Acute Effects of Oral Cannabinoids on Sleep and High‐Density EEG in Insomnia: A Pilot Randomised Controlled Trial

ABSTRACT

Cannabinoids, particularly Δ9‐tetrahydrocannabinol (THC) and cannabidiol (CBD), have gained popularity as alternative sleep aids; however, their effects on sleep architecture and next‐day function remain poorly understood. Here, in a pilot trial, we examined the effects of a single oral dose containing 10 mg THC and 200 mg CBD (THC/CBD) on objective sleep outcomes and next‐day alertness using 256‐channel high‐density EEG in 20 patients with DSM‐5 diagnosed insomnia disorder (16 female; mean (SD) age, 46.1 (8.6) years). We showed that THC/CBD decreased total sleep time (−24.5 min, p = 0.05, d = −0.5) with no change in wake after sleep onset (+10.7 min, p > 0.05) compared to placebo. THC/CBD also significantly decreased time spent in REM sleep (−33.9 min, p < 0.001, d = −1.5) and increased latency to REM sleep (+65.6 min, p = 0.008, d = 0.7). High‐density EEG analysis revealed regional decreases in gamma activity during N2 sleep, and in delta activity during N3 sleep, and a regional increase in beta and alpha activity during REM sleep. While there was no observed change in next‐day objective alertness, a small but significant increase in self‐reported sleepiness was noted with THC/CBD (+0.42 points, p = 0.02, d = 0.22). No changes in subjective sleep quality, cognitive performance, or simulated driving performance were observed. These findings suggest that a single dose of cannabinoids, particularly THC, may acutely influence sleep, primarily by suppressing REM sleep, without noticeable next‐day impairment (≥ 9 h post‐treatment). Australian New Zealand Clinical Trial Registry (ACTRN12619000714189) https://www.anzctr.org.au/.

1. Introduction

Insomnia disorder is a prevalent condition affecting up to 30% of the general population and is characterised by persistent difficulty initiating and/or maintaining sleep for ≥ 3 nights per week for ≥ 3 months, and is associated with significant daytime impairment (Aernout et al. 2021; American Psychiatric Association 2013). While cognitive behaviour therapy for insomnia (CBT‐I) can be an effective treatment, its widespread implementation is limited by barriers to therapist access and delayed perceived benefits (De Crescenzo et al. 2022). As a result, patients frequently seek interim relief through short‐term pharmacological therapies, including sedative drugs and hypnotics (Miller et al. 2017; Begum et al. 2021). However, concerns arise due to significant side effects, including risk of falls, next‐day drowsiness, and impaired driving ability (Vermeeren 2004). This has led to growing interest in alternative therapies, such as medicinal cannabis, particularly its two major phytocannabinoids, ∆⁹‐tetrahydrocannabinol (THC) and cannabidiol (CBD).

Insomnia disorder is one of the most commonly reported conditions for which medicinal cannabis is used in the United States (Schlienz et al. 2021; Bachhuber et al. 2019), Canada (Cahill et al. 2021), and Australia (Suraev et al. 2023; MacPhail et al. 2022). Many individuals report using cannabis as a substitute for prescription medications, such as benzodiazepines (Piper et al. 2017). However, few randomised controlled trials have examined its therapeutic effects in clinical insomnia populations. With the exception of one study (Walsh et al. 2021), most existing literature lacks validated objective sleep measures and involves cohorts with substantial heterogeneity in patient characteristics, as well as variations in dosage, timing, and route of cannabinoid administration (Cousens and DiMascio 1973; Mondino et al. 2021; Gates et al. 2014; AminiLari et al. 2022; Suraev et al. 2020a; Maddison et al. 2022).

Prior research, primarily in individuals with heavy recreational cannabis use, has demonstrated that cannabis—particularly THC—suppresses REM sleep (Feinberg et al. 1975; Freemon 1972; Pivik et al. 1972; Tassinari et al. 1999a; Gillin et al. 1972; Karacan et al. 1976). In a recent trial using a THC‐dominant formulation (oral; 10–20 mg THC per night) over 2 weeks in individuals with insomnia disorder found a non‐significant trend towards reduced REM sleep (−3.5%, p = 0.055, d = 0.4) (Walsh et al. 2021). By contrast, a single 300 mg oral dose of CBD had no significant effect on sleep architecture in healthy volunteers (Linares et al. 2018). Hypnotics, such as benzodiazepines, are known to disrupt sleep architecture, leading to non‐restorative sleep and next‐day impairment (Atkin et al. 2018). However, no study to date has objectively assessed the residual effects of combined THC and CBD on next‐day alertness.

This pilot study aims to address existing knowledge gaps by evaluating the effects of a single oral dose of a combined THC and CBD formulation on objective sleep outcomes and next‐day alertness in adults with insomnia. We employed polysomnography, augmented with 256‐channel high‐density EEG, to comprehensively assess alterations in sleep architecture and regional brain activity following cannabinoid treatment. Although not a primary outcome, subjective sleep effects were also evaluated following the single‐dose treatment.

