Skip to main content

Foundational Trials

RICH

Image: Shutterstock / ChaNaWiT

Publication

  • Title: 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
  • Acronym: RICH
  • Year: 2020
  • Journal published in: JAMA
  • Citation: Zarbock A, Küllmar M, Kindgen-Milles D, Wempe C, Gerß J, Brandenburger T, 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(16):1629-1639.

Context & Rationale

  • Background
    Continuous kidney replacement therapy (CKRT) requires extracorporeal circuit anticoagulation to prevent premature filter clotting, treatment interruptions, blood loss, and reduced delivered dose.
    Systemic unfractionated heparin is familiar and inexpensive but increases bleeding risk and can cause heparin-induced thrombocytopenia.
    Regional citrate anticoagulation (RCA) provides anticoagulation largely confined to the circuit (via chelation of ionised calcium), plausibly reducing bleeding and prolonging circuit life, but adds operational complexity and can cause electrolyte/acid–base derangements (and citrate accumulation in impaired metabolism).
    Prior RCTs and meta-analyses suggested improved circuit patency and less bleeding with RCA, but were underpowered/heterogeneous for patient-centred outcomes (mortality, renal recovery), leaving persistent uncertainty.
  • Research Question/Hypothesis
    In critically ill adults with severe acute kidney injury (AKI) requiring CKRT, does RCA (vs systemic heparin) prolong dialysis filter life span and improve survival (90-day all-cause mortality)?
  • Why This Matters
    Anticoagulation strategy is a universal, high-frequency ICU decision with direct implications for circuit performance, transfusion/bleeding risk, nursing workload, metabolic safety, and potentially outcomes.
    A large, methodologically rigorous trial could justify stronger guideline recommendations and standardise CKRT delivery internationally.

Design & Methods

  • Research Question: Among critically ill adults with severe AKI requiring CKRT, does regional citrate anticoagulation (vs systemic heparin anticoagulation) increase filter life span and reduce 90-day all-cause mortality?
  • Study Type: Investigator-initiated, multicentre, parallel-group, randomised clinical trial with adaptive group-sequential features; open-label; conducted in 26 ICUs (Germany).
  • Population:
    • Setting: adult ICUs; enrolment after clinical decision to start CKRT for AKI.
    • Core inclusion: adults (18–90 years) with severe AKI (KDIGO stage 3) and an indication for CKRT (common indications at initiation included oliguria/anuria <200 mL/12 h, fluid overload with oedema, serum urea >150 mg/dL, potassium ≥6 mmol/L, blood pH <7.15).
    • Additional eligibility conditions were prespecified in the protocol (eg, severe sepsis/septic shock, vasopressor use, refractory fluid overload).
    • Key exclusions (protocol-defined): contraindication to citrate or heparin; high bleeding risk/coagulopathy or severe thrombocytopenia; situations with impaired citrate handling (eg, severe hepatic failure/shock with concern for citrate accumulation); other protocol-specified safety/feasibility exclusions.
  • Intervention:
    • Regional citrate anticoagulation for CKRT, titrated to a low post-filter ionised calcium (target 1.0–1.4 mg/dL; approximately 0.25–0.35 mmol/L), with systemic calcium substitution to maintain physiological systemic ionised calcium; protocolised laboratory monitoring.
    • CKRT prescription standardised: target effluent dose 30 mL/kg/h; blood flow >100 mL/min; routine filter change mandated at 72 hours (earlier if clinically indicated).
  • Comparison:
    • Systemic unfractionated heparin anticoagulation during CKRT with target activated partial thromboplastin time 45–60 seconds, with the same CKRT prescription standards (effluent dose, blood flow, routine 72-hour filter change rule).
  • Blinding: Unblinded (open-label) for clinicians and trial personnel; mortality is objective, but circuit-management decisions (filter changes, downtime) could be influenced by lack of blinding.
  • Statistics: Planned sample size up to 1450 (1260 anticipated analysable) to detect an ~8% absolute reduction in 90-day mortality (from ~48% to ~40%) with two-sided α=0.05; filter life span effect size was also powered (protocol/SAP prespecified co-primary endpoints). Primary analyses were performed on a modified intention-to-treat “primary analysis set” (randomised participants with post-randomisation consent) with time-to-event methods for mortality and non-parametric estimation for filter life span.
  • Follow-Up Period: Co-primary outcomes: filter life span during CKRT and 90-day all-cause mortality; extended follow-up for longer-term outcomes (including 1-year mortality).

Key Results

This trial was stopped early. Recruitment was stopped at an interim analysis (adaptive group-sequential design) after 638 patients were randomised (596 in the primary analysis set) because the filter life span endpoint met a stopping boundary for efficacy while 90-day mortality met a futility criterion (low conditional power for achieving statistical significance).

