Keeping bacteria out of the bag
James P. AuBuchon, MD
Miriam F. Leach, MS, MT(ASCP)SBB
Report of bacterial contamination
The era of modern blood banking began during World War II when anticoagulant solutions were developed that made it possible to store whole blood. Blood was collected in glass bottles in what today would be called an open system, and units in which bacteria had proliferated during storage were often infused, with catastrophic results.1
The development and commercial introduction of plastic bags for blood collection not only made large-scale production of components possible but also reduced the potential for external contamination of the blood. As a result, reports of bacterial contamination dropped sharply, and the problem was generally believed to have disappeared. With viral contamination of the blood supply garnering the lion's share of attention in the 1980s, bacterial contamination received scant notice. However, now that the risks of viral transmission have become vanishingly small, the ever-present threat of bacterial contamination has become recognized as one of the greatest risks of post-transfusion morbidity and mortality.
Sources of contamination
Most bacterial contamination emanates from the collection process rather than from the bags or their manipulation. Occasional instances of lots of collection bags being contaminated with an organism have been reported,2 but, thanks to Good Manufacturing Practices being applied to the bag manufacturing and sterilization processes, this is a rare event.3 Contamination arising from manipulation of units is essentially unknown in recent times because of the short periods during which components can be held after the closed system is violated4 and the ability to connect plastic tubing while maintaining a closed system and thus avoid contamination from the environment.5
Although donors must pass a stringent health history assessment before they are accepted for donation, some apparently healthy donors have a low-level bacteremia. Inoculation of even a few bacteria into a blood component may lead to proliferation to lethal levels during storage. The organisms contaminating blood units through donor bacteremia are usually gram-negative "enteric" pathogens, such as Yersinia enterocolitica, Escherichia coli, or Salmonella. Attempts to identify and interdict donors who are at higher risk of carrying these organisms, through historical screening, have proved to be relatively insensitive and nonspecific.6,7 As a result, such questioning has not been widely adopted. This source of contamination probably remains as the primary source for red blood cell units containing gram-negative organisms.
The most common source of contamination, however, is the skin of the donor. Ineffective antiseptic cleansing of the skin could lead to collection of skin organisms. The dimpling of skin that occurs with frequent donation may make it difficult to sterilize the phlebotomy site. Compounding the problem is the frequent "coring" of skin and subcutaneous structures that occurs with large-bore needles. These "skin plugs" may carry into the unit of blood bacteria harbored in adnexal structures.8 Consideration is being given to altering the construction of collection bags to divert the first 20-100 mL of blood into a separate bag to prevent contamination of the unit. Though this may provide additional safety, it is not likely to eliminate the bacterial contamination problem entirely because of the volumes that can be diverted without causing significant additional donor red cell loss.9,10
Therefore, most bacterial contaminations occur without suspicion when collecting blood from donors who appear to be healthy.
Fate of bacteria
Not all bacteria that enter a unit of blood survive to multiply due to soluble and cellular antibacterial mechanisms. Many gram-negative bacteria are inactivated by a complement-dependent mechanism11 such that more than 99 percent cannot be recovered after two hours at room temperature.12 Donors may also have preformed antibodies directed against specific bacteria or classes of bacteria that may lead to the clearance of contaminating organisms.13
Leukocytes are, of course, an important effector in antimicrobial defenses, and leukocytes remain active at least for several hours post-collection.14 While refrigeration retards their subsequent activity, some European blood collecting agencies have delayed refrigeration for up to 24 hours to allow leukocytes to phagocytize as many contaminating bacteria as possible. As leukocytes undergo cellular death during storage, however, they may release phagocytized bacteria that were not killed in their lysosomes, thus setting up the potential for bacteria to proliferate several days after storage begins and at a time with reduced leukocytic and complement activity to provide an effective defense.