2. Materials and Methods

2.1. Participants

Participants were recruited through self‐referral, recommendations from sleep physicians and psychologists, and media advertisements. The key inclusion criteria were: (i) age between 25 and 65 years; (ii) a clinical diagnosis of insomnia disorder, defined as: (a) self‐reported difficulty initiating and/or maintaining sleep on more than three nights per week and for longer than 3 months, accompanied by daytime impairments despite adequate sleep opportunity, and (b) Insomnia Severity Index (ISI) (Morin et al. 2011) score ≥ 15; and (iii) good health, confirmed through clinical interview and physical examination by a sleep physician. After the trial commenced, the age range was widened (from 35–60 years) to include a broader sample of both younger and older adults (Suraev et al. 2020b). The key exclusion criteria were: (i) cannabis use within the past 3 months (confirmed via urinary drug screening); (ii) clinically significant sleep disorders other than insomnia (identified through an in‐laboratory overnight diagnostic sleep study); (iii) use of any treatment for insomnia, including CBT‐I, within the past 3 months; (iv) excessive caffeine use contributing to insomnia, or inability to abstain from caffeine for ≥ 24 h before each overnight treatment session, as determined by a sleep physician; (v) current use of medications that affect the central nervous system (CNS) (e.g., hypnotics, antidepressants), modulate the cytochrome (CYP) P450 enzyme system, or are metabolised by CYP enzymes inhibited by CBD. The complete exclusion criteria are detailed in the published protocol (Suraev et al. 2020b).

2.2. Study Design and Procedures

This within‐participant, double‐blind, placebo‐controlled, crossover study was approved by Bellberry Human Research Ethics Committee (2018‐04‐284) and prospectively registered with the Australian New Zealand Clinical Trials Registry (ACTRN12619000714189) on 13 May 2019. The study was conducted at the Woolcock Institute of Medical Research (the Sponsor) and adhered to the Consolidated Standards of Reporting Trials (CONSORT) guideline.

All participants were informed of the nature and risk of experimental procedures by the trial coordinator and sleep physician before providing written informed consent. Initial eligibility was assessed during a clinical screening visit, which included an electrocardiogram, urinary drug test (DrugCheck NxStep Onsite Urine Drug Test), and a pregnancy test where applicable (Human Chorionic Gonadotropin Cassette, Alere). Baseline questionnaires included the Pittsburgh Sleep Quality Inventory (PSQI) (Buysse et al. 1989), Patient Health Questionnaire‐9 (PHQ‐9) (Martin et al. 2006), Hospital Anxiety and Depression Scale (HADS) (Zigmond and Snaith 1983), and Epworth Sleepiness Scale (ESS) (Johns 1991).

Participants completed an overnight in‐laboratory diagnostic sleep study using a standard EEG system to rule out other clinically significant sleep disorders and to familiarise them with the sleep laboratory environment. They then attended a separate familiarisation visit to practice wearing the high‐density EEG sensor cap during a brief daytime nap (~20 min). Participants were instructed to abstain from caffeine and alcohol for at least 24 h before each treatment session, and to avoid illicit drugs (including cannabis) and all CNS‐active medications, including hypnotics, for the duration of the trial. They were also encouraged to maintain regular sleep and wake times consistent with their habitual schedule throughout the study period.

On the day of each treatment session, participants underwent a brief medical screening with a sleep physician, and urine drug and pregnancy tests were repeated as applicable. The two 24‐h overnight treatment sessions were separated by a minimum of one week to allow adequate drug washout. To maintain familiarity with the testing environment across visits, participants slept in the same bedroom assigned to them during the diagnostic sleep study.

Participants selected their preferred bedtime based on a seven‐day sleep diary completed prior to their first treatment session, and this bedtime was maintained for both treatment sessions. Sleep opportunity was standardised to an 8‐h period across all sessions due to practical constraints related to the timing of the morning assessment. After lights out, participants slept undisturbed in a private bedroom within the outpatient sleep laboratory. They were permitted to wake naturally or were gently awakened by a sleep technician, who gradually increased the room's ambient light before gently knocking on the door.

2.3. Randomisation

Participants were randomly allocated in a 1:1 ratio to one of two treatment sequences (‘THC/CBD–placebo’ or ‘placebo–THC/CBD’) using a computer‐generated randomisation schedule created by the unblinded study investigator. The investigational product was packaged in sequentially numbered boxes by an independent drug distributor, following the randomisation schedule provided by the unblinded study investigator. The trial coordinator enrolled participants, and the study doctors assigned participants to their treatment sequence. All participants, study investigators, and outcome assessors were blinded to treatment allocation until after the database lock.

2.4. Investigational Product

The investigational product was a plant‐derived oral formulation containing a 1:20 ratio of THC to CBD (i.e., 5 mg/mL THC and 100 mg/mL CBD), suspended in medium‐chain triglyceride (MCT) oil. It was manufactured at a Good Manufacturing Practice (GMP)‐certified facility (Linnea SA, Lavertezzo, Switzerland). Participants received a single fixed oral dose of 2 mL, containing 10 mg THC and 200 mg CBD, or a matched placebo (2 mL without cannabinoids) 1 h before bedtime.

The 1:20 THC to CBD ratio was selected to mitigate potential adverse effects of THC, such as anxiety and memory impairment (Freeman et al. 2019), and aligned with data from naturalistic settings where higher CBD concentrations were reportedly used to effectively manage insomnia symptoms (Belendiuk et al. 2015). Since the design of the trial, new research has shown that at higher oral CBD to THC ratios (e.g., 30:1), similar to those used in the present study, CBD potentiates THC blood concentrations and associated adverse effects due to CYP‐mediated inhibition of Δ9‐THC metabolism (Zamarripa et al. 2023). The 10 mg THC dose was chosen based on prior studies identifying it as a typical starting dose for cannabis‐inexperienced individuals, as it induces noticeable subjective effects, such as increased drowsiness, without significantly impairing cognitive or psychomotor performance (Bhaskar et al. 2021; MacCallum and Russo 2018; Schlienz et al. 2020).

