Dosage exploration of the effects of honey and its derivatives on cardiometabolic outcomes: an overview of systematic reviews and GRADE-assessed updated meta-analysis

Abstract

Cardiovascular diseases and diabetes are associated with significant mortality, morbidity, and economic burden, highlighting the need for alternative preventive strategies. Honey bee products, including honey, royal jelly, and propolis, are considered potential interventions for managing cardiometabolic risk factors. This umbrella review aimed to compare the effectiveness of these products in cardiometabolic health. PubMed, Scopus, and Web of Science databases were systematically searched from inception through 21^ST October 2024 to identify eligible meta-analyses and primary randomized controlled trials (RCTs). Random-effect pairwise analysis combined trial findings, while dose-response and influence analyses assessed result robustness. Evidence quality and certainty were evaluated using AMSTAR-II, ROB, GRADE, and ICEMAN criteria. Analysis of 69 RCTs with 3544 participants revealed that 10 g of honey daily may lower Hemoglobin A1C but adversely affect systolic blood pressure, Aspartate transferase, triglycerides, fasting blood glucose, and high-sensitive C-reactive protein. Royal jelly improved blood pressure, lipid profiles, glycemic indices, and total antioxidant capacity. Propolis demonstrated reductions in anthropometric measures and improvements in lipid profile, glycemic control, liver enzymes, and inflammation and oxidative stress markers. While long-term or high-dose honey consumption in individuals with health concerns warrants caution, RJ and propolis demonstrated dose-dependent benefits for cardiometabolic health with proper certainty. Future research should focus on population-specific characteristics and optimized dosages.

Introduction

Non-communicable diseases (NCDs) are responsible for an estimated 43 million deaths annually, disproportionately affecting low- and middle-income countries where nearly 73% of global NCDs-related mortality was recorded [1]. In high-income countries like the United States, an NCDs-related death occurs every 33 seconds, highlighting their pervasive impact [2]. The World Health Organization (WHO) Sustainable Development Goal 3.4 aims to reduce premature NCDs mortality by 30% by 2030 to address this burden [1]. Complementary and alternative medicine may contribute to this target; however, robust, well-designed research is essential to establish effective interventions [3].

A variety of pharmaceutical and nutraceutical agents are used to manage NCDs, yet their inconsistent efficacy and potential side effects have led to growing interest in complementary strategies [4]. Dietary components with therapeutic potential are of interest, as they are commonly consumed and may have notable effects on cardiometabolic outcomes [3]. Such interventions can help refine dietary and clinical guidelines, especially for individuals at high risk of NCDs [4].

Among natural alternatives, honey bee (Apis mellifera) products, namely honey, royal jelly (RJ), and propolis have a long history of medicinal use dating back to 2000 BC [5–7]. Honey is rich in fructose, trace elements, and bioactive compounds [8, 9], showing promise in preventing and managing NCDs [10–12]. RJ, a nutrient-dense bee secretion, offers antioxidant, anticancer, neuroprotective, and cardioprotective benefits [13–19]. Propolis, a resin-like material collected from plant sources, contains over 300 phytochemicals, including polyphenols, amino acids, vitamins, and aromatic compounds, and is recognized for its potent anti-inflammatory and immune-modulating effects [20–24].

Recent systematic reviews and meta-analyses (SRMAs), and trials underscor honey bee products nuanced and sometimes contrary effects on cardiovascular health. A 2023 meta-analysis reported that honey improved blood glucose and lipid parameters but also increased pro-inflammatory markers [25], whereas a 2021 SRMA found no significant effect on lipid profiles [26]. Hadi et al. observed no overall impact of RJ on serum lipids. However, longer-duration studies (>90 days) showed lipid improvements [27, 28]. While RJ had no significant overall effects on liver enzymes or glycemic indices, subgroup analyses noted reductions in blood glucose in non-healthy individuals treated for ≥8 weeks [29]. Vajdi et al. found no general effect on anthropometric indices, but RJ at doses <3000 mg/day significantly improved anthropometric indices [30]. Regarding propolis, a 2024 SRMA showed significant improvements in serum cholesterol, though not in triglyceride (TG) [31]. Nazari Bonab et al. reported increases in antioxidant markers, but significant effects were only seen at higher doses (≥1000 mg/day) and shorter durations (<11 weeks) [32]. Another SRMA found propolis to significantly reduce glycemic markers, inflammatory indicators, and liver enzymes while showing no notable effect on insulin resistance, oxidative stress markers, lipid profile, or anthropometric outcomes [33].

These findings highlight the variability in outcomes based on dose, duration, and population characteristics, underscoring the need for comprehensive evaluations. To address this, we conducted an umbrella review of existing SRMAs to compare the effects of honey, RJ, and propolis on cardiometabolic factors. To strengthen the validity of our results, we performed dose-response, influence, and subgroup analyses. Additionally, we assessed the certainty and credibility of the evidence using established tools, including GRADE, ICEMAN, and AMSTAR.

Methods

The study protocol has been registered in PROSPERO with the registration number CRD42024601377. We conducted the review following the Cochrane Handbook for Systematic Reviews of Interventions [34]. We also adhered to the Preferred Reporting Items for Overviews of Reviews (PRIOR) statement (Supplementary Table 1) [35]. In addition, we followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Supplementary Table 2) [36].

Systematic search

MN and SBZ independently searched MEDLINE (PubMed), Scopus, and Web of Science until October 2024 to identify relevant SRMAs of randomized controlled trials (RCTs). We initially systematically searched databases from inception to July 5, 2024. Following the completion of screening and data extraction, and prior to analysis, we performed an updated search to capture any newly published studies up to 21^st October 2024. The search strategy combined keywords related to the intervention, outcomes, and study design (Supplementary Table 3). Gray literature was explored using Google Scholar to supplement the electronic database search, and reference lists of relevant reviews, meta-analyses, and citing articles were hand-searched. Only studies with a full published report were eligible for inclusion; abstracts without full-text availability were excluded.

