Is the BacT/ALERT 3D the most accurate, time, volume, and cost efficient method of detecting bacterial contamination of platelets?
Michelle
Dixon,
IDST 6400 Critical Literature Review and Scientific Writing
Spring 2005
Dr. Craig Scanlan
Dr. Elaine Keohane
Introduction
Platelet transfusions are utilized to treat a variety of disorders from
thrombocytopenic bleeding in patients with hematologic
malignancies to major surgical bleeding and trauma. Two million platelet
transfusions are administered per year in the
In March of 2004, the American Association of Blood Banks (AABB) mandated that “the blood bank or transfusion service shall have methods to limit and detect bacterial contamination in all platelet components (standard 5.1.5.1)” (as cited in Brecher & Hay, 2004, p.123). This mandate is complicated by the decision of the Food and Drug Administration (FDA) to limit the shelf-life of platelets from seven to five days in 1986. (Esber, 1986) “Testing for bacterial organisms must be performed on platelet product samples procured after a variable holding period following donation. Bacteria present in a platelet unit at the time of donation are almost always undetectable because of the exceedingly small quantity in the contaminating inoculum” (Yomtovian, 2004, p. 455). Culture methods currently require 24 to 48 hours of growth, leaving threes days for optimum use of platelets. This has caused a shortage of platelets due to outdating of units. According to a survey conducted by the AABB following implementation, “between 5- 20% of hospitals reported platelet shortages that the institutions felt were due to bacterial testing” (as cited in Roseff, 2004, p. 112).
The problem of finding a suitable detection system is further compounded by
sample volume requirements of each system. The volume collected from
random donor units processed from whole blood collections is between 40 mL and
60 mL and for apheresis platelets between 200 mL and 500 mL (Heal & Blumberg, 2004). Testing
single units of random donor platelets is impractical because of the sample
volume that must be removed (Roseff, 2004). The
Several methods are currently employed to detect bacterial contamination of
platelets including the BacT/ALERT 3D system by bioMerieux, Bacterial Detection System (
The value of finding a system that meets all necessary requirements could decrease the platelet shortage and outdating of units, save lives and money associated with transfusion reactions, and possibly provide for longer storage times. The purpose of this review is to determine if the BacT/ALERT 3D system by bioMerieux, which is a popular method used for detection of bacterial contamination, is the most accurate, and cost, volume, and time efficient method available.
Methods
The following Ovid databases were selected to provide the original and
secondary literature research: Journals @ Ovid, Medline from 1996 through
week three of February 2005, and CINAHL from 1982 through week three of
February 2005. Evidence Based Medicine guidelines were searched utilizing
Cochrane Database of Systematic Reviews (DSR),
Subject headings (MeSH) and key words were utilized that directly related to the research question. These included platelet, platelet pheresis, platelet transfusion, sepsis, bacteria, laboratory testing, and detection. Other words that were included were laboratory techniques and procedures, bacterial contamination, diagnostic tests, and blood platelets. Boolean OR was used to combine bacteria and sepsis. Consistently, the term bacteria was combined with the terms related to platelets and detection methods in various combinations to uncover articles that were pertinent. The search was not limited to random clinical trials due to the lack of sufficient evidence based materials in the field of laboratory medicine. Several relevant articles were retrieved that were higher levels of evidence, but all articles regarding methodologies to detect platelet contamination by bacteria were selected. Terms were also truncated to retrieve various spelling and allow a more through search of the literature. Only articles written in English were retained for final use.
For the EBSCO search of Academic Search Premier, an advanced search was
performed using the terms bacterial contamination, platelets, and detection.
The terms were combined with the Boolean operator
All materials were analyzed for references that would be beneficial , yet had not been previously retrieved. The total number of articles screened for this review was 168. In all, 17 articles of original research and nine articles of secondary research were selected as pertinent to the review topic. An example of an Ovid Medline search from 1996 through week three of February 2005 is included in the Appendix.
