Objective: To appraise the literature on the effect of initial renal replacement therapy (RRT) modality on clinical outcomes.
Design: Systematic review and meta-analysis.
Setting: Academic medical center.
Patients and participants: Adult critically ill patients with acute kidney injury.
Interventions: Continuous vs. intermittent RRT.
Measurements and results: MEDLINE, EMBASE, Cochrane Controlled Clinical Trials Register, and other sources were searched. We identified nine unique randomized trials (n = 1,403). No trial satisfied all quality indicators and several had limitations related to selection bias, randomization, imbalances in patient characteristics, and high treatment crossover. No trial standardized the timing, criteria, for initiation or dose of RRT. There was no statistical evidence that initial modality influenced mortality (odds ratio, 0.99; 95% confidence interval, 0.78-1.26, p = .93; I2 = 11%; nine trials, n = 1,403) or recovery to RRT independence (odds ratio, 0.76; 95% confidence interval, 0.28-2.07, p = .59; I2 = 0%; four trials, n = 306). There was suggestion that continuous RRT had fewer episodes of hemodynamic instability and better control of fluid balance.
Conclusions: We identified numerous issues related to study design, conduct, and quality that dispute the validity and question any inferences that can be drawn from these trials. In the context of these limitations, the initial RRT modality did not seem to affect mortality or recovery to RRT independence. There is urgent need for additional high-quality and suitably powered trials to adequately address this issue.