Adsorption filtration to reduce the content of leukocytes in cellular blood components may also remove bacteria from units. Leukoreduction filtration within the first several days of storage may also reduce the risk of leukocytes releasing viable bacteria later in storage. In addition, some bacteria demonstrate adherence to some leukocyte filters directly when suspended in certain solutions.12 Thus, several laboratories have demonstrated that units subjected to adsorption filtration of leukocytes early in storage after intentional inoculation at collection have lower (or at least not higher) rates of bacterial proliferation than those not leukoreduced by filtration.15,17 These salubrious effects of leukocyte filtration cannot, however, guarantee sterility of a leukoreduced unit because not all species of bacteria at all concentrations are reliably removed.16
Leukocyte reduction by other means has the potential to increase the risk of bacterial contamination. Inoculation of bacteria into units that had already undergone leukoreduction resulted in accelerated bacterial growth compared with leukocyte-replete units.18 From a logistic point of view, this observation is not a practical problem for whole blood-derived platelet units because leukoreduction filtration is performed after, rather than before, the point that almost all contaminated units are seeded with bacteria. However, several manufacturers' recent developments in plateletpheresis technology allow a therapeutic dose of platelets to be collected from a single donor with so few leukocytes (usually less than 105) that the unit qualifies as "leukoreduced" without having to undergo filtration. If bacteria have been introduced into the unit via phlebotomy or via the donor's bloodstream, there are fewer leukocytes available to phagocytize and kill them.
From practical experience, however, the removal of leukocytes during apheresis does not appear to pose a greater risk of post-transfusion sepsis. Not only have the rates of sepsis following transfusion of platelets, pheresis not increased during a time that leukoreduction has been achieved to greater degrees, but the risk of morbidity and mortality are reduced 80 to 90 percent when using (primarily leukoreduced) platelets, pheresis rather than platelet concentrates derived from whole blood units.19 Single-donor platelets appear to provide platelet support with a significantly reduced risk of bacterial contamination.
Magnitude of the problem
The exact magnitude of the problem of bacterial contamination is difficult to define. Some studies have performed surveillance cultures on large numbers of units shortly after collection.20,21 The reported rates of contamination (as defined by bacteria recovery) were about 1:1,000 to 1:10,000 units. However, these studies may lack relevance and sensitivity. Detection of an organism does not equate with the clinical consequence of sepsis or even infection since the organism may not have proliferated under blood bank storage conditions. On the other hand, these results may underestimate contamination problems because they usually involve culture of small volumes that may fail to include one of the few contaminating organisms in the unit. These studies do document, however, that the risk of a unit containing bacteria is several orders of magnitude higher than the risk of a unit containing the commonly feared bloodborne viruses.
A more clinically relevant approach to defining the magnitude of the problem emanates from observations of clinical reactions attributable to bacterially contaminated units. However, the problem is likely to be underreported from the bedside of the patient. While contamination of red blood cells with gram-negative organisms and the endotoxin they generate is not likely to go unnoticed, contamination of units with gram-positive organisms may not lead to acute decompensation. Development of symptoms or infection at a later time may then not be etiologically connected with the earlier transfusion. Furthermore, many recipients of platelet units are immunosuppressed and neutropenic; as a result, they are susceptible to infection with the same types of organisms found to contaminate platelet units most often, and the association with transfusion may be missed. Because of their decreased resistance to infection, these same patients are often placed on prophylactic antibiotics, again making it difficult to recognize a contaminated transfusion.
