Publication

• Title: Sodium Bicarbonate for Critically Ill Adults with Metabolic Acidosis and Shock
• Acronym: SODa-BIC
• Year: 2026
• Journal published in: New England Journal of Medicine
• Citation: The SODa-BIC Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group. Sodium bicarbonate for critically ill adults with metabolic acidosis and shock. N Engl J Med. Published online June 12, 2026.

Context & Rationale

• Background

  Metabolic acidosis is common in critical illness and is associated with myocardial depression, reduced catecholamine responsiveness, acute kidney injury, renal replacement therapy, and death.
  
  Binational observational data from Australia and New Zealand suggested that early metabolic acidosis affected approximately 12% of ICU admissions and was associated with markedly higher hospital mortality, renal replacement therapy use, and MAKE30 than absence of metabolic acidosis.¹
  
  Sodium bicarbonate reliably raises extracellular pH, bicarbonate, base excess, and sodium, but previous trials had not established whether biochemical correction translates into patient-centred benefit.
  
  Prior randomised evidence was dominated by severe-acidaemia populations, especially BICAR-ICU, which was open-label and neutral overall but generated a renal-signal in the acute kidney injury subgroup.²
  
  BICARICU-2 subsequently enrolled a more selected population with severe metabolic acidemia and moderate-to-severe acute kidney injury and found no day-90 mortality benefit, despite less kidney replacement therapy in the bicarbonate group.³
  
  Sepsis guidance before SODa-BIC therefore supported only a narrow, low-certainty role for bicarbonate in septic shock with severe metabolic acidaemia and acute kidney injury, not routine bicarbonate for all vasopressor-dependent metabolic acidosis.⁴
• Research Question/Hypothesis

  SODa-BIC asked whether protocolised intravenous sodium bicarbonate, compared with an indistinguishable 5% dextrose placebo, would reduce major adverse kidney events within 30 days in critically ill adults with metabolic acidosis receiving vasopressors.
  
  The key methodological advance was the double-blind design in a population with less severe acidosis than BICAR-ICU and BICARICU-2: pH <7.30 rather than pH ≤7.20.
• Why This Matters

  Sodium bicarbonate is cheap, familiar, physiologically active, and widely used, yet clinicians have had little high-quality evidence to decide whether correcting acidaemia improves outcomes or simply improves the blood gas.
  
  The decision to initiate renal replacement therapy is often influenced by pH; therefore, an unblinded trial can make bicarbonate appear renoprotective by delaying dialysis rather than by improving kidney biology.
  
  SODa-BIC directly tested whether early correction of moderate metabolic acidosis in shock improves a clinically meaningful kidney-centred composite, rather than merely improving acid–base numbers.

