infection or suspected infection after hip replacement surgery with autologous or homologous blood...

Infection or suspected infection after hip replacement surgery with autologous or homologous blood transfusions

P. MURPHY, J.M. HEAL, AND N. BLUMBERG

Homologous blood transfusions have been associated in both animals and humans with an increased risk of acute postoperative infectious complications. Eighty-four patients who underwent hip replacement surgery and were transfused with 2 or 3 units of blood were analyzed to determine whether those receiving homologous transfusions had different outcomes than those receiving autologous blood only. Only patients free of other risks for postoperative infection were studied. Those receiving homologous blood had a 32 percent (1 6/50) rate of proven or suspected infections, which was significantly higher than the 3 percent (1/34) rate in patients receiving autologous blood (p = 0.0029). Wound infections accounted for only a minority (61 17) of the proven or suspected infections, which suggests that nonsurgical factors contributed to these complications. The patients identified as being infected required significantly more antibiotic therapy (mean, 7.6 days) and lengthier hospital stays mean, 15.5 days) than the patients who remained free of evidence of infection I means: 2.3 days of antibiotics and 12.3 days in the hospital) (p=O.OOOl for each

variable). Other potential risk factors for infection, such as duration of surgical procedure, advanced patient age, amount of blood loss, type of anesthesia, surgeon performing the operation, use of a cemented versus porous-coat prosthesis, leuko- cytopenia, anemia, and underlying medical diagnosis, did not account for the differences in infection rates seen in those receiving homologous and autologous transfusions. These results confirm previous reports of an increased risk of postoperative infection in patients receiving homologous transfusions. Homologous transfusion may contribute to an increased risk of infection by immunologic modulation of the recipient. If confirmed by other studies, these results rovide another powerful argument in favor of autologous blood transfusion. TRA&FUSlON 1991 :31:212- 21 7.

RECENTLY, IT HAS BEEN PROPOSED that patients receiving homologous blood transfusions have an increased risk of postoperative bacterial infections. Increasing numbers of homologous transfusions are also a major independent predictor of septic complications after t r a ~ m a . ~ These findings are supported by animal experiments demonstrating adverse effects of allogeneic blood transfusions on host defenses against bacterial in- f e ~ t i o n . ~ . ~ One of these effects may be the impairment of monocyte migration to wound sites. Other proposed clinical consequences of homologous transfusions include improved success rates for renal allograft~,~ increased rates of solid tumor recurren~e,~.~ a decreased rate of recurrence of Crohn's disease,l0 and increased severity of viral infection." The proposed mechanisms for such effects center on impaired immunologic function in the transfused patient. In some animal model systems, the deleterious effects of allogeneic transfu-

From the Transfusion Medicine Unit, Department of Pathology and Laboratory Medicine, and Hematology Unit, Department of Medicine, University of Rochester Medical Center, and Amcrican Red Cross Blood Services, Rochester, New York.

Presented in part at the 42nd Annual Meeting of the American As- sociation of Blood Banks, New Orleans. LA, October 1989.

Received for publication March 2, 1990; revision received May 15, 1990, and accepted May 23, 1990.

sions on host defenses against tumor growth are not observed when syngeneic transfusions are given.l2*l3 In humans, autologous transfusions are the closest analogue of syngeneic transfusions in animals.

Thus, an obvious question regarding impairments of host defenses after transfusion is whether autologous transfusions are associated with outcomes similar to those seen with homologous transfusions. The increase in use of autologous transfusi~ns,'~-'~ largely as a result of con- cerns over transmission of infectious viral agents, makes feasible the testing of the hypothesis that recipients of autologous transfusions might have less likelihood of developing postoperative bacterial infections than recipients of similar amounts of homologous blood. We in- vestigated this hypothesis in patients who underwent hip replacement surgery in our hospital.

