**4. Discussion**

Interest in cannabis preparations as therapeutic agents in neuropsychiatric disorders is growing in both the scientific and lay communities [42,43]. This interest is particularly strong in disorders with substantial unmet needs, such as refractory epilepsy and pediatric ASD, which lacks medications that target its core symptoms [44]. Currently, robust evidence exists only for Epidiolex, a plant-derived pure CBD isolate, to treat the following specific types of refractory epilepsy: Dravet syndrome [45], Lennox–Gastaut syndrome [46], and tuberous sclerosis complex [47]. Nevertheless, full-spectrum extracts of various cannabis strains and synthetic cannabinoids are being widely used to treat adults living with chronic pain [48–50], chemotherapy-induced nausea and vomiting [51,52], sleep disorders [26], depression, anxiety, psychosis [53], PTSD [54], and to treat children with various types of refractory epilepsy [55], and irritability associated with autism spectrum disorder (ASD) [22,24,25,56].

The endocannabinoid system is involved in the pathophysiology of both sleep disorders [9] and ASD [19], which might contribute to the high incidence of sleep disturbances in people with ASD. These associations make the endocannabinoid system an attractive target for the treatment of sleep disturbances in ASD.

There is much anectodical evidence, as well as several reports of uncontrolled case series, suggesting an improvement in sleep disturbances following treatment with various strains of medical cannabis [24,25,57,58]. However, placebo-controlled studies have not been published so far.

In this randomized, placebo-controlled trial we used the following two CBD-rich preparations: a full-spectrum (whole-plant) extract, and purified CBD and THC isolates (pure cannabinoids). This is particularly relevant for sleep disturbances as it is commonly believed that, in addition to the main cannabinoids, other components of the cannabis plant, such as terpenes and flavonoids, also assist in alleviating sleep disturbances (an entourage effect) [59–61].

In our cohort, the effect of these two cannabinoid preparations at a dose of ~5.5 mg CBD and ~0.3 mg THC per kg, per day, was not superior to the placebo in all aspects of sleep measured by the CSHQ.

These findings are in line with a recent report on the negative acute effect of CBD on the sleep–wake cycle of healthy adults in a placebo-controlled study [62]. Of note, while a recent meta-analysis of randomized clinical trials demonstrated that cannabinoids provided a small benefit for impaired sleep, the studies that were analyzed were mainly in adults living with chronic pain, who used THC-rich preparations [26].

Accordingly, future studies of cannabinoid treatment for sleep disorders should consider using a more balanced CBD to THC ratio. Indeed, recreational cannabis strains that contain high THC and low CBD concentrations were associated with serious adverse events when used during youth, including decreased motivation [63–65], addiction [66], mild cognitive decline [64,67–69], and schizophrenia [64,70–72]. However, all of these risks are higher in cannabis strains with a high ratio of THC to CBD [73], than in the more balanced strains.

Consistent with previous studies [39,74], we also found that the severity of sleep disturbances, as indicated by the CSHQ total score at the baseline, correlated with a younger age and with the severity of the autistic core symptoms. Most of the participants in our cohort had severe autistic symptoms (78.7% had a comparison score of 8–10 in the ADOS-2). This might explain the higher rate of participants (86%) who screened positive for sleep disturbances, compared to a rate of approximately 70% that was reported in a US registry study [74] and in a Chinese multicenter survey [39].

Notably, we found a longitudinal association between changes in the CSHQ total score (in participants who received either cannabinoids or the placebo) and changes in the disruptive behavior and the severity of the core symptoms, suggesting the possibility of a cause-and-effect relationship (better sleep leads to lower symptoms). These findings are congruen<sup>t</sup> with associations between sleep quality, behavior, and the severity of autistic core symptoms that have been reported in cross-sectional studies [39,75]. This finding also underscores the importance of sleep quality in children with ASD, the impact of which may exceed that of their typically developing peers.

Our study had several limitations: The study was designed as a cross-over study, which allows within-participant analyses, comparing the two treatments that each participant received. However, a treatment order effect (all treatments were more effective in the first period, probably due to a greater initial placebo effect) made this analysis less accurate. Therefore, we reported the more accurate between-subject analyses of the first treatment period (Table 2). We also present, in the Supplemental Information, the within-participant analyses (Table S3) and the between-participant analyses of Period 2 (Table S3), which yielded similar negative results. Another limitation of this study was the use of a caregiver's report for assessing sleep quality, without more objective measures of sleep such as actigraphy and sleep logs. Additionally, our study was not powered to detect the effects of age, the level of function, and other baseline characteristics on the treatment response.
