Preoperative autologous blood donation (PABD) has become widely used in total hip (THA) and total knee (TKA) arthroplasties (1,2). If this practice reduces allogeneic transfusion (3) and risks from viral transmission (4), it may also increase exposure to all red blood cell (RBC) transfusion (5). Bacterial contamination and clerical errors (6,7) may affect any type of transfusion. Moreover, PABD that is based on standard predicted blood loss may lead to overcollection (8), preoperative anemia (9), and wastage of autologous RBCs (10,11). Patients participating in PABD (donors) may be overtransfused because physicians adopt more liberal criteria for transfusion of the patient’s own blood, especially if this blood is easily accessible (12,13). The question is whether one can use the same criteria for allogeneic and autologous transfusion (AT) (14). The differences in PABD programs and transfusion practices can explain hospital variations in blood used for THA and TKA (15). PABD remains widely used, and two studies showed transfusion (autologous and allogeneic) rates for THA from 53% to 81%(16,17).
Because of these drawbacks, a selection of patients for THA or TKA who will truly benefit from PABD has been recommended (18). This selection had to be based on a patient’s specific factors, such as the baseline hematocrit (Hct) and the estimated blood loss (19–21). However, the potential advantages of such a strategy have not yet been established in the fields of THA or TKA.
For these reasons, we compared the efficiency of two strategies: standard care, including extended indications for PABD, not taking into account individual specific factors and use of liberal AT; and novel care, including indications for PABD on the basis of individual specific factors, refinement of criteria for any transfusion, and homogeneous criteria for any transfusion, i.e., allogeneic or autologous. We therefore conducted a quality assurance assessment on the basis of two consecutive prospective observational cohort studies during primary THA or TKA. The first cohort received standard care and the second one novel care. Our aims were primarily to reduce donors and AT without increasing the risk of allogeneic requirements. For this purpose, we compared the percentages of patients transfused with autologous or allogeneic RBCs in the two studies. The second objective was to assess the evolution of wasted autologous RBC units. The team of physicians as well as anesthesia and surgical practices were unchanged throughout the studies.
Each study is presented separately. Every patient included gave written, informed consent. The study was performed in accordance with the regulations of the official edict of our local committee for the probation of persons involved in medical research.
Study 1 was prospectively conducted during a 12-mo period (from the end of December, 1997, to the beginning of January, 1999) and included every ASA physical status I–III patient scheduled for a cemented primary unilateral THA or TKA in the Department of Orthopedic Surgery of our institution. Refusal to participate was the single criterion for noninclusion.
The preoperative anesthesia evaluation was performed 2 mo before surgery. At this time, the baseline Hct was measured, and a PABD decision was made except in case of medical contraindication (including anemia with Hct <33%) or of refusal to participate in PABD. Both donors and nondonors were included in the study. The collection of three RBC units was prescribed, one each week, the last one at least 7 days before surgery. All patients had 320 mg of oral ferrous sulfate given twice a day beginning 3 wk before starting PABD. A decision of AT was left to the discretion of the anesthesiologist in charge during the perioperative period. If all autologous units had been transfused, or if no autologous unit was available, allogeneic blood was used. Its indication was an individual medical decision based on poor clinical tolerance and a Hct value <30%. In case of previous cardiovascular or cerebral disease, allogeneic blood was transfused to have a Hct value of ≥30%. No erythropoietin was used in any patient with Hct ≤39% because this medication was not yet easily available for THA or TKA at the beginning of Study 1.
Aspirin, antiplatelet, and oral anticoagulant treatments were planned to be stopped 8 days before surgery, and a switch to low-molecular-weight heparin was instituted when necessary after a cardiologist’s evaluation. For other patients, venous thromboembolism prevention by low-molecular-weight heparin was started on the day before surgery and was continued throughout hospitalization.
Neither acute normovolemic hemodilution nor intraoperative autologous salvaged blood was used during the surgical course. For TKA patients, transfusion of postoperative autologous salvaged blood from wound drainage during the first six postoperative hours was performed.