Concerted efforts in many countries in the past several years to set up "hemovigilance" systems to detect all untoward occurrences following transfusion are yielding more complete information about the magnitude of the risk of bacterial contamination. A report from the French system for the first two years of operation (Nov. 1, 1996-Oct. 31, 1998) noted that six of the 10 transfusion-related deaths detected had been attributable to bacterial contamination of units, and the transfusion of 15 percent of all bacterially contaminated units had resulted in fatalities (see box).22
Other European countries have reported even higher contamination rates. Bacterial risk in Germany has been reported to be 400 per million (0.04) for red blood cells and 1,300 per million (0.13 percent) for platelet concentrate units. In Spain the reported risk is 320 per million (0.032 percent) for platelet concentrates.23,24 Recent reports of the results of routine culturing of units have documented that 10 of 5,046 (0.2 percent) plateletpheresis units were documented to contain bacteria in Belgium25, the same frequency noted in a Dutch study.26 To place these results in the context of the U.S. blood banking system, one might extrapolate the French results to expect approximately 30 deaths to occur annually as a result of bacterial contamination in addition to the morbidity associated with less severe reactions, or perhaps even higher if the German and Spanish observations are applicable. These projections would imply that the U.S. Study of Transfusion Reactions Caused by Bacterial Contamination of Blood and Blood Products (BaCon), which was being conducted with the cooperation of all major blood banking associations and agencies and the Centers for Disease Control and Prevention, is not capturing all the occurrences of this problem.27 Data from the first 18 months of the BaCon study extrapolate to a case rate of 9-16 per million platelet units. The fatality rate is projected at 0.25-1 per million units. Concurrently, a study at a U.S. institution documented a contamination rate of one per 1,500 platelet units.28 Thus, U.S. contamination rates are similar to those reported from other countries and document the universal nature of this problem in transfusion medicine.
These risks should be placed in context with other, more commonly reported risks, such as the risk of HIV at approximately 1:1,000,000 units and of hepatitis C at approximately 1:500,000 units.29,30
Methods of detection
A variety of techniques have been proposed to detect units that harbor potentially dangerous levels of bacteria. Those that pertain to platelet units relate primarily to means of detecting bacterial growth through consequences of bacterial metabolism. All detection techniques suffer from the inherent problem that bacterial concentrations are very low early in the storage period; the earlier the method is to be applied during storage, the closer its sensitivity limit must approach one organism/bag.
Bacteria consume glucose and produce organic acids through their metabolic cycles. Although platelets perform the same processes, their greater dependence on aerobic metabolic processes and their use of other substrates allow for detection of decreases in glucose concentration and falls in pH that are more rapid in bacterially contaminated units. Burstain and colleagues and Wagner and colleagues documented greater than 95 percent sensitivity using simple chemistry "dipsticks" to test stored platelet units for decreased glucose and pH levels during five days of storage.31,32 This success must be tempered, however, with the realization that the bacterial levels at which these tests extended beyond the reference range for sterile units approximated at least 107 bacteria/mL, well into the range considered to be potentially lethal for patients. Thus, although these techniques can be used to detect some units that contain bacteria prior to issuance for transfusion, they cannot be relied on in all circumstances. An additional problem is that these tests detect a large number of units that are not bacterially contaminated, adding expense and creating inventory management problems.28
A related approach is to examine platelet units for "swirling." An opalescent, shimmering appearance of the unit is observed with mild agitation and by holding the unit up to the light, so long as the platelets retain their normal, discoid form. With drops in the pH to < 6.0-6.2, platelets become spheroidal and thus lose their ability to align with the flow of the suspending medium. Since bacterial contamination usually leads to a drop in pH, the loss of shimmering can be used as a means to detect contaminated units.32,33 While the technique is not readily suitable for automation, it can be taught so as to be applied in a reproducible manner. However, even in experienced hands, observing platelet units for shimmering has a sensitivity to detect bacterial contamination that is not substantially better than using chemistry dipsticks32 and may yield an unacceptably high false positive rate.34
Taking a sample of the unit shortly before issuance and reviewing it for bacteria by a gram stain or other staining technique has also been reported to be beneficial.35 However, the detection limit for gram stain is approximately 106-107 per mL, and the sensitivities of fluorescent staining techniques, such as acridine orange, are only an order of magnitude better; most do not consider these techniques to be sufficiently sensitive to be relied on to ensure sterility. In addition, these stains can be difficult to interpret because of the small size of both platelets and bacteria, thus leading to reduced sensitivity and poor specificity.
