Among adults with insomnia symptoms, sleep disturbance, or primary insomnia, what are the effects of single orally administered bioactive compounds, standardized botanical extracts, amino acids, minerals, peptides, and nutraceuticals versus placebo or active control on objectively measured sleep-onset latency and sleep architecture, and separately on validated subjective sleep outcomes?
Single-ingredient supplements including melatonin, ashwagandha, standardized valerian, and botanical extracts objectively reduce sleep-onset latency by 6-34 minutes and improve sleep efficiency by 3-7% with varied sleep architecture effects, while subjective sleep quality improvements are demonstrated across a broader range of compounds, though efficacy depends critically on formulation standardization, dosing, and population selection.
Abstract
Thirty-nine randomized controlled trials evaluated single-ingredient supplements for adult insomnia, including 22 trials with polysomnography or actigraphy (Tier 1 evidence) [1–22] and 17 with validated subjective outcomes only (Tier 2 evidence) [23–39]. Prolonged-release melatonin 2mg reduced sleep onset latency by 9 minutes [6] in older adults with primary insomnia, while low-dose melatonin 0.5mg advanced sleep onset by 34 minutes in delayed sleep-wake phase disorder [4]. Ashwagandha extracts consistently reduced sleep onset latency by 10-12 minutes and increased sleep efficiency by 6-8% across two formulations [8, 10]. Valerian 200mg standardized to ≥2% valerenic acid improved sleep onset latency and total sleep time after 2-8 weeks [1, 18], but lower concentrations and shorter durations showed null results [9]. Rice bran extract, lemon verbena, Melissa officinalis Phytosome, saffron, and GABA from rice germ demonstrated improvements in sleep onset latency (5.7-34 minutes), total sleep time (22-84 minutes), and sleep efficiency (3-7%) [2, 3, 12, 13, 15, 16]. Sleep architecture effects were heterogeneous: Melissa officinalis increased slow-wave sleep by 15% while decreasing REM by 10% [2], whereas lemon verbena increased both deep sleep and REM [14]. Seventeen Tier 2 studies reported significant improvements in Pittsburgh Sleep Quality Index or Insomnia Severity Index scores with magnesium [39], saffron [30], vitamin E [38], and botanical extracts [31, 33], though effect sizes and clinical meaningfulness varied. Adverse events were predominantly mild gastrointestinal symptoms and headache, with no serious events attributed to active interventions [1, 6, 8, 30]. Heterogeneity in findings reflected formulation differences (immediate versus prolonged-release, standardization levels), dose-response relationships, population characteristics (primary insomnia versus specialized conditions), and study quality factors including sample size and outcome measurement methods.
Flow Diagram
Paper search
Elicit Corpus
We performed a semantic search across over 138 million academic papers from the Elicit search engine, which includes all of Semantic Scholar and OpenAlex.
We ran these queries:
- "adult insomnia sleep disturbance oral dietary supplement nutraceutical randomized placebo polysomnography actigraphy sleep onset latency wake after sleep onset"
- "adult insomnia standardized botanical extract herbal medicine oral randomized placebo polysomnography actigraphy sleep architecture sleep quality"
The searches returned 500 total results from Elicit.
PubMed Corpus
We performed a keyword search across the PubMed corpus.
We ran this query:
- (adult insomnia OR sleep disturbance) AND (melatonin OR amino acid OR GABA OR magnesium OR zinc OR peptide OR casein hydrolysate) AND randomized AND (polysomnography OR actigraphy OR sleep onset latency OR PSQI OR ISI)
The search returned 250 total results from PubMed.
We retrieved 621 papers most relevant to the query for screening.
Screening
Abstract screening
We screened in sources based on their abstracts that met these criteria:
- Randomized design: Is this a randomized controlled trial, including a randomized crossover trial?
- Primary completed-trial report: Is this a primary report of a completed trial, rather than a review, protocol, editorial, erratum, or conference abstract?
- Adult sleep-disturbance population: Does the title or abstract indicate adults with insomnia, sleep disturbance, poor sleep, or sleep-onset difficulty?
- Eligible single oral intervention: Does the title or abstract indicate a single orally administered nutrient, amino acid, peptide/hydrolysate, isolated phytochemical, or standardized botanical extract rather than a multi-ingredient formula, non-oral intervention, or prescription hypnotic?
Papers that failed any strict criterion were automatically excluded. For the remaining papers, we considered all screening questions together and made a holistic judgement about whether to screen in each paper.
132 papers passed abstract screening and moved to full-text screening.
At abstract screening, the number of papers excluded for each primary reason was:
- Randomized design: n = 9
- Primary completed-trial report: n = 55
- Adult sleep-disturbance population: n = 140
- Eligible single oral intervention: n = 283
- Other / below screening threshold: n = 2
Full-text screening
We then screened papers based on their full text using these additional criteria:
- Eligible population: Are participants adults with insomnia symptoms, sleep disturbance, or primary insomnia, with no excluded secondary condition or setting unless eligible data are separately reported?
- Eligible intervention: Is the intervention a single, orally administered, identifiable nutrient, amino acid, peptide/protein hydrolysate, isolated phytochemical, or standardized botanical extract with stated dose/formulation, not a multi-ingredient product or prescription hypnotic?
- Eligible trial comparison: Is this a completed randomized parallel or crossover trial with placebo or active comparator?
