The use of MElatonin in children with Neurodevelopmental Disorders and impaired Sleep: a randomised, double-blind, placebo-controlled, parallel study (MENDS)

Background

Difficulties in initiating and maintaining sleep are common in children with neurodevelopmental disorders. Melatonin is unlicensed in children yet widely prescribed for sleep problems.

Objective

To determine whether or not immediate-release melatonin is beneficial compared with placebo in improving total duration of night-time sleep in children with neurodevelopmental problems.

Design

Randomised, double-blind, placebo-controlled, parallel study.

Setting

Hospitals throughout England and Wales recruited patients referred by community paediatricians and other clinical colleagues.

Participants

Children with neurodevelopmental problems aged from 3 years to 15 years 8 months who did not fall asleep within 1 hour of lights out or who had < 6 hours of continuous sleep. Before randomisation, patients meeting eligibility criteria entered a 4- to 6-week behaviour therapy period in which a behaviour therapy advice booklet was provided. Sleep was measured using sleep diaries and actigraphy. After this period the sleep diaries were reviewed to determine if the sleep problem fulfilled the eligibility criteria. Eligible participants were randomised and followed for 12 weeks.

Interventions

Melatonin or placebo capsules in doses of 0.5 mg, 2 mg, 6 mg and 12 mg for a period of 12 weeks. The starting dose was 0.5 mg and the dose could be escalated through 2 mg and 6 mg to 12 mg during the first 4 weeks, at the end of which the child was maintained on that dose.

Main outcome measures

The primary outcome was total night-time sleep time (TST) calculated using sleep diaries at 12 weeks compared with baseline. Secondary outcome measures included TST calculated using actigraphy data, sleep-onset latency (SOL) (time taken to fall asleep), sleep efficiency, Composite Sleep Disturbance Index score, global measure of child’s sleep quality, Aberrant Behaviour Checklist, Family Impact Module of the Pediatric Quality of Life Inventory (PedsQL^™), the Epworth Sleepiness Scale, number and severity of seizures and adverse events. Salivary melatonin concentrations and association of genetic variants with abnormal melatonin production were also investigated.

Results

A total of 275 children were screened to enter the trial; 263 (96%) children were registered and completed the 4- to 6-week behaviour therapy period and 146 (56%) children were randomised, of whom 110 (75%) contributed data for the primary outcome. The difference in TST time between the melatonin and placebo groups adjusted for baseline was 22.43 minutes [95% confidence interval (CI) 0.52 to 44.34 minutes; p = 0.04] measured using sleep diaries. A reduction in SOL, adjusted for baseline, was seen for melatonin compared with placebo when measured by sleep diaries (–37.49 minutes, 95% CI –55.27 to –19.71 minutes; p < 0.0001) and actigraphy (–45.34 minutes, 95% CI –68.75 to –21.93 minutes; p = 0.0003). There were no significant differences between the two groups in terms of the reporting of adverse events. The results of other secondary outcomes favoured melatonin but were not statistically significant.

Conclusions

On average, the children treated with melatonin slept 23 minutes longer than those in the placebo group; however, the upper limit of the confidence interval was less than 1 hour, the minimum clinically worthwhile difference specified at the outset of the trial. Melatonin is effective in reducing SOL in children with neurodevelopmental delay by a mean of 45 minutes; a value of 30 minutes was specified a priori to be clinically important. Future studies should be conducted over longer periods and directly compare different formulations of melatonin with conventional hypnotic and sedative medications. It would also be important to study groups of children with specific neurological disorders.

Trial registration

Current Controlled Trials ISRCTN05534585.

Funding

This project was funded by the NIHR Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 16, No. 40. See the HTA programme website for further project information.

Potential risks

Clinical studies in humans (adult volunteers and patients of both sexes and all ages) have not shown any consistent or serious short- or long-term adverse side effects. 35 Most of the reported adverse side effects have been described in very small numbers of patients. 36,37 Although the chronic use of exogenous melatonin for sleep problems in paediatrics appears widespread, there is a paucity of data on its safety. Melatonin is widely distributed at different densities throughout the body and appears to be implicated in various physiological functions other than sleep. There are therefore theoretical risks to the chronic administration of exogenous melatonin to children, and particularly to children with a range of neurological problems, including epilepsy and behavioural problems. The most significant theoretical risks in this population are related to:

• sexual development

• nocturnal asthma

• growth

• seizures.

With age, nocturnal melatonin levels appear to decrease with the most striking falls occurring around puberty. Nocturnal melatonin levels have been assessed in children at various pubertal stages and it is observed that they are higher in the earlier than in the later stages. 38 Whether this is cause or effect is not known but there is a potential risk that exogenous melatonin may delay sexual maturity.

Elevated endogenous melatonin levels have been associated with an increased incidence of nocturnal asthma,39 although there is at least one study in adults that demonstrated an improvement in sleep in adults with asthma following administration of 3 mg of melatonin with no apparent worsening of their asthma symptoms. 40

Melatonin has been observed to have a direct effect on growth hormone. 41 Eight male volunteers received single doses of 0.05 mg, 0.5 mg and 5 mg of melatonin or placebo with serum growth hormone levels measured for up to 150 minutes afterwards. Compared with placebo, growth hormone levels were found to increase for doses of 0.5 mg and 5 mg of melatonin. The exact mechanism is not clear and the effect of increases in growth hormone of this magnitude on longitudinal bone growth in children is not known.

One study has suggested that seizure control may deteriorate in some children with epilepsy42 but this observation has not been confirmed in a number of anecdotal8,17 and limited randomised controlled trials (RCTs). 25–27 There is some anecdotal evidence that seizure control or seizure severity may actually improve as a secondary effect of improved sleep and increased seizure threshold. 8 There have been two spontaneous reports to the Medicines and Healthcare products Regulatory Agency of seizures associated with exogenous melatonin and responders to the survey by Waldron et al. 37 reported an increase in seizure activity or new-onset seizures.

Melatonin oral capsules contain melatonin, lactose and magnesium stearate. Placebo oral capsules contain lactose and magnesium stearate. Individuals with lactose intolerance are able to consume significant quantities of dairy products without manifesting any symptoms of lactose malabsorption and therefore individuals with lactose intolerance are able to consume capsules without adverse effects and were eligible for inclusion into the study.

Potential benefits

There are very few meta-analyses of RCTs of melatonin. 35,36 Those reported have indicated that exogenous melatonin may improve sleep in a number of clinical situations, including:

• children with autism and intellectual disability

• patients with visual impairment [particularly when the visual impairment is due to an abnormality within the anterior visual pathway (specifically in patients with micro-ophthalmia or anophthalmia) rather than cortical visual impairment]

• elderly patients with insomnia.

Reported benefits include a reduced sleep latency time (i.e. reduced time to fall asleep), reduced number of awakenings throughout the night (i.e. increased periods of continuous, uninterrupted sleep throughout the night) and improved behaviour and performance during the day. Seizure frequency and seizure control may also improve, probably as a secondary or indirect effect of improved quality of sleep. 8,43

It is important to emphasise that all of the reported studies show a marked heterogeneity in inclusion and exclusion criteria, the types and causes of impaired sleep in the populations studied, the doses and formulations of melatonin used, the methods of assessment and the reported outcomes. Although the current study includes a heterogeneous group of children with neurodevelopmental delay, all were treated according to a strict protocol and within a dose-escalation framework. Finally, the patient population in this RCT is almost as large as that of the combined studies that were assessed in a recent meta-analysis. 23

Inclusion criteria

• Children aged from 3 years to 15 years and 8 months at screening (the age at screening was set to ensure that all those enrolled in the study were minors, because of the implications for consent in incapacitated adults).

• Diagnosis of a neurodevelopmental disorder by a community paediatrician, paediatric neurologist or paediatric neurodisability consultant, categorised as:

  1. – developmental delay alone

  2. – developmental delay and epilepsy

  3. – developmental delay and autistic spectrum disorder (ASD) (in coding the presence of epilepsy and ASD diagnoses we required sight of documentation from relevant services which demonstrated that appropriate diagnostic assessments and investigations have been used)

  4. – developmental delay with ‘other’ (‘other’ is defined as the child having a specific genetic/chromosomal disorder)

  5. – any combination of the above.

• Adaptive Behaviour Assessment System (ABAS) questionnaire score with a percentile rank < 7.

• Minimum 5 months’ history of impaired sleep at screening as defined by:

  1. – not falling asleep within 1 hour of ‘lights off’ or ‘snuggling down to sleep’ at age-appropriate times for the child [this was the child’s usual bedtime (recorded in the sleep diary) based upon the family’s normal routine; ‘age appropriate’ was defined as a sensible target sleep onset time earlier than 20:30 for children at age 6 years and 15 minutes later per year for older children44] in three nights out of five and/or

  2. – < 6 hours of continuous sleep in three nights out of five .

• Children whose parents were likely to be able to use the actigraph and complete sleep diaries.

• Children who were able to comply with taking the study drug.

• English speaking.

• Children whose parents had completed sleep diaries for an average of 5 out of 7 nights at baseline (T0W).

Exclusion criteria

• Children treated with melatonin within 5 months of screening (T–4W).

• Children who had been taking the following medication for < 2 months:

  1. – any benzodiazepines

  2. – amisulpride (Solian^®, Sanofi-Aventis)

  3. – chlorpromazine (Largactil^®, Sanofi-Aventis)

  4. – haloperidol (Haldol^®, Janssen)

  5. – olanzapine (Zyprexa^®, Lilly)

  6. – risperidone (Risperdal^®, Janssen)

  7. – sertindole (Serdolect^®, Lundbeck)

  8. – sulpiride (Sulpor^®, Rosemont)

  9. – thioridazine (Melleril^®, Novartis)

  10. – trifluoperazine (Stelazine^®, Goldshield).

• Current use of beta-blockers (minimum of 7 days’ washout required).

• Current use of sedative or hypnotic drugs, including chloral hydrate, triclofos, and alimemazine tartrate (Vallergan^®, Sanofi-Aventis) (minimum of 14 days’ washout required).

• Children with a known allergy to melatonin.

• Regular consumption of alcohol (more than three times per week).

• Children for whom there are suggestive symptoms of obstructive sleep apnoea syndrome (OSAS) (including combinations of snoring, gasping, excessive sweating or stopping breathing during sleep), physical signs supportive of OSAS (such as very large tonsils/very small chin) or results of investigations suggesting OSAS (such as overnight pulse oximetry or polysomnography), for which the child should be referred to appropriate respiratory or ear, nose and throat colleagues for specific assessment and treatment.

• Girls or young women who were pregnant at the time of screening (T–4W).

• Currently participating in a conflicting clinical study or participation in a clinical study involving a medicinal product within the last 3 months.

Sleep diaries

Each week parents were asked to complete a two-sided A4 sleep diary that covered a period of 7 days, with one column per day (see Appendix 2). Parents recorded the time that their child went to bed, fell asleep and woke up the next morning. They also recorded any daytime naps, night-time awakenings and the time and duration of any actigraphy removal. Sleep diaries were completed continuously between T–4W and review at T0W, and also continuously throughout the study until study completion (T+12W). A sleep diary records the parental observation and perception of the child’s sleep. Parents were not required to differentiate between periods when the child was actually asleep and periods when the child was awake but quiet (i.e. not disturbing the rest of the household). Consequently, parents did not have to stay awake to complete the sleep diary and were not requested or expected to repeatedly check their child throughout the night.

Sleep diaries were used to calculate TST, SOL (the time taken to fall asleep) and daily global sleep quality in a subjective manner from the parents’ perceptions.