A number of nucleic acid amplification techniques have been developed for this purpose. These are able to detect approximately 104-105 bacteria per mL, which is several orders of magnitude better than staining techniques and probably below the threshold for severe acute morbidity on transfusion.36,37 The techniques are relatively rapid (less than one hour) and could be used in a transfusion service laboratory. None have been commercially developed to the point of FDA approval, however, and are not generally available.
Blood centers have not implemented routine screening in this country because of the inability to culture a large enough volume early in storage to be able to document bacterial sterility. While early cultures might detect some units,23,38,39 only one study40 has documented the ability of such an approach to guarantee sterility by reculturing units at the end of the storage period. Furthermore, U.S. blood centers appear reluctant to release a unit at a time when a test result is still pending, as would have to be the case given the time required (80th percentile: 48 hours) for a culture to be detected as positive when beginning with a low-level inoculum.41
Culture of a component is regarded as the gold standard of highest sensitivity, but it must be applied properly to attain this sobriquet. As the initial bacterial content of a unit may be as low as one organism per 10-100 mL, the volume cultured and the timing of the sampling are critical. Optimal sensitivity would be obtained by culturing the entire unit's volume, but this is obviously not practical. Culturing a smaller volume risks missing the presence of the bacteria. While logistics would recommend culturing at the time all other tests are performed, the largest volume would need to be cultured earliest in the storage period. Delaying culturing to a later time increases the likelihood that the volume cultured would contain bacteria but increases the chance that a unit will be transfused with clinically significant concentrations of bacteria before the culture is obtained or found to be positive.
The surveillance studies reported in the literature21 generally have cultured 2-10 mL at the beginning of the storage period, finding 1-10/1,000 units contaminated. However, postponing culturing to the second day should provide greater predictive value of a negative result. Kinetic studies in plateletpheresis units inoculated with commonly encountered organisms to one colony forming unit (CFU)/mL and then subjected to repetitive 5 mL automated cultures illustrated that cultures were 100 percent sensitive beginning on the second day of storage.33 A similar study using a variety of organisms illustrated growth in all but one sample (sensitivity >99 percent) when cultures were taken at this point.40 However, it is recognized that some slower-growing organisms, such as some Staphylococcus epidermidis, may still occasionally be missed.42 Cultures taken for the purpose of quality control sterility checks can, of course, be collected at the end of the storage period; cultures taken to ensure sterility before release must be done early enough to be useful but late enough to be sensitive. Culturing on Day 2 may be a practical yet sensitive compromise between these competing requirements. In practical application of Day 2 culturing in a hospital transfusion service laboratory, the technique has shown to have a false positivity < 1/200 and allowed safe and effective use of units beyond the usual five days of storage.43
1. Mollison PL, Engelfriet CP, Contreras M. Blood Transfusion in Clinical Medicine. 10th ed. Oxford, England: Blackwell Science; 1997.
2. Heltberg O, Skov F, Gernere-Smidt P, et al. Nosocomial epidemic of Serratia marcescens septicemia ascribed to contaminated blood transfusion bags. Transfusion. 1993;33:221-227.
3. Food and Drug Administration. Current Good Manufacturing Practice for Finished Pharmaceuticals. Washington, DC: US Government Printing Office; April 1, 1999. 21CFR211.
4. Food and Drug Administration. General Biological Products Standards. Dating Periods for Licensed Biological Products. Washington, DC: US Government Printing Office; April 1, 1999. 21CFR610.53.
5. AuBuchon JP, Pickard C, Herschel L. Sterility of plastic tubing welds for components stored at room temperature. Transfusion. 1995;35:303-307.
6. Grossman BJ, Kollins P, Lau PM, et al. Screening blood donors for gastrointestinal illness: a strategy to eliminate carriers of Yersinia enterocolitica. Transfusion. 1991;31:500-501.
