Metabolic complications of citrate anticoagulation in continuous renal replacement therapy (crrt): a delphi consensus on indications, monitoring and management

Abstract

Background

Regional citrate anticoagulation (RCA) is recommended by guidelines over systemic heparinization for continuous renal replacement therapy (CRRT). However, its use in patients with impaired citrate metabolism poses specific challenges and standardized guidance for managing RCA-related metabolic complications remains lacking.

Methods

A modified Delphi study was conducted according to a predefined protocol and reported in adherence with the CREDES (Conducting and REporting of DElphi Studies) checklist. The international expert panel comprised 29 clinicians and researchers from Europe, United States and Canada, with recognized expertise in RCA for CRRT in critically ill patients. Three iterative survey rounds were conducted to obtain agreement with proposed statements.

Results

Twenty-three experts completed all Delphi rounds, achieving consensus on twenty-two statements. RCA was considered feasible in patients with liver dysfunction, severe shock, or hyperlactatemia, with close monitoring and citrate dosing adjustment. Citrate accumulation can be prevented and managed using a stepwise approach, focused on reducing citrate delivery and discontinuing RCA in cases of overt accumulation. Metabolic alkalosis and electrolyte disturbances were identified as relevant but manageable complications, underscoring the need for individualizing CRRT settings.

Conclusion

These consensus statements support the use of RCA during CRRT in critically ill patients with impaired citrate metabolism and provide practical guidance for monitoring and management of metabolic complications. However they reflect expert opinion, especially for questions with limited data and low-level evidence.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-026-06066-x.

Background

Continuous Renal Replacement Therapy (CRRT) is the preferred modality for managing acute kidney injury (AKI) in critically ill patients, particularly those who are hemodynamically unstable as it enables gradual fluid and solute removal [1]. A critical component of CRRT is anticoagulation, necessary to prevent filter clotting. Despite contraindications, regional citrate anticoagulation (RCA) is recommended over heparin, as it prolongs circuit lifespan, reduces bleeding risk and avoids heparin-related complications such as heparin-induced thrombocytopenia (HIT) [2–4].

The body’s capacity to metabolize citrate is limited and capable of being saturated particularly in certain clinical situations where the Krebs cycle is compromised. As a result, RCA poses specific challenges in patients with impaired citrate metabolism, such as those with liver dysfunction, shock-related tissue hypoperfusion, or mitochondrial disorders. In these patients, citrate accumulation may occur, leading to hypocalcemia and metabolic acidosis [5–9]. In contrast, citrate overload refers to a condition in which the metabolic capacity is not exceeded and all citrate-calcium complexes are fully metabolized; according to the Stewart approach, the accompanying net sodium load increases the strong ion difference (SID) resulting in plasma alkalinization [10]. Consequently, RCA requires systematic monitoring to mitigate acid-base disturbances and multiple electrolyte imbalances, and its implementation may be limited by procedural complexities and the need for adequately trained staff [11–14].

Practice patterns differ worldwide. A secondary analysis of the STARRT-AKI- trial demonstrated notable regional variations in anticoagulation strategies. Anticoagulation-free CRRT was more common in North America, whereas RCA is the predominant anticoagulation method in Europe and Australia-New Zealand [15]. Differences in citrate formulations further influence daily practice, as these variations affect the sodium load and buffering capacity [16–18].

Despite its advantages, standardized guidance for managing RCA-related metabolic complications remains lacking. To fill this gap, we conducted a Delphi study to develop expert-based guidance focusing on: (1) evaluation of previously proposed contraindications for citrate use; (2) best practices for monitoring citrate-related complications; and (3) management strategies for citrate-induced metabolic disturbances, such as citrate accumulation and electrolyte imbalances.

Methods

Study design

The Delphi method is a validated, iterative approach designed to achieve expert consensus, particularly in areas with limited or conflicting evidence. It involves multiple rounds of structured surveys with controlled feedback between rounds, allowing experts to refine their opinions and move toward agreement [19–21].

In a modified Delphi approach, candidate statements are developed a priori by the research team based on literature review and expert input, rather than being generated by the panel. Panelists are, however, still able to contribute by suggesting additions or modifications throughout the process, ensuring that their expertise is incorporated while maintaining a structured framework [22].

This modified Delphi study was conducted according to a pre-specified protocol and adhered to the CREDES (Conducting and REporting of DElphi Studies) checklist for reporting Delphi research [23] (see Supplementary Table S1). Ethical approval was obtained from the Research Ethics Board of the Antwerp University prior to study initiation (Project ID 6794 – Edge: N/A).

