Coppola 2004.

Methods	Randomized, double‐blind, cross‐over, placebo‐controlled trial
Participants	Inclusion criteria: mental retardation with/without epileptic seizures; aged > 12 months; diagnosis of sleep disorder, defined according to DSM‐IV criteria as the circadian rhythm sleep disorder; exclusion of medical issues such as gastro‐oesophageal reflux, pain, or epileptic seizures mimicking sleep disorders; persisting sleep disturbances despite maintaining appropriate sleep hygiene; informed consent by parents, carers, or both Exclusion criteria: progressive neurological, systemic, or both, diseases; aged < 12 months; poor compliance from parents/carers with the study requirements before trial entry N: 32 (7 lost to follow‐up); 25 completed the trial M: 16; F: 9 Age: 3.6‐26 years (mean age 10.5 years) 18/25 participants had epilepsy and on concurrent AEDs Type of seizures: complex partial (8), with secondary generalization (5), tonic‐clonic (4), tonic (3), drop‐attacks (2), atypical absences (1), myoclonic seizures (2) Type of epilepsy: partial epilepsy (9); generalized symptomatic (5) or cryptogenic (1) epilepsy; multifocal epileptic encephalopathy (3). A genetic syndrome was diagnosed in 5 participants (20%); they were the following: Angelman syndrome; Saethre‐Chotzen syndrome; 11p13 microdeletion; Leber amaurosis; CHARGE syndrome Concurrent AED treatment: monotherapy (11), bi‐therapy (2;), and tri‐therapy (5) Seizure frequency: seizure‐free (7); sporadic (3); 1‐3/month (3); > 1/week (2); > 1/day (3) Duration of the trial (including follow‐up): 9 weeks; phase 1 (melatonin or placebo): 4 weeks, cross‐over period: 1 week, phase 2 (placebo or melatonin): 2 weeks.
Interventions	Melatonin (fast release) compared with placebo (see text for more details)
Outcomes	Seizure freedom/seizure frequency: out of the 11 seizure‐free participants before starting the study, 9 remained unchanged on melatonin; in the other 2, seizures re‐appeared after 1 month (1 Lennox‐Gastaut syndrome; 1 partial seizures). Soon after discontinuing melatonin, seizures stopped in both these participants. Among the 7 participants with uncontrolled seizures during the baseline phase, 1 became seizure‐free (1 secondarily generalized partial seizure), 2 partially improved (1 partial seizure; 1 myoclonic seizure), and the other 2 were unchanged; the remaining 2 showed a seizure worsening 1 month after starting melatonin phase (1 cryptogenic generalized seizure; 1 secondary generalized partial seizure). In the latter participants, seizures decreased soon after melatonin withdrawal Adverse events: none reported Quality of life: not systematically evaluated. Authors stated that "half the parents/caregivers referred improved behavior and alertness in children who appeared more quiet and better disposed to rehabilitation treatment; familial environment improved concomitantly, as a consequence of a better quality of night time"
Notes	Not all participants had epilepsy Seizure diaries were used to monitor the frequency and type of seizures Reasons for lost to follow‐up (7 participants) were reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated permutation of code numbers for the treatment groups (information provided directly from the principal Investigator)
Allocation concealment (selection bias)	Unclear risk	Method of concealment not described
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Blinding not explicitly reported (insufficient information to permit judgement)
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Reported missing to follow‐up. No intention‐to‐treat analysis
Selective reporting (reporting bias)	High risk	Authors stated that seizures occurred, without further specifying the exact number of seizures. Adverse events not systematically evaluated
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding not explicitly reported (insufficient information to permit judgement)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding not explicitly reported (insufficient information to permit judgement)