Abstract
Obstructive sleep apnoea (OSA) is a highly prevalent but heterogeneous condition which makes identifying patients at risk of vigilance and driving impairment clinically challenging. Resting wake electroencephalography (EEG) is associated with vigilance performance in healthy participants. We examined if rested wake EEG predicted vigilance and driving impairment in OSA following extended wakefulness. Fifty-four patients underwent baseline polysomnography and 28-h extended wakefulness, repeated vigilance assessments (psychomotor vigilance task (PVT), driving simulator) and resting wake EEG (Karolinska drowsiness test). Cluster analysis assigned patients into groups of either resistant (n = 38) or vulnerable (n = 16) to vigilance failure based on PVT and driving performance following extended wakefulness. Backward stepwise regression models and receiver operator characteristics curves were used to determine the strongest clinical and wake EEG predictors of vigilance impairment. The vulnerable OSA group showed impaired PVT and driving performance relative to the resistant group (p < 0.01). Compared with resistant patients, the vulnerable group exhibited increased delta (p < 0.001) and theta (p = 0.003) EEG power across extended wake. Significant predictors of vigilance impairment were (1) baseline wake EEG theta and O(2) Nadir during sleep, explaining 42% of the variance, and (2) delta power and O(2) Nadir explaining 32% of the variance in vigilance performance. ROC analysis showed strong discrimination between vulnerable and resistant patients (AUC 0.85-0.86, sensitivity 73%-87%, specificity 71%-84%). Slow frequency wake EEG activity and sleep hypoxemia at baseline are predictive of subsequent driving simulator and vigilance impairment in patients with OSA following extended wakefulness. This is potentially important for vigilance and fitness-to-drive assessments in OSA. Trial Registration: Australian New Zealand Clinical Trials Registry: ACTRN12613001171707.