Quantitative Aspects of Septicemia

Transcription

1 CLINICAL MICROBIOLOGY REVIEWS, JUly 1990, p Vol. 3, No /90/ $02.00/0 Copyright i 1990, American Society for Microbiology Quantitative Aspects of Septicemia PABLO YAGUPSKY' AND FREDERICK S. NOLTE2* Department of Microbiology and Immunology, University of Rochester Medical Center, Rochester, New York 14642,1 and Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia INTRODUCTION DEVELOPMENT OF QUANTITATIVE BLOOD CULTURE METHODS CLINICAL SIGNIFICANCE OF QUANTITATIVE BLOOD CULTURES Severity of Infection in Children and Adults Efficacy of Antibiotic Therapy Endocarditis Catheter-Related Sepsis (CRS) Contaminants versus True Pathogens in Positive Blood Cultures CONCLUSIONS LITERATURE CITED INTRODUCTION Blood cultures are a crucial part of the evaluation of patients with suspected sepsis, and a positive blood culture is the best criterion for defining that condition. In most clinical microbiology laboratories, routine blood cultures are accomplished by using broth media. Although broth-based blood culture systems provide a sensitive means for recovering microorganisms from blood, they provide no information about the number of microorganisms present in the original blood sample. In contrast, inoculation of blood onto solid media results in growth of individual colonies which can be enumerated and the result can be used to calculate the concentration of bacteria per milliliter of blood. Until recently, the methods available for quantitative blood cultures have been cumbersome and labor intensive and did not find widespread use in clinical microbiology laboratories. In 1976, Dorn et al. (18, 19) described a blood culture technique based on lysis of cellular elements in blood followed by concentration of microorganisms by centrifugation and plating of the sediment on solid media. Improvements were made in the original technique, and a blood culture system based on this method (Isolator Microbial Tube) was made commercially available by E. I. du Pont de Nemours & Co., Inc., Wilmington, Del. A recent survey of blood culture methods used in 288 clinical microbiology laboratories revealed that 13% used the Isolator system for routine bacteriologic culture and 30% used it for fungal culture (K. S. C. Kehl, Clin. Microbiol. Newsl. 8: , 1986). The availability of the Isolator system has sparked a renewed interest in the quantitation of microorganisms in patient blood. In this article, we review the development of quantitative blood culture methods and published applications of these methods. We also critically evaluate the available data on possible diagnostic and prognostic significance of quantitative blood culture results and identify areas for future research. The subject of quantitative blood cultures has also been reviewed recently by Kiehn (42). * Corresponding author. 269 DEVELOPMENT OF QUANTITATIVE BLOOD CULTURE METHODS The first quantitative blood cultures were performed by the pour plate or spread plate techniques. The pour plate technique involves mixing blood and molten agar; with the spread plate technique, blood is spread over the surface of an agar plate. The number of colonies that grow either in or on the agar is enumerated, and the CFU per milliliter of blood are calculated. Both methods suffer because only a small volume, usually 1 ml of blood, can be inoculated. To maximize the recovery of microorganisms, it is recommended that at least 10 ml of blood be cultured from adult patients (36, 89, 94). As a result, these methods are of practical value only when the quantity of bacteria in blood is relatively large. The magnitude of bacteremia in infants and children is generally greater than that in adults so cultures of as little as 0.5 to 1.0 ml of blood are considered adequate in most cases (94). Direct plating methods have been used in pediatrics to enhance the recovery of commonly encountered pathogens such as Haemophilus influenza and Neisseria meningitidis and to study the correlation between the magnitude of bacteremia and clinical manifestations (52, 53). Quantitative blood culture studies in pediatrics have used several methods, including pour plate (1, 25); spread plate (75); quantitative direct plating, using heparin tubes (53); and lysis direct plating, using the Isolator 1.5 Microbial Tube (10-12, 98). La Scolea et al. (53) described a quantitative direct plating method in which 0.5 to 1.0 ml of blood was collected into a heparin tube; upon arrival in the laboratory, the blood was inoculated onto agar plates. The quantitative direct plating method was compared with broth methods for detection of bacteremia in children (53). It detected 83% of the cultures positive for H. influenza and 100% of those positive for N. meningitidis, but only 50% of the cultures positive for Streptococcus pneumoniae. The number of other pathogens recovered was too small for evaluation. Of those cultures positive by both quantitative direct plating and a broth procedure, 55% were positive first by quantitative direct plating. The Isolator 1.5 Microbial Tube was designed for culturing small-volume (0.5- to 1.5-ml) pediatric blood samples directly to agar media. It contains 0.96 mg of the anticoagulant sodium polyanetholsulfonate and 0.1 ml of a saponin-con-

2 270 YAGUPSKY AND NOLTE training solution to lyse the cellular components of the blood. The lysis step allows the recovery of viable phagocytized microorganisms. The Isolator 1.5 tubes have been compared with conventional and radiometric broth methods in several clinical trials (10, 12, 98). In none of the trials did the Isolator system recover significantly more organisms than the comparative method, and in only one trial did the Isolator system provide positive results more rapidly. However, in each trial there were substantial reductions in the time required to obtain isolated colonies with the Isolator system. Large increases in the contamination rate for Isolator cultures were reported in each evaluation and averaged 10.9%. All of the direct plating methods described above are practical only for small volumes of blood. Concentrationdirect plating methods were developed to maximize the detection of low-magnitude bacteremia or fungemia and to remove inhibitory factors that may be present in blood. The basic concept behind these methods is that blood cells are lysed, microorganisms are concentrated either by membrane filtration or centrifugation, and, finally, the concentrate is plated directly to solid media. Since the concentration procedures effectively separate microorganisms from inhibitory substances in blood, these methods potentially improve the recovery of microorganisms. Brown and Kelsh (9) were the first to report use of a membrane filter technique for recovery of Brucella spp. from the blood. Tidwell and Gee (88) found that detection of experimental bacteremia in rabbits was faster and more reliable with a filter technique than with a conventional broth method. In another study (69), the kinetics of Salmonella typhi bacteremia during therapy were followed with a membrane filtration technique. These early studies used cumbersome and time-consuming methods in which only small volumes of blood were processed for culture. Winn et al. (104) and Finegold et al. (27) reported a differential membrane filtration technique which allowed the processing of larger blood volumes and separate culture of different blood components. The formed elements of blood were separated by differential centrifugation and plasma only was passed through a membrane filter. The filter was placed on the surface of an agar plate, and the erythrocytes and leukocytes were cultured separately on solid media. Isolated colonies were available after 24 to 48 h of incubation, even in patients treated with antibiotics. The technique provided quantitative information and also facilitated the diagnosis of polymicrobic bacteremia. The basic membrane filtration technique was modified by Stanaszek (79), Sullivan et al. (82-84), and Zierdt et al. (106) to include a solution that lysed blood cells before filtration. By using this modified technique, transient bacteremia following common urologic procedures was demonstrated, as well as superior recovery of gram-positive bacteria in both experimental and human clinical studies. Komorowski and Farmer (48) found that a lysis-filtration technique was superior to a conventional broth culture for detection of candidemia. Lamberg et