General anesthesia was standardized and used in every case. All anesthetic inductions were performed with propofol, sufentanil, atracurium, tracheal intubation, and controlled ventilation. For maintenance, isoflurane was administered in 50% oxygen/50% N2O, and sufentanil reinjections were given as needed. A three-in-one nerve block with 30 mL of bupivacaine 0.25% and clonidine 1 μg/kg was used for postoperative analgesia. Forced-air warming set at 43°C via a blanket applied on the upper part of the body was used throughout the procedure. Systemic controlled hypotension was not used; the anesthesiologist’s plan was limited to maintaining mean arterial pressure values in the 20%–25% range—less than the usual awake values. No antifibrinolytic drug was used.
The collected data included the patient’s age, sex, weight, ASA class, type of surgery, type of anesthesia, duration of the procedure, and length of hospitalization. Potential clerical errors in transfusion were noted. Hct values were obtained at preoperative anesthesia evaluation and before inclusion for PABD (baseline Hct), the day before surgery (admission Hct), and at Day 5 or 8 after surgery (discharge Hct).
The number of collected autologous units was noted, as was the number of autologous and allogeneic units transfused during the perioperative periods. The volume of the postoperative salvaged autologous blood transfused was also noted, and RBC loss was calculated.
The calculation of total perioperative RBC loss used the formula of Mercuriali and Inghilleri (22):MATHwhere EBV = the patient’s estimated blood volume (body weight in kilograms × 70 mL/kg) and transfused RBCs = 150 mL × number of autologous and/or allogeneic packed RBCs and postoperative autologous salvaged blood transfused (V). The postoperative autologous salvaged RBC was calculated by the following formula: V × 0.3. The mean Hct of this unwashed blood was 30%.
Thus, this calculated RBC loss (mL) included total perioperative blood loss. We selected THA and TKA patients who had no allogeneic transfusion and calculated their mean RBC loss. RBC loss was <800 mL for 83% of THA patients and <1000 mL for 87% of TKA patients. It was thus possible to know the mandatory mean RBC reserve allowing performance of THA or TKA without allogeneic transfusion for the Study 2 design.
We calculated the costs of collected autologous RBC units and transfused allogeneic RBC units. Our local cost for a single unit was $188.13 (US) for an autologous RBC unit and $155.27 (US) for an allogeneic RBC unit.
After evaluation of blood requirements for THA and TKA in our institution and analysis of the results of the first study, a second prospective study was initiated to evaluate the effects of changes in our transfusion policy. This second study was planned to include the same number of consecutive patients as Study 1 and took place from April, 1999, to February, 2000, i.e., a 10-mo period. Every ASA status I–III patient was potentially eligible. Criteria for inclusion and exclusion were the same as those described in Study 1. The same anesthesia and surgical teams were involved in the two studies.
There were three main policy changes: a different selection of donors for PABD and a different autologous RBC collection program were decided at preoperative anesthesia evaluation. During the perioperative period, new criteria for transfusion were introduced. First, indications for PABD were based on each patient’s RBC reserve, baseline Hct, and an estimated life expectancy of more than 10 yr. The patient’s estimated RBC reserve was calculated with the following formula:MATHwhere 30% was used for the target Hct value at discharge. This estimated RBC reserve was then compared with the mean RBC loss observed in Study 1 (800 mL for THA and 1000 mL for TKA). PABD was avoided if RBC reserve was ≥800 mL (THA) or 1000 mL (TKA). PABD was indicated by the anesthesiologist in case of insufficient RBC reserve, baseline Hct of >33%, an estimated life expectancy of >10 yr, and no medical contraindication. The second main change was to limit to 2 U the amount of blood collected preoperatively, with each collection performed a week apart and with 2 wk between the last collection and the day of surgery. The third main change of policy was the criterion for transfusion. Transfusion, either autologous or allogeneic, was indicated in case of Hct <24% or a Hct value between 24% and 30%(23) and one of the following symptoms: dyspnea, excessive weakness impeding deambulation or rehabilitation, evidence of myocardial ischemia or overt congestive heart failure, or postoperative neuropsychological impairment. We planned to have identical criteria for autologous and allogeneic blood transfusion.
