Cannabinol for Acute Treatment of Insomnia Disorder in a Randomized Placebo‐Controlled Crossover Trial

ABSTRACT

Insomnia disorder is harmful and requires novel treatments. Cannabinol, an oxidative by‐product of Δ9‐tetrahydrocannabinol, is claimed to be a hypnotic, but its effects on objective sleep and insomnia remain unknown. This randomized, double‐blind, placebo‐controlled, three‐arm, single‐night crossover trial evaluated the acute efficacy and safety of cannabinol for insomnia disorder. Twenty adults (aged 25–65) with physician‐diagnosed insomnia disorder (meeting DSM‐5 and ICSD‐3 criteria; Insomnia Severity Index ≥ 15) were enrolled at the Woolcock Institute of Medical Research (Sydney, Australia) between August 2022 and September 2023. Participants received a single 2 mL oral dose of 30 mg (1.5%) or 300 mg (15%) cannabinol, or matched placebo (2‐week washout). All participants (17 female and 3 males; mean ± SD age 42 ± 13 years) completed the protocol and were statistically analysed. The primary outcome was wake after sleep onset (WASO) minutes, measured by overnight polysomnography. Cannabinol did not significantly change WASO (300 mg: −6.3 min [95% CI: −18.2, +5.5], p = 0.29, dz = −0.22; 30 mg: −4.0 min [−15.9, +7.9], p = 0.50, dz = 0.11). However, 300 mg cannabinol increased non‐rapid eye movement‐2 sleep (p = 0.03, dz = 0.54), subjective sleep quality (p = 0.005, dz = 0.56); and reduced sleep onset latency (p = 0.004, dz = −0.74) and electroencephalographic arousal indices (p = 0.02, dz = −0.65). There were 247 mild‐to‐moderate adverse events across arms. Larger, longer trials are warranted.

ClinicalTrials.gov Identifier: NCT05344170.

1. Introduction

Insomnia disorder is characterized by subjective sleep disturbance that causes daytime impairment (≥ 3 days/week for ≥ 3‐months) (American Psychiatric Association ²⁰¹³; American Academy of Sleep Medicine ²⁰¹⁴). It is highly prevalent (10% of adults (Morin and Jarrin 2022)), increases the risk of physical and other psychiatric illnesses (notably, cardiovascular disease (Javaheri and Redline 2017)) and depression (Riemann et al. 2020), and imposes a high economic burden (USD$150 billion per annum in the USA) (Reynolds and Ebben 2017). Cognitive‐behavioural therapy for insomnia (CBTi) has demonstrated efficacy (Trauer et al. 2015; van Straten et al. ²⁰¹⁸) but faces accessibility and scalability issues (Koffel et al. 2018) and has significant attrition when delivered via digital means (Zachariae et al. 2016). Weak evidence supports available pharmacotherapies (Sateia et al. 2017), which have limited tolerability and safety risks (e.g., falls and abuse potential) (De Crescenzo et al. 2022). Novel and safe pharmacotherapies are required.

Cannabis sativa contains over 140 known terpenophenolic phytocannabinoids (‘cannabinoids’). The USA cannabis market grew from USD$30 to USD$112 billion between 2022 and 2024 (Long 2024) and insomnia is a leading indication for use and prescription of these products (MacPhail et al. ²⁰²²; Suraev et al. ²⁰²³). Yet the evidence base remains sparse, with very few randomized controlled trials (RCTs) in adults with insomnia disorder (Suraev et al. 2020; Lavender et al. ^{2022, 2024}). Rigorous clinical evaluation of cannabinoids is, therefore, needed. Mostinsomnia studies to‐date have examined Δ9‐tetrahydrocannabinol (THC) and cannabidiol (CBD) (Suraev et al. 2020; Lavender et al. ²⁰²²). However, cannabinol (CBN) is increasingly marketed as a “sleep aid” despite no objective scientific evidence (Corroon 2021; Masataka et al. ²⁰²⁴).

CBN is formed via oxidative degradation of THC, a process accelerated by exposure to oxygen, light, and heat (Banister et al. 2019; Huestis ²⁰⁰⁵). It is a weak partial agonist at the cannabinoid 1 receptor (CB₁R), with lower binding affinity and psychoactive potency than THC (McPartland et al. ²⁰⁰⁷). CB₁R activation is one proposed mechanism through which cannabinoids may influence sleep regulation (Murillo‐Rodriguez ²⁰⁰⁸; Kesner and Lovinger ²⁰²⁰). Human pharmacokinetic data for orally administered CBN isolate are limited (McPartland et al. ²⁰⁰⁷). Cannabinoids generally undergo extensive first‐pass metabolism and distribute widely into fatty tissues because of their high lipophilicity (Huestis 2007). Available evidence indicates CBN is primarily metabolized by cytochrome P450 (CYP450) enzymes to 11‐hydroxy‐CBN (11‐OH‐CBN) (Wall et al. 1976), which demonstrates greater CB₁R activity (Pertwee and Cascio 2014; Rhee et al. ¹⁹⁹⁷).