- Eligible outcome tier: Does the report include either Tier 1 PSG/actigraphy-derived SOL, WASO, TST, sleep efficiency, N3/slow-wave sleep, or REM; or Tier 2 only validated subjective PSQI, ISI, standardized subjective SOL, or standardized sleep logs?
- Primary full report: Is this a full primary report with usable trial methods and results, not a duplicate-only secondary publication, protocol, review, editorial, erratum, or conference abstract?
Papers that failed any strict criterion were automatically excluded. For the remaining papers, we considered all screening questions together and made a holistic judgement about whether to include each paper in the final analysis.
39 papers passed full-text screening and moved to data extraction.
At full-text screening, the number of papers excluded for each primary reason was:
- Eligible population: n = 5
- Eligible intervention: n = 4
- Eligible trial comparison: n = 1
- Eligible outcome tier: n = 1
- No full text: n = 82
Data extraction
We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.
- Evidence tier: Classify as exactly one: Tier 1 objective, Tier 2 subjective-only, or Ineligible. Tier 1 requires PSG or actigraphy; Tier 2 has only validated subjective outcomes.
- Trial identity and design: Extract trial name/acronym if available, registration identifier, parallel/crossover design, randomization/blinding, and follow-up duration.
- Population: Extract sample size randomized/analyzed, age, sex, diagnostic/symptom definition, baseline sleep severity, and any condition relevant to eligibility.
- Intervention identity: Extract exact single substance or standardized extract, plant part/active marker if relevant, formulation, dose, frequency, and administration timing.
- Comparator: Extract placebo or active comparator and matched dosing schedule.
- Objective outcomes: For each PSG/actigraphy outcome, extract device/method, SOL, WASO, TST, sleep efficiency, N3/slow-wave sleep, REM, measurement timepoint, units, and direction of benefit. State not reported if absent.
- Objective arm-level data: For every objective outcome, extract intervention and comparator n, mean baseline, mean follow-up or change, and SD, SE, or CI, including whether results are adjusted. Do not infer missing values.
- Objective comparative estimates: For every objective outcome, extract reported between-group mean difference, 95% CI, p value, and effect size (Cohen d/Hedges g) if reported. State not reported if absent.
- Subjective outcomes: Extract validated subjective instrument (PSQI, ISI, standardized subjective SOL, standardized sleep log), timepoint, and direction of benefit.
- Subjective arm-level and comparative data: For each subjective outcome, extract intervention/comparator n, means/change and SD/SE/CI, plus between-group difference, 95% CI, p value, and effect size if reported. State not reported if absent.
- Sleep architecture: Extract N3/slow-wave sleep and REM duration or percentage data separately, with units, timepoint, arm-level values and comparative statistics where reported.
- Safety and daytime sedation: Extract adverse events, withdrawals due to adverse events, and residual daytime sedation/psychomotor performance measures, with quantitative data where reported.
- Pharmacokinetics and timing: Extract Tmax, absorption route or BBB information, and time-to-bed administration only when reported in the study; otherwise state not reported.
- RoB 2 domains: Assess risk of bias separately for randomization process, deviations from intended intervention, missing outcome data, measurement of outcome, and selection of reported result. Give each domain Low risk, Some concerns, or High risk with brief evidence-based rationale.
- EU regulatory context: Extract any stated information about food supplement, food, novel food, medicinal-product context, or authorised EU health claim. Do not infer regulatory status when the source does not state it.
Results
Characteristics of Included Studies
This review identified 39 randomized controlled trials evaluating single-ingredient supplements for adult insomnia. Twenty-two trials included objective sleep measures (Tier 1 evidence) [1–22], while 17 trials reported only validated subjective outcomes (Tier 2 evidence) [23–39].