al. (50) described an improved filtration method that did not require lysis of blood cells or sophisticated equipment. The blood is applied to a Ficoll-Hypaque gradient, centrifuged, and filtered. In a study with seeded blood samples, this system had a sensitivity equivalent to the broth method used for comparison. In spite of these promising advances, membrane filtration techniques remain cumbersome, expensive, and time-consuming, and no commercially available system is based on this technique. In 1976, Dorn et al. (18, 19) described a concentration technique for blood cultures termed lysis-centrifugation. In CLIN. MICROBIOL. REV. this two-step process, the blood was lysed in one tube and the lysate was transferred to a second tube, where microorganisms were concentrated by centrifugation onto a stabilizing density layer composed of 50% sucrose and 1.5% gelatin. After centrifugation, most of the supernatant was removed with a needle and syringe. Aliquots of the sediment were then plated to agar media. Dorn et al. used the new technique in a study of 1,000 blood cultures from patients suspected of having bacteremia (18). The blood was also cultured in Trypticase soy broth and in pour plates. Of 176 significant positive cultures, the lysis-centrifugation technique recovered 73% and the pour plate and broth methods recovered 49 and 38%, respectively. The lysis-centrifugation technique demonstrated enhanced recovery of fungi, grampositive cocci, and Pseudomonas spp. The superior performance of the lysis-centrifugation technique could be due to the high theoretical dilution of antimicrobial factors (1:400) or to protection provided to cell-wall-damaged bacteria by the hypertonic sucrose-gelatin cushion. Although this new technique showed promise, it was cumbersome and had a high (9.3%) rate of contamination. In 1978, Dom and Smith (22) described a new centrifugation blood culture device. It was essentially a doublestoppered evacuated glass tube containing 0.3 ml of a dense hydrophobic cushion, 1.2 ml of an aqueous solution of sodium polyanetholsulfonate, a lysing agent, and an antifoaming agent. Blood was introduced into the tube and mixed thoroughly, and the tube was centrifuged at 3,000 x g for 30 min in a fixed-angle rotor. After centrifugation, the majority of the supernatant was removed and discarded. The remaining 1.5 ml of solution was equally dispersed to five agar plates. A clinical trial of this new blood culture device was conducted with 3,335 blood cultures (20). All blood samples were cultured concurrently in supplemented peptone broth with sodium polyanetholsulfonate under a C02-rich atmosphere. The centrifugation technique detected 80% of positive cultures compared with 67% for the broth method. Superior recovery of Staphylococcus aureus, Pseudomonas spp., and yeasts was demonstrated, and the time required for detection of a positive blood culture was shorter for the centrifugation method. The contamination rate observed with the new method, 1.4%, was comparable to that of the broth method. A lysis-centrifugation device similar to that developed by Dom et al. is commercially available as the Isolator 7.5- or 10-ml Microbial Tubes. The Isolator system has been compared with both conventional and radiometric broth and biphasic blood cultures in many clinical trials and overall had demonstrated good performance (5-7, 33, 34, 38-40, 46, 63). Certain common findings emerged from these evaluations of the Isolator system. The lysis-centrifugation method recovered significantly more fungi than conventional broth, radiometric broth, or biphasic brain heart infusion bottles. The superiority of the lysis-centrifugation method for the recovery of mycobacteria (30, 45), especially Mycobacterium avium-m. intracellulare complex (31, 43, 44, 58, 104), Cryptococcus neoformans (5, 6), and Histoplasma capsulatum (5, 100), from blood has been useful in diagnosis of these infections in immunocompromised patients. Enhanced isolation of staphylococci and members of the family Enterobacteriaceae was also found in several studies (6, 33, 34, 38, 39, 46,63). However, the Isolator system may not be optimal for recovery of Streptococcus pneumoniae, Pseudomonas aeruginosa, and anaerobic bacteria (6, 33, 34, 39, 46, 63). The Isolator system was superior to conventional broth or

3 VOL. 3, 1990 biphasic broth cultures in terms of speed of detection of positive cultures but, on average, was no faster than the radiometric method. Positive cultures with the lysis-centrifugation system occur as colonies on agar plates, and therefore this system offers potential advantages in the time required to identify and determine antibiotic susceptibility of isolates. However, for laboratories performing direct identification and antibiotic susceptibility tests from positive blood culture bottles, this advantage may be less meaningful. The incidence of polymicrobic bacteremia has been increasing steadily during the last decades, possibly as a result of the expanding population of immunocompromised patients, aggressive surgery, and widespread use of intravascular medical devices (41, 95, 96). Detection of polymicrobic bacteremia is extremely important to select the most appropriate antimicrobial therapy. Blood cultures on solid media have some potential advantages over broth cultures because overgrowth of cultures by one organism is less likely to occur on solid media, and morphologic differences between isolated colonies are easily recognized (34). Clinical trials have confirmed the superiority of blood cultures performed on solid media for recovery of multiple organisms from blood. In one study, 14 of 15 episodes of polymicrobic bacteremia were detected by lysis-centrifugation as opposed to only 3 of 15 detected by the broth culture method (40). In a second study that included 25 episodes of polymicrobic bacteremia, the Isolator system detected 21 episodes and the biphasic bottle detected only 17 (39). Blood cultures on solid media have also been shown to be superior to broth methods for diagnosis of polymicrobic bacteremia in catheter-related infections (70, 72). One significant disadvantage of the Isolator system is the relatively high contamination rate. This occurs because the lysis-centrifugation method requires additional processing steps and the use of petri plates rather than closed bottles, which offer greater opportunity to introduce contaminants. Reported contamination rates for the Isolator system range from 0.3 to 13.8% (81). Rates in different laboratories depend on definition of contamination, experience with the system, use of relatively dry solid media, and inoculation of plates in a laminar flow hood. It appears that use of a laminar flow cabinet is a critical factor for reducing contamination rates to an acceptable level. In one study, the contamination rate dropped from 9.5 to 6.2% (87) and in another study it dropped from 9.6 to 2.5% (94) when lysis-centrifugation blood cultures were processed in a laminar flow cabinet. An important concern for laboratories that use direct plating methods is the length of time that blood is held in the tube before processing. Cashman et al. (13) conducted a study with blood cultures seeded with low numbers of 26 different bacterial species and demonstrated that a holding time of 15 h might be feasible for the Isolator tube. However, the authors observed substantial increases in numbers of some species and significant declines in others over the 15-h period. The quantity of bacteria recovered from specimens held for this length of time would bear little relationship to the quantity of bacteria originally present in the blood. Stockman et al. (80), in a prospective study of 5,125 fungal blood cultures, found that lysis-centrifugation tubes processed within 9 h showed a significantly higher yield of yeasts, filamentous fungi, and bacteria than did those processed after 9 h of hold time. With any of the direct plating blood culture methods, including the Isolator system, specimens should be processed as quickly as possible to enhance QUANTITATIVE ASPECTS OF SEPTICEMIA 271 recovery of pathogens and to ensure that colony counts truly reflect the quantity of bacteria in vivo. CLINICAL SIGNIFICANCE OF QUANTITATIVE BLOOD CULTURES Severity of Infection in Children and Adults The clinical spectrum of bacteremia is extremely wide, ranging from a self-limited septic event after trauma or manipulation of colonized mucosal surfaces