Treatment by low-molecular-weight heparin, type of anesthesia, and type of surgery did not differ in Study 2. Neither acute normovolemic hemodilution nor intraoperative autologous salvaged blood was used during the surgical course. For TKA patients, transfusion of postoperative autologous salvaged blood from wound drainage during the first six postoperative hours was performed. The collected data, the calculation of RBC loss, and costs of RBC units were not different.
Demographic and biological data were expressed as mean ± sd, except for variables not normally distributed (duration; RBC loss; and collected, transfused, or wasted RBC units); for these, median and range values were used. The following tests were performed with the StatView (SAS Institute, Cary, NC) program. Comparisons of quantitative variables, such as Hct of patients with and without autologous or allogeneic transfusion, or Hct of patients between groups, used the Student’s t-test. Comparisons of the qualitative variables with two or more classes, such as autologous and allogeneic transfusions, sex, surgery, and anemia, were assessed with the χ2 test or Fisher’s exact test in case of insufficient calculated values. Quantitative variables not normally distributed were compared by using the Mann-Whitney Wilcoxon test. Changes in Hct value within a group or between groups were analyzed with one-way or two-way analysis of variance and then paired two-sided tests. A P value <0.05 was considered statistically significant.
A total of 364 patients were included and evaluated: 182 consecutive patients in Study 1 (120 THA and 62 TKA) and 182 consecutive patients in Study 2 (101 THA and 81 TKA). Their main characteristics are summarized in Table 1. There were no significant between-studies differences either at study entry or for operative time and RBC loss. The characteristics of patients according to PABD or no PABD are also shown in Table 1. There were no differences in donors. For nondonors, female sex was less frequent (P < 0.0001) and baseline Hct higher (P < 0.0001). Patient selection differences between the studies can explain these results. Study 1 patients’ evaluation showed that the mandatory RBC reserve allowing arthroplasty without allogeneic transfusion was 800 mL for 83% of THA patients and 1000 mL for 87% of TKA patients. At the preoperative anesthesia evaluation of Study 2, we selected 35 (19%) patients who had a sufficient RBC reserve to avoid PABD, mainly men (91%) with a mean baseline Hct of 47% ± 2%. They did not receive any transfusion. The nondonation was also because of an estimated insufficient life expectancy and, as in Study 1, to medical contraindications. The incidence of PABD was thus reduced in Study 2, as compared with Study 1, from 84% to 50% (P < 0.0001).
The characteristics of patients according to procedure are shown in Table 2. The populations of the two studies were comparable, except for THA, for which baseline Hct was slightly increased in the Study 2 population.
The frequency of overall transfusions was less in Study 2 than in Study 1 (43% vs 95% of patients [P < 0.0001]), as shown in Figure 1. The median transfused RBC units was also less in Study 2 (2 [1–5] vs 3 [1–5];P < 0.0001, as shown in Table 3). This was mainly explained by the marked decrease of exclusive AT (80% vs 30% [P < 0.0001]) without an increase in allogeneic transfusion (15% in Study 1 versus 13% in Study 2;Fig. 1). There were 10 times more patients without any transfusion in Study 2 (n = 103) than in Study 1 (n = 10;P < 0.0001) (Table 3). No clerical errors in transfusion practice were noted during the perioperative periods. The frequency of any transfusion when comparing donors and nondonors was increased for donors, both in Study 1 (99% vs 73%;P < 0.0001) and in Study 2 (68% vs 19%;P < 0.0001) (Table 3).
Table 4 shows that the reduction in overall transfusion, autologous and allogeneic, was observed with both types of arthroplasty. Comparisons between THA and TKA show differences only in Study 2 for allogeneic transfused patients and for all RBC units transfused.
The collection was complete for 78% of donors in Study 1 (three RBC units) and for 90% in Study 2 (two RBC units). The median individual number of collected autologous units was reduced in Study 2 versus Study 1 (2 [0–2] vs 3 [0–3];P < 0.0001), as expected by the change of PABD policy. Thus, the number of autologous units collected was 2.5 times less in Study 2 than in Study 1 (172 vs 426), i.e., 254 autologous RBC units were saved.