Outcomes

The primary outcomes examined were cardiometabolic biomarkers, including systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol (TC), TG, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), TC/HDL-C ratio, LDL-C/HDL-C ratio, very low-density lipoprotein-cholesterol (VLDL-C) fasting blood glucose (FBG), two-hour postprandial glucose (2hPG), serum insulin (INS), hemoglobin A1C (HbA1C), homeostatic model assessment for insulin resistance (HOMA-IR), quantitative insulin sensitivity check index (QUICKI), aspartate aminotransferase (AST), alanine aminotransferase (ALT), AST/ALT ratio, alkaline phosphatase (ALP), and gamma-glutamyl transferase (GGT). Furthermore, we examined the impact of honey, RJ, and propolis on anthropometric measurements (body weight (BW), body mass index (BMI), fat mass (FM), total body fat percent (TBFP), fat-free mass (FFM), waist circumference (WC), hip circumference (HC), waist to hip ratio (WHR), waist to height ratio (WHtR)), and inflammatory and oxidative stress biomarkers (C-Reactive Protein (CRP), high sensitive CRP (hs-CRP), interleukin-1β (IL-1β), interleukin-6 (IL-6), interleukin-10 (IL-10), tumor necrosis factor-alpha (TNF-α), glutathione (GSH), glutathione peroxidase (GSHPx), monocyte chemotactic protein 1 (MCP-1), malondialdehyde (MDA), pro-oxidant–antioxidant balance (PAB), glutathione peroxidase (TAC) and superoxide dismutase (SOD)).

Selection of SRMAs and primary trials

We applied specific eligibility criteria to identify SRMAs (as peer-reviewed studies with a systematic search in databases) suitable for our analysis. These criteria included: 1) SRMAs of RCTs conducted in adult populations aged 18 years and older; 2) Studies that evaluated the effects of honey, RJ, and propolis as an intervention compared to a control group; and 3) Studies that reported mean differences (MDs) with corresponding 95% confidence intervals (CIs) for the relevant cardiometabolic factors. We also established the following exclusion criteria to filter out ineligible studies: 1) Narrative reviews or systematic reviews without a meta-analysis, 2) SRMAs focused on pregnant or lactating women, and 3) Primary RCTs with a short duration (less than one week), multicomponent supplementation, or a quasi-experimental or non-randomized design. When multiple SRMAs were available for a specific outcome in a given population, we selected the one with the highest number of primary RCTs or the most comprehensive methodology (e.g., dose-response SRMAs) [37]. We chose one meta-analysis for each outcome and carefully reviewed the reference lists of all previously published SRMAs on the same outcome. This allowed us to identify and include eligible RCTs not part of the selected SRMA and avoid primary study overlap. To further strengthen the robustness of our findings, we incorporated additional primary RCTs that had not been included in previous SRMAs [38].

Data extraction

Data extraction from eligible SRMAs was carried out independently by two authors, SBZ and FHT. From both eligible SRMAs and their included primary trials, we extracted the first author’s name and publication year. From the SRMAs, we also recorded the number of RCTs in the original meta-analysis, the number of additional RCTs identified in similar SRMAs, and the reported mean differences. To resolve discrepancies across SRMAs, we also extracted the following details directly from each primary trial: study location, percentage of female participants, outcomes and their effect sizes, follow-up duration, number of participants, type of intervention and placebo, supplementation dosage, study design, mean BMI, and participants’ health status.

Assessment of methodological quality

The quality of the included studies was independently assessed by two authors, MHR and SBZ. Eligible SRMAs were evaluated using the AMSTAR-2 tool, a critical appraisal tool designed to assess the quality of SRMAs of randomized or non-randomized healthcare interventions [39]. Additionally, the quality of primary RCTs included in each meta-analysis was assessed using the Cochrane Collaboration’s risk of bias tool for randomized trials [40]. Any disagreements between the authors were resolved through consensus.

Data synthesis and analysis

To report the findings of this umbrella review, we used the weighted mean difference (WMD) and 95% confidence intervals (CIs) of change. We directly calculated the mean ± standard deviation (SD) of change from the primary RCTs to reduce bias. The methodology for calculating the mean ± SD of change for both the intervention and control groups is detailed in our previous publication [41]. We applied a random-effects model to incorporate study-specific results as shown in the forest plots and account for both within- and between-study heterogeneity [42].

We used the I² statistic to evaluate heterogeneity quantitatively and conducted a chi-square test for homogeneity, considering a P-heterogeneity threshold above 0.10. As recommended by the Cochrane Handbook, the interpretation of I² values is categorized as follows: values between 0% and 40% might imply negligible heterogeneity, 30–60% could suggest moderate heterogeneity, 50–90% may indicate substantial heterogeneity and 75–100% often points to considerable heterogeneity [43]. To assess the potential presence of publication bias, we visually inspected funnel plots and considered P-values derived from Egger’s test [44] and Begg’s test [45]. Additionally, we performed an influence analysis to determine the impact of each trial on the overall effect size.

We evaluated the credibility of subgroup factors using the ICEMAN Instrument, which provides guidelines for assessing the Credibility of Effect Modification Analysis [46]. This framework offers a systematic method for examining the reliability of subgroup effects, emphasizing key aspects like methodological rigor, hypothesis pre-specification, and the consistency of results [46]. The ICEMAN tool allowed for a thorough review of each subgroup analysis, ensuring that any identified effect modifications were valid and not the result of biased reporting or multiple testing. Using this approach, the credibility of the subgroups was determined to range from very low to high [46]. Consequently, we conducted subgroup analyses of the pooled results for BMI, TG, TC, LDL-C, FBG, HbA1C, HOMA-IR, ALT, AST, and TNF-α, based on the following criteria: health status (diabetic or non-diabetic), obesity status (normal weight, overweight, or obese), intervention duration (≥12 weeks or <12 weeks), and the risk of bias in the included trials (low, some concern, or high).

Finally, we conducted a single-stage weighted mixed-effects meta-analysis to examine the effects of varying honey, RJ, and propolis doses on selected cardiometabolic biomarkers [47]. To explore non-linear dose-response relationships, we used restricted cubic splines with three knots positioned at the centiles suggested by Harrell (10%, 50%, and 90%). The suitability of the non-linear model was determined by evaluating the significance of the Wald test [47, 48]. All statistical analyses were performed using STATA statistical software, version 17.0, developed by StataCorp, with a significance level set at P < 0.05.

