Hemodynamic Effects of Renal Replacement Therapy

Renal replacement therapy changes the circulation while treating the consequences of kidney failure. Blood enters an extracorporeal circuit, water and solutes cross a membrane, and returning blood may differ in temperature, electrolyte composition, and acid–base state. These changes occur in patients whose vascular tone, cardiac reserve, and microcirculation may already be impaired. The result can be beneficial decongestion, circulatory deterioration, or both at different times. Chapter 50 explains transport and fluid accounting; this chapter follows their consequences through the cardiovascular system.

51.1 Hemodynamic instability is a pressure–flow problem

Arterial pressure depends on the interaction between cardiac output and systemic vascular resistance. Mean arterial pressure is therefore an incomplete description of the circulation during dialysis. A pressure fall can reflect reduced output, reduced vascular tone, or both. Conversely, a maintained pressure can conceal falling flow if vasoconstriction compensates. Hemodynamic instability related to renal replacement therapy includes clinically important circulatory deterioration around treatment; it does not identify a single mechanism or prove that the machine caused every deterioration.

For a simplified steady circulation, cardiac output is measured in liters per minute, systemic vascular resistance in mmHg·min/L, and right atrial pressure in mmHg. Ignoring pulsatility and small differences between measured central venous and right atrial pressures:

Suppose right atrial pressure is 10 mmHg, output is 5 L/min, and resistance is 12 mmHg·min/L. The estimated MAP is 10 + (5 × 12) = 70 mmHg. If output falls to 4 L/min with unchanged resistance, MAP becomes 58 mmHg. The same 58 mmHg could occur with output maintained at 5 L/min and resistance falling to 9.6 mmHg·min/L. The pressure is identical; the physiological problem is different. These constructed values illustrate relationships, not diagnostic thresholds.

In Bitker and colleagues’ selected ICU cohort, hypotension occurred during 61 of 107 intermittent dialysis sessions, but only 19% of hypotensive sessions were associated with preload dependence assessed by passive leg raising. This observational finding does not establish the alternative mechanism in every episode. It does demonstrate why dialysis-associated hypotension cannot automatically be translated into inadequate circulating volume. 1

51.2 Connection establishes a new blood compartment

The extracorporeal circuit contains blood outside the patient’s anatomical vasculature. During connection, blood displaces the priming solution and fills the tubing and filter. The circulatory effect depends on the circuit volume, the patient’s blood volume, whether priming fluid enters the patient or is discarded, and concurrent infusions and withdrawals. “Blood outside the body” and “blood volume lost from the patient” are therefore different quantities.

Consider an illustrative circuit containing 200 mL and an adult with an estimated blood volume of 5,000 mL. The circuit volume is 200/5,000 = 4% of that blood volume. This describes the scale of the extracorporeal compartment. It does not establish an abrupt 4% reduction in effective circulating volume: the priming and connection procedure determines what volume is added, displaced, or removed. The same circuit occupies a much larger fraction of a small child’s blood volume, which limits extrapolation between pediatric and adult connection studies.

Once a venovenous circuit is operating, blood withdrawn through its access limb is returned through its return limb, apart from the net fluid balance across the circuit. The displayed blood-pump flow is not a continuing loss of blood and must not be subtracted from cardiac output as though the circuit were an open drain. Access and return pressures largely reflect local catheter and circuit resistance; a very negative access pressure is not a direct measurement of right atrial pressure.

Connection can nevertheless expose limited reserve through volume redistribution, temperature change, a membrane reaction, bleeding, or disturbances in vascular access and medication delivery. In a small adult before-and-after study, Eastwood and colleagues found no hypotensive episodes with either of two pump-start approaches and could not recommend the slower approach from their results. The study supports observation during initiation, not a universally protective pump-start rule. 2

51.3 Net removal and plasma refill operate on different compartments

Net patient fluid removal extracts plasma water at the filter. Much of the excess fluid being treated may reside in interstitial tissue rather than in plasma. Continued removal therefore depends on redistribution into the vascular compartment, often described as plasma refill. This includes transvascular movement and lymphatic return; it is not a fixed pump supplying water from an unlimited edema reservoir.