In transfused patients, a median reduction of one transfused autologous RBC unit was observed between Study 1 and Study 2 (Table 3). Despite fewer units collected, the median of wasted autologous units per donor was larger in Study 2 than in Study 1 (1 [0–2] vs 0 [0–3];P < 0.0001) (Table 3). Forty-six percent (n = 79) of autologous RBC units were wasted in Study 2 and 12% (n = 52) in Study 1 (P < 0.0001). Eighty percent of Study 2 patients with at least one wasted autologous RBC unit had a baseline Hct >39%.
We found a less frequent incidence of allogeneic transfusion when PABD was performed: 3% of donors versus 73% of nondonors in Study 1 (P < 0.0001) and 11% of donors versus 19% of nondonors in Study 2 (P = 0.005). The median number of allogeneic RBC units used in transfused patients was not different in the two studies, either in donors or in nondonors (Table 3). As shown in Figure 2, the more frequent rate of allogeneic transfusion was observed in patients with preoperative anemia, and the less frequent rate was observed in patients with the highest baseline Hct (Fig. 2). Fifty percent (Study 1) and 52% (Study 2) of allogeneic transfusions occurred in preoperative anemic patients. There were 40 anemic patients in Study 1 and 54 in Study 2. They received allogeneic transfusion in respectively 30% and 26% of the cases.
The total cost of RBC units was $88,683.23 in Study 1 and $38,569.16 in Study 2. This represents an average cost savings of $275.35 per patient.
The mean admission Hct was higher in Study 2 than in Study 1 for the whole population, donors and nondonors. The admission Hct was also higher in nondonors than in donors for both studies (P < 0.0001) (Table 5).
The mean discharge Hct was lower in Study 2 than in Study 1, both for the whole population and for donors, whereas it was higher in Study 2 than in Study 1 for nondonors. This last result was in accordance with the design of Study 2, in which 35 nondonors were selected on the basis of a sufficient RBC reserve with a high baseline Hct. The percentage of patients with a discharge Hct <30% was increased in Study 2 (36% vs 25%;P < 0.05). These mean Hct values were not different: 27% ± 2% (range, 22%–30%) in Study 1 and 28% ± 1% (range, 24%–30%) in Study 2. The discharge Hct of allogeneic transfused patients did not differ between the two studies (30% ± 3% in Study 1 and 30% ± 2% in Study 2), whereas there was a significant difference for AT patients (33% ± 3% in Study 1 versus 31% ± 2% in Study 2 [P < 0.0001]). Study 1 discharge Hct was higher in AT patients than in allogeneic transfused patients (33% ± 3% vs 30% ± 3%) (P = 0.0002). The Study 2 discharge Hct did not differ between allogeneic transfused patients (30% ± 2%) and AT patients (31% ± 2%), in accordance with the transfusion criteria of Study 2 (P = 0.09).
The mean length of hospitalization was identical in the two studies: 10.4 ± 2.0 days in Study 1 and 10.3 ± 2.4 days in Study 2.
Our study mainly shows that a quality assurance assessment for PABD and transfusion in primary THA and TKA may decrease overcollection and correct overtransfusion. When a PABD policy based on evaluation of individual RBC reserve and mean expected perioperative bleeding is associated with homogeneous criteria for transfusion (autologous and allogeneic), 10 times more patients are not transfused. This is because of a reduction in PABD and AT, with no subsequent increase in allogeneic requirements. After improvement of transfusion criteria in Study 2, the autologous requirement for primary THA and TKA was one or two RBC units, which is sufficient to discharge with a Hct of 30%. However, 46% of collected RBC units were wasted. This wastage was mainly attributable to patients with a baseline Hct >39%. Refinement of PABD indications in this subpopulation could potentially reduce this wastage.