Design & Methods

• Research Question: Among critically ill adults with metabolic acidosis and vasopressor-dependent shock, does a protocolised sodium bicarbonate infusion, compared with 5% dextrose placebo, reduce major adverse kidney events within 30 days?
• Study Type: Pragmatic, adaptive, international, multicentre, double-blind, placebo-controlled, parallel-group, superiority randomised clinical trial conducted in 55 ICUs across seven countries.
• Population:
  • Adults aged ≥18 years receiving a continuous vasopressor infusion to maintain mean arterial pressure >65 mm Hg, or another clinician-determined target.
  • Metabolic acidosis had to be documented within 2 hours before randomisation: pH <7.30, base excess ≤−4 mmol/L, and PaCO2 ≤45 mm Hg if not intubated or ≤50 mm Hg if intubated.
  • A dedicated central or peripheral line had to be available, or planned within 1 hour after randomisation.
  • Major exclusions included fulfilment of eligibility criteria more than 48 hours earlier; diabetic ketoacidosis; clinically significant digestive or urinary bicarbonate loss; estimated GFR <30 mL/min due to chronic kidney disease; current sodium bicarbonate infusion; current or planned renal replacement therapy within 3 hours; severe dysnatraemia; severe hypokalaemia; hypocalcaemia; pulmonary oedema with PaO2/FiO2 <100; overdose or intoxication; pregnancy or breastfeeding; high risk of cerebral oedema; imminent death; life expectancy <30 days; or clinician judgement that enrolment was not in the patient’s best interests.
• Intervention:
  • Sodium bicarbonate 8.4% was diluted in a 500 mL 5% dextrose bag to a final sodium bicarbonate concentration of 600 mmol/L.
  • The infusion had to start within 1 hour after randomisation.
  • The infusion started at 100 mL/h and was titrated using the same bedside algorithm in both trial groups.
  • The physiological target was pH ≥7.30 and base excess ≥0 mmol/L.
  • If pH was 7.30 to 7.35 and base excess was ≥0 mmol/L, the infusion rate was reduced to 25 mL/h.
  • If pH was >7.35 and base excess was >0 mmol/L, the infusion was stopped.
  • The maximum trial infusion duration was 5 hours; after this, bicarbonate use was at clinician discretion.
• Comparison:
  • Placebo was a standard 500 mL 5% dextrose bag.
  • Placebo was infused using the identical algorithm, with the same blood gas schedule and the same stopping and titration rules.
  • Open-label sodium bicarbonate after enrolment was allowed but discouraged.
  • All other treatments, including fluids, vasopressors, ventilation, renal replacement therapy, and other organ support, were determined by the treating clinical team.
• Blinding: Patients and treating clinicians were blinded to trial allocation. Study drug preparation was separated from outcome data collection. Blinding was strengthened by using macroscopically indistinguishable solutions, but biochemical separation in pH and sodium could plausibly have allowed some bedside inference of allocation.
• Statistics: A total of 470 patients was required to detect a 14 percentage-point absolute reduction in MAKE30 from an assumed placebo-group event rate of 40%, with 90% power at a two-sided alpha of 0.05; the target was increased to 500 to allow for loss to follow-up or withdrawal. The trial used a blinded sample-size re-estimation after 100 patients and a planned maximum cap of 700, but the DSMC recommended continuing with the original sample size. The published protocol and statistical analysis plan specified the adaptive double-blind design and primary MAKE30 outcome.⁵ The primary analysis used a modified intention-to-treat population, excluding patients who withdrew consent for use of data, and modelled adjusted absolute differences with site as a random effect and stratification factors as fixed effects.
• Follow-Up Period: The primary outcome was assessed within 30 days, censored at hospital discharge or day 30. Secondary follow-up included ICU death by day 30, in-hospital death by day 90, and process-of-care outcomes to day 30 or day 90. Planned longer-term and health economic analyses were not reported in the primary manuscript.

Key Results

This trial was not stopped early. One interim review after 125 patients had completed 90-day follow-up recommended continuation without protocol change, and the blinded sample-size re-estimation after 100 patients retained the planned sample size.

• Sodium bicarbonate produced strong biochemical separation: median lowest pH in the first 24 hours was 7.31 vs 7.26, median highest bicarbonate was 26.0 vs 21.0 mmol/L, and median highest base excess was 1.5 vs −4.2 mmol/L.
• Despite this physiological effect, there was no reduction in MAKE30, mortality, acute kidney injury, vasopressor-free days, ICU-free days, or hospital-free days.
• Subgroup results did not demonstrate credible benefit, including in patients with pH <7.25 (45/93 [48.4%] vs 53/99 [53.5%]; adjusted absolute difference −5.1 percentage points; 95% CI −19.1 to 8.9) or acute kidney injury stage 2 or 3 (30/53 [56.6%] vs 33/46 [71.7%]; adjusted absolute difference −15.2 percentage points; 95% CI −34.1 to 3.8).

