Intravenous fluid administration is performed in critically ill patients to maintain fluid volume homeostasis and to achieve many other goals. Intensivists and perioperative physicians who care for the cardiothoracic critically ill patient must appreciate that fluids are ‘drugs’ and therefore it is necesssary to understand their physiology and pharmacology. Appropriate fluid resuscitation is often the first intervention for the haemodynamically unstable patient. There is ongoing debate regarding whether the fluid chosen for these patients should be a crystalloid solution (i.e. isotonic saline, Ringer’s lactate) or a colloid containing solution (i.e. albumin solution, hyperoncotic starch). Large international surveys have shown that choice of fluid is mostly a matter of institutional preference rather than due to specific procedural/patient related factors.
There are several determinants in assessing the need for intravascular volume replacement in the context of cardiac surgery: (i) blood loss, (ii) increasing vascular capacitance as often occurs with patient rewarming, (iii) third space fluid losses secondary to cardiopulmonary bypass (CPB) induced systemic inflammation, and (iv) increased cardiac preload requirements in the setting of transient cardiac ischaemia – reperfusion injury, myocardial stunning and reduced ventricular compliance.
Total body water for the average 70 kg adult male is about 45 l of which 30 l (65%) is intracellular and 15 l (35%) is extracellular fluid. The extracellular space can be subdivided into interstitial space (10l) and intravascular space (5 l). All three compartments also contain electrolytes, including ions and larger molecules. The cellular membranes maintain potential differences via a complex differential concentration between intracellular (potassium) and extracellular (sodium) ions. In addition, proteins and large molecules maintain the colloid distributions between intracellular, interstitial and intravascular spaces.
The fate of any intravenous fluid given into the vascular compartment depends on many factors including the electrolyte and colloid (if any) compositions, the patient’s intravascular volume status, the total volume of fluid administered, and the integrity of endothelial glycocalyx. The endothelial glycocalyx modulates vascular permeability and inflammation and according to the Starling model it is a key determinant of fluid disposition. The primary forces defining transcapillary fluid movement and the counterbalancing process of fluid movement into the vascular space in a normal and damaged vasculature are shown in Figure 14.1. It has been shown that CPB is associated with loss of glycocalyx integrity and microvascular dysfunction. Interestingly, significant shedding of the glycocalyx also occurs during off-CPB procedures. In critical illness, including post-CPB/postoperative states, endothelial glycocalyx denudation causes high transcapillary movement of fluid leading to tissue oedema and inadequate intravascular volume expansion after intravenous fluids are administered.
Figure 14.1 The opposing forces defining the steady state net flow of fluid from the capillary into the interstitial space are defined by the hydrostatic pressure differences between the capillary lumen pressure (Pc) and interstitial pressure (Pi) as opposed by the filtration coefficient (Kf) which itself is a function of the vascular endothelial cell integrity and the intraluminal glycocalyx. This net efflux of fluid out of the capillary into the interstitium is blunted by an opposing oncotic pressure gradient moving fluid in the opposite direction because capillary oncotic pressure (πc) is greater than interstitial oncotic pressure (πi). And like hydrostatic pressure dependent flow, oncotic dependent flow is blunted by the reflection coefficient (σ), which like Kf is a function of the glycocalyx and vascular endothelial integrity. Under normal conditions (left side), both Kf and σ are high, minimising fluid flux resulting in a slight loss of plasma into the interstitium which is removed by lymphatic flow. However, if the vascular endothelium and glycocalyx are damaged (right side), oncotic pressure gradients play a minimal role because a large amount of protein-rich plasma translocates into the interstitial space, minimising the oncotic pressure gradient, whereas the constant Pc promotes massive fluid loss and interstitial oedema.
Fluids in the Cardiothoracic ICU
There is a lack of prospective data with special regard to ‘ideal’ fluid in the cardiac surgical patient population. Colloid fluids would be expected to produce a much larger volume expanding effect. Usual teaching suggests that the colloid to crystalloid volume expansion efficacy is 1:3 (300% greater efficacy for colloids), whereas the colloid to crystalloid ratio for volume expansion in adequately powered and well-designed randomised controlled trials (RCTs) was 1:1.2–1:1.4 (approximately 30% greater efficacy for colloids). The contents of solutes in colloid and crystalloid solutions are presented in Tables 14.1 and 14.2.
It would stand to reason that early during elective cardiac surgery colloid fluids may have a better volume expansion effect than crystalloids as capillary pressures are likely to be normal and loading with crystalloids would produce a dilutional effect on plasma oncotic pressure leading to tissue oedema and a reduced volume expansion effect.
Crystalloids are preferred over colloid containing solutions for the management of patients with severe hypovolaemia not due to bleeding. Saline solutions seem to be as effective as other crystalloid solutions and colloid containing solutions, and are much less expensive. It is recommended that hyperoncotic starch solutions are avoided as they increase the risk of acute kidney injury, need for renal replacement therapy (RRT) and mortality. They may also lead to hypernatraemia, hyperchloraemia and acidosis. There was a black box warning from FDA for heat starch.
In a 9 year multicentre open label trial (CRISTAL), 2857 patients with hypovolaemic shock, due to any cause, were randomly assigned to fluid resuscitation with colloid or crystalloid. There was no difference in 28 day mortality or need for RRT between the two arms. Patients treated with colloids had more mechanical ventilation free days (13.5 versus 14.6 days) and vasopressor therapy (15.2 versus 16.2 days), as well as a lower 90 day mortality (31% versus 34%). However, the open label design and heterogeneity of fluids that were compared between the groups limit confidence in the apparent benefit of colloid solutions in this population.
