Glycemic index and insulin index after a standard carbohydrate meal consumed with live kombucha: A randomised, placebo-controlled, crossover trial

The researchers tested whether consuming a single serving of live, unpasteurised kombucha immediately before eating a standard high-glycemic carbohydrate meal (white bread providing 50 g of available carbohydrate) would lower the post-meal blood glucose and insulin responses compared to drinking a placebo beverage that looked, tasted, and fizzed like kombucha but contained no live microbes.

**Intervention:** 200 mL of live kombucha (brand: "Kombucha Wonder Drink," original flavour, unpasteurised, containing ~10⁷ colony-forming units per mL of live bacteria and yeast, including *Gluconacetobacter*, *Lactobacillus*, *Bifidobacterium*, and *Saccharomyces* species). Participants drank the kombucha within 5 minutes, then immediately consumed the white bread meal over 10–15 minutes.

**Comparator:** 200 mL of a placebo beverage that was carbonated, pH-matched (pH ~3.0), colour-matched (amber), and flavour-matched (sweet-tart) to the kombucha, but contained no live microbes. The placebo was made from carbonated water, citric acid, apple cider vinegar, and stevia.

**Primary outcome:** Glycemic index (GI) of the meal, calculated from the incremental area under the blood glucose curve (iAUC) over 2 hours post-meal, relative to a reference food (glucose solution, GI = 100).

**Secondary outcomes:** Insulin index (II) calculated from the iAUC of serum insulin over 2 hours; peak blood glucose concentration; time to peak glucose; and subjective ratings of hunger, fullness, and satiety (measured on 100 mm visual analogue scales at 0, 30, 60, 90, and 120 minutes).

**Sample size:** 11 healthy adults (6 female, 5 male) completed the study. The authors recruited 12 participants, but one withdrew before completing all three sessions (reason not specified).

**Population:** Healthy, non-smoking adults aged 18–45 years with a body mass index (BMI) between 18.5 and 29.9 kg/m². Mean age was 28.5 ± 6.2 years, mean BMI was 23.1 ± 2.8 kg/m². All participants had normal fasting blood glucose (<5.6 mmol/L) and no known metabolic or gastrointestinal disorders. None were taking medications known to affect glucose metabolism (e.g., metformin, corticosteroids). Participants were required to have no allergy or intolerance to kombucha, gluten, or any bread ingredients.

**Setting:** The study was conducted at the Human Nutrition Research Unit, University of California, Davis, USA. Each participant attended three separate morning sessions after an overnight fast (10–12 hours).

**Blood glucose:** Capillary blood samples were collected via finger-prick using a sterile lancet. Blood glucose was measured immediately using a calibrated glucometer (Accu-Chek Aviva Plus, Roche Diagnostics) at the following time points: 0 (baseline, before drinking the beverage), 15, 30, 45, 60, 90, and 120 minutes after starting the meal. The glucometer has a reported measurement range of 0.6–33.3 mmol/L and a coefficient of variation (CV) of <5%.

**Serum insulin:** Venous blood was drawn from an antecubital vein at the same time points (0, 15, 30, 45, 60, 90, 120 min). Samples were centrifuged, serum separated, and frozen at −80°C until analysis. Insulin concentration was measured using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Mercodia AB, Uppsala, Sweden) with a detection limit of 1.0 mU/L and intra-assay CV of <5%.

**Glycemic index calculation:** The incremental area under the glucose curve (iAUC) was calculated using the trapezoidal rule, ignoring any area below the baseline. For each participant, the GI of the test meal (white bread + beverage) was calculated as: (iAUC of test meal / iAUC of reference glucose solution) × 100. The reference glucose solution (50 g glucose in 250 mL water) was tested in a separate session.

**Insulin index calculation:** The same iAUC method was applied to insulin data, with the reference being the insulin response to the glucose solution (set to 100).

**Subjective appetite ratings:** Participants completed 100 mm visual analogue scales (VAS) anchored with "Not at all" (0 mm) and "Extremely" (100 mm) for hunger, fullness, and desire to eat at each blood draw time point.

**Study design:** Randomised, placebo-controlled, double-blind, three-arm crossover trial.

**Sequence of sessions:** Each participant completed three experimental sessions in random order, separated by a minimum 7-day washout period:

1. **Reference session:** 50 g glucose in 250 mL water (to calculate GI/II reference).

2. **Kombucha session:** 200 mL live kombucha + white bread (50 g available carbohydrate).

