Metabolic Acidosis: Lactate, Unmeasured Anions, and Bicarbonate

Metabolic acidosis describes a process that lowers pH through a nonrespiratory disturbance. The resulting arterial pH depends on respiratory compensation and every accompanying metabolic process. A lactate concentration, an anion gap, a bicarbonate concentration, and a pH therefore describe related but different features of an illness. Understanding their relationships allows the clinician to ask what is accumulating, what is being lost, how the lungs and kidneys are responding, and what a proposed intervention can actually change.

54.1 Separate the acidifying process from the measured acidemia

Acidemia means that blood pH is below its reference range. Metabolic acidosis means that a metabolic process is pushing pH downward. These are not interchangeable: respiratory alkalosis or a concurrent metabolic alkalosis can conceal the acidifying process. A patient with substantial lactate accumulation may consequently have a near-normal arterial pH without having a normal metabolic state.

A low plasma bicarbonate concentration is an important clue, but it also occurs during renal adaptation to sustained respiratory alkalosis. Interpretation therefore begins with pH, arterial carbon dioxide tension, bicarbonate or total carbon dioxide, and their timing. Chapters 52 and 53 describe buffering and the shared constraints of the principal frameworks; the present chapter applies those relationships to the causes and consequences of metabolic acidosis.

The principal mechanisms are net addition or retention of nonvolatile acid, loss of bicarbonate or its metabolic precursors, and changes in electrolyte composition that reduce the strong ion difference. These descriptions overlap rather than identify unrelated chemistries. When a retained acid is buffered, bicarbonate can fall while its accompanying anion remains in extracellular fluid. When bicarbonate is lost with sodium or potassium and chloride is retained or administered, a low bicarbonate concentration may instead accompany relative hyperchloremia.

Production, distribution, metabolism, and excretion must be followed separately. Buffering limits the change in free hydrogen ion activity; it does not automatically eliminate the underlying retained anion. Exhaling additional carbon dioxide changes an open buffer system, but it does not excrete sulfate or replace gastrointestinal bicarbonate losses. Renal acid excretion, metabolism of appropriate organic anions, removal of an exogenous source, or extracorporeal transport may be needed to change those balances.

The immediate pH and the ongoing acid burden can thus move differently. This distinction explains both why severe acidemia can demand attention and why correcting the arterial number may leave its cause active.

54.2 The anion gap is a charge-accounting clue

Plasma remains electrically neutral. The clinical anion gap exists because the routine calculation includes selected ions while omitting others. With potassium omitted, it is sodium minus the sum of chloride and bicarbonate. For these monovalent ions, a concentration in mmol/L has the same numerical value as a charge concentration in mEq/L.

Albumin supplies a substantial fraction of the normally omitted negative charge. Low albumin can therefore lower the observed gap and conceal an increase in other anions. Figge and colleagues derived an approximate adjustment in critically ill patients: add 2.5 mEq/L for each 1 g/dL by which albumin is below the selected reference concentration. 1

Consider constructed concentrations: sodium 140 mmol/L, chloride 104 mmol/L, bicarbonate 18 mmol/L, and albumin 2.0 g/dL. The observed gap is 140 − (104 + 18) = 18 mEq/L. Using an illustrative albumin reference of 4.0 g/dL, the adjustment is 2.5 × (4.0 − 2.0) = 5 mEq/L, giving an adjusted gap of 23 mEq/L. This brings the estimate to a comparable albumin scale; it does not directly measure an additional 5 mEq/L of a particular toxin or acid.

The laboratory method, inclusion of potassium, local reference interval, and the patient’s baseline all matter. The adjustment is approximate, because protein charge also depends on pH and other unmeasured ions vary. It does not replace measuring lactate, ketones, or a suspected toxic substance.

An increased gap is not synonymous with lactic acidosis, and a normal gap does not exclude lactate elevation or retained acids. Several analytes, often obtained through different instruments, enter the calculation; analytical error and samples drawn at different times can create an apparent physiological discrepancy. A useful gap prompts a better explanation rather than providing one by itself.

54.3 Lactate production does not require absent oxygen

Glycolysis produces pyruvate and reduces NAD⁺ to NADH. Continued glycolytic flux requires regeneration of NAD⁺. Lactate dehydrogenase provides one route by transferring reducing equivalents from NADH to pyruvate, forming lactate. In that reaction’s lactate-forming direction, a hydrogen ion is consumed rather than released. The net acid–base consequences of accelerated metabolism depend on the coupled reactions, ATP turnover, buffering, transport, and subsequent substrate use.