Grading of the evidence

The certainty level of evidence (CoE) was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) methodology [49]. RCTs start as high-certainty evidence but may be downgraded based on specific factors. These factors include study limitations (evaluated through the Cochrane Risk of Bias tool) [50], inconsistency [51], indirectness [52], imprecision [53], and significant evidence of publication bias. Conversely, the CoE can be upgraded if there is a dose-response relationship or if large effect sizes meet the minimal clinically important difference (MCID) (Supplementary Table 4). The CoE was categorized into four levels: high, moderate, low, or very low. A very low CoE means substantial uncertainty about the actual effect, which may differ considerably from the estimated effect. Low CoE indicates that the actual impact might significantly differ from the estimate, and additional research is expected to influence our confidence in the forecast and potentially alter the effect size. Moderate CoE suggests that the estimated effect is likely close to the actual effect, but further studies could impact our confidence and possibly change the estimate. High CoE reflects strong confidence that the estimated effect is accurate, with additional research unlikely to change our confidence in the estimate.

Results

Study selection

Figure 1 presents the results of a systematic search across three primary databases. Initially, we screened 772 SRMAs in this field. After removing duplicates (n = 250) and performing an initial screening (n = 522), 55 studies underwent a full-text review for inclusion, from which seven SRMAs [25, 28–33], encompassing 47 independent trials [8, 19, 23, 54–97], were incorporated into this umbrella review. An additional systematic search was conducted to minimize bias and update previous SRMAs. Of 2870 identified records, 2260 were screened after removing duplicates, including 22 trials [98–119]. Overall, this study includes 69 RCTs involving 3544 participants to examine the effects of honey (n = 16), RJ (n = 13), or propolis (n = 40) on 41 cardiometabolic outcomes.

Fig. 1.

The PRISMA flow diagram for the selection of studies and participants.

As shown in Supplementary Table 5, the meta-analysis by Ahmed et al. [25] was selected as the primary reference to identify studies examining the effects of honey on cardiometabolic factors. From this meta-analysis, 12 RCTs were included, along with four [98–101] additional studies from our systematic search. For the effects of RJ on cardiometabolic factors (Supplementary Table 6), the meta-analyses by Bahari et al. [29], Bahari et al. [28], and Vajdi et al. [30] were used, contributing to the extraction of 12 trials, with an additional trial [102] identified through systematic search. For the effects of propolis on cardiometabolic factors, the meta-analyses by Ahmed et al. [31], Hallajzadeh et al. [33], and Nazari Bonab et al. [32] were referenced. Out of the 40 included RCTs, 17 [103–119] were identified through a systematic search (Supplementary Table 7).

SRMAs and RCTs characteristics

Table 1 presents the basic characteristics of the seven included SRMAs. Among these, three focused on RJ [28–30], three on propolis [31–33], and one on honey [25]. Two of the seven SRMAs had a dose-response design [28, 30], and two included non-randomized trials [25, 32], additionally, except for the study by Ahmed et al. [31], which specifically examined individuals with metabolic syndrome and related disorders; other studies focused on general adult populations. The number of trials in each SRMAs ranged from 8 to 18, with participant populations ranging from 512 to 1105 individuals. Notably, one study did not specify its search references or databases [33], and the most recent search date for the study by Ahmed et al. [31] was May 2024. In terms of focus, two SRMAs considered cardiometabolic factors [25, 33], two addressed lipid profiles [28, 31], one analyzed glycemic profiles and liver enzymes [29], one investigated anthropometric factors [30], and one explored oxidative stress indices [32].

Table 1.

Basic characteristics of included SRMAs.

Author, year	Design	Databases searched	Eligibility criteria	Search End	Sample characteristics	Interventions addressed	Primary outcomes	N studies	AMSTAR-II
Hallajzadeh, 2021 [33]	SRMA	–	RCTs	–	812 Adults	Propolis	BW, BMI, FBG, INS, HbA1C, HOMA-IR, TC, LDL-C, HDL-C, TG, CRP, IL-6, TNF-aTNF-α, MDA, TAC, ALT, AST	14	Very Low
Ahmed, [25]	SRMA	PubMed, Embase, and the Cochrane Library	RCTs and nRCTs	January 2021	1105 Adults	Honey	BW, BMI, WC, FBG, INS, HbA1C, HOMA-IR, TC, LDL-C, HDL-C, TG, Apo-B SBP, DBP, TNF-aTNF-α, IL-6, hs-CRP, ALT, UA	18	Low
Bahari, [29]	SRMA	PubMed, Scopus, and WOS	RCTs	June 2023	517 Adults	Royal Jelly	ALT, AST, ALP, GGT, FBG, HbA1C, INS	10	Low
Bahari, [28]	Dose-response SRMA	PubMed, Scopus, and WOS	RCTs	April 2023	361 Adults	Royal Jelly	TC, TG, LDL-C, HDL-C	8	Low
Nazari bonab, 2023 [32]	SRMA	PubMed, Scopus, WOS, Google Scholar, and Cochrane Library	RCTs and nRCTs	October 2022	517 Adults	Propolis	GSH, GSHPx, TAC, SOD, MDA	9	Very Low
Vajdi, [30]	Dose-response SRMA	Embase, PubMed, WOS, and Scopus	RCTs	March 2023	512 Adults	Royal Jelly	BW, BMI, FM, WC	10	High
Ahmed, [31]	SRMA	PubMed, Scopus, WOS, and Cochrane Library	RCTs	May 2024	683 Adults with MetS and related disorders	Propolis	TC, TG, LDL-C, HDL-C	10	Low

ALP Alkaline Phosphatase, ALT Alanine Transaminase, AST Aspartate Transaminase, BMI Body Mass Index, BW Body Weight, CRP C-Reactive Protein, DBP Diastolic Blood Pressure, FBG Fasting Blood Glucose, FM Fat Mass, GGT Gamma-Glutamyl Transferase, GSH Glutathione, GSHPx Glutathione Peroxidase, HbA1C Hemoglobin A1C, HDL-C High-Density Lipoprotein Cholesterol, HOMA-IR Homeostatic Model Assessment for Insulin Resistance, hs-CRP High-Sensitive C-Reactive Protein, IL-6 Interleukin-6, INS Serum Insulin, LDL-C Low-Density Lipoprotein Cholesterol, RCTs Randomized Controlled Trials, nRCTs non-Randomized Controlled Trials, MDA Malondialdehyde, MetS Metabolic Sysndrome, SBP Systolic Blood Pressure, SRMA Systematic Review and Meta-Analysis, SOD Super Oxid Dismutase, TAC Total Antioxidant Capacity, TC Total Cholesterol, TG Triglyceride, TNF-α Tumor Necrosis Factor-alpha, WC Waist Circumference, WOS Web Of Science.

