Fluid Therapy at a Glance

In healthy humans, maintaining the balance between fluid intake and fluid losses is a minor issue. Fluid intake is driven by thirst and hunger (food contains a lot of water) and the kidneys quickly and effectively excrete any fluid overload. Feedback loops then restore the optimal body fluid volumes (Fig. 18.1). However, when our bodies are subjected to extreme environments or physiological strain, it becomes apparent that we only tolerate a deviation of a few percent from our optimal body fluid volumes. For example, during acute exercise, our physiological and mental capacities soon become impaired if fluid losses due to increased evaporation from the airways and sweating are not adequately replaced.

Fig. 18.1 Schematic drawing of the three fluid compartments in the human body.

From Hahn RG, Ed. Fluid Therapy in the Perioperative Setting. 2nd ed., Cambridge, 2016.

The thirst mechanism becomes gradually impaired with age and often deteriorates further in disease states, such as serious infection and dementia. Patients coming to hospital may be in such a debilitated state that they cannot eat and drink adequately by themselves. Intravenous (IV) fluid therapy then becomes a way to support the normal intake of fluid; this is referred to as maintenance fluid therapy. The minimum fluid requirement in the average man is 1 mL/kg/hour, but a rate of 1.3 mL/kg/hour is frequently used, and a rate of 1.5 mL/kg/hour is recommended in children.

The so-called 4/2/1 rule provides a detailed recommendation for suitable water intake in children. This old rule, which is still practised, suggests 4 ml/kg/hour for infants weighing 3 to 10 kg; 40 ml/kg plus 2 ml/kg/hour for each kg over 10 kg for children ranging from 10 to 20 kg; and 60 ml/h plus 1 ml/kg/hour for each kg over 20 kg in children weighing more than 20 kg.

In adults, the maintenance fluid is given as slow IV drip in the form of a 5% glucose solution. Sodium and potassium, at 1 mmol/kg body weight, should be added. The glucose solution is chosen because it provides a minimum amount of energy (200 kcal/L). Glucose is also the only widely available IV solution that partially hydrates the cells; this hydration is needed if therapy spans over several days, as both evaporation and urinary excretion cause losses of intracellular fluid.

The internal control of the fluid balance becomes grossly impaired during anaesthesia, major surgery and intensive care. The normal feedback loops operate less effectively and are disturbed by stress mechanisms, which usually act to retain fluid. For example, the diuretic response to crystalloid fluid during general anaesthesia and surgery is only about 10% of the response seen in conscious subjects [1]. During surgery, fluid evaporation from the skin and open wounds cannot be replaced by oral intake of fluid. Haemorrhage losses should be substituted with crystalloid electrolyte or colloid fluid until the blood haemoglobin concentration has decreased to the point that the ‘transfusion trigger’ is reached. This ‘trigger’ is the haemoglobin concentration where the body is no longer able to compensate for the reduced haemoglobin mass by increasing cardiac output in order to maintain oxygen delivery (in humans, 60–100 g/L). Furthermore, maintenance fluid therapy has to continue as long as patients are unable to eat and drink on their own.

On top of this, a patient might come to surgery in a state of fluid deficit due to vomiting, starvation, ketoacidosis, ileus, or a combination of these factors. Compensation of these losses is called replacement fluid therapy. My practice is to provide at least half of this fluid deficit before general anaesthesia for acute surgery is induced, while giving careful consideration to the cardiovascular situation.

The need for high-rate substitution of fluid during surgical procedures necessitates the use of resuscitation fluids, such as isotonic or nearly isotonic salt-based crystalloid and colloid solutions. These include buffered Ringer’s or Hartmann’s solution, hydroxyethyl starch and gelatin.

Urine Analysis

Detection of a fluid deficit in a healthy human is not as easy as one might think. Anamnesis is helpful, but routinely used clinical tests (such as skin-fold thickness) are poor guides. If the problem is purely a lack of water, the serum osmolality is increased (> 300 mOsmol/kg) but this is a fairly late sign [2]. Serum creatinine rises in a severe fluid deficit, but one must then have another sample to compare with. Lack of diuresis is an even later sign of dehydration (approximately 7% of the body weight) and the anaesthetist should be concerned at a much earlier stage (2–3% of the body weight).

A sensitive and simple method for detection of dehydration is to examine whether the kidneys are set to excrete or retain water. This can be assessed by taking a urine sample and measuring the concentrations of waste products that are excreted at a fairly stable rate. These measurements can include urine specific gravity (summary of solids), osmolality (salt balance), creatinine (from the metabolism of muscle) and urine colour (from the metabolism of erythrocytes).

The ranges of colour, osmolality and creatinine concentrations have been published for subjects aged 17–69 years, and each range paralleled the specific gravity scale [3]. These ranges were assigned a score, where a higher value indicated more severe dehydration (Table 18.1).

