gms | German Medical Science

In a recent letter Costa et al. [Ref. 1] used multi-scale entropy analysis to distinguish healthy hearts from those suffering from congestive heart failure (CHF) or atrial fibrillation (AF) based on heart rate variability (HRV) recordings. They mainly concluded that healthy and pathologic groups can be differentiated consistently using this multi-scale entropy, while conventional algorithms fail. Shortly after publication, this investigation was recommended enthusiastically [Ref. 2]. Here, two major weaknesses of this study shall be pointed out.

First, the age of the patients and healthy subjects they considered differs significantly. Consequently, hypothesis testing between these groups is not appropriate and misleading, because HRV decreases with age, as it is well known in cardiology. Secondly and in contrast to the main statement made by Costa et al., we show that already a rather simple measure of complexity from symbolic dynamics, which detects intermittently decreased beat-to-beat variability, is able to discriminate clearly between CHF patients and even age-matched healthy subjects [Ref. 3]. Moreover, for patients with AF this parameter is zero and we therefore do not have to consider them here.

No sophisticated techniques, such as multi-scale entropy, are necessary. It can only be concluded that their methods robustly separates young healthy volunteers from both elderly cardiac patients and elderly healthy subjects, which is similar to comparing apples with oranges.