Bulletin de la Dialyse à Domicile

Assessment of the Completeness and Representativeness of the RDPLF Registry Using the French National REIN Registry as a Reference

Mon, 15 Jun 2026 00:00:00 +0200

For more than 40 years, the French Language Peritoneal Dialysis (and Home Hemodialysis) Registry (RDPLF) has been a major source of data on home dialysis treatment modalities. Since centers participate on a voluntary basis, the question of its completeness and national representativeness remains essential. This study evaluates these parameters by comparing RDPLF data with those from the 2026 report of the French national REIN registry, which is considered comprehensive for France.
The comparison is based on data as of December 31, 2024; several common indicators were analyzed: number of prevalent and incident patients on peritoneal dialysis, regional distribution of patients, treatment modalities by age, level of autonomy, use of assistance, and parameters related to anemia. The results show that the RDPLF captures between 90% and 97% of prevalent patients recorded in REIN. The number of incident patients is slightly higher in the RDPLF, likely due to differences in the timing of data updates. Despite these discrepancies, the trends observed in the two registries are remarkably consistent.
A comparison of patient characteristics and treatment practices reveals a very high degree of consistency. The distributions of peritoneal dialysis techniques by age, levels of autonomy, and dialysis modalities are virtually identical.
These results confirm the RDPLF’s national representativeness for peritoneal dialysis in France and highlight the complementary nature of the two registries. While REIN provides comprehensive surveillance of chronic kidney disease, the RDPLF offers detailed, specialized clinical data on home-based treatments. This complementarity is a significant asset for evaluating practices and developing home dialysis strategies.

Four Best Practices for Using ChatGPT/Claude as an Effective R Assistant

Claire Della Vedova — Mon, 15 Jun 2026 00:00:00 +0200

Artificial intelligence can be of considerable help to those wishing to publish their work; however, using it without a critical eye can lead to completely false results. We felt it would be useful to ask a biostatistician and trainer to raise our readers’ awareness of the potential pitfalls of such software when analysing their data, using the example of a programme widely used in medical research.
This article presents four best practices for the effective use of language models (LLMs) such as ChatGPT or Claude in an R programming context. The author begins by describing the three most common errors produced by LLMs: the invention of non-existent functions, the use of obsolete syntax and, even more seriously, the suggestion of inappropriate statistical methods. These errors can go unnoticed by beginners and lead to erroneous analyses. It is therefore essential to systematically check the suggested functions, pay attention to R’s warning messages, and understand the conditions under which the recommended statistical tests apply.
The article then details a three-step method for obtaining more reliable results: providing a precise context regarding the data and the working environment, clearly describing the desired objective, and specifying the expected format of the output. The author places particular emphasis on data confidentiality.
A third section explains how to verify that generated code actually works. Four checks are suggested: checking for the absence of errors and “warnings”, verifying the structure of the results, assessing the plausibility of the values obtained, and testing the code on a small set of known data.

Finally, the article specifies when to prioritise an LLM, when to consult the official R documentation, and when to seek the advice of a competent human. The overall aim is to empower the user whilst using LLMs as support tools rather than as substitutes for critical thinking.

Early-Onset Craniofacial Brown Tumor Progressing to Multifocal Skeletal Lesions in Peritoneal Dialysis Patient: A Case Report and Literature Review

Mon, 15 Jun 2026 00:00:00 +0200

Background: Brown tumors (BTs) are rare skeletal manifestations of osteitis fibrosa cystica caused by prolonged hyperparathyroidism. They are most commonly reported in patients with advanced secondary hyperparathyroidism undergoing hemodialysis. Conversely, early presentation in peritoneal dialysis (PD) remains uncommon, and craniofacial involvement is especially rare.
Case Presentation: We report a case of a 27-year-old woman with end-stage renal disease on PD who developed a craniofacial BT shortly after dialysis initiation. A biochemical evaluation revealed markedly elevated parathyroid hormone (PTH) levels (3032 pg/mL). Despite initial partial regression, the BT rapidly progressed to multifocal skeletal lesions, including on the clavicle, humeral head, ribs, and ischiopubic ramus, associated with fractures and tumoral calcinosis. Medical management was initiated, and parathyroidectomy was planned.
Conclusion: This case emphasizes the potential for early craniofacial BT development and rapid multifocal skeletal progression in PD patients with uncontrolled hyperparathyroidism. It further demonstrates that early detection and prompt surgical referral may be vital in preventing severe skeletal complications in PD patients with uncontrolled hyperparathyroidism.