Category Archives: Colloid and Crystalloid



In 1861, Thomas Graham’s investigations on diffusion led him to classify substances as crystalloids or colloids based on their ability to diffuse through a parchment membrane. Crystalloids passed readily through the membrane, whereas colloids (from the Greek word for glue) did not. Intravenous fluids are similarly classified based on their ability to pass through barriers separating body fluid compartments, particularly the one between intravascular and extravascular (interstitial) fluid compartments. This chapter describes the salient features of crystalloid and colloid fluids, both individually and as a group. This is a must-know topic in the care of hospitalized patients, and several reviews are included at the end of the chapter to supplement the text.


The principal component of crystalloid fluids is the inorganic salt sodium chloride (NaCl). Sodium is the most abundant solute in the extracellular fluids, and it is distributed uniformly throughout the extracellular space. Because 75 to 80% of the extracellular fluids are located in the extravascular (interstitial) space, a similar proportion of the total body sodium is in the interstitial fluids. Exogenously administered sodium follows the same distribution, so 75 to 80% of the volume of sodium-based intravenous fluids are distributed in the interstitial space. This means that the predominant effect of volume resuscitation with crystalloid fluids is to expand the interstitial volume rather than the plasma volume.


As indicated by the horizontal bar that is second from the top, infusion of 1 L of 0.9% sodium chloride (isotonic saline) adds 275 mL to the plasma volume and 825 mL to the interstitial volume. Note that the total volume expansion (1100 mL) is slightly greater than the infused volume. This is the result of a fluid shift from the intracellular to extracellular space, which occurs because isotonic saline is actually hypertonic to the extracellular fluids.


The prototype crystalloid fluid is 0.9% sodium chloride (NaCl), also called isotonic saline or normal saline. The latter term is inappropriate because a one normal (1 N) NaCl solution contains 58 g NaCl per liter (the combined molecular weights of sodium and chloride), whereas isotonic (0.9%) NaCl contains only 9 g NaCl per liter.


The pH of isotonic saline is also considerably lower than the plasma pH. These differences are rarely of any clinical significance.


The chloride content of isotonic saline is particularly high relative to plasma (154 mEq/L versus 103 mEq/L, respectively), so hyperchloremic metabolic acidosis is a potential risk with large-volume isotonic saline resuscitation. Hyperchloremia has been reported, but acidosis is rare.


Ringer’s solution was introduced in 1880 by Sydney Ringer, a British physician and research investigator who studied mechanisms of cardiac contraction. The solution was designed to promote the contraction of isolated frog hearts, and contained calcium and potassium in a sodium chloride diluent. In the 1930s, an American pediatrician named Alexis Hartmann proposed the addition of sodium lactate buffer to Ringer’s solution for the treatment of metabolic acidoses. The lactated Ringer’s solution, also known as Hartmann’s solution, gradually gained in popularity and eventually replaced the standard Ringer’s solution for routine intravenous therapy.


Lactated Ringer’s solution contains potassium and calcium in concentrations that approximate the free (ionic) concentrations in plasma. The addition of these cations requires a reduction in sodium concentration for electrical neutrality, so lactated Ringer’s solution has less sodium than isotonic saline. The addition of lactate (28 mEq/L) similarly requires a reduction in chloride concentration, and the chloride in lactated Ringer’s more closely approximates plasma chloride levels than does isotonic saline.

Despite the differences in composition, there is no evidence that lactated Ringer’s provides any benefit over isotonic saline. Furthermore, there is no evidence that the lactate in Ringer’s solution provides any buffer effect.


The calcium in lactated Ringer’s can bind to certain drugs and reduce their bioavailability and efficacy. Of particular note is calcium binding to the citrated anticoagulant in blood products. This can inactivate the anticoagulant and promote the formation of clots in donor blood. For this reason, lactated Ringer’s solution is contraindicated as a diluent for blood transfusions.



The major feature of these solutions is the added buffer capacity, which gives them a pH that is equivalent to that of plasma. An additional feature is the addition of magnesium, which may provide some benefit in light of the high incidence of magnesium depletion in hospitalized patients.


Magnesium administration can promote hypermagnesemia in renal insufficiency and can counteract compensatory vasoconstriction and promote hypotension in low flow states.