Panel selection

An expert panel was assembled to ensure both clinical relevance and subject-matter expertise. Panelists were selected based on their clinical experience in critical care nephrology and their academic contributions [24]. Further selection criteria included authorship of peer-reviewed publications on AKI, CRRT or anticoagulation in critically ill patients.

Efforts were made to ensure broad representation across geographic regions and institutions. Invitations were sent via email and included a summary of the study’s aims, methodology, and expected time commitment. Participation in the first round was considered as provision of informed consent.

Overview of the modified Delphi process

An overview of the Delphi process is depicted in the flowchart provided in Fig. 1.

Fig. 1.

Flow chart of the modified delphi process

Preparatory phase (Round 0)

The process was initiated with a preparatory “Round 0,” which served to define the study framework and the content of candidate statements. During this phase, a core group of four clinical experts (RJ, AS, EH, TR)– representing critical care nephrology- reviewed and revised an initial list of statements derived from a systematic review on metabolic complications associated with RCA in CRRT [25], and further refined through collective expert discussion. The group assessed each item for clinical relevance, clarity, and potential redundancy. Feedback from this stage shaped the structure and content of the initial survey.

Delphi rounds

Three structured, web-based survey rounds were conducted using the SurveyMonkey^® platform between February 28 and October 14, 2025. Email reminders were sent to encourage participation. Panelists who did not participate in Round 1 were excluded from participation in subsequent rounds.

• Round 1 – Initial survey

  • In the first round, participants provided opinions on a series of statements regarding contraindications to RCA use, early detection of metabolic complications, and management strategies for electrolyte disturbances and citrate accumulation. Each statement was rated on a 9-point Likert scale (1 = strongly disagree, 9 = strongly agree). Free-text fields were available for rewording suggestions, clarifications, or comments. The aim of this round was to capture a broad range of expert perspectives and identify areas of divergence.
• Round 2 – Refinement of statements

  • Statements that did not reach consensus in Round 1 were revised based on both quantitative results and qualitative feedback. This round focused on refining statements and promoting convergence of expert opinion.

• Round 3 – Final consensus

  • Prior to Round 3, panelists received aggregated results from the first two rounds. They were given the opportunity to review these results and provide additional feedback. They were then invited to reassess the remaining statements that had achieved only partial agreement (≥60%).

Consensus definition and level of recommendation

Consensus was defined a priori as ≥ 70% of panelists rating a statement between 7 and 9 on the Likert scale [26, 27]. Disagreement was defined as > 15% of raings between 1 and 3 and a median score < 6. Statements with ≥ 60% but less than 70% agreeent were considered to have partial consensus.

This structured approach ensured that conclusions reflected the majority opinion and the distribution of responses, reducing the risk of overinterpreting results with significant variability or hidden disagreement.

Statements meeting these criteria were retained as final consensus statements. Items that did not meet the consensus threshold were revised based on qualitative feedback and re-evaluated in subsequent rounds.

Data handling and confidentiality

All responses were anonymized prior to analysis to ensure confidentiality. Expert identifiers were stored separately from response data and were accessible only to the principal investigator. Aggregated data were used for reporting, and no individual responses were identifiable in the results.

Statistical analysis

Descriptive statistics were used to analyze responses across the three Delphi rounds. For each statement, the following metrics were calculated:

• Agreement: Proportion of respondents rating 7–9.

• Disagreement: Proportion of respondents rating 1–3.

• Median used to assess central tendency and variability.

Software

Survey responses were recorded at the end of each Delphi round and compiled in Microsoft Excel (Microsoft Corp., Redmond, WA, USA). Summary statistics are presented in tabular form, and Likert scale distributions are visualized where appropriate.

Outcomes measurement

The main outcomes of the Delphi study include:

• The number of statements reaching consensus (≥ 70% agreement).

• Identification of areas with persistent disagreement, highlighting topics requiring further research or discussion in clinical practice.

Results

A total of 60 experts were invited to serve as panelists. Of these, 35 agreed to participate, 5 declined and 20 did not respond to repeated invitations. The characteristics of the panelists are summarized in Table S2. Although 35 panelists initially agreed to participate, 29 completed the first round, comprising 22 men and 7 women, 11 were nephrologists and 18 were board-certified intensivists, with 16 from Europe, 9 from the USA and 4 from Canada. Round 2 included 26 participants, and in round 3, 23 panelists provided responses.

