Strongest evidence
Two outcomes have moderate evidence strength: Reduced Fasting Blood Glucose Levels (4 studies, 3 beneficial, small effect size) and Reduced Tumor Necrosis Factor Alpha (3 studies, all beneficial, small-to-moderate effect size). For fasting glucose, the typical effective dose is 200 mg/day, with effects observed around 12 weeks. For TNF-α, doses of 100–200 mg/day showed benefits in as little as 7 days in clinical populations (type 2 diabetes, sepsis).
Mixed or weaker evidence
Three outcomes have low evidence strength: Reduced Depression Symptoms (4 studies, all beneficial but mixed effect sizes, small sample size, many reviews with low evidence scores), Reduced Interleukin-6 Levels (3 studies, 2 beneficial, 1 neutral, inconsistent results), and Reduced Hemoglobin A1c (3 studies, 2 beneficial small reductions, 1 no effect). The depression evidence is promising but preliminary, with no consistent dose or population identified. IL-6 and HbA1c findings are inconsistent across studies.
Effective dose patterns
A dose range of 100–200 mg/day appears across multiple outcomes (fasting glucose, TNF-α, IL-6). For HbA1c, reported doses varied widely (5 mg/day to 1 g/day) without a clear effective range. The only outcome with a specific effective dose is fasting glucose at 200 mg/day.
Population insights
Studies predominantly enrolled clinical populations: people with type 2 diabetes, metabolic syndrome, prediabetes, obese women with type 2 diabetes, and ICU sepsis patients. Generalizability to healthy individuals is unclear. No age-specific or sex-specific patterns emerged beyond these clinical contexts.
Notable caveats
- Publication bias is a concern across all outcomes — null-result studies are less likely to be published.
- The evidence base is small (3–4 studies per outcome), making conclusions preliminary.
- Many studies are meta-analyses or reviews with low evidence scores, and individual study details (dose, duration, sample size) are often unreported.
- Results are inconsistent for IL-6 and HbA1c, with some meta-analyses finding no significant effect.