to overwhelming and often fatal infections. It has been estimated that every year in the United States half of the 200,000 patients with septicemia die in spite of the progress achieved in antibiotic and supportive therapy (92, 94). Great efforts have been made to identify factors that contribute to this high case mortality rate. For more than 30 years, several authors have used quantitative blood cultures to investigate the magnitude of bacteremia in septic patients and its correlation with the occurrence of complications and death. The results of these studies have shown that most episodes of clinically significant bacteremia in adults are characterized by low numbers of circulating bacteria. Werner et al. (99) found that 54% of blood cultures from patients with staphylococcal and streptococcal endocarditis contained between 1 and 30 CFU/ml of blood. In a study of adults, Henry et al. found <1 CFU/ml of blood in 27% of episodes of bacteremia due to Staphylococcus aureus, in 62% of those caused by Escherichia coli, and in 55% of patients with septicemia due to P. aeruginosa (34). Kreger et al. (49) found that 73% of 77 patients with gram-negative bacteremia had blood cultures which contained <10 bacteria per ml of blood. In children, the magnitude of bacteremia is usually much higher than in adults and in general is inversely related to the child's age, with the highest numbers of bacteria found in the blood of septic neonates. In neonatal E. coli sepsis, Dietzman et al. (16) demonstrated that 78% of patients had >5 CFU/ml of blood and one-third had bacterial counts in excess of 1,000 CFU/ml. Studies performed in children with bacteremic diseases caused by encapsulated bacteria have In a series of 20 adults with bacteremia and septic shown bacterial counts higher than 100 CFU/ml in 43% of the patients (3, 51, 52, 54, 62, 86, 98). Marshall and Bell (62) used quantitative blood cultures to study 83 episodes of H. influenzae bacteremia in children and found that those patients with high bacterial densities tended to be younger than those with low-grade bacteremia. A similar relationship between age and magnitude of H. influenzae bacteremia was demonstrated by Moxon and Ostrow in the infant rat model (65). The relationship between magnitude of bacteremia and severity of clinical picture has also been extensively investigated. shock, Hall and Gold (32) found that all six patients with >100 CFU/ml of blood died while only 41% of those with lesser bacteremia died. Although the number of microorganisms in blood showed a moderate correlation with the occurrence and severity of septic shock, some patients with bacterial counts as high as 300 CFU/ml of blood did not suffer shock. In another study, Weil and Spink found a correlation between magnitude of bacteremia in patients with gram-negative sepsis and fatal outcome (95). The mortality rate for eight patients with <5 CFU/ml of blood was 50%, while in 19 cases in which >5 CFU were isolated from patients' blood, mortality increased to 84%. Although in a recently published study no correlation

4 272 YAGUPSKY AND NOLTE between magnitude of bacteremia and mortality was found among 55 patients with Staphylococcus aureus septicemia (61), data on gram-negative sepsis have consistently shown that high-magnitude bacteremia is associated with poor prognosis. DuPont and Spink (24) reviewed 860 cases of gram-negative bacteremia at the University of Minnesota Medical Center from 1958 to Among the 655 adults, there was a definitive correlation between the number of colonies counted in pour plates and mortality. This correlation was seen in patients suffering from underlying diseases of relatively good or intermediate prognosis, whereas patients with poor prognosis usually died regardless of bacterial counts. Similar results were reported in a study of 242 bacteremic patients seen at the University of Maryland Hospital from 1965 to 1972 (47). A statistically significant correlation was found between mortality and magnitude of gram-negative bacteremia, while a similar trend was found among patients with gram-positive and polymicrobic bacteremia. The authors also stressed the role played by the nature and severity of the underlying disease in determining the clinical outcome. The relationship among gram-negative bacteremia, underlying disease, and mortality rate was also investigated by Kreger et al. (49). They found that patients with more severe underlying disease did not necessarily have higher bacterial counts in the blood. However, the density of bacteremia was related to fatality rates among patients with underlying diseases of similar severity, and of all patients combined, the mortality rate was higher in those with > 10 bacteria per ml of blood. It is apparent that the magnitude of bacteremia in adult patients correlates with mortality rate. However, because bacteremia usually results from breakdown of defense mechanisms that control infection, underlying debilitating diseases appear to be a more important factor than the number of CFU per milliliter of blood in determining the ultimate prognosis. Intravascular foci of infection such as endocarditis and catheter-related sepsis may be important exceptions to these generalizations about the relationship between the magnitude of bacteremia and outcome. Whimbey et al. (101) analyzed colony counts in 15 patients with central venous catheter-related Staphylococcus aureus bacteremia and found that the magnitude of bacteremia was directly related to the incidence of metastatic complications. Six of 7 patients with CFU/ml in blood from a vessel had metastatic foci of infection, whereas none of 7 patients with c 10 CFU/ml had metastatic complications (P < 0.01). The remaining patient had 20 CFU/ml of blood and did not develop metastatic infection. Three patients with, and no patients without, metastatic infections died. Correlation between magnitude of bacteremia and severity of clinical disease in pediatric patients has been firmly established. Using the pour plate technique, Dietzman et al. (16) studied the quantitative aspects of E. coli septicemia in 30 neonates and found a bimodal distribution of bacterial density: 65% of the patients had <200 CFU/ml, and 31% had bacterial counts of >1,000 CFU/ml. Only 1 of 9 patients with <1,000 CFU/ml had meningitis as compared with 6 of 11 with colony counts of >1,000 CFU/ml. In a study of 83 H. influenzae type b bacteremic episodes, it was demonstrated that children with high-density bacteremia had a significantly shorter duration of illness before presentation (median, 1 day) than did low-grade bacteremic patients (median, 3 days), and such children had a higher rate of complications including secondary foci of infection (62). Other studies involving older children with bacteremic CLIN. MICROBIOL. REV. infections caused by three major pediatric pathogens, H. influenza type b, Streptococcus pneumoniae, and N. meningitidis, also demonstrated a clear relationship between severity of disease and bacterial counts (3, 25, 54, 55, 62, 75, 85, 86, 98). For H. influenza, the combined results of these studies show that 50 of 72 (69.4%) patients with meningitis and 19 of 28 (67.9%) patients with epiglottitis had bacterial counts of >100 CFU/ml, but only 8 of 42 (19.0%) patients with less severe infections such as pneumonia, cellulitis, osteomyelitis, arthritis, otitis media, or bacteremia without demonstrable focus had bacterial densities as great (3, 54, 55, 62, 86, 98). Among the patients with S. pneumoniae bacteremia, 6 of 8 (75%) with meningitis, but only 3 of 55 (5.4%) with less severe diseases, had counts of >100 CFU/ ml (3, 25, 86, 98). The combined results for children with N. meningitidis infections show that 15 of 24 (62.5%) patients with meningitis and half of those with Waterhouse- Friderichsen syndrome have >100 CFU/ml. Bacterial counts of <100 CFU/ml were observed in 7 of 10 (70%) cases with mild focal meningococcal disease or occult bacteremia (85, 86). In summary, 71 of 104 (68.3%) children with meningitis due to H. influenza, S. pneumoniae, or N. meningitidis but only 37 of 145 (25.5%) with other bacteremic diseases caused by these organisms had bacterial counts of >100 CFU/ml (P < 0.001). Febrile diseases of infectious origin are more common in early childhood than during any other period of human life. Although most febrile children suffer from self-limited illness, 3 to 15% of febrile children between