In adults with primary insomnia, sleep disturbance, or elevated psychological/physiological stress, which single-ingredient oral bioactives, standardized botanicals, amino acids, minerals, and heat-killed postbiotics improve objective sleep-onset latency, slow-wave sleep/sleep architecture, or cardiac vagal autonomic indices versus placebo or active control, and what formulation-specific evidence supports hot-beverage sachet delivery?
Standardized ashwagandha (KSM-66 and Shoden®), saffron (Safr'Inside™ 30 mg), higher-standardization valerian (2% valerenic acid), lemon verbena, corn leaf extract, and low-dose GABA (75 mg) improve objective sleep-onset latency or slow-wave sleep architecture versus placebo in capsule or tablet form, while heat-killed L. gasseri CP2305 and palmitoylethanolamide show preliminary cardiac vagal autonomic benefits; no formulation-specific evidence supports hot-beverage sachet delivery for any of these ingredients.
Abstract
Among 55 randomized, placebo-controlled trials evaluating single-ingredient oral bioactives in adults with insomnia, sleep disturbance, or elevated stress, objective sleep-onset latency improvements are most consistently demonstrated by higher-standardization valerian (2% valerenic acid, 200 mg/day over 8 weeks) [1, 2], ashwagandha (Shoden® 120 mg/day and KSM-66 600 mg/day) [3, 4], saffron (Safr'Inside™ 30 mg/day by EEG-based device in older adults) [5], lemon verbena (400 mg/day over 90 days by actigraphy) [6], and corn leaf extract (500 mg/day, large actigraphy/EEG effect sizes) [7]. GABA from fermented rice germ at 300 mg/day improved PSG sleep latency and efficiency within-group [8], and a 75 mg low dose produced a significant PSG-confirmed increase in N3 slow-wave sleep (8.6% to 12.9%; p=0.0043) [9], representing the only direct slow-wave sleep architecture signal in the natural compound evidence base. For cardiac vagal autonomic indices, heat-killed Lactobacillus gasseri CP2305 acutely increased RMSSD and cortical alpha power in a single-dose crossover EEG/HRV study [10, 10], and palmitoylethanolamide (Levagen+® 600 mg/day) significantly increased SDNN over 6 weeks in stressed women (p=0.024) [11], but no study prespecified peri-sleep-onset HRV as a primary endpoint in a population with insomnia or sleep disturbance. Lower-standardization valerian (0.8% valerenic acid, 300 mg, 2 weeks) produced no significant objective sleep benefit and a numerically adverse WASO trend by PSG [12], while melatonin 5 mg in post-acute substance-use withdrawal showed no significant effect on any outcome [13], illustrating that null results cluster around lower standardization, shorter durations, and populations whose sleep disturbance is secondary to comorbid conditions. Heat-killed postbiotics—particularly CP2305, SNK12, and HK-PS23—show convergent HPA-axis suppression (reduced salivary cortisol and CgA) alongside subjective sleep improvements, but significant objective effects on sleep architecture are confined to the 24-week CP2305 EEG study in chronically stressed students [14, 14].
No trial in this evidence base evaluated any target ingredient delivered via a hot-beverage sachet format with post-brew marker recovery, volatile loss assessment, or matrix interaction data [15–17]; thermal processing conditions are documented only for manufacturing steps (extraction temperatures of 80–140°C for botanical concentrates) [6, 7], and the heat-killed postbiotic delivery studies confirm viable-organism absence after sterilization but provide no reconstituted-beverage bioavailability data [17]. Consequently, the available evidence supports several single ingredients as effective for objective sleep onset and, to a lesser extent, slow-wave sleep architecture when delivered in capsule or tablet form at defined standardization levels, but it does not address whether hot-beverage sachet delivery preserves the bioactive content or clinical efficacy of any of these ingredients.
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 onset latency polysomnography actigraphy randomized placebo glycine GABA L-theanine ashwagandha valerian Melissa Passiflora saffron lime peel palmitoylethanolamide"
"adult stress randomized placebo RMSSD HF-HRV cardiac vagal tone PEP palmitoylethanolamide Levagen LipiSperse heat killed Lactobacillus gasseri CP2305 oral"
The searches returned 500 total results from Elicit.
PubMed Corpus
We performed a semantic search across the PubMed corpus.
We ran these queries:
"adult insomnia sleep onset latency polysomnography actigraphy randomized placebo glycine GABA L-theanine ashwagandha valerian Melissa Passiflora saffron lime peel palmitoylethanolamide"
"adult stress randomized placebo RMSSD HF-HRV cardiac vagal tone PEP palmitoylethanolamide Levagen LipiSperse heat killed Lactobacillus gasseri CP2305 oral"
The searches returned 500 total results from PubMed.
We retrieved 675 papers most relevant to the query for screening.
Screening
Abstract screening
We screened in sources based on their abstracts that met these criteria:
- Randomized primary trial: Is this a completed primary report of a randomized human trial, rather than a protocol, review, editorial, conference abstract, or purely preclinical study?
- Eligible adult population: Does the study include adults with primary insomnia, sleep disturbance/non-restorative sleep, or elevated psychological or physiological stress?
- Eligible oral single ingredient: Does the study evaluate an oral single-ingredient chemical bioactive, amino acid, mineral, standardized botanical extract, or explicitly heat-killed/inanimate postbiotic?
- Comparator: Does the randomized study compare the intervention with placebo or an active control?
Papers that failed any criterion were automatically excluded; all other papers, including borderline ones, were screened in.
113 papers passed abstract screening and moved to full-text screening.
At abstract screening, the number of papers excluded for each primary reason was:
- Randomized primary trial: n = 84
- Eligible adult population: n = 204
- Eligible oral single ingredient: n = 258
- Comparator: n = 16
Full-text screening
We then screened papers based on their full text using these additional criteria:
- Eligible design and population: Is this a randomized parallel or crossover trial with separable data for adults aged 18 or over who have insomnia, sleep disturbance, non-restorative sleep, or elevated psychological/physiological stress; excluding pediatric-only reports and severe psychiatric cohorts unless separable eligible data are available?
- Single-ingredient intervention: Is the intervention an oral single ingredient with an identifiable chemical/botanical/strain identity and dose; excluding multi-ingredient products without a separable target-only arm, live probiotic strains, heat-labile enzymes, and nonoral interventions?
- Eligible comparator: Is there a placebo or active comparator arm that permits comparison of the eligible intervention?
- Relevant reported outcome: Does the full report measure at least one relevant sleep, autonomic, stress-physiology, or validated subjective sleep/stress outcome (PSG, sleep EEG, actigraphy, RMSSD, HF-HRV, PEP, PSQI, ISI, VAS-stress, cortisol, or alpha-amylase)?
Papers were included in the final analysis only if they met every criterion; papers that failed or were borderline on any criterion were excluded.
55 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 design and population: n = 2
- Single-ingredient intervention: n = 2
- Relevant reported outcome: n = 2
- Full text not available: n = 52
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.
- Ingredient identity and standardization: Extract exact chemical name or botanical species and plant part; marker percentage/content; proprietary trade name; strain designation for postbiotics; and delivery technology. State Not reported if absent.
- Dose, timing, duration and route: Extract oral dose, dosing frequency, administration timing relative to bedtime, treatment duration, and dosage form.
- Population and setting: Extract randomized and analyzed sample sizes by arm; age; sex; sleep/stress eligibility condition; and exclusions relevant to applicability.
- Trial design and comparator: Extract parallel/crossover design, allocation ratio, blinding, comparator, washout if crossover, and trial registration/protocol identifier.
- Objective sleep outcomes: For each relevant PSG, sleep EEG, or actigraphy outcome, extract arm-level baseline and post-treatment means/SD or change/SD, units, placebo-adjusted effect and p-value for SOL, WASO, TST, sleep efficiency, N3/SWS, delta power, and REM. Do not infer missing values.
- Autonomic outcomes: Extract arm-level results, units, timepoint, placebo-adjusted effect and p-value for RMSSD, HF-HRV, PEP, HR, and other prespecified autonomic outcomes. Do not infer missing values.
- Subjective and neuroendocrine outcomes: Extract validated PSQI, ISI, VAS stress, cortisol, and alpha-amylase results with arm-level values/effect/p-value where reported.
- Effect sizes and uncertainty: Extract reported Cohen d, Hedges g, confidence intervals, standard errors, and p-values for eligible outcomes. If no effect size is reported, state Not reported; do not calculate.
- Adverse events and daytime function: Extract adverse events, serious adverse events, withdrawals for adverse events, residual daytime sedation, daytime sleepiness, and psychomotor/cognitive outcomes.
- Pharmacokinetics and proposed mechanism: Extract human Tmax or bioavailability information if reported, administration-window rationale, and proposed mechanism. Clearly distinguish measured human target engagement from hypothesis.
- Randomization process facts: Extract sequence generation, allocation concealment, and baseline imbalance information relevant to RoB 2 domain 1.
- Deviations from intended intervention facts: Extract participant/personnel blinding, adherence, cointerventions, protocol deviations, and analysis population relevant to RoB 2 domain 2.
- Missing outcome data facts: Extract outcome-specific missingness by arm, reasons, and any sensitivity analyses relevant to RoB 2 domain 3.
- Outcome measurement facts: Extract outcome ascertainment method, assessor blinding, and measurement comparability relevant to RoB 2 domain 4.
- Selective reporting facts: Extract pre-specified analysis plan/registration availability and concordance with reported outcomes relevant to RoB 2 domain 5.
- Funding and conflicts: Extract funding source, sponsor role, conflicts of interest, and author employment.
- Thermal and processing evidence: Extract any ingredient-specific evidence about heat treatment, stability at 60–90 C, matrix interactions, delivery-system integrity, volatile loss, or post-brew assay recovery. State Not reported if not studied; do not infer from clinical efficacy.
Results
Characteristics of Included Studies
The 55 included studies span a wide range of single-ingredient interventions tested against placebo or active control in adults with insomnia, non-restorative sleep, sleep disturbance, or elevated stress. The table below captures ingredient class, primary research question focus, trial design, population, treatment duration, and whether full text was retrieved for each study.