Early human work (1970–80s) demonstrated promising safety and non‐intoxicating effects with orally administered CBN in doses up to 1200 mg (Karniol et al. 1975; Bird et al. ¹⁹⁸⁰; Perez‐Reyes et al. ¹⁹⁷³; Gong Jr. et al. ¹⁹⁸⁴; Hollister ¹⁹⁷³), but modern data are scarce. A recent study in Long‐Evans rats with implanted biotelemetry probes found 4‐h delayed increases in REM and non‐REM (NREM) sleep at the expense of wakefulness following treatment with CBN, or its active metabolite 11‐hydroxy‐CBN (11‐OH‐CBN); the increase in NREM sleep was comparable in magnitude to the more commonly used sedative‐hypnotic zolpidem (Arnold et al. 2024). This suggests CBN could improve wake after sleep onset (WASO). Since 2023, three studies have tested oral CBN (15–100 mg/day) on self‐reported sleep quality in healthy individuals (Bonn‐Miller et al. ²⁰²⁴; Saleska et al. ²⁰²⁴; Kolobaric et al. ²⁰²³). The only study in adults with insomnia disorder showed significant improvements to insomnia severity with an oral liquid formulation of THC, CBN, and CBD (THC/CBN/CBD 20/2/1 mg nightly for two weeks) (Walsh et al. 2021); though, CBN's isolated effects on objective sleep and/in adults with insomnia disorder remain unclear.

Access to CBN is tightly regulated in many regions. For example, it is classified as a Schedule 9 (‘Prohibited Substance’) in Australia where it can only be legally administered in a clinical trial with Government approval. Given these restrictions, we conducted a single‐night 3‐arm in‐laboratory study utilizing gold‐standard measures to initially examine efficacy and safety. This aimed to characterize the acute effects of oral CBN (30 mg and 300 mg) on objective and subjective sleep and next‐day neurobehavioral function in a sample of adults with sleep physician‐diagnosed insomnia disorder using a crossover design. We hypothesized that both CBN doses would improve polysomnographic indices of sleep disturbance (primary outcome: WASO) compared to placebo.

2. Method

2.1. Trial Design

This was a randomized, double‐blind, placebo‐controlled, single‐dose, three‐arm, crossover, proof‐of‐concept clinical trial, where participants received two CBN doses and placebo in random order (≥ 2‐weeks washout). The 30 mg dose corresponds to typical community use of over‐the‐counter CBN products (Corroon 2021; Full Spectrum CBN Oil (60mg/mL) ²⁰²²) while the 300 mg dose equates to the preclinical study of CBN (Arnold et al. 2024) (10 mg/kg in rats adjusted for species using interspecies scaling (FDA ²⁰⁰⁵) and route of administration [i.e., low oral bioavailability]). The indication and treatment both lend themselves to a crossover design, which is more statistically efficient and precise, requiring fewer participants. The washout period is a trade‐off between the known persistence of cannabinoids in plasma (McCartney et al. ²⁰²³; Johansson et al. ¹⁹⁸⁷), the low probability of a two‐week carryover effect with a single‐dose hypnotic, and minimizing the time that participants remained untreated. In‐laboratory polysomnography was necessary because CBN's effects on sleep architecture were unknown and subjective‐objective sleep misalignment is well described (Manconi et al. 2010); with subjective sleep susceptible to placebo effects (Yeung et al. 2018). Ethical approval was granted by Bellberry Limited (2021‐08‐907; 04AUG22). The study was conducted and sponsored by the Woolcock Institute of Medical Research (Sydney, Australia) in accordance with ICH Good Clinical Practice guidelines, the Declaration of Helsinki (2013) and Federal and local regulations. New South Wales Health granted Authority for the Principal Medical Investigator to prescribe a Prohibited Drug (H22/48537‐2; August 02, 2022). We prospectively registered the trial on ClinicalTrials.gov (Identifier: NCT05344170; April 25, 2022) and published the study methodology elsewhere (Lavender et al. 2023). We include the Participant Information Statement, Statistical Analysis Plan, and protocol as Supporting Information. No significant protocol changes occurred after registration of the trial.

2.2. Participants

The recruitment and screening procedures are published in detail elsewhere (Lavender et al. 2023). We recruited adults aged 25–65 years with sleep physician‐diagnosed insomnia disorder, as per Diagnostic and Statistical Manual of Mental Disorders Version 5 (DSM‐V) and International Classification of Sleep Disorders Version 3 (ICSD‐3) criteria, and Insomnia Severity Index (ISI) score ≥ 15. Participants were excluded for any (1) medical condition/medication impacting sleep; (2) known cannabinoid sensitivity; (3) cannabis/cannabinoid use in 3 months prior (confirmed via urinary drug screen); (4) sleep apnoea (defined as Apnoea Hypopnea Index [AHI] > 15 and Oxygen Desaturation Index [ODI] > 10 on overnight in‐laboratory polysomnography); (5) sleep‐related movement disorder; (6) delayed sleep phase syndrome; (7) medical condition producing abnormal electroencephalography; (8) clinically relevant cardiovascular abnormalities (via electrocardiogram); (9) shift‐work or trans meridian travel over the last month; (10) psychiatric disorder in the past 12 months, except clinically managed mild depression and/or anxiety; (11) historical suicide attempt or current ideation; (12) pregnancy, lactation, or lack of contraception; (13) drug or alcohol dependence prior 2 years; (14) central nervous system‐active drug use prior 3 months (confirmed via urinary drug screen); (15) use of medication impacting metabolism or excretion of cannabinoids (i.e., CYP450 enzyme inducers/inhibiters); (16) excessive caffeine use; (17) daily nicotine use; (18) frequent need to get out of bed (e.g., nocturia); (19) psychological or behavioural treatment for insomnia within 3 months (including CBTi); and (20) mandatory drug alcohol screening (e.g., occupational). Pre‐screening occurred before an on‐site medical screening visit where participants were briefed on the protocol and risks before providing written informed consent. Participants then underwent thorough screening medical assessment with a study physician including electrocardiography, urinary drug and alcohol screening, pregnancy testing, actigraphy (1‐week; GENEActiv; Activinsights Ltd., UK), and in‐laboratory overnight diagnostic polysomnography.

