New England Journal of Medicine October 14, 2026

Intermittent or Continuous Renal-Replacement Therapy in Acute Kidney Injury

Stéphane Gaudry, Marouane Boubaya, Guillaume Louis, Khalil Chaïbi, Etienne de Montmollin, et al. (ICRAKI Trial Group)

Bottom Line

In a randomized noninferiority trial of critically ill adults with severe acute kidney injury, an initial strategy of intermittent hemodialysis was not noninferior to continuous renal-replacement therapy for 90-day major adverse kidney events and was associated with a significantly higher 90-day mortality.

Key Findings

1. The primary composite outcome of a major adverse kidney event at 90 days (MAKE90) occurred in 78.1% of patients in the intermittent hemodialysis (IHD) group compared to 71.6% in the continuous renal-replacement therapy (CRRT) group.

2. Noninferiority of IHD was not established, with an adjusted absolute risk difference of 6.28 percentage points (95% CI, -1.22 to 13.79), as the upper boundary of the confidence interval exceeded the pre-specified noninferiority margin.

3. At 90 days, all-cause mortality was significantly higher in the intermittent therapy group (63.3%, 164 of 260 patients) than in the continuous therapy group (53.6%, 147 of 275 patients), resulting in a hazard ratio of 1.30 (95% CI, 1.04 to 1.63).

4. Hemodynamic complications were highly prevalent but similar between arms; hypotension occurred in 68.8% of the IHD group and 66.5% of the CRRT group.

5. The trial was terminated prematurely by the Data and Safety Monitoring Board after 554 patients of a planned 1,000 were randomized due to the strong mortality signal favoring the continuous therapy arm.

Study Design

Design

RCT

Open-Label

Sample

554

Patients

Duration

90 days

Median

Setting

Multicenter, France

Population Critically ill adult patients with severe acute kidney injury requiring renal-replacement therapy.

Intervention Initial strategy of intermittent hemodialysis (IHD) for at least the first 4 consecutive days unless kidney recovery occurred.

Comparator Initial strategy of continuous renal-replacement therapy (CRRT) for at least the first 4 consecutive days unless kidney recovery occurred.

Outcome Major Adverse Kidney Event at 90 days (MAKE90), defined as a composite of death, dependence on renal-replacement therapy, or an increase in serum creatinine of more than 25% from baseline.

Study Limitations

• Premature termination of the trial (enrolling 554 of the planned 1000 patients) may exaggerate the precision and magnitude of the observed mortality difference.

• The trial utilized an open-label design, which is inherent to renal-replacement modality studies but allows potential bias in adjunctive patient management.

• There was a high rate of modality crossover after the protocol-mandated initial 3 days, with 21.9% of the IHD group and 42.5% of the CRRT group switching treatments based on clinician preference.

• Minor baseline imbalances existed between the groups, including more chronic kidney disease and emergency surgery patients in the intermittent arm, and more ARDS in the continuous arm.

• The intervention was limited to 38 ICUs exclusively in France, potentially limiting generalizability to healthcare systems with different ICU practices or resources.

• Intermittent dialysis dosing in the trial was reported to be slightly lower than standard recommended targets.

Clinical Significance

This trial challenges the long-standing paradigm that intermittent and continuous modalities provide equivalent clinical outcomes for severe acute kidney injury in the ICU. The failure of intermittent hemodialysis to meet noninferiority criteria, combined with a 9.7% absolute increase in 90-day mortality, strongly supports utilizing continuous renal-replacement therapy as the preferred initial modality for critically ill patients requiring renal support.

Historical Context

The optimal modality for renal replacement therapy in the intensive care unit has been heavily debated for decades. Prior historical randomized trials and meta-analyses generally suggested no significant difference in overall survival or long-term renal recovery between CRRT and IHD, leading to variable utilization driven largely by local expertise, cost, and machine availability, though CRRT has traditionally been favored for patients with profound hemodynamic instability. The ICRAKI trial was designed as a definitive, modern noninferiority study to ascertain whether IHD could be safely applied as an initial, universal strategy to reduce resource utilization. Its early halt due to excess mortality in the IHD arm marks a pivotal shift in critical care nephrology evidence, formally demonstrating outcome superiority for CRRT.

Guided Discussion

High-yield insights from every perspective

Med Student

Medical Student

Why might a critically ill patient with severe acute kidney injury tolerate continuous renal-replacement therapy (CRRT) better than intermittent hemodialysis (IRRT) from a hemodynamic and physiological standpoint?

Key Response

IRRT involves rapid fluid and solute shifts over 3-4 hours, which can lead to intravascular volume depletion, osmolar shifts, and subsequent hypotension. In contrast, CRRT provides gentle, continuous ultrafiltration and solute clearance, maintaining cardiovascular stability and preventing secondary ischemic insults to the recovering kidneys, which is critical in shock states.

