The effect of hydrogen-rich water interventions on lipid profiles in adults with overweight or obesity and associated metabolic disorders: a systematic review and meta-analysis of randomized controlled trials

Abstract

Objective

Obesity-related dyslipidemia remains a significant health challenge, contributing to increased cardiovascular risk. Hydrogen-rich water (HRW) has been proposed as a potential adjunctive intervention due to its antioxidant and anti-inflammatory properties. This systematic review and meta-analysis aimed to evaluate the effects of HRW on lipid profiles in adults with overweight or obesity.

Methods

We searched Scopus, PubMed, Web of Science, and Embase for randomized controlled trials (RCTs) published up to November 2025 that compared HRW with placebo or standard water in adults with mean body mass index (BMI) ≥ 25 kg/m². Random-effects meta-analyses were performed to estimate pooled weighted mean differences (WMDs) and 95% confidence intervals (CIs).

Results

Twelve articles comprising 13 RCTs (n = 757 participants) were included. HRW significantly reduced total cholesterol (TC) (WMD: − 6.71 mg/dL; 95% CI: − 10.38 to − 3.04; P < 0.001) and low-density lipoprotein cholesterol (LDL-C) (WMD: − 3.21 mg/dL; 95% CI: − 6.31 to − 0.10; P = 0.043). A small reduction in high-density lipoprotein cholesterol (HDL-C) was observed (WMD: − 1.16 mg/dL; 95% CI: − 1.92 to − 0.40; P = 0.003). Triglycerides showed no significant change (WMD: − 4.21 mg/dL; 95% CI: − 17.75 to 9.32; P = 0.54). Most trials enrolled overweight or obese adults with cardiometabolic comorbidities. Subgroup analyses were underpowered and considered hypothesis-generating.

Conclusion

HRW supplementation in overweight or obese adults produces small, statistically detectable but clinically modest reductions in TC and LDL-C, with minimal effects on TG and HDL-C. These effects are below thresholds considered meaningful for cardiovascular risk reduction and should not justify routine use of HRW as a lipid-lowering therapy outside of research settings and these findings primarily apply to adults with preexisting metabolic conditions and should not be generalized to all overweight/obese individuals. Any potential role remains limited to an investigational adjunct, and further adequately powered, longer-term RCTs are needed before HRW can be considered in dyslipidemia management.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13098-026-02124-0.

Introduction

The cornerstone of management for obesity-related dyslipidemia is therapeutic lifestyle change, encompassing dietary modification, increased physical activity, and weight reduction [1–3]. While these interventions are effective, their real-world utility is often limited by poor long-term patient adherence [4]. This creates a clear unmet clinical need for safe, well-tolerated, and easily adhered-to adjunctive therapies that can address lipid abnormalities in overweight and obese populations.

Oxidative stress and inflammation play key roles in the pathophysiology of obesity-related dyslipidemia [5, 6], making these pathways attractive therapeutic targets. Hydrogen-rich water (HRW) has been proposed as a well-tolerated intervention with potential antioxidant effects [7, 8]; HRW selectively neutralizes harmful reactive oxygen species involved in lipid peroxidation [7, 9–11], while sparing essential redox-signaling molecules [12]. In addition, HRW exhibits anti-inflammatory and cytoprotective effects [8, 13, 14], and may support endogenous antioxidant defenses [15].

Preclinical and early clinical studies suggest that HRW may benefit metabolic health, including potential improvements in lipid metabolism [9]. Several trials have reported reductions in serum total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C), decreases in oxidative stress markers [16], and improved high-density lipoprotein cholesterol (HDL-C) functionality [17]. However, evidence from randomized controlled trials (RCTs) remains inconsistent [14].

A major challenge in interpreting existing evidence is the substantial heterogeneity across studies. Prior trials have enrolled diverse populations, including individuals with metabolic syndrome, type 2 diabetes, hyperuricemia, and non-alcoholic fatty liver disease (NAFLD), and have varied widely in hydrogen dose (< 3 to > 12 mg/day) and intervention duration (8–24 weeks). Although recent meta-analyses have synthesized these findings [14, 18], pooling such heterogeneous populations may obscure effects within specific risk groups. Consequently, the efficacy of HRW in individuals with overweight or obesity, those at high risk for developing atherogenic dyslipidemia, has not been clearly defined.

Previous meta-analyses, including those by Todorovic et al. [18] and Jamialahmadi et al. [14], evaluated HRW across heterogeneous clinical populations without restriction by adiposity status, limiting inference for overweight or obese adults specifically. By restricting inclusion to RCTs with mean baseline BMI ≥ 25 kg/m², the present review focuses on a more homogeneous, clinically relevant population at elevated cardiometabolic risk. This approach enables clearer interpretation of lipid-specific effect sizes in individuals most likely to consider non-pharmacologic adjuncts before initiating drug therapy. Importantly, this narrower focus was designed to yield effect-size estimates that are more directly interpretable for overweight or obese adults and may differ from those reported in prior reviews. Any potential clinical relevance of HRW, if present, would most plausibly be limited to pre-pharmacologic or adjunctive contexts rather than therapeutic lipid lowering.