the ages of 6 and 24 months have bacteremic infections that may result in severe complications and death (3, 37). Clinical signs are not always reliable criteria for the identification of bacteremia. Occasionally, an ill child managed as an outpatient with a "benign" infection is already bacteremic and will later develop meningitis (86). Identification of febrile children at risk of complications of bacteremia is important to ensure appropriate evaluation and antibiotic therapy. In one report, four pediatric outpatients who presented with seemingly mild respiratory infections and otitis media had blood cultures obtained because of high fever (85). Three developed meningitis due to S. pneumoniae or N. meningitidis, and the fourth developed H. influenza epiglottitis. In each case, counts on the initial blood culture were >100 CFU/ml. In a second study, three children initially admitted without evidence of focal infection, in whom N. meningitidis was recovered from blood in excess of 500 CFU/ml, developed meningitis (85). Modern quantitative blood culture techniques allow both rapid detection of bacteremia and provide information on its magnitude. When high bacterial counts in the blood of a child are found, careful reevaluation is necessary to identify and treat potential life-threatening complications especially in those children sent home after initial evaluation. Although a relationship has been established between the severity of disease and bacterial density in the blood for H. influenza, S. pneumoniae, and N. meningititis, more studies are needed to establish this relationship for other pathogens such as group B beta-hemolytic streptococci and Staphylococcus aureus. Efficacy of Antibiotic Therapy Detection of bacteremia by a positive blood culture followed by antibiotic susceptibility testing of the isolate are crucial for confirmation of the diagnosis and rational selection of antibiotics. It is accepted that administration of

5 VOL. 3, 1990 appropriate antibiotic therapy should result in negative blood cultures. Persistence of positive blood cultures is interpreted as evidence of treatment failure, and changes in antibiotic therapy are usually made. However, in certain clinical situations such as infections in immunocompromised patients, M. avium-m. intracellulare complex infections in patients with acquired immunodeficiency syndrome, or the slowly resolving bacteremia of typhoid fever, prompt sterilization of the blood does not always occur. Serial quantitative blood cultures have potential value as an in vivo test of antibiotic efficacy. Using a membrane-filtration technique, Randriambololona and Dodin (69) demonstrated that administration of chloramphenicol to patients with typhoid fever results in gradual decrease in the magnitude of bacteremia over 3 to 8 days, followed by negative cultures. Normalization of body temperature occurs close to the time of clearance of Salmonella typhi from the blood. Marshall and Bell (62) recently reported the results of a quantitative blood culture study in 81 bacteremic children. They showed that children with partially treated H. influenzae meningitis had lower bacterial counts in blood than untreated controls, and another report (102) demonstrated that, in five appropriately treated patients with persistently positive blood cultures, broth cultures concealed a prompt decline in the number of bacteria circulating in the blood seen in quantitative blood cultures. The declining colony counts supported the clinical assessment that patients responded to therapy. In 1982, Fojtasek and Kelly (30) reported that quantitative blood cultures were useful for monitoring the response to antibiotic therapy in a renal transplant recipient with M. chelonae bacteremia. Mycobacterial counts initially ranged from 50 to 70 CFU/ml and dropped to 45 CFU/ml after 2 weeks of amikacin therapy. Cefotaxime was added to the antibiotic regimen and the density dropped to 5 CFU/ml after several days of combined therapy. The counts rose to 25 to 30 CFU/ml when cefotaxime alone was continued. By this time, MIC data were available and showed that the organism was resistant to cefotaxime and susceptible to amikacin and cefoxitin. The antibiotic therapy was changed to cefoxitin; the level of mycobacteremia decreased (0.2 to 2 CFU/ml) and remained at that level for several weeks. Blood cultures became negative after removal of the transplanted kidney and one of the arteriovenous fistulas used for hemodialysis. With the emergence of acquired immunodeficiency syndrome, disseminated M. avium-m. intracellulare complex infections have become common. The disease is characterized by heavy infiltration of the reticuloendothelial system by the bacterium and by prolonged high-level mycobacteremia (44, 58, 105) resembling that seen in lepromatous leprosy (23). Due to difficulties in assessing the clinical response to antibiotic therapy, serial quantitative blood cultures have been used to monitor the effectiveness of antibiotic treatment (58, 105). In some patients bacterial counts decreased or blood cultures became negative with appropriate therapy, but in other patients counts remained stable or even increased. Regardless of the quantitative changes in the magnitude of bacteremia, diffuse infiltration of various organs by mycobacteria was observed in all patients at autopsy, indicating that, as in lepromatous leprosy, controlling bacteremia does not guarantee control of tissue infection (105). Some patients with septicemia deteriorate clinically in spite of antibiotic therapy that is highly effective against the infecting organism in vitro. Using quantitative blood cultures and a Limulus lysate gelation assay, Shenep et al. (76) showed that the amount of total circulating endotoxin in QUANTITATIVE ASPECTS OF SEPTICEMIA 273 patients with gram-negative septicemia was related to the magnitude of bacteremia and that the level of free endotoxin rose sharply after administration of antibiotics. They postulated that destruction of bacterial cells by antimicrobial agents releases endotoxin and other bacterial products sequestered within the intact bacterial cell into the bloodstream, aggravating the host cell injury and worsening the clinical disease. The serial quantitation of plasma endotoxin levels and bacteremia helps explain the paradoxical response sometimes seen in patients treated appropriately for gramnegative sepsis. However, neither the level of bacteremia nor the level of endotoxemia alone determined the clinical severity of gram-negative sepsis in this study. Quantitative blood cultures can provide useful information about in vivo efficacy of antibiotic therapy in patients with persistently positive blood cultures; however, further studies are needed to extend these observations to larger groups of patients with septicemia. Serial quantitative blood cultures obtained from patients receiving antibiotics could be used to study the in vivo relevance of certain in vitro phenomenon, such as antibiotic tolerance and antibiotic synergy. Endocarditis Bacterial endocarditis, a uniformly fatal disease before the antibiotic era, was the subject of several early quantitative blood culture studies. More than 70 years ago, Warren and Herrick (90) used the pour plate technique to show that a broad range of bacterial counts, from one to several hundred CFU per milliliter, occurred in the blood of patients with endocarditis. In 1925, Libman demonstrated that magnitude of bacteremia in a particular patient remains fairly constant during the course of the disease (56). Seven years later, Weiss and Ottenberg (97) studied the relationship between number of circulating bacteria and febrile response in patients with endocarditis. The results showed that there was a constant and relatively uniform number of bacteria released into the blood from vegetations, with periodic liberation of larger numbers of bacteria into the circulation ("bacterial showers"). In four cases of subacute endocarditis, the rise and fall in the magnitude of bacteremia was followed by corresponding changes in patient body temperature. However, fever spikes lagged behind the peak levels of bacteremia by several hours. The authors suggested that diagnostic blood cultures should be obtained at periods of low temperature just prior to the next expected temperature rise to take advantage of the large number of bacteria present in blood and thus maximize the chances to detect the septic episode. They also confirmed previous observations that the severity of disease, measured by patient survival period, was not related to bacterial counts in peripheral blood. A unique study was performed by Beeson et al. (2) shortly before the advent of effective antibiotic therapy for endocarditis when the disease course was not substantially modified by available therapy. Patients frequently suffered for months before dying from cardiac or embolic complications, allowing the opportunity for studies that provided important clues for understanding the pathophysiology of bacteremia. Blood was drawn simultaneously from catheterized femoral arteries and veins, hepatic and renal veins, superior vena cava, and right auricle in a group of six patients. In general, bacterial counts in the femoral vein were slightly lower than in the corresponding arterial cultures, indicating that some circulating bacteria were retained in the peripheral vascular bed, a fact consistent with the

