Personalizing dialysis treatment: haemodiafiltration or haemodialysis

In CHAMPION we will leverage routine clinical data to identify patients who benefit from the new therapy, i.e. haemodiafiltration, expanding beyond RCTs to provide robust, generalizable evidence. In our unique collaboration, we will be able to investigate this in datasets that spans multiple countries and includes extensive clinical data, offering a unique opportunity to address unresolved questions from previous trials.

Kidney failure patients have a poor prognosis in their usual care, i.e., hemodialysis (HD), highlighting the need for advancements in dialysis.  HDF offers potential benefits, however, these results, derived from strict inclusion and exclusion criteria, may not generalize to all patients. Personalized treatment approaches are needed to maximize HDF benefits and optimize resource allocation, enhancing patient outcomes and healthcare efficiency.

We are pioneering the use of target trial emulation studies, an innovative approach to address real-world clinical questions regarding treatment efficacy and safety. Randomised controlled trials, while invaluable, often lack generalizability due to their strict patient selection criteria and standardized protocols, which do not always reflect clinical practice. We will use data from over 500,000 dialysis patients to personalize treatment using machine learning techniques, considering time-varying exposures and individual patient characteristics.

We will provide insights into individual treatment benefits as well as shed light on generalisability issues of randomised controlled trials compared with real-world data. Our study will explore the theoretical understanding regarding the real-world effects of high-dose HDF in clinical practice. We will inform patients and clinicians on the expected individual treatment estimates, which might will be calculated through personalised prediction models.