Therefore, the primary objective of this systematic review and meta-analysis was to quantitatively evaluate the effects of HRW on TC, triglycerides (TG), HDL-C, and LDL-C in adults with overweight or obesity, predominantly in the presence of cardiometabolic comorbidities. By synthesizing evidence exclusively from RCTs conducted in this target population, we aimed to provide a robust estimate of treatment effects, explore potential sources of heterogeneity, and clarify the potential role of HRW in managing obesity-related dyslipidemia.

Materials and methods

This study was designed as a systematic review and meta-analysis of RCTs conducted in human adults. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. Rather than representing a single prospective or retrospective cohort, this work synthesizes evidence from multiple independently conducted RCTs across different countries and clinical settings, providing a multicenter evidence base for the effects of hydrogen-rich water (HRW) on lipid outcomes.

Search strategy

A comprehensive literature search was carried out across multiple electronic databases, including Scopus, Web of Science, Embase, and PubMed/MEDLINE, to identify relevant studies published up to November 20, 2025. Both Medical Subject Headings (MeSH) and free-text keywords were used to maximize search sensitivity. The main terms included various names and synonyms of HRW and its related compounds (see Supplementary Table 1 for details). No restrictions were placed on language or publication date to ensure a broad scope. The search was not restricted to human studies; however, during screening, only human RCTs were included according to the predefined eligibility criteria. Additionally, reference lists of all included studies were manually screened to capture any further eligible articles. Although no language restrictions were applied, all eligible studies identified and included in the review were published in English. Non-English articles retrieved during the search were screened, but none met the inclusion criteria at full-text review.

Inclusion and exclusion criteria

Study eligibility was determined using the PICOS framework: Population (P): Eligible studies were required to include adult participants (≥ 18 years) with a mean baseline body mass index (BMI) ≥ 25 kg/m², corresponding to overweight or obesity. Trials enrolling participants with associated cardiometabolic conditions (e.g., metabolic syndrome, NAFLD, hyperuricemia, impaired fasting glucose (IFG), or type 2 diabetes) were eligible provided that the reported mean BMI met the predefined threshold. Studies including children, normal-weight populations (mean BMI < 25 kg/m²), animals, or in vitro designs were excluded; Intervention (I): Studies evaluating the effects of HRW; Comparison (C): A placebo or no-intervention control group; Outcomes (O): Studies had to report mean and standard deviation (SD) values for at least one lipid profile parameter, TC, TG, HDL-C, or LDL-C, both before and after the intervention; Study design (S): Only RCTs were included. Exclusion criteria were as follows: (1) editorials, letters, or reviews; (2) absence of a comparator group; and (3) lack of relevant lipid outcome data. Although some eligible RCTs enrolled participants with broader metabolic conditions (e.g., metabolic syndrome, NAFLD, hyperuricemia), all included studies reported mean baseline BMI values ≥ 25 kg/m², indicating that their study populations met our predefined criteria for overweight or obesity. No trial included a participant subgroup with a mean BMI below this threshold, and no subgroup extraction was required. Additional exclusion criteria were: non-randomized designs, retrospective observational studies, lack of a control group, absence of extractable lipid data, conference abstracts without full text, and duplicate publications.

Data extraction

Two reviewers independently extracted data using a standardized form. Information collected included: first author, publication year, sample size, number of participants in intervention and control groups, mean age and BMI of participants, intervention duration, study location, study design, and dosage of HRW. Outcome measures included mean and SD values for lipid parameters (TG, HDL-C, TC, and LDL-C), along with participants’ health status. Any discrepancies were resolved by discussion, with unresolved disagreements referred to a senior author for final decision-making.

Definition and measurement of outcomes

The primary outcomes of this meta-analysis were changes in LDL-C and TC, given their established causal associations with cardiovascular disease risk. Secondary outcomes included TG and HDL-C. All lipid parameters were extracted as continuous variables and expressed in mg/dL as reported in the original trials (or converted when necessary). Outcomes were defined as the mean change from baseline to the end of the intervention period. Baseline measurements were obtained prior to randomization in each RCT, and follow-up measurements were collected at the end of the assigned HRW or control intervention period (ranging from 8 to 24 weeks across studies). Lipid concentrations in the original trials were measured using standard enzymatic or automated clinical chemistry assays on fasting blood samples, as reported by each study.

Definition of exposure (hydrogen-rich water)

The exposure of interest was hydrogen-rich water (HRW). HRW was operationally defined as drinking water containing dissolved molecular hydrogen (H₂), delivered either as pre-packaged HRW or generated in situ using hydrogen-producing tablets or caplets dissolved in water. Across trials, participants consumed between 500 mL and 2,000 mL per day. Hydrogen concentration ranged approximately from 0.5 to 7.5 ppm, corresponding to estimated daily hydrogen doses of ~ 3 to > 12 mg/day. For quantitative synthesis and subgroup analyses, daily hydrogen exposure was standardized and categorized as < 10 mg/day versus ≥ 10 mg/day based on reported volume and hydrogen concentration in each study. Intake was typically distributed across the day according to trial protocols.

Definition of covariates and subgroup variables

Pre-specified covariates for subgroup analyses included baseline lipid status, hydrogen dose, and intervention duration. Baseline TC was categorized using conventional clinical thresholds (< 200 vs. ≥ 200 mg/dL), and baseline LDL-C was categorized as < 130 versus ≥ 130 mg/dL, consistent with widely used dyslipidemia definitions. Intervention duration was classified as short-term (< 12 weeks) or longer-term (≥ 12 weeks). Hydrogen dose was categorized as < 10 mg/day versus ≥ 10 mg/day based on the empirical distribution of exposure across included trials. These cut-off values were selected a priori and aligned with clinical practice guidelines and prior HRW literature.