| Study | Full text retrieved? | Ingredient (Proprietary name) | Ingredient class | Primary focus | Design | N randomized | Population | Duration |
|---|---|---|---|---|---|---|---|---|
| Langade et al., 2019 [3] | Yes | Ashwagandha root (KSM-66) [3] | Standardized botanical | Sleep outcomes in insomnia and anxiety [3] | Parallel, double-blind [3] | 60 [3] | Adults 18–60 yrs, DSM-IV insomnia [3] | 10 weeks [3] |
| Deshpande et al., 2020 [4] | Yes | Ashwagandha root/leaf (Shoden®) [4] | Standardized botanical | Sleep quality in non-restorative sleep [4] | Parallel, double-blind [4] | 150 [4] | Healthy adults 18–65 yrs, RSQ-W ≤50 [4] | 6 weeks [4] |
| O'Connor et al., 2022 [18] | Yes | Ashwagandha root (Gaia Herbs) [18] | Standardized botanical | Sleep and stress in college students [18] | Parallel, double-blind [18] | 60 [18] | Healthy college students ≥18 yrs [18] | 30 days [18] |
| Pandit et al., 2024 [19] | Yes | Ashwagandha root/leaf (Sensoril®) [19] | Standardized botanical | Stress reduction; dose-response [19] | Parallel, double-blind [19] | 131 [19] | Adults 20–58 yrs with chronic stress (PSS ≥28) [19] | 8 weeks [19] |
| Choudhary et al., 2017 [20] | Yes | Ashwagandha root (KSM-66) [20] | Standardized botanical | Stress and weight management [20] | Parallel, double-blind [20] | 52 [20] | Adults 18–60 yrs with chronic stress (PSS >20) [20] | 8 weeks [20] |
| Schuster et al., 2025 [21] | Yes | Saffron (Safr'Inside™) [21] | Standardized botanical | Insomnia and stress [21] | Parallel, 3-arm, double-blind [21] | 165 [21] | Adults 18–65 yrs, moderate insomnia [21] | 4 weeks [21] |
| Pachikian et al., 2021 [22] | Yes | Saffron (Saffr'activ®) [22] | Standardized botanical | Sleep quality with mild-to-moderate sleep disorder and anxiety [22] | Parallel, double-blind [22] | 66 [22] | Adults 25–70 yrs, ISI 7–21 [22] | 6 weeks [22] |
| Láng et al., 2025 [5] | Yes | Saffron (Safr'Inside™) [5] | Standardized botanical | Sleep and gut microbiome in older adults [5] | Parallel, double-blind [5] | 52 [5] | Adults 55–85 yrs, PSQI >5 or ISI >10 [5] | 4 weeks [5] |
| Shekhar et al., 2023 [1] | Yes | Valerian (Sleeproot®, 2% valerenic acid) [1] | Standardized botanical | Sleep quality with complaints [1] | Parallel, double-blind [1] | 80 [1] | Adults 18–50 yrs, PSQI ≥5 [1] | 8 weeks [1] |
| Thomas, 2024 [2] | Yes | Valerian (Sleeproot®, 2% valerenic acid) [2] | Standardized botanical | Acute and sustained sleep effects of single dose [2] | Parallel, double-blind [2] | 80 [2] | Adults 18–50 yrs, PSQI ≥5 [2] | 8 weeks (acute Day 1 endpoint) [2] |
| Taibi et al., 2009 [12] | Yes | Valerian root (Nature's Resource, 0.8% valerenic acid) [12] | Standardized botanical | Insomnia in older women [12] | Crossover, double-blind, 13-day washout [12] | 16 [12] | Women aged 55–80 yrs, PSQI ≥5 [12] | 2 weeks per phase [12] |
| Oxman et al., 2007 [23] | Yes | Valerian (Valerina Forte®) [23] | Standardized botanical | Self-reported insomnia [23] | Parallel, double-blind [23] | 405 [23] | Adults 18–75 yrs, PSQI >5, insomnia ≥1 month [23] | 2 weeks [23] |
| Mirmohammadali et al., 2014 [24] | Yes | Valerian extract 700 mg [24] | Standardized botanical | Sleep in postmenopausal women [24] | Crossover, triple-blind, 2-week washout [24] | 144 [24] | Postmenopausal women 45–60 yrs [24] | 1 month per phase [24] |
| Di Pierro et al., 2024 [25] | Yes | Melissa officinalis phytosome (Meloff) [25] | Standardized botanical (phytosome) | Sleep quality and insomnia [25] | Crossover, double-blind, 5-day washout [25] | 30 [25] | Adults 18–65 yrs with non-restorative sleep [25] | 2 weeks per phase [25] |
| Bano et al., 2023 [26] | Yes | Melissa officinalis phospholipid extract (Relissa™) [26] | Standardized botanical (phytosome) | Emotional distress and sleep [26] | Parallel, double-blind [26] | 100 [26] | Healthy adults 18–60 yrs with moderate emotional distress or PSQI >5 [26] | 3 weeks [26] |
| Ngan & Conduit, 2011 [15] | Yes | Passiflora incarnata herbal tea [15] | Standardized botanical (tea) | Subjective sleep quality in healthy adults [15] | Crossover, double-blind, 1-week washout [15] | 41 [15] | Healthy adults 18–35 yrs with mild sleep fluctuations [15] | 1 week per phase [15] |
| Khalid et al., 2025 [16] | Yes | Chamomile (Matricaria chamomilla) vs. Passiflora incarnata powder tea [16] | Standardized botanical (tea) | Primary insomnia [16] | Parallel, 3-arm (no explicit blinding stated) [16] | 90 [16] | Adults 25–45 yrs, primary insomnia [16] | 8 weeks [16] |
| Zick et al., 2011 [27] | Yes | Chamomile extract (Matricaria recutita) [27] | Standardized botanical | Chronic primary insomnia [27] | Parallel, double-blind [27] | 34 [27] | Adults 18–65 yrs, DSM-IV primary insomnia ≥6 months [27] | 28 days [27] |
| Ha et al., 2019 [28] | Yes | Polygonatum sibiricum rhizome [28] | Standardized botanical | Mild insomnia [28] | Parallel, double-blind [28] | 80 [28] | Adults 20–65 yrs, mild insomnia ≥2 months [28] | 4 weeks [28] |
| Pérez-Piñero et al., 2024 [6] | Yes | Aloysia citrodora (lemon verbena) extract [6] | Standardized botanical | Sleep quality in healthy adults with disturbances [6] | Parallel, double-blind [6] | 71 [6] | Healthy adults (mean age 29.5 yrs), poor sleep quality [6] | 90 days [6] |
| Martínez-Rodríguez et al., 2022 [29] | Yes | Lemon verbena (Lippia citriodora, PLX/RelaxPLX) [29] | Standardized botanical | Stress and sleep quality [29] | Parallel, double-blind [29] | 40 [29] | Adults (mean 38.7 yrs), PSS >15 and PSQI >5 [29] | 8 weeks + 4-week washout [29] |
| Di Minno et al., 2025 [30] | Yes | Scutellaria lateriflora extract (BlueCALM®, 10% baicalin) [30] | Standardized botanical | Primary insomnia [30] | Crossover, double-blind, 28-day washout [30] | 66 [30] | Adults 18–70 yrs, primary moderate insomnia ≥1 month [30] | 56 days per phase [30] |
| Jiang et al., 2015 [31] | Yes | Black cohosh (Cimicifuga racemosa, isopropanolic extract) [31] | Standardized botanical | Sleep in postmenopausal women [31] | Parallel, double-blind [31] | 48 [31] | Postmenopausal women 45–60 yrs with sleep disturbance ≥1 month [31] | 6 months [31] |
| Rao et al., 2021 [32] | Yes | PEA (Levagen+®, LipiSperse® CWD technology) [32] | Endocannabinoid-related fatty acid amide | Sleep disturbance (PSQI >5) [32] | Parallel, double-blind [32] | 103 [32] | Adults >18 yrs, PSQI >5 [32] | 8 weeks [32] |
| Deb et al., 2025 [11] | Yes | PEA (Levagen+®, LipiSperse® CWD technology) [11] | Endocannabinoid-related fatty acid amide | Stress (HRV and subjective) in university students [11] | Crossover, double-blind, 6-week washout [11] | 16 [11] | Female university students (mean age 22 yrs), PSS >13 [11] | 6 weeks per phase [11] |
| Sarris et al., 2019 [33] | Yes | L-theanine (450–900 mg) [33] | Amino acid | Anxiety and sleep in GAD (adjunctive) [33] | Parallel, double-blind [33] | 46 [33] | Adults with DSM-5 GAD, HAMA ≥16, on stable antidepressant [33] | 8 weeks [33] |
| Byun et al., 2018 [8] | Yes | GABA from fermented rice germ (Lactobacillus sakei B2-16), 15% GABA [8] | Amino acid | Insomnia (PSG) [8] | Parallel, double-blind, 3:1 allocation [8] | 40 [8] | Adults mean age 49 yrs, PSQI >5, ISI >8 [8] | 4 weeks [8] |
| Shin et al., 2016 [34] | Yes | GABA from fermented rice germ (100 mg or 300 mg) [34] | Amino acid | Insomnia (dose-response) [34] | Parallel, 3-arm, double-blind [34] | 114 [34] | Adults ≥30 yrs, DSM-IV insomnia [34] | 1 week [34] |
| Yoon et al., 2022 [9] | Yes | GABA from fermented rice germ (75 mg low-dose) [9] | Amino acid | Low-dose GABA in insomnia (PSG) [9] | Parallel, double-blind [9] | 50 [9] | Patients with insomnia [9] | 4 weeks [9] |
| Kärppä et al., 2020 [35] | Yes | Lemborexant (LEM5 and LEM10) [35] | Dual orexin receptor antagonist | Insomnia disorder (long-term) [35] | Parallel, 3-arm, double-blind (SUNRISE 2) [35] | 949 [35] | Adults ≥18 yrs, DSM-5 insomnia, ISI ≥15 [35] | 6-month placebo-controlled phase [35] |
| Huang et al., 2023 [36] | Yes | Dimdazenil (EVT201), 2.5 mg, partial GABA-A PAM [36] | GABA-A receptor modulator | Insomnia disorder (Phase III) [36] | Parallel, double-blind, 2:1 allocation [36] | 546 [36] | Adults ≥18 yrs across 66 hospitals in China [36] | 14 days [36] |
| Luthringer et al., 2009 [37] | Yes | Melatonin 2 mg prolonged-release (Circadin) [37] | Melatonin | Primary insomnia in elderly [37] | Parallel, double-blind [37] | 40 [37] | Adults ≥55 yrs, DSM-IV primary insomnia [37] | 3 weeks [37] |
| Wade et al., 2011 [38] | Yes | Melatonin 2 mg prolonged-release (Circadin) [38] | Melatonin | Primary insomnia, age cut-off analysis [38] | Parallel, double-blind [38] | 930 [38] | Adults 18–80 yrs, primary insomnia, SL >20 min [38] | 3 weeks + 26-week extension [38] |
| Etedali et al., 2022 [39] | Yes | Melatonin 6 mg/day (3 mg twice daily) [39] | Melatonin | ADT-induced sleep problems in prostate cancer [39] | Parallel, double-blind [39] | 43 [39] | Men ≥18 yrs with prostate cancer on ADT, mean age ~70 yrs [39] | 4 weeks [39] |
| Sletten et al., 2018 [40] | Yes | Melatonin 0.5 mg fast-release [40] | Melatonin | Delayed sleep-wake phase disorder (DSWPD) [40] | Parallel, double-blind [40] | 116 [40] | Adults 16–65 yrs with clinically diagnosed DSWPD [40] | 4 weeks [40] |
| Bondi et al., 2018 [13] | Yes | Melatonin 5 mg [13] | Melatonin | Post-acute withdrawal sleep in substance use recovery [13] | Parallel, double-blind [13] | 70 [13] | Males ≥18 yrs in residential substance use recovery [13] | 28 days [13] |
| Ballester et al., 2019 [41] | Yes | Agomelatine 25 mg (MT1/MT2 agonist, 5HT2C antagonist) [41] | Melatonin receptor agonist | Sleep and circadian rhythm in ASD+ID adults [41] | Crossover, triple-blind, 2-week washout [41] | 23 [41] | Adults 18–65 yrs with ASD and intellectual disability [41] | 12 weeks per phase [41] |
| Um et al., 2019 [42] | Yes | Rice bran extract (γ-oryzanol 4.5 mg/g) [42] | Standardized botanical | Sleep disturbance (PSG) [42] | Parallel, double-blind [42] | 50 [42] | Adults, mean age ~44 yrs, PSQI ≥5 [42] | 2 weeks [42] |
