Clinical Question
Among critically ill patients with severe metabolic acidemia (pH ≤7.20) and moderate to severe acute kidney injury, does sodium bicarbonate infusion reduce 90-day mortality compared to usual care?
TL;DR
Sodium bicarbonate didn't reduce mortality in critically ill patients with severe acidemia and acute kidney injury. However, it did reduce the need for dialysis by 15% and delayed time to dialysis initiation-without increasing adverse events. The trial suggests bicarbonate may be a safe temporizing measure to avoid or delay kidney replacement therapy, but it won't keep more patients alive.
The Deets
Sodium bicarbonate for severe acidemia is one of those interventions that everyone uses but nobody really knows if it works. We give it reflexively when pH drops below 7.20, especially in patients with acute kidney injury, under the assumption that buffering the acid will improve outcomes. But does it?
BICARICU-2 was designed to answer that question. The trial enrolled 627 critically ill patients across 43 French ICUs with severe metabolic acidemia (pH ≤7.20) and moderate to severe AKI (KDIGO stage 2 or 3). Patients were randomized to receive either 4.2% sodium bicarbonate infusions targeting a pH ≥7.30 or no bicarbonate (usual care). The median pH at enrollment was 7.15 in both groups, with lactate levels around 5.7-5.9 mmol/L. This was a sick population—77% were on mechanical ventilation, 80% were on vasopressors, and median SOFA scores were 10-11.
The intervention group received a median of 750 mL of bicarbonate over the first 48 hours, administered in 125-250 mL boluses targeting pH normalization. The control group received standard care without bicarbonate infusions—though notably, 15% of control patients crossed over and received bicarbonate anyway, typically as rescue therapy.
The primary outcome was 90-day all-cause mortality. The result? No difference. Mortality was 62.1% in the bicarbonate group versus 61.7% in controls (absolute difference 0.4%, 95% CI -7.2 to 8.0). The trial was powered to detect a 10% absolute reduction in mortality, which, in retrospect, was wildly optimistic. A more realistic effect size would have required a much larger trial.
Where bicarbonate did make a difference was in kidney replacement therapy (KRT) use. Only 35% of the bicarbonate group required dialysis compared to 50% of controls—a 15.5% absolute reduction. Among patients who did require KRT, the bicarbonate group had it initiated later (median 31 hours vs 16 hours). The main indications for starting dialysis were persistent oliguria, refractory acidemia, and elevated creatinine, criteria that bicarbonate could plausibly influence.
Secondary outcomes were mostly neutral. There was no difference in vasopressor use, mechanical ventilation days, ICU length of stay, or hospital length of stay. One intriguing finding: ICU-acquired bloodstream infections were 50% lower in the bicarbonate group (4% vs 9%). The investigators speculate this may be related to reduced dialysis catheter exposure, though the mechanism remains unclear.
This trial differs significantly from BICARICU-1, the predecessor study that generated the hypothesis for BICARICU-2. In BICARICU-1, moderate to severe AKI was only present in 47% of patients and was a randomization stratum, not an inclusion criterion. In that trial's AKI subgroup, bicarbonate appeared to reduce 28-day mortality from 63% to 46%—a large effect that prompted this confirmatory trial. BICARICU-2 enrolled a more targeted population with universal AKI, but mortality was lower than expected (62% instead of the assumed 70-80%), and the hypothesized mortality benefit didn't materialize.
The per-protocol analysis is revealing. Among the 47 control patients who crossed over to receive bicarbonate, 90-day mortality was 72%-higher than both the control group (62%) and the bicarbonate group (62%). This suggests bicarbonate crossover was a marker of disease severity, used as rescue therapy in the sickest patients.
Design
Multicenter, open-label, randomized controlled trial
N=627
Sodium bicarbonate infusion (n=314)
Usual care (n=313)
Follow-up: 90 days
43 ICUs across France
Population
Inclusion Criteria
Adults ≥18 years old
SOFA score >4 or lactate ≥2 mmol/L within 48 hours of ICU admission
Within 6 hours before enrollment:
pH ≤7.20
Serum bicarbonate ≤20 mEq/L
PaCO₂ ≤45 mm Hg (excluding respiratory acidosis)
Moderate to severe AKI (KDIGO stage 2 or 3)
Exclusion Criteria
Respiratory acidosis (PaCO₂ >45 mm Hg while breathing spontaneously)
Proven bicarbonate wasting (GI or urinary losses >1500 mL/24h)
Baseline GFR ≤10 mL/min
Ketoacidosis
Already received sodium bicarbonate or KRT within 24 hours
KRT planned for next 6 hours
Exogenous acid poisoning (metformin, salicylate, methanol, ethylene glycol)
Pregnancy or breastfeeding
Life expectancy <48 hours
Baseline Characteristics
From the bicarbonate group
Demographics: Median age 67 years (IQR 59-74), 62% male, BMI 27
Severity: SAPS II 61 (IQR 52-75), SOFA score 10 (IQR 8-13)
Comorbidities: Chronic hypertension 51%, alcohol misuse 28%, diabetes 27%, immunodeficiency 17%
Admission type: Medical 68%, surgical 32%
Main cause of acidemia: Septic shock 55%, hemorrhagic shock 14%, AKI 8%, cardiac arrest 6%
Organ support: Invasive mechanical ventilation 77%, vasopressors 81%
Lab values: pH 7.15 (IQR 7.08-7.18), lactate 5.9 mmol/L (IQR 3.0-10.8), serum bicarbonate 13 mEq/L (IQR 10-15), creatinine 2.2 mg/dL (IQR 1.5-3.2)
Interventions and Controls
Randomly assigned to a group:
Sodium bicarbonate infusion - 4.2% intravenous sodium bicarbonate targeting arterial pH ≥7.30 throughout the next 28 days or until ICU discharge. Each infusion consisted of 125-250 mL within 30 minutes, maximum 1000 mL within 24 hours. Blood gas analysis performed 1-4 hours after each infusion. Median total volume received: 750 mL (IQR 500-1000 mL) in first 48 hours, initiated 15 minutes (IQR 6-36 minutes) after enrollment.
