Delirium and sleep in intensive care II – monitoring and diagnostic options

Overview

Monitoring sleep quality and delirium are essential in providing modern intensive care. They present both equipment and personnel challenges. Not only because certain monitoring methods, such as polysomnography, affect monitored sleep themselves. Although new alternatives exist, polysomnography remains the gold standard in diagnosing and researching sleep disorders for the validity of the data obtained. Without monitoring and screening methods, delirium and sleep disorders cannot be reliably diagnosed in intensive care. Without a clearly established diagnosis, the outcomes of delirium and reduced sleep quality cannot be investigated. This article summarizes various options for monitoring both sleep and delirium, their advantages and limitations in the critical care setting.

Keywords:

monitoring – intensive care – sleep – delirium – polysomnography – sleep quality – actigraphy

This is an unauthorised machine translation into English made using the DeepL Translate Pro translator. The editors do not guarantee that the content of the article corresponds fully to the original language version.

 

Introduction

Despite the high prevalence of intensive care unit (ICU) delirium and the identification of a large number of risk factors [1], the consequences of delirium development in the short and long term cannot be consistently determined. This may be explained by the different aetiologies of delirium, and therefore the results of individual studies investigating the consequences of delirium are often incomparable [2]. On the other hand, the acute and long-term consequences of delirium experienced in the ICU are substantial for the patient and his/her health. Without a prior and correct diagnosis of delirium, these consequences cannot be identified. The clinical practice of delirium diagnosis in neurointensive care in the Czech Republic was mapped in a recent questionnaire study. Although a significant majority of respondents (97%) considered delirium to be an important clinical and research problem, the majority of departments (89%) agreed on the problem of underdiagnosis of delirium. Only slightly more than half (57%) perform the screening itself and less than a quarter (21%) use validated tools for diagnosis [3]. Disturbances in the sleep-wake cycle are associated with the development of delirium and are related to disruption of circadian rhythms; therefore, sleep and sleep quality should be assessed when further studying the impact of delirium development [4]. A number of sleep and delirium monitoring methods are available for this purpose, but each has its limitations in the intensive care setting [5].

 

Options for monitoring sleep and sleep quality

Sleep and its quality and quantity can be monitored objectively by instrumental monitoring with polysomnography (PSG), other EEG-based methods and actigraphy. The validity of the measured data increases with the technical and personal complexity of the method, and the PSG has the highest validity. Some questionnaire techniques that assess sleep subjectively have also been validated against PSG. The questionnaires are answered directly by the patient or nursing staff as part of clinical observation. The advantages and limitations of the methods are summarized in Table 1.

 

Polysomnography

The PSG [6] is considered the gold standard for monitoring aspects of sleep quality and quantity, monitoring multiple parameters, namely: EEG, ECG, electrooculography (EOG), EMG from the lower extremities and chin, respiratory effort using chest and abdominal plethysmographic strips, pulse oximetry, expiratory airflow and nasal pressure, body position during sleep, snoring monitoring, and a night vision camera. To objectify sleep quality and quantity, we monitor these PSG parameters [7]:

               Total Recording Time (TRT) - total recording time;

    Total Bed Time (TBT) - total time on bed - often the same as TRT;

    Sleep Latency (SL) - latency to fall asleep - the time from putting the patient to sleep

from "lights out") to falling asleep, usually given in minutes, 5-15 minutes to fall asleep is considered the norm;

    Total Sleep Time (TST) - the total duration of sleep - during which N1-3 and REM (rapid eye movement) phases take place;

    Sleep Efficiency (SE) - Sleep efficiency is the ratio of time spent asleep to time spent in bed, considered normal to be 80% or more;

    Wake after Sleep Onset (WASO) - the length of wakefulness after first falling asleep. WASO time reflects sleep fragmentation along with the number of awakenings;

    REM latency - REM sleep latency - the time from falling asleep to the first period of REM sleep, 70-110 min.