| Study | Full text retrieved? | Evidence Tier | Design | Population | Intervention | Comparator | Duration | Registration |
|---|---|---|---|---|---|---|---|---|
| Khalid et al., 2024 [23] | Yes | Tier 2 [23] | Parallel RCT, single-blind [23] | 290 adults with diabetes and insomnia (93 male, 197 female) [23] | Mg 250mg, K 250mg, or Mg+K 250mg monthly [23] | Placebo (starch) [23] | 2 months [23] | NCT04642313 [23] |
| Shekhar et al., 2023 [1] | Yes | Tier 1 [1] | Parallel RCT, double-blind [1] | 72 adults aged 18-50 with sleep complaints (PSQI≥5) [1] | Valeriana officinalis 200mg (2% valerenic acid), 1h before bed [1] | Placebo (microcrystalline cellulose) [1] | 8 weeks [1] | CTRI/2022/05/042818 [1] |
| Nielsen et al., 2010 [24] | Yes | Tier 2 [24] | Parallel RCT, double-blind [24] | 96 adults aged 51-85 with poor sleep (PSQI>5) [24] | Magnesium citrate 320mg/day [24] | Sodium citrate placebo [24] | 7 weeks [24] | NCT00833092 [24] |
| Di Pierro et al., 2024 [2] | Yes | Tier 1 [2] | Crossover RCT, double-blind [2] | 30 adults aged 18-65 with unrefreshing sleep [2] | Melissa officinalis Phytosome 400mg, 30min before bed [2] | Placebo [2] | 2 weeks each arm [2] | NCT05950932 [2] |
| Esmaeilzadeh et al., 2025 [25] | Yes | Tier 2 [25] | Parallel RCT, triple-blind [25] | 80 infertile women aged 18-45 with endometriosis and PSQI≥5 [25] | Melatonin 5mg prolonged-release, 2h before bed [25] | Placebo (starch) [25] | 2 months [25] | IRCT20171209037794N4 [25] |
| Um et al., 2019 [3] | Yes | Tier 1 [3] | Parallel RCT, double-blind [3] | 45 adults with PSQI≥5 [3] | Rice bran extract 1000mg (4.5mg/g γ-oryzanol), 30-60min before bed [3] | Placebo (maltodextrin) [3] | 2 weeks [3] | KCT0001893 [3] |
| Sletten et al., 2018 [4] | Yes | Tier 1 [4] | Parallel RCT, double-blind [4] | 116 adults aged 16-65 with DSWPD and delayed melatonin rhythm [4] | Melatonin 0.5mg fast-release, 1h before desired bedtime [4] | Placebo [4] | 4 weeks [4] | ACTRN12612000425897 [4] |
| Baskett et al., 2003 [5] | Yes | Tier 1 [5] | Crossover RCT, double-blind [5] | 34 adults aged ≥65 with age-related sleep problems [5] | Melatonin 5mg immediate-release at bedtime [5] | Placebo (glucose) [5] | 4 weeks each arm [5] | Not reported [5] |
| Luthringer et al., 2009 [6] | Yes | Tier 1 [6] | Parallel RCT, double-blind [6] | 40 adults aged ≥55 with primary insomnia (DSM-IV) [6] | Prolonged-release melatonin 2mg, 2h before bed [6] | Placebo [6] | 3 weeks [6] | Not reported [6] |
| Gendy et al., 2020 [26] | Yes | Tier 2 [26] | Parallel RCT, double-blind [26] | 60 adults with alcohol use disorder and sleep problems [26] | Melatonin 5mg fast-dissolving, 1h before bed [26] | Placebo [26] | 4 weeks [26] | NCT03043443 [26] |
| Kim et al., 2021 [27] | Yes | Tier 2 [27] | Parallel RCT, double-blind [27] | 38 women aged ≥55 with insomnia (PSQI≥5) [27] | Melatonin 2mg prolonged-release, 2h before sleep [27] | Placebo [27] | 6 weeks [27] | YUHS 4-2016-0418 [27] |
| Ha et al., 2019 [7] | Yes | Tier 1 [7] | Parallel RCT, double-blind [7] | 80 adults aged 20-65 with mild insomnia [7] | Polygonatum sibiricum 500mg, 30-60min before bed [7] | Placebo [7] | 4 weeks [7] | NCT03337789 [7] |
| Zick et al., 2011 [28] | Yes | Tier 2 [28] | Parallel RCT, double-blind [28] | 34 adults aged 18-65 with DSM-IV primary insomnia ≥6 months [28] | Chamomile 270mg twice daily (2.5mg α-bisabolol, ≥2.5mg apigenin) [28] | Placebo [28] | 28 days [28] | NCT01286324 [28] |
| Deshpande et al., 2020 [8] | Yes | Tier 1 [8] | Parallel RCT, double-blind [8] | 144 adults aged 18-65 with non-restorative sleep (RSQ-W≤50) [8] | Ashwagandha 120mg (Shoden) once daily, 2h before evening meal [8] | Placebo (rice powder) [8] | 6 weeks [8] | CTRI/2017/02/007801 [8] |
| Taibi et al., 2009 [9] | Yes | Tier 1 [9] | Crossover RCT, double-blind [9] | 16 women aged 55-80 with insomnia (PSQI≥5) [9] | Valerian 300mg (0.8% valerenic acid), 30min before bed [9] | Placebo [9] | 2 weeks each arm [9] | Not reported [9] |
| Langade et al., 2019 [10] | Yes | Tier 1 [10] | Parallel RCT, double-blind [10] | 58 adults aged 18-60 with DSM-IV insomnia [10] | Ashwagandha 300mg (KSM-66) twice daily [10] | Placebo (starch) [10] | 10 weeks [10] | Not reported [10] |
| Oxman et al., 2007 [29] | Yes | Tier 2 [29] | Parallel RCT, double-blind [29] | 405 adults aged 18-75 with insomnia [29] | Valerian 600mg (3×200mg), 1h before bed [29] | Placebo [29] | 14 days [29] | ISRCTN72748991 [29] |
| Schuster et al., 2025 [30] | Yes | Tier 2 [30] | Parallel RCT, double-blind [30] | 165 adults aged 18-65 with moderate insomnia (RIS>12) [30] | Saffron 20mg or 30mg (Safr'Inside), 60min before bed [30] | Placebo (maltodextrin) [30] | 4 weeks [30] | DRKS00033435 [30] |
