The efficacy and safety of herbal medicines for glycaemic control and insulin resistance in individuals with type 2 diabetes: an umbrella review

Abstract

Aim

The global prevalence of diabetes is increasing rapidly. While glycaemic control remains the cornerstone of type 2 diabetes (T2DM) management, hypoglycaemic medications are associated with side effects. Many individuals with T2DM use herbal medicines despite potential risks. This review aims to comprehensively evaluate the efficacy and safety of herbal medicines for T2DM.

Methods

Four electronic databases were searched from August 2018-August 2023 for umbrella reviews, meta-analyses, and systematic reviews on herbal medicine for T2DM. The primary outcome was glycaemic control assessed by glycated haemoglobin (HbA1c); secondary outcomes included measures of insulin resistance, quality of life and adverse events. Quality assessment was performed using A Measurement Tool to Assess Systematic Reviews Version 2 (AMSTAR 2) and certainty of evidence for the primary outcome was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach.

Results

Eight meta-analyses and two systematic reviews were included. Ginger, turmeric, and Jinlida granules significantly decreased HbA1c compared to placebo (mean differences (MD): -0.467%, -0.486%, -0.283%, respectively) with moderate to high certainty evidence. Berberis vulgaris, Gegen Qinlian decoction and pycnogenol may potentially decrease HbA1c (very low certainty evidence). Ginger, turmeric, cardamom, Berberis vulgaris, Nigella sativa, and various Chinese herbal medicines demonstrated improvements in insulin resistance measures. However, overall confidence in these findings is limited due to the low to very low quality of the included reviews. Few mild adverse effects were reported, but data was incomplete.

Conclusions

Current evidence supports the use of ginger and turmeric for glycaemic control in type 2 diabetes, however, given the high clinical heterogeneity and low quality of the review, our confidence in this finding is somewhat limited. Herbal medicines should be used only as an adjunct to conventional treatment, with shared decision-making between clinicians and patients. Careful monitoring for hypoglycaemia is essential, particularly when herbal medicines are used concurrently with antidiabetic drugs, given a current lack of data on herb-drug interactions. Future research should clarify optimal regimens, safety profiles, and include diverse populations to enhance generalisability.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12906-025-05059-7.

Background

The global prevalence of diabetes has risen from 200 million in 1990 to 830 million in 2022 [1]. More than 95% of people with diabetes have Type 2 Diabetes Mellitus (T2DM) [1]. T2DM is characterised by insulin resistance in normally insulin-sensitive tissues and pancreatic β-cell dysfunction [2], with the resultant hyperglycaemia leading to serious long-term sequelae such as cardiovascular disease and blindness due to retinopathy [3]. Glycaemic control is a cornerstone of treatment and can be achieved through both pharmaceutical and non-pharmaceutical interventions, such as lifestyle changes. However, despite their efficacy, hypoglycaemic medications are associated with varying costs and side effect profiles [4] that may influence a patient’s decision to seek other interventions, including herbal medicine.

Medicinal plants have been used for the treatment of metabolic diseases since ancient times. Phytochemicals including polysaccharides, flavonoids, phenolic compounds and alkaloids have been identified as having antidiabetic effects [5, 6]. An estimated 53% of individuals with T2DM use herbal medicine worldwide [7], although this figure may be underestimated due to high rates of non-disclosure by consumers [8, 9]. Many users perceive herbal medicine as safe simply because it is natural and not synthetic [10]. However herbal medicines can pose risks in various ways, including toxicity from the known herbal ingredient, substitution with other herbal material, contamination with additional substances, risks in certain populations and potential herb-drug interactions [11, 12]. There is limited evidence on how herbal medicine may interact with Western pharmaceuticals [12]. One systematic review [13] highlighted that some herbs have either synergistic or antagonistic effects on antidiabetic drugs, although clinical trial evidence in humans remains limited. While no severe adverse effects have been reported with many herbs [14], one case report [15] documented hypoglycaemia with concurrent use of ginseng and oral hypoglycemics. Informed decision-making on use of herbal medicines in T2DM is therefore important, given the potential for adverse herb-drug interactions [16]. Concurrently, the World Health Organisation calls for a greater integration of traditional, complementary and integrative medicine into health service delivery, as this may improve health outcomes in regions where traditional medicines are widely used or where access to pharmaceuticals and health services is limited [17]. However, evidence regarding the efficacy and safety of herbal medicines for people with T2DM remains limited [18]. The high accessibility and acceptance of herbal medicines, along with their potential for glycaemic control, suggest a possible role in reducing the global burden of T2DM.

Clinicians require access to the best available and most recent evidence to support informed decision-making with their patients. We conducted an umbrella review to synthesise and evaluate the highest levels of evidence currently available [19, 20], with the aim of providing up-to-date clinical guidance on the efficacy and safety of herbal medicines in adults with T2DM.

Methods

An umbrella review, or overview of reviews, aims to present a comprehensive picture of the results of studies, typically of systematic reviews and meta-analyses, within a certain topic [20]. Our protocol was developed a priori adhering to PRIOR guidelines [21] and was prospectively registered on Open Science Framework (https://osf.io/nvedx, 29 August 2023). No changes were made to the protocol after registration. There is no clinical trial number to report.

Search strategy

A comprehensive search strategy (see Supplementary Table 1) was developed by one author (EL) based on a previous protocol [22]. A systematic search of MEDLINE, EMBASE, CINAHL and Cochrane databases was conducted on August 25th, 2023. Three authors (EL, CE, CP) independently screened articles in duplicate using Covidence based on title and abstract, then by full text review. All articles were reviewed by EL at both stages, with the second independent review of each article conducted by either CE or CP. Disagreements were resolved via discussion.

Selection criteria

Study design

All English-language umbrella reviews, meta-analyses (MA), and systematic reviews (SR) of data from placebo-controlled randomized controlled trials (RCTs) were considered. Studies were only included if they were published in the last five years from the search date, had clear inclusion criteria, reported a systematic search and data extraction procedure, and conducted a quality assessment. Studies were excluded if they were narrative or non-systematic reviews.

Our research question was framed using the following population, intervention, control, outcome (PICO) criteria.

Population

We included studies on adults with T2DM. Studies were excluded if their population included individuals with another medical condition, such as metabolic syndrome, dyslipidaemia, or another type of diabetes, unless subgroup analysis was provided for individuals with T2DM only.

Intervention

All herbal medicines were included as interventions. Herbal medicines were defined using the full Therapeutic Goods Administration (TGA) definition [23] as “all or part of a plant or substance (other than a pure chemical or a substance of bacterial origin)”.

Comparison

Only placebo-controlled studies were included. Studies with co-interventions were included only if the same co-interventions were added to both treatment and placebo groups.