Quality assessment

The risk of bias (RoB) for included studies was assessed using the Cochrane RoB 2 tool, which evaluates the adequacy of randomization, adherence to interventions, completeness of outcome data, accuracy and objectivity of outcome measurement, risk of selective reporting, and overall study integrity. Based on these domains, studies were classified as having low RoB, some concerns, or high RoB [19, 20]. To evaluate the certainty of evidence, the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) framework was applied. Evidence quality was rated as very low, low, moderate, or high depending on study limitations, result consistency, precision, and potential publication bias [20]. Most studies enrolled volunteers from clinical or specialty-care settings, which may introduce spectrum bias, a situation in which study participants differ systematically from the general population, potentially limiting the generalizability of findings. This factor was considered when interpreting pooled estimates. This factor was considered when interpreting pooled estimates.

Statistical analysis

All analyses were conducted using Stata version 15 (StataCorp, College Station, TX, USA). Pooled effect sizes were estimated with the DerSimonian and Laird random-effects model, which accounts for between-study variability. Results were expressed as weighted mean differences (WMDs) with 95% confidence intervals (CIs) [21, 22]. Lipid outcomes were assessed at baseline and at the end of each intervention period, and pooled estimates reflect between-group differences in mean change over these predefined follow-up intervals.

When standard deviations (SDs) of change scores were not directly reported, they were estimated using the formula: √[(SD baseline² + SD end²) – (2 × R × SD baseline × SD end)], where R represents the assumed correlation between baseline and final measurements. In the primary analysis, a correlation coefficient of R = 0.5 was assumed in accordance with Cochrane Handbook recommendations. To evaluate the robustness of this assumption, sensitivity analyses were additionally performed using alternative plausible values (R = 0.25 and R = 0.75). The overall results were stable across these assumptions, indicating that the pooled estimates were not materially influenced by the choice of correlation coefficient.

Handling of missing data was addressed at both the trial and meta-analysis levels. When studies did not report standard deviations (SDs) for change scores, SDs were calculated from available baseline and follow-up data using recommended formulas with empirically derived correlation coefficients. If essential outcome data (means, SDs, or sample sizes) were not reported and could not be reconstructed from the manuscript or supplementary materials, the study or specific outcome was excluded from quantitative synthesis. Attrition within trials was considered in the risk-of-bias assessment under the domain of missing outcome data. Sensitivity analyses were performed to examine the robustness of pooled estimates to the inclusion or exclusion of individual studies.

A p-value of less than 0.05 was considered statistically significant. Between-study heterogeneity was assessed using Higgins’ I² statistic and the chi-squared (χ²) test. I² values were interpreted as: <30% (low), 31–74% (moderate), and > 75% (high). Although the review was not prospectively registered (e.g., PROSPERO), the analysis plan, outcome hierarchy, and subgroup definitions were documented internally prior to data extraction. For continuous study-level characteristics, subgroup cut-points were defined a priori using the median value across included trials to ensure balanced comparisons and minimize analytic flexibility.

Subgroup analyses were performed to explore potential sources of heterogeneity. To enhance transparency, we have included the detailed protocol, including the outcome hierarchy and subgroup analysis plan, in Appendix 1. Subgroup analyses based on baseline lipid levels, hydrogen dose, and intervention duration were pre-specified; however, several strata contained a limited number of trials and participants (see Appendix 2). Accordingly, these subgroup findings should be interpreted as hypothesis-generating rather than confirmatory and require validation in larger, adequately powered randomized trials.

To reduce ambiguity regarding analytic prioritization, we established an outcome hierarchy a priori. The primary outcomes of interest were LDL-C and TC, given their established causal associations with cardiovascular disease. Secondary outcomes included TG and HDL-C, which were assessed to provide a broader characterization of lipid metabolism but were not considered primary endpoints. Subgroup analyses were pre-specified prior to data extraction and limited to variables identified in prior HRW studies as potential effect modifiers: (1) baseline lipid concentrations (e.g., TC ≥ 200 vs. < 200 mg/dL; LDL-C ≥ 130 vs. < 130 mg/dL), (2) hydrogen dose (< 10 mg/day vs. ≥ 10 mg/day), and (3) intervention duration (< 12 vs. ≥ 12 weeks). No exploratory post hoc subgroups were added after reviewing results. Given the evaluation of four lipid outcomes and multiple subgroup strata, we recognize that the number of comparisons increases the potential for type I error. All results, especially for secondary outcomes and subgroups, should therefore be interpreted with appropriate caution. To minimize analytic flexibility, the statistical analysis plan was developed prior to data extraction. Primary outcomes (LDL-C and TC), secondary outcomes (TG and HDL-C), and all subgroup analyses (baseline lipid levels, hydrogen dose, and intervention duration) were specified a priori based on existing HRW literature. No additional outcomes, subgroup strata, or exploratory analyses were introduced after reviewing the study data. Publication bias was assessed visually with funnel plots and statistically using Egger’s regression test. Sensitivity analyses were conducted by sequentially excluding individual studies to examine the robustness of results [23]. In cases where publication bias was detected, the trim-and-fill method was applied to estimate the impact of potentially missing studies and adjust pooled effect sizes accordingly [24]. Because individual participant data were unavailable, no multivariable regression modeling, non-linearity assessment, or covariate adjustment could be performed.