The percentages of overall transfusions in THA and TKA are highly dependent on interhospital variations (15,16) because of the use of PABD and transfusion thresholds (17). Our first study shows that the rate of overall transfusion was frequent (95%). This was not because of an unusual rate of allogeneic transfusion, because our percentage of 15% was in the range of rates previously published (16). This frequent transfusion was, rather, explained by the liberal use of AT alone in 80% of our Study 1 patients. We suspected that this percentage of AT was a result of excessive collection of PABD units with admission Hct less than baseline Hct and liberal transfusion of autologous blood resulting in discharge Hct >30%. In the 1990s, all of these points were questioned, leading to new recommendations (19–21). The risk of allogeneic blood-induced development of acquired viral disease in older patients now seems to be infrequent enough to be challenged with the risks of PABD. Routine autologous collection of excessive blood leads to discard of up to half (4,10). PABD induces a preoperative anemia, and patients are more likely to receive any type of transfusion (3,17). Study 1 results show that our initial transfusion policy was not in agreement with recent recommendations of transfusion. On the basis of a quality assurance principle, Study 2 was designed to evaluate the following changes: 1) PABD indicated by estimated RBC reserve and life expectancy; 2) collection of autologous RBC limited to 2 U; 3) two weeks between last collection and time of surgery; and 4) the same criteria for any transfusion. Study 2 results support the conclusion that each of these methodological endpoints had been fulfilled: donors decreased from 83% to 50% (Table 3); overall AT decreased from 84% to 34% of patients (Fig. 1); the median number of collected RBC units per donor decreased from three to two with no increase in allogeneic requirements; the admission Hct of donors was higher in Study 2 (38%) than in Study 1 (36%) (Table 5); and, finally, discharge Hct did not differ according to PABD or not, or type of transfusion (autologous or allogeneic) (Table 5). In addition, transfusion cost was less in Study 2 because of an average cost savings of $275.35 per patient, i.e., 56% of the average Study 1 cost.
Avoidance of PABD in case of sufficient RBC reserve is often advocated as a logical guideline (18,21,24); however, its feasibility remains unproven in the field of primary joint orthopedic surgery. Our Study 2 shows that patients with a sufficient preoperative RBC reserve need no allogeneic transfusion. This result was obtained on a rather small population (n = 35), but it demonstrates the feasibility of such a policy during primary THA or TKA. Studies 1 and 2 are consistent with the Belgium BIOMED study (17) because we found that autologous donors received fewer allogeneic transfusions than nondonors, especially in Study 1 (3% vs 73%). In Study 2, the difference of percentages in allogeneic transfusion was less marked (11% of donors versus 19% of nondonors). This can be explained by the different selection of patients between the two studies. In Study 2, the nondonors were not only patients who could not go through a PABD program for medical contraindications, but also patients with a sufficient RBC reserve to avoid transfusion. In accordance with Forgie et al. (3), we also found that liberal use of PABD favors exposure to overall transfusions. Most Study 1 patients (84%) had PABD, and the incidence of any transfusion was 95%, with a different discharge Hct between patients who received AT (33%) and those with allogeneic transfusion (30%). In Study 2, PABD and overall transfusions were reduced, and discharge Hct did not differ (31 for AT and 30 for allogeneic transfusion). Thus, in Study 2, criteria for transfusion were comparable for autologous or allogeneic blood. The reduction in overall transfusions was a result of the marked reduction in exclusive AT and adhesion to clinical individual criteria for transfusion. Our Study 2 shows that such a reduction can be obtained without inducing marked postoperative anemia. In this study, indications for transfusion were based on the fact that a Hct value of 30% in impaired cardiac function, and values near 21%–24% in otherwise healthy patients, are generally recognized as appropriate transfusion triggers for surgical procedures (23). Our within-hospital outcome data and mainly identical length of hospitalization suggest that in-hospital postoperative outcome was not different between the two studies. Because we did not study out-of-hospital outcome, our study cannot, however, determine which method is the better one. Study 2 patients had a higher admission Hct and a lower discharge Hct (31% ± 3%) (Table 5), and 36% of them had a discharge Hct in the range of 24%–30%. Thus, the question raised is, is there evidence that maintaining a discharge Hct of 30% or more has a positive effect on morbidity or mortality? Neither our study nor the existing literature can give a definitive answer to this question. However, a postoperative hemoglobin value in the range of 9.0–11.0 g/dL does not affect recovery after an orthopedic operation, according to Green et al. (24). Other studies showed that a Hct value <30% might be even beneficial to some patients, including those with coronary disease (25,26). Thus, we believe it is unlikely that a discharge Hct of 31% ± 3% may have favored morbidity in our patients.