| Di Minno et al., 2025 [31] | Yes | Tier 2 [31] | Crossover RCT, double-blind [31] | 66 adults aged 18-70 with mild-moderate primary insomnia (ISI<22) [31] | Scutellaria lateriflora 400mg (10% baicalin) daily [31] | Placebo [31] | 56 days each arm [31] | ISRCTN12126820 [31] |
| Pigeon et al., 2010 [32] | Yes | Tier 2 [32] | Crossover RCT, double-blind [32] | 15 adults aged ≥65 with chronic insomnia (ISI≥10) [32] | Tart cherry juice 16oz/day (morning and 1-2h before bed) [32] | Placebo beverage [32] | 2 weeks each arm [32] | Not reported [32] |
| Jiang et al., 2015 [11] | Yes | Tier 1 [11] | Parallel RCT, double-blind [11] | 42 postmenopausal women aged 45-60 with sleep disturbance [11] | Black cohosh 40mg/day (20mg crude drug per tablet) after meals [11] | Placebo [11] | 6 months [11] | Not reported [11] |
| Pérez-Piñero et al., 2024 [12] | Yes | Tier 1 [12] | Parallel RCT, double-blind [12] | 71 adults aged ≥18 with sleep disturbances [12] | Lemon verbena 400mg (≥24% verbascoside), 1h before sleep [12] | Placebo (cellulose) [12] | 90 days [12] | NCT06154629 [12] |
| Pachikian et al., 2021 [13] | Yes | Tier 1 [13] | Parallel RCT, double-blind [13] | 66 adults aged 25-70 with mild-moderate insomnia (ISI 7-21) and anxiety [13] | Saffron 15.5mg/day (Saffr'activ) in evening [13] | Placebo (maltodextrin) [13] | 6 weeks [13] | NCT04750681 [13] |
| Martínez-Rodríguez et al., 2022 [14] | Yes | Tier 1 [14] | Parallel RCT, double-blind [14] | 40 adults with stress (PSS>15) and poor sleep (PSQI>5) [14] | Lemon verbena 400mg (≥28% phenylpropanoids), 1-2h before sleep [14] | Placebo [14] | 8 weeks + 4-week washout [14] | Not reported [14] |
| Bano et al., 2023 [33] | Yes | Tier 2 [33] | Parallel RCT, double-blind [33] | 100 adults aged 18-65 with moderate emotional distress or PSQI>5 [33] | Melissa officinalis 400mg/day (17-23% hydroxycinnamic acids) after meals [33] | Placebo [33] | 3 weeks [33] | NCT05602688 [33] |
| Byun et al., 2018 [15] | Yes | Tier 1 [15] | Parallel RCT, double-blind [15] | 40 adults with insomnia symptoms (PSQI≥5, ISI≥8) [15] | GABA 300mg from rice germ, 1h before sleep [15] | Placebo (maltodextrin) [15] | 4 weeks [15] | Not reported [15] |
| Mahmoudi et al., 2020 [34] | Yes | Tier 2 [34] | Parallel RCT, double-blind [34] | 96 postmenopausal women aged 45-65 with PSQI<5 [34] | Jujube seed 250mg twice daily at bedtime [34] | Placebo [34] | 21 days [34] | IRCT20180307038992N1 [34] |
| Losso et al., 2017 [16] | Yes | Tier 1 [16] | Crossover RCT, double-blind [16] | 8 adults aged ≥50 with ICSD-2 insomnia [16] | Tart cherry juice 240ml twice daily (morning and 1-2h before bed) [16] | Placebo beverage [16] | 2 weeks each arm [16] | NCT01669317 [16] |
| Rao et al., 2021 [17] | Yes | Tier 1 [17] | Parallel RCT, double-blind [17] | 103 adults with PSQI>5 [17] | Palmitoylethanolamide 350mg (Levagen+), 1h before sleep [17] | Placebo (maltodextrin) [17] | 8 weeks [17] | ACTRN12618001339246 [17] |
| Thomas, 2024 [18] | Yes | Tier 1 [18] | Parallel RCT, double-blind [18] | 72 adults with mild insomnia [18] | Valeriana officinalis 200mg (2% valerenic acid), 1h before sleep [18] | Placebo (cellulose) [18] | 56 days [18] | CTRI/2022/05/042818 [18] |
| Nagase et al., 2020 [35] | Yes | Tier 2 [35] | Crossover RCT, double-blind [35] | 20 adults aged 28-73 with PSQI 6-10 [35] | Porcine placental extract 300mg/day [35] | Placebo [35] | 2 weeks each arm [35] | UMIN000026468 [35] |
| Breus et al., 2024 [19] | Yes | Tier 1 [19] | Crossover RCT, double-blind [19] | 31 adults with nonclinical insomnia symptoms [19] | Magnesium 1g/day (Upgraded Magnesium) [19] | Placebo [19] | 2 weeks each arm [19] | ISRCTN 70584524 [19] |
| Hausenblas et al., 2024 [20] | Yes | Tier 1 [20] | Parallel RCT, double-blind [20] | 56 adults with nonclinical poor sleep [20] | Dichrostachys glomerata 300mg/day (DYG-400) [20] | Placebo (rice protein) [20] | 60 days [20] | ISRCTN10099861 [20] |
| Kim et al., 2019 [21] | Yes | Tier 1 [21] | Crossover RCT, double-blind [21] | 43 adults with mild-moderate sleep disturbance [21] | Alpha-s1 casein hydrolysate 300mg/day (Lactium), 1h before bed [21] | Placebo (maltodextrin) [21] | 4 weeks each arm [21] | KCT0001867 [21] |
| Rao & Briskey, 2025 [36] | Yes | Tier 2 [36] | Crossover RCT, double-blind [36] | 26 adults with difficulty initiating sleep (>20min) [36] | Pistachio-derived melatonin 1mg (Prosomnial), 1h before bed [36] | Synthetic melatonin 1mg [36] | 7 days each arm [36] | Not reported [36] |