Participants were instructed to consume a peppermint lozenge (Fisherman's Friend Mint; Lofthouse of Fleetwood, England) before ingesting the oil to mask any potential perceived differences in taste or smell. To assess the success of blinding, participants were asked to guess the treatment they had received the night before (i.e., ‘THC/CBD’, ‘Placebo’, or ‘Not sure’) and to indicate their level of certainty on a 4‐point Likert scale: ‘Not at all’, ‘Somewhat’, ‘Moderately’, or ‘Extremely’.

2.5. Co‐Primary and Secondary Outcomes

The two primary outcomes were total sleep time (TST) and wake after sleep onset (WASO), both measured in minutes using in‐laboratory polysomnography. Key secondary outcomes included: (a) sleep macroarchitecture metrics measured via polysomnography; (b) global and regional EEG power spectral analysis measured through high‐density EEG; (c) objective next‐day alertness measured using the Maintenance of Wakefulness Test (MWT) and the Karolinska Drowsiness Test (KDT); (d) subjective ratings of sleep–wake behaviours; and (e) adverse event profiling. No changes to the trial outcomes were made after the study commenced.

2.6. Data Collection

2.6.1. Overnight Polysomnography With High‐Density EEG

Participants underwent in‐laboratory polysomnography combined with 256‐channel high‐density EEG (Magstim EGI, Eugene, OR, USA). Recordings were scored according to the American Academy of Sleep Medicine (AASM) criteria (American Academy of Sleep Medicine 2007) by an experienced sleep scientist using ProFusion PSG v4 software (Compumedics, Abbotsford, Australia). Sleep staging—including wake, N1, N2, N3 and REM—was performed in 30‐s epochs based on the six high‐density EEG channels (F3, F4, C3, C4, O1, O2), re‐referenced to the mastoids. Overnight high‐density EEG signals were recorded using NetStation Software (Magstim EGI, Eugene, OR, USA) at a sampling rate of 500 Hz. Impedances were maintained below 50 kΩ throughout the NetStation acquisition.

2.6.2. High‐Density EEG Pre‐Processing Analysis

A first‐order high‐pass filter (0.1 Hz) was initially applied in NetStation to mimic common hardware analogue filters and remove low frequency drift. The data were then band‐pass filtered (Kaiser type, 0.3–50 Hz) and a 50 Hz notch filter was used to minimise power line noise in NetStation. Raw EEG signals were analysed in MATLAB (The MathWorks Inc., Natick, MA) using custom‐built functions based on the EEGLAB toolbox. Channel quality and artefact epochs were determined through semi‐automatic identification and visual inspection. Excluded channels were interpolated using spherical splines, and artefact time segments were removed.

The data were average‐referenced to the mean voltage across all channels, with separate datasets generated for each sleep stage. Data quality was visually confirmed prior to spectral power computation. Power spectral density was calculated using Welch's method in 6‐s data segments (Hamming windows, 50% overlap, frequency resolution 0.12 Hz) across six frequency ranges in each sleep stage: low‐delta, 0.5–1 Hz; delta, 1–4.5 Hz; theta, 4.5–8 Hz; alpha, 8–12 Hz; sigma, 12–15 Hz; beta, 15–25 Hz; gamma, 25–40 Hz. Channels overlying the neck and cheeks were excluded in the topographical analyses.

2.6.3. Next Day Objective Assessments of Alertness

Alertness was evaluated using the following assessments:

• Maintenance of Wakefulness Test (MWT): A validated measure of the ability to stay awake in a quiet, darkened environment. Participants were instructed to remain awake for as long as possible during four 40‐min trials conducted at 10:00, 12:00, 14:00, and 16:00 the day after treatment, following standard guidelines (Krahn et al. 2021). The key outcome variable was the mean sleep latency (minutes) across the trials.

• Psychomotor Vigilance Task (PVT): This 10‐min simple reaction‐time task assesses sustained attention and is sensitive to sleep loss (Dorrian et al. 2004). The key outcome variables were the mean reciprocal reaction time (msec) and the number of lapses (response time > 500 msec).

• Karolinska Drowsiness Test (KDT): Administered with high‐density EEG, this test evaluated physiological sleepiness 1 h (PM; before lights out) and 9 h (AM; after waking) post‐treatment. The key outcome variables were resting wake EEG power under eyes‐closed and eyes‐open conditions.

2.6.4. Subjective Next‐Day Alertness and Sleep–Wake Behaviour

The Karolinska Sleepiness Scale (KSS) (Åkerstedt and Gillberg 1990) was used to assess subjective sleepiness at six timepoints: 0.5 h (before lights out), 9 h (upon waking), and at 12, 14, 16, and 18 h post‐treatment. The Leeds Sleep Evaluation Questionnaire (LSEQ) (Parrott and Hindmarch 1980) and the Daytime Insomnia Symptom Response Scale (DISRS) (Carney et al. 2013) were administered approximately 9–10 h post‐treatment.

