Timing of kidney replacement therapy in critically ill patients: A call to shift the paradigm in the era of artificial intelligence

Abstract

Acute kidney injury (AKI) is a common condition in intensive care units (ICUs) and is associated with high mortality rates, particularly when kidney replacement therapy (KRT) becomes necessary. The optimal timing for initiating KRT remains a subject of ongoing debate. Emerging tools and methodologies, such as machine learning and advanced sub-phenotyping, offer promising insights into refining AKI management. Moving beyond the traditional “early” versus “delayed” paradigm and the heavy reliance on serum creatinine measurements, there is an opportunity to develop treatment strategies tailored to the unique pathophysiological and medical context of each patient. Such individualized approaches could potentially improve outcomes and transform AKI management into ICUs. However, KRT is not without risks. Hemodynamic instability poses a significant challenge, complicating the management of critically ill patients. The selection of a KRT modality—whether intermittent hemodialysis, continuous renal replacement therapy, or peritoneal dialysis—introduces additional complexities. Each modality has distinct advantages and limitations, requiring a careful, patient-specific approach to ensure optimal care. This decision-making process is further influenced by the availability of specialized equipment and trained personnel, resources that may be limited in some settings. Notably, current evidence does not demonstrate a clear survival or recovery benefit from the early initiation of dialysis. This narrative review explores the ongoing debates surrounding KRT timing and methodology, highlighting the importance of adopting patient-centric, individualized strategies to navigate the evolving landscape of AKI management.

Keywords

Acute kidney injury kidney replacement therapy critical care timing patient-centered care

Introduction

Acute kidney injury (AKI), characterized by its significant heterogeneity, has profound implications for both short- and long-term patient outcomes.¹ Within the intensive care unit (ICU) environment, AKI manifests at an alarming prevalence, affecting 30–60% of admissions and mortality rates over 50% when necessitating kidney replacement therapy (KRT).^2,3 While current clinical practice guidelines articulate the initiation of KRT in addressing severe AKI-related complications, such as hyperkalemia, metabolic acidosis, and fluid overload-induced pulmonary edema, there remains a lack of consensus on the optimal timing for KRT initiation when these severe complications are not present. This lacuna in the guidelines has instigated extensive deliberations and divergence of opinions among ICU practitioners.^4–8 The confluence of this lack of directive clarity and apprehensions regarding the potential adverse events of KRT has culminated in a cautious approach to KRT initiation by many clinicians.⁸ Expanding the introduction further, it is crucial to balance the benefits and risks associated with KRT. While KRT can be life-saving in acute settings, its adverse effects, such as hemodynamic instability and the challenge of choosing the appropriate modality, must be considered. Major clinical trials have provided insights but have also highlighted the complexities and nuances of KRT management in different patient populations.

In this narrative review, we aim to examine the current evidence on the ongoing controversies around KRT timing in critically ill patients with AKI, discuss why solely relying on the “early vs. delayed” paradigm may be insufficient, and consider how emerging biomarkers and machine learning (ML) approaches might inform more individualized care.

Methods

In accordance with the Scale for the Assessment of Narrative Review Articles (SANRA)⁹ and to ensure methodological rigor, a structured literature search was conducted using PubMed, Scopus, and Web of Science to identify studies published from January 2018 through July 2024. We used the following search terms (including both free-text and, where applicable, MeSH terms): “acute kidney injury,” “AKI,” “kidney replacement therapy,” “renal replacement therapy,” “timing of dialysis,” “early initiation,” “late initiation,” “indications,” “sub-phenotyping,” “critical care,” and “ICU.” To ensure comprehensive coverage, we also screened reference lists from key research articles, reviews, and meta-analyses for additional sources. Only peer-reviewed English language-publications were included. Conference abstracts, commentaries without primary data, and papers focusing exclusively on pediatric populations were excluded.

An initial screening based on titles and abstracts was conducted by the first author, and it was followed by full-text review of potentially relevant articles by two authors. Studies were selected if they addressed the timing, modality choice, or individualized management of KRT in adult ICU patients with AKI, with particular emphasis on hemodynamic implications, outcome measures, and novel methodologies (e.g. ML-based approaches). Where multiple publications from the same research group overlapped, the most recent and comprehensive article was chosen.

Acute kidney injury in the intensive care unit

Prevalence and setting

AKI is characterized by a rapid decline in estimated glomerular filtration rate, signifying impaired kidney function.¹⁰ AKI occurring in the ICU stands as an independent risk factor for mortality with rates ranging from 40 to 55% exceeding the mortality rates observed in serious conditions such as myocardial infarction, sepsis in the absence of AKI, and acute respiratory distress syndrome.^10,11 The pathogenesis of AKI in ICU patients is multifactorial. Disturbances in glomerular blood flow can arise from hypovolemic and ischemic conditions exacerbated by factors such as periglomerular shunting and endothelial dysfunction.¹² Additionally, inflammatory responses triggered by damage-associated and pathogen-associated molecular patterns (DAMPs and PAMPs) lead to immune cell activation and subsequent harmful effects. Tubular disruptions through processes such as ferroptosis and mitochondrial dysfunction add to AKI's varied presentations.¹³ The overall complexity is further compounded by organ crosstalk and nephrotoxic effects of some medications.¹⁴