As plasma water is removed, reduced capillary hydrostatic pressure and increased plasma protein concentration can favor vascular volume preservation. Interstitial pressure, interstitial protein concentration, endothelial permeability, the glycocalyx, and lymphatic drainage modify that response. Direct measurements in maintenance dialysis patients demonstrate that plasma and interstitial volumes and their colloid osmotic relationships change together during ultrafiltration. 3 These observations establish redistribution, not a reliable bedside refill rate for septic capillaries.

Define the removal rate as the net volume extracted from the patient per hour, after the circuit’s replacement-fluid accounting. Define refill here as net fluid entry into the plasma from outside that compartment. If removal is illustratively 120 mL/h and refill is 80 mL/h, with no other vascular inputs, losses, or osmotic redistribution, the plasma compartment loses 40 mL/h. Over two hours the calculated reduction is 80 mL. Neither rate is a prescription. In a living patient both vary, and a small plasma deficit can matter when venous return is already precarious.

Edema does not guarantee rapid refill. Conversely, a fall in plasma volume need not cause hypotension if venoconstriction and cardiac responses preserve flow. A positive whole-patient fluid balance also does not prove adequate intravascular filling: infused water may leave the circulation while vascular extraction continues. Whole-body balance, plasma volume, and stressed venous volume answer different questions.

Hypoalbuminemia is relevant to these exchanges but does not, by itself, quantify refill failure or establish that an albumin infusion will restore tolerance. The endothelial barrier and downstream tissue conditions remain part of the mechanism.

51.4 Venous return depends on reserve, not simply on filling pressure

The circulation accommodates a reduction in blood volume partly by recruiting venous blood into the stressed compartment. Sympathetic venoconstriction reduces venous capacitance and can help preserve the upstream pressure driving blood toward the right atrium. When this compensation is inadequate, fluid removal reduces venous return, right ventricular filling, and eventually left ventricular output.

The simplified venous-return relation uses the difference between mean systemic filling pressure and right atrial pressure, divided by resistance to venous return. Neither upstream pressure nor that resistance is fixed during dialysis. Sepsis, sedation, autonomic dysfunction, vasoactive treatment, and abdominal pressure alter the relation. If external pressure partly collapses a vein, a local flow-limiting segment can become the effective downstream constraint rather than right atrial pressure alone. Central venous pressure cannot identify that constraint, the amount of recruitable venous blood, or the cardiac response to losing it.

Preload dependence means that a reversible increase in cardiac preload increases stroke volume. It identifies an operating position on the cardiac function relationship, not a diagnosis of total-body dehydration. A congested patient with a poorly functioning right ventricle can remain preload dependent. Giving additional volume might increase output briefly while worsening venous pressure, pulmonary edema, or organ congestion.

In a 39-patient physiological study, Monnet and colleagues found that a pre-treatment passive leg raise, assessed by changes in cardiac index, predicted subsequent intolerance of intermittent fluid removal. The study was small and used an intervention-based definition of hypotension; its diagnostic threshold is not a universal removal criterion. 4 Its teaching value is that the response to a transient preload change can be more informative than a static pressure. The maneuver requires an appropriate technique and prompt, reliable flow measurement; abdominal pressure can limit the blood volume it mobilizes.

A negative test is not permanent protection. Removal, vasodilation, new arrhythmia, or altered ventilation may move the patient to a different operating point. The converse also matters: identifying preload dependence explains a vulnerability but does not establish that fluid administration has the best overall benefit. Need for decongestion, evidence of impaired perfusion, and capacity to tolerate added fluid must still be considered together.

51.5 The two ventricles can respond differently to decongestion

Fluid removal can improve a congested circulation before it threatens filling. Lower right-sided volume may reduce tricuspid regurgitation, right ventricular distension, and the septal displacement that impairs left ventricular filling. Lower left atrial pressure can reduce pulmonary hydrostatic filtration and pulmonary venous loading. These benefits explain why a patient with edema and venous congestion may improve while total circulating volume decreases.

