Publication

• Title: Sodium bicarbonate therapy for patients with severe metabolic acidaemia in the intensive care unit (BICAR-ICU): a multicentre, open-label, randomised controlled, phase 3 trial
• Acronym: BICAR-ICU
• Year: 2018
• Journal published in: The Lancet
• Citation: Jaber S, Paugam C, Futier E, Lefrant JY, Lasocki S, Lescot T, et al. Sodium bicarbonate therapy for patients with severe metabolic acidaemia in the intensive care unit (BICAR-ICU): a multicentre, open-label, randomised controlled, phase 3 trial. Lancet. 2018;392(10141):31-40.

Context & Rationale

• Background

  • Severe metabolic acidaemia in critical illness is associated with impaired cardiac contractility, vasoplegia, arrhythmias, hyperkalaemia, reduced catecholamine responsiveness, and high mortality.
  • Intravenous sodium bicarbonate is widely used to raise arterial pH and to temporise acidaemia-related complications (particularly hyperkalaemia) while treating the underlying cause, but robust randomised evidence in modern ICU cohorts was limited and practice varied.
  • Potential harms include hypernatraemia, fluid overload, metabolic alkalosis, and reduced ionised calcium; bicarbonate therapy can also alter clinician behaviour (eg, renal replacement therapy decisions) by modifying pH and potassium.
• Research Question/Hypothesis

  • In critically ill adults with severe metabolic acidaemia (pH ≤7.20), does IV sodium bicarbonate titrated to a pH target (>7.30) reduce death by day 28 and/or persistent organ failure at day 7 compared with no sodium bicarbonate?
  • Does any treatment effect differ in the pre-specified subgroup with moderate-to-severe acute kidney injury (AKIN stage 2–3), where bicarbonate could plausibly avert or delay renal replacement therapy?
• Why This Matters

  • Defines whether alkalinisation is a disease-modifying intervention or mainly a physiological “bridge” with uncertain patient-centred benefit.
  • Clarifies whether bicarbonate can safely reduce or delay renal replacement therapy (RRT), an invasive therapy with important resource and complication implications.
  • Provides an evidence base for guideline thresholds (particularly in sepsis/AKI-associated acidaemia) and informs bedside risk–benefit discussions.

Design & Methods

• Research Question: Among ICU patients with severe metabolic acidaemia, does IV sodium bicarbonate (target arterial pH >7.30) reduce the composite of 28-day mortality and/or organ failure at day 7 compared with no sodium bicarbonate?
• Study Type: Multicentre, open-label, randomised controlled, investigator-initiated, phase 3 trial in 26 ICUs in France; 1:1 allocation with stratification/minimisation by centre, age ≥65 years, sepsis status, and AKIN stage 2–3.
• Population:
  • Setting/timing: Adult ICU patients enrolled within 48 hours of ICU admission.
  • Key inclusion: arterial pH ≤7.20; PaCO₂ ≤45 mm Hg; serum bicarbonate ≤20 mmol/L; plus SOFA ≥4 or arterial lactate ≥2 mmol/L.
  • Key exclusions: respiratory acidosis; large bicarbonate losses (digestive/urinary, ≥1500 mL/24 h); stage IV chronic kidney disease (eGFR <30 mL/min/1.73 m²); ketoacidosis; sodium bicarbonate infusion or RRT in the preceding 24 hours; other protocol exclusions (not fully enumerated in the main paper).
• Intervention:
  • IV sodium bicarbonate 4.2% solution.
  • Delivered as 125–250 mL over 30 minutes, repeated as needed to target arterial pH >7.30.
  • Maximum permitted dose 1000 mL within the first 24 hours after enrolment.
• Comparison:
  • No sodium bicarbonate infusion protocol (usual care).
  • RRT co-intervention framework (both groups): RRT was recommended when ≥2 of the following were present: urine output <0.3 mL/kg/h for 24 h; arterial pH <7.20 despite resuscitation; serum potassium >6.5 mmol/L; RRT was mandated immediately for hyperkalaemia with ECG changes and/or pulmonary oedema with anuria.
• Blinding: Unblinded (open-label); objective outcomes (mortality) less susceptible, but clinician-driven outcomes (eg, RRT initiation) potentially influenced.
• Statistics: A total of 376 patients were required to detect a 15% absolute difference in the primary composite outcome with 80% power at α=0.03; target enrolment was 400 to accommodate losses; primary analysis was intention-to-treat; one planned interim analysis occurred after 200 patients.
• Follow-Up Period: Day 7 (organ failure) and day 28 (mortality and secondary outcomes through ICU stay/discharge as applicable).