AKI in the context of sepsis is particularly significant, as sepsis is the most common cause of AKI in ICU settings and the focus of many clinical trials. The inflammatory and hemodynamic alterations induced by sepsis compound the risk of AKI, leading to worse outcomes. Studies have shown that septic AKI is associated with higher mortality and prolonged hospital stays compared to non-septic AKI.² Following AKI onset, its repercussions extend beyond immediate threats. Long-term AKI predisposes patients to recurrent episodes, increased chances of hospital readmission, cardiovascular complications, oncological issues, and hastened progression to chronic kidney disease.¹¹

Basic management

AKI management involves carefully considering volume status and hemodynamic management. Both volume depletion and fluid overload can harm kidney function, and striking the right balance is crucial.¹⁵ Intravenous fluids can be beneficial in cases of true hypovolemia, but diagnosing this condition is challenging. Empiric fluid resuscitation based solely on serum creatinine (sCr) levels or oliguria can be misguided, especially in patients with fluid intolerance syndrome. Although fluid responsiveness tests can identify patients who are unlikely to benefit from fluid administration, the presence of fluid responsiveness does not necessarily mean that fluid therapy is indicated or safe. Individualized care, including monitoring blood pressure and cardiac function, is crucial for AKI.¹⁶ Vasopressors are essential in managing hypotension when volume resuscitation alone is insufficient or in severe vasodilatory shock, including septic shock, where they help preserve renal perfusion only after intravascular volume is restored. Norepinephrine is first-line, but vasopressin may better protect kidneys by minimizing afferent arteriolar constriction while constricting efferent arterioles. Other vasopressors offer inconsistent kidney outcomes. Goal-directed therapy, combining fluid and vasopressor titration, can further prevent AKI.¹⁷ Nephrotoxic management of AKI involves the prompt identification and discontinuation of all potentially nephrotoxic drugs and agents. Avoidance of nephrotoxic agents is a crucial element in preventing AKI, especially if associated with risk factors (pre-existing CKD, dehydration, other nephrotoxic medications). However, it is necessary to emphasize the importance of adapting dosages to renal function in order to avoid underdosing or overdosing which can lead to renal toxicity and the occurrence of AKI. When such drugs are essential for treatment, prophylactic volume expansion with crystalloids should be used to prevent AKI, with careful monitoring of the drug concentrations and kidney function. Contrast agents in the context of investigations or interventions should not be delayed for potential preventative measures.¹⁸

While these management approaches are vital, an equally critical intervention in AKI is the initiation of KRT. The dilemma is not just about deciding to initiate KRT but more pertinently, when to initiate KRT. A plethora of research exists, yet a clear consensus has yet to be reached. The discrepancies in various studies underline the need to further explore this topic.

Evidence on KRT timing: What do recent research reveal?

A large number of recent systematic reviews and meta-analyses, summarized in Table 1, revealed a predominant result on the impact of early KRT initiation. A significant majority of these studies, hailing from diverse global locations and encompassing varying numbers of randomized controlled trials (RCTs), converge on the notion that early KRT initiation does not confer a notable survival advantage in critically ill patients with AKI.^18–34 This consensus remains robust, even when studies dissect the results in varying populations, such as those with septic AKI or those in surgical ICUs.^20,23 However, nuances emerge, most prominently from the Cochrane reviews by Fayad et al., where an initial 2018 study suggested the potential benefits of early KRT initiation,¹⁹ but only for a subsequent 2022 study to echo the prevailing sentiment.³⁰ Examination of the previously mentioned meta-analyses necessitated a comprehensive review of the primary RCTs from 2016 to 2020, providing an in-depth perspective on the dynamics of early KRT initiation in patients with AKI (Table 2). Published in esteemed academic journals, these seminal studies, while united in their focus, manifest marked heterogeneity in their definitions of “early” initiation, reflecting the complexities and nuances of the issue at hand.^35–40 For instance, “early” in AKIKI STUDY is delineated as a mere 6-h post-diagnosis of Kidney Disease: Improving Global Outcomes (KDIGO) stage 3 AKI,³⁵ while the STARRT AKI study perceives early as within 12 h following a KDIGO stage 2 or 3 diagnosis.³⁸ Notably, the ELAIN study, which sets its early initiation threshold at 8 h for stage 2 AKI, stands apart by indicating a benefit associated with the early initiation strategy³⁶; a finding that contrasts with the majority of other studies, such as the FST, which defines “early” within 6 h of any KDIGO stage of AKI without a response to the Furosemide Stress Test.³⁹ This variability in defining “early” illuminates the absence of a standardized, universally accepted criterion, a challenge that significantly impacts clinical decision-making and research outcomes synthesis. In the 2021, AKIKI 2 study conducted across 39 French ICUs, comparing “delayed” versus “more-delayed” KRT initiation strategies revealed that postponing KRT for an extended duration was associated with an increased risk of 60-day mortality.⁴¹ Beyond these definitions, the interventions present their own tapestry of diversity (Table 2), with modalities mostly left to the physician's discretion, ranging from intermittent hemodialysis and Continuous Veno-venous Hemodiafiltration to Continuous Veno-venous hemofiltration and Sustained Low-Efficiency Dialysis.^35–41 This landscape of findings underscores the multifaceted nature of the impact of KRT initiation and accentuates the need for nuanced interpretations tailored to specific patient profiles.

Table 1.

Recent metanalysis on the timing of initiating kidney replacement therapy in AKI.