Supplementary Tables 8–10 indicate the primary characteristics of the included trials. Among the 69 included studies, research was conducted across various regions: America (n = 10), Europe (n = 6), Asia (n = 50), and Africa (n = 3). Of these, 51 studies included participants of both sexes (50.2% female), 10 focused solely on females, and eight exclusively on males. The study populations comprised healthy individuals (n = 14), cancer patients (n = 5), individuals with chronic kidney disease (n = 3), non-alcoholic fatty liver disease (n = 4), diabetes (n = 23), and other health conditions. The average duration of intervention on cardiometabolic factors was 7.6 weeks for honey, 8.3 weeks for RJ, and 13.5 weeks for propolis. Honey intervention doses ranged from 5 to 175 g, with a mean of 57.1 g. RJ doses varied between 150 and 1500 mg per day (mean: 820 mg), while propolis doses ranged from 660 to 7200 mg, with an average of 2800 mg.

Meta-analysis

Effect of honey, RJ, and propolis on blood pressure

As shown in Fig. 2 and Supplementary Table 11, propolis had no significant effect on blood pressure. However, RJ significantly reduced DBP (WMD: −2.48 mmHg, 95% CI: −3.99 to −0.97; I²: 0.00%, CoE: Moderate), while honey significantly increased both SBP (WMD: 4.24 mmHg, 95% CI: 2.03–6.45; I²: 0.00%, CoE: Moderate) and DBP (WMD: 1.94 mmHg, 95% CI: 0.40 to 3.48; I²: 0.00%, CoE: Moderate), as shown in Supplementary Figs. 1–3. (Corresponding forrest plots indicated as Supplementary Figs. 1–3).

Fig. 2. Pairwise analysis for the effects of honey, royal jelly, and propolis on cardiometabolic biom arkers.

Pairwise analysis for the effects of honey, royal jelly, and propolis on a anthropometric indices, b blood pressure, c glycemic profile, d liver enzymes, and e lipid profile.

Effect of honey, RJ, and propolis on anthropometric measures

According to Fig. 2 and Supplementary Table 11, honey significantly reduced both BW (WMD: −0.19 kg, 95% CI: −0.24 to −0.15; I²: 0.00%, CoE: Low) and BMI (WMD: −0.41 kg/m², 95% CI: −0.53 to −0.28; I²: 0.00%, CoE: Very Low). While RJ did not significantly affect anthropometric indicators, propolis significantly reduced BW (WMD: −0.59 kg, 95% CI: −1.14 to −0.05; I²: 60.6%, CoE: Moderate), BMI (WMD: −0.11 kg/m², 95% CI: −0.13 to −0.09; I²: 0.00%, CoE: Very Low), TBFP (WMD: −1.00%, 95% CI: −1.38 to −0.62; I²: 0.00%, CoE: Moderate), and FM (WMD: −0.42 kg, 95% CI: −0.74 to −0.10; I²: 0.00%, CoE: Low). Corresponding forest plots are shown in Supplementary Figs. 4–6.

Effect of honey, RJ, and propolis on liver enzymes

As shown in Fig. 2 and Supplementary Table 12, supplementation with honey and RJ did not significantly affect serum liver enzyme levels. In contrast, propolis supplementation was associated with a reduction in AST (WMD: −2.21 IU/L, 95% CI: −3.54 to −0.88; I²: 41.1%, CoE: High) and ALT levels (WMD: −3.31 IU/L, 95% CI: −5.97 to −0.65; I²: 63.5%, CoE: High). Corresponding forest plots can be found in Supplementary Figs. 7–9.

Effect of honey, RJ, and propolis on lipid profile

As shown in Fig. 2 and Supplementary Table 12, honey supplementation was associated with increased TG levels (WMD: 8.09 mg/dL, 95% CI: 1.22 to 14.9; I²: 7.78%, CoE: Low). Conversely, RJ supplementation reduced TC (WMD: −11.0 mg/dL, 95% CI: −18.0 to −3.96; I²: 13.4%, CoE: Moderate) and TG levels (WMD: −5.26 mg/dL, 95% CI: −10.3 to −0.16; I²: 4.40%, CoE: Low). Propolis supplementation also led to reductions in LDL-C (WMD: −4.78 mg/dL, 95% CI: −8.86 to −0.70; I²: 61.3%, CoE: Low), the LDL-C/HDL-C ratio (WMD: −0.20, 95% CI: −0.36 to −0.05; I²: 25.2%, CoE: Very Low), and the TC/HDL-C ratio (WMD: −0.24, 95% CI: −0.46 to −0.02; I²: 35.9%, CoE: Very Low). Corresponding forest plots are available in Supplementary Figs. 10–12.