Key Results

This trial was not stopped early. 400 patients were randomised; 389 were analysed after withdrawal of consent (control n=194; sodium bicarbonate n=195).

• Despite achieving biochemical separation (eg, arterial pH >7.30 at 12 h: 142/195 [73%] vs 84/194 [43%]; pH >7.30 maintained ≥36 h: 118/195 [60%] vs 51/194 [26%]), the primary composite outcome did not differ significantly overall.
• RRT utilisation was lower and RRT initiation occurred later in the bicarbonate group (overall RRT by day 28: 35% vs 52%; median time-to-RRT: 19 h vs 7 h), with the strongest signal in the pre-specified AKIN 2–3 stratum.
• Metabolic complications were more frequent with bicarbonate (hypernatraemia, alkalosis, hypocalcaemia), supporting the need for close electrolyte and acid–base monitoring.

Internal Validity

• Randomisation and allocation: Central allocation with minimisation/stratification (centre, age ≥65, sepsis, AKIN 2–3) supports balance and reduces selection bias.
• Dropout/exclusions: 11/400 (2.8%) withdrew consent after randomisation (7 control; 4 bicarbonate), leaving 389 analysed; attrition was small but not zero.
• Protocol violations: Violations of inclusion/exclusion criteria occurred (control 22/194 [11%]; bicarbonate 15/195 [8%]), which can dilute effects and introduce heterogeneity.
• Performance/detection bias: Open-label design increases risk of differential co-interventions and outcome ascertainment for clinician-mediated endpoints (especially RRT initiation).
• Protocol adherence and contamination: Cross-over occurred (47/194 [24%] controls received bicarbonate; 1/195 [1%] in the bicarbonate arm did not receive it), likely biasing towards the null for between-group comparisons.
• Baseline comparability (selected): Median age 66 vs 65 years; sepsis 119/195 (61%) vs 118/194 (61%); AKIN 2–3 92/195 (47%) vs 90/194 (46%); SOFA 10.0 (IQR 8.0–12.0) vs 10.0 (8.0–13.0); arterial pH 7.15 (7.09–7.18) vs 7.15 (7.11–7.18); invasive mechanical ventilation 160/195 (82%) vs 160/194 (83%); vasopressors 159/195 (82%) vs 155/194 (80%).
• Timing: The intervention was delivered early (time to first bicarbonate injection: 0.2 h [0.1–0.4] vs 7.0 h [1.6–15.5]), supporting biological plausibility for early pH correction.
• Dose and separation of the variable of interest: Bicarbonate volume from enrolment to 24 h: 500 mL (250–750) vs 0 mL (0–0); biochemical target separation at 12 h and ≥36 h was substantial.
• Heterogeneity: The overall trial was neutral for the primary composite endpoint, while the pre-specified AKIN 2–3 stratum showed benefit signals for several outcomes; this pattern increases the importance of interaction testing, multiplicity considerations, and replication.
• Adjunctive therapy use: Clinician-triggered interventions (notably RRT) are intrinsically linked to acid–base status; because acidaemia criteria were part of RRT recommendations, bicarbonate therapy plausibly alters the threshold for “meeting” RRT triggers.
• Outcome assessment: Mortality is objective; organ failure relied on SOFA-based assessment; RRT initiation is clinically meaningful but partially discretionary and vulnerable to bias in open-label designs.
• Statistical rigour: The trial nearly met its planned sample size (389 analysed vs 400 target); intention-to-treat analysis was used; the primary outcome was negative, reducing risk of overinterpretation of secondary endpoints, but multiple secondary and subgroup analyses remain interpretation-sensitive.

