Revisiting the link between sodium-glucose cotransporter-2 inhibitors and urinary tract infections

This commentary refers to ‘SGLT2 inhibitors and urinary tract infections: resolving the dilemma through aetiological classification’, by Y. Doi et al., https://doi.org/10.1093/eurheartj/ehag266 and the discussion piece ‘Urinary tract infection and continuation of sodium-glucose cotransporter-2 inhibitors in diabetic patients’, by M.-z. Wu et al., https://doi.org/10.1093/eurheartj/ehaf788. We are grateful to Dr Doi and colleagues for their interest in our paper1 and are pleased to respond to the points raised in the discussion article. A summary hazard ratio can be challenging to interpret when the proportional hazards assumption is violated, as it represents an averaged estimate that may obscure meaningful time-varying treatment effects. To formally address potential non-proportionality, we incorporated a time-dependent covariate into the Cox regression model. For clearer visualization of the outcome dynamics, we also plotted cumulative incidence curves for recurrent urinary tract infection (UTI), comparing patients who discontinued vs continued sodium-glucose cotransporter-2 (SGLT2) inhibitors therapy. As shown in Figure 1, the cumulative incidence curves for both groups are nearly superimposable and exhibit parallel trajectories throughout the 1-year follow-up. The parallel curves further underscore the lack of time-varying effects, emphasizing the robustness of our conclusion that discontinuation does not reduce recurrent UTI risk.