Outcome Regional citrate anticoagulation Systemic heparin anticoagulation Effect p value / 95% CI Notes
Filter life span (hours), median (IQR) 47 (19–70) 26 (12–51) Median difference 15 h 95% CI 11 to 20; P<0.001 Hodges–Lehmann estimate; filters were routinely changed at 72 h per protocol.
90-day all-cause mortality 150/300 (KM 51.2%) 156/296 (KM 53.6%) Adjusted HR 0.79 95% CI 0.63 to 1.004; P=0.054 Unadjusted HR 0.91; 95% CI 0.72 to 1.13; P=0.38 (not co-primary).
Total treatment downtime (minutes), median (IQR) 120 (0–720) 300 (0–930) Difference −120 min 95% CI −240 to 0; P=0.01 Less cumulative interruption time with citrate.
Bleeding complications, n (%) 15 (5.1) 50 (16.9) OR 0.27 95% CI 0.15 to 0.49; P<0.001 Marked reduction in clinically adjudicated bleeding complications.
New culture-proven infection since start of dialysis, n (%) 204 (68.0) 164 (55.4) OR 1.71 95% CI 1.23 to 2.39; P=0.001 Direction opposite to bleeding benefit; causality uncertain (open-label, multiple secondary endpoints).
Hypophosphataemia, n (%) 44 (15.4) 18 (6.2) OR 2.74 95% CI 1.58 to 4.74; P<0.001 Electrolyte derangement more frequent with citrate-based protocols.
Severe alkalosis, n (%) 7 (2.4) 1 (0.3) OR 7.21 95% CI 0.89 to 58.30; P=0.04 Imprecise estimate (rare events).
Persistent kidney dysfunction at day 90 (survivors with available data), n (%) 27/95 (28.4) 15/100 (15.0) OR 2.25 95% CI 1.11 to 4.56; P=0.02 Supplementary outcome; denominators reflect survivors with creatinine follow-up available.
  • RCA achieved clear biological/technical separation: longer filters (median 47 vs 26 h) and less downtime (120 vs 300 min).
  • Mortality signal favoured citrate in the adjusted model (HR 0.79) but did not reach conventional statistical significance (P=0.054) and the trial was stopped early for futility on this endpoint.
  • Trade-off pattern: substantially fewer bleeding complications with citrate, but more culture-proven infections and more metabolic/electrolyte derangements (notably hypophosphataemia).

Internal Validity

  • Randomisation and allocation: Centralised randomisation using a minimisation approach with balancing factors (eg, centre and key prognostic covariates); allocation concealment was operationally feasible because assignment occurred centrally.
  • Post-randomisation exclusions: 42/638 randomised patients (6.6%) were excluded from the primary analysis set because consent could not be obtained after randomisation (19 citrate; 23 heparin), introducing potential (though modest) selection bias.
  • Follow-up completeness (mortality): 90-day mortality analyses censored 19 participants (17 withdrew consent; 2 other reasons), with vital status otherwise broadly complete.
  • Performance/detection bias (open-label): Clinicians were unblinded; circuit-related behaviours (thresholds for filter change, handling of access issues) could be influenced, although mortality is objective and circuit life span is largely time-stamped.
  • Protocol adherence and timing: Time from randomisation to CKRT initiation was similar (median 2.4 vs 2.5 h), supporting prompt delivery of assigned strategy.
  • Baseline comparability: Groups were clinically well matched at randomisation (eg, age 67.5 vs 67.6 years; APACHE II 28.4 vs 28.5; SOFA 11.5 vs 11.5; vasopressor use 92.6% vs 90.5%; invasive ventilation 80.7% vs 84.1%).
  • Heterogeneity: Multicentre design improves robustness; however, practice variation in CKRT delivery and infection ascertainment can still contribute to heterogeneity (partly mitigated by protocolised CKRT dose and filter-change rules).
  • Dose and separation of variable of interest: CKRT prescription was standardised (target effluent 30 mL/kg/h; blood flow >100 mL/min; routine 72-hour filter change); achieved separation included filter life span 47 vs 26 h and downtime 120 vs 300 min.
  • Co-interventions: Both arms received CKRT within the same protocol framework; nonetheless, differential electrolyte replacement practices (particularly phosphate) and catheter handling could plausibly influence secondary outcomes such as hypophosphataemia and infection.
  • Statistical rigour: Adaptive group-sequential stopping rules were prespecified, but early stopping materially reduced the precision for mortality and other patient-centred outcomes; co-primary endpoint structure imposes a conservative evidentiary bar.

Conclusion on Internal Validity: Overall, internal validity is moderate: randomisation and baseline balance were strong and technical separation was clear, but open-label delivery, post-randomisation consent exclusions, and early stopping limit the certainty of inferences for mortality and several secondary clinical outcomes.