The first round included 51 statements, of which 7 reached consensus according to our pre-specified criteria. In the second round, 22 statements were presented, and 12 were accepted. The reduction in the number of statements from the first to the second round was based on participant feedback that the first round contained too many statements. The statements were therefore revised in accordance with their suggestions and comments. In the third round, only statements that had achieved partial consensus (> 60%) were resubmitted. However, 2 statements that met this criterion were not resubmitted following consultation within the research team. One statement regarding the measurement of post-filter ionized calcium (pf-iCa) was not included in the final statements as it was determined by the Delphi panel as largely redundant. Of the 6 statements presented, 3 were accepted. Overall, of the 22 statements that ultimately achieved consensus, 4 related to indications, 8 to monitoring, 3 to management of citrate accumulation, and 7 to metabolic complications. All accepted statements along with their corresponding percentage of consensus and median score, are listed in Table 1. Statements presented in Rounds 1, 2 and 3, along with their corresponding percentage of consensus, median scores, and levels of disagreement, are provided in Supplementary Tables S3, S4, S5, respectively.

Table 1.

Final Delphi consensus statements

Domain	Statement	Agreement % (score ≥ 7)	Median	Disagreement % (score ≤3)
Indications	RCA can be used in severe liver failure, with regular routine monitoring and load adjustments	92.3	8	3.8
	No anticoagulation may be considered in severe liver failure, particularly with accompanying significant coagulopathy	76.9	7.5	7.6
	RCA can be used in severe shock when regular, routine monitoring is achieved	84.6	8	3.8
	RCA can be used though cautiously in case of a high lactate level, but rising levels indicate a higher risk of CA	88.5	8	3.8
Monitoring	pf-iCa should be systematically and regularly monitored in all patients receiving RCA	86	9	10
	In patients at high-risk of CA, target pf-iCa should remain within the standard therapeutic range, towards the higher end	80.8	7	11.5
	Although a validated predictive score does not exist, perform regular clinical evaluation to assess the risk of CA, before initiating RCA in liver failure	88.5	7.5	3.8
	Routine monitoring of the total-to-ionized calcium ratio to detect CA is mandatory with RCA	86	8	7
	A rising total-to-ionized calcium ratio is a reliable marker for diagnosing CA, with a ratio ≥ 2.5 indicating its presence	100	8	0
	CA is suspected in case of a low systemic ionized calcium, increased calcium supplementation needs, elevated calcium ratio (≥ 2.5) or high anion gap metabolic acidosis	100	8	0
	In patients at high-risk of CA systemic ionized calcium monitoring according to a standardized protocol should be performed	95.7	8	0
	In patients at high-risk of CA, treatment modalities that provide higher citrate clearance (e.g., CVVHD, CVVHDF) may be preferred	82.6	7	13
Management of CA	Initial management of CA includes reducing citrate load -achieved by decreasing blood flow and/or increasing effluent flow- to enhance clearance	92.3	8	3.8
	In patients at high-risk of CA, treatment with settings aiming for a lower citrate administration to the patient should be initiated	72	8	7
	Persistent CA despite adjustments, requires stopping RCA and switching anticoagulation	92.3	8	3.8
Complications	Metabolic alkalosis is a relevant complication of RCA	71	7	10
	Strategies to correct metabolic alkalosis may include increasing the effluent rate or lowering bicarbonate in the replacement solutions	73.9	7	13
	Metabolic acidosis during RCA-CRRT, if CA is excluded, may be managed by increasing citrate delivery, adjusting infusion rate/ target dose. To reduce citrate and bicarbonate losses, strategies such as lowering dialysate/ replacement flow rate, or increasing bicarbonate concentration in these fluids may be used	77	7	7.7
	Hypernatremia may occur with hypertonic citrate solutions (e.g., 4% TSC)	84.6	7	0
	Adjusting CRRT solution composition to correct electrolyte imbalances may be considered but requires careful monitoring	73.1	7	3.8
	Hyperkalemia can be managed by increasing dialysate or replacement flow rates	75	7.5	7
	Calcium supplementation in dialysate/replacement should be avoided during RCA-CRRT	75	8	10

CVVHD: continuous venovenous hemodialysis; CVVHDF: continuous venovenous hemodiafiltration; CRRT: continuous renal replacement therapy; CA: citrate accumulation; pf-iCa: post-filter ionized calcium; RCA: regional citrate anticoagulation; severe liver failure: serum transaminases > 5 times upper limit of normal or > 500 IU/L, INR > 1.5 ; bilirubin > 2 mg/dl (Girish V, Royer A, John S. Acute Liver Failure. [Updated 2025 Jul 6]. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2025 Jan-. Available from: https://www.ncbi.nlm.nih.gov/books/NBK482374/

Discussion

Domain 1: Indications

Consensus statement 1: RCA can be used in patients with severe shock when regular, routine monitoring is achieved.

Consensus statement 2: No anticoagulation may be considered for patients with severe liver failure, particularly those with accompanying significant coagulopathy.