This is why the phrase “lactic acid production” can obscure more than it explains. Lactate is predominantly dissociated at physiological pH. Its accumulation often accompanies an acidifying disturbance, but the circulating lactate concentration is not a direct count of the hydrogen ions generated by the entire metabolic process. Equally, emphasizing that LDH consumes a proton does not make severe lactate-associated acidosis harmless or deny a hypoxic mechanism when one is present.

Oxygen limitation can impede oxidative metabolism and favor lactate accumulation. However, rapid glycolysis can also supply pyruvate and reducing equivalents faster than their alternative pathways handle them while oxygen consumption continues. Catecholamine stimulation, seizures, intense muscle activity, and inflammatory metabolic changes can all contribute. Experimental endotoxemia in human volunteers increased leg lactate release under aerobic conditions, providing direct evidence that oxygen deficiency is not required for every rise. This controlled volunteer model does not reproduce the full physiology of septic shock. 2

The relevant clinical question is therefore why lactate is increasing in this patient. Low oxygen delivery, regional ischemia, adrenergic acceleration, mitochondrial or enzyme dysfunction, impaired disposal, and combinations of these mechanisms can produce overlapping blood concentrations. Normal global oxygen delivery does not exclude regional hypoxia; elevated lactate does not establish it.

Lactate remains a useful marker of serious physiological disturbance. Its value increases when interpreted with perfusion, organ function, medications, substrate availability, and the trajectory of illness rather than assigned one automatic mechanism.

54.4 Disposal and anion fate determine what a lactate trend means

The circulating lactate pool reflects entry, removal, and distribution. Liver and kidney participate in disposal through oxidation and glucose synthesis, while other tissues can take up and oxidize lactate as fuel. A concentration can rise because entry increases, disposal decreases, distribution changes, or several of these occur together.

In a physiological study of stable septic patients, Levraut and colleagues used lactate-loading kinetics to distinguish production from utilization. Mild hyperlactatemia was associated with lower modeled clearance rather than greater production in the compared groups. This selected population and kinetic method do not establish the cause of severe hyperlactatemia during evolving shock, but they show why a concentration alone cannot identify the flux responsible. 3

A serial percentage decline also differs from measured metabolic clearance. If lactate falls from 8 to 6 mmol/L, its concentration has decreased by (8 − 6)/8 × 100 = 25%. That calculation does not measure the volume of plasma cleared of lactate per unit time. Reduced production, increased disposal, dilution, and redistribution can all contribute. Continued production may remain substantial even while the measured pool shrinks.

The acid–base consequences depend on what happens to the accompanying charge. Metabolism of a retained organic anion can restore bicarbonate equivalents in the overall balance, whereas losing that anion in urine with sodium or potassium removes potential base from the body. Lactate delivered as a sodium salt is therefore not equivalent to an endogenous acidifying process accompanied by lactate accumulation. The amount of counterion administered and the capacity to metabolize the anion matter.

Falling lactate can consequently coexist with persistent metabolic acidosis from chloride loading, renal acid retention, gastrointestinal losses, or previously lost organic anions. Conversely, pH can improve because ventilation changes while lactate remains elevated. Following lactate together with chloride, bicarbonate, albumin, and organ function identifies the changing mixture more reliably than demanding that all variables normalize together.

54.5 Ketoacids link substrate metabolism to potential base loss

Insulin deficiency or a low insulin-to-counterregulatory-hormone signal favors lipolysis and hepatic ketogenesis. Acetoacetate and β-hydroxybutyrate can accumulate when production exceeds utilization and excretion. The β-hydroxybutyrate-to-acetoacetate relationship also changes with hepatic mitochondrial redox conditions, so “total ketones” cannot be inferred reliably from one component without assumptions.

Ketoacidosis is therefore a disorder of substrate flux as well as acid–base balance. The glucose concentration does not quantify ketogenesis: diabetic, fasting, alcohol-associated, and medication-associated presentations can have different degrees of hyperglycemia. A relatively modest glucose value does not establish the absence of a clinically important ketone burden. The mechanism and exposure history guide interpretation; the acid–base pattern alone cannot identify the initiating cause.