2.6.5. Plasma Cannabinoids

Blood samples were collected 11 h post‐treatment via venipuncture into EDTA vacutainer tubes (Becton, Dickinson and Company, New Jersey, USA). Samples were centrifuged at 1500 × g for 10 min at 4°C, and the supernatant plasma was aliquoted into 1.8 mL cryotubes and stored at −80°C until analysis. Plasma concentrations of cannabinoids (CBD, THC) and their metabolites (11‐OH‐THC, THC‐COOH; 7‐COOH‐CBD, and 7‐OH‐CBD) were quantified using LC–MS/MS, according to previously published methods (Kevin et al. 2020).

2.6.6. Adverse Events

Adverse events were recorded at the end of each treatment session through open‐ended questioning.

2.6.7. Statistical Analyses

Data analysis began on 17 February 2022, following the a priori statistical analysis plan. As this was a pilot study, no formal sample size calculation was performed. No interim analyses were planned or conducted, and there were no early stopping guidelines. Subgroup or adjusted analyses were not performed. All data were analysed under the intention‐to‐treat principle by blinded investigators using SPSS version 26 (IBM Corp., Armonk, NY). Figures were generated using GraphPad Prism version 9 (GraphPad Inc., San Diego, CA).

Descriptive statistics were calculated for demographic variables and adverse events. Linear mixed‐effects models were used to evaluate differences between treatment and placebo. Fixed factors included Treatment (2 levels: THC/CBD and placebo) and Order (2 levels: ‘THC/CBD‐placebo’ or ‘placebo‐THC/CBD’). Participant was included as a random effect. For the KSS, which was measured at multiple timepoints, Time (6 levels: 0.5, 9, 12, 14, 16, and 18 h post‐treatment) was also included as a fixed factor. The least‐squares means procedure in the mixed‐model analyses was used to handle missing data. Statistical significance was set at p < 0.05.

For high‐density EEG analysis, differences in absolute spectral power between THC/CBD and placebo were assessed using a two‐step process. First, paired t‐tests were applied to each EEG channel to generate topographic maps comparing THC/CBD with placebo. Then, to correct for type I error in multiple testing, we used statistical non‐parametric mapping (SnPM) to identify significant clusters at an α‐level < 0.05 (Jones et al. 2014). We selected an appropriate threshold to identify clusters (i.e., neighbouring channels with a critical t‐value i.e., two‐tailed alpha level of < 0.05). SnPM randomly shuffles the dependent variable between conditions (10,000 times). For each reshuffling, the size of the largest cluster above the threshold was used to generate a maximal cluster size distribution (null distribution). The p‐value for the original cluster is then calculated as the probability of the actual cluster size given the maximal cluster size null distribution.

The high‐density EEG data recorded during the KDT were analysed similarly. A mixed‐effects generalised linear model was applied to each channel, including fixed effects factors: Timepoint (AM or PM), Treatment (THC/CBD or placebo), and their interaction, with Participant as a random intercept. An alpha threshold of 0.05 was used to identify clusters, and SnPM was applied to control for type I error using permutation tests (10,000 times). Clusters were considered significant at a cluster‐alpha threshold of 0.05. EEGLAB was used to project the t‐statistic onto topographic maps. All HD‐EEG statistical analyses were performed using MATLAB.

Cohen's d_z effect sizes were calculated by standardising the mean difference between THC/CBD and placebo against the standard deviation (SD) of change (Lakens 2013). The 95% confidence intervals (95% CI) were calculated using the Hedges and Olkin approximation, adapted for a repeated‐measures design (Goulet‐Pelletier and Cousineau 2018), as demonstrated previously (McCartney et al. 2022).

3. Results

3.1. Participant Characteristics

Between August 2019 and October 2021, 571 individuals were pre‐screened for eligibility, with 38 individuals proceeding to medical screening (Figure 1). Of these, 20 participants (16 female; mean (SD) age, 46.1 (8.6) years) were randomised, and all completed the trial (Table 1). Primary outcome data were available for all 20 participants, who were analysed according to their randomisation sequence. The trial concluded once the target sample size was achieved.

FIGURE 1.

CONSORT flow diagram.

TABLE 1.

Participant demographics and characteristics.

Characteristic	N = 20
Sex (M/F), n	4/16
Age, mean (SD), y	46.1 (8.6)
Weight, mean (SD), kg	70.6 (14.7)
BMI, mean (SD), kg/m 2	25.1 (3.7)
Participants with at least some tertiary education, n	18
Participants with current employment, n	15
Weekly standard drinks, median [IQR]	2.5 [0.0]
Subjective sleep scales, mean (SD)
ISI total score	20.8 (2.5)
PSQI total score	12.6 (3.2)
ESS total score	4.4 (4.2)
Psychiatric scales, mean (SD)
HADS total score
Anxiety	5.3 (3.5)
Depression	3.8 (3.5)
PHQ‐9 total score	7.5 (4.1)
a Baseline PSG‐derived objective sleep metrics, mean (SD)
Total sleep time, min	316.7 (83.9)
Wake after sleep onset, min	92.5 (63.0)
Sleep onset latency, min	19.5 (18.6)
NREM sleep, min	253.2 (57.5)
REM sleep, min	67.4 (31.0)
REM latency, min	121.3 (59.4)
Sleep efficiency (%)	73.0 (16.4)
AHI, events/h	3.9 (4.0)

Abbreviations: AHI Apnea‐Hypopnea Index; BMI Body Mass Index; ESS Epworth Sleepiness Scale; HADS Hospital Anxiety and Depression Scale; ISI Insomnia Severity Index; IQR interquartile range; NREM non‐rapid eye movement sleep; PHQ‐9 Patient Health Questionnaire‐9; PSQI Pittsburgh Sleep Quality Inventory; REM rapid eye movement sleep; SD standard deviation.

