Unrecognised sleep disturbances in patients with cirrhosis diagnosed with a portable electroencephalogram device

Abstract

Objectives

We aimed to compare sleep characteristics between patients with liver cirrhosis and healthy controls using a standardised protocol and portable electroencephalogram (EEG) devices.

Methods

We enrolled patients with early stage cirrhosis at low risk for sleep disorders (no apnoea, insomnia, alcohol use, pruritus or major portosystemic shunt; body mass index (BMI) ≤31 kg/m²). Using propensity score matching (age, sex, BMI), 18 patients with cirrhosis were compared with 18 healthy older adults from a 95-person cohort. Sleep was assessed at home using portable EEG devices measuring total sleep time, sleep latency, wake after sleep onset, sleep efficiency, sleep stages (N1–N3, rapid eye movement (REM)) and REM latency. Questionnaires were also administered.

Results

Questionnaires indicated no major sleep complaints. However, EEG revealed longer sleep latency, increased wakefulness and lower sleep efficiency in cirrhosis. N1 sleep time and percentage were higher, REM sleep was reduced and REM latency was prolonged.

Discussion

Traditional assessments rely on subjective reports, while polysomnography is often impractical. Our portable EEG approach revealed distinct disturbances—fragmented REM and delayed onset—undetectable by questionnaires alone.

Conclusion

Home EEG monitoring uncovered previously unrecognised sleep abnormalities in cirrhosis, suggesting utility for early detection and management.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Sleep disturbances are common in patients with liver cirrhosis.

• To date, prior studies have relied almost exclusively on subjective questionnaires, and objective polysomnographic assessments are rarely performed because of their complexity and limited feasibility in clinical research.

WHAT THIS STUDY ADDS

• Using a portable electroencephalogram (EEG) device with matched healthy controls, we objectively identified previously unrecognised alterations in sleep architecture—marked by increased sleep latency, reduced sleep efficiency, prolonged rapid eye movement latency and shifts in stage distribution—in patients with cirrhosis.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our findings demonstrate that portable EEG is a practical, scalable tool for sleep assessment in patients with cirrhosis, highlighting the need for routine screening for sleep disorders in this population and informing future guidelines for sleep evaluation and management of chronic liver disease.

Introduction

Cirrhosis has become increasingly prevalent globally, partly due to population growth and ageing. Deaths due to cirrhosis accounted for 2.4% of global deaths in 2017, compared with 1.9% in 1990. Sleep-wake disturbance is common in patients with cirrhosis; 50%–80% of them experience disturbances five times higher than healthy individuals do. However, polysomnography, the gold standard for assessing sleep architecture, is extremely difficult to conduct in research due to instrumental complexity and insurance coverage; moreover, knowledge about it is rather limited.¹

Medical history and physical examinations can help identify patients at risk of cirrhosis. Patients with cirrhosis frequently experience muscle cramps (64%), pruritus (39%), poor sleep quality (63%) and sexual dysfunction (53%).² However, many of the risk factors, such as diabetes or alcohol use, along with the symptoms, are neither sensitive nor specific.³ Unlike polysomnography, the use of portable electroencephalogram (EEG) devices does not require large machines or hospitalisation, and can be applied at home, reflecting normal sleep patterns. The performance of the specific device used in this study has only been studied in healthy control groups and patients with sleep apnoea syndrome. However, it has not been studied in individuals with cirrhosis, who might have a high prevalence of comorbid sleep disorders.⁴

This study conducted a comparison of patients with pure cirrhosis with healthy controls on the American Academy of Sleep Medicine’s (AASM) 2017 criteria.

Methods

Ethics statement

The study protocol conformed to the guidelines of the Declaration of Helsinki (as revised in Fortaleza, Brazil, October 2013). Additionally, we used the data from the study by Seol et al.⁵

Patients were enrolled from April 2021 to September 2022 in the cirrhosis group and from October 2020 to December 2021 in the healthy control group.

The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting observational studies, and a complete checklist is provided as online supplemental materialonline supplemental material.