The beneficial direction is conditional on the starting state. A right ventricle with severe pulmonary vascular load may require adequate preload to maintain stroke volume. If removal reduces filling without reducing that load, output can fall. Right ventricular dilation, pericardial constraint, and ventricular interdependence also mean that measured filling pressures cannot be interpreted as isolated left- or right-heart volume gauges.

Pulmonary vascular resistance is influenced by lung volume, alveolar and vascular pressures, hypoxia, hypercapnia, acidemia, and pulmonary vascular disease. Removing pulmonary edema may improve oxygenation and some components of vascular loading. It does not necessarily reverse pulmonary vascular obstruction or overdistension caused by ventilation. A lower wedge pressure may reduce one burden while reduced pulmonary vascular distension changes resistance and compliance in another direction. A falling pulmonary artery pressure can even accompany worsening output rather than improved right ventricular afterload.

Left ventricular hypertrophy and diastolic dysfunction introduce another vulnerability. A stiff ventricle can require higher filling pressure to achieve a given end-diastolic volume. Removal may reduce stroke volume when filling falls below its useful operating range, even while elevated pressures and congestion remain. At the same time, lower arterial pressure can reduce coronary perfusion, particularly when intracardiac diastolic pressures remain elevated. Thus, lower congestion pressure and adequate myocardial perfusion are related goals that can diverge during treatment.

Serial assessment asks whether chamber size, septal configuration, regurgitation, forward flow, and organ perfusion move together. A single ejection fraction or filling pressure cannot describe the balance between unloading benefit and loss of preload reserve.

51.6 Positive pressure and position change the response to removal

Positive-pressure ventilation changes both the external pressure around the heart and the vascular conditions through which venous blood returns. Increasing pleural pressure may raise measured right atrial pressure while reducing transmural filling. It can also alter mean systemic filling pressure through pressure transmission and redistribution. Canine experiments show that PEEP can raise both mean systemic filling pressure and right atrial pressure; their relative changes depend on the preparation and accompanying conditions. 5 Consequently, subtracting a fixed fraction of PEEP from CVP does not reveal a reliable “true preload.”

Fluid removal and a ventilator change may therefore converge on the same limited reserve. A patient tolerating treatment can become unstable after recruitment, increased PEEP, or deeper sedation, even if net removal has not changed. Conversely, improved lung aeration and reduced respiratory effort may improve cardiac loading. The relevant direction depends on lung recruitment, overdistension, pleural pressure transmission, and right ventricular function.

Position adds a second perturbation. Raising the torso redistributes blood toward dependent capacitance vessels. Proning may change abdominal pressure, venous drainage, chest-wall mechanics, and right ventricular loading; its effect cannot be reduced to one expected CVP change. Compression of the abdomen can impede venous return while increasing measured venous pressures. Turning may also alter catheter function and produce interruptions that change actual treatment delivery.

Pressure measurements must retain their hydrostatic reference. If the patient’s position changes without appropriate transducer leveling, an apparent pressure change can be partly a measurement artifact. A CVP rise during positive pressure can coexist with reduced output, and a CVP fall during decongestion can coexist with preserved output. Neither direction alone establishes tolerance.

The sequence of events is therefore valuable physiology. Recording ventilation, position, sedation, and circuit changes alongside the hemodynamic trend helps identify a combined perturbation that would be missed by inspecting the cumulative removal volume alone.

51.7 Solute removal changes water distribution

Dialysis can lower extracellular solute concentrations more rapidly than all tissue compartments equilibrate. Water then redistributes according to the transient effective osmotic gradients across their barriers. A reduction in extracellular osmolality may shift water into cells and reduce the volume available to support the circulation, even without substantial net extraction of water from the body.

Urea illustrates why time and membrane permeability matter. At equilibrium, urea crosses many cell membranes and contributes less to sustained tonicity than sodium salts. During rapid removal, however, slower equilibration in particular tissues can create transient gradients. Measured plasma osmolality, calculated osmolality, and effective tonicity are related but different quantities; a fall in urea concentration must not be converted directly into a predicted plasma-volume loss.