^a Diagnostic sleep study obtained at baseline using a standard EEG system.

3.2. Objective Sleep Outcomes

3.2.1. Sleep Macroarchitecture

Compared to placebo, THC/CBD significantly decreased TST by 24.5 min (p = 0.047, d = −0.49 [95% CI −0.9, −0.03]) with no change to WASO (Figure 2 and Table 2). THC/CBD also reduced time spent in REM sleep by 33.9 min (p < 0.001, d = −1.49 [95% CI −2.1, −0.9]) and increased latency to REM sleep (+65.6 min, p = 0.008, d = 0.68 [95% CI 0.2, 1.2]). Post hoc analysis showed that THC/CBD consistently decreased time spent in REM sleep during the first (−6.4 min, p = 0.025, d = −0.56 [95% CI −1.03, −0.09]), second (−12.7 min, p = 0.01, d = −0.62 [95% CI −1.10, −0.14]), and third tertiles of sleep (−14.9 min, p = 0.002, d = −0.77 [95% CI −1.3, −0.3]) (Supplementary Table 1). No treatment‐by‐order effects were observed for any of the objective sleep outcomes or other outcomes analysed (all p's > 0.05).

FIGURE 2.

Acute effects of THC/CBD and placebo on polysomnography‐derived sleep macroarchitecture. Panels show mean (SD) values for (A) total sleep time (TST; min), (B) wake after sleep onset (WASO; min), (C) REM sleep latency (min), and (D) REM sleep (min) during THC/CBD and placebo treatments.

TABLE 2.

Polysomnography‐derived objective sleep outcomes during THC/CBD and placebo treatment (n = 20).

	Placebo	THC/CBD	p value a	Cohen's d [95% CI]
Total sleep time, min	396.3 (48.0)	371.8 (62.7)	0.047	−0.49 [−0.95, −0.03]
Wake after sleep onset, min	61.5 (39.6)	72.2 (46.4)	0.422	0.19 [−0.25, 0.63]
Sleep onset latency, min	20.7 (19.3)	28.8 (25.6)	0.189	0.31 [−0.14, 0.76]
N1 sleep, min	58.8 (24.2)	64.5 (29.6)	0.341	0.22 [−0.23, 0.66]
N2 sleep, min	158.1 (36.9)	166.9 (37.1)	0.420	0.19 [−0.25, 0.63]
N3 sleep, min	91.1 (37.6)	86.1 (41.5)	0.516	−0.15 [−0.59, 0.29]
NREM sleep, min	308.0 (41.4)	317.4 (48.4)	0.377	0.21 [−0.24, 0.65]
REM sleep, min	88.3 (20.9)	54.4 (25.9)	< 0.001	−1.49 [−2.13, −0.85]
REM latency, min	128.2 (58.7)	193.8 (74.8)	0.008	0.68 [0.19, 1.16]
Sleep efficiency (%)	82.8 (9.9)	78.6 (11.0)	0.119	−0.37 [−0.83, 0.08]
AHI, events/h	8.4 (7.8)	8.2 (7.9)	0.900	−0.03 [−0.47, 0.41]

Note: Obtained using a 256‐channel high‐density EEG system. Values are presented as mean (SD).

Abbreviations: AHI Apnea‐Hypopnea Index; CI confidence interval; NREM non‐rapid eye movement sleep; REM rapid eye movement sleep.

^a p‐value based on linear mixed model analysis for THC/CBD versus placebo.

3.2.2. Sleep Microarchitecture

3.2.2.1. Global and Regional EEG Power During Sleep

No significant differences in global EEG spectral power were observed between treatments during NREM and REM sleep (Supplementary Figure 1). However, topographical analysis revealed that THC/CBD reduced both gamma EEG activity during N2 sleep in the central‐posterior region (cluster size n = 20, average t‐value = −2.6 and p = 0.02) and delta activity during N3 sleep in the posterior region (cluster size n = 18, average t‐value = −3.0 and p = 0.01) compared to placebo (Figure 3).

FIGURE 3.

Acute effects of THC/CBD and placebo on global EEG spectral power. Topographic t‐value maps showing channel‐by channel comparisons for normalised power between THC/CBD and placebo treatments for N2, N3, and REM sleep. White dots indicate significant electrode clusters identified by the statistical nonparametric mapping suprathreshold (SnPM) cluster test (p < 0.05). Yellow dots represent electrodes with differences detected by paired t‐tests (uncorrected p < 0.05). Frequency bands include low‐delta, 0.5–1 Hz, delta, 1–4.5 Hz; theta, 4.5–8 Hz; alpha, 8–12 Hz; sigma, 12–15 Hz; beta, 15–25 Hz; gamma, 25–40 Hz.

During REM sleep, THC/CBD increased both alpha (cluster size n = 24, average t‐value = 2.9 and p = 0.02) and beta EEG activity in posterior regions (cluster size n = 7, average t‐value = 2.3 and p = 0.04). A channel‐by‐channel topographical comparison of normalised power between THC/CBD and placebo treatment for NREM and REM sleep is provided in Supplementary Figure 2.