| Kim et al., 2019 [43] | Yes | Alpha-s1 casein hydrolysate (Lactium®) [43] | Bioactive peptide | Mild-to-moderate sleep disturbance [43] | Crossover, double-blind, 4-week washout [43] | 48 [43] | Adults mean age 49 yrs, PSQI >5 [43] | 4 weeks per phase [43] |
| Kim et al., 2007 [44] | Yes | Alpha-s1 casein hydrolysate (Lactium) [44] | Bioactive peptide | Stress-related symptoms in women [44] | Crossover, double-blind, 3-week washout [44] | 63 [44] | Females >18 yrs with stress-related symptoms [44] | 30 days per phase [44] |
| Eckert et al., 2024 [45] | Yes | Fish hydrolysate (Peptidyss®, sardine peptides) [45] | Bioactive peptide | Poor sleep quality (PSQI >6) [45] | Crossover, double-blind, 4-week washout [45] | 35 (PP) [45] | Healthy German adults 35–60 yrs, PSQI >6 [45] | 4 weeks per phase [45] |
| Ayabe et al., 2024 [46] | Yes | β-lactolin (whey peptide, 1.6 mg/day) [46] | Bioactive peptide | Trait anxiety and subjective stress [46] | Parallel, double-blind [46] | 56 [46] | Adults 45–64 yrs with self-reported low psychological health [46] | 6 weeks [46] |
| Tsukamoto-Sen et al., 2026 [47] | Yes | Piceatannol (PIC) 10 mg/day from passion fruit seeds [47] | Polyphenol | Sleep quality (WASO) in healthy adults [47] | Parallel, double-blind [47] | 76 [47] | Healthy adults 20–59 yrs [47] | 4 weeks [47] |
| Doma et al., 2026 [7] | Yes | Corn leaf extract (CLE, 0.2–0.3% 6-MBOA) [7] | Standardized botanical | Sleep quality in adults with difficulty falling/staying asleep [7] | Parallel, triple-blind [7] | 80 [7] | Healthy adults 18–65 yrs with ≥2 sleep disturbance episodes/week ≥1 month [7] | 28 days [7] |
| Lee et al., 2025 [48] | Yes | Heat-treated Limosilactobacillus fermentum PS150 (HT-PS150) [48] | Heat-killed postbiotic | Poor sleep (PSQI ≥5) [48] | Parallel, double-blind [48] | 84 [48] | Adults 20–60 yrs, PSQI ≥5, ISI <22 [48] | 8 weeks [48] |
| Nishida et al., 2019 [14] | Yes | Heat-killed Lactobacillus gasseri CP2305 (tablets) [14] | Heat-killed postbiotic | Sleep quality and anxiety in stressed students [14] | Parallel, double-blind [14] | 60 [14] | Medical students under chronic examination stress [14] | 24 weeks [14] |
| Nishida et al., 2017 [17] | Yes | Heat-killed Lactobacillus gasseri CP2305 (fermented milk beverage) [17] | Heat-killed postbiotic | Stress-associated sleep in students [17] | Parallel, double-blind [17] | 32 [17] | Undergraduate medical students (cadaver dissection stress) [17] | 5 weeks [17] |
| Tanihiro et al., 2026 [10] | Yes | Heat-killed Lactobacillus gasseri CP2305 (tablets, single-dose EEG/HRV study) [10] | Heat-killed postbiotic | Acute brain alpha oscillations and stress [10] | Crossover, double-blind, ≥3-day washout [10] | 28 [10] | Healthy adults 20–44 yrs [10] | Single dose (acute) [10] |
| Sawada et al., 2022 [49] | Yes | Lactobacillus gasseri CP2305 (tablets) [49] | Heat-killed postbiotic | Menopausal symptoms in middle-aged women [49] | Parallel, double-blind [49] | 80 [49] | Women 40–60 yrs [49] | 6 menstrual cycles [49] |
| Watanabe et al., 2025 [50] | Yes | Heat-killed Lactiplantibacillus plantarum SNK12 [50] | Heat-killed postbiotic | Sleepiness on rising; HPA-axis modulation [50] | Parallel, double-blind [50] | 56 [50] | Healthy Japanese adults (mean age ~49 yrs) with morning fatigue [50] | 4 weeks [50] |
| Wu et al., 2022 [51] | Yes | Heat-killed Lactobacillus paracasei PS23 (HK-PS23) [51] | Heat-killed postbiotic | Stress and anxiety in nurses (PSS ≥27) [51] | Parallel, double-blind [51] | 70 [51] | Female nurses, PSS ≥27 [51] | 8 weeks [51] |
| Hirose et al., 2013 [52] | Yes | Heat-killed Lactobacillus plantarum L-137 (LP20 tablet) [52] | Heat-killed postbiotic | URTI and immune function under job stress [52] | Parallel, double-blind [52] | 78 [52] | Adults mean age 50.6 yrs, psychological stress (BJSQ >41) [52] | 12 weeks [52] |
| Komano et al., 2018 [53] | Yes | Heat-killed Lactococcus lactis JCM 5805 (LC-Plasma) [53] | Heat-killed postbiotic | Immunity and fatigue in athletes [53] | Parallel, double-blind [53] | 51 [53] | Healthy male athletes (~20 yrs) [53] | 13 days [53] |
| Eungpinichpong, 2018 [54] | Yes | Kaempferia parviflora extract 360 mg/day [54] | Standardized botanical | Physiological and psychological stress [54] | Parallel, double-blind [54] | 80 [54] | Healthy adults 24–46 yrs, moderate stress [54] | 14 days [54] |
| Ito et al., 2014 [55] | Yes | Enzyme-treated asparagus extract (ETAS) [55] | Standardized botanical | Stress hormones and sleep [55] | Study 1: parallel; Study 2: crossover, 2-week washout [55] | Study 1: 20; Study 2: 18 [55] | Healthy adult males [55] | 7 days [55] |
The 55 trials cover 14 broad ingredient categories. Ashwagandha is represented by five trials using three extracts (KSM-66, Shoden®, Sensoril®) [3, 4, 18–20]. Saffron appears in three trials, all using the Safr'Inside™ or Saffr'activ® standardized extracts [5, 21, 22]. Valerian is represented by four trials employing two standardization levels (0.8% and 2% valerenic acid) [1, 2, 12, 23]. GABA from fermented rice germ features in three dose-ranging trials across different dose levels (75, 100, and 300 mg) [8, 9, 34]. Heat-killed Lactobacillus gasseri CP2305 is assessed across four trials with varying designs and populations [10, 14, 17, 49]. Pharmaceutical comparators include lemborexant (SUNRISE 2, n=949) [35] and dimdazenil (Phase III, n=546) [36], providing large-trial benchmarks. Sample sizes range from 16 [11] to 949 [35], and trial durations range from a single acute dose [10] to 24 weeks [14]. Crossover designs were used in 12 trials [10–12, 15, 24, 25, 30, 41, 43–45, 55]. Most studies were funded by ingredient manufacturers or commercial entities, a consideration that applies broadly across this evidence base.
Effects
Objective Sleep Outcomes
The table below presents the available objective sleep data extracted from PSG, sleep EEG, and actigraphy, organized by ingredient. Where arm-level means and SD are available these are reported; where only significance or direction is stated, that is noted. Studies reporting no objective sleep data are listed with a corresponding notation.
| Ingredient | Study | Measurement method | SOL (active vs. placebo) | WASO (active vs. placebo) | TST (active vs. placebo) | Sleep efficiency (active vs. placebo) | Other architecture findings |
|---|---|---|---|---|---|---|---|
| Ashwagandha (KSM-66, 600 mg/day) [3] | Langade et al., 2019 [3] | Actigraphy | Post-tx: 29.00 (7.14) vs. 33.94 (7.65) min; p=0.019 [3] | Not reported [3] | Not reported [3] | Baseline: 75.63% (2.70) vs. 75.14% (3.73); Post-tx: 83.48% (2.83) vs. 79.68% (3.59); p<0.001 [3] | Not reported [3] |
| Ashwagandha (Shoden®, 120 mg/day) [4] | Deshpande et al., 2020 [4] | Actigraphy | Baseline: 13.6 (8.0) vs. 14.3 (9.9); Post-tx: 9.9 (5.5) vs. 12.2 (6.4) min; p<0.01 [4] | Baseline: 82.6 (35.4) vs. 80.7 (31.3); Post-tx: 70.4 (32.1) vs. 82.7 (36.2) min; p<0.05 [4] | Baseline: 359.8 (50.4) vs. 370.6 (46.8); Post-tx: 377.1 (46.8) vs. 351.7 (41.5) min; p<0.001 [4] | Baseline: 78.9% (8.4) vs. 79.6% (6.8); Post-tx: 82.6% (7.0) vs. 78.9% (7.4); p<0.01 [4] | Not reported [4] |
| Ashwagandha (Gaia Herbs, 700 mg/day) [18] | O'Connor et al., 2022 [18] | Questionnaire only | Not reported objectively [18] | Not reported [18] | Not reported [18] | Not reported [18] | Not reported [18] |
| Ashwagandha (Sensoril®, 125/250/500 mg/day) [19] | Pandit et al., 2024 [19] | Questionnaire only | Not reported [19] | Not reported [19] | Not reported [19] | Not reported [19] | Not reported [19] |
| Ashwagandha (KSM-66, 600 mg/day) [20] | Choudhary et al., 2017 [20] | Questionnaire only | Not reported [20] | Not reported [20] | Not reported [20] | Not reported [20] | Not reported [20] |
| Saffron (Safr'Inside™, 20 mg or 30 mg/day) [21] | Schuster et al., 2025 [21] | Questionnaire only | Not reported objectively [21] | Not reported [21] | Not reported [21] | Not reported [21] | Not reported [21] |
| Saffron (Saffr'activ®, 15.5 mg/day) [22] | Pachikian et al., 2021 [22] | Actigraphy | Baseline: 16.4 (18.9) vs. 9.6 (11.4); Week 6: 14.7 (16.1) vs. 7.5 (6.3) min; no sig. group effect [22] | Baseline: 97.7 (30.7) vs. 83.4 (25.8); Week 6: 103.3 (30.5) vs. 81.8 (26.6) min; no sig. group effect [22] | Baseline: 368.8 (48.3) vs. 387.8 (27.3); Week 6: 376.7 (42.5) vs. 379.7 (34.9) min; no sig. group effect [22] | Baseline: 76.0% (8.2) vs. 80.4% (5.4); Week 6: 75.4% (7.8) vs. 80.2% (5.7); no sig. group effect [22] | TIB increased significantly with saffron (p=0.023) [22] |
| Saffron (Safr'Inside™, 30 mg/day) [5] | Láng et al., 2025 [5] | EEG-based sleep tracker (Dreem 3) | SOL: Baseline 14.3 (6.9) vs. 14.7 (7.4); FU: 11.3 (5.2) vs. 17.5 (9.3) min; p=0.03. LtPS: 15.2 (7.1) vs. 23.7 (12.6) min; p=0.003 [5] | FU: 24.5 (12.7) vs. 42.6 (17.5) min; p=0.0004 [5] | No significant difference (p=0.6) [5] | SE increased with saffron; p=0.04 (subjective) [5]; objective SE p=0.2 [5] | REM and N3/SWS: no sig. difference [5] |
| Valerian (Sleeproot®, 2% VA, 200 mg/day) [1] | Shekhar et al., 2023 [1] | Wrist actigraphy + PSG subset (n=40) | SOL change: VE −10.89 (4.98) vs. placebo +5.34 (4.17) min; p<0.05 [1] | Not reported [1] | TST change: VE +75.21 (34.05) vs. placebo −74.58 (31.33) min; p<0.05 [1] | Improved; p<0.05 [1] | REM and NREM3 trends, not significant [1] |
| Valerian (Sleeproot®, 2% VA, 200 mg, single-dose) [2] | Thomas, 2024 [2] | Wrist actigraphy | Day 1: SOL VE 76.47 (9.74) vs. placebo (no direct value); change −12.91 (8.67) vs. +52.34 (12.14) min; p<0.05 [2] | Not reported [2] | Day 1: TST VE 328.93 (10.54) vs. baseline 306.09 (7.61) min; change vs. placebo p<0.05 [2] | Day 1: SE 68.76% (2.07) vs. baseline 66.40% (1.67); p<0.05 vs. placebo [2] | Not reported [2] |
| Valerian (0.8% VA, 300 mg, crossover) [12] | Taibi et al., 2009 [12] | PSG + actigraphy | No significant differences on any measure [12] | WASO increased with valerian (+17.7 ± 25.6 min; p=0.02 vs. baseline; NS vs. placebo) [12] | Not significant [12] | Not significant [12] | No significant differences in SWS or REM [12] |
