Is dynamic desaturation better than a static index to quantify the mortality risk in heart failure patients with Cheyne-Stokes respiration?

Cheyne-Stokes respiration (CSR) is a periodic, highly dynamic, respiratory pattern and a known comorbidity in congestive heart failure (CHF) patients. It is generally seen as an indicator for a negative prognosis, even if no distinction in degree is known or understood. This paper aims to improve on existing attempts by creating a quantification of the behavior of the dynamic desaturation process of oxygen in the blood. We performed this work on a cohort of $11$ subjects with CHF, reduced left ventricular ejection fraction, and CSR. The dynamic desaturation process was evaluated according to changes to peripheral capillary oxygenation $SpO2$ resulting from highly nonlinear relationships in the ventilatory system perturbed by periodic breathing. Hypoxaemic burden expressed as a static index $T90$ was compared to a novel relative desaturation index $RDI$⁠, developed in this paper. While $T90$ represents a single value calculated using a static cut-off value of $90%$ $SpO2$⁠, the $RDI$ is more sensitive to dynamic influences as it uses the specific maximum change in saturation for each CSR episode. The threshold of $T90=22$min$per night$ as suggested by Oldenburg et al. could not be confirmed to predict survival, but all central apneas resulting in a relative desaturation of $SpO2$ above a cut-off value of $8%$ were a $100%$ positive predictor of mortality. The $RDI$ proved sufficiently stable in intraindividual measurements across CSR epochs. Across the cohort, it showed a bimodal distribution for the deceased group, indicative of a possible aetiological difference. Hence, it is our conclusion that a dynamic approach to analyse desaturation of oxygen during Cheyne-Stokes respiration is to be strongly favoured over a static approach to analysis.