Fluid management during mechanical ventilation is complicated by both the influence of positive-airway pressure on normal homeostatic control of bodily fluids, and the interaction of mechanical ventilation with fluid status. Hypovolemia may lead to hemodynamic intolerance of positive-airway pressure, and fluid overload may result in both impaired gas exchange and respiratory mechanics and deleterious systemic effects. In patients with acute lung injury (ALI) and the acute respiratory distress syndrome (ARDS), positive fluid balance has been associated with both fewer ventilator-free days, and longer intensive care unit (ICU) stay, and with mortality in prospective randomized1 and observational2 studies, respectively.

Fluid Compartments

See reference 3.

In the normal adult male, total body water accounts for approximately 60% of body weight. In turn, approximately 40% of body weight is intracellular water and approximately 20% is distributed into the extracellular fluid volume, made up of interstitial fluid (approximately 16%), plasma volume (approximately 4%), and usually negligible volumes of lymph and transcellular fluid (cerebrospinal fluid and pericardial, intrapleural, and peritoneal fluid). Tissues such as brain, kidney, liver, and muscle have high water contents (70% to 80%) but adipose tissue has low water content (approximately 10%). Consequently, women, who tend to have more adipose tissue, have a lower total body water (approximately 50% of body weight). Total body water decreases in the elderly because of a loss of muscle mass.

The extracellular volume is distributed in interstitial fluid and plasma volume, and consists of two compartments. Seventy percent of the volume is rapidly equilibrating (approximately 20 minutes), and the remainder slowly equilibrates (approximately 24 hours) in dense connective tissue and bone. Sodium balance regulates the extracellular volume, whereas water balance regulates the intracellular volume.

Water Balance

Water balance is primarily determined by thirst and the renal action of arginine vasopressin, also termed antidiuretic hormone, which is secreted from the posterior pituitary following synthesis in the hypothalamus, in response to a wide variety of stimuli, particularly plasma osmolality. Vasopressin activates V2 receptors on the basolateral surface of the distal renal tubule and collecting duct, leading to an increase in water permeability, and reabsorption of filtrate, through fusion of aquaporin-2 with the luminal membrane. Vasopressin also reduces water clearance by decreasing renal medullary blood flow, and independently increases the renal medullary concentration gradient by stimulating a urea transporter.4

Under normal circumstances, a plasma osmolality of 280 mOsm/kg suppresses vasopressin secretion allowing maximal urinary dilution. As osmolality progressively rises to 295 mOsm/kg, so does the secretion of vasopressin, with an associated reduction in free water clearance. The kidney can normally concentrate filtrate up to 1200 mOsm/kg under the influence of vasopressin, although this tends to deteriorate with age and renal dysfunction. Table 65-1 lists other stimuli that influence vasopressin secretion. High-pressure stretch receptors in the aortic arch and carotid sinus sense a significant (>10%) fall in blood pressure (BP), leading to an increase in vasopressin release. As vasopressin also causes vasoconstriction through stimulation of V1 receptors, this is an important homeostatic response in shock,4 but appears to be reset within 32 hours of sustained hypovolemia.5 Stimulation of low pressure stretch receptors in the atria primarily results in an increase in both sympathetic tone and renin, and a decrease in atrial natriuretic peptide (ANP), with vasopressin release unaffected until the systemic BP falls.

Sodium Balance

The extracellular volume is primarily regulated through control of sodium balance, which is, in turn, regulated through control of effective plasma volume and its composition. The total body sodium content is approximately 4000 mmol; most of which is found extracellularly, and about half is rapidly exchangeable. Although the standard Western diet contains approximately 150 mmol of sodium per 24 hours, this varies widely and urinary sodium excretion varies between 0.2 and 242 mmol per 24 hours,6 reflecting a balance between sodium input and output. Although a moderate range of total body sodium content is well tolerated, once effective plasma volume is significantly affected, short-term and longer-term homeostatic responses are initiated.