Outcome

The primary outcome measure was glycaemic control measured using glycated haemoglobin (HbA1c). Secondary outcomes measured included fasting blood glucose (FBG), 2-h plasma glucose concentration (2hPG), measures of insulin secretion and resistance (serum insulin, Homeostatic Model Assessment of Beta Cell Function [HOMA-β], Homeostatic Model Assessment for Insulin Resistance [HOMA-IR], Homeostatic Model Assessment of Insulin Sensitivity [HOMA-S]), quality of life and adverse effects.

Hierarchical gathering of evidence

Due to the broad scope of our research question, it would not be feasible to gather data from every available study. We have therefore developed a systematic approach (see Supplementary Fig. 1) to gather the highest levels of evidence [24, 25], preferentially including umbrella reviews, followed by meta-analyses where these are not available, then systematic reviews where no meta-analyses exist.

In cases where a newer meta-analysis covered an individual aspect of an older umbrella review, the meta-analysis was used to update the overall findings of the umbrella review. Where several potentially overlapping meta-analyses were identified, the most recent and comprehensive review was retained. Where the amount of overlap was uncertain, primary study overlap was assessed by producing a citation matrix and calculating the corrected covered area [26].

Data extraction

One reviewer (EL) independently extracted relevant information from the included studies, including study design, population, intervention and comparator details, and outcome measures. To ensure consistency, a second reviewer (CE) extracted 20% of data in duplicate and verified the remaining data. We extracted number and type of all-included trials and relevant trials only. Relevant trials were defined as any randomised control trial (RCT) included in a meta-analysis or systematic review that matched the PICO criteria for our review. Outcomes were extracted as weighted or standard mean differences (WMD or SMD) with confidence intervals (CIs) or standard errors, or were summarised qualitatively by direct quotation or narrative description where a subgroup analysis of relevant trials was not conducted. Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) assessments for HbA1c were extracted when available from the original study authors. All reported values are HbA1c change using % as a unit of measurement, not % change in HbA1c.

Quality assessment

All included studies were critically appraised by two independent reviewers (JL, AM) using A Measurement Tool to Assess Systematic Reviews Version 2 (AMSTAR 2) which was used to generate an overall confidence rating in the review results from “High” to “Critically Low” [27]. Disagreements were resolved by discussion or referred to a third reviewer (CE) if agreement could not be reached.

Certainty of evidence

The GRADE framework [28] was used to assess the certainty of evidence for our primary outcome, HbA1c. Where GRADE assessments were not provided by the original study authors, they were conducted in duplicate by two independent reviewers (JL, AM), with any discrepancies resolved through discussion.

Results

Search results

After screening 437 titles and abstracts, 77 studies underwent full text review. Of these, ten studies involving 5,454 participants with T2DM were included, while 64 studies were excluded (Fig. 1). A complete list of excluded studies, along with reasonsfor exclusion, is provided in Supplementary Table 2.

Fig. 1.

Flow chart of search results

Study characteristics

See Table 1 for full study characteristics and extracted data. Two studies were systematic reviews [29, 30], and eight were meta-analyses [31–38]. One review pooled studies on Chinese herbal medicine (CHM) formulations including Tang Min Ling pill, Tangke soft capsules, Gegen Qinlian decoction, berberine, JianYuTangKang tablets, Tianqi capsule, and Jinlida granules [35]. Other meta-analyses examined specific interventions: Tianqi capsule [32], Jinlida granules [37], Gegen Qinlian decoction [38], Momordica charantia [31, 39], Berberis vulgaris [34], Nigella sativa [29], ginger, turmeric and cardamom [36], pycnogenol [33], blueberries and cranberries [30].

Table 1.