Author/Year	Journal	Country	Type of included studies	Number of included studies	Enrollment	Author's conclusions
Fayad et al. 2018¹⁷	Cochrane Database of Systematic Reviews	Argentina	RCTs	5	1084	“early RRT may reduce the risk of death and may improve the recovery of kidney function”
Cui et al. 2018¹⁸	BioMed Research International	China	RCTs	4	355	“Early initiation of RRT had no beneficial impacts on outcomes in patients with CSA-AKI.”
Moreira et al. 2018¹⁹	Rev Bras Ter Intensiva	Brazil	RCTs	6	1292	“Early initiation of renal replacement therapy is not associated with improved survival”
Xiao et al.2019²⁰	PLOS ONE	China	RCTs	11	2242	“early compared with late initiation of RRT is not associated with favorable mortality outcomes”
Y. Li et al. 2019²¹	Medicine	China	RCTs	5	900	“early initiation of CRRT could not reduce mortality in patients with septic AKI”
Gaudry et al. 2020²²	The Lancet	France	RCTs	10	2143	“The timing of RRT initiation does not affect survival in critically ill patients”
Zhang et al. 2020²³	Renal Failure	China	RCTs	18	2856	“no evidence to show any benefit of an early RRT strategy”
Bhatt et al. 2021²⁴	Nephron	India	RCTs	14	5234	“There is no added benefit of early initiation of RRT in patient with AKI”
Pan et al. 2021²⁵	Critical Care	Taiwan	RCTs	10	4753	“benefit from accelerated RRT initiation regarding all-cause mortality and being eventually free of dialysis only if they were surgical ICU…”
X. Li et al. 2021²⁶	Critical Care	China	RCTs	11	5086	“early initiation of RRT was not associated with survival benefit in critically ill patients with AKI”
Xiao et al. 2022²⁷	Frontiers in medicine	China	RCTs	9	5423	“early RRT had no significant survival benefit for ill patients with AKI”
Fayad et al. 2022²⁸	Cochrane Database of Systematic Reviews	Argentina	RCTs	12	4880	“early KRT has no beneficial effect on death and may increase the recovery of kidney function”
Castro et al. 2022²⁹	Clinical Kidney Journal	Portugal	RCTs	13	5193	“Early RRT initiation does not improve the 28-day and overall mortality”
Lan et al. 2022³⁰	Medicine	China	RCTs	12	5279	“Accelerated RRT does not reduce the risk of death and does not improve the recovery of kidney function”
Y. Li 2022³¹	Ther Apher Dial	China	RCTS	15	5395	“Early initiation of RRT cannot improve the prognosis and benefit patients”
J.-H. Li 2023³²	Euro Rev Med Pharmacol Sci	China	RCTs	15	5625	“early RRT initiation was not associated with survival benefits”

The authors’ conclusions are quoted from the listed sources.

RCTs: randomized controlled trials; KRT/RRT: kidney /renal replacement therapy; CSA-AKI: cardiac surgery-associated acute kidney injury; ICU: intensive care unit.

Table 2.

Comparison of main recent trials in timing of initiating kidney replacement therapy in the ICU.

	AKIKI³³	ELAIN³⁴	IDEAL ICU³⁵	STARRT AKI³⁶	FST³⁷	EARLYRRT³⁸
Journal / year	The New England journal of medicine 2016	JAMA 2016	The New England journal of medicine 2018	The New England journal of medicine 2020	Critical Care 2018	Journal of Critical care 2017
Definition of early	Within 6 h of diagnosis of stage 3 AKI	within 8 h of diagnosis of stage 2 AKI	Within 12 h of failure-stage kidney disease with early-stage septic shock and without life-threatening complications	within 12 h of KDIGO stage 2 or 3	Within 6 h of any KDIGO stage of AKI and no response to furosemide stress test	within 12 h of AKI according to RIFLE and pNGAL ≥ 400 ng/mL
Intervention Early delayed	56% IHD, 44% CRRT 55% IHD, 45% CRRT	100% CVVHDF 100% CVVHDF	43% IHD, 57% CRRT 45% IHD, 55% CRRT	27% IHD, 68.4% CRRT, 4.6% SLED 25.3% IHD, 70.6% CRRT, 4.1% SLED	100% CVVH 100% CVVH	100% CRRT 100% CRRT
Primary outcome	Overall survival at 60 days	90-day mortality from all causes	death at 90 days	death at 90 days	28-day all-cause mortality	death upon day 28 after enrollment

AKI: acute kidney injury; KDIGO: Kidney Disease Improving Global Outcome; IHD: intermittent hemodialysis; CRRT: continuous renal replacement therapy; CVVHDF: continuous Veno-venous hemodiafiltration; SLED: slow low-efficiency dialysis; ICU: intensive care unit.

