Anion Gap Calculator

Five panels, five gaps — and what the correction does to each

Every figure below was run through the calculator above, starting from its default panel — sodium 140, chloride 100, bicarbonate 24, albumin 2.0 — and changing one thing at a time. The formula never changes: anion gap equals sodium minus the sum of chloride and bicarbonate, all in mEq/L, and the albumin-corrected gap adds 2.5 mEq/L for every 1.0 g/dL that albumin sits below 4.0 g/dL. The usual reference range quoted on the page is 8 to 12 mEq/L, but reference ranges vary by laboratory and by assay method — always compare a result with the range printed on your own lab report. Nothing here is a diagnosis: this tool only does the arithmetic, and a single number without clinical context never identifies a cause.

The default row is the page’s own worked example, and it is worth following digit by digit: 140 − (100 + 24) = 140 − 124 = 16.0 mEq/L, above the usual 8–12 range, so the band reads “High”. The correction then adds 2.5 × (4.0 − 2.0) = 5.0, giving a corrected gap of 21.0 mEq/L. The correction does not change the measured gap; it estimates what the gap would be if albumin were at its normal level, on the reasoning that albumin itself is a negatively charged protein making up a large part of the normal gap. When albumin is low, the uncorrected gap can look less abnormal than the underlying chemistry, because part of the gap that albumin would normally supply is missing — and the correction restores it arithmetically.

The next three rows are the teaching set, because in each of them the uncorrected gap sits inside the usual range while the corrected gap sits outside it. Raise chloride from 100 to 105 and the gap falls by exactly 5, from 16.0 to 11.0 — each unit of chloride moves the gap by one unit in the opposite direction, since chloride is subtracted — while the corrected gap falls by the same 5, from 21.0 to 16.0, still above the usual range. Raise bicarbonate from 24 to 28 instead and the gap falls by 4, from 16.0 to 12.0, the very top of the usual range, with a corrected gap of 17.0. Raise chloride to 108 and the gap falls to 8.0, the very bottom of the usual range, with a corrected gap of 13.0 — still above it. In all three cases, a reader who looked only at the uncorrected gap would call the panel normal, while the corrected figure points above the range. That disagreement is not a contradiction between two calculators; it is the correction doing the job it exists to do, revealing unmeasured anions that low albumin was hiding. Which figure a clinician weighs more, and what either figure means for a particular patient, is a clinical judgement made with the full panel, the history and the symptoms — not something this page, or any arithmetic, can decide.

The last row isolates the correction itself. With albumin at 4.0 — the level the correction treats as normal — the correction term is 2.5 × (4.0 − 4.0) = 0, and the corrected gap equals the uncorrected gap exactly: 16.0 mEq/L in both columns. The correction only ever adds, never subtracts, on this page’s rule, because it only applies when albumin sits below 4.0; at or above that level there is nothing to add back. That asymmetry is worth remembering when comparing panels: two patients with the same measured gap of 16.0 can have corrected gaps of 16.0 and 21.0 depending on albumin alone, and the difference is entirely the albumin term, not any difference in sodium, chloride or bicarbonate. The page’s scale graphic makes the same point visually, marking the usual 8–12 band on a 0–30 scale and placing the result against it — but the graphic, like the band label, describes the uncorrected gap, and the corrected figure must be read alongside it, not instead of it.

Why albumin changes the answer, and when to leave it blank

Blood must stay electrically neutral, and that single fact is the whole idea behind the anion gap. The routine electrolyte panel measures the main positively charged ion, sodium, and the main measured negatively charged ions, chloride and bicarbonate. If those were the only ions present, sodium would exactly equal chloride plus bicarbonate — but it does not, because blood also contains unmeasured anions: proteins, phosphates, sulfates and organic acids that the routine panel does not report directly. The gap between sodium and the measured anions is therefore an estimate of those unmeasured anions, not a measurement of any single one of them. A gap in the usual range means the unmeasured anions are roughly where they are usually found; a gap above it means they are raised relative to bicarbonate; a gap below it is less common and can reflect low albumin, since albumin contributes much of the normal gap, or laboratory variation.

