Table 1.
Characteristics and main findings of reviews included.
| Author and Year (Review Type) | N of Participants (Trials) Inclusion Criteria | Intervention (Control Type) | Outcomes—Glycaemic Control | Outcomes—Insulin Resistance | Certainty of Evidence for the Primary Outcome: HbA1c (GRADE) |
|---|---|---|---|---|---|
| VITAMIN C | |||||
| Ashor, A., et al. 2019 [31] (UR) Inception–Feb 2018. Date of search for most recent MA (Ashor 2017) was February 2016. |
N= 6409 (10 SR/MA; 3 on T2D) P = SRs and Mas of RCTs in adults of any health status; I = vitamin C administered alone; C = placebo; O = biomarkers of CVD risk (arterial stiffness, blood pressure, endothelial function, glycaemic control, and lipid profile). |
Ashor, et al. 2017: duration of 28 to 120 days, dose administered varied between 500 and 2000 mg per day [45] Tabatabaei-Malazy, et al. 2014: duration 4 weeks to 9 years, dose between 200 to 1000 mg/day [46] Khodaeian, et al. 2015: duration 4 to 16 weeks, dose between 800–1000 mg/day [47] |
↓FBG: SMD −20.59 (CI −40.77 to −0.4, 5 trials, n = NR) [46] ↓FBG: WMD −0.41 mmol/L (CI −0.78 to −0.04, 15 trials, n = 469) [45] HbA1c: NS (9 trials, n = NR) [45] |
Insulin: NS except for ↓fasting insulin in Ashor, et al. 2017: WMD: −15.67 pmol/L (CI −31.61 to 0.27, 3 trials, n = NR) [45] HOMA-IR: NS (3 trials, n = 92) [47] |
NA |
| Mason, S., et al. 2021 [43] (MA) Inception–September 2020 |
N = 1574 (28 trials) P = RCTs in people with T2D; I = vitamin C; C = placebo; O = HbA1c, FBG, PPG, FI, HOMA-IR, clamp insulin sensitivity, lipids, BP, oxidative stress markers 10 trials published after the Ashor umbrella review. |
Oral vitamin C supplementation Dose range: 200 to 3000 mg daily Duration: 2 weeks to 1 year, majority of studies <6 months duration (Placebo) |
↓FBG: MD −0.74 mmol/L (CI −1.17 to −0.31, 19 trials, n = 1305) ↓HbA1c: MD −0.54% (CI −0.90 to −0.17, 16 trials, n = 1133) ↓PPG: MD −0.95 mmol/L (CI −1.83 to −0.06, 4 trials, n = 235). No modifying effect from baseline Vitamin C concentration |
Fasting insulin (9 trials, n = 436), HOMA-IR (5 trials, n = 263) and clamp insulin sensitivity (3 trials, n = 86): NS |
Very low certainty * Evidence rated down for inconsistency (1 level), imprecision (1 level), and indirectness (1 level). |
| CHROMIUM | |||||
| Zhao, F., et al. 2021 [40] (MA) Inception–July 2020 |
N = 509 (10 trials) P = RCTs in people with T2D with lab values FBG ≥ 140 mg/dL, HbA1c ≥ 6.9%, triglyceride ≥ 125 mg/dL; I = chromium supplementation; C = placebo; O = HbA1c, FBG, triglycerides |
5 different forms of chromium supplements: Cr- containing milk powder, Cr-enriched yeast, chromium nicotinate, brewer’s yeast and chromium picolinate. Doses of chromium ranged from 42 to 1000 μg per day. Duration of intervention ranged from 90 days to 25 weeks. (Placebo) |
FBG: NS (10 trials, n = 522) ↓HbA1c: MD −0.54% (CI −0.98 to −0.09, 9 trials, n = 481) |
Very low certainty Quality of the evidence was downgraded one level for “risk of bias” due to high risk of bias for blinding and unclear allocation concealment; two levels for “inconsistency” due to varying point estimates, inconsistent direction of effect, limited overlap of confidence intervals and point estimates, and high heterogenetiy; one level for “imprecision” due to small sample sizes, and one level for “publication bias” |
|
| PROBIOTICS | |||||
| Cao, D., et al. 2021 [38] (MA) From inception–May 2020 |
N = 1948 (31 trials, 17 in people with T2D) P = RCTs in people with prediabetes, T2D or GDM; I = probiotics or synbiotics; C = placebo; O = FBG, HbA1c, fasting insulin, HOMA-IR, HOMA-B, and QUICKI Search: May 2020 |
Single-strain formulation was used in 4 studies and bacteria from Lactobacillus (including Lactobacillus sporogenes) and Bifidobacterium genera were included in probiotic formulations in all 17 and 9 of the included studies, respectively. 7 studies—synbiotics 10 studies—probiotics Dose range: 1 × 108 CFU to 1.00001 × 1012 CFU per day. Duration: 6 weeks to 6 months (Placebo) |
