The results of the initial database search yielded 557 references, 39 of which met the inclusion criteria and were subsequently included in the review. Articles were excluded mostly because the intervention was not melatonin or because they did not report on sleep related outcomes. Of these 39 included articles, four [25–28] reported on different outcomes of the same study and were, therefore, combined with their appropriate counterpart. Ultimately, 35 RCTs, with a total of 2,356 subjects, were included in this review (see Figure 1 for the flow chart of included studies).
Table 2 describes the characteristics of the individual studies (grouped by shift workers, jet lag, insomnia, and healthy volunteers) and overall SIGN 50 score. Table 3 describes the GRADE analysis results of the overall literature pool, and Table 4 illustrates the key design issues considered important for dietary interventional studies. Table 5 presents the number and types of subjective and/or objective assessment tools utilized in this review.
Table 2
Characteristics and quality score of included studies
Table 3
Grading of Recommendations Assessment, Development and Evaluation (GRADE) analysis: quality of the overall literature pool assessing melatonin for the promotion of healthy sleep patterns
Table 4
Reporting of dietary supplement design elements
Table 5
Objective and subjective outcome measures captured in the review
Overall quality assessment of individual included studies
The overall methodological quality of the RCTs was evaluated as being of the highest (++) quality, high (+) quality or poor (−) quality, according to the SIGN 50 criteria indicated in Table 1 (see Table 2 for quality scores). The majority (80%) of the 35 included RCTs were high (+) quality, with one study (3%) being of the highest (++) quality [34]. Conversely, 17.0% of studies were scored as poor quality. An appropriate and clearly focused research question was adequately addressed in 80.0% of the trials; the remaining (20.0%) addressed this area well. Over half (51.0%) of the studies had dropout rates less than 10.0% and were, therefore, considered well-covered for this criterion, whereas 31.0% of studies either did not mention dropout rates or reported rates greater than 20.0%. Regarding intention-to-treat analysis, 54.0% of the included studies were classified as poorly addressed because such analyses were not mentioned or described; 3.0% of the studies adequately addressed intention-to-treat analysis; and 43.0% addressed it well. Although the majority (86.0%) of studies poorly described their methods of allocation concealment, all (100%) studies adequately addressed blinding methods regarding treatment allocation. Methods of randomization were described poorly by 48.0% of the studies; 46.0% adequately described this process and only 6.0% did it well. Nearly all (97.0%) and the majority (55.0%) of the trials adequately addressed differences between treatment groups and baseline similarities, respectively. Results indicated that 51.0% of articles adequately covered outcome validity and reliability, whereas 37.0% covered this criterion well. Three studies [28, 29, 62] were multi-site, and one [28] poorly addressed the comparability of results among sites; the remaining two were well covered [29] or adequately addressed [62] in this area.
Adverse events
Of the total 35 studies included in our analysis, 15 [29, 30, 33, 34, 36, 38, 39, 41–47, 52] included information on adverse events. No serious adverse events were reported. One [36] study reported that adverse events occurred, but did not describe them, and two [34, 45] reported no adverse events occurred at all. The most common adverse events were headache [29, 33, 44, 46, 47] and somnolence [33, 41, 44, 48]. Palpitations [42, 43] and abdominal pain [33, 43] were each reported in two studies. The remaining adverse events were reported infrequently, and each occurred in only one of the multiple studies: nasopharyngitis [46], arthralgia [46], tachycardia [39], dizziness [33], nausea [33], vomiting [33], nightmares [33], difficulty swallowing and breathing [38], hypnotic activity [39], heavy head [39], heartburn [43], flatulence [43], swelling of arms/legs [43], sweating/hot flash [43], exanthema [43], sleeping difficulties [44], depression [44], problems with the rectal probe [52], and sleep walking [42].
Effectiveness of melatonin for promoting healthy sleep outcomes
Included studies were categorized according to the intended use of melatonin in 1) shift workers and individuals with jet lag to rebalance the sleep-wake cycle; 2) persons with insomnia to promote sleep; and 3) in healthy volunteers to improve outcomes of sleep efficacy, somnolence, and/or hormonal phase shift changes (see Table 2 for full description of included studies and Table 3 for GRADE Analysis).