Actigraphy

Actigraphy is the use of accelerometers to measure human movement. The actigraph is worn on the wrist and the movement of the wrist is monitored continuously whilst it is being worn. The actigraph is very lightweight and can be used on individuals of all ages for long periods of time. Wrist movement data are stored within the unit and processed using software programmes to give an indication of the activity levels of the wearer. Analysis of frequency and pattern of movement by means of validated algorithms permits detection of basic sleep–wake patterns. 48

The actigraph was worn continuously day and night for the behaviour therapy phase and for the final week of the study period. The actigraph could be removed or worn during bathing or showering. It could be worn on either wrist but it was emphasised that the same wrist should be used throughout the study. The actigraph used in this study was the MicroMini-Motionlogger^®, supplied by Ambulatory Monitoring Inc.

The actigraph measures and stores data on movements. Frequency of movements above a preset threshold are scored in 1-minute epochs; all epochs that are scored above a preset threshold (sensitivity level) are scored as ‘wake’ and those that are below this threshold are scored as ‘sleep’. The threshold is not set on an individual basis.

In line with existing guidance,49 interpretation of the actigraphy data was informed by the sleep diaries. The sleep diaries recorded the child’s ‘snuggle down to sleep’ time, and the start of sleep was determined from the actigraph as the first 10-minute interval after ‘snuggle-down’ time when there was no more than one epoch that was above the threshold (automatically calculated by the software) for determining wakefulness. The software then considered the first minute of this 10-minute period as the time of sleep onset.

Any sleep interruptions were determined from the actigraph by searching for 10-minute intervals in which activity in more than one epoch was above the threshold set automatically for determining wakefulness or ‘wake’. Final wake-up time was recorded by parents in the sleep diary to the nearest minute. Sleep offset was determined to be the last 10-minute period before final wake-up time in which there was no more than one epoch that was above the threshold for determining ‘wake’.

Total night-time sleep was calculated as the sum of all epochs scored as sleep from sleep onset to sleep offset.

As the actigraphy watch defines periods of sleep as periods with little/no activity, it is acknowledged that those periods of restless sleep may be interpreted by the unit as periods of ‘wake’. This may be a particular issue for children with motor problems, including cerebral palsy.

Children’s Sleep Habits Questionnaire

A comprehensive, parent-reported sleep-screening instrument designed for school-age children, the Children’s Sleep Habits Questionnaire (CSHQ)50 yields both a total score and eight subscale scores, reflecting key sleep domains that encompass the major medical and behavioural sleep disorders in this age group. The questionnaire takes 10 minutes to complete. It was undertaken at T–4W and T0W.

Pediatric Quality of Life Inventory Family Impact Module

The PedsQL Family Impact Module51 is designed to measure the impact of paediatric chronic health conditions on parents and the family. It measures parent self-reported physical, emotional, social and cognitive functioning, communication and worry. The module also measures parent-reported family daily activities and family relationships. Scores range between 0 and 100 and higher scores indicate better functioning. The questionnaire takes approximately 5 minutes to complete. It was undertaken at T0W and T+12W.

Epworth Sleepiness Scale

The Epworth Sleepiness Scale (ESS)52 is a simple, self-administered questionnaire that provides a measurement of the caregiver’s general level of daytime sleepiness. Scores range between 0 and 24 and higher scores indicate poorer functioning. The questionnaire takes approximately 3 minutes to complete. It was undertaken by one caregiver and the same caregiver completed the questionnaire at T0W and T+12W.

Aberrant Behaviour Checklist

The Aberrant Behaviour Checklist (ABC)53,54 is an instrument for assessing individual baseline behaviour and for evaluating behavioural change. The ABC contains five subscales and higher scores indicate poorer functioning. The checklist takes approximately 20 minutes to complete. It was undertaken at T0W and T+12W.

Composite Sleep Disturbance Index

A Composite Sleep Disturbance Index (CSDI), based on allocating scores according to the frequency and duration of sleep problems reported by parents in questionnaires, was first used by Richman and Graham55 and has since been used in many other studies, including that by Quine,10 who reported high internal reliability and showed that the measure was sensitive to change. 56 The questionnaire takes 3 minutes to complete and was undertaken at T–4W, T0W and T+12W.

Applying the scoring criteria of Table 2 the CSDI was calculated as follows: settling problems, night waking, early waking (before 0500) and co-sleeping were each assigned a score of 0–2 based upon their reported weekly frequency; settling and night-waking problems were also assigned a score of 0–2 based upon the reported duration of the problem, when it occurred. Total scores were derived by adding the scores assigned for these six items. The scores ranged from 0 to 12 and higher scores indicate greater sleep disturbance.

Salivary melatonin assay

Salivary melatonin levels were measured for each patient at two time points. Saliva samples were collected hourly from 17:00 until the child’s usual bedtime at:

• T–1W, on the night before the randomisation clinic visit

• T+10W, on the night after a dose of trial treatment had been omitted and on which night no trial medication was given (i.e. two doses were missed at the beginning of the eleventh week of study treatment).

This is a very similar methodology to that described by Keijzer et al. 57 in which five evening collections were effective in the majority of cases when evaluating dim-light melatonin onset (DLMO) timing for patients with circadian rhythm disorders. A minimum of 2 ml of saliva was obtained by asking the child to spit into a tube or by placing a saliva sponge in the buccal cavity (cheek pouch) of the child’s mouth (the space between the gums and the inner cheek).

Salivary samples were collected and stored by the parent in a domestic freezer at a maximum temperature of –18°C. Samples were collected by the research practitioner for storage until trial completion, when they were placed in dry ice and transported to the School of Biomedical and Molecular Sciences, University of Surrey, Guildford for blinded analysis. The theoretical basis for these measurements is that they should allow accurate categorisation of which children are physiologically phase delayed at the beginning of the study, which may prove to be an important variable when comparing responders to non-responders in any secondary analysis.

Baseline salivary melatonin levels are at their lowest during the day. During the evening, as light levels decrease, there is a natural rise in melatonin levels (usually between 2000 and 2200 depending on age) that starts to peak around midnight. The time when the melatonin levels naturally start to rise from the low daytime baseline is called the DLMO period. This is measured biochemically as the time when melatonin levels first start to rise by at least two standard deviations above the mean baseline level. An 8-year-old child who has a DLMO of midnight, for example, would be classified as having a delayed sleep phase and, according to recent research, should respond better to exogenous melatonin than another child with a normal DMLO of 2000. 44

If sampling is taken regularly for a 24-hour period the DLMO should be measurable, whenever it occurs. However, for practical reasons, it is common in paediatric populations to take five swabs before bedtime. It is therefore possible to ‘miss’ a DLMO that precedes sampling or occurs when sampling has finished.

Deoxyribonucleic acid analysis

Salivary deoxyribonucleic acid (DNA) was collected from 186 patients [with some samples taken from children who participated only in the behaviour therapy part of the study and not in the interventional (randomised controlled) part of the trial] using the DNA collection kit from DNA Genotek (OG-250^®, DNA Genotek, Kanata, ON, Canada). DNA was extracted according to the manufacturer’s instruction in a small volume of 200 µl of tris–ethylenediaminetetraacetic acid (EDTA; TE) buffer 10 : 1 to increase the quality of the high-throughput DNA genotyping. Samples passing DNA quality control were subjected to genome-wide genotyping using several different Illumina single nucleotide polymorphism (SNP) arrays and sample processing was in accordance with the manufacturer’s protocol. Each array contained a minimum of 600,000 SNPs. Genotype data were generated using Illumina’s^® BeadStudio software (Illumina Inc., Chesterford, UK). All copy-number variants (CNVs) were detected using the QuantiSNP algorithm based on the signal intensity and the B allele frequency values of each SNP. Visualisation was undertaken using the SnipPeep software.

After fully blinded genotyping had taken place, each genetic variant was tested for association with each outcome of interest. Full details of the outcomes investigated and the statistical methods used in the genetics substudy are provided in the statistical analysis plan (see Appendix 3). These methods were agreed in advance before undertaking any analyses.

In addition to genome-wide genotyping, all coding exons of AANAT (arylalkylamine N-acetyltransferase) and ASMT (N-acetylserotonin O-methyltransferase), the two genes of the melatonin synthesis pathway, were also sequenced. Sequencing conditions and primers have been described previously. 58 The impact of mutation on protein function was addressed in silico using the Polyphen2 algorithm (http://genetics.bwh.harvard.edu/pph2/) and/or according to previously published in vitro experiments. 59

A partial duplication of the ASMT gene, involving exons 1–7, has been described as occurring more frequently in patients with autism spectrum disorders than in the general population. 60 As this CNV cannot be detected using genotyping arrays, it was necessary to use a polymerase chain reaction (PCR)-based genotyping test. The CNV breakpoint was amplified together with a positive control PCR using the Qiagen Multiplex PCR kit (Crawley, UK) according to the manufacturer’s instruction. The annealing temperature was 66°C. The CNV-specific primers were as follows: forward primer: 5′–GTGGTGACAGATCTCGGCTCCCTTCAA–3′; reverse primer: 5′–GTCTGGCAGGACGGTTTCAG–3′. The positive control primers were as follows: forward primer: 5′–TGGTGCAATCTCATTTGACTCTG–3′.; reverse primer: 5′–GGGTTCATGCCATTCTCCTG–3′. The presence of PCR products was assessed by migration on 2% agarose gels.

Primary outcome

The primary outcome was TST, calculated using parentally completed diaries. The total amount of sleep for 1 night was calculated as the amount of time between the time that the child went to sleep and the time that the child woke up the following morning minus any night-time awakenings. The baseline measurement was calculated using the average total amount of night-time sleep in the 7 days before randomisation and the post-treatment measurement was calculated as the average total amount of night-time sleep from day 77 to day 84 post randomisation (this corresponds to the final 7 days of treatment because patients received enough drug supply only for 84 days).

A minimum of 5 nights of sleep from each time period was required for the data to contribute to the primary outcome. If a child had less than 5 out of 7 nights completed the data were regarded as missing and were not included in the primary analysis.

Secondary outcomes

• TST calculated using actigraphy data.

• SOL (the time taken to fall asleep) calculated using actigraphy.

• SOL (the time taken to fall asleep) calculated using sleep diaries (the number of minutes between lights out/‘snuggle-down’ time and sleep start time).

• Sleep efficiency calculated using actigraphy: (number of minutes spent sleeping in bed/total number of minutes spent in bed) × 100.

• CSDI score.

• Daily global measure of parental perception of child’s sleep quality (a ‘smiley face’ scale).

• Behavioural problems assessed using the ABC.

• Quality of life of the parent assessed using the Family Impact Module of the PedsQL.

• Level of daytime sleepiness in caregivers assessed using the ESS.

• Number and severity of seizures evaluated using seizure diaries throughout trial follow-up.

• Adverse effects of melatonin treatment assessed weekly between weeks T0W and T+12W using TESS.

• Salivary melatonin concentration.

• Associations between genetic variants and abnormal melatonin production.

Original sample size calculations

For the outcome total night-time sleep, the change between the total amount of sleep before randomisation and the total amount of sleep following randomisation will be calculated for each child. The titration period will not be used for the analysis of change. The null hypothesis is that there is no difference in the total amount of sleep between the melatonin and the placebo groups. The alternative hypothesis is that there is a difference in the total amount of sleep between the groups. The study is designed to detect a difference of 1 hour TST between the melatonin group and the placebo group. Assuming a common standard deviation of 1.7 (based on published data in similar populations/settings8,21), a sample size of 57 per group, increasing to 63 per group to allow for an estimated 10% loss to follow-up, will be required to provide 80% power using a t-test with a 0.025 two-sided significance level (adjusted to allow for multiple outcomes).

