Early initiation of renal replacement therapy in critically ill patients: a meta-analysis of randomized clinical trials

BACKGROUND: Acute kidney injury (AKI) is strongly associated with high morbidity and mortality of critically ill patients. In the last years several different biological markers with higher sensitivity and specificity for the occurrence of renal impairment have been developed in order to promptly recognize and treat AKI. Nonetheless, their potential role in improving patients’ outcome remains unclear since the effectiveness of an “earlier” initiation of renal replacement therapy (RRT) is still debated. Since one large, high-quality randomized clinical trial has been recently pubblished, we decided to perform a meta-analysis of all the RCTs ever performed on “earlier” initiation of RRT versus standard RRT in critically ill patients with AKI to evaluate its effect on major outcomes. METHODS: Pertinent studies were independently searched in BioMedCentral, PubMed, Embase, and Cochrane Central Register of clinical trials. The following inclusion criteria were used: random allocation to treatment (“earlier” initiation of RRT versus later/standard initiation); critically ill patients. RESULTS: Ten trials randomizing 2214 patients, 1073 to earlier initiation of RRT and 1141 to later initiation were included. No difference in mortality (43.3% (465 of 1073) for those receiving early RRT and 40.8% (466 of 1141) for controls, p = 0.97) and survival without dependence on RRT (3.6% (34 of 931) for those receiving early RRT and 4.2% (40 of 939) for controls, p = 0.51) were observed in the overall population. On the contrary, early initiation of RRT was associated with a significant reduction in hospital length of stay. No differences in occurrence of adverse events were observed. CONCLUSIONS: Our study suggests that early initiation of RRT in critically ill patients with AKI does not provide a clinically relevant advantage when compared with standard/late initiation. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s12871-019-0733-7) contains supplementary material, which is available to authorized users.