Study design, participants and settings

Inclusion criteria

Patients with cirrhosis were enrolled from Niigata University Graduate School of Medical and Dental Sciences; healthy control participants were recruited from the University of Tsukuba between 2021 and 2023. Eligible patients included outpatients with Child-Pugh class A cirrhosis who had not used sleeping pills in the past 6 months and had a body mass index (BMI) of ≤31 kg/m² (a common cut-off for sleep apnoea diagnosis).^{6 7} Participation was voluntary; all participants provided written informed consent.

All participants were fitted with the same portable EEG device (Insomnograf; S’UIMIN, Tokyo, Japan). The screening procedures (in-clinic anthropometrics and in-person clinical interview) were identical at both sites. Patients with cirrhosis were outpatients attending follow-up at Niigata University Medical and Dental Hospital; EEG recordings were performed at home over seven consecutive days, of which the mean values were used for analysis. The control group comprised a cohort from the Sleep Science Center at the University of Tsukuba, who underwent home recordings with the same device for 7 days, where 7-day averages were calculated. In addition to EEG recording, patients with cirrhosis also underwent serum ammonia testing, hepatic reserve assessment and the Epworth Sleepiness Scale (ESS). The collected data were then compared between the patient and control groups.

The sleep data of each patient were recorded for 7 days; the average values were used for the analysis. If data collection on a given day was inadequate, it was repeated on a different day.

Exclusion criteria

To minimise confounding from conditions associated with sleep disturbances, we excluded patients (1) With a major shunt (mainly splenorenal) on CT, which may cause hyperammonaemia;⁸ (2) With moderate to very severe itching based on the Verbal Rating Scale (none, mild, moderate, severe, very severe) that interfered with daily activities;⁹ (3) With clinically apparent encephalopathy or psychiatric comorbidities; and (4) With alcohol use disorder or reported daily alcohol intake of 20 g within the last 6 months.

Reference population

Overall, 95 healthy volunteers (75% male, mean age=75.9 years, mean body BMI=23.6 kg/m²) were recruited from the University of Tsukuba. Objective daily sleep parameters were recorded using the same portable EEG device (Insomnograf; S’UIMIN, Tokyo, Japan) for both patients with cirrhosis and healthy controls to ensure consistency in measurement procedures. Using propensity score matching based on sex, age and BMI, the study participants were matched with a calliper of 1. No healthy volunteers had reported evidence of alcohol abuse, chronic liver disease or neurological/psychiatric disorders; consumed >20 g of alcohol daily; or took prescription drugs.

Epworth Sleepiness Scale

Conventional sleepiness was assessed using the ESS (figure 1A). Participants were asked to rate their likelihood of ‘falling asleep’ in eight different situations during the day on a scale of 0 (unlikely) to 3 (very likely); the eight scores were summed up to result in a total score between 0 and 24. A score higher than 11 points was considered as strong daytime sleepiness.¹⁰

Figure 1. (A) Epworth Sleepiness Scale. (B) Characteristics of sleep stages.

Determining sleep stages in adults

The AASM published a scoring manual on non-rapid eye movement (REM) sleep, which is now recommended for determining sleep stages.¹¹ According to the AASM 2007 Scoring Manual, adult sleep stages are classified as follows. Wakefulness is designated as Stage W, which means wakefulness. Non-REM sleep is divided into two stages: Stage N1, which is light sleep (Stage 1 of non-rapid eye movement (NREM)sleep) and Stage N2, which is intermediate sleep (Stage 2 of NREM sleep). Deep sleep, also known as slow wave sleep, is classified as Stage N3 (Stage 3 of NREM sleep). REM sleep is designated as Stage R (figure 1B).