Characteristics and relevant findings of included studies

Author and Year (Review Type) Date of Last Search	N of participants in main review (N and type of trials; N and type of trials on adults with T2DM) Inclusion Criteria	Intervention details (duration of intervention) Control details	Primary outcome—HbA1c	Secondary outcomes – glycaemic control (FBG, 2hPG), insulin resistance (insulin, HOMA-IR, HOMA-S, HOMA-β), adverse effects	GRADE analysis for HbA1c
Chinese herbal medicine
Tian J. et al. 2019 (SR, MA) [35] December 2017	N = 1249 (26 RCTs; 8 RCTs) P = Individuals with prediabetes, diabetes, DN, DR or diabetic peripheral neuropathy I = Chinese herbal medicine C = Western medicine or placebo O = rate of developing diabetes (prediabetes), HbA1C (diabetes), 24 h urinary albumin (DN) NB: Study with T2DM + dyslipidaemia population [Zhang et al. 2008] included due to inclusion in HbA1c subgroup analysis	Tang Min Ling pill, Tangke soft capsules, Gegen Qinlian decoction, berberine, JianYuTangKang tablets, Tianqi capsule, Jinlida granules + metformin (in 3 studies) (12–26 wks) Placebo (5 RCTs), placebo + metformin (in 3 RCTs)	↓HbA1c: SMD = −0.49% (95% CI:−0.70;−0.28, I2 = 39%, 5 RCTs, n = 912; SMD = −0.39%, 95% CI:−0.61;−0.17, I2 = 11%, 3 RCTs with metformin as co-comparator, n = 377 NB: unclear whether outcome is measured using SMD or WMD; SMD stated in forest plot but WMD stated in methodology	FBG: statistically significant decrease (3 RCTs, n = 598, p < 0.05), NS (1 RCT, n = 178) ↓2hPG: statistically significant decrease (2 RCTs, n = 577, p < 0.05) ↓HOMA-β: statistically significant decrease (2 RCTs, n = 185, p < 0.05) Insulin: “Tianqi promoted insulin secretion” (1 RCT, n = 40, p NR) NB: no subgroup analysis of placebo-controlled T2DM RCTs provided, but individual numeric data on FBG and 2hPG available for each study. Qualitative analysis from results section quoted above. No numeric data provided for HOMA-β	N/A
Piao C. et al. 2020 (SR, MA) [32] August 10, 2019	N = 1029 (8 RCTs; 3 RCTs) P = Individuals with T2DM I = Tianqi Jiangtang (TJT) capsule C = Routine comprehensive treatment or routine comprehensive treatment + other treatment except TJT capsule O = HbA1c, FBG, 2hPG, TG, LDL-c, insulin	Cai et al.: Tianqi capsule 1.6 g, oral TID + lifestyle modification Lian et al.: Tianqi capsule 1.6 g, oral TID + metformin + lifestyle modification Li et al.: Tianqi capsule 1.6 g, oral TID + metformin 500 mg oral BD + lifestyle modification Cai et al.: Placebo, 5 capsules oral TID + lifestyle modification Lian et al.: Placebo 1.6 g oral TID + metformin + lifestyle modification Li et al.: Placebo 1.6 g oral TID + metformin + lifestyle modification NB: description/nature of lifestyle modification and duration of studies not provided	HbA1c: NS (3 RCTs, n = 249)	FBG: MD = −0.76, 95% CI-1.44,−0.08, 1 RCT, n = 90, p NR; MD = NS (3 RCTs, n = 159) 2hPG: MD = −1.46, 95% CI-2.3-,−0.62, 1 RCT, n = 90, p NR; MD = NS (3 RCTs, n = 159) Fasting insulin: NS (2 RCTs, n = 169) Adverse effects: “the reported AEs were not serious”, details NR NB: no subgroup analysis for placebo RCTs only	⨁⨁⨁⨁ High* (other considerations: all plausible residual confounding would reduce the demonstrated effect)
Zhao X. et al. 2022 (SR, MA) [37] June 1, 2022	N = 1555 (22 RCTs; 3 RCTs) P = Adults with T2DM without severe renal or hepatic insufficiency or severe diabetic complications I = Jinlida granules with metformin C = Metformin with or without placebo O = HbA1C, FPG, 2hPG	Jinlida granules (9 g TID) + metformin (0.5 g BD or 0.5 g TID) (12wks to 16wks) Placebo (9 g TID) + metformin (0.5 g BD or 0.5 g TID)	↓HbA1c: WMD = −0.283%, 95% CI:−0.561;−0.004, I2 = 0%, 3 RCTs, n = 411, p = 0.046)	FBG: NS (3 RCTs, n = 411) 2hPG: NS (3 RCTs, n = 411)	⨁⨁⨁◯ Moderate* (serious imprecision)
Zheng Y. et al. 2021 (SR, MA) [38] June 1, 2020	N = 650 (5 RCTs; 1 RCT) P = Adults with T2DM I = CHM C = Western medicine, placebo, or no intervention O = gut microbiota modulation, glucose metabolism	Gegen Qinlian decoction, 150 mL BD (12 wks) Placebo NB: concentration of Gegen Qinlian decoction used not provided	↓HbA1c: WMD = −0.88 ± 0.14% (placebo: −0.36 ± 0.15, p < 0.05, 1 RCT, n = 187) NB: between-group analysis for relevant RCT not available	↓FBG: −1.46 ± 0.23 mmol/L (placebo: −0.24 ± 0.24, p < 0.001, 1 RCT, n = 187) 2hPG: NS (1 RCT, n = 187) ↓HOMA-β: “significantly improved as per ANCOVA analysis” (1 RCT, n = 187) Adverse effects: NR NB: between-group analysis for relevant RCT not available	N/A
Momordica charantia
Peter E. et al. 2019 (SR, MA) [31] Search date not provided	N = 1045 (10 RCTs; 8 RCTs) P = Adults with T2DM or prediabetes I = Momordica charantia C = Unspecified O = Glucose control	M. charantia seed, fruit powder, fruit pulp, dried unripe fruit juice, extract, mixture, between 500 mg-6 g/day (4wks-4mths) Placebo	↓HbA1c: WMD = −0.26%, 95% CI:−0.49;−0.03, I2 = 0%, 5 RCTs, n = 243, p = 0.03 for monoherbal formulation, WMD = −0.66%, 95% CI:−1.26;−0.06, 1 RCT, n = 50, p = 0.03 for polyherbal formulation	↓FBG: WMD = −0.72 mmol/L, 95% CI:−1.33;−0.12, I2 = 14%, 5 RCTs, n = 231, p = 0.02 Adverse effects: heartburn, nausea and/or vomiting, diarrhoea, abdominal discomfort, enhanced appetite, constipation, flatulence (4 RCTs, n = 154), headache and dizziness (1 RCT, n = 24). Diarrhoea, rumbling, nausea, vomiting, more common in M. charantia group compared to placebo (1 study, n = 52). No serious adverse effects reported (8 RCTs, n = 447)	⨁◯◯◯ Very low (very serious risk of bias e, serious imprecision f)
Berberis vulgaris
Safari Z. et al. 2020 (SR, MA) [34] January 2020	N = 452 (7 RCTs; 2 RCTs) P = Adults I = Barberry C = Not stated O = Glycaemic control	Shidfar et al.: 3 g/day Berberis vulgaris extract (12 wks) Mansouri et al.: 200 mL/day Berberis vulgaris juice + 0.5% citric acid + 5% sugar (8 wks) Placebo	HbA1c: WMD = −2.32%, 95% CI:−3.18;−1.46, 1 RCT, n = 60, p NR; NS (1 RCT, n = 42) NB: No subgroup analysis of placebo controlled T2DM RCTs provided	↓FBG: WMD = −24.20 mg/dL, 95% CI:−28.39;−20.01, 1 RCT, n = 42, p NR ↓Serum insulin: WMD = −1.38mIU/mL, 95% CI:−2.04;−0.72, 1 RCT, n = 42, p NR ↓HOMA-IR: WMD = −1.10, 95% CI:−1.21;−0.99, 1 RCT, n = 42, p NR NB: fasting status for insulin unclear	⨁◯◯◯ Very low (serious risk of bias g, very serious inconsistency h, serious indirectness i, very serious imprecision j)
