Key Takeaways
Extended Takeaways
- The kidney replacement therapy signal was reproduced in propensity-matched analyses: after matching (n=3,908 per group), 5-year event-free survival was 87.93% with GLP-1 RAs versus 84.63% with DPP-4 inhibitors (absolute risk reduction 3.30%; χ21=15.40; P <.001).
- Against a chronic kidney disease standard-of-care comparator, GLP-1 RAs also showed higher 5-year kidney replacement therapy-free survival than RAAS inhibitors (94.68% vs 89.77%; log-rank χ21=17.66; P <.001), corresponding to a 4.91 percentage point absolute reduction and number needed to treat of 21.
- Health care utilization findings were less robust than the kidney outcome: although the primary Cox model favored GLP-1 RAs versus DPP-4 inhibitors (aHR 0.74, 95% CI 0.71–0.78; P <.001), propensity-matched Kaplan–Meier analyses found no significant difference versus DPP-4 inhibitors (58.06% vs 60.64%; χ21=0.018; P =.893) or RAAS inhibitors (60.62% vs. 59.24%; log-rank χ21=2.09; P=.148).
- Individual-agent sensitivity analyses suggest the association was not driven by one GLP-1 receptor agonist alone: semaglutide was associated with lower observed risk of kidney replacement therapy (aHR 0.56, 95% CI 0.47–0.67; P<.001) and tirzepatide showed a similar association (aHR 0.40, 95% CI 0.25–0.62; P<.001).
- The negative control outcomes were null, which supports—but does not prove—reduced residual confounding in this observational design: acute appendicitis showed aHR 1.09, 95% CI 0.49–2.42; P=.827 and benign skin neoplasm showed aHR 1.20, 95% CI 0.85–1.68; P=.305.
- For psychiatrists managing lithium or antipsychotic treatment in bipolar disorder with chronic kidney disease, the observed kidney replacement therapy association persisted across psychotropic regimens, including lithium (aHR 0.41, 95% CI 0.24–0.71; Q=.001) and antipsychotics (aHR 0.53, 95% CI 0.47–0.61; Q=.001).
