Cannabidiol for moderate–severe insomnia: a randomized controlled pilot trial of 150 mg of nightly dosing

Abstract

Study Objectives:

Low-dose cannabidiol (CBD) has become readily available in numerous countries; however, little consensus exists on its efficacy as a sleep aid. This trial explored the efficacy of 150 mg of CBD (n = 15) compared with placebo (n = 15) as a sleep aid in primary insomnia. CBD supplementation was hypothesized to decrease insomnia symptoms and improve aspects of psychological health, relative to placebo.

Methods:

Using a randomized, placebo-controlled, parallel design featuring a single-blind placebo run-in week followed by a 2-week double-blind randomized dosing phase, participants consumed the assigned treatment sublingually 60 minutes before bed nightly. Wrist-actigraphy and sleep diaries measured daily sleep. Sleep quality, sleep effort, and well-being were measured weekly over 4 in-laboratory visits. Insomnia severity and trait anxiety were measured at screening and study conclusion.

Results:

Insomnia severity, self-reported sleep-onset latency, sleep efficiency, and wake after sleep onset did not differ between treatments throughout the trial (all P > .05). Compared with placebo, the CBD group reported greater well-being scores throughout the trial (trial end mean difference = 2.60; standard error: 1.20), transient elevated behavior following wakefulness scores after 1 week of treatment (mean difference = 3.93; standard error: 1.53), and had superior objective sleep efficiency after 2 weeks of treatment (mean difference = 6.85; standard error: 2.95) (all P < .05). No other significant treatment effects were observed.

Conclusions:

Nightly supplementation of 150 mg CBD was similar to placebo regarding most sleep outcomes while sustaining greater well-being, suggesting more prominent psychological effects. Additional controlled trials examining varying treatment periods and doses are crucial.

Clinical Trial Registration: Registry: Australian New Zealand Clinical Trials Registry; Name: Cannabidiol (CBD) treatment for insomnia; URL: https://anzctr.org.au/Trial/Registration/TrialReview.aspx?ACTRN=12620000070932; Identifier: ACTRN12620000070932.

Citation:

Narayan AJ, Downey LA, Rose S, Di Natale L, Hayley AC. Cannabidiol for moderate–severe insomnia: a randomized controlled pilot trial of 150 mg of nightly dosing. J Clin Sleep Med. 2024;20(5):753–763.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Primary insomnia accounts for approximately 25% of sleep-related diagnoses and is linked to decreased mood and quality of life. Emerging anecdotal evidence suggests that cannabidiol (CBD) is an effective sleep aid with a safe side-effect profile, leading to its growing accessibility. Yet, there is insufficient evidence concerning its efficacy from randomized controlled clinical trials, leading to unclear dosage and therapeutic guidelines for sleep benefits.

Study Impact: This pilot trial was the first to explore the maximum dose of over-the-counter CBD in Australia (150 mg) over 2 weeks of nightly dosing in people with primary insomnia (n = 15) compared with placebo (n = 15). While CBD did not reliably enhance sleep, the significant preservation of well-being suggests some positive therapeutic potential.

INTRODUCTION

Globally, over one-third of all adults report insomnia symptoms throughout their lifetime.¹ Primary insomnia, defined as persistent insomnia symptomology in the absence of underlying medical or psychiatric pathology, accounts for an estimated 25% of all sleep-related diagnoses.^1,2 Sleep disturbances and affect are intrinsically and bidirectionally linked, and their shared pathology is often established and strengthened via persistent negative feedback loops of physiological hyperarousal.^3,4 Long-term persistence of this bidirectional relationship can lead to comorbid health conditions and decreased quality of life.^5,6 The severity and longevity of these comorbid symptoms are currently ineffectively managed by conventional pharmacological treatments, warranting exploration into alternative treatments such as cannabinoids.⁷

Cannabinoids are a promising treatment for various clinical conditions, including disrupted sleep and elevated anxiety.⁸ Cannabidiol (CBD), the nonintoxicating constituent of the cannabis plant, is theorized to produce sleep-enhancing effects via neuromodulatory action on shared receptor sites, including CB1, GABAa, 5HT-1A, and TRPV1 receptors.^9,10 Pharmacokinetic studies of CBD in human adults have revealed highly variable mean times to maximum concentration (range: 1 to 6.13 hours), with evidence for sublingual CBD reaching a maximum concentration earlier than oral formulations in doses between 20 mg and 6,000 mg.¹¹ Maximum peak concentration was also suggested to be dose-dependent and not dose-proportional, with formulations of oral CBD suggested to play an important role.^11,12 Limited studies assessing the half-life of CBD have produced evidence suggesting a half-life of up to 2 days for single dose in humans,¹³ with high-fat meals prior to administration resulting in increased CBD exposure and modifying the half-life by an estimated 20% in healthy participants due to its highly lipophilic nature.¹⁴ Last, CBD drug–drug interaction effects are largely unknown, with some pediatric studies reporting increased serum concentrations of antiseizure medication¹⁵ and methadone¹⁶ when taken with CBD, further suggesting these potential interactions need to be taken into consideration when administering CBD treatments.