2.3. Randomization and Allocation Concealment

Participants were assigned to one of six possible treatment sequences (1:1:1:1:1:1 ratio) using a pre‐established randomization timetable algorithmically generated by the study epidemiologist (N.S.M.) who never met a participant and played no role in participant selection. The sequences formed a balanced Latin square. The study epidemiologist and drug manufacturer were the only people with access to the randomization schedule. The principal investigators could request their unblinding if a serious adverse event occurred, but this never happened. The principal investigators, study physicians, and trial coordinator confirmed participant consent and eligibility, then randomized them, which was irrevocably recorded in our trial database. The randomization schedule remained undisclosed until after database lock and completion of primary statistical analysis in accordance with the Statistical Analysis Plan.

2.4. Intervention and Blinding

All study personnel, outcome assessors, and participants were blinded to the allocation. The investigational product was an oral liquid formulation of CBN suspended in medium chain triglyceride oil, containing paprika oleoresin colourant, to maintain blinding and prevent oxidation (‘ECS310’). The placebo contained the excipient only. Participants received a single 2.0 mL dose of the following in random order: 1.5 mg/mL CBN (30 mg CBN), (2) 15 mg/mL CBN (300 mg CBN), and (3) matched placebo. The maximum detectable THC content was ≤ 150 μg. Medropharm (Zürich, Switzerland) manufactured the investigational products, which were sourced by BOD Australia (Sydney, Australia); neither of which had input into the study. The investigational products were identical in visual appearance, smell, and taste and the packaging contained no treatment‐identifying information (i.e., labelled with randomization code and visit number only).

2.5. Experimental Procedures

We used actigraphy and daily sleep diaries (Akerstedt et al. 1994) to ensure participants maintained a consistent sleep schedule (i.e., lights out) in the 7‐nights prior to each treatment session. The study physician conducted a medical assessment, urinary drug and alcohol screen, and pregnancy test at the outset of each treatment session to confirm eligibility before prescribing and dispensing the blinded investigational product, which was stored in a locked safe until drug administration. Participants self‐administered the study interventions under observation two hours prior to habitual lights out, which was followed by a fixed 8‐h sleep opportunity. Habitual lights out time was extrapolated from the pre‐treatment actigraphy and sleep diary data (averaged across the 7 days) which was kept consistent across all treatment sessions. Standardized pre‐packaged meals were provided two hours prior to drug administration and one hour after wake. Each participant received the same pre‐dosing meal across all three treatment sessions. Vegetarian and non‐vegetarian meal options were matched for total energy (~500 kcal) and macronutrient composition (approximately 45% carbohydrate, 30% fat, 25% protein) to minimize variability in cannabinoid absorption related to dietary fat. Blinding effectiveness was assessed prior to discharge using a standardized questionnaire to assess participants' guess of the allocated treatment the night prior.

2.6. Outcomes

The primary outcome was WASO minutes from the first epoch after lights out until the last epoch scored as any stage of sleep in the polysomnogram. Secondary outcomes were traditional sleep staging (i.e., the proportion of the sleep opportunity scored as wake, N1, N2, N3, and REM sleep); sleep onset latency (SOL) minutes; and absolute spectral electroencephalographic (EEG) power across five frequency bands (beta: 15–32 Hz; sigma: 12–15 Hz; alpha: 8–12 Hz; theta: 4.5–8 Hz; and delta: 0.5–4.5 Hz) during NREM sleep. Tertiary outcomes included sleep spindle density (11–16 Hz) and slow oscillation/wave power (0.5–2 Hz), arousal indices (count/h), absolute EEG power during REM sleep, and subjective sleep quality. Safety and exploratory outcomes included neurobehavioral function and plasma cannabinoid concentrations, respectively. The description of the registered outcomes in the trial registry was edited by blinded study team members during the trial to remove grammatical ambiguity without any change to the order of the primary, secondary, and tertiary outcomes (ClinicalTrials.gov Identifier: NCT05344170).

2.7. Data Collection

2.7.1. Polysomnography

Participants underwent in‐laboratory polysomnography during the 8‐h sleep opportunity (GraelV2 and ProfusionPSG4, Compumedics, Victoria, Australia) at electroencephalographic (EEG) derivations (F3‐M2, Fz‐M2, F4‐M1, C3‐M2, Cz‐M2, C4‐M1, Pz‐M2, O1‐M2, Oz‐M2, O2‐M1), left and right electrooculogram, chin electromyogram, and lead II ECG. All sleep studies were scored by a single, experienced, blinded polysomnographic technician in accordance with American Academy of Sleep Medicine (AASM) 2020 Sleep Scoring criteria (Version 2.6) (American Academy of Sleep Medicine ²⁰⁰⁷).

2.7.2. Quantitative EEG (qEEG) Microarchitecture

Polysomnography recordings were exported as European Data Format (EDF) and analysed using an in‐house‐built program (D’Rozario et al. ²⁰¹⁴). Absolute EEG spectral power (delta, theta, alpha, sigma, beta), spindle density, and slow oscillation power were analysed per stage of sleep (Lavender et al. 2023 or see Supporting Information).