For the outcome SOL, the null hypothesis is that there is no difference in the time to sleep onset between the melatonin and the placebo groups. The alternative hypothesis is that there is a difference in the time to sleep onset. A sample size of 78 in each group (86 per group with estimated loss to follow-up of 10%) will have 80% power to detect a difference in means of 30 minutes, assuming a common standard deviation of 60 minutes using a two-group t-test with a 0.025 two-sided significance level.

Randomising a total of 172 children, 86 into each of the study arms, satisfies both sample size calculations. The sample size calculations are based on the use of nightly sleep diaries for total night-time sleep and actigraphy for time to sleep onset. Both outcomes will be analysed using analysis of covariance (ANCOVA), which will give an additional increase in statistical power.

Sample size calculation revision

The original trial recruitment target was 172 randomised patients, which was the maximum of the sample size calculations for the outcomes SOL (n = 172) and TST (n = 126). The original sample size calculation for TST was powered at 80% to detect a difference of 1 hour between the melatonin and the placebo groups using a common standard deviation of 1.7. Using a t-test with a 0.025 two-sided significance level, a sample size of 57 per group, increasing to 63 per group to allow for estimated 10% loss to follow-up, was required. Following the amendment to move SOL from a primary to a secondary outcome the sample size for the trial was recalculated based on the TST outcome. The revised calculation required 47 per group based on a 0.05 two-sided significance level as the multiplicity adjustment was no longer required; this was increased to 57 per group to allow for 20% missing data based on observed rates at the time of the amendment.

Interim monitoring

The estimate of the common standard deviation used in the sample size calculation was checked after the first 20 participants had been randomised and completed follow-up. This blinded internal pilot is not deemed to have any significant impact on the final analysis and no between-group comparisons were made. If the standard deviation had been found to be smaller than that used in the sample size calculation, suggesting that fewer patients were required than initially proposed, then no action would have been taken and the size of the study would have remained as originally planned. If the standard deviation was found to be larger than assumed, suggesting the need for more patients, then, on the advice of the Data Monitoring Committee (DMC), the TSC would have aimed to increase recruitment and consider implications for funding and existing resources. The DMC was presented with the results of the blinded internal pilot and recommended no change to the sample size based on these results.

Levels of missing data were monitored throughout and strategies developed to minimise its occurrence; however, as much information as possible was collected about the reasons for missing data.

Analysis plan

All analyses were conducted according to the statistical analysis plan (see Appendix 3), which provides a detailed and comprehensive description of the main, preplanned analyses for the study. Analyses were performed with standard statistical software [Statistical Analysis Software (SAS^®) 9.1.3; SAS Institute Inc., Cary, NC, USA] apart from those in the genetic substudy, which were undertaken using specialist genetic association software (see Appendix 3 for details).

The main features of the analysis plan are summarised below.

The Consolidated Standards of Reporting Trials (CONSORT) flow diagram is used to summarise representativeness of the study sample and patient throughput. Baseline characteristics are presented by treatment group and overall, with continuous variables presented with means and standard deviations and categorical variables with numbers and percentages.

The intention-to-treat principle is used as far as practically possible, with a two-sided p-value of 0.05 (5% level) used to declare statistical significance and 95% CIs reported throughout.

All continuous study outcomes are presented with means and standard deviations at T0 and T+12 and for the change over baseline (T+12 – T0) for each treatment group. ANCOVA is used to present results adjusted for baseline values. Reasons for missing data are provided (see Appendix 5). Sensitivity analyses are used to investigate the robustness of the primary outcome results to missing data (see Appendix 6).

Salivary melatonin assay

Tables 10 and 11 provide a summary of the results obtained for DLMO.

At T–1W the time to reach DLMO is shown in Figure 3 (excludes ‘none’ and ‘no samples’).

FIGURE 3.

Kaplan–Meier graph of time to reach DLMO at T–1W.

When producing the Kaplan–Meier curve, in those patients with results between two times, the latest time was taken.

It is of note that, at T+10W, only seven participants on melatonin provided samples from which DLMO time could be calculated compared with 41 for placebo. This difference between the two groups was not detected at T–1W. Of the samples classified as ‘none’ for DLMO at T+10W, 26 were possibly contaminated, two had a high baseline and four had a low volume (Table 12).

Some of the contamination and high baseline values in the melatonin arm are likely to reflect children who are poor metabolisers and whose levels of exogenous melatonin had accumulated during the study. Unfortunately, limited numbers at T+10W prevent meaningful analysis of the impact of this phenomenon on treatment response.

For the genetic substudy we defined DLMO categories using quartiles [1: DLMO ≤ 1930; 2: 1930 < DLMO ≤ 2030; 3: 2030 < DLMO < 2200; 4: DLMO ≥ 2200 (or no melatonin peak)] or medians [1: DLMO < 2100; 2: DLMO ≥ 2100 (or no melatonin peak)].

Genetic investigations

After applying the quality control filters, genotype data were available for 125 individuals. The principal component analysis indicated that the genotyped cohort was relatively homogeneous, with only eight outliers (Figure 4). All analyses were performed both with and without the outliers, and this did not affect the results.

FIGURE 4.

Multidimensional scaling plot of the MENDS sample to identify outliers (x) according to their identity by state with their nearest neighbour.

Outcome 1(i): association between genetic variations and sleep-onset latency (by sleep diary at T0)

It was possible to assess this outcome for 106 of the genotyped patients. A Manhattan plot showing results for association with this outcome is given in Figure 5. The different colours in the Manhattan plots are used to distinguish between chromosomes. Each point on the plot represents the p-value for an individual SNP. The p-values (indicated on the y-axis) are log-transformed to base 10 for ease of interpretation. Thus, a p-value of 10^–8 (which is typically taken as the threshold for significance in a genome-wide study) would have a value of 8 when log-transformed. None of the SNPs reached genome-wide significance.

FIGURE 5.

Manhattan plot for association with SOL at T0W.

Outcome 1(ii): association between genetic variations and amount of total sleep (by sleep diary at T0W)

It was possible to assess this outcome for 107 of the genotyped patients. A Manhattan plot showing results for association with this outcome is given in Figure 6. None of the SNPs reached genome-wide significance.

FIGURE 6.

Manhattan plot for association with amount of total sleep at T0W.

Outcome 4(i): association between genetic variations and difference in sleep-onset latency between T0W and T+12W

A Manhattan plot showing results for association with this outcome is given in Figure 7. One SNP [chromosome 4, rs17580458 in gene ANK2 (ankyrin 2)] reached genome-wide significance (p = 1.05 × 10^–9).

FIGURE 7.

Manhattan plot for association between genetic variations and difference in SOL between T0W and T+12W.

Outcome 4(ii): association between genetic variations and difference in amount of total sleep between T0W and T+12W

A Manhattan plot showing results for association with this outcome is given in Figure 8. None of the SNPs reached genome-wide significance.

FIGURE 8.

Manhattan plot for association between genetic variations and difference in amount of total sleep between T0W and T+12W.

Copy-number variant detection

Detection of CNVs was highly variable, depending on the genotyping array used. Chromosomal aneuploidy and known deleterious CNVs were present in several patients. Among the main findings, five patients had Down syndrome, one had DiGeorge syndrome, one had Cornelia de Lange syndrome and one had Smith–Magenis syndrome. Several CNVs were identified that affect genes known to be associated with developmental delay and/or ASD as well as genes involved in the clock/circadian pathway. Validation of these CNVs is in progress.

Exomic sequencing of AANAT and ASMT

Deoxyribonucleic acid of sufficient quality for sequencing was available for 134 individuals. The sequencing phase is completed for the ASMT gene. Four individuals were identified as carrying non-synonymous variations (Table 13). Two novel variations, G32C and H264D, were identified and a previously reported damaging mutation N17K was observed in two female patients. The G32C mutation is predicted to be ‘probably damaging’ whereas the H246D mutation is predicted to be benign. The overall frequency of ASMT mutations among the MENDS patients (0.029) does not seem to be different from that observed in the general population.

The sequencing phase is in progress for the AANAT gene but preliminary results are available for 77 patients. Three individuals were identified as carrying non-synonymous variations (see Table 13). Two patients were both carriers of a single variation, either T3M or A163V. A further patient was carrying two variations, T3M and A163V. The A163V mutation has been previously described and shown to moderately affect protein function. The T3M mutation is predicted to be benign. All patients with ASMT or AANAT mutations and treated with melatonin responded to treatment (Table 14).

Detection of ASMT copy-number variants

The presence of the ASMT CNV was assessed in 126 patients. Three of them were carriers of the CNV (Table 15). The overall frequency of this CNV among the MENDS patients (0.024) does not seem to be different from that observed in the general population (0.036, unpublished data, p = 0.80). One patient carrying the ASMT CNV was treated with melatonin and responded to treatment (see Table 14).

Number and severity of seizures evaluated using seizure diaries throughout trial follow-up

A total of 16 children (eight melatonin and eight placebo) had a diagnosis of epilepsy before randomisation. Thirteen children experienced seizures in the period between randomisation and the end of the study. There were a total of 411 seizures post randomisation and 123 seizures pre randomisation.

Table 16 shows the number of seizures by type and the number of children experiencing the seizure for each treatment group for both pre and post randomisation. Severity data were not recorded. The pre-randomisation data were based on seizure activity in a 4-week period and the post-randomisation data were based on seizure activity in a 12-week period. No formal statistical analyses were undertaken because of the limited data, particularly on the different seizure types.

Treatment-emergent signs and symptoms

Adverse effects were assessed weekly between baseline and the final visit using TESS.

The numbers (and percentages) of participants experiencing each aspect of TESS are presented for each treatment arm in Table 17. Table 18 presents TESS categorised by severity. For each participant only the maximum severity experienced of each symptom is displayed. Originally there were 14 aspects of TESS (somnolence, increased excitability, mood swings, seizures, rash, hypothermia, cough, increased activity, dizziness, hung-over feeling, tremor, vomiting, nausea and breathlessness); however, on 27 April 2009 TESS was reduced to seven domains (somnolence, increased excitability, mood swings, seizures, rash, hypothermia and cough).

No formal statistical testing was undertaken on these data.

From a careful evaluation of the data in Tables 17 and 18, there did not appear to be any major events in either of the treatment groups and no obvious differences between the two treatment groups.

Serious adverse events and suspected unexpected serious adverse reactions

There were five serious adverse events and two suspected unexpected serious adverse reactions during the course of the trial. Two serious adverse events were deemed to be unrelated and three unlikely to be related. There was one suspected unexpected serious adverse reaction in each treatment group. Details are given in Table 19.

Withdrawals

There were a total of 13 withdrawals from the trial: seven in the placebo group and six in the melatonin group. The reasons for withdrawal are shown in Table 20 by time point. Three participants provided data following withdrawal; these are indicated by an asterisk.

Primary outcome

The MENDS study is the largest RCT of melatonin in children with neurodevelopmental disabilities, including children with ASD, which was powered to detect a minimum clinically worthwhile change in TST of 1 hour. The main findings based on a blinded evaluation of the primary end point of mean change in TST at 12 weeks compared with baseline, measured using sleep diaries, showed that melatonin does increase TST but the increase is not clinically worthwhile. The upper limit of the CI was < 1 hour, the minimum clinically worthwhile difference specified at the outset of the trial.

The trial included a heterogeneous group of children covering a wide age range. Of the 10 children who did achieve a 1-hour increase in TST there were six on placebo and four on melatonin.

