Background A current phenomenon is the constant increase in life expectancy with polarised consequences for health. Among these, dementia represents a major public health impact on individuals and governments around the world. The search for modifiable risk factors remains a high priority. Many longitudinal studies have shown that education is associated with cognitive functioning in midlife, but the existing evidence regarding its protection against faster cognitive decline is relatively mixed. The aim of these analyses was to examine cognitive decline on both fluid and crystallised functions in preclinical stages of dementia and the role of education on these trajectories.

Methods The participants were drawn from the Swedish Octogenarian Twin study, a sample of individuals, aged 80+ years at baseline, who were interviewed biannually up to a maximum of 5 times. Only those who developed dementia after the study entry were included in these analyses (n=205). Cognitive functioning was assessed at each wave with the Mini-Mental State Examination and other neuropsychological tests such as verbal memory, inductive reasoning, language, information task and verbal meaning. Random effects models were fitted to each cognitive outcome as a function of time to dementia diagnosis, accounting for education, age and socio-economic position. The multilevel model was characterised by a fixed part which contained average effects for the intercept (initial status) and slope (rate of change from the baseline to the time to dementia diagnosis) and a random part which contained individual differences (variance) in the intercept, slope, and the within person residual. Sensitivity analysis were conducted (Tobit mixed models) to account for the ceiling/floor effects in the measures of global function.

Results The results indicate a steeper cognitive decline prior to dementia diagnosis in most of the fluid cognitive abilities measured with MMSE, Prose Recall, Block Design, Digit Symbol, Memory Recognition, Memory Correspondence and Information tests. Education had a significant effect on the level of cognitive functioning at the time of dementia diagnosis on both crystallised cognitive measures: synonym (β=0.80, SE=0.35, p = 0.02) and information tests (β=1.72, SE=0.54, p = 0.001), but not on the rate of cognitive decline in any of the cognitive measures investigated.

Conclusion These findings suggest that higher education is associated with better cognitive functioning on crystallised cognitive measures (e.g. information and verbal meaning) which tend to be preserved as we age; but did not support the cognitive reserve hypothesis, which stipulates a protective role of education against steeper cognitive decline before dementia diagnosis.