Colloids versus crystalloids in fluid resuscitation: an analysis of randomized controlled trials.
Methods. Mortality rates taken from randomized controlled trials were analyzed to compare colloidal and crystalloidal fluid for resuscitation efforts. We controlled for the underlying pathological process by categorizing subjects into three groups: (1) surgical stress, (2) hypovolemia, and (3) severe pulmonary failure. A cost-effectiveness analysis also was performed.
Results. No statistically significantly differences in mortality rates were found. The cost of each life saved using crystalloids is $45.13, and the cost of each life saved using colloidal solutions is $1493.60.
Conclusions. Because there is no significant mortality-rate advantage to using colloids, and because the cost-effectiveness ratio for crystalloids is much lower than for colloids, it is concluded that crystalloids should always be used in resuscitation efforts.
Key words. Colloids, resuscitation, mortality, costeffectivenss. J Fam Pract 1991; 32:387-390.
Controversy about fluid therapy in resuscitation has existed since the 1960s. The difficulty in predicting the effects of colloidal and crystalloidal fluids may be caused by the fact that fluid behavior at the lung capillary membrane level may vary depending on the patient's particular pathology.
Many laboratory and clinical studies addressing this topic have been reported in the past thee decades. Some of them have shown advantages for crystalloids, [1-8] some have demostrated advantages for colloids, [9-15] and still others have demonstrated no difference between the two. [16,17]
In evaluating the effectiveness of fluid resuscitation, several authors have studied a wide variety of physiologic variables and different physiologic endpoints. Surprisingly, mortality has not been the primary dependent measure. The sole exception is the work of Velanovich.  Velanovich used a meta-analytic approach and found that the overall treatment effect, when data from several studies were pooled, showed a 5.7% relative difference in mortality rates in favor of crystalloid therapy. He pooled data from randomized the nonrandomized clinical trials. His decision to include the nonrandomized study of Shoemaker et al  played a large role in his results.
The present article describes a data compilation from several randomized controlled clinical trials (Table 1), an analysis of mortality rates of different patient populations, and a cost-effectiveness analysis of different types of fluid therapy in resuscitation.
Material and Methods
To determine whether there is a difference between colloid and crystalloid fluid resuscitation, data from rigorously designed studies, ie, randomized clinical trials, [16,19-24] were pooled. Thereafter, mortality rates were analyzed by separating trials according to the severity of the underlying process of the patients. The following three groups were used:
1. Patients with surgical stress only (major operative procedures on the abdominal aorta). Patients in this group had no hypovolemic shock, no pulmonary failure, and no sepsis.
2. Injured patients, subject to surgical stress and presenting with hypovolemia.
[TABULAR DATA OMITTE]
3. Patients with established severe pulmonary failure (intrapulmonary shunt of greater than 20% and a roentgenogram of the chest demonstrating interstitial and intra-alveolar edema). Fluid therapy in these patients was administered after the diagnosis of pulmonary failure was made.
In all studies, subjects were randomized into groups receiving either colloidal or crystalloidal treatment. The proportion of subjects receiving colloidal treatment, however, was not exactly .5 (Table 1). This is because in two studies, the probability of a subject receiving colloidal therapy was not .5. In the study by Boutros et al  there were 26 subjects randomized (with equal probability) into three groups. Two of these groups received crystalloid solutions (one group received normal saline, the other, Ringer's lactate solution). In the same way, Rackow et al  randomized subjects into three groups. Two of these groups received colloids (one received albumin, the other, hetastarch).
Chi-square tests were used to determine whether differences in mortality rates among colloid and crystalloid subgroups were statistically significant.
Cost-effectiveness analysis was performed using a commercially available computer program called SMLTREE. 
Fluid cost estimates were obtained locally. At Oklahoma Memorial Hospital, the cost of 1 g of albumin was $5.00 and the cost of 1 L of Ringer's lactate solution was $5.95.
When all the trials were pooled, the mortality rate for crystalloid therapy was 13.4%, and the mortality rate for colloid therapy was 21.25% (P was not significant at .01 level). Thereafter, mortality rates were compared between colloid and crystalloid therapy subgroups. This method of comparison was done for each of the three patient groups described previously (surgical stress, hypovolemic status, and pulmonary failure). The results are displayed in Figure 1. Significant difference were not found between colloid and crysalloid therapy in any of the patient groups.
Furthermore, in order to test whether the severity of the underlying process played an important role, group 2 was subcategorized into group 2.1, hypovolemic status, or shock without sepsis, or pulmonary failure; and group 2.2, serious hypovolemic shock with complications such as sepsis or pulmonary failure. Fluid therapy in these patients was administered before complications developed.
Afterward, colloid and crystalloid subgroups were analyzed for these patients. The results are displayed in Figure 2. There were no significant differences between the colloidal mortality rate and the crystalloidal mortality rate for either group 2.1 or group 2.2.
