Exclusion of EEG-based arousals in wake epochs of polysomnography leads to underestimation of the arousal index

Abstract

Study objectives: There is an internal contradiction in current American Academy of Sleep Medicine standards for arousal index (AI) calculation in polysomnography: Arousals in sleep and wake epochs are counted, but only sleep time is used in the denominator. This study aimed to investigate the impact of including arousals scored in wake epochs on the AI.

Methods: We compared AIs including (AI_inc) vs excluding (AI_exc) awake-epoch arousals from 100 consecutive polysomnograms conducted for investigation of possible obstructive sleep apnea. To determine the AI that most closely approximated "truth," AI_inc and AI_exc were compared to an AI calculated from continuous sleep analysis (AI_cont) in a 20-polysomnogram subgroup of patients.

Results: The median (interquartile range) increase in AI_inc was 5.2 events/h (3.5-8.1) vs AI_exc (AI_inc = 28.0 events/h [18.4-38.9] vs AI_exc = 22.9 events/ h [13.1-31.3]), equating to an increase of 25.3% (15.6-40.8). As the AI increased, the difference increased (P < .001), with decreasing sleep efficiency and an increasing apnea-hypopnea index as the strongest predictors of the difference between AI_exc and AI_inc. The absolute AI_exc-AI_cont difference (7.7 events/h [5.1-13.6]) was significantly greater than the AI_inc-AI_cont difference (1.2 events/h [0.6-5.7]; z = -3.099; P = .002).

Conclusions: There was a notable increase in AI when we included wake-epoch arousals, particularly in patients with more severe obstructive sleep apnea or reduced sleep efficiency. However, the AI including wake-epoch arousals best matched the "true" continuous sleep-scoring AI. Our study informs clinical and research practice, highlights epoch scoring pitfalls, and supports the current American Academy of Sleep Medicine standard arousal reporting approach for future standards.

Citation: Wilson DL, Tolson J, Churchward TJ, Melehan K, O'Donoghue FJ, Ruehland WR. Exclusion of EEG-based arousals in wake epochs of polysomnography leads to underestimation of the arousal index. J Clin Sleep Med. 2022;18(5):1385-1393.

Keywords: analysis; arousal; continuous; epoch; polysomnography; wake.

Figure 1. An arousal from sleep can be scored in wake and sleep epochs.

(A) Thirty-second epoch during PSG showing how an arousal from sleep can be scored in an epoch marked as “W” (wake). The green underline indicates the 20-second arousal scored after 10 seconds of sleep as per AASM guidelines. (B) 30-second epoch during PSG showing an arousal during an epoch marked as “N2” (sleep). The green underline indicates the 6-second arousal scored as per AASM guidelines. The panel shows the signals Chin1-Chin2 = EMG; E1-M2 and E2-M2 = left and right electrooculogram; F4-M1, C4-M1, and O2-M1 = frontal, central, and occipital EEG and ECG across a 30-second window. AASM = American Academy of Sleep Medicine, ECG = electrocardiogram, EEG = electroencephalogram, EMG = electromyogram, PSG = polysomnography.

Figure 2. Bland-Altman plot illustrating agreement between AI_inc and AI_exc.

(A) Linear scales and (B) log scales show the median, 5th, and 95th percentiles. The raw (A) and transformed (B) data show that as the average of AI_inc and AI_exc increases, the discrepancy between these AIs becomes larger with a wider spread. AI = arousal index, AI_exc = arousal index excluding arousals scored in epochs of wake, AI_inc = arousal index including arousals scored in epochs of wake.

Figure 3. Comparison of the AI_cont to the AI_inc and AI_exc.

Difference between the AI_cont, and (A) AI_inc, and (B) AI_exc. The median (with 5th and 95th percentiles) indicates that the AI_inc is consistently closer to the true AI_cont. n = 20. AI = arousal index, AI_cont = AI based on continuous sleep scoring, AI_exc = arousal index excluding arousals scored in epochs of wake, AI_inc = arousal index including arousals scored in epochs of wake.