6 274 YAGUPSKY AND NOLTE frequent occurrence of embolic lesions in the extremities of patients with endocarditis. The greatest reduction in bacterial counts was found in the hepatic veins (97% decrease), showing the importance of the liver and spleen as sites of removal of circulating microorganisms. In another study, the sensitivity of arterial, venous, and bone marrow cultures in the detection of bacteremia was examined in 88 patients with endocarditis due to viridans streptococci (74). By using the pour plate technique, it was shown that bacterial counts in arterial and venous blood were approximately the same magnitude. In this study, cultures of bone marrow were superior to arterial and venous cultures in detecting bacteremia, but no quantitative information for bone marrow cultures was provided. In 1967, Werner et al. (99) published a retrospective review of 206 cases of bacterial endocarditis seen over 17 years at the New York Hospital-Cornell Medical Center. They showed that the magnitude of bacteremia in streptococcal and staphylococcal endocarditis was generally of low order and the number of bacteria per milliliter of blood was relatively constant. In this study, the cumulative detection rates of the first two cultures was 98.5% in cases of streptococcal endocarditis and 100% in cases caused by other organisms. The clinical implication is that, in patients with endocarditis, blood cultures should be persistently positive; therefore, it is seldom necessary to obtain more than two or three blood cultures over a 24-h period to make the diagnosis. Quantitative blood cultures have also been used to localize the site of infection in patients with endocarditis, and especially in those with involvement of the right side of the heart. At the time when these studies were conducted, modern sonographic techniques were not available, and diagnosis of right-sided endocarditis was particularly difficult. In one study, published in 1973, Reyes et al. performed intraoperative quantitative blood cultures from each cardiac chamber during a valvulotomy on a grossly infected tricuspid valve (71). Culture of the right ventricle yielded 19 CFU of P. aeruginosa per ml, while all other cultures were negative. In 1974, Pazin et al. performed quantitative blood cultures of cardiac chambers and great vessels during the course of cardiac catheterization of two intravenous drug abusers with bacteremia (68). Finding of unusually high bacterial counts distal to a valve was considered diagnostic of involvement of that particular valve in the disease. The technique succeeded in confirming endocarditis as the cause of patient bacteremia, in excluding other potential sources of sepsis, and in localizing the site of the infection. Quantitative blood cultures have contributed to our understanding of the basic pathophysiology of bacterial endocarditis. The bacteremia is continuous, with the number of bacteria per milliliter of blood usually low and relatively constant. Fever spikes lag behind the peak levels of bacteremia by several hours in patients with endocarditis. The early quantitative studies of bacteremia in endocarditis helped form the basis for the current recommendations for the number and timing of blood cultures (89, 91, 93). The use of serial quantitative blood cultures to monitor the efficacy of antibiotic therapy of endocarditis should be investigated. Catheter-Related Sepsis (CRS) Intravenous infusion therapy has become a cornerstone of modern medical management. Total parenteral nutrition, prolonged chemotherapy for cancer, and intensive care of critically ill patients have led to the development of new CLIN. MICROBIOL. REV. techniques for establishing access to central veins. In 1973, Broviac et al. reported the use of a silicone-rubber catheter that was inserted into a central vein but exited at a distant site after passing through a subcutaneous tunnel (8). A small cuff encircling the catheter at the exit site became fixed in place by fibrotic tissue and provided mechanical stability and a barrier to infection. Unfortunately, emergence of a new sort of intravascular infection has been one of the untoward effects of central venous catheterization. Infections of these tunneled, cuffed silicone catheters are categorized into three nonexclusive groups: septic infections, exit site infections, and tunnel infections. Although the risk of death from CRS is low due to the low pathogenicity of the microorganisms usually involved, infection of intravascular devices is today's leading cause of nosocomial bacteremia (15). Catheter-related exit site and tunnel infections are readily diagnosed on the basis of local inflammation, purulence, fever, and leukocytosis. The diagnosis of septic infections, however, is more difficult due to the lack of localizing signs and the fact that patients requiring central venous access are among the most debilitated patients, often with multiple potential sources of bacteremia. Several techniques have been developed to help diagnose CRS in these cases, including the semiquantitative catheter tip culture pioneered by Maki et al. (60), the use of Gram stain or impression smears of the catheter, and quantitative cultures of the catheter obtained by flushing the catheter with nutrient broth (14). A major drawback with these techniques is that the catheter must be removed from the patient before the diagnosis can be made. In one report, more than two-thirds of the central venous catheters removed because of suspected CRS were found to be sterile or proved not to be the source of the septicemia (64). Quantitative blood cultures have been advocated for diagnosis of CRS. In 1979, Wing and colleagues (103) documented CRS in a patient with a Broviac catheter, using pour plate blood cultures. Blood drawn from a peripheral vein grew 25 CFU of Enterobacter cloacae and Citrobacter freundii per ml. Blood drawn from the catheter grew >10,000 CFU of the same organisms per ml. The catheter was removed and found to have an adherent clot present at the tip. Culture of the clot also grew E. cloacae and C. freundii. Whimbey et al. (102) reported two cases in which large differences in colony counts in blood obtained simultaneously from a peripheral vein and the Broviac catheter were used to identify the catheter as the source of infection. In one series of 28 children with Broviac catheters, 100 surveillance blood cultures were obtained through the catheter in afebrile, clinically well patients and from both a peripheral vein and the central catheter during 30 febrile episodes (70). All blood samples were cultured both by the spread plate method and in nutrient broth. Of the 100 surveillance cultures, 79 were negative by both methods. Eighteen cultures showed growth, but the catheters were not considered infected because <10 colonies grew on the plates, multiple organisms or organisms of doubtful clinical significance were recovered, or concordance between plates and broths was lacking. Subsequent catheter surveillance cultures were negative in all 18 patients. The remaining three positive surveillance cultures were from the same child and grew >100 CFU of Bacillus cereus per ml. Although no illness resulted, the catheter was removed and a Bacillus species and P. aeruginosa were recovered from the tip. These findings were considered consistent with contamination of the cultures or asymptomatic catheter colonization. In the group of febrile patients, CRS was diagnosed nine

