Individual Variation in Diet Responses

Variation in a single meal

In the PREDICT 1 study, 1,002 twins and unrelated healthy adults in the United Kingdom ate identical meals. The population coefficient of variation, a measure of spread, was 103% for blood triglyceride, 68% for glucose and 59% for insulin. [1] In an earlier study of 800 people, researchers recorded responses to 46,898 meals and also found high variability in the response to identical meals. [2]

Evidence level: these are human observational and measurement studies. They show that short-term responses differ. They do not show that the differences matter for long-term health. See Glucose Food Responses and Individual Variation.

What explains the differences

In PREDICT 1, person-specific factors such as the gut microbiome explained more of the variation in blood fat after a meal than the meal's macronutrients did (7.1% against 3.6% of variance), but not for blood sugar, where meal content explained 15.4% and person-specific factors 6.0%. [1] Genetic variants had a modest effect on predictions: 9.5% for glucose, 0.8% for triglyceride and 0.2% for C-peptide. [1] Several of the authors are consultants to or employees of Zoe Global Ltd, a company. [1]

Responders and non-responders

In research, a "responder" is simply someone whose measure changed most after an intervention. The groups depend on the cutoff chosen. In the Food4Me study, people were classed by the change in total cholesterol from baseline to month 6, with the lower and upper quartiles called responders and non-responders. [3] The groups did not differ in demographic or body measures. [3] They did differ in baseline fatty acid profiles; for example the responder group had higher stearic acid. [3]

In the same trial, people who gained more from personalised advice were more often older, women, and those with lower baseline Healthy Eating Index scores. [4] Characteristics of those who benefited did not differ by other demographic, health-related, body or genotype characteristics. [4]

These are exploratory comparisons. They suggest which people may change most, not what any one person should do.

Variation does not always mean a better diet exists for you

The DIETFITS trial randomized 609 adults to a healthy low-fat or low-carbohydrate diet. Weight change at 12 months was -5.3 kg and -6.0 kg, and neither genotype pattern nor insulin secretion identified which diet worked better for whom. [5] So people can vary widely, yet the factors tested in that trial did not sort them into better-suited diets.

How to read variation claims

• A large spread between people is not the same as a stable personal trait. See Misinterpreting Normal Variation in Nutrition Data.
• A short-term marker is not a health outcome. See Biomarker Changes vs Meaningful Health Outcomes.
• Check whether the research was funded or run by the company selling the test. [1]