Materials and Methods W e retrospectively reviewed the medical records of all adult

patients who undewent total hip replacement surgery in our institution between January 1986 and March 1987. Because of the small number of patients meeting the criteria for this study who received autologous transfusions during that period (n = 16), we also reviewed the charts of 36 patients who received autologous blood from January to December 1988. This step made it possible to have an 80 percent or greater chance (power)

212

TRANSFUSION 1991-Vol. 31. No. 3 AUTOLOGOUS OR HOMOLOGOUS TRANSFUSIONS 213

of detecting a true 20-percent difference in the infection rates (5% in autologous blood recipients versus the 25% expected in homologous blood recipients). This difference has been seen in previous work comparing nontransfused and transfused recipients.1-3

We collected information on the patient's age and gender, the etiology of joint disease, orthopedic diagnosis, underlying medical problems, type of appliance, use or nonuse of methyl- methacrylate cement, type of anesthesia, attending surgeon, admission hematocrit (Hct) and total white blood cell count (WBC), prophylactic or therapeutic use of antibiotics, duration of operative procedure, estimated blood loss, transfusions during the periods before, during, and after operation, results of surveillance or diagnostic bacterial cultures, and other clinical or laboratory evidence for infection, including days of fever >37.5"C (99.5"F), days of fever >38.3"C (101"F), volume and number of days of wound drainage (and the presence or absence of blood), length of stay, and Hct at discharge. We include in the reported analysis only patients who underwent primaty hip replacements, and we specifically exclude patients who underwent revision of failed previous operations. Patients with malignant, autoimmune, or preexisting infectious conditions who developed infections were excluded from the analysis of frequency of infectious complications, which left only those patients in whom traumatic or primary degenerative arthritis was the cause of the disability. Patients with postoperative urinary retention were also excluded because of their high rate of developing urinary tract infections.

Patients who received more than three transfusions are not included in the detailed analysis in order to restrict comparisons to patients with comparable transfusion therapy. As previous

had shown no association between single-unit transfusions and an increased rate of infection, we compared only those patients receiving 2 or 3 units of autologous or homologous blood. No patient received blood components other than red cells (RBCs) or whole blood, and none of the patients received both autologous and homologous blood. Transfusions were, with rare exception, all given intraoperatively or within 24 hours after operation.

We reviewed the medical records of 190 patients and analyzed for this study 34 patients who received 2 or 3 units of autologous blood and 50 who received 2 or 3 units of homologous blood. Twenty patients were excluded because of preexisting infection or risks for infection, such as rheumatoid arthritis, a history of chronic urinary tract infection, or postoperative urinary retention requiring catheterization. Other patients were not analyzed because they received no transfusions (n = 18), only 1 unit of autologous or homologous blood (n=27), or 2 4 units of autologous or homologous blood (n =41). Of the patients excluded because of infection risks unrelated to transfusion, 11 were recipients of autologous blood, 6 were recipients of homologous blood, and 3 received no transfusions. Thus, the autologous blood recipients had a higher incidence (11/55; 20%) of increased risk of infection than those who did not donate for themselves (9/135; 7%) (p=O.Ol).

We employed commonly accepted and validated criteria for ascertaining infection through retrospective medical record reviews.'* For example, criteria for proven or suspected bacterial infection included, as a minimum, strong clinical or laboratory evidence of infection and either a positive bacterial culture or antibiotic therapy for at least 5 days. Only patients considered by both their physician and the reviewer of the record to have been infected were categorized as infected for the purposes of this study.

All patients had received 2 days of prophylactic antibiotics as part of their routine management. This consisted of an in-

traoperative dose of a long-acting cephalosporin and one fur- ther dose the next day. As would be expected after the administration of prophylactic antibiotics, many patients with compelling clinical and laboratory evidence for postoperative infection, which led their physicians to initiate therapeutic courses of antibiotics, never had positive bacterial cultures. Previous research has shown that in this setting, consideration of culture- proven infections only will lead to a great underestimation of the number of clinical events that are due to infection.'* Con- versely, signs and symptoms that suggest infection in the absence of antibiotic therapy or positive cultures were not considered as infections, particularly when other causes (e.g., wound hematoma) were considered more likely by the clinical staff. An example would be postoperative fever with wound erythema that remitted without sequelae or need for antibiotic therapy.

We chose to analyze in detail only the patients meeting the above criteria who received either 2 or 3 units of blood. Our intent in choosing a case-control type of methodology was to compare patients who were as similar as possible given the nonrandom nature of selection by themselves or their surgeon for autologous donation.