Nigella sativa
Mahomoodally M. et al. 2022 (SR) [29] January 2022	N = 881 (14 RCTs, 3 NRCTs; 5 RCTs) P = Adults with T1DM, T2DM, non-diabetic conditions but known diabetic parameters, eg. Glucose level, healthy subjects I = Nigella sativa (NS) ± oral hypoglycemic drug C = Placebo ± oral hypoglycemic drug O = HbA1c, FBG, 2hPG, insulin levels, insulin resistance HOMA-IR, HOMA-S, HOMA-β, adverse effects	Nigella sativa oil or powder in capsule form, dosage either 5 mL or 1 g-3 g daily + oral hypoglycemic drug (in 3 studies) (8–52 wks) Placebo (included equivalent amounts of other oil or activated charcoal capsules (260 mg)) + oral hypoglycemic drug (in 3 studies) NB: concentration of oil for 5 mL/day study not provided, type of oral hypoglycemic drug not noted	HbA1c: After 3 months, MD = −0.71 to −0.3% for treatment groups vs. baseline (treatment group at beginning of study), MD = −0.09% to + 0.3% for placebo vs. baseline (placebo group at beginning of study) (3 RCTs, n = 245) After 6 months, MD = −0.84% for treatment vs. baseline, MD = + 0.14% for placebo vs. baseline (1 RCT, n = 103) After 9 months, MD = −0.66% for treatment vs. baseline, MD = + 0.27% for placebo vs. baseline (1 RCT, n = 103) After 12 months, MD = −0.4% for treatment vs. baseline, MD = + 0.28% for placebo vs. baseline (1 RCT, n = 103) NB: only within-group analysis provided by article; between-group analysis not available	FBG: MD = −65.4 to −3.0 mg/dL for treatment groups versus baseline, MD = −8.5 to + 23.8 mg/dL for placebo versus baseline (5 RCTs, n = 335) 2hPG: MD = −15.1 mg/dL for treatment vs. baseline, MD = + 2.5 mg/dL for placebo vs. baseline (1 RCT, n = 70) Insulin: MD = −1.2 to −0.26 mg/dL for treatment vs. baseline, MD = −1.44 to + 3.4 mg/dL for placebo vs. baseline (2 RCTs, n = 122) HOMA-IR: MD = −0.8 to −0.1 for treatment vs. baseline, MD = −0.51 to + 1.4 for placebo vs. baseline (2 RCTs, n = 122) HOMA-S: MD = −2.36 for treatment group vs. baseline, MD = + 9.24 for placebo vs. baseline (1 RCT, n = 40) HOMA-β: MD = −1.95 for treatment group vs. baseline, MD = + 1.03 for placebo vs. baseline (1 RCT, n = 40) NB: only within-group analysis provided by article; between-group analysis not available. Fasting status for insulin unclear	N/A
Ginger, turmeric, cardamom
Zhang. X et al. 2022 (SR, MA) [36] October 18, 2021	N = 2154 (34 RCTs; 34 RCTs) P = Adults with T2DM I = Medical plants from Zingiberaceae family C = Placebo O = body composition, glycaemic control, insulin sensitivity, lipid profiles-related indicators, BP, inflammatory biomarkers	Ginger, turmeric, or cardamom extracts. Dosage ranged from 80-4000 mg/day (4–25 wks) Placebo	↓HbA1c (%): WMD = − 0.432%, 95% CI: − 0.607;−0.257, I2 = 61.7%, 22 studies, n = 1653, p < 0.001, WMD = − 0.467%, 95% CI: − 0.760; − 0.174, I2 = 64.7%, p = 0.004 for ginger, WMD = − 0.486%, 95% CI: − 0.766; − 0.206, I2 = 56.4%, p = 0.011 for turmeric), WMD NS for cardamom	↓FBG: WMD = − 14.292 mmol/L (95% CI:−18.588;−9.995, I2 = 33.3%, 26 studies, n = 1815, p < 0.001), WMD = − 13.283 mmol/L (CI: − 18.529; − 8.036, I2 = 35.3% for turmeric), WMD = − 20.117 mmol/L (CI: − 25.975; − 14.258, I2 = 1.4% for ginger), NS for cardamom ↓Serum insulin: WMD = − 2.036 mU/L, 95% CI: − 2.857; − 1.216, I2 = 53.9%, 14 studies, n = 1077, p = 0.005. WMD = − 2.561 mU/L 95% CI: − 3.406; − 1.716, I2 = 0.0% for ginger, WMD − 1.777 mU/L, 95% CI: − 3.544; − 0.009, I2 = 70.5% for turmeric ↓HOMA-IR: WMD = − 0.886 mg/dL, 95% CI: − 1.375; − 0.398,, I2 = 53.9%, 15 studies, n = 1184, p ≤ 0.001. WMD = − 0.966 mg/dL, 95% CI: − 1.863; − 0.069, I2 = 79.3% for turmeric, WMD = −1.000 mg/dL, 95% CI:−1.775;−0.225, I2 NR for cardamom) Adverse effects: few minor effects including gastrointestinal symptoms, headache, sleepiness, bruising, fever, dermatitis, glossitis. No serious adverse effects (34 RCTs, n = 2154) NB: fasting status for insulin unclear	⨁⨁⨁◯ Moderate* (serious inconsistency)
Pycnogenol (pine bark extract)
Robertson NU. et al. 2020 (MA) [33] Sept 30, 2019	N = 1641 (27 RCTs; 1 RCT) P = Adults or children with any chronic disorder I = Supplements containing pine bark extract alone, or in combination with other supplements as long as comparison received same treatment apart from pine bark extract supplement C = Placebo, no intervention, other supplements (excluding any with antioxidant properties) O = Any disorder-related clinical outcome, all-cause mortality, adverse events, biomarkers of oxidative stress Concomitant use of 1 st and 2nd generation sulphonylureas, metformin, thiazolidinediones permitted	Zibadi et al. (2008): Pycnogenol 25 mg oral tablets 5x/day (12 wks) Placebo	↓HbA1c: MD:−0.90%, 95% CI:−1.78;−0.02, 1 RCT, n = 24, p NR)	↓FBG: MD:−1.00 mmol/L, 95% CI:−1.09;−0.91, 1 RCT, n = 24, p NR Adverse effects: NS (1 RCT, n = 24)	⨁◯◯◯ Very low* (serious risk of bias, indirectness, imprecision)
Blueberry, cranberry
Rocha D. et al. 2019 (SR) [30] June 11, 2017	N = 272 (7 RCTs; 5 RCTs) P = Adults with T2DM I = Any oral dose and form of cranberry or blueberry. Studies on both berries or combinations with other berries excluded C = Placebo, no treatment O = blood glucose, insulin, HOMA-IR, HbA1c, other glucose metabolic markers Concomitant use of insulin and/or oral hypoglycemic agents included	Blueberry hydroalcoholic extract or powder (containing Vaccinium myrtillus L., Vaccinium arctostaphylos L.) (3 RCTs), cranberry juice or extract (2 RCTs) (1 day-12wks) Placebo NB: Hoggard et al. was single-day study	HbA1c: ↓significant decrease using extract, 1 RCT, n = 74, p NR; NS using powder (1 RCT, n = 176 for blueberry); NS (1 RCT, n = 30 for cranberry)	FBG: statistically significant decrease using extract or powder (2 RCTs, n = 176 for blueberry, p NR); statistically significant decrease using juice (1 RCT, n = 58, p NR); NS using extract (1 RCT, n = 30 for cranberry) ↓2hPG: statistically significant decrease using extract or powder (2RCTs, n = 176 for blueberry, p NR) Fasting insulin: NS (2 RCTs, n = 132 for blueberry and cranberry) HOMA-IR: NS (2 RCTs, n = 132 for blueberry and cranberry) ↓insulin iAUC: −18% compared to placebo (1 single-day study, n = 8 for blueberry) ↓glucose iAUC: “improved postprandial glucose response” (1 single-day study, n = 8 for blueberry)	N/A