Countries including Australia have recently permitted the sale of over-the-counter CBD in pharmacies; yet, studies have not demonstrated the efficacy of this cannabinoid for these target conditions.¹⁷ A recent review investigated the efficacy of CBD with various doses taken during a single dosing session or longer periods of up to 6 months of daily dosing.⁶ A maximum of 175 mg administered daily for 1 month showed decreased and preserved anxiety levels with fluctuating self-reported sleep improvements.¹⁸ Similar results were reported for single doses between 18 mg and 800 mg and treatment periods over 1 month, suggesting more prominent anxiolytic, rather than sleep, effects.⁶ Differing administration routes and target doses distributed among heterogeneous clinical populations have meant that the direct mechanism contributing to CBD’s proposed therapeutic benefit has not yet been clearly defined. Indeed, it is currently unclear if the proposed sleep-enhancing effects can be adequately replicated at therapeutically indicated doses in primarily sleep-disordered cohorts under highly controlled conditions. Thus, additional, well-controlled, randomized trials are urgently needed to determine the efficacy of CBD as a treatment for sleep in these clinical populations.

A daily maximal dose of 150 mg was recently permitted for sale in Australia; however, an approved product remains unavailable and treatment efficacy is yet to be confirmed, and the therapeutic window is yet to be defined. Although the effects of CBD-only treatments on sleep indices have been evaluated in prior studies, these studies have not examined a patient population with primary insomnia.^5,6 In contrast, recent studies assessing sleep in primary insomnia have focused on the effects of combined cannabinoid treatments only.^7,19 Populations with primary insomnia remain greatly understudied, and considering the frequency of cannabinoids used for insomnia symptoms,¹⁷ this trial aimed to provide the first examination as to the effects of nightly supplementation of 150 mg of CBD objective and self-reported (subjective) sleep effects in adults with moderate to severe primary insomnia. It secondarily aimed to assess CBD effects on well-being and trait anxiety. It was hypothesized that CBD would improve sleep and have beneficial effects on mood and well-being in a population without pathological mood scores.

METHODS

A 1-week, single-blind placebo run-in, followed by a 2-week randomized, double-blind, placebo-controlled, parallel trial was used to compare the effects of CBD and placebo on sleep, trait anxiety, and well-being. A comprehensive and multistage screening procedure was followed prior to enrollment, as outlined below, and shown in Figure 1. This trial was prospectively registered online with the Australia and New Zealand Clinical Trials Registry (ID: ACTRN12620000070932) and approved by the Swinburne University of Technology’s Human Research Ethics Committee (approval granted: January 20, 2020; Ref: 20220392-9708). All data were collected and stored at the Centre for Mental Health and Brain Sciences located at Swinburne University of Technology in Hawthorn, Melbourne, Australia. This trial was conducted in adherence with Good Clinical Practice guidelines and the ethical standards of the Declaration of Helsinki. Ongoing consent was taken at each of the following weekly in-laboratory testing visits.

Figure 1. Study schedule: study schedule, testing visits, and measures at each trial phase.

BAI = Beck Anxiety Inventory, BDI = Beck Depression Inventory, GSES = Glasgow Sleep Effort Scale, ISI = Insomnia Severity Index, LSEQ = Leeds Sleep Evaluation Questionnaire, STAI-T = State-Trait Anxiety Index (Trait anxiety only), STOP-BANG = snoring history, tired during the day, observed stop of breathing while sleeping, high blood pressure, WHO-5 = World Health Organization Well-Being Index-5.

Telephone screening (prescreening), in-laboratory screening visit (V0)

Informed consent was obtained verbally during the prescreening call, with males and females aged between 18 and 45 years, with ongoing moderate–severe insomnia symptom severity (Insomnia Severity Index [ISI] score ≥ 15)²⁰ scheduled for the in-laboratory screening visit (V0). Participants reported any prior, recent, or ongoing use of medicinal or recreational drugs during an extensive medical history interview conducted by the trial nurse. Those who self-reported medical conditions (eg, parasomnias, psychiatric, or clinical conditions) or took medication likely to affect sleep (eg, antidepressants, opioids, benzodiazepines), who were currently engaged in shift work, or who consumed excessive amounts of caffeine (> 400 mg caffeine/4 cups of coffee) were excluded during prescreening. Written consent was acquired at the in-laboratory screening (V0), with those with severe depressive symptoms (Beck Depression Inventory score ≥ 20),²¹ severe anxiety symptoms (Beck Anxiety Inventory score ≥ 16),²¹ or at moderate to severe risk of obstructive sleep apnea (STOP-BANG [snoring history, tired during the day, observed stop of breathing while sleeping, high blood pressure] score ≥ 5)²² excluded during this visit (Figure 1). A washout period was determined by the trial doctor for any self-reported changes to medication that may interfere with sleep throughout the trial. Any self-reported use of illicit drugs prior to all visits resulted in exclusion from the trial.

In-laboratory testing visit—placebo run-in week (V1–V2)

All eligible participants meeting the initial screening criteria were enrolled in a single-blind, placebo run-in week (V1) with instructions to continue with their normal routines and fill in the provided sleep diaries and treatment compliance logs in addition to nightly dosing (Figure 1). GENEActiv actigraphy watches (version 1.1, Activinsights, Kimbolton, United Kingdom) were required to be worn daily for the placebo run-in week, with non-wear to be recorded in the provided pen-and-paper watch log. Data from this week were used to confirm eligibility by identifying and excluding placebo responders and treatment noncompliers. The weekly average of each sleep parameter was calculated and participants with 1 or fewer self-reported sleep parameters (self-reported sleep efficiency [SE] > 85%, sleep-onset latency [SOL] < 31 minutes, wake after sleep onset [WASO] < 31 minutes) were classified as placebo responders. Treatment noncompliers were those missing 20% or more of the doses, excessive use of treatment (> 15 mL used), or who had > 20% of actigraphy data missing. Both placebo responders and treatment noncompliers were excluded prior to randomization at the next in-laboratory testing visit (V2), scheduled at the end of the placebo run-in week (Figure 1).

