Anion gap. AG = Na⁺ − (Cl⁻ + HCO₃⁻). Including potassium, (Na⁺ + K⁺) − (Cl⁻ + HCO₃⁻), which adds about 4. The commonly quoted reference range is 8–16 mmol/L without potassium, but modern ion-selective-electrode analysers report chloride lower than older flame-photometry methods did, and many laboratories now quote 3–11. Always use your own laboratory's range; a gap of 14 is normal in one lab and raised in another. Both forms are shown here.

Albumin-corrected anion gap. Albumin is the main unmeasured anion, so a low albumin hides a gap. Corrected AG = observed AG + 2.5 × (4.0 − albumin in g/dL), which in SI is observed AG + 0.25 × (40 − albumin in g/L). At an albumin of 20 g/L this adds 5 mmol/L — enough to turn an apparently normal gap into a high one in a critically ill patient.

Delta ratio. Δ ratio = (AG − reference AG) ÷ (24 − HCO₃⁻), using your laboratory's midpoint gap. Under 0.4 suggests a hyperchloraemic normal-gap acidosis; 0.4–0.8 a mixed high-gap and normal-gap acidosis; 0.8–2.0 a pure high-gap acidosis; above 2.0 a high-gap acidosis on top of a pre-existing high bicarbonate (metabolic alkalosis, or compensation for chronic respiratory acidosis).

Corrected sodium. Hyperglycaemia pulls water out of cells and dilutes sodium. Katz (1973): add 1.6 mmol/L for every 5.6 mmol/L (100 mg/dL) of glucose above 5.6 mmol/L. Hillier (1999) measured a larger effect and argued for 2.4 mmol/L per 5.6 mmol/L; both are shown because the correction matters most in DKA and HHS, exactly where the choice changes whether you call the sodium normal.

Corrected calcium. Payne's formula: corrected calcium (mmol/L) = measured calcium + 0.02 × (40 − albumin in g/L), or in mg/dL, measured + 0.8 × (4.0 − albumin in g/dL). About half of plasma calcium is protein-bound, so a low albumin lowers total calcium without changing the ionised fraction that actually matters — which is why an ionised calcium is the better test whenever the correction is large or the patient is unwell.

Osmolality. Calculated osmolality = 2 × Na⁺ + glucose + urea (all mmol/L). Osmolar gap = measured − calculated, normally under 10 mmol/kg. A raised gap raises the possibility of an unmeasured osmole — ethanol, methanol, ethylene glycol, isopropanol, acetone, or mannitol — but the calculated figure ignores potassium, calcium, magnesium and protein, so a gap of 10–15 on its own is weak evidence.

Urine anion gap. Urine (Na⁺ + K⁺) − Cl⁻, used in a normal-gap metabolic acidosis to separate the causes. A negative gap means the kidney is excreting ammonium appropriately, pointing to a gastrointestinal bicarbonate loss such as diarrhoea; a positive gap means it is not, pointing to a renal tubular acidosis. It is invalid in volume depletion, ketonuria (where the unmeasured anion is the ketone) and renal impairment.

Worked panel. Na 130, K 4.0, Cl 100, HCO₃⁻ 12, albumin 25 g/L, glucose 28 mmol/L, urea 8 mmol/L, calcium 2.10 mmol/L. Anion gap = 130 − (100 + 12) = 18; with potassium, 134 − 112 = 22. Albumin correction adds 0.25 × (40 − 25) = 3.75, giving a corrected gap of 21.8. Delta ratio = (18 − 12) ÷ (24 − 12) = 0.50 — a mixed high-gap and normal-gap acidosis, so there is a hyperchloraemic component alongside the gap. Corrected sodium = 130 + 1.6 × (28 − 5.6) ÷ 5.6 = 136.4 by Katz and 139.6 by Hillier, so the apparent hyponatraemia is largely dilutional. Corrected calcium = 2.10 + 0.02 × (40 − 25) = 2.40 mmol/L — normal total calcium was masking a genuinely raised corrected value. Calculated osmolality = 260 + 28 + 8 = 296 mmol/kg.

Traps. The gap is only as good as the chloride: a chloride reported from a different sample, or a lipaemic or icteric sample, produces a gap that does not exist. Bromide, iodide and nitrate falsely raise chloride and so falsely lower the gap; lithium and some paraproteins do the opposite. Correcting sodium for glucose does not correct it for the osmotic shift of mannitol, IVIG or contrast. Payne's formula assumes a normal acid-base state — acidosis itself lowers protein-bound calcium, so in a sick patient with a low albumin the correction is doubly unreliable and an ionised calcium should be measured instead. All of these corrections are estimates about a single blood sample, not about the patient.

Disclaimer. Arithmetic for qualified clinicians — not medical advice and not a diagnosis. Interpret against your laboratory's own reference ranges. Sources: Kraut & Madias on the anion gap and delta ratio (Nature Reviews Nephrology) · Katz, Diabetes Care 1973 · Hillier et al., American Journal of Medicine 1999 · Payne et al., Lancet 1973 · Rodriguez-Soriano on the urine anion gap. Last reviewed: 2026-09-21.