is likely to be more appropriate to assess QT dispersion from standardised constant sets of electrocardiographic leads.