2.7.3. Subjective Sleep Quality

The Computerized Leeds Sleep Evaluation Questionnaire (Shahid et al. 2011) and Richard Campbell Sleep Questionnaire (Richards et al. 2000) were administered upon wake.

2.7.4. Neurobehavioral Function

Safety outcome measures were published in detail elsewhere (Lavender et al. 2023 or see Supporting Information). Next‐day driving performance was assessed (08:00 h) using a fixed‐based driving simulator equipped with standard vehicle controls (Lavender et al. 2023; McCartney et al. ²⁰²⁰). The pre and post sleep Karolinska Drowsiness Test (KDT) (Putilov and Donskaya 2013) was analysed using an in‐house‐built program (D’Rozario et al. ²⁰¹⁴).

2.7.5. Plasma Cannabinoid Concentrations

Supernatant plasma was collected at pre‐dosing and next‐day timepoints (Lavender et al. 2023 or see Supporting Information [Method]). Samples underwent liquid chromatography–mass spectrometry (LC‐MS/MS) analysis (Kevin et al. 2017, 2021) for cannabinoids (CBN, THC, CBD) and their primary phase I metabolites (e.g., 11‐OH‐CBN, 11‐COOH‐CBN).

2.8. Adverse Events

Adverse events (AEs) were collected spontaneously and via a standardized questionnaire administered pre‐discharge, and ~7‐h and 7‐days post‐discharge. The Principal Medical Investigator (BJY) assessed the severity and causality of each AE. Standardized questionnaire items were reported verbatim. Non‐standardized AEs were coded by a study sleep physician by Preferred Term (PT) within System Organ Class (SOC) according to the Medical Dictionary for Regulatory Activities (MedDRA) (Brown et al. 1999). AEs are reported per participant report. AEs that resolved and then reoccurred within the same collection period are presented as separate items. No formal hypothesis testing was conducted on AE data.

2.9. Statistical Analysis

We arbitrarily selected a sample size of 20 participants for this proof‐of‐concept study given funding constraints. No formal power calculations were applied. All analyses were conducted according to the intention‐to‐treat principle by investigators who remained blinded to treatment allocation. Data analysis was performed using SPSS Statistics version 26 (IBM Corp., Armonk, NY), and figures were created with GraphPad Prism version 9 (GraphPad Software, San Diego, CA). Study outcomes were analysed using linear mixed models, with comparisons made between each CBN dose and placebo. Participants were coded as random effects to account for repeated measurements, with treatment (30 mg CBN, 300 mg CBN, or placebo) and visit (treatment session 1, 2, or 3) coded as fixed effects. For multiple timepoint outcomes, the timepoint (pre‐dose, time 1, or 2) was also included as a fixed effect in the model. The critical p‐value for hypothesis testing was set at 0.05 (two‐tailed). Differences in least squared means within the treatment main effect were only investigated for the hypotheses that the two active doses were superior to placebo. No comparisons were made between active doses. The study did not include an early stopping guideline, nor were interim, subgroup, or adjusted analyses planned or conducted. A detailed Statistical Analysis Plan (see Supporting Information) was finalized by the trial coordinator (I.L.) and Study Epidemiologist (N.S.M) and signed by the Principal Investigators (October 19, 2023) prior to database lock (November 06, 23) (Gamble et al. 2017). Our analysis of the arousal index stratified by sleep type and Leeds Sleep Evaluation Questionnaire subscales were post hoc decisions driven by observation that the overall variable was affected by CBN. Post hoc analyses of plasma CBN/metabolite levels and safety outcomes were conducted after visually observing possible correlations between extreme values. These were not listed in the Statistical Analysis Plan. Effect sizes were calculated using the same repeated‐measures Cohen's dz approach described in Suraev et al. (2026), standardizing the mean difference between placebo and each active dose (30 mg CBN vs. placebo; 300 mg CBN vs. placebo) and the standard deviation of the corresponding change scores.

3. Results

Recruitment began on August 24, 2022, and concluded upon achieving the pre‐specified sample size on August 29, 2023. The first participant was randomized on October 13, 2022, and the final participant visit occurred on August 29, 2023. A total of 20 participants were randomized, received all three treatments, and analysed in the groups they were randomized to (Intention to Treat analysis; Figure 1). Baseline characteristics of the randomized participants are reported in Table 1. All 20 participants were included in primary outcome analysis. Model estimates and data completeness are reported in Table S1.

FIGURE 1.

Consort flowchart. 30 = 30 mg CBN; 300 = 300 mg CBN; 0 = placebo. Participants often met multiple exclusion criteria.

TABLE 1.

Baseline characteristics of study participants (N = 20).

Variable	Descriptive statistics	Range
Sex (F/M), No.	17/3
Age (years, mean ± SD)	41.5 ± 13.0	26–65
BMI (kg/m2, mean ± SD)	23.8 ± 4.1	17.5–32.0
Sleep scales (mean ± SD)
ISI (range 0–28)	20.4 ± 3.2	16–26
ESS (range 0–24)	6.5 ± 6.0	0–17
Psychiatric scales (mean ± SD)
HADS–Anxiety (range 0–21)	6.1 ± 4.1	0–13
HADS–Depression (range 0–21)	2.8 ± 2.4	0–8
PHQ‐9–Suicidal ideation (range 0–3)	0.1 ± 0.2	0–1
Education (≥ university), No. (%)	15 (75)
Currently employed, No. (%)	17 (85)
Country of birth, No. (%)
Australia	10 (50)
England	4 (20)
Hong Kong	2 (10)
Other	4 (20)
Lifetime cannabis uses, No. (%)
0 times	2 (10)
1–10 times	12 (60)
10–100 times	5 (25)
> 100 times	1 (5)
Standard drinks per week (mean ± SD)	2.8 ± 3.5	0–12

Abbreviations: BMI, body mass index; F, female; HADS, hospital anxiety and depression scale; ISI, insomnia severity index; M, male; PHQ‐9, patient health questionnaire.

