What the osmolal gap actually measures
Serum osmolality is the total concentration of dissolved particles in plasma water, reported in milliosmoles per kilogram of water. Three solutes account for almost all of it in a healthy person: sodium with its accompanying anions, glucose, and urea. Add those three up and you have the calculated osmolality. Send the same tube to an osmometer and you get the measured osmolality, which counts every particle present, named or not.
The difference between the two is the osmolal gap, and it is a detector for solutes nobody ordered a test for. A methanol or ethylene glycol molecule contributes to the measured value and contributes nothing to the calculation, so it shows up entirely as gap. That is the whole clinical point: in an emergency department at three in the morning you cannot get a methanol level back in twenty minutes, but you can get a sodium, a glucose, a BUN and an osmometer reading, and their arithmetic tells you whether an unmeasured osmole is present.
The gap is a screening quantity, not a diagnosis. It is sensitive to a large recent ingestion and blind to a late one, because once methanol has been oxidised to formate the osmoles are gone and only the acid remains. That is why the gap is always read next to the anion gap and the arterial pH: early poisoning shows a large osmolal gap with a normal anion gap, and late poisoning shows the reverse, with the crossover somewhere in the middle.
Why the formula has the shape it has
Each term converts a laboratory concentration into millimoles of particles per litre, and osmolality counts particles. The default equation here is the Smithline–Gardner form, 2 × Na + glucose/18 + BUN/2.8, published alongside the original description of the osmolal gap in 1976 and still the version most toxicology sources use. The alternative offered in the advanced panel is the Dorwart–Chalmers regression, 1.86 × Na + glucose/18 + BUN/2.8 + 9, fitted in 1975 against measured osmolalities.
2 × Na. Sodium is the dominant extracellular cation, and electroneutrality means every sodium ion travels with an anion, mostly chloride and bicarbonate. Doubling sodium therefore stands in for the whole sodium-plus-anion pair. It is an approximation with a known bias: because potassium and its anions are ignored and the activity coefficient of a real electrolyte solution is below one, the doubled term slightly overstates in some patients and understates in others, which is exactly the residual that regression-fitted alternatives such as the 1.86 × Na + 9 form were built to absorb.
Glucose ÷ 18. Glucose has a molecular weight of 180.16 g/mol. A concentration in mg/dL divided by one tenth of the molecular weight gives mmol/L, so 180.16/10 rounds to 18. Glucose is a non-dissociating molecule, so one millimole is one milliosmole.
BUN ÷ 2.8. BUN reports the nitrogen in urea, not urea itself. Urea contains two nitrogen atoms with a combined weight of 28.02, so dividing mg/dL by 2.8 converts BUN directly to millimoles of urea per litre. Laboratories reporting in SI already give urea in mmol/L, and the divisor disappears.
Ethanol ÷ 3.7. Ethanol weighs 46.07 g/mol, so the strict conversion divisor is 4.607. The toxicology literature commonly uses 3.7 instead, which credits ethanol with more osmotic effect per mg/dL and so removes more of the gap. The choice matters: at an ethanol of 300 mg/dL the two divisors differ by 16 mOsm/kg, which is enough to move a patient across a screening threshold. Both are offered above; state which one you used when you document the number.
Sodium, glucose and urea are the only three terms because everything else in plasma is either present in trivial molar amounts or is a large molecule. Albumin at 4 g/dL sounds substantial but weighs about 66,000 g/mol, so it contributes well under one milliosmole per kilogram. Osmolality is a molar census, and mass is irrelevant to it.
Worked example: an intoxicated patient with a metabolic acidosis
A 46-year-old arrives confused. The chemistries come back sodium 142 mEq/L, glucose 216 mg/dL, BUN 22 mg/dL, and the laboratory measures osmolality by freezing-point depression at 322 mOsm/kg.
- Sodium term. 2 × 142 = 284 mOsm/kg.
- Glucose term. 216 ÷ 18 = 12.0 mOsm/kg.
- Urea term. 22 ÷ 2.8 = 7.86 mOsm/kg.