The mean of the four scores is termed the fluid retention index (FRI). In sports medicine, urine sampling has been evaluated as a means of detecting a fluid deficit, but only one marker is then commonly employed [4–6]. In hospital care, potential confounders due to diet and disease might affect the result. The use of an aggregate index is therefore recommended to obtain a more robust measure. An index value of ≥ 4.0 corresponds to the degree of renal water conservation that accompanies dehydration amounting to 3% of the body weight (specific gravity ≥ 1.02, creatinine ≥ 12 mmol/L, colour ≥ 4 and osmolality ≥ 600 mOsmol/kg) [3].

In hospital care, a high content of metabolic waste products in the urine is associated with a greater plasma volume expansion in response to crystalloid fluid [7]. For example, a lower cardiac index and greater need for fluid optimization has been reported before abdominal surgery, where the waste products show high values [8]. In hip fracture surgery, a high preoperative FRI has been associated with a greater rise in neutrophil gelatinase-associated lipocalin; a sensitive biomarker of kidney injury (NGAL) [9] and a higher number of postoperative complications [10]. In acute geriatric care, an admission FRI of ≥4.0 was associated with higher 30-day mortality [11].

The FRI score after surgery can less clearly be attributed to dehydration (indeed, a previous name was the Dehydration Index) because of the overlapping effects of trauma-induced fluid retention. However, in a study of hip fracture patients, the FRI hardly changed at all during surgery [9]. The FRI score also remained unchanged during the course of lengthy (3–4 h) abdominal cancer surgery when hydroxyethyl starch was used as infusion fluid, whereas a minor decrease occurred when Ringer’s solution was administered [12]. Therefore, FRI could be a useful index to the fluid status in a fasting patient before surgery while its value during surgery is more uncertain.

BioImpedance

Whole-body bioimpedance (BIA) is a method of estimating the volumes of the body fluid spaces (intra- and extracellular) by analysing a series of weak electrical currents of different frequency that run between the arm and the foot [13]. The method is based on the fact that electric currents have more difficulty passing through large amounts of water than through small amounts. Moreover, the preference for the current to pass through or outside the cells varies with the frequency. The patient must be at complete rest (no movement allowed) during the time when the series of currents are sent through the body. The risk of mechanical and electrical interference must be considered.

The best use of bioimpedance for the anaesthetist would be for detection of pre- and postoperative dehydration [14]. Of course, the purpose is to compensate for any deficit that is detected. This author has found the method to be useful for groups but perhaps not sufficiently accurate for clinical use in individuals.

Blood Volume Changes

The blood volume changes assessed from variations in haemoglobin (Hb) concentration can serve as a practical guide to fluid therapy during and after surgery. The basic idea is to calculate the total amount of Hb in the circulation, which is then corrected for dilution and estimated Hb losses [15]. For example, assume that Hb_mass(0) is the amount of Hb in the circulation at time zero (usually about 1 kg), BV₀ is the estimated blood volume at baseline and Hb₀ is the blood Hb concentration, and that the measurements are repeated at a later time (t). Then:

 Hbmass(0)= BV0Hb0BV(t) = (Hbmass(0)–Hbloss(0-t)) / Hb(t)ΔBV(t) = BV(t) – BV0(1)

Hb_loss is obtained as the product of the Hb₀ and the volume of the surgical bleeding if haemorrhage occurs rapidly. When the bleeding is slow, one multiplies the bled volume by the average of Hb₀ and Hb(t).

BV₀ is most conveniently obtained by some anthropometric measure. A simple equation is that BV₀ is 7% of the body weight, while regression equations provide more precise measures. These are typically based on tracer measurements of body fluid volumes performed in a large number of humans, and usually employ the gender, body weight and height of the subject as predictors. Below are examples of these equations for estimation of the blood volume in women and men [16].

These equations are easy to programme into a pocket calculator for use in the clinic. The following are three examples of how they can be applied. The examples may look similar, but the Hb calculations suggest three completely different treatments.

Our pocket calculator says that BV₀ should be approximately 5.0 L; therefore, the Hb_mass(0) should be 5 × 145 = 725 g. The bleeding of 500 mL was slow, and Hb_loss should therefore be calculated based on the mean of the pre- and postoperative Hb values. Hb_loss is then 0.5 × (145+110)/2 = 64 g. Further is BV(t) = (725 – 64) / 110 = 6.0 L.

The blood volume of this patient is 1.0 L larger than before the surgery. Hence, the patient should not have any more resuscitation fluid but should only receive a restrictive amount of maintenance fluid. An injection of furosemide would probably improve this patient’s general condition, and this is recommended. A bolus infusion to challenge the patient’s fluid status in this setting could possibly have precipitated pulmonary oedema.