Relative to placebo, the primary outcome of WASO did not change with 300 mg CBN (−6.3 min [95% CI: −18.2, 5.5], p = 0.29, dz = −0.22) or 30 mg CBN (−4.0 min [−15.9, 7.9], p = 0.50, dz = 0.11; Figure 2). Regarding secondary outcomes, 300 mg CBN reduced sleep onset latency (−7.0 min [−11.6, −2.3], p = 0.004, dz = −0.74), while 30 mg CBN showed no change (Figure 3). Neither dose altered the proportion of wake, N1, N3, or REM sleep. CBN (300 mg) increased N2 sleep proportion (+3.8% [+0.5, 7.1], p = 0.025, dz = 0.54), whereas 30 mg showed no change (Figure 4). CBN (300 mg) reduced absolute NREM delta power (−140.2 μV² [−246.7, −33.7], p = 0.011, dz = −0.51), while 30 mg CBN showed no change. Absolute NREM theta, alpha, sigma, and beta power were unchanged by CBN.

FIGURE 2.

Polysomnographic wake after sleep onset minutes in adults with insomnia disorder (N=20) treated with cannabinol (CBN; 30 and 300 mg) and placebo across three overnight treatment arms. Wake after sleep onset is assessed as minutes per 30‐s epoch scored as wake in the polysomnogram between the first and last epochs scored as any stage of sleep. Bars represent effect estimates for placebo, 30 mg cannabinol (CBN), and 300 mg CBN. Error bars represent 95% confidence intervals. Effect estimates were derived from mixed models of repeated measures with the participant included as a random effect and treatment night (1, 2, 3) and intervention (placebo, 30 mg, 300 mg) included as fixed effects. Comparisons against placebo only. Raw individual participant data is presented in multicoloured scatter plots. *p < 0.05, †p < 0.001.

FIGURE 3.

Polysomnographic sleep onset latency minutes in adults with insomnia disorder (N=20) treated with cannabinol (CBN; 30 and 300 mg) and placebo across three overnight treatment arms. Sleep onset latency is assessed in minutes between lights out and the first epoch scored as any stage of sleep in the polysomnogram. Bars represent effect estimates for placebo, 30 mg cannabinol (CBN), and 300 mg CBN. Error bars represent 95% confidence intervals. Effect estimates were derived from mixed models of repeated measures with the participant included as a random effect and treatment night (1, 2, 3) and intervention (placebo, 30 mg, 300 mg) included as fixed effects. Comparisons against placebo only. Raw individual participant data is presented in multicoloured scatter plots. *p < 0.05, †p < 0.001.

FIGURE 4.

Polysomnography and quantitative electroencephalographic assessed sleep architecture in adults with insomnia disorder (N = 20) treated with cannabinol (CBN; 30 and 300 mg) and placebo across three overnight treatment arms. Panels A–D present effect estimates for placebo (black circles), 30 mg CBN (blue squares), and 300 mg CBN (purple triangles). (A) proportion of wake, N1, N2, N3, and rapid eye movement (REM) sleep across the 8‐h sleep opportunity; (B) Arousal index (cortical electroencephalographic [EEG] arousals detected during each hour) according to REM and non‐REM (NREM) sleep stages; Absolute EEG spectral power during (C) rapid eye movement (REM) and (D) non‐REM sleep. Error bars represent 95% confidence intervals. Effect estimates were derived from mixed models of repeated measures with the participant included as a random effect and treatment night (1, 2, 3) and intervention (placebo, 30 mg, 300 mg) included as fixed effects. Comparisons against placebo only. *p < 0.05, †p < 0.001.

Regarding tertiary outcomes, the arousal index was reduced by 300 mg CBN (−1.5 events/h [−2.8, −0.2], p = 0.021, dz = −0.65), while 30 mg CBN showed no change. Spindle density and slow oscillation power were unchanged by CBN. For REM spectral power, both 30 mg and 300 mg CBN reduced REM theta (−3.0 μV² [−5.5, −0.5], p = 0.021, dz = −0.48; and −2.6 μV² [−5.1, −0.1], p = 0.041, dz = −0.38, respectively) and REM beta power (−1.3 μV² [−2.2, −0.4], p = 0.005, dz = −0.61; and −1.8 μV² [−2.6, −0.9], p < 0.001, dz = −0.77). REM delta and REM alpha power did not significantly change with either dose. CBN (30 mg) reduced REM sigma power (−0.4 μV² [−0.7, −0.1], p = 0.018, dz = −0.60), while 300 mg showed no change. CBN (300 mg) increased subjective sleep quality on both the LSEQ (+11.6 points [+3.7, +19.4], p = 0.005, dz = 0.56) and the RCSQ (+16.6 points [+4.7, +28.4], p = 0.007, dz = 0.66), while 30 mg CBN showed no change (Figure 5).