Other outcomes

Sleep-onset latency measured using actigraphy was a primary outcome at the outset of the trial but became a secondary outcome because of the large proportion of missing data; however, SOL remains an important end point for which a reduction of > 30 minutes at 12 weeks compared with baseline was considered to be clinically worthwhile. The results were both clinically important and statistically significant; however, the mean size of the effect was approximately 10 minutes larger when SOL was measured using actigraphy than when it was measured using sleep diaries. The CIs of SOL measured using actigraphy and measured using sleep diaries largely overlapped and each contained clinically worthwhile values. The differences in the results may be a reflection of the subgroup of children who were able to wear actigraphy monitors or differences between the subjective and objective methods of measuring sleep, or both.

Changes in the 12-week scores for each of the sleep and behaviour questionnaires were small but the direction of the change tended to favour melatonin. The magnitude of the change may be a reflection of the sensitivity of the instruments to change over a 12-week period, or the improvements in sleep seen were not sufficient to impact on the domains being assessed. With each of the instruments there is a lack of guidance on the size of change that would be considered worthwhile.

This study raised no safety concerns in relation to the medication, at least in the short term, with only five serious and two suspected unexpected serious adverse reactions, none of which was considered to be related to the active medication (melatonin). Seizure exacerbation was not seen in the children with an established diagnosis of epilepsy. However, because of the small numbers involved in the trial no firm conclusions can be drawn from the study.

Genetic investigations

Among the main findings of the genetic study, specific genetic disorders were identified in five patients with Down syndrome and one each with DiGeorge syndrome, Cornelia de Lange syndrome and Smith–Magenis syndrome; none of these disorders had been diagnosed in these patients before enrolling in MENDS. Several CNVs that affect genes known to be associated with developmental delay and/or ASD as well as genes involved in the clock/circadian pathway were detected and the validation of these CNVs is in progress. Partial deletions in the ASMT gene involving exons 1–7 have been described as occurring more frequently in patients with ASD than in the general population. 60 The sequencing phase for ASMT is complete and four individuals were identified as carrying non-synonymous variations. The sequencing phase for AANAT is still in progress with three non-synonymous variations identified thus far. The rates of these findings in our study population did not exceed those in the general population. It is of interest that all patients with ASMT or AANAT mutations treated with melatonin responded to treatment.

Design

The MENDS study was designed as a parallel RCT, although many of the previous RCTs that had assessed the use of melatonin (against placebo) in children had utilised a crossover design. We believe that the possible residual effect of either a behavioural or a melatonin intervention on sleep patterns and circadian rhythm make this particular intervention poorly suited for a crossover study.

The MENDS study incorporated a Dose escalation phase during the first 4 weeks of the study following randomisation. Although this cannot be considered a dose-ranging study, it demonstrated that patients who received placebo titrated more rapidly up to the maximum dose and also remained on the maximum dose at the end of the double-blind phase of the study. However, these results do not allow us to conclude which dose is the most effective.

The definition of sleep disorder used within MENDS was broad and did not make allowances for the age of the child in determining whether or not they had a sleep disorder. Data were not collected on the specific nature of the clinical diagnosis of the sleep disorder at baseline (delayed sleep onset or poor TST, or a combination of both) and this is being collected retrospectively to facilitate secondary analyses.

The length of follow-up post randomisation was short and this may have impacted on the ability of the quality-of-life instruments to detect change. Longer-term follow-up should be considered for future studies and use appropriate quality-of-life tools as well as qualitative interviews to fully explore the impact of impaired (and improved) sleep on family life.

Recruitment and retention

Recruitment into the trial was slower than expected. This may have reflected the widespread availability and prescribing of melatonin outside of the trial and its perceived excellent safety profile. The sample size target for the primary outcome TST was achieved over an extended recruitment period. The trial was supported by the MCRN LRNs and this aided trial recruitment, motivation at sites and data collection. The trial showed a low withdrawal rate such that once participants were randomised into the trial they continued until the completion of the follow-up period. This is considered to reflect not only the commitment of the families and the motivation and hard work of the research practitioners, but also the design of the study, which maintained weekly contact with the families, an important activity that was supported by the LRNs.

Outcomes

The study was ambitious in its goals in that the initial design had two primary outcomes and a large number of secondary outcomes. The decision to use two primary outcomes reflected the perceived characteristic sleep problems in children with neurodevelopmental delay, namely their difficulty in falling asleep (measured by actigraphy) but also their difficulty in sleeping continuously throughout the night (measured by sleep diaries); many children experience both problems.

The trial used both subjective and objective measures of sleep as recommended by Sadeh. 47 There were potential benefits of each approach and underlying reasons why the results would not be expected to be concordant. These included that the sleep diaries would not detect periods when the child was awake but not disturbing the household (a particular concern for determining SOL), and that the actigraph would interpret restless sleep as being awake (a concern for children with motor problems, including cerebral palsy). Dayyat et al. 63 have also considered these objective and subjective measures and their findings concur with our observed differences when measuring SOL and TST using actigraphy and sleep diaries.

It was clear that many children were unable to co-operate with wearing actigraphs and these missing actigraphy data were compounded by a significant rate of actigraph failure. Consequently, SOL measured using actigraphy was moved to a secondary outcome.

The impact on generalisability of the actigraphy outcomes needs to be considered when interpreting results; however, the results were largely consistent with the results for the same outcomes calculated using data from sleep diaries.

The ability of the potentially sleep-deprived families to complete the sleep diaries continuously over the 16-week period required was a concern. The completion rates of the diaries were monitored, and improvements made to the sleep diary to highlight key fields led to improvements in their completion. Overall, completion of sleep diaries following the amendment was very good and parents did complete diaries for the full 16 weeks. This trial further demonstrates the acceptability of such data collection to parents and carers, as has been reported previously. 64

Parents and carers were requested to obtain saliva samples from the children for salivary melatonin assays to determine the time of DMLO. The salivary melatonin analysis was undertaken primarily as an exploratory or hypothesis-generating approach. This was an attempt to enable biochemical phenotyping of those children with a genuinely delayed sleep phase and who might be expected to be better responders to melatonin. However, the limitations of trying to collect saliva in the home and without causing the children any distress were reflected in the limited data obtained. Therefore, it is very difficult to interpret our salivary melatonin data. The salivary melatonin results of those children who do not display a DLMO before sampling stopped may reflect a delayed DLMO or a lack of melatonin production or a combination of both; unfortunately, we are not able to say which is more likely.

The number of sleep awakenings was not a prespecified end point of the study as the impact of night awakenings was expected to impact on TST. However, this end point will be considered in secondary analyses.

We did not measure ‘before and after’ cognitive abilities of the children in this study. Experimentally, sleep restriction can affect certain cognitive processes and we might arguably have been expected to have found changes in these abilities. However, many of the children in this study were low or very low functioning academically, and IQ tests are less reliable for this population and require experienced testers to perform reliable and interpretable assessments. Our compromise was to try to use the computer-based, picture-formatted and child-specific MARS test but even this was too difficult to undertake for most of the children. Participation (compliance) was so poor that the MARS test was omitted from the battery of secondary outcomes.

Genotyping

Because of the use of several different genotyping arrays, it was only possible to undertake analyses of association on approximately 200,000 SNPs that were common to all arrays. Genotype imputation is due to start shortly in order to increase the number of SNPs investigated, and thus coverage of the genome.

Comparison with other studies

A number of systematic reviews with meta-analyses are available that include trials relevant to the MENDS population. 23,24,28,35,36,65 There are many differences between these systematic reviews, including the robustness of the eligibility criteria for inclusion of the trials and heterogeneity between the trials. A relevant protocol for a systematic review of randomised evidence of melatonin in children with neurodevelopmental disorders,66 poignant to the MENDS trial, is published on The Cochrane Library and, following publication, the MENDS trial will be included in an update.

The most relevant systematic reviews with meta-analyses23,24 have been published since the inception of MENDS. Although MENDS had a primary focus on children with neurodevelopmental disorders, approximately one-third had ASD. We have therefore also carefully studied the recently published meta-analysis by Rossignol and Frye24 of melatonin use in individuals with ASDs. It is important to note that the meta-analyses by Braam et al. 23 and Rossignol and Frye24 contain overlapping studies, and the total number of participants in the included trials is small.

The MENDS study has shown that, with melatonin, total night-time sleep duration on average increased by 23 minutes, a clinically unimportant increase based on our prior power calculations, and SOL was reduced on average by 38 minutes when measured using sleep diaries. The meta-analysis of the effect of melatonin in individuals with intellectual disability23 found that melatonin increased TST by a mean of 50 minutes and reduced SOL by 34 minutes. Rossignol and Frye24 found a mean increase in TST of 44 minutes and a reduction in SOL of 39 minutes. The size of the effect for SOL is similar and consistent between MENDS and these meta-analyses; however, it is estimated that there is a larger average effect for TST in the meta-analyses than in MENDS.

We have some concerns about the robustness of these meta-analyses because of a number of shared methodological issues. Many of the issues of concern have been discussed in a critique of the Buscemi review67 and are applicable to the two meta-analyses. This included concerns around the reasons why studies were excluded, the differences between studies, the information reported, the quality/risk of bias of included studies, and outcome reporting bias.

All of the included studies in the Rossignol and Frye meta-analysis24 and seven of nine studies in the Braam et al. meta-analysis23 were crossover designs. It is likely that the primary reason for choosing this design was to reduce sample size. Unfortunately, the possible residual effects of either a behavioural or a melatonin intervention on sleep patterns and circadian rhythm make this particular intervention poorly suited for a crossover study. The reasons for the persistence of the effect of melatonin on body clock synchronisation are summarised in the meta-analysis of Braams et al. 23 The difficulties that this posed to the researchers are exemplified in their choice of variable washout periods of just 3 days to 1 month. There are simply inadequate data to be confident that a change in timing of sleep onset and offset will securely ‘wash out’ to baseline timings after any specified time period.

The actual definition of a sleep problem (nature, persistence or severity) justifying treatment with melatonin also varies widely between studies. In the study by Wirojanan et al. ,18 entry depended exclusively on parents reporting a sleep disorder. In the study by Wright et al. ,68 children were included who manifested any type of sleep problem that persisted after a behavioural intervention. In the study by Garstang et al. ,69 children were required to have a sleep latency of > 1 hour for inclusion. In addition, this last study was suspended after only seven children completed, because of the discovery that some placebo capsules were empty. Consequently, little can be concluded and therefore generalised about the nature of the underlying sleep problem from these few studies.

Different doses of different preparations of melatonin were used in most of the studies, with some relying on a fixed dose (from 2 mg to 9 mg) and others allowing dose escalation.

Very few of the studies included objective outcome measures such as actigraphy. In most studies outcome measures were largely based on subjective parental diaries. The lack of any objective measures in most studies, and differences between objective and subjective measures, was a problem acknowledged by the reviewers.

It is interesting that we found a similar magnitude of change in SOL as did a robust study on ‘normally’ developing children with sleep-onset insomnia. 70 This study also failed to demonstrate any significant increase for total sleep duration. The correlation between sleep latency and sleep duration identified in MENDS is important and supports a hypothesis that melatonin exerts its main effects by reducing sleep latency, and that this on its own would increase sleep duration if sleep offset (time of wakening) remains the same.

