We thank Li et al. 1 for their thoughtful comments regarding our study evaluating the oncological impact of prophylactic urethrectomy (PU) at the time of radical cystectomy (RC) 2. We welcome the opportunity to clarify key methodological aspects and to further contextualise our findings. First, the authors raise the important point that an observed overall survival (OS) benefit without a parallel cancer-specific survival (CSS) advantage raises concern for residual confounding and/or competing risks. We acknowledged this limitation and did not interpret the OS association as definitive evidence of a direct oncological effect. Rather, we reported an association that persisted after adjustment for measured confounders and emphasised that this finding should be regarded as hypothesis-generating. In surgical oncology, OS is sensitive to baseline health status and non-cancer mortality, and discordance between OS and CSS has been described in retrospective comparative effectiveness studies. Second, regarding the propensity score matching (PSM) strategy, variable selection was restricted to factors plausibly influencing both treatment allocation (PU vs non-PU) and tumour-driven outcomes at the time of RC, including age, tumour stage, nodal status, prostatic urethral involvement, carcinoma in situ, smoking status, and neoadjuvant chemotherapy. Methodological guidance cautions against incorporating downstream consequences of treatment selection or proxies of global frailty, such as American Society of Anesthesiologists (ASA) score or Charlson Comorbidity Index (CCI), as this may induce over-adjustment or collider bias, obscuring true associations rather than reducing bias 3, 4. In this context, the ASA score and CCI primarily reflect competing mortality risk rather than indication for PU, and their inclusion risked degrading balance on tumour-related covariates central to the study question. Importantly, comorbidity burden was not ignored: the CCI was incorporated into multivariable Cox regression for OS, where it emerged as an independent predictor alongside age, nodal status, margin status, and PU. This allowed us to account for frailty in outcome modelling while avoiding inappropriate adjustment in treatment assignment modelling—an approach aligned with contemporary recommendations for observational survival analyses 3. Third, the ‘healthy user’ or treatment-selection effect highlighted by Li et al. 1 is an inherent limitation of retrospective surgical cohorts and cannot be fully eliminated by PSM, which balances only measured covariates. We addressed this by presenting matched and unmatched analyses, performing multivariable modelling after matching, and cautioning that the OS finding should be interpreted as hypothesis-generating in the absence of concordant improvements in recurrence-free survival (RFS) or CSS. Despite adjustment for comorbidity burden, PU remained independently associated with OS, suggesting the association is not solely attributable to frailty imbalance, although residual confounding cannot be excluded. Finally, the lack of a significant difference in urethral RFS after matching reinforces current European Association of Urology (EAU) guideline recommendations and supports our principal conclusion that routine PU should not be universally advocated but rather considered selectively in high-risk patients. In summary, our study does not claim a definitive survival benefit of PU or advocate routine application. Instead, it highlights the complexity of survival endpoint interpretation after RC, methodological challenges inherent to observational surgical research, and need for individualised clinical decision-making. We appreciate the authors’ engagement, which contributes meaningfully to this discussion. The authors have no conflicts of interest.
Akand et al. (Tue,) studied this question.