| Baradari et al., 2018 [37] | Yes | Tier 2 [37] | Parallel RCT, double-blind [37] | 53 ICU nurses with PSQI>5 and zinc deficiency [37] | Zinc sulfate 220mg every 72h before sleep [37] | Placebo (starch) [37] | 1 month [37] | Not reported [37] |
| Thongchumnum et al., 2023 [38] | Yes | Tier 2 [38] | Parallel RCT, double-blind [38] | 160 postmenopausal women with chronic insomnia disorder [38] | Vitamin E 400 units (mixed tocopherol) daily [38] | Placebo [38] | 1 month [38] | TCTR20220405002 [38] |
| Schuster et al., 2025a [39] | Yes | Tier 2 [39] | Parallel RCT, double-blind [39] | 155 adults aged 18-65 with poor sleep (RIS>12) [39] | Magnesium bisglycinate 250mg elemental Mg, 30-60min before bed [39] | Placebo (cellulose) [39] | 4 weeks [39] | DRKS00031494 [39] |
| Thomas et al., 2023 [22] | Yes | Tier 1 [22] | Crossover RCT, double-blind [22] | 13 physically active males aged 24±4 with sleep complaints (AIS≥6) [22] | Collagen peptides 15g/day, 1h before bed [22] | Placebo [22] | 7 nights each arm [22] | osf.io/vjf8y [22] |
Studies were predominantly conducted in adults with primary insomnia or sleep complaints, though specialized populations included postmenopausal women [11, 34, 38], individuals with diabetes [23], endometriosis [25], alcohol use disorder [26], delayed sleep-wake phase disorder [4], and physically active males [22]. Sample sizes ranged from 8 [16] to 405 [29] participants. Most trials employed parallel-group designs, with 10 using crossover methodology [2, 5, 9, 16, 19, 21, 22, 31, 32, 35, 36]. Intervention durations varied from single-dose assessment [18] to 6 months [11], with most trials lasting 2-8 weeks.
Effects on Objective Sleep Outcomes
Melatonin
Five trials evaluated melatonin formulations using polysomnography or actigraphy. Prolonged-release melatonin 2mg administered 2 hours before bedtime reduced sleep onset latency by 9 minutes (p=0.02) in 40 adults aged ≥55 years with primary insomnia [6]. No significant changes in sleep architecture were observed [6]. In adults with delayed sleep-wake phase disorder, melatonin 0.5mg given 1 hour before desired bedtime advanced sleep onset by 34 minutes (95% CI -60 to -8) and increased sleep efficiency in the first third of the night by 5.14% (95% CI 2.52-7.76) [4]. Actigraphic sleep onset latency decreased by 11.9 minutes (95% CI -19.54 to -4.39) [4].
In contrast, immediate-release melatonin 5mg at bedtime showed no significant effects on sleep parameters in a crossover trial of 34 older adults with age-related sleep problems [5]. Normal sleepers experienced a median decrease of 4.4 awakenings, but problem sleepers showed no benefit [5].
Ashwagandha
Two trials evaluated standardized ashwagandha root extracts with actigraphy. In 144 adults with non-restorative sleep, Shoden extract (120mg/day containing 21mg withanolide glycosides) significantly improved sleep efficiency (p<0.01), total sleep time (p<0.001), sleep onset latency (p<0.01), and wake after sleep onset (p<0.05) after 6 weeks [8]. Self-reported sleep quality increased by 72% versus 29% in placebo (p<0.001) [8].
KSM-66 extract (300mg twice daily) in 58 adults with DSM-IV insomnia reduced sleep onset latency from 41.61±6.84 to 29.00±7.14 minutes in the intervention group versus 41.94±6.98 to 33.94±7.65 minutes in placebo (p=0.019) [10]. Sleep efficiency increased from 75.63±2.70% to 83.48±2.83% versus 75.14±3.73% to 79.68±3.59% in placebo [10].
Valerian
Three trials assessed Valeriana officinalis with contrasting results. A single-dose study of 200mg extract (2% valerenic acid) in 72 adults with sleep complaints significantly increased actual sleep time by 22.84±9.82 minutes versus -39.28±9.13 minutes with placebo (p<0.05) [18]. Over 8 weeks, the same formulation reduced sleep latency by 10.89±4.98 minutes and increased total sleep time by 75.21±34.05 minutes [1]. Polysomnography after 56 days showed significant improvements in total sleep time, sleep latency, and sleep efficiency (all p<0.05) [1].
However, a crossover trial in 16 older women using 300mg valerian extract (0.8% valerenic acid) found no improvement in sleep latency, wake after sleep onset, or sleep efficiency after 2 weeks [9]. Wake after sleep onset actually increased by 17.7±25.6 minutes compared to baseline (p=0.02) [9].
Botanical Extracts
Rice bran extract standardized to 4.5mg/g γ-oryzanol (1000mg/day) in 45 adults with sleep disturbance decreased sleep latency (adjusted p=0.047), increased total sleep time (p=0.019), and improved sleep efficiency (p=0.010) after 2 weeks [3]. Stage 2 sleep duration increased significantly (p=0.022) [3].