2hPG 2-h postprandial glucose, ANCOVA analysis of covariance, BD twice a day, BP blood pressure, CHM Chinese herbal medicine, CI 95% confidence interval, DN diabetic nephropathy, DPN diabetic peripheral neuropathy, DR diabetic retinopathy, FBG fasting blood glucose, HbA1c glycated hemoglobin, HOMA-IR Homeostatic Model Assessment for Insulin Resistance, HOMA-S Homeostatic Model Assessment for Insulin Sensitivity, HOMA-β Homeostatic Model Assessment of Beta Cell Function, iAUC incremental area under curve, LDL-c low-density lipoprotein cholesterol, MA meta-analysis, MD mean difference, mths months, N/A we were unable to conduct a GRADE due to a lack of relevant forest plots or sub-analysis for HbA1c outcome, NR not reported, NRCT non-randomised control trial, NS not statistically significant, QE quasi-experimental study, RCT(s) randomised control trial(s), SR systematic review, TDS/TID three times a day, TG triglycerides, wks weeks, WMD weighted mean difference

 ± standard error

^*GRADE was conducted by original study authors

↑significant increase in outcome measure when intervention was compared to control

↓significant decrease in outcome measure when intervention was compared to control

All studies included analysis of glycated HbA1c and FBG. Six studies [29, 30, 32, 35, 37, 38] assessed 2hPG, and seven studies [29, 30, 32, 34–36, 38] reported measures of insulin resistance (e.g. fasting insulin, HOMA-β, HOMA-IR or HOMA-S). Adverse events were reported in five studies [31–33, 36, 38], while none assessed quality of life.

Quality assessment using AMSTAR 2

Four studies [31–33, 36] which investigated Momordica charantia, Tianqi Jiangtang, pine bark extract and the combination of ginger, turmeric, and cardamom, were assessed as having low methodological quality due to the presence of one critical flaw. The remaining reviews were assessed as having critically low methodological quality, having more than one critical flaw (see Table 2). The most commonly identified critical flaws included a lack of justification for excluding individual studies (item 7), a partially inadequate literature search strategy (item 4), absence of prospective protocol registration (item 2), and insufficient assessment of publication bias (item 15).

Table 2.

Summary of AMSTAR 2 rating and critical flaws of included reviews

Herbal medicine	Reference	Critical flaws	AMSTAR 2 rating
Gegen Qinlian decoction, Tang Min Ling pill, Tangke soft capsules, berberine	Tian J. et al. 2019 [29]	2, 4 (partial yes), 7	Critically Low
Tianqi capsule	Piao C. et al. 2020 [26]	7	Low
Jinlida granules	Zhao X. et al. 2022 [31]	7 (partial yes), 15	Critically Low
Gegen Qinlian decoction	Zheng Y. et al. 2021 [32]	4 (partial yes), 7, 15	Critically Low
Momordica charantia (bitter melon)	Peter E. et al. 2019 [25]	15	Low
Berberis vulgaris (barberry)	Safari Z. et al. 2020 [28]	2, 4 (partial yes), 7	Critically Low
Nigella sativa	Mahomoodally M. et al. 2022 [29]	2, 4 (partial yes), 7, 9 (9a yes, 9b no), 13	Critically Low
Ginger, turmeric, cardamom	Zhang. X et al. 2022 [30]	7	Low
Pycnogenol (pine bark extract)	Robertson NU. et al. 2020 [27]	15	Low
Blueberry, cranberry	Rocha D. et al. 2019 [30]	2 (partial yes), 4 (partial yes), 7, 13	Critically Low

Critical Flaws [27]: item 2: protocol registered before commencement of the review, item 4: adequacy of the literature search, item 7: justification for excluding individual studies, item 9: risk of bias from individual studies being included in the review, item 11: appropriateness of meta-analytical methods, item 13: consideration of risk of bias when interpreting the results of the review, item 15: assessment of presence and likely impact of publication bias

Efficacy of herbal medicine in glycaemic control

Overall, there was moderate to high certainty evidence that ginger, turmeric and Jinlida granules decreased HbA1c levels compared to placebo, but not cardamom or Tianqi capsule. Very low certainty evidence suggests that Berberis vulgaris and pycnogenol may lower HbA1c. Other herbal medicines such as Tang Min Ling pill, Tangke soft capsules, berberine, JianYuTangKang tablets, blueberry, and Nigella sativa may potentially decrease HbA1c, however, GRADE assessments could not be conducted for the associated studies. Momordica charantia and cranberry are unlikely to have a significant effect on HbA1c levels. Mild adverse effects were reported for Momordica charantia, ginger, turmeric, and cardamom, while no severe adverse effects were reported in any study. Further, adverse effects were not reported for several interventions, including various CHM, Berberis vulgaris, Nigella sativa, blueberry, and cranberry. The compositions of all included herbal formulations are detailed in Supplementary Table 3. A summary of the findings for each herb is provided below.

Chinese herbal medicine

A 2019 meta-analysis [35] pooled data from studies on various forms of CHM and included eight studies investigating seven different CHM formulae. CHM was found to decrease HbA1c compared to placebo alone (SMD = −0.49%) and when used as an adjunct to metformin (SMD = −0.39%). These effects may approach clinical significance. A decrease in 2hPG, insulin secretion and HOMA-β was also reported, although effect sizes could not be determined due to the absence of a meta-analysis of placebo-controlled trials in T2DM. Findings on FBG were inconsistent.

High certainty evidence indicated that Tianqi capsules had no effect on HbA1c, either as an adjunct to lifestyle modification alone or in combination with metformin [32]. Evidence for effects on FBG and 2hPG was inconsistent, and two studies reported no effects on fasting insulin. No significant adverse effects were reported (n = 1029).

Moderate certainty evidence supported the efficacy of Jinlida granules in reducing HbA1c as an adjunct to metformin (WMD = −0.283%, CI −0.561 to −0.004) [37], although this decrease may not be clinically significant. No significant changes in FBG or 2hPG were found. For Gegen Qinlian decoction, within-group comparisons suggested a decrease in HbA1c (WMD = −0.88 ± 0.14%); however no between-group comparisons were reported. No serious adverse effects were reported.

Momordica charantia (bitter melon)

Low certainty evidence [31] suggests that Momordica charantia may lead to a modest decrease in HbA1c (WMD: −0.26%, CI −0.49 to −0.03) and FBG (WMD: −0.72 mmol/L, CI −1.33 to −0.12). However, substantial heterogeneity was observed across RCTs regarding the form of the herb, dosage, and duration of treatment.

Minor gastrointestinal adverse effects were reported [31], with one RCT finding an increase in symptoms such as diarrhoea, abdominal rumbling, nausea, vomiting in the Momordica charantia group compared to placebo. However, a causal link between Momordica charantia and these side effects could not be established. No serious adverse effects were reported.

