Monitoring of nighttime EEG slow‐wave activity during dexmedetomidine infusion in patients with hyperactive ICU delirium: An observational pilot study

Abstract Background The disturbance of sleep has been associated with intensive care unit (ICU) delirium. Monitoring of EEG slow‐wave activity (SWA) has potential in measuring sleep quality and quantity. We investigated the quantitative monitoring of nighttime SWA and its association with the clinical evaluation of sleep in patients with hyperactive ICU delirium treated with dexmedetomidine. Methods We performed overnight EEG recordings in 15 patients diagnosed with hyperactive delirium during moderate dexmedetomidine sedation. SWA was evaluated by offline calculation of the C‐Trend Index, describing SWA in one parameter ranging 0 to 100 in values. Average and percentage of SWA values <50 were categorized as poor. The sleep quality and depth was clinically evaluated by the bedside nurse using the Richards‐Campbell Sleep Questionnaire (RCSQ) with scores <70 categorized as poor. Results Nighttime SWA revealed individual sleep structures and fundamental variation between patients. SWA was poor in 67%, sleep quality (RCSQ) in 67%, and sleep depth (RCSQ) in 60% of the patients. The category of SWA aligned with that of RCSQ‐based sleep quality in 87% and RCSQ‐based sleep depth in 67% of the patients. Conclusion Both, SWA and clinical evaluation suggested that the quality and depth of nighttime sleep were poor in most patients with hyperactive delirium despite dexmedetomidine infusion. Furthermore, the SWA and clinical evaluation classifications were not uniformly in agreement. An objective mode such as practical EEG‐based solution for sleep evaluation and individual drug dosing in the ICU setting could offer potential in improving sleep for patients with delirium.

software to monitor slow waves as a surrogate for recovering sleep. Sleep quality was generally poor as assessed by both the presence of slow waves and the Richards-Campbell Sleep Questionnaire, though the two assessment methods were not strongly concordant. The study corroborates previous findings of poor sleep quality in patients with ICU delirium, and the challenges for assessing sleep quality in these patients.

| INTRODUCTION
Delirium is as a generalized dysfunction of cerebral cortical processes with disturbed sleep-wake cycle, disorientation, and attention deficits.
The pathophysiological mechanism of delirium is poorly understood, but inflammation, multiple organ failure, use of benzodiazepines, and sepsis are known risk factors for intensive care unit (ICU) delirium. [1][2][3] ICU delirium is related to undesirable outcomes such as prolonged hospitalization, increased costs, higher mortality, and long-term cognitive impairment. [4][5][6] Sleep disturbances are considered to be important risk factors for delirium. 7 Diminished total sleep time and reduced non-rapid eye movement (NREM) sleep are associated with delirium. 8 Loss of stage N2 electroencephalogram (EEG) features (K-complexes and spindles) also are common in critically ill patients with delirium and associated with more severe encephalopathy and higher odds of death. 9 These studies suggest a crucial role for NREM sleep in avoiding and recovering from ICU delirium. In healthy volunteers, dexmedetomidine promotes NREM sleep and produces a state closely resembling physiological sleep in humans, activating normal NREM sleeppromoting pathways. 10,11 Slow waves are the most important EEG features of NREM sleep. 12,13 Originating from the oscillatory activity of neurons in the neocortex and thalamus, slow-wave activity (SWA) can be measured from the EEG with its power focused below 1 Hz. 14 Because it is associated with the deep recovering stages of sleep, SWA offers an objective neurophysiologic measure of sleep quality and quantity. Our previous research shows that EEG SWA can be captured reliably using forehead electrodes in the ICU setting. 15 In this observational pilot study in patients with hyperactive ICU delirium receiving nighttime dexmedetomidine infusion, we investigated the possibility of quantitatively monitoring nighttime EEG SWA and the association of SWA with the clinical evaluation of sleep. We hypothesized that, in these patients, nighttime EEG SWA would offer an objective quantitative measure for evaluating sleep quality and depth.

| METHODS
This prospective observational study was conducted in a 26-bed medical-surgical ICU in a tertiary-level university hospital in Finland.
The study was approved by the Northern Ostrobothnia Hospital district review board (135/2017) and local ethics committee (47/2017).
Written informed consent for participation was obtained from the next of kin.