Consensus statement 3: RCA can be used in patients with severe liver failure if risks are carefully managed through regular, routine monitoring and load adjustments.

Consensus statement 4: RCA can be used cautiously in patients with elevated lactate levels, but worsening hyperlactatemia suggests a high risk of citrate accumulation.

• Remarks

Citrate is metabolized via the Krebs cycle in the liver, skeletal muscle, and kidneys, generating sodium bicarbonate. Under normal conditions, this process contributes to plasma alkalinization, mainly due to the sodium load, which increases the SID according to the Stewart concept [28, 29]. Citrate metabolism is saturable and may be impaired in cases of liver failure, tissue hypoperfusion or mitochondrial dysfunction [5].

RCA in liver failure

Acute liver failure is characterized by a rapid deterioration of liver function in patients without pre-existing liver disease or cirrhosis and is defined by the presence of hepatic encephalopathy and an international normalized ratio (INR) ≥ 1.5 within 26 weeks of symptom onset [30]. On the other hand, chronic liver failure represents a syndrome of progressive and persistent hepatic dysfunction resulting from underlying chronic liver disease, leading to life-threatening complications due to impairment of the liver’s synthetic, metabolic and detoxification functions [31]. High-risk patients are those who, according to KDIGO guidelines, have relative or absolute contraindications for RCA [3].

Severe liver failure is considered a contraindication to RCA due to concerns regarding impaired citrate metabolism. However, pharmacokinetic data suggest that RCA is feasible in critically ill patients, including those with decompensated cirrhosis, if the amount of citrate delivered to the patient is adapted and ionized calcium closely monitored [8]. Importantly, standard liver function tests are poor predictors of citrate accumulation. In contrast, elevated serum lactate (≥ 3.4 mmol/L) and reduced prothrombin time (≤ 26%) provide a greater sensitivity and specificity for detecting patients at risk of citrate accumulation [32].

In a prospective observational study of 133 patients with varying degrees of liver impairment (the L-Cat study), citrate accumulation was rare and occurred in those with severe hyperlactatemia [33]. Similarly, meta-analysis confirmed that the risk of citrate accumulation was not higher in patients with liver dysfunction compared to those normal liver function, while also prolonging filter lifespan [34].

In a study involving patients with acute liver failure or cirrhosis with presumed severe impaired citrate metabolism, the use of RCA was shown to be safe and effectively maintained circuit patency [35, 36]. This Shock protocol employs a fixed, high citrate-to-blood flow ratio to ensure effective circuit anticoagulation pf-iCa level below 0.4 mmol/L. Very high effluent flow rates relative to circuit plasma flow result in a single-pass citrate extraction ratio (ECit) exceeding 70–75%, hile delivering a CRRT dose of approximately 38–42 ml/kg/h in patients with severe shock. This high fractional citrate removal results in regional citrate anticoagulation, with negligible citrate levels (< 2 mmol/L) returning to the patient. As a result, citrate accumulation and toxicity are avoided regardless of the patient’s citrate-metabolizing capacity. However, this strategy may also increase clearance of critical solutes, including antibiotics and other medications, potentially affecting therapeutic efficacy [37, 38]. In centers with experienced teams, broader use of RCA for CRRT led to longer filter lifespan without an increase in rates of citrate accumulation, rates of electrolyte disturbances or significant bleeding [14].

Role of lactate

In a large retrospective cohort (1059 patients), the incidence of citrate accumulation within 48 h was rare (2.26%) but increased with rising baseline lactate, reaching 6.3% when lactate was ≥ 4 mmol/L. Importantly, dynamic lactate monitoring proved more predictive: a 12-hour lactate clearance below 24.3% was a stronger indicator of accumulation risk than the initial lactate level alone [9]. Similar findings have been reported in smaller cohorts, where citrate accumulation was strongly associated with hyperlactatemia, hyperbilirubinemia, and increased calcium infusion requirements [39].

These observations are supported by systematic review data of seven observational studies (n = 1573) demonstrating that patients with citrate accumulation had higher baseline and peak lactate concentrations and a greater risk of mortality (OR 5.24). Although the level of evidence remains low, elevated lactate levels appear to be a useful predictor of impaired citrate metabolism and increased risk of accumulation [40]. In a recent retrospective cohort study of 911 critically ill patients with a high burden of liver dysfunction (~ 50%) undergoing RCA-based CRRT, signs of citrate accumulation occurred in 17% and was associated with liver dysfunction rather than circulatory shock; after filter adjustments most of these patients stabilized, so that at the end only 5.2% hd overt citrate accumulation requiring its discontinuation. Lactate proved to be a strong independent predictor, with the odds of accumulation increasing by 2.34(CI 1.94–2.85, p < 0.001) for each log-unit rise in lactate, highlighting its reliability for risk assessment and individualized anticoagulation management [41].