Usual care - No sodium bicarbonate infusions or placebo. Standard ICU management according to local protocols. Capillary refill time measured only at baseline and 6 hours. Note: 47 patients (15%) in the control group crossed over and received bicarbonate, typically as rescue therapy.
KRT recommendations (standardized in both groups):
Immediate initiation recommended for: serum potassium >6.5 mEq/L with ECG changes, cardiogenic pulmonary edema with oliguria and P/F ratio <200 on high settings
KRT also recommended 24 hours after enrollment if meeting ≥2 of: urine output <0.3 mL/kg/h for 24 hours, pH ≤7.20 despite resuscitation, serum potassium >6.5 mEq/L
Outcomes
Comparisons are sodium bicarbonate vs. usual care.
Primary Outcome
90-day all-cause mortality
195/314 (62.1%) vs. 193/313 (61.7%)
Absolute difference: 0.4% (95% CI -7.2 to 8.0, P=0.91)
HR 0.97 (95% CI 0.80-1.19, P=0.78)
Secondary Outcomes
Outcome | Sodium Bicarbonate | Usual Care |
|---|---|---|
28-day mortality | 171/314 (54%) | 170/313 (54%) |
180-day mortality | 204/314 (65%) | 196/313 (63%) |
Use of KRT by day 28 | 109/314 (35%) | 157/313 (50%) |
Time to KRT initiation, median | 30.9 hours (IQR 12-70) | 15.5 hours (IQR 5-32) |
Vasopressor use by day 28 | 305/314 (97%) | 293/313 (94%) |
Invasive mechanical ventilation by day 28 | 287/314 (91%) | 275/313 (88%) |
ICU length of stay, median | 5 days (IQR 2-11) | 5 days (IQR 1-12) |
Hospital length of stay by day 90, median | 10 days (IQR 2-28) | 8 days (IQR 1-26) |
ICU-acquired bloodstream infections | 14/314 (4%) | 28/313 (9%) |
Major adverse kidney events at day 90 | 199/240 (83%) | 195/232 (84%) |
Absolute difference in KRT use: -15.5% (95% CI -23.1 to -7.8)
Main reasons for KRT initiation: Persistent oliguria, refractory acidemia, elevated creatinine/BUN (less common for acidemia indication in bicarbonate group)
Subgroup Analysis
No significant treatment effect modification observed across any prespecified subgroups:
pH (≤7.10 vs 7.11-7.20)
Age (<65 vs ≥65 years)
Sepsis (yes vs no)
SOFA score (<11 vs ≥11)
Type of admission (medical vs surgical)
Cardiac arrest
Hemorrhagic shock
Kidney SOFA score (≤2 vs ≥3)
Serum lactate level
Center (BICARICU-1 participants vs new centers)
Adverse Events
No evidence of group effect on adverse events. Sodium bicarbonate therapy appeared safe with no increase in complications.
Criticisms
Open-label design could have introduced detection bias, particularly for the subjective decision of when to initiate KRT. Although the trial recommended standardized criteria for dialysis initiation, clinicians were unblinded and may have had lower thresholds for starting KRT in controls with persistent acidemia.
Trial was dramatically underpowered for the primary outcome. The sample size calculation assumed 80% mortality in controls and a 10% absolute reduction with bicarbonate-an implausibly large effect. Observed mortality was 62%, and a realistic treatment effect (1-2% absolute reduction based on target trial emulation data) would have required 5,000-10,000 patients to detect.
Fifteen percent crossover from control to bicarbonate group suggests equipoise was fragile. These crossover patients had 72% mortality compared to 62% in both primary groups, indicating bicarbonate was used as rescue therapy in the sickest patients. This complicates interpretation—was bicarbonate ineffective, or was it just deployed too late in the crossover group?
The reduction in KRT use could be confounded by the intervention itself. Bicarbonate increases pH, which directly addresses one of the criteria for dialysis initiation (refractory acidemia). This doesn't necessarily mean bicarbonate improved kidney function—it may have just made the indication for dialysis less apparent. The fact that oliguria and acidemia were less common reasons for KRT in the bicarbonate group supports this interpretation.