Results

Study selection

The initial database search yielded 2821 records. As outlined in Figs. 1 and 372 duplicates were removed, leaving 2449 unique studies. After screening titles and abstracts, these were excluded for not meeting the eligibility criteria. A total of 27 articles were retrieved for full-text assessment, of which 12 articles comprising 13 trials met the inclusion criteria for the final meta-analysis. Among these trials, 13 RCTs reported outcomes for HDL-C, 11 for LDL-C, 13 for TC, and 12 for TG [25–36]. Although the initial database search was broad and captured animal and non-human studies, these were excluded during full-text screening according to our predefined eligibility criteria, which required RCTs conducted in human participants. Because this work is a systematic review and meta-analysis, participants were not directly recruited by the present investigators. Recruitment was conducted independently within each included RCT. According to trial reports, participants were typically recruited from outpatient clinics, hospital registries, or community-based advertisements targeting adults with overweight or obesity and related cardiometabolic conditions. Consequently, the synthesized evidence primarily reflects adults with overweight/obesity who are already engaged in specialty care or have established metabolic disorders, rather than the general population of overweight/obese adults. Eligibility screening, informed consent, and enrollment procedures were performed at the level of each original study. Consequently, the present analysis synthesizes previously recruited trial populations rather than prospectively enrolling new participants.

Fig. 1.

Flow chart for study examined and included into the meta-analysis

Characteristics of the included studies

A comprehensive overview of the study attributes is presented in Table 1. A total of 12 articles with 13 RCTs, published between 2015 and 2025, were included in this review. The studies were conducted across diverse regions, including China, Japan, India, Serbia, and Slovakia, reflecting a broad international research interest in the effects of HRW.

Table 1.

Characteristics of randomized controlled trials evaluating hydrogen-rich water (HRW) on lipid outcomes in adults with overweight or obesity

Author, year	Country	Age (years)	Population	Sample size (HRW/Control)	BMI (kg/m²)	Duration	Baseline lipid levels(TC, LDL-c, [I/C])	Outcome	Hydrogen-rich water dosage (mL/day)	Control group
Kuzmanovic, 2025	Serbia	57.6	Men and women over 50 years old	14/13	25.4	12 weeks	TC: 247.5/228.2 mg/dL LDL-C: 150.8/143.1 mg/dL	TC, TG, LDL-C, HDL-C	Two 500 mL servings per day, administered every 12 h (12 mg of dihydrogen per serving, taken twice per day)	Control water containing < 0.1 ppm of dihydrogen.
Moribe, 2024	Japan	57.74	Individuals with metabolic syndrome or pre-metabolic syndrome	91/90	27.44	12 weeks	TC: 217.44/202.53 mg/dL LDL-C: 127.96/121.59 mg/dL	TC, TG, LDL-C, HDL-C	1050 mL per day (divided into three 350 mL servings)	Filtered water (FW)
Wu(a), 2024	China	35.36	Patients with hyperuricemia	33/32	27.44	8 weeks	TC: 176.0/169.0 mg/dL LDL-C: 107.1/104.8 mg/dL	TC, TG, LDL-C, HDL-C	Three 330 mL bottles of HRW daily (total H_2 intake approx. 4.61 mg/day)	Placebo (ordinary drinking water, three 330 mL bottles daily)
Wu(b), 2024	China	34.23	Patients with hyperuricemia	35/32	28.16	8 weeks	TC: 176.3/169.0 mg/dL LDL-C: 114.1/104.8 mg/dL	TC, TG, LDL-C, HDL-C	Two 330 mL bottles of HRW and one 330 mL bottle of placebo water daily. Total H 3.06 mg/day	Placebo (ordinary drinking water, three 330 mL bottles daily)
Sumbalova,2023	Slovakia	52.6	Patients with non-alcoholic fatty liver disease (NAFLD)	17/13	35.5	12 weeks	TC: 179.8/179.4 mg/dL LDL-C: NR	TC, TG, HDL-C	Three tablets daily, each dissolved in 330 mL of water	Placebo tablets (producing CO₂)
Liang,2023	China	46.16	Patients with impaired fasting glucose (IFG)	32	25.26	12 weeks	TC: 197.6/203.4 mg/dL LDL-C: 121.0/117.6 mg/dL	TC, TG, LDL-C, HDL-C	1000 mL/day	Placebo pure water
Ogawa,2021	Japan	Median 66.0 years (IQR 64.5, 72.5)	Patients with type 2 diabetes mellitus (T2DM)	23/20	25.9	12 weeks	TC: 205.0/183.7 mg/dL LDL-C: NR	TC, TG, LDL-C, HDL-C	1,500-2,000 mL/day	Filtered water (FW)
Kura,2022	Slovakia	52.65	Subjects with non-alcoholic fatty liver disease (NAFLD)	17/13	35.52	12 weeks	TC: 180.32/179.34 mg/dL LDL-C: 116.33/114.34 mg/dL	TC, TG, HDL-C	Three tablets daily, each dissolved in 330 mL of water(tablets producing > 4 mg/L H₂)	Placebo tablets (producing CO₂)
Zanini,2021	Serbia	76	Older adults aged 70 years and over	16/18	28.6	24 weeks	TC: 232.0/208.8 mg/dL LDL-C: 143.1/119.9 mg/dL	TC, TG, LDL-C, HDL-C	0.5 L/day of HRW (15 ppm of hydrogen)	Control drink (0 ppm of hydrogen)
LeBaron,2020	India	43.4	Men and women with metabolic syndrome	30/30	28.9	8 weeks	TC: 187.7/184.3 mg/dL LDL-C: 109.0/105.5 mg/dL	TC, TG, LDL-C, HDL-C	> 5.5 millimoles of H₂ per day (from 3 tablets/day, each dissolved in 250 mL water)	Placebo
Gu, 2020	China	> 40	Men at risk of peripheral arterial disease (PAD)	30/29	> 27	10 weeks	TC: 202.6/208.4 mg/dL LDL-C: 46.8/47.9 mg/dL	TC, TG, LDL-C, HDL-C	735 mL/day (245 mL, 3 times/day)	Placebo (bottled pure water)
Korovljev,2017	Serbia	56.4	Middle-aged overweight women	10/10	29.33	12 weeks	TC: 228.2/216.6 mg/dL LDL-C: 143.1/127.6 mg/dL	TC, TG, LDL-C, HDL-C	Caplets supplying approximately 6 ppm of H_2 per day	Placebo (cellulose) caplets
Song, 2015	China	NR	Patients with untreated isolated hypercholesterolemia	34/34	27.3	10 weeks	TC: 234.0/234.7 mg/dL LDL-C: 166.7/163.6 mg/dL	TC, LDL-C, HDL-C	0.9 L/day, administered in three 0.3 L servings.	Placebo water