In-laboratory testing visits (V3, V4)—randomized dosing weeks 1 and 2

Eligible participants were randomized and given 2 weeks’ worth of active or placebo treatments at the end of the placebo run-in week (V2), with the next in-laboratory visits (V3 and V4) scheduled 1 week and 2 weeks after daily dosing (Figure 1). Participants followed the same procedures as the placebo run-in week throughout the trial, with self-reported sleep diary entries, treatment compliance logs, and treatment bottle weight checked at all weekly visits. All sleep data and logs were collected at the end of placebo run-in, 1 week after dosing, and 2 weeks after dosing, with computerized questionnaires used during the weekly in-laboratory testing visits (Figure 1).

G*Power (version 3.1) was used to determine the number of participants to detect a small–moderate effect size (0.3) across a 2-way between-subjects design on reduction in ISI score. A small effect size, 80% power, and a .05 P value required 23 completed participants; therefore, this study recruited 15 completed participants per treatment group (n = 30).

Study treatments

CBD and the placebo oil vehicle (corn oil) were provided by Brains Bioceutical, UK. Treatments were compounded into identical 30-mL bottles of 100-mg/mL CBD treatments by Aspa Pharmacy, Prahran, Melbourne, before being stored securely on-site at Swinburne University prior to dispensing. Participants were instructed to ingest 1.5 mL of the oil treatment (150 mg CBD or corn-oil placebo) nightly, 60 minutes before bed, using a supplied 3-mL graduated dosing syringe. Both oil treatments were identical visually and in odor without any alterations to their taste.

Researchers recruited and enrolled all participants using Swinburne trial recruitment databases, word of mouth, posters, and ads (both physical and online via social media). Laboratory staff outside the trial coded and maintained the key to treatment coding until data collection was completed. Treatments were allocated randomly using randomization software (Research Randomizer Software, version 4.0). Treatment bottles were given numbers and an order for dispensing was successfully kept by staff outside the trial to maintain blinding regardless of receiving placebo or CBD.

Primary outcomes

Insomnia severity symptomology (ISI) data were collected and analyzed at prescreening and after 2 weeks of dosing. Daily self-reported SOL (minutes), WASO, and SE (%) (total reported hours of sleep divided by total time spent in bed [time in bed minus time out of bed], multiplied by 100) data were collected using daily pen-and-paper sleep diary entries. Sleep diaries were collected during the weekly visits (Figure 1).

Secondary outcomes

Sleep diary–derived daily sleep quality, total sleep time (TST), and number of nightly awakenings in addition to actigraphy-derived SOL, SE, WASO, TST, and number of awakenings after sleep onset were collected. In addition to the weekly collection of sleep diaries, actigraphy data were also downloaded during the weekly visits (Figure 1).

Furthermore, computerized questionnaires were used to measure changes in well-being (World Health Organization Well-being Index-5 [WHO-5]), sleep quality (Leeds Sleep Evaluation Questionnaire [LSEQ]), and sleep effort (Glasgow Sleep Effort Scale [GSES]) and which were recorded at all in-laboratory visits (screening, start and end of placebo run-in, after 1 and 2 weeks of dosing). The WHO-5 is a short 5-question scale designed to measure levels of self-reported well-being, with higher scores reflecting greater well-being.²³ The LSEQ consists of ten 100-mm visual analog scales evaluating aspects of sleep and waking. It is divided into subscales assessing the ease of getting to sleep, quality of sleep, ease of awakening from sleep, and behavior following wakefulness, with higher scores indicating greater sleep quality.²⁴ The GSES is composed of 7 items, each with a 3-point Likert scale. It was developed to measure the direct and voluntary attempts to control sleep that contrarily worsen and perpetuate insomnia symptoms, with higher scores indicating greater sleep effort.²⁵ Additionally, trait anxiety (State-Trait Anxiety Index [STAI-T]) was measured during screening and after 2 weeks of dosing (Figure 1) to assess changes in underlying and more stable anxiety characteristics.²⁶

Analytical plan

Actigraphy data were extracted and cleaned using GGIR (version 2.8-2)²⁷ on R software (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria). Watches were configured with a sampling rate of 100 Hz for 7 days and worn on the nondominant hand, with potential sleep periods calculated using the default GGIR algorithm (5 minutes of inactivity, ±5 degrees).²⁷ A minimum number of 16 valid hours per night (within 24 hours measured from noon to noon) was considered for analyses. Watches were at default sensitivity (12 bit [3.9 g] resolution and a range of ± 8 g) and analyzed with GGIR using 5-second epochs.²⁷ Data from participant sleep diaries were used to clean actigraphy data for non-wear and naps in addition to being used as a guide for producing nondefault GGIR parameters, including the duration of spent time in bed, SOL, and SE.²⁷ Primary outcomes of sleep diary–derived SOL, WASO, and SE; secondary outcomes of sleep diary–derived sleep quality and TST, actigraphy-derived SOL, SE, WASO, and TST; in addition to weekly LSEQ categories (overall, getting to sleep, awakening from sleep, and behavior following wakefulness), GSES, and WHO-5 were each analyzed using a separate linear mixed effect (LME) model with maximum likelihood estimation for each outcome to compare CBD with placebo changes over time. Each separate model had treatment and time entered as repeated measures and fixed effects, with participants entered as random effects. The likelihood ratio statistic was used to determine the best-fitting variance structure with compound symmetry used for all between-treatment analyses. Daily actigraphy and sleep diary outcomes with nonnormal distribution (determined by Shapiro-Wilk [P < .05] and distribution plots) were transformed using log or square root transformations with original data used if transformed data remained non–normally distributed.²⁸ Weekly questionnaire data (LSEQ, GSES, WHO-5) did not require transformations. A main effect of treatment, time, or the interaction of treatment and time resulted in the examination of post hoc paired t tests with Bonferroni corrections for multiple comparisons. LME model–generated weekly means were used to compare treatments for all daily sleep measures.