Polygonatum sibiricum rhizome extract (500mg/day) improved total sleep time by actigraphy after 4 weeks (group-by-visit interaction p=0.046) [7]. Melissa officinalis formulated as a Phytosome preparation (400mg/day) reduced Insomnia Severity Index scores from 9.7±3.7 to 6.8±4.1, a mean difference of 2.9 points (p=0.003) [2]. Slow-wave sleep duration increased by 15% while REM decreased by 10% [2].
Lemon verbena extract (400mg/day, ≥24% verbascoside) administered for 90 days improved PSQI sleep latency (1.6±1.0 vs 1.9±0.7, p=0.027) and sleep efficiency (84.5±12.8 vs 79.8±13.6, p=0.023) [12]. Actigraphy showed significant improvements in latency, efficiency, and wakefulness (p=0.001) [12]. A parallel study using the same extract demonstrated increased deep sleep and REM percentages by wearable device tracking [14].
Saffron extract (15.5mg/day) increased time in bed by 16 minutes compared to baseline and by 25 minutes versus placebo (p=0.023) [13]. Black cohosh standardized extract (40mg/day) administered for 6 months increased sleep efficiency and decreased wake after sleep onset duration by 15.8% [11].
Amino Acids and Peptides
GABA extracted from rice germ (300mg/day) reduced sleep latency from 13.4±15.7 to 5.7±6.2 minutes (p=0.001) and increased sleep efficiency from 79.4±12.9% to 86.1±10.5% (p=0.018) after 4 weeks [15]. Alpha-s1 casein hydrolysate (300mg/day) showed significant improvements in total sleep time, sleep efficiency, sleep latency, and wake after sleep onset during the active phase in sleep diaries (all p<0.001), though polysomnography showed no significant between-group differences [21].
Collagen peptides (15g/day) reduced polysomnographic awakenings from 29.3±13.8 to 21.3±9.7 counts (p=0.028) [22]. Subjective awakenings decreased from 1.9±0.6 to 1.3±1.5 (p=0.023) [22]. Baseline Stroop test performance was higher with collagen (1.00±0.00) than placebo (0.97±0.05, p=0.009) [22].
Other Compounds
Tart cherry juice (240ml twice daily) increased polysomnographic sleep time by 84 minutes (p=0.0182) and improved sleep efficiency (p=0.03) in 8 adults with insomnia [16]. Palmitoylethanolamide (350mg/day) reduced sleep onset latency at weeks 4 and 8 [17]. Dichrostachys glomerata extract (300mg/day) significantly improved sleep score, deep sleep duration, sleep latency, and time awake over 60 days [20].
Summary of Objective Effects
Among Tier 1 studies with objective measures, sleep onset latency improvements ranged from 5.7 minutes [15] to 34 minutes [4]. Total sleep time increases ranged from 75 minutes [1] to 84 minutes [16]. Sleep efficiency gains ranged from 3-7% across interventions [8, 10, 11, 15]. Effects on sleep architecture were inconsistently reported, with some studies showing increases in slow-wave sleep [2, 14] and others finding no significant changes [6].
Effects on Subjective Sleep Outcomes
Tier 2 Studies (Subjective Only)
Seventeen trials reported validated subjective outcomes without objective sleep measures. Magnesium citrate (320mg/day) improved overall PSQI scores from 10.4 to 6.6 regardless of treatment assignment (p<0.0001), but magnesium supplementation specifically decreased plasma C-reactive protein in participants with baseline values >3.0mg/L [24]. Magnesium bisglycinate (250mg elemental magnesium) reduced ISI scores by 3.9 points (95% CI -5.8 to -2.0) versus 2.3 points with placebo (95% CI -4.1 to -0.4, p=0.049, Cohen's d=0.2) [39].
Melatonin trials in specialized populations showed mixed results. In 80 infertile women with endometriosis, 5mg prolonged-release melatonin improved overall sleep quality by 1.7 points (95% CI -2.63 to -0.77, p<0.001, Cohen's d=1) and reduced chronic pelvic pain [25]. However, in 60 adults with alcohol use disorder, 5mg melatonin showed no significant difference from placebo in PSQI reduction [26]. In 38 women aged ≥55 years, 2mg prolonged-release melatonin improved PSQI from 11 to 8 (p=0.01), though no significant between-group difference was observed [27].
Saffron extract (20-30mg/day) reduced Athens Insomnia Scale scores by 1.96 points with 30mg and 1.54 points with 20mg versus 0.71 points with placebo (effect size d=0.45) [30]. Sleep quality improved significantly after 3 weeks (p<0.05 for 30mg, p<0.01 for 20mg) [30]. Perceived stress decreased more with both saffron doses than placebo (p=0.01) [30].
Scutellaria lateriflora extract (400mg/day) significantly decreased PSQI scores and improved sleep onset latency, sleep efficiency, and total sleep time in a crossover design [31]. Melissa officinalis phospholipid formulation (400mg/day) significantly improved depression, anxiety, stress, positive and negative affect, mental wellbeing, quality of life, and sleep quality (all p<0.001) [33].