A recently published, randomised, placebo-controlled trial explored the dose of melatonin and response in typically (‘normally’) developing children aged from 6 to 12 years with sleep-onset insomnia. 68 The children received either melatonin (0.05 mg/kg, 0.1 mg/kg and 0.15 mg/kg) or placebo for 1 week (to allow some comparison of dose ranges, an average 9-year-old boy in MENDS on this regime weighing 30 kg would receive 1.5 mg, 3 mg or 4.5 mg, and no child would have received the 6-mg or 12-mg MENDS dose). The authors did not include sleep duration in their results having previously shown that this does not change in this group of children. 70 Even with these relatively low doses of melatonin, the authors did find that melatonin significantly advanced DLMO by approximately 60 minutes and decreased SOL by 35 minutes. They were unable to find any dose–response relationship, but did show that the circadian time of administration played a significant role. The fact that this study did not report on sleep duration, and all children were typically (‘normally’) developing, limits our ability to directly compare this study with MENDS. Clearly, the 6-mg and 12-mg doses in MENDS were much higher than the doses in this recent study. 68 In addition, their study predominantly focused on melatonin’s ability to phase shift DLMO and sleep onset, rather than its soporific effects, although these were briefly discussed. It is possible that, for children with often profound developmental delay, higher doses are required based on their sedating rather than phase-shifting role.

Generalisability

It is important to appreciate that, to date, MENDS is the largest RCT undertaken in children with neurodevelopmental delay and, although the population studied was heterogeneous, the results for both SOL and TST are similar to those reported for a total of 183 patients in the meta-analysis carried out in 2009. 23 These observations would suggest that the results of MENDS could reasonably be extrapolated to a much larger paediatric population with a range of neurological disorders, including neurodevelopmental delay.

The MENDS study was intended to be pragmatic and, as far as possible, to reflect usual clinical practice for this group of children. This strategy has resulted in strengths and weaknesses. We have attempted to faithfully mirror current clinical practice and believe our results are generalisable. However, the sheer heterogeneity of the population studied has inevitably limited our ability to accurately estimate the impact of melatonin treatment for individual groups of patients with specific clinical (genetic), behavioural or developmental presentations. To a certain extent, this also applies to the different preparations (formulations) of melatonin that are currently in use throughout the UK. At the time that this study was under design, no licensed slow-release preparations of melatonin were available. Reports at the time cast significant doubt and concern over the ‘slow-release’ properties of the available non-pharmaceutical-grade products. Although there is now one slow-release preparation available (Circadin), its slow-release properties rely on the tablet being swallowed whole, which was not a skill possessed by many of the children in the study population. In the future, more flexible slow-release preparations might be available. Head-to-head trials comparing fast- and slow-release preparations are warranted.

Interpretation

The results of MENDS have confirmed the results of previous studies which have shown that melatonin is effective in reducing SOL (the time taken to fall asleep) in children with neurodevelopmental delay, reducing this time by a mean of 45 minutes. The results of meta-analysis demonstrated that melatonin reduced SOL by a mean of 34 minutes. 29

A clinically significant increase in TST remains an important target for sleep interventions. Although the total sleep duration did not increase by 1 hour, we acknowledge that an increase in sleep duration of < 1 hour may still be clinically important. There is a lack of robust information to guide a clinical decision in this context about the importance of a small increase in total duration of night-time sleep. Much of the data that argue about the importance of sleep duration are based on experimental situations in which sleep duration is acutely and artificially reduced. In these experimental models, sleep has an impact on daytime learning and behaviours, although the size of this impact is unclear and inconsistent. 48 At a neurobiological level, some researchers speculate that, over ‘time’ (unspecified), sleep loss may cause actual neuronal loss. 71 If this hypothesis of the effect of cumulative sleep loss on neuronal loss is confirmed, then even a small increase in TST that is achieved consistently, night after night, week after week, may be worthwhile.

Our findings show that the strongest effects of exogenous melatonin are on SOL. In exploratory analyses we have found a strong correlation between those children with later DLMO peaks and those children who fall asleep later. We have also found that the amplitude of treatment response is strongly correlated with the initial severity of the sleep disorder. Thus, as has been described in typically developing children, but not replicated in this population, children who have later DLMO times fall asleep later and respond better to exogenous melatonin. As SOL and sleep duration are related, so an improvement in SOL may also lengthen sleep duration, but this depends on whether or not sleep offset (the time that the child wakes up) alters. This does, however, support the possible utility of pretreatment DLMO measurement to predict the better treatment responders. However, this will clearly require further evaluation as well as a full economic assessment, which was beyond the scope of MENDS. As far as we are aware, no controlled trial has demonstrated a reduction in night awakenings on active treatment, and, although this was not one of our specified secondary outcome measures, it is an important question that should be included in any future study.

The genetic analysis substudy has allowed high-throughput genome-wide genotyping to be completed for all 125 samples that passed the quality control for DNA genotyping. We were fortunate to be able to use the latest high-density Illumina arrays measuring up to 2.5 million SNPs. SNP association analysis was undertaken with several important outcomes, including severity of sleep disorder, melatonin levels and response to melatonin treatment. The analysis also allowed an investigation of the incidence of rare CNVs in the MENDS sample in comparison to the general population control rate. Specific sequencing was undertaken of all exons of AANAT and ASMT, two important and putative enzymes for which rare deletions (seen more commonly in children with autism) have been reported to account for reduced melatonin levels. Although preliminary results for this part of the analysis are available, the volume and the complexity of the data demand that analysis needs to be carefully undertaken by specific software and there are limited resources for such work. Consequently, this is currently still ‘work in progress’, both in Liverpool and at the Pasteur Institute (Paris).

There were only five serious and two suspected unexpected serious adverse reactions during the study period, none of which was considered to be related to the active medication (melatonin). In the 16 patients with epilepsy, equally distributed between the two treatment groups, none showed any deterioration in seizure control, emergence of a new seizure type or de novo epilepsy. This concurs with the vast majority of previous reports, including one which suggested that melatonin may have an inherent anticonvulsant effect. 72 Only one report published in 1998 has suggested that it may have a proconvulsant effect. 42

Implications for health care

Sleep disorders are a common presentation in children with a wide variety of neurodevelopmental conditions. Medication should not be the first-line intervention and, as has been shown previously, our behavioural run-in period was successful, with many children no longer meeting the inclusion criteria for the study. This effect was seen after a relatively short period using a specific evidence-based behaviour therapy advice booklet and monitoring but with no direct contact with psychology or other sleep behavioural specialists. However, it is possible that the relatively large ‘dropout’ of patients in the 4- to 6-week behaviour intervention (therapy) phase may also reflect parental perceptions of the child’s sleep problem. The process of formally observing and documenting their child’s sleep pattern in sleep diaries may have unmasked a significant gap between their perceived interpretation of their child’s sleep problem and their child’s actual sleep problem. It would be relatively easy to test this hypothesis in a future RCT of a behavioural intervention compared with no intervention in this type of population. The results of MENDS would also suggest that in routine clinical practice families should be asked to complete a 2- to 4-week sleep diary before melatonin is prescribed.

Melatonin is more effective than placebo for children with neurodevelopmental delay who have trouble falling asleep. This is a common presenting complaint and melatonin reduces this period by an average of 37 minutes. This is helpful for parents desperate to settle their child with neurodevelopmental delay and may result in a calmer evening for themselves or for siblings (and other family members) or both. However, we found no evidence that this reduction in sleep latency measurably improved the family’s quality of life or children’s behaviour over the 3-month period. It did seem to reduce parents’ reports of daytime fatigue, which is an interesting finding that should be explored in future studies.

Although the children fell asleep earlier, they gained very little extra sleep when measured over the whole night. An extra 23 minutes of sleep over a whole night is small and was deemed not to be clinically significant for our study. The increase does of course vary with individuals and is likely to be cumulative. Experimentally, an artificial reduction of this amount of sleep every night has been shown over the very short term to affect daytime behaviour and cognition. 73 However, in the 3-month study period of MENDS we were unable to identify any improvement in children’s behaviour. Cognition was not formally assessed in MENDS.

Implications for research

• Our trial compared melatonin only with placebo. There are a number of other licensed and unlicensed medications (including hypnotics and sedatives) for children with sleep problems. 74 Head-to-head trials are needed to help clinicians and families decide which medication(s) is (are) likely to be the safest and most helpful.

• Further studies need to be undertaken to try and establish the most appropriate dose of melatonin, incorporating the child’s age, weight, 24-hour endogenous melatonin profile and DLMO and whether they are a fast or slow metaboliser. Both fast- and slow-release preparations should be considered. A dose-ranging study in children with sleep-onset disorder has recently been published;75 however, no dose–response relationship of melatonin in advancing DLMO or reducing SOL was found within a dosage range of 0.05–0.15 mg/kg.

• Patients’ cognition was not directly measured in MENDS. Given that cognitive function may reflect important end points around learning potential, this outcome will be important to explore in future intervention trials, however difficult this is likely to be.

• Future RCTs that assess the effectiveness and safety of melatonin should be undertaken over a longer period and should include both appropriate quality of life and economic evaluations.

Total night-time sleep calculated using sleep diaries

The total amount of sleep for 1 night will be calculated in minutes using the amount of time between the time that the child went to sleep and the time that the child woke up the following morning minus any night-time awakenings that the child has had. The baseline measurement will be calculated using the average total amount of sleep in the 7 days before randomisation and the post-treatment measurement will be the average total amount of sleep from day 77 to day 84 post randomisation (this corresponds to the final 7 days of treatment as patients received enough drug supply only for 84 days).

An example of the data used to calculate TST for a participant randomised on 20 January 2010 is given in Box 1.

A minimum of 5 nights of sleep from each time period is required for the data to contribute to the primary outcome. If a child has < 5 out of 7 nights completed the data will be regarded as missing and the remaining data will not be included in the primary analysis. Methods for handling missing data are discussed in Analysis of missing data.

The mean total night sleep (and standard deviation) for the week before randomisation (T0W) and the final 7 days of treatment (days 77–84, T+12W) and the mean change over baseline (T+12W – T0W) will be presented by treatment group. The mean difference in change over baseline between the two groups will be reported with 95% CIs. A two-sided p-value of 0.05 (5% level) will be used to declare statistical significance and will be reported alongside the CI.

The method of ANCOVA will be used to adjust for baseline sleep, the outcome measure being the average total amount of sleep in the final week of treatment (days 77–84) and the covariates that will be used in the model being treatment group (melatonin or placebo) and the baseline measurement (the average total amount of sleep in the 7 days before randomisation).

Randomisation is stratified by centre. However, because of the large number of small centres, the centre will not be included in the model as a covariate. Including a large number of small centres may lead to unreliable estimates of the treatment effect and p-values that may be too large or too small.

Analysis of missing data

Analysis of missing data will be restricted to the primary outcome only; no imputation methods will be used on any of the secondary outcomes.

The number of completed sleep diary days available for the 7-day period before T0W and T+12W will be reported for each group. A chi-squared test will be used to test for differences between the groups.

The primary analyses for the primary outcome will contain data only on participants who have ≥ 5 days of complete data during the 7 days immediately before randomisation and ≥ 5 days during days 77–84 post randomisation. Those participants who have < 5 days at either time period will not be included in the primary analyses.

The sensitivity of the results to those with missing data will be assessed within two population groups:

1. those contributing data to the primary outcome with ≥ 5 nights observed (missing data for 1 or 2 nights only) at T0W and T+12W

2. those contributing data to the primary outcome with ≥ 2 nights observed (missing data for up to 5 nights) at T0W and T+12W.

For each missing data point that a person has at both T0W and T+12W the worst recorded night of sleep from baseline will be imputed.