↓FBG: WMD −9.48 mg/dL (CI −16.24 to −2.72, n of trials NR, n = 1016) ↓HbA1c: WMD −0.43% (CI −0.69 to −0.18, n of trials NR, n = 635) |
Not provided by review authors, and GRADE assessment not possible due to insufficient reporting of individual RCT risk of bias for subgroups with T2D. Most (>50%) included trials for all disease groups were at unclear or high risk of bias, publication bias could not be ruled out, and heterogeneity was low. |
|
| ZINC | |||||
| Wang, X., et al. 2019 [42] (MA) From inception–February 2019 |
N-1700 (32 trials, 19 in people with T2D) P = RCTs in people with T2D, GDM, obesity, prediabetes; I = Zinc supplementation; C = placebo or co-supplementation only; O = FBG, 2-h postprandial glucose (2h-PG), fasting insulin, HOMA-IR, HbA1c, or hs-CRP. |
Zinc sulphate, gluconate, amino chelate, oxide, and acetate; in some cases, the anion was not specified. Dose range: 4–240 mg/d; median: 30 mg/d); mean 35 mg/d. Duration: 1 to 12 months. (Placebo or co-supplement only) |
↓FBG: WMD −20.34 mg/dL (CI −29.04 to −11.64, 12 trials, n = 752) HbA1c: NS (8 trials, n = 639) PPG: NS (5 trials, n = 256) |
HOMA-IR (4 trials, n = 234), fasting insulin (5 trials, n = 292): NS |
Very low certainty Quality of the evidence was downgraded one level for “risk of bias” due to limited appropriate sequence generation, blinding, and reporting of withdrawals and dropouts; two levels for “inconsistency” due to high heterogeneity, no consistent direction of effect, and varying point estimates; one level for “indirectness” due to insufficient timeframe in some trials; and one level for “imprecision” due to small sample sizes |
| MAGNESIUM | |||||
| Verma, H. and Garg, R. 2017 [39] (MA) Inception–June 2016 |
N = 1694 (28 trials; 17 T2D) P = RCTs in people with T2D/high risk of T2D; I = Magnesium (organic or inorganic) for at least 1 month; C = placebo; O = T2D associated CVD risk factors (FBG, FPI, HbA1C, TC, HDL, LDL, TG, SBP and DBP) |
Form of magnesium supplementation included Mg pidolate, citrate, aspartate, chloride, lactate, sulphate and oxide. Dosage ranged from 31.5 mg to 1006 mg of elemental Mg. Duration of intervention ranged from 4 to 24 weeks (Placebo) |
↓FBG: WMD −6.253 mg/dL (CI −10.602 to −1.904, 15 trials, n = 773) HbA1c: NS (7 trials, n = 505) |
Fasting insulin: NS | Not provided by review authors, and GRADE assessment not possible due to insufficient reporting of individual RCT risk of bias for subgroups with T2D. All studies were at unclear risk of bias for allocation concealment, and none were at high risk of bias for sequence generation or selective reporting. Risk of bias was generally low for blinding of participants, personnel and outcome assesors. Heterogeneity was moderate, and there was no evidence of publication bias. |
| POLYPHENOLS | |||||
| Jeyaraman, M., et al. 2020 [32] (MA) For Cochrane database: Inception –December 2018 Other databases: Inception–April 2018 |
N = 50 (3 trials) P = RCTs in people with T2D (for mixed studies, at least 80% of participants had to be adults with T2D); I = oral resveratrol (any regimen); C = placebo, anti-diabetic medications (OHA, insulin, herbal or nutrient supplements), diet/exercise or no treatment; O = adverse events (primary); diabetes-related or all-cause mortality, diabetes complications, HbA1c, FBG, insulin sensitivity, etc. (secondary). |
Resveratrol Dose range: 10 mg, 150 mg, or 1000 mg daily. Duration: 4 to 5 weeks. (Placebo) |
FBG: NS (2 trials, n = 33) HbA1c: NS (2 trials, n = 33) |
HOMA-IR: NS (2 trials, n = 36) |
Very low certainty * Downgraded by one level because of indirectness (surrogate outcome and insufficient time frame) and by two levels because of serious imprecision (low median sample size and small number of studies, CI ranging between benefit and harm) |
| Raimundo, A., et al. 2020 [33] (MA) Initial search: Inception–November 2016, and updated in Jane 2018 |