Albumin’s role is what makes the correction necessary. Albumin is itself a negatively charged protein and makes up a large part of the normal gap, so when albumin falls, the gap falls with it for a reason that has nothing to do with acids: part of the gap’s normal contents are simply absent. A patient with raised unmeasured acids and low albumin can therefore show a gap that looks normal, because the acids push the gap up while the missing albumin pulls it down, and the two effects partly cancel. The correction — adding 2.5 mEq/L for every 1.0 g/dL that albumin sits below 4.0 — is an attempt to undo that cancellation arithmetically: it asks what the gap would be if albumin were normal, so that the acid contribution can be judged on its own. It is an estimate built on an average relationship, not a measurement, and different laboratories and textbooks may use slightly different correction factors; the page uses 2.5, states it openly, and shows its working (16.0 + 2.5 × (4.0 − 2.0) = 21.0) so that any reader can check or replace it.

Leaving the albumin box blank is a legitimate choice, and the page treats it as one: blank means the correction is skipped and only the uncorrected gap is reported. The default value of 2.0 in the box exists to show the corrected example, not to assert anything about any real patient — if your albumin is not 2.0, the corrected figure shown on load is not yours, and entering your own albumin, or leaving the box blank, is the correct response. The same discipline applies to every input here: the calculator computes from the numbers typed in, it does not know whether they came from a lab report, and a gap computed from mistyped electrolytes is precisely wrong in the amount of the mistyping. Sodium outside 100 to 180 mEq/L even earns a warning on the page to double-check the value — a small guard against the most consequential kind of error on a page like this, which is not a subtle formula error but a plain transcription error.

What the gap is not: diagnosis, cause, or verdict

The page states its own limit in a section titled “What this calculator cannot do”, and it is worth restating here because it governs everything else on the page. This tool only does the arithmetic. It cannot diagnose a condition, tell you why a gap is high or low, or replace a clinician reading the full lab panel with your history and symptoms. A single number without that context is never a diagnosis, and treatment decisions belong with your clinician. That limit is not modesty; it is accuracy. The page’s own FAQ names textbook settings in which a high gap is described — lactic acidosis, ketoacidosis, kidney failure and certain ingested toxins — and immediately adds that the number alone does not identify the cause. Four different causes, one number: the gap can raise the question, but it cannot answer it, and any page that claimed otherwise from three electrolytes and an albumin would be inventing the answer.

The low end of the range carries its own cautions, in the opposite direction. A low gap is less common than a high one and can reflect low albumin or laboratory variation, and the page’s FAQ advises that very low or negative values are often worth rechecking with the laboratory before any interpretation. That advice generalises: any surprising gap — high, low, or inconsistent with the rest of the panel — should first be checked for transcription (a chloride of 105 typed as 150 moves the gap by 45), for the albumin question (was the correction applied, and with the right albumin?), and for the laboratory’s own reference range, which varies by assay method and may not be 8–12 on your report at all. The commonly quoted 8–12 range on this page is a starting point for reading the band label, not a substitute for the range printed on your own lab report, and a gap of 12.0 or 8.0 — the very edges of the range, both verified above as “Normal” on this page — sits exactly where laboratory variation matters most.

Used within those limits, the gap is a genuinely useful piece of arithmetic. It turns four routine numbers into one summary that a clinician can read at a glance, it flags panels worth a second look, and with the albumin correction it avoids one well-understood trap — the normal-looking gap that hides raised acids behind low albumin. The calculator’s contribution is to do that arithmetic transparently, showing the subtraction (140 − (100 + 24) = 16.0) and the correction (16.0 + 2.5 × (4.0 − 2.0) = 21.0) in the open, so that nothing is hidden inside a black box. What it cannot do — interpret the result for a person, identify a cause, or recommend anything — it leaves, correctly, to the clinician with the full picture. A tool that knows the boundary of its own usefulness is more trustworthy, not less, than one that pretends the boundary is not there.

