Strongest evidence: Two outcomes—reduced body mass index (BMI) and reduced hemoglobin A1c—carry moderate evidence strength, each supported by three meta-analyses. In both cases, two of three studies found beneficial effects: BMI reductions ranged from small (weighted mean difference -0.51 kg/m²) to moderate, while HbA1c showed moderate-sized improvements. Effective doses span 200–4600 mg/day for these outcomes, though most studies lasted only 7 days, which is too short for reliable assessment of these metrics. Reduced fasting blood glucose also has moderate evidence, with two of three meta-analyses reporting mixed effect sizes (small to large) and a dose range of 200–4600 mg/day.
Mixed or weaker evidence: The outcomes with low evidence strength include improved insulin sensitivity, reduced body weight, and reduced postprandial glucose. All three studies reported beneficial effects on insulin sensitivity, but effect sizes varied (small to moderate) and the small evidence base raises concern about publication bias. For body weight, two of three meta-analyses found small-to-moderate benefits, but one in type 2 diabetes patients found no effect. Postprandial glucose showed the weakest support: only one of three studies found a moderate benefit, with two neutral results.
Effective dose patterns: Across multiple outcomes—fasting blood glucose, insulin sensitivity, body weight, postprandial glucose, and BMI—the most commonly tested dose range was 200 to 4600 mg/day. However, no single effective dose was identified, and dose reporting was inconsistent.
Population insights: The research predominantly involves people with type 2 diabetes or metabolic diseases. For BMI and body weight, benefits were seen in general adult populations with cardiovascular risk factors, while people with type 2 diabetes showed neutral results for some outcomes (BMI, body weight, postprandial glucose). This suggests potential population-specific differences in response.
Notable caveats: All six outcomes are based on only three meta-analyses each, making conclusions preliminary. The median study duration across many syntheses was only 7 days, far too short to capture meaningful changes in HbA1c, body weight, or BMI. The overwhelmingly positive results in some areas (e.g., insulin sensitivity) may reflect publication bias, as null studies are less likely to be published. Dosing was often unreported or widely variable.