Among 58,754 adults aged 65 or older with type 2 diabetes treated at a regional US healthcare system between 2018 and 2026, GLP-1 receptor agonist initiators (n = 5,095) were meaningfully younger (mean age 75.2 vs 77.7–79.5 years), far more likely to be obese (mean BMI 33.8 vs ~29 kg/m²; 72% vs 42% with BMI ≥30), and carried substantially fewer comorbidities than peers starting SGLT2 inhibitors or DPP-4 inhibitors. Baseline HbA1c was virtually identical across groups (~7.4–7.7%), confirming that glycemic severity is not driving drug selection. LASSO regression identified BMI above 27 kg/m² as the dominant positive predictor of GLP-1RA initiation; heart failure and prior hospitalization were the strongest negative predictors.
These findings matter because GLP-1 drugs now carry compelling cardiovascular and renal outcome trial data — yet the patients most likely to benefit from those effects (frailer, older, with heart failure or CKD) are the least likely to receive them. This "healthy-user" prescribing pattern is a well-documented confound in pharmacoepidemiology and could exaggerate apparent benefits in future observational studies. For clinicians, the data highlight a potential gap: older, frail patients with T2D may be systematically under-treated with agents shown to reduce cardiovascular events and weight. Limitations include the single regional health system, retrospective design, and reliance on an LLM to assess cognitive status. As a preprint not yet peer-reviewed, these results should be interpreted cautiously and may change after formal review.