ditorial 

A-E-I-O-U and Sometimes Why—Dialysis in Acute Kidney Injury

Sushrut S. Waikar

JAMA Published Online: November 7, 2025

doi: 10.1001/jama.2025.21618

Students memorize the indications for dialysis in acute kidney injury (AKI) as the vowels of the alphabet, A-E-I-O-U: acidosis, electrolytes, ingestions, overload, and uremia. In a study published in JAMA, Liu and colleagues¹ ask clinicians to consider the last vowel, sometimes Y (“why”), in their report of the LIBERATE-D trial testing a conservative vs conventional thrice-weekly strategy for dialysis for patients with AKI. A brief historical review of the evolution of strategies for kidney replacement therapy (KRT) will help put LIBERATE-D into context.

Just over 80 years have passed since the hemodialysis pioneer Willem Kolff, MD, saved the life of a 67-year-old woman with AKI by using a rotating drum dialyzer during an 11-hour dialysis session.² In the early days of dialysis, long dialysis sessions were required because of the limits of technology. The invention of Belding Scribner, MD, of a vascular shunt for repeated bloodstream access permitted the development of maintenance hemodialysis at the University of Washington in Seattle. Dr Scribner’s first patient underwent a 76-hour dialysis session for the initial treatment.³

Over time, technologic advances in hemodialysis machinery, membranes, and vascular access allowed shorter, more frequent sessions, culminating in the long-term hemodialysis prescription familiar to many: 3- to 4-hour sessions thrice weekly.⁴ Although thrice-weekly hemodialysis has become the standard for chronic kidney failure, its applicability to AKI has long been questioned, given the many differences between chronic kidney failure and AKI. In the early 2000s, concern regarding underdialysis predominated because AKI is a setting of increased catabolism and hence a theoretic need for more efficient waste product removal and extracellular fluid volume management. In a series of randomized trials beginning with a comparison of daily vs thrice-weekly hemodialysis,⁵ investigators tested higher- vs lower-intensity KRT. It gradually became clear that daily dialysis, or higher dosing of continuous KRT, was no better than conventional thrice-weekly dialysis or standard-dose continuous KRT.^6,7 After this finding was demonstrated, the field moved toward answering the question about timing of dialysis, with the hypothesis that earlier initiation of KRT might avoid uremic complications. Several trials found that earlier initiation was no better than standard initiation of KRT in AKI.^8-10 Increasing the intensity of KRT for AKI, whether by frequent dialysis sessions, enhanced delivery during continuous KRT, or earlier initiation, proved to be of no additional benefit for patients, although excessive delay in KRT initiation has demonstrated risk.¹¹ In some of the trials, more intensive KRT had adverse effects on kidney recovery, whether measured by urine output or dialysis dependence.

Liu and colleagues¹ take this story one step further in LIBERATE-D, addressing the question of when to discontinue KRT. Should patients with AKI requiring dialysis be treated with the thrice-weekly schedule used for maintenance dialysis, with discontinuation after clear evidence of kidney function recovery? Or should clinicians aim to discontinue dialysis early by reassessing biochemical and clinical criteria and continuing only if needed? The trial tested a conventional vs conservative strategy in a 4-center study involving 220 patients with AKI severe enough to require at least 1 dialysis session. In the conservative group, patients were treated with subsequent KRT only if they developed any of the following A-E-I-O-U criteria (while allowing for physician discretion): metabolic acidosis with arterial pH less than 7.15, hyperkalemia with potassium level greater than 6 mmol/L or 5.5 mmol/L despite medical treatment, serum urea nitrogen level greater than 112 mg/dL, or volume overload with clinically significant hypoxemia. In the conventional group, patients were assigned a thrice-weekly schedule, with dialysis discontinued only after demonstration of recovery of kidney function. The primary outcome—kidney function recovery at hospital discharge, defined as being alive and not receiving dialysis for at least 14 days (including after discharge)—occurred more frequently for participants randomized to the conservative (70 of 109 [64%]) vs conventional (55 of 109 [50%]) strategy. Participants assigned to the conservative strategy received fewer dialysis sessions (1.8 vs 3.1) and recovered kidney function earlier (2 vs 8.5 days). Patients treated with the conservative strategy also had no discernible increase in adverse events, all of which suggests that the standard of care should change toward a more conservative approach to (dis)continuing KRT in AKI.

