PACULit
← September 2026 edition

A19 · PMID:42171427

Sodium Bicarbonate for Acute Metabolic Acidosis in Critically Ill Adults: A Meta-Analysis of Randomized Clinical Trials

Open the original article record

Bottom line

In a meta-analysis of 4 RCTs (1,111 critically ill adults with acute metabolic acidosis), sodium bicarbonate significantly reduced renal replacement therapy (RRT) need, while a favorable mortality trend remained statistically unconfirmed.

PICO at a glance

Clinical question and design

Acute metabolic acidosis is common in critically ill patients and has historically been managed with volume replacement, sodium bicarbonate, and renal replacement therapy. Bicarbonate's role has long been controversial: older, smaller, heterogeneous trials failed to show consistent benefit, and theoretical concerns about paradoxical intracellular acidosis, hypernatremia, and volume overload have made many clinicians reluctant to use it routinely. The BICAR-ICU and BICAR-ICU-2 trials provided newer randomized evidence on bicarbonate therapy in severe acidemia, prompting this updated meta-analysis.

The authors systematically searched PubMed, Embase, Medline, and the Cochrane Library, restricting inclusion to RCTs. Analyses used Hartung-Knapp random-effects meta-analysis with trial sequential analysis (TSA) for the primary outcome, plus Bayesian random-effects meta-analyses with noninformative and weakly informative priors as sensitivity analyses. Four trials enrolling 1,111 patients were included. The four trials were Chen 2013 (China, n=65, sepsis-lactate acidosis), BICAR-ICU 2018 (France, n=389), BICARICU-2 2025 (France, n=627) and the SODaBIC pilot 2023 (Australia, New Zealand and Japan, n=30). Enrolling pH thresholds ranged from 7.15 to 7.30 and the treatment target was pH 7.25 in one trial and 7.30 in three. Mean age was 65.8 years and 40.1% were women. Three trials were at low risk of bias; Chen 2013 was at high risk for randomization and selective reporting. Chen 2013 titrated to target pH rather than dosing by fixed volume, and the SODa-BIC pilot used 8.4% sodium bicarbonate diluted in D5W to a final concentration of 600 mEq/L.

Intervention and outcomes measured

The intervention was sodium bicarbonate versus placebo or control care; regimens varied across trials, with the two BICAR-ICU trials using the regimen described above. BICARICU-2 also required moderate-to-severe AKI, KDIGO stage 2 or 3, at entry, so most of the pooled population already had kidney injury. No AKI subgroup or effect-modification test was run, so this describes who was enrolled, not where benefit is largest. The primary outcome was RRT use.

Results

Sodium bicarbonate significantly reduced RRT use (RR, 0.69; 95% CI, 0.61-0.78), and TSA supported firm, adequately powered evidence of benefit for this outcome. Bayesian analysis estimated a 94.6% posterior probability of reduced RRT use.

Secondary findings

Across the 4 trials, sodium bicarbonate showed a nonsignificant trend toward lower mortality (RR, 0.84; 95% CI, 0.55-1.30); TSA indicated the cumulative sample size remained far below the required information size, so the mortality effect is inconclusive rather than negative. Bayesian analysis estimated a 90.4% posterior probability of any mortality reduction. Secondary outcomes remained imprecise: ICU length-of-stay estimates did not demonstrate a clear difference; ventilator-free days, -0.05 day (95% CI, -11.45 to 11.34); and vasopressor-free days, 1.7 days (95% CI, -4.5 to 7.91), the only secondary outcome with a notable posterior probability of clinically meaningful benefit at 72.9% against a 1-day threshold. Heterogeneity separates the two headline findings: I-squared was 0% for RRT and 55.7% for mortality.

Uncertainty and limitations

Only 4 trials and 1,111 patients were pooled, a relatively small evidence base. Removing Chen 2013 collapses ICU length-of-stay heterogeneity from 92.5% to 0.0%. BICARICU-2 alone supplies 627 of the 1,111 patients, and with BICAR-ICU 2018 (n=389) the two French trials supply the large majority, so the pooled estimate is close to being those two trials. Excluding the high-risk-of-bias Chen 2013 trial, the posterior probability of any mortality benefit falls from 90.4% to 80.0% under noninformative priors. The two trials carrying the pooled result were both open-label, and Chen 2013 was rated high risk of bias: no information on the randomization sequence or a pre-specified analytic plan.

Literature review and evidence synthesis

This trial-level meta-analysis sits alongside a 2026 individual patient data meta-analysis in Critical Care that pools the BICAR-ICU and BICAR-ICU2 trials directly, the same anchor trials that prompted this updated review, offering a patient-level complement to this study-level synthesis. A 2026 Cochrane review on buffered solutions versus 0.9% saline for resuscitation in critically ill adults and children addresses an adjacent but distinct question, chronic buffered-fluid exposure during resuscitation rather than bolus or infusion bicarbonate therapy for established acidosis, and should not be conflated with this trial's intervention. Because RRT is this meta-analysis's primary outcome, a 2021 Critical Care systematic review and meta-analysis with trial sequential analysis on the timing of RRT initiation for acute kidney injury provides useful adjacent methodologic context, even though it addresses a different clinical question.

How this fits with the broader evidence

The RRT reduction is statistically significant by conventional frequentist criteria, has TSA-confirmed adequate information size, and carries a high Bayesian posterior probability of benefit, a comparatively strong, converging signal across three analytic lenses. Mortality is directionally favorable but the frequentist confidence interval crosses 1, and TSA explicitly flags the current trial evidence as underpowered for a firm mortality conclusion. This is a useful model for interpreting any meta-analysis where a primary and secondary outcome diverge in statistical maturity: reasonably confident bicarbonate reduces RRT need, encouraging but not yet established that it reduces mortality.

ACPE UAN: 0683-0000-26-036-H01-P

Study source: PMID:42171427