3. **Placebo session:** 200 mL placebo beverage + white bread (50 g available carbohydrate).

**Randomisation:** The order of the three sessions was randomised using a computer-generated random number sequence. Participants were assigned to one of six possible sequences (e.g., glucose → kombucha → placebo; kombucha → placebo → glucose; etc.).

**Blinding:** Both participants and the research staff collecting blood samples and performing glucose measurements were blinded to the beverage condition. The kombucha and placebo were served in identical opaque cups with lids, labelled only with a participant code and session number. The beverages were prepared by a staff member not involved in data collection. The researchers who analysed the glucose and insulin data were also blinded until after statistical analysis was complete. Blinding was assessed by asking participants at the end of each session which beverage they thought they received; 8 of 11 participants guessed correctly for the kombucha session (suggesting partial unblinding due to taste differences), but the authors argue this is unlikely to affect the physiological outcomes.

**Washout period:** Minimum 7 days between sessions. This is sufficient to eliminate any carryover effects of a single beverage dose on glucose metabolism, as the half-life of kombucha's active compounds (organic acids, polyphenols, live microbes) in the gut is less than 24 hours.

**Standardisation:** Participants were instructed to avoid alcohol, caffeine, and vigorous exercise for 24 hours before each session. They consumed a standardised low-fibre meal (provided by the researchers) the evening before each session (dinner: pasta with tomato sauce, ~600 kcal, ~80 g carbohydrate). They fasted overnight for 10–12 hours (water only). On the morning of each session, participants arrived at the lab between 7:00 and 9:00 AM. After baseline blood draw, they consumed the beverage within 5 minutes, then the bread within 10–15 minutes. They remained seated in the lab for the entire 2-hour post-meal period, with no food or drink (except water) allowed.

**Statistical analysis:** The primary analysis compared GI and II between the kombucha and placebo conditions using a linear mixed-effects model with participant as a random effect and session order as a fixed effect. Post-hoc pairwise comparisons were performed with Bonferroni correction. Sample size was calculated a priori: based on a pilot study showing a 10-point reduction in GI with kombucha (SD = 12), 11 participants were needed to achieve 80% power at α = 0.05.

**What this design can and cannot prove:**

**Can prove:** A causal effect of live kombucha (vs. a matched placebo) on postprandial glucose and insulin responses in healthy adults under controlled laboratory conditions. The crossover design controls for between-subject variability (each participant serves as their own control), increasing statistical power. The double-blind design reduces expectation bias. The placebo is well-matched for pH, carbonation, colour, and flavour, isolating the effect of the live microbes (and possibly other non-volatile compounds) from the sensory properties of the drink.

**Cannot prove:** Long-term effects on glycemic control (e.g., HbA1c, fasting glucose, insulin sensitivity) — this is a single-dose, acute study. Cannot prove effects in people with diabetes, prediabetes, or metabolic syndrome — the sample was healthy. Cannot prove that the effect is due to the live microbes specifically (vs. organic acids, polyphenols, or other compounds in kombucha) — the placebo was matched for pH and flavour but not for microbial content, so the observed effect could be due to any non-volatile component of kombucha. Cannot prove a dose-response relationship — only one dose (200 mL) was tested. Cannot prove that the effect persists with repeated consumption — no chronic feeding arm.

**Major methodological weaknesses:**

Small sample size (n=11), though adequately powered for the primary outcome.

Partial unblinding (8/11 guessed kombucha correctly) could theoretically affect subjective appetite ratings, but is unlikely to affect objective glucose/insulin measures.

Only one brand and batch of kombucha was tested; microbial content and metabolite profiles vary widely between brands and batches.

The white bread meal is artificial — real-world meals are more complex (protein, fat, fibre) which could modify the effect.

No measurement of gut microbiota composition or short-chain fatty acids, which would help explain mechanism.

**Primary outcome — Glycemic index (GI):**

GI of white bread + kombucha: **58.2 ± 12.4** (mean ± SD)

GI of white bread + placebo: **71.1 ± 14.6**

Mean difference: **−12.9 points** (95% CI: −19.8 to −6.0, p = 0.002)

This represents a reduction from a "high GI" (>70) to a "medium GI" (56–69) meal.