Shift workers
Eight [29–36] RCTs with 300 total participants assessed the efficacy of melatonin for promoting sleep in shift workers. The majority of studies were of either high (+) [29, 30, 33, 35, 36] [or highest (++) quality [34], with two [31, 32] poor (−) quality RCTs suffering from inadequate reporting of dropout rates, concealment methods, and baseline differences between groups. Results indicated that both of the poor quality studies favored melatonin [31, 32] however, all of the high and highest quality studies were inconclusive [29, 30, 33–36] in that they favored neither melatonin nor the control. Based on the five [29, 30, 33, 34, 36] studies that reported adverse events, melatonin appears to be relatively safe, with frequent but not serious adverse events and interactions. Despite its apparent safety, and the general high quality of this literature pool, sample sizes were generally small, and results inconclusive, with no magnitude of an estimate of effect size reported. Consequently, the SMEs were not able to give any recommendation for the use of melatonin in shift workers at this time.
Jet Lag
Eight [37–44] RCTs with 972 total participants characterized melatonin use for counteracting jet lag. Almost all of the studies were of high (+) quality [37, 39–44], with the exception of one poor (−) quality study [38], which favored neither melatonin nor control, despite a large sample size (n = 339). Of the seven high (+) quality studies, one [40] favored neither melatonin nor control. The remaining six [37, 39, 41–44] RCTs favored melatonin, including two [42, 43] large studies (n = 320 [44] and n = 160 [41]) and one [39] which noted a limitation that melatonin increased tiredness the next morning. Melatonin appears to be relatively safe based on the six [38, 39, 41–44] studies that reported adverse events, citing occasional, but not serious adverse events and interactions. Based on the high quality and favorable results reported, the SMEs concluded that in a jet lagged population, further research may have an impact on the confidence in the estimate of the effect, and as such, provide a weak recommendation in favor of melatonin use for rebalancing the sleep-wake cycle in people with jet lag.
Insomnia
Four studies of high (+) quality [45–48] with 845 total participants assessed the efficacy of melatonin in promoting better sleep in persons with insomnia. Two [45, 48] of these studies favored neither melatonin nor control, while the remaining two [46, 47] including one large study (n = 791) [46] favored melatonin, Similar to the results in jet lag studies, limitations in sample size compromised the power to produce an effect in populations with insomnia. Despite the trend in small sample sizes and lack of effect size reporting, all four studies were high quality, showing positive effects and infrequent, non-serious adverse events; as a result, the SMEs give a weak recommendation in favor of melatonin when used to promote sleep in persons with insomnia, with the understanding that the introduction of more large, high quality studies may have an important impact on this recommendation, and potentially change the confidence in the estimate of the effect size.
Healthy volunteers
Fifteen [28, 49–61, 63] RCTs with a total of 223 participants described melatonin use for promoting sleep in healthy volunteers. Of the 15 total studies, 12 [49–53, 55–59, 62, 63] were high (+) quality and the remaining three [54, 60, 61] were poor (−) quality. Two [54, 61]of the poor quality studies favored melatonin, whereas the third [60] favored neither melatonin nor control. Of the high quality studies, eight [49, 51–53, 56–58, 62] indicated favorable results for melatonin, although six [51–53, 57, 58, 62] had small sample sizes (n = 6 to 23; total subjects for six studies = 71). The remaining four [50, 55, 59, 63] high quality studies showed no beneficial effects for either melatonin or control groups.
Healthy volunteers were furthered subdivided into three groups based on the sleep outcome under evaluation: initiation of sleep/sleep efficacy [36, 49–51, 53–55], occurrence of daytime sleepiness/somnolence [50, 51, 56–58], and induction of phase shift/hormone changes [59–63]. Two studies [50, 51] included both initiation of sleep/sleep efficacy and occurrence of daytime sleepiness/somnolence outcomes, and were consequently included in both categories. Results for each group are described below.
Initiation of sleep/sleep efficacy
All except one [54] of the seven studies investigating the effect of melatonin on initiation of sleep or sleep efficacy were scored high quality, and five [49, 51–54] of them showed results in favor of melatonin. Because only one [52] study in this group reported on adverse events, citing a problem with the rectal probe, safety is not well understood. Similarly, effect sizes were not reported. Despite the lack of safety and effect size reporting and small sample sizes, however, most of the studies were high quality, reporting favorable results for melatonin use. Subsequently, the SMEs provide a weak recommendation in favor of melatonin use in a healthy population for promoting sleep.
