Secondary Logo

Journal Logo

General Article

Maximum Blood Savings by Acute Normovolemic Hemodilution

Feldman, Jeffrey M. MD, MSE; Roth, Jonathan V. MD; Bjoraker, David G. MD

Author Information
  • Free


Acute normovolemic hemodilution (ANH) is advocated as a technique for reducing homologous blood transfusion requirements during surgery [1-3]. The process entails collecting the patient's blood immediately before surgery with concurrent fluid infusion to maintain intravascular volume constant. The purpose is to reduce the red cell mass (RCM) lost during surgery by reducing the hematocrit of the lost blood. Blood collected during ANH is reinfused as needed to replace surgical blood loss (SBL). If a minimum safe hematocrit (Hm) is defined for a patient, it should be possible to lose more blood during surgery if ANH is used without falling below that minimum hematocrit than if ANH is not used. The difference between the SBL possible when ANH is used and the SBL possible without ANH is the net savings associated with the technique.

For any given patient, the total RCM saved by ANH, and the extent to which the technique may reduce the need for homologous blood transfusion, depends upon three factors: 1) the initial hematocrit (Hi); 2) the amount of blood removed during the hemodilution process; and 3) the amount of SBL. Clinical studies of the efficacy of ANH for reducing homologous blood transfusion have not examined the relative impact of these factors on RCM savings. The optimal method for performing ANH is not apparent from existing studies, since recommendations for how much blood to remove, and when the saved blood should be replaced, are quite varied Table 1.

Table 1:
Differences in Techniques for Performing Acute Normovolemic Hemodilution Between Representative Clinical Studies

We present a mathematic model of ANH which calculates the maximum possible RCM savings using ANH given the patient's weight, Hi, and Hm. The potential impact of this technique on the need for homologous blood transfusion is discussed.


If one identifies the minimum safe hematocrit (Hm) desirable for a given patient, the maximum allowable surgical blood loss when losses are replaced with acellular fluid to maintain euvolemia (BLs) can be found by the following equation [4-5]: Equation 1 where EBV is the estimated blood volume (defined here as 70 mL/kg) and Hi is the patient's initial hematocrit. All hematocrits are expressed as fractions (e.g., 0.40) rather than percent (e.g., 40%) to eliminate the need for conversions in the calculations and to simplify the presentation of the equations. Figure 2

Figure 2

If ANH is used, the maximum red blood cell (RBC) savings will be obtained by diluting the patient's concentration of red cells to the minimum safe hematocrit prior to the start of surgery. This will remove the maximum amount of RCM for transfusion. Since the process of ANH is essentially controlled hemorrhage, the calculation of BLS in Equation 1 applies to the hemodilution process as well, so that the volume of blood removed during ANH to the Hm is the same as BLS. Assuming that each unit removed by hemodilution has a volume of 450 mL (the actual volume of a unit will vary somewhat since completion of collection into a typical bag is based upon weight not volume), the number of units that need to be removed to hemodilute to the minimum safe hematocrit (ANHu) can be found by, Equation 2 Since the model assumes hemodilution to the Hm prior to surgery, retransfusion of blood obtained by hemodilution must begin when SBL begins. The RCM available for retransfusion after ANH (RCMH) can be calculated from the patient's Hi and the final hematocrit after hemodilution (Hm), Equation 3 The maximum SBL that is possible when ANH is used without falling below Hm (BLH) is found by assuming that all the blood removed during ANH is returned to the patient at a rate sufficient to maintain the hematocrit at the minimum safe level: Equation 4 If ANH is used, as long as SBL does not exceed BLH, there will not be any need for homologous blood transfusion to maintain a safe hematocrit.

Based upon the underlying rationale for ANH, BLH should exceed BL (S) Figure 1. The difference between these two quantities is the additional amount of blood the patient can lose before homologous blood transfusion becomes necessary. This difference can be termed the incremental surgical blood loss (BLI) possible when using ANH, Equation 5 BLI can be expressed in terms of red cell mass by Equation 6 where RCMI is the red cell mass that would have to be administered using homologous blood to maintain the Hm if ANH is not used and blood loss equals BLH. RCMI can be expressed in packed red blood cell unit equivalents (PRBCe) by assuming that the volume of a unit of PRBC is 250 mL and that each unit has a hematocrit of 0.60. (These assumptions are reasonable, but in reality there is some variation of the hematocrit and volume of a given unit of PRBCs.) The RCM per unit of packed cells is therefore 150 mL and PRBCe is, Equation 7

Figure 1:
The path from Points A to B depicts the change in hematocrit when acute normovolemic hemodilution (ANH) is performed to the minimum safe hematocrit (Hm) prior to the start of surgery. Points B to D depict subsequent administration of the blood saved during ANH to maintain Hm during surgery. BLH is the volume of blood that can be lost during surgery when ANH is used before homologous blood is required. The path from A to C depicts the change in hematocrit when surgical blood loss is replaced with acellular fluid. BLS is the maximal blood loss possible without ANH before homologous blood is required. The difference between BLH and BLS is BLI, the incremental blood loss possible when using ANH. Alternatively, BLI is the volume of blood that would have to be replaced using homologous transfusion to maintain Hm if ANH is not used and blood loss equals BLH.