7. Katz L, MacPherson JL, Zuck TF. Yersinia and blood donation [letter]. Transfusion. 1992;32:191.
8. Perez P, Bruneau C, Chassaigne M, et al. Multivariate analysis of determinants of whole blood donation bacterial contamination [abstract]. Transfusion. 1999;39(suppl):S116-S117.
9. Wagner SJ, Robinette D, Fridman LI, et al. Use of a precollection diversion arm to reduce bacterial load in whole blood units. Transfusion. 1999;39(suppl):S12.
10. Bruneau C, Perez P, Chassaigne M, et al. Efficacy of a new collection procedure for preventing bacterial contamination of whole blood donations [abstract]. Transfusion. 1999;39(suppl):S116.
11. Gong J, Rawal BD, Hogman CF, et al. Complement killing of Yersinia enterocolitica and retention of the bacteria by leucocyte removal filters. Vox Sang. 1994;66:166-170.
12. AuBuchon JP, Pickard C. Leukodepletion and bacterial proliferation [letter]. Transfusion. 1993;33:533-534.
13. Gong J, Hogman CF, Hambraeus A, et al. Transfusion-transmitted Yersinia enterocolitica infection. Protection through buffy coat removal and failure of the bacteria to grow in platelet-rich or platelet-poor plasma. Vox Sang. 1993;65:42-46.
14. Wagner S, Moroff G, Katz A, et al. Bacterial levels in components, prepared from deliberately inoculated whole blood held for 8 and 24 hours at room temperature [abstract]. Transfusion. 1994;34(suppl):S9.
15. Buchholz DH, AuBuchon JP, Snyder EL, et al. Removal of Yersinia enterocolitica from AS-1 red cells. Transfusion. 1992;32: 667-672.
16. Buchholz DH, AuBuchon JP, Snyder EL, et al. Effects of leukocyte reduction on the resistance of blood components to bacterial multiplication. Transfusion. 1994;34:852-857.
17. Kim DM, Brecher ME, Bland LA, et al. Prestorage removal of Yersinia enterocolitica from red cells with white cell-reduction filters. Transfusion. 1992;32: 658-662.
18. Hogman CF, Gong J, Eriksson L, et al. White cells protect donor blood against bacterial contamination. Transfusion. 1991;31: 620-626.
19. Ness PM, Braine HG, King K, et al. Single-donor platelets reduce the risk of septic platelet transfusion reactions. Transfusion. 2001;41:857-861.
20. Blajchman MA. Bacterial contamination of blood products and the value of pre-transfusion testing. Immunol Invest. 1995;24:163-170.
21. Goldman M, Blajchman MA. Blood product-associated bacterial sepsis. Transfus Med Rev. 1991;5:73-83.
22. Perez P, Salmi RL, Folléa G, et al. Transfusion-associated bacterial contamination: description of 41 cases from the French National Case Control study (BACTHEM) [abstract]. Transfusion. 1999;39(suppl):S2.
23. Walther-Wenke G, Doerner R, Montag-Lessing T, et al. Bacterial contamination of blood com-ponents: results of a two-year
survey in Germany working party "Microbiological Investigations in Transfusion Medicine" of the Advisory Board of the German Ministry of Health ("Arbeitskreisblut") [abstract]. Transfusion. 1999;39(suppl):S34.
24. Castro EM, Bueno JL, Gonzalez R, et al. Implementation of an automated bacterial culture of platelet products in the blood bank routine [abstract]. Transfusion. 1999;39(suppl):S75.
25. Schelstraete B, Bijnens B, Wuyts G. Prevalence of bacteria in leucodepleted pooled platelet concentrates and apheresis platelets: Current status in the Flemish Blood Service [abstract]. Vox Sang. 2000;78(S1):370.
