Study links microbe-rich diets with better insulin and cholesterol markers

By Tarun Sai LomteReviewed by Susha Cheriyedath, M.Sc.Feb 16 2026

New Australian research suggests that foods naturally rich in live microbes may be associated with better metabolic health, offering fresh insight into diet, microbiome interactions, and chronic disease risk. 

Study: Association between dietary intake of foods estimated to contain live microbes and health indicators in Australian adults: An exploratory analysis. Image Credit: UliAb / Shutterstock

In a recent study published in the journal Nutrition Research, researchers investigated associations between the intake of foods containing live microbes (LMs) and health indicators in Australians, with a primary aim of developing a database to estimate LM content in Australian foods and beverages and a secondary exploratory aim of examining health associations.

LMs are naturally present in many common foods, including raw vegetables and fruits, fermented foods, and probiotics. Interest in the intake of LM-containing foods has surged in recent years, given their associations with health and disease risk. High consumption of beneficial LMs has been associated with a lower risk of mortality in previous observational studies, particularly analyses of US population cohorts rather than Australian samples. However, most research on dietary LM intake has focused on American populations or specific foods rather than the whole diet.

Development of an Australian Live Microbe Food Database and Study Design

In the present study, researchers assessed associations between dietary LM intake and health indicators in Australians. First, they developed a database of LM content for common foods and beverages from the Australian Food and Nutrient (AUSNUT) database, relevant to the Australian Eating Survey (AES). Next, food and beverage items were stratified into low, medium, or high LM categories based on the expected prevalence of viable microbes, using previously published methods; microbial levels were estimated indirectly rather than directly measured in individual food samples.

The low category had an estimated microbial count of < 10⁴ colony-forming units per gram (CFU/g); the medium and high categories had counts of 10⁴-10⁷ CFU/g and > 10⁷ CFU/g, respectively. These data were then used in an exploratory cross-sectional analysis to investigate the relationship between estimated dietary LM intake and health indicators. Data from adults recruited in 2019-20 from the Newcastle region of Australia were analyzed.

Participants were aged 18 years or older and had a stable weight over the past two months. Individuals who were trying to conceive, pregnant, or breastfeeding, those taking medications affecting weight, fluid balance, or metabolic rate, and those with food allergies, chronic medical conditions, certain implanted medical devices, claustrophobia, or other protocol-specified exclusions were excluded. Participants reported demographic data and dietary intake using the AES Food Frequency Questionnaire, a validated instrument that may nonetheless overestimate some dietary intakes due to self-reporting.

The following cardiometabolic health indicators were measured: body mass index (BMI), blood pressure (BP), waist circumference, fasting plasma glucose, total cholesterol, triglycerides, fasting insulin, low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C). Estimated inflammatory markers included interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and C-reactive protein (CRP).

Differences in consumption of LM-containing foods by gender, smoking status, and ethnicity were assessed using the Kruskal-Wallis test and the Mann-Whitney U test. To explore relationships between estimated LM content categories and health indicators, Spearman’s rank correlation was first used to assess the direction of association. Subsequently, weighted least-squares (WLS) regression was used, adjusting for relevant covariates, including gender, smoking status, and energy intake, to account for potential confounding; however, residual confounding cannot be excluded in observational analyses.

Associations Between LM Food Categories and Health Indicators

The team categorized more than 200 food items from the AUSNUT database to create the LM database. About 229 items were classified as having low LM content, including vegetables, cereal-based products, and meat, poultry, and game products. Additionally, 21 items, including fruits, vegetables, and milk products, had medium LM content.

Of the five fermented foods, two were classified as high LM content and three as medium. Given the limited number of high-LM foods, the medium- and high-LM food groups were aggregated (Med/Hi) to improve statistical power, with yogurt remaining the only clearly high-LM food after grouping. The study included 58 adults, predominantly Caucasian (86%) and female (69%), with a mean age of 38.16 years and a BMI of 26.18 kg/m². Participants reported relatively higher fruit and vegetable intake than typically observed in the broader Australian population.

Participants primarily consumed the low LM food group (mean daily intake, 1,902 g), followed by the medium LM group (253.6 g/day). Males consumed significantly more low-LM foods than females, and non-smokers had significantly higher intake of Med/Hi LM foods than smokers. Consumption of the low-LM food group was positively correlated with BP.

By contrast, intakes of the medium and Med/Hi-LM food groups were positively correlated with HDL-C and negatively correlated with BMI, fasting insulin, body weight, waist circumference, CRP, and IL-6, although inflammatory marker associations did not remain statistically significant after covariate adjustment, and HDL-C associations remained statistically significant in adjusted analyses. WLS regression showed that consumption of the Med/Hi-LM food group was significantly inversely associated with BMI, insulin, and waist circumference, and positively associated with HDL-C. No significant adjusted associations were observed with fasting glucose, triglycerides, LDL cholesterol, total cholesterol, or TNF-α.

Interpretation, Limitations, and Future Research Needs

In summary, more frequent intake of foods with high or medium LM content was positively associated with HDL-C and inversely associated with insulin levels, BMI, waist circumference, and body weight in this Australian sample.

Further research is needed to corroborate these findings across larger, diverse populations and to determine whether dietary LM intake is associated with changes in the gut microbiota, particularly given the exploratory cross-sectional design, relatively small sample size, potential dietary reporting bias, and the inability of observational studies to establish causal relationships.