3.2.2.2. Objective Alertness Outcomes

THC/CBD had no significant effect compared to placebo on average latency to sleep (minutes) in the MWT or on any of the outcome variables of the PVT (all p's > 0.05) (Table 3).

TABLE 3.

Objective alertness and subjective sleep and alertness outcomes following THC/CBD and placebo.

	Observations (Actual/Total)	Placebo	THC/CBD	p value	Cohen's d [95% CI]
MWT
Average sleep latency, min	40/40	31.2 (11.9)	33.2 (10.2)	0.331	0.23 [−0.22, 0.67]
PVT
Mean RRT, ms	39/40	3.6 (0.5)	3.5 (0.6)	0.468	0.16 [−0.29, 0.61]
Lapses, median [IQR], n	39/40	2.0 [1.0−3.0]	2.0 [1.0‐4.0]	0.283	0.26 [−0.19, 0.72]
LSEQ, mm (0–100)
Getting to sleep	40/40	50.5 (19.1)	48.0 (12.7)	0.620	−0.11 [−0.55, 0.33]
Quality of sleep	40/40	43.2 (22.4)	41.1 (25.7)	0.777	−0.06 [−0.50, 0.34]
Awakening from sleep	40/40	39.7 (19.1)	34.7 (18.9)	0.400	−0.19 [−0.64, 0.25]
Behaviour following wakefulness	40/40	28.8 (15.9)	29.3 (14.9)	0.924	0.02 [−0.42, 0.46]
DISRS
Total score	40/40	41.5 (11.1)	42.4 (11.5)	0.581	0.13 [−0.31, 0.57]

Note: Values are mean (SD) unless otherwise stated. The 'Observations' column indicates the number of recorded observations used in the mixed‐effects model relative to the total possible observations. Higher scores on the LSEQ indicate better sleep–wake behaviour. Lower scores on the DISRS indicate less daytime sleep‐related rumination.

Abbreviations: DISRS Daytime Insomnia Symptom Response Scale; IQR interquartile range; LSEQ Leeds Sleep Evaluation Questionnaire; MWT Maintenance of Wakefulness Test; PVT Psychomotor Vigilance Test; RRT reciprocal reaction time.

No between‐treatment differences were observed in global resting wake EEG spectral power during the KDT (Supplementary Figure 3). However, topographical analysis revealed that THC/CBD increased delta EEG activity during the eyes‐closed condition in small, discrete clusters over central and right temporal regions and reduced beta EEG activity during the eyes‐open condition in the frontal region (Supplementary Figure 4). No significant Treatment‐by‐Time interactions were detected.

3.2.2.3. Subjective Sleep and Alertness Outcomes

No significant differences were observed between THC/CBD and placebo on any domain of the LSEQ or the overall DISRS score (all p's > 0.05) (Table 3). However, a small overall increase in sleepiness was observed with THC/CBD compared to placebo (+0.42 points, p = 0.02, d = 0.219) (Supplementary Figure 5).

3.2.2.4. Plasma Cannabinoid Concentration

Supplementary Table 2 presents the mean (SD) plasma concentrations (ng/mL) of CBD, THC, and their metabolites in plasma by treatment order. No residual THC was detected the morning after drug administration. However, residual 11‐COOH‐THC and 11‐OH‐THC were detected in plasma in 8 out of 20 and 10 out of 20 participants, respectively.

3.2.2.5. Adverse Events and Blinding Integrity

No serious adverse events were reported, and no participant withdrew from the trial following randomisation. A total of 85 mild, self‐limiting adverse events were recorded: 55 during THC/CBD treatment across 16 participants, and 30 during placebo treatment across 13 participants. The most common adverse event associated with THC/CBD was dry mouth (Supplementary Table 3). Regarding blinding integrity, 14 of 20 participants (70%) correctly guessed that they had received THC/CBD. When given placebo, 12 of 20 participants (60%) correctly guessed the treatment, while two participants (10%) were uncertain (Supplementary Table 4).

4. Discussion

In individuals with insomnia, a single oral dose of THC/CBD was found to reduce total sleep time and REM sleep by a clinically significant amount. Additionally, the time taken to reach REM sleep was delayed by approximately 1 h. This is the first study to demonstrate these effects in a clinical sample following acute use, aligning with previous research showing that cannabis suppresses REM sleep. These effects are primarily attributed to THC (Feinberg et al. 1975; Freemon 1972; Pivik et al. 1972; Tassinari et al. 1999b), as CBD has not been reported to disrupt sleep architecture (Linares et al. 2018). The high‐density EEG findings suggest that while THC/CBD may reduce cortical hyperarousal during lighter sleep stages, it also disrupts deeper and REM sleep stages, which are essential for sleep quality and restorative functions. Interestingly, despite disruption in sleep architecture, there were no observed effects on next‐day alertness. None of the participants in this pilot study were regular cannabis users, making these findings particularly compelling as they demonstrate the acute impact of THC/CBD on sleep architecture in cannabis‐naïve or infrequent cannabis users.

