Oliveira, 2018¹²

Portugal

RCT

2 primary study eligible for this report

Studies published between January 1970 to July 2017 were searched

N = 39

Inclusion criteria: Adults ≥18 years with Schizophrenia and comorbid insomnia

Included participants had Schizophrenia and Insomnia defined using DSM-IV in one study and outpatients with stable Schizophrenia and initial insomnia of ≥2 weeks in the other

Intervention:

Melatonin

Daily dose: 2 mg of controlled release and 3–12 mg standard melatonin

Duration of the included studies were 7 weeks (included 2 × 3 week intervention periods with one week between in one study, and 15 days in the other study

Comparator:

No change in treatment

Sleep Outcomes

Sleep efficiency measured by Actigraphy

Quality and depth of nighttime sleep assessed by unspecified questionnaire

No description of measures, minimal clinically important difference, or description of what direction represents clinical improvements or worsening was provided.

Follow-up for 3 days following each treatment period

Auld, 2017⁵

UK

Placebo-controlled RCTs: single or double-blind; cross-over or parallel

6 studies examined primary insomnia (only 5 included in meta-analysis)

Studies published between 1950 and 2015 were searched

N = 1,315

Inclusion criteria: Adults ≥18 years with a primary sleep disorder; subsample with primary insomnia is included in this report

Eligible age ranges spanned 18 years to ≥80 years (an upper limit was not provided for 2 studies)

Participant descriptive statistics not provided

Intervention:

Melatonin

Daily doses were: 0.01 mg, 0.3 mg, 1.0 mg, 2.0 mg, and 3.0 mg

Timing was 2 hours before bed, where reported

Study duration ranged from 7 days to 9 weeks, including washout periods, where they were included

Comparator:

Placebo

Doses, frequency, and duration same as intervention group

Sleep Outcomes

Sleep onset latency

No description of measure, minimal clinically important difference, or description of what direction represents clinical improvements or worsening was provided.

Follow-up immediately post intervention

McCleery, 2016¹³

UK

Placebo-controlled RCTs were eligible

1 study was relevant to this report: a multi-centre, single blind, 2-arm parallel treatment group RCT

Studies published between 2012 to March 2016 were searched

N = 13

Inclusion criteria: Adults with dementia (MMSE score ≥15) and a sleep disturbance;

Subsample of 13 patients with comorbid insomnia included in this report

Other information:

• Participants had been taking stable doses of acetylcholinesterase inhibitor with or without memantine for 2 months prior to recruitment;
• Participants were instructed to spend 2 hours per day outdoors during daylight hours

Intervention:

2 mg melatonin PR, once daily, 1 to 2 hours before bed

Comparator:

Placebo

Duration of treatment: 6 months (i.e., 2 weeks run-in phase, 24-week double-blind randomized treatment phase, 2-week placebo run-out phase)

Sleep outcomes

Carer-rated sleep quality

Cognition assessed with ADAS-cog and MMSE

Functioning, Mood, QoL:

IADL

No description of measures, minimal clinically important difference, or description of what direction represents clinical improvements or worsening was provided.

Follow-up immediately post intervention

Brasure, 2015⁹

US

SRs, RCTs, and longitudinal studies were eligible

1 study (published in two manuscripts) was relevant to this report

Studies published between 2004 to January 2015

N = 711

Inclusion criteria for SRs and RCTs: Adults (age ≥18) with insomnia disorder according to unspecified diagnostic criteria; studies with at least 4 weeks follow-up; reported global or sleep outcomes

Inclusion criteria for observational studies that reported harms: adults with chronic insomnia without major comorbidities; ≥6 month duration; N ≥100

Adults with comorbid medical or mental health disorders were eligible

Actual demographic data for the 711 participants included in this report not available. Data for the 722 of the full study: mean age 62 years, 99% white, 69% female

Intervention:

Melatonin prolonged release

Dose = 2mg

Comparator:

