| Postoperative Complications | Respiratory Care| Cardiovascular Therapy| Fluid Management| Nutrition in Critical Patients |Traumatic Brain Injury-Stroke-Brain Death |

The fluid management in patients who have central nervous system (CNS) pathology presents special challenges for anesthesiologists and intensivists. These patients often receive diuretics (e.g., mannitol, furosemide) to treat cerebral edema and to reduce intracranial hypertension. At the same time, neurosurgical patients may require large volumes of either intravenous fluid or blood as part of an initial resuscitation, treatment of cerebral vasospasm, correction of preoperative dehydration, or maintenance of intraoperative and postoperative hemodynamic stability.
Fluid restriction had been a cornerstone of the treatment of patients with CNS pathology for many years. This practice arose from the concern that the administration of fluid might exacerbate cerebral edema and intracranial hypertension. Not only has the efficacy of fluid restriction remained unproven, even after all this time, but also the consequences of fluid restriction, if pursued to the point of hypovolemia, can be devastating. A recent study of head trauma patients demonstrated that negative fluid balance (<594 mL) was associated with an adverse effect on outcome, independent of its relationship to intracranial pressure (ICP), mean arterial pressure, or cerebral perfusion pressure.
Although few human data exist concerning the impact of exogenous fluids on the injured brain to guide rational fluid management in the neurosurgical patient, it is possible to examine the factors that influence water movement into the brain and make some reasonable recommendations.

I. Osmolality/osmolarity, colloid oncotic pressure (COP), crystalloid, and colloid

Osmotic pressure. This is the hydrostatic force acting to equalize the concentration of water (H2O) on both sides of a membrane that is impermeable to substances dissolved in that water. Water moves along its concentration gradient. This means that if a solution containing 10 mosmol of sodium (Na) and 10 mosmol of chloride (Cl) are placed on one side of a semipermeable membrane with H₂O on the other, water will move "toward"� the saline (NaCl) solution. The saline solution has a concentration of 20 mosmol/L, and the force driving water will be approximately 19.3 mm Hg/mosmol, or 386 mm Hg. Note that the driving force is proportional to the gradient across the membrane; if two solutions of equal concentration are placed across a membrane, there is no driving force. Similarly, if the membrane is permeable to the solutes (e.g., Na and Cl), this reduces the gradient and hence the osmotic forces.
Osmolarity and osmolality. Osmolarity describes the molar number of osmotically active particles per liter of solution. In practice, this value is typically calculated by adding up the milliequivalent (mEq) concentrations of the various ions in the solution. Osmolality describes the molar number of osmotically active particles per kilogram of solvent. This value is directly measured by determining either the freezing point or the vapor pressure of the solution (each of which is reduced by a dissolved solute). Note that osmotic activity of a solution demands that particles be "independent." As NaCl dissociates into Na and Cl, it creates two osmotically active particles. If electrostatic forces act to prevent dissociation of the two charged particles, osmolality is reduced. For most dilute salt solutions, osmolality is equal to or slightly less than osmolarity. For example, commercial lactated Ringer's solution has a calculated osmolarity of approximately 275 mosmol/L but a measured osmolality of approximately 254 mosmol/kg, indicating incomplete dissociation.
Calculated versus measured serum osmolality is relevant in the clinical setting. The reference method for measuring serum osmolality is the delta-cryoscopic technique. However, the technology may not be available, and/or it is not always possible to obtain an emergency measurement 24 hours a day. In such cases, osmolality can be calculated from the osmoles that are routinely measured, such as sodium, potassium, urea, and glucose. Be advised, however, that calculation of osmolality introduces a bias, overestimating osmolality in the lower ranges and underestimating it in the higher ranges.
COP. Osmolarity and osmolality are determined by the total number of dissolved "particles"� in a solution regardless of their size. COP is the osmotic pressure produced by large molecules (e.g., albumin, hetastarch, dextran). This factor becomes particularly important in biological systems in which vascular membranes are often permeable to small ions but not to large molecules (typically plasma proteins). In such situations, proteins might be the only osmotically active particles. Normal COP is approximately 20 mm Hg (or equal to approximately 1 mosmol/kg).
Starling's hypothesis. In 1898, Starling published his equations describing the forces driving water across vascular membranes. The major factors that control the movement of fluids between the intravascular and extravascular spaces are the transcapillary hydrostatic gradient, the osmotic and oncotic gradients, and the relative permeability of the capillary membranes that separate these spaces. The Starling equation is as follows:
FM = k (Pc + π - Pi - πc)
where FM is fluid movement, k is the filtration coefficient of the capillary wall (i.e., how leaky it is), Pc is the hydrostatic pressure in the capillaries, Pi is the hydrostatic pressure (usually negative) in the interstitial (extravascular) space, and πi and πc are interstitial and capillary osmotic pressures, respectively.
Fluid movement is therefore proportional to the hydrostatic pressure gradient minus the osmotic pressure gradient across a vessel wall. The magnitude of the osmotic gradient depends on the relative permeability of the vessels to solute. In the periphery (muscle, bowel, lung, etc.), the capillary endothelium has a pore size of 65 Å and is freely permeable to small molecules and ions (Na, Cl) but not to large molecules, such as proteins (Fig.-1). As a result, π is defined only by colloids, and the Starling equation can be simplified by saying that fluid moves into a tissue whenever either the hydrostatic gradient increases (either intravascular pressure rises or interstitial pressure falls) or the osmotic gradient decreases.
In normal situations, the intravascular protein concentration is higher than the interstitial concentration, acting to draw water back into the vascular space. If COP is reduced (e.g., by dilution with large amounts of isotonic crystalloid), fluid begins to accumulate in the interstitium, producing edema. This fact is familiar to all anesthesiologists who have seen marked peripheral edema in patients given many liters of crystalloid during surgery or resuscitation. By contrast, the blood-brain barrier (BBB) is impermeable to both ions and proteins so that osmotic pressure is determined by the total osmotic gradient, of which COP contributes only a tiny fraction.