HRW, hydrogen-rich water; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TG, triglycerides; BMI, body mass index; IFG, impaired fasting glucose; NAFLD, non-alcoholic fatty liver disease; T2DM, type 2 diabetes mellitus; I/C, intervention/control; NR, not reported.Estimated daily hydrogen dose calculated from reported hydrogen concentration (ppm or mg/L) × daily water volume. When hydrogen concentration was not reported, dose category was based on tablet manufacturer specifications

The sample sizes varied considerably, ranging from small pilot studies with fewer than 30 participants to larger trials enrolling over 180 individuals. The mean age of participants spanned from young adults in their mid-30s to elderly populations above 70 years, while baseline BMI values generally fell within the overweight or obese range (25–35 kg/m²). The study populations represented a range of metabolic and chronic conditions, including type 2 diabetes mellitus, metabolic syndrome, IFG, hyperuricemia, NAFLD, untreated hypercholesterolemia, and age-related risk groups.

Intervention durations ranged from 8 to 24 weeks. Most studies administered hydrogen in the form of HRW, with daily dosages varying between 500 mL and 2,000 mL. Several studies employed hydrogen-generating tablets or caplets, which were dissolved in water prior to consumption, ensuring controlled hydrogen release. The estimated hydrogen doses across trials ranged from approximately 3 mg/day to more than 12 mg/day. Control groups typically received placebo water, standard drinking water, or tablets without hydrogen release.

Primary outcomes were consistent across studies and focused mainly on lipid metabolism, including changes in TC, TG, LDL-C, and HDL-C. A few trials additionally examined disease-specific markers, such as fasting glucose in patients with IFG or broader cardiometabolic risk indicators in individuals with metabolic syndrome. Overall, the included studies demonstrated methodological diversity in terms of sample characteristics, intervention forms, and hydrogen dosages, but shared a common aim of assessing HRW’s potential benefits on metabolic and lipid-related outcomes. Supplementary Table 2 provides an evaluation of the RoB using the Cochrane framework. Across the 13 included trials, 11 were judged to have low RoB, while 2 were rated as having some concerns, primarily related to incomplete reporting of the randomization process or measurement procedures. No study was assessed as having high RoB. According to GRADE criteria, the certainty of evidence was rated moderate for the effects on TC and LDL-C, and low for TG and HDL-C, reflecting imprecision and heterogeneity across studies. Supplementary Table 3 presents the characteristics of the HRW intervention and the estimated daily dosages. Dose categories were defined as < 10 mg/day and ≥ 10 mg/day based on the empirical distribution observed across included RCTs (Supplementary Table 3). Most trials (approximately 77%) delivered daily hydrogen doses between 1 and 7.5 mg/day, whereas only two trials provided substantially higher hydrogen exposure (≥ 10 mg/day). Therefore, 10 mg/day served as a natural boundary separating the ‘standard-dose’ HRW studies from a small subset of higher-dose interventions. This threshold also aligns with prior human HRW literature, where daily intakes approaching or exceeding ~ 10 mg/day are typically classified as high-dose protocols. Of the 13 included trials, only three involved overweight or obese adults without a specified metabolic diagnosis (Kuzmanović 2025; Korovljev 2017; Zanini 2021). The remaining ten trials enrolled participants with established cardiometabolic conditions, including metabolic syndrome (two trials), NAFLD (two trials), hyperuricemia (two trials), IFG (one trial), type 2 diabetes (one trial), untreated isolated hypercholesterolemia (one trial), and men at risk for peripheral arterial disease (one trial). A stratified overview is presented in Supplementary Tables 4 to illustrate the distribution of metabolic comorbidities across trials.

