(A-153) The association between metformin-based regimens and cognitive performance in participants with type2 diabetes in the UK Biobank

Background: Metformin may help delay cognitive decline, but findings remain inconsistent; the effects of different doses of metformin have been rarely investigated.

Objectives: To investigate the association between varying dosage levels of metformin and cognitive performance in people with type 2 diabetes (T2DM).

Methods: We conducted a retrospective cohort study using UK Biobank participants aged ≥40 years (2006–2010) with T2DM, no prior dementia, valid Fluid Intelligence (FI) scores (0–13) at baseline, and linked primary care data. Participants were followed up to 2015; a subset had more than one FI test. Metformin intensity was defined by prescriptions in two years before each FI assessment and categorised into 6 daily dose groups: none, ≤500mg, 500–1000mg, 1000–1500mg, 1500–2000mg, and >2000mg. Other antidiabetic use was defined as having ≥3 prescriptions in two years, including sulfonylurea and insulin (binary: yes/no), and a count of remaining drug classes (0–6). Confounders included age, sex, ethnicity, education, socioeconomic deprivation, smoking, alcohol use, body mass index, and diabetes duration. We also considered comorbidities in a five-year look-back period (e.g., stroke, heart disease, depression, hearing loss, hypertension, hyperlipidaemia, brain injury, and microvascular complications), and concurrent use of antipsychotics and anticholinergics in two-year look-back period. A mixed-effects linear regression model with participant-level random intercepts was used to assess the association between metformin use and FI score.


Results: We included 4851 participants, of whom 1197 had more than one FI assessment. Metformin use was significantly associated with higher FI scores (i.e, better cognition): 1000–1500 mg (coefficient: 0.23, 95%CI: [0.04, 0.42]), 1500–2000 mg (0.21 [0.03, 0.39]), > 2000 mg (0.24 [0.04, 0.44]). No association was observed in low doses (≤500 mg & 500–1000 mg).
Greater education years (19–20 vs 7 years: 1.66 [1.51, 1.9]), former smoker (0.14 [0.03, 0.25]), current drinker (0.44 [0.22, 0.66]), and diabetic microvascular complications (0.73 [0.03, 1.43]) were also associated with better cognitive performance.
Conversely, sulfonylurea use was related to lower FI scores (−0.32 [−0.48, −0.16]), alongside older age (−0.01 [−0.02, -0.002]), stroke (−0.44 [−0.81, −0.07]), anticholinergic use (−0.31 [−0.61, −0.01]), non-white ethnicity (non-white vs white: −2.04 [−2.22, −1.86]), and higher deprivation (quintile 5 vs 1: −0.66 [−0.84, −0.48]).

Conclusions: Metformin daily dose more than 1000 mg was linked to better cognitive performance, with a similar effect size across dosage levels.