Characteristics of the sleep stages

The characteristics of each sleep stage are summarised in figure 1A. Sleep staging was performed in 30 s epochs. Stage N1 shows reduced alpha activity (8–13 Hz) compared with wakefulness, with predominant 4–7 Hz activity. Stage N2 is defined by sleep spindles and K-complexes, with <20% slow wave activity. Stage N3 is characterised by slow wave activity (0.5–2.0 Hz, >75 µV) occupying >20% of an epoch. Stage R shows low-frequency activity similar to N1 (alpha 8–13 Hz <50% of the time, dominant 4–7 Hz) and is accompanied by sawtooth waves, REMs on electro-oculography and reduced muscle tone on electromyography compared with non-REM sleep.¹²

Sleep variables

Objective daily sleep parameters were recorded using a portable EEG device (Insomnograf; S’UIMIN, Tokyo, Japan) (figure 2A). The device is lightweight (162 g) and easy to attach/remove; different montages are available, and electrooculography (EOG)/electromyography (EMG). components can be displayed by adjusting filtering conditions. Compared with typical polysomnography, it showed high concordance (0.80 in healthy controls and 0.78 in patients with sleep apnoea).¹³

Figure 2. (A) Features of the portable electroencephalogram (EEG) device potential data to be measured. (B,C) Smartphone screen display. REM, rapid eye movement.

Participants attached adhesive electrodes after showering and pressed the recording button immediately before sleep and after waking. Bedtime and final morning wake-up time were determined from these data; recording ended when the power button was pressed or a charging cable was inserted. We crosschecked device-derived bed/wake times with self-reported questionnaires to secure timing data even if participants forgot to press the button after waking.

The system comprised four EEG electrodes (Fp1, Fp2, M1 and M2) and one reference electrode (Fpz) placed according to the 10–20 system. Four derivations were generated (Fp1–M2, Fp2–M1, Fp1–average M and Fp2–average M). Fp1–Fp2 and Fp2–Fp1 were used for left and right electro-oculography, respectively, and M1–M2 for chin electromyography for sleep staging. Records were scored in 30 s epochs as wakefulness (stage W), non-REM (stage N: N1, N2, N3) and REM; sleep-onset latency epochs were classified as stage W. REM latency was defined as time to the first REM epoch after sleep onset, and wake after sleep onset was defined by standard criteria (AASM 2017).

Statistical analysis

Propensity score matching was performed based on a previous study (calliper=1.0).⁴ Group differences were evaluated using analysis of variance followed by paired-samples t-tests in the matched cohorts, with mean differences and 95% CIs reported. Effect sizes were calculated using Cohen’s d (small, 0.2; medium, 0.5; large, 0.8).¹⁴ Bonferroni correction was applied for multiple comparisons (n=13), and both adjusted and unadjusted values of p were reported. All analyses were performed using IBM SPSS Statistics (V.29.0; IBM Corporation, Armonk, New York, USA) and Microsoft Excel V.2408 was used for illustration; statistical significance was set at p<0.05 (two-tailed).

Details of inclusion and exclusion criteria and the matching procedure are summarised in online supplemental figure S1.

Results

Patient characteristics

Propensity score matching was used to select 18 from the 20 patients with cirrhosis and 18 from the 95 healthy controls. Overall, 18 participants were included (15 male individuals, mean age=72.9 years). Furthermore, 12 participants were diagnosed with metabolic dysfunction-associated steatohepatitis; 8 were treated for liver cancer. All participants had a Child-Pugh A reserve capacity, and no patient had more than first-degree encephalopathy according to the West Haven criteria classification, with a mean ammonia value of 70.3±5.5 µg/dL. The ESS was 6.2±2.3 points; no patient showed severe sleep disturbance in the ESS results (table 1).

Table 1. Comparison of sleep parameters in patients with liver cirrhosis and healthy controls.

	Patients with liver cirrhosis (n=18)	Healthy controls (n=18)	P values
Age (years)	72.9±8.3	75.9±3.9	0.924
Sex, male/female	15 (83%)	15 (83%)	1
BMI (kg/m2)	24.5±4.3	23.6±3.3	0.945
Aetiology, HBV/HCV/MASH/AIH/PBC	2/2/12/1/1
HCC, presence/absence	8/10	0
ALBI Score	−2.26±0.31	*
AST (U/L)	25.1±2.2	19.2±3.5	0.723
ALT (U/L)	21.3±3.3	15.2±2.3	0.546
NH3 (μg/dL)	70.3±5.5	*
ESS (points))	6.2±2.3	*

Values are presented as mean±SD or n (%), unless otherwise noted.