 

The use of the PSG in the ICU is complex in terms of technical and staffing issues. In addition, the use of PSG in inpatient monitored patients has been questioned because the assessment algorithms (the original Rechtschaffen and Kalesa, then its successor used since 2007: the American Academy of Sleep Medicine; AASM) [8,9] were developed for outpatients and ICU patients have an atypical record of brain activity [5]. Sedatives, opioids, stress response, and artificial pulmonary ventilation (UPV) affect brain activity [10], muscle tone, limb and eye movements, which tend to be monitored during PSG across modalities [11]. The ICU environment is also the source of numerous artifacts (motion artifacts, artifacts from UPV, contamination of the sensed electrical signals by other devices). In their study where ventilated ICU patients were monitored while awake, Drouot et al. reported that standard criteria could not be used to assess sleep in 28% and postulated two new conditions: atypical sleep (sleep with atypical EEG recording) and pathological wakefulness (slow EEG recording when the patient was awake) [12]. Although new assessment algorithms and methods are being developed to evaluate a subpopulation of ICU patients [13], no assessment algorithm is currently considered standardized in the critical care setting. Another limitation of PSG may be the difficulty of monitoring with all modalities when the patient is critically ill. The recording itself may exacerbate patient discomfort, and if taken during ICU operation, compliance with sleep laboratory parameters (control of noise, ambient noise, light exposure and other environmental aspects) must be ensured.

 

Bispectral Index (BIS)

A modified EEG modality, more suited to the ICU environment and critically ill patients, is the BIS (bispectral index) monitor, which records the EEG using only four electrodes and processes it into a numerical scale of 100 (corresponding to full wakefulness) - 0 (no brain activity). The BIS index values are shown in more detail in Table 2. Some studies verifying the possibility of using BIS to monitor sleep stages report that BIS values decrease during NREM sleep. However, the BIS index is falsely higher during REM sleep artifacts caused by rapid eye movements and overlap with the N1 stage, thus reducing the sensitivity of BIS [14]. Another study confirms the satisfactory specificity and sensitivity of the use of BIS to identify N3 stage sleep (BIS index 60-40 is also the target value during general anaesthesia). The other NREM (non rapid eye movement) stages cannot be individually distinguished, nor does the study admit an inability to distinguish between REM sleep and wakefulness. For further studies, the authors recommend the use of a combination of BIS and EMG or EOG [15]. The use of a combination was also recommended in a validation study of BIS and PSG in patients with 40h sleep deprivation where NREM stages correlated and the use of EMG activity was suggested to differentiate REM sleep [16]. Algorithmically processed EEG has the potential to become a modality capable of measuring sleep quality in the ICU, but so far studies have not achieved sufficient correlation between BIS and PSG [5]. The advantage of BIS over PSG is its lower technical and personnel requirements.

 

Actigraphy

An actigraph is a motion-tracking sensor - an accelerometer - nowadays used in most mobile phones and smart devices that the user wears. It uses algorithms to distinguish sleep and wake times according to different epochs of activity [17]. Sleep is described according to the aforementioned time parameters. The actigraph can be placed most often on the wrist or around the ankle, thus avoiding the introduced inputs or other devices used to monitor the vital signs of ICU patients. The main advantage is simplicity over PSG, but at the cost of lower method validity. The validity of the measurements is reduced or disappears completely in deeply sedated patients, absolutely in patients requiring myorelaxation, where there is no movement on the monitored limb/limbs. This overestimation of sleep time in sedated patients, and thus sleep efficiency compared to PSG, is shown in several comparative studies [17,18]. The same trend appears in a comparison with a questionnaire-based systematic observation of nursing staff [19]. Nevertheless, actigraphy achieves up to 90% concordance with PSG [20,21] and in 2007, the AASM recognized it as reliable and valid for the assessment of circadian rhythm disorders, whereas its validity has not yet been demonstrated for other sleep disorders (parasomnias, insomnias, hypersomnias, sleep-disordered ventilation, and others) [22].

 

Systematic observation

These are methods based on the collection of data by an observer, in the case of the ICU, by nursing staff who observe the patient's behaviour. They can take the form of continuous monitoring or retrospective data on sleep, e.g. from a video recording. Methods used include, for example, the Sleep Observational Tool, which has been validated with PSG. The attending staff performs a 5-second observation every 15 min for 4 h and assesses whether the patient is asleep, awake, or unable to express themselves. In 81.9%, the sleep assessment by the attending staff was consistent with the PSG [23]. Although other observational methods are available, their validation with other sleep assessment methods has not been conclusive [24] or only a weak correlation has been demonstrated [17]. By systematically observing sleep assessments, a cohort of patients who are unable to give subjective sleep assessments due to their condition can be followed. A limitation is the inability to distinguish some aspects of sleep, e.g. number of awakenings, time to fall asleep, and there may also be non-negligible interindividual differences in sleep ratings between raters.

