The Population Representativeness of Randomized Controlled Trials and Real‐World Evidence Studies for Patients With End‐Stage Kidney Disease in the United States

OBJECTIVE: To assess the population representativeness of randomized controlled trials (RCTs) and real-world evidence (RWE) studies in end-stage kidney disease (ESKD) relative to the United States Renal Data System (USRDS) target population. The primary aim was to evaluate the generalizability of US-based evidence, and the secondary aim to examine the transportability of non-US evidence to the US dialysis population. METHODS: We searched PubMed for RCTs and RWE studies in ESKD dialysis populations (2007-2024). Random-effects models were used to pool effect modifiers and mortality rates separately for US-based and non-US RCTs and RWE studies. The primary outcome was a composite representativeness measure, defined as no meaningful deviation across all effect modifiers and mortality rate relative to USRDS benchmarks. Secondary outcomes were single-component representativeness measures for age, diabetes prevalence, and mortality rate. Meaningful deviation was defined as absolute standardized mean difference (SMD) >0.25. Subgroup and sensitivity analyses were conducted. RESULTS: We included 194 RCTs (n = 71,119) and 173 RWE studies (n = 50,931,214). For the primary aim, neither US-based RCT samples nor US-based RWE samples were representative. RCT participants were younger (SMD = -0.27) and had lower mortality rates (SMD = -1.52) than the USRDS population, whereas RWE samples showed no meaningful deviations in age, diabetes prevalence, or mortality rate (SMDs = 0.00, -0.11, and 0.08). Meaningful deviations in other effect modifiers remained in both study types. Non-US samples also showed limited representativeness. CONCLUSIONS: RCTs and RWE studies demonstrate limited representativeness, underscoring the need to validate and adapt evidence before clinical application.