postdoc University of North Carolina at Chapel Hill, Department of Epidemiology Chapel Hill, United States
Background: Numerous observational studies suggest that metformin is associated with a reduced risk of dementia and metformin’s effect on preventing Alzheimer’s disease is being evaluated by the MAP (Metformin in Alzheimer’s Dementia Prevention) clinical trial.
Objectives: To examine whether metformin decrease risk of Alzheimer’s disease and related dementia compared with other antidiabetics in real-world data.
Methods: We implemented an active comparator, new user cohort design identifying initiators of metformin and dipeptidyl peptidase 4 inhibitors (DPP4i), using a US nationwide 20% random sample of fee-for-service Medicare beneficiaries aged 70+ with parts A, B, and D coverage from 2007-2019. We required patients to have 2nd prescription of the same drug class and continuous enrollment of parts A, B and be free of dementia in the 3-years prior to drug initiation. The dementia outcome was defined as 1) 1 dementia claim in 1 year AND 2) 1 dementia claim in another year or nursing home stay 6 months in the 3-year follow-up period. A total of 80 covariates were measured in the 1-year baseline period, including demographics, neurological conditions, diabetes comorbidities, cardiovascular comorbidities, other comorbidities, neurological disorder medications and other drug classes, durable medical equipment claims, and measures of healthcare utilization. We estimated propensity scores (PS) to balance measured confounders across cohorts, and applied asymmetric PS trimming using a cut point corresponding to the 5th and 95th percentiles of the PS distribution in the treated and untreated patients, respectively. We estimated adjusted risk difference (RD) and 95% CI using inverse probability of treatment weights.
Results: The initiators of metformin (n=34,886) and DPP4i (n=8,009) were followed for a median (IQR) duration of 3.0 (1.8 to 3.0) and 2.8 (1.4 to 3.0) years, respectively. During 3-year follow-up, 7,077 and 1,838 met the dementia definition in metformin and DPP4i, and the incident rate was 85 and 103 per 1000 person-years, respectively. Adjusted RD (%) was 0.3 (95%CI 1.3 to 0.7) comparing metformin to DPP4i. Results were consistent across sensitivity analyses with varying exclusion criteria, baseline period, and outcome definitions.
Conclusions: Our active comparator, new user cohort study does not suggest metformin leading to a clinically relevant decrease in risk for dementia compared with DPP4i. This study is limited by the short duration of treatments and outcome misclassification of dementia using claims data.