7 VOL. 3, 1990 times in seven children. In each case, colony counts in the blood drawn through the catheter were at least 10 times greater than those found in peripheral blood. In six cases, blood obtained from the catheter contained >2,000 CFU/ml, of which four were >5,000 CFU/ml. Quantitative cultures in two patients with septicemia not attributable to the catheter had low colony counts in blood from the catheter. Surveillance blood cultures from the catheters were not helpful in predicting subsequent septicemia in any patients. In a short communication by Myint and Lowes (S. Myint and A. Lowes, Lancet ii: , 1985), bacterial counts of >1,000 CFU of coagulase-negative staphylococci and Enterococcus faecium per ml were found in culture of the central venous catheters in two different patients while the peripheral cultures yielded 100 and 1 CFU/ml, respectively. In another report, quantitative blood cultures were obtained from paired peripheral veins and central catheters in 63 infants with umbilical artery indwelling catheters or Broviac central venous catheters in whom sepsis was suspected (72). Infants in whom both central and peripheral cultures were positive were considered to have proven sepsis, whereas infants with positive central catheter cultures but negative peripheral cultures were diagnosed as having colonized catheters without sepsis. In 10 of 21 children with confirmed sepsis, colony counts in cultures drawn from central lines were >50% higher than the corresponding peripheral counts, suggesting CRS. Paired colony counts were approximately equal in one case, suggesting that sepsis was not related to the catheter. Confluent growth in the remaining 10 pairs prohibited precise colony counts and any conclusions about the source of bacteremia. In 1987, Mosca et al. (64) reported their experience with the use of quantitative blood cultures in 26 intravascular devices suspected of being the source of sepsis. A fivefold or greater central-to-peripheral bacterial count ratio was considered confirmatory of CRS. In the eight instances in which the diagnostic criterion was fulfilled, removal of the catheter resulted in prompt defervescence of the septic patients. In half of the remaining 18 cases in which the central venous cultures were sterile or the differences in colony counts were small, the catheter was removed but the patients' conditions did not improve. The remaining nine catheters were left in place and another source of sepsis was ultimately found. In the studies of CRS cited in the preceding paragraphs, the central venous catheter/peripheral bacterial counts ratio used to identify CRS ranged from 1.5- to several hundredfold. Establishing an appropriate ratio is complicated by the fact that central venous blood passes through the lungs before it reaches the peripheral veins. Since bacteria are removed from the blood by reticuloendothelial cells present in the lung, the final concentration of microorganisms in peripheral blood is less than the concentration in catheter blood even in those patients with sepsis from other sources. To address this problem, Flynn et al. (29) used simultaneous quantitative blood cultures of samples obtained from the superior vena cava and the femoral vein of rabbits with non-catheter-associated bacteremia. As expected, significantly larger bacterial concentrations were found in blood from the superior vena cava compared with concentrations in peripheral blood. Based on data collected in this animal model, the authors calculated that 99% of non-catheterrelated bacteremia should show differences of <fivefold in bacterial concentrations of central and peripheral blood and proposed this criterion for diagnosing CRS. In two clinical studies they found that, in patients with CRS, the observed QUANTITATIVE ASPECTS OF SEPTICEMIA 275 central-to-peripheral bacterial count ratio far exceeded the proposed value and ranged from 57 to >100 fold (28, 29). In spite of the potential advantages of quantitative blood cultures as a means of identifying CRS and evaluating the efficacy of antibiotic therapy without removing the catheter, few studies comparing quantitative blood cultures and the semiquantitative method of Maki et al. (60) have been published. In 1983, Moyer et al. (66) compared results of peripheral nonquantitative blood cultures, quantitative cultures obtained through the catheter, and semiquantitative cultures of the catheter tips. Among nonbacteremic patients, all central line blood cultures were negative or grew <24 CFU/ml, but 15 of 68 catheter tips grew >15 colonies. Among the five bacteremic patients, four catheter blood cultures grew >25 CFU/ml and >15 colonies were recovered from all catheter tips. In a prospective study of total parenteral nutrition-related infections, a pour plate method was used to culture blood from the central line and the semiquantitative technique of Maki et al. (59) was used to culture intravascular line segments when it was removed (78). Blood from the catheter was collected twice a week for pour plate cultures. Twelve of 100 courses of total parenteral nutrition resulted in infections: five patients had sepsis and seven had local infections. In 5 of 12 patients, pour plate cultures were positive, with 8 CFU of Candida parapsilosis per ml and 2 25 CFU/ml when bacteria were recovered. In 9 of 12, cultures of the intravascular segments were positive and grew >50 CFU/plate. All of the septic patients had positive line segment cultures (50 to >100 CFU/plate), but only two of the five had positive pour plate cultures of central line blood. No quantitative peripheral blood cultures were done. Despite the infrequent positive pour plate cultures results, the authors concluded that these cultures were helpful in implicating the total parenteral nutrition line in three instances because the pour plate culture results antedated positive peripheral blood culture results or removal of the line. The correlation between the results of quantitative blood culture technique and the semiquantitative catheter tip culture method for the diagnosis of CRS has been more recently examined by Fan et al. (26) and Paya et al. (67). In the study by Fan et al., semiquantitative tip cultures and quantitative blood cultures performed by the pour plate method were compared with 24 catheters removed from patients with suspected CRS. CRS was defined as a central-to-peripheral bacterial count ratio of >7 or growth of >15 colonies from the tip of the catheter (26). Seven of the nine catheters positive by the semiquantitative method were also positive by the quantitative blood culture criterion. For none of the patients with sterile catheters was the central-to-peripheral bacterial count ratio significant. In the study by Paya et al. (67) that involved 52 intravascular cannulae from patients with sepsis, CRS was defined as the isolation of >15 CFU from the removed cannulae. The quantitative blood culture method detected only 7 of the 15 cases in which >15 CFU were isolated from the cannula. On the other hand, a high central-to-peripheral bacterial count ratio was found in four cases that did not fulfill the criterion for the diagnosis of CRS. The authors suggested the following explanations for the conflicting results: (i) colonization of intravascular devices may occur intraluminally with no involvement of the outer surface or the tip (77); (ii) blood may be contaminated during phlebotomy or, alternatively, the catheter tip may be contaminated during its removal through an infected exit site; and (iii) lysis-centrifugation

