IJCS | Volume 33, Nº3, May / June 2020

266 test. The Kolmogorov-Smirnov was used to assess the normality of the continuous variables. To identify the best QT dispersion cutoff point for the diagnosis of obstructive coronaropathy, ROC curves were used both for QT dispersion values and for rest and stress QTc. We calculated the QTd, the QTc, and conventional stress test sensitivity and specificity for the diagnosis of chronic obstructive coronaropathy. We also found a QTc delta – the difference between rest-stress QTc intervals, as well as the QT delta - the difference between rest-stress QTd intervals. Subsequently, patients were divided into three groups: true positive (TP) – patients with positive stress test for ischemia and coronary angiography showing stenosis ≥ 70% of at least one major epicardial artery, except for left coronary trunk lesions which were considered to be significant when the obstruction was > 50%. The false positive (FP) group was composed of patients with positive stress test and stenosis less than 70% in at least one major epicardial artery, except for left coronary trunk lesions which were considered to be significant when the obstruction was < 50%. Finally, the true negative (TN) group was composed of negative stress test patients and coronary angiography showing stenosis less than 70% in any epicardial coronary artery, except for left coronary trunk lesions which were considered to be significant when the obstruction was < 50%. p < 0.05 was considered to be statistically significant and one-way-variance analysis (ANOVA) was used to compare the three groups. A paired t-test was used to assess the QTc behavior during rest and effort. All the patients signed the free and clarified term of consent and the research Project was approved by the Ethics Committee on Human Research of Federal University of Espirito Santo (UFES), by the protocol number: 06177412.1.0000.5071. Results The difference between the mean QTd values obtained by both observers, at rest, was only 0,8 ± 18,3 ms, which is quite satisfactory. However, we observed some data points of standard deviation away from the mean, which demonstrated the low reproducibility of the measures at rest. The difference between the means obtained by the two observers was 9,5 ± 12,5 ms in peak stress measurements. Additionally, we also found data points of standard deviation away from the mean, which confirmed the low reproducibility of QT dispersion for peak stress measures as well. The variability between the two observers was also analyzed using Pearson’s correlation coefficient. For the baseline value of QTd, the value of “r” found was 0.36 (p = 0.25). For the peak QTd, the value of “r” was 0.73 (p = 0.007). The correlation between the measures obtained by the two observers was positive weak at rest and positive moderate on exertion. These results indicate a poor reproducibility of QT interval dispersion measures. Out of the 74 patients initially selected, after we excluded those who presented any of the exclusion criteria, and after the false-positives were removed – because theywere of no interest for the research (negative stress test and stenosis greater than 70% in epicardial arteries, except for trunk lesions which were considered to be significant when the obstruction was > 50%), there were 63 patients who fulfilled the criteria to be in the three research groups: TP, FP and TN. The three groups studied (TP, FP and TN) were similar in relation to general characteristics (Table 1), including comorbidities and medicaments in use. The cutoff value found with the ROC curve was the point where better QTd and QTc sensitivity and specificity for the diagnosis significant coronaropathy were achieved (46ms and 57ms, respectively). Regarding Table 1 - General characteristics of the groups TP (n = 26) FP (n = 23) TN (n = 14) p-value Age (years) 58 ± 10 54 ± 12 56 ± 11 0.43 Male sex (%) 81% 71% 71% 0.453 Diabetes mellitus (%) 27% 13% 7% 0.213 Arterial hypertension (%) 58% 58% 79% 0.372 Beta-blockers (%) 23% 33% 21% 0.633 ACEI/ARB (%) 42% 21% 21% 0.188 Statins (%) 12% 8% 21% 0.199 Calcium antagonists (%) 12% 0% 29% Values expressed as mean ± SD or percentages; TP: true positive; FP: false positive; TN: true negative. Barcelos et al. QT interval dispersion in exercise test Int J Cardiovasc Sci. 2020; 33(3):263-271 Original Article