• Stéphane Gaudry, Jean-Damien Ricard, Clément Leclaire, Cédric Rafat, Jonathan Messika, Alexandre Bedet, Lucile Regard, David Hajage, and Didier Dreyfuss.
• AP-HP, Hôpital Louis Mourier, Service de Réanimation Médico-Chirurgicale, 178 rue des Renouillers, F-92700, Colombes, France; Univ Paris Diderot, Sorbonne Paris Cité, ECEVE, UMRS 1123, F-75010 Paris, France; INSERM, ECEVE, U1123, F-75010 Paris, France.
• J Crit Care. 2014 Dec 1;29(6):1022-7.

PurposeRenal replacement therapy (RRT) is a major supportive treatment of acute kidney injury (AKI) in intensive care unit (ICU), but the timing of its initiation remains open to debate.Materials And MethodsWe retrospectively analyzed ICU patients who had AKI associated with at least one usual RRT criteria: serum creatinine concentration greater than 300 μmol/L, serum urea concentration greater than 25 mmol/L, serum potassium concentration greater than 6.5 mmol/L, severe metabolic acidosis (arterial blood pH<7.2), oliguria (urine output<135 mL/8 hours or <400 mL/24 hours), overload pulmonary edema. To estimate the risk of death associated with RRT adjusted for risk factors, we performed a marginal structural Cox model with inverse-probability-of-treatment-weighted estimator.ResultsAmong 4173 patients admitted to the ICU, 203 patients fulfilled potential RRT criteria. Ninety-one patients (44.8%) received RRT and 112 (55.2%) did not. Non-RRT and RRT patients differed in terms of severity of illness: Simplified Acute Physiology Score II (55±17 vs 60±19, respectively; P<.05) and Sequential Organ Failure Assessment score (8 [5-10] vs 9 [7-11], respectively; P=.01). Crude analysis indicated a lower ICU mortality for non-RRT compared with RRT patients (18% vs 45%; P<.001). In the marginal structural Cox model, RRT was associated with increased mortality (P<.01).ConclusionA conservative approach of AKI was not associated with increased mortality.Copyright © 2014 Elsevier Inc. All rights reserved.

29 keywords...

hide…