For clinical trial studies in which the criterion for the end of the resuscitation period was clearly stated, data for our cost-effectiveness analysis were pooled (Table 2). The per-patient intake of fluids in the resuscitation period averaged 4.70 L for the colloid group and 6.57 L for the crystalloid group. The per-patient average amount of albumin used in the colloid group was 229.77 g. Therefore, the per-patient crystalloidal therapy cost was estimated to have been $39.09; the estimate for the per-patient colloidal therapy cost was $1176.21. For crystalloidal therapy, the cost-effectiveness ratio is $45.13 per life saved; for colloidal therapy, the cost-effectiveness ratio is $1493.60 per life saved.
In resuscitation research, study results are frequently dependent on the design and conditions of the experiment and the underlying pathological process. Increased microvascular permeability to albumin and protein macromolecules has been extensively documented and underlies the concern theat colloidal solutions may be harmful or ineffective in clinical disorders.
Robin et al  demonstrated that in septic patients the concentration of solutes (such as albumin, globulin, and dextran) was similar in both pulmonary edema fluid and plasma. The authors concluded that permeability to macromolecules was increased in sepsis. Lung-lymph
[TABULAR DATA OMITTED]
models have thus been used to show increased microvascular permeability in laboratory animals following intravenous injection of Eschevichia coli endotoxin  and live Pseudomonas bacteria.  Haupt concluded that "the concern that resuscitations with colloidal fluids might be detrimental in conditions associated with increased microvascular permeability to macromolecules is valid." 
On the other hand, when microvascular permeability was normal, Tranbaugh et al  found no increment of extravascular lung water after hemodilution with crystalloidal solutions in severe hypovolemic shock. They concluded that lung contusion or sepsis are the primary determinants of interstitial fluid accumulation.
Our results are consistent with these trials. In patients presumably without altered microvascular perme-ability (group 1 and group 2.1), outcomes are independent of the kind of fluids used. The overall mortality rate in group 1 was higher than the mortality rate in group 2.1, and this difference could be explained by the average age in group 1 (mean age = 60) being higher than the average age in group 2.1 (mean age = 30).
In cases with definitely altered microvascular perme-ability (group 3), or in patients who were developing microvascular permeability alterations (group 2.2), colloidal treatment groups showed even higher mortality rates than crystalloidal treatment groups, but the difference was not enough to reach statistical significance. Therefore, colloids failed to show superiority over crystalloids, regardless of the severity of the underlying process.
Colloids could have a partial extravascular distribution that increases pulmonary edema (capillary leak syndrome). This could explain the lack of expected improvement for this type of therapy. In addition, there are clinical data indicating deleterious effects of albumin resuscitation on renal function,  cardiac function,  and coagulation.  There is also a small but definite incidence of serious allergic reactions  when colloids are used. Our results encourage the use of crystalloids in all cases. Furthermore, cost-effectiveness analysis showed that using crystalloids cost $45.13 per life saved as compared with using colloidal solutions at a cost of $1493.60 per life saved.
In choosing among feasible alternatives, the physician and the health policy analyst should consider some means of evaluation that compares the relative cost and effectiveness of these competing strategies. One appropriate measure is the cost to effectiveness ratio, which, in this case, could express how many additional patients' survival can be "bought" for each additional dollar expended. Since colloids cost more than crystalloids and are no more effective, the colloid strategy in fluid resuscitation can be rejected from further consideration.
 Weaver DW, Ledgerwood AM, Lucas CE, et al. Pulmonary effects of albumin resuscitation for severe hypovolemic shock. Arch Surg 1978; 113:387-92.
 Collins JA, Braitberg A, Butcher HR. Changes in lung and body weight and lung water content in rats treated for hemorrhage with various fluids. Surgery 1973; 73:401-11.
 Dahn MS, Lucas CE, Ledgerwood AM, Higgins RF. Negative inotropic effect of albumin resuscitation for shock. Surgery 1979; 86:235-41.
 Moss GS, Siegel DC, Cochin A, Fresquez V. Effects of saline and colloid solutions on pulmonary function in hemorrhagic shock. Surg Gynecol Obster 1971; 33:53-8.
 Weedn RJ, Cook JH, McElreath RL, Greenfield LJ. Wet lung syndrone after crystalloid and colloid repletion. Curr Top Surg Res 1970; 2:335-9.
 Schloerb PR, Hunt PT, Plummer JA, Cage GK. Pulmonary edema after replacement of blood loss by electrolyte solutions. Surg Gynecol Obster 1972; 135:893-6.
 Gaisford WD, Pandey N, Jensen CG. Pulmonary changes in treated hemorrhagic shock. Am J Surg 1972; 24:738-43.
 Litwin MS. Comparison of effects of dextran 70 and dextran 40 on postoperative animals. Ann Surg 1972; 71:295-9.
 Skillman JJ, Restall DS, Salzman EW. Randomized trial of albumin vs electrolyte solutions during aortic operations. Surgery 1975; 78:291-303.