Despite these risks, RCA appears safe even in most hyperlactatemic patients requiring CRRT, particularly when lactate levels are stable or declining. In a study on critically ill patients with hyperlactatemia requiring CRRT, compared with low molecular weight heparin (LMWH), RCA provided a superior filter lifespan and a lower bleeding risk without increasing the risk of metabolic complications [42]. Moreover, a retrospective study reported an association with reduced 28-, 60-, and 90-day mortality in patients with hyperlactatemia treated with RCA compared to heparin [43].

RCA in shock

Patients with circulatory shock may exhibit decreased cellular oxygen delivery, resulting in reduced Krebs cycle activity due to impaired mitochondrial oxidative function [2, 5]. Nevertheless, the presence of shock itself should not be viewed as absolute contraindication to RCA; rather, clinicians should focus on markers of impaired citrate metabolism, particularly lactate levels and their trends. However, the lactate threshold above which RCA should be avoided had yet to be established [40].

Anticoagulation-free strategies

In patients with severe liver failure and high bleeding risk, CRRT without anticoagulation has been evaluated, with evidence showing increased circuit failure compared with RCA, but overall supporting anticoagulant-free strategies as a viable option [44]. A key challenge in clinical practice is that it remains unclear when CRRT can be safely initiated without anticoagulation, as no specific cut-off values have been established for platelet count, APTT, INR, fibrinogen or other coagulation factors.

Domain 2: Monitoring

Consensus statement 5: We should measure postfilter ionized calcium.

Consensus statement 6: In patients at high-risk of citrate accumulation, target postfilter ionized calcium level should remain within the standard therapeutic range, aiming toward the higher end. Consensus statement 7: In patients at high-risk of citrate accumulation, systemic ionized calcium monitoring according to a standardized protocol should be performed.

• Remarks

Measurement of post-filter ionized calcium (pf-iCa) is a principal component of most RCA-CRRT protocols, primarily used to assess anticoagulation efficiency within the filter and to guide citrate infusion adjustments. Target pf-iCa levels of 0.25–0.35 mmol/L are generally recommended to prevent filter clotting while maintaining circuit patency and optimizing filter lifespan [45–47].

The clinical value of pf-iCa measurement remains debated. Unlike systemic ionized calcium, which is essential for patient safety, pf-iCa is less critical for direct patient monitoring. Concerns exist regarding reproducibility across blood gas analyzers, particularly at low concentrations, which may lead to unreliable adjustments of citrate dosing [48]. Consequently, pf-iCa monitoring requires careful interpretation by experienced staff and adds complexity, cost, and workload to RCA protocols. Some groups have proposed simplifying RCA management by reducing the frequency of pf-iCa monitoring or omitting it altogether [49, 50].

Despite these limitations, pf-iCa monitoring remains useful. Maintaining pf-iCa within the standard therapeutic range (0.25–0.35 mmol/L) and aiming toward the higher end (up to 0.40 mmol/L) strikes a practical balance between minimizing citrate exposure and ensuring effective anticoagulation [51]. Citrate continues to exert anticoagulant activity even at relatively higher pf-iCa levels. In vitro data from Calatzis et al., demonstrated meaningful inhibition of coagulation up to pf-iCa values up to 0,5mmol/L [45]. Increasing the pf-iCa target from 0.25 to 0.35 mmol/L to 0.30–0.40 mmol/L does not compromise filter lifespan but may reduce unnecessary citrate exposure. However, optimal pf-iCa targets should be individualized based on the patient’s clinical and biological status [51].

Consensus statement 8:Routine monitoring of the calcium ratio (total calcium/ionized calcium) is mandatory when using RCA to detect citrate accumulation.

Consensus statement 9: A rising total-to-ionized calcium ratio is a reliable marker for diagnosing citrate accumulation, with a value of 2.5 or higher indicating its presence.

Consensus statement 10: Citrate accumulation should be suspected in the presence of a combination of decreased systemic ionized calcium, increased calcium supplementation requirements, an elevated total-to-ionized calcium ratio (≥ 2.5) or metabolic acidosis.