Dextrose is a common additive in intravenous solutions, for reasons that are unclear. A 5% dextrose-in-water solution is not an effective volume expander. The use of 5% dextrose solutions was originally intended to supply nonprotein calories and thus provide a protein-sparing effect. However, total enteral and parenteral nutrition is now the standard of care for providing daily energy requirements, and the use of 5% dextrose solutions to provide calories is obsolete.


A 5% dextrose solution (50 g dextrose per liter) provides 170 kcal per liter (3.4 kcal/g dextrose).


The addition of dextrose to intravenous fluids increases osmolarity (50 g of dextrose adds 278 mosm to an intravenous fluid) and creates a hypertonic infusion when 5% dextrose is added to lactated Ringer’s solution (525 mOsm/L) or isotonic saline (560 mOsm/L). If glucose use is impaired (as is common in critically ill patients), the infused glucose accumulates and creates an undesirable osmotic force that can promote cell dehydration.

Other undesirable effects of glucose infusions in critically ill patients include enhanced CO2 production (which can be a burden in ventilator-dependent patients), enhanced lactate production, and aggravation of ischemic brain injury.

Lactate Production

The proportion of a glucose load that contributes to lactate formation can increase from 5% in healthy subjects to 85% in critically ill patients. This can produce an increase in circulating lactate levels, even when infusing 5% dextrose solutions. Patients undergoing abdominal aortic aneurysm surgery were given either a Ringer’s solution or a 5% dextrose solution intraoperatively to maintain normal cardiac filling pressures. As shown, the 5% dextrose infusions were associated with a 125% increase in arterial lactate levels (from 1.85 to 4.15 mmol/L). Thus, in patients with circulatory compromise, abnormal glucose metabolism can transform glucose from a source of useful energy to a source of toxin production.

The disadvantages noted above, when combined with a lack of documented benefit, favor the recommendation that the routine use of 5% dextrose infusions be abandoned in critically ill patients.


As mentioned earlier, colloids are large molecules that do not pass across diffusional barriers as readily as crystalloids. Colloid fluids infused into the vascular space therefore have a greater tendency to stay put and enhance the plasma volume than do crystalloid fluids. The colloid fluid in this case is 5% albumin, and as demonstrated, the plasma expansion with this colloid fluid is nearly twice that produced by an equivalent volume of isotonic saline (500 mL versus 275 mL, respectively). This is the principal benefit of colloid fluid resuscitation: more effective resuscitation of plasma volume than that produced by crystalloid fluids. Much of this potency is related to the colloid osmotic pressure exerted by each fluid.


Large solute molecules that do not move freely across barriers separating fluid compartments create a force that draws water into the large solute compartment. This force opposes the hydrostatic pressure (which favors the movement of water out of a fluid compartment) and is called the colloid osmotic pressure (COP) or oncotic pressure. As would be expected, the ability of each fluid to expand the plasma volume is directly related to the COP; that is, the higher the COP, the greater the volume expansion. If the COP of a colloid fluid is greater than the COP of plasma (i.e., greater than 25 mm Hg), the plasma volume expansion exceeds the infused volume. The 25% albumin solution, which has a COP of 70 mm Hg and a plasma volume expansion that is 4 to 5 times the infused volume.


Albumin is a transport protein that is responsible for 75% of the oncotic pressure of plasma. Heat-treated preparations of human serum albumin are commercially available in a 5% solution (50 g/L) and a 25% solution (250 g/L) in an isotonic saline diluent. The 25% solution is given in small volumes (50 to 100 mL) and because the accompanying sodium load is small, 25% albumin is also called salt-poor albumin.


A 5% albumin solution (50 g/L or 5 g/dL) has a COP of 20 mm Hg and thus is similar in oncotic activity to plasma. Approximately half of the infused volume of 5% albumin stays in the vascular space. The oncotic effects of albumin last 12 to 18 hours.

The 25% albumin solution has a COP of 70 mm Hg and expands the plasma volume by 4 to 5 times the volume infused. Thus, infusion of 100 mL of 25% albumin can increase the plasma volume 400 to 500 mL. This plasma volume expansion occurs at the expense of the interstitial fluid volume, so 25% albumin should not be used for volume resuscitation in hypovolemia. It is intended for shifting fluid from the interstitial space to the vascular space in hypoproteinemic conditions, although the wisdom of this application is questionable.