Dissecting the problematic of AKI

The Kidney Disease: Improving Global Outcomes definition

The KDIGO group established a standard AKI definition, emphasizing the changes in sCr levels over a week or the onset of oliguria. This consensus has played a key role in shaping research on AKI.⁷ While sCr remains a conventional marker for kidney function, its use as a sole determinant for initiating KRT in critically ill patients is fraught with limitations. The inherent variability in creatinine production in critically ill individuals coupled with fluctuating fluid balances makes its interpretation challenging. However, the advent of “KDIGO AKI stages” based on sCr and urine output fails to capture the nuanced etiological, anatomical, and cellular responses of kidney injuries.⁴² Such categorizations, particularly in critically ill patients, risk oversimplifying a multifaceted condition, which could lead to suboptimal clinical decisions. In the context of KRT for critically ill patients, there is an urgent need to move beyond the sCr-centric paradigm.^43,44

A dichotomous approach: Early vs delayed initiation of dialysis

AKI is not a singular, homogenous entity, but rather a loose collection of syndromes, each with unique pathophysiological underpinnings. Recognizing this heterogeneity is crucial when formulating therapeutic strategies and interpreting clinical research.¹² The prevailing dichotomic classification of KRT initiation into “early” versus “delayed” categories, while offering some semblance of clinical simplicity, becomes particularly limited in the context of AKI's diverse spectrum. In a sense, such binary classification seeks to fit a multifaceted clinical phenomenon into a one-size-fits-all framework, which can be problematic.⁴⁵ The lack of universally consistent definitions for “early” and “delayed” across studies amplifies these challenges. For instance, while AKIKI and STARRT-AKI studies employ differing timelines for “early” initiation, their definitions become even more convoluted when considered in meta-analyses.^35,38 Collating data from studies with divergent definitions dilutes the precision of the findings, potentially hindering scientific progress and yielding less actionable insights (Figure 1). Approaching a therapy as sensitive as KRT with dichotomous reasoning for a heterogeneous clinical syndrome, such as AKI, can lead to oversimplification. This may culminate in premature interventions for some patients and undue delays in essential treatment for others. The variability inherent in AKI mandates an approach that goes beyond dichotomies, emphasizing the need for individualized, patient-centric strategies.

Figure 1.

“Early” definitions in major recent clinical trials (based on Table 2).

Can biomarkers help in initiating KRT?

sCr has been a conventional marker for kidney function, yet its use as the sole determinant for initiating KRT in critically ill patients is fraught with limitations. The variability in creatinine production in critically ill individuals, coupled with fluctuating fluid balances, makes its interpretation challenging. As such, relying exclusively on sCr can lead to suboptimal clinical decisions.⁴⁶ Therefore, several urine and plasma biomarkers have been explored in the setting of AKI, and a three-tiered biomarker framework has emerged, encompassing functional, stress, and damage biomarkers.

Functional biomarkers (for example, cystatin C and proenkephalin A) track glomerular filtration rather than tubular injury. They offer a more sensitive indication of renal function changes than sCr and can detect subclinical AKI in vulnerable populations with reduced muscle mass.

Stress biomarkers (including tissue inhibitor of metalloproteinases-2 [TIMP-2], insulin-like growth factor binding protein 7 [IGFBP7], and Dickkopf-3 [DKK3]) signal imminent tubular injury before permanent damage occurs. Their elevations forecast a high risk of AKI progression, guiding early hemodynamic optimization and nephroprotective strategies.

Damage biomarkers (for instance, neutrophil gelatinase-associated lipocalin [NGAL], C-C motif chemokine ligand 14 [CCL14], kidney injury molecule-1 [KIM-1], interleukin-18 [IL-18], liver-type fatty acid–binding protein [L-FABP], and N-acetyl-β-D-glucosaminidase [NAG]) signify established renal injury. These markers help assess AKI severity, elucidate the underlying cause (e.g. sepsis, ischemia, nephrotoxins), and determine the urgency of therapeutic interventions.^46,47

However, their accuracy in anticipating the requirement for KRT at the bedside remains under investigation. For instance, a previous systematic review and meta-analysis included 41 studies evaluating 13 different biomarkers showed promising results in predicting the need for KRT but the strength of evidence could not recommend their routine use in clinical decision-making regarding KRT initiation.⁴⁸ A further meta-analysis from 2020 proposed threshold values for both urinary and plasma NGAL that correspond to an increased risk of severe AKI or AKI requiring dialysis. These cutoffs could aid clinical research and practice by highlighting patients at greater renal risk.⁴⁹ Moreover, exosomes are small messengers that help cells communicate and play a role in both damage and healing in AKI. Researchers are using multi-omics (studying genes, proteins, and metabolites together) to better understand AKI, detect it earlier, and find new treatments. There is also growing interest in using exosomes to deliver medicine directly to the kidneys, which could improve recovery and reduce complications.⁵⁰

The furosemide stress test has potential, especially if integrated with others tests and clinical features in critically ill patients.^51–53 The Doppler-based resistive index has not been successful in differentiating transitory AKI from persistent AKI.^54,55 Finally, biomarkers may prove valuable when used in conjunction with other diagnostic methods and clinical features.⁴⁶