Incomplete data on AKI diagnostic criteria. The trial didn't consistently report whether AKI was diagnosed by creatinine, urine output, or both. This matters because the mechanism and reversibility of AKI differ depending on etiology, and treatment responses may be heterogeneous.
Limited information on etiology of acidemia beyond gross categorization. Most patients had high anion gap lactic acidosis or mixed acidosis, but the underlying causes (sepsis, hemorrhagic shock, cardiogenic shock, etc.) have different prognoses and may respond differently to bicarbonate.
Patients in the bicarbonate group received more total fluid (median 2805 mL vs 2200 mL by day 2). It's unclear whether this additional fluid represented clinically meaningful volume differences that could have influenced decisions about KRT initiation for fluid overload.
Follow-up was limited to mortality and organ support metrics. The trial didn't assess functional outcomes, quality of life, or post-ICU syndrome-outcomes that matter to patients and families beyond just survival.
Dashevsky's Dissection
For patients: This trial tells us that sodium bicarbonate won't help you survive severe acidemia in the ICU. The mortality rate was the same whether you got bicarbonate or not—about 62% died within 90 days regardless. However, bicarbonate did reduce the chance you'd need dialysis by about 15%, and if you did need it, you'd likely get it later rather than sooner. That's not nothing. Dialysis comes with risks-infections, hypotension, vascular access complications—so avoiding it when possible has real value, even if it doesn't change your ultimate survival.
For pulmonary and critical care physicians: BICARICU-2 forces us to reckon with the limits of our reflexive interventions. We've been giving bicarbonate for decades based on physiologic rationale: acidemia impairs cardiac contractility, increases arrhythmia risk, and causes pulmonary vasoconstriction. But BICARICU-2 shows that buffering the pH doesn't translate into mortality benefit, even in the sickest patients with the worst acidemia.
The 15% reduction in dialysis use is the more interesting finding. Is that a real benefit, or is it just masking the indication for KRT? I lean toward the former. The time to KRT initiation was doubled in the bicarbonate group (31 hours vs 16 hours), and many patients avoided dialysis altogether. In an era where timing trials like STARRT-AKI and AKIKI have taught us that delayed KRT is often safe and sometimes preferable, using bicarbonate as a bridge makes sense, especially if it helps some patients recover renal function without ever needing dialysis.
The 50% reduction in ICU-acquired bloodstream infections is provocative but needs cautious interpretation. The investigators speculate it's related to fewer dialysis catheters, but only 26 of 266 patients (10%) who received KRT developed bloodstream infections compared to 16 of 361 (4%) who didn't-a plausible mechanism. Still, this was a secondary outcome in an unblinded trial. We shouldn't over-interpret it.
What bothers me most about this trial is the crossover rate. Fifteen percent of controls received bicarbonate anyway, typically as rescue therapy, and their mortality was 72%-higher than both randomized groups. This tells me clinicians didn't really believe in equipoise. When faced with a crashing patient and a pH of 7.05, they gave bicarbonate, protocol be damned. That's understandable, but it muddies the science.
The trial also highlights our ongoing struggle with KRT timing. The recommendations for dialysis initiation were standardized but not mandatory, leaving room for clinical judgment. That's appropriate—dialysis decisions should be individualized—but it also opens the door to bias. Did intensivists have a lower threshold for starting KRT in controls with refractory acidemia? Probably. Does that invalidate the finding? Not entirely, but it complicates things.
For the health system: From a resource standpoint, reducing KRT use by 15% is meaningful. Dialysis is expensive, labor-intensive, and requires specialized nursing. In a system where ICU capacity is often strained, interventions that reduce the need for high-intensity support have value even if they don't reduce mortality. Bicarbonate is cheap, about $10-20 per dose, and the infrastructure to administer it is already in place.
But we need to be honest about what we're buying. We're not saving lives. We're temporizing. We're buying time for kidneys to recover, delaying the need for dialysis, possibly reducing catheter-related infections. Those are real benefits, but they're incremental, not transformative.
The broader lesson from BICARICU-2 is about realistic expectations. We've chased mortality reduction in critical care for decades, and trials keep coming back neutral or with modest effects. Maybe that's because mortality in severe septic shock with multiorgan failure is driven more by the underlying disease than by our ICU interventions. Bicarbonate can buffer pH, but it can't reverse septic shock, fix failing kidneys, or undo the damage from prolonged hypoperfusion.
What BICARICU-2 does give us is permission to use bicarbonate pragmatically, as a temporizing measure, a bridge therapy, a tool to delay or avoid dialysis, without expecting it to be a mortality-reducing magic bullet. That's a more modest role than we hoped for, but it's a role nonetheless.
Jung B, Jabaudon M, De Jong A, et al. Sodium Bicarbonate for Severe Metabolic Acidemia and Acute Kidney Injury: The BICARICU-2 Randomized Clinical Trial. JAMA. 2025;334(22):2000-2010. doi:10.1001/jama.2025.20231