| Nighttime sedation
The nighttime sleep of the patients was supported following the ICU's standard protocol with dexmedetomidine infusion (4 μg/ml solution), which was started at 9 p.m. with 0.2 μg/(kg*h) infusion (no bolus) and continued until 7 a.m. The bedside nurse targeted the sedation level by adjusting the dexmedetomidine infusion rate to achieve a Richmond Agitation-Sedation Scale (RASS) of À3 to À2. RASS score was performed every 2 h and prior to possible propofol infusion. A maximum dexmedetomidine infusion rate of 1.4 μg/(kg*h) was not exceeded following the ICU's standard practice. If needed for an adequate level of sedation, propofol was combined with dexmedetomidine with a dose of 1-4 mg/(kg*h). Other sedatives, analgesics, melatonin, and antipsychotics were administered if needed at the discretion of the attending intensivist. Nursing disturbances (taking of laboratory samples, changing of patient position, and changes in drug, fluid, or nutrition administration) between 12 a.m. and 5 a.m. were recorded. The ICDSC score was also obtained 2 h after the discontinuation of dexmedetomidine infusion in the morning.

| Clinical data acquisition
All clinical data were input into the ICU electronic clinical data management system (Centricity Critical Care*(8.1) SP7 (8.17.034); GE Healthcare, Barrington, IL, USA). Patient age, sex, body mass index, reason for ICU admission, severity of illness score (Acute Physiology and Chronic Health Evaluation II [APACHE II]) and need for vasoactive or sedative agents were retrieved. Length of stay in the hospital and in the ICU as well as days on mechanical ventilation were also obtained, and the need for additional sedatives or antipsychotics was recorded. Outcomes after 3 months following the hospitalization were obtained via telephone contact and defined using the cerebral performance category score. Good recovery was defined as a score of 1 or 2. In addition to the objective quantitative evaluation using EEG, the night shift nurse that spent the night bedside also clinically evaluated sleep at 6 a.m. using the Richards-Campbell Sleep Questionnaire (RCSQ). 18 For this analysis, the RCSQ parameters "sleep depth" and Age, years 59 (15) Sex, male, n (%) 9 (60) Body mass index 30 (7) APACHEII score on admission 18 [11][12][13][14][15][16][17][18][19][20][21] Time in ICU prior to the study, days 2.4 [1][2][3][4][5][6] Time in mechanical ventilation prior to the study, days 0.49 [0.  Emergency admission, n (%) 11 (73) Operated, n (%) 6 (40)

| RESULTS
During the study period, we had 134 eligible patients with hyperactive delirium, of whom 15 were included in the study (Figure 1)   However, only half of the patients in that study slept more than 50% of the night, despite dexmedetomidine administration. In another report, the median percentage of stage N2 sleep with dexmedetomidine infusion was 44% in non-cardiac surgical patients in the ICU. 21 Another group assessing 10 patients receiving mechanical ventilation found that sleep efficacy was 50% of the nighttime period during dexmedetomidine infusion, with slow-wave sleep in only three patients and rapid eye movement sleep in only two. 22  Of interest, in our series, the total administered dexmedetomidine dose was higher in patients without delirium the next morning, suggesting a beneficial effect unrelated to sleep. There was no statistically significant difference in sleep depth, sleep quality, average of SWA, or percentage of SWA between those with or without delirium in the following morning. We note that the role of sleep disturbance in the development of ICU delirium is not undisputed. In contrast to EEG-based studies, polysomnography studies have yielded contradictory results regarding the association between disturbed sleep and development of delirium. 24,25 Many other pathophysiologic mechanisms contribute to the development of delirium and could be affected by dexmedetomidine. 26 Furthermore, reduced delirium risk with dexmedetomidine may arise from avoidance of benzodiazepines. 27 In our series, the only patient who also received diazepam experienced both low SWA and poor sleep quality according to the RCSQ.
We also found a discrepancy between the clinical evaluation of sleep based on RCSQ parameters and the EEG SWA. To our knowledge, this has not been studied earlier in patients suffering from ICU delirium. However, in a wider sense, subjective perception of sleep has been reported to be inconsistent with objective EEG indicators of sleep 28 as well as sleep evaluation based on polysomnography. 29 Furthermore, in intensive care, nurses tend to give higher ratings on the RCSQ compared with patient-reported outcomes, suggesting overestimation of sleep quality. 30 A recent randomized controlled trial showed that although dexmedetomidine administration reduced the incidence of delirium, it did not affect patient-reported sleep quality. 31 Taken together, questionnaires do not seem to be sufficiently robust for reliable assessment of sleep quality and quantity.
All patients in the current study were disturbed at least once during the night by treatment events, but SWA did not correlate significantly with the number of disturbances. We did not quantify the effect of environmental factors in the ICU, such as noise and lighting.
Others have shown that patient sleep in the ICU is disrupted by noise, light, and treatment events. [32][33][34] We also note that clinical assessment of sedation or sleep can disturb the patient and interrupt sleep, which also argues for easy-to-use bedside technology for sleep evaluation.
Indeed, monitoring could even be used to alert the bedside nurse to delay unnecessary treatment events when the patient is in deep sleep states.

CONFLICT OF INTEREST
Jukka Kortelainen is a co-founder of Cerenion Oy, Oulu, Finland.
All other authors have no conflict of interest.