Placebo

3 weeks randomized to melatonin or placebo

After 3 weeks, placebo group re-randomized to melatonin or placebo and melatonin group remained on melatonin for 26 week extension. Only extension included in the review (n = 711)

Global outcomes (sleep symptoms and daytime functioning or distress associated with sleep symptoms)

Assessed by PSQI

Sleep outcomes:

Sleep onset latency (minutes); Assessed by sleep diary

AE

Method of assessment not reported

Grima, 2018¹⁰

Australia

Multi-centre, randomized, double-blind, placebo-controlled, two-period, two-treatment crossover

The study was conducted from August 2011 to November 2016

N = 33

Mean age = 37

Female, 33%;

Male, 67%

Median (IQR Q1 to Q3) months since TBI = 46 (13 to 102)

Inclusion criteria: Community dwelling adults (aged 18 to 65) with mild to severe acquired TBI (history of blunt head trauma with loss of consciousness, initial Glasgow Coma Scale of 3 to 14; post-traumatic amnesia)

Sleep complaint corroborated by PSQI global score ≥8 and confirmed diagnosis of chronic insomnia (ICSD, 3^rd ed)

Exclusion criteria: Sleep problems, fatigue or neurological conditions before their TBI; Pregnant; Trans meridian travel across >1 time zone or worked night shifts in previous 3 months; high risk of obstructive sleep apnea; Use of non-prescription sleep medication in previous 6 weeks; Psychoactive substances in previous 12 months

Intervention:

Prolonged-release melatonin

Dose; 2 mg

Frequency: 1/night

Duration: 4 weeks

Timing: ~same time each night within 2 hours of bed time

Comparator:

Placebo

Dose: Mannitol (106 mg)

Acacia (11 mg)

Pure icing sugar (106 mg)

Duration: 4 weeks

Timing: ~same time each night within 2 hours of bed time

Sleep outcomes:

Global sleep outcomes were assessed over the previous over the previous month using the 19-item PSQI global score, which combines subdomains of sleep duration, sleep disturbance, sleep latency, sleep efficiency, daytime dysfunction, overall sleep quality and medication use (Score range 0 to 21) higher values indicate poorer sleep quality

No description of minimal clinically important difference or measurement properties were provided.

Sleep onset latency (time elapsed between the start of the rest interval relative to the sleep start time) was measured using wrist Actigraphy (Actiwatch-2, Phillips Respironics), collected in 1 minute epochs at medium sensitivity; the average value across nights measured was taken

Authors reported validity for wrist Actigraphy generally but not for Actiwatch specifically. Cut points used to classify activity counts as sleep were not provided by authors

No description of minimal clinically important difference, description of what direction represents clinical improvements or worsening were provided

Sleep efficiency was measured by Actigraphy and calculated as total sleep time / total duration of the sleep episode (%) averaged across nights

Daytime sleepiness was assessed with a modified version of the ESS to assess “daytime somnolence in the preceding 4 weeks”

Anxiety was assessed by self-report with the HADS

Depression was assessed by self-report with the HADS

Subjective fatigue was assessed by self-report with the FSS

Health related QoL was assessed by self-report with the SF-36 v1, which included the following subscales: Physical functioning; Role-physical; Role-emotional; Vitality; Mental health; Social functioning; Bodily pain; General health

Adverse events were assessed by showing participants a list of symptoms and asking if they had experienced any in the previous period

Follow-up was immediately post-test

Kurdi, 2016¹¹

India

N = 50

Mean age = 52.42

Female, 48%;

Male, 52%;

Unclear reporting of cancer stage overall and per group

Inclusion criteria: Age: 20 to 65 years Cancer patients; Primary insomnia (DSM IV criteria) with sleep complaints >1 month; not using sleep medication in previous 2 weeks

Exclusion criteria: Pregnancy or lactation; cognitive impairment; History of psychiatric disorders; On antipsychotic treatment; Language or communication difficulties; Other sleep disorders; Liver and renal dysfunction