Measurement matters. Nitroprusside-based qualitative ketone tests chiefly detect acetoacetate and do not quantify β-hydroxybutyrate. During recovery, β-hydroxybutyrate may be converted to acetoacetate, making a qualitative test appear persistently positive or stronger despite improvement in the dominant circulating ketoanion. Specific assays also require attention to their analytical performance; a primary laboratory comparison demonstrated assay-dependent interference. 4

The subsequent acid–base course depends on anion conservation. If accumulated ketoanions remain available for metabolism, their utilization can restore bicarbonate equivalents. If ketoanions have already been excreted with sodium or potassium, those potential bicarbonate equivalents have been lost. Chloride-containing fluid can further shift the pattern toward hyperchloremic acidosis as ketogenesis subsides.

Serial human measurements during recovery from diabetic ketoacidosis demonstrated that organic anions and the gap could decline more than bicarbonate rose. Anion loss before treatment and differences in distribution contributed to the persistence of acidosis. 5 Thus, a persistent low bicarbonate concentration need not mean ongoing ketone production, while a falling gap does not by itself prove complete metabolic recovery.

54.6 Renal acid retention and bicarbonate loss produce different mixtures

The kidneys reclaim filtered bicarbonate and add new bicarbonate to the body when acid is excreted as ammonium and titratable acid. Reduced filtration, impaired ammonium generation or transport, limited distal hydrogen-ion secretion, and altered delivery can reduce that capacity. These functions can fail to different degrees, so serum creatinine is not a direct measurement of net acid excretion.

Retention of phosphate, sulfate, and organic anions can contribute to an increased gap in advanced kidney dysfunction. Urea itself is not the acid anion responsible for “uremic acidosis.” Tubular dysfunction or impaired ammonium excretion can instead produce predominantly hyperchloremic acidosis, including before filtration falls enough to produce a large accumulation of gap anions. Hyperkalemia can impair ammonium handling and therefore participate in the mechanism rather than merely accompany it.

In a study of critically ill patients with acute renal failure, quantitative analysis identified acidifying contributions from unmeasured anions and phosphate together with an opposing effect from hypoalbuminemia. Half had an observed gap within its conventional range. This retrospective, model-based analysis demonstrates simultaneous processes rather than identifying the chemical composition of every omitted anion. 6

Loss of bicarbonate-rich gastrointestinal fluid creates a different starting problem. Bicarbonate leaves the body, and electrolyte and renal responses determine the resulting plasma composition. Renal bicarbonate wasting likewise differs from failure to excrete a new daily acid load. A low urine pH cannot alone distinguish them: urine can be acidic while containing too little ammonium and carrying an inadequate total quantity of acid out of the body.

Urinary indices are therefore interpreted through flow, ammonium, accompanying anions, kidney function, and ongoing treatment, rather than as isolated labels. Indirect estimates of urine ammonium become less reliable when unusual organic anions are being excreted. The mechanism determines whether a changing bicarbonate reflects continued loss, reduced acid excretion, retained anions, or a combination.

54.7 Unmeasured anions include endogenous metabolites and exposures

“Unmeasured” describes what the selected assay panel omitted, not a single substance. Once lactate or a ketoanion is measured, it can explain part of the omitted charge, but the residual estimate may still include phosphate, other organic metabolites, drug-related products, and measurement uncertainty. The strong ion gap developed in Chapter 53 and the albumin-adjusted anion gap are related estimates with different conventions, not interchangeable chemical assays.

Direct plasma assays in small groups with metabolic acidosis found elevations of several intermediary metabolites, including isocitrate, α-ketoglutarate, and malate. These observations support a heterogeneous endogenous contribution; they do not establish that one metabolite explains the entire clinical gap or identify a universal metabolic block. 7 A large residual gap may therefore warrant a broader search even when lactate is elevated.

Exogenous compounds introduce both the parent molecule and its metabolites. In methanol exposure, the parent alcohol contributes to an osmolal gap, whereas accumulated formate contributes to the anion gap and toxicity. The osmolal gap compares measured osmolality with an estimate from routinely measured solutes; the calculation and unit conventions must be checked. As metabolism proceeds, the parent-molecule signal may diminish while acidic metabolites accumulate.

In a 28-patient methanol outbreak study, three patients had an osmolal gap within the reference area despite formate-associated gap elevation, illustrating why a reassuring value cannot exclude the later metabolic phase. 8 The exposure history, timing, direct assays, and clinical findings remain necessary.

The source can also be treatment-related. Propylene glycol in some intravenous formulations can accumulate and contribute to an abnormal osmotic or metabolic picture, particularly when exposure and disposal are mismatched. Human ICU observations linked exposure to measured propylene glycol and osmotic abnormalities, with imperfect relationships to lactate. 9 Reviewing the actual formulation can therefore be as important as reviewing the active drug. An unexplained gap is a prompt for causal investigation, not a reason to assume that all residual charge is harmless “critical illness.”