- Calculated osmolality. 284 + 12.0 + 7.86 = 303.86 mOsm/kg.
- Osmolal gap. 322 − 303.86 = 18.14 mOsm/kg.
- What that gap would represent. Methanol weighs 32.04 g/mol, so 18.14 × 3.204 = 58.1 mg/dL. Ethylene glycol weighs 62.07 g/mol, so 18.14 × 6.207 = 112.6 mg/dL. Either concentration is well above the level at which antidote therapy is considered.
Now suppose the ethanol comes back at 120 mg/dL. Its contribution is 120 ÷ 3.7 = 32.4 mOsm/kg, so the calculated osmolality rises to 303.86 + 32.4 = 336.3. Against a measured 322 the gap becomes 322 − 336.3 = −14.3 mOsm/kg. Ethanol has more than accounted for the excess, and the case for a co-ingested toxic alcohol weakens considerably — while raising the separate question of whether the 3.7 divisor has over-credited it, since the strict 4.607 divisor would give 26.0 mOsm/kg and a gap of −7.9.
That reversal is the single most useful thing this calculator does. An unadjusted gap of 18 in a patient who smells of alcohol invites a fomepizole decision that the ethanol level alone can resolve.
How to read the number you get
Most laboratories print a reference range of roughly −10 to +10 mOsm/kg and many clinicians use +10 as a single cut-point. That cut-point is convenient and imprecise. Hoffman and colleagues re-examined the normal values and found the distribution in healthy people is centred slightly below zero with a spread wide enough that individuals with nothing wrong return values from about −14 to +10; a single patient's baseline is not knowable, so a gap of +9 in someone whose true baseline is −8 already represents 17 mOsm/kg of unmeasured osmoles.
Read the result in three bands, and read all three against the clinical picture:
Gap under about 10 mOsm/kg. Consistent with the healthy distribution. It does not exclude a toxic alcohol, particularly late in the course after the parent compound has been metabolised, and it does not exclude a small but dangerous ethylene glycol ingestion — 20 mg/dL of ethylene glycol, around the concentration at which antidote therapy is considered under the American Academy of Clinical Toxicology guidance, produces only 3.2 mOsm/kg of gap, which is inside the noise.
Gap of 10 to 25 mOsm/kg. Worth explaining. Toxic alcohols are one explanation among several. Severe ketoacidosis and lactic acidosis both raise the gap, as do mannitol, glycerol, sorbitol irrigation fluid, intravenous immunoglobulin in a sucrose vehicle, and the propylene glycol carrier in intravenous lorazepam and phenytoin infusions.
Gap above 25 mOsm/kg. Rarely explained by measurement scatter. In a patient with an unexplained high anion gap metabolic acidosis this is grounds to treat empirically while the confirmatory assay is pending, rather than to wait for it.
A negative gap deserves as much attention as a positive one, because it usually means an analytic problem rather than a physiological one: the osmolality and the chemistries came from different draws, the sodium is falsely low from severe hyperlipidaemia or paraproteinaemia in an indirect ion-selective electrode, or a unit has been entered in the wrong system. Check the specimen before you check the patient.
Osmolal contribution of the alcohols and glycols
| Substance | Molecular weight (g/mol) | Divisor (mg/dL → mOsm/kg) | mOsm/kg from 25 mg/dL | mOsm/kg from 100 mg/dL |
|---|---|---|---|---|
| Methanol | 32.04 | 3.20 | 7.8 | 31.2 |
| Ethanol | 46.07 | 4.61 (3.7 by convention) | 5.4 | 21.7 |
| Acetone | 58.08 | 5.81 | 4.3 | 17.2 |
| Isopropanol | 60.10 | 6.01 | 4.2 | 16.6 |
| Ethylene glycol | 62.07 | 6.21 | 4.0 | 16.1 |
| Propylene glycol | 76.09 | 7.61 | 3.3 | 13.1 |
The heavier the molecule, the less gap a given concentration in mg/dL produces. Ethylene glycol is nearly twice the weight of methanol, so 20 mg/dL — the concentration at which antidote therapy is considered — raises osmolality by only 20 ÷ 6.207 = 3.2 mOsm/kg, well inside normal variation. That is why a normal gap never rules ethylene glycol out.