Statistical analysis of the data employed a microcomputer (Macintosh IIcx, Apple Computer, Cupertino, CA) and soft- ware (Statview SE+ Graphics, v. 1.02, Abacus Concepts, Berkeley, CA). Unless otherwise noted, all tests were performed by both parametric and nonparametric analyses and yielded similar results with both.

Results Table 1 shows the results of a comparison of patients who

received 2 or 3 units of autologous or homologous blood. Thosc who received homologous blood showed little or no difference from those who received autologous blood in potentially important variables such as age, gender distribution, duration of surgery, estimated blood loss, type of anesthesia, days and volume of postoperative wound drainage, admission WBC count, or days with fever. The patients who had donated blood for their own use had a small but significantly lower mean admission Hct, as might be expected (38% for autologous donation versus 41% for homologous blood).

The most striking difference between the two groups was the prevalence of proven or suspected infection, with 16 (32%) of the 50 recipients of homologous blood developing proven or clinically suspected infections versus 1 (3%) of the 34 who received autologous blood (p = 0.0029 by chi-square with correction for continuity). As would be expected given the difference in proven or suspected infection rate, the recipients of homologous blood received more antibiotic therapy and had significantly longer hospital stays than those who received autologous blood. The infections were not clinically severe and no deaths occurred. The infectious complications were pre- dominately (11/17) away from the wound site (Table 2). Two patients were readmitted because of infection. Three patients had postoperative fever of unknown origin, which was treated with extended courses of intravenous antibiotics. None of these episodes bore an obvious relationship to a preexisting risk of infection other than the receipt of homologous blood. We have not yet studied these patients for long-term, more serious infectious complications such as osteomyelitis.

Two variables in addition to type of transfusion were significantly associated in the initial univariate analysis (one-way analysis of variance [ANOVA] or its nonparametric equivalent) with the likelihood of infection-that is, the attending surgeon responsible for the patient and whether a cemented prosthesis

214 MURPHY, HEAL, AND BLUMBERG TRANSFUSION Val. 31, No. 3-1991

Table 1 . Characteristics of recipients of autolonous or homologous transfusions* Autologous Homologous

Variable recipients recipients p value Number of patients 34 50

Men 21 (62%) 30 (60%) N S t Women 13 20

Units transfused 2.3 t 0.5 2.3 f 0.5 NS

Age (years) 58 t 1 1 59 t 15 NS Preoperative hematwrit (%) 38 f 3.8 41 f 4.7 0.0008 Discharge hematccrit (%) 32 f 4.0 33 t 3.1 NS Admission white cell count per FL 8300 f 5600 7700 5 2200 N S Length of stay (days) 12.1 t 2.5 13.5 f 3.6 0.0427 Days of fever > 38.3OC 1.3 -c 1.4 1.1 f 1.5 NS Days of antibiotics$ 2.6 t 1.3 3.9 5 3.6 0.0607 Days of wound drainage 2.9 f 1.5 3.0 t 1.5 NS Duration of surgely (minutes) 148 -c 32 143 f 32 NS Blood loss (mL) 890 t 300 lo00 f 400 NS ‘The values shown are mean f SD unless otherwise noted. Differences in proportions were determlned by chi-square, with continuity correction If appropriate. The other p values were determined by one-way analysis of variance. +Not significant (p > 0.10). Wl patients in both groups received prophylactic intraoperative and postoperative cephalosporin for 2 days.

Number with infections 1 (3%) 16 (32%) 0.0029

was employed. Two surgeons performed 67 (80%) of the 84 operations, and one surgeon’s patients had a much higher rate of infectious complications. However, the differing infection rates were associated with variations in the use of autologous transfusions. For surgeon A’s patients, infections occurred in 13 (41%) of 32 homologous blood recipients, but in none of 7 autologous blood recipients. For surgeon B’s patients, infections occurred in 2 (33%) of 6 homologous recipients, versus 1 (5%) of 22 autologous recipients. Thus, the apparent difference between the complication rates found with these two surgeons is almost completely related to differences in the proportion of each surgeon’s patients who received autologous blood.

