In a dialysis registry, haemodiafiltration was tied to fewer deaths

A large multinational analysis of dialysis registry data found that patients treated with high-volume haemodiafiltration died at a lower rate than those on standard high-flux haemodialysis — but the study was observational, not a randomised trial, and several of its authors work for the company that makes the equipment and owns the registry [s1]. The direction of the finding matches CONVINCE, the randomised trial that first showed a survival edge for the more intensive treatment, though the size of the benefit in the registry analysis was smaller [s1][s2].

Both treatments filter the blood of people whose kidneys have failed. Conventional high-flux haemodialysis removes waste mainly by diffusion across a membrane; haemodiafiltration adds convection, pushing a large volume of fluid across the membrane to drag out larger molecules, then replacing it [s2]. The open question has been whether that extra clearance actually helps people live longer, and if so by how much.

What the study did

The analysis, published in the Journal of the American Society of Nephrology, used a method called target-trial emulation: researchers take observational data and analyse it as though it were a randomised trial, to reduce the biases that plague simple comparisons [s1]. The data came from EuCliD, a registry of dialysis patients treated across eight European countries, covering adults on thrice-weekly in-centre dialysis between 2014 and 2019 [s1]. Inverse-probability weighting was used to balance the two groups, and kidney transplantation was treated as a competing event [s1].

Among 19,539 eligible patients, the weighting produced a pseudo-population of 19,758 (8,641 on haemodiafiltration, 11,117 on haemodialysis) [s1]. Over a median follow-up of 16 months, 4,282 deaths occurred [s1]. Haemodiafiltration was associated with a lower risk of death from any cause, with a hazard ratio of 0.72 (95% CI 0.67 to 0.77) [s1]. In absolute terms, the weighted cumulative incidence of death at two years was 20.6% with haemodiafiltration and 22.3% with haemodialysis — a difference of 1.7 percentage points [s1]. The association was broadly consistent across patient subgroups, with a stronger relative benefit among those who already had cardiovascular disease (interaction p<0.001) [s1].

Why the caveats do heavy lifting

A hazard ratio of 0.72 sounds large, but the absolute gap it describes here is modest, and the design cannot rule out the oldest problem in dialysis research: healthier, more stable patients are more likely to be placed on, and to tolerate, high-volume haemodiafiltration in the first place [s1]. Target-trial emulation and weighting are built to blunt exactly that confounding-by-indication, but they can only adjust for factors that were measured; unmeasured differences in how sick patients were can still flatter the treatment that fitter patients received [s1]. The authors themselves cautioned that the apparent dose-response — more convective volume, more benefit — "should be interpreted cautiously," because higher volumes may reflect patient stability and centre expertise rather than a true causal effect [s1].

Then there is who did the work. Several of the study's authors are employees of Fresenius Medical Care, a company that manufactures dialysis machines and consumables and operates the EuCliD registry the analysis drew on [s1]. That does not make the result wrong, and the analysis is transparent about its methods — but a positive finding for a treatment, produced from a manufacturer's own registry by some of its own staff, is exactly the kind of result that should lead with independent evidence rather than stand on its own.

The randomised benchmark

That independent evidence exists. CONVINCE, a pragmatic randomised trial funded by the European Commission, assigned 1,360 patients on high-flux haemodialysis to switch to high-dose haemodiafiltration (a convective volume of at least 23 litres per session) or to continue as they were [s2]. Over a median 30 months, death from any cause occurred in 17.3% of the haemodiafiltration group and 21.9% of the haemodialysis group, a hazard ratio of 0.77 (95% CI 0.65 to 0.93) [s2].

The registry emulation and the trial point the same way, which is reassuring; a randomised result is the stronger of the two, and the observational study is best read as evidence that the trial's benefit may extend to routine practice, not as an independent discovery [s1][s2]. Health Newspapers has separately covered how the blood test behind kidney-function estimates actually works and an experimental transplant approach aimed at replacing dialysis entirely.

What to watch

The practical questions are cost and capacity: high-dose haemodiafiltration needs ultrapure water and higher-specification machines, and whether health systems invest in it will turn on whether a roughly 1.7-to-4.6 percentage-point survival difference is judged worth the outlay [s1][s2]. Future analyses that draw on registries not owned by a manufacturer would strengthen the real-world case considerably.

This article describes research findings and is not medical advice. Dialysis modality is a decision for patients and their nephrology teams.

Sources

• Hemodiafiltration versus High-flux Hemodialysis and Risk of Mortality: A Multinational Target Trial Emulation — Journal of the American Society of Nephrology, 13 August 2026
• Effect of Hemodiafiltration or Hemodialysis on Mortality in Kidney Failure (CONVINCE) — New England Journal of Medicine, 16 June 2023