A fall in effective plasma volume leads to activation of baroreceptors with augmentation of myocardial performance and peripheral vascular tone, and defense of plasma volume through shift of fluid from the interstitium. Longer-term responses include reduced sodium loss by the kidney and sweat glands, through a direct effect of aldosterone. When the baroreceptors are stimulated the increase in sympathetic tone reduces sodium loss through reduced glomerular filtration rate, and through increased tubular sodium reabsorption, both through a direct effect and through the actions of increased renin, angiotensin II, and aldosterone. In addition, ANP is released from the cardiac atria in response to stretch, and directly increases glomerular filtration rate through afferent arteriolar vasodilation, increases renal medullary blood flow, antagonizes vasoconstriction secondary to angiotensin II, and decreases sodium reabsorption by the collecting duct. Dopamine is produced in the kidney following conversion from l-dopa under the action of the cytosolic enzyme l–amino acid decarboxylase present in the proximal tubules.7 This is upregulated following a high-salt diet, leading to increased urinary sodium loss as dopamine acts to inhibit sodium reabsorption in the proximal tubule,8 and contributes to the increase in urine output sometimes seen following administration of low-dose dopamine. The renal synthesis of prostaglandins, such as prostaglandin E2 and prostacyclin (PGI2), tends to maintain renal blood flow and glomerular filtration rate through vasodilation, and directly increase water and sodium excretion. Consequently, in stressed patients cyclooxygenase inhibitors can precipitate renal dysfunction. Dopaminergic renal vasodilation in part acts through release of PGI2, because administration of dopaminergic antagonists leads to reduced urinary prostaglandins and loss of dopaminergic vasodilation,9 perhaps explaining why low-dose dopamine appears to be ineffective in septic ICU patients10 who already have a prostaglandin-driven kidney.

Influence of Positive-Pressure Ventilation on Fluid Balance

Positive-pressure ventilation and positive end-expiratory pressure (PEEP) raise intrathoracic pressure, resulting in reduced venous return and transmural pressure (see Chapter 36), with consequent complex neurohumoral responses leading to sodium and water retention. Because assisted, supported, and spontaneous modes of ventilation progressively ameliorate the elevation of intrathoracic pressure and its consequences, different ventilator modes variably reduce venous return. Reductions in stroke volume, cardiac output, and BP then lead to stimulation of high-pressure baroreceptors, and altered regional blood flow. Both low- and high-pressure baroreceptor stimulation lead to increased sympathetic outflow, and release of renin, aldosterone, and ANP. Renal denervation does not prevent sodium and water retention.11 Angiotensin-converting enzyme inhibitors12 and deliberate hypervolemia,11 however, reduce sodium and water retention during positive-pressure ventilation.

Right atrial transmural pressure and stretch are also reduced by PEEP and positive-pressure ventilation, and this leads to reduced secretion of ANP,13,14 with consequent reduction in water and sodium excretion reversed by restoration of venous return with lower body positive pressure. PEEP levels above 10 cm H2O may lead to an increase in central venous pressure (CVP), and regional venous pressures, which in the kidney contribute to reduced sodium and water excretion, independent of neurohumoral effects.15

In summary, various neurohumoral responses to positive-pressure ventilation lead to retention of sodium and water, as a homeostatic response to raised intrathoracic pressure. A major consequence of this response is expanded plasma volume, and a tendency toward systemic and pulmonary edema.

The Starling Equation

The major difference between plasma volume and interstitial fluid is the lower concentration of plasma proteins in the interstitial fluid, typically 40% of their plasma concentration. Although this concentration difference has little effect on the osmotic pressure between these two compartments, it leads to an important difference in oncotic pressure, and 80% of this is attributed to differences in albumin concentration. The normal plasma osmotic pressure is approximately 5500 to 6000 mm Hg, with approximately 28 mm Hg contributed by plasma proteins, despite having an osmolality of approximately 1.2 mOsm/kg.

