A79-15 Discordance Between Apnea-hypopnea Index And Hypoxic Burden-based Severity Classification In Obstructive Sleep Apnea

Rationale

The apnea-hypopnea index (AHI) is widely used to classify the severity of obstructive sleep apnea (OSA), yet it reflects event frequency rather than the cumulative burden of nocturnal hypoxemia. Hypoxic burden (HB), which integrates both the depth and duration of oxygen desaturations, may better capture the physiologic impact of OSA. We examined the degree of agreement between AHI-based and HB-based severity classification.

Methods

We conducted a retrospective analysis of consecutive adult in-laboratory polysomnography studies, performed at a single center between November 1 and November 30, 2025. OSA severity by AHI was classified as normal (<5 events/hour), mild (5-14.9), moderate (15-29.9), or severe (≥30). Hypoxic burden severity was defined using a validated 0-10 composite score and categorized as mild (<3.34), moderate (3.34-6.66), or severe (≥6.67). Severity categories were compared using cross-tabulation. Among patients with OSA (AHI ≥5), discordance was defined as HB severity being higher, equal to, or lower than AHI severity. Agreement was evaluated using observed agreement and Cohen’s kappa (κ).

Results

A total of 95 patients were included (mean age 58.4 ± 18.3 years; mean body mass index 33.2 ± 7.9 kg/m²). Among 73 patients with OSA, hypoxic burden severity was higher than AHI-based severity in 27.4% of cases, concordant in 61.6%, and lower in 11.0%. Overall agreement between AHI- and HB-based severity classifications was poor, with an observed agreement of 30.1% and a low kappa value (κ = 0.15, p = 0.0004). Patients in whom hypoxic burden exceeded AHI severity demonstrated lower nocturnal oxygen saturation, indicating greater physiologic impairment despite similar event frequency.

Conclusions

Severity classification based on AHI and hypoxic burden shows poor agreement, underscoring important limitations of AHI as a standalone metric for assessing OSA severity. Incorporating hypoxic burden may provide a more physiologically meaningful assessment of disease severity and improve risk stratification beyond event counts alone.

This abstract is funded by: None