In theory, albumin has two possible advantages over crystalloid solutions: (i) more rapid plasma volume expansion, since the colloid solution remains intravascularly (in contrast to saline, three quarters of whichenters the interstitium) and (ii) lesser risk of pulmonary oedema, since dilutional hypoalbuminaemia will not occur. Well-designed RCTs and meta-analyses have failed to demonstrate benefits from the use of albumin. However, in some cardiothoracic intensive care units albumin remains the colloid of choice for patients undergoing lung transplantation and pulmonary endarterectomy when volume expansion is needed in the immediate postoperative period, in an attempt to reduce the risk of early reperfusion lung injury.
Use of hyperoncotic starch solutions has been associated with increased incidence of acute kidney injury (AKI) and high mortality in some studies. In a RCT, 7000 patients were randomised to receive 6% HES or normal saline for all fluid resuscitation until ICU discharge. There was an increased incidence of AKI requiring RRT in the HES group compared with the saline group (5% versus 5.8%). Two meta-analyses, one of which excluded seven trials that were retracted due to scientific misconduct of one investigator, found that compared to conventional fluid resuscitation regimens, HES was associated with increased risk of mortality and need for RRT. HES associated AKI appears to be related to pinocytosis of metabolites into renal proximal tubular cells after glomerular filtration. HES has also been associated with increased risk of bleeding, which is probably due to impaired fibrin polymerisation and decreased factor VIII, vWF and XIII levels and not associated with haemodilution. Given the fact that the cardiac surgical patient population is at risk of AKI and coagulopathy anyway, the use of HES in the setting of cardiac surgery is not recommended.
One of the most commonly used crystalloid solutions is 0.9% sodium chloride or ‘normal’ saline, which is anything but ‘normal’ or ‘physiological’ as it is slightly hypertonic and contains equal amounts of sodium and chloride making it hypernatraemic and very hyperchloraemic relative to plasma. Administration of high volume ‘normal’ saline will result in hypernatraemia, hyperchloraemic acidaemia leading to renal vasoconstriction and reduction in renal blood flow. This has led to suggestions that physiologically buffered fluids with a chemical composition that approximates extracellular fluid (e.g. Ringer’s lactate solution, Hartman’s solution, Plasma-lyte 148, Accusol) are used instead of ‘isotonic’ saline for large volume resuscitation. A recent RCT which enrolled healthy volunteers, and large observational series, have demonstrated that adverse effects associated with saline use were not evident when physiologically buffered fluids were used.
Hypertonic solutions typically used in cardiac surgery include 2–7% saline. They have been used in an attempt to minimise volume overload and produce cellular dehydration in the context of cerebral oedema. The majority of the trials of hypertonic solutions have been in patients with traumatic brain injury. The results cannot be extrapolated for the general cardiac surgical population. However, cardiac surgical patients with cerebral oedema may benefit from hypertonic saline solutions on an individual basis. Studies examining use of hypertonic solutions did not examine their effect on renal function and reported a high incidence of hypernatraemia.
Positive fluid balance has been associated with worse outcomes in some but not all critically ill patient populations. A recent observational study that included a large cohort of medical, surgical and cardiothoracic ICU patients showed that positive fluid balance at the time of ICU discharge was associated with increased risk of death, after adjusting for markers of illness severity and chronic medical conditions, particularly in patients with underlying heart or kidney disease.
A secondary analysis of the FACTT trial demonstrated that patients with ARDS who developed AKI had a higher mortality (independent of a conservative or liberal fluid administration strategy). This study demonstrated some evidence of causality, as diuretic use was associated with a protective effect, potentially due to its effect on fluid balance. When the diuretic effect was adjusted for fluid balance, the protective effect was attenuated, thus potentially suggesting that the modulation of fluid balance promoted the benefit.
Intravenous fluid administration is a common intervention used for both the prevention and treatment of AKI. The pathogenesis of AKI in cardiac surgical patients is multifactorial, not only due to perturbed haemodynamics but also the result of direct cellular injury as well as indirect injury from inflammation and microcirculatory changes. AKI with oliguria as well as fluid resuscitation often results in accumulation of excess total body fluid. This fluid accumulates in all tissues of the body, and through the interstitial space, as well as remaining within the vascular space, resulting in increased venous pressure. The presence of oliguria is associated with a poor prognosis; however, it remains unclear whether this is due to severity of injury or due to fluid overload. Most studies to date have been conducted in the general ICU population. Similar findings have been reproduced in the cardiac surgery population, as early administration of fluid can lead to AKI. In a prospective observational cohort of 100 patients undergoing cardiac surgery, those patients in the quartile receiving the highest volume of fluid suffered the highest degree of AKI. This study was limited by the small number of patients included.
It has been demonstrated that positive fluid balance in the first 3 postoperative days is associated with primary graft dysfunction after lung transplantation. Patients undergoing pulmonary endarterectomy constitute a unique patient population at risk of development of ARDS due to high permeability lung injury. Therefore restrictive fluid management strategies and adequate diuresis in the first 48 hours (aiming for 2 l negative fluid balance) is of paramount importance. Negative fluid balance should also be maintained during the first week after lung resection surgery. Fluids should be limited because lung resection decreases the pulmonary vascular bed and overhydration will result in impaired gas exchange, need for prolonged mechanical ventilation, prolonged length of stay and high mortality.