Concerns about Hospitalization on Cystatin C– and Creatinine-Based eGFR

We read with great interest the JASN article by McCoy et al. titled “Differential Effect of Hospitalization on Cystatin C– and Creatinine-Based Estimated GFR,” which highlights the differential effect of hospitalization on estimated GFR based on cystatin C (eGFRCys) and estimated GFR based on serum creatinine (eGFRCr), with prolonged hospitalizations leading to greater decreases in eGFRCys compared with eGFRCr.1 Although the findings are compelling, we would like to raise several points for further discussion and consideration. First, the authors used hospitalization days as the primary exposure variable, which raises questions about the granularity of the data. Although the authors categorized hospitalization days into six groups, this approach may not fully capture the heterogeneity of hospitalizations. For instance, were the hospitalizations primarily due to AKI, infections, or cardiovascular events? Each of these conditions could have distinct effects on muscle mass, inflammation, and cystatin C levels, potentially confounding the observed associations. Would a more detailed stratification of hospitalization causes provide additional clarity? Second, the study assumes that eGFRCys is a more accurate reflection of true kidney function in the posthospitalization setting, based on prior evidence suggesting that eGFRCr may overestimate GFR due to reduced muscle mass during critical illness.2 However, cystatin C is also influenced by non-GFR factors such as inflammation, obesity, and thyroid dysfunction.3 Given that hospitalized patients often experience systemic inflammation, could the observed decreases in eGFRCys partly reflect inflammatory processes rather than true declines in kidney function? Incorporating biomarkers of inflammation, such as C-reactive protein, into the analysis might help disentangle these effects. Third, the study used linear mixed-effects models to estimate changes in eGFR, which is appropriate for longitudinal data, but the potential for residual confounding remains. Although the authors adjusted for baseline characteristics in a supplemental model, time-varying covariates such as changes in medication use (e.g., angiotensin-converting enzyme inhibitors or diuretics) or nutritional status during follow-up were not accounted for. Could these factors have influenced the observed trends in eGFRCys and eGFRCr? In conclusion, McCoy et al. provided a thought-provoking analysis of the differential effect of hospitalization on eGFRCys and eGFRCr. We believe that addressing the above questions in future research could further enhance our understanding of this important topic and its implications for clinical practice and research.