Conclusion on Internal Validity: Overall internal validity is moderate; randomisation and baseline balance were strong, but open-label conduct with substantial control-group cross-over and the clinician-mediated nature of RRT outcomes meaningfully increase risk of bias and dilute treatment separation.

External Validity

• Population representativeness: Enrolled patients were severely unwell (high vasopressor and ventilation use; SOFA around 10) with frequent sepsis, aligning with many real-world ICU acidaemia presentations.
• Important exclusions: Findings do not directly apply to diabetic ketoacidosis, predominant respiratory acidosis, large bicarbonate loss states, or advanced chronic kidney disease (eGFR <30 mL/min/1.73 m²).
• System/setting dependence: Conducted in French ICUs with defined RRT recommendation rules and ready access to blood gas/electrolyte monitoring; practice patterns, thresholds, and resource constraints may differ internationally.
• Intervention feasibility: Requires frequent arterial blood gas and electrolyte reassessment, and capacity to manage hypernatraemia/alkalosis/hypocalcaemia; feasibility and safety may be reduced in resource-limited environments.

Conclusion on External Validity: External validity is moderate; applicability is strongest to well-resourced ICUs managing severe metabolic acidaemia (especially with AKI), but generalisability is limited by key exclusions and by RRT practice variability across systems.

Strengths & Limitations

• Strengths:
  • Large, pragmatic, multicentre ICU randomised trial addressing a common, high-mortality physiological derangement.
  • Pre-specified stratification for AKI severity (AKIN 2–3) with clinically coherent mechanistic rationale (potential to avert/delay RRT).
  • Clearly specified bicarbonate concentration, dosing approach, pH target, and dose cap, enabling protocol reproducibility.
  • Clinically relevant secondary endpoints including RRT utilisation, timing, and dialysis dependence at ICU discharge.
• Limitations:
  • Open-label design with potential performance bias, especially for clinician-mediated endpoints (eg, RRT initiation).
  • Substantial control-group contamination (24% received bicarbonate), likely attenuating between-group differences.
  • Primary composite endpoint integrates outcomes with different causal proximity to bicarbonate (mortality vs day-7 organ failure), and day-7 organ failure was assessed only in survivors.
  • Subgroup signals (AKIN 2–3) occurred in a trial with a neutral primary outcome, increasing the need for cautious interpretation and replication.
  • Metabolic adverse events (hypernatraemia, alkalosis, hypocalcaemia) were more frequent with bicarbonate, constraining indiscriminate use.

Interpretation & Why It Matters

• Clinical meaning of a neutral primary endpoint

  Routine bicarbonate infusion for all ICU patients meeting this “severe metabolic acidaemia” phenotype is not supported; the primary composite (66% vs 71%) and 28-day mortality (45% vs 53%) did not achieve statistical separation overall.
• Where the signal lies

  Patients with moderate-to-severe AKI (AKIN 2–3) demonstrated lower 28-day mortality (46% vs 63%) and markedly less RRT exposure (51% vs 73%), suggesting bicarbonate may be most relevant as a temporising or RRT-sparing strategy in severe acidaemia with AKI.
• Risk–benefit and bedside implementation

  The metabolic harms (eg, hypernatraemia 49% vs 29%; alkalosis 16% vs 9%; hypocalcaemia 24% vs 15%) make careful monitoring and avoidance of overcorrection essential; bicarbonate is a targeted adjunct, not a substitute for prompt treatment of the underlying aetiology of acidaemia.