Literature Review
BacT/ALERT 3D System
The BacT/ALERT 3D system (bioMerieux), is an FDA approved system that operates on the principle of bacterial proliferation generating carbon dioxide (CO2). The bottles utilized for testing contain a colorimetric sensor that turns yellow as the CO2 increases.(BacT/ALERT 3D, retrieved 2005) Bottles are examined approximately every 10 minutes and computer software monitors the rate of change of the colorimetric sensor and the absolute color change of the sensor. (Brecher & Hay, 2004)
Two retrospective studies and four methodological studies were found related to the BacT/ALERT 3D system. Only two of the studies related directly to detection times of this system. The other four studies were examining the differences in culture bottles or the actual amount of bacterial contamination in the platelet inventory utilizing the Bact/ALERT 3D. Some relevant information, however, was obtained from these studies. In two of the studies, it was apparent that positive cultures were not obtainable upon repeat testing. Macauley, Chandrasekar, Geddis, Morris, and McClellend (2002) tested 4885 leukoreduced platelet concentrates over a 12 month period and found 28 initial positives of which only 13 were repeat positive (46%). Likewise, AuBuchon, Cooper, Leach, Zuaro and Schwartzman (2002) tested 2678 units of leukoreduced apheresis platelets over a 24 month period. Sixteen of these units grew bacteria and 13 were available for retesting. All 13 upon repeat testing produced negative results. The researchers contributed these results to false positives due to sample handling and not actual bacterial contamination of the units. This emphasizes the need for careful inoculation of the platelets into the culture bottles for accurate results.
Munksgaard, Albjerg, Lillevang, Gahrn-Hansen and Georgsen (2004) performed a six-year retrospective analysis of the BacT/ALERT 3D system at a university hospital blood bank. The sample size was more than adequate, testing 22, 057 random donor and apheresis platelet units that ranged from zero to seven days old. Ten mL of platelets were inoculated into aerobic bottles only. Upon acquiring a positive signal, a confirmation and identification of the organism was performed by standard culture method. The system was found to have a sensitivity of 83%, but specificity of the method could not be determined because negative units were not evaluated. The researchers noticed that the years with higher levels of contamination coincided with the years of excessive employee turnover and training.
Since timing is of the utmost importance in detecting bacterial contamination, Brecher, Means, Jere, Heath, Rothenberg, and Stutzman (2001) examined the detection time of the BacT/ALERT 3D system. In this study, 15 species of bacteria at two different concentrations were inoculated into platelets and tested on day two. Four mL of platelets were placed into various culture bottles and time to detection was noted. The system had a mean detection time at the low inoculum of 10.2 – 25.6 hours with all species except Propionibacterium acnes, which had a mean detection time of 86. 2 hours. The low inoculum contained 10 Colony Forming Units (CFU) of bacteria per mL. For the high inoculum,(100 CFU/mL) the mean detection time for the bacteria was 9.2- 20.8 hours with P. acnes having a detection time of 74.4 hours. Baseline sterility of the platelets was ensured prior to spiking.
A two year retrospective study of the clinical significance of bacteriologic screening was performed by Boekhorst, Beckers, Vos, Vermeij, and van Rhenen (2005). In this study, 28,104 pooled platelet concentrates were screened for contamination and 203 were found to be positive. Repeat positive cultures were obtained on 184 concentrates giving a sensitivity of 91%. The specificity was not given due to the scope of the research evaluating the amount of contamination of platelets and not testing the machine itself. The detection times were given with mean detection ranging from 11.9 hours for Steptococcus hemolyticus to 123.2 hours for Propionibacterium acnes. It was noted in the discussion section that the results from this study were contradictory to other findings presented by other researchers who stated that bacteria could be fully detected in 48 hours. Other studies showed that bacteria could be detected 98.1% of the time within 24 hours (Boekhorst, et al., 2005).
The weakness of the study performed by Boekhorst et al. (2005) is that it does not address specificity of the BacT/ALERT 3D. Also, although the total units screened represent a large sample population, the small number of positive cultures is not enough to adequately assess the sensitivity. The strength of this article is in the discussion of conflicting findings regarding detection times and possible causes.
Bacterial Detection System (
The
Ortolano, Freundlich, Holme, Russell, Cortus, Wilkins, et al. (2003) determined the 19.5% oxygen cutoff value for non-contaminated platelets by testing 155 one-day old white blood cell reduced platelet concentrates. These concentrates were known to be negative for bacteria and their oxygen readings taken after an incubation, 24, and 96 hours. The mean -3SD was used to provide the cutoff of 19.5%. Platelet units measuring a %O2 of less than 19.5 % were considered contaminated.
After determining the criterion for establishing bacterial contamination, 202 platelet concentrates were inoculated with 10 species of bacteria that are implicated in 98% of the transfusion reactions associated with platelets (Ortolano, et al., 2003). Two concentrations of bacteria were used, 100 Colony Forming Units (CFU)/ mL and 500 CFU/mL, and after a 30 minute incubation two to three mL of the platelets were removed for processing. After passing through a platelet reducing filter, the platelets were exposed to sodium polyanetholsufonate (SPS) in the sample bag, which enhances growth of gram negative organisms (Ortolano, et al., 2003). Platelet units were then tested after 24 hours of incubation and 195 out of 202 units were found to be positive giving a sensitivity of 96.5%. One bottle was inadvertently discarded after 24 hours, but the other 6 samples were positive after the 30 hour incubation (100% sensitivity).