Demographic characteristics were analyzed to determine sample size, percentage of sample size (categorical variables), means and standard deviations (continuous variables) for each treatment group and the full sample (Table 1). The primary outcome of ISI changes and secondary mood outcome of STAI-T changes were analyzed using independent-samples t tests (Table 2). Cohen’s d was calculated to measure the effect size between all CBD and placebo group outcomes. All statistical analyses were performed using SPSS (version 29; IBM Corporation, Armonk, NY) with tests being 2-tailed and statistical significance defined as P < .05. LME model interaction effects, within treatment raw means and standard deviations (SDs), are reported in Table 3, Table 4, Table 5 and Table 6, with only between-group outcomes comprising the body of the results presented here.

Table 1.

Demographic characteristics.

Baseline Characteristic	CBD		Placebo		Full Sample
	n	%	n	%	n	%
Sex
Female	8	53	7	47	15	50
Male	7	47	8	53	15	50
Height, cm, mean (SD)	173.8 (7.1)		175.7 (8.5)		174.8 (9.6)
Weight, kg, mean (SD)	70.2 (19.2)		75.1 (17.3)		72.7 (18.5)
Age, y, mean (SD)	33.5 (7.1)		29.7 (6.0)		31.6 (6.8)
Handedness
Left	—	—	2	13.3	2	6.7
Right	14	93	12	80	26	86.7
Ambidextrous	1	6.7	1	6.7	2	6.7
Total years of education, mean (SD)	17.4 (3.2)		15.7 (2.3)		16.6 (2.9)
Highest educational levela
Secondary	2	13.3	3	20	5	16.7
Tertiary	9	60	10	66.7	19	63.3
Postgraduate	3	20	2	13.3	5	16.7
Employment
Full-time	7	46.7	9	60	—	53.3
Part-time	6	40	2	13.3	—	26.7
Studying	1	6.7	3	20	—	13.3
Unemployed	1	6.7	1	6.7	—	6.7
Ethnicity
European/European descent	15	100	12	80	27	90
Indian	—	—	2	13.3	2	6.7
Chinese	—	—	1	6.7	1	3.3
First language
English	15	100	14	93.3	29	96.7
Other	—	—	1	6.7	1	3.3

Baseline demographic and clinical group characteristics are shown. ^an = 1 missing. CBD = cannabidiol, SD = standard deviation.

Table 2.

ISI and trait anxiety outcomes.

Measure	Treatment	Baseline/Pretreatment	After 2 Weeks of Dosing
ISI	CBD	18.6 (2.5)	11.6 (4.2)
	PLA	17.4 (2.3)	9.7 (4)
STAI-T	CBD	42.2 (5.7)	43.7 (4.4)
	PLA	42.5 (3.2)	45.3 (4.3)

ISI and trait anxiety absolute mean scores and standard deviations at baseline and after 2 weeks of daily dosing are shown. CBD = cannabidiol treatment group, ISI = Insomnia Severity Index, PLA = placebo treatment group, STAI-T = State-Trait Anxiety Inventory—Trait Anxiety.

Table 3.

Sleep diary outcomes.

Measure	Treatment	After Placebo Run-In	After 1 Week of Dosing	After 2 Weeks of Dosing	F	P
Sleep quality	CBD	2.8 (1)	3.4 (1)	3 (1)	8.1(2, 582.5)	<.001*
	PLA	2.9 (1.1)	2.9 (1)	3.1 (0.9)
No. of awakenings	CBD	2.1 (1.9)	1.5 (1.6)	1.5 (1.7)	0.4(2, 569.1)	.69
	PLA	1.7 (1.3)	1.4 (1.7)	1.3 (1.3)
WASO (decimal hours)	CBD	0.6 (0.5)	0.5 (0.6)	0.6 (0.7)	1.3(2, 544.1)	.27
	PLA	0.6 (0.7)	0.4 (0.7)	0.4 (0.6)
SE (%)	CBD	86.8 (3.1)	86.4 (4.2)	87 (2.7)	1.5(2, 567.1)	.22
	PLA	87 (3.8)	85.9 (6)	85.7 (6.3)
SOL (decimal hours)	CBD	0.82 (0.9)	0.6 (0.7)	0.6 (0.7)	0.6(2, 576)	.63
	PLA	0.9 (0.7)	0.6 (0.8)	0.5 (0.4)
TST (decimal hours)	CBD	7.94 (1.5)	8 (1.6)	8 (1.5)	1.8(2, 566.3)	.17
	PLA	8 (1.5)	7.8 (1.8)	7.6 (1.8)

Weekly sleep diary–derived raw means, standard deviation, and interaction terms are shown. *Statistically significant effect (P < .05). “No. of awakenings” indicates number of awakenings after sleep onset. CBD = cannabidiol treatment group, PLA = placebo treatment group, SE = sleep efficiency, SOL = sleep-onset latency, TST = total sleep time, WASO = wake after sleep onset.

Table 4.

Actigraphy outcomes.

