Calculated serum osmolality ≈ (2 × Na) + (glucose ÷ 18) + (BUN ÷ 2.8) Osmolar gap = measured osmolality − calculated osmolality Effective tonicity ≈ (2 × Na) + (glucose ÷ 18)

Urea contributes to measured osmolality but crosses cell membranes sufficiently readily that it is not a major sustained determinant of transcellular water movement. Effective tonicity is therefore approximated by two times sodium plus glucose divided by eighteen. Once exogenous mannitol is present, however, mannitol itself is an effective extracellular osmole and has to be conceptually added back into that picture. This is why two patients can both have measured serum osmolalities of 325 milliosmoles per kilogram and have completely different physiology. A uremic patient may have a high measured osmolality with much less effective tonicity, whereas a patient whose osmolality is 325 because of mannitol or hypernatremia has a much larger effective osmotic gradient.

This also explains why the osmolar gap is particularly useful with mannitol but much less interesting with hypertonic saline. Sodium delivered with hypertonic saline is already incorporated into the calculated osmolality through the two-times-sodium term. Mannitol is an unmeasured osmole in the standard calculation, so as circulating mannitol accumulates, the osmolar gap rises. Conceptually, the gap is therefore a rough bedside surrogate for circulating mannitol concentration rather than simply another arbitrary number to keep below a threshold.

Both mannitol and hypertonic saline depend substantially on exclusion of the administered osmole from brain tissue. Water crosses the blood–brain barrier readily; sodium and mannitol are substantially reflected by an intact barrier. Raising extracellular tonicity therefore creates an osmotic pressure gradient favoring water movement from brain toward blood. But notice the crucial limitation: as blood–brain barrier integrity deteriorates, the effective reflection of the osmole falls. Osmotherapy becomes less efficient exactly where severe structural injury may be worst. Much of its volume-reducing effect may consequently occur in relatively preserved brain surrounding the lesion rather than magically extracting water from every injured cell.

Mannitol has a second physiologic effect that occurs before substantial whole-brain water movement could plausibly occur. Its initial intravascular osmotic effect expands plasma volume and decreases blood viscosity. With functioning autoregulation, the decrease in viscosity can be accompanied by compensatory vasoconstriction that decreases cerebral blood volume and ICP while preserving flow. Then the renal effect takes over. Mannitol is freely filtered and poorly reabsorbed, producing osmotic diuresis. That latter feature is both useful and dangerous. A patient who is volume overloaded with preserved kidney function may tolerate it beautifully. A patient with MAP 68, ICP 28, and therefore a CPP around 40 does not have much reserve for several liters of subsequent urine output. Lowering ICP from 28 to 18 while simultaneously dropping MAP from 68 to 58 has not meaningfully rescued CPP.

Hypertonic saline attacks that situation from the opposite direction. It raises extracellular tonicity but also expands the intravascular compartment rather than producing the same obligatory osmotic diuresis. Consequently, it can simultaneously lower ICP and support MAP. Because CPP is MAP minus ICP, hypertonic saline can improve both sides of the equation. This is the most compelling physiologic reason that many neuro-ICUs favor hypertonic saline when CPP is tenuous. It is not because randomized trials have established that saline improves neurological outcome. They have not.

The current September 2024 ENLS 6.0 intracranial-hypertension protocol describes mannitol and hypertonic saline as having equivalent efficacy for lowering ICP. Its mannitol dose is 0.5 to 1 gram per kilogram intravenously over roughly 5 to 15 minutes, with repeat dosing potentially every four to six hours when physiology and laboratory monitoring justify it. The companion ENLS 6.0 pharmacotherapy protocol lists 3% saline at 5 milliliters per kilogram, 5% at 3 milliliters per kilogram, 7.5% at 2 milliliters per kilogram, and 23.4% saline as a 30-milliliter dose. For concentrations above 3%, ENLS recommends central administration. Individual institutional rescue protocols vary, particularly in how rapidly 23.4% saline is given, so those local protocols still matter.