Because albumin preparations are heat-treated, there is no risk of viral transmission (including human immunodeficiency virus). Allergic reactions are rare, and although coagulopathies can occur, most are dilutional and not accompanied by bleeding.


Hetastarch is a synthetic colloid available as a 6% solution in isotonic saline. It contains amylopectin molecules that vary in size from a few hundred to over a million daltons. The average molecular weight of the starch molecules is equivalent to that of albumin, and the colloid effects are equivalent to those of 5% albumin. The main advantage of hetastarch over albumin is its lower cost.


Hetastarch is slightly more potent than 5% albumin as a colloid. It has a higher COP than 5% albumin (30 versus 20 mm Hg, respectively) and causes a greater plasma volume expansion (up to 30% greater than the infused volume). It also has a long elimination half-life (17 days), but this is misleading because the oncotic effects of hetastarch disappear within 24 hours.


Hetastarch molecules are constantly cleaved by amylase enzymes in the bloodstream before their clearance by the kidneys. Serum amylase levels are often elevated (2 to 3 times above normal levels) for the first few days after hetastarch infusion, and return to normal at 5 to 7 days after fluid therapy. This hyperamylasemia should not be mistaken for early pancreatitis. Serum lipase levels remain normal, which is an important distinguishing feature.

Anaphylactic reactions to hetastarch are decidedly rare (incidence as low as 0.0004%). Laboratory test coagulopathy (prolonged partial thromboplastin time from an interaction with Factor VIII) can occur, but is not accompanied by bleeding. Coagulopathy claims have dogged hetastarch for years, without evidence of hetastarch-induced bleeding.


Pentastarch is a low-molecular-weight-derivative of hetastarch that is available as a 10% solution in isotonic saline. Although it is not currently approved for clinical use in the United States, there is considerable evidence indicating that pentastarch is an effective and safe plasma volume expander.


Pentastarch contains smaller but more numerous starch molecules than hetastarch, and thus has a higher colloid osmotic pressure. It is more effective as a volume expander than hetastarch, and can increase plasma volume by 1.5 times the infusion volume. The oncotic effects dissipate after 12 hours. Pentastarch shows less of a tendency to interact with coagulation proteins than hetastarch, but the significance of this tendency is unclear.


The dextrans are glucose polymers produced by a bacterium (Leuconostoc) incubated in a sucrose medium. First introduced in the 1940s, these colloids are not popular (at least in the United States) because of the perceived risk of adverse reactions. The two most common dextran preparations are 10% dextran-40 and 6% dextran-70, both diluted in isotonic saline.


Both dextran preparations are hyperoncotic to plasma (COP = 40 mm Hg). Dextran-40 causes a larger increase in plasma volume than dextran-70, but the effects last only a few hours. Dextran-70 is the preferred preparation because of its prolonged action.


Dextrans produce a dose-related bleeding tendency by inhibiting platelet aggregation, reducing activation of Factor VIII, and promoting fibrinolysis. The hemostatic defects are minimized by limiting the daily dextran dose to 20 mL/kg.

Anaphylactic reactions were originally reported in as many as 5% of patients receiving dextran infusions. However, this has improved considerably in the last 20 years because of improvements in antigen detection and desensitization and improvements in preparation purity. The current incidence of anaphylaxis is 0.032%.

Dextrans coat the surface of red blood cells and can interfere with the ability to cross-match blood. Red cell preparations must be washed to eliminate this problem. Dextrans also increase the erythrocyte sedimentation rate as a result of their interactions with red blood cells.

Finally, dextrans have been implicated as a cause of acute renal failure. The proposed mechanism is a hyperoncotic state with reduced filtration pressure. However, this mechanism is unproven, and renal failure occurs only rarely in association with dextran infusions.


There is considerable disagreement about the most appropriate fluid for volume resuscitation in critically ill patients. The following is a brief description of the issues involved in the colloid-crystalloid debate.


Because crystalloid fluids fill primarily the interstitial space, these fluids are not useful for filling the vascular space. The early popularity of crystalloid fluid resuscitation in hypovolemia stems from two observations made about 40 years ago. The first is the response to mild hemorrhage, which involves a shift of fluid from the interstitial space to the vascular space. The second observation stems from studies in an animal model of hemorrhagic shock, where survival was much improved if a crystalloid fluid was given along with reinfusion of the shed blood volume. The combination of these two observations has been interpreted as indicating that the major consequence of hemorrhage is an interstitial fluid deficit, and that replacement of interstitial fluid with crystalloid fluids is important for survival.