Internal Validity

• Randomisation and Allocation: Randomisation used a secure web-based system, permuted blocks, and stratification by centre, pH (<7.25 vs ≥7.25), and creatinine (<150 vs ≥150 μmol/L). Allocation concealment appears strong.
• Drop out or exclusions: A total of 500 patients underwent randomisation; 2 withdrew consent, leaving 498 in the modified intention-to-treat population. Primary outcome data were available for all 498 analysed patients. Five analysed patients did not receive any infusion: 1/244 in the bicarbonate group and 4/254 in the placebo group.
• Performance/Detection Bias: Blinding was a major strength relative to BICAR-ICU and BICARICU-2. Treating clinicians were blinded, but the acid–base response itself could have partly revealed allocation because pH, bicarbonate, base excess, and sodium diverged quickly.
• Protocol Adherence: Trial infusion was delivered to 493/498 patients (99.0%). Median infusion duration was 5.0 hours in both groups. Protocol deviations were not absent: infusion rate was not adjusted according to protocol in 33/244 (13.5%) vs 7/254 (2.8%), and ineligible patients were randomised in 12/244 (4.9%) vs 15/254 (5.9%).
• Baseline Characteristics: Baseline balance was acceptable. Median age was 65 vs 68 years; female sex 45.5% vs 40.9%; median APACHE II 21 vs 22; median SOFA 7 vs 7; median pH 7.26 vs 7.26; median base excess −9.2 vs −8.8 mmol/L; acute kidney injury 113/243 (46.5%) vs 115/251 (45.8%); septic shock 38.9% vs 40.9%; and median noradrenaline dose 0.14 vs 0.15 μg/kg/min.
• Heterogeneity: Clinical heterogeneity was substantial but expected: sepsis/septic shock, cardiogenic shock, hypovolaemia, cardiac arrest, postoperative states, vasoplegia, and other causes were represented. Stratification and mixed-effects modelling addressed some statistical heterogeneity, but biological heterogeneity remains important because bicarbonate plausibly has different effects in lactic acidosis, renal acidosis, hyperchloraemic acidosis, and mixed disorders.
• Timing: Treatment was delivered early. Median time from eligibility determination to randomisation was 39.9 minutes vs 37.8 minutes, and median time from randomisation to infusion start was 0.5 vs 0.4 hours. Median time from ICU admission to randomisation in the supplement was 8.8 vs 8.7 hours, supporting enrolment during the early shock trajectory.
• Dose: The sodium bicarbonate group received a median total bicarbonate dose of 300.0 mmol (IQR 197.4 to 300.0), while the placebo group received 0 mmol. The tested dose was sufficient to correct acid–base variables but may not answer whether repeated or longer-duration bicarbonate exposure would change outcomes.
• Separation of the Variable of Interest: Treatment separation was robust. Median lowest pH in the first 24 hours was 7.31 vs 7.26, median highest pH was 7.43 vs 7.37, median lowest bicarbonate was 19.0 vs 17.0 mmol/L, median highest bicarbonate was 26.0 vs 21.0 mmol/L, median lowest base excess was −6.0 vs −9.0 mmol/L, and median highest base excess was 1.5 vs −4.2 mmol/L.
• Key Delivery Aspects: The protocol corrected acidaemia quickly, reduced recurrence of metabolic acidosis, and reduced open-label bicarbonate exposure, but did not alter vasopressor exposure, renal outcomes, or mortality. This is the core internal signal: the intervention was delivered and biologically active, but the biological surrogate did not translate into clinical benefit.
• Crossover: Open-label sodium bicarbonate was more common in placebo recipients: 17/243 (7.0%) in the bicarbonate group vs 38/250 (15.2%) in the placebo group, adjusted absolute difference −8.2 percentage points (95% CI −13.6 to −2.8). This may have attenuated treatment contrast, although it also confirms that clinicians perceived more need for rescue bicarbonate in placebo patients.
• Adjunctive therapy use: Other organ support was clinician-directed. Renal replacement therapy initiation was not protocolised, and physiological data at the time of renal replacement therapy initiation were not collected, limiting mechanistic interpretation of the renal replacement therapy signal.
• Outcome Assessment: MAKE30 is objective and patient-centred but composite. Its components are not equally patient-centred: mortality is unequivocal, persistent renal dysfunction is biologically meaningful, and renal replacement therapy is clinically important but partly clinician-dependent.
• Statistical Rigor: The SAP was published before enrolment was completed, the primary analysis adjusted for stratification factors and centre clustering, and sensitivity analyses were consistent with the primary analysis. Secondary and subgroup analyses were not multiplicity-adjusted and should be interpreted as exploratory.