Figure-1 Schematic diagram of a peripheral capillary.

The vessel wall is permeable to both water (H2O) and small ions but not to proteins (P)

Interstitial clearance. Peripheral tissues have a net outward movement of fluid (i.e., the value of FM is positive). Edema is not normally present, however, because this extravasated fluid is cleared by the lymphatics. While many researchers agree that there is some lymphatic drainage of the brain, most interstitial fluid in the brain is cleared either by bulk fluid flow into the cerebrospinal fluid (CSF) spaces or via pinocytosis back into the intravascular compartment. This is a slow process and probably does not counteract rapid fluid movement into the interstitial space.
Hydrostatic forces and interstitial compliance. In the tissues, the net hydrostatic gradient is determined by (a) intravascular pressure and (b) interstitial tissue compliance. Normally, the direction is outward (capillary to interstitium). There is no question that in the brain (or in any organ), elevated intravascular pressure, such as that produced by either high jugular venous pressure or a head-down posture, can increase edema formation. However, an often overlooked factor that influences the pressure gradient is the interstitial compliance (i.e., the tendency of tissue to resist fluid influx). The loose interstitial space in most peripheral tissues does little to impede the influx of fluid. This explains the ease with which edema develops around, for example, the face and the eyes, even with minor hydrostatic stresses (e.g., a facedown posture).
By contrast, the interstitial space of the brain is extremely noncompliant, resisting fluid movement. As a result, minor changes in driving forces (either hydrostatic or osmotic/oncotic) do not produce measurable edema. However, a vicious cycle can develop so that, as edema forms in the brain, the interstitial matrix is disrupted, the compliance increases, and additional edema forms more easily. In contrast, the closed cranium and ICP can act to retard fluid influx. This may partially explain the exacerbation of edema formation that can occur after rapid decompression of the intracranial space.
Can we explain the influence of certain fluids on the brain? In contrast to capillaries elsewhere in the body, the endothelial cells in the brain are held together by continuous tight junctions to form the BBB. There are no intercellular gaps: the membranes are not fenestrated and do not have channels or chains of vesicles that form transendothelial pathways. The effective pore size of the BBB is only 7 to 9 Ã, making this unique structure normally impermeable to large molecules (e.g., plasma proteins and synthetic colloids, such as hetastarch and dextrans) and relatively impermeable to many small polar solutes (Na, K, Cl) (Fig.-2). The BBB functions as a semipermeable membrane that allows only water to move freely between the brain's interstitial space and the vasculature. This should serve to make the brain an exquisitely sensitive osmometer where water moves according to osmotic gradients.

