Low Bicarbonate and Anion Gap: What It Means and When to Worry

Blood pH is kept within a very narrow range (about 7.35–7.45), because enzymes, heart muscle and the brain are sensitive to even small fluctuations. The main buffering system is the pair carbon dioxide and bicarbonate: the lungs regulate the amount of CO2 (exhaled), and the kidneys regulate the amount of bicarbonate (conserving it and generating new bicarbonate while excreting hydrogen ions). When excess acid enters the blood, bicarbonate is 'used up' neutralizing it and its concentration falls. That is why low bicarbonate is the basic biochemical sign of metabolic acidosis.

Two practical consequences follow for the reader. First, low bicarbonate is not a diagnosis but a signal, and it requires establishing whether the acidosis is acute or chronic and where the excess acid comes from. Second, what is done in acute acidosis (treating the cause, fluids, insulin in diabetic ketoacidosis) looks different from chronic acidosis (for example in kidney disease), where treatment with base is considered; recommendations must therefore not be transferred from one situation to the other.

Low bicarbonate can also result from a response to a respiratory disorder. With chronic hyperventilation (for example in lung disease, anxiety, high altitude, pregnancy) the lungs exhale excess CO2, pH rises, and the kidneys respond by lowering bicarbonate to restore balance. This is not metabolic acidosis, yet the result shows the same low bicarbonate. Telling them apart requires a blood gas test (pH, pCO2).

The anion gap is the difference between the main cation in the blood (sodium) and the sum of the two main measured anions (chloride and bicarbonate): anion gap = Na+ − (Cl− + HCO3−). It is not a real 'void' in the blood, because charges always balance; the gap represents the anions we do not measure routinely: mainly albumin, and also phosphates, sulfates and organic acids. When an excess of an organic acid (for example a ketoacid or lactate) appears in the blood, its anions take the place of bicarbonate, so bicarbonate falls and the gap rises.

Anion gap norms depend on the analyzer: older textbooks give about 12 ± 4 mmol/l, while in many modern laboratories the range is lower (often about 8–12 mmol/l, sometimes even lower), because chloride measurement methods have changed. The most important rule is therefore to compare the result with the range given by your own laboratory, not with one universal number. Some laboratories report an anion gap that includes potassium; the range is then several mmol/l higher.

The division into acidoses with a raised and a normal gap is the most useful diagnostic step. In acidoses with a raised gap, a new acid enters the blood whose anion is not measured. In acidoses with a normal (hyperchloremic) gap, the body loses bicarbonate (through the gut or kidneys) or receives an excess of chloride, and the lost bicarbonate is replaced by chloride, so the gap does not change.

The table shows that quite different situations can stand behind the same low bicarbonate: from life-threatening ketoacidosis to mild, slowly progressing acidosis in kidney disease. Bicarbonate alone is therefore not enough to make a diagnosis; the anion gap, the clinical context and, if acute acidosis is suspected, a blood gas with assessment of pH and pCO2 are needed. It is also worth remembering that hyperphosphatemia, which we describe in the article on phosphorus, can raise the anion gap.

The biggest pitfall is the effect of albumin. Albumin is the main unmeasured anion in serum, so with low albumin (malnutrition, liver disease, severe illness, nephrotic syndrome) the anion gap is underestimated even if acid anions are indeed circulating in the body. A 'normal' gap can then mask significant acidosis.

The second pitfall is paraproteins. Cationic paraproteins (for example in multiple myeloma) can lower the anion gap, as can lithium or bromide; conversely, very high albumin (dehydration) or a laboratory error raises it. The third pitfall concerns sampling: CO2 escapes from underfilled or long-open tubes and with delayed analysis, which falsely lowers bicarbonate. That is why an unexpected, isolated low bicarbonate without any symptoms and without changes in the other parameters is worth repeating first in a properly collected and promptly analyzed sample.

This is the question readers ask us most often, but the answer requires distinguishing two phenomena. Dietary ketosis, that is, raised ketone levels with carbohydrate restriction, is a physiological state in which insulin still works and the body can maintain pH. Ketoacidosis is a pathological accumulation of ketoacids beyond buffering capacity, with lowered pH and bicarbonate and a raised anion gap. In healthy people on a typical ketogenic diet, ketoacidosis is rare, but it has been described in particular situations when several factors overlap.

