Dairy intake and cognitive function in older adults in three cohorts: a mendelian randomization study.

BACKGROUND

Meta-analyses of observational studies on the effect of dairy on cognitive function have yielded inconclusive results, potentially due to unmeasured confounding. To avoid the no-unmeasured confounding assumption, we used lactase persistence genetic variant as an instrumental variable, for which the CC genotype is associated with lower lactase production and, consequently, lower dairy consumption. We used it to assess the effect of long-term consumption of total and non-fermented dairy on cognitive function.

METHODS

We included 43,836 individuals over 55 years old with genotyping, dietary data, and cognitive function measurements from three population-based studies: CoLaus|PsyCoLaus (Switzerland), the Rotterdam Study (the Netherlands) and the Canadian Longitudinal Study on Aging (CLSA - Canada). We performed a one-sample Mendelian randomization using two-stage least-squares regression. First, we estimated total and non-fermented dairy consumption by T-allele frequency. Second, we used the estimated dairy consumption in linear regression models on general cognition, assessed by the Mini-Mental State Examination (MMSE) and the Mental Alternation Test, executive function, verbal fluency, verbal learning, and memory.

RESULTS

Per T-allele, total dairy intake and non-fermented was 24.8 and 15.3 g/day higher in PsyCoLaus, 57.9 and 49.8 g/day in the Rotterdam Study, and 0.31 and 0.29 times/day in CLSA, respectively. We found no association between the genetically predicted difference and the MMSE in PsyCoLaus and the Rotterdam Study. However, lactase persistent individuals scored 3.4 (95% CI 2.1- 4.7) and 3.5 (95% CI 2.3-4.7) points more in the Mental Alternation Test for total and fermented dairy, respectively, in CLSA. Similarly, lactase persistent participants in CLSA had higher verbal fluency, verbal learning and executive function, but no differences were found in the other cohorts. Such inconsistencies might stem from different FFQs across cohorts and consumption ranges. Nonetheless, the generally small magnitude of effect sizes may suggest that there is no real effect between total or non-fermented dairy intake and cognitive function.

CONCLUSION

The evidence for a causal effect of dairy consumption on general cognitive function is weak, consistent with previous results from classic analysis from observational studies. Interventions targeting dairy are unlikely to have a relevant effect on cognitive function.