Surrogate can be a verb (to appoint a representative), an adjective (providing or receiving nurture or parental care though not related by blood or legal ties), or a noun (a stand-in or substitute). Several laboratory tests function as surrogates for analytes that are difficult or impractical to measure. Examples include anion gap as a measure of "unmeasured" anions, foam stability, once used as a measure of lung surfactants in amniotic fluid; erythrocyte sedimentation rate, as an indirect measure of fibrinogen or immunoglobulins; and ferritin, which reflects the amount of iron stored in tissues. Osmolality is a surrogate measure of the total concentration of dissolved substances in plasma or urine.

Osmolality (or osmolarity, which differs from osmolality only by the units in which it is measured—osmolality is in units per weight [Kg] of solute, whereas osmolarity is in units per volume [L] of solute) is, itself, measured indirectly. It is a surrogate twice removed!

Osmolality is a colligative (from the Latin colligere, meaning “to gather”) property of solutions that is influenced by the number of dissolved particles. Vapor pressure (and hence, freezing and boiling temperatures) are colligative properties, as is osmotic pressure, from whence the names osmolality and osmolarity are derived. Osmos, in Greek, means “to push or impel.” The relationship between osmolality and vapor pressure is described by Raoult’s Law; solutes decrease the molar fraction of solvent, and therefore its vapor pressure. Most osmometers calculate osmolality based on freezing point depression.

Plasma solutes can be divided into two broad categories based on their concentrations: solutes in millimolar concentrations (eg, electrolytes, glucose, urea, creatinine, organic acids, phosphate), and solutes in micromolar or less concentrations (eg, proteins, hormones, free amino acids). Only the former contribute substantially to the osmolality of plasma, which can be estimated by familiar equation: (2 x Na) + (Glucose/18) + (BUN/2.8). Using the concentration of sodium (at least 95% of all the cations in plasma) to approximate the concentration of charge balancing anions (mostly chloride and bicarbonate), the equation usually provides an estimate, within 5-12 mosmol, of the measured osmolality. The difference is the “osmolal gap” (usually shortened to “osmol gap”) which accounts for all of the solutes other than electrolytes, glucose, and urea.

The first problem with the osmol gap is that the calculated osmolality ignores Raoult’s assumption that the solution is ideal, meaning none of the solutes interact chemically, and that assumption is only true for very dilute solutions. Linear regression analysis of measured osmolality versus sodium concentration consistently produces a coefficient of less than 2, reflecting the fact that sodium in plasma has an activity coefficient less than unity. Sodium coefficients as low as 1.9 have been proposed, and many algorithms for calculation of the estimated osmolality use something less than 2.

It is rare for any physiological component of plasma other than electrolytes, glucose, or urea to reach a high enough concentration to significantly affect the osmolality, but certain ingestibles can. Toxicologists use the mnemonic “ME DIE” to remind them that methanol, ethanol, osmotic diuretics (such as mannitol and isosorbide), isopropanol, and ethylene glycol can reach millimolar concentrations in plasma, resulting in significant elevations in the osmol gap. To aid the differential diagnosis of alcohol poisoning, most clinical laboratories only offer quantitative measurements of ethanol. When ethanol is absent, or is present in plasma concentrations that do not explain the patient’s symptoms, non-ethanol alcohol poisoning often is suspected, and clinical toxicologists turn to the osmol gap to estimate the concentration of the alcohol. Rapid assessment is critical because potentially fatal non-ethanol alcohol intoxication will prompt invasive (hemodialysis) or expensive [fomepizole, an alcohol dehydrogenase (ADH) inhibitor] therapy.

The calculation should be simple, based on the molecular weights of methanol, isopropanol, and ethylene glycol. Their concentrations, per mg/dL, should contribute 0.31, 0.17, and 0.16 mosmol/Kg to the osmol gap, respectively. Given a reasonable estimate of the normal osmol gap, you should be able to estimate the alcohol concentration, right?

Not so fast.

If the osmol gap can reliably predict alcohol concentrations, then it ought to be able to do that for ethanol, the one alcohol we routinely measure. Ethanol, with a molecular weight of 46, should contribute 0.22 mosmol/Kg to the plasma osmolality per mg/dL ethanol concentration. But linear regression analyses of osmol gap versus ethanol concentration consistently produce a higher conversion factor. Ethanol concentration underestimates the elevation in the osmol gap most likely because ethanol metabolites—acetaldehyde and acetate—contribute to osmolality, as well. All of the alcohols undergo metabolic oxidation to toxic products.(1)

Surrogate measurements can be a useful and economical alternative to direct assay of clinically relevant analytes, but their limitations should be recognized.


 1. Baselt, RC. Disposition of Toxic Drugs and Chemicals in Man. Seventh Edition. 2004. Biomedical Publications, Foster City, CA.