Meta-Analysis findings

Effect on TG levels

A total of 12 RCTs, including 687 subjects with overweight or obesity (346 interventions, 341 control), demonstrated that HRW non-significantly reduced serum TG levels (WMD: − 4.21 mg/dL; 95% CI: − 17.75 to 9.32; P = 0.542). Considerable heterogeneity was observed (I² = 60%, P_het < 0.001) (Fig. 2). Subgroup analyses did not reveal any significant impact based on dose, duration, and baseline value (Supplementary Fig. 1).

Fig. 2.

Forest plot of randomized controlled trials investigating the effects of hydrogen-rich water (HRW) on serum triglyceride (TG) levels (mg/dL). The weighted mean difference (WMD) is presented with 95% confidence intervals (CI). Negative values indicate a non-significant reduction in TG. The direction of effect is labeled as “Favours HRW” (left) and “Favours Control” (right) for consistency across figures. TG, triglycerides; WMD, weighted mean difference; CI, confidence interval

Effect on TC levels

A total of 13 RCTs, including 757 subjects with overweight or obesity (382 interventions, 375 control), showed that HRW significantly lowered TC (WMD: − 6.71 mg/dL; 95% CI: − 10.38 to − 3.04; P < 0.001), with low heterogeneity (I² = 10.3%, P_het = 0.342) (Fig. 3). In subgroup analyses, short-term interventions (< 12 weeks) led to significant reductions (WMD: − 6.63 mg/dL; 95% CI: − 11.9 to − 2.07; P = 0.004), whereas longer interventions (≥ 12 weeks) did not (WMD: − 7.49 mg/dL; 95% CI: − 16.81 to 1.83; P = 0.115. Subgroup analyses based on baseline lipid concentrations and HRW dose were conducted to explore potential effect modifiers; results are presented in Supplementary Fig. 1. These analyses were underpowered and hypothesis-generating, and should not guide clinical decision-making.

Fig. 3.

Forest plot of randomized controlled trials investigating the effects of hydrogen-rich water (HRW) on total cholesterol (TC) levels (mg/dL). The weighted mean difference (WMD) is presented with 95% confidence intervals (CI). Negative values indicate a reduction in TC. The direction of effect is labeled as “Favours HRW” (left) and “Favours Control” (right) for clarity. TC, total cholesterol; WMD, weighted mean difference; CI, confidence interval

Effect on HDL-C levels

Across 13 RCTs (435 subjects with overweight or obesity), HRW produced a significant decrease in HDL-C (WMD: − 1.16 mg/dL; 95% CI: − 1.92 to − 0.40; P = 0.003), with substantial heterogeneity (I² = 0.0%, P_het = 0.483) (Fig. 4). Subgroup analyses based on baseline lipid concentrations and HRW dose were conducted to explore potential effect modifiers; results are presented in Supplementary Fig. 1. These analyses were underpowered and hypothesis-generating, and should not guide clinical decision-making.

Fig. 4.

Forest plot of randomized controlled trials investigating the effects of hydrogen-rich water (HRW) on high-density lipoprotein cholesterol (HDL-C) levels (mg/dL). The weighted mean difference (WMD) is presented with 95% confidence intervals (CI). Negative values indicate a reduction in HDL-C. The direction of effect is labeled as “Favours HRW” (left) and “Favours Control” (right) for clarity. HDL-C, high-density lipoprotein cholesterol; WMD, weighted mean difference; CI, confidence interval

Effect on LDL-C levels

In 11 RCTs 682 subjects with overweight or obesity (340 interventions, 342 control), HRW significantly reduced LDL-C levels (WMD: − 3.21 mg/dL; 95% CI: − 6.31 to − 0.10; P = 0.043), with moderate heterogeneity (I² = 0.0%, P_het = 0.554) (Fig. 5). Subgroup analyses based on baseline lipid concentrations was conducted to explore potential effect modifiers; results are presented in Supplementary Fig. 1. These analyses were underpowered and hypothesis-generating, and should not guide clinical decision-making.

Fig. 5.

Forest plot of randomized controlled trials investigating the effects of hydrogen-rich water (HRW) on low-density lipoprotein cholesterol (LDL-C) levels (mg/dL). The weighted mean difference (WMD) is presented with 95% confidence intervals (CI). Negative values indicate a reduction in LDL-C. The direction of effect is labeled as “Favours HRW” (left) and “Favours Control” (right) for clarity. LDL-C, high-density lipoprotein cholesterol; WMD, weighted mean difference; CI, confidence interval

Sensitivity analyses and publication bias

Sensitivity analyses, conducted by sequentially omitting individual studies, confirmed that no single study unduly influenced the pooled results (Supplementary Fig. 2). Publication bias was assessed using funnel plots and Egger’s regression test. No significant bias was detected for any lipid outcomes (LDL-C, TC, HDL-C, or TG) (Supplementary Fig. 3).