Reading a panel as a whole, rather than as four isolated numbers, is the skill that turns the gap from a party trick into a useful summary — and the page’s own structure encourages it. Sodium, chloride and bicarbonate move together in real panels more often than they move alone: a bicarbonate that falls while chloride rises leaves the gap changed by the sum of both moves, while a sodium that falls with them can leave the gap unchanged even though every number on the panel moved. That is why the gap is computed from all three at once rather than read off any single one, and why a panel in which sodium, chloride and bicarbonate all shifted in the same direction can still show the same gap as before. The gap summarises the relationship between the numbers, not the numbers themselves, and two quite different panels can honestly share a gap of 16.0 while differing in every input — the default panel’s 140/100/24 and a hypothetical panel with higher sodium balanced by higher chloride and bicarbonate would both compute to the same gap, for entirely different physiological reasons that only the full clinical picture could separate.

The sensitivity of the gap to each input is worth stating as a rule, because it governs how carefully each number must be transcribed. Every 1 mEq/L error in sodium moves the gap by 1 in the same direction; every 1 mEq/L error in chloride or bicarbonate moves it by 1 in the opposite direction; and every 0.4 g/dL error in albumin moves the corrected gap by 1 (since 2.5 per 1.0 g/dL means 1.0 per 0.4 g/dL). A chloride mistyped by 5 — 105 entered as 100 — moves the gap by 5, from 11.0 to 16.0 on the panels above, crossing the entire usual range on a keystroke. An albumin mistyped by 1.0 g/dL moves the corrected gap by 2.5, which is a quarter of the usual range’s width. Those sensitivities are not a criticism of the formula; they are the reason the page warns about out-of-range sodium, shows its subtraction and correction in the open, and advises rechecking very low or negative values with the laboratory. On a page where a single keystroke can move the answer across a clinical boundary, the discipline of checking the inputs matters more than the sophistication of the formula.

Finally, the units themselves deserve a sentence, because unit errors are the quietest way to get a confident wrong answer. Sodium, chloride and bicarbonate are entered in mEq/L, the units of the routine electrolyte panel, and the gap inherits those units: 16.0 mEq/L, not 16.0 of anything else. Albumin is entered in g/dL, a different unit for a different kind of quantity, and the correction factor of 2.5 carries the conversion between them implicitly — 2.5 mEq/L of gap per 1.0 g/dL of albumin below 4.0. Mixing the two — entering albumin in g/L, for instance, which would make a normal albumin of 40 g/L look like a catastrophic 40 g/dL — would corrupt the correction completely while leaving the uncorrected gap untouched, a failure mode that shows up only in the corrected column. The page labels each box with its unit for exactly this reason, and the habit of reading the unit label before typing the number is, on this page, worth more than any other single precaution.

What a clinician does with the gap, in broad outline, is also worth understanding — not so that a reader can imitate it, but so that the gap’s limits are concrete rather than abstract. The first step is always the one this page cannot take: checking the numbers against the patient, the sample and the rest of the panel, because a gap computed from a haemolysed sample, a mistyped chloride or a panel drawn at a different time from the albumin is a gap about the wrong blood. The second step is the albumin question, which this page models: is the gap, corrected or not, actually abnormal for this laboratory, and does the albumin explain any discrepancy between the two figures? The third step leaves arithmetic entirely: the history and examination, the other laboratory results, and the trajectory — whether the gap is rising, falling or stable across successive panels — all bear on what the gap means in a way no single panel can. A gap of 21.0 corrected, like the default panel’s, raises a question that only that fuller picture can answer, and the same corrected gap in two different patients can lead to two different investigations, because the gap is a clue, not a conclusion.

The trajectory point deserves emphasis, because it is the use of the gap that a single-panel calculator most obscures. One panel gives one gap; successive panels give a direction, and direction often matters more than level. A gap falling from 21.0 toward the usual range across days of treatment tells a different story from a gap stuck at 21.0, and a gap rising from 11.0 to 16.0 tells a different story again — even though the final values in the second and third stories are the same number. This calculator cannot show trajectories, because it computes one panel at a time, and any reader tracking a changing gap should record each panel’s inputs and outputs together — sodium, chloride, bicarbonate, albumin, gap and corrected gap, with dates — rather than carrying a single number forward in memory. A table of successive gaps, each computed transparently from its own panel, is a far better basis for a clinical conversation than any one gap remembered in isolation, and it is also the record that lets a clinician see at a glance whether the correction, the albumin, or the electrolytes themselves are what changed.