Measure	Treatment	After Placebo Run-In	After 1 Week of Dosing	After 2 Weeks of Dosing	F	P
No. of awakenings	CBD	13.8 (6.8)	14.4 (6)	14.6 (5.1)	3.2(2, 525.6)	.04*
	PLA	16.7 (5.9)	15.6 (6.3)	14.3 (6.8)
WASO (decimal hours)	CBD	1.0 (1.1)	1.2 (1.3)	0.9 (0.5)	0.2(2, 532.5)	.78
	PLA	1.4 (1)	1.6 (1.5)	1.3 (1.4)
SE (%)	CBD	77 (17.4)	76.8 (15.8)	80.2 (7.6)	0.1(2, 533.6)	.91
	PLA	74.3 (12.9)	70.4 (20.4)	72.2 (17.7)
SOL (decimal hours)	CBD	0.3 (0.8)	0.4 (0.4)	0.4 (0.4)	0.3(2, 539.3)	.76
	PLA	0.4 (0.4)	0.4 (0.5)	0.5 (0.6)
TST (decimal hours)	CBD	6 (2)	6.1 (1.7)	6.3 (1.2)	1.2(2, 531)	.29
	PLA	6.1(1.5)	5.9 (2.1)	5.6 (2.1)

Weekly actigraphy-derived raw means, standard deviation, and interaction terms are shown. *Statistically significant effect (P < .05). “No. of awakenings” indicates number of awakenings after sleep onset. CBD = cannabidiol treatment group, PLA = placebo treatment group, SE = sleep efficiency, SOL = sleep-onset latency, TST = total sleep time, WASO = wake after sleep onset.

Table 5.

LSEQ outcomes.

Categories	Treatment	Screening	Placebo Run-In Start	Placebo Run-In Start	After 1 Week of Dosing	After 2 Weeks of Dosing	F	P
Overall	CBD	33 (12.7)	35.2 (10.9)	52.3 (13.2)	58.6 (12.7)	59.6 (14.6)	0.1 (4,111.3)	.95
	PLA	29 (8.9)	30.3 (10.9)	47 (10.5)	50.4 (9.1)	54 (11.2)
Getting to sleep	CBD	11.2 (5)	12.2 (5.4)	16.3 (6.2)	18.3 (4)	19.2 (5.2)	0.1(4,111.3)	.97
	PLA	9.2 (3.6)	10.8 (5)	15.5 (4.6)	16.2 (5)	17.3 (3.7)
Quality of sleep	CBD	3.9 (2.7)	5 (2.8)	11.3 (4.5)	11.9 (3.7)	12.8 (4.3)	0.4(4,111.6)	.81
	PLA	4.6 (2.8)	4.7 (2.2)	9.9 (3.9)	10.9 (2.9)	12.2 (3.2)
Awake following sleep	CBD	7.1 (4.9)	8.8 (4.7)	9.9 (4.3)	10.9 (3.2)	10.2 (3.9)	0.6(4,111.3)	.67
	PLA	7.9 (4.6)	7.1 (4.5)	9.2 (2.5)	9.7 (2.6)	9.7 (3.5)
Behavior following awakening	CBD	10.8 (5.7)	9.1 (4.1)	14.9 (3.7)	17.5 (4.3)	17.4 (4.9)	0.7(4,110.9)	.59
	PLA	7.4 (3.4)	7.7 (3.2)	12.4 (3.7)	13.6 (4.2)	14.8 (3.9)

Weekly raw means, standard deviation, and interaction terms for each LSEQ category are shown. CBD = cannabidiol treatment group, LSEQ = Leeds Sleep Evaluation Questionnaire, PLA = placebo treatment group.

Table 6.

GSES and WHO-5 outcomes.

Categories	Treatment	Screening	Placebo Run-In Start	Placebo Run-In Start	After 1 Week of Dosing	After 2 Weeks of Dosing	F	P
GSES	CBD	9.4 (2.2)	9 (2.5)	6.7 (3)	6 (3.2)	5.8 (3.4)	2(4, 111.1)	.11
	PLA	8.8 (2.2)	9.3 (2.8)	7.3 (3.3)	7 (3.3)	6.2 (3.3)
WHO-5	CBD	13.9 (3.7)	14.2 (4.8)	15.7 (2.4)	17.3 (2.5)	17.5 (2.6)	0.4(4, 111.2)	.82
	PLA	11.3 (3)	12.3 (3.1)	14.3 (3.8)	14.6 (3.5)	14.9 (3.1)

Weekly raw means, standard deviation, and interaction terms for GSES and WHO-5 are shown. CBD = cannabidiol treatment group, GSES = Glasgow Sleep Effort Scale, PLA = placebo treatment group, WHO-5 = World Health Organization Well-being Index-5.

RESULTS

Recruitment and sample characteristic

Participant information and consent forms were sent to 810 registrants between February 2021 and September 2022, of whom 88 passed the prescreening criteria and 76 were excluded. Of the eligible participants, 18 participants withdrew due to time commitments, with 70 attending the in-laboratory screening visit (V0). A further 15 participants were excluded for high Beck Anxiety Inventory/Beck Depression Inventory, and 3 had other reasons for not wanting to continue. Prior to the placebo run-in visit, an additional 8 of the 52 remaining participants withdrew from the trial due to other commitments, leaving 44 participants starting the placebo run-in period (V1) and being assessed for eligibility at the end of the placebo run-in week (V2). A total of 8 participants were identified as placebo responders and 2 withdrew due to time commitments and an adverse event of anxiety and paranoia. The remaining 34 participants were randomized to receive either CBD (n = 18) or placebo (n = 16). At the end of 1 week of randomized dosing (V3), 1 participant in the CBD group withdrew due to side effects (ongoing restlessness) and 1 participant in the placebo group withdrew for other reasons, resulting in 32 participants attending the final visit after 2 weeks of continuous dosing and completing the trial.