• Remarks

The total-to-ionized calcium ratio (tCa/iCa) is widely used as a surrogate marker to detecting citrate accumulation during RCA-CRRT. A value ≥ 2.5 is generally considered suggestive of citrate accumulation (citrate binds to ionized calcium, lowering its levels, while the total calcium which includes the citrate-bound calcium increases due to supplementation) [52]; however, it citrate accumulation should not be interpreted as a standalone diagnostic criterion. Rather, citrate accumulation should be suspected when a rising calcium ratio coincides with decreased systemic ionized calcium, increased calcium requirements, or progressive metabolic acidosis [7]. Evaluating these parameters together provides a practical and reliable approach to identifying citrate toxicity in critically ill patients. In addition, interpretation of tCa/iCa ratio may be influenced by factors such as phosphate concentration, pH, and serum albumin, which are frequently altered in critically ill patients [53]. The interpretation of the total-to-ionized calcium ratio may be influenced by serum albumin levels, as total calcium is partly protein-bound and therefore reduced in hypoalbuminemic states commonly seen in critically ill patients. Recent evidence suggests that using albumin-corrected total calcium or applying lower thresholds of the tCa/iCa ratio may improve the detection of clinically relevant citrate accumulation [54]. However, Bidar et al. have raised concerns that albumin correction may artificially elevate the ratio in hypoalbuminemic patients, potentially misclassifying them as having citrate accumulation, when the elevation may instead reflect the underlying severity of illness [55]. The optimal frequency of tCa/iCa monitoring remains unknown and current practices vary widely across centers. Nonetheless, elevated tCa/iCa ratios have been associated with adverse outcomes; in critically ill patients with AKI receiving RCA-CRRT, a tCa/iCa ratio ≥ 2.4 independently predicted higher 28-day mortality [56].

Although direct measurement of serum citrate would provide a more definitive evaluation, its widespread use is limited, and validated thresholds as well as established clinical utility remain undefined [57]. Therefore, dynamic monitoring of the tCa/iCa ratio, interpreted alongside systemic ionized calcium and acid–base status, remains the most practical strategy for detecting citrate accumulation.

It is also important to note that citrate itself is not inherently toxic; the main risks arise from its downstream effects—hypocalcemia and metabolic acidosis. With adequate calcium replacement and careful acid–base management, even higher calcium ratios can sometimes be tolerated, if patients are closely monitored and hemodynamically stable [58].

Consensus statement 11: Although a validated predictive score does not exist, a careful clinical evaluation to assess the risk of citrate accumulation can be performed before initiating RCA in liver failure patients.

Consensus statement 12: In patients at high-risk of citrate accumulation, treatment modalities that provide higher citrate clearance (e.g., CVVHD, CVVHDF) may be preferred.

• Remarks

Although no universally validated risk score for citrate accumulation exists, several studies have attempted to identify predictors highlighting its multifactorial nature. Lactate is consistently identified as the strongest predictor, although bilirubin and CRRT-settings also contribute while vasopressor requirement is not identified as a reliable screening variable citrate accumulation [41], Hong et al. developed a nomogram for citrate accumulation in patients with hepatic insufficiency, incorporating six independent risk factors including sex, INR, norepinephrine dose, arterial PO₂, peripheral blood citrate concentration, and dialysate flow rate, with INR and peripheral citrate levels being the strongest contributors [59]. Other models highlighted hepatic insufficiency, albumin, mechanical ventilation and CRRT settings as relevant [60]. In a retrospective study, severe metabolic acidosis with hyperlactatemia at CRRT initiation (i.e. lactate > 10 mmol/L or pH < 7.14), predicted citrate accumulation more reliably than liver function markers [61] [9]. Overall, these findings emphasize that risk assessment should rely on a multimodal and dynamic evaluation, rather than a single parameter.

Citrate clearance increases with higher dialysate flow in diffusive modalities and higher filtration flow in convective modalities [5, 62]. In CVVHD, citrate requirements are lower, as equivalent effluent doses can be achieved with lower blood flows, and diffusion allows a greater proportion of infused citrate to be removed due to the higher effluent-to-blood flow ratio [2, 63]. In CVVH, citrate removal relies entirely on convective clearance, determined by the filtration fraction. Maintaining an acceptable filtration fraction requires higher blood flows, and achieving therapeutic anticoagulation under these conditions necessitates higher citrate infusion rates, which together increase the net citrate load to the patient. Increasing ultrafiltration to enhance clearance inevitably raises filtration fraction, promoting hemoconcentration, higher transmembrane pressures, and an increased risk of filter clotting [2, 28, 64]. An in vitro study on citrate clearance found that high-flux filters combined with elevated dialysate flow rates may be preferred in patients with impaired liver function to reduce the risk of accumulation [62].

Domain 3: Managing citrate accumulation

Consensus statement 13: When encountering citrate accumulation, initial management should include reducing the citrate load by decreasing blood flow and/or increasing effluent flow- to enhance clearance.

Consensus statement 14: For patients at high risk of citrate accumulation, we should start treatment with settings aiming for lower citrate administration to the patient.