There may be other limitations. First, ours was not a randomized study. We chose to adopt a quality assurance assessment method that was based on two successive prospective studies. After analysis of Study 1 results and educational department meetings, we believed that it was important to modify the whole team’s clinical practices. It seemed the most appropriate method to address a question that concerned the whole team. We tried to minimize this methodological weakness by consecutive inclusion, without any change in either members or clinical practices of surgical and anesthesia teams. Second, because we did not stratify our two studies according to the type of surgery, more TKA patients were included in Study 2. Because RBC loss was slightly higher during TKA than THA (Table 2), allogeneic transfusion was more frequent in TKA than in THA patients within Study 2 (Table 4). However, these facts had no effect on the validity of our conclusions. Indeed, our primary aim was not to reduce allogeneic transfusion, but to show that a reduction in AT with a patient-based personalized policy was possible without inducing more allogeneic requirements. Our study shows that despite the inclusion of more TKA patients in Study 2, and thus more potential bleeding and transfusion, the decrease in AT remains possible without increasing allogeneic requirements.
We believe that two reasons may explain why no change in allogeneic requirements was observed despite the use of a more restrictive transfusion threshold in Study 2. Unpredictable significant bleeding and insufficient RBC reserve are the two main triggers of allogeneic transfusion. The most frequent rate of allogeneic transfusion occurred in patients with preoperative anemia, as shown in Figure 2. In fact, roughly 50% of allogeneic transfusions were observed in patients with preoperative anemia in both studies. This fact suggests that, when no erythropoietin is used, there might be an incompressible percentage of allogeneic requirements, whatever the transfusion policy. Goodnough et al. (27) made a similar suggestion: they found that 12% of patients still received allogeneic blood for hip arthroplasty despite the use of a homogeneous trigger for any transfusion. Third, despite a 2.5-fold reduction of total autologous collected RBC units in Study 2, relative wastage was increased (46% of collected units). Mercuriali and Inghilleri (22) showed that a wastage rate of <15% can be achieved with a personalized prediction of a patient’s transfusion requirement. Wastage in Study 2 was mainly because of unnecessary PABD in patients with baseline Hct >39%. This is suggested by the fact that roughly 80% of Study 2 patients with at least one autologous RBC unit discarded had baseline Hct >39% (Fig. 2), whereas allogeneic transfusion was infrequent (9%) in this subpopulation. Thus, two further improvements in transfusion policy might be useful in our institution: first, the use of recombinant erythropoietin without PABD in case of baseline Hct <37% to reduce allogeneic requirements; and second, avoidance of PABD for the nonanemic population (baseline Hct >39%) to reduce autologous wastage. These refinements in transfusion policy are now applied in our institution.
In summary, we evaluated the consequences of changing a liberal policy for PABD and autologous RBC transfusion for a personal requirement policy during primary THA and TKA. Such changes produce 10 times more patients not transfused, with no change in allogeneic requirements, no overtransfusion, a 44% reduction in costs linked to RBC unit utilization, and no uncontrolled anemia on discharge. Despite a 2.5-fold reduction of total autologous collected RBC units, units were still wasted, probably because of the unpredictable effect of restriction in transfusion criteria. Further evaluation is thus needed, including the effect of recombinant human erythropoietin in case of baseline anemia and reduction of autologous individual needs in case of baseline Hct >39%.
We are indebted to Andrée Verrier for typing the manuscript.
1. Consensus conference: perioperative red blood cell transfusion. JAMA 1988; 260: 2700–3.
2. Welch HG, Meechan KR, Goodnough LT. Prudent strategies for elective red blood cell transfusion. Ann Intern Med 1992; 116: 393–402.
3. Forgie M, Wells P, Laupacis A, Fergusson D. Preoperative autologous donation decreases allogeneic transfusion but increases exposure to all red blood cell transfusion: results of a meta-analysis—International Study of Perioperative Transfusion Investigators. Arch Intern Med 1998; 158: 610–6.
4. Goodnough LT, Brecher ME, Kanter MH, Aubuchon JP. Transfusion medicine: second of two parts—blood conservation. N Engl J Med 1999; 340: 525–33.