Figure-2. Schematic diagram of a cerebral capillary.

The blood-brain barrier (BBB) is impermeable to small ions and proteins (P) but not to water (H2O)

We would then predict that reducing serum osmolality (e.g., by infusing water or large volumes of nonisotonic crystalloid solution) increases cerebral edema; conversely, increasing osmolality reduces brain water content. This experiment was first done in 1919 by Weed and McKibben, who showed a large and rapid increase in brain volume with the reduction of serum osmolality. Since that time, numerous studies have demonstrated the exquisite sensitivity of the brain's water content to changes in serum osmolality, in which even small changes can produce measurable changes in the brain water content. This makes sense considering that a 5 mosmol/kg gradient is equivalent to an almost 100 mm Hg force driving water (see preceding text). In fact, in experimental animals, a reduction in plasma osmolality of as little as 5% under otherwise normal conditions causes brain edema and increases ICP.
What about changes in COP? As mentioned, colloidally active molecules contribute to only a tiny fraction of the total osmolality and, when the BBB is intact, can be responsible for only a small driving force. Normal plasma COP is approximately 20 mm Hg, whereas that in the brain interstitium is approximately 0.6 mm Hg. This is equal to the force that could be generated by a change in the capillary/tissue osmotic gradient of only 1 mosmol/kg. We would therefore predict that changes in COP have only minimal effect on brain water content. This hypothesis has been tested directly in several animal experiments, which have demonstrated that normal brain water content can be altered by small changes in osmolality but not by clinically achievable changes in COP. There was also no increase in brain water content (i.e., edema) in any region with an intact BBB, even after a very large reduction (approximately 50%) in COP.
However, what about more clinically relevant conditions, with varying degrees of BBB abnormality? We can predict that if the BBB is made completely permeable to both small and large molecules (i.e., complete breakdown of the BBB as is common with several experimental and clinical injuries), it should be impossible to maintain any form of osmotic or oncotic gradient between the intravascular compartment and the brain interstitium. As a result, no changes in brain water content would be expected with a change in either gradient. Indeed, in several animal models resembling human brain injuries (e.g., implanted glioma and freezing lesion, which are similar to brain tumor and trauma, respectively), changes induced by a reduction in total serum osmolality were seen only in apparently normal regions of the brain relatively distant from the focus of injury. In keeping with this, several studies have shown that acute hyperosmolality (as with mannitol, urea, glycerol, hypertonic saline [HS]) reduces water content only in relatively normal brain tissue where the BBB is intact. Conversely, when the BBB is severely damaged, investigations have failed to demonstrate that reducing the COP affects cerebral edema. In the presence of mild injury to the BBB in experimental animals, a reduction in COP may potentially aggravate brain edema. Therefore, it is possible that, with a less severe injury, the BBB may function similarly to the peripheral tissue.
Summary. Injury to the brain interferes with the integrity of the BBB to varying degrees, depending on the severity of the damage. Regions in which there is a complete breakdown of the BBB, there will be no osmotic/oncotic gradient. Water accumulation (i.e., brain edema) occurs because of the pathologic process and cannot be directly influenced by the osmotic/oncotic gradient. In other regions where there is moderate injury to the BBB (i.e., a mild opening rendering pore size similar to the periphery), it is possible that the colloid oncotic gradient is effective as in the peripheral tissue. Finally, the BBB is normal in a significant portion of the brain. The presence of a functionally intact BBB is essential if osmotherapy is to be effective.
Fluids for intravenous administration. Table-1 lists a variety of solutions suitable for intravenous use.