Consensus statement 15: In cases of persistent citrate accumulation despite treatment adjustments, RCA should be discontinued and replaced with an alternative anticoagulation strategy, while continuing the most appropriate RRT modality based on patient hemodynamic status.

• Remarks

When citrate accumulation occurs, the net citrate load delivered to the patient should be promptly reduced. This can be achieved by (1) lowering the blood flow rate to decrease citrate delivery through blood flow–citrate coupling, (2) increasing citrate removal by increasing the dialysate flow in CVVHD, whereas in CVVH increasing filtration rate must be balanced against high filtration fraction-related hemoconcentration and clotting risk, or (3) reducing the targeted citrate concentration within the filter [5, 7, 29, 52, 56]. These adjustments should be guided by biochemical monitoring, including pf-iCa (aiming towards the upper normal range), systemic ionized calcium, and the tCa/iCa ratio. Correction of hypocalcemia with calcium infusion is essential. If accumulation persists i.e., (biochemical abnormalities (i.e., tCa⁺/iCa⁺ ratio ≥ 2.5, severe hypocalcemia, worsening metabolic acidosis) fail to improve despite these measures, RCA should be stopped and alternative strategies should be considered., including different anticoagulants and delivering RRT without anticoagulation for a limited period.

Domain 4: Metabolic complications

Consensus statement 16:Metabolic alkalosis is a relevant complication of RCA.

Consensus statement 17: Strategies to correct metabolic alkalosis may include increasing the effluent rate or lowering bicarbonate in the replacement solutions.

Consensus statement 18: When metabolic acidosis occurs during RCA-CRRT and citrate accumulation has been excluded, management may include to increase citrate delivery by adjusting the infusion rate or target dose. To reduce citrate and bicarbonate losses, strategies including lowering the dialysate or replacement fluid flow rate, increasing the bicarbonate concentration in these fluids can be considered.

Consensus statement 19:Hypernatremia can occur as complication of hypertonic citrate solutions, particularly with high concentration trisodium citrate (e.g., 4% TSC)."

Consensus statement 20:Adjusting CRRT solution composition to correct electrolyte imbalances may be considered but requires careful monitoring.

Consensus statement 21: In case of hyperkalemia, the rate of the dialysate flow or replacement flow should be increased.

Consensus statement 22: Calcium supplementation in the dialysate/replacement should be avoided for patients on RCA-CRRT.

• Remarks

Metabolic alkalosis during RCA primarily results from the metabolism of trisodium citrate, which delivers a substantial alkali load and alters the strong ion difference (SID) [64–67]. When citrate is fully metabolized, each millimole of trisodium citrate yields three millimoles of sodium bicarbonate, thereby providing a buffering effect like that of bicarbonate. The persistence of sodium after citrate metabolism increases the SID, driving alkalinization [68–70].

In daily practice, metabolic alkalosis may develop from excessive citrate administration and intact citrate metabolism or impaired citrate clearance, for example when membrane performance declines (termed clogging) [71]. This condition is usually correctable by increasing dialysate or replacement flow, reducing the citrate infusion rate, or using solutions with lower sodium or higher chloride content [72–75]. In a recent retrospective analysis, patients undergoing CVVH with RCA who developed metabolic alkalosis refractory to adjustments of CVVH settings, switching to a low bicarbonate replacement fluid resulted in significant reduction in serum bicarbonate levels [76].

The risk of alkalosis is also influenced by the citrate formulation. Hypertonic citrate solutions, with high sodium concentrations, are typically infused separately and often require hyponatremic replacement or dialysate with lower bicarbonate concentrations to mitigate electrolyte imbalances [16]. Isotonic citrate solutions, closer to physiologic sodium levels, are commonly delivered as calcium-free predilution replacement fluids. Comparative studies using both hypertonic and isotonic formulations have shown that, despite theoretical concerns, the incidence of hypernatremia and metabolic alkalosis remains low and not significantly different from heparin-based anticoagulation strategies using carefully designed RCA prescriptions [25].

Hyperkalemia, although not a complication of RCA-CRRT is a common complication of AKI and poses significant risk to cardiac conduction. When IHD is unavailable, not tolerated or if the hyperkalemia is life-threatening, CRRT can also be used for hyperkalemia using higher volumes of replacement fluid or dialysate with low or zero potassium concentrations [13, 77].

Although studies show that RCA-CRRT with calcium-containing solutions is feasible and safe in critically ill patients, concerns are that they may reduce anticoagulation efficacy or increase citrate accumulation by necessitating higher citrate doses [78, 79].