At the bedside, I would think of agent selection primarily as a hemodynamic, renal, and electrolyte decision. Low MAP, hypovolemia, or threatened CPP pushes strongly toward hypertonic saline. Significant preexisting hypernatremia or a problematic chloride burden makes mannitol more attractive if renal function and intravascular volume permit it. Renal failure, progressive oliguria, or anuria makes mannitol particularly unattractive because the drug cannot be cleared normally and can accumulate. Conversely, significant heart failure or pulmonary edema makes large-volume hypertonic saline problematic, although mannitol is not automatically safe either because its initial osmotic effect also expands plasma volume and accumulated mannitol in renal failure can worsen volume overload. This is why “heart failure equals mannitol” and “renal failure equals saline” are useful tendencies, not absolute rules.

The disease-specific recommendations are similarly nuanced. The still-current Neurocritical Care Society cerebral-edema guideline conditionally favors hypertonic sodium solutions over mannitol for initial ICP or edema management in traumatic brain injury and also favors hypertonic therapy in intracerebral hemorrhage, although the evidence is low or very low quality. In acute ischemic stroke it regards either agent as reasonable initially and specifically cautions against prophylactic scheduled mannitol. In subarachnoid hemorrhage it favors symptom-driven hypertonic boluses rather than simply chasing a predetermined sodium concentration. Contrast that with the current Brain Trauma Foundation severe-TBI guideline, which concludes that evidence is insufficient to recommend a specific hyperosmolar agent based on clinical outcomes. Those positions are not really contradictory. NCS is making a conditional bedside choice partly from ICP, CPP, and safety physiology; BTF is saying that those physiologic advantages have not demonstrated superiority in patient-centered outcomes.

Monitoring mannitol is where one of the most persistent neuro-ICU myths deserves to be discarded. A measured serum osmolality of 320 milliosmoles per kilogram is not an evidence-based brick wall. NCS specifically recommends following osmolar gap preferentially over a simple serum-osmolality threshold because the gap tracks mannitol accumulation more directly. More importantly, NCS concluded that there is insufficient evidence for any particular osmolar-gap cutoff. The often-used gap of 20 has not been validated as a toxicity threshold. ENLS 6.0 currently says clinicians may use an osmolar-gap range of approximately 20 to 55, while noting that additional therapeutic benefit near the upper end is unlikely.

Mannitol-associated AKI appears related to several interacting mechanisms: intravascular depletion from osmotic diuresis, extremely high circulating mannitol concentrations, renal hemodynamic effects, and osmotic tubular injury. The key warning pattern is therefore not merely “osmolality above 320.” It is declining clearance, rising osmolar gap, worsening creatinine, falling urine output, and unfavorable volume physiology. Giving more mannitol because the ICP has rebounded while the kidney is no longer clearing the previous dose is exactly how a useful rescue drug turns into an accumulating osmotic toxin.

Hypertonic saline requires a different surveillance strategy. ENLS recommends checking sodium approximately every four to six hours during active therapy and keeping sodium below 160 mEq/L. NCS suggests that avoiding severe hypernatremia and hyperchloremia is prudent and proposes an upper sodium range around 155 to 160 and chloride around 110 to 115 as reasonable ranges associated with less AKI. But the evidence behind those numbers is low to very low quality. They are safety warning zones, not therapeutic targets and not absolute toxicity thresholds. In a rapidly changing patient receiving repeated concentrated boluses, every four to six hours may also be too slow; NCS explicitly allows monitoring frequencies ranging all the way to every two hours according to the rate of change.

Chloride deserves more attention than it historically received. Hypertonic NaCl delivers enormous sodium and chloride loads together. Because normal plasma contains far more sodium than chloride, adding them in equal proportions narrows the plasma strong-ion difference, lowering bicarbonate and generating hyperchloremic metabolic acidosis. Observational neurocritical-care data associate higher chloride exposure, particularly during continuous hypertonic saline therapy, with AKI. That association does not establish chloride as the sole causal mechanism, but it is enough that rising chloride, falling bicarbonate, worsening creatinine, and a diminishing neurological benefit should make you reconsider repeated chloride-heavy therapy. This is also why some centers use mixed chloride/acetate hypertonic formulations when cumulative chloride becomes problematic, although comparative outcome evidence for that strategy remains limited.

