Mediterranean-diet scores were almost identical in 177 women with and without endometriosis. One spoonful of the questionnaire stood apart

Diet advice around endometriosis can become restrictive quickly. A study published in Nutrients on 31 October 2026 tested a quieter question: could a short Mediterranean-diet questionnaire distinguish women with diagnosed endometriosis from women without it, once their ages were reasonably comparable? Researchers surveyed 177 women in Spain — 77 members of an endometriosis patient association and 100 controls recruited from the community.

On the 14-point PREDIMED diet screen, the averages were almost the same: 8.6 in the endometriosis group and 8.5 in the control group. About 70% of each group reached the study's threshold for adequate adherence. Physical activity did not clearly differ after age adjustment either. The study therefore did not find a Mediterranean-diet pattern peculiar to endometriosis.

The one food question that held up after correcting for 13 comparisons was sofrito, the slow-cooked mixture of vegetables and olive oil used as a base for many Spanish dishes. Women with endometriosis reported eating it less often. But this was one answer in a cross-sectional survey, not an experiment in which anybody was assigned sofrito, and the paper does not claim that the dish prevents or treats the condition.

How the groups entered the study makes that lone difference especially hard to interpret. The cases came through a patient association and the controls through community recruitment. Women already living with a diagnosis may have changed what they eat, encountered dietary advice or reported their food differently. Geography and culture also differed between the groups. Diagnoses and disease stage were reported by participants rather than checked against medical records.

A machine-learning exercise produced an impressive-looking test-set AUC of 0.90 to 0.93 when symptoms and menstrual quality of life were included. When the researchers limited the model to earlier-life and dietary variables, performance fell to 0.67 in cross-validation and 0.73 in the held-out test set. In other words, the model mostly recognised the burden of an established condition; it was not an early-detection or risk tool.

Daytime sleepiness also looked a little higher before adjustment — an average Epworth score of 10.4 versus 8.8 — but the difference was no longer statistically clear after accounting for age. The useful result here is the null one: a brief Mediterranean-diet score and broad activity categories did not separate the groups. That offers no support for diagnosing endometriosis by diet, or for cutting out foods in hopes of treating it. Diet may still matter to an individual's overall health and symptoms, but that requires intervention studies designed to test it.