Heart Rate Influence on the QT Variability Risk Factors

Abstract

QT interval variability, mostly expressed by QT variability index (QTVi), has repeatedly been used in risk diagnostics. Physiologic correlates of QT variability expressions have been little researched especially when measured in short 10-second electrocardiograms (ECGs). This study investigated different QT variability indices, including QTVi and the standard deviation of QT interval durations (SDQT) in 657,287 10-second ECGs recorded in 523 healthy subjects (259 females). The indices were related to the underlying heart rate and to the 10-second standard deviation of RR intervals (SDRR). The analyses showed that both QTVi and SDQT (as well as other QT variability indices) were highly statistically significantly (p < 0.00001) influenced by heart rate and that QTVi showed poor intra-subject reproducibility (coefficient of variance approaching 200%). Furthermore, sequential analysis of regression variance showed that SDQT was more strongly related to the underlying heart rate than to SDRR, and that QTVi was influenced by the underlying heart rate and SDRR more strongly than by SDQT (p < 0.00001 for these comparisons of regression dependency). The study concludes that instead of QTVi, simpler expressions of QT interval variability, such as SDQT, appear preferable for future applications especially if multivariable combination with the underlying heart rate is used.

Keywords: QT variability; QT variability index; RR variability; sequential analysis of regression variance; underlying heart rate.

Figure 1

Values of measured indices. For each of the investigated indices (see the labels of the horizontal axes of individual panels) the figure shows the cumulative distributions of intra-subject means calculated over electrocardiograms (ECGs) with heart rate between 50 and 75 bpm (full lines) and between 75 and 100 bpm (dashed lines). The red and blue lines correspond to female and male subjects, respectively.

Figure 2

Relationship to age. Scatter diagrams of the relationship between age of the study subjects (horizontal axes) and RR interval coefficient of variance (top panels), QT coefficients of variance (middle panels), and QT variability index (bottom panels). The panels on the left and on the right show the relationship of age to intra-subject means calculated over ECG with heart rate between 50 and 75 bpm and between 75 and 100 bpm, respectively. In each panel, the red circles and blue squares correspond to female and male subjects, respectively. The solid red and solid blue lines show the linear regressions between the ages and the intra-subject mean values. The red shaded and blue shaded areas are the 95% confidence bands of the regression lines; the violet areas are the overlaps between the confidence bands of the sex-specific regressions.

Figure 3

Intra-subject coefficients of variance. For each of the investigated indices (see the labels of the horizontal axes of individual panels) the figure shows the cumulative distributions of intra-subject coefficient of variance of the given index calculated over ECGs with heart rate between 50 and 75 bpm (full lines) and between 75 and 100 bpm (dashed lines). The red and blue lines correspond to female and male subjects, respectively.

Figure 4

Intra-subject rank correlations. For each of the investigated indices (see the labels of the horizontal axes of individual panels) the figure shows the cumulative distributions of intra-subject Spearman rank correlation coefficients (calculated over all available ECGs in the given subject) between the given index and the underlying heart rate (full lines), standard deviation of RR intervals (dashed line), and standard deviation of QT intervals (dotted line). The red and blue lines correspond to female and male subjects, respectively.

Figure 5

Inter-subject relationship of the measured indices. Scatter diagrams of the inter-subject relationships between intra-subject means of SDRR and SDQT (panels on the top), RR_cvar and QT_cvar (panels in the middle), and SDQT and QT variability index (panels at the bottom). Panels at the left and on the right show the indices calculated over ECGs with heart rates 50–75 bpm and 75–100 bpm, respectively. In each panel, the red circles and blue squares correspond to female and male subjects, respectively. The solid red and solid blue lines show the linear regressions between the intra-subject mean values of the compared indices. The red shaded and blue shaded areas are the 95% confidence bands of the regression lines; the violet areas are the overlaps between the confidence bands of the sex-specific regressions.

Figure 6

Sequential analysis of the regression variance. Results of the sequential analysis of the regression variance (see the text for details). Each panel corresponds to a given linear regression estimate (see the formulas at the top right of the panels) and shows the relative residuals, that is the proportions (e_A − e_AB)/e_A or (e_AB − e_ABC)/e_AB as explained in the text. Scatter diagrams in the separate panels show the relationship between relative residuals of predictors used in the multivariable regression. The labels of the axes have the form “Relative A residuals” or “Relative A+B residuals” meaning the proportions (e_A − e_AB)/e_A or (e_AB − e_ABC)/e_AB. In each panel the dashed line shows the line of identity. In each panel, the red circles and blue squares correspond to female and male subjects, respectively (note that the multivariable linear regressions and their residuals regressions were evaluated in each subject separately using all the ECGs available for the given subject).