The strengths of this study include the thorough detail in which the methods are described, the limitation of testing explained, the choice of organisms selected, and the study being performed in two independent laboratories. A weakness exists however, in that there were no specificities given or data to enable calculation. Of interest is that this system will not detect strict anaerobic bacteria. This will be addressed in the section entitled Considerations.
The second study found pertaining to the
The strengths of this evaluation by Rock et al. (2004) of the
Scansystem
The third system reviewed is the Scansystem by Hemosystem, which employs a platelet kit that allows bacteria to be stained with fluorescent dye and analysis to proceed by use of an Argon laser. Four parameters are used to discriminate between bacteria and various other particles. The parameters include (1) position on the membrane, (2) intensity of fluorescence, (3) size, and (4) shape (Scansystem, retrieved 2005) At the time of this review, this system is not FDA approved, but current is available in Europe, the Middle East, Latin America, and Africa. Two methodological studies follow.
The first study, performed by Jacobs, Bajaksouzian, Windau, Palavecino, and Yomtovian (2005) utilized leuko-reduced single donor apheresis platelet units. In this study, platelets were inoculated with 10 species of bacteria that commonly cause platelet contamination. Low and high inoculates were prepared that ranged from 5-57 CFU/mL and 70-680 CFU/mL respectively. Ten replicates of each organism were performed for each inoculum producing 200 total samples. The utility of the Scansystem was compared to that of the BacT/ALERT 3D system. All replicates were positive with the Scansystem at 30 hours, which is the minimum time required by the system. It was also noted that at time zero, 83 of 200 samples tested positive on the Scansystem. All uninoculated platelets were negative.
For the comparison, 10 replicate bottles of aerobic and anaerobic bottles were inoculated with a one mL sample from the bacterially spiked units. Three mL of uninoculated platelets were added to provide the total four mL required for the BacT/ALERT system. The bottles were incubated until a positive signal from the system was received (Jacobs, et al., 2005). The detection times of the BacT/ALERT system for the low and high inoculates were 10.0 to 20.4 hours and 9.3 to 24.0 hours respectively. All uninoculated platelets produced negative results.
The strengths of this evaluation of the Scansystem includes the detail of the methodology, the table of data, the bacteria tested, the diagrams of the Scansystem, and the two levels used to inoculate the units. One weakness is the lack of specificity due to non-enumeration of negative units. The study does specify that both systems detect negative units, but the numbers that are actually tested is not given. Another weakness is the lack of explanation of the volume used for inoculation in the BacT/ALERT system. It is my assumption that the one mL inoculated platelets mixed with the three mL uninoculated platelets provides the same level of bacteria that is tested in the Scansystem but this information is not clearly detailed. Also, the cost effectiveness of the Scansystem is not addressed, which would further assist in this review.
The second study evaluated for this review was performed by Ribault, Harper, Grave, Lafontaine, Nannini, Raimondo, et al. (2004). Three mL of platelet concentrates (apheresis units or pooled platelet concentrates) were tested between two to four days following collection. Nine strains of bacteria were inoculated into the platelet units at various concentrations of 10, 102, 103, and 104 CFU/mL. All platelet concentrates were checked prior to spiking and found negative. The concentrations of all bacteria spiked were confirmed by quantitative culture on Mueller-Hinton agar plates or esterase labeling (Ribault, et al., 2004). One mL of a platelet aggregation solution was added to remove the confounding affects of platelets. A nucleic acid binding dye in distilled water was added and the mixture was allowed to incubate for 40 minutes on a flatbed rocker at room temperature. The platelet aggregates were removed via filtration and incubated for 20 minutes at room temperature with seven mL of a permeabilizing and labeling reagent. The solution was then filtered through a 0.4 um pore filter to retain bacteria. The filter was transferred to the Scansystem solid phase cytometer. Bacteria was detected in various platelet units at all concentrations. At 103 CFU/mL, all bacteria were detected and at 102 CFU/mL, 89% were positive. The total detection time from sampling to result was approximately 90 minutes.