8 276 YAGUPSKY AND NOLTE may be sensitive enough to detect low-grade bacteremias that are associated with cannula tip counts of <15 CFU. In the past it was assumed that CRS could not be cured as long as the catheter remained in place (59, 73). However, in eight reported series of Broviac-type catheter infections reviewed by Decker and Edwards (15), eradication of the organism causing sepsis was accomplished without removal of the central line in 75% of the infections and, more recently, in 61% of episodes reported by Benezra et al. (4). Most of the current methods require that the catheter be removed to confidently characterize a septic episode as catheter related. However, identifying the catheter as the source of the bacteremia is immaterial if clinical management and prognosis are the same regardless of the source. Comparison of paired quantitative central catheter and peripheral venous blood cultures shows promise as a means of diagnosing CRS, and it is the only available method that does not require removal of the catheter. However, for this approach to gain universal acceptance, additional studies comparing this method with other techniques are needed. In addition, larger numbers of actual catheter infections need to be studied to determine the range of central-to-peripheral bacterial count ratios that occur so that the sensitivity, specificity, and predictive values of this method for the diagnosis of CRS can be established. Processing blood cultures by quantitative methods provides additional information, but that information has not yet been demonstrated to be necessary to management or to correlate with outcome of CRS (15). Contaminants versus True Pathogens in Positive Blood Cultures Isolation of "classic" pathogens such as Streptococcus pneumoniae or Salmonella typhi from blood culture is always considered evidence of true bacteremia. However, when microorganisms of low virulence, especially skin flora, are recovered from blood cultures, they are traditionally dismissed as contaminants (57). Advances in medical therapeutics have resulted in an increasing population of immunocompromised patients due to antineoplastic chemotherapy, organ transplantation, and survival of extremely premature neonates. In addition, the use of intravascular catheters and prosthetic devices has become part of standard medical practice (17). As a result of these changes, microorganisms that were once considered nonpathogenic have emerged increasingly as causes of disease. At present, coagulase-negative staphylococci are leading causes of bacteremia in neonatal intensive care units and in patients with infected prosthetic heart valves, orthopedic prostheses, and ventriculoperitoneal shunts (17). A definition of contamination based on clinical criteria, the number of positive cultures, and the identity of the microorganism can often be used in immunocompetent patients, but in the immunocompromised host, bacteremia due to an opportunistic pathogen cannot be easily excluded. Finding reliable criteria to distinguish true positive and contaminated blood cultures would be of great value. False-positive blood cultures can result in wasted laboratory effort, the overuse of antibiotics, untoward drug effects, and prolonged hospitalization. Quantitative information provided by the lysis-centrifugation method has been proposed as an objective means to distinguish true bacteremia from contamination. For coagulase-negative staphylococci, Propionibacterium acnes, and aerobic diphtheroids, Dorn et al. (21) found that colony counts of <1 CFU/ml of blood were invariably associated CLIN. MICROBIOL. REV. with contamination, as determined after evaluation of the patients' medical records. However, 16% of contaminated blood cultures contained >1 CFU/ml. In a study of neonates and young infants with umbilical and Broviac catheters, isolation of skin flora from blood cultures or recovery of microorganisms after day 3 of incubation were considered evidence of contamination (72). Colony counts of <10 CFU/ ml were observed in 6 of 7 contaminated cultures but in only 1 of 21 blood cultures from septic patients. McLaughlin et al. (63) found 1 CFU/ml of blood in 98 of 138 (71%) blood cultures contaminated with coagulase-negative staphylococci, alpha-hemolytic streptococci, or diphtheroids, while the majority of true pathogens were isolated at a level of >1 CFU/ml. However, not all cultures from which <1 CFU/ml were isolated represented contamination. Similar results were obtained by Hill et al. (35), Welch et al. (98), Kelly et al. (40), and Campos and Spainhour (10). In summary, a clear trend exists toward low colony counts in contaminated blood cultures compared with true positive cultures. However, no quantitative criterion can invariably distinguish contamination from bacteremia. Available data suggest that quantitative blood cultures are useful in interpreting positive blood culture results when considered along with clinical criteria and the identity of the microorganism. CONCLUSIONS For many years quantitative blood cultures found only limited use in the diagnosis and management of septicemia because the available methods were cumbersome, labor intensive, and practical only for small blood volumes. Despite limitations, early studies of the quantitative aspects of bacteremia added to our understanding of sepsis and its complications. The development and commercial availability of lysis-centrifugation and lysis-direct plating methods have provided clinical microbiologists convenient and sensitive methods for quantitative blood cultures and have renewed interest in the quantitation of microorganisms in patient blood. Recent attention has been focused on the diagnostic and prognostic significance of determining the magnitude of bacteremia in various clinical circumstances and patient populations. Although no study has shown unequivocally that quantitative blood culture results are necessary for accurate diagnosis or appropriate management of septic patients, there are certain circumstances in which bacterial counts can be helpful. Two of the most promising areas for additional research are for diagnosis of central venous catheter-associated bacteremia and assessment of the efficacy of antibiotic therapy. There are several fundamental issues that need to be addressed before small differences in the concentration of bacteria in blood can be interpreted meaningfully. The reproducibility of the colony counts obtained with the quantitative blood culture methods and the normal variation in the level of bacteremia that occurs over time need to be defined. In addition, these observations need to be extended to larger groups of septic patients before the utility of quantitative blood culture results can be adequately assessed. LITERATURE CITED 1. Albers, W. H., C. W. Tyler, and B. Boxerbaum Asymptomatic bacteremia in the newborn infant. J. Pediatr. 69: Beeson, P. B., E. S. Brannon, and J. V. Warren Observations on the sites of removal of bacteria from the blood in patients with bacterial endocarditis. J. Exp. Med. 81: Bell, L. M., G. Alpert, J. M. Campos, and S. A. Plotkin

9 VOL. 3, 1990 Routine quantitative blood cultures in children with Haemophilus influenzae or Streptococcus pneumoniae bacteremia. Pediatrics 76: Benezra, D., T. E. Kiehn, J. W. M. Gold, A. F. Brown, A. D. M. Turnbull, and D. Armstrong Prospective study of infections in indwelling central venous catheters using quantitative blood cultures. Am. J. Med. 85: Bille, J., L. Stockman, G. D. Roberts, C. D. Horstmeier, and D. M. Ustrup Evaluation of a lysis-centrifugation system for recovery of yeast and filamentous fungi from blood. J. Clin. Microbiol. 18: Brannon, P., and T. E. Kiehn Large-scale clinical comparison of lysis-centrifugation and radiometric systems for blood culture. J. Clin. Microbiol. 22: Brannon, P., and T. E. Kiehn Clinical comparison of lysis-centrifugation and radiometric resin systems for blood culture. J. Clin. Microbiol. 24: Broviac, J. W., J. J. Coyle, and B. H. Schribner A silicone rubber arterial catheter for prolonged parenteral alimentation. Surg. Gynecol. Obstet. 136: Brown, W., and J. Kelsh Improved method for cultivation of Brucella from blood. Proc. Soc. Exp. Biol. Med. 85: Campos, J. M., and J. R. Spainhour Comparisons of the Isolator 1.5 Microbial Tube with a conventional blood culture broth system for detection of bacteremia in children. Diagn. Microbiol. Infect. Dis. 3: Campos, J. M., and J. R. Spainhour Rapid detection of bacteremia in children with a modified lysis direct plating method. J. Clin. Microbiol. 22: Carey, R. B Clinical comparison of the Isolator 1.5 Microbial Tube and the BACTEC radiometric system for detection of bacteremia in children. J. Clin. Microbiol. 19: Cashman, J. S., R. Boshard, and J. M. Matsen Viability of organisms held in the Isolator blood culture system for 15 hours and then rapid detection by acridine orange staining. J. Clin. Microbiol. 18: Cleri, D. J., M. L. Corrado, and S. J. Seligman Quantitative culture of intravenous catheters and other intravenous inserts. J. Infect. Dis. 141: Decker, M. D., and K. M. Edwards Central venous catheter infections. Pediatr. Clin. North Am. 35: Dietzman, D. E., G. W. Fischer, and F. D. Schoenknecht Neonatal Escherichia coli septicemia-bacterial counts in blood. J. Pediatr. 