Consistent with a priori trial registration, a participant pool of n = 30 completers were analyzed and are presented, with the 4 extra participants randomized to mitigate late-stage attrition (n = 2) and significant coronavirus disease 2019 (COVID-19) delays that risked failure to meet recruitment targets prior to investigational product expiry (n = 2) (see Figure 2). All adverse events were considered mild and transient in nature, and affected individuals were debriefed and put in contact with any necessary services.

Figure 2. Participant flow diagram: adapted CONSORT participant flow diagram showing recruitment at each trial phase.

BAI = Beck Anxiety Inventory, BDI = Beck Depression Inventory, CONSORT = Consolidated Standards of Reporting Trials, ISI = Insomnia Severity Index.

The final sample consisted of 15 males and 15 females, with an overall mean age of 31.6 years (SD ±6.84 years), a mean height of 174.8 cm (SD ± 9.6 cm), and weight of 72.7 kg (SD ± 18.5 kg). Most participants had completed a tertiary degree (63.3%), had full-time employment (53.7%), were right-handed (86.7%), and spoke English as their first language (96.7%) (Table 1).

Primary outcomes

ISI scores were comparable at the prescreening visit (baseline mean difference = –1.2; standard error: 0.9; P > .05; confidence interval [95% CI]: –3, 0.6; d = –0.5), with no differences noted between groups after 2 weeks of dosing at study conclusion (mean difference = –1.9; standard error: 1.5; P > .05; 95% CI: –5, 1.1; d = –0.5) (Table 2). Similarly, sleep diary–derived SOL, SE, and WASO did not differ as a function of treatment at any point within the trial period including conclusion (all P > .05) (Table 3).

Secondary outcomes

A significant main effect for the interaction of time and treatment [F_{(2, 582.5)} = 8.1, P < .001] was observed for self-reported daily sleep quality (Table 3). Post hoc analyses showed that participants in the CBD group reported improved sleep quality compared with placebo after 1 week of dosing (mean difference = 0.5; standard error: 0.2; P = .025; 95% CI: 0.1, 0.9; d = 0.5), but not after 2 weeks of dosing at study conclusion (P > .05). No significant condition effects were observed for the self-reported number of awakenings after sleep onset or TST (Table 3).

An interaction effect of treatment and time was observed for the objective number of awakenings after sleep onset [F_{(1, 519.5)} = 3.3; P = .039] (Table 4). The participants in the CBD group showed fewer awakenings than placebo at the trend level at the end of the placebo run-in week (mean difference = –2.9; standard error: 1.5; P = .054; 95% CI: –5.9, 0.1; d = 0.5). The number of awakenings became similar across treatments groups after 1 and 2 weeks of dosing (all P > .05).

A significant main effect for treatment [F_{(1, 28.6)} = 5.4; P = .027] for objective WASO relative to placebo was observed, with participants receiving CBD presenting less WASO relative to placebo at the end of the placebo run-in only (mean difference = –0.2; standard error: 0.1; P = .023; 95% CI: –0.4, –0.0; d = 0.3) (Table 4). No difference was noted between treatments at any other time point (P > .05).

A significant main effect for treatment [F_{(1, 27.9)} = 6.8; P = .01] was observed for objective SE (Table 4). Participants receiving CBD presented significantly greater SE than those receiving placebo after 1 week of dosing (mean difference = 6.7; standard error: 2.8; P = .02; 95% CI: 1.2, 8.3; d = 0.4) and at study conclusion (mean difference after 2 weeks of dosing = 6.9; standard error: 3; P = .02; 95% CI: 1, 12.7; d = 0.6).

No statistically significant condition effects were noted for objective SOL and TST at any time point (all P > .05) (Table 4).

Weekly sleep questionnaire outcomes

A significant main effect for treatment was observed for total LSEQ score [F_(1,28.1) = 4.5; P = .04] (Table 5). Post hoc analyses showed increased total LSEQ after 1 week of dosing for participants in the CBD group failing to reach significance relative to placebo (mean difference = 8.2; standard error: 4.2; P = .054; 95% CI: –0.2, 16.6; d = 0.8), with no other significant effects observed throughout the trial period (all P > .05). An observed main effect of treatment for the LSEQ subcategory of behavior following wakefulness [F_(1,28.8) = 6.1; P = .016] was noted, with post hoc analyses showing that those receiving CBD reported greater LSEQ behavior following wakefulness at the in-laboratory screening visit (baseline mean difference = 3.4; standard error: 1.5; P = .03; 95% CI: 0.3, 6.4; d = 0.7) and after 1 week of dosing only, relative to placebo (mean difference = 3.9; standard error: 1.5; P = .01; 95% CI: 0.9, 67; d = 0.93) (Table 5). There were no observed condition effects for LSEQ subscales of getting to sleep, quality of sleep, and awake following sleep at any assessment time point throughout the trial period (all P > .05) (Table 5). Sleep effort scores (GSES) did not differ as a function of treatment at any assessment time point throughout the trial period (P > .05) (Table 6).