Conclusion on Internal Validity: Internal validity appears strong. Randomisation, allocation concealment, blinding, near-complete follow-up, prespecified analysis, and clear physiological separation support confidence that the neutral clinical result is real for the tested intervention, dose, duration, and population.

External Validity

• Population Representativeness: The trial population maps well to ICU patients with vasopressor-dependent shock and moderate metabolic acidosis. Median pH was 7.26, median base excess was approximately −9 mmol/L, median lactate was 2.5 mmol/L, 78–82% were mechanically ventilated, and roughly two thirds had sepsis.
• Screening and exclusions: Of 3060 assessed patients, 500 were randomised. Major exclusions included current or planned renal replacement therapy (641), chronic kidney disease with estimated GFR <30 mL/min (443), imminent or inevitable death (223), eligibility criteria fulfilled >48 hours earlier (205), acute brain injury or intracranial hypertension risk (167), diabetic ketoacidosis (145), current bicarbonate therapy (109), and participation not in the patient’s best interest (106).
• Geography: The study was international, but enrolment was dominated by Australia and New Zealand: 359/498 (72.1%) from Australia and 109/498 (21.9%) from New Zealand. Other countries contributed small numbers: Oman 16 (3.2%), India 8 (1.6%), Saudi Arabia 3 (0.6%), Brazil 2 (0.4%), and Japan 1 (0.2%).
• Applicability: Findings apply best to high-resource ICUs with frequent blood gas monitoring, vasopressor-dependent patients, and moderate metabolic acidosis where immediate renal replacement therapy is not already planned.
• Limited applicability: The results should not be overextended to diabetic ketoacidosis, toxicological indications, bicarbonate-losing states, severe chronic kidney disease, established or imminent renal replacement therapy, pregnancy, paediatric critical care, traumatic brain injury, or patients with very severe acidaemia for whom clinicians believe bicarbonate is mandatory.
• Resource-limited settings: Sodium bicarbonate itself is inexpensive and widely available, but the tested approach required timely randomisation, dedicated infusion, repeated blood gases, and protocolised titration. Implementation may be more difficult where arterial blood gas availability is limited.

Conclusion on External Validity: External validity is moderate to strong for vasopressor-dependent ICU patients with moderate metabolic acidosis in high-resource settings. Generalisability is more limited for severe acidaemia with advanced AKI, bicarbonate-loss states, current dialysis planning, and settings where blood gas-guided titration cannot be delivered reliably.

Strengths & Limitations

• Strengths:
  • Large double-blind placebo-controlled trial in an area previously dominated by open-label trials.
  • International multicentre enrolment across 55 ICUs.
  • Clinically relevant primary composite outcome with complete primary-outcome ascertainment in the modified intention-to-treat population.
  • Rapid delivery of the assigned intervention and clear biochemical separation.
  • Published protocol and statistical analysis plan before enrolment completion.⁵
  • Low rate of consent withdrawal and minimal missingness.
• Limitations:
  • Powered for a large 14 percentage-point absolute difference; smaller benefits or harms cannot be excluded.
  • Single short 5-hour infusion may not test longer or repeated bicarbonate strategies.
  • Physiological response may have partially revealed allocation despite formal blinding.
  • Open-label bicarbonate was allowed and was more common in placebo patients, potentially diluting treatment contrast.
  • Renal replacement therapy initiation was not protocolised and may vary by clinician and site.
  • Physiology at renal replacement therapy initiation was not collected.
  • Exclusions limit applicability to severe chronic kidney disease, current or planned dialysis, diabetic ketoacidosis, bicarbonate losses, intoxications, and high cerebral oedema risk.
  • Geographic enrolment was heavily weighted toward Australia and New Zealand.