FIGURE 5.

Subjective sleep quality in adults with insomnia disorder (N = 20) treated with cannabinol (CBN; 30 and 300 mg) and placebo across three overnight treatment arms. Leeds Sleep Evaluation Questionnaire (LSEQ) Score Subscales of getting to sleep (GTS), quality of sleep (QOS), awakenings following sleep (AFS), and behaviour following wake (BFW) where larger values represent higher sleep quality: where larger values represent better sleep quality. Effect estimates for placebo (black circles), 30 mg (blue squares), and 300 mg (purple triangles) cannabinol (CBN). Error bars represent 95% confidence intervals. Effect estimates were derived from mixed models of repeated measures with the participant included as a random effect and treatment night (1, 2, 3), intervention (placebo, 30 mg, 300 mg), and timepoint (1, 2) included as fixed effects. Comparisons against placebo only. *p < 0.05, †p < 0.001.

Of the 54 safety outcomes assessed per treatment arm, 3 and 7 differed from placebo with 30 mg and 300 mg CBN, respectively, see Table S2. CBN did not change lateral vehicular control (i.e., standard deviation of lateral position) on the simulated drive (Figure 6); however, some longitudinal vehicular control measures increased with 300 mg CBN (i.e., distance headway, speed coherence, and speed variability), see Supporting Information. Overnight procedural memory consolidation and postural sway (anterior–posterior; next‐day) decreased with 30 mg CBN; and Psychomotor Vigilance Test reaction time (time 2) increased with 300 mg CBN. Next‐day mood disturbance significantly reduced with both CBN doses. There was a significant increase in subjective drug effect ‘feel effect’ subscale (1‐h only) and ‘like effect’ (next‐day) with 300 mg CBN but no change to other subscales, see Figure 7.

FIGURE 6.

Next day neurobehavioral function in adults with insomnia disorder (N = 20) treated with cannabinol (CBN; 30 and 300 mg) and placebo across three overnight treatment arms. Panels A–C present effect estimates for placebo (black circles), 30 mg (blue squares), and 300 mg (purple triangles) cannabinol (CBN). (A) Standard deviation of lateral position (SDLP) per epoch (car following task, standard freeway drive, and divided attention task [DAT]) assessed next‐day 08:00 h using a fixed‐based driving simulator; (B) Psychomotor Vigilance Test assessing reaction time in milliseconds administered twice next day post wake; (C) Profile of Mood States Questionnaire (POMS) assessed 1‐h post drug administration (~21:00 h) and next‐day (~09:30 h) with higher scores indicating worse mood. Error bars represent 95% confidence intervals. Effect estimates were derived from mixed models of repeated measures with the participant included as a random effect and treatment night (1, 2, 3), intervention (placebo, 30 mg, 300 mg), and timepoint (1, 2) included as fixed effects. Comparisons against placebo only. *p < 0.05, †p < 0.001.

FIGURE 7.

Subjective drug effects in adults with insomnia disorder (N = 20) treated with cannabinol (CBN; 30 and 300 mg) and placebo across three overnight treatment arms. Drug Effects Questionnaire Subscales (‘Feel effect’, ‘Feel high’, ‘Like effects’, ‘Dislike effects’, ‘Want more’, and ‘Feel sedated’) assessed using 0–100 mm visual analogue scales 1‐h post‐administration and next‐day. Effect estimates for placebo (black), 30 mg (blue), and 300 mg (purple) cannabinol (CBN). Error bars represent 95% confidence intervals. Effect estimates were derived from mixed models of repeated measures with the participant included as a random effect and treatment night (1, 2, 3), intervention (placebo, 30 mg, 300 mg), and timepoint (1, 2) included as fixed effects. Comparisons against placebo only. *p < 0.05, †p < 0.001.

No cannabinoids or major phase I metabolites were detected in plasma at pre‐dosing timepoints, but CBN and its metabolites were detected post‐wake in 30 mg and 300 mg arms (Table S3). THC, CBD, and their phase 1 metabolites were not detected in plasma at any time‐point. Forty percent of participants correctly guessed they had received 30 mg and 300 mg CBN (35% were correct in the placebo arm) (Table S4).

3.1. Adverse Events

No serious AEs such as unplanned hospitalizations or deaths occurred. There were no participant withdrawals due to AEs. A total of 246 mild AEs (91 in 300 mg CBN, 73 in 30 mg CBN, and 82 in placebo arms) were reported by 20/20 participants, with one moderate AE reported (in the placebo arm) (Table S5). The most common AEs reported were standardized questionnaire items (e.g., fatigue, drowsiness, lethargy, dry mouth, dizziness, nausea/vomiting, drunk/intoxicated).

4. Discussion

This is the first investigation of CBN isolate in adults with sleep physician‐diagnosed insomnia disorder. We tested two acute oral doses of CBN (30 and 300 mg) versus placebo, on objective and subjective sleep and next‐day function in a three‐arm crossover RCT. The primary outcome, wake‐after‐sleep‐onset, numerically improved with CBN. However, this improvement was neither statistically nor clinically significant (Buysse, Cheng, et al. ²⁰¹⁰; Bettica et al. ²⁰¹²). The effect sizes were small, the confidence intervals were wide, and WASO was on average relatively mild across all study arms (34 min) relative to other trials of hypnotics in insomnia disorder (e.g., 114 min (Rosenberg et al. 2019)) which often exclude objectively mild insomnia.

