Feasibility and Acceptability of Using Wireless Limited Polysomnography to Capture Sleep Before, During, and After Hospitalization for Patients with Planned Cardiothoracic Surgery

Abstract

Background:

Sleep disruption, a common symptom among patients requiring cardiovascular surgery, is a potential risk factor for the development of post-operative delirium (POD). POD is a disorder of acute disturbances in cognition associated with prolonged hospitalization, cognitive decline, and mortality.

Objective:

To evaluate the feasibility and acceptability of using polysomnography (PSG) to capture sleep in patients with scheduled cardiothoracic surgery.

Methods:

Wireless limited PSG assessed sleep at baseline (pre-surgery at home), post-operatively in the intensive care unit (ICU), and at home post-hospital discharge. Primary outcomes were quality and completeness of PSG signals, and acceptability by participants and nursing staff.

Results:

Among 15 patients, PSG data were of high quality, and mean percent of un-scorable data was 5.5±11.1%, 3.7±5.4%, and 3.7±8.4% for baseline, ICU, and post-hospitalization measurements respectively. Nurses and patients found the PSG monitor acceptable.

Conclusions:

Wireless, limited PSG to capture sleep across the surgical continuum was feasible and data were of high-quality. Future studies will evaluate associations of sleep indices and development of post-operative delirium in this high-risk population.

Patients requiring cardiovascular surgery are at high risk for developing post-operative delirium (POD).¹ Delirium is characterized as an acute form of cognitive dysfunction that fluctuates over time and is associated with prolonged intensive care unit (ICU) and hospital lengths of stay and post-hospitalization mortality.² Importantly, delirium is an independent risk factor for incident dementia and long-term cognitive decline post-hospitalization.^2,3 Incidence of POD among patients requiring cardiac surgery ranges from 11–52%, where patients of older age with pre-existing comorbidities and a complicated post-operative course are at highest risk.^1,4 Although some risk factors may be non-modifiable and mechanistic pathways are complex, it is hypothesized that development of POD is preventable in many cases, or at least the severity and duration could be attenuated. One important potential contributor to the development of POD is sleep disruption.

High quality sleep of sufficient duration is a cornerstone of overall health and wellbeing.⁵ Insufficient and poor-quality sleep are associated with a myriad of both short- and long-term negative health outcomes including impaired cognitive function, increased symptom burden, incident dementia, and increased all-cause mortality risk.^6–12 Sleep disruption, which can be defined as sleep that is abnormal in duration, timing, continuity, or quality, is commonly experienced by patients with cardiovascular disease, especially if advanced enough to require surgical intervention.^6,7,13–15 In the immediate post-operative period, patients are at high risk for sleep disruption caused by the hospital environment, illness severity, and pharmacologic agents.^16,17 Patients also report poor sleep quality following hospitalization for cardiothoracic surgery.^18–20 For example, in a study of N=52 patients undergoing coronary artery bypass graft surgery, sleep quality (assessed using the Pittsburg Sleep Quality Index where a score >5 indicates poor sleep quality²¹) was significantly worse one month after surgery relative to pre-operation (from approximately 3.7 vs 7.5).

Sleep disruption is hypothesized to contribute to the development of delirium through several mechanisms. Acute disruption of the sleep-wake cycle, especially in the presences of sleep loss, is linked with increased oxidative stress, reduced clearance of brain extracellular metabolites, impaired blood-brain barrier regulation, neurotransmitter dysregulation (especially acetylcholine suppression), and activation of inflammatory pathways, all of which are recognized pathways of delirium pathogenesis.^2,22 Sleep-wake disturbance is also a key symptom exhibited by patients with delirium,² and baseline (pre-operative) sleep disruption has been linked to development of POD.²³ However, few studies have evaluated sleep in patients requiring cardiovascular surgery, and even fewer have included objective sleep parameters.