Intervention:

Melatonin

Dose: 3 mg daily

Timing: 2 hours before bedtime

Duration = 14 days

Comparator:

Multivitamin tablets

Duration = 6 months

Global Sleep Outcomes:

Sleep quality/severity of insomnia assessed with the self-report 8-item AIS. Scales range from 0 (no problem) to 3 (did not sleep); Individual components assessed by AIS (subscales: Sleep induction; Awakenings during the night; Final awakening; Total sleep duration; Sleep quality; Well-being; Functional capacity; Sleepiness during the day)

Measurement properties were not reported

Follow-up was immediately post-test

• The research question and inclusion criteria for the review included all of the components of PICO
• Four databases were searched and key words were provided; reference lists of included studies were hand searched; search was completed within 24 months of completion of the review
• Review authors reported using Cochrane’s Risk of Bias tool to assess the risk of bias in individual studies included in the review
• Review authors reported no competing interests, or described a funding source as a potential conflict. However, the author did not describe how this potential conflict of interest was managed.

• Review report did not contain an explicit statement that the review methods were established prior to the conduct of the review.
• Review authors did not explain their selection of the study designs for inclusion in the review
• Additional exclusions (i.e., language) and the decision to limit the search dates were not justified
• Study selection was not performed in duplication
• Number of data extractors was not reported
• A list of excluded studies was not provided, but reasons for exclusions were
• Included studies were described in partial detail. Usual treatment of comparators, duration of one included intervention, study outcomes and outcome measures, and time frame for follow up were not reported or poorly reported
• Review authors did not report on the sources of funding for the studies included in the review
• Review authors did not account for risk of bias in individual studies when discussing the results of the review
• Review authors reported no competing interests, or described a funding source as a potential conflict. However, the author did not describe how this potential conflict of interest was managed.

• Taken together, the research question and inclusion criteria for the review included all of the components of PICO
• Four databases were searched and key words were provided; no additional publication restrictions were identified; search was conducted within 24 months of completion of review
• Review authors performed study selection in duplicate
• Review authors performed data extraction in duplicate
• A list of all exclusions of studies that were read in full-text form was provided and reasons for exclusions were justified.
• Authors used appropriate methods for statistical combination of results
• Authors assessed the risk of bias in individual studies that were included in the review using the 3-item Jadad scale
• Authors accounted for risk of bias in individual studies when discussing the results
• Review authors discussed reasons for low heterogeneity in the results of the review
• Review authors provided funnel plots and discussed the likelihood (i.e., low) of publication bias
• Review authors indicated no conflict of interest

• There was no explicit statement that the review methods were established prior to the conduct of the review. Several sleep-related study outcomes were extracted from included studies, but only sleep onset latency was synthesized or discussed, suggesting the methods may not have been established prior to the conduct of the review.
• There was no explanation of the selection of study designs for inclusion in the review
• Authors did not report searching reference lists of included studies, trial / study registries, or grey literature
• Included studies superficially described eligible populations, interventions, comparators, outcomes, and research designs. Adequate detail was not provided regarding included participant characteristics, study settings, or timeframe for follow up.
• Authors used the 3-item Jadad scale to assess risk of bias in individual studies, which assesses randomization, blinding, and reporting of withdrawals and dropouts. The Jadad scale has been shown to have poor inter-rater agreement in one study¹⁴ and does not consider allocation concealment, which is described as critical to avoiding bias by the Cochrane Collaboration.¹⁵
• Authors did not report on the sources of funding for the studies included in the review
• Review did not assess the potential impact of risk of bias in individual studies on the results of the meta-analysis