Secondary outcomes showed acute CBN (300 mg) halved sleep onset latency (−7 min) in a laboratory setting, which is consistent with that reported in meta‐analyses of orexin antagonists (−4.1 to −11.9 min) (Rocha et al. 2023), benzodiazepines (−4.2 min) (Holbrook et al. 2000), and melatonin (−6.3 min) (Maruani et al. 2023). The SOL effect appeared to be driven by individuals with clinically elevated objective SOL. This effect may be most desirable in those who report sleep‐onset insomnia (> 80% of insomnia cases (Pillai et al. 2015)).

The acute improvement in subjective sleep quality (with 300 mg CBN) identified in two validated questionnaires is promising, given the subjective nature of insomnia. However, we found some modest subjective drug effects with 300 mg (although ~10 times milder than THC (Spindle et al. 2018)), which contradicts most (Karniol et al. 1975; Bird et al. ¹⁹⁸⁰; Gong Jr. et al. ¹⁹⁸⁴; Hollister ¹⁹⁷³), but not all (Perez‐Reyes, Timmons, et al. ¹⁹⁷³), historic studies describing CBN as a non‐intoxicating compound. Subjective sleep quality improvements were statistically robust to this effect; however, the potential influence of placebo effects cannot be ruled out. Post hoc analyses showed minor subjective sleep improvement with the 30 mg dose, suggesting lower doses should not be precluded from future repeated dosing investigations where possible accumulation (Table S3) might help overcome low oral bioavailability (4%–19%) (Huestis 2005, 2007; Jung et al. ²⁰²²). In an earlier study, comparable doses of oral CBN (20 mg/day for 10 days) improved some PROMIS subscales in non‐clinical participants (Buysse, Yu, et al. ²⁰¹⁰).

CBN may have reduced central nervous system arousability without impairing important sleep stages. CBN produced a modest reduction in arousal count (300 mg only) and in high frequency EEG activity during REM sleep. Conversely, commonly used sedative‐hypnotics (e.g., benzodiazepines) (de Mendonça et al. ²⁰²³; Bastien et al. ²⁰⁰³; Kang et al. ²⁰²²) and other cannabinoids, such as THC (Walsh et al. 2021; Suraev et al. ²⁰²⁶), have been shown to increase high‐frequency EEG activity and suppress or delay REM sleep. These effects may disrupt neuroplasticity and overnight distress adaptation, perpetuating the cycle of arousability seen in insomnia and anxiety disorders (Feige et al. 2023; Wassing et al. ²⁰¹⁶). In this regard, it is promising that both CBN doses improved next‐day mood. CBN produced other microarchitectural sleep changes, including reduced REM theta power and NREM delta (300 mg only), which could have functional implications (e.g., memory (de Mendonça et al. ²⁰²³; Kang et al. ²⁰²²)) or indicate treatment response (Krystal and Edinger 2010). While qEEG outcomes are not used in clinical practice, these findings warrant longer‐term investigation.

CBN demonstrated a promising safety signal, but the study duration was insufficient to fully evaluate long‐term safety. Lateral vehicular control in simulated driving (i.e., SDLP; ‘weaving’) is recognized as the most sensitive measure of drug and sleepiness induced driving impairment (Vinckenbosch et al. 2020; Irwin et al. ²⁰¹⁷), and this was not impaired by CBN, unlike zopiclone (Leufkens and Vermeeren 2014) and benzodiazepines (Stone et al. 2015; Verster and Roth ²⁰¹³). However, 300 mg CBN produced subtle changes in longitudinal driving control, including increased coherence with the preceding vehicle and greater headway distance, which does not necessarily reflect impairing behaviour. Within the divided attention task, 300 mg CBN increased speed variability without affecting average speed or SDLP, suggesting mild attentional disruption under cognitive load. Participants also reported mild subjective drug effects with 300 mg CBN; however, comparable in magnitude to placebo responses with THC (0 mg oral vaporized THC) (Spindle et al. 2018) and low alcohol doses (0.4 g/kg) (King et al. 2011). These findings align with historic reports of weak psychoactivity following high CBN doses (Perez‐Reyes, Timmons, et al. ¹⁹⁷³) and contradict reports of CBN being non‐psychoactive (Bird et al. 1980; Gong Jr. et al. ¹⁹⁸⁴; Hollister ¹⁹⁷³). However, we recruited cannabis naïve participants who likely had lower tolerance to cannabinoid‐intoxication than participants in 20th century studies. Repeated exposure could also contribute to tolerance or dependence, as reported with other CB₁R‐acting cannabinoids (Colizzi and Bhattacharyya 2018). Given the long half‐life and potential for accumulation of oral cannabinoids, patients should be cautioned about next‐day impairment with high doses, particularly for safety‐critical tasks such as driving. Pharmacokinetic studies of oral CBN isolate are needed to inform optimal dosing strategies and alternative formulations (e.g., sublingual or oromucosal delivery) that may achieve faster onset and shorter duration of effect.