Interpretation & Why It Matters

• Biochemistry is not outcome

  Sodium bicarbonate corrected pH, bicarbonate, and base excess, but did not reduce MAKE30, mortality, acute kidney injury, vasopressor duration, ICU stay, or hospital stay.
• Renal signal remains uncertain

  Renal replacement therapy use was numerically lower with bicarbonate (16.8% vs 20.9%), but this was imprecise and did not translate into a significant reduction in MAKE30 or dialysis dependence.
• Routine use is not supported

  For vasopressor-dependent patients with pH <7.30 and base excess ≤−4 mmol/L, routine short-course bicarbonate to normalise acid–base variables should not be expected to improve kidney-centred outcomes.
• Selective rescue remains plausible

  SODa-BIC does not remove established or conventional uses of bicarbonate for selected indications such as severe hyperkalaemia, bicarbonate-loss acidosis, toxicological indications, or clinician-judged rescue in extreme acidaemia; these were not the tested question.

Controversies & Other Evidence

• Severe acidaemia with acute kidney injury remains a distinct phenotype. BICAR-ICU and BICARICU-2 enrolled much more severely acidotic patients than SODa-BIC, especially pH ≤7.20 populations, and BICARICU-2 specifically targeted moderate-to-severe acute kidney injury.²³ SODa-BIC therefore should not be used to claim definitive futility in severe acidaemia with AKI stage 2–3, but it substantially weakens the argument for routine bicarbonate in the broader pH <7.30 vasopressor population.
• Renal replacement therapy is a vulnerable outcome. The BICARICU-2 editorial emphasised that lower KRT use may be clinically important but is exposed to bias in open-label trials because acidaemia itself often triggers dialysis decisions.⁸ SODa-BIC improves on this through blinding, but treatment-induced pH separation could still have influenced clinicians’ beliefs about renal replacement therapy need.
• Guidelines lag the new evidence. Current sepsis guidance before SODa-BIC supported bicarbonate only weakly for septic shock with severe metabolic acidaemia and AKI, and suggested against bicarbonate for improving haemodynamics or reducing vasopressors in hypoperfusion-induced lactic acidaemia.⁴ SODa-BIC provides high-quality evidence against expanding bicarbonate routinely to less severe vasopressor-dependent acidosis.
• Observational evidence has been more favourable than randomised evidence. A 2025 target trial emulation of 6157 eligible Australian ICU admissions found bicarbonate administration associated with a small 1.9% absolute mortality reduction and a risk ratio of 0.86 (95% CI 0.80 to 0.91), but residual confounding and treatment-selection bias remain central concerns.⁹
• Meta-analytic inference will change. A 2026 RCT meta-analysis before SODa-BIC concluded that bicarbonate reduced renal replacement therapy requirement but mortality evidence remained inconclusive.¹⁰ SODa-BIC adds a large blinded neutral trial in moderate-acidaemia shock and should push future pooled estimates toward a more nuanced phenotype-specific interpretation.
• The “treating the number” critique is strengthened. A 2026 Journal of Intensive Care comment argued that correcting extracellular pH may not correct the intracellular variable of interest and that making pH a target risks Goodhart’s law: the surrogate becomes a treatment goal detached from meaningful clinical outcome.¹¹ SODa-BIC is consistent with this concern: the blood gas improved, but patients did not.
• Practice variability remains important. Surveys of clinicians before and during the SODa-BIC era showed heterogeneity in thresholds, dosing, infusion strategy, perceived benefits, and perceived harms of bicarbonate therapy.⁷¹² SODa-BIC should reduce routine use in moderate acidosis but is unlikely to eliminate clinician discretion in severe or mixed indications.
• The pilot trial predicted the physiology but not the clinical effect. The SODa-BIC pilot confirmed feasibility, faster pH and base-excess correction, and lower recurrence of acidosis, but was not powered for patient-centred outcomes.⁶ The phase 3 trial reproduced the physiological effect and tested the clinical hypothesis directly.