One more practical note, for anyone who will use this calculator more than once: keep the panel and the gap together. A gap written down without its sodium, chloride, bicarbonate and albumin — “my gap was 16” — cannot be checked, corrected or compared, because the four inputs that produced it are gone, and a later panel’s gap cannot be compared with it on equal terms. The calculator shows its working for exactly this reason: the subtraction line (140 − (100 + 24) = 16.0) and the correction line (16.0 + 2.5 × (4.0 − 2.0) = 21.0) are the record, not decoration. Copy the whole block — inputs, gap, corrected gap, and the laboratory’s own range for that date — into whatever record you keep, and the next person to read it, clinician or patient, can reproduce the arithmetic in seconds and see at once whether the inputs, the correction, or the range is what changed since last time.

And if anything on this page ever seems to contradict your lab report — a band label that disagrees with the report’s flag, a corrected gap that differs from the laboratory’s own corrected figure — trust the report and ask the laboratory. Laboratories differ in assay methods, in reference ranges, and sometimes in the correction factor they apply, and a difference between this page and a report is usually a difference in one of those choices, not an error in either arithmetic. The page’s job is to make its own choices visible — 8 to 12 for the range, 2.5 for the correction, 4.0 for normal albumin — so that any such difference can be found and explained in minutes, rather than argued about in the dark.

Two formulas, side by side: with and without potassium

The anion gap is computed two ways, and the difference between them is one ion: potassium. The formula this calculator uses leaves potassium out: anion gap equals sodium minus the sum of chloride and bicarbonate. The alternative formula puts potassium in: anion gap equals sodium plus potassium, minus the sum of chloride and bicarbonate. Because potassium in blood is normally around 4 to 5 mEq/L, the formula that includes it returns a gap about 4 higher than the formula that leaves it out, for the same patient, on the same panel. Neither formula is wrong. They are two conventions, and the reference range belongs to the convention: the range most often quoted for the formula without potassium is 8 to 12 mEq/L, and the corresponding range for the formula with potassium is 12 to 16 mEq/L. A gap of 14 is high on the first convention and ordinary on the second, and the number alone, without the convention, cannot be read.

The practical rule is to know which convention the laboratory and the calculator use, and to compare like with like. This page uses the formula without potassium, states the 8 to 12 range for that formula, and applies the albumin correction to it. A laboratory report that states its own range, computed with potassium, should be read against that range, not against this page’s. A clinician comparing a gap across two panels, from two laboratories, should confirm that both used the same formula before concluding that the gap changed. The companion guide on low anion gap causes explains what a low result usually means under this page’s convention, and the companion guide on high anion gap causes explains the high-gap groups; both guides use this page’s formula and range throughout, so their numbers and this page’s numbers can be compared directly.

Reading a result against your own laboratory’s range

Reference ranges vary between laboratories, because the analysers, the assays, and the populations the ranges were derived from vary. A laboratory’s range for the anion gap may sit a point or two above or below the 8 to 12 range quoted on this page, and a result that sits just outside this page’s range may sit inside the reporting laboratory’s range, or the reverse. The reporting laboratory’s range is the range the result should be read against first, because it is the range derived for the assay that produced the result. This page’s range is the commonly quoted range for the formula without potassium, and it is a guide, not a verdict on any individual panel.

Edge values deserve a repeat before they deserve an interpretation. A gap one point above or below a range boundary, on a single panel, can reflect assay variation, a specimen handling effect, or a transient change, and the first step for a borderline result is a repeat electrolyte panel, with albumin, read alongside the clinical picture, rather than a conclusion drawn from the single number. A gap well outside the range, or a gap that moves sharply between panels, deserves prompt clinical attention, and a gap accompanied by symptoms deserves urgent assessment, whatever the number. The gap is a derived number, computed from three measured values, and its precision is the precision of those measurements combined; treating a one-point edge as a diagnosis mistakes the arithmetic for the medicine. The calculator shows the gap, the corrected gap, and where the result sits against the range, and the interpretation belongs to the clinician who knows the patient.