External Validity

  • Population representativeness: Critically ill ICU patients with severe AKI requiring CKRT and high acuity (APACHE II ~28; SOFA ~11.5; vasopressors >90%) are typical of CKRT populations in high-resource systems.
  • Key exclusions: Patients with major contraindications to citrate or heparin (including those at very high risk for citrate accumulation or severe bleeding) were excluded, limiting applicability to these important subgroups.
  • Setting and implementation burden: Conducted in German ICUs with established CKRT capability; RCA requires protocolised monitoring and staff competence, which may be variably available in resource-limited environments.
  • Transferability across CKRT platforms: Findings are most applicable to settings with comparable effluent dosing, monitoring frequency, and filter-change practices (routine 72-hour changes), and may differ where practices vary.

Conclusion on External Validity: External validity is good for high-acuity ICU patients receiving CKRT in centres capable of delivering RCA with close biochemical monitoring, but is limited for patients with severe hepatic dysfunction/shock physiology and for systems without robust citrate protocols.

Strengths & Limitations

  • Strengths: Multicentre ICU RCT; protocolised CKRT prescription in both arms; clinically meaningful co-primary outcomes (circuit performance and mortality); clear separation in technical endpoints; pragmatic relevance to routine ICU CKRT.
  • Limitations: Stopped early (596 analysed vs planned up to 1450), reducing power/precision for mortality and other patient-centred outcomes; open-label design with potential influence on circuit management and secondary outcomes; post-randomisation exclusions due to delayed consent; multiple secondary endpoints increase the risk of chance findings (eg, infection and renal recovery signals).

Interpretation & Why It Matters

  • Circuit efficacy
    RCA materially improves circuit performance (longer filters; less downtime), which is likely to increase delivered dose reliability and reduce blood loss from premature circuit failure.
  • Patient-centred outcomes
    The trial did not demonstrate a statistically significant mortality benefit at 90 days; the near-threshold adjusted hazard ratio suggests a possible modest benefit, but early stopping and wide confidence intervals mean a clinically relevant effect (benefit or no effect) remains plausible.
  • Safety trade-offs
    Lower bleeding risk with RCA is counterbalanced by higher rates of biochemical complications (hypophosphataemia; alkalosis) and a higher rate of culture-proven infection, reinforcing the need for protocolised monitoring, electrolyte replacement strategies (including phosphate), and careful catheter handling.

Controversies & Subsequent Evidence

  • Early stopping complicates mortality inference: The trial crossed an efficacy boundary for the technical endpoint (filter life span) while stopping for futility on mortality; this design decision makes the mortality estimate intrinsically less precise and heightens type II error risk despite the adjusted HR favouring citrate. 1
  • Discordant secondary signals (infection; renal recovery): The increase in culture-proven infection and persistent kidney dysfunction at day 90 emerged from secondary/supplementary analyses in an open-label trial, and should be interpreted cautiously in the context of multiplicity and potential measurement/practice effects.
  • Post-RICH evidence synthesis: Later systematic reviews/network meta-analyses incorporating RICH continue to show improved circuit life and reduced bleeding with citrate compared with heparin; effects on mortality remain uncertain and sensitive to between-trial heterogeneity and small-study influence. 2
  • Guideline positioning after RICH: Contemporary expert guidance continues to recommend RCA as first-line anticoagulation for CKRT when no major contraindications exist, explicitly emphasising protocolised monitoring for citrate accumulation and electrolyte/acid–base derangements. 345

Summary

  • Multicentre ICU RCT comparing regional citrate vs systemic heparin anticoagulation during CKRT for severe AKI.
  • Stopped early at an interim analysis (596 analysed), limiting precision for mortality and other patient-centred outcomes.
  • RCA substantially prolonged filter life (median 47 vs 26 hours) and reduced downtime (120 vs 300 minutes).
  • RCA reduced bleeding complications (5.1% vs 16.9%) but increased culture-proven infection (68.0% vs 55.4%).
  • No statistically significant difference in 90-day mortality (adjusted HR 0.79; 95% CI 0.63 to 1.004; P=0.054); biochemical complications (notably hypophosphataemia) were more frequent with RCA.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Prespecified subgroup analysis (surgical vs non-surgical): filter life span benefit with citrate was consistent; 90-day mortality did not show a statistically significant interaction; infection difference appeared larger in non-surgical patients (interaction P reported in supplementary material).
  • Interpretation of infection and renal recovery signals should consider multiplicity and open-label delivery alongside clear benefits in bleeding and circuit performance.

Overall Takeaway

RICH is a landmark CKRT anticoagulation trial because it demonstrated, at multicentre scale, that regional citrate meaningfully improves circuit performance and reduces bleeding compared with systemic heparin in severely ill ICU patients with AKI. However, early stopping limited the trial’s ability to definitively establish whether these technical and safety advantages translate into improved survival or renal recovery, underscoring the continuing need to balance circuit efficacy against metabolic and infectious complications.

Overall Summary

  • RCA improves CKRT delivery (longer filter life; less downtime) and lowers bleeding compared with systemic heparin.
  • Mortality benefit was not proven; early stopping reduced precision for patient-centred endpoints.
  • RCA requires vigilant monitoring due to higher rates of hypophosphataemia/alkalosis and a higher rate of culture-proven infection in this trial.

Bibliography