In 1,298 Swedish 70-year-olds followed up to 20 years, a higher Mediterranean Diet Score (eight food groups, alcohol excluded) was linked to lower dementia morbidity in men (adjusted HR 0.83, 95% CI 0.71-0.96) but not in women. The signal did not survive multiple-testing correction, and the sex-by-diet interaction (p=0.048) was only nominal. Cardiovascular disease, cancer and all-cause mortality showed no significant associations; the 10-year mortality link in men (HR 0.90) lost significance after full adjustment (HR 0.92, 0.84-1.01).
Treat this as a cautionary null rather than a sex-difference discovery. A nominal p-value of 0.048 for interaction, plus a result that fails correction, is the classic profile of a chance finding in subgroup analysis. The cohort is also modest for dementia and cancer endpoints, and the authors concede that registers miss over half of dementia cases, especially among people living at home, which biases outcomes toward later, institutionalised disease. Diet was measured once, at 70, so it cannot capture midlife eating, when dietary influence on dementia is plausibly strongest. Larger cohorts and trials have generally shown modest or inconsistent Mediterranean diet benefits for dementia, with stronger evidence for cardiovascular outcomes, though those typically involve younger or higher-risk participants. The sensible reading: starting or sustaining a Mediterranean-style pattern remains reasonable for general health, but this study gives no grounds to expect a dementia payoff specific to men, or a mortality payoff when begun in the eighth decade. Confirmatory in its nulls, incremental overall.