| Valerian (Valerina Forte®, 600 mg) [23] | Oxman et al., 2007 [23] | Sleep diary only (no actigraphy/PSG) | Not measured objectively [23] | Not measured [23] | Not measured [23] | Not measured [23] | Not measured [23] |
| Valerian 700 mg/day [24] | Mirmohammadali et al., 2014 [24] | Questionnaire only | Not measured objectively [24] | Not measured [24] | Not measured [24] | Not measured [24] | Not measured [24] |
| Melissa officinalis phytosome (Meloff, 400 mg/day) [25] | Di Pierro et al., 2024 [25] | Wearable device (Garmin VenuSq) + ISI | SWS increased 15% with treatment [25] | Not reported specifically [25] | Not reported specifically [25] | Not reported specifically [25] | REM decreased 10% with treatment; SWS +15% [25] |
| Melissa officinalis phospholipid extract (Relissa™, 400 mg/day) [26] | Bano et al., 2023 [26] | Questionnaire only | Not measured objectively [26] | Not measured [26] | Not measured [26] | Not measured [26] | Not measured [26] |
| Passiflora incarnata tea (2 g) [15] | Ngan & Conduit, 2011 [15] | PSG + sleep diary | SOL: Passiflora Md=15 vs. Placebo Md=8 min; NS (p>0.01) [15] | Not reported [15] | TST: 411.89 (37.65) vs. 398.33 (48.35) min; NS [15] | SE: 87.73% (9) vs. 85.72% (11.85); NS; Cohen's d=0.34 [15] | SWS and REM: no significant differences [15] |
| Chamomile (Matricaria recutita, 270 mg twice daily) [27] | Zick et al., 2011 [27] | Sleep diary only | Not measured objectively [27] | Not measured [27] | Not measured [27] | Not measured [27] | Not measured [27] |
| Chamomile vs. Passiflora tea (1 g/day) [16] | Khalid et al., 2025 [16] | PSQI + biomarkers (no PSG/actigraphy) | Not measured objectively [16] | Not measured [16] | Not measured [16] | Not measured [16] | Not measured [16] |
| Polygonatum sibiricum rhizome (500 mg/day) [28] | Ha et al., 2019 [28] | Actigraphy | Not significant [28] | WASO: no significant interaction (p=0.275) [28] | Baseline PS: 365 (66.0); Week 4: 373 (68.9) min vs. placebo 354 (71.0) and 346 (60.7); p interaction=0.046 [28] | SE: no significant interaction (p=0.522) [28] | Not reported [28] |
| Lemon verbena (Aloysia citrodora, 400 mg/day) [6] | Pérez-Piñero et al., 2024 [6] | Actigraphy | Baseline: 3.7 (0.9) vs. 3.0 (0.9); Post: 2.9 (0.8) vs. 3.2 (0.8) (actigraphy units); p=0.001 [6] | FU: 28.7 (9.2) vs. 35.0 (13.5) min; p=0.001 [6] | Not reported [6] | FU: 93.3% (2.5) vs. 91.5% (3.4); p=0.001 [6] | Awakenings: 2.0 (0.6) vs. 2.2 (0.6); p=0.001 [6] |
| Lemon verbena (RelaxPLX, 400 mg/day) [29] | Martínez-Rodríguez et al., 2022 [29] | Wearable Fitbit | % awake time: 11.6% vs. 11.3% (NS trend) [29] | Not specifically reported [29] | % minutes asleep: 88.4% vs. 88.6% (NS) [29] | Not reported [29] | REM 23.0% vs. 19.0%; deep sleep 18.6% vs. 16.8% (no p-values) [29] |
| Scutellaria lateriflora (BlueCALM®, 400 mg/day) [30] | Di Minno et al., 2025 [30] | Not specified (VAS + PSQI as primary) | Significant improvement (p<0.001) [30] | Not reported [30] | Significant improvement (p<0.001) [30] | Significant improvement (p<0.001) [30] | Not reported [30] |
| Black cohosh (isopropanolic extract) [31] | Jiang et al., 2015 [31] | PSG | Not reported [31] | Decreased 15.8% vs. placebo; r=0.404, p=0.009 [31] | Not reported [31] | Increased vs. placebo; r=0.391, p=0.010 [31] | Not reported [31] |
| Rice bran extract (RBS, γ-oryzanol, 1000 mg/day) [42] | Um et al., 2019 [42] | PSG | Decreased; p=0.047 [42] | Non-significant decrease [42] | +11.7 (29.9) min; p=0.019 (adjusted) [42] | +2.9% (7.7); p=0.010 (adjusted) [42] | Stage 2 sleep increased (p=0.022); delta power trend p=0.067 [42] |
| Corn leaf extract (CLE, 500 mg/day, 60 min pre-bed) [7] | Doma et al., 2026 [7] | Actigraphy + EEG/EOG | SOL: Day 14 p=0.008 (d=0.54); Day 28: p=0.054 (d=0.57) [7] | Day 14: 81.74 (42.49) vs. 129.45 (67.90) min; p=0.001 (d=0.84). Day 28: 73.71 (39.79) vs. 111.41 (67.92) min; p=0.007 (d=0.67) [7] | Significant improvement Day 28 [7] | Day 14: 80.96% (7.77) vs. 71.39% (10.91); p=0.001 (d=1.01). Day 28: 81.03% (9.63) vs. 74.04% (13.46); p=0.007 (d=0.59) [7] | Non-REM sleep +35.7 min CLE vs. −10.6 min placebo at Day 28; REM improved Days 14 and 28 [7] |
| PEA (Levagen+®, 350 mg/day) [32] | Rao et al., 2021 [32] | Actigraphy + sleep diary | Significant reduction in SOL in PEA group vs. placebo [32] | Not reported [32] | No significant difference between groups [32] | No significant difference between groups [32] | Not reported [32] |
| L-theanine (450–900 mg/day adjunctive) [33] | Sarris et al., 2019 [33] | No objective sleep measurement | Not measured [33] | Not measured [33] | Not measured [33] | Not measured [33] | Not measured [33] |
| GABA fermented rice germ (300 mg/day) [8] | Byun et al., 2018 [8] | PSG | Pre: 13.4 (15.7); Post: 5.7 (6.2) min; p=0.001 (within-group) [8] | Not reported [8] | Not reported [8] | Pre: 79.4% (12.9); Post: 86.1% (10.5); p=0.018 (within-group) [8] | Not reported [8] |
| GABA fermented rice germ (100 mg or 300 mg) [34] | Shin et al., 2016 [34] | Actigraphy (1 week) | Post: 300 mg 30.92 (17.66); 100 mg 36.49 (33.68); placebo 37.74 (33.35) min; p=0.162 (group effect NS) [34] | Not reported [34] | Post: 300 mg 314.0 (61.6); 100 mg 305.1 (73.1); placebo 298.5 (62.7) min; p=0.413 [34] | Post: 300 mg 76.06%; 100 mg 74.11%; placebo 71.34%; p=0.146 [34] | Not reported [34] |
| GABA fermented rice germ (75 mg low-dose) [9] | Yoon et al., 2022 [9] | PSG | Baseline: 9.0 (12.6); Post: 4.8 (4.6) min; p=0.0038 (within-group) [9] | Not reported [9] | Not reported [9] | Baseline: 86.5%; Post: 88.0%; p=0.2972 (NS) [9] | N3 sleep: baseline 8.6% → post 12.9%; p=0.0043. Arousal index improved p=0.0316 [9] |
| Alpha-s1 casein hydrolysate (Lactium®, 300 mg/day) [43] | Kim et al., 2019 [43] | Actigraphy + PSG | SOL: all phases improved; no significant group difference (p=0.437) [43] | WASO: improved in both phases; NS group difference (p=0.551) [43] | TST increased both phases; NS group difference (p=0.729) [43] | SE: actigraphy significantly improved with 4 weeks ACH; Cohen's d=0.47 [43] | N3 increased after placebo phase; NS [43] |
| Fish hydrolysate (Peptidyss®, 1396 mg/day) [45] | Eckert et al., 2024 [45] | Wearable + sleep diary | Not reported [45] | Not reported [45] | Not reported [45] | Not reported [45] | Deep sleep reduced during fish hydrolysate phase (noted in PSQI component) [45] |
| Piceatannol (PIC, 10 mg/day) [47] | Tsukamoto-Sen et al., 2026 [47] | Actigraphy | Not significant [47] | Significant reduction vs. placebo [47] | Not reported [47] | Not reported [47] | Not reported [47] |
| Rice bran (RBS, gamma-oryzanol) [42] noted above. Asparagus extract (ETAS, 150 mg/day) [55] | Ito et al., 2014 [55] | Actigraphy | NS overall; subgroup with SE <90%: trend toward improvement [55] | Not reported [55] | NS (405.6 vs. 399.3 min) [55] | NS overall; subgroup interaction [55] | Not reported [55] |
| HK-PS150 (L. fermentum) [48] | Lee et al., 2025 [48] | Fitbit Inspire 3 actigraphy | Not significant (full cohort) [48] | Not significant (full cohort) [48] | Not significant (full cohort) [48] | Subgroup ISI ≥8: significantly greater improvement in sleep duration and efficiency [48] | Nocturnal melatonin up-regulated in higher proportion of HT-PS150 participants [48] |
| L. gasseri CP2305 (tablets, 24 weeks) [14] | Nishida et al., 2019 [14] | Single-channel EEG | SOL to N3 significantly shortened [14] | WASO significantly decreased [14] | Not reported [14] | Not reported [14] | Delta power ratio in first sleep cycle significantly increased [14] |
| L. gasseri CP2305 (fermented milk, 5 weeks) [17] | Nishida et al., 2017 [17] | Questionnaire only (PSQI) | Not measured objectively [17] | Not measured [17] | Not measured [17] | Not measured [17] | Not measured [17] |
| L. gasseri CP2305 (acute EEG/HRV) [10] | Tanihiro et al., 2026 [10] | EEG alpha power + HRV | Not measured [10] | Not measured [10] | Not measured [10] | Not measured [10] | Alpha power significantly greater post-CP2305 vs. placebo [10] |
| L. gasseri CP2305 (menopause, 6 cycles) [49] | Sawada et al., 2022 [49] | Questionnaire only (SMI/GCS) | Not measured objectively [49] | Not measured [49] | Not measured [49] | Not measured [49] | Not measured [49] |
| L. plantarum SNK12 (4 weeks) [50] | Watanabe et al., 2025 [50] | OSA-MA subjective scale only | Not measured objectively [50] | Not measured [50] | Not measured [50] | Not measured [50] | Not measured [50] |
| HK-PS23 (L. paracasei, 8 weeks) [51] | Wu et al., 2022 [51] | Fitbit wearable | WASO: shorter in HK-PS23 group vs. placebo [51] | WASO shorter [51] | TST: mean difference −50.24 min (SE 14.50); p=0.002 [51] | Not reported [51] | REM% decreased; light sleep decreased in HK-PS23 group [51] |
| Melatonin 2 mg PRM (Circadin, 3 weeks) [37] | Luthringer et al., 2009 [37] | PSG | PRM: Post 13.7 (6.6) vs. placebo 22.6 (13.4) min; p=0.02; Δ=−9 min [37] | NS [37] | NS [37] | NS [37] | No significant differences in sleep stages [37] |
| Melatonin 2 mg PRM (Circadin, 26 weeks) [38] | Wade et al., 2011 [38] | Sleep diary (subjective) | Significant improvement in SL (diary) in 55–80 yr group (−15.4 vs. −5.5 min; p=0.014) [38] | Not measured objectively [38] | Not measured objectively [38] | Not measured objectively [38] | Not measured objectively [38] |
| Melatonin 6 mg/day, prostate cancer ADT [39] | Etedali et al., 2022 [39] | Questionnaire only (PSQI) | Not measured objectively [39] | Not measured [39] | Not measured [39] | Not measured [39] | Not measured [39] |
| Melatonin 0.5 mg fast-release, DSWPD [40] | Sletten et al., 2018 [40] | Actigraphy | Actigraphic SOL 15.7 min shorter (95% CI −24.41 to −6.89) [40]; Sleep onset time 34 min earlier (95% CI −60 to −8) [40] | Not reported [40] | Not reported [40] | SE T1: +5.14% (95% CI 2.52–7.76; p<0.001) [40] | Not reported [40] |
| Melatonin 5 mg, post-acute withdrawal [13] | Bondi et al., 2018 [13] | Questionnaire only | Not measured objectively [13] | Not measured [13] | Not measured [13] | Not measured [13] | Not measured [13] |