• The research question and inclusion criteria for the review included all of the components of PICO
• The report contained an explicit statement that review methods were published prior to the conduct of the original review, which was published in 2012. The updated review justified deviations from the original protocol and indicated the deviations were made before the updated search was conducted.
• The review authors used a somewhat comprehensive literature search strategy. They searched the ALOIS database, which searches 5 databases and trials registries; the search was conducted within 2 months of completion of the report; grey literature was not included
• The ALOIS database selects studies using a combination of machine learning an manual selection
• Review authors performed data extraction in duplicate
• Review authors provided a list of excluded studies and justified the exclusions
• Review authors described the included studies in detail
• Review authors used Cochrane’s risk of bias tool to assess risk of bias in individual studies included in the review
• Review authors reported the sources of funding for the studies included in the review. Note – the study of relevance to this report was industry funded
• Review authors used appropriate methods for statistical combination. Note - one subsample of one study was eligible for this report
• Review authors accounted for risk of bias in individual studies when interpreting the results
• Heterogeneity is not of relevance to this report as only one study was included
• Authors reported there were insufficient publications to investigate publication bias using graphic tests
• Authors reported no competing interests

• Review authors did not explain their selection of the study designs for inclusion in the review

• The research question and inclusion criteria for the review included all of the components of PICO
• The report contained an explicit statement that review methods were completed prior to the conduct of the review
• Review authors explained their selection of the study designs for inclusion in the review
• Review authors used a comprehensive literature search strategy
• Review authors performed study selection in duplicate
• Review authors performed data extraction in duplicate
• Review authors provided a list of excluded studies and justified the exclusions
• Review authors described the included studies in adequate detail
• Review authors used a satisfactory technique for assessing the risk of bias in individual studies that were included in the review
• Review authors reported the sources of funding for the studies included in the review
• Review authors accounted for risk of bias in individual studies when discussing the results
• Potential for reporting bias was examined and discussed
• Review authors reported potential sources of conflict of interest and how it was addressed

• Allocation sequence was random
• Allocation sequence was concealed until participants were enrolled and assigned to interventions
• Baseline characteristics were not analyzed for statistical differences between those who started in the intervention group and those who started in the comparator group. However, the crossover design was employed to minimize the influence of any between group baseline differences.

• Participants were not aware of their assigned intervention during the trial
• Carers and people delivering the intervention were not aware of participants’ assigned intervention during the trial
• No deviations from the intervention arose because of the experimental context
• Appropriate analysis was used to estimate the effect of assignment to intervention

• Data were available for nearly all participants randomized

• Measurement of the main outcomes was appropriate.
• Measurement of the outcomes could not have differed between intervention groups
• Outcome assessors were not aware of the intervention received by study participants

• The trial was analyzed in accordance with a pre-specified plan that was registered a priori
• The numerical result being assessed is not likely to have been selected, on the basis of the results, from multiple outcome measurements within the outcome domain or multiple analyses of the data

• Secondary outcomes may have been appropriately assessed; however measurement properties were not reported.

• Allocation sequence was reported to be random using the fishbowl method

• Participants were not likely aware of their assigned interventions
• Carers and trial personnel were not likely aware of participants’ assigned intervention during the trial
• No participants were analysed in a group different from the ones to which they were assigned
• Analysis was appropriate to estimate the effect of assignment to the intervention

• Outcome data were available for nearly all participants randomized

• Outcome assessors were not aware of the intervention received by study participants

• Reported outcomes are not likely to have been selected on the basis of the results from multiple outcome measurements within the outcome domain or multiple analyses of the data

• Baseline differences between groups were not statistically assessed. There were similarities between groups that suggest there may have been a problem with the randomization process. I.e., there were equal numbers of males and females in each group and there was no mention of stratification by gender or sex in the allocation process. Additionally, frequencies tables suggest the melatonin group included 19 patients with stage I and II cancer and 19 had stage III and IV cancer, while 6 in the placebo group had each type. The number of patients in each group would add up to 38 intervention and 12 comparator, which is not consistent with the 25 reported participants in each group.
• Allocation concealment was not reported

