A 65-year-old woman has undergone 12 hours of an expected 16-hour radical neck dissection and flap construction. Estimated blood loss is 1,000 mL. She has received three units of packed red blood cells and 6 L of 0.9% saline. Her blood pressure and heart rate have remained stable while anesthetized with 0.5% to 1% isoflurane in 70:30 nitrous oxide and oxygen. Urinary output is adequate. Arterial blood gas levels are shown in Table 14-8.

The step-by-step interpretation is as follows:

A 35-year-old man, 3 days after appendectomy, develops nausea with recurrent emesis persisting for 48 hours. An arterial
blood gas reveals the results shown in the middle column of Table 14-9.

This sequence illustrates the important concept that the final pH, PaCO₂, and [HCO₃⁻] represent the result of all of the vectors operating on acid–base status. Complex or “triple disturbances” can only be interpreted using a thorough, stepwise approach.

Accurate replacement of fluid deficits necessitates an understanding of the expected distribution spaces of water, sodium, and colloid. The sum of intracellular volume (ICV), which constitutes 40% of total body weight, and extracellular volume, which constitutes 20% of body weight, comprises total body water (TBW), which therefore approximates 60% of total body weight. PV, equals about one-fifth of ECV, the remainder of which is interstitial fluid volume (IFV). Red cell volume, approximately 2 L, is part of ICV.

The distribution volume of sodium-free water is TBW. The distribution volume of infused sodium is ECV, which contains equal sodium concentrations ([Na⁺]) in the PV and IF. Plasma [Na⁺] is approximately 140 mEq/L. The predominant intracellular cation, potassium, has an intracellular concentration ([K⁺]) approximating 150 mEq/L. Albumin, the most important oncotically active constituent of ECV, is unequally distributed in PV (∼4 g/dL) and IFV (∼1 g/dL). The IFV concentration of albumin varies greatly among tissues; however, ECV is the distribution volume for colloid solutions.

Conventionally, clinical prediction of PV expansion after fluid infusion assumes that body fluid spaces are static. Kinetic analysis of PV expansion replaces the static assumption with a dynamic description. As an example of the static approach, assume that a 70 kg patient has suffered an acute blood loss of 2,000 mL, approximately 40% of the predicted 5 L blood volume. The formula describing the effects of replacement with 5% dextrose in water (D₅W), lactated Ringer solution, or 5% or 25% human serum albumin is as follows:

Rearranging the equation yields the following:

To restore blood volume using D₅W, assuming a distribution volume for sodium-free water of TBW, requires 28 L:

where 2 L is the desired PV increment, 42 L = TBW in a 70 kg person, and 3 L is the normal estimated PV.

To restore blood volume using lactated Ringer solution requires 9.1 L:

where 14 L = ECV in a 70 kg person.

If 5% albumin, which exerts colloid osmotic pressure similar to plasma, were infused, the infused volume initially would remain in the PV, perhaps attracting additional interstitial fluid intravascularly. Twenty-five percent human serum albumin, a concentrated colloid, expands PV by approximately 400 mL for each 100 mL infused.

However, these static analyses are simplistic. Infused fluid does not simply equilibrate throughout an assumed distribution
volume but is added to a highly regulated system that attempts to maintain intravascular, interstitial, and intracellular volume. A more comprehensive kinetic model was proposed by Svensén and Hahn.²⁰ Kinetic models of intravenous fluid therapy allow clinicians to predict more accurately the time course of volume changes produced by infusions of fluids of various compositions. Kinetic analysis permits estimation of peak volume expansion and rates of clearance of infused fluid and complements analysis of “pharmacodynamic” effects, such as changes in cardiac output or cardiac filling pressures.²¹

Using a kinetic approach to fluid therapy permits analysis of the effects of common physiologic and pharmacologic influences on fluid distribution in experimental animals or humans. For example, in chronically instrumented sheep, isoflurane anesthesia and the conscious state were associated with similar kinetics of PV expansion after fluid infusion, but reduced urinary output in anesthetized sheep demonstrated that expansion of extravascular volume was relatively greater during anesthesia; this effect was attributable to isoflurane and not to mechanical ventilation during anesthesia.²² Also in chronically instrumented sheep, administration of catecholamine infusions before and during fluid infusions profoundly altered intravascular fluid retention, with phenylephrine diminishing and isoproterenol enhancing intravascular fluid retention (Fig. 14-1).²³

TBW content is regulated by the intake and output of water. Water intake includes ingested liquids plus an average of 750 mL ingested in solid food and 350 mL that is generated metabolically. Insensible losses are normally 1 L/day and gastrointestinal losses are 100 to 150 mL/day. Thirst, the primary mechanism of controlling water intake, is triggered by an increase in body fluid tonicity or by a decrease in ECV.