| Agomelatine 25 mg, ASD+ID [41] | Ballester et al., 2019 [41] | Ambulatory circadian monitoring (wrist) | Not specifically reported [41] | Not reported [41] | TST: +83 min (post 532±121 vs. pre 449±177 min); p=0.016; d=0.55 [41] | Not reported [41] | Circadian phase correction; rhythm stability improved [41] |
| Lemborexant 5 mg/10 mg (SUNRISE 2) [35] | Kärppä et al., 2020 [35] | Electronic sleep diary (subjective) | sSOL significantly decreased vs. placebo (LEM5 and LEM10) at Month 6 and most timepoints [35] | sWASO significantly decreased vs. placebo [35] | Not reported in PSG [35] | sSE significantly increased [35] | Not measured by PSG in this report [35] |
| Dimdazenil 2.5 mg (Phase III) [36] | Huang et al., 2023 [36] | PSG | LPS reduced 21.65 min vs. baseline; 6.46 min vs. placebo (p=0.023) [36] | WASO reduced 49.67 min vs. baseline; 20.16 min vs. placebo (p<0.001) [36] | TST increased 71.09 min vs. baseline; 31.68 min vs. placebo (p<0.001) [36] | SE: +13.26% vs. baseline; +5.55% vs. placebo (p<0.001) [36] | Stage 2 longer; Stage 3 and REM shorter in dimdazenil group [36] |
| Kaempferia parviflora 360 mg/day [54] | Eungpinichpong, 2018 [54] | HRV only; no sleep measure | Not measured [54] | Not measured [54] | Not measured [54] | Not measured [54] | Not measured [54] |
| ETAS (asparagus extract) [55] | Ito et al., 2014 (Study 2) [55] | Actigraphy | NS (15.2 vs. 15.4 min) [55] | Not reported [55] | NS overall; subgroup effect [55] | NS overall [55] | Not reported [55] |
| L. plantarum L-137 (LP20, 12 weeks) [52] | Hirose et al., 2013 [52] | No sleep measurement | Not measured [52] | Not measured [52] | Not measured [52] | Not measured [52] | Not measured [52] |
| LC-Plasma (L. lactis JCM 5805, 13 days) [53] | Komano et al., 2018 [53] | No sleep measurement | Not measured [53] | Not measured [53] | Not measured [53] | Not measured [53] | Not measured [53] |
| β-lactolin (whey peptide, 6 weeks) [46] | Ayabe et al., 2024 [46] | Questionnaire only | Not measured objectively [46] | Not measured [46] | Not measured [46] | Not measured [46] | Not measured [46] |
Discussion of Objective Sleep Effects
Across the 55 trials, only a subset provide Tier 1 objective sleep data (PSG, validated EEG-based devices, or actigraphy against SOL, WASO, TST, SE, or sleep architecture). The pharmaceutical benchmarks are instructive: dimdazenil 2.5 mg produced a PSG-confirmed +71.09 min TST, −49.67 min WASO, −21.65 min LPS, and +13.26% SE versus baseline, with placebo-adjusted effects of +31.68 min TST, −20.16 min WASO, −6.46 min LPS, and +5.55% SE [36]. Lemborexant 5 and 10 mg similarly produced sustained, statistically significant improvements in subjective SOL and WASO over 6 months [35], though the SUNRISE 2 publication did not include PSG data for this report period [35]. These figures frame the magnitude of effects in the natural compound literature.
Among the botanicals with multi-parameter actigraphy or PSG evidence, corn leaf extract (CLE) demonstrated the largest effect sizes: Cohen's d=1.01 for sleep efficiency at Day 14, d=0.84 for WASO, and d=0.54 for SOL, sustained to Day 28 [7]. Non-REM sleep increased by 35.7 min in the CLE group compared to a decrease of 10.6 min in the placebo group [7]. These effect sizes are larger than those observed with pharmaceutical agents in this review and should be interpreted with caution given the modest sample size (n=40 per arm) [7] and retrospective trial registration [7].
Ashwagandha (Shoden® 120 mg/day) produced significant actigraphy improvements across all four primary sleep parameters (SOL p<0.01, WASO p<0.05, TST p<0.001, SE p<0.01) in 144 subjects with non-restorative sleep [4]. The higher-dose KSM-66 formulation (600 mg/day) in patients with diagnosed insomnia showed significant SOL shortening (29.00 vs. 33.94 min; p=0.019) and SE improvement (83.48% vs. 79.68%; p<0.001) [3]. Of note, the Shoden® dose is achieved through an 8% withanolide glycoside standardization concentrated into 120 mg, versus 5% withanolides at 600 mg for KSM-66 [3, 4], meaning the absolute withanolide contents are not directly comparable.
Saffron (Safr'Inside™ 30 mg/day) in 52 older adults provided objective EEG-based evidence of reduced SOL (p=0.03) and latency to persistent sleep (p=0.003), and reduced WASO (p=0.0004), without significant effects on TST, REM, or N3 [5]. By contrast, the Saffr'activ® preparation at 15.5 mg/day in 66 adults with mild-to-moderate sleep disorders showed no significant group effects on any actigraphy parameter, though TIB increased (p=0.023) and questionnaire-based outcomes improved [22, 22]. This dose-standardization discrepancy may account for some of the divergence: Saffr'activ® delivers 0.9 mg crocins and 0.7 mg safranal per capsule [22], while Safr'Inside™ is standardized to >3% crocins, >0.2% safranal, >1% picrocrocin derivatives, and >0.1% kaempferol derivatives [5, 21].
Valerian evidence is split. The high-standardization Sleeproot® preparation (2% valerenic acid, 200 mg/day administered 1 hour before bed) produced significant actigraphy improvements in SOL and TST in subjects with PSQI ≥5 [1], and even a single dose produced significant improvements in TST on Day 1 [2]. By contrast, Taibi et al. found no significant benefit with 300 mg of 0.8% valerenic acid extract over 2 weeks in 16 older women assessed by PSG, actigraphy, and sleep diary [12]; WASO increased numerically (though not significantly versus placebo) with valerian [12]. The large internet-based Oxman trial (n=405) found only a marginally significant global self-assessment effect (difference 5.5%; p=0.04) with no significant improvement in the primary sleep-quality diary endpoint [23]. Several factors—dose standardization (0.8% versus 2% valerenic acid), population severity (older women with insomnia versus a general adult population self-reporting insomnia), and trial duration (2 weeks versus 8 weeks)—likely contribute to this divergence.
Lemon verbena (Aloysia citrodora/Lippia citriodora) at 400 mg/day for 90 days produced significant actigraphy improvements in SOL, WASO, SE, and number of awakenings (all p=0.001) [6] alongside increased nocturnal melatonin levels (199.7 vs. 174.7 pg/mL; p=0.048) [6]. The Martínez-Rodríguez et al. trial used the same ingredient and dose but relied on consumer-grade Fitbit wearables, reporting directionally consistent trends in REM (23.0% vs. 19.0%) and deep sleep (18.6% vs. 16.8%) without reporting statistical comparisons for these parameters [29].
GABA from fermented rice germ at 300 mg/day produced within-group PSG improvements in SOL (13.4 to 5.7 min; p=0.001) and SE (79.4% to 86.1%; p=0.018) in Byun et al. [8], with the absence of between-group reporting a limitation (the placebo arm had only n=10 due to 3:1 allocation) [8]. The low-dose 75 mg preparation yielded significant within-group reductions in PSG sleep latency (p=0.0038) and a notable increase in N3 sleep (8.6% to 12.9%; p=0.0043) [9], a finding not reported in the higher-dose trials. The dose-response study by Shin et al. found significant improvements in PSQI and ISI scores but non-significant actigraphy effects on SOL, TST, and SE [34, 34].
Heat-killed L. gasseri CP2305 is the only postbiotic with EEG-level sleep architecture data. In 60 medical students over 24 weeks, CP2305 significantly shortened SOL to N3 sleep, decreased WASO, and increased the delta power ratio in the first sleep cycle [14] with partial eta-squared values of 0.085–0.120 [14]. The same strain in an acute single-dose crossover design (n=28) significantly increased alpha power in the EEG after ingestion, with a concomitant increase in HRV RMSSD [10, 10]. These are the only trials in this review to directly demonstrate a neurophysiological mechanism—increased cortical alpha activity and cardiac vagal tone—acutely following a heat-killed postbiotic. Other postbiotics (HT-PS150, HK-PS23, SNK12) produced improvements only on subjective scales or wearable-based estimates [48, 50, 51].
Autonomic Outcomes
Autonomic data are sparse across the evidence base. The table below summarizes all studies that reported RMSSD, SDNN, HF-HRV, LF/HF ratio, or related markers.
| Ingredient | Study | Outcome | Direction and magnitude | p-value |
|---|---|---|---|---|
| L. gasseri CP2305 (acute) [10] | Tanihiro et al., 2026 | RMSSD | Significantly greater increase vs. placebo [10] | Not reported numerically [10] |
| PEA (Levagen+®, 600 mg/day) [11] | Deb et al., 2025 | SDNN | +9.70 ± 6.02 ms (PEA) vs. −5.72 ± 3.14 ms (placebo) [11] | p=0.024 [11] |
| PEA (Levagen+®, 600 mg/day) [11] | Deb et al., 2025 | RMSSD | Trend toward increase [11] | p=0.087 [11] |
| PEA (Levagen+®, 600 mg/day) [11] | Deb et al., 2025 | HF-HRV | No significant change [11] | p=0.965 [11] |
| Kaempferia parviflora 360 mg/day [54] | Eungpinichpong, 2018 | RMSSD | Higher in KP group [54] | p=0.033 [54] |
| Kaempferia parviflora 360 mg/day [54] | Eungpinichpong, 2018 | LF/HF ratio | Lower in KP group [54] | p=0.036 [54] |
| ETAS (asparagus extract) [55] | Ito et al., 2014 | Total power and ANS balance | Significant improvement [55] | p<0.05 and p<0.01 respectively [55] |
Only seven measurements from four trials provide autonomic endpoint data. PEA at 600 mg/day for 6 weeks in 16 stressed female university students increased SDNN (a global HRV marker associated with autonomic resilience) by 9.70 ms versus a reduction of 5.72 ms with placebo (p=0.024) [11], though RMSSD (reflecting cardiac vagal tone more specifically) only showed a non-significant trend (p=0.087) [11]. This is notable because the smaller sample size and lower statistical power constrain interpretation. The acute single-dose CP2305 study showed a significantly greater RMSSD increase, an important finding, but the study design (n=28 crossover, single dose, EEG study not a sleep trial) limits direct inference to sleep outcomes [10]. No study in this review prespecified RMSSD or HF-HRV as a primary sleep endpoint or measured peri-sleep-onset cardiac vagal tone in a dedicated sleep protocol.