Patients receiving a cemented joint prosthesis also had an increased rate of postoperative infection in the univariate analysis (one-way ANOVA). However, of the 18 patients with cemented prostheses, none of the 4 who received autologous blood had infections, whereas 8 (57%) of the 14 who received homologous blood did so. Thus the small number of patients with cemented prostheses does not account for the overall differences in infection prevalence seen in the recipients of autologous and homologous blood. When we performed a three- way ANOVA, employing autologous versus homologous transfusions, attending surgeon, and cement versus no cement as predictive variables for postoperative infection, the only significant predictor of infection was transfusion type (p=0.0161). The corresponding values for cement status (p = 0.3553) and surgeon (p =0.7965) suggest that the variance in infection rates seen among these patients is related primarily to whether autologous transfusions were employed and not to which surgeon performed the operation or the type of prosthesis used.

The mean length of stay for the patients with proven or suspected infections is 15.7 days (n = 17), which is more than 3 days longer than the mean stay of patients without evidence of infection (12.3 days; n=67) (p=O.OOOl) (Table 3). The infected patients account almost completely for the difference in length of stay of homologous and autologous blood transfusion recipients. It can be seen from the data in Table 3 that, as expected, infected recipients had more days with fever and days on antibiotics than patients who were free of the evidence of infection. The data in Table 3 also support the contention

that the patients included in this study were quite similar in many important respects, with the exception of the likelihood of infection after homologous transfusion. The infected patients were not significantly older, more anemic, more leu- kocytopenic, or more heavily transfused than their uninfected counterparts. They also did not have more wound drainage, longer operations, or greater blood loss during operation than the uninfected patients. The only factor that distinguishes the infected from uninfected patients in this study is the proportion in each group who received homologous blood.

We excluded a number of patients from this analysis because of increased risks of infection unrelated to transfusion. Four patients receiving 2 or 3 units of homologous blood were excluded from the data: two with rheumatoid arthritis, one with systemic lupus erythematosus, and one with urinary retention. Two of these patients developed postoperative infections. Seven patients receiving 2 or 3 units of autologous blood were excluded from the data: three with urinary retention, three with preexisting chronic urinary tract infections, and one with rheumatoid arthritis. One of these patients developed a new infection after operation, and three others received antibiotic prophylaxis or treatment because of previous urinary tract infections. If all these patients were included in the analysis, the rate of new infections would still be significantly different in the homologous transfusion recipients (18/54; 33%) and the autologous recipients (2/41; 5%) (p=O.OOOS by Fisher’s exact test). Even if preexisting urinary tract infections in the autologous blood recipients were to be included in the analysis, the difference in infection rate would remain significant (18/54 in homologous recipients and 5/41 in autologous recipients; p = 0.0146).

Discussion This is the first study of which we are aware that

compares postoperative infectious complications in recipients of autologous or homologous blood transfusions. Thus, our findings must be considered preliminary. Given the uncontrolled nature of our observations, it remains possible that the differences we observed are due to inapparent but important differences in the pop-

AUTOLOGOUS OR HOMOLOGOUS TRANSFUSIONS TRANSFUSION 1991-Vol. 31. No. 3 215

Table 2. Patients develooina postoperative infection after transfusions ~ ~~

Number Age1 Underlying Type of Culture Length of of units Gender disease‘ lnfectiont Antibiotics given* results5 stay (days)