The Starling equation quantitates the transvascular flux of fluids across the microcirculation (Jv). It is usually written as:

where Kf,c is the capillary filtration coefficient, Pcap is the hydrostatic microvascular pressure, Pint is the interstitial pressure, πcap is the plasma oncotic pressure, πint is the interstitial oncotic pressure, and σ is the osmotic reflection coefficient. Kf,c is determined by both endothelial hydraulic conductance and endothelial surface area, and σ is a measure of protein selectivity. The osmotic reflection coefficient is thought to be 1 in the cerebral microcirculation where the blood–brain barrier effectively prevents protein flux, and approximately 0.7 to 0.8 in the normal pulmonary microcirculation, although this is markedly reduced during lung injury.14 In a typical systemic microcirculatory bed, the arterial end of Pcap is 30 mm Hg and the venous end is 10 mm Hg. Assuming σ equals 1, Pint is −3 mm Hg, πcap is 28 mm Hg, and πint is 8 mm Hg, the net driving pressure out of the capillary at the arterial end of the microcirculation will be ([30−3] − [28−8]) or 13 mm Hg, although the effective πint may be a little lower.16 At the venous end of the microcirculation the net driving pressure into the capillary will be 7 mm Hg, and most of the filtered fluid is reabsorbed, with the lymphatics draining the remainder.

In the healthy lung, Pcap is usually assumed to be 7 mm Hg, Pint as −8 mm Hg, and πint as 14 mm Hg. Consequently, the driving pressure across the pulmonary circulation is thought to be positive (Fig. 65-1), leading to net filtration of fluid, with lymphatic absorption usually estimated to be approximately 20 mL/hour. The final filtration rate is determined by the capillary surface area, convective forces, and diffusive forces. When Pcap is suddenly raised, the filtration rate may increase more than threefold.16

There are a number of safety factors that are thought to help prevent alveolar edema. These include low epithelial permeability (pore size approximately 10% that of the endothelium), low alveolar surface tension reflecting normal surfactant function, active vectorial transport of ions across the epithelium (see Fig. 65-1), and favorable interstitial function and lymphatic drainage (edema fluid moves along a negative-pressure gradient centrally away from the alveolus, while an associated reduction in the negative pressure and dilution of interstitial oncotic pressure reduce filtration).17 Changes in either Starling forces or these safety factors can lead to pulmonary edema. For example, an increase in Pcap is the basis for hydrostatic pulmonary edema; a decrease in Pint, as might be seen with an obstructed airway and vigorous respiratory efforts, is thought to cause postobstructive pulmonary edema; although a decrease in σ is the basis for permeability pulmonary edema, concurrent surfactant dysfunction leads to an increase in surface tension and a decrease in Pint, also favoring the development of edema.18–20 Remodeling of the lung parenchyma in chronic heart failure and conditions of persistently elevated Pcap such as mitral stenosis allow patients to tolerate a relatively high Pcap without developing marked pulmonary edema. Models of chronic heart failure suggest important homeostatic responses including a reduction in Kf,c, which together with further reduction in surface tension associated with increased surfactant content, can prevent pulmonary edema despite elevated Pcap.21,22

In formulating an approach to fluid management in the ventilated patient, a balance between parsimonious and generous fluid therapy needs to be considered. Although this is commonly termed the “dry or wet” approach, and this may be an appropriate general description for particular groups of patients, most clinicians classify fluid therapy as (a) maintenance, (b) replacement, and (c) resuscitation fluids. In general, the tendency of ventilated patients to retain fluids, and the benefits of fluid restriction, argue for the dry approach provided there is due attention to adequate resuscitation. A consensus statement classified fluid restriction as grade IIa evidence in ALI and ARDS.23

