Is your nutrigenetic test reliable? A checklist to assess it
Fifteen questions, from the variants and their evidence to the weights, recommendations and your data, to assess a nutrigenetic test before trusting it.
When someone offers you a nutrigenetic test, how do you know whether it is worth it? A 2026 review of direct-to-consumer tests that include food recommendations found 104 companies with 204 panels. Only 56 of them (54%) publicly disclosed which genes they used to make recommendations, and 55 (53%) offered support after the result, often at extra cost and with very different qualifications among those providing it (McCartney, 2026). With an offer that variable, it helps to have an orderly way of reading a report before making decisions with it. This is the checklist I use, in fifteen questions.
Where the checklist comes from
Almost thirty frameworks exist for evaluating a genetic test. A systematic review located 29, most of them based on the ACCE model, and the most widely used criteria are analytic validity, clinical validity, clinical utility and ethical, legal and social implications (Pitini, 2018). That same review points out that less attention is paid to context: how the result is delivered, who interprets it and what the person receiving it thinks. The checklist covers the four classic criteria and adds a fifth block with that context.
The fifteen questions
| Block | Question | A good answer |
|---|---|---|
| What it measures | 1. Which variants does it analyse, with their identifier (rsID)? | A complete list you can check in a public database, with the strand stated. |
| What it measures | 2. With what technique and what quality controls? | The platform and the sample quality criteria. For variants that are not a letter of DNA, such as the copy number of a gene, a specific technique. |
| What it measures | 3. What happens if a variant cannot be read? | The report says "not determined" instead of inventing a result. |
| What has been observed | 4. Does it cite the studies for each variant? | References you can open, with the population and size of each study. |
| What has been observed | 5. What level of evidence does it give each result? | It separates what has been replicated from what is exploratory. A systematic review using the GRADE method on genes and the response to omega-3 found insufficient evidence for most interactions and strong evidence for two (Keathley, 2022). |
| What has been observed | 6. Has it been replicated in people like you? | It states the population of the studies. The scores available today are several times more accurate in people of European ancestry than in others (Martin, 2019). |
| How it is combined | 7. How much does each variant weigh and where does that weight come from? | It says whether the weight comes from the measured effect or from the strength of the evidence (see the second part of the carbohydrate series). |
| How it is combined | 8. What does the final number mean? | It makes clear whether it is an index, a percentile or an absolute risk, and relative to which population. |
| How it is combined | 9. Was the combination checked in an independent group? | A trial or an external validation. In the DIETFITS trial, with 609 adults, a three-gene pattern showed no interaction between genotype and diet (P = 0.20) (Gardner, 2018). |
| What to do | 10. Is there a recommendation for each genotype, including "no variant"? | A concrete one, with the basis that supports it: mechanism, observation in studies or an intervention trial. |
| What to do | 11. Are the figures in the recommendations labelled? | A figure per genotype is useful if it says it is the provider's estimate and shows the calculation. |
| What to do | 12. Does it separate what is proven from what you should try and observe? | It marks what is exploratory as such. |
| Framework and context | 13. What is the product in regulatory terms and how is it sold? | It states whether it is an in vitro diagnostic medical device. In Spain, Royal Decree 942/2025 treats genetic tests with that status as prescription products and bans selling them to the public by mail or by electronic means (article 19.5). |
| Framework and context | 14. Who interprets the result and what support is there afterwards? | A qualified professional, with the chance to resolve questions after receiving the report. |
| Framework and context | 15. What happens to your data and who gains what from the test? | Where it is stored and how it is deleted, and a clear statement of interests: who sells it and what else they offer. |
How to use it
Score each question 0 if the report does not say, 1 if it says so partly and 2 if it says so in detail and with a source. The total runs from 0 to 30. The cut-offs in the next table are a CESGEN teaching convention, not a validated instrument, and the right-hand column is a proposal to try and observe.