In-laboratory polysomnography (PSG), including brain activity monitoring via electroencephalography (EEG), is the gold standard to assess physiologic sleep.²⁴ However, use of traditional PSG requires expertise to apply, is impractical long-term, and has limited capabilities for routine use in most settings. Alternatives to assess objective sleep, such as actigraphy, are of limited utility as they rely on gross body movements to infer sleep, and lack the ability to capture EEG features that may elucidate differences in sleep-wake states important to delirium identification.²⁵ Therefore, there is a need to identify additional approaches to capture sleep that could be used in both the home and hospital environment as a first step to identifying sleep-related indices associated with POD risk. The purpose of this pilot study was to determine the feasibility and acceptability of using wireless, limited PSG to capture signals for objective sleep assessments at baseline (pre-surgery at home), post-operatively in the ICU, and at home post-hospital discharge. Results from this pilot will be used for a full-scale observational study evaluating relationships between sleep indices and development of POD.

METHODS

Adults (≥18y) with a scheduled cardiothoracic surgery who were free of diagnosed sleep disorders, not taking habitual sleep aids, and not hospitalized at time of enrollment were eligible to participate. Exclusion criteria were allergy to adhesive, an inability to sleep on back or side, and scheduled surgery for which post-operative ICU admission was not anticipated. This was a single-center study where participants were recruited pre-operatively from a Heart and Vascular Center, and in-hospital sleep assessments were performed in a Heart and Vascular ICU at a large academic medical center in Philadelphia, PA. Due to the Covid-19 pandemic, all subject recruitment was done virtually over the phone or via videoconferencing. Participants were identified through medical record review of patients with a scheduled open cardiothoracic procedure (e.g. coronary artery bypass graft, valve repair/replacement, aortic repair/replacement, etc.) within the next 30 days and were screened for initial eligibility. Qualified patients were contacted via phone to determine interest and full eligibility. If interested and qualified, study materials were sent in a pre-paid package to participant’s home address. Data collection took place from March 2021-August 2022. This study was approved by the Institutional Review Board at the University of Pennsylvania and subjects provided written informed consent prior to study participation.

Sleep assessments were performed during a single overnight sleep period at three timepoints; 1) baseline (approximately 2 weeks pre-operatively in the home), 2) in the ICU postoperatively (approximately 2200–0600 on day 0 or 1), and 3) approximately 2–4 weeks post-hospital discharge at home. Sleep was assessed using the Prodigy Sleep System (Cerebra Health, Winnipeg, Canada) (Figure 1A). The head mounted unit (weight 70g) was attached to the forehead by 4 disposable snap electrodes (2 Neuroline 720 00-S electrodes [Ambu, Copenhagen, Denmark] for EEG; and 1 Nicolet disposable electrode Duo [NT-429400, Natus, Pleasanton, CA] for ground and bias) and powered by 2AAA batteries (battery life ~12h). Four additional Neuroline electrodes connected to the head-mounted unit through touch-proof connectors were attached near the left and right eyes, chin, and left mastoid to monitor electrooculography (EOG), electromyography (EMG), and reference respectively. All signals were transmitted wirelessly at 120 Hz to a tabletop unit. For each in-home assessment, the PSG monitor was mailed to participants and was self-applied following written and video instructions. Participants were asked to return study equipment following each in-home assessment (baseline and post-hospitalization) using pre-paid postage. Study personnel applied the PSG monitor in the ICU. Prodigy has been shown to provide quality signals comparable to full PSG in other populations.^26,27 Primary outcomes were quality and completeness of PSG signals.

Figure.

Depiction of sleep monitor and example signal output and interpretation

A: The head-mounted unit attached to the forehead showing standard electrode placement; B: An example of data collected by the sleep monitor as displayed in the scoring viewer showing a 30-sec epoch from the 6 data signals; C: Example of hypnogram showing sleep stages across the night derived from polysomnographic signals at C1: Baseline (at home pre-hospitalization); C2: In the intensive care unit post-operatively; and C3: Post-hospitalization at home.