Emerging perspectives: Phenotyping and machine learning

The management and understanding of AKI are witnessing significant advancements propelled by the integration of clinical insights, novel sub-phenotyping methods, and innovative ML techniques. Sitbon et al. highlighted the significance of clinical judgment in predicting KRT needs in patients with AKI. Their study underscored that model incorporating physician judgment, especially in conjunction with metrics such as SOFA score, consistently outperformed those that did not.⁵⁶ Additionally, Pike et al. highlighted that conventional clinical models that consider parameters such as age and mean arterial pressure are valuable for predicting KRT outcomes.⁵⁷ In a more recent study, the authors applied unsupervised consensus clustering, a ML technique, to discern distinct sub-phenotypes among patients with dialysis-requiring sepsis-associated AKI, showing differing clinical outcomes. Intriguingly, one identified subgroup; with a higher severity of acute illness; displayed a heightened risk of mortality and a diminished chance of being free from KRT compared to others, even though they were younger and had fewer comorbidities.⁵⁸ This ML enabled sub-phenotyping resonated in other studies, further drawing attention to its potential to elucidate AKI's intricacies.^59,60 While the value of prognostic enrichment in AKI management is recognized, the overarching goal of sub-phenotyping diverse disorders leans towards predictive enrichment or tailoring specific treatments to individuals; however, these sub-phenotyping tools have not made it to clinical practice because of challenges such as wide-scale validation, timely availability of biomarkers, and model complexities⁴⁵

Beyond the realm of sub-phenotyping, recent advances in artificial intelligence by ML have shown promise in providing accurate, data-driven predictions across a range of clinical domains, including sepsis management, ventilator weaning, and acute organ failure.^61–64 By integrating numerous clinical variables (such as laboratory parameters, physiological measures, and patient demographics), ML models can potentially detect subtle patterns that might be overlooked during routine clinical assessment. In the context of AKI, emerging evidence suggests ML-based predictive tools could offer earlier and more precise estimation of which patients are likely to require KRT, thus enhancing resource allocation and patient care.⁶⁵ A proof-of-concept study demonstrated the feasibility of using a recurrent neural network to predict, up to 48 h in advance, the need for KRT, admission to intensive care, or death in patients with AKI. Trained on over 21,000 AKI episodes (25 million data points), the model exceeded 99% in both specificity and negative predictive value. Its positive predictive value for dialysis reached 62%.⁶⁶ Collectively, these studies highlight the evolving trajectory of research and management of AKI. The combination of clinical insights, advanced sub-phenotyping, and ML promises the future of more tailored and precise interventions for AKI.

Perspectives: From dilemma to direction

Sean Bagshaw et al. in 2009 proposed an algorithm for starting KRT in adult critically ill patients, giving a starting point to a more patient-centric approach. In this strategy, the authors integrate two important aspects that may influence the decision of initiating kidney support, especially “dynamic factors” like the severity of the illness by a rapid progression of kidney and non-kidney organ dysfunction. In addition, there is the need for co-interventions that could increase kidney striving, such as surgery, large fluid volumes, mechanical ventilation, and corticosteroids, etc.⁶⁷ In 2012, Ostermann and colleagues proposed another algorithm to guide decision-making regarding AKI-KRT that emphasizes the importance of regular evaluation of patients’ features and trends.⁶⁸ The latter was readapted by adding the notion of kidney baseline function and reserve and also highlighting the need for consideration of all patients’ trajectories rather than relying only on laboratory threshold values.⁶⁹

The concept of a balance between kidney capacity and metabolic demands has been suggested to be taken into consideration when deciding to initiate KRT. However, the exact method used to assess this balance at the bedside is still unclear.^7,43,70 Another major point of focus is the implications of patient and family in such decisions by delivering clear simple language concerning the risks and potential benefits of such therapy and overall prognosis.⁷ A more recent proposition by Bagshaw et al. summarized the prevailing approach of the “watch and wait strategy” by incorporating all patient's features.⁷¹ These algorithms have the merit of proposing a patient-centered approach, taking into consideration key factors (Figure 2), such as the severity of causal disease, kidney reserve, clinical and biological trends, overall prognosis, and patient and family preferences rather than the urine output/ sCr-centric paradigm.^67–71 Nevertheless, the vast heterogeneity of AKI syndrome still stands against precision medicine approaches. The challenge that remains before the scientific community is to identify relevant clinical phenotypes based on “evidence-based medicine” and thus develop distinct management algorithms.^72,73

Figure 2.

Assessment metrics for KRT initiations in AKI in the absence of life-threatening complications. AKI: acute kidney injury; KRT: kidney replacement therapy.

As a narrative review, our search and selection processes were not entirely systematic, increasing the possibility of selection bias. We did not perform a formal risk-of-bias assessment or quantitative synthesis of the available studies, which limits our ability to draw firm conclusions regarding the strength of the evidence. Moreover, we focused primarily on frequently cited and recent research; thus, relevant but less prominent studies may be underrepresented. These factors should be considered when interpreting our findings.

Conclusion

AKI in the ICU underscores the critical need to refine KRT initiation strategies. Current data question the presumed benefits of an “early” approach, advocating instead for individualized, evidence-based interventions that account for the complexity and heterogeneity of AKI. While many pivotal trials show no definitive advantage of early over delayed KRT, their methodological inconsistencies make broad generalizations risky. Even if consensus emerges regarding the optimal timing of KRT, additional questions on the modality (continuous vs. intermittent) and the appropriate timing to discontinue dialysis remain unresolved. Continued advancements in biomarkers, ML and sub-phenotyping promise new opportunities for personalized diagnosis and treatment. Ultimately, a paradigm shift is needed; one that moves beyond rigid dichotomies and leverages innovative technologies and clinical expertise to deliver more precise, patient-centered AKI care.

Footnotes

ORCID iDs

Es-saad Ounci

Youssef Motiaa

Ethical considerations

Not applicable, because this article does not contain any studies on human or animal subjects.

