Among the 1,355 children with ASD in the ANCAN database (November 2021), there an indication for melatonin recommendation or use for only 107 (8%) children (Figure S1, available online). A comparison of the sociodemographic and clinical characteristics between children with ASD who were treated with melatonin and those who were not is presented in Table 1. Children who were given melatonin were diagnosed, on average, six months earlier than children who were not (35.1±15.3 months vs. 41.4±19.3 respectively; p-value = 0.002). Children who were treated with melatonin had more severe ASD symptoms, according to both DSM-5 criteria and the ADOS-2 comparison score, and they also had lower cognitive and adaptive behavior abilities (Table 1). As expected, children who were treated with melatonin had more severe sleep disturbances, according to the CSHQ total and subscales scores, compared to their counterparts who were not.

Table 1 Comparison of Sociodemographic and Clinical Characteristics between Children with ASD with and without a Melatonin Recommendation Characteristics of the melatonin-treatment cohort

Overall, the parents of 78 children (73%) for whom there was an indication for melatonin recommendation or use completed our phone survey and reported adherence to melatonin treatment for at least one week. Notably, the median adherence time for those children exceeded 88 months (Fig. 1), with 26 children (33.5%) discontinuing the treatment for various reasons, including lack of effectiveness (14.1%), natural improvement (8.9%), loss of effectiveness (7.7%), side effects (3.8%), starting other medications (3.8%), child’s refusal (2.6%), doctor’s recommendation (2.6%), medication cost (1.3%), or starting a behavioral sleep intervention (1.3%) (Table S1, available online).

Fig. 1

Kaplan-Meier Curve of Melatonin Treatment Adherence over Time. The cumulative adherence probability to melatonin treatment over time is depicted by a continuous black line with 95% confidence intervals in black dashed lines

The sociodemographic and clinical characteristics of the children who were treated with melatonin, and the association of these characteristics with treatment discontinuation, are presented in Table 2. None of these characteristics was significantly associated with treatment discontinuation, although the use of other psychoactive drugs concurrently with melatonin treatment was marginally associated with a lower probability of treatment discontinuation (HR = 0.452, 95%CI = 0.194–1.052, respectively). Characteristics associated with melatonin treatment and their effect on treatment discontinuation are also presented in Table 2. The age at treatment initiation ranged between 1.6 and 9.6 years, and the mean final melatonin dosage was 3.9 (±2.7) mg. Most of the parents (77%) gave the treatment every day before bedtime, while the others gave melatonin only on school days, or when the child had serious sleep onset difficulties. Interestingly, the parents of 14 children (18%) purchased an imported over-the-counter (OTC) formulation of melatonin, without prescription. The parents of 11 children (14%) reported mild side effects of melatonin treatment, including crying, irritability/hyperactivity, morning drowsiness, rash, vomiting, fever, increased appetite, and abdominal pain (Table S1, available online). None of the treatment characteristics or its side effects was significantly associated with treatment discontinuation.

Table 2 Characteristics Associated with Discontinuation of Melatonin Treatment Effect of melatonin treatment

The parents of 70 children (90%) reported that melatonin improved their children’s sleep, with 86%, 53%, and 45% reporting that melatonin treatment influenced the sleep onset, sleep duration, and night awakenings, respectively (Fig. 2A). In addition, the parents of 27 children (35%) reported that melatonin had an additional effect on their children’s daytime behavior, with better educational functioning, improved moods, reduction in tantrums, better communication abilities and better sensory regulation being reported in 28%, 21%, 10%, 9%, and 6% of children, respectively (Fig. 2B). Of note, a moderate and statistically significant concordance was seen between the effect of melatonin treatment on different sleep and daytime behaviors, with the most significant associations being between sleep duration and night awakening (kappa = 0.459; p-value < 0.001), and between tantrum reduction and mood improvement (kappa = 0.571; p-value < 0.001) (Fig. 2C).

Fig. 2

Effect of Melatonin on Sleep and Daytime Behaviors of Children with ASD. (A) & (B) Rates of children whose parents reported an effect of melatonin treatment on their sleep parameters (A) and daytime behaviors (B). (C) Kohen’s kappa coefficients for the concordance between different effects of melatonin treatment

Finally, both univariable and multivariable Cox regression models were used to assess the association between the reported effect of melatonin treatment on sleep and daytime behavior and treatment adherence (Table 3). Improvements in nighttime awakening, sleep duration and educational functioning were significantly associated with a decreased likelihood to discontinue melatonin treatment (HR = 0.208, 95%CI = 0.071–0.610; HR = 0.376, 95%CI = 0.161–0.878; and HR = 0.271, 95%CI = 0.080–0.91, respectively) in the univariate analysis. However, because of the relationship between the effects of melatonin treatment on sleep duration and night awakenings (see Fig. 2C), only the effects of night awakenings and educational functioning remained significantly associated with treatment adherence in the multivariate analysis (aHR = 0.142, 95%CI = 0.036–0.565; and aHR = 0.195, 95%CI = 0.047–0.806, respectively).

Table 3 Characteristics of Treatment Effectiveness and its Effect on Adherence