The model assumes that ANH is used for a 70-kg patient with an estimated blood volume of 70 mL/kg (4900 mL). A range of Hi and Hm were evaluated to understand conditions where hemodilution is most likely to be beneficial to the patient, and the degree of hemodilution necessary to achieve that benefit. The model was implemented in Quick BASIC for MS-DOS (Microsoft Corp., Redmond, WA).


The results of the model calculations are presented in Table 2 for a range of Hi s from 0.30 to 0.50 with ANH performed to minimum hematocrits from 0.30 to 0.15. The line in Table 2 indicated by the arrow will be used to illustrate how to interpret these data. Given an Hi of 0.40, if the Hm is assumed to be 0.25, ANH is not necessary if BLS does not exceed 2303 mL, since the hematocrit will not fall below Hm. Slightly more than 5 units of blood (5.1 units to be exact) must be removed during hemodilution under these conditions to achieve the maximum benefit from the technique and, ANH is used, no homologous blood will be required to maintain the Hm if blood loss does not exceed 2940 mL. In this case, ANH can save a maximum of 1.1 packed red cell unit equivalents, i.e., the amount of homologous blood that would be required to maintain the Hm if ANH was not used and a blood loss of 2940 mL occurred. If surgical blood loss exceeds 2940 mL, homologous blood transfusion will be necessary to maintain Hm even if ANH is used.

Table 2:
Model Calculations for Hypothetical 70-kg Patienta

The model can be used more generally to understand when ANH may be of benefit for a given patient and the degree of ANH necessary to maximize that benefit. For example, if Hi is 0.30 or less, it is simply not possible to save a red cell mass equivalent to 2 units of homologous PRBC even if the patient is hemodiluted to an Hm of 0.15. If Hi is 0.40, one must remove at least 7.5 units of blood during ANH, resulting in an Hm of 0.20, to save 2 unit equivalents. Clearly, the greater the Hi and the greater the number of units removed during hemodilution, the more effective ANH is for preventing homologous blood transfusion.

The results of the model calculations are designed to allow the clinician to decide whether ANH may be beneficial for a patient based upon knowledge of the Hi, the potential for SBL, and an estimate of the Hm. Although the results presented here are for a 70-kg patient, these results can readily be generalized for patients of any size. To find the results for a different body weight, any of the values BLS, BLH, ANHu, or PRBCe given in Table 2 need only be multiplied by the factor (70 times patient weight in kg) divide 4900.


The results of this analysis demonstrate that ANH can reduce the need for homologous blood transfusion if used appropriately. One striking finding is the number of units of the patient's blood that must be removed during ANH to realize a significant benefit form the technique. While some authors recommend that 2 or 3 units be removed during the dilution process Table 1, the model does not support this approach as likely to have a significant impact upon the requirement for homologous blood transfusion in the average adult patient.

It is important to emphasize that these conclusions are based solely upon a mathematic model. We believe these results can be used to guide clinical decision-making but the assumptions of the model, and the implications of those assumptions, must be clearly understood. The RCM is assumed to exist within a single compartment. This assumption is based upon the fact that red cells do not readily move in and out of an intact intravascular compartment. During ANH, blood is removed from this compartment while fluid is added to maintain euvolemia. During surgery, fluid and salvaged blood are added while the "patient's" blood is lost. This approach does not account for either losses of red cells due to hemolysis, or to release of red cells into the circulation from the spleen or bone marrow. There is no literature to document the impact of these other processes on RCM during surgery, although they are unlikely to be major determinants of hematocrit in the relatively short time period of surgical hemorrhage and volume replacement.

As mentioned earlier, the model is designed to predict the maximum RCM that can be saved by ANH. Approaches to ANH that advocate moderate dilution (e.g., 2-unit collections), would not dilute the patient to a Hm prior to surgery (assuming the Hi is in the normal range). The result is loss of a greater RCM in the blood shed during surgery than would otherwise occur if more extreme dilution was achieved via ANH. Indeed, the model demonstrates that more extreme degrees of dilution save a greater RCM. If one decides to use ANH to reduce RCM loss during surgery, the goal should be to remove the maximum RCM during dilution. That is, the patient should be diluted to Hm during the collection phase of ANH.