631M DJDq FUO“ Cefazolinl Negative 14 Trimethopriml

211

2 2

2 2 2 2 2 2

2 3

3 3

3

3 3 3

891F 36lM

731M 301M 641F 69lF 40lM 831F

57lM 74lF

74lF 7OlF

74lF

42/M 66lM 661F

DJD Trauma**

DJD DJD DJD DJD DJD DJD

DJD DJD

DJD DJD

DJD

DJD DJD DJD

Wound Wound

Wound Wound Wound FUO Cellulitis Dermatitis

FUO Urinary

Urinary Boil

Urinary

Wound Cellulitis Urinary

Sulfamethoxazole Cefazolin Trimethoprid

Sulfamethoxazole; Dicloxacillin

Cefazolin Dicloxacillin Dicloxacillin Cefazolin Cefazolin Topical

antifungals Cefazolin Trimethopriml

Sulfamethoxazole Clindamycin Trimethoprid

Sulfamethoxazole Trimethoprid

Sulfamethoxazole; Cephalexin

Cefazolin Dlcloxacillln None recorded

Negative NT#

NT NT NT Negative NT NT

Negative NT

Negative NT

E. coli

S. aureus Negative P. aetuginosa

15 13

13 13 12 16

23 955

15 16

25 11

16

2268 18 13

‘Patients with malignancies. autoimmune disease, diabetes, and other predisposing causes for infection were excluded from the analysis. +Patients with fevers of short duration who were not given antibiotics and with other likely causes of pyrexia were not considered to have Infections. Unless specified, cellulitis and dermatitis were at locations distant from the operative wound. *Only therapeutic courses of 5 days or more of entiblotics are included. All patients received 2 days of prophylactic cephalosporin. §For positive results, only those that were thought to be clinically relevant are listed. All patients had negative intraoperative hip cultures. Blood, sputum, urine, and wound cultures were negative in patients with fevers of unknown origin. IlReceived autologous blood (all other patients listed received homologous transfusions). qDegenerative arthritis. “Fever of unknown origin. ttPosttraumatic arthritis. #Not tested. §§Patients who were readmitted with infection.

ulations of patients who receive autologous or homologous transfusions. We found no evidence to support this possibility, but it is evident that the two patient populations cannot be completely identical. Some patients are simply not candidates for autologous donation because of their degree of disability or other intercurrent ill- nesses, anemia, or logistic constraints. The autologous and homologous transfusion recipients that we studied were, in some cases, treated by different surgeons and at different periods, but these factors were not significant in predicting infection. The surgical, antibiotic, and other treatment protocols were identical during both periods. A patient’s overall clinical condition might be expected to be prognostic of postoperative infection; however, we excluded from analysis all patients with underlying ill- nesses or conditions that might predispose to infection. Thus, it is striking that a tenfold difference in infection rate should be seen in the absence of obvious differences in the patient populations involved. Moreover, the total group of patients who donated autologous blood had a

higher rate of preexisting medical conditions predisposing to infection than the group that did not donate for themselves. This finding suggests that autologous donors were not remarkably “healthier” than patients receiving homologous blood. Randomized controlled trials of autologous transfusions may be difficult, if not impossible, to perform, despite their obvious advantages over our case-control approach. Many patients and physicians would consider randomized studies clinically and ethi- cally unacceptable. Thus future studies of this issue, al- though they might include larger numbers of subjects and prospective evaluation of infections, may also have to rely on essentially uncontrolled observations.

One might argue reasonably about whether infection indeed occurred in some of the patients we classified as such. Only 3 of 17 infections were proven by culture, but we employed accepted criteria for performing retrospective medical record reviews to detect nosocomial infections.’* The use of culture data as the sole standard for infection in patients receiving prophylactic antibiot-

216 MURPHY, HEAL, AND BLUMBERG TRANSFUSION

Vol. 31, No. 3-1991

Table 3. Characteristics of patlents with infection or suspected infection as compared with those without evidence of infection*

Variable Infection No infection p value

Number of patients

Preoperative hematocrit (%) Discharge hematocrit (%) Units transfused Proportion of patients receiving homologous blood Length of stay (days) Days of fever > 38.3"C Days of antibiotics* Days of wound drainage Duration of surgery (minutes) Blood loss (mL) Admission white cell count per FL

Age 17

63 ? 17

34 2 3.1 2.4 f 0.5 16/17 (94%) 15.5 f 4.3 1.8 f 2.1 7.6 f 4.4 3.1 r 2.3 140 f 24 lo00 t 430 7900 f 2500

42 t 4.8

67 57 2 13 39 2 4.5 32 2 3.5 2.3 f 0.5 34/67 (51%) 12.3 f 2.6 1.0 f 1.2 2.3 f 0.7 2.9 ? 1.9 147 f 32 950 t 340 7900 f 4200