Summary

• SODa-BIC randomised 500 critically ill adults with metabolic acidosis and vasopressor-dependent shock to sodium bicarbonate or 5% dextrose placebo; 498 were analysed.
• The intervention was biologically active: bicarbonate improved pH, bicarbonate, base excess, sodium, and recurrence of metabolic acidosis.
• The primary outcome was neutral: MAKE30 occurred in 40.2% vs 39.4%; adjusted absolute difference 1.2 percentage points; 95% CI −7.1 to 9.4; P=0.78.
• Mortality, renal replacement therapy dependence, acute kidney injury, vasopressor-free days, ICU-free days, and hospital-free days were not improved.
• The trial argues against routine short-course bicarbonate for moderate metabolic acidosis in vasopressor-dependent ICU patients, while leaving selected severe-acidaemia or specific-indication use unresolved.

Overall Takeaway

SODa-BIC is a practice-shaping trial: it shows that rapid correction of moderate metabolic acidosis with sodium bicarbonate in vasopressor-dependent ICU patients does not improve 30-day major adverse kidney events. Its importance lies not in proving bicarbonate is ineffective, but in separating biochemical correction from clinical benefit and narrowing routine use to more selective indications.

Bibliography

• 1.Mochizuki K, Fujii T, Paul E, Anstey M, Pilcher DV, Bellomo R. Early metabolic acidosis in critically ill patients: a binational multicentre study. Crit Care Resusc. 2021;23:67-75.
• 2.Jaber S, Paugam C, Futier E, Lefrant JY, Lasocki S, Lescot T, et al; BICAR-ICU Study Group. 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;392:31-40.
• 3.Jung B, Jabaudon M, De Jong A, Bitker L, Audard J, Klouche K, et al; BICARICU-2 Study Group. Sodium bicarbonate for severe metabolic acidemia and acute kidney injury: the BICARICU-2 randomized clinical trial. JAMA. 2025;334:2000-2010.
• 4.Evans L, Rhodes A, Alhazzani W, Antonelli M, Coopersmith CM, French C, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med. 2021;47:1181-1247.
• 5.Serpa Neto A, McNamara M, Cooper J, Fujii T, Higgins A, Hodgson C, et al. Protocol summary and statistical analysis plan for the sodium bicarbonate for metabolic acidosis in the intensive care unit (SODa-BIC) trial. Crit Care Resusc. 2025;27:100108.
• 6.Serpa Neto A, Fujii T, McNamara M, Moore J, Young PJ, Peake S, et al. Sodium bicarbonate for metabolic acidosis in the ICU: results of a pilot randomized double-blind clinical trial. Crit Care Med. 2023;51:e221-e233.
• 7.Serpa Neto A, Fujii T, El-Khawas K, Udy A, Bellomo R. Sodium bicarbonate therapy for metabolic acidosis in critically ill patients: a survey of Australian and New Zealand intensive care clinicians. Crit Care Resusc. 2020;22:275-280.
• 8.Kusirisin P, Zampieri FG, Bagshaw SM. Sodium bicarbonate in severe acidemia and acute kidney injury—turning the tide or chasing a myth? JAMA. 2025;334:1986-1987.
• 9.Blank SP, Blank RM, Laupland KB, Tabah A, Gill D, Kumar A, et al; Queensland Critical Care Research Network. Sodium bicarbonate administration for metabolic acidosis in the intensive care unit: a target trial emulation. Intensive Care Med. 2025;51:1078-1086.
• 10.Chen JJ, Lee TH, Chang CH, Lai PC, Tu YK, Huang YT. Sodium bicarbonate for acute metabolic acidosis in critically ill adults: a meta-analysis of randomized clinical trials. Crit Care Med. Published online May 22, 2026.
• 11.Diop S, Aparicio M, Mounier R, Premachandra A. Sodium bicarbonate for “metabolic acidosis”: an illustration of the Goodhart’s law. J Intensive Care. 2026;14:23.
• 12.Caroli A, Zhou L, Chaba A, Eastwood G, Hikasa Y, Kitisin N, et al. Metabolic acidosis and sodium bicarbonate therapy in critically ill patients: a multinational survey of international and Chinese practitioners. Crit Care Resusc. 2026;28:100167.