Controversies & Subsequent Evidence

• Composite endpoint selection (death by day 28 and/or day-7 organ failure) and the survivor-only assessment of day-7 organ failure complicate interpretation of the composite; the accompanying Lancet commentary emphasised that bicarbonate corrects a physiological variable and may not alter the underlying illness trajectory¹.
• Open-label design and substantial control-group cross-over (24%) raise concern for treatment dilution and clinician-behaviour effects; this is particularly relevant to RRT outcomes, because acidaemia and potassium are part of typical RRT triggers and are directly modified by bicarbonate.
• Guidelines published after BICAR-ICU incorporated a conditional role for bicarbonate in severe metabolic acidaemia—especially when accompanied by AKI—while generally avoiding endorsement of routine buffer therapy for all causes of lactic acidosis²³.
• Systematic reviews and meta-analyses published after BICAR-ICU have generally found uncertain effects on mortality with low-to-moderate certainty evidence, while supporting a plausible reduction in RRT utilisation or delayed initiation in selected phenotypes⁴⁵.
• Contemporary observational work (including target-trial emulation approaches) has reported associations between bicarbonate administration and improved outcomes in selected cohorts, but remains limited by residual confounding and treatment-by-indication bias⁶.
• The subsequent large RCT BICARICU-2 (AKI-focused) found no difference in day-90 mortality (62.1% vs 61.7%), while kidney replacement therapy use was lower (35% vs 50%), supporting an RRT-sparing effect but tempering claims of a consistent mortality benefit⁷⁸.

Summary

• BICAR-ICU randomised 400 ICU patients with severe metabolic acidaemia (pH ≤7.20) to IV 4.2% sodium bicarbonate (target pH >7.30; max 1 L/24 h) vs no bicarbonate.
• The primary composite endpoint (death by day 28 and/or organ failure at day 7) was not significantly reduced overall (66% vs 71%).
• Bicarbonate was associated with less RRT use overall (35% vs 52%) and later RRT initiation (median 19 h vs 7 h).
• In the pre-specified AKIN 2–3 stratum, bicarbonate was associated with lower 28-day mortality (46% vs 63%) and less RRT (51% vs 73%), but subgroup inference remains interpretation-sensitive given an overall neutral primary outcome.
• Metabolic adverse events were more frequent with bicarbonate (hypernatraemia, alkalosis, hypocalcaemia), mandating careful monitoring and avoidance of overcorrection.

Overall Takeaway

BICAR-ICU provided the first large modern ICU RCT to test protocolised bicarbonate therapy for severe metabolic acidaemia and showed no overall benefit on its composite primary endpoint, despite clear biochemical separation. Its most influential contribution was the signal that bicarbonate may reduce or delay RRT—especially in patients with moderate-to-severe AKI—while also clearly demonstrating meaningful metabolic harms that mandate careful monitoring and selective use.

Bibliography

• 1Kraut JA, Madias NE. Sodium bicarbonate for severe metabolic acidaemia. Lancet. 2018;392(10141):3-4.
• 2Jung B, Martinez M, Claessens YE, et al; Société de Réanimation de Langue Française (SRLF); Société Française de Médecine d’Urgence (SFMU). Diagnosis and management of metabolic acidosis: guidelines from a French expert panel. Ann Intensive Care. 2019;9(1):92.
• 3Evans L, Rhodes A, Alhazzani W, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock 2021. Crit Care Med. 2021;49(11):e1063-e1143.
• 4Lo KB, Garcıa L, DeJoy R III, et al. Outcomes in patients treated with sodium bicarbonate for metabolic acidosis: a meta-analysis. Heart Lung. 2020;49(2):91-96.
• 5Korang SK, Juul S, Nielsen EE, et al. Sodium bicarbonate supplementation for critically ill patients with metabolic acidosis. Cochrane Database Syst Rev. 2021;CD014371.
• 6Blank SP, Blank RM, Laupland KB, et al; Queensland Critical Care Research Network (QCCRN). Sodium bicarbonate administration for metabolic acidosis in the intensive care unit: a target trial emulation. Intensive Care Med. 2025;51(6):1078-1086.
• 7Jung B, Jabaudon M, De Jong A, et al; for the BICARICU-2 Study Group. Sodium bicarbonate for severe metabolic acidemia and acute kidney injury: the BICARICU-2 randomized clinical trial. JAMA. 2025;334;(22):2000-2010.
• 8Kusirisin P, Zampieri FG, Bagshaw SM. Sodium Bicarbonate in Severe Acidemia and Acute Kidney Injury—Turning the Tide or Chasing a Myth. JAMA. 2025;334;(22):1986-1987.