What question did this study set out to answer?

This study aims to evaluate the diagnostic accuracy of hypoxic burden in patients with obstructive sleep apnea to improve severity classification.

April 3, 2026Open Access

Hypoxic burden calculation in patients with obstructive sleep apnea diagnosed by peripheral arterial tonometry: diagnostic accuracy and clinical implications

Key Points

This study aims to evaluate the diagnostic accuracy of hypoxic burden in patients with obstructive sleep apnea to improve severity classification.
Conducted a retrospective analysis of adults undergoing Peripheral Arterial Tonometry Home Sleep Apnea Testing (PAT-HSAT) from 2016 to 2025.
Included 1,171 patients with an apnea-hypopnea index (AHI3%) ≥ 5 events/h after exclusions.
Calculated hypoxic burden as a normalized measure of oxygen desaturation events (ODI4%) per hour of total sleep time (TST).
Assessed diagnostic accuracy with receiver operating characteristic (ROC) analyses and standard performance metrics.
Hypoxic burden showed strong discrimination with AUC values of 0.867 for moderate-to-severe OSA and 0.897 for severe OSA.
30% of mild, 49.2% of moderate, and 26.6% of severe OSA patients had discordant hypoxic burden values compared to AASM-defined severity categories.
Hypoxic burden thresholds ≥ 16.6%·min/h had 90.1% sensitivity for moderate-to-severe OSA, and ≥ 29.5%·min/h had 88.3% specificity for severe OSA.

Abstract

The apnea–hypopnea index (AHI) is the standard metric for grading obstructive sleep apnea (OSA) severity but does not capture cumulative oxygen desaturation; hypoxic burden (HB), integrating desaturation area per hour, may better reflect physiological load, but its diagnostic performance remains uncertain. We conducted a retrospective diagnostic study of adults undergoing Peripheral Arterial Tonometry—Home Sleep Apnea Testing (PAT-HSAT) (WatchPAT®) between 2016 and 2025. After prespecified exclusions, 1,171 patients with AHI3% ≥ 5 events/h were included. HB (%·min/h) was derived from Oxygen Desaturation Index (ODI4%) linked desaturation events and normalized by total sleep time (TST). Diagnostic accuracy for moderate-to-severe (AHI3% ≥ 15 events/h) and severe OSA (AHI3% ≥ 30 events/h) was assessed using receiver operating characteristic (ROC) analyses and standard performance metrics. HB distributions were examined across American Academy of Sleep Medicine (AASM) categories. Substantial heterogeneity in HB was observed within AASM severity categories. HB values discordant with AASM-defined severity were present in 30.0% of mild, 49.2% of moderate, and 26.6% of severe OSA patients, affecting 35.4% of the overall cohort. HB showed good discrimination for moderate-to-severe OSA (AUC 0.867, 95% CI 0.846–0.888) and strong discrimination for severe OSA (AUC 0.897, 95% CI 0.879–0.916). An HB threshold ≥ 16.6%·min/h showed high sensitivity (90.1%) for moderate-to-severe OSA, while ≥ 29.5%·min/h showed high specificity (88.3%) for severe OSA. HB derived from PAT-HSAT identifies clinically relevant heterogeneity within AASM severity categories and provides good diagnostic discrimination. By capturing cumulative hypoxemic load beyond event frequency, HB complements AHI-based classification and may refine severity stratification. Not applicable.

Mark Helpful

Bookmark

Relay

View Full Paper