Mood questionnaire outcomes

Trait anxiety (STAI-T) was comparable between treatments at the in-laboratory screening visit (baseline mean difference = 0.3; standard error: 1.7; P > .05; 95% CI: –3.1, 3.8; d = 0.1) and after 2 weeks of dosing at study conclusion (mean difference = 1.7; standard error: 1.6; P > .05; 95% CI: -1.6, 4.9; d = 0.4) (Table 2).

There was a significant main effect for treatment [F_(1,28.1) = 6.2; P = .02] for well-being (WHO-5). Statistically significant group contrasts were observed at the in-laboratory screening visit (baseline mean difference = 2.6; standard error: 1.2; P = .03; 95% CI: 0.2, 5; d = 0.8), after 1 week of dosing (mean difference = 2.7; standard error: 1.2; P = .03; 95% CI: 0.3, 5.1; d = 0.9), and at study conclusion after 2 weeks of dosing (mean difference = 2.6; standard error: 1.2; P = .03; 95% CI: 0.2, 5; d = 0.9), with those receiving CBD reporting consistently higher well-being compared with placebo (Table 6).

DISCUSSION

This randomized, placebo-controlled, parallel pilot study provided the first empirical evidence that nightly sublingual supplementation of 150 mg of CBD oil sustained greater well-being compared with placebo in addition to temporary benefits for feeling alert, balanced, and coordinated upon waking (as measured by the LSEQ subcategory behavior following wakefulness) and objective SE among individuals with primary insomnia, whereas insomnia severity, secondary sleep outcomes, and mood remained unaffected following a 2-week dosing period. Additionally, results indicated that a lower objective number of awakenings after sleep onset and WASO noted at the end of the placebo run-in week did not persist in the CBD group during the 2 weeks of active dosing relative to placebo. Past evidence has suggested that CBD may have wake-promoting properties; yet, these findings do not strongly support this.²⁹

Enhanced objective SE and modest improvements in alertness, balance, and coordination upon awakening indicated that, at 150 mg, CBD may be more therapeutically beneficial for these specific measures. Other objective sleep outcomes, including number of awakenings after sleep onset and WASO, were largely comparable between CBD and placebo treatments or else did not change substantially from baseline. With insomnia symptoms known to vary heterogeneously across the condition, improvement within distinctive insomnia symptomology time frames (acute, subacute, chronic) and phenotypes of sleep (difficulty staying asleep, early awakenings, sleep onset, or a combination) may result in inconsistently measured outcomes in sleep assessments when insomnia is treated uniformly.³⁰ The transient changes in behavior following wakefulness and consistent objective SE improvements suggested that the sample utilized may have had more prominent difficulties staying asleep and/or sleep-onset insomnia characteristics. The observed changes in both measures are not exclusive to 1 phenotype, and therefore without the distinction between symptom time frames and sleep phenotypes in the present study, it is uncertain if present improvements were due to baseline differences in insomnia symptomology or if CBD was therapeutically beneficial for these specific parameters. It is necessary for future trials to treat insomnia as a heterogenous condition based on predefined symptom categories to accurately assess if therapeutic benefits observed are uniformly generalizable across insomnia populations, or if subgroup differences in insomnia symptomology affect measurable treatment outcomes, as well as to explore appropriate doses and dose administration timing for the differing phenotypes.⁷ Additionally, with only objective SE outcomes showing a consistent difference between groups over the dosing period, it is difficult to determine if CBD improved the utilization of time in bed for sleep specifically without consequentially affecting other daily objective and subjective parameters compared with placebo. Weekly self-reported SE averages demonstrated improvements for the CBD group; however, treatments were statistically comparable, indicating either a discrepancy between self-report and objective measures or heightened actigraphy sensitivity in detecting SE more so than other parameters.⁷

Results suggested that mood outcomes were not affected by the initially elevated scores observed for behavior following wakefulness and objective SE in the CBD group, as expected. Higher baseline well-being in the CBD group indicated initial group differences prior to receiving the active treatment; yet, elevated scores remained consistent after 1 and 2 weeks of active dosing, which suggested that treatment with CBD did not compromise well-being outcomes over the trial period. Similarly, trait anxiety outcomes were unlikely to be affected when comparing treatments despite past evidence of more prominent anxiolytic effects when treated with CBD.^31–35 This could be attributed to dose size and administration route, considering past studies have observed a dose-dependent inverted U-shaped relationship in acute CBD doses between 300 and 400 mg in healthy participants, suggesting that the dose used in this trial may have been too low for the occurrence of anxiolytic effects.^36,37 Additionally, a single dose of 150 mg CBD oil administered via capsule demonstrated faster peak and oral bioavailability compared with a powdered vehicle,¹² with no changes in sedation or anxiety noted within 1 dosing session, whereas limited mood improvements were observed in as little as 1 session in vaporized and inhaled doses under 150 mg.³⁸ In contrast, daily intake of up to 175 mg for 4 weeks¹⁸ or 300 mg for 12 weeks³¹ has also shown limited and fluctuating improvements in sleep satisfaction in populations with sleep conditions, whereas anxiety had more significant and prolonged improvements in a sample more prone to anxiety.¹⁸ Taken together, these observations highlighted overall discrepancies in what could be considered therapeutically beneficial doses, resulting in the lack of concise therapeutic guidelines currently available.⁵ The accumulation of CBD through daily dosing was also suggested to play a role in beneficial outcomes due to nightly administration, but cannot be reliably distinguished from the current study’s results,⁷ with past evidence also strongly suggesting that administration routes influenced measured outcomes.¹² Without pharmacokinetic measures, it is uncertain if peak plasma concentrations and time to peak plasma concentration reached significant levels to produce the hypothesized effects, particularly when considering no data were collected for factors such as food intake prior to taking the treatment in addition to individual pharmacokinetic variations that could affect treatment outcomes.¹¹ Moreover, little is known about CBD-related tachyphylaxis in sleep medicine and its potential role in modulating longer-term treatment efficacy and tolerance. CBD tolerance has not been noted in humans for epilepsy³⁹ or in the medicinal use of combined THC (tetrahydrocannabinol)/CBD treatments for insomnia⁷ and neuropathic pain,⁴⁰ with sleep improvements occurring in 2⁷ and 4 weeks,⁴⁰ respectively. Preclinical evidence additionally suggested that CBD may induce THC antinociception tolerance.⁴¹ Therefore, with conflicting preclinical and clinical evidence, tachyphylaxis cannot be ruled out at present and more research is necessary to understand its mechanisms and occurrence in long-term, sustained medicinal use of CBD alone for insomnia.⁴² Given the additional variations in dosing, treatment periods, and administration routes and among broader trial conditions reported to peripherally increase anxiety levels,⁴³ future investigations should prioritize a comprehensive exploration of these factors for sleep and mood outcomes.