Permission: Figure 1A used with permission of J Clin Sleep Med, from Performance of a New Portable Wireless Sleep Monitor, Younes M, et al., 2017;13(3):245–258; permission conveyed through Copyright Clearance Center, Inc.

For each sleep assessment, PSG signals were downloaded from the tabletop unit using Prodigy Sleep Monitor software (version 1.00.008) and channel (2 EEG, mastoid, EOG, and EMG) outputs were reviewed and exported in the European data format to the Michele Sleep Scoring software which applies an automatic scoring algorithm based on 2007 American Academy of Sleep Medicine criteria.²⁶ Scored PSG files were then visually inspected by a trained PSG technologist for accuracy with adjustments to scored sleep stages made as needed. After editing, a report was generated for each study containing numerical and graphic summaries of results.

Feasibility and acceptability from both participants and primary ICU nursing staff were assessed. If able, participants were asked about comfort and ability to sleep while wearing the monitor in the ICU. Following hospitalization, participants were interviewed regarding their experience using the monitor, their overall sleep experience, and provided any additional feedback. The ICU nurses were asked if the monitor interfered with patient care, if they had any concerns with the monitor, and provided additional feedback regarding study methods. Primary outcomes were patient acceptability and willingness to wear the monitor, and acceptability by ICU nursing staff.

RESULTS

A total of 20 patients agreed to participate in this study. One patient became ill pre-operatively requiring hospitalization, three patients withdrew consent prior to surgery due to high anxiety regarding their surgery, and one patient failed to return the sleep monitor following the baseline assessment. Of the 15 included participants, average age (±standard deviation) was 59.5 (±16.2) with 66.7% male (n=10). Types of cardiovascular surgery varied, with 2, 8, and 5 patients receiving a coronary artery bypass graft, valve repair/replacement, or aortic arch/ascending aortic repair respectively. Postoperatively, all patients received analgesic medication, and 6 (40%) were on mechanical ventilation with sedation during the sleep assessment. A total of 2 patients were found to have delirium during their hospitalization (13.3%), and median hospital length of stay was 6 (Q1 5.25, Q3 8) days. The majority of patients were discharged to home (73.3%, n=11), with the remainder being discharged to an acute rehabilitation center (26.7%, n=4).

Overall data from PSG channels were of high quality (Figure 1B), with only one instance of signal dropout (<1 sec) and no noted interference with medical equipment or adverse events. Average ± standard deviation percent of “bad” or un-scorable PSG was 5.5 ±11.1%, 3.7 ±5.4%, and 3.7 ±8.4% for the baseline, ICU, and post-hospitalization measurements respectively. Reasons for poor signal quality included EEG electrodes lifting from the patient’s forehead during the night. This was especially prevalent in the ICU where presence of forehead moisture contributed to poor overnight electrode adherence to the skin which was identified early-on in the study. Sub-optimal electrode adherence was remedied with additional securement using medical tape (3M, 0.5 in Durapore^™) placed over forehead electrodes. The chin lead was dislodged in 2 instances in the ICU. Alternative placement to the muscle (sternocleidomastoid below the mastoid process) behind the right ear per manufacturer’s recommendation was found to be preferred and did not negatively affect data quality. Thus, alternative placement was used for subsequent ICU assessments. Two patients in the ICU were successfully extubated with the sleep monitor in place, which did not interfere with patient care.

As expected, staging of sleep was more challenging when patients were in the ICU compared to home assessments given the presence of sedating and analgesic medications that can cause atypical sleep patterns. Thus, more manual scoring was needed for these assessments. Compared to baseline, patients in the ICU spent more time in lighter stages of sleep (stages 1/2), with a decrease in stage 3 and rapid eye movement (REM) sleep (Table and Figure 1C). Following hospitalization, average proportion of time spent in each sleep stage generally returned to baseline.

Table.