Reabsorption of filtered water and sodium is enhanced by changes mediated by the hormonal factors antidiuretic hormone (ADH), atrial natriuretic peptide (ANP), and aldosterone. Renal water handling has three important components: (1) Delivery of tubular fluid to the diluting segments of the nephron, (2) separation of solute and water in the diluting segment, and (3) variable reabsorption of water in the collecting ducts. In the descending loop of Henle, water is reabsorbed while solute is retained to achieve a final osmolality of tubular fluid of approximately 1,200 mOsm/kg (Fig. 14-2). This concentrated fluid is then diluted by the active reabsorption of electrolytes in the ascending limb of the loop of Henle and in the distal tubule, both of which are relatively impermeable to water. As fluid exits the distal tubule and enters the collecting duct, osmolality is approximately 50 mOsm/kg. Within the collecting duct, water reabsorption is modulated by ADH (also called vasopressin). Vasopressin binds to V₂ receptors along the basolateral membrane of the collecting duct cells, then stimulates the synthesis and insertion of the aquaporin-2 water channel into the luminal membrane of collecting duct cells.²⁴

Plasma hypotonicity suppresses ADH release, resulting in excretion of dilute urine. Hypertonicity stimulates ADH secretion, which increases the permeability of the collecting duct to water and enhances water reabsorption. In response to changing plasma [Na⁺], changing secretion of ADH can vary urinary osmolality from 50 to 1,200 mOsm/kg and urinary volume from 0.4 to 20 L/day (Fig. 14-3).²⁵ Other factors that stimulate ADH secretion, although none as powerfully as plasma tonicity, include hypotension, hypovolemia, and nonosmotic stimuli such as nausea, pain, and medications, including opiates.

Two powerful hormonal systems regulate total body sodium. The natriuretic peptides, ANP, brain natriuretic peptide, and C-type natriuretic peptide, defend against sodium overload²⁶,²⁷,²⁸ and the renin–angiotensin–aldosterone axis defends against sodium depletion and hypovolemia. ANP, released from the cardiac atria in response to increased atrial stretch, exerts vasodilatory effects and increases the renal excretion of sodium and water. ANP secretion is decreased during hypovolemia. Even in patients with chronic (nonoliguric) renal insufficiency, infusion of ANP in low, nonhypotensive doses increased sodium excretion and augmented urinary losses of retained solutes.²⁹

Aldosterone is the final common pathway in a complex response to decreased effective arterial volume, whether decreased effective arterial volume is true or relative, as in edematous states or hypoalbuminemia. In this pathway, decreased stretch in the baroreceptors of the aortic arch and carotid body and stretch receptors in the great veins, pulmonary vasculature, and atria result in increased sympathetic tone. Increased sympathetic tone, in combination with decreased renal perfusion, leads to renin release and formation of angiotensin I from angiotensinogen. Angiotensin-converting enzyme (ACE) converts angiotensin I to angiotensin II, which stimulates the adrenal cortex to synthesize and release aldosterone.³⁰ Acting primarily in the distal tubules, high concentrations of aldosterone cause sodium reabsorption and may reduce urinary excretion of sodium nearly to
zero. Intrarenal physical factors are also important in regulating sodium balance. Sodium loading decreases colloid osmotic pressure, thereby increasing the glomerular filtration rate (GFR), decreasing net sodium reabsorption and increasing distal sodium delivery, which, in turn, suppresses renin secretion.

Calculation of maintenance fluid requirements is of limited value in determining intraoperative fluid requirements. However, calculation of maintenance fluid requirements (Table 14-10) is useful for estimating water and electrolyte deficits that result from preoperative restriction of oral food and fluids and for estimating the ongoing requirements for patients with prolonged postoperative bowel dysfunction. In healthy adults, sufficient water is required to balance gastrointestinal losses of 100 to 200 mL/day, insensible losses of 500 to 1,000 mL/day (half of which is respiratory and half is cutaneous), and urinary losses of 1,000 mL/day. Urinary losses exceeding 1,000 mL/day may represent an appropriate physiologic response to ECV expansion or pathophysiologic inability to conserve salt or water.

