• Anion gap (AG) = [Na] – [Cl] – [HCO3].
• Normal AG = 8-12 mEq/L with a serum albumin of 4g/dl.
  • The normal AG depends on serum albumin. The expected "normal" range for AG can be calculated by 2.5 x albumin (g/dl) +/- 2.
    • For severe hypoalbuminemia (2.0 g/dl), the expected gap would be 5 +/- 2, and an AG of 8-12 would be abnormally elevated.
  • Paraproteinemia (especially polyclonal or monoclonal IgG) can lower anion gap due to the abundance of cations added to the system; this should be accounted for when evaluating anion gap since it could obscure a truly elevated AG.
  • For calculation, >12 is generally used as the cutoff for a high AG. Because 12 is the upper limit of normal, keep in mind that approach might increase specificity for elevated AG at the cost of sensitivity. Because of person-to-person variation in “normal” AG, it can be useful to review a patient’s previous values of AG to establish a baseline and give context to small, acute changes in AG. 
• If the AG is elevated, there is an anion gap metabolic acidosis present. Proceed to step 2b.
• If the AG not elevated, there is not an anion gap metabolic acidosis present. Proceed to step 3.

Step 2b: Calculate and Interpret the Excess Anion Gap

• If there is an anion gap, determine whether the excess in AG fully explains the drop in bicarbonate.
  • For every 1 unit increase in AG (△AG↑) from its upper limit of normal value (12), there is an expected 1 drop in HCO3 (△HCO3↓) from its “normal” value (24). 
  • If the AG and HCO3 are balanced, (△HCO3 ± 5mmol/L ≈ △AG), there is no concurrent metabolic process.
  • If the HCO3 is lower than expected (△HCO3 ± 5mmol/L > △AG), there is a concurrent non-elevated anion gap metabolic acidosis as evidenced by a reduction of HCO3.
  • If the HCO3 is higher than expected (△HCO3 ± 5mmol/L < △AG), there is a concurrent metabolic alkalosis as evidenced by an excess of HCO3.
  • The 5mmol/L correction accounts for the normal range of bicarbonate.
• Alternate approach to answer this question: “correcting” the bicarb.
  • Corrected HCO3 = HCO3 + (AG - 12).
  • If the sum is greater than the usual bicarbonate range (> 28), there is a concomitant metabolic alkalosis. If the sum is less than the usual bicarbonate range (< 22), there is a concomitant non-elevated anion gap metabolic acidosis.
• Finally, there is a third approach to this same problem, “the delta-delta,” which is not discussed here.

Step 3: Evaluate for Compensation

Evaluate for physiologic compensation for the acid-base disorder. Any values above or below expected suggest an additional acid-base disturbance; a compensation should not normalize or overcorrect the pH. The following table gives a quick rule-of-thumb for evaluating compensation. More detailed formulas are listed below. 

• Metabolic acidosis: 
  • There are several methods for estimating pCO2 compensation:
    • Most precise: Winter’s formula.
      • Expected pCO2 = [(1.5 x serum HCO3) + 8] ± 2.
        • If measured pCO2 > expected pCO2, concomitant respiratory acidosis is present.
        • If measured pCO2 < expected pCO2 concomitant respiratory alkalosis is present.
  • Less precise: expected pCO2 = [serum bicarbonate) + 15.
  • Least precise: expected pCO2 = last two digits of pH (e.g. pH 7.15, pCO2 should be 15).
• Metabolic alkalosis:
  • pCO2 increases 0.7 mmHg for each mmol/L increase in HCO3.
  • Expected pCO2 = 0.7([HCO3] - 24) + 40 ± 2.
• Respiratory acidosis:
  • Acute (<3-5 days): HCO3 increases 0.1 mmol/L for every mmHg increase in pCO2.
  • Chronic (>3-5 days): HCO3 increases 0.35 mmol/L for every mmHg increase in pCO2.
  • The reason respiratory disorders have acute/chronic phases is that it takes the kidneys several days to fully compensate. Respiratory compensation of metabolic pH disorders is nearly immediate.
• Respiratory alkalosis:
  • Acute (<3-5 days): HCO3 decreases 0.22 mmol/L for every mmHg decrease in pCO2.
  • Chronic (>3-5 days): HCO3 decreases 0.4 mmol/L for every mmHg decrease in pCO2.
• No need to calculate compensation if you have a primary respiratory disorder as well as a gap acidosis, as that has already been done in step 2b.
• Key point: physiologic compensation never normalizes the pH; a normal pH indicates at least 2 primary acid-base disorders.