The interstitial fluid deficit is predominant only when blood loss is mild (less than 15% of the blood volume), and in this situation, no volume resuscitation is necessary (because the body is capable of fully compensating for the loss of blood volume). When blood loss is more severe, the priority is to keep the vascular space filled and thereby support the cardiac output. Because colloid fluids are about three times more potent than crystalloid fluids for increasing vascular volume and supporting the cardiac output, colloid fluids are more effective than crystalloid fluids for volume resuscitation in moderate to severe blood loss. Crystalloid resuscitation can achieve the same endpoint as colloid resuscitation, but larger volumes of crystalloid fluid (about three times the volume of colloid fluids) must be used. This latter approach is less efficient, yet it is the one favored by crystalloid users.


Despite the superiority of colloid fluids for expanding plasma volume, colloid fluid resuscitation does not confer a higher survival rate in patients with hypovolemic shock. This lack of improved outcomes is a major rallying point for crystalloid users, but it does not negate the fact that colloid fluids are more effective for maintaining blood volume in patients who are actively bleeding.


The biggest disadvantage of colloid resuscitation is the higher cost of colloid fluids. Using equivalent volumes of 250 mL for colloid fluids and 1000 mL for crystalloid fluids, the cost of colloid resuscitation is three times as high (if hetastarch is used) to six times as high (if albumin is used) than volume resuscitation with isotonic saline.


The risk of edema has been used to discredit each type of fluid. Because crystalloid fluids distribute primarily in the interstitial space, edema is an expected feature of crystalloid fluid resuscitation. However, edema is also a risk with colloid fluid resuscitation. This is particularly true with albumin-containing fluids; even though albumin is the principal oncotic force in plasma, over half of the albumin in the human body is in the interstitial fluid. Therefore, a large proportion of infused albumin eventually finds its way into the interstitial fluid and promotes edema. Furthermore, this egress of albumin from the bloodstream is magnified when capillary permeability is disrupted, which is a common occurrence in critically ill patients. Despite this risk, troublesome edema (e.g., pulmonary edema) is not common with either type of fluid resuscitation when capillary hydrostatic pressure is not excessive.


The following analogy helped me resolve the colloid-crystalloid conundrum. Assume that the goal is to recreate the performance of crystalloid and colloid fluids in expanding the plasma volume by filling a bucket. Because the volume of crystalloid fluids needed to expand the plasma volume (fill the bucket) is three times larger than the volume of colloid fluid that fills the bucket, holes will need to be punched in the bucket while it is filled with crystalloid fluids (to allow the extra fluid to escape). Therefore, the question is this: If the goal is to fill a bucket with fluid, do you want to punch holes in the bucket (and make the bucket more difficult to fill)? Seen in this light, it is more efficient to use colloid fluid resuscitation to expand the plasma volume.


An interesting approach to volume resuscitation that has stalled in recent years is the use of small-volume hypertonic saline solutions. A 7.5% sodium chloride solution is given either in a fixed volume of 250 mL or in a volume of 4 mL/kg. The volume increments in both fluid compartments are similar to those produced by 1 L of 5% albumin. Thus, hypertonic saline resuscitation can produce equivalent volume expansion to colloid fluids, but at one-fourth the infused volume. Note that the total volume expansion (1235 mL) produced by 7.5% saline is far greater than the infused volume (250 mL). The additional volume comes from intracellular fluid that moves out of cells and into the extracellular space. This movement of intracellular fluid points to one of the feared complications of hypertonic resuscitation: cell dehydration.


Since the first report of its successful use in 1980, hypertonic saline has been shown repeatedly (but not unanimously) to be safe and effective in the early resuscitation of hypovolemia. However, there is little evidence that hypertonic resuscitation is superior to standard volume resuscitation. Hypertonic resuscitation seems best suited for prehospital resuscitation in cases of trauma, but studies in trauma resuscitation fail to document a clear benefit with this approach in most patients. Select subgroups of patients (e.g., those with penetrating truncal injuries who required surgery) may benefit from hypertonic resuscitation, but these subgroups are small. Thus, after over 15 years of evaluating this technique, hypertonic resuscitation has few advocates.