Discussion

This meta-analysis indicates that HRW supplementation leads to modest, statistically detectable changes in lipid profiles that are unlikely to confer clinically meaningful cardiovascular risk reduction. Pooled analyses showed statistically significant reductions in TC and LDL-C. From a clinical perspective, the magnitude of LDL-C reduction observed (− 3.2 mg/dL) is far below guideline-defined minimally important differences, which generally correspond to reductions of ≥ 10–15 mg/dL from lifestyle intervention and ≥ 30 mg/dL with first-line statin therapy [37]. Epidemiologic data suggest that each 38–40 mg/dL reduction in LDL-C is associated with an approximately 20–25% reduction in atherosclerotic cardiovascular disease (ASCVD) risk [38]. By extrapolation, an LDL-C reduction of ~ 3 mg/dL would be expected to correspond to an absolute ASCVD risk reduction of approximately 1–2% or less over the long term, which is clinically negligible. Therefore, the LDL-C changes associated with HRW in overweight or obese adults would be expected to translate into negligible cardiovascular benefit.

A modest reduction in HDL-C (~ 1 mg/dL), mainly at higher HRW doses (≥ 10 mg/day), is paradoxical and should be interpreted cautiously. Possible explanations include chance variation, regression to the mean, analytic multiplicity, or shifts in HDL-C subclass distribution and functionality rather than overall HDL-C quantity. Although the clinical impact of this small reduction is likely minimal, future HRW trials should pre-specify HDL-C functionality and safety monitoring as outcomes to clarify whether higher doses of HRW exert biologically meaningful effects on HDL-C composition or cardiovascular risk.

Consistent with the pre-specified outcome hierarchy, the primary lipid outcomes (TC and LDL-C) represent the most robust findings of this analysis. Secondary outcomes and subgroup analyses, including HDL-C changes and dose-stratified effects, should be regarded as exploratory and should not drive clinical decision-making at this stage.

The strength of these conclusions is supported by both risk-of-bias assessment and certainty-of-evidence grading. Most included trials were judged to have low RoB, with only minor concerns in a small number of studies. According to the GRADE framework, the certainty of evidence was moderate for TC and LDL-C and low for TG and HDL-C. While moderate-certainty evidence suggests that HRW likely produces small reductions in TC and LDL-C, the absolute magnitude of these effects remains clinically modest. Accordingly, this evidence does not justify guideline-level recommendations for HRW in dyslipidemia management.

From a clinical perspective, the findings of this meta-analysis do not support HRW as a lipid-lowering therapy in the conventional sense. The magnitude of LDL-C and TC reduction observed is far below thresholds associated with meaningful cardiovascular risk reduction. If HRW has a role, it would most plausibly be as an investigational adjunct in early or pre-pharmacologic stages of cardiometabolic risk, or among individuals who are statin-intolerant or decline pharmacologic therapy, rather than as a substitute for guideline-recommended treatments.

Heterogeneity was observed across trials, particularly for TG outcomes (I² = 60%), likely due to differences in population metabolic status, baseline lipid concentrations, intervention dose, and duration. Exploratory subgroup analyses were conducted to explore potential effect modifiers, but these analyses were underpowered, not corrected for multiple comparisons, and are presented in Supplementary Fig. 1. Therefore, they should be interpreted cautiously and cannot guide clinical decision-making.

Several included trials also evaluated non-lipid cardiometabolic markers, including fasting glucose, insulin resistance, hepatic fat content, and inflammatory biomarkers. While HRW’s lipid effects were modest, some studies reported more consistent improvements in these secondary outcomes, such as reductions in fasting glucose, HOMA-IR, and pro-inflammatory cytokines (IL-6, TNF-α). These findings suggest that HRW may exert broader metabolic and anti-inflammatory effects beyond lipid modulation. However, because these outcomes were heterogeneously reported and were not primary endpoints in most trials, definitive conclusions remain limited.

Compared with prior systematic reviews of HRW and lipid outcomes [14, 18], this meta-analysis offer several novel contributions. First, it focuses exclusively on adults with BMI ≥ 25 kg/m², whereas prior reviews included broader clinical populations with established metabolic disorders. Second, it includes a larger number of RCTs, enabling more robust pooling of primary lipid outcomes (TC and LDL-C). Third, it incorporates pre-specified subgroup analyses and formal GRADE assessment, providing clearer context for interpreting the clinical relevance of HRW-associated lipid changes (Supplementary Table 5).

These trends align with key randomized trials. LeBaron et al. (2020) reported approximately 4–10% reductions in TC after 24 weeks of high-concentration HRW in individuals with metabolic syndrome [15], Song et al. (2013) observed significant decreases in LDL-C and improved HDL-C functionality after 10 weeks of HRW in patients with potential metabolic syndrome [17].

Korovljev et al. (2018) reported improvements in TG and mitochondrial function in overweight women following HRW supplementation [36], whereas their 2019 NAFLD trial demonstrated reductions in hepatic fat and liver enzymes despite minimal changes in serum lipids [39]. Collectively, these findings are consistent with our pooled results showing small decreases in TC and LDL-C and inconsistent or negligible effects on TG and HDL-C.

Preclinical studies suggest that HRW may exert antioxidant, anti-inflammatory, and mitochondrial-modulating effects, including reductions in reactive oxygen species and suppression of NF-κB–mediated cytokine production [40–44]. While these mechanisms provide a plausible basis for potential metabolic effects, the modest lipid changes observed in this meta-analysis are far smaller than those required for clinically meaningful cardiovascular benefit, and current clinical evidence does not support broad cardiometabolic effects. Therefore, HRW should remain investigational, and its mechanistic effects should not be interpreted as demonstrating substantive lipid-lowering or cardiovascular efficacy. Importantly, mechanistic plausibility does not equate to clinically significant effects, and any observed changes in lipid profiles should be interpreted within the context of modest effect sizes and limited trial evidence.