A single case report cannot estimate how often this happens, but it illustrates the mechanism: a high demand for glucose (lactation), very little carbohydrate in the diet and ketogenesis lead to an accumulation of ketones beyond buffering capacity. Similar risk situations are the combination of a low-carbohydrate diet with SGLT2-inhibitor drugs (which increase urinary glucose loss) and with alcohol, starvation, severe illness or pregnancy. This phenomenon is called euglycemic ketoacidosis.

Metformin is sometimes associated with lactic acidosis, but the risk should be assessed realistically. A 2016 review in Metabolism stresses that metformin raises lactate in a concentration-dependent manner by inhibiting mitochondrial respiration mainly in the liver, and that the development of metformin-associated lactic acidosis usually needs both high drug concentrations (as in kidney failure) and an additional event disrupting lactate production or clearance (for example cirrhosis, sepsis, hypoperfusion). The authors note that the reported incidence is very low (fewer than 10 cases per 100,000 patient-years), but the mortality of MALA approaches 50%, which is why moderate and severe renal impairment is a contraindication.

Other common causes of low bicarbonate include chronic kidney disease (reduced ability of the kidneys to excrete acids and generate bicarbonate), diarrhea (loss of bicarbonate in intestinal fluid), renal tubular acidosis, some drugs (acetazolamide, topiramate, misused salicylates), alcohol misuse and starvation. In advanced kidney disease low bicarbonate is one element of the picture of uremia and mineral disturbances; it is interpreted together with creatinine, eGFR, potassium and phosphate.

For a patient taking metformin the practical conclusions are simple: monitor creatinine and eGFR as advised, temporarily stop the drug (after consulting the doctor) in acute illness with dehydration, vomiting, diarrhea or sepsis, or before contrast studies, and do not stop it on your own out of fear of acidosis. The benefits of metformin in diabetes are well documented, as we also describe in the article on metformin and vitamin B12.

Bicarbonate slightly below normal that persists for months is usually a sign of chronic kidney disease, and less often of renal tubular acidosis or habitual hyperventilation. It is notable that even in healthy older people low bicarbonate can have prognostic significance. In the prospective Health ABC study of 2,287 well-functioning people aged 70–79, bicarbonate below 23 mEq/l was associated with higher mortality compared with the reference group of 23–27.9 mEq/l (HR 1.24; 95% CI 1.02–1.49) after accounting for eGFR, albuminuria, pH and other factors. The association applied both to metabolic acidosis and to respiratory alkalosis as the cause of low bicarbonate; metabolic alkalosis was also associated with higher mortality.

The conclusion needs caution in both directions. The Kraut and Madias review indicates that giving base in chronic acidosis is associated with improved cellular function and few complications, and low bicarbonate is linked to muscle breakdown and bone loss. The BiCARB trial, however, showed no benefit from treatment in a specific group of older people with very advanced kidney disease and mild acidosis. The evidence is therefore ambiguous, and the decision about treatment with base belongs to the nephrologist and depends on the degree of acidosis, the stage of kidney disease, blood pressure and the risk of sodium overload. Taking baking soda on your own to 'de-acidify' the body is not recommended.

Mildly lowered bicarbonate usually causes no symptoms. In acute acidosis the body tries to compensate by increasing breathing: characteristic is deep, rapid breathing (Kussmaul breathing), often with a feeling of breathlessness. It is accompanied by nausea, vomiting, abdominal pain, weakness, drowsiness, confusion and, in severe cases, hypotension and heart rhythm disturbances. Chronic acidosis is quiet: it can cause fatigue, muscle weakness and, in the longer term, bone disturbances and loss of muscle mass.

Interpretation always begins with the question of whether the patient is stable. In a person with symptoms or suspected acute acidosis the order of action is reversed: blood gas and treatment first, detailed differentiation afterwards. In a person without symptoms and with isolated low bicarbonate, it makes sense to repeat calmly and analyze the gap and kidney function within a few days.

Low bicarbonate and the anion gap are a pair of parameters that carry more information in a single line of a result than you might think. For a person without symptoms a sensible plan is: repeat the test in a well-collected sample, calculate the gap (and correct it for albumin), check the kidneys and glucose, and review drugs and diet. For a person with symptoms or suspected acute acidosis the plan is simpler and more urgent: medical help without delay.

If you follow a ketogenic or very low-carbohydrate diet, especially together with SGLT2-class drugs, in pregnancy or while breastfeeding, discuss it with your doctor and know the warning symptoms. If you have kidney disease, check bicarbonate regularly and do not try to treat it on your own: a mild lowering of bicarbonate requires a nephrologist's decision, and the evidence for benefit from treatment with base is not unambiguous.