- Identical stool samples sent to seven direct-to-consumer microbiome testing companies produced substantially different bacterial profiles and health assessments.
- Across 18 commonly reported microbial genera, no company's results matched the consensus; and only three genera of 1,208 identified taxa appeared in every report.
- Researchers attributed the discrepancies to differences in laboratory methods and analysis pipelines, and say the results underscore the need for standardized testing and quality controls.
Direct-to-consumer gut microbiome tests produced markedly different results -- even when analyzing the same stool sample, researchers found.
Identical fecal samples sent via 21 home-testing kits to seven anonymized direct-to-consumer testing companies yielded a wide variation in reported bacterial abundance and in the health assessments generated from those data, reported Stephanie L. Servetas, PhD, of the National Institute of Standards and Technology (NIST) in Gaithersburg, Maryland, and colleagues in Communications Biology.
In some cases, there was not even agreement among kits produced by the same company. Of three identical samples analyzed by one provider, one was categorized as "unhealthy" while the other two were labeled "healthy."
The study is believed to be the first systemic evaluation of the analytical performance of these direct-to-consumer tests. Its findings raise questions about consistency and reliability in a rapidly expanding consumer testing market that operates outside traditional clinical laboratory regulation amid a generally favorable approach to the wellness industry by HHS under the Trump administration.
"We didn't do this to try to disparage the microbiome scientific movement," said co-author Scott Jackson, PhD, speaking during a briefing to announce the findings. "I am a proud member of this community, and I want only the best for it. We did this to really highlight the state of this field, and specifically, probably the most highest-stakes applications of this field, which are therapeutics and diagnostics."
Direct-to-consumer tests are not considered diagnostic or regulated as such, but "there's obviously a lot of similarity here," he said. "We want the industry and the scientific community to do better, and we're helping champion that movement to do better," added Jackson, formerly of the NIST and now founder of the NEST biotechnical consulting firm.
When researchers compared 18 commonly reported microbial genera across companies, no single provider aligned with the consensus profile for all 18. Across the full dataset, 1,208 unique taxa were reported, but only three genera appeared in every company's results.
The genus Clostridium was among the most variable. While the American Gut Project has reported average Clostridium abundance at just over 2.5% of the overall gut microbiome, one company in the study reported roughly five times that level, and three others failed to detect it in one or more samples. The genus includes species associated with human disease, including Clostridioides difficile, a cause of diarrhea.
The authors said the discrepancies likely stem from differences in sample processing, sequencing methods, bioinformatics pipelines, reference databases, reporting thresholds, and quality control standards. They argued that standardized sampling protocols and analytical benchmarks are needed to ensure reproducibility and consumer confidence.
"These tests have become popular, partly because people, I think, are increasingly interested in health and wellness, and partly because the gut microbiome has been linked -- at least in the public imagination -- to the idea that you can improve a whole range of conditions through diet and lifestyle changes," said co-author Diane Hoffmann, MS, JD, of the University of Maryland in Baltimore.
"There's been a lot of hype around that, but the hype doesn't really match the evidence. These tests often have limited evidence behind them, especially when it comes to informing clinical decisions or even basic dietary recommendations," she added. "So the marketing can be questionable, and consumers can end up misinterpreting or over-trusting results that aren't very reliable."
To create a standardized reference material, researchers homogenized stool from a single donor, prepared 100 mg/mL slurries, lyophilized the material, and stored it at -80°C to maintain consistency. These identical samples were then submitted to companies offering either 16S rRNA gene sequencing or whole metagenome shotgun sequencing workflows.
"What this material is meant to do is really to say how reproducible are the results, either between companies or within a company," said NIST microbiologist Servetas. "But it's not going to be able to tell us who was closest to the correct answer."
Disclosures
Jackson reported no competing interests.
One co-author founded LUCA Biologics.