Daily adult requirements for sodium and potassium are approximately 75 and 40 mEq/kg respectively, although wider ranges of sodium intake than potassium intake are physiologically tolerated because renal sodium conservation and excretion are more efficient than potassium conservation and excretion. Therefore, healthy, 70 kg adults require 2,500 mL/day of water containing [Na⁺] of 30 mEq/L and [K⁺] of 15 to 20 mEq/L. Intraoperatively, fluids containing sodium-free water (i.e., [Na⁺] < 130 mEq/L) are rarely used in adults because of the necessity for replacing isotonic losses and the risk of postoperative hyponatremia.

Traditionally, glucose-containing intravenous fluids have been given in an effort to prevent hypoglycemia and limit protein catabolism. However, because of the hyperglycemic response associated with surgical stress, only infants and patients receiving insulin or drugs that interfere with glucose synthesis are at risk for hypoglycemia. Iatrogenic hyperglycemia can limit the effectiveness of fluid resuscitation by inducing an osmotic diuresis and, in animals, may aggravate ischemic neurologic injury.³¹ Although associated with worsened outcome after subarachnoid hemorrhage³² and traumatic brain injury³³ in humans, hyperglycemia may also constitute a hormonally mediated response to more severe injury. In critically ill patients, some evidence suggests that tight control of plasma glucose (maintenance of plasma glucose between 80 and 110 mg/dL) is associated with better outcomes, but other evidence does not, and in fact, suggests that tight control may increase mortality.³⁴,³⁵,³⁶,³⁷

Surgical patients require replacement of PV and ECV losses secondary to wound or burn edema, ascites, and gastrointestinal secretions. Wound and burn edema and ascitic fluid are protein-rich and contain electrolytes in concentrations similar to plasma. Although gastrointestinal secretions vary greatly in composition, the composition of replacement fluid need not be closely matched if ECV is adequate and renal and cardiovascular functions are normal. Substantial loss of gastrointestinal fluids requires more accurate replacement of electrolytes (i.e., potassium, magnesium, phosphate). Chronic gastric losses may produce hypochloremic metabolic alkalosis that can be corrected with 0.9% saline; chronic diarrhea may produce hyperchloremic metabolic acidosis that may be prevented or corrected by infusion of fluid containing bicarbonate or bicarbonate substrate (e.g., lactate). If cardiovascular or renal function is impaired, more precise replacement may require frequent assessment of serum electrolytes.

Conventionally, intraoperative fluid management has included replacement of fluid that is assumed to accumulate extravascularly in surgically manipulated tissue.³⁸ Until recently, perioperative clinical practice included, in addition to maintenance fluids and replacement of estimated blood loss, 4 to 6 mL/kg/hr for procedures involving minimal tissue trauma, 6 to 8 mL/kg/hr for those involving moderate trauma, and 8 to 12 mL/kg/hr for those involving extreme trauma.

However, recent clinical trials strongly link perioperative fluid management to potentially important alterations of both minor and major morbidities. Moreover, the influence of fluid volume may be specific to the type of surgery and to the types of fluid used. Maharaj et al.³⁹ randomized 80 ASA I-II patients scheduled for gynecologic laparoscopy either to large volume, defined as 2 mL/kg/hr of fasting over 20 minutes preoperatively (e.g., 1,440 mL/60 kg in a patient who had been fasting for 12 hours) or small volume, defined as total fluid of 3 mL/kg over 20 minutes preoperatively. In patients receiving the higher dose, postoperative nausea and vomiting and pain were significantly reduced (Fig. 14-4).³⁹ Holte et al.⁴⁰ randomized 48 ASA I-II patients undergoing laparoscopic cholecystectomy to receive either 15 or 40 mL/kg of lactated Ringer solution intraoperatively; the higher dose of fluid was associated with improved postoperative pulmonary function and exercise capacity, reduced neurohumoral stress response, and improvements in nausea, general sense of well-being, thirst, dizziness, drowsiness, fatigue, and balance function. Holte et al.⁴¹ randomized 48 ASA I-III patients undergoing fast-track elective knee arthroplasty under intraoperative epidural/spinal anesthesia and postoperative epidural analgesia to either liberal or restricted fluids. Median intravenous fluid administered intraoperatively and in the postanesthesia care unit in the restrictive group was 1,740 mL (range: 1,100 to 2,165 mL) of lactated Ringer solution and in the liberal group was 3,275 mL (range: 2,400 to 4,000 mL). Restrictive fluid administration was associated with a higher incidence of vomiting but less hypercoagulability and no difference in short-term postoperative mobility or ileus. Therefore, fluid restriction appears to be less well tolerated than more liberal fluid
therapy in patients undergoing surgery of limited scope, but perhaps at the expense of hypercoagulability.