Differential Diagnosis for Each Disorder

Respiratory alkalosis

• Anything that causes hyperventilation:
  • CNS: anxiety, pain, primary CNS disorders, Cheyne-Stokes respirations.
  • Pulmonary receptor stimulation (asthma, pneumonia, pulmonary edema, or embolism).
  • Systemic: chronic liver failure, pregnancy, sepsis, hyperthyroidism.
  • Drugs (salicylates, theophylline).

Respiratory acidosis

• Respiratory center inhibition (sedatives, excessive supplemental O2 administration in chronic hypercarbic respiratory failure).
• Neuromuscular disorder (Guillain–Barré, myasthenia gravis, myopathies, hypokalemia).
• Chest wall or pleural disorders (scoliosis, ankylosing spondylitis, pneumothorax).
• Airway obstruction (tracheal/laryngeal/bronchial).
• Acute and chronic lung disease through several of the above mechanisms (obstructive sleep apnea, obesity hypoventilation syndrome, COPD, interstitial fibrosis).
• Iatrogenic hypoventilation (insufficient respiratory rate or tidal volumes on mechanical ventilation).

Metabolic alkalosis

• Saline–responsive (urine chloride < 25 mEq/L): volume depletion, vomiting, anorexia, NG drainage, diuretic use, lingering compensation for resolved hypercapnia, cystic fibrosis, villous adenoma.
• Saline–resistant (urine chloride > 40 mEq/L): hypokalemia (K<2), primary hyperaldosteronism (Conn’s syndrome), secondary hyperaldosteronism (CHF, cirrhosis, ascites), Cushing’s disease, Bartter’s syndrome, Gitelman’s syndrome, Liddle’s syndrome, licorice ingestion.
• Miscellaneous: poorly resorbed anion (high dose carbenicillin or other penicillin derivatives), refeeding alkalosis, administration of alkali (excessive treatment for acidosis, massive transfusions with citrate anticoagulant, milk alkali).

High anion gap metabolic acidosis

• Mnemonic for causes: the mnemonic GOLDMARK replaces the older MUDPILERS and is a more accurate reflection of common modern causes of anion gap acidosis. The starred conditions are the most common causes. 
  • G    glycols (ethylene and propylene)
  • O   5-oxoproline (from chronic acetaminophen use, often in those with malnutrition) 
  • L    L-lactic acid*
  • D    D-lactic acid (from short-gut syndrome)
  • M   Methanol
  • A    Aspirin
  • R    Renal failure*
  • K    Ketones* (from diabetes, alcohol, or starvation)
• Helpful studies in building differential:
  • Serum ketones (β-hydroxybutyrate level) and serum lactate.
  • Urine toxicology screen, serum salicylate levels, ethanol levels and possibly acetaminophen levels (if history/concern for ingestion).
  • Serum CK if there is suspicion for massive rhabdomyolysis.
  • If concerned for an ingestion, calculate the osmolal gap to assess for toxins such as glycols and methanol.
    • Osmolal gap = measured serum osmolality – calculated serum osmolality.
    • Calculated serum osmolality = (2 x Na) + (BUN / 2.8) + (Glucose / 18) + (EtOH / 3.7).
    • Normal gap = 0-6 (can be ~10 in ICU given other unknown osmoles).
    • If osmolal gap > 10, consider ingestion of toxic alcohol, but note that osmolal gap is NOT very sensitive.
      • Isopropyl alcohol causes an osmolal gap but does NOT cause AG metabolic acidosis.
      • As toxic alcohol is metabolized, anion gap will increase as osmolal gap slowly decreases -- so an elevated anion gap with low osmolal gap can be seen in late phase of intoxication. Likewise, a very early ingestion may present with elevated osmolal gap but normal anion gap.

Normal anion gap metabolic acidosis

• Renal causes
  • Chronic kidney disease.
  • Renal tubular acidosis (more detail in RTA section).
  • Urinary tract diversions (uretosigmoidostomy or fistula, ileal conduit).
• GI causes
  • Functional: diarrhea, ileus.
  • Structural: pancreatic fistula, villous adenoma.
• Iatrogenic causes
  • Administration of chloride (e.g. large volume resuscitation with normal saline), plasma exchange.
  • TPN.
  • Cholestyramine.
  • Acetazolamide.
• Determining the urine anion gap (UAG) helps to distinguish renal from GI causes. 
  • UAG is a surrogate for urine NH4+, the unmeasured cation in the urine.
  • UAG = UNa + UK – UCl.
    • UAG < 0 suggests extrarenal cause: the kidney is appropriately compensating for the acidosis by secreting NH4+. Mnemonic: UAG is ne-GUT-ive in GI causes. 
    • UAG > 0 suggests renal cause (UAG may be negative in some cases of proximal RTA).
  • Note: UAG should not be used if there is excretion of another anion (lactate, DKA anions, etc.) OR if urine sodium <20 mEq/L (insufficient Na+ delivery to the distal tubule does not allow for H+ exchange required for urinary acidification).
    • If urine sodium <20 mEq/L, consider calculating urine osmolal gap (UOG) instead.
      • UOG = 2(UNa +UK) + Uurea/2.8 + Uglucose/18. 
      • UOG <50 is suggestive of RTA.