54.8 Chloride and concurrent processes reshape the pattern over time

An acid–base sample records the net result of several balances. The same bicarbonate concentration can occur with predominantly lactate accumulation, chloride-associated acidosis, retained renal anions, or mixtures partly opposed by low albumin or metabolic alkalosis. A familiar numerical pattern should therefore be treated as a hypothesis about mechanisms rather than a unique diagnosis.

Chloride-rich fluid illustrates a treatment-related change. Administering proportionally more chloride than is present relative to sodium in plasma reduces the strong ion difference. Bicarbonate and hydrogen-ion activity then change within the shared physicochemical constraints. Calling this solely “dilution of bicarbonate” misses the importance of the infused electrolyte composition. In a randomized physiological study during gynecologic surgery, saline produced hyperchloremic acidosis whereas lactated Ringer’s did not under the tested conditions. This establishes a compositional mechanism, not a universal outcome advantage for every patient or infusion. 10

Consider another constructed transition with sodium held at 140 mmol/L. Initially chloride is 102 mmol/L and bicarbonate 14 mmol/L, giving a gap of 24 mEq/L. Later chloride is 110 mmol/L and bicarbonate 16 mmol/L, giving a gap of 14 mEq/L. The gap has fallen by 10 mEq/L while bicarbonate has risen by only 2 mmol/L. That is compatible with a shrinking gap-anion component and a growing chloride-associated component; it is not proof of either process without the accompanying measurements.

Comparing a gap increase with a bicarbonate decrease can help detect such mixtures, but exact one-to-one matching is not guaranteed. Baseline values, albumin, distribution, renal anion loss, fluid balance, and time alter the relationship. The detailed mixed-disorder approach belongs in Chapter 57. Here the essential step is to follow measured components over time and avoid declaring persistent disease or treatment failure from bicarbonate alone.

54.9 Respiratory compensation changes the immediate physiological burden

Metabolic acidosis stimulates ventilation, reducing arterial carbon dioxide tension and limiting the pH fall. The compensation depends on respiratory drive, muscle capacity, dead space, lung mechanics, and the ventilatory support available. It is a physiological workload, not an automatic correction of the nonvolatile disturbance. Respiratory fatigue, sedation, or an interruption in ventilation can remove that compensation quickly.

For illustration, use a bicarbonate concentration of 12 mmol/L and the conventional relationship at approximately 37°C. Carbon dioxide solubility is taken as 0.03 mmol/L/mmHg and the apparent pK′ as 6.1:

At a PaCO₂ of 24 mmHg, pH is approximately 6.1 + log₁₀(12/0.72) = 7.32. If PaCO₂ rises to 40 mmHg while bicarbonate is held at 12 mmol/L for this comparison, calculated pH becomes 6.1 + log₁₀(12/1.20) = 7.10. The approximately 0.22-unit difference illustrates the importance of ventilation; it is not a prescribed PaCO₂ target or a prediction that bicarbonate would remain unchanged in vivo.

Severe acidemia can impair myocardial performance, modify calcium handling and vascular responsiveness, and increase pulmonary vascular load. The magnitude and direction of individual responses depend on the acidifying agent, carbon dioxide, extracellular and intracellular conditions, and underlying disease. A systemic arterial pH does not directly measure myocardial intracellular pH or establish the cause of circulatory failure.

These interactions create competing constraints. Increasing ventilation may be limited by lung protection and mechanical load. A treatment that generates additional carbon dioxide may require more elimination at the same time that respiratory reserve is failing. Evaluating pH together with PaCO₂, respiratory effort, ventilation, circulation, and the trend in the metabolic disturbance gives a more useful assessment than categorizing severity by pH alone.

54.10 Bicarbonate changes buffering, sodium balance, and carbon dioxide handling

Administered sodium bicarbonate supplies both sodium and bicarbonate. Bicarbonate can accept hydrogen ions, forming carbonic acid and then carbon dioxide and water. Its immediate effect therefore depends on buffering, distribution, ongoing acid generation, and carbon dioxide elimination. It does not directly stop ketogenesis, restore hepatic lactate disposal, remove formate, or repair impaired renal acid excretion.

If ventilation and perfusion permit carbon dioxide transport and excretion, extracellular pH can rise. If elimination is constrained, PaCO₂ may rise and offset some of the expected benefit. Carbon dioxide crosses cell membranes readily, so extracellular and intracellular responses need not match. The possibility of intracellular acidification is mechanistically relevant but should not be presented as an inevitable outcome in every tissue or clinical setting.