Subjective and Neuroendocrine Outcomes
The table below presents PSQI, ISI, cortisol, and other validated subjective outcomes where reported with placebo-comparison data.
| Ingredient | Study | PSQI (active vs. placebo) | ISI (active vs. placebo) | Cortisol/neuroendocrine | Key finding |
|---|---|---|---|---|---|
| Ashwagandha KSM-66 (600 mg/day) [3] | Langade et al., 2019 | Baseline 13.07 (1.51) vs. 13.47 (1.38); 10-week: 9.15 (1.82) vs. 11.8 (1.46); p<0.0001 [3] | Not reported [3] | HAM-A improved; p=0.002 [3] | Significant PSQI improvement |
| Ashwagandha Shoden® (120 mg/day) [4] | Deshpande et al., 2020 | Not reported [4] | Not reported [4] | Not reported [4] | RSQ-W: 72% vs. 29% increase in sleep quality (p<0.001) [4] |
| Ashwagandha KSM-66 (600 mg/day) [20] | Choudhary et al., 2017 | Not reported [20] | Not reported [20] | Serum cortisol: 16.05% reduction at 4 wks; 22.2% at 8 wks (p<0.05) [20] | Significant PSS and cortisol improvement |
| Ashwagandha Sensoril® (125–500 mg/day) [19] | Pandit et al., 2024 | Not reported [19] | Not reported [19] | Cortisol: dose-dependent reduction (all three doses vs. placebo; p<0.01 to p<0.001). Alpha-amylase: dose-dependent reduction [19] | Dose-dependent HPA suppression |
| Saffron Safr'Inside™ (20 or 30 mg/day) [21] | Schuster et al., 2025 | Not reported [21] | AIS improved vs. placebo (p<0.05); Cohen's d=0.45 [21] | Not reported [21] | SQS improved at 3 weeks (p<0.05 for 30 mg; p<0.01 for 20 mg); PSS reduced (p=0.01) [21] |
| Saffron Saffr'activ® (15.5 mg/day) [22] | Pachikian et al., 2021 | PSQI: sleep quality p=0.014; latency p=0.032; duration p=0.013; global p=0.001 improved within saffron arm [22] | Not reported [22] | Not reported [22] | LSEQ ease of sleep improved; no between-group actigraphy effect [22] |
| Saffron Safr'Inside™ (30 mg/day) [5] | Láng et al., 2025 | PSQI global: 21.3% reduction (saffron) vs. 8.6% increase (placebo); p=0.02 [5] | ISI: no significant effect [5] | Not reported [5] | Sleep efficiency subjective: improved (p=0.04) |
| Valerian Sleeproot® (200 mg/day) [1] | Shekhar et al., 2023 | PSQI significantly decreased vs. placebo on Days 14, 28, 56 [1] | Not reported [1] | Not reported [1] | BAI reduced; ESS reduced; VAS improved |
| Valerian (300 mg, 0.8% VA) [12] | Taibi et al., 2009 | Used in eligibility; no significant change reported [12] | Not reported [12] | Not reported [12] | No significant benefit on any measure |
| Valerian (600 mg/night) [23] | Oxman et al., 2007 | Mean 11.6 (2.7) vs. 11.5 (2.6); no significant change [23] | Not reported [23] | Not reported [23] | Global self-assessment marginally better (5.5%; p=0.04) |
| Valerian 700 mg/day [24] | Mirmohammadali et al., 2014 | Baseline 10.9 ± 3.6; valerian phases 7.8 ± 3.4 and 7.4 ± 3.0; p<0.001 [24] | ISI: 2.87 ± 0.62 after valerian vs. 4.02 ± 0.5 after placebo phase; p<0.001 [24] | Not reported [24] | Significant PSQI and ISI improvement in postmenopausal women |
| Melissa officinalis phytosome (400 mg/day) [25] | Di Pierro et al., 2024 | Not reported (ISI used) [25] | ISI: 6.8 ± 4.1 (treated) vs. 9.7 ± 3.7 (placebo); p=0.003 [25] | Not reported [25] | 87% vs. 30% self-reported sleep improvement (χ²; p=0.0003) [25] |
| Melissa officinalis (Relissa™, 400 mg/day) [26] | Bano et al., 2023 | Baseline 10.9 ± 4.3 vs. 8.6 ± 3.8; FU: 3.5 ± 2.6 vs. 6.6 ± 3.3; p<0.001 [26] | Not reported [26] | Not reported [26] | Significant PSQI and wellbeing improvement |
| Passiflora incarnata tea [15] | Ngan & Conduit, 2011 | Not reported [15] | Not reported [15] | Not reported [15] | Sleep diary quality: t(40)=2.70; p<0.01 |
| Chamomile 270 mg twice daily [27] | Zick et al., 2011 | Not reported [27] | Not reported [27] | Not reported [27] | No significant sleep diary differences; moderate effect sizes (d=0.47–0.61) for SOL, awakenings, fatigue |
| Chamomile vs. Passiflora tea [16] | Khalid et al., 2025 | PSQI significantly better with chamomile vs. passionflower and control [16] | Not reported [16] | Cortisol reduced with chamomile (5.3→3.1 µg/dL; p<0.01); not with passionflower [16] | Melatonin increased with chamomile (p<0.001) [16] |
| Scutellaria lateriflora (400 mg/day) [30] | Di Minno et al., 2025 | PSQI improved (p<0.001) [30] | Not reported [30] | Not reported [30] | VAS stress decreased (p<0.001) |
| Black cohosh (6 months) [31] | Jiang et al., 2015 | Decreased in BC group (r=0.601; p=0.001); not sig. vs. placebo (r=0.301; p=0.051) [31] | Not reported [31] | Not reported [31] | PSG WASO and SE significantly improved vs. placebo |
| Polygonatum sibiricum (500 mg/day) [28] | Ha et al., 2019 | AIS: significant group × visit interaction (z=−2.1; p=0.035) [28] | Not reported [28] | Not reported [28] | mPFC perfusion increased (p=0.001) |
| Lemon verbena (400 mg/day, 90 days) [6] | Pérez-Piñero et al., 2024 | PSQI overall: significant (p=0.008); latency p=0.027; efficiency p=0.023 [6] | Not reported [6] | Melatonin: 199.7 vs. 174.7 pg/mL (p=0.048); cortisol NS [6] | Significant actigraphy and questionnaire improvements |
| Lemon verbena (RelaxPLX, 400 mg/day, 8 weeks) [29] | Martínez-Rodríguez et al., 2022 | PSQI: 12.2% decrease vs. baseline (p<0.05) [29] | Not reported [29] | Cortisol: 15.6% decrease vs. baseline [29] | Perceived stress PSS: 10.7% decrease (p<0.05) |
| PEA (Levagen+®, 350 mg/day) [32] | Rao et al., 2021 | PSQI improved similarly in both groups [32] | Not reported [32] | Not reported [32] | SOL reduced; cognition on waking improved |
| L-theanine (450–900 mg/day) [33] | Sarris et al., 2019 | Not reported [33] | ISI overall NS (p=0.35); ISI item 4 (satisfaction) improved (p=0.015); ISI ≤14 subgroup significant (p=0.007) [33] | Not reported [33] | No effect on anxiety (HAMA) |
| GABA 300 mg/day (fermented) [8] | Byun et al., 2018 | PSQI: 11.0→9.8 (p=0.003); components C1 (p=0.002) and C3 (p=0.004) improved [8] | ISI: 14.6→11.5; p<0.0001 [8] | Not reported [8] | Objective PSG SOL and SE also improved |
| GABA (100 mg and 300 mg) [34] | Shin et al., 2016 | PSQI total: both doses significantly decreased vs. placebo (p=0.017 low, p=0.004 high) [34] | ISI: dose-dependent decrease (p=0.037) [34] | Not reported [34] | Sleep latency subscale improved; WHO-5 wellbeing improved |
| GABA 75 mg low-dose [9] | Yoon et al., 2022 | PSQI: 11.2→8.7 (p<0.0001 within-group) [9] | ISI: 15.3→10.4 (p<0.0001 within-group) [9] | Not reported [9] | Similar improvements in placebo group (PSQI 10.3→8.3; p=0.002); no between-group tests reported |
| Alpha-s1 casein hydrolysate (Lactium®) [43] | Kim et al., 2019 | PSQI: 9.79→8.51 (ACH); 9.79→8.55 (placebo); NS group difference [43] | ISI: 12.08→10.04 (ACH); NS group difference [43] | Not reported [43] | Sleep diary: TST and SE improved with ACH; actigraphy SE improved at 4 weeks (d=0.47) |
| Fish hydrolysate (Peptidyss®) [45] | Eckert et al., 2024 | PSQI improved with intervention (p=0.002); not during placebo phase [45] | Not reported [45] | Salivary cortisol: no significant changes [45] | Significant within-group improvement; no significant between-group difference |
| β-lactolin (whey peptide, 6 weeks) [46] | Ayabe et al., 2024 | Not reported [46] | Not reported [46] | Cortisol: NS; alpha-amylase: NS [46] | PSS trait anxiety improved (p=0.046); subjective stress PSS improved (p=0.043) |
| Melatonin 2 mg PRM (3 weeks, elderly) [37] | Luthringer et al., 2009 | Not reported [37] | Not reported [37] | Not reported [37] | LSEQ sleep quality: 50% vs. 15% substantial improvement (p=0.018); PSG SOL −9 min (p=0.02) |
| Melatonin 2 mg PRM (26 weeks) [38] | Wade et al., 2011 | PSQI improved significantly with PRM [38] | Not reported [38] | Not reported [38] | Improvement maintained over 6 months; no rebound |
| Melatonin 6 mg (ADT-induced insomnia) [39] | Etedali et al., 2022 | PSQI: 4 domains improved (p=0.04) [39] | Not reported [39] | Not reported [39] | Anxiety and depression: NS |
| Melatonin 0.5 mg (DSWPD) [40] | Sletten et al., 2018 | PSQI lower with melatonin (p=0.037) [40] | ISI lower with melatonin (p=0.035) [40] | Not reported [40] | 52.8% vs. 24.0% clinician-rated much improved (p<0.05) |
| Melatonin 5 mg (post-acute withdrawal) [13] | Bondi et al., 2018 | PSSQ: no significant between-group difference [13] | Not reported [13] | Not reported [13] | No significant differences on any outcome |
| Agomelatine 25 mg (ASD+ID) [41] | Ballester et al., 2019 | Not reported [41] | Not reported [41] | Not reported [41] | TST +83 min; circadian phase correction |
| Lemborexant 5 mg/10 mg [35] | Kärppä et al., 2020 | Not reported [35] | Not reported [35] | Not reported [35] | Significant improvements in all subjective diary endpoints |
| Dimdazenil 2.5 mg [36] | Huang et al., 2023 | Not reported [36] | ISI significantly improved [36] | Not reported [36] | All subjective and objective endpoints improved |
| Rice bran extract (1000 mg/day) [42] | Um et al., 2019 | PSQI decreased in both groups; no significant between-group difference [42] | Not reported [42] | Not reported [42] | ESS significantly lower in RBS group |
| Corn leaf extract (CLE, 500 mg/day) [7] | Doma et al., 2026 | PSQI: both groups decreased (NS group difference) [7] | Not reported [7] | Melatonin decreased in CLE at Day 14 (p=0.002); GABA and serotonin NS [7] | Actigraphy/EEG showed large effects |
| HT-PS150 (L. fermentum, 8 weeks) [48] | Lee et al., 2025 | PSQI: improved over time in both groups; no significant interaction [48] | ISI: improved over time; no significant interaction [48] | Cortisol: 69.2% vs. 52.5% showed reduction; p=0.128 [48] | Subgroup ISI ≥8: greater improvement |
| L. gasseri CP2305 (24 weeks, students) [14] | Nishida et al., 2019 | PSQI significantly improved vs. placebo [14] | Not reported [14] | Salivary CgA significantly lower; cortisol NS [14] | STAI trait anxiety significantly reduced |
| L. gasseri CP2305 (5 weeks, students) [17] | Nishida et al., 2017 | PSQI improved in CP2305 group (5.44→4.04; p=0.038); significant in males (p=0.004) but not females (p=0.760) [17] | Not reported [17] | CgA, cortisol, alpha-amylase: NS [17] | Sex-specific PSQI response |
| L. gasseri CP2305 (menopausal women) [49] | Sawada et al., 2022 | Not reported [49] | Not reported [49] | Not reported [49] | SMI and GCS menopausal scores improved |
| HK-PS23 (L. paracasei, 8 weeks) [51] | Wu et al., 2022 | Not reported [51] | ISI used for screening; specifics NS [51] | Cortisol: significant reduction in HK-PS23 group (p=0.043) [51] | Anxiety improved in high-anxiety subgroup (STAI ≥103) |
