Microbiome Scores and Indexes

What goes into a score

Scores typically combine some of these:

• Diversity: how many kinds of microbes are present and how evenly they are spread.
• Named groups: amounts of certain bacteria that studies linked to disease or its absence.
• Function: genes or products, such as those linked to short-chain fatty acids (see Microbial Metabolites and Short-Chain Fatty Acids).

Each ingredient has weaknesses, covered below.

A research example: the Gut Microbiome Wellness Index 2

In 2024, researchers described GMWI2, an open-source, disease-agnostic indicator built from 8,069 stool shotgun metagenomes from 54 published studies in 26 countries. It reached a cross-validation balanced accuracy of 80% for separating people with no disease from people with disease, and nearly 75% on average in inter-study and external validation. [2] Label: promising but limited.

How to read that:

• "Healthy" here means "no disease recorded in the dataset", not good health or a high score of well-being. [2]
• Accuracy refers to sorting groups of samples in research datasets. It does not tell what a number means for one person's symptoms or future.
• It is built from taxonomic profiles, meaning which microbes were found, in stool. [2]

Diversity numbers are not a health grade

Diversity is not a verdict (see Microbial Diversity and Health Associations). A 2021 review of 59 case-control studies in psychiatric conditions found small, inconsistent differences in diversity across diagnoses. Richness was lower in patients overall, yet consistently decreased only in bipolar disorder when individual diagnoses were examined. [3] Label: insufficient evidence for diversity as a personal health indicator.

Why a single score is hard to trust

There are no validated normal ranges. A 2019 expert workshop said microbiome-based indicators must still be determined and validated, along with normal ranges, using approaches similar to those for other biomarkers. [1]

Everyday factors move the numbers. Stool consistency and medication were among the largest sources of variation in a study of over 2,200 people. [4] Alcohol intake and bowel movement quality also differ between healthy and diseased groups and can confound studies. [5]

Labs differ. In a blinded 15-laboratory study, variability depended most on the sample type and origin, then DNA extraction, handling and analysis. [6] Seven consumer services given the same standardized fecal material showed major discrepancies, with differences between providers on the same scale as differences between donors. [7]

A score can look more official than it is. Veterinary medicine offers a contrast: a "Dysbiosis Index" exists as a validated quantitative PCR test for dog feces, with reference intervals established for dogs. [8] No equivalent validated consumer index with reference intervals for people was found in the sources read for this page.

Questions to ask about any score

• Who was the reference group, and was it like you?
• Was the score checked in a separate group of people it was not built on?
• Does the company show that people who changed their score improved a health outcome (not only the score)? Proper evidence not found for consumer microbiome scores; the research, consensus and regulator material cited on these pages was checked.
• Does the score change if the same sample goes to another lab?

A score that rises after a product is a change in a number, not a health outcome. See Surrogate vs Patient-Important Outcomes.

Related guides

Commercial Stool Tests, Dysbiosis in Research and Marketing.