In patients undergoing major intra-abdominal surgery, recent randomized controlled trials also suggest that restrictive fluid administration is associated with a combination of positive and negative effects. Brandstrup et al.⁴² randomized 172 elective colon surgery patients to either restrictive perioperative fluid management or standard perioperative fluid management, with the primary goal of maintaining preoperative body weight in the fluid-restricted group. By design, the fluid-restricted group received less perioperative fluid and acutely gained <1 kg in contrast to >3 kg in the standard therapy group. More importantly, cardiopulmonary complications, tissue-healing complications, and total postoperative complications were significantly fewer in the fluid-restricted group. In 152 patients undergoing intra-abdominal surgery, including colon surgery, Nisanevich et al.⁴³ reported less prompt return of gastrointestinal function and longer hospital stays in patients receiving conventional fluid therapy (10 mL/kg/hr of lactated Ringer solution) than in patients receiving restricted fluid therapy (4 mL/kg/hr). In a small clinical trial comparing gastric emptying in patients randomized to receive the postoperative fluids at a restricted (≤2 L/day of water; ≤77 mEq/day) or liberal regimen (≥3 L/day of water; ≥154 mEq/day), gastric emptying time for both liquids and solids was significantly reduced in patients receiving restricted fluids (Fig. 14-5).⁴⁴ Khoo et al.⁴⁵ randomized 70 ASA I-III patients undergoing elective colorectal surgery to conventional perioperative management, including intraoperative
fluid management at the discretion of the anesthesiologist, or to multimodal perioperative management, including intraoperative fluid restriction, unrestricted postoperative oral intake, prokinetic agents, early ambulation, and postoperative epidural analgesia. Multimodal perioperative management was associated with a reduced median stay (5 vs. 7 days) and fewer cardiorespiratory and anastomotic complications, but more hospital readmissions. Holte et al.⁴⁶ randomized 32 ASA I-III patients undergoing “fast-track” colon resection under combined epidural/general anesthesia to intraoperative fluid administration using either a restrictive (median: 1,640 mL; range: 935 to 2,250 mL) or liberal (median: 5,050 mL; range: 3,563 to 8,050 mL) regimen. Fluid-restricted patients had significantly improved postoperative forced vital capacity and fewer, less severe episodes of oxygen desaturation but at the expense of increased stress responses (aldosterone, ADH, and angiotensin II measurements) and a statistically insignificantly increased number of complications. In a recent meta-analysis, Corcoran et al. reviewed 23 randomized trials involving 3,861 patients assigned to liberal or goal-directed therapy during major surgery.⁴⁷ Patients in both the liberal and goal-directed therapy groups received more fluid during surgery than their respective comparative groups (restrictive fluid administration). However, the patients in the liberal groups had a higher risk of pneumonia (risk ratio 2.2), pulmonary edema (risk ratio 3.8), and longer hospital stay (mean difference 2 days) than their comparative groups. The patients in the goal-directed therapy groups had a lower risk of pneumonia and renal complications (risk ratio 0.7), and shorter hospital stay (mean difference 2 days) compared to the patients in the non goal-directed therapy group. These authors conclude that goal-directed fluid therapy is associated with fewer adverse outcomes than liberal fluid administration that is not goal directed. In addition, they conclude that whether goal-directed therapy is superior to restrictive fluid therapy remains uncertain.

Critically ill patients with acute lung injury represent an important group that may benefit from careful regulation of fluid intake. The ARDS Clinical Trials Network⁴⁸ randomized 1,000 patients with acute lung injury to a 7-day trial comparing a conservative fluid strategy with a liberal fluid strategy. Over the course of the trial the conservative strategy group had a cumulative net fluid balance that was slightly negative in comparison to a mean net cumulative fluid balance in the liberal group of nearly 7 L. Although overall mortality was no different in the two groups, the conservative fluid group had improved oxygenation and required fewer days of mechanical ventilation and intensive care. Despite achieving a negative fluid balance, the conservative strategy group had no greater incidence of acute renal failure.

Osmotically active particles attract water across semipermeable membranes until equilibrium is attained. The osmolarity of a solution refers to the number of osmotically active particles per liter of solvent; osmolality, a measurement of the number of osmotically active particles per kilogram, can be estimated as follows:

where osmolality is expressed in mmol/kg, [Na⁺] is expressed in mEq/L, serum glucose is expressed in mg/dL, and BUN is blood urea nitrogen expressed in mg/dL. Sugars, alcohols, and radiographic dyes increase measured osmolality, generating an increased “osmolal gap” between the measured and calculated values.