Effect of honey, RJ, and propolis on glycemic profile

Honey and RJ demonstrated favorable effects on glycemic profiles. Honey supplementation was associated with a reduction in HbA1C (WMD: −0.25, 95% CI: −0.45 to −0.05; I²: 58.6%, CoE: Very Low), while RJ lowered INS levels (WMD: −3.63 µU/mL, 95% CI: −6.77 to −0.49; I²: 27.8%, CoE: Very Low). Notably, propolis supplementation improved glycemic control by reducing FBG (WMD: −10.9 mg/dL, 95% CI: −17.3 to −4.51; I²: 94.4%, CoE: Moderate), 2hPG (WMD: −38.7 mg/dL, 95% CI: −54.4 to −29.9; I²: 52.5%, CoE: Very Low), INS (WMD: −1.55 µU/mL, 95% CI: −2.47 to −0.63; I²: 88.4%, CoE: Very Low), HbA1C (WMD: −0.49, 95% CI: −0.76 to −0.21; I²: 85.2%, CoE: Low), and HOMA-IR (WMD: −0.66, 95% CI: −0.96 to −0.36; I²: 83.0%, CoE: Low). Refer to Fig. 2 and Supplementary Table 12, with corresponding forest plots in Supplementary Figs. 13–15.

Effect of honey, RJ, and propolis on inflammatory markers

As shown in Fig. 2 and Supplementary Table 13, honey elevated hs-CRP levels (WMD: 0.26 mg/dl, 95% CI: 0.06 to 0.47; I²: 92.4%, CoE: Very Low). In contrast, propolis further reduced IL-6 (WMD: −5.48 pg/mL, 95% CI: −7.10 to −3.86; I²: 97.1%, CoE: Moderate), MCP-1 (WMD: −44.3 pg/mL, 95% CI: −61.4 to −27.2; I²: 0.00%, CoE: Low), and TNF-α (WMD: −5.29 pg/mL, 95% CI: −7.08 to −3.49; I²: 98.9%, CoE: Very Low), while increasing IL-10 (WMD: 10.0 pg/mL, 95% CI: 5.56 to 14.5; I²: 99.5%, CoE: Very Low). For further details, refer to Supplementary Figs. 16–19.

Effect of honey, RJ, and propolis on oxidative stress markers

Figure 2 and Supplementary Table 13 show that RJ enhanced TAC levels (WMD: 0.97 mmol/L, 95% CI: 0.31–1.64; I²: 94.1%, CoE: Very Low). Propolis also elevated TAC (WMD: 0.60 mmol/L, 95% CI: 0.53–0.67; I²: 93.8%, CoE: Low) and increased GSHPx activity (WMD: 25.6 IU/L, 95% CI: 6.64–44.5; I²: 0.00%, CoE: Very Low), while lowered MDA (WMD: −0.16 µmol/L, 95% CI: −0.29 to −0.02; I²: 99.9%, CoE: Moderate). Further details are provided in Supplementary Figs. 17–19.

Publication bias

According to Supplementary Tables 11–13, among the 83 pairwise associations, ~15% displayed evidence of publication bias. Specifically, significant publication bias was observed for the effects of honey on BMI, ALP, GGT, and TC, as well as for the effects of RJ on TC. Additionally, publication bias was detected in the effects of propolis on LDL-C, HDL-C, the TC/HDL-C ratio, INS, HbA1C, HOMA-IR, and TNF-α. This level of publication bias suggests that smaller, non-significant studies may be missing from the analysis (see Supplementary Figs. 20–38 for details).

Influence analysis

Supplementary Tables 14–23 present the influence analysis results for the effects of honey, RJ, and propolis on cardiometabolic outcomes. This analysis revealed that the observed increase in DBP associated with honey supplementation was primarily attributable to the study by Tang et al. [62], with a recalculated WMD of 1.49 mmHg (95% CI: −0.19 to 3.18). Furthermore, after excluding the study by Fujisue et al. [19], a significant reduction in SBP with RJ supplementation was observed, yielding a WMD of −4.32 mmHg (95% CI: −6.96 to −1.69). Additionally, when the study by Rasad [8] was excluded, the effects of honey on BW (WMD: −0.17 kg, 95% CI: −1.12 to 1.46) and BMI (WMD: 0.03 kg/m², 95% CI: −0.44 to 0.54) were no longer significant. Similarly, after excluding the studies by Moayedi [85] and Miryan [114], the effects of propolis on BW (WMD: 0.04 kg, 95% CI: −0.00 to 0.09) and BMI (WMD: −0.02 kg/m², 95% CI: −0.18 to 0.12), respectively, were also rendered non-significant.

Regarding the effect of honey on lipid profile, upon excluding the study by Rasad [8], a significant increase in HDL-C levels was observed (WMD: 0.72 mg/dl; 95% CI: 0.14–1.31). Additionally, the removal of studies by Fujisue et al. [19] and Morita [69] revealed a significant decrease in TG (WMD: −7.53 mg/dl; 95% CI: −13.2 to −1.78) and LDL-C levels (WMD: −3.94 mg/dl; 95% CI: −7.75 to −0.32) following RJ supplementation, respectively. The observed effect of honey on reducing HbA1C levels appears to be driven by a single study [56]; when this study was excluded, there was no significant change in HbA1C levels (WMD: −0.23 mg/dl; 95% CI: −0.51 to 0.03).

Subgroup analysis

Supplementary Tables 24 and 25 present the results of subgroup analyses conducted according to the ICEMAN guidelines. Long-term honey interventions (12 weeks or more) were associated with increased FBG and AST levels. Studies with a moderate risk of bias indicated that RJ supplementation reduced BMI (Moderate credibility). Similarly, with moderate credibility, RJ demonstrated a more significant reduction in FBG in diabetic and individuals with overweight compared to non-diabetic and normal-weight individuals. Studies with low to moderate risk of bias also showed that propolis supplementation decreased BMI, FBG, and HbA1C. Propolis appeared more effective in reducing HbA1C and AST levels in obese participants compared to overweight individuals. Also, propolis led to a more substantial reduction in HOMA-IR among individuals with diabetes than in non-diabetic ones. Furthermore, the decrease in TNF-α levels was more pronounced in studies with intervention durations of less than 12 weeks.