Maintenance Fluids

The volume of appropriate maintenance fluids in a ventilated patient is usually the most contentious of these parameters. As noted in Table 65-2, approximately 1150 mL of fluid per day should be adequate maintenance in ventilated patients; this amount is less than in nonventilated subjects because all gases are humidified. In practice, metabolic production of water is usually ignored, allowing a total intake of 1500 mL/day; this represents a baseline volume that needs to be reviewed following clinical and biochemical assessment of plasma volume and total body water. Nevertheless, in balance, this volume should be sufficient to generate a urine output of approximately 800 mL/day or 0.5 mL/kg/hour; because the normal daily solute load excreted by the kidney is approximately 600 mOsm, this is easily achieved by the normal kidney with this urine volume by concentrating urine to 700 to 800 mOsm/L. The planned sum of enteral and parenteral maintenance fluids, however, may be less than this because of obligatory fluids given with drug infusions and hydrostatic pressure transducer flush (approximately 3 mL/hour per transducer). Other factors influencing maintenance fluids include the size of the patient, covert losses, such as a diaphoresis, fever that increases fluid loss by approximately 10 mL/kg/day for each degree of temperature elevation, and the ease of supplying adequate nutrition. Although 1500 mL is given as an adequate maintenance fluid volume, many centers use greater volumes, and many patients tolerate greater maintenance fluid volumes.

The Dry Approach

The neurohumoral response to positive-pressure ventilation tends to retain sodium and water, which may lead to both peripheral and pulmonary edema. Daily weights are inconvenient and infrequently performed in ventilated patients. Peripheral edema needs to be carefully sought, and is often evident in the limbs of bedridden patients, and as a wedge-shaped swelling in the flanks. Pulmonary edema is typically detected as dependent crackles on auscultation, but this is a late sign and requires an appropriately placed stethoscope. Other techniques include chest radiographs and measurement of extravascular lung water. In addition to lack of sensitivity, these methods may be troublesome to interpret in disease states such as ALI.

Moderate hypohydration (mean 4.5% loss in body weight) leads to a reversible improvement in lung volume and airflow resistance in normal subjects.24 Excess fluids may lead to pulmonary edema, which is associated with impaired oxygenation, prolonged ventilation and ICU stay, and difficulty weaning. In both ALI and ARDS,25 however, and in acute cardiogenic pulmonary edema,26 extravascular lung water does not correlate with CVP or pulmonary artery occlusion pressure (PAOP). Although the PAOP is an important determinant of the pulmonary capillary filtration pressure, the extravascular lung water is also influenced by permeability and temporal effects. In acute pulmonary edema, empiric treatment with diuretics, nitrates, and ventilator support will often lead to marked reduction of CVP and PAOP before resolution of the pulmonary edema.26 Indeed, there may be transient hypovolemia secondary to extravasation of fluid in the lung, requiring volume loading.

In ALI and ARDS, extravascular lung water is usually elevated despite normal filtration pressure. Excess lung water portends a worse outcome.27 Patients with ARDS who achieve a significant reduction in PAOP28 or total body water, as estimated from weight loss or cumulative fluid balance,29 are more likely to survive. Prospective, randomized studies report improved oxygenation in hypoproteinemic patients with ALI when a negative fluid balance was produced using furosemide and concentrated albumin;30 and management of pulmonary edema according to lung water, as compared to PAOP, can lead to a lower cumulative fluid balance, fewer ventilator days, and shorter ICU stay.31 Prevention of acute left-ventricular failure by diuresis allows weaning in some ventilator-dependent patients with chronic obstructive pulmonary disease.32 Positive fluid balance is associated with increased mortality in both ALI,2 and in critically ill patients with acute kidney injury.33 Liberal fluid management in ALI reduces ventilator-free days and increases ICU stay,1 with a trend to increased dialysis (14% vs. 10%, p = 0.06) consistent with reduced recovery of renal function with fluid overload in acute kidney injury.33 An important aspect of these data is that the patients were enrolled following the diagnosis of ALI in ventilated patients, and may not apply to the initial resuscitation phase. Taken together, these data suggest that accumulation of excess fluid is common, often difficult to detect, and may be associated with serious adverse events, and mortality.

Replacement Fluids