Heart Rate Turbulence: Higher Predictive Value Than Other Risk Stratifiers?
To the Editor:
We have read the article by Barthel et al1 describing the first prospective trial to determine the predictive value of heart rate turbulence (HRT) in patients after acute myocardial infarction. In previous studies, the ability of HRT to predict risk was only determined retrospectively.2 We would like to critically discuss here the uniqueness of the emerging risk factor, HRT, in comparison with other parameters. Interestingly, Barthel et al1 found that HRT was the strongest ECG-based risk predictor. This conclusion is surprising for 2 reasons, as follows. First, in an editorial comment3 on the original article by Schmidt et al,2 it was noted that the positive predictive value of HRT is only moderately higher than other ECG risk parameters, and it was suggested that some of the tests should be combined. However, Barthel et al1 considered only the heart rate variability (HRV) index and, as representatives, 3 other time domain parameters, but they did not analyze frequency domain, nonlinear HRV calculations, late potentials, prolonged QT interval, or T-wave alternans. Second, in a recent study,4 we investigated the suitability of short-term HRT (30 minutes) versus HRV analyses to characterize the regulatory differences in patients with dilated cardiomyopathy (DCM, n=37) and healthy controls (n=167). Although premature beats were excluded before HRV analysis, the highest correlation of HRT to HRV parameters was 0.94 in controls and 0.87 in DCM patients. The discrimination rate between DCM patients and controls was 86.3% for the complete data set (without HRT parameters). This rate was comparably high (88.0%) for the subgroup where HRT was applicable (only 14% of all data). The results of this study4 indicated that HRV and HRT have at least the same prognostic value, but HRV parameters have a significantly higher applicability. Moreover, we showed in 1998 that the predictive value of HRV using sophisticated parameters is significantly higher than the HRV index alone.5 Thus, taking only the HRV index as a reference parameter does not prove that HRT is the strongest risk predictor. We therefore think that the HRV parameters with the best predictive values as well as other ECG-based risk stratifiers should be retrospectively determined and then compared with HRT measurements. Thus, these observations cast doubt on the conclusion of Barthel et al1 that HRT is the strongest ECG-based risk predictor.
We prospectively assessed heart rate turbulence (HRT) parameters in 1455 survivors of acute myocardial infarction1 and stated that, “in our patients, as in the MPIP, EMIAT, and ATRAMI populations, HRT was the strongest ECG-based risk predictor.” This exactly corresponds to the observations made and can hardly be called an inappropriate generalization.
Dr Wessel and his coworkers assessed heart rate turbulence in only 37 patients suffering from dilated cardiomyopathy, ie, in a different patient group. Besides, the goal of the study by Dr Wessel et al was different from that of our work. We investigated the prognostic value of different parameters on long-term prognosis after acute myocardial infarction, whereas Dr Wessel’s group researched the association with a certain pathology, that is, dilated cardiomyopathy. It is not at all surprising that stratification of high- and low-risk patients with ischemic heart disease leads to different perceptions than a simple comparison of cardiac nonischemic patients with healthy controls.
In addition, Dr Wessel et al used a nonvalidated algorithm for the detection of ventricular premature complexes based on arterial pressure tracings rather than on ECG recordings.2 Moreover, they restricted the duration of the recordings to 30 minutes, that is, <3% of the duration proposed in our studies.1,3 Again, it is not surprising that such a restriction diminishes the power of HRT.
Hence, to summarize, Dr Wessel et al (1) abridged the HRT method in a hardly acceptable way, (2) investigated only a tiny population of different patients and healthy controls, and (3) restricted their study to simple diagnostic separation of the subjects rather than investigating their prognostic stratification.