Study of 1.8 million older adults compares safety of GLP-1 drugs and other diabetes medicines

A study of 1.8 million adults aged 65 and older with type 2 diabetes found that GLP-1 receptor agonists and SGLT2 inhibitors were linked to significantly lower risks of hypoglycemia (low blood sugar) and hyperkalemia (high blood potassium) than sulfonylureas ^[1]. The same analysis linked GLP-1 receptor agonists to a lower risk of peripheral edema than DPP-4 inhibitors, and linked SGLT2 inhibitors to a higher risk of diabetic ketoacidosis than either GLP-1 receptor agonists or sulfonylureas ^[1][2].

What the study looked at

The analysis pooled nine databases: five US insurance claims databases and four electronic health record systems in the US, Germany and Spain ^[1]. It included 1,808,003 people aged 65 or older with type 2 diabetes who started a second-line antihyperglycemic drug. Of those, 984,590 (54.5%) were male, 1,211,729 (67.0%) were 65 to 74, and 596,274 (32.9%) were 75 or older ^[1].

Sulfonylureas made up most of the group: 1,075,919 people (59.5%). Another 485,016 (26.8%) started a DPP-4 inhibitor, 173,465 (9.6%) an SGLT2 inhibitor, and 73,603 (4.1%) a GLP-1 receptor agonist ^[1]. The paper cites global estimates that about 19.3% of adults over 65, or 135.6 million people, were living with diabetes in 2019, a figure projected to reach 276.2 million by 2045 ^[1].

Which drugs were involved

The most-prescribed GLP-1 receptor agonists in the cohort were dulaglutide (45.4%), semaglutide (36.1%) and exenatide (17.8%). The most-prescribed SGLT2 inhibitors were empagliflozin (58.6%), dapagliflozin (22.0%) and canagliflozin (18.9%). Sitagliptin accounted for 84.3% of DPP-4 inhibitor use, and glipizide (53.9%) and glimepiride (35.3%) for most sulfonylurea use ^[1].

Results were reported by drug class, not by individual product. Separate estimates for Ozempic, Wegovy or Rybelsus and for individual SGLT2 inhibitor brands have not been reported. The study covered treatment of type 2 diabetes; it did not examine use of these medicines for weight loss ^[1]. Results for tirzepatide, which is not a GLP-1 receptor agonist, were not reported.

How the study was designed, and its limits

Researchers compared eighteen safety outcomes using propensity score adjustment, empirical calibration and prespecified diagnostics, and estimated hazard ratios ^[1]. Median on-treatment follow-up was shorter for the newer classes: 55 to 202 days for GLP-1 receptor agonists and 59 to 246 days for SGLT2 inhibitors, compared with 86 to 719 days for DPP-4 inhibitors and 84 to 741 days for sulfonylureas ^[1].

One planned comparison did not survive the study's quality checks. The all-cause mortality analysis between GLP-1 receptor agonists and sulfonylureas was not presented because no valid result remained after the diagnostic process ^[1]. The published abstract also does not give the size of the risk differences, only that they were significant; the precise estimates are in supplementary tables that are not included in the abstract ^[1].

Several authors disclosed financial ties outside the study, including research support through Yale University from Novo Nordisk, employment at Johnson & Johnson and IQVIA, and consulting or patent interests ^[2].

Why it matters for patients

The paper notes that older adults are especially vulnerable to adverse drug events because of multimorbidity, frailty and polypharmacy, and that they are underrepresented in randomized trials, leaving clinicians with limited guidance when starting a second-line diabetes drug ^[1]. The authors write that the results "support informed, safety-conscious prescribing in the older adults underrepresented in clinical trials yet highly susceptible to adverse events" ^[1].

For someone 65 or older weighing these classes with a clinician, the study adds real-world comparisons on two safety problems — hypoglycemia and hyperkalemia — where GLP-1 receptor agonists and SGLT2 inhibitors came out ahead of sulfonylureas ^[1]. It also adds a caution about diabetic ketoacidosis with SGLT2 inhibitors relative to the other two classes ^[1].

What the study does not do: it reports average risks across large groups, not individual risk, and it reports by class rather than brand, so it does not show how semaglutide compares with dulaglutide, or how one SGLT2 inhibitor compares with another. It does not cover weight-loss use of these medicines, and the mortality comparison between GLP-1 receptor agonists and sulfonylureas has not been reported ^[1].