Abstract
Introduction
Sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1 RA) are increasingly combined in type 2 diabetes mellitus (T2DM). SGLT2i cause glycosuria and raise genitourinary tract infection risk, driven mainly by genital infection; GLP-1 RA are non-glycosuric and improve glycemic control and weight. While single-agent comparisons are well established, the literature lacks direct combination-therapy comparisons, and whether adding a second agent, and the order in which it is added, raises or lowers this risk remains debatable. We tested whether genitourinary tract infection risk tracks the glycosuric drug by examining both directions of combination therapy.
Methods
This retrospective, new-user, active-comparator cohort study used the TriNetX US Collaborative Network. Adults aged ≥18 years with T2DM receiving metformin were included; those with type 1 diabetes, end-stage renal disease, or tirzepatide use were excluded. Two bidirectional contrasts were constructed: adding a GLP-1 RA to established SGLT2i therapy versus continued SGLT2i alone (Contrast 1), and adding an SGLT2i to established GLP-1 RA therapy versus continued GLP-1 RA alone (Contrast 2). Combination arms were indexed at initiation of the second agent, with ≥3 months of prior first-agent therapy required to align index timing; comparators lacked the second agent within this interval. Each contrast underwent separate 1:1 propensity score matching on demographics; renal, genitourinary, and cardiometabolic diagnoses; glucose-lowering and other medications; hemoglobin A1c; estimated glomerular filtration rate; and body mass index. After matching, standardized mean differences were below 0.1 for all covariates. Genital and urinary tract infections, defined separately by International Classification of Diseases, Tenth Revision, Clinical Modification codes, were assessed as incident events over 90 days. Conjunctivitis and low back pain served as negative controls. Risk ratios (RR) with 95% confidence intervals (CI) are reported.
Results
After matching, Contrast 1 included 85,649 and Contrast 2 included 80,740 patients per arm. Adding a GLP-1 RA to an SGLT2i was associated with lower genital infection risk (0.4% versus 0.8%; RR 0.52, 95% CI 0.46 to 0.60). Conversely, adding an SGLT2i to a GLP-1 RA was associated with higher genital infection risk (1.0% versus 0.4%; RR 2.62, 95% CI 2.28 to 3.01). The opposite directions indicate that risk tracked the glycosuric agent rather than treatment escalation itself. Urinary infection was minimally affected, with a modest reduction in one direction (Contrast 1 RR 0.84, 95% CI 0.75 to 0.94; Contrast 2 RR 1.04, 95% CI 0.93 to 1.15). Negative control outcomes were near null (RR 0.87 to 0.97), arguing against surveillance bias.
Conclusion
Genitourinary tract infection risk during combination SGLT2i and GLP-1 RA therapy was determined by the glycosuric agent. Adding a GLP-1 RA to an SGLT2i roughly halved genital infection risk, whereas adding an SGLT2i to a GLP-1 RA more than doubled it, with urinary infection unaffected throughout. These findings suggest GLP-1 RA co-therapy does not compound, and may reduce, the genital infection risk of SGLT2i, whereas initiating an SGLT2i in a GLP-1 RA-treated patient markedly increases it. Confounding by indication remains a limitation.