Author + information
- Michael Dommasch, MD†,
- Daniel Sinnecker, MD†,
- Petra Barthel, MD†,
- Alexander Müller, MSc†,
- Ralf J. Dirschinger, MD†,
- Alexander Hapfelmeier, PhD‡,
- Katharina M. Huster, MD†,
- Karl-Ludwig Laugwitz, MD†,§,
- Marek Malik, MD, PhD‖ and
- Georg Schmidt, MD†,§∗ ()
- †1. Medizinische Klinik der Technischen Universität München, Munich, Germany
- ‡Institut für Medizinische Statistik und Epidemiologie der Technischen Universität München, Munich, Germany
- §German Centre for Cardiovascular Research (DZHK), partner site Munich Heart Alliance, Munich, Germany
- ‖St. Paul’s Cardiac Electrophysiology, University of London, and Imperial College, London, England
- ↵∗1. Medizinische Klinik, Klinikum rechts der Isar der Technischen Universität München, Ismaninger Strasse 22, 81675 München, Germany
To the Editor:
Survivors of acute myocardial infarction (MI) are at substantial risk of subsequent death due to progressive heart failure, arrhythmic events, and/or reinfarction. We have recently shown that respiratory rate measured in 10-min supine position by a piezoelectric thoracic sensor provides powerful and independent prognostic information in these patients (1). The goal of this study was to assess the prognostic potential of the nocturnal respiratory rate (NRR), assessed from standard Holter electrocardiogram recordings in a larger patient cohort. Additionally, we analyzed the association of NRR and different death modes, especially sudden (SCD) and non–sudden cardiac death (N-SCD).
Data from the previously reported cohort (1) were used to develop an algorithm deriving the NRR from Holter recordings, and to determine NRR optimum risk-predicting dichotomy of ≥18.6/min. Briefly, the algorithm determines the respiration rate from QRS amplitudes, QRS vectors, and heartbeat intervals, all of which are influenced by respiration. NRR was subsequently prospectively assessed in 1,538 MI survivors enrolled between 1996 and 2005 in 2 Munich centers. Standard Holter electrocardiograms were recorded at a median of 5 days (interquartile range: 4 to 7 days) after MI. The primary study endpoint was 5-year cardiac mortality. Secondary endpoints were N-SCD, SCD, all-cause mortality, and noncardiac mortality. An independent endpoint committee assigned the modes of death. The study was approved by the local ethics committee.
Within the first 5 years, 146 patients (9.5%) died. Of these, 82 patients suffered cardiac deaths (39 N-SCD and 43 SCD) and 64 noncardiac deaths.
NRR was ≥18.6/min in 384 patients and <18.6/min in 1,154 patients. Five-year cardiac mortality rates in these groups were 13.7% and 3.3%, respectively (p < 0.001). In multivariable Cox analysis, cardiac mortality and N-SCD, but not SCD, were significantly associated with NRR (Fig. 1A). NRR ≥18.6/min indicated a hazard ratio for N-SCD of 4.56 (95% confidence interval: 2.21 to 9.43; p < 0.001). The association of NRR and N-SCD was independent from established Holter-based parameters of cardiac autonomic function such as heart rate variability parameters or heart rate turbulence.
A total of 155 patients had left ventricular ejection fraction (LVEF) ≤35%. Of these, 64 had an NRR <18.6/min, and 91 had an NRR ≥18.6/min. In the group of patients with NRR <18.6/min, cardiac mortality was 12.5%, 3.1% attributable to N-SCD and 9.4% attributable to SCD. In the group of patients with NRR ≥18.6, cardiac mortality was 28.6%, 16.5% attributable to N-SCD, and 12.1% attributable to SCD. Differences were statistically significant for cardiac mortality and N-SCD (p < 0.001 and p = 0.009, respectively), but not for SCD (p = 0.595). Figure 1B depicts the proportions of N-SCD among all cardiac deaths as a function of the LVEF and NRR dichotomies.
We assume that increased NRR is an early sign of impending heart failure prompted by either subclinical pulmonary congestion or by altered autonomic control of respiratory activity (1). We foresee potential roles of NRR for future risk stratification approaches. Currently, LVEF is the mainstay of post-MI risk assessment, as the decision whether to implant a defibrillator is based mainly on LVEF. NRR might prove helpful by identifying patients who are less likely to benefit from ICD implantation because of a high N-SCD prevalence. Among patients with LVEF ≤35% (i.e., ICD candidates), increased NRR identifies a group of patients with frequent N-SCD (i.e., a mode of death likely not implantable cardioverter-defibrillator [ICD] preventable). Moreover, it cannot be excluded that ICD implantation might be harmful in some of these patients.
Independent validation of this new risk stratification approach is necessary before it can be incorporated into clinical decision making. Most importantly, the post-hoc assigned cause-specific mortalities can only provide an approximation of deaths that may and may not be prevented by ICD implantation. While a definite proof of the hypothesis would require a prospective study performed in ICD candidates, the methodology presented here also allows for independent confirmatory analyses in datasets from ICD primary prevention studies, provided that digitized Holter recordings are available.
Please note: This work was supported by the Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie (13N/7073/7) and by the Deutsche Forschungsgemeinschaft (Si 1747/1-1). Drs. Dommasch and Sinnecker contributed equally as first authors. Drs. Malik and Schmidt contributed equally as senior authors.
- American College of Cardiology Foundation