Abstract Introduction Repetitive hypoxia serves as the principal mechanistic bridge between Obstructive Sleep Apnoea and cardiovascular disease (CVD). Desaturation area-based parameters, like Hypoxic Burden (HB) and the Respiratory Event Desaturation Transient Area (REDTA), are promising metrics for predicting CVD mortality. These metrics are not without limitations. HB calculation is complex and noise sensitive, while REDTA may be overly simplistic. We investigated six area-based methods, each with a different choice of sampling window and baseline. Methods The oximetry signals from the Sleep Heart Health Study (SHHS) with CVD mortality outcomes and covariate information was used. The HB method was implemented and denoted as ARE. It uses a record-specific sampling window and event-specific baseline. We extended the experiment to include two additional baseline methods: a record-specific baseline (ARR) and a fixed baseline (ARF). Similarly, REDTA which uses a fixed sampling window and baseline (AFF), was extended to an event-specific baseline (AFE) and a record-specific baseline (AFR). Six desaturation area-based methods were performed in this study. The performance of each parameter in predicting CVD mortality was assessed using an adjusted Cox proportional hazard ratio(HR) analysis. Results ARR and AFR outperformed other methods The HRs were ARR:1.79 (95% CI: 1.00-3.19, p<0.05), and AFR:1.64 (95% CI: 1.16-2.32, p<0.01). Discussion Computational complexity was also considered. AFR was easier to calculate than ARR as it used a fixed sampling window. We believe AFR provides a step towards a novel parameter with easy calculation providing early risk stratification in cardiovascular patients.
Read full abstract