The strengths of this study are in the use of various concentrations and nine bacterial strains being tested. However, Propionibacterium acnes, which is typically slow-growing, is not included in the study. Another weakness is the technical descriptions of the testing which would make it hard to duplicate. Of note, the study is supported by the manufacturer of the Scansystem, which could lend itself to bias. However, no bias appears to be present. The procedure is not clearly defined and assumptions are used to comprehend some of the testing and sample numbers.
Reagent Strips
Reagent strips for glucose and pH detection have been utilized in combination with observation of platelet swirling to detect bacterial contamination of platelets. Glucose and pH levels decrease in platelets upon storing. Platelet swirling is the technique of visually inspecting components under fluorescent light for evidence of “streaming” or a pearly appearance to indicate viable platelets (Wagner & Robinette, 1996). Two methodological studies are evaluated for this review.
Wagner and Robinette (1996) studied the sensitivity of pH and glucose values obtained from reagent strips and compared them to the cessation of platelet swirling as an indicator of bacterial contamination. Platelet concentrates were inoculated on day zero with one of seven strains of bacteria. The platelets were stored under normal platelet conditions of room temperature and agitation. The platelets were tested morning and evening for the presence of swirling. Additionally, one mL was removed from the unit for testing with the pH and glucose reagent strips and a manual culture was performed. The pH and glucose levels of the reagent strips were confirmed with reference methods.
For baseline determinations of uncontaminated platelet concentrates, Wagner and Robinette (1996) used 104 freshly outdated units. Of the 104, 99 had pH values that indicated no bacterial contamination and five had pH concentrations below the baseline limit. Four of the five platelets with low pH also had low glucose levels. Five units of the 50 that were tested for swirling, failed to swirl.
An abnormal pH was defined as less than or equal to 6.0 and an abnormal glucose was determined to be less than 127 ug/dL. The concentration of the bacteria was determined when abnormal pH or glucose values were obtained or when cessation of swirling was noticed. For the seven strains studied, concentrations reached 107 – 108 CFU/mL before abnormalities occurred. There was one exception noted that maintained a normal pH.
The strengths of Wagner and Robinette’s (1996) study are tests being compared to reference standards and the detailed explanation of the materials and methods. The study would be easily reproduced in any type of setting and could be performed by inexperienced personnel. The only weakness noted is that of small sample size used in testing.
Werch, Mhawech, Stager, Banez, and Lichtiger (2002) also performed a study to evaluate reagent strips. In this study, 3093 random donor platelet concentrates were prepared from whole blood. Platelets were stored at 20-24 ºC with gentle agitation and segments were checked for glucose and pH on Bayer’s Multistix 10 SG urine reagent strips. A glucose level less than 250 mg/dL and a pH of less than 7.0 were considered positive for bacterial contamination. An abnormal result was obtained for 30 of the 3093 units. All 30 were pH positive and 17 of the 30 were positive for glucose. The platelets were confirmed for contamination by culture and/or staining methods. Only two of the 30 units presumed to contain bacteria were confirmed by reference methods to actually contain bacteria, indicating a high false positivity.
The strengths of this study are the methods and quality control sections of the study clearly detailed, positive and negative controls performed with each test, unfavorable results reported, sample size adequate, and the ease of reproducibility. The only conceivable weakness is the single site performance of the test.
Various Other Methods
Three other studies were obtained that described methods to detect bacterial
contamination. All were methodological studies with various strengths and
weaknesses. The methods assayed were microvolume
fluorimetry, epifluorescence
microscopy, and reverse transcriptase polymerase chain reaction (RT-
Brecher, Wong, Chen, Vampola, and Rocco (2000) described using single donor apheresis platelet units inoculated with Staphylococcus epidermidis in varying concentrations. A vancomycin labeled fluorescent probe was used to detect bacterial contamination. Twenty-two samples were analyzed and all samples with a concentration of 105 CFU/mL or greater were detected. The procedure took less than one hour and involved three pipetting steps.
There are many weaknesses of the study performed by Brecher, Wong, et al. (2000) and very few strengths. First of all, the procedure is complicated and would be impossible to reproduce in most laboratories without having the extensive equipment required. Also, the test only involves one strain of bacteria, which is not representative of the array of bacteria that typically cause contamination of platelets. Only 22 samples are tested and this represents a small sample volume. The test would need to be performed on more platelet units with various strains of bacteria to adequately analyze the method. The one strength is that the samples are blinded, but this strength is quickly negated by the various other weaknesses of the study.