Author contributions

SE contributed to concept and design, literature search, article writing and drafting. AD and RA involved in literature search and review. YM and SM contributed to critical revision. All authors approved the final version of the article submitted for publication.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Verma

Kellum

. Defining acute kidney injury. Crit Care Clin 2021; 37: 251–266.

Hoste

EAJ

Kellum

Selby

, et al. Global epidemiology and outcomes of acute kidney injury. Nat Rev Nephrol 2018; 14: 607–625.

Faubel

Shah

. Immediate consequences of acute kidney injury: the impact of traditional and nontraditional complications on mortality in acute kidney injury. Adv Chronic Kidney Dis 2016; 23: 179–185.

Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO Clinical practice guideline for acute kidney injury. Kidney Int Suppl 2012; 2: 1–138.

Vinsonneau

Allain-Launay

Blayau

, et al. Renal replacement therapy in adult and pediatric intensive care: recommendations by an expert panel from the French intensive care society (SRLF) with the French society of anesthesia intensive care (SFAR) French group for pediatric IntensiveCare emergencies (GFRUP) the French dialysis society (SFD). Ann Intensive Care 2015; 5: 58.

National Institute for Health and Care Excellence (NICE). Acute kidney injury: prevention, detection and management. 2019. NICE Guideline, No. 148. Accessed July 17, 2021. https://www.ncbi.nlm.nih.gov/books/NBK552160/

Ostermann

Bellomo

Burdmann

, et al. Controversies in acute kidney injury: conclusions from a kidney disease: improving global outcomes (KDIGO) conference. Kidney Int 2020; 98: 294–309.

Nanchal

Subramanian

Karvellas

, et al. Guidelines for the management of adult acute and acute-on-chronic liver failure in the ICU: cardiovascular, endocrine, hematologic, pulmonary and renal considerations: executive summary. Crit Care Med 2020; 48: 415–419.

Baethge

Goldbeck-Wood

Mertens

. Sanra—a scale for the quality assessment of narrative review articles. Res Integrity Peer Rev 2019; 4: 5.

10.

Griffin

Liu

Teixeira

. Critical Care Nephrology: Core Curriculum 2020. Am J Kidney Dis 2020; 75: 435–452.

11.

James

Bhatt

Pannu

, et al. Long-term outcomes of acute kidney injury and strategies for improved care. Nat Rev Nephrol 2020; 16: 193–205.

12.

Kellum

Prowle

. Paradigms of acute kidney injury in the intensive care setting. Nat Rev Nephrol 2018; 14: 217–230.

13.

Pickkers

Darmon

Hoste

, et al. Acute kidney injury in the critically ill: an updated review on pathophysiology and management. Intensive Care Med 2021; 47: 835–850.

14.

Ounci

Boukabous

Bkiyar

, et al. Acute kidney injury in critically ill patients with COVID-19: prevalence, risk factors and mortality in eastern Morocco. J Nephrol 2022; 35: 2383–2386.

15.

Kellum

Romagnani

Ashuntantang

, et al. Acute kidney injury. Nat Rev Dis Primers 2021; 7: 52.

16.

Argaiz

Rola

Haycock

, et al. Fluid management in acute kidney injury: from evaluating fluid responsiveness towards assessment of fluid tolerance. Eur Heart J Acute Cardiovasc Care 2022; 11: 786–793.

17.

Tamargo

Hanouneh

Cervantes

. Treatment of acute kidney injury: a review of current approaches and emerging innovations. J Clin Med 2024; 13: 2455.

18.

Joannidis

, et al. Prevention of acute kidney injury and protection of renal function in the intensive care unit: update 2017: expert opinion of the working group on prevention, AKI section, European society of intensive care medicine. Intensive Care Med 2017; 43: 730–749. Epub 2017 Jun 2. PMID: 28577069; PMCID: PMC5487598.

19.

Fayad

AII

Buamscha

Ciapponi

. Timing of renal replacement therapy initiation for acute kidney injury. Cochrane Database Syst Rev 2018; 12: CD010612.

20.

Cui

Tang

Chen

, et al. Impact of early versus late initiation of renal replacement therapy in patients with cardiac surgery-associated acute kidney injury: meta-analysis with trial sequential analysis of randomized controlled trials. Biomed Res Int 2018; 2018: 6942829.

21.

Moreira

Palomba

Chaves

RCF

, et al. Early versus delayed initiation of renal replacement therapy for acute kidney injury: an updated systematic review, meta-analysis, meta-regression and trial sequential analysis of randomized controlled trials. Rev Bras Ter Intensiva 2018; 30: 376–384.

22.

Xiao

Jia

, et al. Early versus late initiation of renal replacement therapy for acute kidney injury in critically ill patients: a systematic review and meta-analysis. PLoS One 2019; 14: e0223493.

23.

Zhang

. Timing of continuous renal replacement therapy in patients with septic AKI: a systematic review and meta-analysis. Medicine (Baltimore) 2019; 98: e16800.

24.

Gaudry

Hajage

Benichou

, et al. Delayed versus early initiation of renal replacement therapy for severe acute kidney injury: a systematic review and individual patient data meta-analysis of randomised clinical trials. Lancet 2020; 395: 1506–1515.

25.

Zhang

Chen

Tang

, et al. Timing of initiation of renal replacement therapy in acute kidney injury: an updated meta-analysis of randomized controlled trials. Ren Fail 2020; 42: 77–88.