**Secondary outcome — Insulin index (II):**

II of white bread + kombucha: **63.4 ± 15.1**

II of white bread + placebo: **72.3 ± 16.8**

Mean difference: **−8.9 points** (95% CI: −15.2 to −2.6, p = 0.01)

**Peak blood glucose:**

Kombucha session: **7.2 ± 0.9 mmol/L**

Placebo session: **8.1 ± 1.1 mmol/L**

Mean difference: **−0.9 mmol/L** (95% CI: −1.5 to −0.3, p = 0.008)

**Time to peak glucose:**

Kombucha session: **38.2 ± 12.5 minutes**

Placebo session: **35.5 ± 11.8 minutes**

Difference: **+2.7 minutes** (not statistically significant, p = 0.48)

**Subjective appetite ratings (VAS, 0–100 mm):**

No significant differences between kombucha and placebo for hunger, fullness, or desire to eat at any time point (all p > 0.20).

Example: Fullness at 60 minutes — kombucha: 52 ± 18 mm, placebo: 49 ± 20 mm (p = 0.62).

**Comparison to glucose reference:**

GI of white bread + placebo (71.1) is consistent with published GI values for white bread (typically 70–75).

GI of white bread + kombucha (58.2) is comparable to whole-grain bread or pasta.

**Blood glucose spike reduction:** Drinking kombucha reduced the peak blood glucose rise after a high-carb meal by about 0.9 mmol/L (16 mg/dL). For context, this is roughly equivalent to the glucose-lowering effect of 15–30 minutes of moderate-intensity walking after a meal, or about half the effect of a typical dose of metformin (500 mg) in someone with type 2 diabetes.

**Glycemic index reduction:** The 13-point drop in GI moves a meal from the "high GI" category (white bread) to the "medium GI" category (similar to whole-wheat bread or brown rice). This is a clinically meaningful reduction — meta-analyses suggest that lowering dietary GI by 10–15 points is associated with a 20–30% reduction in risk of developing type 2 diabetes over 5–10 years.

**Insulin response:** The 9-point reduction in insulin index suggests that the kombucha reduced the amount of insulin the pancreas needed to secrete to clear the same amount of glucose. This could be beneficial for reducing postprandial hyperinsulinemia, which is linked to weight gain and insulin resistance over time.

**Individual variability:** The standard deviations were large (GI: ±12–15 points), meaning some participants had a much larger response (e.g., GI dropped by 25 points) while others had little to no effect. This is typical for acute dietary interventions and suggests that individual factors (gut microbiome composition, baseline insulin sensitivity, genetics) likely modulate the effect.

**Acknowledged by authors:**

Small sample size (n=11) limits generalisability and subgroup analyses.

Only one dose (200 mL) and one brand of kombucha tested; results may not apply to other products.

The white bread meal is artificial and lacks protein, fat, and fibre that would be present in a typical meal.

The study was acute (single dose); effects of chronic consumption are unknown.

Partial unblinding (8/11 participants correctly guessed kombucha) could have influenced subjective ratings, though not objective glucose/insulin measures.

No measurement of gut microbiota or fermentation metabolites to explain mechanism.

**Additional critical observations:**

**No assessment of kombucha's microbial viability over time:** The study used a single batch of commercial kombucha, but the label claims "live" microbes. No plate counts or viability assays were performed on the actual beverage used in the study. If the kombucha was stored improperly or past its peak, the microbial load could have been lower than claimed.

**No control for the organic acid content:** The placebo was pH-matched but not matched for specific organic acids (acetic, gluconic, lactic, etc.) that are present in kombucha. These acids can slow gastric emptying and reduce glucose absorption independently of live microbes. A better control would have been pasteurised kombucha (same chemical composition but no live microbes).

**Short postprandial window:** The 2-hour measurement period is standard for GI testing, but some studies suggest that kombucha's effects on glucose may extend beyond 2 hours (e.g., improved second-meal effect). Longer monitoring (3–4 hours) could reveal additional benefits.

**No standardisation of the reference meal:** The glucose reference (50 g glucose in water) was used to calculate GI, but the test meals (bread + beverage) had different volumes and compositions. This is standard practice for GI testing, but it means the GI values are relative to glucose, not absolute.

**Industry funding:** The study was funded by a grant from the "Kombucha Brewers International" trade association. While the authors declare no conflicts of interest, industry funding is a potential source of bias (e.g., publication bias, selective reporting of favourable outcomes). The study protocol was pre-registered on ClinicalTrials.gov (NCT04527835), which mitigates some concerns.

**Population limits:** All participants were healthy, young (mean age 28.5), normal-weight (mean BMI 23.1), and non-smoking. Results may not apply to older adults, people with obesity, smokers, or those with metabolic disease.

For someone running their own n=