Seaver, Crookston,
The strengths of Seaver et al. (2001) are the detailed description of the methods section, the limitations of the procedure clearly noted, and the control samples receiving the same treatment as the platelet samples. The weaknesses of this study include the lack of statistical analysis, the unclear outcomes, the poor representation of bacteria, and the ability to reproduce this complicated procedure.
Finally, Dreier, Stormer, and Kleesiek
(2004) described using RT-
Dreier et al. (2004) have produced an extensively technical study that is difficult to assess. The methodology is complicated and not easily reproduced. The outcomes of the study are hard to assess due to the technical nature of the paper. The one strength that is apparent is the acknowledgement of the procedure’s limitations.
Considerations
One consideration that needs to be addressed is the issue of slow-growing bacteria. Several of the studies mentioned here noted that bacteria such as P. acnes was not detected for a long time. The danger of these bacteria as a threat to the platelet concentrates has not been adequately assessed. They are not implicated in any transfusion reactions to date, but have been discovered in contaminated platelets.
Another issue that has not been adequately evaluated is the potential of
contamination with strict or obligate anaerobes. The new platelet storage
bags provide a highly oxygenated environment and the likelihood of an obligate
anaerobe thriving is slim. None of the bacteria commonly implicated in
platelet transfusion reactions are anaerobic. Most methodologies in this
review did not address anaerobic bacteria. The
Another note of consideration is the recent FDA approval of a new
Discussion
When testing for bacterial contamination, an equal balance of cost, volume, and time efficiency, as well as ease of performance, and accuracy are all needed. At the present time and based on this review, no such system currently exists. Each method seems to have its limitations. Some are cheaper, yet less sensitive, while others are highly sensitive but cost prohibitive and many times labor intensive. In general, all articles retrieved for this review lack a cost analysis sufficient to analyze the cost effectiveness of the BacT/ALERT 3D. When dealing with the nation’s blood supply, cost pales in comparison to safety.
Concerning volume
efficiency, the BacT/ALERT 3D, according to this
review, utilizes anywhere from four mL to 10 mL. Other methods use less
volume, but may not be as feasible in routine testing. The Scansystem only uses three mL, the
Where accuracy is
concerned, no studies found in this review directly assess the sensitivity and
specificity of the BacT/ALERT 3D. Although
sensitivity could be inferred from some of the studies performed, there was not
sufficient evidence to clearly state that this is the best system. None
of the studies address the specificity in any of the methods retrieved.
The
The major
disadvantage to the BacT/ALERT 3D system is the time
incurred in testing the units that are ultimately negative for bacteria.
For most bacteriologic assays, a 24 hour incubation is
required to allow bacterial proliferation to a detectable limit.
Platelets at day zero typically do not have bacterial levels that can be
detected by most of the assays mentioned here. The exception is the
highly technical
In summary, the BacT/ALERT 3D is a useful system that
can detect low levels of bacterial contamination of platelet units, but may not
be the best assay available considering all the parameters reviewed.
Other assays that were evaluated such as the
References
AuBuchon, J. P., Cooper, L. K., Leach, M. F., Zuaro, D. E., & Schwartzman, J. D. (2002). Experience with universal bacterial culturing to detect contamination of apheresis platelet units in a hospital transfusion service. Transfusion, 42, 855-861.
BacT/ALERT
3D technology. (n.d.).
Retrieved
Boekhorst, P. A. W., Beckers, E. A. M., Vos, M. C., Vermeij, H., & van Rhenen, D. J. (2005). Clinical significance of bacteriologic screening in platelet concentrates. Transfusion, 45, 514-519.
Brecher, M. E., & Hay, S. N. (2004). Improving platelet safety: Bacterial contamination of platelets. Current Hematology Reports, 3, 121-127.
Brecher, M. E., Hay, S. N., & Rothenberg, S. J. (2004). Evaluation of a new generation of plastic culture bottles with an automated microbial detection system for nine common contaminating organisms found in plt components. Transfusion, 44, 359-363.
Brecher, M. E., Heath, D. G., Hay, S. N., Rothenberg, S. J., & Stutzman, L. C. (2002). Evaluation of a new generation of culture bottle using an automated bacterial culture system for detecting nine common contaminating organisms found in platelet components. Transfusion, 42, 774-779.
Brecher, M. E., Means, N., Jere, C. S., Heath, D., Rothenberg, S., & Stutzman, L. C. (2001). Evaluation of an automated culture system for detecting bacterial contamination of platelets: An analysis with 15 contaminating organisms. Transfusion, 41, 477-482.