The concept of Hm warrants closer consideration to understand the implications of the results presented here to the clinical situation. Although H (m) exists for every patient, the actual value cannot be determined prospectively. Nevertheless, it is a useful concept since clinicians often approach patient management with an estimate of the Hm for a given patient. It is not until the physiologic compensation mechanisms for anemia are exceeded, however, that the Hm is identified. Although results are presented for a range of Hm's, for some patients an estimated minimum value may not be safe. Conversely, some patients physiologically may be able to tolerate even greater hemodilution than anticipated at the outset of the procedure. For purposes of this analysis, however, the concept of Hm is useful to illustrate the theoretical maximum RBC savings possible when using ANH.

There are some additional practical issues to be considered when applying the results of the model to the clinical situation. The blood which is collected during ANH is a source of platelets and coagulation factors in addition to RBC. No attempt has been made here to evaluate the use of this technique when platelet and coagulation factor salvage is desirable. In the case of massive blood transfusion, coagulopathy is a potential complication and salvage of platelets and coagulation factors may reduce the total blood loss. Homologous blood transfusion would not be prevented by ANH in that setting, however, since coagulopathy secondary to massive transfusion does not typically occur until an entire blood volume is replaced.

The model also assumes that euvolemia is maintained constant throughout surgery which assumes a more accurate assessment of blood losses and more steady replacement than is possible in the clinical situation. It is difficult to maintain precise control of hematocrit during retransfusion also because each unit of blood removed during ANH will have a different RCM depending upon the degree of hemodilution present when the unit is removed. Furthermore, fractions of units will not be given when homologous blood is administered. As a result, when the model indicates that ANH saves only 1.5 units of blood, 2 units of homologous blood would be given to maintain the hematocrit above Hm. Nevertheless, the results of the calculations should represent the maximum red cell savings possible. Clinical application of the technique under the conditions used for the model would likely result in less, rather than more, blood saved when using ANH.

The literature which purports to document that ANH is efficacious for reducing homologous blood transfusion is extensive. As much as a 90% reduction in homologous blood requirements has been claimed [6]. How can it be that the clinical studies predict much greater savings than are estimated here? The answer lies in study design since flaws are readily apparent in the clinical studies upon close inspection. Indeed, the authors of a recent review of ANH note that "…carefully controlled, randomized studies of ANH are few and far between," [7]. In fact, such studies do not exist.

One would like to have a prospective study of ANH used for similar patients, undergoing similar surgical procedures, randomized to receive ANH or conventional fluid replacement. A very important feature of the ideal study would be to blind the individual making decisions about the need for transfusion to any knowledge of whether ANH is used. The availability of autologous blood, and belief in the efficacy of ANH, may well influence the decision whether or not to transfuse homologous blood. Many studies do not ascertain the true impact of the technique on transfusions throughout the perioperative period since hematocrit and homologous blood transfusion are not documented past the operative period. Furthermore, clear and consistent guidelines for transfusion are used infrequently.

Given the perspective of the desirable attributes of a clinical study to determine the utility of ANH, it is useful to examine the literature which supports ANH as a method to reduce homologous blood transfusion. There are many studies which conclude that homologous blood transfusion can be reduced by ANH. Some representative prospective studies will be reviewed to exemplify the inconclusive nature of this literature.

One study of patients undergoing major vascular surgery reported homologous blood transfusion requirements of 2.6 units in the hemodiluted patients compared with 6.0 units in the nonhemodiluted group [8]. These investigators did not, however, apply the same transfusion criteria to each group. The hemodiluted patients received homologous blood to maintain a hematocrit above 0.30, whereas the other group was transfused to maintain a hematocrit above 0.35.

Another prospective study reported experience with 19 patients who were hemodiluted prior to "major" surgical procedures, primarily total hip replacement and aortic reconstruction [9]. Criteria for transfusion were established and a mean of 1.5 units of homologous blood was reported as the homologous blood used for these patients. The range of transfusion was not reported, nor was the SBL. Most importantly, a control group was not used to evaluate the impact of following strict transfusion criteria on homologous blood requirements in a similar population not undergoing ANH.

One study designed to evaluate the SBL associated with induced hypotension compared to hemodilution concluded that "the need for donor blood transfusion was reduced in patients undergoing hemodilution" [10]. This study was prospective but did not establish uniform transfusion criteria. In fact, the authors state in the results section that "criteria for blood transfusions were different in each group, so that comparison of the amount of transfused blood is only of incidental interest."