Valerian 600mg (3×200mg tablets) showed a marginally significant benefit, with 5.5% more participants reporting better or much better sleep (95% CI 0.2 to 10.8, p=0.04) [29]. Chamomile extract (270mg twice daily) showed trends favoring treatment for sleep latency (Cohen's d=0.47) and night awakenings (Cohen's d=0.61), though not statistically significant [28].
Jujube seed capsules (250mg twice daily) improved PSQI scores more than placebo in 96 postmenopausal women (p<0.001) [34]. Vitamin E (400 units/day) reduced PSQI scores from 13 to 6 in the intervention group versus 11 to 9 in placebo (p=0.012) [38]. The improvement score was significantly higher with vitamin E (5 vs 1, p<0.001) [38]. Zinc sulfate (220mg every 72 hours) significantly lowered PSQI total scores and improved subjective sleep quality and sleep latency (p=0.008) [37].
Tart cherry juice in a pilot crossover trial (n=15) significantly reduced Insomnia Severity Index and wake after sleep onset (both p<0.05) but showed no improvement in sleep latency, total sleep time, or sleep efficiency compared to placebo [32]. Porcine placental extract (300mg/day) improved subjective sleep depth and wellness on the St. Mary's Hospital Sleep Questionnaire but showed no objective effects [35].
Pistachio-derived melatonin (Prosomnial 1mg) showed comparable absorption to synthetic melatonin and significantly improved sleep latency, total sleep time, sleep quality, and feeling rested over 7 days [36].
Safety and Tolerability
Adverse events were generally mild and infrequent across studies. The most common reported events included gastrointestinal discomfort [30, 34], headache [6, 15], and daytime drowsiness [4, 15]. No serious adverse events were attributed to active interventions in most trials [1, 2, 7, 8, 10, 11, 16, 19, 38].
Melatonin was associated with light-headedness, daytime sleepiness, and decreased libido in one study [4], though rates were similar between treatment groups [4]. Two participants withdrew from an older adult study due to excessive drowsiness [5]. Prolonged-release melatonin 2mg showed no impairment in psychomotor performance, with motor reaction time 60ms faster than placebo (p=0.004) [6].
Valerian was generally well-tolerated with minor side effects [9]. Chamomile reported 6 adverse events versus 10 with placebo, all mild and transient [28]. Ashwagandha studies reported no treatment-related adverse events [8, 10]. Saffron extract caused mild gastrointestinal discomfort in 10 cases (30mg group), 6 cases (20mg group), and 4 cases (placebo) [30]. One withdrawal occurred due to palpitations in a saffron trial [13].
Alpha-s1 casein hydrolysate caused one adverse event (itching and urticaria) in the placebo group, with no events in the active group [21]. GABA from rice germ caused mild abdominal discomfort, headache, and drowsiness in 10% of subjects [15]. Collagen peptides and palmitoylethanolamide reported no adverse effects [17, 22].
Synthesis
Heterogeneity in Melatonin Findings
Melatonin trials demonstrate substantial heterogeneity in outcomes that can be explained by dose-response relationships, formulation differences, and population characteristics. Prolonged-release formulations (2mg) consistently showed efficacy in older adults with primary insomnia [6, 27], with Tmax occurring at 3 hours [6] enabling sustained nighttime concentrations. The 9-minute reduction in sleep onset latency with prolonged-release 2mg [6] aligns with regulatory approval thresholds and produced no psychomotor impairment [6].
In contrast, immediate-release formulations yielded inconsistent results dependent on population and dose. Fast-release 0.5mg administered 1 hour before desired bedtime showed pronounced efficacy in delayed sleep-wake phase disorder (34-minute advance in sleep onset) [4], a population with documented circadian phase delays where melatonin's chronobiotic effects are mechanistically relevant [4]. However, immediate-release 5mg at bedtime failed to benefit older adults with age-related sleep problems [5], potentially reflecting the brief half-life (42.6 minutes) [4] in a population requiring sustained sleep maintenance rather than phase shifting.
The failure of 5mg melatonin in alcohol use disorder patients [26] may reflect distinct pathophysiology—alcohol disrupts sleep architecture through GABAergic mechanisms that melatonin does not address [26]. Conversely, 5mg prolonged-release melatonin's large effect size (Cohen's d=1) in endometriosis patients [25] likely reflects dual benefits on both sleep and pain, as melatonin demonstrates anti-inflammatory properties relevant to pelvic pain [25].
Valerian's Contradictory Evidence
Valerian trials present a methodological hierarchy that explains discordant findings. The null result in older women [9] came from a small crossover trial (n=16) using 300mg extract (0.8% valerenic acid) for only 2 weeks [9, 9]. Wake after sleep onset actually worsened by 17.7 minutes [9], suggesting insufficient dosing or duration for age-related sleep maintenance problems.
In contrast, positive findings emerged from larger, adequately powered trials using higher doses or concentrations. The 200mg extract standardized to 2% valerenic acid (2.5-fold higher active marker) showed immediate single-dose effects on total sleep time (+22.84 minutes, p<0.05) [18] and sustained multi-week improvements in sleep latency (-10.89 minutes) and total sleep time (+75.21 minutes) [1]. These effects manifested only after 14-56 days [1], consistent with accumulation hypotheses for GABAergic herbs.
The 600mg dose in a large web-based trial (n=405) showed marginal benefit (5.5% improvement, p=0.04) [29], suggesting a dose-response plateau. Valerian's apparent efficacy window appears to be 200mg of highly standardized extract (≥2% valerenic acid) with minimum 2-week duration, particularly in younger-to-middle-aged adults with sleep-onset rather than maintenance complaints.