N = 1200 (20 total,14 in people with T2D) P = RCTs in people with prediabetes or T2D; I = pure (poly)phenol or an enriched fraction of (poly)phenols (4 weeks or more for glucose, 12 weeks or more for HbA1c); C = placebo; O = FBG, HbA1c, insulin, HOMA-IR, IAPP/amylin, glucagon, and C-peptide. Included cross-over trials. |
3 trials: Polyphenol mixture (from passion fruit, grape, pine tree park, among others- doses of 125–2093 mg/day) 5 trials: Resveratrol (doses of 40–1000 mg/day) 3 trials: Isoflavones (doses of 33–100 mg/day) 2 trials: Flavanols (doses of 560–1270 mg/day) 1 trial: Anthocyanin (392 mg/day) Duration: 4 to 52 weeks. (Placebo) |
↓FBG: MD −5.86 mg/dL (CI −11.34 to −0.39, 13 trials, n = 740) Subgroup analysis: ↓FBG in those taking anti-diabetic medication: MD − 10.17 (CI −17.59 to −3.75, 6 trials, n = 378) No subgroup analysis conducted in people with T2D for HbA1c |
Insulin (9 trials, n = 552) and HOMA-IR (7 trials, n = 489): NS | Not provided by review authors, and GRADE assessment not possible due to insufficient reporting of individual RCT risk of bias for subgroups with T2D. 12/14 trials were at moderate-high risk of bias. Heterogeneity was moderate and there was no evidence of publication bias for the included trials for all disease groups. |
| Ω-3 PUFAs | |||||
| Gao, C., et al. 2020 [35] (MA) Inception–May 2019 |
N = 820 (12 trials) P = RCTs in people with T2D; I = Fish oil supplementation alone; C = placebo; O = TG, TC, HDL-C, LDL-C, FBG, FPI, HbA1c, and HOMA-IR. |
n3-PUFAs in fish oil Dose range n-3 PUFA: 0.3 g/d to 10.08 g/d. Duration: 3 weeks to 6 months. (Placebo) |
FBG: NS at any time point; ≤1 month (3 trials, n = 102); 1–3 months (6 trials, n = 451) or >3 months (4 trials, n = 381) HbA1c: NS at any time-point; ≤1 month (1 trial, n = 20), 1–3 months (5 trials, n = 329), or >3 months (4 trials, n = 309) |
HOMA-IR: NS at any time-point; ≤1 month (2 trials, n = 82), 1–3 months (2 trials, n = 224) or >3 months (4 trials, n = 381) Fasting insulin: NS at any time-point; ≤1 month (2 trials, n = 61), 1–3 months (2 trials, n = 224), or >3 months (4 trials, n = 381) |
Very low certainty Quality of the evidence was downgraded one level for “risk of bias” due to unclear risk of bias across random sequence generation, allocation concealment, and blinding; one level for “inconsistency” due to limited overlap of confidence intervals and point estimates and inconsistent direction of the effect; one level for “imprecision” due to small sample sizes; and one level for “publication bias” |
| O’Mahoney, L., et al. 2018 [36] (MA and MR) Inception–July 2017 |
N = 1187 T2D (45 total, 31 with glycaemic outcomes) P = Parallel or cross-over RCTs in people with T2D; I = n3-PUFAs including in diet as long as dosage and duration could be determined; C = placebo; FBG, HbA1c, fasting insulin, HOMA-IR and C-peptide, lipid profile, inflammatory markers, BP. |
Ω-3 PUFAs in capsule/liquid or diet (sardine-enriched) form. Origin of n3-PUFAs not reported. All trials used EPA, DHA, or a combination. Dose range: 0.40 to 18.00 g. Duration: 2 to 104 weeks (14/33 RCTs were >3 months duration) (Placebo) |
FBG: NS (28 trials, n = 1702) ↓HbA1c: Effect size −0.27 (CI −0.48 to −0.06, 31 trials, n = 2021) In “leave one out” sensitivity analysis for HbA1c, removal of two RCTs attenuated the statistical significance of the results to non-significant. |
HOMA-IR, fasting insulin: NS |
Very low certainty Quality of the evidence was downgraded one level for “inconsistency” due to high heterogeneity, inconsistent direction of effect and limited overlap of confidence intervals and point estimates; one level due to insufficient timeframe in some trials; and one level for “imprecision” due to small sample sizes and confidence interval not consistent with benefit |
| Ebada, M., et al. 2019 [34] (MA) Inception–May 2017, and updated on April 2018 |
N = 553 (10 trials, 8 T2D) P = RCTs in people with T1D or T2D; I = 1) RCTs with DM patients (both T1 and T2); alpha-lipoic acid (ALA); C = placebo; HbA1c, FBG, PPG, HDL, LDL, TG, TC, HOMA, Glutathione peroxidase, and waist circumference. |
