Beat-to-beat signatures of sleep apnea across cardiovascular diseases.

BACKGROUND

Obstructive sleep apnea (OSA) is closely linked with cardiovascular disease (CVD), but different cardiovascular phenotypes may influence ECG-derived RR interval dynamics differently. Previous work has shown that scale-dependent detrended fluctuation analysis (sDFA) can detect OSA across broad CVD status, but it remains unclear whether this relationship differs between CVD subtypes.

METHODS

Overnight polysomnography recordings from the Sleep Heart Health Study were analyzed in participants with a history of myocardial infarction (MI), congestive heart failure (CHF), or angina pectoris (ANG). Participants with more than one of these conditions were excluded. OSA was analyzed as a binary outcome using an apnea-hypopnea index threshold of 5 events/h. Propensity score matching was performed within each subgroup using age and body mass index. ECG-derived RR interval dynamics were analyzed using cHRV metrics and second-order sDFA, with discrimination assessed using ROC-AUC.

RESULTS

The matched subgroups consisted of 112 participants with MI, 18 with CHF, and 54 with ANG. sDFA achieved higher ROC-AUC values than cHRV metrics in all subgroups. The highest sDFA ROC-AUC was observed in ANG (0.79), compared with 0.66 in MI and 0.69 in CHF. MI and ANG showed broadly similar scale-dependent profiles, whereas CHF appeared distinct; however, the CHF analysis was substantially underpowered.

CONCLUSION

ECG-derived sDFA features showed subtype-dependent behavior for OSA discrimination, with the strongest performance observed in ANG. These exploratory findings suggest that cardiovascular phenotype may influence OSA-related RR interval dynamics. Binary OSA grouping may mask severity-related differences, while exclusion of participants with multiple cardiovascular conditions limits clinical applicability.