Data available from all sources will be considered to inform any further imputation strategies. This includes use of actigraphy for total night-time sleep and the daily global measure. The imputation strategies will be within patient rather than averages across patients. The potential for this approach will be informed by completion rates and the demonstration of a relationship but it will be performed only if the robustness of the results to the approach above is questionable.

Further sensitivity analyses were identified after the preliminary results were presented. These were to include:

• An analysis of those patients who had ≥ 5 completed sleep diary days at T+12W only. This will compare the mean TST between the treatment groups at T+12W only. A t-test will be used to compare the difference between the two groups and a p-value and 95% CI will be presented.

• An analysis that will be same as the primary analysis but which will include patients who have a minimum of 4 out of 7 completed sleep diary days. The analysis will be the same as that described in Analysis of primary outcome and the results will be presented similarly.

Total night-time sleep calculated using actigraphy data

The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average total amount of sleep and the treatment group. Those participants who have < 5 days of data will not be included in the analysis. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12W and the change over baseline (T+12 – T0W) for each treatment group.

Reasons for missing data will be documented and the results interpreted as appropriate.

Sleep efficiency calculated using actigraphy data [(number of minutes spent sleeping in bed/total number of minutes spent in bed) × 100]

The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average total amount of sleep and the treatment group. Those participants who have < 5 days of data will not be included in the analysis. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12 and the change over baseline (T+12 – T0W) for each treatment group.

Reasons for missing data will be documented and the results interpreted as appropriate.

Sleep-onset latency (the time taken to fall asleep) calculated using actigraphy data

The baseline measurement will be calculated using the average sleep latency in the 7 days before randomisation and the post-treatment measurement will be the average sleep latency from day 77 to day 84 post randomisation (this corresponds to the final 7 days of treatment as the patients received enough drug supply only for 84 days).

The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average total amount of sleep and the treatment group. Those participants who have < 5 days of data will not be included in the analysis. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12W and the change over baseline (T+12 – T0W) for each treatment group.

Sleep-onset latency (the time taken to fall asleep) calculated using sleep diaries

The SOL will be calculated using the sleep diary that has been completed. The baseline measurement will be calculated using the average sleep latency in the 7 days before randomisation and the post-treatment measurement will be the average sleep latency from day 77 to day 84 post randomisation (this corresponds to the final 7 days of treatment as the patients received enough drug supply only for 84 days).

The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average total amount of sleep and the treatment group. Those participants who have < 5 days of data will not be included in the analysis. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12W and the change over baseline (T+12 – T0W) for each treatment group.

Composite Sleep Disturbance Index

The CSDI is based on six items and includes measures on settling frequency and duration, night-waking frequency and duration, frequency of early waking and frequency of co-sleeping. Settling problems, night waking, early waking and co-sleeping are measured in terms of weekly frequency, and settling and night waking problems are also assessed in terms of nightly duration of the problem. The questions are scored as in Table 22.

The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average and the treatment group. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12W and the change over baseline (T+12W – T0W) for each treatment group. If the data are found not to be normally distributed, then the equivalent non-parametric test will be used.

Daily global measure of parental perception of child’s sleep quality

The daily global measure of parental perception of a child’s sleep quality is measured on the sleep diary with the use of ‘smiley faces’ on a scale of 1–7, with 1 being a very good night’s sleep and 7 being a very bad night’s sleep. To be included in the analysis the global measure must be completed for a minimum of 5 out of 7 nights.

The score will be calculated in two ways at T0W and T+12W:

1. the percentage of night sleeps with which the parent was dissatisfied (faces 5–7)

2. the mean of the scores for each night.

The change between T0W and T+12W will be calculated and presented with 95% CIs; for analysis 1 the mean percentage and standard deviation of night sleeps with which the parent was dissatisfied will be reported for T0, T+12W and the change over baseline (T+12W – T0W) for each treatment group, and for analysis 2 the mean score and standard deviation for T0, T12 and the change over baseline (T+12W – T0W) will be reported for each treatment group.

Behavioural problems assessed using the Aberrant Behaviour Checklist

The ABC consists of 58 items, each scored on a 4-point scale (0 = not a problem to 3 = problem is severe in degree). The items fall into five subscales: (1) irritability, agitation, crying (15 items), (2) lethargy, social withdrawal (16 items), (3) stereotypical behaviour (7 items), (4) hyperactivity, non-compliance (16 items) and (5) inappropriate speech (4 items).

The ABC manual does not describe methods of analysis of data collected using this tool, nor does it discuss methods for handling missing data. The manual does present the results of data in subscales and not as a total and therefore we will adopt this method for the presentation of the results of this tool.

No imputation will be made for missing items within a subscale and therefore subscales must be complete. The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average and the treatment group. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12W and the change over baseline (T+12W – T0W) for each treatment group. If the data are found not to be normally distributed then the non-parametric equivalent test will be used.

Quality of life of the caregiver assessed using the Family Impact Module of the Pediatric Quality of Life Inventory

The scoring instructions for the PedsQL Family Impact Module were taken from Varni et al. 51

The PedsQL Family Impact Module consists of 36 items. The module is split into two sections: six scales measure parent self-reported functioning and two scales measure parent-reported family functioning. The six scales that measure parent self-reported functioning are as follows:

1. physical functioning (6 items)

2. emotional functioning (5 items)

3. social functioning (4 items)

4. cognitive functioning (5 items)

5. communication (3 items)

6. worry (5 items).

The two scales that measure parent reported family functioning are as follows:

1. daily activities (3 items)

2. family relationships (5 items).

To score the module, a 5-point scale is used (0 = if it is never a problem, 1 = if it is almost never a problem, 2 = if it is sometimes a problem, 3 = if it is often a problem and 4 = if it is almost always a problem) and items are reverse scored and linearly transformed to a 0–100 scale (0 = 100, 1 = 75, 2 = 50, 3 = 25 and 4 = 0); therefore, higher scores indicate better functioning.

Each scale is scored and the scale score is calculated as the sum of the items divided by the number of items answered. If > 50% of the items in the scale have not been answered then the scale score should not be computed. A sensitivity analysis will impute the mean of the completed items in the scale (if the scale does not contain > 50% missing data).

The PedsQL Family Impact Module total scale score is calculated as the sum of all 36 items divided by the number of items answered. If more than 50% of the items have not been answered then the total scale score should not be computed. A sensitivity analysis will impute the mean of the completed items in the scale (if the total scale score does not contain > 50% missing data). The parent health-related quality-of-life summary score (20 items) is computed as the sum of the items divided by the number of items answered in the physical, emotional, social and cognitive functioning scales. The family functioning summary score (8 items) is computed as the sum of the items divided by the number of items completed in the daily activities and family relationships scales. The same procedure for handling missing data will used for these scales as described above.

The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average and the treatment group. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12W and the change over baseline (T+12W – T0W) for each treatment group. If the data are found not to be normally distributed then the non-parametric equivalent test will be used.

Level of daytime sleepiness of caregiver assessed using the Epworth Sleepiness Scale

The ESS is used to determine the parent’s (not the child’s) level of daytime sleepiness. There are eight questions and a 4-point scale for each question. The total ESS score is the sum of the eight questions and can range between 0 and 24. If a question has been left blank then the ESS is not valid and no methods of imputation will be used; this is as recommended by the author of the ESS (http://epworthsleepinessscale.com/1997-version-ess/).

The analysis will use the method of ANCOVA and will not adjust for any missing data and the model will contain only two covariates: the baseline average and the treatment group. The adjusted and unadjusted results will be presented as well as means and standard deviations for T0W, T+12W and the change over baseline (T+12W – T0W) for each treatment group. If the data are found not to be normally distributed then the non-parametric equivalent test will be used.

Number and severity of seizures evaluated using seizure diaries throughout trial follow-up

The numbers and percentages of patients (who have epilepsy) experiencing a seizure will be presented for each treatment arm pre and post randomisation. The type of epilepsy that each patient had at the initial screening visit will also be presented. For each patient, only the maximum frequency experienced of each seizure will be displayed. No formal statistical testing will be undertaken.

Adverse effects assessed weekly between weeks T0W and T12+W using treatment-emergent signs and symptoms

The numbers (and percentages) of patients experiencing each aspect of TESS will be presented for each treatment arm categorised by severity. Originally there were 14 aspects of TESS (somnolence, increased excitability, mood swings, seizures, rash, hypothermia, cough, increased activity, dizziness, hangover feeling, tremor, vomiting, nausea and breathlessness). On 27 April 2009 TESS was reduced to seven aspects (somnolence, increased excitability, mood swings, seizures, rash, hypothermia and cough). Results will be presented for all data that have been collected during the study period.

For each patient, only the maximum severity experienced of each type of TESS will be displayed. The numbers (and percentages) of occurrences of each type will also be presented for each treatment arm. No formal statistical testing will be undertaken.

Salivary melatonin concentrations

From the saliva concentrations measured, the DLMO time will be calculated for T0W and T+12W as the time when saliva melatonin levels reach 2 × SD of baseline values.

Kaplan–Meier curves and the log-rank test will be used to compare time to DLMO between the melatonin and the placebo groups at T0W and T+12W.

Outcomes

For the genetic substudy, the associations between genetic variants and the following outcomes were of interest:

1. severity of sleep disorder, assessed using two different measures:

   1. SOL (as assessed by sleep diary at T0W)

   2. TST (as assessed by sleep diary at T0W)

2. melatonin level, defined as salivary melatonin concentration at 2000 at T–1W

3. ability to synthesise melatonin

4. response to melatonin treatment, assessed using two different measures:

   1. difference in SOL between T0W and T+12W (as assessed by sleep diaries at these two time points)

   2. difference in TST between T0W and T+12W (as assessed by sleep diaries at these two time points).

For the biochemical investigation, the relationship between melatonin levels and sleep disorders was of interest. The variables of interest here were SOL and TST, both as defined above, and DLMO, which was represented as a categorical covariate in two different ways:

1. using quartiles: 1: DLMO ≤ 1930; 2: 1930 < DLMO ≤ 2030; 3: 2030 < DLMO < 2200; 4: DLMO ≥ 2200 (or no melatonin peak)

2. using medians: 1: DLMO < 2100; 2: DLMO ≥ 2100 (or no melatonin peak).

Genotyping

Before the analyses of association, the genotype data were subjected to a number of quality control filters. Patient samples with a call rate < 98% were excluded from further analysis, as were SNPs with call rates < 98% and SNPs showing departure from the Hardy–Weinberg equilibrium (p < 10^–6). To check for genetic diversity within the genotyped cohort, a principal component analysis was undertaken using the identity by state metrics, which estimates genetic distances between individuals. 76 To identify outliers, the distance of each individual to their fifth nearest neighbour was calculated and those having the largest distance flagged as outliers. When population outliers were detected, all analyses were performed both with and without the outliers, and unless otherwise stated this did not affect the results.

Analyses of association

The association analyses between genetic variants and each outcome were performed using the software package PLINK. 76 For the outcomes of severity of sleep disorder, a linear regression model was fitted for each SNP in turn. As the outcomes are age dependent, a covariate to represent age was included in the models. An additive mode of inheritance was assumed, with SNP represented by a single covariate. For the outcomes of response to melatonin treatment, a similar approach was taken; however, rather than adjusting for age the regression models were this time adjusted for severity of sleep disorder at baseline (i.e. SOL at T0W was taken as a covariate when outcome was difference in SOL, and sleep duration at T0W was taken as a covariate when outcome was difference in sleep duration). For the outcome of melatonin levels, an unadjusted linear regression model was fitted for each SNP in turn.