One study was eligible for this report and narratively described

Sleep efficiency

Assessed in 1 study

Improved sleep efficiency with melatonin (significant)

Subgroup: improved sleep efficiency in poorer sleepers (significant)

Quality and depth of nighttime sleep

Assessed in one study

Relative to placebo, melatonin significantly improved the quality and depth of nighttime sleep without producing hangover.” (p.3)

5 studies included in the meta-analysis

Sleep onset latency

Total MD = −5.05 minutes; 95% CI: −8.51 to 1.59

Overall effect of melatonin

Z = 2.86, P = 0.004

Subgroup of 13 patients from 1 study eligible for this report

Melatonin (n = 7) vs placebo (n = 6)

Carer-rated sleep quality, change from baseline

M(SD) = 5.29 (3.45) vs 2.83 (2.56)

Standard MD = 0.74; 95% CI: −0.40 to 1.89

MMSE, change from baseline

M(SD) = 1.5 (2.9) vs −2.8 (2.9)

Standard MD = 4.30; 95% CI: 0.86 to 7.74

Favours melatonin

ADAS-cog, change from baseline

M(SD) = −2.5 (3.1) vs 1 (6)

Standard MD = −3.50; 95% CI: −9.31 to 2.31

IADL, change from baseline

M(SD) = 0.67 (1.75) vs 1.8 (1.3)

Standard MD = −0.66; 95% CI: −1.90 to 0.58

“From the studies we identified for this review, we found no evidence that melatonin (up to 10mg) helped sleep problems in patients with moderate to severe dementia due to AD.” (p.4)

Note, conclusion applies to the overall study and “sleep problems” includes insomnia among other sleep disorders.

711 patients from 1 study eligible for this report

Global outcomes

PSQI global score (n = 700)

MD = −0.39; 95% CI: −0.71 to −0.08

Sleep Outcomes

Sleep onset latency self-report, minutes (n = 700)

MD = −6; 95% CI: −10 to −2.1

Adverse Effects

Overall withdrawals

MD = 0.87; 95% CI: 0.64); NS

Withdrawals due to AE

MD = 0.86; 95%CI: 0.42 to 1.75; NS

Participants with ≥1 AE

MD = 0.96; 95%CI: 0.87 to 1.06; NS

Melatonin vs placebo

Global Sleep Outcomes

PSQI global (n = 33)

Adjusted mean (95% CI)

7.68 (6.34 to 9.02) vs 9.47 (8.13 to 10.81)

Treatment effect estimate = −1.79 (−2.70 to −0.88)

Cohen’s d = 0.46; P < 0.0001

Sleep Outcomes

Sleep onset latency, minimum (n = 31)

Adjusted mean (95% CI)

1.37 (1.26 to 1.48) vs 1.42 (1.31 to 1.53)

Treatment effect estimate = −0.05 (−0.14 to 0.03)

Cohen’s d = 0.18; P = 0.23

Sleep efficiency (n = 31)

Adjusted mean (95% CI)

−3.22 (−3.61 to −2.82) vs −3.54 (−3.94 to −3.13)

Treatment effect estimate = 0.32 (0.01 to 0.63)

Cohen’s d = 0.28; P = 0.04

Daytime sleepiness, ESS (n = 33)

Adjusted mean (95% CI)

2.36 (2.00 to 2.73) vs 2.53 (2.17 to 2.90)

Treatment effect estimate = −0.17 (−0.40 to 0.06)

Cohen’s d = 0.17; P = 0.15

Functioning, Mood, QoL

HADS anxiety (n = 32)

Adjusted mean (95% CI)

7.84 (6.23 to 9.45) vs 9.00 (7.39 to 10.61)

Treatment effect estimate = −1.15 (−1.97 to −0.34)

Cohen’s d = 0.27; P = 0.006

HADS depression (n = 32)

Adjusted mean (95% CI)

8.53 (6.93 to 10.13) vs 8.34 (6.75 to 9.94)

