Key Takeaways
Extended Takeaways
- The study emulated 5 separate target trials, comparing semaglutide/tirzepatide against metformin, insulin, sulfonylureas, DPP4i’s, and SGLT2i’s over 12 months, which is clinically useful because overdose associations were not uniform across diabetes drug classes.
- Patients with ≤30-day cumulative prescription coverage or <2 fills were excluded, so these findings apply most directly to Veterans who maintained at least minimal early treatment persistence rather than to patients who stopped after an initial prescription.
- Propensity matching substantially improved baseline balance across cohorts, with median standardized mean differences falling from 0.12, 0.11, 0.09, 0.10, and 0.09 to 0.03, 0.05, 0.06, 0.03, and 0.03 in the insulin, SGLT2i, DPP4i, metformin, and sulfonylurea comparisons, respectively.
- Because event counts were small, the outcome was broadened to all-cause overdose and analyzed with unadjusted Cox models after matching; clinicians should interpret this as a signal for overdose risk reduction overall, not proof of a specifically opioid-overdose effect.
- The sensitivity analysis strengthens the insulin and SGLT2i findings: an unmeasured confounder would need hazard ratio associations of 5.70 and 7.46, respectively, with both treatment assignment and overdose to fully explain away those observed associations.
- Comparator choice may matter clinically when interpreting null findings, because the authors note that metformin and DPP4i’s may themselves have antiaddiction properties, which could reduce the apparent advantage of semaglutide/tirzepatide in head-to-head observational comparisons.