Crystalloid is the term commonly applied to solutions that do not contain any high molecular weight compound and have an oncotic pressure of zero. Crystalloids can be hyposmolar, iso-osmolar, or hyperosmolar, and may or may not contain glucose. Crystalloids can be made hyperosmolar by the inclusion of electrolytes (e.g., Na and Cl, as in HS) and low molecular weight solutes such as mannitol (molecular weight 182) or glucose (molecular weight 180).
Colloid is the term used to denote solutions that have an oncotic pressure similar to that of plasma. Some commonly administered colloids include 5% or 25% albumin, plasma, 6% hetastarch (hydroxyethyl starch, molecular weight 450), pentastarch (a low molecular weight hydroxyethyl starch, molecular weight 264), and the dextrans (molecular weights 40 and 70). Dextran and hetastarch are dissolved in normal saline so that the osmolarity of the solution is approximately 290 to 310 mosmol/L with a sodium and chloride content of approximately 154 mEq/L, which makes them hyperoncotic and hypertonic. Hetastarch should be limited and used with caution, however, because of the depletion of factor VIII and possible coagulation difficulties encountered with volumes exceeding 1,500 mL. Dextran-40 interferes with normal platelet function and is therefore not advisable for patients who have intracranial pathology other than to improve rheology as in ischemic cerebrovascular diseases. Although small volumes of hetastarch and dextran can restore normovolemia rapidly and without increasing ICP, it is unclear whether they offer any advantage over the combination of isotonic fluids and osmotic diuretics.

II. Clinical fluid management of neurosurgical patients

Fluid restriction. Despite the lack of convincing experimental evidence that iso-osmolar crystalloids are detrimental, fluid restriction is widely practiced in patients who have mass lesions or cerebral edema or are at risk for intracranial hypertension. The only directly applicable data indicate that clinically acceptable restriction has little effect on edema formation. However, "modest"� fluid restriction has some logic. One of the few human studies on fluid therapy in neurosurgical patients has demonstrated that patients given standard "maintenance"� amounts of intravenous fluid (e.g., 2,000 mL/day of 0.45 normal saline in 5% dextrose) in the postoperative period develop a progressive reduction in serum osmolality (Fig.-3). On the other hand, patients given half of this volume over a period of several days (about a week) show a progressive increase in serum osmolality, which could account for dehydration of the brain (Fig. 20-3). While no CNS parameters were measured, the results suggest that usual maintenance fluids contain excess free water for the typical postoperative craniotomy patient. In this regard, fluid restriction can be viewed as "preventing"� hyposmotically driven edema. However, this does not imply that even greater degrees of fluid restriction are beneficial or that the administration of a fluid mixture that does not reduce osmolality is detrimental.

Intraoperative volume replacement/resuscitation. As a general rule, intraoperative fluids should be given at a rate sufficient to replace the urinary output and insensible losses (e.g., skin and lungs). Table-2 illustrates the intravascular volume expansion obtained with different types of fluids. Volume replacement with crystalloid solutions, geared to maintain the hematocrit at approximately 33%, is calculated on a 3:1 ratio (crystalloid to intraoperative blood loss) because of the larger distribution space of the crystalloids.