| L. plantarum SNK12 (4 weeks) [50] | Watanabe et al., 2025 | Not reported [50] | Not reported [50] | Salivary cortisol: −0.132 µg/dL (95% CI −0.239 to −0.026); p=0.016. TNF-α: reduced (p=0.037) [50] | OSA-MA Sleepiness on Rising improved (p=0.032); Initiation/Maintenance improved (p=0.010) |
| L. plantarum L-137 (LP20, 12 weeks) [52] | Hirose et al., 2013 | Not reported [52] | Not reported [52] | Not reported [52] | URTI incidence lower; immune function improved |
| LC-Plasma (L. lactis JCM 5805, 13 days) [53] | Komano et al., 2018 | Not reported [53] | Not reported [53] | Not reported [53] | Fatigue symptom days reduced |
| Kaempferia parviflora (360 mg, 14 days) [54] | Eungpinichpong, 2018 | Not reported [54] | Not reported [54] | Alpha-amylase: mentioned but no quantitative values [54] | HAM-A and SPST-20: both groups improved, NS between groups |
| ETAS (asparagus, 150 mg/day, 7 days) [55] | Ito et al., 2014 | Not reported [55] | Not reported [55] | Salivary CgA significantly decreased (9.4→5.7 pmol/mL; p<0.01) [55]. Cortisol unchanged with ETAS, increased with placebo [55] | AIS "early awakening" improved (p<0.05) |
| Piceatannol (PIC, 10 mg/day) [47] | Tsukamoto-Sen et al., 2026 | Not reported [47] | Not reported [47] | Not reported [47] | WASO reduced on actigraphy; subjective NS |
| Alpha-s1 casein hydrolysate (150 mg/day) [44] | Kim et al., 2007 | Not reported [44] | Not reported [44] | Not reported [44] | Stress-related symptom questionnaire: digestion (p<0.01), cardiovascular (p<0.05), intellectual (p<0.01), emotional (p<0.05), social problems (p<0.05) improved |
Discussion of Subjective and Neuroendocrine Effects
Across subjective outcomes, the most consistent signal emerges for ashwagandha and saffron. Ashwagandha preparations using KSM-66 or Shoden® standardization consistently improve self-reported sleep quality, with the KSM-66 10-week study showing a PSQI decrease from 13.07 to 9.15 versus 11.8 in placebo (p<0.0001) [3]. Cortisol reduction with ashwagandha is dose-dependent in the Sensoril® dose-ranging trial, reaching statistical significance at all three doses (125, 250, 500 mg/day) with an ascending magnitude [19]. The KSM-66 study by Choudhary et al. reported a 22.2% serum cortisol reduction after 8 weeks (p<0.05) [20], providing objective neuroendocrine evidence for HPA-axis modulation.
Saffron at 30 mg/day (Safr'Inside™) produced the most complete evidence profile in older adults, with simultaneous improvements in EEG-based objective sleep (SOL, LtPS, WASO), PSQI global score (21.3% reduction versus 8.6% increase in placebo; p=0.02), and microbiome shifts in SCFA-producing bacteria [5, 5]. The two-arm Schuster et al. trial similarly found significant improvements in AIS (Cohen's d=0.45), SQS, and PSS with both 20 mg and 30 mg doses [21, 21], suggesting the active standardization is more determinative than dose within this range.
For valerian, the subjective evidence is more favorable than the objective data: PSQI and ISI improved significantly in postmenopausal women using 700 mg/day over two crossover phases [24], and the higher-standardization Sleeproot® preparation produced consistent multi-timepoint PSQI improvement in a parallel trial [1]. The Taibi et al. crossover study in 16 older women and the large Oxman internet trial both failed to demonstrate significant PSQI improvement [12, 23], with the Taibi trial uniquely showing WASO numerically worsening with valerian [12]. The population and dose differences between these trials are discussed in the synthesis section.
For heat-killed postbiotics, the L. gasseri CP2305 trials present a coherent mechanistic picture: reduced salivary chromogranin A (a sympathetic stress marker), improved PSQI (particularly in males) [17], improved STAI anxiety scores, shortened EEG-measured sleep latency, increased delta power [14], and acute enhancement of cortical alpha power and RMSSD [10, 10] —all pointing toward HPA-axis and autonomic modulation via the gut-brain axis. Heat-killed SNK12 (L. plantarum) separately demonstrated reduced salivary cortisol (−0.132 µg/dL; p=0.016) and TNF-α (p=0.037) with concomitant improvement in two OSA-MA sleep factors [50]. HK-PS23 (L. paracasei) reduced blood cortisol significantly (p=0.043) and improved anxiety in a high-anxiety subgroup [51]. These convergent findings across distinct postbiotic strains suggest a class effect involving HPA-axis dampening and pro-vagal modulation.
Adverse Events
The table below summarizes the adverse event profile across included studies.
| Ingredient | Study | Notable adverse events | Serious AEs | Withdrawals for AE |
|---|---|---|---|---|
| Ashwagandha KSM-66 (600 mg/day) [3] | Langade et al., 2019 | None reported [3] | None [3] | Not reported [3] |
| Ashwagandha Shoden® (120 mg/day) [4] | Deshpande et al., 2020 | Mild, unrelated to treatment [4] | None [4] | None [4] |
| Ashwagandha Sensoril® (125–500 mg/day) [19] | Pandit et al., 2024 | 48% reported AEs: heartburn, abdominal discomfort, insomnia [19] | None [19] | Not specified [19] |
| Saffron Safr'Inside™ (20–30 mg/day) [21] | Schuster et al., 2025 | GI discomfort (dose-dependent: 10 vs. 6 vs. 4 cases) [21] | None [21] | None reported [21] |
| Saffron Saffr'activ® (15.5 mg/day) [22] | Pachikian et al., 2021 | One palpitation event [22] | None [22] | One (possibly product-related) [22] |
| Valerian Sleeproot® (200 mg/day) [1] | Shekhar et al., 2023 | Upper respiratory, headache, GI (judged unrelated) [1] | None [1] | None [1] |
| Valerian (300 mg, 0.8% VA) [12] | Taibi et al., 2009 | Minor, no significant differences vs. placebo [12] | None [12] | None [12] |
| Valerian (600 mg/night) [23] | Oxman et al., 2007 | No serious AEs; no significant differences in minor AEs [23] | None [23] | Not reported [23] |
| Melissa officinalis phytosome (400 mg/day) [25] | Di Pierro et al., 2024 | Not reported [25] | Not reported [25] | Not reported [25] |
| Melissa officinalis (Relissa™) [26] | Bano et al., 2023 | Mild stomach upset (equally distributed) [26] | None [26] | None [26] |
| Passiflora incarnata tea [15] | Ngan & Conduit, 2011 | Not reported [15] | Not reported [15] | Not reported [15] |
| Scutellaria lateriflora (400 mg/day) [30] | Di Minno et al., 2025 | None reported [30] | None [30] | None [30] |
| PEA (Levagen+®, 350 mg/day) [32] | Rao et al., 2021 | Loss of taste, reduced sleep quality (PEA); GI, dizziness, worse sleep (placebo) [32] | Not reported [32] | 6 total (2 PEA, 4 placebo) [32] |
| PEA (Levagen+®, 600 mg/day) [11] | Deb et al., 2025 | No adverse effects [11] | None [11] | None [11] |
| L-theanine (450–900 mg/day) [33] | Sarris et al., 2019 | Sleep disturbance, drowsiness, GI (mild) [33] | None [33] | Not explicitly stated [33] |
| GABA 300 mg/day [8] | Byun et al., 2018 | Abdominal discomfort, headache, drowsiness (4/40 subjects) [8] | Not reported [8] | Not reported [8] |
| GABA 100/300 mg/day [34] | Shin et al., 2016 | 5 patients (4.4%) reported AEs [34] | Not reported [34] | Not reported [34] |
| GABA 75 mg [9] | Yoon et al., 2022 | None reported [9] | None [9] | None [9] |
| Rice bran extract (1000 mg/day) [42] | Um et al., 2019 | 16 mild AEs total; none serious [42] | None [42] | Not reported [42] |
| Alpha-s1 casein hydrolysate [43] | Kim et al., 2019 | Itching/urticaria in one placebo subject; none in ACH [43] | None [43] | None [43] |
| Fish hydrolysate (Peptidyss®) [45] | Eckert et al., 2024 | 211 total; 2 serious (unrelated); 195 mild [45] | 2 (unrelated) [45] | Not reported [45] |
| Piceatannol (PIC, 10 mg/day) [47] | Tsukamoto-Sen et al., 2026 | 10 (PIC) vs. 24 (placebo); all mild [47] | None [47] | None [47] |
| Corn leaf extract (500 mg/day) [7] | Doma et al., 2026 | One diarrhea case in placebo group [7] | None [7] | None [7] |
| Melatonin 2 mg PRM (Circadin, 3 weeks) [37] | Luthringer et al., 2009 | Headaches (equal in both groups); no treatment-related AEs [37] | None [37] | None [37] |
| Melatonin 2 mg PRM (26 weeks) [38] | Wade et al., 2011 | Most mild; no sig. differences vs. placebo [38] | 42 reported; 1 possibly PRM-related; 1 death (placebo) [38] | 59 total; rates similar [38] |
| Melatonin 6 mg (prostate cancer ADT) [39] | Etedali et al., 2022 | Excessive daytime sleepiness in 3 (NS) [39] | None [39] | None [39] |
| Melatonin 0.5 mg (DSWPD) [40] | Sletten et al., 2018 | Light-headedness, daytime sleepiness, decreased libido (similar rates both arms) [40] | None [40] | None [40] |
| Melatonin 5 mg (substance use recovery) [13] | Bondi et al., 2018 | Fatigue, headache, nightmares (greater frequency in melatonin group) [13] | None [13] | 1 per group [13] |
| Agomelatine 25 mg (ASD+ID) [41] | Ballester et al., 2019 | Mild and transient; one aminotransferase elevation [41] | None [41] | 1 (hepatic enzymes) [41] |
| Lemborexant 5/10 mg [35] | Kärppä et al., 2020 | Somnolence most common (higher in LEM10); low serious AE rate [35] | Low rate, no deaths [35] | Higher in LEM10 due to somnolence [35] |
| Dimdazenil 2.5 mg [36] | Huang et al., 2023 | Dizziness 12.53% vs. 5.06%; headache 4.36% vs. 2.25% [36] | 0.54% vs. 1.12% [36] | 3 (dizziness, vertigo, skin rash) vs. 1 [36] |
| All heat-killed postbiotics (multiple) [14, 49, 50] | Multiple | No adverse events observed across CP2305, SNK12, and Sawada CP2305 trials [14, 49, 50] | None [14] | None [14] |
| Kaempferia parviflora (360 mg) [54] | Eungpinichpong, 2018 | One flatulence [54] | None [54] | None [54] |
Across the botanical and postbiotic interventions, the safety profile is broadly favorable. The most notable finding is that the Sensoril® ashwagandha dose-ranging trial reported AEs in 48% of participants, primarily heartburn, abdominal discomfort, and trouble sleeping [19], though no serious events occurred. This rate is higher than reported in other ashwagandha trials, possibly reflecting the three-arm design with more systematic event capture, the use of a combined root and leaf extract, or dose-related effects. By contrast, heat-killed postbiotics across all strains showed no adverse events in trials that systematically monitored them [14, 49, 50]. The pharmaceutical benchmarks demonstrated dose-related somnolence with lemborexant (particularly at 10 mg) [35] and dizziness with dimdazenil (12.5% versus 5.1% placebo) [36], consistent with their GABAergic and orexin-blocking mechanisms of action. Melatonin at 5 mg in a substance-use recovery population showed a numerically higher frequency of adverse events (fatigue, nightmares) compared to placebo, though not statistically significant [13]. Daytime psychomotor performance was not impaired and was better with prolonged-release melatonin 2 mg compared to placebo in the Luthringer et al. PSG trial [37].
