Nighttime respiratory stability in heart failure: A promising signal requiring rigorous validation

To the Editor,

We read with great interest the article by Tamaru et al. in DIGITAL HEALTH, which explores the association between nighttime respiratory rate stability (RRS), measured by a novel non-contact sensor, and adverse outcomes in patients with heart failure (HF). ¹ The authors are to be commended for tackling a critical unmet need in remote HF management: the identification of non-invasive biomarkers with higher specificity than conventional metrics like daily weight. Their finding that lower RRS during hospitalization is associated with a significantly increased risk of one-year mortality and rehospitalization is both intriguing and promising for future telemonitoring strategies.

However, while RRS demonstrates prognostic potential, its physiological underpinnings remain uncalibrated. A key limitation is the lack of concurrent validation against the gold standard of polysomnography (PSG). Consequently, it is unclear whether the detected respiratory instability is a surrogate for specific sleep-disordered breathing patterns, such as the prognostically potent Cheyne–Stokes respiration, which is highly prevalent in HF and a known independent predictor of mortality. ² Establishing a quantitative correlation between RRS and the apnea–hypopnea index would be a crucial step to confirm its validity and clarify its mechanistic basis, elevating it from a statistical association to a clinically interpretable measure.

Furthermore, the clinical utility of a single RRS measurement obtained during an acute decompensation hospitalization is debatable. This period reflects a state of maximum physiological stress and fluid overload, which may not represent the patient's baseline respiratory pattern in a stable, euvolemic state. The true promise of telemonitoring lies not in static risk stratification, but in the dynamic tracking of physiological parameters to preemptively detect clinical deterioration, as successfully demonstrated by invasive hemodynamic monitoring. ³ Whether longitudinal changes in RRS measured in a home setting can predict impending HF exacerbations remains the pivotal, unanswered question. To date, non-invasive telemonitoring strategies have yielded inconsistent results, largely due to the low predictive power of the monitored signals. ⁴

Therefore, the path toward clinical implementation of this technology requires further rigorous investigation. Future research should prioritize two key areas: first, multicenter validation studies incorporating simultaneous PSG to establish the external and physiological validity of RRS. Second, prospective longitudinal studies are needed to assess whether dynamic changes in home-monitored RRS, perhaps integrated with other biosignals within a machine learning framework, can provide actionable alerts for impending decompensation. ⁵

In conclusion, Tamaru et al. have introduced a potentially valuable tool for HF risk stratification. Their work provides a strong impetus for future research. However, before RRS can be integrated into clinical practice, its physiological basis must be validated, and its utility for dynamic, predictive monitoring in the ambulatory setting must be demonstrated. Bridging these evidence gaps is essential to transform this promising signal into a meaningful advance in the remote management of HF.