 

Subjective assessment by the patient

When assessing sleep using subjective methods, we obtain data on the patient's perception of sleep and its quality. Several questionnaires are used to obtain this data. For example, Jeffs et al. in their review article analyzed 17 methods from 38 studies, most of which were not validated, and they recommend the use of the Richards Campbell Sleep Questionnaire (RCSQ) for further studies of sleep in the ICU [25].

The RCSQ is composed of five visual analog scales (VAS) that measure five areas of sleep: latency, efficiency, depth, number of awakenings, and overall quality. The RCSQ has been validated against the PSG [26]. Its use is also recommended by the PADIS guidelines for monitoring sleep disorders in the ICU [27]. The Czech translation of the RCSQ has been validated [28]. Among other questionnaires, the Verran Snyder-Halpern Sleep Scale (VSH) has been used in studies. It is a VAS composed of 9-15 items, originally developed and validated for the healthy population [29], however, it has also been used in the ICU setting [20]. Another scale is the Numeric Rating Scale for Sleep (NRS-Sleep) [30] a single-item scale from 0 (worst night's sleep) to 10 (best night's sleep), which was validated with the RCSQ in 2019. Unfortunately, the ability of the patient to answer the questionnaire alone is an unavoidable condition and limitation in the ICU setting.

 

Delirium monitoring

The diagnosis of delirium is facilitated by screening questionnaires, an overview of which is given in Table 3. The most commonly used questionnaire is the Confusion Assessment Method for the ICU (CAM-ICU) [31], translated into 19 languages [32], where a standardised translation into Czech has been validated [33], or the Intensive Care Delirium Screening Checklist (ICDSC) [34]. The use of the ICDSC appears to be more appropriate in patients with communication disorders (e.g. aphasia). The ICDSC has also been translated into English and published as part of the recommended practices of the Czech Neurological Society [35]. Screening for delirium using these methods is rapid (2-5 min) and shows high specificity and sensitivity [36]. One of the prerequisites for the use of the aforementioned questionnaire techniques is the need to assess patients' sedation/agitation using the Richmond Agitation-Sedation Scale (RASS).

Numerous studies in the ICU and beyond have found that without validated screening tools, bedside nurses and physicians unfortunately do not always recognize delirium [27]. The benefit of delirium screening tools has been documented in many cohorts, e.g., an association has been found between better treatment outcome in ventilated patients who were screened for delirium [37]. The above screening methods do not provide information on the severity of delirium, which is usually related to the outcome of patients; the Delirium Rating Scale - Revised-98 (DRS-R-98) [38] and the more recent CAM-ICU-7 [39] were developed to provide this information. Efforts are also underway to identify a biomarker associated with the risk or development of delirium. Frequently cited hypotheses in the development of delirium tend to be decreased cholinergic activity [40], melatonin deficiency, increased glutamate, noradrenaline or dopamine levels, and fluctuations in GABA (gamma-aminobutyric acid), histamine and serotonin levels [4]. Inflammation has also been considered in the etiopathogenesis; for example, C-reactive protein has been investigated [41]. Although many biomarkers have been studied, none has been validated for clinical use [32].

 

Paediatric population

Delirium is a common problem in paediatric intensive care, with an incidence of up to 49% according to various sources [42]. As in adult patients, its development has an impact on the length of ICU stay, mortality, length of UPV and the development of long-term cognitive impairment [43]. In pediatric patients, the recommended score for delirium screening is The Preschool Confusion Assessment Method for the ICU (psCAM-ICU) for children 6 months to 5 years and The Pediatric Confusion Assessment Method for the ICU (pCAM-ICU) for patients older than 5 years [44]. These methods are considered to be reliable, rapid, and have high specificity and sensitivity [44,45]. Other possible scoring systems include The Cornell Assessment of Paediatric Delirium (CAPD), which does not require the cooperation of the paediatric patient and can be used at any age, but has the disadvantage of being time consuming; the assessment should be preceded by observation of the patient by a nurse for 4-6 h [46].