Several considerations need to be kept in mind when analyzing LIBERATE-D. First, statistical significance was narrowly met for the unadjusted primary end point (with a fragility index of 0), and prespecified adjustment for baseline clinical variables led to a nonsignificant result, leading to some uncertainty about the statistical robustness of the findings. The more important consideration relates to the provision of dialysis in the conventional group, which was designed to mimic the standard of care but deserves scrutiny: were the control patients actually receiving standard of care, or were they receiving unnecessary dialysis as specified in a randomized trial? According to the protocol, dialysis was to be delivered thrice weekly and discontinued only when specific criteria were met, according to 1 of the following: timed creatinine clearance of at least 8 mL/min; 24-hour urine output of 1 L/d without diuretics or 2 L/d with diuretics; or spontaneous decrease in creatinine level of greater than 0.3 mg/dL during 12 hours. Although these criteria were modeled after a previous clinical trial of early vs delayed dialysis initiation⁸ and clinicians were free to use clinical judgment to discontinue KRT off-protocol, the criteria are perhaps too strict and may not resemble clinical practice; LIBERATE-D may arguably be considered a trial of sensible discontinuation vs unnecessary continuation rather than conservative vs conventional dialysis strategies. In this regard, it is notable that outcomes between the 2 groups diverge immediately on randomization after day 0: as shown in the authors’ Table 2, more than 35% of patients in the conservative group recovered kidney function and did not need a single subsequent dialysis session, whereas less than 10% of patients in the conventional group recovered kidney function after day 0.

LIBERATE-D tested biochemical and volume indications for dialysis against a rigid kidney recovery goalpost; the latter appears to cause injury, and the former appears to be safe, at least for non–critically ill patients with recovering AKI. The trial results should change clinical practice by emphasizing that KRT with hemodialysis, although life-saving, is nephrotoxic. Each dialysis session exposes liters upon liters of blood to plastic tubing, an artificial membrane, and rapid solute/volume fluxes that do not resemble anything in normal human physiology. Previous AKI studies have also made this point: earlier, more intense, and more frequent dialysis may delay recovery of kidney function^8,12,13; the type of dialysis membrane (bioincompatible vs biocompatible) can injure the kidneys¹⁴; and, in large animal models, continuous KRT leads to an abrupt decrease in urine output without a change in kidney blood flow.¹⁵ Investigators are now revisiting the concept of minimizing hemodialysis in the long-term setting for new patients with residual kidney function: rather than automatically assigning a thrice-weekly prescription to new patients, incremental dialysis starts with 1 or 2 weekly sessions as a method to gradually introduce patients to treatment while sustaining residual kidney function.¹⁶ Both incremental dialysis initiation and LIBERATE-D rest on the same recognition: residual kidney function is superior to anything we can currently replace with machines, and relying on and sustaining residual kidney function should be a consideration in weighing risks and benefits of KRT. Future research should explore the mechanisms beyond just hypotension and membrane-induced inflammation that might account for the nephrotoxicity of KRT. For example, what are the consequences of the unintended filtration of phosphate, amino acids, and trace elements, which the kidney is smart enough to reabsorb (at great energetic cost) but which the KRT machine expels as wastewater? What future technologic advances can allow the dialysis membrane and filtration procedure to better replicate the glomerular capillary endothelium and the tubular epithelium?

The LIBERATE-D A-E-I-O-U criteria should not be extrapolated to critically ill patients with AKI who are receiving vasopressors or mechanical ventilation, which were exclusion criteria for trial participation. The LIBERATE-D trial patients were far less sick than many of the patients in the intensive care unit who receive KRT, many of whom may in fact need frequent and intensive KRT for optimal management of volume and metabolic needs. But for hemodynamically stable patients with hope for kidney recovery, excessive hemodialysis can set recovery back many days. Students and all clinicians managing dialysis for patients with AKI should still remember the acronym A-E-I-O-U but always ask why.