Statistical modeling is commonly used in both predictive and explanatory studies in health research. Its use in Cuba continues to grow, although it is sometimes employed inappropriately, which can lead to errors that imperil validity. This article attempts to shed light on faulty practices in statistical modeling by examining and discussing the main differences between explanatory and predictive models, with reference to the following: study objectives, theoretical considerations in model-building, aspects requiring assessment, variable and algorithm selection, analysis of confounders, treatment of multicollinearity, and reporting results.

KEYWORDS Prognosis, risk factor, protective factor, causality, statistical models, linear models, predictive models, explanatory models, logistic regression, Cuba

INTRODUCTION EKG remains a highly valuable tool for heart disease management. Corrected QT interval dispersion is a useful EKG parameter to assess prognosis in ischemic heart disease and specifically acute coronary syndrome. Understanding QT interval physiopathology helps assess importance of QT measurement in this context. Although increased QT dispersion is an ominous prognostic marker, its utility has not been evaluated for all types of acute coronary syndrome, even though in many circumstances it is the only tool available for diagnosing patients with equivocal EKG signs and/or atypical symptoms.

OBJECTIVE Describe corrected QT interval dispersion in acute coronary syndrome in three groups of patients—with ST elevation, without ST elevation, and without ST elevation with equivocal EKG signs—admitted to the intensive care unit of Celestino Hernández Robau University Hospital in Santa Clara, Cuba, from January 2010 through June 2011.

METHODS A descriptive retrospective study was conducted in 194 patients admitted with diagnosis of acute coronary syndrome. QT interval was measured and its dispersion calculated for the first EKG after symptom onset. Patterns were assessed for typical and atypical clinical presentations, and unequivocal and equivocal EKG signs.

RESULTS Nonclassifiable acute coronary syndrome was found in 6.7% of patients (13/194), the majority of whom had increased QT dispersion (76.9%, 10/13). There were significant differences in QT dispersion patterns between patients with typical and atypical presentations and between patients with equivocal and unequivocal EKG findings. In non-ST elevation acute coronary syndrome and nonclassifiable acute coronary syndrome with increased dispersion, atypical presentation was the most common (65.5%, 19/29; and 90%, 9/10, respectively).

CONCLUSION Corrected QT interval dispersion is a useful diagnostic tool for acute coronary syndrome, especially when patients present with atypical symptoms and equivocal EKG findings. Thus, it is a low-cost alternative in management of acute coronary syndrome in resource-poor settings.

KEYWORDS Risk factors, ischemic heart disease, acute coronary syndrome, myocardial ischemia, electrocardiography, QT interval dispersion, prognosis, Cuba