The available data indicate that intravascular volume replacement and expansion will have no effect on cerebral edema as long as normal serum osmolality is maintained and cerebral hydrostatic pressure is not markedly increased (e.g., owing to true volume overload and elevated right heart pressures). Whether this is achieved with crystalloid or colloid seems irrelevant, although the osmolality of the selected fluid is crucial. With respect to this issue, it should also be noted that lactated Ringer's solution is not strictly isosmotic (measured osmolality 252 to 255 mosmol/kg), particularly when administered to patients whose baseline osmolality has been increased by either fluid restriction or hyperosmolar fluids (mannitol, HS, etc.).
It is recommended that serum osmolality be checked repeatedly with the goal of either maintaining this value or increasing it slightly. Fluid administration that results in a reduction in osmolality should be avoided. Small volumes of lactated Ringer's (1 to 3 L) are unlikely to be detrimental and can be used safely, for example, to compensate for the changes in venous capacitance that typically accompany the induction of anesthesia. If large volumes are needed (either to replace blood loss or to compensate for some other source of volume loss), a change to a more isotonic fluid such as normal saline is probably advisable. It is also important to remember that rapid infusion of a large volume of 0.9% NaCl can induce a dose-dependent hyperchloremic metabolic acidosis. Whether this acid-base abnormality is, in fact, harmful remains unclear, although animal studies suggest that hyperchloremia causes renal vasoconstriction. If large volumes are needed, a combination of isotonic crystalloids and colloids may be the best choice (Table 20-1).
These recommendations should not be interpreted, however, as a license to "give all the isotonic fluid you like."� Volume overload can have detrimental effects on ICP by increasing either cerebral blood volume (CBV) or hydrostatically driven cerebral edema formation.
Postoperative period. In the postoperative period, the patient no longer needs large volumes of fluid. The recommendation of Shenkin to give approximately 1,000 mL/day is probably reasonable (Fig. 20-3); we would again recommend periodic measurement of serum osmolality, particularly if the patient's neurologic status deteriorates. If cerebral edema does develop, further fluid restriction is unlikely to be of value and can cause hypovolemia. Instead, treatment consists of mannitol or furosemide and maintenance of normovolemia with fluids to sustain the increased osmolality. Inducing hypovolemia so that vasopressors are required to maintain acceptable hemodynamic parameters has little advantage (and some disadvantage).
HS solutions. HS solutions have been evaluated for use in fluid resuscitation since the 1960s, particularly because hemodynamic improvement can be achieved with very small volumes given very quickly. Because hyperosmolality is known to reduce brain volume, HS has been used in patients who are at risk for increased ICP. In humans, acute resuscitation from hemorrhagic shock with 7.5% HS is associated with improved outcome in patients with multiple trauma and head injuries. Clinical studies suggest that HS may be better in hypotensive, brain-injured patients during transportation to the hospital.
There is no question that HS can quickly restore intravascular volume while reducing ICP through brain water reduction in uninjured brain. Additional benefit may be derived from decreases in CSF production. Unfortunately, what remains unclear is whether this approach is unique or whether the identical CNS benefit could be achieved with any resuscitation fluid that increases osmolality.
The principal disadvantage of HS is the danger of hypernatremia. In a recent study of neurosurgical patients undergoing elective supratentorial procedures, we have shown that equal volumes of 20% mannitol and 7.5% HS reduce brain bulk and CSF pressure to the same extent (Fig.-3). However, serum sodium increased during the administration of HS and peaked at more than 150 mEq/L at the conclusion of the infusion.
Mannitol and furosemide. Both mannitol and furosemide (and occasionally other diuretics) are used to control ICP and brain swelling in neurosurgical patients. Mannitol acts by establishing an osmotic gradient between blood and brain in the presence of a relatively intact BBB. This promotes removal of water from areas of normal brain. Mannitol might elevate ICP transiently from the vasodilator effects of hyperosmolality, resulting in an increase in CBV. In both dogs and humans, this phenomenon occurs neither in the presence of intracranial hypertension nor when mannitol is given at moderate rates. Mannitol may, therefore, be given to most neurosurgical patients. The exception might be patients who have significant cardiovascular disease in whom the transient volume expansion might precipitate congestive heart failure.
The other important concern is that the repeated use of mannitol may cause excessive hyperosmolality, which can be deadly. In addition, mannitol accumulates in the interstitium with repeated doses. If interstitial osmolality rises excessively, it is possible for the normal brain-blood gradient to be reversed with resultant exacerbation of the edema. Furthermore, if brain osmolality is increased, edema may be enhanced by the subsequent normalization of serum osmolality. The recommended dose of mannitol is therefore 0.25 to 1 g/kg. The smallest possible dose is selected and infused over 10 to 15 minutes.
The exact mechanism of action of furosemide remains controversial, although it certainly is related to the drug's ability to block Cl transportation. Furosemide and similar drugs might also act primarily by reducing cellular swelling rather than by changing the extracellular fluid volume. Several studies have demonstrated that furosemide decreases CSF production. This effect can explain the synergistic effect of mannitol and furosemide on intracranial compliance. However, the maximal effect of furosemide compared with that of mannitol is delayed. For this reason, mannitol probably remains the drug of choice for rapid control of ICP.
Glucose-containing solutions. Salt-free solutions containing glucose should be avoided in patients who have intracranial pathology. Free water reduces serum osmolality and increases brain water content. Furthermore, solid evidence in animals and humans indicates that excessive glucose exacerbates neurologic damage and can worsen the outcome from both focal and global ischemia. This happens because glucose metabolism enhances tissue acidosis in ischemic areas. The reduction of adenosine levels from hyperglycemia could also be detrimental. Adenosine inhibits the release of excitatory amino acids, which play a major role in ischemic cell damage.
Although clinical investigations have indicated a negative relationship between patients' plasma glucose on admission and outcome after stroke, cardiac arrest, and head injury, more recent studies suggest that this correlation is not necessarily one of cause and effect because the high glucose may be a concomitant of more severe CNS damage. Nor has it been possible to demonstrate that the administration of glucose to humans is detrimental. Nevertheless, withholding glucose from adult neurosurgical patients is not associated with hypoglycemia. It may therefore be prudent to withhold glucose-containing fluids from acutely injured and elective surgical patients. This caveat does not apply to the use of hyperalimentation in such patients, perhaps because the administration of these hyperglycemic solutions typically begins several days after the primary insult.
Should insulin be administered to correct hyperglycemia in patients who have intracranial pathology? While laboratory evidence suggests that the correction of hyperglycemia with insulin before the ischemic insult occurs improves outcome, this has not been studied in humans.
Summary. Blood sugar in neurosurgical patients should be controlled carefully to avoid both hypo- and hyperglycemia and to maintain glucose between 100 and 150 mg/dL. Glucose-containing solutions should be withheld, except in the case of neonates and patients who have diabetes in whom hypoglycemia can occur very rapidly and be detrimental.

