Author's Reply: Methodological Considerations to Enhance the Relevance of Causal Observational Research in the Field of Kidney Disease

We express our gratitude to Creon et al.1 for their interest in our recent JASN article on statin use in the patients with kidney failure.2 Their correspondence highlights that the emphasis on incident statin users limits its relevance, as most patients with kidney failure already use statins or have cardiovascular disease. As noted by Creon et al., only 17% of them initiated statins in our study. The low proportion of statin initiators in the electronic health record reflects a paucity of robust clinical evidence to recommend initiating statin in kidney failure for primary prevention. Despite a strong recommendation to initiate statin or statin/ezetimibe combination in GFR <60 ml/min per 1.73 m2 by Kidney Disease Improving Global Outcomes,3 some previous evidence suggested that the effects of LDL lowering on cardiovascular disease prevention might be primarily driven by the patients with GFR ≥30 ml/min per 1.73 m2.4 Our study supplements the existing data and demonstrates the benefit of statin beyond moderate CKD. We agree that polypharmacy is prevalent across the spectrum of kidney disease due to the high prevalence of comorbid conditions, and there is a lack of guidelines targeted toward deprescribing with minimal available evidence.5 Future investigation on the effect of statin discontinuation is warranted. Regarding the dialysis status, as noted in the limitations section, we concur with Creon et al. that the absence of dialysis status information would impede the examination of potential benefits of statin initiation across various dialysis statuses. With respect to transplant status, the incidence of kidney transplantation in Hong Kong is 7.3 per million population (56 donors) in 2022,6 which represents the lowest rate globally. In addition, concomitant use of glucocorticoid and calcineurin inhibitors among kidney transplant recipients would be important confounders. Consequently, transplant status was not included in our original manuscript. The exclusion of the first-year incidence is based on the consideration that some patients with undiagnosed cardiovascular diseases at the start of the “trials” were more likely to receive treatment due to some unrecorded reasons.7 The length of this period is chosen based on the duration of asymptomatic period for the diseases, with a sensitivity analysis shortening this period to 6 months in our manuscript. As suggested by Creon et al., we recalculated the absolute risk after including these cases and removing the criterion of another follow-up visit after baseline. The new estimates of 5- and 10-year standardized absolute cardiovascular disease risk difference were −5.8% (95% confidence interval [CI], −10.3% to −1.4%) and −11.3% (95% CI, −19.6% to −2.9%) in the per-protocol analysis, which also supports our original research findings. Regarding the adjustment for competing risk, we acknowledged that our adopted approach can only obtain estimates in a hypothetical scenario where no individual died. This limitation is inherent to this method. We additionally calculated 5- and 10-year absolute risk reduction for the composite outcome of overall cardiovascular disease and all-cause mortality, which were −9.1% (95% CI, −12.8% to −5.3%) and −10.4% (95% CI, −19.5% to −1.4%) in the per-protocol analysis, respectively. The estimates of this composite outcome also imply substantial overall benefits of statin use among patients with kidney failure.