Clinical relevance

In overweight or obese adults, HRW does not meaningfully lower LDL-C or TC and should not be recommended for lipid control. The observed LDL-C reduction (~ 3 mg/dL) is far below guideline-defined thresholds [45, 46] considered minimally important for ASCVD risk modification (generally ≥ 10–15 mg/dL with lifestyle intervention and ≥ 30 mg/dL with statin therapy) and would be expected to translate into an absolute ASCVD risk reduction of approximately 1–2% or less. Accordingly, HRW should not delay or replace evidence-based lifestyle or pharmacologic interventions in individuals who meet established treatment criteria.

Limitations

Our analysis has several important limitations that should be considered when interpreting the findings. First, many of the included trials had relatively small sample sizes (often fewer than 50 participants per arm), which limits statistical power and increases the likelihood of imprecise or unstable estimates. In addition, most studies had short intervention durations (typically 8–12 weeks), restricting inference about the long-term effects of HRW on lipid metabolism and cardiovascular risk. Consequently, the observed associations may reflect short-term biochemical changes rather than sustained clinical effects.

Second, because this review focused on adults with BMI ≥ 25 kg/m², findings may not generalize to normal-weight populations or pediatric cohorts.

Third, information bias and measurement limitations should be acknowledged. Although all studies used standard clinical chemistry assays on fasting samples, there was variability in laboratory instruments, calibration procedures, assay kits, and timing of blood collection across trials. Such heterogeneity may introduce non-differential measurement error, potentially attenuating true associations toward the null, or, in some settings, amplifying apparent effects. In addition, HRW exposure itself was measured indirectly through reported volume and hydrogen concentration, and actual hydrogen intake may vary depending on preparation methods, storage conditions, and participant adherence, introducing further exposure misclassification.

Fourth, heterogeneity across studies, particularly for TG outcomes (I² ≈ 60%), reflects differences in hydrogen dose (ranging from approximately 500 mL to 2 L/day with concentrations of 5–15 ppm), intervention duration, delivery form (pre-packaged HRW vs. tablets), and participant clinical characteristics. While reductions in TC and LDL-C showed low heterogeneity, this methodological and clinical diversity complicates interpretation and may influence effect size magnitude in either direction.

Fifth, although most trials were rated as having low RoB, incomplete reporting of randomization procedures or attrition in a minority of studies may affect internal validity. Missing outcome data were handled using statistical imputation for unavailable standard deviations, which is methodologically accepted but introduces additional uncertainty. If missingness was related to treatment response, this could bias pooled estimates upward or downward.

Sixth, because this meta-analysis used aggregated study-level data rather than individual participant data, residual confounding cannot be fully excluded, despite randomization within trials. Differences in adherence, background diet, physical activity, concomitant medication use, and co-interventions across studies may influence observed associations. These unmeasured factors could partially account for the modest lipid changes attributed to HRW.

Seventh, although funnel plots and Egger’s regression tests did not indicate publication bias, the limited number of available studies reduces the power of these tests, and selective reporting cannot be completely ruled out. Additionally, although subgroup analyses were pre-specified, they remain exploratory and underpowered. Because multivariable meta-regression was not feasible, potential effect modification by dose, baseline lipid status, disease comorbidity, or intervention duration cannot be reliably disentangled, and observed subgroup patterns may reflect confounding by trial design or population characteristics rather than true biological interactions.

Finally, the protocol was not prospectively registered (e.g., PROSPERO), which may increase the risk of analytic flexibility. The evaluation of multiple lipid outcomes and subgroup strata also increases the possibility of type I error, meaning that some statistically significant findings may occur by chance. Collectively, these limitations indicate that the findings should be interpreted cautiously as modest associations rather than definitive evidence of clinically meaningful lipid modification.

Conclusion

HRW intake in overweight or obese adults, predominantly those with coexisting metabolic disorders, results in small but statistically significant reductions in TC and LDL-C. The magnitude of these changes is below clinically meaningful thresholds, and current evidence does not support routine use of HRW as a lipid-lowering therapy in clinical practice. Effects on TG and HDL-C were minimal, with only a modest reduction in HDL-C observed.

These findings do not generalize to all overweight/obese adults, particularly those without preexisting metabolic conditions. Any potential role for HRW is limited to an adjunctive or investigational therapy in populations with metabolic risk. Future research should evaluate HRW effects in otherwise healthy overweight or obese adults to determine preventive benefits, alongside larger, longer-term RCTs in adults with metabolic disorders to assess sustained effects on lipid profiles and cardiovascular outcomes.

Author contributions

The authors’ contributions were as follows: H.Y, J.F, LH., M.S, and P.J carried out the research, data analysis, and statistical assessment. P.J, H.F, H.Y, M.S, and K.P were involved in writing and structuring the manuscript. All authors reviewed and approved the final version of the manuscript.

Funding

This study was supported by grants from Zhejiang Provincial Natural Science Foundation of China (LTGY24H080009).

Data availability

All data generated or analyzed during this study are included in this published article, and you can find these data in references.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Human ethics and consent to participate 

Not applicable.

Competing interests

The authors declare no competing interests.

Clinical trial number

Not applicable.