A hyperosmolar state occurs whenever the concentration of osmotically active particles is high. Both uremia (increased BUN) and hypernatremia (increased serum sodium) increase serum osmolality. However, because urea distributes throughout TBW, an increase in BUN does not cause hypertonicity. Sodium, largely restricted to the ECV, causes hypertonicity, that is, osmotically mediated redistribution of water from ICV to ECV. The term tonicity is also used colloquially to compare the osmotic pressure of a parenteral solution to that of plasma.

Although only a small proportion of the osmotically active particles in blood consist of plasma proteins, those particles are essential in determining the equilibrium of fluid between the interstitial and plasma compartments of ECV. The reflection coefficient (σ) describes the permeability of capillary membranes to individual solutes, with 0 representing free permeability and 1 representing complete impermeability. The reflection coefficient for albumin ranges from 0.6 to 0.9 in various capillary beds. Because capillary protein concentrations exceed interstitial concentrations, the osmotic pressure exerted by plasma proteins (termed colloid osmotic pressure or oncotic pressure) is higher than interstitial oncotic pressure and tends to preserve PV. The filtration rate of fluid from the capillaries into the interstitial space is the net result of a combination of forces, including the gradient from intravascular to interstitial colloid osmotic pressures and the hydrostatic gradient between intravascular and interstitial pressures. The net fluid filtration at any point within a systemic or pulmonary capillary is represented by Starling’s law of capillary filtration, as expressed in the equation:

where Q is the fluid filtration, k is the capillary filtration coefficient (conductivity of water), A is the area of the capillary membrane, P_c is the capillary hydrostatic pressure, P_i is the interstitial hydrostatic pressure, σ is the reflection coefficient for albumin, π_i is the interstitial colloid osmotic pressure, and π_c is the capillary colloid osmotic pressure.

The IFV is determined by the relative rates of capillary filtration and lymphatic drainage. P_c, the most powerful factor promoting fluid filtration, is determined by capillary flow, arterial resistance, venous resistance, and venous pressure. If capillary filtration increases, the rates of water and sodium filtration usually exceed protein filtration, resulting in preservation of π_c, dilution of π_i, and preservation of the oncotic pressure gradient, the most powerful factor opposing fluid filtration. When coupled with increased lymphatic drainage, preservation of the oncotic pressure gradient limits the accumulation of IF. If P_c increases at a time when lymphatic drainage is maximal, then IFV accumulates, forming edema.

If membrane permeability is intact, colloids such as albumin or hydroxyethyl starch preferentially expand PV rather than IFV. Concentrated colloid-containing solutions (e.g., 25% albumin) exert sufficient oncotic pressure to translocate substantial volumes of IFV into the PV, thereby increasing PV by a volume that exceeds the original infused volume. PV expansion unaccompanied by IFV expansion offers apparent advantages: Lower fluid requirements, less peripheral and pulmonary edema accumulation, and reduced concern about the cardiopulmonary consequences of later fluid mobilization (Table 14-11).

However, exhaustive research has failed to establish the superiority of either colloid-containing or crystalloid-containing fluids for either intraoperative or postoperative use. Moretti et al.⁴⁹ reported that patients who were randomized to receive
6% hetastarch had less postoperative nausea and vomiting than those who received lactated Ringer solution without colloid. In addition, colloid administration appears to have been an essential component of perioperative management strategies that demonstrated improved morbidity after colon surgery⁴² and after major surgery in conjunction with goal-directed fluid challenges.⁵⁰,⁵¹

In critically ill patients and patients undergoing more extensive surgery, systematic reviews of available comparisons of colloid versus crystalloid and albumin versus crystalloid suggested that the choice of fluid did not influence mortality.⁵² A recent randomized controlled trial comparing 4% albumin with 0.9% saline for fluid maintenance in 6,997 critically ill patients supports the conclusion that choice of colloid or crystalloid does not influence mortality.⁵³ Baseline serum albumin concentration did not alter the lack of effect of albumin management on outcome.⁵⁴ However, subgroup analyses suggested that crystalloid treatment could be superior in patients after trauma and that colloid could be superior in patients with severe sepsis.⁵³ Subsequent 2-year follow-up of a subset of 460 patients with traumatic brain injury (Glasgow Coma Scale score ≤13) demonstrated a nearly twofold increased risk of death in patients receiving colloid fluid management.⁵⁵