For scale, suppose plasma urea falls from 30 to 15 mmol/L while sodium and glucose remain unchanged. Urea’s contribution to calculated osmolarity falls by approximately 15 mOsm/L, numerically close to 15 mOsm/kg under the dilute-plasma approximation. That arithmetic describes a concentration contribution. It does not establish that an effective 15 mOsm/kg gradient persists across every cell membrane, and it cannot calculate the volume of water leaving plasma without a compartmental model and equilibration data.

The relation is also more complex than “dialysis impairs refill whereas isolated ultrafiltration preserves it.” Rodriguez and colleagues found lower plasma osmolality during regular dialysis, but comparable plasma-volume and colloid osmotic changes during matched fluid removal in a small maintenance-dialysis comparison. 6 Osmotic redistribution is a plausible contributor whose magnitude depends on the treatment and patient; it is not a complete explanation of every pressure response.

Longer delivery can reduce the speed of concentration change. Increasing sodium delivery may support extracellular osmolality during treatment but can add sodium and water burden. These competing effects prevent the mechanistic argument from establishing a universally favorable dialysate composition.

51.8 Temperature and vascular tone can uncouple pressure from perfusion

Blood exchanges heat with tubing, the filter, dialysate, and replacement fluid. Net thermal balance depends on those exchanges, warming devices, ambient conditions, and the patient’s heat production. A displayed fluid temperature is not identical to core temperature or to the thermal energy transferred to the patient.

Warming can favor cutaneous vasodilation and oppose the vasoconstriction needed to compensate for falling volume. Cooling may increase vascular resistance, recruit venous volume, and reduce heart rate and metabolic demand. A pressure increase during cooling therefore does not necessarily mean increased cardiac output or oxygen delivery. In nine septic ventilated patients, Rokyta and colleagues observed increased resistance and MAP during CVVH-induced cooling, accompanied by lower cardiac output and systemic oxygen delivery; measured hepatosplanchnic indices did not deteriorate. This small physiological experiment cannot establish universal organ protection. 7

The distinction between immediate hemodynamics and clinical outcomes remains essential. The MyTEMP cluster-randomized trial included 15,413 maintenance hemodialysis patients and found no reduction in its major cardiovascular composite with a center-wide personalized cooler-dialysate policy. It studied chronic outpatient dialysis, not septic ICU patients, and does not invalidate every temperature-dependent vascular effect. 8 It does prevent a physiological rationale from being presented as established survival benefit.

Vasoplegia may predate treatment and continue to evolve during it. Sepsis-related vasodilation, sedative effects, and impaired sympathetic compensation can make a modest treatment-related change consequential. An abrupt pressure collapse also raises possibilities such as bleeding, arrhythmia, severe membrane-associated reaction, or a problem with vasoactive-drug delivery. These mechanisms require different responses from simple volume depletion.

Thus, “stable pressure” achieved through vasoconstriction is only one part of tolerance. Temperature, peripheral perfusion, cardiac output, rhythm, and the changing need for circulatory support determine what that pressure means.

51.9 Acid–base and electrolyte changes influence contraction and rhythm

Correction of severe acidemia may improve cardiovascular function by changing myocardial performance, vascular responsiveness, and pulmonary vascular loading. The result is not determined by serum bicarbonate alone. Carbon dioxide generation and removal, ventilation, intracellular buffering, and the speed of correction influence the final acid–base response. Bicarbonate delivery can increase the carbon dioxide load that must be eliminated through the lungs.

Alkalemia increases calcium binding to albumin and can lower ionized calcium despite an unchanged total calcium concentration. Calcium transfer across the dialysis membrane may reinforce or oppose that change. Because ionized calcium participates in excitation–contraction coupling, a change can alter stroke volume and vascular responses independently of removal volume. A randomized crossover study in stable maintenance-dialysis patients showed distinct hemodynamic effects of changing bicarbonate and calcium concentrations; those patients and tested prescriptions do not define an ICU optimum. 9

Regional citrate anticoagulation introduces a related pathway. Citrate binds calcium in the circuit; systemic calcium balance depends on citrate delivery and removal, calcium replacement, and citrate metabolism. If systemic ionized calcium falls, impaired contraction, hypotension, or arrhythmia may follow. If citrate metabolism is impaired, citrate–calcium complexes can accumulate; if citrate is metabolized, its alkali effect also contributes to acid–base balance. The RICH randomized trial documented metabolic adverse events during citrate treatment, underscoring that anticoagulation and circuit patency do not remove the need to monitor systemic chemistry. 10

Potassium and magnesium shifts can alter membrane excitability and repolarization. Rapid potassium removal may correct an immediate threat while subsequent hypokalemia introduces another risk. Phosphate depletion during prolonged treatment can contribute to respiratory and myocardial weakness, although a low measured concentration is not proof that it caused a particular episode.