26. Claeys H, Verhaegle B. Bacterial screening of platelets [abstract]. Vox Sang. 2000;78(S1):374.
27. From the Centers for Disease Control and Prevention. Red blood cell transfusions contaminated with Yersinia enterocolitica—United States, 1991-1996, and initiation of a national study to detect bacteria-associated transfusion reactions. JAMA. 1997;278:196-197.
28. Mhawech PY, Werch J, Stager C, et al. Detecting bacterial contamination in platelet concentrates using reagent strips—application in a major cancer center blood bank [abstract]. Transfusion. 1999;39(suppl):S36.
29. Hotaling S, Kolk D, Baker D, et al. Significant closure of the HCV detection window with the TMA HIV-1/HCV assay [abstract]. Transfusion. 1999;39(suppl):S67.
30. Kolk D, Hotaling S, Baker D, et al. Significant closure of the HIV-1 detection window with the TMA HIV-1/HCV assay [abstract]. Transfusion. 1999; 39(suppl):S70.
31. Burstain JM, Brecher ME, Workman K, et al. Rapid identification of bacterially contaminated platelets using reagent strips: glucose and pH analysis as markers of bacterial metabolism. Transfusion. 1997;37:255-258.
32. Wagner SJ, Robinette D. Evaluation of swirling, pH, and glucose tests for the detection of bacterial contamination in platelet concentrates. Transfusion. 1996;36:989-993.
33. Leach MF, Pickard CA, Herschel LH, et al. Evaluation of glucose and pH test strips for detection of microbial contaminants in apheresis platelets [abstract]. Transfusion. 1998;38:S89.
34. Bertolini F, Murphy S. A multicenter inspection of the swirling phenomenon in platelet concentrates prepared in routine practice. Transfusion. 1996;36:128-132.
35. Yomtovian R, Lazarus HM, Goodnough LT, et al. A prospective microbiologic surveillance program to detect and prevent the transfusion of bacterially contaminated platelets. Transfusion. 1993;33:902-909.
36. Brecher ME, Hogan JJ, Boothe G, et al. Platelet bacterial contamination and the use of a chemiluminescence-linked universal bacterial ribosomal RNA gene probe. Transfusion. 1994;34:750-755.
37. Brecher ME, Boothe G, Kerr A. The use of a chemiluminescence-linked universal bacterial ribosomal RNA gene probe and blood gas analysis for the rapid detection of bacterial contamination in white cell-reduced and nonreduced platelets. Transfusion. 1993;33:450-457.
38. De Korte D, Welle F, Marcelis J, et al. Determination of the prevalence of bacterial contamination of whole blood collections using Bacti/Alert. Transfusion. 1999;39(suppl):S20.
39. Schelstraete B, Bijnens BJ, Wuyts G. Prevalence of bacteria in leuco-depleted pooled platelet concentrates and apheresis platelets: a 3 year experience [abstract]. Transfusion. 1999;39(suppl):S88-S89.
40. Castro EM, Bueno JL, Gonzalez R, et al. Implementation of an automated bacterial culture of platelet products in the blood bank routine. Vox Sang. 2000;74(suppl 2).
41. Brecher ME, Means N, Jere CS, et al. Evaluation of BacT/Alert (BTA) microbial detection system testing of platelets [abstract]. Transfusion. 1999;39(suppl):S75.
42. Brecher ME, Means N, Jere CS, et al. Evaluation of an automated culture system for detecting bacterial contamination of platelets: an analysis with 15 contaminating organisms. Transfusion. 2001;41:477-482.
43. AuBuchon JP, Cooper LK, Leach MF, et al. Bacterial culture of platelet units and extension of storage to seven days. Transfusion. 2001;41:1S.
Dr. AuBuchon and Miriam Leach are in the Department of Pathology, Dartmouth-Hitchcock Medical Center, Lebanon, NH.
Dr. AuBuchon is vice chair of the CAPTransfusion Medicine Resource Committee.