Characteristics of Populations Included in Randomized Controlled Trials of Hemodiafiltration and Registry Real-Life Populations: A Systematic Review.

Selection criteria in randomized trials (RCTs) can lead to differences in key characteristics between trial participants and real-life populations. We evaluated reporting of population characteristics in existing RCTs of hemodiafiltration (HDF) and in real-life populations included in kidney registries to descriptively identify key differences. We used systematic review methodology to identify existing RCTs of HDF vs hemodialysis (HD) (1966 to May 2024). We also searched the Fresenius Quantitative Market Analysis team registry database (2024 update) for existing registries from Europe, the Asia-Pacific region and America including populations on HDF. Patient characteristics from RCTs and registries were extracted, summarized and compared descriptively. Eleven RCTs (N=5108) and eight registries (N=1,147,167) were identified. There were no RCTs in the United States and only two small RCTs from Australia (N=124) and Brazil (N=195). Most trials were from Europe. Key characteristics consistently reported in both RCTs and registries were only age, sex, diabetes, cardiovascular disease, vascular access type and dialysis vintage. There was moderate to high heterogeneity for these patient characteristics in RCTs, indicating enrolment of a broad array of people. The proportion of people with diabetes was 26% in RCTs and 43% in registries. The prevalence of arteriovenous fistulas/graft was 90% in RCTs and 70% in registries. There was a broad but incomplete array of patient characteristics in existing RCTs and real-world registries of HDF vs HD. Data were primarily limited to Europe and only a core set of demographic and clinical variables. Apart for age, sex, diabetes, cardiovascular disease, vascular access type and dialysis vintage, other patient and treatment relevant characteristics were erratically or not at all reported in RCTs as well as in real-world registries. With potential differences in patient populations, we support the need for studies examining HDF in real world settings, e.g. with target emulation trials.