Dose-response analysis

Figure 3, Supplementary Tables 26–28, and Supplementary Figs. 39–59 present the dose-response analysis results for the effects of honey, RJ, and propolis on selected cardiometabolic markers. The analysis revealed a linear relationship between honey intake and SBP, with each gram of honey consumed resulting in an SBP increase of 0.09 mmHg (95% CI: 0.03–0.14). In contrast, the effect of honey on AST followed a non-linear pattern, peaking at a daily dose of 25 g (WMD: 16.7 IU/L; 95% CI: 4.22–29.3) before declining with higher doses. Similarly, honey demonstrated a non-linear effect on FBG: it increased at a daily dose of 20 g (WMD: 3.81 mg/dl; 95% CI: 0.01–7.62) but decreased at doses of 75 grams or more (WMD: −4.41 mg/dl; 95% CI: −8.59 to −0.23). Additionally, the most significant reduction in HbA1C was observed with a daily dose of 10 g (WMD: −0.37; 95% CI: −0.56 to −0.18), though this effect diminished at doses of 50 g or higher.

Fig. 3. Non-linear dose–response analysis of honey and its derivatives on cardiometabolic outcomes.

Shown are the effects of a honey on SBP, b honey on AST, c honey on FBG, d royal jelly on TG, e royal jelly on LDL-C, and f propolis on MDA.

It is estimated that each gram increase in RJ intake decreases LDL-C levels by a WMD of −1.75 mg/dl (95% CI: −2.86 to −0.65). The most significant reduction in TG was observed with a daily dose of 500 mg, yielding a WMD of −10.5 mg/dl (95% CI: −18.7 to −2.37).

Figure 4 illustrates a linear dose-response relationship for propolis, where each 1-gram increase in dose is associated with reductions in ALT (WMD: −4.27 IU/L; 95% CI: −7.93 to −0.60), AST (WMD: −2.54 IU/L; 95% CI: −4.43 to −0.65), FBG (WMD: −17.2 mg/dl; 95% CI: −27.8 to −6.71), HbA1C (WMD: −0.74; 95% CI: −1.09 to −0.38), HOMA-IR (WMD: −0.74; 95% CI: −0.95 to −0.52), INS (WMD: −1.64 µU/mL; 95% CI: −2.53 to −0.75), IL-6 (WMD: −6.79 pg/mL; 95% CI: −9.92 to −3.67), TNF-α (WMD: −3.27 pg/mL; 95% CI: −5.26 to −1.28) and increase in TAC (WMD: 0.46 mmol/L; 95% CI: 0.42–0.50). Additionally, a non-linear effect of propolis on MDA levels was observed, with the maximum reduction occurring at a daily dose of 300 mg (WMD: −0.28 µmol/L; 95% CI: −0.51 to −0.04). However, at 1000 mg and higher doses, MDA levels increased (WMD: 0.05 µmol/L; 95% CI: 0.01–0.10).

Fig. 4.

Linear dose-response analysis for each 1gr increase in propolis intake on ALT, AST, FBG, HbA1C, HOMA-IR, Insulin, IL-6, TAC, TNF-α.

Methodological quality

Supplementary Table 29 evaluates the quality of the included SRMAs using the AMSTAR-II criteria. One study [30] was rated high quality, fulfilling all domain requirements. In contrast, four SRMAs were rated as low quality due to their failure to account for the impact of risk of bias on the pooled results [25, 28, 29, 31]. The remaining two SRMAs [32, 33] were rated as very low quality, attributed to the absence of a registered protocol and the aforementioned concern regarding the risk of bias.

Supplementary Tables 30–32 detail the quality assessment of the included RCTs. Among the RCTs involving honey supplementation, 87.5% had concerns regarding the risk of bias, primarily due to issues arising from the randomization process or deviations from the intended interventions. At the same time, the remaining trials were classified as low risk. For RCTs with RJ supplementation, the distribution was equal between those with low risk and those with some concerns regarding bias, with only one trial categorized as high risk due to biases across multiple domains. Regarding propolis supplementation, 60% of the trials were considered low-risk. At the same time, the rest had some concerns, primarily due to the randomization process and the selection of reported results. A smaller proportion had a high risk of bias due to worries across multiple domains.

Grading the evidence

As shown in Supplementary Tables 33–35, the primary reasons for downgrading the CoE for blood pressure and anthropometric measures were considerable risk of bias, indirectness due to limited studies, and imprecision from statistically or clinically non-significant results. For glycemic and lipid profiles, publication bias was detected as the most common reason for downgrading the CoE. In propolis supplementation, RCTs, ALT, and AST were rated as high quality, with the dose-response effect mitigating the downgrading due to imprecision. IL-6 and MDA were rated as moderate, owing to the modifiable impact of the dose-response relationship. However, IL-10, MCP-1, and TAC in propolis supplementation RCTs were rated low, while the remaining outcomes in this group were rated as very low certainty. Major indirectness and inconsistency were the additional common reasons for downgrading in these cases.

As shown in Supplementary Tables 33–35, the primary reasons for downgrading the CoE for blood pressure and anthropometric measures were considerable risk of bias, indirectness due to the limited number of studies, and imprecision stemming from statistically or clinically non-significant results. In addition, detected publication bias was the most common reason for downgrading CoE for glycemic and lipid profiles. ALT and AST in propolis supplementation RCTs were high quality. The dose-response effect was covered. The downgrade occurred due to imprecision. Although IL-6 and MDA in propolis supplementation RCTs rated moderate due to the modifiable effect of the dose-response relationship, IL-10, MCP, and TAC in propolis supplementation RCTs rated low, and the remaining outcome of this group had very low certainty. Major indirectness and inconsistency, in addition, were the most common reasons for downgrading.

Discussion

Honey consumption was associated with adverse effects on SBP, AST, TG, FBG, and hs-CRP. Conversely, RJ improved BP, lipid profile (TG and LDL-C), glycemic indices (INS and FBG), and TAC. Propolis supplementation reduced BMI, FM, and TBFP, enhanced lipid profile (LDL-C, LDL-C/HDL-C, TC/HDL-C), and improved glycemic markers (FBG, 2hPG, INS, HbA1C, HOMA-IR) and inflammatory markers (decreased IL-6, MCP-1, TNF-α, MDA; increased IL-10, TAC, and GSHPx).