 

Predictive models of delirium

Given the recommendation to use multimodal non-pharmacological strategies to prevent the development of delirium [27], there is a tendency to predict the development of delirium based on risk factors. Two models, the PREdiction of Delirium in ICu patients (PRE-DELIRIC) [47,48] with 10 predictors and the E-PRE-DELIRIC model with nine predictors, have been developed and validated to identify patients at high risk of developing delirium for subsequent prevention of delirium development. Their limitations are that many of the risk factors are not modifiable, the so-called predisposing factors (e.g., age, APACHE II score), and there are no consistent data showing improvement in the patient's health status after discharge (outcome) when known risk factors are reduced. The majority of risk factors arise during the course of the disease itself, during the ICU hospitalization (e.g., metabolic abnormalities or environmental factors) [1]. These risk factors are referred to as precipitating factors. Precipitating factors could be monitored during hospitalization and completely avoided or at least reduced. With the ongoing development of artificial intelligence, machine learning models that process health data from electronic health records are beginning to be used in the problem of predicting delirium development. Their ability to predict delirium is accurate, with the advantage of continuous dynamic assessment over time [49].

 

Conclusion

Currently, there are several validated options for screening, monitoring and diagnosing sleep in the ICU setting, but each method has its limitations and cannot be universally applied to all ICU patients. It is also one of the reasons why there is limited identification of the clear impact of poor sleep quality and the development of delirium on patient outcomes, both in the short and long term. The use of multiple modalities in parallel and the systematic conduct of further research are essential to answering any questions regarding the relationship between sleep and delirium in the ICU.

 

Grant support

Supported by the Ministry of Health of the Czech Republic - RVO (FNBr, 65269705), the Ministry of Education and Science of the Czech Republic through the VVI CZECRIN project (LM2023049) and the MU Brno specific research project (MUNI/A/1186/2022, MUNI/A/1109/2022 and MUNI/A/1105/2022).

 

Conflict of interest

The authors declare that they have no conflict of interest in relation to the topic of the paper.

 

Tables

Table 1: Comparison of sleep quality and quantity monitoring methods.

Name of method		Benefits	Limitations
Monitoring	PSG	* gold standard * robust method * highest validity of the data obtained	* demanding in terms of personnel and technical requirements * in an intensive care setting: - Lack of a specific assessment algorithm, available methods are developed for the sleep laboratory - abnormal records often occur during sleep in the ICU - interference of PSG modalities with other devices - PSG monitoring alone can exacerbate patient discomfort
	EEG modifications (e.g. BIS index)	* less demanding than PSG * fast data interpretation and the ability to react to changes in status	* during REM sleep the BIS index is falsely elevated, mimicking the N1 phase, to differentiate it is necessary to use another method EOG, EMG * still no clear correlation with PSG * disposable set of EEG electrodes
	Actigraphy	* low intensity compared to PSG and BIS * the device can be used repeatedly	* necessity to preserve the patient's spontaneous momentum, otherwise non-valid - cannot be used in deeply sedated patients, - in patients requiring muscle relaxation, - patients with neurological deficits - other conditions where the patient's spontaneous movement is restricted
Questionnaire studies	systematic observation	* no technical equipment required * can be performed by nursing staff * can be done retrospectively (from video footage)	* only one test is validated against PSG (Sleep Observational Tool) * fail to distinguish some aspects of sleep - number of awakenings - time of falling asleep * inter-individual differences in ratings between individual raters arise
	subjective assessment by the patient	* is the only method that provides data on the patient's perception of sleep quality	* the patient must have sufficiently preserved consciousness qualitatively and quantitatively to complete the questionnaire

BIS - bispectral index; EMG - electromyography; EOG - electroculography; ICU - intensive care unit; PSG - polysomnography; REM - rapid-eye movement

 

 

Table 2: BIS index range.

BIS index	Depth of seating/anesthesia
100	full consciousness
80-60	lightweight seating
60-40	target values for general anaesthesia
40-20	deep anaesthesia
20-0	burst suppression
0	isoelectric EEG line

BIS - bispectral index; EEG - electroencephalography