The physiological lesson is to connect the electrocardiogram, ionized calcium, potassium, magnesium, phosphate, blood gases, and treatment timing. Electrolyte-mediated instability may occur when net removal is zero and will not be explained by a fluid balance chart.

51.10 Myocardial dysfunction may be missed by global pressure monitoring

Myocardial stunning denotes persistent, potentially reversible contractile dysfunction after a transient ischemic insult without requiring completed infarction. Dialysis can create conditions favoring myocardial oxygen supply–demand imbalance: lower coronary perfusion pressure, reduced diastolic filling time, elevated ventricular diastolic pressure, anemia, microvascular dysfunction, or limited coronary reserve. Regional contractile impairment can then reduce output and further compromise perfusion.

In a four-patient maintenance-hemodialysis PET study, McIntyre and colleagues linked regional wall-motion abnormalities to reductions in regional myocardial blood flow, despite absence of angiographically significant coronary disease. This supplies direct physiological support for an ischemic mechanism in that selected population. 11 It does not establish the prevalence or mechanism of all cardiac dysfunction in critical illness.

Small acute-kidney-injury studies extend the concern. Mahmoud and colleagues reported new regional strain abnormalities during intermittent dialysis in 11 patients. Slessarev and colleagues found new regional dysfunction in 10 of 11 critically ill patients after starting continuous therapy, despite stable global hemodynamics. 12 13 These uncontrolled studies identify a signal, not a reliable incidence estimate or proof that continuous therapy caused ischemic stunning in every affected segment.

Strain and wall motion are also load dependent. Changes in preload and afterload, sepsis-associated myocardial dysfunction, image quality, and the natural evolution of illness complicate interpretation. Demonstrating a new echocardiographic abnormality does not by itself prove reversible postischemic dysfunction. Nevertheless, stable MAP should not be used to exclude treatment-associated cardiac stress.

The practical inference is a broader assessment when perfusion deteriorates without an obvious volume explanation. New regional or global dysfunction, right ventricular failure, ischemia, and arrhythmia belong in that assessment. The evidence does not justify labeling every pressure fall “stunning” or prescribing a single cardioprotective dialysis strategy.

51.11 Modality and removal strategies must be judged by their actual evidence

Spreading an equivalent treatment over more time can reduce the hourly demands on plasma refill and slow solute shifts. Continuous and prolonged intermittent treatments therefore offer different temporal patterns from conventional intermittent dialysis. Interruptions, circuit changes, fluid inputs, and attempts to recover missed removal can make actual exposure quite different from the modality label.

Older randomized evidence, including the 360-patient HEMODIAFE trial using measures to optimize tolerance in both groups, did not demonstrate a survival difference between intermittent and continuous strategies. 14 More recent evidence prevents an unconditional equivalence claim. In ICRAKI, reported in 2026, 554 patients were randomized and 535 analyzed. Its composite of major adverse kidney events included death, RRT dependence, or serum creatinine more than 25% above baseline by day 90. The initial intermittent strategy did not demonstrate noninferiority for that outcome and was associated with higher mortality: 63.1% versus 53.5%, with a mortality hazard ratio of 1.30 (95% CI 1.04–1.63). The trial stopped early. Reported hypotension proportions were similar, so the outcome difference cannot simply be assigned to counted pressure events. 15

Net-removal rate is a separate exposure from clearance intensity. Secondary analyses of RENAL and other cohorts associate higher delivered net-removal rates with worse outcomes. Clinicians did not randomize those rates, and congestion, severity, treatment duration, and changing tolerance influence their selection. The reported categories are not validated safety boundaries. 16 17