Current evidence suggests that honey increases TG levels, and although it may slightly elevate HDL-C, the change is insufficient for therapeutic recommendations. In contrast, RJ improves lipid profiles by reducing TC, TG, and LDL-C. A dose-response analysis revealed that ~2200 mg of RJ is needed to achieve an MCID of ~4 mg/dL in LDL-C, while a dose of 500 mg significantly reduces TG beyond the MCID (10.5 vs. 8 mg/dL). Propolis has also been shown to lower serum cholesterol and its ratio to HDL-C, further supporting its lipid-modulating benefits.

Despite its potential health benefits, honey consists predominantly of carbohydrates, with fructose and glucose accounting for 95% of its composition [26]. Fructose, a key dietary sugar, plays a significant role in promoting lipid synthesis. High fructose intake may enhance lipid production, elevate plasma lipid levels, and increase the risk of cardiovascular diseases [120]. Mechanistically, fructose bypasses the phosphofructokinase regulatory step in glycolysis, which can result in hypertriglyceridemia [121]. While RJ is known to influence human lipoprotein metabolism, the exact mechanisms remain unclear. RJ’s cholesterol-lowering effects are partly attributed to bile acid-binding proteins and major RJ protein 1 (MRJP1), which exhibits hypocholesterolemic properties [122]. Meanwhile, propolis modulates lipid metabolism by upregulating ATP-binding cassette transporters (ABCA1 and ABCG1) in liver cells, enhancing cholesterol efflux and increasing HDL-C levels [123]. Additionally, propolis reduces TG levels by improving insulin sensitivity and activating lipoprotein lipase [124]. It also regulates peroxisome proliferator-activated receptors (PPARs) by upregulating PPARα and downregulating PPARγ, further contributing to its lipid-lowering effects [125].

Our findings indicate that honey intake increases FBG levels, particularly with long-term consumption. While daily honey consumption of up to 10 g (0.5 tablespoon) reduces HbA1C, higher amounts lead to an increase. Conversely, RJ supplementation reduces insulin and FBG levels. Propolis administration significantly improves glycemic markers, particularly in obese and individuals with diabetes. A dose-dependent relationship was observed, with 1700 mg and 3000 mg of propolis required to achieve clinically significant reductions in FBG (29 mg/dL) and INS (5 pmol/L), respectively.

While small amounts of honey have been suggested as a potential sugar substitute for individuals with diabetes [126], its high carbohydrate content raises concerns about its glycemic control effects. The fructose-to-glucose ratio in honey varies by type and may significantly influence its impact on metabolic parameters, which warrants further investigation [127]. Besides, key bioactive components of RJ, such as 10-hydroxy-2-decanoic acid (10H2DA), may enhance insulin activity and glycemic control by activating intracellular AMP-activated protein kinase (AMPK) and increasing GLUT4 gene expression [128, 129]. Additionally, RJ contains insulin-like compounds, such as sebacic acid, that further support its hypoglycemic effects [130, 131]. An SRMA by Karimian et al. demonstrated that propolis effectively manages glycemic levels in patients with diabetes mellitus [132]. Another SRMA confirmed these findings, reporting significant reductions in serum FBG and HbA1c levels with propolis consumption, though no changes were observed in INS levels or HOMA-IR [133]. The glycemic control benefits of propolis are attributed to multiple mechanisms: enhancing glycolysis and glucose utilization in liver cells, reducing carbohydrate absorption in the gastrointestinal tract, activating insulin-sensitive glucose transporters (GLUT-4) to promote glucose uptake in peripheral tissues like skeletal muscle, and inhibiting hepatic glucose release into the bloodstream [134].

Our study found that honey increased hs-CRP levels, while RJ elevated TAC. Propolis intake showed anti-inflammatory and antioxidant benefits by increasing IL-10, TAC, and GSHPx and reducing IL-6, MCP-1, TNF-α, and MDA. Linear dose-response analysis revealed that the effect sizes for IL-6 and TAC per gram of propolis exceeded their MCIDs. However, approximately 1900 mg of propolis was needed to achieve the MCID for TNF-α (~6.5 pg/mL).

Honey supplementation has shown mixed effects on inflammatory markers, with an increase in TNF-α and a decrease in hs-CRP, suggesting its impact on inflammation in chronic smokers remains inconclusive [57]. Its role in wound healing may stem from activating inflammatory cytokines in monocytic cells, which are crucial for tissue repair [135]. Animal studies highlight honey’s anti-inflammatory properties, partly due to its ability to suppress nitric oxide release. Honey also reduces inflammatory mediators such as COX-2 and TNF-α by inhibiting NF-κB nuclear translocation. Specifically, Gelam honey suppresses the NF-κB pathway, demonstrating antioxidant effects attributed to its phenolic compounds and radical scavenging activity [136].

A systematic review concluded that RJ improves the inflammatory response to muscle injuries. This effect is attributed to RJ fatty acids, particularly 10-H2DA, which modulate key inflammatory genes [137]. In vitro studies demonstrate that RJ fatty acids regulate proteins involved in the MAPK and NF-κB signaling pathways [138]. RJ also exhibits antioxidative properties, reducing lipid peroxidation caused by free radicals in vitro, suggesting its potential for treating inflammatory diseases [139]. An SRMA of 14 trials found that propolis positively affects CRP and TNF-α levels [33]. Propolis is rich in flavonoids and polyphenols, such as pinocembrin, quercetin, terpenes, and caffeic acid phenyl ester (CAPE), which reduce inflammation and oxidative stress by inhibiting the NF-κB pathway [140, 141]. Propolis polyphenols act as potent scavengers of reactive oxygen and nitrogen species, mitigating oxidative damage by suppressing the xanthine oxidase reaction [86].