Treatment effect estimate = 0.18 (−0.70 to 1.07)

Cohen’s d = 0.04; P = 0.03

Fatigue Severity, FSS (n = 32)

Adjusted mean (95% CI)

−4.18 (−4.74 to −3.62) vs −3.73 (−4.28 to −3.17)

Treatment effect estimate = −0.45 (−0.86 to −0.04)

Cohen’s d = 0.29; P = 0.03

Health Related QoL (SF-36) (n = 33):

  Physical functioning

Adjusted mean (95% CI)

43.17 (39.15 to 47.20) vs 41.72 (37.69 to 45.75)

Treatment effect estimate = 1.45 (−0.33 to 3.24)

Cohen’s d = 0.13; P = 0.11

  Role-physical

Adjusted mean (95% CI)

43.17 (39.15 to 47.20) vs 41.72 (37.69 to 45.75)

Treatment effect estimate = 1.45 (−0.33 to 3.24)

Cohen’s d = 0.13; P = 0.11

  Role-emotional

Adjusted mean (95% CI)

37.58 (33.11 to 42.06) vs 36.85 (32.38 to 41.32)

Treatment effect estimate = 0.73 (−3.38 to 4.84)

Cohen’s d = 0.05; P = 0.73

  Vitality

Adjusted mean (95% CI)

42.43 (38.97 to 45.90) vs 38.76 (35.30 to 42.22)

Treatment effect estimate = 3.67 (0.36 to 6.98)

Cohen’s d = 0.35; P = 0.35

  Mental health

Adjusted mean (95% CI)

43.60 (40.00 to 47.24) vs 41.09 (37.45 to 44.73)

Treatment effect estimate = 2.51 (0.58 to 4.42)

Cohen’s d = 0.23; P = 0.01

  Social functioning

Adjusted mean (95% CI)

37.09 (33.05 to 41.13) vs 34.82 (30.78 to 38.86)

Treatment effect estimate = 2.27 (−1.36 to 5.90)

Cohen’s d = 0.19; P = 0.22

  Bodily pain

Adjusted mean (95% CI)

44.07 (39.85 to 48.30) vs 43.27 (39.05 to 47.50)

Treatment effect estimate = 0.80 (−1.39 to 2.99)

Cohen’s d = 0.06; P = 0.48

  General health

Adjusted mean (95% CI)

40.96 (36.95 to 44.97) vs 40.29 (36.28 to 44.30)

Treatment effect estimate = 0.67 (−0.97 to 2.30)

Cohen’s d = 0.06; P = 0.42

Global Sleep Outcomes:

AIS

Melatonin (n = 25) vs placebo (n = 25)

Mean scores at days 1, 7 and 14:

1 day

Mean (SD) = 17.88 (2.03) vs 16.28 (2.62)

t = 2.4136; P = 0.0197

7 days

Mean (SD) = 14.32 (1.49) vs 16.12 (2.52)

t = −3.0714; P = 0.0035

14 days

Mean (SD) = 9.56 (2.58) vs 14.44 (4.69)

t = −4.5562; P = 0.00001

Change scores after 1^st week, 2^nd week, and from day 1 to day 14:

Change from day 1 to day 7

Mean (SD) = 3.56 (2.58) vs 0.16 (0.69)

t = 6.3591; P = 0.00001

Day 8 to day 14

Mean (SD) = 4.76 (2.26) vs 1.68 (5.60)

t = 2.5490; P = 0.0141

Day 1 through day 14

Mean (SD) = 8.32 (3.77) vs 1.84 (5.42)

t = 4.9059; P = 0.00001

Percent sleep improvement:

Day 1 to day 7

19.91% (P = 0.00001) vs .98% (P = 0.2563)

Day 8 to day 14

33.24% (P = 0.00001) vs 10.42% (P = 0.1469)

Day 1 to Day 14

46.53%; P = 0.00001 vs 11.30% (P = 0.1026)