Alpha-lipoic acid oral or intravenous. Dose range: 300–600 mg/d. The follow-up duration ranged from three weeks to six months. Duration of intervention NR. (Placebo) |
FBG: NS (6 trials, n = 322), HbA1c: NS (6 trials, n = 316) PPG: NS (3 trials, n = 190): |
Very low certainty Quality of the evidence was downgraded by one level for “inconsistency” due to limited overlap of confidence intervals and point estimates, moderate heterogeneity and inconsistent direction of effect; one level for “indirectness” due to limited applicability and insufficient timeframe for some trials; one level for “imprecision” due to small sample sizes; and one level for “publication bias” |
|
| VITAMIN D | |||||
| Hu, Z., et al. 2019 [37] (MA) From inception of database–March 2018 |
N = 747 (19 trials) P = RCTs in people with T2D; I = Vitamin D; C = placebo; O = FBG, insulin, HbA1c, HOMA-IR. |
Vitamin D (type NR) Dose range: 1000 IU/d to 300,000 IU single IM injection Duration: 4 wk to 12 mo. Duration <6 mo considered short-term. (Placebo) |
FBG: NS (14 trials, n = 289), HbA1c: NS (19 trials, n = 747) Short-term (<6 mo): ↓ HbA1c: SMD −0.17% (CI −0.27 to −0.04, 15 trials, n = 1059) Long term: NS. |
↓ HOMA-IR: SMD −0.60 (CI −0.79 to −0.42, 9 trials, n = 425) ↓Fasting insulin: SMD −0.49 (CI −0.68 to −0.31, 9 trials, n = 436) Short-term (<6 mo): ↓ HOMA-IR: SMD −0.75 (CI −0.97 to −0.53, 8 trials, n = 405) ↓ Insulin: SMD −0.57 (CI −0.78 to −0.35, 8 trials, n = 389). Long term: NS. |
Very low certainty Quality of the evidence was downgraded one level for “risk of bias” because allocation concealment was unclear; two levels for “inconsistency” because the direction of the effect was inconsistent and there was high heterogeneity; and one level for “imprecision” due to small sample sizes |
| FOLATE | |||||
| Lind, M., et al. 2019 [41] (MA) Pubmed from 1953–March 2018 Web of Science from 1900–March 2018 EMBASE from 1974–March 2018. |
N = 572 (29 total, 8 T2D) P = Parallel and cross-over RCTs with no restriction on health condition; folate supplementation; C = placebo; O = glucose, insulin, HOMA-IR, or HbA1c. |
Folate given as adjuvant therapy (alongside antidiabetic medication ± insulin). Two studies combined folate with B12 and B6. Dose range: 0.25 mg folate and 5 mg folic acid/d, with most studies using dosage of 5 mg. Duration: 2 weeks to 2 years, with the majority of studies lasting between 4 and 8 weeks. (Placebo) |
FBG: NS (6 trials, n = 309) HbA1c: NS (7 trials, n = 482). No differences found on subgroup analysis by baseline folate concentration, however this was conducted on the pooled sample which included people without T2D. |
HOMA-IR: NS (9 trials, n = 431) |
Very low certainty Quality of the evidence was downgraded two levels for “risk of bias” due to unclear allocation concealment and blinding of outcome assessment, and high risk of selective reporting; one level for “inconsistency” due to varying point estimates and limited overlap of confidence intervals and point estimates; one level for “indirectness” due to insufficient timeframe and inclusion of co-interventions in some trials; and one level for “imprecision” due to small sample sizes. |
Legend-Review Type: MA: Meta-analysis; MR: Meta-regression; UR: Umbrella review; SR: Systematic Review. Interventions: DHA: docosahexaenoic acid; EPA: eicosapentaenoic acid. Outcome Measures: FBG: fasting blood glucose; PPG: post-prandial glucose; HbA1c: glycosylated haemoglobin; HOMA-IR: Homeostatic Model Assessment for Insulin Resistance; FPI: fasting plasma insulin; QUICKI: Quantitative Insulin-Sensitivity Check Index; HDL-C: high density lipoprotein cholesterol; LDL-C: low density lipoprotein cholesterol; NR = not reported; NS = results not statistically significant; OHA: oral hypoglycaemic agents; TG: triglycerides; TC: total cholesterol; SBP: systolic blood pressure; DBP: diastolic blood pressure. * GRADE assessment conducted by systematic review authors; ↓=decrease in outcome measure when intervention was compared to control.