Our findings indicate that honey consumption may nonlinearly increase AST levels, particularly at doses of 25 g/day over 12 weeks. Notably, Tang et al.‘s study in individuals with HIV with six months of honey supplementation has a great share in our results [62]. In contrast, in line with our results, a meta-analysis by Bahari et al. demonstrated that RJ does not significantly impact liver function or enzymes [29]. Propolis supplementation, however, showed dose-dependent improvements in AST and ALT levels, especially in individuals with overweight and obesity. Based on the MCIDs and effect size from linear dose-response analysis, a dose of 1700–1800 mg/day of propolis induces clinically significant changes in liver enzyme levels. A meta-analysis by Hallajzadeh et al. [33] and another SRMA of 21 RCTs confirmed the positive effects of propolis supplementation on liver enzyme levels [142]. The flavonoids in propolis may reduce cholesterol synthesis by inhibiting hepatic enzymes such as acyl CoA cholesterol o-acyltransferase and HMG-CoA reductase [143]. Propolis also regulates SREBP-1-responsive lipogenic genes, including Stearoyl-Coenzyme A desaturase I and Fatty acid-binding protein 5, promoting fat oxidation and reducing liver fat accumulation [144–146]. Additionally, its rich antioxidant and flavonoid profile—including galangin, naringin, pinocembrin, and chrysin—improves liver enzyme function [147].

Meanwhile, our study found that honey supplementation increased blood pressure, while RJ reduced blood pressure, and propolis showed no significant effect. Despite evidence of publication bias, previous SRMA reported no significant impact of honey on blood pressure [25]. However, our dose-response analysis revealed a linear relationship, with daily honey consumption of ~22 gr (1 tablespoon) increasing SBP by two mmHg. Experimental studies suggest that honey may improve blood pressure by reducing oxidative stress in hypertensive rat kidneys [148, 149]. Nonetheless, honey showed no beneficial effects on blood pressure in this study, possibly due to limited data, higher carbohydrate intake, elevated doses, and varying pathological conditions. Of note, honey’s effects on inflammation [57] and nitric oxide (NO) suppression [150] may underlie its blood pressure influence. Although observed blood pressure increases remained within normal limits (<120/80 mmHg), caution is advised for individuals with HTN. Experimental studies suggest RJ exhibits antihypertensive effects by enhancing NO production [151]. RJ contains muscarinic receptor agonists, likely acetylcholine, which induce vasorelaxation through NO release from vascular endothelium [152].

Additionally, Ile-Val-Tyr (IVY) peptides in RJ can lower BP by inhibiting angiotensin-converting enzyme (ACE) activity [153]. While our study did not find a significant effect of propolis on blood pressure, a previous SRMA reported that propolis supplementation improved SBP but not DBP [154]. These cardiovascular benefits are likely linked to its antioxidant, anti-inflammatory properties and enhancement of endothelial function [155].

Pairwise and influence analyses showed no significant effects of honey or RJ on anthropometric factors, while propolis significantly reduced FM and TBFP. Subgroup analysis indicated that propolis also reduced BMI in high-quality studies. Previous SRMA found no significant impact of propolis on anthropometric indices [33]. Although its modest anti-obesity effects may lack clinical significance [156], propolis may reduce fat absorption by promoting fecal fat excretion. Additionally, it can enhance thermogenesis by modulating gut microbiota, improving lipid metabolism, and stimulating brown adipose tissue development [157].

This study has several strengths, including a comprehensive evaluation of systematically identified SRMAs, robust statistical validation of results, and the application of diverse quality assessment tools. Nonetheless, limitations exist. High heterogeneity, likely due to variability in participants’ dosages, durations, and health statuses, was addressed through influence and subgroup analyses. The small number of studies on specific outcomes restricted definitive conclusions in these areas, and publication bias, though present, appeared to have minimal impact on the results. Additionally, the limited number of studies precluded dose-response analyses for all outcomes. Many studies also overlooked lifestyle factors, such as diet, physical activity, and smoking, which are crucial for interpreting health markers. Lastly, small sample sizes in some trials may have affected the overall effect estimates.

Despite promising therapeutic effects, the clinical application of honey bee products requires rigorous evaluation of their purity, ensuring the absence of pharmaceutical adulterants and heavy metals, and potency, defined by standardized concentrations of active constituents. Many commercially available preparations lack such quality control, and the precise doses of bioactive compounds required for efficacy remain unclear. Although our analysis identified an effective intervention dose, the variability in constituent content underscores the need for standardization. Future studies should prioritize these quality parameters to ensure safety and reproducibility.

Among honey bee products, RJ appears particularly promising for its antihypertensive and lipid-lowering properties, while propolis shows greater potential for improving glycemic control and exerting anti-inflammatory, antioxidant, and hepatoprotective effects. These attributes support their use in clinical practice for individuals with cardiovascular diseases, metabolic syndrome, hypertension, diabetes, or inflammatory conditions. However, given the limited evidence base, particularly concerning honey’s effects on blood pressure and inflammatory markers, further high-quality trials are warranted. Investigating combination therapies may uncover synergistic benefits, and future trials and meta-analyses should focus on high-risk populations.

Conclusions

In conclusion, consuming 10 g (0.5 tablespoon) of honey daily as a substitute for simple sugars may positively influence HbA1c. However, caution is advised for long-term use due to potential adverse effects on FBG, TG, liver enzymes, BP, and inflammatory markers. RJ at a daily dose of ~2200 mg can enhance antioxidant capacity, reduce blood sugar, improve insulin resistance, and particularly control the lipid profile. In addition to improving lipid indices, propolis, at 1700–1900 mg daily, specifically improved liver enzyme levels, glycemic control, inflammation, and oxidative stress. Further studies focusing on population-specific characteristics and optimized dosage are warranted.

Author contributions

MN: Conceptualization, Methodology, Formal analysis, Investigation, Validation, Writing - Review & Editing. SBZ: Conceptualization, Data Curation, Writing - Original Draft. MHR: Data Curation, Methodology, Writing - Review & Editing. SJ: Writing - Original Draft, Writing - Original Draft. FHF: Data Curation, Writing - Original Draft. ZM: Writing - Review & Editing, Validation. FT: Supervision. MM: Supervision. All authors read and approved the final version of the manuscript.

Data availability

The data supporting this study’s findings are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study followed the Declaration of Helsinki and received approval from the Ethics Committee of the Iran University of Medical Sciences (IR.IUMS.REC.1404.036). The review followed the Cochrane Handbook for Systematic Reviews of Interventions and adhered to the PRIOR and PRISMA guidelines, as detailed in Supplementary Tables 1 and 2.

Supplementary information

The online version contains supplementary material available at 10.1038/s41387-025-00403-9.