The LIfestyle for BRAin Health (LIBRA) score, consisting of twelve factors, highlights individuals’ potential for dementia risk reduction through lifestyle. The LIBRA score includes modifiable protective factors such as low to moderate alcohol consumption, and risk factors such as hypertension.

We studied whether LIBRA scores are longitudinally associated with cognition, and to what extent this is due to between-person differences or within-person changes in LIBRA scores.

Individuals were included from four Dutch community-based cohorts: Doetinchem Cohort Study (DCS; n = 4770), Maastricht Aging Study (MAAS; n = 1295), Longitudinal Aging Study Amsterdam (LASA; n = 2391) and the Rotterdam Study (RS; n = 5205). The number of available LIBRA components (range 7–11) and timepoints (range 3–9) differed per cohort. Outcomes were standardized processing speed (LDST), memory (15-word delayed recall of the verbal learning test (VLT)) and verbal fluency. Hybrid mixed models were fit for the association of 1) mean LIBRA score and 2) change in LIBRA between subsequent timepoints. Models were adjusted for age, sex, education and learning effects. Interactions of the mean LIBRA score with age, and change in LIBRA score with age were tested in two separate models.

Higher (i.e., unhealthier) mean LIBRA scores were associated with worse cognitive speed (lower LDST z-score per 1-point higher LIBRA, range between cohorts: 0.039 – 0.0587), memory (VLT, 0.026 – 0.035), and fluency (0.020 – 0.033). Associations of mean LIBRA scores with cognitive function were stronger with older age (LDST: significant age-interaction, 2 out of 4 cohorts; VLT and fluency: 1 out of 4 cohorts). Relative to 65-year-old individuals with a mean LIBRA score at the 50th percentile, individuals at the 90th percentile of the LIBRA score showed an estimated 1.9–3.2 years more advanced cognitive ageing for LDST, 1.9 – 5.3 years for VLT and 1.4 – 1.7 years for fluency. Within-person change in LIBRA showed no consistent associations with cognitive decline.

An individual's mean LIBRA score, but not their change in LIBRA score over time, was longitudinally associated with cognitive functioning. In the general population, the investigated version of the LIBRA score is possibly not suitable to capture how cognition (as a proxy for dementia risk) changes with improvements in lifestyle.