Ashwagandha's Consistent Signal
Both ashwagandha trials demonstrated clinically meaningful improvements despite different extracts and dosing regimens. Shoden (120mg/day, 21mg withanolide glycosides) [8] and KSM-66 (600mg/day) [10] both reduced sleep onset latency by 10-12 minutes [8, 10] and increased sleep efficiency by 6-8% [8, 10]. This consistency across formulations suggests the withanolide class, rather than a specific proprietary extraction, mediates sleep benefits through adaptogenic stress reduction [8, 10]. Both trials enrolled adults with baseline anxiety or stress components [8, 10], indicating ashwagandha may preferentially benefit stress-related insomnia over primary sleep disorders.
Botanical Extract Patterns
Lemon verbena trials consistently improved sleep metrics across both actigraphy [12, 12] and wearable device tracking [14], despite different manufacturers and slightly varying standardization (≥24% vs ≥28% phenylpropanoids) [12, 14]. The 90-day trial showed progressive benefits peaking at endpoint [12], while the 8-week trial with washout demonstrated lasting effects extending beyond cessation [14], suggesting both immediate and cumulative mechanisms.
Melissa officinalis results depend critically on formulation. The Phytosome preparation (phospholipid carrier) designed for enhanced bioavailability showed objective sleep improvements [2, 2], whereas the standard aqueous extract showed only subjective benefits [33]. This formulation-dependence likely reflects the poor oral bioavailability of hydroxycinnamic acids, which Phytosome technology addresses [2].
Saffron's two trials used different formulations (15.5mg vs 20-30mg) and outcome measures (actigraphy vs questionnaires) [13, 30], yet both demonstrated benefits on sleep quality and stress [13, 30]. The dose-response was modest—20mg and 30mg differed minimally [30] —suggesting a threshold effect rather than linear scaling.
Amino Acids and Specialized Compounds
GABA from rice germ showed large improvements in sleep latency (7.7-minute reduction, p=0.001) [15], seemingly contradicting conventional wisdom about GABA's inability to cross the blood-brain barrier [15]. However, the trial did not measure cerebral spinal fluid GABA [15], leaving the mechanism uncertain. The significant PSG-confirmed benefits [15] may reflect peripheral effects on the gut-brain axis or small amounts crossing compromised barriers in sleep-deprived states.
Alpha-s1 casein hydrolysate demonstrated a clear placebo-sensitive split between subjective (significant) and objective (non-significant) outcomes [21]. Sleep diary improvements were pronounced and sustained [21], while polysomnography showed trends without reaching significance [21]. This pattern suggests expectancy effects or that subjective sleep quality improved through anxiolytic mechanisms not captured by sleep architecture measures.
Population-Specific Effects
Postmenopausal women responded distinctly across interventions. Black cohosh's 15.8% reduction in wake after sleep onset [11] aligns with its established effects on vasomotor symptoms that disrupt sleep [11]. Vitamin E's large PSQI improvement (7-point reduction) [38] and reduction in sedative drug use [38] may reflect antioxidant effects on age-related sleep changes, though the trial enrolled women with chronic insomnia disorder rather than healthy aging [38].
Specialized populations showed both enhanced and diminished responses. ICU nurses with zinc deficiency showed significant improvement with zinc supplementation [37], indicating correction of a specific deficiency rather than a primary sleep effect. Physically active males showed reduced awakenings with collagen peptides [22] and improved cognition [22], possibly reflecting recovery-enhancement mechanisms distinct from sedation.
Quality and Design Factors
Study quality substantially influenced effect detection. The three trials failing to show valerian benefits [5, 9, 29] were either underpowered crossover designs (n=16, n=34) [5, 9] or relied solely on self-report in a web-based format [29]. In contrast, adequately powered parallel trials (n=45-80) with objective measures consistently detected effects for valerian [1, 18], ashwagandha [8, 10], and botanical extracts [3, 7, 12].
Crossover designs introduced period effects evident in alpha-s1 casein hydrolysate, where improvements occurred in both phases [21], and in porcine placental extract, where subjective but not objective benefits emerged [35]. Parallel designs avoided these carryover effects while crossovers offered within-subject power advantages when washout periods were adequate (≥2 weeks) [2, 16].
Mechanistic Implications
The divergent effects on sleep architecture provide mechanistic insights. Melissa officinalis increased slow-wave sleep by 15% while decreasing REM by 10% [2], consistent with GABAergic enhancement promoting deeper NREM at REM's expense. Lemon verbena increased both deep sleep and REM [14], suggesting a distinct mechanism possibly involving monoaminergic pathways. Rice bran extract specifically increased stage 2 sleep [3], the most abundant sleep stage, without altering REM or slow-wave proportions.
These architecture differences matter clinically. Interventions preserving or enhancing REM (ashwagandha, lemon verbena) may benefit conditions requiring intact REM-dependent memory consolidation. Those increasing slow-wave sleep (Melissa officinalis, potentially magnesium) may preferentially aid physical recovery. The absence of REM suppression across most interventions [6, 9, 21] distinguishes these compounds from benzodiazepines and Z-drugs, which consistently reduce REM percentage.