^*Missing values were considered within normal limits based on clinical context.

AIH, autoimmune hepatitis; ALBI, Albumin-Bilirubin Score; ALT, alanine transaminase; AST, aspartate aminotransferase; BMI, body mass index; ESS, Epworth Sleepiness Scale; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MASH, metabolic dysfunction-associated steatohepatitis; NH₃, ammonia; PBC, primary biliary cholangitis.;

Sleep quality evaluation

A feature of this device allows data analysis to be performed immediately on recharging with a smartphone application. Data were collected from all 20 patients.

No significant differences were observed between the healthy controls and patients with cirrhosis in age, sex and BMI. However, patients with cirrhosis had lower Albumin-Bilirubin Scores and higher ammonia levels than the control group.

Figure 2B shows a really good quality case, deep sleep EEG Score of 910/1000 (placing in the top 5%), with a total sleep duration of 7 hours and 1 min. Overall sleep quality was 88/100, took 14 min to fall asleep, spent 51 min awake during the night and sleep efficiency was 87% (figure 2B).

Figure 2C shows a bad quality case, EEG Score was 190/1000 (placing in the bottom 10%), with a total sleep duration of 3 hours and 55 min. Overall sleep quality was 41/100, took 11 min to fall asleep, spent 6 hours and 2 min awake during the night and the sleep efficiency was 39% (figure 2C). More than half of the patients experienced a phenomenon known as sleep misrecognition, where they were objectively sleep deprived even though they thought they were actually sleeping well. In the control group, no participant had >11 points on the ESS. However, the portable EEG device revealed a high degree of dissociation between the two groups regarding sleep efficiency. Although 90% of the respondents had scores of ≤5, indicating almost no sleep disturbance, more than half had a decrease in sleep efficiency (table 2).

Table 2. Comparison of sleep parameters in patients with liver cirrhosis and healthy controls.

	Patients with liver cirrhosis (n=18)	Healthy controls (n=18)	Differences (95% CI)	Effect size (Cohen’s d)	t value	DF	P value	Bonferroni-adjusted P value
Total sleep time (min)	365.6±74.6	382.9±34.6	−17.2 (−56.7 to 22.3)	−0.298	−0.920	17	0.383	1.000
Sleep latency (min)	45.4±31.4	15.5±8.6	29.9 (15.0 to 44.8)	1.299	4.223	17	<0.001	0.007
WASO (min)	168.3±68.0	71.1±29.3	97.2 (70.0 to 124.3)	1.856	7.549	17	<0.001	<0.001
Sleep efficiency, %	63.0±11.0	81.41±5.6	−18.4 (−24.0 to 12.8)	−2.109	−6.964	17	<0.001	<0.001
N1 (min)	62.0±37.5	18.2±11.1	43.8 (23.1 to 64.5)	1.584	4.470	17	<0.001	0.004
N1 (%TST)	17.9±10.7	4.7±2.8	13.1 (7.4 to 18.8)	1.688	4.846	17	<0.001	0.002
N2 (min)	222.2±78.4	231.6±36.8	−9.4 (−51.8 to 33.0)	−0.153	−0.468	17	0.649	1.000
N2 (%TST)	59.5±12.4	61.2±9.4	−1.6 (−8.9 to 5.6)	−0.155	−0.477	17	0.659	1.000
N3 (min)	20.2±34.3	20.9±22	−0.7 (−18.1 to 16.8)	−0.024	−0.079	17	0.946	1.000
N3 (%TST)	5.8±10.9	5.6±6.2	0.2 (−5.3 to 5.8)	0.023	0.090	17	0.937	1.000
REM (min)	58.9±29.5	109.8±42.3	−50.9 (−77.4 to 24.3)	−1.396	−4.045	17	<0.001	0.011
REM (%TST)	16.1±7.7	28.5±10.6	−12.4 (−19.1 to 5.6)	−1.338	−3.868	17	<0.001	0.016
Stage R (min)	154.9±80.1	43.5±21.5	111.4 (73.3 to 149.5)	1.900	6.167	17	<0.001	<0.001

REM, rapid eye movement; TST, total sleep time; WASO, wakefulness after sleep onset.