Berberis vulgaris (barberry)

Very low certainty evidence suggests that daily consumption of Berberis vulgaris juice for eight weeks may result in a clinically significant decrease in HbA1c (WMD: −2.32%, CI −3.18 to −1.46) [34], whereas taking Berberis vulgaris extract daily for twelve weeks may not have the same effect [40]. However, one RCT [40] found that Berberis vulgaris extract significantly reduced FBG, HOMA-IR, and serum insulin levels compared to placebo.

Nigella sativa

A 2022 systematic review [29] on Nigella sativa oil or powder capsules reported only within-group mean differences, with no between-group analyses provided. Two studies reported decreases in HbA1c within the intervention groups, while HbA1c increased in the placebo groups. Another study reported decreases in HbA1c in both the intervention and placebo groups. Additionally, single studies reported improvements in 2hPG, HOMA-S, HOMA-β within the intervention groups only. However, improvements in FBG, insulin and HOMA-IR were observed in in both groups across several studies.

Ginger, turmeric, cardamom

A meta-analysis of 22 RCTs investigating the effects of ginger, turmeric or cardamom extracts found moderate certainty evidence of a reduction in HbA1c when all interventions were pooled (WMD = −0.432%, CI −0.607 to −0.257), for ginger alone (WMD = −0.467%, CI −0.760 to −0.174) and turmeric alone (WMD = −0.486%, CI −0.766 to −0.206), approaching clinical significance [36]. In contrast, there was moderate certainty evidence that cardamom had no effect in reducing HbA1c. Ginger and turmeric were also associated with improvements in FBG, while cardamom was not. Additionally, ginger and turmeric improved serum insulin levels, and reductions in HOMA-IR were reported for both turmeric and cardamom. High heterogeneity was observed across studies regarding dosages and treatment durations.

A small number of mild adverse effects were reported, including gastrointestinal symptoms, headache, sleepiness, bruising, fever, dermatitis, and glossitis; however, these were not stratified by herb type [36]. No serious adverse effects were reported.

Pycnogenol (pine bark extract)

Very low certainty evidence [33] suggests that pycnogenol may decrease HbA1c (mean difference (MD) = −0.90%, CI −1.78 to −0.02) compared to placebo. A reduction in FBG was also reported (MD = −1.00 mmol/L), though this was not considered clinically significant. No serious adverse effects were reported [33].

Blueberry, cranberry

A systematic review [30] on blueberry and cranberry supplements reported a decrease in HbA1c following eight weeks of supplementation with blueberry extract capsules containing 9.1 mg of anthocyanins (1 RCT, n = 74) [41], but no effect after twelve weeks of blueberry powder capsules containing 9.8 mg of anthocyanins (1 RCT, n = 102) [42]. Two RCTs (n = 176) [41, 42] reported significant reductions in FBG and 2hPG compared to placebo, while one study [42] found no significant changes in fasting insulin or HOMA-IR. A single-day study [43] (n = 8) found that ingestion of a glucose load along with blueberry extract capsules containing 169 mg of anthocyanins led to a greater reduction in the insulin and glucose incremental area under the curve (iAUC) compared to placebo. Single studies [44, 45] found no difference between cranberry extract and placebo for HbA1c and HOMA-IR, and reported inconsistent findings for FBG.

Sensitivity analysis

We conducted a post hoc sensitivity analysis according to methodological quality of the reviews, as assessed using the AMSTAR-2 tool. Reviews examining Momordica charantia, Tianqi capsule, pycnogeneol, ginger, turmeric, and cardamom were rated as low methodological quality due to having one critical flaw, while all other reviews were rated as very low quality due to having more than one critical flaw. Taking this into consideration, the very low certainty evidence suggesting that Berberis vulgaris may lower HbA1c, combined with very low methodological quality of the review, substantially limits confidence in this finding. Similarly, since GRADE assessments could not be conducted for reviews on Tang Min Ling pill, Tangke soft capsules, JianYuTangKang tablets, blueberry, and Nigella sativa, all of which were rated as very low quality, we have low confidence in the reported findings that these interventions may reduce HbA1c.

Discussion

To the best of our knowledge, this is the first umbrella review to provide a comprehensive appraisal and synthesis of top-tier evidence on the use of herbal medicine for glycaemic control in T2DM. Overall, we found that ginger and turmeric probably result in clinically significant decreases in HbA1c and FBG, Jinlida granules probably result in a modest decrease in HbA1c, and cardamom and Tianqi capsule probably have no effect on HbA1c compared to placebo. Berberis vulgaris, Gegen Qinlian decoction and pycnogenol may also reduce HbA1c, although the evidence is very uncertain, while Momordica charantia and cranberry may result in little to no difference in HbA1c. Evidence for the effect of other herbal medicines on HbA1c could not be assessed for certainty. Ginger, turmeric, cardamom, Berberis vulgaris, Nigella sativa, and various Chinese herbal medicines also demonstrated improvement in insulin resistance measures. Notably, none of the studies that reported on adverse events documented any serious side effects. Only Momordica charantia, ginger, turmeric and cardamom were associated with any adverse effects, all of which were mild. However, overall confidence in these findings is limited due to the low to very low methodological quality of the included reviews. In particular, we have very low confidence in findings on Berberis vulgaris, Tang Min Ling pill, Tangke soft capsules, JianYuTangKang tablets, blueberry, and Nigella sativa.

Ginger and turmeric probably decrease HbA1c and FBG compared to placebo [36]. These effect sizes (MD −0.467% and −0.486% respectively) may be clinically significant as they approach the efficacy of most oral hypoglycaemic drugs, which on average reduce HbA1c by a mean difference of 0.5–1.0% [46] compared to placebo. However, given the high clinical heterogeneity and low methodological quality of the review, our confidence in this finding is somewhat limited. Both herbs belong to the Zingiberaceae family, which contain polyphenols [36] that inhibit pancreatic β-cell apoptosis and key enzymes responsible for carbohydrate metabolism, thereby improving regulation of glycogenesis and glycolysis [47]. Adverse effects were few and minor, making these herbs likely safe for use in the general population. However, clinicians should be aware of the potential for synergistic effects glucose-lowering effects and monitor carefully for possible hypoglycaemia when used concurrently with antidiabetic medications. In the same meta-analysis, cardamom was found to decrease HOMA-IR, but not HbA1c, FBG, or serum insulin [36]. Although cardamom also belongs to the Zingiberaceae family [36], turmeric and ginger extracts have a significantly higher total phenolic content per gram and greater antioxidant activity compared to cardamom [48], which may explain why cardamom did not demonstrate a significant improvement in HbA1c.