III. Hemodilution

One common accompaniment of fluid administration is a reduction in the hemoglobin and hematocrit. With active blood loss, the use of asanguineous fluids can cause marked anemia. An increase in cerebral blood flow (CBF) typically accompanies this hemodilution, and physicians have long argued whether the hemodilution is beneficial, benign, or detrimental. The answer probably depends on the degree of hemodilution and on the state of the disease.
From a theoretical vantage, a hematocrit of 30% to 33% gives the optimal combination of viscosity and oxygen-carrying capacity. In the normal brain, the increase in CBF produced by hemodilution is almost certainly an active compensatory response to a decrease in arterial oxygen content; this response is essentially identical to that seen with hypoxia. With a brain injury, however, the normal CBF response to hypoxia and hemodilution is attenuated, and both conditions can contribute to secondary tissue damage.
The one situation in which hemodilution might be beneficial is the period during and immediately after a focal cerebral ischemic event. Several animal studies have shown that regional oxygen delivery may be increased (or at least better maintained) in the face of modest hemodilution to a hematocrit of approximately 30% with improvement in CBF and reduction in infarction volume. In spite of this, several clinical trials have failed to demonstrate any benefit from hemodilution in stroke patients, except in those who were polycythemic to begin with. The lack of success, however, may reflect the delay in instituting therapy or inadequate hematocrit reduction.
What clinical lesson can be learned from the work on hemodilution? Our opinion is that for elective neurosurgical patients and patients suffering from head injuries, hemodilution to a hematocrit below 30% to 35% is unlikely to be any more “beneficial� than hypoxia. Hemodilution to 30% to 35% might be better tolerated in patients who are at risk for focal ischemia. Nevertheless, active attempts to reduce the hematocrit below 30% are probably not advisable at the present time.
Water and electrolyte disturbances. Table-3 summarizes the principal differences among the most common water and electrolyte disturbances in patients who have intracranial pathology.
Diabetes insipidus (DI). DI is a common sequela of pituitary and hypothalamic lesions, but it can also occur with other cerebral pathology, such as head trauma, bacterial meningitis, intracranial surgery, phenytoin use, and alcohol intoxication. Patients who have markedly elevated ICP and brain death also commonly develop DI.

