A population-based study of over 13,500 older diabetic patients found no statistically significant difference in dementia risk between SGLT2 inhibitors and the GLP-1 RA dulaglutide over a median follow-up of 4.4 years.
Risk ratio 0.81 (not significant)SGLT2 inhibitors showed a non-significant 19% lower dementia risk compared to dulaglutide over 4.4 years. The confidence interval (0.56-1.16) means the true effect could range from 44% lower to 16% higher risk, so no clear winner emerged.
What the researchers found
In a propensity score-matched cohort using South Korean national health data (2010-2022), 12,489 patients initiating SGLT2 inhibitors (dapagliflozin or empagliflozin) and 1,075 patients initiating dulaglutide were compared. Over a median follow-up of 4.4 years:
- Dementia events: 69 in the SGLT2 inhibitor group vs. 43 in the dulaglutide group
- Risk difference: -0.91 percentage points (95% CI: -2.45 to 0.63) — not statistically significant
- Risk ratio: 0.81 (95% CI: 0.56 to 1.16) — not statistically significant
The data were compatible with dementia risk being anywhere from 2.5 percentage points lower to 0.6 percentage points higher for SGLT2 inhibitors compared to dulaglutide.
Why it matters
Type 2 diabetes approximately doubles the risk of dementia, making neuroprotection a critical consideration when choosing diabetes medications. Both SGLT2 inhibitors and GLP-1 RAs have shown brain-protective signals in preclinical studies, but real-world comparative data have been scarce. This study — published in the prestigious Annals of Internal Medicine — provides some of the first head-to-head evidence comparing these two drug classes on dementia outcomes.
How the study worked
Target trial emulation study using nationwide South Korean health care data from the National Health Insurance Service (2010-2022). Patients aged 60+ with type 2 diabetes initiating SGLT2 inhibitors or dulaglutide were included. Propensity score matching (1:2 ratio) was used to adjust for confounders. The primary outcome was presumed clinical onset of dementia, defined as one year before diagnosis. Five-year risk ratios and risk differences were estimated using Cox models.
What this study cannot tell us
The study has several important limitations: residual confounding is possible since HbA1c levels and diabetes duration were not adjusted for. The dulaglutide group was much smaller than the SGLT2 inhibitor group (1,075 vs 12,489), limiting statistical power. Newer GLP-1 RAs like semaglutide were not included. Dementia onset was estimated (one year before diagnosis) rather than directly measured. The South Korean population may not be representative of other ethnic groups. The study was observational, not a randomized trial.
How to read the evidence
This is a well-designed population-based cohort study using target trial emulation methodology and propensity score matching, published in Annals of Internal Medicine. However, it is observational, has an unbalanced group size, and lacks adjustment for key confounders like HbA1c. It provides moderate-quality evidence.
When this study was published
Published in 2024, this study is very recent and addresses a cutting-edge clinical question. However, its data (2010-2022) predate the widespread use of newer GLP-1 RAs like high-dose semaglutide, which may have different neuroprotective properties.
The bigger picture
As the diabetes population ages and dementia rates climb, finding medications that can treat diabetes while also protecting the brain is a high priority. This study suggests both SGLT2 inhibitors and GLP-1 RAs may have comparable neuroprotective profiles. However, the field is evolving rapidly — newer, more potent GLP-1 RAs like semaglutide were not well represented in this study and may have different neuroprotective effects. Large randomized trials specifically testing dementia prevention with these drug classes are urgently needed.
Questions still open
- Would newer, more potent GLP-1 RAs like semaglutide show a stronger neuroprotective effect against dementia compared to SGLT2 inhibitors?
- Is a longer follow-up period needed to detect meaningful differences in dementia risk between these drug classes?
- Should dementia prevention be considered when choosing between SGLT2 inhibitors and GLP-1 RAs for older diabetic patients?
Common questions
Can diabetes medications protect against dementia?
Does this study apply to newer GLP-1 medications like semaglutide (Ozempic)?
Read the original research
Sodium-Glucose Cotransporter-2 Inhibitors, Dulaglutide, and Risk for Dementia : A Population-Based Cohort Study.
Annals of internal medicine, 177(10), 1319-1329
Citation
Hong, Bin; Bea, Sungho; Ko, Hwa Yeon; Kim, Woo Jung; Cho, Young Min; Shin, Ju-Young. (2024). Sodium-Glucose Cotransporter-2 Inhibitors, Dulaglutide, and Risk for Dementia : A Population-Based Cohort Study.. Annals of internal medicine, 177(10), 1319-1329. https://doi.org/10.7326/M23-3220