Common Questions in Metabolic Acidosis

What is the pathophysiology of saline-induced normal anion gap (hyperchloremic) metabolic acidosis? 

The normal anion gap metabolic acidosis resulting from large volume administration of normal saline (NS) can be explained as a dilutional effect on the existing bicarbonate level. This is because NS contains a supraphysiologic concentration of chloride (154 mEq/L) and no bicarbonate; thus it will raise serum chloride while diluting other anions - namely lowering bicarbonate - inducing a metabolic acidosis. This may be avoided by using a balanced intravenous solution such as Lactated Ringer’s or Plasmalyte. These solutions have a lower [Cl-] concentration compared to NS and, most importantly, contain lactate and acetate respectively, which are rapidly metabolized to an equimolar amount of bicarbonate.  

What are the disadvantages of giving bicarbonate in the setting of acute metabolic acidosis? 

• Can cause an initial transient worsening of intracellular acidosis.
• Can lead to generation of increased CO2 which causes respiratory acidosis (especially in patients with respiratory failure such as in ARDS).
• Can represent a large sodium load that can exacerbate hypervolemia.
• Can worsen hypokalemia.

The 2008 Surviving Sepsis guidelines recommend against its use in sepsis if pH > 7.15. One can consider temporary NaHCO3 administration in the setting of severe metabolic acidosis (pH < 7.1) or to facilitate permissive hypercarbia. In most cases of normal anion gap metabolic acidosis, it is probably safe. Consider renal replacement therapy in cases of severe acidosis refractory to medical therapy.

Key Points

• It is important to follow a systematic approach each time interpreting a blood gas. Establish the primary disorder. Calculate AG to reveal an anion gap metabolic acidosis (can be hidden when pH is normal but mixed disorders are present). Compare △AG and △HCO3 to look for concurrent metabolic alkalosis or normal anion gap metabolic acidosis. Use a compensation chart/formula to reveal “overcompensation” or “undercompensation” which indicates the presence of another disorder.
• The presence of a normal pH with abnormal pCO2 and bicarbonate suggests a mixed acid-base disorder with counterbalancing acidosis and alkalosis.
• The pCO2 and serum bicarbonate typically move in parallel with an isolated acid-base disorder; both are high OR both are low. For example, a respiratory acidosis will have an increased pCO2 with a compensatory increase in serum bicarbonate. If the pCO2 and serum bicarbonate move in opposite directions (one high and one low), then you should consider the possibility of two simultaneous primary acid-base disorders (i.e., a mixed acid-base disorder).
• A mixed acid-based disorder consists of any combination of at least two disorders: two metabolic disturbances OR one respiratory and one metabolic. Triple acid-base disorders include one respiratory disorder (acidosis or alkalosis) with two metabolic disorders (high gap and normal gap metabolic acidosis OR high gap metabolic acidosis and metabolic alkalosis).

Berend K, de Vries APJ, Gans ROB. Physiological Approach to Assessment of Acid–Base Disturbances. New England Journal of Medicine. 2014;371(15):1434-1445. Doi :10.1056/NEJMra1003327

Haber RJ.  A practical approach to acid-base disorders.  West J Med 1991;155:146-151.

Kraut JA, Nagami GT. The serum anion gap in the evaluation of acid-base disorders: what are its limitations and can its effectiveness be improved? Clin J Am Soc Nephrol 2013;8(11):2018–24.

Kraut JA, Madias NE. Serum anion gap: its uses and limitations in clinical medicine. Clin J Am Soc Nephrol 2007;2(1):162–74.

Kraut JA, Mullins ME. Toxic Alcohols. New England Journal of Medicine. 2018;378(3):270-280. doi:10.1056/NEJMra1615295 

Rose & Post. Clinical Physiology of Acid-Base and Electrolyte Disorders: 5th Edition. 2001