5. Cohen J, Brecher ME. Preoperative autologous blood donation: benefit or detriment? A mathematical analysis. Transfusion 1995; 35: 640–4.
6. Goldman M, Rémy-Prince S, Trépanier A, Décary F. Autologous donation error rates in Canada. Transfusion 1997; 37: 523–7.
7. Linden JV, Kruskall MS. Autologous blood: always safer? Transfusion 1997; 37: 455–6.
8. Goodnough LT, Verbrugge D, Vizmeg K, Riddell J. Identifying elective orthopedic surgical patients transfused with amounts of blood in excess of need: the transfusion trigger revisited. Transfusion 1992; 32: 648–53.
9. Goodnough LT, Brittenham GM. Limitations of the erythropoietic response to serial phlebotomy: implications for autologous blood donor programs. J Lab Clin Med 1990; 115: 28–35.
10. Etchason J, Petz L, Keeler E, et al. The cost effectiveness of preoperative autologous blood donations. N Engl J Med 1995; 332: 719–24.
11. Birkmeyer JD, Goodnough LT, AuBuchon JP, et al. The cost-effectiveness of preoperative autologous blood donation for total hip and knee replacement. Transfusion 1993; 33: 544–51.
12. Kanter M, Van Maanen D, Anders K, et al. Preoperative autologous blood donations before elective hysterectomy. JAMA 1996; 276: 798–801.
13. Wasman J, Goodnough LT. Autologous blood donation for elective surgery: effect on physician transfusion behavior. JAMA 1987; 258: 3135–7.
14. Spahn D, Casutt M. Eliminating blood transfusions: new aspects and perspectives. Anesthesiology 2000; 93: 242–55.
15. Churchill WH, McGurk S, Chapman RH, et al. The Collaborative Hospital Transfusion Study: variations in use of autologous blood account for hospital differences in red cell use during primary hip and knee surgery. Transfusion 1998; 38: 530–9.
16. Bierbaum BE, Callaghan JJ, Galante JO, et al. An analysis of blood management in patients having a total hip or knee arthroplasty. J Bone Joint Surg Am 1999; 81: 2–10.
17. Baele P, Beguin C, Waterloos H, et al. The Belgium BIOMED Study about transfusion for surgery. Acta Anaesthesiol Belg 1998; 49: 243–303.
18. Goodnough L. Autologous blood procurement in surgery. TATM 2000; S1: 22–6.
19. Larocque BJ, Gilbert K, Brien WF. A point score system for predicting the likelihood of blood transfusion after hip or knee arthroplasty. Transfusion 1997; 37: 463–7.
20. Larocque BJ, Gilbert K, Brien WF. Prospective validation of a point score system for predicting blood transfusion following hip or knee replacement. Transfusion 1998; 38: 932–7.
21. Nuttall GA, Santrach PJ, Oliver WC, et al. The predictors of red cell transfusions in total hip arthroplasties. Transfusion 1996; 36: 144–9.
22. Mercuriali F, Inghilleri G. Proposal of an algorithm to help the choice of the best transfusion strategy. Curr Med Res Opin 1996; 13: 465–78.
23. Goodnough LT, Brecher ME, Kanter MH, Aubuchon JP. Transfusion medicine: first of two parts—blood transfusion. N Engl J Med 1999; 340: 438–47.
24. Green D, Lawler M, Rosen M, et al. Recombinant human erythropoietin: effect on the functional performance of anemic orthopedic patients. Arch Phys Med Rehabil 1996; 77: 242–6.
25. Hébert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. N Engl J Med 1999; 340: 409–17.
26. Spiess BD, Ley C, Body SC, et al. Hematocrit value on intensive care unit entry influences the frequency of Q-wave myocardial infarction after coronary artery bypass grafting. J Thorac Cardiovasc Surg 1998; 116: 460–7.
27. Goodnough LT, Despotis GJ, Merkel K, et al. A randomized trial comparing acute normovolemic hemodilution and preoperative autologous blood donation in total hip arthroplasty. Transfusion 2000; 40: 1054–7.