The osmometer method decides whether the number means anything
Freezing-point depression and vapour-pressure osmometry give the same answer for salt, sugar and urea, and completely different answers for alcohols. A volatile solute lowers the freezing point but also raises the vapour pressure of the solution, so a vapour-pressure instrument under-reads or misses it entirely. A vapour-pressure osmometer can therefore return a reassuringly normal gap in a patient with a lethal methanol level. Before you act on a small gap, confirm with the laboratory which instrument produced it.
Mistakes that make an osmolal gap misleading
- Using chemistries from a different draw. The gap is a difference between two large numbers. A sodium drawn two hours earlier, after two litres of saline, moves the calculated value by several mOsm/kg on its own.
- Confusing osmolality with osmolarity. Osmolality is per kilogram of solvent water and is what an osmometer reports; osmolarity is per litre of solution and is what the equation strictly returns. In plasma they differ by roughly the water fraction, about 93%, which is small but not zero.
- Forgetting ethanol. The commonest false positive. Any patient in whom you suspect methanol is a patient who may also have drunk ethanol, and 100 mg/dL of ethanol alone produces about 22–27 mOsm/kg of gap depending on the divisor.
- Treating a normal gap as a negative screen. The gap detects parent alcohol only. Once it has been metabolised, the diagnosis lives in the anion gap and the arterial blood gas.
- Ignoring iatrogenic osmoles. Mannitol given for raised intracranial pressure, and the propylene glycol vehicle in continuous lorazepam infusions, both produce a real, drug-related gap that has nothing to do with a poisoning.
- Comparing a gap against the wrong reference interval. The 2 × Na and 1.86 × Na + 9 equations do not produce the same calculated osmolality, so they do not share a cut-point.
Where the osmolal gap sits among the other bedside gaps
The osmolal gap is one of three arithmetic screens that a chemistry panel supports, and each has a different blind spot. The anion gap finds unmeasured anions, so it turns positive once methanol becomes formate and ethylene glycol becomes glycolate and oxalate — precisely when the osmolal gap is fading. Running both in sequence gives a rough clock on the ingestion: a wide osmolal gap with a normal anion gap points to a recent one, matched gaps to an intermediate presentation, and a wide anion gap with a normal osmolal gap to a late one or to a different diagnosis entirely.
The third screen is the corrected sodium in hyperglycaemia, which is a different use of the same physiology: glucose that cannot enter cells pulls water into the extracellular space and dilutes sodium without changing total osmolality. That is why a patient in hyperosmolar hyperglycaemic state can have a sodium of 125 and an osmolality of 340 at the same time. Calculators for those adjacent numbers are worth keeping open beside this one, along with corrected calcium for the albumin-bound fraction and estimated GFR or Cockcroft–Gault creatinine clearance when renal handling of the same solutes is in question.
When you have an ethanol level, treat the Widmark estimate of blood alcohol as a cross-check on plausibility rather than a substitute for the assay: an estimated concentration far from the measured one usually means the history is wrong, and the history is what the toxic-alcohol decision rests on.
None of this replaces a serum methanol or ethylene glycol assay. The gap buys you the hours before that assay returns, and it does so honestly only if you know its reference interval is wide, its sensitivity depends on molecular weight, and its validity depends on an osmometer method you have not personally chosen.
Key terms
- Osmolality
- Particle concentration expressed per kilogram of solvent water, in mOsm/kg. This is what a freezing-point osmometer measures and it is independent of temperature.
- Osmolarity
- Particle concentration per litre of whole solution, in mOsm/L. Because plasma is about 93% water, osmolarity runs a few per cent below osmolality in the same specimen.
- Osmolal gap
- Measured osmolality minus calculated osmolality. It quantifies osmotically active particles that the sodium, glucose and urea terms do not account for.
- Tonicity
- The osmolality of only those solutes that cannot cross the cell membrane. Urea crosses freely, so it adds to osmolality without adding to tonicity, which is why uraemia raises osmolality without shrinking cells.