The results of ANH used in 26 patient's undergoing the Harrington procedure were compared to 12 patients undergoing the same procedure without hemodilution [11]. This study reported a mean intraoperative plus postoperative transfusion requirement of 750 mL in the hemodiluted group versus 4370 mL in the control group. However, transfusion was started in the hemodiluted group with autologous blood at a hematocrit of 0.15, whereas the control group received homologous blood when the hematocrit was 0.27. The transfusion trigger for homologous blood in the hemodiluted group was not defined. Although the results are striking, the variation in transfusion threshold makes interpretation difficult.

Although most publications on the efficacy of ANH indicate significant blood savings are possible, this has not been a universal finding. One retrospective study of 16 patients undergoing radical prostatectomy found very modest RBC savings associated with ANH [12]. The patients in that study had a mean Hi of 0.42 (range 0.36-0.48), an average of 976 mL of blood removed during ANH (range 500-1000 mL), and an estimated mean RBC volume savings of 95 mL (range 25-204 mL). Given our assumption of a RBC volume of 150 mL per unit of PRBCs, the results of that clinical study are consistent with the data from our model. Another recent study used mathematic modeling techniques to emphasize the progressive reduction of hematocrit that occurs during ANH. The results of that study also corroborated those presented here [13].

The risks of ANH seem to be small but there are some potential risks that remain to be documented in the literature. Units removed from the patient may become contaminated or inadvertently administered to a different patient. Since ANH removes a significant RCM from the patient, if the removed blood is wasted or becomes unusable, the patient will be more likely to need a homologous blood transfusion. These considerations of risk notwithstanding, ANH is relatively safe and may offer physiologic advantages to some patients, such as conservation of platelets and coagulation factors, improved tissue blood flow, and decreased thromboembolic complications. As a method for eliminating homologous blood transfusion, it appears to be of limited value based upon the calculations presented here. Modest reductions in homologous blood requirements may be possible with extreme degrees of dilution that are greater than typically advocated. Strategies currently available to eliminate homologous blood transfusion which are likely to be more effective than ANH are autologous predonation and intraoperative salvage. In the future, recombinant erythropoietin used to increase the patient's RCM prior to surgery and hemodilution with oxygen-carrying blood substitutes may also prove useful. The most important technique for minimizing homologous blood transfusion at present is sound clinical decision-making based upon a thorough understanding of the risks to the patient and the techniques available to minimize transfusion.


1. Stehling L. Autotransfusion and hemodilution. In: Miller RD, ed. Anesthesia. 3rd ed. New York: Churchill Livingstone, 1990:1501-13.
2. Messmer K. Acute preoperative hemodilution: physiological basis and clinical application. In: Tuma RF, White JV, Messmer K, eds. The role of hemodilution in optimal patient care. Munich: W. Zuckschwerdt Verlag, 1989:54-73.
3. The National Blood Resource Education Program Expert Panel. The use of autologous blood. JAMA 1990;263:414-7.
4. Bourke DL, Smith TC. Estimating allowable hemodilution. Anesthesiology 1974;41:609-12.
5. Gross JB. Estimating allowable blood loss: corrected for dilution. Anesthesiology 1983;58:277-80.
6. Martin E, Hansen E, Peter K. Acute limited normovolemic hemodilution: a method for avoiding homologous transfusion. World J Surg 1987;11:53-9.
7. Stehling L, Zauder H. Acute normovolemic hemodilution. Transfusion 1991;31:857-68.
8. Davies MJ, Cronin KD, Domaingue C. Haemodilution for major vascular surgery--using 3.5% polygeline (Haemaccel). Anaesth Intensive Care 1982;10:265-70.
9. Laks H, Handin RI, Martin V, Pilon RN. The effects of acute normovolemic hemodilution on coagulation and blood utilization in major surgery. J Surg Res 1976;20:225-30.
10. Barbier-Bohm G, Desmonts JM, Couderc E, et al. Comparative effects of induced hypotension and normovolemic haemodilution on blood loss in total hip arthroplasty. Br J Anaesth 1980;52:1039-43.
11. Martin E, Ott E. Extreme hemodilution in the Harrington procedure. Bibl Haematol 1981;47:322-37.
12. Goodnough LT, Grishaber JE, Monk TG, et al. Acute preoperative hemodilution in patients undergoing radical prostatectomy: a case study analysis of efficacy. Anesth Analg 1994;78:932-7.
13. Brecher ME, Rosenfeld M. Mathematical and computer modeling of acute normovolemic hemodilution. Transfusion 1994;34:176-9.
© 1995 International Anesthesia Research Society