Compared with past studies, a strength of this trial was the use of daily actigraphy measures in addition to daily and weekly self-reported sleep measures. Past studies exploring CBD-only treatments have largely been limited to self-reported sleep questionnaires measured either weekly or monthly in both healthy or clinical populations.⁶ While studies are increasingly utilizing objective measures, including polysomnography and actigraphy in sleep conditions, it has largely been to explore the efficacy of combined cannabinoid treatments only.^7,19,44 Notably, there is limited exploration for CBD as a standalone treatment, specifically in the context of clinically diagnosed primary insomnia.⁵ Furthermore, the design of the current study was strengthened by rigorous screening and weekly eligibility procedures, successful detection and removal of placebo responders and treatment noncompliers after the placebo run-in period, and ongoing treatment checks through bottle-weight measurements at weekly visits and daily self-reported treatment logs. The design did not measure changes to other peripheral factors known to influence sleep, such as sleep hygiene, which consists of comfortable sleeping environments, exercising earlier, and avoiding phone or television use prior to bedtime.⁴⁵ Unassessed changes to sleep hygiene or similar factors could have served as uncontrolled variables further influencing outcomes.^46,47 Moreover, LME post hoc analyses and Bonferroni corrections were used to mitigate the inflation of type I errors due to the multitude of comparisons of outcome measures. Nonetheless, we acknowledge that the strength of these results may have been impacted due to the number of comparisons explored.⁴⁸ Future trials are therefore urged to take these into consideration when measuring and interpreting sleep results, in addition to addressing the gaps in pharmacokinetics previously discussed while exploring differences in administration routes, longer treatment periods within larger samples, and including rigorous prespecified subgroup analyses to distinguish the high heterogeneity within primary insomnia and how CBD may affect these.

The anti-inflammatory, neuroprotective, and anxiolytic effects of CBD have been explored in numerous intervention studies⁴⁹; however, additional research is needed to explicitly outline mechanistic pathways relevant to sleep. Its low toxicity and broad pharmacological and neurological action are proposed as an ideal endocannabinoid system supplement, primarily enhancing its functioning to regulate homeostasis resulting in therapeutic benefits.⁴⁹ As yet, consistent linkage between specific treatments and positive symptom outcomes is not evident. Moreover, its specific mechanisms of action in sleep are unclear, with studies suggesting it plays a role in regulating hypothalamic-pituitary-adrenal axis hyperarousal and increased melatonin production.^44,49 Thus, the exploration of its therapeutic potential within and beyond populations with primary insomnia is highly recommended as a standalone treatment and in combination with cannabinoids including delta-9-tetrahydrocannabinol and cannabinol, which may act synergistically to enhance therapeutic benefits.^7,29

In summary, apart from well-being, behavior following wakefulness and objective SE outcomes, daily supplementation of 150 mg of CBD had similar beneficial therapeutic effects on sleep and mood for individuals with moderate–severe primary insomnia as placebo. Previous evidence indicated that benefits may be improved with higher doses and longer treatment periods; however, additional high-quality research is necessary to confirm this. Future studies are recommended to compare combined cannabinoid formulations and dosing ranges for potential synergistic effects between cannabinoids, as well as to establish effective dosing guidelines for beneficial sleep and mood effects within primary insomnia and its subgroups.

DISCLOSURE STATEMENT

All listed authors have seen and approved this manuscript. Work for this study was performed at Swinburne University of Technology, Melbourne, Australia. This study was supported by Cannvalate and the Barbara Dicker Brain Sciences Foundation. The sponsors provided funding for the study only. The study treatments were provided by Brains Bioceuticals. A.C.H. is currently supported by a Rebecca L. Cooper Al and Val Rosenstrauss Fellowship (GNT: F2021894) and was previously supported by a National Health and Medical Research Council (NHMRC) Early Career Fellowship (GNT: 1119960) during the conduct of this study. The other authors report no conflicts of interest.

ABBREVIATIONS

CBD

cannabidiol

CI

confidence interval

GSES

Glasgow Sleep Effort Scale

ISI

Insomnia Severity Index

LME

linear mixed effects

LSEQ

Leeds Sleep Evaluation Questionnaire

SD

standard deviation

SE

sleep efficiency

SOL

sleep-onset latency

STAI-T

State-Trait Anxiety Index (Trait anxiety only)

TST

total sleep time

WASO

wake after sleep onset

WHO-5

World Health Organization Well-Being Index-5