Why Osms… aren’t so awesome

Author: Justin Corcoran

Internal Reviewer: Annie Arens

Osms… great for a blog title, not so great for managing patients.

Set the stage: you’re a shiny new PGY-1, and you get your first (maybe) toxic alcohol poisoned patient. The patient states that they drank some antifreeze, but aren’t sure how much or what the formulation was (could be ethylene glycol or propylene glycol). Having just come off studying in medical school, you know what you need to do! Get a serum osmolality, sodium, glucose, ethanol, and BUN at the same time, and calculate an osmolar gap:

Osm gap = Osm measured -Osm calculated

Osm calculated = 2 * [Na] + [BUN]/2.8 + [Glucose]/18 + [Ethanol]/3.7¹ *

*note – this is just one of many formulas, some use 4.6 for the ethanol correction, which we’ll discuss later

You tell your attending what you’ve ordered… and they cancel your serum osmolality, and they order a dose of fomepizole and direct methanol/ethylene glycol concentrations (despite the fact that it’s going to take 6-12 hours to get those levels back at your facility).

What happened?

First, let’s start with some basics.

What the heck is an osmole?

Simply put, an osmole is 1 mole of fully dissociated particle in solution.

Why don’t we just use regular moles? It turns out that sometimes we care more about the total number of particles in solution than the number of moles of any one species. Usually this occurs when we are using colligative properties of chemistry: freezing point depression, boiling point elevation, lowering of vapor pressure, or osmotic pressure – all of which depend on the concentration of all particles in solution rather than the molar concentration of the solution. What’s nice is that for these properties, one particle is the same is the next; it’s the number that matters rather than the specific chemical species.

For an example, if you were to calculate the freezing point depression of a 100 mM NaCl solution, you would need to figure out how many osmotically active discrete particles there are – in this case, it’d be twice the molarity (because NaCl dissociates into one Na+ and one Cl- in solution, each of which are osmotically active) – the osmolarity then would be 200 mOsm/L, which you would use to calculate your expected freezing point depression.

So…what is serum osmolality? First of all, osmolarity vs osmolality…a favorite distinction for argument between nerds for decades.

• Serum osmolarity: the concentration of a solution expressed as osmoles of solute per liter of solution (osmol/L)
• Serum osmolality: the concentration of a solution expressed as osmoles of solute particles per kg of solvent (mOsm/kg)

Technically speaking, when you are calculating your osm gap you subtract the calculated osmolarity from the measured osmolality.²

When you order a “serum osm” from your lab, you get back the measured serum osmolality – which is essentially giving you an idea of the number of osmotically active particles per kg of your solvent. In the normal human body, glucose, urea, and sodium account for almost all of the osmotically active particles in serum.³ Thus, when you add their molar concentrations you get a pretty close estimate of osmolality. An osm gap represents other unmeasured osmotically active compounds like: ethanol, methanol, ethylene glycol, acetone, paraldehyde, hyperlipidemia, or hyperproteinemia, just to name a few.³

There are a few reasons why the osmolar gap just sucks at being clinically useful in predicting toxic alcohol ingestion.

First, the osmolar gap depends on the time course. As the alcohol is metabolized, it becomes first an aldehyde and then a carboxylic acid – these are accounted for in the anion gap, and thus stop showing up in the osmolar gap. This leads to an inverted relationship between the osmolar gap and the anion gap over time, where the osmolar gap falls as the anion gap increases.

If you measure the osmolar gap early (green line) – you might find a large osmolar gap and a small or normal anion gap. This would potentially be helpful information. However, later in the course, once the toxic alcohol has been metabolized (purple line), you will have a low osmolar gap and high anion gap. You may conclude that the patient doesn’t have poisoning with a toxic alcohol, when in fact this is the most dangerous time for the patient – the cat is already out of the bag, so to speak – they’ve already metabolized a lot of the relatively non-toxic parent compound to the toxic metabolites (the carboxylic acids that actually cause the end-organ toxicity).