DI is a metabolic disorder caused by the decreased secretion of antidiuretic hormone (ADH). This results in the failure of the tubular reabsorption of water. Polyuria (>30 mL/kg/hour or, in an adult, >200 mL/hour), progressive dehydration, and hypernatremia occur subsequently. DI is present when the urine output is excessive, the urine osmolality is inappropriately low relative to serum osmolality (which is above normal because of water loss), and the urine specific gravity is <1.002.
Management. The management of DI requires restoration of normal serum sodium and carefully balancing intake and output to avoid fluid overload. The patient should receive hourly maintenance fluids plus either three-fourths of the previous hour's urine output or the previous hour's urine output minus 50 mL (Table-4). Half-normal saline and free water are commonly used as replacement fluids, with appropriate potassium supplementation. Serum sodium, potassium, and glucose should be checked frequently.
If the urine output is >300 mL/hour for 2 hours, it is now standard practice to administer aqueous vasopressin, 5 to 10 international units of drug (IU), intramuscularly (i.m.) or subcutaneously (s.c.) every 6 hours or the synthetic analog of ADH, desmopressin acetate, 0.5 to 2 mcg intravenously (i.v.) every 8 hours or by nasal inhalation, 10 to 20 mcg.
Syndrome of inappropriate antidiuretic hormone secretion (SIADH). Various cerebral pathologic processes (mostly head trauma) can cause excessive release of ADH, which leads to the continued renal excretion of sodium (>20 mEq/L), despite hyponatremia and associated hypo osmolality. Urine osmolality is therefore high relative to serum osmolality. SIADH can also result from the excessive administration of free water in patients who cannot excrete free water because of excess ADH.
Management. The mainstay of treatment of SIADH is fluid restriction to 1,000 mL/24 hours of iso-osmolar solution. If hyponatremia is severe (<110 to 115 mEq/L), the administration of hypertonic (3% to 5%) saline and furosemide might be appropriate. Because rapid correction of hyponatremia has been associated with the occurrence of central pontine myelinolysis, restoring serum sodium at a rate of approximately 2 mEq/L/hour is advisable.

Cerebral salt wasting syndrome (CSW). CSW is characterized by hyponatremia, volume contraction, and high urine sodium concentration (>50 mEq/L). This syndrome is frequently seen in patients after subarachnoid hemorrhage (SAH). The causative factor seems to be the increased release of a natriuretic factor from the brain.
Management. The therapy is to reestablish normovolemia with the administration of sodium-containing solutions.
The distinction between SIADH and CSW is very important because treatment of these two syndromes is quite different: fluid restriction versus fluid infusion, respectively. It should be stressed that in patients who have SAH for whom normo- to hypervolemia is advocated, fluid restriction (i.e., further volume contraction) might be especially deleterious.

IV. Conclusion

As neuroanesthesiologists and intensivists, we should always remember that we treat the whole patient, not only the brain. Therefore, with the exception of patients with SIADH, we should abandon the old dogma that patients who have intracranial pathology must be "run dry"� and replace it with "run them isovolemic, isotonic, and iso-oncotic."

V. Key points

Movement of water between the normal brain and the intravascular space depends on osmotic gradients.
Reducing serum osmolality by administering free water or hypotonic crystalloid solutions (0.45% NaCl) results in edema formation in all tissues, including normal brain tissue.
Reduction of COP with maintenance of serum osmolality is associated with increased water content in many tissues, but not in the normal brain. Colloid solutions exert little influence on brain water content and ICP.
In the setting of brain injury, reducing serum osmolality increases edema and ICP. Therefore, the goal of fluid management in neurosurgery is to avoid the reduction of serum osmolality. Reduction of COP, with careful maintenance of osmolality, does not increase edema in the injured brain.
Hypertonic solutions: mannitol decreases brain water content in the normal brain and is commonly used to reduce ICP. HS decreases brain water content and ICP but can cause hypernatremia.
Glucose-containing solutions should not be used in patients who have brain pathology and should be avoided in patients at risk for cerebral ischemia.
Fluid restriction minimally affects cerebral edema and, if overzealously pursued, can lead to hemodynamic instability, which is detrimental in neurosurgical patients.
Isotonic crystalloid solutions are widely used to maintain and/or restore intravascular volume.