The reduction in REM sleep in the present study confirms and extends prior findings, employing a randomised controlled trial design with a pharmaceutical‐grade cannabinoid product and a clinical population of infrequent cannabis users. The REM‐suppressant effect is comparable to that observed with certain hypnotics, such as benzodiazepines (Borbély et al. 1985; Feinberg et al. 2000; Arbon et al. 2015; Aeschbach et al. 1994; Johnson et al. 1983; de Mendonça et al. 2023). Similarly, a recent trial reported a non‐significant trend towards reduced REM sleep (−3.5%, p = 0.055, d = 0.4) following a 2‐week treatment of combined oral 10 mg THC, 1 mg cannabinol (CBN), and 0.5 mg CBD in 24 individuals with insomnia (Walsh et al. 2021). The smaller effect observed in that study may reflect the development of tolerance to THC's REM‐suppressant properties with repeated use (Colizzi and Bhattacharyya 2018). Notably, abrupt cessation of cannabis use is commonly associated with sleep disturbances, including insomnia, vivid or unusual dreams, and poor sleep quality (Budney et al. 2004). These symptoms, likely related to REM rebound, are key risk factors for relapse due to their intensity and associated distress (Babson et al. 2013). To date, no studies have investigated the potential sleep‐related withdrawal symptoms following the discontinuation of repeated medicinal cannabis use, highlighting a critical gap in our understanding of its long‐term safety of cannabinoids in treating insomnia disorder. Moreover, emerging evidence suggests that THC and CBD can have pharmacokinetic and pharmacodynamic interactions, with a higher oral CBD to THC ratio potentiating THC blood concentrations and associated impairment due to a CYP‐mediated inhibition of THC metabolism (Zamarripa et al. 2023). It is therefore plausible that the 10 mg THC dose in combination with 200 mg CBD in this study may have produced a stronger than anticipated drug effect, which may have interrupted sleep rather than facilitating it in our cohort of cannabis‐naive or infrequent cannabis users.

High‐frequency EEG activity is commonly reported as a physiological correlate of insomnia disorder and serves as a marker of cortical hyperarousal (Zhao et al. 2021; Merica et al. 1998). The observed reduction in gamma EEG activity during N2 sleep suggests that THC/CBD may reduce cortical arousal (Cervena et al. 2004; Riedner et al. 2016). However, the reduction in delta EEG activity during N3 sleep indicates reduced sleep depth, which could diminish the restorative quality of sleep. Interestingly, despite the reduction in delta activity, this did not translate into a decrease in slow wave sleep at the macro level, nor did it impair next‐day alertness. The increase in alpha and beta EEG activity during REM sleep points to heightened cortical arousal in this stage, consistent with changes in sleep macro‐architecture, such as prolonged REM latency and reduced time spent in REM sleep (Zhao et al. 2021). Disruption of REM sleep is recognised as a hallmark of insomnia and has been proposed to result from dysregulation of locus coeruleus (LC) activity (Wassing et al. 2019; Cabrera et al. 2024; Van Someren 2021). Cannabinoid receptor 1 (CB₁R) agonists like THC have been shown to increase LC activity, which elevates noradrenaline (NA) levels (Cohen et al. 2019). This heightened noradrenergic activity partitions sleep cycles into NREM and REM phases, acting as a gatekeeper for REM sleep onset (Osorio‐Forero et al. 2023). Thus, it is plausible that THC‐induced LC activation may contribute to delayed and reduced REM sleep. The clinical significance of the findings warrants further investigation through repeated dosing studies conducted in larger clinical samples.

THC/CBD administration was associated with reduced frontal beta activity during eyes‐open conditions, indicating decreased alertness, and increased centro‐temporal delta activity during eyes‐closed conditions, suggesting heightened drowsiness, as measured by the KDT. A small increase in subjective drowsiness was also observed, with a 0.42‐point rise on the KSS compared to placebo, reflecting a mild degree of daytime sleepiness. However, this increase is not considered clinically significant, as monotonous tasks—such as driving a train through long stretches of homogeneous forest—can increase KSS scores by 1–2 units (Åkerstedt et al. 2014; Ingre et al. 2004). Importantly, no significant impairment was observed on the MWT, a validated objective measure of daytime sleepiness. This indicates that although participants reported a slight subjective increase in drowsiness, it did not translate into impaired performance on objective alertness assessments. This finding is further supported by the absence of significant deficits in objective measures of cognitive and psychomotor function, as well as simulated driving performance, all assessed ≥ 9 h post‐treatment (Suraev et al. 2024).

This study has several limitations that should be considered. First, it focused on a well‐defined clinical insomnia population, which may limit generalisability to the broader insomnia patient population, many of whom may use prescription sleep aids and/or have comorbid sleep disorders. Although necessary to accommodate the morning assessment schedule and maintain consistency, the fixed 8‐h sleep opportunity may have led to an underestimation of total sleep time and REM duration by limiting participants' ability to recover from potential delayed sleep onset or increased night‐time awakenings. As a pilot study, only the effects of a single dose were assessed, limiting conclusions about the impact of repeated cannabinoid use on objective sleep outcomes. This design was intentional, as a longer protocol would have restricted participants from driving under current state government regulations. Additionally, it is important to note that the generalisability of these findings may be limited to cannabis products with similar THC:CBD ratios, as formulations available in legal or recreational markets—particularly those with higher THC and lower CBD concentrations—may have differing pharmacological effects. Finally, cannabinoids are lipophilic and tend to persist in plasma for extended periods (Johansson et al. 1989; McCartney et al. 2023). Small concentrations of residual CBD (and its metabolites) and THC metabolites—but not THC—were detected during the placebo condition in participants who received THC/CBD first, despite a minimum one‐week washout period. The pharmacological significance of low residual plasma levels of 7‐COOH‐CBD remains unclear and warrants further investigation, though evidence suggests that the analogous THC metabolite, 11‐COOH‐THC, does not produce subjective or physiological effects (Ujváry and Grotenhermen 2014).