Our pocket calculator says that BV₀ should be approximately 5.5 L; therefore, the Hb_mass(0) should be 5.6 × 114 = 638 g. The bleeding of 800 mL was rapid, which means that Hb_loss can be set to 0.8 × 114 = 91 g. The BV(t) then becomes = (638 – 91) / 105 = 5.2 L.

The change in blood volume, which is given by ∆BV(t) = BV(t) – BV₀, shows that the patient is hypovolaemic by 300 mL. Slight hypervolaemia in this setting would be beneficial, due to the vasodilatation caused by the epidural analgesia and the fact that an internal organ has been removed, which would be expected to cause some exudation and fluid accumulation in the surgical area. An increase in the blood volume by at least 0.5 L is warranted. However, this cannot be achieved without reducing the Hb concentration to below 100 g/L, which was decided as the transfusion trigger for this patient. My choice would be to prescribe 1 unit of erythrocytes, accompanied by the same amount of buffered Ringer’s solution, and to reassess her status thereafter with a new Hb analysis. A likely option is that a second unit of erythrocytes would be prescribed later in the evening.

The BV₀ should be approximately 4.2 L and Hb_mass(0) then 4.2 × 140 = 588 g. The bleeding of 800 mL occurred rapidly, which means that Hb_loss can be set to 0.8 × 140 = 112 g. The BV(t) then becomes (588 – 112) / 132 = 3.6 L.

The diagnosis is that the patient is hypovolaemic by 0.6 L. Treatment consists of clear fluids – either 500 mL of colloid or 1.5 L of buffered Ringer’s solution. I would prefer the colloid if the patient has already received 2 L of crystalloid during the surgery, because adding even more crystalloids would cause gastrointestinal oedema [17, 18]. This patient has no risk of hitting the transfusion trigger (in his case, probably 90 g/L) due to fluid prescription. One can actually estimate how much the Hb concentration would decrease by adding 500 mL of clear fluid to the BV(t) in the equations above. If more Hb is lost by haemorrhage, that amount is subtracted from Hb_mass. Transfused erythrocytes are simply added in the same fashion. It is important to note that the usefulness of the subsequent estimates depends on the accuracy of the quantitation of Hb losses and additions.

The Hb method for estimating blood volume changes is a simplified form of volume kinetics that has also been used in many scientific studies. For example, the BV change that is required before a drop in arterial pressure occurs during a transurethral prostatic resection is determined to be –0.5 L with the legs in stirrups but is +0.5 when the legs are lowered [19]. For this reason, it is good practice to give the patient a fluid bolus just before the legs are lowered at the end of a surgery performed in the lithotomy position when blood loss has occurred.

The Hb method has also been applied to show that as much as 60% of administered buffered Ringer’s solution is retained in the bloodstream during ongoing surgery [20] as well as to demonstrate that arterial hypotension acts to retain fluid during induction of epidural anaesthesia [21].

The lesson learnt from these blood volume assessments is that Hb should change in proportion to the haemorrhage (i.e. the lost blood volume) if normovolaemia is to be maintained. The patient is hypovolaemic when the haemorrhage, in proportion to the initial blood volume, is greater than the relative drop in Hb. Conversely, a marked drop in Hb level without a proportional blood loss indicates hypervolaemia. Most anaesthetists make this judgement by mentally extrapolating from the patient’s laboratory data, but erroneous conclusions can easily be drawn if the simple equations shown above are not entered into a pocket calculator (or an Excel spreadsheet) to obtain more precise estimates.

Electrolyte Shifts

The electrolyte shifts occurring when a fluid bolus is given can be used to analyse a clinical situation and to refine the fluid therapy. The assumptions in these calculations are quite similar to those made for Hb monitoring of the blood volume, except that ions that distribute across the entire ECF volume are used (i.e. interstitial fluid is included).

When an isotonic fluid is infused, the dilution of ions that are present in blood but not in the fluid indicates the size of the ECF volume. For example, an infusion of 1 L isotonic (5%) mannitol without electrolytes can be used to estimate the ECF space based on the accompanying dilution of serum sodium [22]. However, the calculation is not straightforward because some fluid and sodium are excreted during the experimental period. These losses by excretion can, and should, be accounted for. Assuming that the ECF volume of water at baseline is 20% of the body weight (BW) and serum sodium is S-Na, the following changes would occur in the ECF content of Na and water in response to an infusion:

Total amount of Na before infusion=ECF S-NaTotal amount of Na after infusion=ECF S-Na +infused Na – voided NaTotal amount of water before infusion=ECF Expected amount of water in ECFafter infusion =ECF+infused volume – voided volume(4)

Na ions remain outside the cells, while water can either enter or leave the cells. If no flow of water occurs, the S-Na after infusion, S-Na(t), is given by the ratio of the amount of Na in the ECF after the infusion and the expected amount of water in the ECF. Hence:

S −Na(t) at no flow = ECF × S​ -Na + infused Na −voided Na​ECF ​+ infused volume −voided volume(5)

Using this equation, one can predict the change in serum sodium (or any other electrolyte that distributes in the ECF volume) in response to any type of infusion. Interestingly, by measuring the serum sodium also after the infusion, we can compare the measured and predicted S-Na and thereby calculate the flow of fluid between the ICF and the ECF volumes that occurs in response to the infusion. This can be done by saying that the measured S-Na after this infusion must be the same as for the theoretical S-Na(t) at no flow, but adding the flow of fluid between the ECF and ICF as a new term in the denominator (∆ICF). Hence:

Measured S -Na(t) = ECF × S​ -​​​ Na + infused Na – voided Na​ECF ​+ infused volume  – voided volume +ΔICF(6)

By re-arrangement:

ΔICF=ECF+(infused – voided)volume – (ECF × S-​  Na – (added – voided)Na)Measured S-  Na(t)(7)

This fluid shift is difficult to calculate by alternative methods in a living human. This mass balance equation has mostly been used to estimate how fast and how extensively electrolyte-free irrigating fluid (e.g. used during endo-urologic surgery) distributes to the ICF space [23–25]. It has also demonstrated that a rapid infusion of buffered Ringer’s solution in volunteers actually recruits fluid from the ICF to the ECF, because the kidneys in healthy humans are not set to excrete as high an amount of sodium as is contained in this crystalloid fluid [26]. This runs counter to the “logical” and “intuitive” assumption that the slight hypo-osmolarity of the Ringer’s solution causes this fluid to hydrate the cells.

The equation shown above can also provide an explanation for the development of rebound oedema when plain 15% mannitol is infused. This infusion causes osmotic diuresis whereby sodium ions are lost. As long as the infused fluid does not contain any sodium, the therapy will reduce serum sodium, which will cause water to diffuse into the cells, i.e. increase the ICF volume (causing cellular edema). The magnitude of the oedema can be estimated from the equation. The amount of sodium that should be infused to withdraw the oedema can also be calculated using the same equation.

The implications of the electrolyte equation can be developed even further. Assume that we have a patient with diabetic ketoacidosis whose fluid balance and glucose concentration has to be corrected. We assume that ECF is low and that circulatory shock can develop if we administer too much insulin. When glucose enters the cells, the osmotic forces drive fluid in along with it, causing a positive ∆ICF (and thus cause cellular edema). The therapy must therefore be cautious while still being effective.

By combining two electrolytes, we can actually estimate both ECF at baseline and the changes in ECF and ICF that accompany an infusion of fluid (2 equations, 2 unknowns). Fluid therapy in ketoacidosis usually begins with 1 L of isotonic saline; therefore, the most suitable electrolytes to monitor are sodium and chloride. If we create two equations, one with sodium and the other with chloride, they should both indicate the same ∆ICF. For this purpose, one may use the same equation as the one derived above for the calculation of ∆ICF based on changes in serum sodium. Based on this equivalence, we combine the equations and mathematically re-arrange them to have ECF on one side of the equation:

ECF = S-Na(t) × Cl (infused–voided)  –  S-Cl(t)  ×  (infused – voided) NaS-Na × S-Cl(t)  – S-Na(t) × S-Cl(8)

In the next step, ∆ICF is calculated using the equation shown above, and ∆ECF is simply the difference between infused volume, urinary excretion and ∆ICF.

The following examples illustrate how this ‘electrolyte maths’ can be applied.

This patient is expected to have an ECF volume amounting to 20% of his body weight, which is 19 L. Isotonic saline contains 152 mmol/L of both sodium and chloride. Insertion of the available data into the equation shows that ECF was only 14.4 L, which is consistent with severe dehydration. All of the infused fluid volume ended up in the ECF space and, worse yet, 0.78 L of the infused fluid entered the ICF, which means that the severe extracellular dehydration was somewhat worsened, despite the infusion of saline (1 L given, of which 0.30 L became urine and 0.64 L became ∆ICF). Our conclusion should be that the insulin was administered somewhat too quickly and should be slowed down to prevent circulatory shock due to depletion of the ECF volume. Another way to handle the situation would be to intensify the fluid treatment, perhaps by doubling the rate of infusion of the crystalloid electrolyte fluid for 2–3 hours, after which the situation should be re-assessed.

The ECF is now 16.3 L and the ∆ICF response to the infusion +0.15 L. Some extracellular dehydration still remains, but no hydration of the ECF was achieved during this fluid challenge. There is still room for more resuscitation fluid.

To summarize, the above concepts can greatly help the clinician personalize care of the patient. Unfortunately, evaluations of the fluid balance based on changes in Hb and electrolyte concentrations are grossly underused in clinical medicine, even though the data required are often on hand for other purposes: the changes in body fluid volumes that they reflect are often simply not appreciated as such.