NSt 0.045 0.0531

NS 0.0029 0.0001 0.0478 0.0001 NS NS NS NS

'The values shown are mean t SD unless otherwise noted. Differences in proportions were determined by chi-square, with continuity correction if appropriate. The other p values were determined by analysis of variance. tNot significant (p > 0.10). Wl patients in both groups received 2 days of prophylactic intraoperative and postoperative cephalosporins.

ics may lead to many false-negative diagnoses. The patients studied all were thought by their physician to be infected, were given extended antibiotic therapy, and had longer hospital stays due to the infection. The elim- ination of several of these patients from the infected category would not alter the conclusions of this study. In any event, these complications, whether truly infectious or not in all instances, occurred at a rate 10 times lower in patients receiving autologous transfusions. We hypothesize that the observed differences in complication rate are due in large part to the deleterious effects of homologous transfusion on resistance to infection, rather than to inapparent surgical or other clinical variables that differ among the groups. Evidence exists of impaired immunity, including decreased phagocytic cell function, after homologous transfusions in both humans and animal^.^

Differences in iron status, particularly iron over- load, have been hypothesized as one possible cause of this immunologic dysfunction. We did not measure serum iron or ferritin; future studies of these variables would be of interest. However, autologous transfusion recipients probably are relatively iron loaded as compared with homologous recipients. Autologous donors commonly take oral iron for several weeks prior to operation. In any case, the blood removed at donation is usually entirely transfused intraoperatively or post- operatively. We doubt that decreased circulating iron is characteristic of autologous transfusion recipients, or that this factor could explain the differences we observed in infection prevalence.

Our findings in patients receiving homologous transfusions are strikingly similar, both quantitatively and qualitatively, to those of Tartter and his associates.'-' In those studies, postoperative bacterial infection rates of approximately 25 percent were observed in patients transfused with more than 1 unit of homologous blood

while undergoing abdominal surgery. Rates of only 4 to 8 percent were seen in those receiving no transfusions or 1 unit of homologous blood. Most of the infections occurred away from the abdominal wound site. These investigators searched for but did not find known surgical risk factors to account for the increased risk of bacterial infection in their patients who received homologous transfusions. Their studies did not include patients who were transfused with autologous blood. Our data demonstrate that recipients of two or more homologous transfusions, who are free of other predisposing infection risks, are much more likely to develop proven or suspected postoperative bacterial infections than are otherwise similar recipients of equivalent amounts of autologous blood. Our autologous and homologous recipients were similar or identical to each other in measures such as age, gender, admission WBC count, duration of operation, estimated blood loss, and amount and duration of wound drainage. Both groups were selected for freedom from conditions that might predispose to infectious complications. The infections that occurred were predomi- nately away from the wound site.

Our finding of a reduced risk of proven or suspected postoperative bacterial infection in autologous blood recipients is also of biologic interest. These data provide support for the hypothesis that the newly proposed complications of transfusion**9 are mediated by immunologic mechanisms. Immunologic differences between donor and recipient may be central to the diminished host defenses against infection seen after transfusion. It was hypothesized previously that the process of storing blood and transfusing the storage-damaged cells and debris could compromise host defense^,^ and this may yet be proven to be so. However, the present results argue against al- terations of self-antigens during blood storage, microag- gregate debris, or surgical stress as the predominant mediators of posttransfusion recipient immune dysfunc-

TRANSFUSION 1991-Vol. 31. No. 3 AUTOLOGOUS OR HOMOLOGOUS TRANSFUSIONS 217

tion. Autologous transfusion recipients are presumably subject to these influences to the same degree as are homologous transfusion recipients.

These data are the initial clinical observations sug- gesting that autologous transfusions are associated with a decreased risk of postoperative bacterial infection as compared with homologous blood transfusions. These results are consonant with findings in animal studies, but they must be confirmed by others. The use of autologous blood transfusions in patients undergoing elective surgery may be one way of decreasing morbidity due to infection.

Acknowledgments The authors are grateful for the clerical and secretarial assistance

of Carol Cole and the clerical and technical assistance of the staffs of the Transfusion Medicine Unit and Blood Bank at Strong Memorial Hospital.