Daytime sleepiness/somnolence
All five studies investigating daytime sleepiness or somnolence were high quality, and four [51, 56–58] of the five small studies favored melatonin over the control. The one study not favoring melatonin [50] was poorly powered, with a sample size of (n = 10 subjects). Because no information was reported on the frequency or severity of adverse events in any of these studies, safety is not well understood. Although this group of studies suffered from small sample sizes, methodological quality was high. As a result, the SMEs provided a weak recommendation in favor of melatonin use to improve daytime sleepiness in healthy people.
Phase shift/hormone changes
The five studies investigating the effects of a nighttime dose of melatonin on phase shift/ hormone changes in healthy populations were more physiologically-based with primary outcomes being a change in the biomarkers being studied, and had severe limitations in study quality compared to the other two groups. Two [60, 61] of the five studies were low quality due to methodological flaws in reporting of randomization, concealment, and dropout rates. The remaining three [59, 62, 63] studies were high quality; however, the sample sizes for all five studies were fairly low. Because neither adverse events nor effect sizes were reported in any of the studies, this information remains unknown. Given this lack of information, the SMEs could not provide any recommendation for the use of melatonin to improve hormonal phase shift changes in healthy people.
Additional dietary supplement design elements
The authors looked at additional design elements thought to be important for understanding the specific effects related to dietary supplements (Table 4). None of the studies in shift workers or jet lagged populations reported information on baseline diet exposures, but two [46, 47] studies on insomnia and two [52, 62]in healthy populations reported this information. Four [38–40, 43] jet lag studies controlled for background diets during the study, compared to nine [49, 51, 53, 57–59, 61–63] studies in healthy volunteers, none in insomnia, and three [29, 33, 35] studies in shift workers. Several studies reported that they did not control for background diets: one jet lag [42]; one insomnia [46]; one shift worker [34], and two [52, 55] healthy volunteer studies; the remaining did not report on this information. In four [29, 40, 53, 57] studies, subjects abstained from caffeine and three [43, 58, 62] studies allowed, but limited caffeine use. Formulation of the melatonin for the intervention was described in 57% of the papers, including five [29, 30, 33, 34, 36] shift worker, one [40] jet lag, three [45–47] insomnia, and 11 [49, 50, 52, 53, 55–57, 59, 60, 62, 63] healthy volunteer studies. Melatonin supplement purity was analyzed in one [33] shift worker, two [38, 40] jet lag, no insomnia, and three [50, 53, 60] healthy volunteer studies. Finally, analysis of proper absorption of melatonin was conducted in one [35] shift worker, one [37] jet lag, two [45, 46] insomnia, and nine [45, 49, 52, 53, 55, 58–60, 63] healthy volunteer studies.
Outcome measures
A total of 31 unique assessment tools, including 19 subjective and 12 objective measures, were utilized to measure sleep outcomes (see Table 5). Although three of the 12 objective measures were of great interest, they were not relevant outcomes of interest in this review - melatonin measurements in saliva, blood, and urine. Thirteen studies [32, 35, 36, 40, 42, 45, 48, 49, 53, 54, 58, 59, 62] used a combination of both subjective and objective assessment tools to evaluate outcomes and 85.0% of these studies received high quality scores. Twenty studies [29–31, 33, 34, 37–39, 41, 42, 44, 46, 47, 50–52, 55–57, 61], 80% of which included high quality scores, used only subjective assessment tools; four [50, 51, 60, 63] studies, three of which were high quality, used only objective assessment tools.
Dosing as reported in the literature
The amount of melatonin provided, and frequency of administration reported in the included studies varied greatly. Oral preparations were used in amounts ranging from 0.3 to 10.0 mg/day. All except two [53, 61] studies used capsules, with 72% failing to indicate the type of capsule (e.g. hard or soft) used. Two [33, 43] studies utilized fast-release preparations in amounts ranging from 3.0 to 5.0 mg. Six [35, 45, 46, 49, 62, 63] studies utilized a sustained-release formulation in amounts ranging from 0.3 to 6.0 mg, and one [53] study utilized a patch preparation providing 2.1 mg. Only one [61] study utilized a drink preparation, where 2 mg of melatonin was provided in 100 ml of 1% ethanol in water.