Abstract
OBJECTIVE: Apnea/hypopnea index (AHI), especially without arousal criteria, does not adequately risk stratify patients with mild obstructive sleep apnea (OSA). We describe and test scoring reliability of an event, Flow Limitation/Obstruction With recovery breath (FLOW), representing obstructive airflow disruptions using only pressure transducer and snore signals available without electroencephalography. METHODS: The following process was used (i) Development of FLOW event definition, (ii) Training period and definition refinement, and (iii) Reliability testing on 10 100-epoch polysomnography (PSG) samples and two 100-sample tests. Twenty full-night in-laboratory baseline PSGs in OSA patients with AHI with ≥4% desaturations <15 were rescored for FLOW events, traditional hypopneas with desaturations, respiratory-related arousal (RRA) events (hypopneas with arousals and respiratory-effort related arousals) and non-respiratory arousals (NRA). RESULTS: Scoring of FLOW events in 100-epoch samples had good reliability with intraclass correlation (ICC) of 0.91. The overall kappa for presence of events on two sets of 100 sample events was 0.84 and 0.87 demonstrating good agreement. Moreover, 80% of RRA and 8% of NRA were concurrent with FLOW events. Furthermore, 56% of FLOW events were independent of RRA events. FLOW stratifies patients in traditional AHI categories with 50%/8% of AHI with ≥3% desaturations (AHI3) <5 and 12%/63% of AHI3 >5 in lowest/highest tertiles of AHI3 plus FLOW index. CONCLUSIONS: Scoring of FLOW after training is reliable. FLOW scores a high proportion of RRA and many currently unrepresented obstructive airflow disruptions. FLOW allows for stratification within the current normal-mild OSA category, which may better identify patients who will benefit from treatment.