A new classification for sleep analysis in critically ill patients.

Abstract Background Patients in intensive care units (ICUs) experience severe sleep alterations and conventional sleep scoring rules are difficult to use in these patients. In a previous study, we showed that abnormal sleep EEG and wake EEG patterns could predict the outcome of noninvasive ventilation in a group of patients treated for acute respiratory failure. Our aims were to assess the prevalence of these abnormal sleep/wake EEG patterns in a larger group and search for objective parameters to help their identification. Methods We reviewed sleep studies previously performed with full polysomnography during 17-h in conscious nonsedated ICU patients receiving invasive ventilation during weaning or noninvasive ventilation for acute respiratory failure. Results We included 57 patients. Sleep scoring using conventional rules was not feasible in 16 (28%) patients due to the absence of stage-2 markers. Wake EEG in these 16 patients, although recognizable, showed abnormal features, including decreased reactivity to eye opening and slower peak EEG frequency compared to patients with normal sleep–wake EEGs. Conclusion In almost one third of awake mechanically ventilated ICU patients, sleep cannot be classified with standard criteria. Two new states, atypical sleep and pathologic wakefulness, need to be added. We suggest rules for scoring these states. The origin and links with outcomes of these abnormal EEG patterns deserve investigation.