1.

2.

3.

4.

5.

6.

7.

References Tartter PI, Quintero S, Barron DM. Perioperative blood transfusion associated with infectious complications after colorectal cancer operations. Am J Surg 1986;152:479-82. Tartter PI. Dreifuss RM, Malon AM, Heimann TM, Aufses AH. Relationship between postoperative septic complications and blood transfusions i n patients with Crohn’s disease. Am J Surg

Tartter PI. Blood transfusion and infectious complications follow- ing colorectal cancer surgery. Br J Surg 1988;75:789-92. Nichols RL, Smith JW, Klein DB, et al. Risk of infection after penetrating abdominal trauma. N Engl J Med 1984;311:1065-70. Waymack JP, Rapien J, Garnett D. Tweddell JS, Alexander JW. Effect of transfusion on immune function in a traumatized animal model. Arch Surg 1986;121:50-5. Waymack JP. Balakrishnan K. McNeal N, et al. Effect of blood transfusions on macrophage-lymphocyte interaction in an animal model. Ann Surg 1986;204:681-5. Opelz G, Terasaki PI. Improvement of kidney-graft survival with increased numbers of blood transfusions. N Engl J Med

1988;155:43-8.

1978;299:799-803.

8. Blumberg N, Heal JM, Murphy P, Agarwal MM, Chuang C. Association between transfusion of whole blood and recurrence of cancer. Br Med J [Clin Res] 1986;293:530-3.

9. Blumberg N, Heal JM. Transfusion and recipient immune function. Arch Pathol Lab Med 1989;113:246-53.

10. Williams JG, Hughes LE. Effect of Ixrioperative blood transfusion on recurrence of Crohn’s disease. Lancet 1989;2:131-3.

11. Blumberg N, Heal JM. Evidence for plasma-mediated immunom- odulation-transfusions of plasma-rich blood components are associated with a greater risk of acquired immunodeficiency syndrome than transfusions of red blood cells alone. Transplant Proc 1988;20:1138-42.

12. Jakobisiak M, Wlodarski K, Lasek W, Gorecki D, Podziszewska M. Transfusions of syngeneic blood do not seem to induce sup pressive effect on antitumor or transplantation immuniry in mice. Arch Immunol Ther Exp (Warsz) 1985;33:543-7.

13. Waymack JP, Chance WT. Effect of blood transfusions on immune function: IV. Effect on tumor growth. J Surg Oncol

14. Toy PT, Strauss RG, Stehling LC, et al. Predeposited autologous blood for elective surgery. A national multiccnter study. N Engl

15. Kruskall MS, Glazer EE, Leonard SS, et al. Utilization and ef- fectiveness of a hospital autologous preoperative blood donor pro- gram. Transfusion 1986;26:335-40.

16. Wasman J, Goodnough LT. Autologous blood donation for elective surgery. Effect on physician transfusion behavior. JAMA

17. Kay LA. The need for autologous blood transfusion (editorial). Br Med J [Clin Res] 1987;294:137-9.

18. Haley RW, Schaberg DR, McClish DK, et al. The accuracy of retrospective chart review in measuring nosocomial infection rates. Results of validation studies in pilot hospitals. Am J Epidemiol

1988;39: 159-64.

J Med 1987;316:517-20.

1987;258:3135-7.

1980;111:516-33.

Paul John Murphy, MD, Senior Resident, Transfusion Medicine Unit and Blood Bank, University of Rochester Medical Center; current address: Blood Bank, S U M Health Sciences Center at Syracuse, Syr- acuse, NY.

Joanna Mary Heal, MBBS. MRCP, Associate Clinical Professor of Medicine, Hematology Unit, and Associate Medical Director, Amer- ican Red Cross Blood Sewices, Rochester Region, Rochester. NY.

Neil Blumberg MD. Director, Transfusion Medicine Unit and Blood Bank, and Associate Professor of Pathology and Laboratory Medicine, University of Rochester Medical Center, Box 608, 601 Elmwood Av- enue, Rochester, NY 14642. [Reprint requests]

infection or suspected infection after hip replacement surgery with autologous or homologous blood...

Documents