QT Interval Dispersion 36 Years On: Why the Concept Failed and What Replaced It

In 1990, the difference between the longest and the shortest QT interval measured across the twelve leads of the standard electrocardiogram was proposed as a simple and inexpensive index of the spatial heterogeneity of ventricular repolarization. QT dispersion was adopted rapidly and applied to almost every cardiac and to many non-cardiac conditions and was described by its proponents as a potential Holy Grail for the prevention of sudden cardiac death. Within a decade, however, disappointing results from large studies and a series of methodological and conceptual objections had led others to characterise it as the greatest fallacy in electrocardiography of the 1990s. This narrative review reappraises QT dispersion thirty-six years after its description. We examine four separate lines of criticism: the poor reproducibility of T-wave offset determination and the consequent amplification of measurement error in a parameter of small absolute magnitude; the absence of any demonstrated rate dependence, which invalidates the widespread practice of heart-rate correction; the competing local and vectorial explanations of the phenomenon, together with the experimental and clinical evidence that favours projection over regional information; and the inconsistent prognostic performance of the parameter across major cohort studies. We further argue that the failure of QT dispersion is not a failure of the underlying premise that repolarization heterogeneity is arrhythmogenic, but of the specific surface measurement chosen to capture it, and we summarise the contemporary non-invasive markers that have since been developed for the same purpose. We conclude that QT dispersion should not be interpreted as a measure of the dispersion of repolarization duration, and that its residual association with clinical risk is more plausibly explained by projection effects and measurement inaccuracy than by regional electrophysiological information. QT dispersion is still reported without acknowledgement of these objections. Setting them out together with the markers that have replaced the parameter should help readers judge what it can and cannot support.