Formulation-Specific Evidence for Hot-Beverage Sachet Delivery
Among the 55 included studies, three employed beverage-matrix delivery: the Passiflora incarnata herbal tea at 2 g/teabag steeped in boiling water [15], the Khalid et al. chamomile and passionflower powder at 1 g in hot water [16], and the Nishida et al. 2017 fermented milk beverage containing heat-killed CP2305 [17]. None of these trials reported post-brew assay recovery, volatile loss quantification, marker stability data, or matrix interaction experiments at 60–90°C. The Nishida et al. 2017 beverage study confirmed absence of live bacteria after sterilization at 95°C for 30 seconds (continuous sterilizer) [17], establishing delivery-system integrity in the sense that the heat-killed postbiotic state was maintained through thermal processing, but no bioavailability or recovery data were reported.
For botanical extracts used in capsule form, the lemon verbena extraction process involved maceration at <80°C followed by vacuum drying at 60–80°C [6, 29], and the corn leaf extract underwent ethanol extraction at 80°C and spray-drying at 140°C [7]. The asparagus-based ETAS ingredient was extracted with hot water at 121°C for 45 minutes [55]. No study assessed how these thermal processing conditions affect final bioactive content in a reconstituted beverage medium, and no trial in this review evaluated a hot-beverage sachet format for any of the primary target ingredients (PEA, L-theanine, standardized valerian, ashwagandha, GABA, Melissa officinalis phytosome, or heat-killed postbiotics). Accordingly, no clinical or physicochemical evidence exists in this dataset to support or refute the feasibility of hot-beverage sachet delivery for any of these ingredients.
Synthesis
The Central Tension: Standardization, Population, and the Meaning of "Null"
Several apparent contradictions across the trial evidence resolve when population severity, ingredient standardization, and outcome measurement methodology are considered systematically.
Valerian: standardization and population severity drive divergence.
The single clearest example involves valerian. Taibi et al. found no significant effect—and a numerically adverse WASO trend—using 300 mg of 0.8% valerenic acid in 16 older women with PSG-confirmed insomnia, assessed over 2 weeks [12, 12]. Oxman et al. found only marginal self-reported benefit in 405 adults who self-identified as having insomnia but were recruited via television [23]. By contrast, two trials using the Sleeproot® preparation (2% valerenic acid) in younger adults (18–50 years) with PSQI ≥5 showed significant actigraphy improvements in SOL and TST over 8 weeks, including an acute Day 1 TST improvement with a single dose [1, 2]. Mirmohammadali et al. using 700 mg in postmenopausal women found significant PSQI and ISI improvements [24]. The 2.5-fold difference in valerenic acid content between preparations (0.8% versus 2%) likely produces meaningfully different GABA receptor effects given the known dose-dependence of valerenic acid binding [1, 2]. Additionally, the older women (mean age 69.4 years) in Taibi et al. may represent a population with more entrenched insomnia pathophysiology less responsive to mild GABAergic modulation, while the younger adult populations in the Sleeproot® trials may have a more functional insomnia phenotype with greater treatment headroom. Notably, the Taibi trial also had a 2-week duration, while the positive valerian trials ran 8 weeks [1, 2, 12] —a timeline consideration given valerian's proposed mechanism of gradual GABA modulation.
GABA: within-group versus between-group significance and a low-dose architecture signal.
The three GABA trials collectively illustrate several methodological tensions. Byun et al. used PSG and a 3:1 allocation ratio (30 GABA, 10 placebo), reporting only within-group improvements for their primary endpoints (SOL p=0.001; SE p=0.018) [8]; the underpowered placebo arm prevents inference about placebo-adjusted effects. Shin et al. used actigraphy over 1 week and found significant PSQI and ISI improvements (dose-dependently, p=0.017–0.037) but non-significant actigraphy effects on SOL, TST, and SE [34, 34] —suggesting questionnaire sensitivity may exceed objective measurement sensitivity over short timeframes at these doses. The most interesting finding comes from Yoon et al., where low-dose 75 mg GABA showed a PSG-confirmed increase in N3 sleep (8.6% to 12.9%; p=0.0043) and reduced arousal index (p=0.0316) [9], whereas SE improvement was non-significant [9]. The parallel placebo arm showed comparable ISI and PSQI improvements [9], raising the possibility that the within-group improvements reflect regression to the mean or non-specific effects, but the N3 architecture finding has no obvious regression explanation and warrants replication with between-group testing. If the N3 and arousal index findings are specific to GABA, they suggest a different mechanism and dose-response curve than the global sleep efficiency endpoint used in higher-dose studies.
Saffron: dose standardization predicts outcome measurement selectivity.
The Saffr'activ® preparation at 15.5 mg/day showed questionnaire improvements but no actigraphy effects [22, 22]. Safr'Inside™ at 30 mg/day showed EEG-confirmed objective sleep improvements alongside PSQI improvements in older adults [5, 5]. Both preparations are standardized to different marker profiles [5, 21, 22], and the higher-dose/better-standardized preparation demonstrates the more complete evidence profile. The Schuster et al. trial (n=165) provides the most rigorous dose comparison (20 mg versus 30 mg), finding similar effects on AIS and PSS but differing timing of SQS response (30 mg significant at 3 weeks, 20 mg at 3 weeks for a different metric) [21] —insufficient to establish a clear dose threshold but suggesting that standardized delivery within the 20–30 mg range is effective for subjective endpoints.
Heat-killed postbiotics: converging mechanistic evidence with population specificity.
A consistent pattern emerges across heat-killed postbiotic trials: effects are most evident in populations with a defined stressor (students under examination stress, stressed nurses, menopausal women, people with ISI ≥8). The L. gasseri CP2305 24-week trial in 60 medical students showed significant EEG sleep architecture changes and PSQI improvement [14, 14]. The 5-week CP2305 fermented milk study in undergraduate students showed PSQI improvement significant in males but not females [17]. The HT-PS150 trial in unselected poor sleepers (PSQI ≥5) found no significant group-time interaction on PSQI or ISI in the full cohort, but the subgroup with ISI ≥8 showed significantly greater PSQI improvements [48]. Wu et al. found significant cortisol reduction overall but anxiety improvement only in the high-anxiety subgroup (STAI ≥103) [51]. Watanabe et al. targeting morning fatigue and sleep dissatisfaction (a more specific phenotype) found two significant OSA-MA factors alongside cortisol and TNF-α reductions [50]. This pattern is consistent with a mechanism where HPA-axis modulation is most detectable against a background of elevated hypothalamic-pituitary-adrenal activation—as observed in populations under psychological stress—and less detectable in mild, non-specific poor sleepers without elevated cortisol at baseline.
The acute single-dose CP2305 EEG/HRV study [10, 10] provides direct neurophysiological support for this mechanistic pathway: RMSSD and alpha power both increased within 60 minutes of a single dose in healthy young adults. Alpha power enhancement is associated with reduced cortical arousal and parasympathetic upregulation. That these effects occur acutely and in healthy volunteers suggests the mechanism is not exclusively dependent on chronic gut microbiome remodeling but may also involve rapid gut-vagal signaling via serotonin secretion from enterochromaffin cells, as the in vitro data from the same study showed elevated 5-HT levels under CP2305 exposure [10].
Pharmaceutical benchmarks and the magnitude problem.
The dimdazenil Phase III data provide the clearest effect-size anchors: PSG-confirmed −20.16 min WASO, +31.68 min TST, and +5.55% SE versus placebo [36]. The corn leaf extract CLE trial reported effect sizes (d=0.84 for WASO, d=1.01 for SE) numerically exceeding those of dimdazenil, which—given CLE's proposed mechanism via 6-MBOA and melatonin receptor binding [7] and the modest trial size (n=40 per arm) [7] —should be regarded as preliminary pending larger, independently conducted replication. The ashwagandha actigraphy effects (SOL, WASO, TST, SE all improved in two trials with n≥60) occupy a plausible intermediate range consistent with a mild GABAergic and cortisol-lowering mechanism. The natural compound trials with the most extensive objective evidence—saffron (Safr'Inside™ 30 mg), ashwagandha (Shoden® and KSM-66), lemon verbena (400 mg, 90 days), valerian (2% VA, 8 weeks), and GABA (300 mg PSG, 75 mg N3 architecture data)—collectively demonstrate effects on SOL, WASO, or TST that are statistically significant but in most cases smaller in absolute magnitude than the pharmaceutical comparators.
The main unresolved question for the cardiac vagal autonomic endpoint is whether any of the botanical or postbiotic ingredients produce a clinically meaningful, sustained RMSSD or HF-HRV increase in a dedicated sleep protocol. The PEA SDNN finding (p=0.024, n=16) [11] and the acute CP2305 RMSSD finding [10] are promising but neither involves peri-sleep-onset autonomic recording or a population with primary insomnia. A study design combining actigraphy, sleep-period HRV, and a standardized stressed or insomnia population would be needed to resolve whether the observed autonomic effects translate to improvements in sleep-onset cardiac vagal modulation, which is the Tier 1 autonomic endpoint specified in this review protocol.
