Efficacy of renal replacement therapy in critically ill patients: a propensity analysis.

INTRODUCTION: Although renal replacement therapy (RRT) is a common procedure in critically ill patients with acute kidney injury (AKI), its efficacy remains uncertain. Patients who receive RRT usually have higher mortality rates than those who do not. However, many differences exist in severity patterns between patients with and those without RRT and available results are further confounded by treatment selection bias since no consensus on indications for RRT has been reached so far. Our aim was to account for these biases to accurately assess RRT efficacy, with special attention to RRT timing. METHODS: We performed a propensity analysis using data of the French longitudinal prospective multicenter Outcomerea database. Two propensity scores for RRT were built to match patients who received RRT to controls who did not despite having a close probability of receiving the procedure. AKI was defined according to RIFLE criteria. The association between RRT and hospital mortality was examined through multivariate conditional logistic regression analyses to control for residual confounding. Sensitivity analyses were conducted to examine the impact of RRT timing. RESULTS: Among the 2846 study patients, 545 (19%) received RRT. Crude mortality rates were higher in patients with than in those without RRT (38% vs 17.5%, P < 0.001). After matching and adjustment, RRT was not associated with a reduced hospital mortality. The two propensity models yielded concordant results. CONCLUSIONS: In our study population, RRT failed to reduce hospital mortality. This result emphasizes the need for randomized studies comparing RRT to conservative management in selected ICU patients, with special focus on RRT timing.