Results were presented as Manhattan plots, which are plots of –log₁₀ of the p-value from the test of association with each SNP against its chromosomal location. A p-value ≤ 1 × 10^–8 [or –log₁₀(p-value) ≤ 8] is widely accepted as representing statistical significance at the genome-wide level.

To investigate the association between melatonin levels and sleep disorders, JMP statistical software (version 9; SAS, NC, USA) was used. The Wilcoxon test was used when the outcome was categorical and linear regression was used when the outcome was continuous.

Seventh substantial amendment version 6.2 (23 June 2009) to version 7.0 (1 April 2010)

Substantial amendment version 6.1 (4 March 2009) to version 6.2 (10 July 2009)

Fifth substantial amendment version 5.0 (9 July 2008) to version 6.1 (4 March 2009)

Version 6.0 (27 January 2009) was submitted to the Multicentre Research Ethics Committee and required additional amendments before approval.

Forth substantial amendment version 4.0 (6 May 2008) to version 5.0 (9 July 2008)

Third substantial amendment version 3.0 (25 January 2008) to version 4.0 (6 May 2008)

Second substantial amendment version 2.3 (3 December 2007) to version 3.0 (25 January 2008)

First substantial amendment Version 1.0 (26 April 2007) to Version 2.0 (17 August 2007)

Prioritisation Group

1. Director, NIHR HTA programme, Professor of Clinical Pharmacology, Department of Pharmacology and Therapeutics, University of Liverpool

2. Professor Imti Choonara, Professor in Child Health, Academic Division of Child Health, University of Nottingham

   Chair – Pharmaceuticals Panel

3. Dr Bob Coates, Consultant Advisor – Disease Prevention Panel

4. Dr Andrew Cook, Consultant Advisor – Intervention Procedures Panel

5. Dr Peter Davidson, Director of NETSCC, Health Technology Assessment

6. Dr Nick Hicks, Consultant Adviser – Diagnostic Technologies and Screening Panel, Consultant Advisor–Psychological and Community Therapies Panel

7. Ms Susan Hird, Consultant Advisor, External Devices and Physical Therapies Panel

8. Professor Sallie Lamb, Director, Warwick Clinical Trials Unit, Warwick Medical School, University of Warwick

   Chair – HTA Clinical Evaluation and Trials Board

9. Professor Jonathan Michaels, Professor of Vascular Surgery, Sheffield Vascular Institute, University of Sheffield

   Chair – Interventional Procedures Panel

10. Professor Ruairidh Milne, Director – External Relations

11. Dr John Pounsford, Consultant Physician, Directorate of Medical Services, North Bristol NHS Trust

    Chair – External Devices and Physical Therapies Panel

12. Dr Vaughan Thomas, Consultant Advisor – Pharmaceuticals Panel, Clinical

    Lead – Clinical Evaluation Trials Prioritisation Group

13. Professor Margaret Thorogood, Professor of Epidemiology, Health Sciences Research Institute, University of Warwick

    Chair – Disease Prevention Panel

14. Professor Lindsay Turnbull, Professor of Radiology, Centre for the MR Investigations, University of Hull

    Chair – Diagnostic Technologies and Screening Panel

15. Professor Scott Weich, Professor of Psychiatry, Health Sciences Research Institute, University of Warwick

    Chair – Psychological and Community Therapies Panel

16. Professor Hywel Williams, Director of Nottingham Clinical Trials Unit, Centre of Evidence-Based Dermatology, University of Nottingham

    Chair – HTA Commissioning Board

    Deputy HTA Programme Director

HTA Commissioning Board

1. Professor of Dermato-Epidemiology, Centre of Evidence-Based Dermatology, University of Nottingham

2. Professor of Bio-Statistics, Department of Public Health and Epidemiology, University of Birmingham

3. Professor of Clinical Pharmacology, Director, NIHR HTA programme, Department of Pharmacology and Therapeutics, University of Liverpool

1. Professor Zarko Alfirevic, Head of Department for Women’s and Children’s Health, Institute of Translational Medicine, University of Liverpool

2. Professor Judith Bliss, Director of ICR-Clinical Trials and Statistics Unit, The Institute of Cancer Research

3. Professor David Fitzmaurice, Professor of Primary Care Research, Department of Primary Care Clinical Sciences, University of Birmingham

4. Professor John W Gregory, Professor in Paediatric Endocrinology, Department of Child Health, Wales School of Medicine, Cardiff University

5. Professor Steve Halligan, Professor of Gastrointestinal Radiology, Department of Specialist Radiology, University College Hospital, London

6. Professor Angela Harden, Professor of Community and Family Health, Institute for Health and Human Development, University of East London

7. Dr Joanne Lord, Reader, Health Economics Research Group, Brunel University

8. Professor Stephen Morris, Professor of Health Economics, University College London, Research Department of Epidemiology and Public Health, University College London

9. Professor Dion Morton, Professor of Surgery, Academic Department of Surgery, University of Birmingham

10. Professor Gail Mountain, Professor of Health Services Research, Rehabilitation and Assistive Technologies Group, University of Sheffield

11. Professor Irwin Nazareth, Professor of Primary Care and Head of Department, Department of Primary Care and Population Sciences, University College London

12. Professor E Andrea Nelson, Professor of Wound Healing and Director of Research, School of Healthcare, University of Leeds

13. Professor John David Norrie, Director, Centre for Healthcare Randomised Trials, Health Services Research Unit, University of Aberdeen

14. Professor Barney Reeves, Professorial Research Fellow in Health Services Research, Department of Clinical Science, University of Bristol

15. Professor Peter Tyrer, Professor of Community Psychiatry, Centre for Mental Health, Imperial College London

16. Professor Martin Underwood, Professor of Primary Care Research, Warwick Medical School, University of Warwick

17. Professor Caroline Watkins, Professor of Stroke and Older People’s Care, Chair of UK Forum for Stroke Training, Stroke Practice Research Unit, University of Central Lancashire

18. Dr Duncan Young, Senior Clinical Lecturer and Consultant, Nuffield Department of Anaesthetics, University of Oxford

1. Dr Tom Foulks, Medical Research Council

2. Dr Kay Pattison, Senior NIHR Programme Manager, Department of Health

HTA Clinical Evaluation and Trials Board

1. Director, Warwick Clinical Trials Unit, Warwick Medical School, University of Warwick and Professor of Rehabilitation, Nuffield Department of Orthopaedic, Rheumatology and Musculoskeletal Sciences, University of Oxford

2. Professor of the Psychology of Health Care, Leeds Institute of Health Sciences, University of Leeds

3. Director, NIHR HTA programme, Professor of Clinical Pharmacology, University of Liverpool

1. Professor Keith Abrams, Professor of Medical Statistics, Department of Health Sciences, University of Leicester

2. Professor Martin Bland, Professor of Health Statistics, Department of Health Sciences, University of York

3. Professor Jane Blazeby, Professor of Surgery and Consultant Upper GI Surgeon, Department of Social Medicine, University of Bristol

4. Professor Julia M Brown, Director, Clinical Trials Research Unit, University of Leeds

5. Professor Alistair Burns, Professor of Old Age Psychiatry, Psychiatry Research Group, School of Community-Based Medicine, The University of Manchester & National Clinical Director for Dementia, Department of Health

6. Dr Jennifer Burr, Director, Centre for Healthcare Randomised trials (CHART), University of Aberdeen

7. Professor Linda Davies, Professor of Health Economics, Health Sciences Research Group, University of Manchester

8. Professor Simon Gilbody, Prof of Psych Medicine and Health Services Research, Department of Health Sciences, University of York

9. Professor Steven Goodacre, Professor and Consultant in Emergency Medicine, School of Health and Related Research, University of Sheffield

10. Professor Dyfrig Hughes, Professor of Pharmacoeconomics, Centre for Economics and Policy in Health, Institute of Medical and Social Care Research, Bangor University

11. Professor Paul Jones, Professor of Respiratory Medicine, Department of Cardiac and Vascular Science, St George‘s Hospital Medical School, University of London

12. Professor Khalid Khan, Professor of Women’s Health and Clinical Epidemiology, Barts and the London School of Medicine, Queen Mary, University of London

13. Professor Richard J McManus, Professor of Primary Care Cardiovascular Research, Primary Care Clinical Sciences Building, University of Birmingham

14. Professor Helen Rodgers, Professor of Stroke Care, Institute for Ageing and Health, Newcastle University

15. Professor Ken Stein, Professor of Public Health, Peninsula Technology Assessment Group, Peninsula College of Medicine and Dentistry, Universities of Exeter and Plymouth

16. Professor Jonathan Sterne, Professor of Medical Statistics and Epidemiology, Department of Social Medicine, University of Bristol

17. Mr Andy Vail, Senior Lecturer, Health Sciences Research Group, University of Manchester

18. Professor Clare Wilkinson, Professor of General Practice and Director of Research North Wales Clinical School, Department of Primary Care and Public Health, Cardiff University

19. Dr Ian B Wilkinson, Senior Lecturer and Honorary Consultant, Clinical Pharmacology Unit, Department of Medicine, University of Cambridge

1. Ms Kate Law, Director of Clinical Trials, Cancer Research UK

2. Dr Morven Roberts, Clinical Trials Manager, Health Services and Public Health Services Board, Medical Research Council

Diagnostic Technologies and Screening Panel

1. Scientific Director of the Centre for Magnetic Resonance Investigations and YCR Professor of Radiology, Hull Royal Infirmary

2. Professor Judith E Adams, Consultant Radiologist, Manchester Royal Infirmary, Central Manchester & Manchester Children’s University Hospitals NHS Trust, and Professor of Diagnostic Radiology, University of Manchester

3. Mr Angus S Arunkalaivanan, Honorary Senior Lecturer, University of Birmingham and Consultant Urogynaecologist and Obstetrician, City Hospital, Birmingham

4. Dr Diana Baralle, Consultant and Senior Lecturer in Clinical Genetics, University of Southampton

5. Dr Stephanie Dancer, Consultant Microbiologist, Hairmyres Hospital, East Kilbride

6. Dr Diane Eccles, Professor of Cancer Genetics, Wessex Clinical Genetics Service, Princess Anne Hospital

7. Dr Trevor Friedman, Consultant Liason Psychiatrist, Brandon Unit, Leicester General Hospital

8. Dr Ron Gray, Consultant, National Perinatal Epidemiology Unit, Institute of Health Sciences, University of Oxford

9. Professor Paul D Griffiths, Professor of Radiology, Academic Unit of Radiology, University of Sheffield

10. Mr Martin Hooper, Public contributor

11. Professor Anthony Robert Kendrick, Associate Dean for Clinical Research and Professor of Primary Medical Care, University of Southampton

12. Dr Nicola Lennard, Senior Medical Officer, MHRA

13. Dr Anne Mackie, Director of Programmes, UK National Screening Committee, London

14. Mr David Mathew, Public contributor

15. Dr Michael Millar, Consultant Senior Lecturer in Microbiology, Department of Pathology & Microbiology, Barts and The London NHS Trust, Royal London Hospital

16. Mrs Una Rennard, Public contributor

17. Dr Stuart Smellie, Consultant in Clinical Pathology, Bishop Auckland General Hospital

18. Ms Jane Smith, Consultant Ultrasound Practitioner, Leeds Teaching Hospital NHS Trust, Leeds

19. Dr Allison Streetly, Programme Director, NHS Sickle Cell and Thalassaemia Screening Programme, King’s College School of Medicine