Average sleep stages for baseline, Intensive Care Unit (ICU), and post-hospitalization sleep assessments

Sleep stage,α mean (standard deviation)	Baseline (pre-hospitalization)	ICU (post-operative)	Post-hospitalization
N1 %	13.7 (4.4)	10.4 (15.8)	18.1 (10.2)
N2 %	65.3 (10.3)	81 (19.2)	66.1 (8.6)
N3 %	8.9 (5.4)	3.1 (4.1)	8 (11.5)
REM1 %	12 (9)	1.3 (2.6)	8.9 (8.3)
NREM2 %	88 (9)	98.4 (3.0)	91.1 (8.3)

^αUsing American Academy of Sleep Medicine 2007 scoring criteria,²⁴ percentage based on total sleep time

¹Rapid eye movement sleep

²Non-rapid eye movement sleep

Included patients and ICU nursing staff found the sleep monitor acceptable. No patients asked for, or removed, the monitor overnight due to discomfort, and non-sedated patients in the ICU stated they could sleep with the monitor in place. Those who were able to be interviewed post-hospitalization stated they would be willing to wear the sleep monitor in the ICU. Nursing staff reported no interference with patient care.

DISCUSSION

In this pilot study of 15 patients with planned cardiothoracic surgery, use of wireless, limited PSG to capture signals for sleep assessments overnight in both the home and hospital setting was found to be feasible and acceptable for both patients and ICU nursing staff. The PSG signals were of high quality with use of medical tape to secure forehead leads, enabling scoring of brain electrical activity as stages of sleep or wake. Further, there were no instances of signal interference with other devices in the home or ICU, and no adverse events occurred.

Although sleep is a foundational component of health and recovery, sleep is not routinely assessed in either the outpatient or inpatient setting.^28,29 Thus, the extent to which sleep is disrupted following cardiothoracic surgery, and sleep’s relationship with post-operative outcomes such as POD, is unknown. In the outpatient setting, it may be possible to assess aspects of sleep by self-report or by using ambulatory methods such as actigraphy.^30,31,23 However, these tools are of limited use in hospital settings, especially in the ICU, as patients are often unable to self-report sleep, and relying on gross body movements required for actigraphy has been shown to consistently overestimate sleep duration,^25,32 especially if patients are immobile. Further, actigraphy does not capture aspects of sleep architecture such as sleep stages or EEG features such as sleep spindles that maybe important markers of delirium risk.^33,34 In order then for a comparable assessment of sleep across the cardiothoracic trajectory, objective measures that capture aspects of physiologic sleep are likely optimal.

Full PSG, which includes a number of EEG electrodes as well as EOG, EMG, limb leads and thoracic/abdominal movement monitoring, is logistically challenging and not well tolerated by patients in ICU settings.³⁵ Limited PSG, which uses fewer EEG electrodes without limb and thoracic/abdominal monitoring, has been shown to provide data quality allowing for assessment of objective sleep similar to full PSG in both the home and hospital setting.^26,36,37 The present study demonstrates that wireless limited PSG (2 frontal EEG channels with EOG and EMG) captured high quality electrical signals, at least overnight, in both sedated and awake patients, and in spontaneously breathing and mechanically ventilated patients in the ICU, and in the home setting that was self-applied by the patient, from which objective sleep could be assessed. Further, wireless sleep monitoring was well tolerated by participants and did not interfere with patient care in the ICU. Although the limited number of electrodes may not capture sleep disorders or oscillatory sleep activity in other brain regions,³⁸ wireless, limited PSG represents a feasible and acceptable approach to capture stages of sleep and objective components of sleep quality. Capturing changes in physiologic sleep would then allow for the identification of sleep indices associated with development of POD, and would serve as targets for intervention development to reduce POD risk.