Jinlida granules are a Chinese herbal medicine formula containing seventeen herbs, including Panax ginseng and Polygonati rhizoma, that probably decrease HbA1c but not FBG or 2hPG [37]. However, the clinical significance of this reduction could not be confirmed due to a wide confidence interval (−0.561 to −0.004%). A recent network analysis [49] found significant overlap between the molecular targets of Jinlida granules and T2DM, and identified seven key targets implicated in cell growth and apoptosis, metabolism, and inflammation [49]. This may account for the mechanism of action of Jinlida granules in T2DM, which is a chronic low-grade inflammatory state [50–54]. The evidence is of moderate certainty due to serious imprecision, suggesting the need for more studies on Jinlida granules with larger sample sizes. The safety of Jinlida granules remains unclear, as adverse effects were not reported.

Metformin, the first-line pharmacological treatment for T2DM, is widely recognized for its ability to reduce HbA1c levels by approximately 1.0–1.5%, along with a favourable safety profile, minimal risk of hypoglycaemia, and low cost [55]. In contrast, the current review highlights that ginger and turmeric show promising reductions in HbA1c that are approaching the threshold for clinical significance (≥ 0.5%). While these reductions are generally more modest than those achieved with metformin, they may hold value as adjunctive therapies. A cross-sectional study of 739 individuals in Thailand found that those using Momordica charantia in combination with metformin (500 mg/day) were nearly three times more likely to achieve good glycemic control than those relying on metformin alone (adjusted odds ratio (OR) = 2.921, 95% CI = 1.227–6.952, p = 0.015) [56]. Further, herbal medicines may offer potential alternatives for individuals who experience gastrointestinal side effects, have contraindications to metformin (e.g., renal impairment), or prefer natural or culturally traditional treatments. However, the evidence base for herbal interventions remains limited in both scope and methodological rigour. Additional high-quality RCTs are needed to determine whether specific herbs can reliably match or complement the glycaemic effects of metformin.

There was substantial heterogeneity in terms of a wide range of dosages and formulations for several interventions including Berberis vulgaris, Nigella sativa, ginger/turmeric/cardamom, and Momordica charantia. This variability limits our ability to provide clinical recommendations and contributes to lower certainty in the evidence for these interventions. The positive findings for ginger, turmeric, Berberis vulgaris, and Nigella sativa need to be interpreted with this in mind. There was generally greater consistency in dosages for Chinese herbal medicine, largely due to the use of patent formulations and standardised doses, which offers clearer guidance for both clinicians and patients [57]. We strongly recommend that herbal medicine dosages and formulations be standardised in both clinical practice and research. Standardisation would reduce clinical heterogeneity, increase the certainty of evidence, and provide healthcare consumers with the necessary guidance on appropriate dosages and formulations to ensure both efficacy and safety [57].

The safety of herbal medicines is an important consideration when considering their use. Only half of the included studies in our review collected data on adverse events. These data suggest that Tianqi capsule, Gegen Qinlian decoction, Momordica charantia, ginger, turmeric, cardamom and pycnogenol were not associated with serious adverse events. However it is important to note that the reviews on Tianqi capsule and ginger/turmeric/cardamom were of low quality, while those on Gegen Qinlian and Momordica charantia were of very low quality, limiting confidence in these findings. The safety of other Chinese herbal medicines, blueberry, Berberis vulgaris and Nigella sativa cannot be determined from the available evidence. Preclinical and limited clinical evidence suggests that curcumin is generally safe, although may result in gastrointestinal symptoms such as nausea and vomiting, dermatitis, reduced sperm motility, and clinically significant herb-drug interactions, including with anticoagulants and chemotherapeutic agents [58]. Regulatory warnings have also been issued regarding the rare but potential risk of liver injury associated with curcumin use [59]. Reports have linked Momordica charantia to adverse effects including hypoglycemic coma, convulsions, and reduced fertility with [60]. Chinese herbal medicines may carry additional risks such as contamination with heavy metals, deliberate adulteration with pharmaceutical drugs, or intrinisic toxicity (e.g. nephrotoxicity) [61]. Further, the varying regulation of herbal medicines both within and between countries reduces the assurance that consumers can access herbal medicines safely [61]. Finally, the potential for hypoglycemic effects from herbal medicines should be closely monitored.

Many of the herbal medicine interventions in our review are commonly used in cultures such as the Chinese culture (Chinese herbal medicine) and in people from the Indian subcontinent (ginger, turmeric, cardamom, Momordica charantia). These may be viewed as highly acceptable forms of healthcare interventions being aligned with cultural values and beliefs. Indeed, TCM (Traditional Chinese Medicine) and AYUSH (Ayurveda, yoga, naturopathy, Unani, Siddha, Sowa Rigpa and homeopathy) are traditional medicine systems that were the original healthcare systems in China and India respectively, are well-established (including providing healthcare services in large hospitals) and are usually integrated with Western medical systems in what is known as “medical pluralism” [62, 63]. In 2023, 1.28 billion people in China accessed TCM services through healthcare institutions such as TCM hospitals [64]. These traditional medicine systems may be particularly important in areas of reduced access to other forms of health care. For example, in India 70% of the rural population depends on Ayurvedic medicine as their main form of health care [65]. Similarly, individuals of lower socio-economic status, particularly those who are unemployed, live in rural areas, or report poor health status, are more likely to use traditional medicine [66]. Strengthening the evidence base for traditional herbal medicines, given their cultural acceptability, affordability, and widespread use, is essential to support safe, effective, and accessible healthcare for disadvantaged populations. This approach aligns with the World Health Organization’s goal of improving health outcomes in regions where conventional medical care is limited or difficult to access.

Our umbrella review adds to and updates findings from a 2021 overview of meta-analyses [14] on the effectiveness of medicinal plants for glycaemic control in T2DM. The main findings from this earlier review suggested that aloe vera leaf gel, psyllium fibre and fenugreek seeds led to substantial decreases in HbA1c ranging from −0.85% to −0.99%, while four other herbal medicines including Nigella sativa and Gegen Qinlian decoction also achieved clinically significant reductions in HbA1c. A key difference between the two reviews is that our analysis includes only placebo-controlled RCTs. Our review is also significantly more recent as we included only reviews published from 2018 to 2023, whereas the earlier overview included reviews published prior to 2020. Consequently, there is little overlap between our reviews, with only one shared review on Momordica charantia [31]. Our review also differs by including systematic reviews as well as meta-analyses, in contrast to the earlier review which focused only on meta-analyses.

