Strongest evidence: The most robust evidence is for reduced vomiting frequency, supported by 3 studies at moderate evidence strength. All studies reported beneficial effects, with doses up to 1 g/day showing significant reductions (OR 0.30–0.41). The strongest support is for chemotherapy-induced vomiting and hyperemesis gravidarum in pregnancy.
Mixed or weaker evidence: Three outcomes have low evidence strength. For improved insulin sensitivity, all 3 studies reported small beneficial effects, but only 2 were statistically significant; dosing and forms were inconsistent. For reduced hemoglobin A1c, only 1 of 3 studies reported a beneficial moderate-sized effect, while 2 higher-quality meta-analyses found neutral effects. For reduced pain intensity, results are mixed: 1 study showed a large beneficial effect for primary dysmenorrhea, but 2 others found neutral effects in different populations.
Effective dose patterns: A consistent dose range was identified only for vomiting (up to 1 g/day). For insulin sensitivity, hemoglobin A1c, and pain, no consistent effective dose emerged across studies, as many syntheses did not report specific doses or forms.
Population insights: The strongest population-specific findings are for chemotherapy patients and pregnant women (vomiting), as well as women with primary dysmenorrhea (pain). Populations studied for insulin sensitivity and hemoglobin A1c included individuals with type 2 diabetes, metabolic dysfunction-associated steatotic liver disease (MASLD), and polycystic ovary syndrome (PCOS).
Notable caveats: The evidence base across all outcomes is small (only 3 studies per outcome). Publication bias is a concern for vomiting and insulin sensitivity studies, as null results may be underrepresented. Doses, forms, and study durations were often not fully reported, limiting the ability to make specific recommendations. Many individual studies did not reach statistical significance.