Our study is a prospective investigation of CBD use in a real‐world setting at a tertiary center for rare epilepsy. Several key points can be highlighted. First, over 50% of CBD use in our center was for off‐label conditions. Second, CBD demonstrated good tolerability, with few major side effects. Third, its efficacy, particularly in reducing seizures and improving quality of life, may take up to 6 months to become evident. Lastly, the 18‐month retention rate was 56%, with higher rates observed in older patients, males, and those using CBD for off‐label conditions.

Off‐label use of CBD in children with drug‐resistant epilepsy is well‐documented, with real‐world data indicating usage rates between 13% and 66% [21, 22, 23, 24, 25]. Our findings place us at the higher end of this range. While randomized controlled trials (RCTs) remain the gold standard, observational studies provide invaluable insights into treatment effectiveness in clinical practice, particularly for identifying populations, such as those with SYNGAP1, who may benefit from CBD and are unlikely to be included in RCT due to recruitment challenges [26]. Moreover, our real‐world data confirm that individuals who exhibit a good response in seizure frequency reduction often also show improvements in non‐seizure outcomes, such as emotional and behavioral functioning [20]. We hypothesize that better seizure control contributes to improvements in the epileptic components of DEE. Understanding the extent of off‐label use in these populations can help refine clinical practice and guide treatment decisions for individuals with limited therapeutic options.

Given the known pharmacokinetic interaction between CBD and clobazam, CBD increasing the levels of Nor‐CLB [27, 28, 29], we performed stratified analyses to evaluate the potential impact of clobazam co‐medication. Although nearly half of the cohort received clobazam, caregiver‐reported outcomes showed no statistically significant difference after Bonferroni correction. Our findings suggest that clobazam co‐medication may not substantially influence the perceived clinical benefit of CBD in this cohort, though further studies with larger samples and pharmacokinetic evaluation are needed. In addition, therapeutic drug monitoring (TDM) was not implemented in this study. While logistically difficult in routine practice, TDM may help better define dose–response relationships and reduce interindividual variability, particularly in patients receiving polytherapy, especially with clobazam, valproate, stiripentol, or fenfluramine. Recent studies highlighted substantial interindividual variability in plasma concentrations of CBD and its major metabolites, particularly 7‐OH‐CBD and 7‐COOH‐CBD [30, 31]. This may underscore the role of close clinical monitoring and the integration of TDM to support individualized CBD treatment strategies and improve therapeutic precision.

In our study, we observed that improvement, mainly of non‐seizure outcomes, could emerge up to 6 months after treatment initiation. This medium‐term window may be crucial for accurately assessing CBD's efficiency. Discontinuing treatment before 6 months may deprive highly pharmacoresistant individuals, who have already undergone multiple ASM trials, of a potential therapeutic benefit. Indeed, some individuals exhibited consistent benefits between 2 and 6 months after CBD initiation. This delayed response underscores the need for patients' and families' education and careful evaluation of treatment impact and highlights the importance of longer‐term assessments to fully capture the therapeutic potential of CBD, particularly in complex and pharmacoresistant epilepsies. Our study extends the evaluation of the efficiency beyond the 14 weeks of the RCTs upon which the drug has been approved [4, 5, 8, 12, 32]. Furthermore, our cohort experienced good tolerability overall, with a few major side effects. The most reported adverse effects included changes in appetite, behavioral disturbances, and, to a lesser extent, sleep issues. Although sleep disturbances are less documented, they can significantly impact quality of life and should not be overlooked in clinical assessments. Importantly, we observed no significant hepatic adverse events both biologically (liver enzyme activity) and clinically, indicating that CBD can be administered safely. The median dose in our population was 14 mg/kg/day, lower than the target dose of 20 mg/kg/day in LGS and DS and 25 mg/kg/day in TSC [12, 21, 33].