This study has some limitations including the small sample size, large number of study outcomes, and single‐night design, which may not capture night‐to‐night variability in insomnia (Buysse, Cheng, et al. ²⁰¹⁰). However, this approach was appropriate to facilitate extended in‐laboratory observation of a novel drug. Given the high proportion of young female participants, it is relevant to note that we did not standardize for menstrual cycle, although this does not typically influence clinical treatment. We used a fixed‐dose design which may have caused impairment in some individuals (e.g., younger females with low body mass indices). Regarding generalisability, high risk individuals (i.e., those with comorbid sleep disorders or concomitant medication use) were excluded to isolate the drug's effect on sleep and ensure participant safety. Studies exploring CBN in more diverse populations are warranted, as insomnia patients often present with comorbidities and use of concomitant medications that modulate CYP450 enzyme activity (e.g., antidepressants) (Bhaskar et al. 2016). This could alter metabolism and exacerbate either drug exposure or next‐day impairment.

The study findings are strengthened by the rigorous screening procedures and a tightly controlled crossover design, testing two active doses. The 2‐week washout was successful, with no plasma cannabinoids detected pre‐dosing in any participant. Blinding was relatively effectively maintained, with less than half of participants accurately ‘guessing’ their treatment allocation in each study arm.

5. Conclusions

This trial was the first investigation of CBN on objective sleep and in adults with physician‐diagnosed insomnia disorder. Although the primary outcome of wake after sleep onset was not improved, CBN improved other objective and subjective sleep outcomes without impairing sleep architecture, suggesting there is potential for this compound as a hypnotic. There was no compelling evidence of a next‐day ‘hangover’ effect, although longer‐term studies are needed. This trial does not provide direct evidence that CBN is effective for insomnia in routine clinical care. This study provides impetus for larger repeated dosing trials to investigate the efficacy of CBN in treating insomnia disorder, particularly sleep onset insomnia.

Author Contributions

I.G.L., N.S.M., D.M., A.S., J.C.A., C.I., A.L.D'.R., C.J.G., R.R.G., B.J.Y., I.S.M., and C.M.H. were involved in methodological design and creation of the study protocol. B.J.Y. and C.M.H. were the principal investigators (medical and non‐medical, respectively) who were responsible for the design, conduct, and decision to submit for publication. N.S.M. was the study epidemiologist and statistician who designed the analysis plan in collaboration with I.G.L. I.G.L. was the trial coordinator responsible for collecting trial data. Data cleaning, processing, and statistical analyses were conducted by I.G.L., N.S.M., and G.C. I.G.L. drafted the manuscript. All authors have read and approved the final manuscript. Ben Zhang scored polysomnograms, and Sarah Cleophas, Isabella Valenzuela, and Jack Lavender assisted with data collection.

Funding

The study was sponsored by the Woolcock Institute of Medical Research, a specialist sleep and respiratory research institute and clinic (Sydney, Australia). The study was funded by the Lambert Initiative for Cannabinoid Therapeutics, within the Brain and Mind Centre at the University of Sydney, a philanthropically funded centre for cannabinoid research. Lambert Initiative activities extend from plant science and cannabinoid production, through cellular and preclinical pharmacology, to medicinal chemistry and drug discovery, including human laboratory studies and clinical trials. Lambert Initiative members played a role in the design, conduct, and reporting of the study (I.G.L., D.M., R.G., J.C.A., I.S.M.). The decision to fund the study was overseen by a scientific committee known as the Internal Management Group comprising members external to the Lambert Initiative, including the Pro‐Vice Chancellor and the Faculty of Science Dean at the University of Sydney. The investigational product was purchased from Medropharm (Zürich, Switzerland) through BOD Australia (Sydney, Australia) who were not involved in the conception, design, or conduct of the study. Patients and/or the public were not involved in the design, conduct, reporting, or dissemination plans of this research. I.G.L. was supported by a University of Sydney Postgraduate Award Stipend and Barry and Joy Lambert Postgraduate Research Scholarship. C.M.H. is funded by a National Heart Foundation Future Leader Fellowship. A.L.D'.R. (2008001) and RRG (1197439) are funded by National Health and Medical Research Council‐Australian (NHMRC) Investigator Grants.

Conflicts of Interest

I.S.M. is the Academic Director of the Lambert Initiative for Cannabinoid Therapeutics, Brain and Mind Centre, University of Sydney. I.S.M. is a consultant for Kinoxis Therapeutics, Psylo, and Emyria and is an inventor on several patents relating to novel cannabinoid and non‐cannabinoid therapeutics unrelated to insomnia. I.S.M. has received consulting fees from the Medicinal Cannabis Industry Australia (MCIA) and acts as an expert witness in legal cases involving cannabis‐related issues. J.C.A. is Deputy Academic Director of the Lambert Initiative for Cannabinoid Therapeutics, University of Sydney. He has served as an expert witness in various medicolegal cases involving cannabis and has received consulting fees from the World Health Organization (WHO), MCIA and Haleon (consumer healthcare subsidiary of Glaxo Smith‐Kline). He reports research grants from the Australian National Health and Medical Research Council (NHMRC) and from the Lambert Initiative for Cannabinoid Therapeutics. He is an inventor on patents WO2019227167 and WO2019071302 issued, which relate to cannabinoid therapeutics but are unrelated to sleep. A.S. has received consulting fees from the Medicinal Cannabis Industry Australia (MCIA) and Haleon Australia. D.M. has received consulting fees from MCIA and acted as an expert witness in a legal case involving cannabis. R.R.G. has received funding for lectures for Teva, Jazz and Eisai in the past 3 years. The other authors have no competing interests to disclose. The Woolcock Institute Sleep and Chronobiology Research Group has received research support from Eli Lilly, Alkermes, Avadel, Nyxoah, Idorsa, ResMed, BOD Australia and Philips.

Supporting information

Data S1: Supporting Information.

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.