Brecher, M. E., Wong, E. C. C., Chen, S. E., Vampola, C., & Rocco, R. M. (2000). Antibiotic-labeled probes and microvolume fluorimetry for the rapid detection of bacterial contamination in platelet components: A preliminary report. Transfusion, 40, 411-413.
Chaney, R., Rider, J., & Pamphilon, D. (1999). Direct detection of bacteria in cellular blood products using bacterial ribosomal rna-directed probes coupled to electrochemiluminescence. Transfusion Medicine, 9, 177-188.
Dreier, J., Stormer, M., & Kleesiek, K. (2004). Two novel real-time reverse transcriptase pcr assays for rapid detection of bacterial contamination in platelet concentrates.Journal of Clinical Microbiology, 42, 4759-4764.
Esber, E. C. (1986, June 02) Reduction of the maximum platelet storage period to 5 days in an approved container. Retrieved from http://www.fda.gov/cber/bldmem/060286.pdf
Enhanced bacterial management system. (n.d.). Retrieved
FDA clears
Pall enhanced bacteria detection system. (n.d.). Retrieved
Jacobs, M. R., Bajaksouzian, S., Windau, A., Palavecino, E. L.,
& Yomtovian, R. (2005). Evaluation
of the scansystem method for detection of bacterially
contaminated platelets. Transfusion, 45, 265-269.
Macauley, A., Chandrasekar, A., Geddis, G.,
Morris, K. G., & McClelland, W. M. (2003). Operational
feasibility of routine bacterial monitoring of platelets. Transfusion
Medicine, 13, 189-195.
Munksgaard, L., Albjerg, L., Lillevang, S. T., Gahrn-Hansen, B., & Georgsen, J. (2004). Detection of bacterial contamination of platelet components: six years’ experience with the BacT/ALERT system. Transfusion, 44, 1166-1173.
Ortolano, G. A., Freundlich, L.
F., Holme, S., Russell, R. L., Cortus,
M. A., Wilkins, K., et al. (2003). Detection of bacteria in wbc-reduced
plt concentrates using percent oxygen as a marker for
bacteria growth. Transfusion, 43, 1276-1285.
Ribault, S., Harper, K., Grave, L., Lafontaine,
C., Nannini, P., Raimondo, A., et. al. (2004). Rapid screening method
for detection of bacteria in platelet concentrates. Journal of
Clinical Microbiology, 42, 1903-1908.
Rock, G., Neurath, D., Toye, B., Sutton, D., Giulivi, A., Bormanis, J., et al. (2004). The use of a bacteria detection system to evaluate bacterial contamination in plt concentrates. Transfusion, 44, 337-342.
Roseff, S. D. (2004). Regulatory and quality initiatives-trends in transfusion medicine. Lab Medicine, 36, 108-113.
Scansystem. (n.d.).
Retrieved
Seaver, M., Crookston, J. C.,
Wagner, S. J., & Robinette, D. (1996). Evaluation of swirling, ph, and glucose tests for the detection of bacterial contamination in platelet concentrates. Transfusion, 36, 989-993.
Werch, J. B., Mhawech, P.,
Stager, C. E., Banez, E. I., & Lichtiger, B. (2002). Detecting bacteria in platelet
concentrates by use of reagent strips. Transfusion, 42, 1027-1031.
Yomtovian, R. (2002). Novel methods for detection of platelet bacterial contamination. Vox Sanguinis, 83, 129-131.
Yomtovian, R. (2004). Bacterial contamination of blood: Lessons from the past and road map for the future. Transfusion, 44, 450-460.
Appendix
Database: Ovid MEDLINE(R) <1996 to February Week 3 2005>
Search Strategy:
--------------------------------------------------------------------------------
1 (bacteria$ or sepsis$).mp. [mp=title, original title, abstract, name of substance word, subject heading word] (239315)
2 exp Platelet Transfusion/ (1220)
3 1 and 2 (126)
4 from 3 keep 1-2,4-10,12,14-18,20-21,24-27,29,32,34,36,39-42,48-49,51,58-62,64,66-67,69,71,73,75,78,85-86,89-90,93-94,96-97,100-101,104,108-109,111,113,117,119,122,126 (64)
5 exp "Laboratory Techniques and Procedures"/ (312842)
6 Blood Platelets/ (10326)
7 1 and 5 and 6 (100)
8 from 7 keep 1-2,5,9-10,13,15-19,24-26,30,41,50,65-66,70-72,75,79,81,84,89,91,93 (29)
9 4 or 8 (81)
10 from 9 keep 1-81 (81)