20. Dr Matthew Thompson, Senior Clinical Scientist and GP, Department of Primary Health Care, University of Oxford

21. Dr Alan J Williams, Consultant Physician, General and Respiratory Medicine, The Royal Bournemouth Hospital

1. Dr Tim Elliott, Team Leader, Cancer Screening, Department of Health

2. Dr Joanna Jenkinson, Board Secretary, Neurosciences and Mental Health Board (NMHB), Medical Research Council

3. Professor Julietta Patrick, Director, NHS Cancer Screening Programme, Sheffield

4. Dr Kay Pattison, Senior NIHR Programme Manager, Department of Health

5. Professor Tom Walley, CBE, Director, NIHR HTA programme, Professor of Clinical Pharmacology, University of Liverpool

6. Dr Ursula Wells, Principal Research Officer, Policy Research Programme, Department of Health

Disease Prevention Panel

1. Professor of Epidemiology, University of Warwick Medical School, Coventry

2. Dr Robert Cook, Clinical Programmes Director, Bazian Ltd, London

3. Dr Colin Greaves, Senior Research Fellow, Peninsula Medical School (Primary Care)

4. Mr Michael Head, Public contributor

5. Professor Cathy Jackson, Professor of Primary Care Medicine, Bute Medical School, University of St Andrews

6. Dr Russell Jago, Senior Lecturer in Exercise, Nutrition and Health, Centre for Sport, Exercise and Health, University of Bristol

7. Dr Julie Mytton, Consultant in Child Public Health, NHS Bristol

8. Professor Irwin Nazareth, Professor of Primary Care and Director, Department of Primary Care and Population Sciences, University College London

9. Dr Richard Richards, Assistant Director of Public Health, Derbyshire County Primary Care Trust

10. Professor Ian Roberts, Professor of Epidemiology and Public Health, London School of Hygiene & Tropical Medicine

11. Dr Kenneth Robertson, Consultant Paediatrician, Royal Hospital for Sick Children, Glasgow

12. Dr Catherine Swann, Associate Director, Centre for Public Health Excellence, NICE

13. Mrs Jean Thurston, Public contributor

14. Professor David Weller, Head, School of Clinical Science and Community Health, University of Edinburgh

1. Ms Christine McGuire, Research & Development, Department of Health

2. Dr Kay Pattison, Senior NIHR Programme Manager, Department of Health

3. Professor Tom Walley, CBE, Director, NIHR HTA programme, Professor of Clinical Pharmacology, University of Liverpool

External Devices and Physical Therapies Panel

1. Consultant Physician North Bristol NHS Trust

2. Reader in Wound Healing and Director of Research, University of Leeds

3. Professor Bipin Bhakta, Charterhouse Professor in Rehabilitation Medicine, University of Leeds

4. Mrs Penny Calder, Public contributor

5. Dr Dawn Carnes, Senior Research Fellow, Barts and the London School of Medicine and Dentistry

6. Dr Emma Clark, Clinician Scientist Fellow & Cons. Rheumatologist, University of Bristol

7. Mrs Anthea De Barton-Watson, Public contributor

8. Professor Nadine Foster, Professor of Musculoskeletal Health in Primary Care Arthritis Research, Keele University

9. Dr Shaheen Hamdy, Clinical Senior Lecturer and Consultant Physician, University of Manchester

10. Professor Christine Norton, Professor of Clinical Nursing Innovation, Bucks New University and Imperial College Healthcare NHS Trust

11. Dr Lorraine Pinnigton, Associate Professor in Rehabilitation, University of Nottingham

12. Dr Kate Radford, Senior Lecturer (Research), University of Central Lancashire

13. Mr Jim Reece, Public contributor

14. Professor Maria Stokes, Professor of Neuromusculoskeletal Rehabilitation, University of Southampton

15. Dr Pippa Tyrrell, Senior Lecturer/Consultant, Salford Royal Foundation Hospitals’ Trust and University of Manchester

16. Dr Nefyn Williams, Clinical Senior Lecturer, Cardiff University

1. Dr Kay Pattison, Senior NIHR Programme Manager, Department of Health

2. Dr Morven Roberts, Clinical Trials Manager, Health Services and Public Health Services Board, Medical Research Council

3. Professor Tom Walley, CBE, Director, NIHR HTA programme, Professor of Clinical Pharmacology, University of Liverpool

4. Dr Ursula Wells, Principal Research Officer, Policy Research Programme, Department of Health

Interventional Procedures Panel

1. Professor of Vascular Surgery, University of Sheffield

2. Consultant Colorectal Surgeon, Bristol Royal Infirmary

3. Mrs Isabel Boyer, Public contributor

4. Mr Sankaran Chandra Sekharan, Consultant Surgeon, Breast Surgery, Colchester Hospital University NHS Foundation Trust

5. Professor Nicholas Clarke, Consultant Orthopaedic Surgeon, Southampton University Hospitals NHS Trust

6. Ms Leonie Cooke, Public contributor

7. Mr Seumas Eckford, Consultant in Obstetrics & Gynaecology, North Devon District Hospital

8. Professor Sam Eljamel, Consultant Neurosurgeon, Ninewells Hospital and Medical School, Dundee

9. Dr Adele Fielding, Senior Lecturer and Honorary Consultant in Haematology, University College London Medical School

10. Dr Matthew Hatton, Consultant in Clinical Oncology, Sheffield Teaching Hospital Foundation Trust

11. Dr John Holden, General Practitioner, Garswood Surgery, Wigan

12. Dr Fiona Lecky, Senior Lecturer/Honorary Consultant in Emergency Medicine, University of Manchester/Salford Royal Hospitals NHS Foundation Trust

13. Dr Nadim Malik, Consultant Cardiologist/Honorary Lecturer, University of Manchester

14. Mr Hisham Mehanna, Consultant & Honorary Associate Professor, University Hospitals Coventry & Warwickshire NHS Trust

15. Dr Jane Montgomery, Consultant in Anaesthetics and Critical Care, South Devon Healthcare NHS Foundation Trust

16. Professor Jon Moss, Consultant Interventional Radiologist, North Glasgow Hospitals University NHS Trust

17. Dr Simon Padley, Consultant Radiologist, Chelsea & Westminster Hospital

18. Dr Ashish Paul, Medical Director, Bedfordshire PCT

19. Dr Sarah Purdy, Consultant Senior Lecturer, University of Bristol

20. Dr Matthew Wilson, Consultant Anaesthetist, Sheffield Teaching Hospitals NHS Foundation Trust

21. Professor Yit Chiun Yang, Consultant Ophthalmologist, Royal Wolverhampton Hospitals NHS Trust

1. Dr Kay Pattison, Senior NIHR Programme Manager, Department of Health

2. Dr Morven Roberts, Clinical Trials Manager, Health Services and Public Health Services Board, Medical Research Council

3. Professor Tom Walley, CBE, Director, NIHR HTA programme, Professor of Clinical Pharmacology, University of Liverpool

4. Dr Ursula Wells, Principal Research Officer, Policy Research Programme, Department of Health

Pharmaceuticals Panel

1. Professor in Child Health, University of Nottingham

2. Senior Lecturer in Clinical Pharmacology, University of East Anglia

3. Dr Martin Ashton-Key, Medical Advisor, National Commissioning Group, NHS London

4. Dr Peter Elton, Director of Public Health, Bury Primary Care Trust

5. Dr Ben Goldacre, Research Fellow, Division of Psychological Medicine and Psychiatry, King’s College London

6. Dr James Gray, Consultant Microbiologist, Department of Microbiology, Birmingham Children’s Hospital NHS Foundation Trust

7. Dr Jurjees Hasan, Consultant in Medical Oncology, The Christie, Manchester

8. Dr Carl Heneghan, Deputy Director Centre for Evidence-Based Medicine and Clinical Lecturer, Department of Primary Health Care, University of Oxford

9. Dr Dyfrig Hughes, Reader in Pharmacoeconomics and Deputy Director, Centre for Economics and Policy in Health, IMSCaR, Bangor University

10. Dr Maria Kouimtzi, Pharmacy and Informatics Director, Global Clinical Solutions, Wiley-Blackwell

11. Professor Femi Oyebode, Consultant Psychiatrist and Head of Department, University of Birmingham

12. Dr Andrew Prentice, Senior Lecturer and Consultant Obstetrician and Gynaecologist, The Rosie Hospital, University of Cambridge

13. Ms Amanda Roberts, Public contributor

14. Dr Gillian Shepherd, Director, Health and Clinical Excellence, Merck Serono Ltd

15. Mrs Katrina Simister, Assistant Director New Medicines, National Prescribing Centre, Liverpool

16. Professor Donald Singer, Professor of Clinical Pharmacology and Therapeutics, Clinical Sciences Research Institute, CSB, University of Warwick Medical School

17. Mr David Symes, Public contributor

18. Dr Arnold Zermansky, General Practitioner, Senior Research Fellow, Pharmacy Practice and Medicines Management Group, Leeds University

1. Dr Kay Pattison, Senior NIHR Programme Manager, Department of Health

2. Mr Simon Reeve, Head of Clinical and Cost-Effectiveness, Medicines, Pharmacy and Industry Group, Department of Health

3. Dr Heike Weber, Programme Manager, Medical Research Council

4. Professor Tom Walley, CBE, Director, NIHR HTA programme, Professor of Clinical Pharmacology, University of Liverpool

5. Dr Ursula Wells, Principal Research Officer, Policy Research Programme, Department of Health

Psychological and Community Therapies Panel

1. Professor of Psychiatry, University of Warwick, Coventry

2. Consultant & University Lecturer in Psychiatry, University of Cambridge

3. Professor Jane Barlow, Professor of Public Health in the Early Years, Health Sciences Research Institute, Warwick Medical School

4. Dr Sabyasachi Bhaumik, Consultant Psychiatrist, Leicestershire Partnership NHS Trust

5. Mrs Val Carlill, Public contributor

6. Dr Steve Cunningham, Consultant Respiratory Paediatrician, Lothian Health Board

7. Dr Anne Hesketh, Senior Clinical Lecturer in Speech and Language Therapy, University of Manchester

8. Dr Peter Langdon, Senior Clinical Lecturer, School of Medicine, Health Policy and Practice, University of East Anglia

9. Dr Yann Lefeuvre, GP Partner, Burrage Road Surgery, London

10. Dr Jeremy J Murphy, Consultant Physician and Cardiologist, County Durham and Darlington Foundation Trust

11. Dr Richard Neal, Clinical Senior Lecturer in General Practice, Cardiff University

12. Mr John Needham, Public contributor

13. Ms Mary Nettle, Mental Health User Consultant

14. Professor John Potter, Professor of Ageing and Stroke Medicine, University of East Anglia

15. Dr Greta Rait, Senior Clinical Lecturer and General Practitioner, University College London

16. Dr Paul Ramchandani, Senior Research Fellow/Cons. Child Psychiatrist, University of Oxford

17. Dr Karen Roberts, Nurse/Consultant, Dunston Hill Hospital, Tyne and Wear

18. Dr Karim Saad, Consultant in Old Age Psychiatry, Coventry and Warwickshire Partnership Trust

19. Dr Lesley Stockton, Lecturer, School of Health Sciences, University of Liverpool

20. Dr Simon Wright, GP Partner, Walkden Medical Centre, Manchester

1. Dr Kay Pattison, Senior NIHR Programme Manager, Department of Health

2. Dr Morven Roberts, Clinical Trials Manager, Health Services and Public Health Services Board, Medical Research Council

3. Professor Tom Walley, CBE, Director, NIHR HTA programme, Professor of Clinical Pharmacology, University of Liverpool

4. Dr Ursula Wells, Principal Research Officer, Policy Research Programme, Department of Health