An important strength of our umbrella review is the use of a hierarchical evidence synthesis approach, which ensures that only the most recent, comprehensive and high-level evidence is presented without the risk of overlapping primary studies. Where possible, we also conducted GRADE assessments to evaluate the certainty of evidence, adding greater rigour and confidence to our conclusions. An inherent limitation of umbrella reviews is the reliance on the quality of the trials included in the systematic reviews. Issues such as poor reporting and over-interpretation of conclusions are likely to have impacted our findings. Moreover, due to the significant workload involved, we did not check all RCTs from the included systematic reviews for concerns related to compromised research integrity, such as retracted publications. This study is also limited by the small number of recent systematic reviews and meta-analyses, which limited the range of herbal medicines included and limited our ability to compare findings with earlier studies [14]. Where evidence indicated that a particular herb had a significant effect, high heterogeneity within intervention characteristics such as dosage and duration of treatment, limited our ability to provide specific guidance, and to confidently recommend herbal medicine use in individuals with T2DM. However, our findings can nonetheless inform shared decision-making by providing the best available evidence on the benefits and harms of herbal medicines for T2DM. Using an umbrella review approach to informing clinical practice is an efficient and pragmatic way of synthesising evidence to support clinical practice.

There is a clear need for additional rigorous evidence on the benefits and harms of herbal medicines for glycaemic control in individuals with T2DM to support safe, effective and accessible healthcare for all populations. Future RCTs should prioritise the use of standardised herbal interventions and clearly define the dosage, form (e.g. capsule, decoction, tincture) and duration of administration for each individual herb as recommended by the CONSORT guidelines for herbal interventions trials [67]. This level of detail is essential for ensuring reproducibility, improving comparability across studies, minimising clinical heterogeneity, and strengthening the validity of reviews in the field of herbal medicine. High-quality RCTs are needed to determine whether specific herbs can reliably match or complement the glycaemic effects of antidiabetic medications such as metformin. A number of interventions including Berberis vulgaris, Gegen Qinlian decoction and pycnogenol are promising in terms of efficacy for glycaemic control but require additional robust evidence in order to reduce uncertainty. We suggest that optimal study designs include longer duration of treatments and longer-term follow up e.g. 12 months, as well as long term follow up of differences between groups for complications such as cardiovascular disease and renal impairment. We note also that no studies measured patient-reported outcomes such as quality of life; these outcomes are essential to collect in future trials in order to provide clinicians and patients with evidence on meaningful outcomes. Further elucidation of the mechanisms of action of herbal medicines is important in order to move the field forward. Understanding the role of these herbal agents within the broader therapeutic landscape will help guide clinicians and patients in making informed, individualised treatment decisions for T2DM management. Where possible, studies should be conducted in a range of different populations to improve their external validity, allowing generalisability to a wider range of individuals worldwide. This may include patients with comorbidities or existing complications, in order to provide more personalised insights into efficacy. This would also allow for the identification of specific populations that may benefit more from herbal medicine, such as individuals from lower socio-economic backgrounds or those with limited access to conventional pharmaceuticals, who may be more likely to use herbal medicine [66]. Further studies elucidating the active compounds and mechanisms of action of herbal medicines would additionally provide a stronger pharmacological basis for the use of herbs in a medical setting. Rigorous RCT methods need to be adhered to, in particular, adequate randomisation sequence generation and allocation concealment as well as blinding of participants and personnel. The routine collection of adverse events during clinical trials is essential to provide robust and reliable evidence on the safety of herbal medicines, with particular focus on identifying potential herb-drug interactions. This is especially important in populations concurrently using conventional antidiabetic medications, as interactions may alter drug efficacy or increase the risk of adverse effects. In addition, long-term safety data is required and could be obtained through adverse event registries and cohort studies. To ensure patient safety and support informed clinical decision-making, future studies must rigorously document and report all adverse events. Lastly, systematic reviews and meta-analyses synthesising data from RCTs should follow the PRISMA reporting guidelines and include a list of reasons for exclusion of studies at full-text screening stage, prospective protocol registration, and assessment of publication bias.

Conclusions

Currently, sufficient evidence only exists to support the use of ginger and turmeric, which probably improve glycaemic control without any severe side effects, although confidence in these findings is limited due to low quality of the included reviews. Clinicians and patients should engage in shared decision-making about the benefits and harms of herbal medicines, accounting for personal preferences, cultural values, and the certainty of available evidence. We recommend that herbal medicines should only be used as an adjunct to conventional treatment for type 2 diabetes, as specific therapeutic dosages and durations are not clearly established. Monitoring for hypoglycaemia is essential, particularly when herbal medicines are used alongside antidiabetic medications, due to their potential glucose-lowering effects. Future research should clarify the benefits and harms of herbal medicines for glycaemic control in T2DM through high-quality, placebo-controlled trials using standardised formulations, robust adverse event reporting—including herb-drug interactions—and exploration of subgroups most likely to benefit, to support safe, effective, and accessible care for all populations.

Abbreviations

2hPG

2-Hour plasma glucose concentration

AMSTAR 2

A Measurement Tool to Assess Systematic Reviews Version 2

CHM

Chinese herbal medicine

CI

Confidence intervals

FBG

Fasting blood glucose

GRADE

Grading of Recommendations, Assessment, Development, and Evaluation

HbA1c

Glycated haemoglobin

HOMA- β

Homeostatic Model Assessment of Beta Cell Function

HOMA-IR

Homeostatic Model Assessment for Insulin Resistance

HOMA-S

Homeostatic Model Assessment of Insulin Sensitivity

MA

Meta-analysis

RCT

Randomised controlled trial

SMD

Standard mean difference

SR

Systematic review

T2DM

Type 2 diabetes mellitus

TGA

Therapeutic Goods Administration

WMD

Weighted mean difference

Authors’ contributions

C.E.: conceptualisation, methodology, analysis, manuscript review and editing. E.L.: conceptualisation, analysis, manuscript original draft, review and editing. A.M., J.L., C.P.: analysis, review and editing. A.C., A.F.: review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. This work was supported by a philanthropic donation by an anonymous donor to Western Sydney University. The donor had no input into the design of the study, interpretation of the findings or preparation of the manuscript for publication. Open access publication fees were provided by Western Sydney University.

Data availability

The authors confirm that the data supporting the findings of this study are available within the manuscript and its supplementary materials.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

CE declares that she is the Jacka Foundation Principal Research Fellow, Chair of the RACGP Integrative Medicine Specific Interest Network (voluntary role), Program Lead of an academic integrative healthcare centre (no financial interest), past GP Advisory Board member for Blackmores Research Institute, has received industry funding from nutraceutical and acupuncture device companies to conduct clinical trials, and has received honoraria and had travel expenses covered for presenting at complementary medicine events. As a medical research institute, NICM Health Research Institute receives research grants and donations from foundations, universities, government agencies, and industry. Sponsors and donors provide untied and tied funding for work to advance the vision and mission of the Institute. AM declares that she is a naturopathic practitioner at a clinic in Sydney, Australia. She is the recipient of a scholarship from the Jacka Foundation of Natural Therapies for her PhD. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.