26.

Bhatt

Das

Satapathy

. Early versus late initiation of renal replacement therapy: have we reached the consensus? An updated meta-analysis. Nephron 2021; 145: 371–385.

27.

Pan

Chen

Tsai

, et al. Accelerated versus standard initiation of renal replacement therapy for critically ill patients with acute kidney injury: a systematic review and meta-analysis of RCT studies. Crit Care 2021; 25: 5.

28.

Liu

Mao

, et al. Timing of renal replacement therapy initiation for acute kidney injury in critically ill patients: a systematic review of randomized clinical trials with meta-analysis and trial sequential analysis. Crit Care 2021; 25: 15.

29.

Xiao

Cheng

, et al. The efficacy and safety of early renal replacement therapy in critically ill patients with acute kidney injury: a meta-analysis with trial sequential analysis of randomized controlled trials. Front Med (Lausanne) 2022; 9: 820624.

30.

Fayad

Buamscha

Ciapponi

. Timing of kidney replacement therapy initiation for acute kidney injury. Cochrane Database Syst Rev 2022; 11: CD010612.

31.

Castro

Relvas

Gameiro

, et al. The impact of early versus late initiation of renal replacement therapy in critically ill patients with acute kidney injury on mortality and clinical outcomes: a meta-analysis. Clin Kidney J 2022; 15: 1932–1945.

32.

Lan

Lai

Chang

, et al. Accelerated-strategy renal replacement therapy for critically ill patients: a systematic review and meta-analysis. Medicine (Baltimore) 2022; 101: e29747.

33.

Zhang

, et al. Timing of initiation of renal replacement therapy for patients with acute kidney injury: a meta-analysis of RCTs. Ther Apher Dial 2023; 27: 207–221.

34.

Cai

Wang

, et al. Early strategy vs. late initiation of renal replacement therapy in adult patients with acute kidney injury: an updated systematic review and meta-analysis of randomized controlled trials. Eur Rev Med Pharmacol Sci 2023; 27: 6046–6057.

35.

Gaudry

Hajage

Schortgen

, et al. AKIKI study group. Initiation strategies for renal-replacement therapy in the intensive care unit. N Engl J Med 2016; 375: 122–133.

36.

Zarbock

Kellum

Schmidt

, et al. Effect of early vs delayed initiation of renal replacement therapy on mortality in critically ill patients with acute kidney injury: the ELAIN randomized clinical trial. JAMA 2016; 315: 2190–2199.

37.

Barbar

Clere-Jehl

Bourredjem

, et al. IDEAL-ICU trial investigators and the CRICS TRIGGERSEP network. Timing of renal-replacement therapy in patients with acute kidney injury and sepsis. N Engl J Med 2018; 379: 1431–1442.

38.

STARRT-AKI Investigators; Canadian Critical Care Trials Group; Australian and New Zealand Intensive Care Society Clinical Trials Group; United Kingdom Critical Care Research Group; Canadian Nephrology Trials Network; Irish Critical Care Trials Group; Bagshaw

Wald

Adhikari

NKJ

, et al. Timing of initiation of renal-replacement therapy in acute kidney injury. N Engl J Med 2020; 383: 240–251. Erratum in: N Engl J Med. 2020 Jul 15.

39.

Lumlertgul

Peerapornratana

, et al. FST study group. Early versus standard initiation of renal replacement therapy in furosemide stress test non-responsive acute kidney injury patients (the FST trial). Crit Care 2018; 22: 101.

40.

Srisawat

Laoveeravat

Limphunudom

, et al. The effect of early renal replacement therapy guided by plasma neutrophil gelatinase associated lipocalin on outcome of acute kidney injury: a feasibility study. J Crit Care 2018; 43: 36–41.

41.

Gaudry

Hajage

Martin-Lefevre

, et al. Comparison of two delayed strategies for renal replacement therapy initiation for severe acute kidney injury (AKIKI 2): a multicentre, open-label, randomised, controlled trial. Lancet 2021; 397: 1293–1300.

42.

Barasch

Zager

Bonventre

. Acute kidney injury: a problem of definition. Lancet 2017; 389: 779–781.

43.

Bouchard

Mehta

. Timing of kidney support therapy in acute kidney injury: what are we waiting for? Am J Kidney Dis 2022; 79: 417–426.

44.

Sparrow

Swan

Moore

, et al. Disparate outcomes observed within kidney disease: improving global outcomes (KDIGO) acute kidney injury stage 1. Kidney Int 2019; 95: 905–913.

45.

Vaara

Bhatraju

Stanski

, et al. Subphenotypes in acute kidney injury: a narrative review. Crit Care 2022; 26: 251.

46.

Ostermann

Zarbock

Goldstein

, et al. Recommendations on acute kidney injury biomarkers from the acute disease quality initiative consensus conference: a consensus statement. JAMA Netw Open. 2020; 3: e2019209. Erratum in: JAMA Netw Open. 2020 Nov 2;3(11):e2029182. doi: https://doi.org/10.1001/jamanetworkopen.2020.29182.

47.

Ostermann

Legrand

Meersch

, et al. Biomarkers in acute kidney injury. Ann Intensive Care 2024; 14: 145.

48.

Klein

Brandtner

Lehner

, et al. Biomarkers for prediction of renal replacement therapy in acute kidney injury: a systematic review and meta-analysis. Intensive Care Med 2018; 44: 323–336.