 Shoemaker WC, Schluchter M, Hopkins JA, et al. Comparison of the relative effectiveness of colloids and crystalloids in emergency resuscitation. Am J Surg 1981; 142:73-84.
 Nees JE, Hauser CJ, Shippy C, et al. Comparison of cardiorespiratory effects of crystalline hemoglobin, whole blood, albumin and Ringer's lactate in the resuscitation of hemorrhagic shock in dogs. Surgery 1978; 83:639-47.
 Hauser CJ, Shoemaker WC. Albumin resuscitation for shock. Surgery 1980; 88:183-4.
 Dawidson I, Eriksson B, Gelin LE, et al. Oxygen consumption and recovery from surgical shock in rats: a study on the efficacy of different plasma substitutes. Crit Care Med 1979; 7:460-5.
 Dawidson I, Gelin LE, Haglind E. Plasma volume, intravascular protein content, hemodynamic and oxygen transport changes during intestinal shock in dogs. Crit Care Med 1980; 8:73-80.
 Dawidson I, Gelin LE, Hedman L, Soderberg R. Hemodilution and recovery from experimental intestinal shock in rats. Crit Care Med 1981; 9:42-6.
 Virgilio RW, Rice CL, Smith DE, et al. Crystalloid vs colloid resuscitation: is one better? Surgery 1979; 85:129-38.
 Moss GS, Lowe RJ, Tilek J, Levine HD. Colloid or crystalloid in the resuscitation of hemorrhagic shock; a controlled clinical trial. Surgery 1981; 89:434-8.
 Valanovich V. Crystalloid versus colloid fluid resuscitation: a meta-analysis of mortality. Surgery 1989; 105:65-71.
 Boutros AR, Ruess R, Olson L, et al. Comparison of hemodynamic, pulmonary, and renal effects of use of three types of fluids after major surgical procedures on the abdominal aorta. Crit Care Med 1979; 7:9-13.
 Lowe RJ, Moss GS, Jilek J, Levine HD. Crystalloid versus colloid in the etiology of pulmonary failure after trauma-a randomized trial in man. Crit Care Med 1979; 7:107-12.
 Moss GS, Lowe RJ, Jilek J, Levine HD. Colloid or crystalloid in the resuscitation of hemorrhagic shock: a controlled clinical trial. Surgery 1981; 89:434-9.
 Rackow EC, Falk JL, Fein IA, et al. Fluid resuscitation in circulatory shock: a comparison of the cardiorespiratory effects of albumin, hetastarch, and saline solutions in patients with hypovolemic and septic shock. Crit Care Med 1983; 11:839-50.
 Lucas CE, Weaver D, Higgins RF, et al. Effects of albumin versus non-albumin resuscitation on plasma volume and renal excretory function. J Trauma 1978; 18:564-70.
 Metildi LA, Shackford SR, Virgilio RW, Peter RM. Crystalloid versus colloid in fluid resuscitation of patients with severe pulmonary insufficiency. Surg Gynecol Obstet 1984; 158:207-12.
 Hollenberg J. SMLTREE. Version 2.2. Roslyn, NY:Jim Hollenberg, 1987.
 Robin ED, Carey LC, Grenvik A, et al. Capillary leak syndrome with pulmonary edema. Arch Intern Med 1972; 130:66-71.
 Brighman KL, Bowers RE, Hayes J. Increased sheep lung vascular permeability caused by Escherichia coli endotoxin. Circ Res 1979; 45:292-7.
 Brighman KL, Woolverton WC, Blake LH, et al. Increased sheep lung vascular permeability caused by Pseudomonas bacteria. J Clin Invest 1974; 54:792-804.
 Haupt MT. Colloidal and crystalloidal fluid resuscitation in shock associated with increased capillary permeability. Curr Stud Hematol Blood Transfus 1986; 53:86-100.
 Tranbaugh RF, Elings VB, Christensens J, Lewis FR. Determinants of pulmonary interstitial fluid accumulation after trauma. J Trauma 1982; 22:820-5.
 Lucas CE. Renal considerations in the injured patient. Surg Clin North Am 1982; 62:133-48.
 Lucas CE, Ledgerwood AM, Mammon EF. Altered coagulation protecin content after varied resuscitation regimens. J Trauma 1982; 22:1-5.
 Ring J, Messmer K. Incidence and severity of anaphylactoid reactions to colloid volume substitutes. Lancet 1977; 1:466-9.
|Printer friendly Cite/link Email Feedback|
|Author:||Bisonni, Roberto S.; Holtgrave, David R.; Lawler, Frank; Marley, Daniel S.|
|Publication:||Journal of Family Practice|
|Date:||Apr 1, 1991|
|Previous Article:||Accuracy and precision of HDL cholesterol measurements using an office chemistry analyzer.|
|Next Article:||Relationship between panic attacks and health locus of control.|