| Total | What you can do |
|---|---|
| 24 to 30 | It is a well-documented report. Use it to rank priorities and to try the proposals for a few weeks, with your blood tests and measurements as the check, and run the checklist again if the provider changes version. |
| 12 to 23 | Use it with caution. Ask in writing for what is missing (variants, weights, reference population) and start with the proposals that already fit the general guideline: more fibre, protein spread across meals, strength work and walking. |
| 0 to 11 | Treat it as interesting reading and not as a basis for changing your diet or starting a supplement. Ask the provider for the method and discuss the report with your health professional. |
Warning signs, and what to do if the test is already done
Some signs weigh more than any score: a report with no rsIDs; traffic-light scores that add up variants with different levels of evidence as if they were worth the same; doses or supplements recommended from a single gene; products for sale tied to your result; no data-deletion policy, and promises of diagnosis or guaranteed weight loss. If the test is already done, ask for the report and, if it exists, the raw data file. Remember that a missing position is not the same as a "normal" genotype, and that any finding with possible clinical consequences is confirmed in a clinical laboratory before acting on it.
An example with the algorithm from the series
If you run the checklist on the algorithm in the first part, it does well on which variants it uses, how much each one weighs and where the weights come from. It does not include a check in an independent group or a reference population, which is why that post presents it as a teaching example and adds that today I would review each variant against recent literature before using it.
What to bear in mind
A high score on the checklist says the report is transparent, not that its recommendations will work for you: what works is checked with your own measurements over a few weeks. Nor is there a single closed framework. For polygenic scores there are reporting standards from the scientific community, such as those of ClinGen and the PGS Catalog, which define minimum requirements on the population, the development and evaluation of the model and its generalisation beyond the population studied (Wand, 2021), and the checklist shares their spirit.
CESGEN teaches nutrigenomics courses, where this evaluation is practised with real reports. The checklist describes how to read a report and does not replace a professional's assessment.
What you can do
Before buying a test, or before acting on a report you already have, go through the fifteen questions. Start with the first five: if the provider cannot tell you which variants it analyses and which studies it relies on, the other answers carry little weight. In the carbohydrate series (first part and second part) you can see a complete example of how a score is calculated, and in this post how an algorithm turns genes into a diet.
Sources
- How is genetic testing evaluated? A systematic review of the literature, Pitini E et al. European Journal of Human Genetics 26:605-615 (PMID 29422659), 2018
- A Scoping Review of Direct-to-Consumer Nutrigenetic Testing: Mapping Genes and Associated Nutrition Recommendations, McCartney C et al. Advances in Nutrition 17:100687 (PMID 42309301), 2026
- Nutrigenetics, omega-3 and plasma lipids/lipoproteins/apolipoproteins with evidence evaluation using the GRADE approach: a systematic review, Keathley J et al. BMJ Open 12:e054417 (PMID 35193914), 2022
- The need for polygenic score reporting standards in evidence-based practice: lipid genetics use case, Wand H, Knowles JW, Clarke SL. Current Opinion in Lipidology 32:89-95 (PMID 33538426), 2021
- Effect of Low-Fat vs Low-Carbohydrate Diet on 12-Month Weight Loss in Overweight Adults and the Association With Genotype Pattern or Insulin Secretion: The DIETFITS Randomized Clinical Trial, Gardner CD et al. JAMA 319:667-679 (PMID 29466592), 2018
- Clinical use of current polygenic risk scores may exacerbate health disparities, Martin AR et al. Nature Genetics 51:584-591 (PMID 30926966), 2019
- Real Decreto 942/2025, de 21 de octubre, por el que se regulan los productos sanitarios para diagnóstico in vitro (Spanish Royal Decree on in vitro diagnostic medical devices), Ministerio de Sanidad. BOE no. 255, 23 October 2025, 2025
Educational content from CESGEN, based on the evidence published up to the review date. It does not replace individual medical or dietary advice: to apply it to your own case, talk it through with your health professional. Nutrigenetics is an evolving field, and we review each piece when the evidence changes.