Now you might be able to get around this by measuring the osmolar gap multiple times – a falling osmolar gap with a rising anion gap is concerning for toxic alcohol poisoning. Sometimes we consider doing for toxic alcohols that we can’t test directly, like diethylene glycol, or if a patient is not in an area that is able to test for toxic alcohols in a timely fashion. If you find yourself in this situation… call your poison center (or toxicologist)!

Second, the osmolar gap varies within the population, and that variation is… not well defined. A common range cited is -14 to +10 mOsm/L.² However, depending on the method by which you calculate the osmolar gap, the mean gap changes. Let that sink in – the reference range for the test (osmolar gap) is going to depend on which equation you used to get there.

Third… even just calculating the osmolar gap is hard. Which equation should you use?

Osm calculated = 2 × [Sodium] + (1.15 * [Glucose]/18) + ([BUN]/2.8) + (1.2 * [ETOH]/4.6) ⁴ 

Or

Osm calculated = 2 * [Na] + [BUN]/2.8 + [Glucose]/18 + [Ethanol]/3.7 ¹

Or

Osm calculated = 1.86 * [Na] + [BUN]/2.8 + [Glucose]/18 + [Ethanol]/4.6 ²

Or

Any of a multitude of other equations ²

Even once you get it nailed down which equation you’re going to use, and which set of normal values you’re going to reference, these are population values – difficult to use for a single patient.

Imagine a patient who shows up with a possible toxic alcohol ingestion, and you calculate their osmolar gap to be +10 mOsm/L. If this particular patient’s normal osmolar gap was -2 mOsm/L (the mean value from the aforementioned Hoffman et al (1993)) then the actual change of +12 mOsm/L would correlate to a serum ethylene glycol concentration of:

12 mOsm/L * 62.07mg/mmol EtGlycol = 744.84 mg/L EtGlycol

In more familiar units, that’s 74 mg/dL of ethylene glycol – well over the threshold at which we’d treat with fomepizole. Even methanol, being approximately half the molecular mass of ethylene glycol, would produce a mg/dL concentration in the treatable range.

What about using the osmolar gap to predict how much of the toxin is around, rather than ruling in/out disease? Even that doesn’t perform well – Greene and Krasowski (2019) looked at how well the osmolar gap correlated with actual measured concentrations of toxic alcohol. See an example below for ethylene glycol:

Greene & Krasowski (2019) ⁵

Although there was a linear relationship between the expected concentration calculated from the osmolar gap and the actual real-life concentration, the calculated value tended to systematically over-estimate the actual value.

So what do you take away from all of this?

1. The ability of your osm gap to suggest a toxic alcohol ingestion depends on how long ago your patient drank his antifreeze. Are you feeling lucky?
2. Your normal range for osm gaps varies WILDLY between patients, AND is dependent on which equation you use! Your osm gap of 10 mOsm/L may be normal, or reflect a treatable toxic alcohol concentration.
3. Even if you use your super smart brain to calculate an estimated toxic alcohol concentration from your osm gap, you’re probably still wrong.

1. Purssell RA, Pudek M, Brubacher J, Abu-Laban RB. Derivation and validation of a formula to calculate the contribution of ethanol to the osmolal gap. Ann Emerg Med 2001;38(6):653–9.

2. Hoffman RS, Smilkstein MJ, Rowland MA, Goldfrank LR. Osmol gaps revisited: Normal values and limitations. J Toxicol Clin Toxicol 1993;31(1):81–93.

3. Worthley LI, Guerin M, Pain RW. For calculating osmolality, the simplest formula is the best. Anaesth Intensive Care. 1987 May;15(2):199-202.

4. Khajuria A, Krahn J. Osmolality revisited—Deriving and validating the best formula for calculated osmolality. Clin Biochem 2005;38(6):514–9.

5. Greene HR, Krasowski MD. Correlation of osmolal gap with measured concentrations of acetone, ethylene glycol, isopropanol, methanol, and propylene glycol in patients at an academic medical center. Toxicol Rep 2019;7:81–8.