Pharmacokinetics of intravenous fluid therapy

Robert G. Hahn, professor, Karolinska institute, Stockholm, Sweden

 

Administration of fluid by intravenous infusion is a cornerstone in the treatment of sick patients in hospital. How the infusion technique is mastered, however, has long been a matter of personal experience and rules-of-thumb.

A central concept when infusing fluid is the "volume effect", which is how much the plasma volume increases in response to the infusion. One approach to quantify the volume effect is by using an isotope, such as radioiodine-labeled human serum albumin, to measure the plasma volume before and after the infusion [1]. The difference in plasma volume before and after the infusion is then divided by the amount of infused fluid to obtain 0.2-0.3 for normal saline [1,2] and 0.8 for 4% albumin [3,4]. One problem connected with such calculations is that the distribution of a crystalloid fluid changes quite rapidly and that isotopes require approximately 15-30 minutes of stable fluid distribution to correctly indicate the blood volume. Furthermore, when measuring before and after infusion, only two points of measurement are obtained, which does not give a dynamic profile of how the fluid is distributed in the blood volume. More sophisticated methods for measuring blood volume now exist, such as the use of indocyanine green, but a dynamic picture is still be difficult to provide.

Another way to assess the efficiency of fluid therapy is to use physiological endpoints. When blood loss can be assessed the amount of fluid required to restore impaired physiological parameters to a normal state represents the efficacy of the fluid. Examples of such parameters are cardiac output and pulmonary artery pressure [5]. Only a single figure for the fluid to restore an altered hemodynamic situation is then obtained.

A third method to quantify the volume effect of an infused fluid is to measure the corresponding change in blood hemoglobin (Hb) concentration following infusion. A common way to use such data is to assume that the Hb dilution is equivalent to the plasma volume expansion [6]. The strength of analyzing Hb, however, is that it can be elaborated upon. The first true pharmacokinetic method for infusion fluid, called volume kinetics, is based on repeated measurements of Hb over time. The model was developed by my group in 1997 [7] and has since then been the subject of approximately 30 studies.

Certain features differ between volume kinetics and conventional pharmacokinetics. Infusing fluid provides water to the blood, which already contains 85% of water. To quantify the fluid load, we therefore use the dilution of the plasma which can be obtained from the change in Hb after correcting for the baseline hematocrit. The plasma dilution, which has no unit, is measured every 5-10 min during and after infusing fluid. The model that is fitted to the 25 and 40 measurements of the plasma dilution assumes that infused fluid is distributed between one, two or three expandable body fluid spaces (V₁, V₂, and V₃). In contrast to conventional kinetics, the sizes of these spaces do become larger when the substance is administered, and are then termed v₁, v₂, and v₃. Exchange of fluid between the first two spaces occurs by virtue of an equilibration process which is proportional by a constant k_t to the difference in dilution between them. Exchange with the most remote space occurs by osmotic forces. The body always strives to regain the baseline sizes of the expanded body fluid spaces by two mechanisms, which both act of the central body fluid space. The first one (k_b) reflects baseline losses, such as evaporation. In the second, the plasma dilution is proportional to the rate of elimination by a constant k_r.

For all practical purposes, k_r can be regarded as the renal clearance of the infused fluid. This means that the measured urinary exceretion might be used to "check" the output of the curve-fitting procedure which, in the two-volume model shown below, yields estimates of V₁, V₂, k_t, and k_r.  Marked deviations between k_r and the urinary excretion indicate an escape of fluid into deep body spaces, such as the gastrointestinal tract [8]. The excess fluid volumes in V₁, V₂, andV₃ over time can be obtained by multiplying the dilution of these spaces by the model-estimated baseline volumes [9]. Naturally, the excess volume in the functional fluid compartments can be predicted by computer and different infusion types, such as albumin and Ringer´s solution, be compared (figure) [10].

A bi-exponential plasma dilution curve corresponds to the two-volume model (figure) and typically applies when fluid excretion is retarded due to fasting, disease, anesthesia, hemorrhage, or trauma.  In healthy volunteers, k_r sometimes approaches k_t and which means that only V₁ becomes expanded. The three-volume model is applicable when glucose solution or non-isotonic fluids are infused.

The volume of V₁ is usually 3 L, which is very close to the expected size of the plasma volume. V₂  is typically about 2/3 of the interstitial fluid space, which indicates that the entire space does not become expanded by infused crystalloid fluid.

In general, the sizes of the fluid spaces are very stable under various physiological conditions while the rate constants k_r and k_r become more affected. The most remarkable finding in our studies so far is the profound reduction of k_r that develops during anesthesia and surgery. The renal clearance is only between 10% and 20% of the value found in volunteers [11,12] which emphasizes the need for precision when infusing crystalloid fluid under such circumstances. Although volume kinetics can be used to predict the infusion rate and volume required to obtain and manitain a predetermined plasma dilution, it is still unclear which degree of plasma dilution that is assocuated with a favorable outcome. Work is in progress to clarify this issue.

 

REFERENCES