Cardiovascular effectiveness and safety of SGLT2 inhibitors vs DPP4 inhibitors by dementia status: a cohort study of older adults with diabetes

Aims/hypothesis — People with dementia are at a higher risk of inappropriate medication use; however, limited data inform clinical outcomes of sodium–glucose cotransporter 2 inhibitor (SGLT2i) use in people with diabetes and comorbid dementia. We aim to investigate cardiovascular effectiveness and safety of SGLT2i vs dipeptidyl peptidase-4 inhibitors (DPP4i) in older community-dwelling adults with diabetes by dementia status.

Methods — This population-based, target trial emulation cohort study used linkable administrative datasets of residents of Ontario, Canada. Eligible initiators of SGLT2i or DPP4i with diabetes aged ≥66 years (2016–2022) were stratified by dementia status and matched 1:1 on propensity score. The primary cardiovascular effectiveness outcome was composite of all-cause mortality or hospitalisation for ischaemic stroke, myocardial infarction or heart failure. We also investigated eight safety outcomes potentially related to SGLT2i. Incidence rate differences (IRDs) per 1000 person-years and the 95% CIs were estimated. Homogeneity in IRDs between strata was assessed using the Cochran’s Q statistic. One year number needed to treat (NNT) or harm (NNH) was also estimated using the Aalen-Johansen estimator with death as a competing risk.

Results — We analysed 2481 pairs with dementia and 52,196 pairs without dementia. The absolute reduction in the composite effectiveness endpoint with SGLT2i vs DPP4i initiation was greater among those with dementia (IRD [95% CI] −61.1 [−78.2, −43.9]; NNT: 20), compared to no dementia (IRD −21.7 [−23.7, −19.7]; NNT: 43; homogeneity: p<0.001). Across the individual effectiveness endpoints, SGLT2i vs DPP4i initiation in people with dementia was associated with reduced all-cause mortality (IRD −51.2 [−65.7, −36.7]; NNT: 25) and hospitalisation for heart failure (IRD −16.4 [−24.6, −8.2]; NNT: 99). For safety outcomes, SGLT2i vs DPP4i initiators with dementia showed less acute kidney injury (IRD −39.7 [−55.0, −24.4]; NNT: 76) and increased genital infection (IRD 9.7 [3.7, 15.6]; NNH: 64). The absolute increase in diabetic ketoacidosis (IRD 1.1 [0.7, 1.6] vs 5.9 [2.1, 9.8]; NNH: 785 vs 109; homogeneity: p=0.015) with SGLT2i was greater among those with dementia vs no dementia.

Conclusions/interpretation — While SGLT2i might provide cardiorenal protection among those with dementia, closer monitoring may be warranted due to greater susceptibility to diabetic ketoacidosis.