Sodium and water delivery can add extracellular volume and osmotic burden. A rise in pH can lower ionized calcium through increased protein binding and can alter potassium distribution. These changes may interact with any benefit from reduced acidemia. The response is consequently not captured by assuming a fixed bicarbonate distribution volume and converting a calculated deficit directly into a universal dose.

In a randomized crossover study of 14 patients with lactate-associated acidosis, Cooper and colleagues observed higher arterial pH but also increased PaCO₂ and lower ionized calcium after bicarbonate. Hemodynamic responses were not better than with the sodium-chloride comparator. 11 A separate ten-patient crossover study found no improvement in hemodynamics or measured indices of tissue oxygenation, while also not demonstrating worsening oxygenation. These small short-term experiments do not determine outcomes in every form of severe acidemia. 12

Replacing actual bicarbonate losses and buffering retained-acid states have different causal rationales. In both situations, a biochemical response is one observable effect. Whether it improves circulation, avoids an invasive support procedure, or changes survival requires clinical evidence beyond the buffer reaction itself.

54.11 Trials distinguish pH correction, kidney support, and survival

BICAR-ICU randomized 389 patients with severe metabolic acidemia. Its primary composite of death by day 28 or organ failure at day 7 was not significantly reduced overall. A prespecified moderate-to-severe AKI stratum showed a favorable survival signal, which generated a clinically important hypothesis rather than a universal indication. Metabolic alkalosis, hypernatremia, and hypocalcemia were more frequent with bicarbonate. 13

BICARICU-2 then specifically enrolled patients with severe metabolic acidemia and moderate-to-severe AKI: 640 were randomized and 627 analyzed. Day-90 mortality was 62.1% with bicarbonate and 61.7% in controls, an absolute difference of 0.4 percentage points (95% CI −7.2 to 8.0). Kidney replacement therapy was used less often, 35% versus 50%. This open-label trial therefore did not confirm a mortality benefit. Less initiation of kidney support is a distinct finding, influenced by treatment of acidemia that itself contributes to initiation decisions; it does not establish repaired kidney tissue or improved long-term renal recovery. 14

In 2026, SODa-BIC randomized 500 vasopressor-treated patients with metabolic acidosis; 498 were analyzed. Its double-blind comparison found major adverse kidney events by day 30 in 40.2% versus 39.4%, with an adjusted difference of 1.2 percentage points (95% CI −7.1 to 9.4). The composite included death, kidney replacement therapy, or persistent renal dysfunction. Its broader acidosis population and different treatment strategy prevent treating it as a duplicate of BICARICU-2. It did not establish benefit for the primary outcome. 15

Guideline statements must likewise be read within their scope. The 2026 Surviving Sepsis Campaign conditionally advises against bicarbonate to improve hemodynamics or reduce vasopressor requirements in septic shock with hypoperfusion-induced lactic acidemia, with low certainty. It conditionally favors bicarbonate in septic shock with pH ≤ 7.2 and an AKIN score of 2 or 3, with very low certainty. These are contextual recommendations, not proof of a mortality benefit or universal pH-normalization protocol. 16

54.12 Integrate the measurements into a causal explanation

Consider a constructed patient with septic illness, reduced kidney function, recent chloride-rich fluid exposure, and ongoing adrenergic stimulation. Sodium is 139 mmol/L, chloride 108 mmol/L, bicarbonate 14 mmol/L, albumin 2.0 g/dL, and lactate 6 mmol/L. PaCO₂ is 28 mmHg. These values describe the starting sample; none is a treatment target.

The observed gap is 139 − (108 + 14) = 17 mEq/L. Using the same illustrative albumin reference of 4.0 g/dL, the adjusted gap is 17 + [2.5 × (4.0 − 2.0)] = 22 mEq/L. Lactate accounts for some omitted charge, but subtracting it does not precisely identify the remaining acids because baseline lactate, albumin charge, other ions, and analytical uncertainty remain relevant.

The calculated pH is approximately 6.1 + log₁₀[14/(0.03 × 28)] = 7.32. This is compatible with a metabolic acidifying disturbance whose arterial expression is moderated by low PaCO₂. It does not mean that the underlying metabolic balances are normal. Chloride-associated effects, renal acid retention, lactate production, and impaired lactate disposal may coexist; their relative importance requires the clinical course and additional measurements.