Importantly, clinicians cannot accurately evaluate the efficacy and safety of antiseizure medications (ASMs) on both seizure‐related and non‐seizure outcomes without the active involvement of caregivers. Engaging parents in the evaluation process fosters a meaningful partnership that is essential for assessing children with DEEs, particularly in areas where standardized scoring systems may be unavailable, difficult to implement, or time‐consuming. It is important to note that seizures and non‐seizure symptoms often fluctuate significantly, and caregivers' observations provide valuable insights into these variations, which may not always be captured during the limited time of the clinical visits. Additionally, clinical assessments in this population are often compromised by the challenges of testing outside the child's usual environment, where difficulties can be exaggerated.

Caregiver‐reported questionnaires can be as informative as clinical assessments, providing critical data that reflect the child's daily experiences. However, this raises the question of which treatment aspects should be prioritized for improvement; caregivers' impressions alone are insufficient. This underscores the need for collaborative work with families to establish patient‐centered treatment goals. Relying solely on broad assessment scales may lack the granularity needed for meaningful comparisons within patient groups. Training parents in structured evaluation methods can help minimize subjectivity and improve the reliability of caregivers‐reported data. By integrating parental input into the evaluation process, clinicians can gain a more comprehensive understanding of a child's needs, allowing tailored treatment approaches that ultimately enhance care quality and outcomes for children with drug‐resistant epilepsy. In future studies, caregiver‐reported impressions could be usefully complemented by validated structured tools such as the Aberrant Behavior Checklist or the PedsQL Epilepsy Module, which require less than 15 min to complete [34]. These instruments would improve the objectivity of reported outcomes while remaining feasible for families and would facilitate comparisons across studies and populations.

Our 18‐month retention rate of 56% aligns with findings from the open‐label extension program and other real‐world data cohorts [22, 23, 24]. This rate reflects the balance between CBD treatment efficacy and safety. However, unlike previous studies that did not investigate factors affecting retention, we identified that off‐label use, male gender, and older age at treatment initiation may be positively associated with higher retention rates. We were particularly surprised by the effect of gender, as we did not expect this outcome in our population. We investigated potential cofactors to explain this difference but found no significant variations in terms of the number of treatments, dosages, the presence or absence of clobazam, or etiology. In the study by Georgieva et al., females made up 64% of the discontinuation group, while only 49% of those who continued CBD were female, mirroring our findings [24]. This gender difference is further supported by preclinical studies in healthy volunteers, which have shown that females exhibit greater relative exposure to CBD metabolites in plasma over time, as well as gender‐specific variations in cannabinoid pharmacokinetics. Specifically, females demonstrated higher Cmax and AUC values than males. Similar trends were observed in preclinical studies, where females had a 2.25‐fold higher Cmax and a 1.97‐fold higher AUC for 7‐COOH‐CBD after normalizing for body weight by Day 4 of administration [35, 36]. Female subjects had a normalized Cmax of 39.81 ng/mL/kg and AUC of 217.70 ng/mL.h/kg, nearly double that of males. Additionally, a significant correlation between 7‐OH‐CBD AUC and body weight was observed in females. Authors have suggested that sex differences in UGT activity, such as slower UGT2B15 metabolism in females, may contribute to these pharmacokinetic differences [37]. Our study is among the first to highlight a gender effect in the long‐term retention of CBD treatment. While the mechanisms remain uncertain, emerging pharmacokinetic data suggest that females may have greater exposure to active metabolites, possibly affecting tolerability. This observation, consistent with previous real‐world and preclinical findings, adds to the ongoing debate regarding sex differences in CBD efficacy and metabolism [38]. Further sex‐specific studies are needed to clarify these differences and support more personalized approaches to CBD therapy. Such research should aim to elucidate the pharmacokinetic and pharmacodynamic mechanisms underlying sex‐related variability, assess their impact on both efficacy and tolerability, and determine whether dose adjustments or monitoring strategies should differ between males and females in clinical practice.

This study, although prospective and including a relatively large number of patients, has some limitations. First, as it was conducted in a single tertiary center, the findings may not be fully generalizable to other settings with different patient populations. Additionally, the significant proportion of off‐label CBD use introduces variability in treatment protocols, which could influence retention rates and outcomes. While caregiver‐reported outcomes are valuable, they may introduce some subjectivity, particularly in assessing non‐seizure outcomes. Lastly, the absence of a control group limits direct comparison of CBD effectiveness to other treatments.