Randomized removal-strategy evidence now exists. GO NEUTRAL randomized 55 vasopressor-treated patients to monitored active fluid removal or minimal removal. It improved 72-hour fluid balance in its selected modified intention-to-treat population, without a detected difference in hemodynamic episodes or mortality; it was a small proof-of-concept trial, not definitive evidence of survival benefit or universal safety. 18 RELIEVE-AKI randomized removal strategies in 97 patients but achieved limited separation in overall delivered rates, with frequent rescue treatment and physiological overrides. Its feasibility findings do not establish a preferred fixed rate. 19

51.12 Assess tolerance as a changing balance of perfusion and congestion

Consider a ventilated patient with septic AKI, edema, elevated venous pressures, and continuous therapy. For an illustrative hour, non-circuit inputs are 200 mL, urine and other measured losses total 40 mL, and net circuit removal is 120 mL. Whole-patient balance is 200 − 40 − 120 = +40 mL/h. The patient is still accumulating fluid despite active extraction. These constructed rates teach accounting, not treatment selection.

If pressure falls two minutes after connection, only 120 × (2/60) = 4 mL has been removed at that setting. This timing weakens an explanation based solely on cumulative extraction. Connection balance, vascular tone, cardiac function, rhythm, chemistry, ventilation, and unrelated deterioration deserve attention. Positive overall balance does not exclude vascular underfilling from redistribution, and later cumulative removal can introduce a new mechanism.

Suppose echocardiography shows preserved forward flow initially, but pressure falls with warmer blood and unchanged rhythm. A vascular explanation becomes more plausible. If output later falls, a valid passive leg raise increases it, and chamber filling decreases, loss of preload reserve becomes more plausible. If the right ventricle instead dilates with septal displacement after a ventilator change, additional fluid could aggravate the loading problem. These are physiological interpretations requiring corroboration, not automatic bedside algorithms.

Monitoring combines the timing and severity of pressure changes, support requirements, peripheral perfusion, serial flow or echocardiographic findings, temperature, chemistry, and delivered treatment. Relative blood-volume monitors infer concentration changes rather than measuring absolute circulating volume; bleeding, infusions, and redistribution complicate their interpretation. An ICU randomized study did not show that adding online blood-volume and temperature control reduced hypotension over its background dialysis regimen. 20

Tolerance also includes whether congestion is resolving without new organ hypoperfusion. A cause-specific change may involve removal intensity, treatment duration, composition, thermal balance, circulatory support, ventilation, or correction of another acute problem. Reassessment determines whether the predicted mechanism actually changed. Neither a fluid bolus for every pressure fall nor persistence with an unchanged removal plan follows from the physiology.

References

1. Bitker L, Bayle F, Yonis H, et al. Prevalence and risk factors of hypotension associated with preload-dependence during intermittent hemodialysis in critically ill patients. Critical Care. 2016;20:44. doi:10.1186/s13054-016-1227-3

2. Eastwood GM, Peck L, Young H, et al. Haemodynamic impact of a slower pump speed at start of continuous renal replacement therapy in critically ill adults with acute kidney injury: a prospective before-and-after study. Blood Purification. 2012;33:52–58. doi:10.1159/000333838

3. Fauchald P. Transcapillary colloid osmotic gradient and body fluid volumes in renal failure. Kidney International. 1986;29:895–900. doi:10.1038/ki.1986.83

4. Monnet X, Cipriani F, Camous L, et al. The passive leg raising test to guide fluid removal in critically ill patients. Annals of Intensive Care. 2016;6:46. doi:10.1186/s13613-016-0149-1

5. Fessler HE, Brower RG, Wise RA, Permutt S. Effects of positive end-expiratory pressure on the gradient for venous return. American Review of Respiratory Disease. 1991;143:19–24. doi:10.1164/ajrccm/143.1.19

6. Rodriguez M, Pederson JA, Llach F. Effect of dialysis and ultrafiltration on osmolality, colloid osmotic pressure, and vascular refilling rate. Kidney International. 1985;28:808–813. doi:10.1038/ki.1985.202