Continuous 3% saline deserves the same skepticism as scheduled mannitol. Inducing a sodium of 150 to 155 because the patient currently has an ICP crisis is one thing. Maintaining every patient with cerebral edema at 150 to 155 prophylactically because “high sodium prevents swelling” is a different intervention with a much weaker evidence base. NCS found insufficient evidence that continuous infusion targeted to a sodium goal improves neurological outcomes. The physiologic endpoint should remain ICP, clinical evidence of tissue shift or herniation, and the underlying disease—not achievement of a pretty laboratory number.

This leads directly to osmotic demyelination, because the risk is often misunderstood in the opposite direction. Hypernatremia itself is not synonymous with osmotic demyelination syndrome. ODS is predominantly a problem of rapidly raising tonicity in a brain that has already adapted to chronic hypotonic hyponatremia. If a previously normonatremic patient with TBI goes from sodium 140 to 150 during rescue hypertonic therapy, ODS is not the major neurologic concern. If the patient arrived chronically at sodium 112, however, the calculation changes completely.

The American expert-panel recommendations limit correction to 8 mEq/L in any 24 hours for patients at high risk for ODS and 10 to 12 in 24 hours, with 18 in 48 hours, for average-risk patients; European guidance uses 10 in the first 24 hours and 8 during subsequent 24-hour periods. Particularly high-risk features include extremely low starting sodium, hypokalemia, alcohol use disorder, malnutrition, and advanced liver disease. American expert-panel guidance and European Society of Endocrinology guidance remain the relevant frameworks.

Rebound edema is almost the mirror-image concept. With repeated mannitol exposure and a severely disrupted blood–brain barrier, mannitol can enter injured brain tissue. When circulating mannitol is subsequently cleared, the original outward osmotic gradient disappears and can theoretically reverse, favoring water movement back into tissue. That is the classic mechanistic explanation for mannitol-associated rebound. Hypertonic saline is not absolutely immune to rebound physiology either. During prolonged induced hypernatremia the brain adapts to the hyperosmolar environment; abruptly allowing serum sodium and tonicity to fall while substantial edema remains can favor renewed brain water accumulation. This is why prolonged hyperosmolar regimens are generally de-escalated thoughtfully rather than simply turning a sodium of 155 into 140 overnight. But a rise in ICP when a drug wears off does not prove “rebound.” Often the underlying edema is simply still evolving and the temporary treatment effect has disappeared.

The living evidence layer has refined rather than overturned this physiology. A 2024 updated meta-analysis of 15 randomized trials involving 624 patients with TBI found no significant difference between hypertonic saline and mannitol in mortality, neurological outcome, successful ICP treatment, magnitude of ICP reduction, or treatment failure. Hypertonic saline did produce an approximately 5.5-mm-Hg greater CPP at 30 to 60 minutes. That is almost exactly what the physiology predicts: saline’s volume-expanding effect can improve CPP, yet improving an intermediate physiologic variable has not translated into demonstrated neurological-outcome superiority. The 2024 meta-analysis is here. More recent pediatric evidence remains concordant rather than disruptive. A 2025 prospective multicenter cohort of 445 children found no significant mortality or functional-outcome advantage for 3% saline over 20% mannitol, and a 2026 pediatric meta-analysis likewise found broadly comparable mortality and ICP efficacy. Those pediatric data should not be extrapolated to determine adult practice, but they reinforce how difficult it has been to convert better-looking ICP physiology into better patient-centered outcomes.

That last question is the practical failure test. If a properly delivered hyperosmolar bolus produces little or transient improvement, repeatedly escalating osmolar exposure is not definitive treatment. Failure may mean that the blood–brain barrier is profoundly disrupted, that cerebral compliance is exhausted, that hydrocephalus needs CSF diversion, that a hematoma or contusion requires evacuation, or that malignant hemispheric swelling requires decompression.

The clean mental model is therefore this: choose hypertonic saline when maintaining intravascular volume, MAP, and CPP is particularly important; choose mannitol when avoiding additional sodium or chloride is advantageous and renal and volume physiology permit it; never select either because of an assumed neurological-outcome advantage; monitor the osmolar gap and kidney/volume physiology rather than worshiping 320 with mannitol; monitor sodium, chloride, acid–base status, renal function, and volume with hypertonic saline; distinguish ODS from rebound because they represent essentially opposite osmotic problems; and judge every dose by its cerebral physiologic response rather than the laboratory value it produces.