49.

Albert

Zapf

Haase

, et al. Neutrophil gelatinase-associated lipocalin measured on clinical laboratory platforms for the prediction of acute kidney injury and the associated need for dialysis therapy: a systematic review and meta-analysis. Am J Kidney Dis 2020; 76: 826–841.e1.

50.

Zhou

Huang

, et al. Emerging frontiers in acute kidney injury: the role of extracellular vesicles. Bioact Mater 2025; 48: 149–170.

51.

Chen

Chang

Huang

, et al. Furosemide stress test as a predictive marker of acute kidney injury progression or renal replacement therapy: a systemic review and meta-analysis. Crit Care 2020; 24: 202.

52.

Chawla

Davison

Brasha-Mitchell

, et al. Development and standardization of a furosemide stress test to predict the severity of acute kidney injury. Crit Care 2013; 17: R207.

53.

Koyner

Davison

Brasha-Mitchell

, et al. Furosemide stress test and biomarkers for the prediction of AKI severity. J Am Soc Nephrol 2015; 26: 2023–2031.

54.

Ninet

Schnell

Dewitte

, et al. Doppler-based renal resistive index for prediction of renal dysfunction reversibility: a systematic review and meta-analysis. J Crit Care 2015; 30: 629–635.

55.

Darmon

Bourmaud

Reynaud

, et al. Performance of Doppler-based resistive index and semi-quantitative renal perfusion in predicting persistent AKI: results of a prospective multicenter study. Intensive Care Med 2018; 44: 1904–1913.

56.

Sitbon

Darmon

Geri

, et al. Accuracy of clinicians’ ability to predict the need for renal replacement therapy: a prospective multicenter study. Ann Intensive Care 2022; 12: 95.

57.

Pike

Murugan

Keener

, et al. Biological markers for recovery of kidney (BioMaRK) study investigators. Biomarker enhanced risk prediction for adverse outcomes in critically ill patients receiving RRT. Clin J Am Soc Nephrol 2015; 10: 1332–1339.

58.

Lai

Liu

Tseng

, et al. Unsupervised clustering identifies sub-phenotypes and reveals novel outcome predictors in patients with dialysis-requiring sepsis-associated acute kidney injury. Ann Med 2023; 55: 2197290.

59.

Bhatraju

Zelnick

Herting

, et al. Identification of acute kidney injury subphenotypes with differing molecular signatures and responses to vasopressin therapy. Am J Respir Crit Care Med 2019; 199: 863–872.

60.

Wiersema

Jukarainen

Vaara

, et al. Two subphenotypes of septic acute kidney injury are associated with different 90-day mortality and renal recovery. Crit Care 2020; 24: 150.

61.

Abgrall

Holder

Chelly Dagdia

, et al.

Should AI models be explainable to clinicians?

Crit Care 2024; 28: 301.

62.

Lim

Gim

Cho

, et al. Real-time machine learning model to predict short-term mortality in critically ill patients: development and international validation. Crit Care 2024; 28: 76.

63.

O'Reilly

McGrath

Martin-Loeches

. Optimizing artificial intelligence in sepsis management: opportunities in the present and looking closely to the future. J Intensive Med 2023; 4: 34–45.

64.

Stivi

Padawer

Dirini

, et al. Using artificial intelligence to predict mechanical ventilation weaning success in patients with respiratory failure, including those with acute respiratory distress syndrome. J Clin Med 2024; 13: 1505.

65.

Mohamadlou

Lynn-Palevsky

Barton

, et al. Prediction of acute kidney injury with a machine learning algorithm using electronic health record data. Can J Kidney Health Dis 2018; 5: 2054358118776326.

66.

Selvaskandan

Gaultney

Heath

, et al.

Can artificial intelligence predict the need for acute renal replacement therapy among inpatients with acute kidney injury?

Future Healthc J 2023; 10: 19–20.

67.

Bagshaw

Cruz

Gibney

, et al. A proposed algorithm for initiation of renal replacement therapy in adult critically ill patients. Crit Care 2009; 13: 317.

68.

Ostermann

Dickie

Barrett

. Renal replacement therapy in critically ill patients with acute kidney injury–when to start. Nephrol Dial Transplant 2012; 27: 2242–2248.

69.

Bagshaw

Wald

. Strategies for the optimal timing to start renal replacement therapy in critically ill patients with acute kidney injury. Kidney Int 2017; 91: 1022–1032.

70.

Ostermann

Joannidis

Pani

, et al. 17th Acute disease quality initiative (ADQI) consensus group. Patient selection and timing of continuous renal replacement therapy. Blood Purif 2016; 42: 224–237.

71.

Bagshaw

Hoste

Wald

. When should we start renal-replacement therapy in critically ill patients with acute kidney injury: do we finally have the answer? Crit Care 2021; 25: 79.

72.

Neyra

Chen

Bagshaw

, et al. Risk classification and subphenotyping of acute kidney injury: concepts and methodologies. Semin Nephrol 2022; 42: 151285.

73.

Kusirisin

Kaewdoungtien

Thanapongsatorn

, et al. Innovations in intensive care nephrology. In: Bezerra da Silva Junior

Nangaku

(eds) Innovations in nephrology: breakthrough technologies in kidney disease care. Springer Cham, Switzerland: Springer Nature, 2022, pp.343–363.