Comparison of sleep variables

Cumulative displays show sleep architecture, illustrating the percentage of individuals in each sleep stage, N1, N2, N3 and REM (Y-axis) according to the time from sleep onset (X-axis) by EEG-based sleep clusters. In patients with cirrhosis, we observed differences in sleep architecture, including the low percentage of N3 in the middle and worse sleep groups and less distinct REM sleep cycles in the worse sleep group (figure 3A).

Figure 3. (A) Comparison of sleep variables. (B) Box and whisker diagram comparing patients with cirrhosis at each sleep stage with healthy controls. REM, rapid eye movement; WASO, wakefulness after sleep onset.

Patients with cirrhosis also had a significantly longer sleep latency (45.4 min vs 15.5 min, adjusted p=0.007) and wakefulness after sleep onset (WASO) (168.3 min vs 71.1 min, adjusted p<0.001) than did healthy controls, as presented in table 2. However, the total sleep time was similar between the groups (356.6 min and 382.9 min, respectively), as shown in figure 3A,B.

Regarding sleep structure, compared with healthy controls, patients with cirrhosis had more total time and proportion of awakening and N1 (62 min), and less total time and proportion of REM (16.1 min), indicating more shallow sleep throughout the night. Additionally, patients with cirrhosis had a markedly long REM latency (adjusted p<0.001) and decreased total REM time (adjusted p=0.011) and percentage (adjusted p=0.016) compared with the healthy controls (table 2, figure 3B). Effect sizes for sleep latency, WASO, sleep efficiency, N1 time and percentage, REM time and percentage, and REM latency all exceeded 0.8, suggesting large group differences according to Cohen’s criteria (table 2).