Suppose lactate subsequently falls while chloride rises and bicarbonate changes little. Possible explanations include reduced glycolytic production or improved disposal accompanied by a growing chloride-associated component. If PaCO₂ then rises, pH may worsen despite improving lactate. If bicarbonate is administered and pH rises, that observation establishes a buffer-system response; it does not by itself demonstrate improved perfusion, renal repair, or removal of all accumulated anions.

A coherent interpretation follows the processes that can change: source production, substrate and redox conditions, liver and kidney handling, gastrointestinal or urinary losses, electrolyte delivery, ventilation, and extracorporeal transport. Serial measurements test whether those mechanisms are behaving as expected. The resulting explanation guides attention to the cause and the patient’s physiological reserve, while the clinical evidence defines what an added therapy can reasonably be expected to achieve.

References

1. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Critical Care Medicine. 1998;26:1807–1810. doi:10.1097/00003246-199811000-00019

2. Bundgaard H, Kjeldsen K, Suarez Krabbe K, et al. Endotoxemia stimulates skeletal muscle Na⁺-K⁺-ATPase and raises blood lactate under aerobic conditions in humans. American Journal of Physiology–Heart and Circulatory Physiology. 2003;284:H1028–H1034. doi:10.1152/ajpheart.00639.2002

3. Levraut J, Ciebiera JP, Chave S, et al. Mild hyperlactatemia in stable septic patients is due to impaired lactate clearance rather than overproduction. American Journal of Respiratory and Critical Care Medicine. 1998;157:1021–1026. doi:10.1164/ajrccm.157.4.9705037

4. Porter WH, Yao HH, Karounos DG. Laboratory and clinical evaluation of assays for β-hydroxybutyrate. American Journal of Clinical Pathology. 1997;107:353–358. doi:10.1093/ajcp/107.3.353

5. Oh MS, Banerji MA, Carroll HJ. The mechanism of hyperchloremic acidosis during the recovery phase of diabetic ketoacidosis. Diabetes. 1981;30:310–313. doi:10.2337/diab.30.4.310

6. Rocktaeschel J, Morimatsu H, Uchino S, et al. Acid–base status of critically ill patients with acute renal failure: analysis based on Stewart–Figge methodology. Critical Care. 2003;7:R60–R66. doi:10.1186/cc2333

7. Forni LG, McKinnon W, Lord GA, et al. Circulating anions usually associated with the Krebs cycle in patients with metabolic acidosis. Critical Care. 2005;9:R591–R595. doi:10.1186/cc3806

8. Hovda KE, Hunderi OH, Rudberg N, Froyshov S, Jacobsen D. Anion and osmolal gaps in the diagnosis of methanol poisoning: clinical study in 28 patients. Intensive Care Medicine. 2004;30:1842–1846. doi:10.1007/s00134-004-2373-7

9. Barnes BJ, Gerst C, Smith JR, Terrell AR, Mullins ME. Osmol gap as a surrogate marker for serum propylene glycol concentrations in patients receiving lorazepam for sedation. Pharmacotherapy. 2006;26:23–33. doi:10.1592/phco.2006.26.1.23

10. Scheingraber S, Rehm M, Sehmisch C, Finsterer U. Rapid saline infusion produces hyperchloremic acidosis in patients undergoing gynecologic surgery. Anesthesiology. 1999;90:1265–1270. doi:10.1097/00000542-199905000-00007

11. Cooper DJ, Walley KR, Wiggs BR, Russell JA. Bicarbonate does not improve hemodynamics in critically ill patients who have lactic acidosis: a prospective, controlled clinical study. Annals of Internal Medicine. 1990;112:492–498. doi:10.7326/0003-4819-112-7-492

12. Mathieu D, Neviere R, Billard V, Fleyfel M, Wattel F. Effects of bicarbonate therapy on hemodynamics and tissue oxygenation in patients with lactic acidosis: a prospective, controlled clinical study. Critical Care Medicine. 1991;19:1352–1356. doi:10.1097/00003246-199111000-00008

13. Jaber S, Paugam C, Futier E, 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. doi:10.1016/S0140-6736(18)31080-8

14. Jung B, Jabaudon M, De Jong A, 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. doi:10.1001/jama.2025.20231

15. 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. New England Journal of Medicine. 2026;395:1062–1074. doi:10.1056/NEJMoa2600526

16. Prescott HC, Antonelli M, Alhazzani W, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock 2026. Critical Care Medicine. 2026;54:725–812. doi:10.1097/CCM.0000000000007075. Official SCCM recommendations