The aim of this study was not to evaluate the physiologic parameters of sleep itself. In this small sample, we did however observe sub-optimal physiologic sleep in the home environment (baseline and post-hospitalization). For example, in healthy adults, the percent of total sleep that should be spent in REM is approximately 25%.³⁹ In our sample, we observed only 9% REM sleep at baseline and 8.3% post-hospitalization. This may be because we only assessed sleep for a single night and participants were not habituated to the monitor. It may also represent the known poor quality of sleep experienced by patients with cardiovascular disease,¹⁵ especially when advanced enough to require surgical intervention. Further, although we aimed to capture baseline sleep approximately 2 weeks prior to the scheduled surgery, participants reported feeling very anxious about their upcoming surgery which may have negatively affected their sleep.⁴⁰ Post-hospitalization, patients reported some ongoing pain and difficulty “getting comfortable” which may somewhat explain the continued sleep disruption.⁴¹

Nurses play an important role in both sleep promotion and delirium prevention for patients following cardiovascular surgery.^29,42,43 It is well-recognized that patients in the ICU experience significant sleep disruption caused by the hospital environment (noise, nursing care activities, etc.), illness severity, and pharmacologic agents.^16,17,29 These sleep disrupting factors are also known contributors to delirium development and severity.^2,29 Concomitant activities for sleep promotion and delirium prevention are often nursing centered. For example, orienting patients to a day-night schedule, mobilization, providing sleep hygiene, managing pain, clustering care activities overnight, managing unit noise, etc., are all non-pharmacologic nurse-driven interventions.^2,42,43 Ongoing assessment and monitoring of sleep and delirium are also part of nursing clinical practice. However, specific aspects of sleep that may contribute to POD have not been rigorously investigated. Thus, the learnings from this pilot study will be applied to a full-scale observational study with the aim of identifying specific sleep indices that are associated with development of POD. We hope to use these findings to develop interventions that promote sleep and mitigate delirium risk which would be directly applicable to nursing care of patients following cardiovascular surgery.

There are several notable limitations. We included a single patient population from a single center, and thus wireless sleep monitoring may not be feasible in other populations or at other institutions. Patients with scheduled cardiovascular surgery generally receive standard medications with a standard recovery progression which provided consistency across patients, and allowed us to capture both mechanically ventilated and spontaneously breathing patients. As most patients are neurologically intact post-hospitalization, we were also able to capture sleep recovery after hospitalization, and interview participants regarding their experience using the monitor. Importantly, all patients received analgesic medications, and some received sedatives during the ICU sleep assessment which may have impaired their awareness of the monitor at that timepoint. We did not assess sleep during the daytime or across multiple days, and thus total sleep time, in the ICU in particular, was not captured. Future planned studies will assess physiologic sleep for consecutive nights to minimize the first night effect. As this study focused on feasibility, we did not capture reasons why sleep may or may not be disturbed, such as environmental contributors, specific medications, or nursing care activities, which limit interpretability of sleep findings. Full PSG is the gold standard to assess physiologic sleep. As we only utilized limited-channel PSG, we may have missed important EEG signals that would otherwise be captured with the full range of electrodes.

CONCLUSIONS

Capturing physiologic signals for sleep assessments using wireless, limited PSG in both the home and hospital setting in patients with planned cardiothoracic surgery is feasible. Polysomnographic data were complete and of high-quality. Both ICU nurses and patients found the PSG monitor acceptable. Findings from this pilot study will inform a larger planned cohort study evaluating associations of physiologic sleep and POD risk in patients requiring cardiovascular surgery.

What’s New.

• Patients requiring cardiothoracic surgery are at high risk of experiencing postoperative delirium, which is associated with prolonged hospitalization, mortality, and is an independent risk factor for dementia.

• Sleep disruption during the pre-surgical and post-operative period may contribute to development of post-operative delirium, however, the sleep-delirium relationship in the cardiothoracic surgical population has not been rigorously investigated, and novel approaches to capture physiologic sleep have not been evaluated.

• In this pilot study, use of wireless limited polysomnography to capture physiologic sleep in the home (at baseline pre-surgery and post-hospitalization) and in the hospital for patients with planned cardiothoracic surgery was feasible and found to be acceptable by both patients and clinical nursing staff.