A key strength of this study is its rigorous randomised, placebo‐controlled, crossover design, which effectively minimises potential confounding factors inherent in observational studies (D'Onofrio et al. 2020). Additionally, a quality‐assured, regulated cannabinoid formulation was administered to adults with physician‐confirmed insomnia. Notably, although most participants correctly identified their treatment order, no improvements in subjective sleep outcomes were observed, and no significant treatment order effects were detected, suggesting that positive expectancy effects did not meaningfully influence the primary outcomes. To minimise negative expectancy effects, participants with a history of adverse reactions to cannabis were excluded, and subjective drug effects and mood were monitored throughout the study, with no indication of elevated anxiety or mood disturbance in either condition (Suraev et al. 2024). The use of diagnostic sleep studies to rule out other sleep disorders, along with ample habituation to the testing environment, further strengthens the robustness of the study design.

5. Conclusion

This study is the first to use high‐density EEG to explore the acute effects of oral THC/CBD on objective sleep outcomes in individuals with insomnia. A single oral dose significantly reduced total sleep time and REM sleep, without impairing next‐day alertness. High‐density EEG analysis revealed frequency‐ and region‐specific effects, suggesting that THC/CBD reduces cortical hyperarousal during lighter sleep stages while disrupting deeper and REM sleep stages. Further research with repeated dosing is needed to clarify the clinical significance of these findings and to evaluate the broader impact of cannabinoids on sleep macro‐ and micro‐architecture, as well as insomnia symptoms.

Author Contributions

Anastasia Suraev: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, project administration, data curation, resources, funding acquisition. Iain S. McGregor: conceptualization, investigation, funding acquisition, writing – review and editing, methodology, project administration, supervision, resources, data curation. Danielle McCartney: conceptualization, investigation, writing – review and editing, methodology, project administration, supervision, data curation, resources, formal analysis. Nathaniel S. Marshall: conceptualization, writing – review and editing, methodology, formal analysis, funding acquisition, investigation, project administration, supervision, resources, data curation, validation. Chien‐Hui Kao: conceptualization, writing – review and editing, data curation, supervision, resources, software, formal analysis, methodology, visualization. Rick Wassing: conceptualization, writing – review and editing, data curation, supervision, resources, software, formal analysis, methodology, visualization. Angela L. D'Rozario: conceptualization, writing – review and editing, funding acquisition, methodology, project administration, supervision, resources. Keith K. H. Wong: conceptualization, writing – review and editing, supervision, resources, project administration, investigation. Brendon J. Yee: conceptualization, writing – review and editing, investigation, supervision, resources, project administration. Sheila Sivam: conceptualization, writing – review and editing, project administration, supervision, resources, investigation. Richard C. Kevin: conceptualization, writing – review and editing, methodology, formal analysis, resources. Ryan Vandrey: conceptualization, writing – review and editing, methodology, supervision. Christopher Irwin: conceptualization, writing – review and editing, investigation, methodology, project administration, data curation, supervision, resources, software. Christopher J. Gordon: conceptualization, writing – review and editing, funding acquisition, methodology. Delwyn Bartlett: conceptualization, writing – review and editing, methodology, funding acquisition. Jonathon C. Arnold: conceptualization, writing – review and editing, methodology, funding acquisition. Ronald R. Grunstein: conceptualization, investigation, funding acquisition, resources, supervision, data curation, project administration, writing – review and editing, methodology, formal analysis, validation. Camilla M. Hoyos: conceptualization, investigation, funding acquisition, resources, supervision, data curation, project administration, writing – review and editing, methodology, formal analysis, validation.

Conflicts of Interest

R. R. G. has received a speaker fee from Eisai. A. S. has received consulting fees from the Medical Cannabis Industry Australia (MCIA) for a commissioned review and Haleon. I. S. M. has served as an expert witness in various medicolegal cases involving cannabis and has received consulting fees from Medical Cannabis Industry Australia (MCIA) and Janssen. ISM has patents WO2019227167 and WO2019071302 issued, which relate to cannabinoid therapeutics. R. V. has received compensation as a consultant or advisory board member for Jazz Pharmaceuticals, Mira1A Pharmaceuticals, Charlotte's Web, WebMD, Syqe Medical Ltd., and Schedule 1 Therapeutics. J. C. A. has served as an expert witness in various medicolegal cases involving cannabis and has received consulting fees from Medical Cannabis Industry Australia (MCIA) and Haleon. J. C. A. is an inventor on patents WO2019227167 and WO2019071302 issued, which relate to cannabinoid therapeutics. All other authors have no competing financial or non‐financial interests to declare. The investigational product was purchased from BOD Australia, who were not involved in the conception or design of this study, data analysis (with no access to the data) or the decision to publish. All other commercially available equipment was purchased.

Supporting information

Data S1 Supporting Information.

Suraev, A. , McGregor I. S., McCartney D., et al. 2026. “Acute Effects of Oral Cannabinoids on Sleep and High‐Density EEG in Insomnia: A Pilot Randomised Controlled Trial.” Journal of Sleep Research 35, no. 1: e70124. 10.1111/jsr.70124.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.