Khalil Chaïbi, Yanis Akrour, Guillaume Louis, Laurent Martin Lefevre, Bertrand Pons, Sébastien Moschietto, Eric Boulet, David Hajage, Saïd Lebbah, Jean-Pierre Quenot, Frederic Sangla, Didier Dreyfuss, Stéphane Gaudry, David Legouis, on behalf of the STOPDAPT-3 Investigators, Stéphane Gaudry, David Hajage, Frédérique Schortgen, Laurent Martin Lefèvre, Bertrand Pons, Eric Boulet, Alexandre Boyer, Guillaume Chevrel, Nicolas Lerolle, Dorothée Carpentier, Nicolas de Prost, Alexandre Lautrette, Anne Bretagnol, Julien Mayaux, Saad Nseir, Bruno Mégarbane, Marina Thirion, Jean-Marie Forel, Julien Maizel, Hodane Yonis, Philippe Markowicz, Guillaume Thiery, Florence Tubach, Jean-Damien Ricard, Didier Dreyfuss, Saïd Lebbah, Guillaume Louis, Sébastien Moschietto, Dimitri Titeca-Beauport, Béatrice La Combe, Sébastien Besset, Alain Combes, Adrien Robine, Marion Beuzelin, Julio Badie, Julien Bohé, Elisabeth Coupez, Nicolas Chudeau, Saber Barbar, Christophe Vinsonneau, Didier Thévenin, Karim Lakhal, Nadia Aissaoui, Steven Grange, Marc Leone, Guillaume Lacave, Florent Poirson, Karim Asehnoune, Guillaume Géri, Kada Klouche, Laurent Argaud, Bertrand Rozec, Cyril Cadoz, Pascal Andreu, Jean Reignier, Jean-Pierre Quenot, Jean-Marie Forel, Guillaume Thiery
INTRODUCTION: Renal replacement therapy (RRT) is a life-saving intervention in severe acute kidney injury (AKI). After a clinician-initiated discontinuation, the likelihood of sustained liberation remains uncertain. We aimed to identify predictors of successful RRT weaning and to develop a pragmatic bedside tool (the UNDERSCORE) to support post-discontinuation management. METHODS: This post-hoc analysis of two multicenter randomized trials (AKIKI and AKIKI2) included ICU patients with KDIGO stage 3 AKI managed with a conservative initiation approach. Patients were eligible if they underwent an RRT weaning attempt, defined as discontinuation for ≥ 3 consecutive days. The primary outcome was successful weaning, i.e., no RRT resumption within seven days. Independent predictors were identified using multivariable logistic regression, and the resulting model (UNDERSCORE) was externally validated in an independent Swiss ICU cohort. RESULTS: Among 554 patients who received RRT, 180 underwent a weaning attempt and 101 (56%) were successfully weaned. Six predictors were retained to construct the UNDERSCORE: RRT duration before the attempt, septic shock on admission, baseline serum creatinine, and three clinical variables assessed after the weaning attempt (use of vasopressors, invasive mechanical ventilation, and urine output). The score showed strong discrimination in the derivation cohort (AUC 0.86, 95% CI 0.80-0.91). In the external Swiss cohort (n = 415), 338 patients (81%) were successfully weaned, with fair performance across a broader case mix (AUC 0.73, 95% CI 0.66-0.80). CONCLUSION: The UNDERSCORE, derived from a homogeneous conservative RRT initiation cohort and validated in a diverse ICU population, provides a bedside tool to estimate the probability of sustained RRT discontinuation after an initial clinician-initiated stop.