7. Rokyta R Jr, Matejovic M, Krouzecky A, et al. Effects of continuous venovenous haemofiltration-induced cooling on global haemodynamics, splanchnic oxygen and energy balance in critically ill patients. Nephrology Dialysis Transplantation. 2004;19:623–630. doi:10.1093/ndt/gfg615

8. MyTEMP Writing Committee. Personalised cooler dialysate for patients receiving maintenance haemodialysis (MyTEMP): a pragmatic, cluster-randomised trial. Lancet. 2022;400:1693–1703. doi:10.1016/S0140-6736(22)01805-0

9. Gabutti L, Bianchi G, Soldini D, Marone C, Burnier M. Haemodynamic consequences of changing bicarbonate and calcium concentrations in haemodialysis fluids. Nephrology Dialysis Transplantation. 2009;24:973–981. doi:10.1093/ndt/gfn541

10. Zarbock A, Küllmar M, Kindgen-Milles D, et al. Effect of regional citrate anticoagulation vs systemic heparin anticoagulation during continuous kidney replacement therapy on dialysis filter life span and mortality among critically ill patients with acute kidney injury: a randomized clinical trial. JAMA. 2020;324:1629–1639. doi:10.1001/jama.2020.18618

11. McIntyre CW, Burton JO, Selby NM, et al. Hemodialysis-induced cardiac dysfunction is associated with an acute reduction in global and segmental myocardial blood flow. Clinical Journal of the American Society of Nephrology. 2008;3:19–26. doi:10.2215/CJN.03170707

12. Mahmoud H, Forni LG, McIntyre CW, Selby NM. Myocardial stunning occurs during intermittent haemodialysis for acute kidney injury. Intensive Care Medicine. 2017;43:942–944. doi:10.1007/s00134-017-4768-2

13. Slessarev M, Salerno F, Ball IM, McIntyre CW. Continuous renal replacement therapy is associated with acute cardiac stunning in critically ill patients. Hemodialysis International. 2019;23:325–332. doi:10.1111/hdi.12760

14. Vinsonneau C, Camus C, Combes A, et al. Continuous venovenous haemodiafiltration versus intermittent haemodialysis for acute renal failure in patients with multiple-organ dysfunction syndrome: a multicentre randomised trial. Lancet. 2006;368:379–385. doi:10.1016/S0140-6736(06)69111-3

15. Gaudry S, Boubaya M, Louis G, et al; ICRAKI Trial Group. Intermittent or continuous renal-replacement therapy in acute kidney injury. New England Journal of Medicine. 2026;395:1677–1690. doi:10.1056/NEJMoa2608397

16. Murugan R, Kerti SJ, Chang CH, et al. Association of net ultrafiltration rate with mortality among critically ill adults with acute kidney injury receiving continuous venovenous hemodiafiltration: a secondary analysis of the Randomized Evaluation of Normal vs Augmented Level (RENAL) of Renal Replacement Therapy trial. JAMA Network Open. 2019;2:e195418. doi:10.1001/jamanetworkopen.2019.5418

17. Naorungroj T, Neto AS, Zwakman-Hessels L, et al. Early net ultrafiltration rate and mortality in critically ill patients receiving continuous renal replacement therapy. Nephrology Dialysis Transplantation. 2021;36:1112–1119. doi:10.1093/ndt/gfaa032

18. Bitker L, Dupuis C, Pradat P, et al. Fluid balance neutralization secured by hemodynamic monitoring versus protocolized standard of care in patients with acute circulatory failure requiring continuous renal replacement therapy: results of the GO NEUTRAL randomized controlled trial. Intensive Care Medicine. 2024;50:2061–2072. doi:10.1007/s00134-024-07676-1

19. Murugan R, Talisa V, Chang CH, et al. Alternative net ultrafiltration rate strategies in acute kidney injury: a feasibility randomized clinical trial. Critical Care. 2026;30:228. doi:10.1186/s13054-026-06040-7

20. du Cheyron D, Terzi N, Seguin A, et al. Use of online blood volume and blood temperature monitoring during haemodialysis in critically ill patients with acute kidney injury: a single-centre randomized controlled trial. Nephrology Dialysis Transplantation. 2013;28:430–437. doi:10.1093/ndt/gfs124