Limitations

Though rigorously performed according to the guidelines and including the opinion of a variety of experts, this study has several limitations inherent to the Delphi methodology. Although the panel comprised internationally recognized experts in critical care nephrology, participants were exclusively from Western countries, with a predominance of European participants. This uneven geographic representation may limit the generalizability of conclusions to regions with different CRRT practices, patient populations, or resource availability, such as those used in Asian countries or developing nations. A broader, more internationally diverse panel would improve the generalizability and help mitigate potential biases. A potential limitation of this study is the attrition of six experts during the Delphi process, which may have introduced bias by reducing the diversity of opinions and potentially affecting the stability of the consensus. Additionally, variation in citrate formulations across regions could influence the applicability of consensus statements.

Additionally, Delphi studies reflect expert opinion rather than high-quality randomized controlled trial data. As such, the statements based on clinical experience reflect low-level of evidence, highlighting an important evidence gap, underscoring the need for prospective validation. The anonymous format minimizes peer pressure but may reduce interactive debate, limiting exploration of controversial or nuanced topics.

Certain operational aspects were not fully addressed. Since the study focused on metabolic complications, the frequency of monitoring was not assessed, and technical or nursing aspects of RCA implementation were not comprehensively evaluated.

Additional limitations include potential bias from panel selection, as experts with strong prior experience or specific clinical perspectives may influence outcomes. The use of the pre-specified consensus threshold of ≥ 70%—though commonly used—stays somewhat arbitrary and may have influenced the inclusion or exclusion of statements. This threshold may have led to the rejection of statements with moderate but clinically relevant agreements, while statements just above the cutoff may reflect only marginal consensus. The choice of this cutoff, therefore, has a direct effect on the interpretation of consensus strength and should be considered when applying these statements in clinical practice.

Future directions

This study highlights the need for internationally harmonized protocols and structured training frameworks for RCA in CRRT, particularly as its use extends beyond traditionally low-risk populations. Future research should focus on.

• Prospective evaluation of RCA protocols in patients with liver failure or shock.

• Development of validated risk stratification tools to predict citrate accumulation.

• Optimal monitoring frequency in high-risk populations.

• Multicentre implementation studies assessing clinical outcomes following adoption of consensus-based statements.

• Several statements did not reach consensus among the expert panel. This highlights areas for further prospective studies to provide standardized guidance.

Conclusion

This modified Delphi study provides consensus-based guidance supporting the use of RCA during CRRT in critically ill patients with liver dysfunction, severe shock or hyperlactatemia, provided that metabolic parameters are closely monitored and citrate dosing is individualized. These consensus-based statements reflect expert opinion in areas where data are limited or low-level evidence exits. They also address gaps in existing guidelines and offer a framework for more standardized RCA use across institutions., Prospective studies are required to confirm safety and applicability across diverse ICU populations.

Take home messages (see Fig. 2 flow chart)

Fig. 2.

Flow chart: RCA-CRRT in patients with impaired citrate metabolism

• RCA in high-risk populations: Although traditionally contraindicated in severe liver dysfunction or shock, RCA can be used with caution in these patients with enhanced monitoring and citrate dose adjustments.

• Calcium monitoring: Postfilter ionized calcium measurement is essential for assessing RCA efficacy, and the total-to-ionized calcium ratio serves as a sensitive early marker for citrate accumulation.

• Metabolic management: Metabolic alkalosis, electrolyte disturbances, and hypernatremia are relevant but manageable complications, emphasizing the need to individualize CRRT settings.

• Protocol-driven adjustments: Reducing citrate load by modifying CRRT parameters should be the first response to suspected citrate accumulation; persistent issues warrant switching to an alternative anticoagulation strategy.

Abbreviations

ABG

Arterial blood gas

AKI

Acute kidney injury

CA

Citrate accumulation

CRRT

Continuous renal replacement therapy

CVVH

Continuous veno-venous hemofiltration

CVVHD

Continuous veno-venous hemodialysis

CVVHD

Continuous veno-venous hemodiafiltration

ECiT

Citrate extraction ratio

HIT

Heparin-induced thrombocytopenia

pf-iCa

Post-filter ionized calcium

RCA

Regional citrate anticoagulation

Author contributions

Conceptualization: RJ, PJMethodology: RJ, MG, AS, TR, EH, PJ.Expert Panel Coordination / Data Curation: RJFormal Analysis: RJWriting – Original Draft: RJWriting – Review & Editing: All authors approved the final draft. All experts who completed all subsequent Delphi rounds and approved the final draft were included as co-authors.

Funding

This research received no external funding.

Data availability

The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Ethical approval was obtained from the Research Ethics Board of the Antwerp University prior to study initiation (Project ID 6794 – Edge: N/A). Participation in at least the first round was considered as provision of informed consent.

Competing interests

The authors declare no competing interests.