Discussion

We compared sleep disturbances in patients with early cirrhosis with those in healthy individuals, with exclusion criteria including previously reported causes of sleep disturbances in patients with cirrhosis, such as severe hepatic encephalopathy (HE) and itching. Patients with cirrhosis showed longer sleep latency, more frequent awakenings during the night and increased amounts of N1 sleep than did healthy individuals, exhibiting decreased REM sleep, longer REM latency and reduced sleep efficiency. Although no significant difference in the amount of N3 sleep was observed between the two groups (table 2), regarding the temporal changes in its occurrence (figure 3B), patients with cirrhosis showed an absence of N3 sleep during the first 1–2 hours after sleep onset compared with healthy adults. Instead, a higher percentage of N1 sleep and wakefulness was observed (figure 3B). According to the non-REM-REM sleep cycle theory, disruption in non-REM sleep by wakefulness resets the cycle before reaching REM sleep, resulting in delayed REM latency and poorer sleep quality. In this study, none of the patients with cirrhosis were identified as having sleep apnoea syndrome based on ESS questionnaire assessment. Possible reasons for the observed sleep disturbances include changes in intestinal bacteria, worsening of liver reserve function, chronic inflammation and the psychological burden caused by disease distress, all of which have been reported to contribute to HE, delayed peak secretion and poor sleep quality. The prevalence of sleep disturbance in patients with cirrhosis ranges between 48% and 55%.^{15 16} A sleep-wake cycle reversal is considered an early sign of HE and is associated with daytime dysfunction.¹⁷ In animal models and human studies, HE is associated with circadian cycle disturbances. Patients with cirrhosis exhibit higher daytime melatonin levels and lower night-time melatonin clearance, as altered melatonin secretion patterns are associated with delayed peak secretion and sleep quality.¹⁸ Changes in intestinal bacteria, worsening with each reserve, chronic inflammation and psychological burden due to disease distress could also cause HE and delayed peak secretion and affect sleep quality.¹⁹ Furthermore, primary biliary cholangitis, itching, sleep disturbances and chronic liver diseases, including hepatitis and cirrhosis, can cause itching as symptoms progress. Intense itching can make sleeping at night difficult. Approximately 70% of patients with primary biliary cholangitis experience sleep disturbances.⁹ To date, the ESS has provided an unequivocal survey tool,²⁰ as studies have consistently used it as a subjective instrument to assess daytime sleepiness in patients with chronic liver disease, with comparable results.^{21 22} Many of these studies were published in Athens, and most were based on the quantification of sleep disturbances using sleep questionnaires. In previous research, sleep progressed from non-REM sleep stages 1, 2 and 3 to REM sleep, with each cycle lasting approximately 90 min and being repeated three to five times a night, depending on the individual.²³ In the first half of sleep, non-REM sleep 3 was more frequent, and REM sleep increased in the second half. These sleep categories were analysed based on brain waves, with amplitude, including fast and slow waves, and alpha, beta and delta waves used to define shallow or deep sleep. With ageing, sleep changes occur, including decreased deep sleep (non-REM sleep 3) and increased percentage of shallow sleep.²⁴ Thus, in a study population of older individuals, adjusting for age, sex, race and environment is necessary when comparing them with patients with cirrhosis. This study examined poor sleep efficiency in patients with cirrhosis compared with healthy controls by analysing the same model in Japanese participants of the same age, sex and body size while using the same equipment. Despite the high prevalence of sleep disorders in patients with cirrhosis, evidence regarding their management remains limited. HE and sleep disorders may have overlapping mechanisms of action. Therefore, in a patient with both cirrhosis and daytime sleepiness with hyperammonaemia, treating hyperammonaemia first is the most reasonable action.^{16 25} Furthermore, melatonin metabolism is impaired in patients with cirrhosis; therefore, melatonin administration could improve their sleep quality and decrease daytime sleepiness.²⁶ Behavioural therapies, including exposure to bright light early in the morning and avoidance of bright light at night, may also aid treatment.²⁷ However, data and results are limited, as this study included a small sample. Therefore, the treatment of sleep disturbances in patients with cirrhosis should be further investigated. Sleep characteristics should be monitored, and adequate treatment should be provided to patients with sleep disturbances.

Limitations

This study has some limitations. First, the small sample size may limit the generalisability of the findings, emphasising the need for larger studies. Although this study provides valuable insights, unlike previous studies that relied primarily on questionnaires, this study used innovative scientific and objective devices to reveal latent sleep disorders in healthy individuals and those in the early stages of liver cirrhosis. Based on these findings, further large-scale investigations are anticipated in the future.

Conclusion

This study sheds light on the notable features of unrecognised sleep disturbances experienced by patients with early stage cirrhosis, compared with healthy controls, and highlights the importance of addressing these disturbances as part of cirrhosis management. By employing advanced portable EEG technology and adhering to standardised sleep architecture scoring criteria, the research provides a comprehensive analysis of sleep patterns, revealing key differences such as increased sleep latency, frequent night-time awakenings, reduced REM sleep and a higher proportion of shallow sleep (N1) among patients with cirrhosis. These findings emphasise the prevalence of sleep misrecognition, where patients with cirrhosis perceive their sleep as adequate despite objectively poor sleep efficiency.

The paper also identified potential contributors to these sleep disruptions, including hyperammonaemia, altered melatonin metabolism and disease-related psychological burdens, all of which may have overlapping mechanisms with HE. Although highlighting the clinical relevance of these findings, this research underscores the need for multifaceted management strategies, including treating hyperammonaemia, administering melatonin and adopting behavioural therapies such as light exposure modifications, to improve sleep quality and daytime functionality in patients with cirrhosis.

Data availability statement

Data are available in a public, open access repository.