Resident

Given the higher 90-day mortality associated with initial IRRT in this trial, how should this influence your decision-making when a newly admitted ICU patient with severe AKI requires dialysis, and are there still clinical scenarios where IRRT is preferred?

Key Response

CRRT should now be strongly favored as the initial modality for critically ill patients with severe AKI, especially those with hemodynamic instability. However, IRRT may still be indicated for life-threatening hyperkalemia, acute severe toxic ingestions like lithium where rapid clearance is required, or when patients have stabilized hemodynamically and are transitioning out of the ICU.

Fellow

The trial demonstrated a failure of noninferiority for MAKE90 and significantly higher 90-day mortality with IRRT. What are the proposed pathophysiological mechanisms by which intermittent hemodialysis might worsen both renal recovery and overall survival in this population?

Key Response

Intradialytic hypotension during IRRT causes recurrent ischemic injury to the renal medulla, delaying tubular recovery and increasing the likelihood of MAKE90, which includes death, new dialysis dependence, or persistent renal dysfunction. Additionally, rapid fluid shifts can induce myocardial stunning, systemic hypoperfusion, and exacerbate systemic inflammatory cascades, contributing to increased overall mortality.

Attending

Historically, meta-analyses suggested no significant mortality difference between CRRT and IRRT, leading to modality choices driven largely by resource availability. How does this trial challenge that paradigm, and how should ICU directors use this data to advocate for resource allocation?

Key Response

This trial shifts the paradigm by demonstrating a clear mortality benefit of CRRT over IRRT in severe AKI, contradicting the older 'equivalence' narrative. ICU directors must use this data to justify the higher upfront costs, equipment needs, and specialized nursing ratios required for CRRT, framing it as a life-saving intervention rather than just a logistical preference.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD

This study utilized a noninferiority design but ultimately found that IRRT was associated with significantly higher mortality. Statistically and methodologically, how do investigators handle the transition from noninferiority testing to concluding outright harm, and what are the implications for the trial's alpha spending?

Key Response

If the upper bound of the confidence interval for the treatment difference or hazard ratio falls entirely above the noninferiority margin and crosses the threshold of no effect, the active control (CRRT) is deemed superior. Because noninferiority and superiority can be tested sequentially in a closed testing procedure without inflating the type I error rate, the finding of harm is statistically robust, though evaluating whether the trial was adequately powered for this specific secondary superiority outcome is essential.

Journal Editor

In critical care trials comparing RRT modalities, blinding is inherently impossible. As a reviewer, how would you scrutinize the potential for performance bias, specifically regarding asymmetric crossover rates or subjective thresholds for initiation and discontinuation of the modality?

Key Response

Unblinded clinicians might have lower thresholds to switch a deteriorating patient from IRRT to CRRT, or conversely, might aggressively volume deplete IRRT patients, driving intradialytic hypotension. A rigorous review must evaluate the per-protocol versus intention-to-treat analyses, exact crossover rates, standardized protocols for fluid management, and whether indication bias influenced the delivery of the assigned therapy.

Guideline Committee

The 2012 KDIGO AKI guidelines state that CRRT and IRRT are considered complementary, though CRRT is suggested for hemodynamically unstable patients. Based on this new evidence demonstrating a 90-day mortality harm with IRRT, how should the strength of recommendation and level of evidence be updated for initial modality selection?

Key Response

Current guidelines treat the modalities as roughly equivalent for overall mortality (KDIGO 5.6.1/5.6.2). This trial provides high-quality randomized evidence that IRRT increases 90-day mortality and MAKE90 in critically ill severe AKI patients. The committee should likely upgrade the recommendation to strongly recommend CRRT as the initial modality for critically ill adults with severe AKI, upgrading the evidence level from low/moderate to high based on these definitive RCT findings.

Clinical Landscape

Noteworthy Related Trials

2006

Hemodiafe Trial

n = 360 · Lancet

Tested

Continuous venovenous hemodiafiltration (CVVHDF)

Population

Critically ill patients with acute renal failure and multiple-organ dysfunction syndrome

Comparator

Intermittent hemodialysis (IHD)

Endpoint

60-day survival

Key result: No significant difference in 60-day survival between the continuous and intermittent renal replacement therapy groups.

2008

ATN Trial

n = 1,124 · NEJM

Tested

Intensive renal replacement therapy

Population

Critically ill patients with acute kidney injury

Comparator

Less-intensive renal replacement therapy

Endpoint

Death from any cause by day 60

Key result: Intensive renal replacement therapy did not decrease mortality or improve recovery of kidney function compared with less-intensive therapy.

2020

STARRT-AKI Trial

n = 2,927 · NEJM

Tested

Accelerated initiation of renal-replacement therapy

Population

Critically ill patients with severe acute kidney injury

Comparator

Standard strategy (delaying initiation)

Endpoint

Death from any cause at 90 days

Key result: Accelerated initiation of renal-replacement therapy did not lower the risk of death at 90 days but was associated with a higher incidence of adverse events.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis