Nearly 122,000 participants were included in the analysis, with a median follow-up of 9.4 years. At least two (and as many as 5) 24-hour validated food frequency questionnaires were used to assess dietary intake, and to calculate a “flavodiet score,” a composite total of the top food contributors to each flavonoid subclass. Genetic risk for dementia was considered high based upon either the presence of an APOE ε4 allele or being in the highest quintile of risk, (based upon a polygenic risk score defined elsewhere).
The median intake of flavonoids was 4.3 servings per day, 2.7 of which were from tea. People with the highest flavodiet scores also tended to exercise more, have a lower BMI, and have less socioeconomic deprivation. After multivariate adjustment for these factors, those in the highest quintile of intake (7.3 servings) had a 28% lower risk for dementia than those in the lowest quintile (1.4 servings). Tea, red wine, and berries appeared to account for the majority of the association with the flavodiet score; when these foods were removed from the analysis, it was no longer significant. The dementia risk reduction was even greater in those at high genetic risk (43% lower with high flavonoid intake), with depressive symptoms (48% lower), or hypertension (30% lower). Indeed, in the absence of these factors, no protective effect was observed. The most protective subclasses of flavonoids included anthocyanins, flavan-3-ols, flavonols, and flavones.