We obtained data from the PAQUID (Personnes Agées QUID) epidemiological study of cognitive and functional ageing (www.healthandage.net/html/min/paquid/entrance.htm
). During the third wave of the study (1991-2) investigators visited 1674 people aged 68 and over without dementia and living at home in 75 parishes in southwestern France and recorded their frequency of consumption of meat and fish or seafood: daily, at least once a week (but not every day), from time to time (but not every week), never. Participants were followed up two, five, and seven years afterwards: 1416 (84.6 %) had at least one follow up visit. All the participants who had lost three points or more on the mini-mental state examination since a previous visit or were suspected of having dementia according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders
, third edition, revised (DSM-III-R) were visited by a neurologist to confirm the diagnosis.
We calculated the incidence of dementia per 100 person years. We used a Cox proportional hazards model with delayed entry to estimate the relative risk of dementia, taking into account age, sex, and education (at least the French primary school diploma “Certificat d'Etudes Primaires” versus less education3
During the seven years of follow up 170 new cases of dementia occurred, including 135 cases of Alzheimer's disease. The table shows a significant trend between increasing consumption of fish or seafood and decreasing incidence of dementia (P for trend=0.0091). Frequency of fish or seafood consumption was higher in the participants with higher education (879/1051 (83.6%) v 262/365 (71.8%) consuming fish at least weekly; P<0.0001). Participants who ate fish or seafood at least once a week had a significantly lower risk of being diagnosed as having dementia in the seven subsequent years (age and sex adjusted hazard ratio 0.66, 95% confidence interval 0.47 to 0.93). When we added education into the model the hazard ratio was almost unchanged (0.73) but the 95% confidence interval (0.52 to 1.03) slightly overlapped 1.00, indicating that the “protective” effect of weekly fish or seafood consumption was partly explained by higher education of regular consumers. Participants who ate fish or seafood at least once a week had a hazard ratio, adjusted for age and sex, of 0.69 for developing Alzheimer's disease in the seven following years, with borderline significance (95% confidence interval 0.47 to 1.01). We found no significant association between meat consumption and risk of dementia—P for trend=0.59; age and sex adjusted hazard ratio for weekly consumers 0.56 (0.26 to 1.20).