85: Donowitz, L. G., C. E. Haley, W. W. Gregory, and R. P. Wenzel Neonatal intensive care unit bacteremia: emergence of gram-positive bacteria as major pathogens. Am. J. Infect. Control 15: Dorn, G. L., G. G. Burson, and J. R. Haynes Blood culture technique based on centrifugation: clinical evaluation. J. Clin. Microbiol. 3: Dorn, G. L., J. R. Haynes, and G. G. Burson Blood culture technique based on centrifugation: developmental phase. J. Clin. Microbiol. 3: Dorn, G. L., G. A. Land, and G. E. Wilson Improved blood culture technique based on centrifugation: clinical evaluation. J. Clin. Microbiol. 9: Dorn, G. L., Y. Lemeshev, and J. M. Hill Quantitative blood cultures-their value with laboratory contaminants. J. Clin. Hematol. Oncol. 11: Dorn, G. L., and K. Smith New centrifugation blood culture device. J. Clin. Microbiol. 7: Drutz, D. J., T. S. N. Chen, and W. H. Lu The continuous bacteremia of lepromatous leprosy. N. Engl. J. Med. 287: DuPont, H. L., and W. W. Spink Infections due to gram-negative organisms: an analysis of 860 patients with bacteremia at the University of Minnesota Medical Center, Medicine (Baltimore) 48: QUANTITATIVE ASPECTS OF SEPTICEMIA Durbin, W. A., E. G. Szymczak, and D. A. Goldmann Quantitative blood cultures in childhood bacteremia. J. Pediatr. 92: Fan, S. T., C. H. Teoh-Chan, and K. F. Lau Evaluation of central venous catheter sepsis by differential quantitative blood culture. Eur. J. Clin. Microbiol. Infect. Dis. 8: Finegold, S. M., M. I. White, I. Ziment, and W. R. Winn Rapid diagnosis of bacteremia. J. Clin. Microbiol. 18: Flynn, P. M., J. L. Shenep, and F. F. Barrett Differential quantitation with a commercial blood culture tube for diagnosis of catheter-related infection. J. Clin. Microbiol. 26: Flynn, P. M., J. L. Shenep, D. C. Stokes, and F. F. Barrett In-situ management of confirmed central venous catheter-related bacteremia. Pediatr. Infect. Dis. 6: Fojtasek, M. F., and M. T. Kelly Isolation of Mycobacterium chelonei with the lysis centrifugation blood culture technique. J. Clin. Microbiol. 16: Gill, V. J., C. H. Park, F. Stock, L. L. Gosey, F. G. Witebsky, and H. Masur Use of lysis-centrifugation (Isolator) and radiometric (BACTEC) blood culture systems for the detection of mycobacteremia. J. Clin. Microbiol. 22: Hall, W. H., and D. Gold Shock associated with bacteremia. Arch. Intern. Med. 96: Henry, N. K., C. M. Grewell, P. E. Van Grevenhof, D. M. Ilstrup, and J. A. Washington II Comparison of lysiscentrifugation with a biphasic blood culture medium for medium for the recovery of aerobic and facultatively anaerobic bacteria. J. Clin. Microbiol. 20: Henry, N. K., C. A. McLimans, A. J. Wright, R. L. Thompson, W. R. Wilson, and J. A. Washington II Microbiological and clinical evaluation of the Isolator lysis-centrifugation blood culture tube. J. Clin. Microbiol. 17: Hill, J. M., Y. Lemeshev, and A. S. Pardue Septicemia: early pure isolation and quantitation. A new blood culture method. J. Clin. Hematol. Oncol. 11: Ilstrup, D. M., and J. A. Washington II The importance of volume of blood cultured in the detection of bacteremia and fungemia. Diagn. Microbiol. Infect. Dis. 1: Jaffe, D. M., R. R. Tanz, A. T. Davis, F. Henretig, and G. L. Fleisher Antibiotic administration to treat possible occult bacteremia in febrile children. N. Engl. J. Med. 317: Kellogg, J. A., J. P. Manzella, and J. H. McConville Clinical laboratory comparison of the 10-ml Isolator blood culture system with BACTEC radiometric blood culture media. J. Clin. Microbiol. 20: Kelly, M. T., G. E. Buck, and M. F. Fojtasek Evaluation of a lysis-centrifugation and biphasic bottle blood culture system during routine use. J. Clin. Microbiol. 18: Kelly, M. T., M. F. Fojtasek, T. M. Abbott, D. C. Hale, J. R. Dizikes, R. Boshard, G. E. Buck, W. J. Martin, and J. M. Matsen Clinical evaluation of a lysis-centrifugation technique for the detection of septicemia. J. Am. Med. Assoc. 250: Kiani, D., E. L. Quinn, K. H. Baruch, T. Madhavan, L. D. Saravolatz, and T. R. Neblett The increasing importance of polymicrobial bacteremia. J. Am. Med. Assoc. 242: Kiehn, T. E Quantitative blood cultures. A review of 52 years, p In A. E. Brown and D. Armstrong (ed.), Infectious complications of neoplastic diseases: controversies in management. Yorke Medical Books, New York. 43. Kiehn, T. E., and R. Cammarata Comparative recoveries of Mycobacterium avium-m. intracellulare from Isolator lysis centrifugation and BACTEC 13A blood culture systems. J. Clin. Microbiol. 26: Kiehn, T. E., F. F. Edwards, P. Brannon, A. Y. Tsang, M. Maio, J. W. M. Gold, E. Whimbey, B. Wong, J. K. McClatchy, and D. Armstrong Infections caused by Mycobacterium avium complex in immunocompromised patients: diagnosis by blood culture and fecal examination, antimicrobial susceptibility tests, and morphological and seroagglutination characteristics. J. Clin. Microbiol. 21: Kiehn, T. E., J. W. M. Gold, P. Brannon, R. J. Timberger, and D. Armstrong Mycobacterium tuberculosis bacteremia

10 278 YAGUPSKY AND NOLTE detected by the Isolator lysis-centrifugation blood culture system. J. Clin. Microbiol. 21: Kiehn, T. E., B. Wong, F. F. Edwards, and D. Armstrong Comparative recovery of bacteria and yeasts from lysiscentrifugation and a conventional blood culture system. J. Clin. Microbiol. 18: Kluge, R. M., and H. L. DuPont Factors affecting mortality of patients with bacteremia. Surg. Gynecol. Obstet. 137: Komorowski, R. A., and S. G. Farmer Rapid detection of candidemia. Am. J. Clin. Pathol. 59: Kreger, B. E., D. E. Craven, P. C. Carling, and W. R. McCabe Gram-negative bacteremia. III. Reassessment of etiology, epidemiology and ecology in 612 patients. Am. J. Med. 68: Lamberg, R. E., R. F. Schell, J. L. Schell, and J. L. Le Frock Detection and quantitation of simulated anaerobic bacteremia by centrifugation and filtration. J. Clin. Microbiol. 17: La Scolea, L. J., Jr., and D. Dryja Quantitation of bacteria in cerebrospinal fluid and blood of children with meningitis and its diagnostic significance. J. Clin. Microbiol. 19: La Scolea, L. J., Jr., D. Dryja, and E. Neter Comparison of the quantitative direct plating method and the BACTEC procedure for rapid diagnosis of Haemophilus influenzae bacteremia in children. J. Clin. Microbiol. 14: La Scolea, L. J., Jr., D. Dryja, T. D. Sullivan, L. Mosovich, N. Ellerstein, and E. Neter Diagnosis of bacteremia in children by quantitative direct plating and a radiometric procedure. J. Clin. Microbiol. 13: La Scolea, L. J., Jr., S. V. Rosales, and P. L. Ogra Haemophilus influenzae type be infection in childhood: history of bacteremia and antigenemia. Infect. Immun. 50: La Scolea, L. J., Jr., S. V. Rosales, R. C. Welliver, and P. L. Ogra Mechanisms underlying the development of meningitis or epiglottitis in children following Haemophilus influenzae type b bacteremia. J. Infect. Dis. 151: Libman, E A consideration of the prognosis in subacute bacterial endocarditis. Am. Heart J. 1: MacGregor, R. R., and H. N. Beaty Evaluation of positive blood cultures. Arch. Intern. Med. 130: Macher, A. M., J. A. Kovacs, V. Gill, G. D. Roberts, J. Ames, C. H. Park, S. Straus, H. C. Lane, J. E. Parrilo, A. S. Fauci, and H. Masur Bacteremia due to Mycobacterium aviumintracellulare in the acquired immunodeficiency syndrome. Ann. Intern. Med. 99: Maki, D. G., D. A. Goldmann, and F. S. Rhame Infection control in intravenous therapy. Ann. Intern. Med. 79: Maki, D. G., C. E. Weise, and H. W. Sarafin A semiquantative culture method for identifying intravenous catheter-related infection. N. Engl. J. Med. 296: Manzella, J. P., J. A. Kellogg, and J. K. Clark Quantitative colony counts in patients with Staphylococcus aureus bacteremia. J. Infect. Dis. 155: Marshall, G. S., and L. M. Bell Correlates of high grade and low grade Haemophilus influenzae bacteremia. Pediatr. Infect. Dis. J. 7: McLaughlin, J. C., P. Hamilton, J. V. Scholes, and R. C. Bartlett Clinical laboratory comparison of lysis-centrifugation and BACTEC radiometric blood culture techniques. J. Clin. Microbiol. 18: Mosca, R., S. Curtas, B. Forbes, and M. M. Meguid The benefits of Isolator cultures in the management of suspected catheter sepsis. Surgery 102: Moxon, E. R., and P. T. Ostrow Haemophilus influenzae meningitis in infants rats: role of bacteremia in pathogenesis of age-dependent inflammatory responses in cerebrospinal fluid. J. Infect. Dis. 735: Moyer, M. A., L. D. Edwards, and L. Farley Comparative culture methods on 101 intravenous catheters; routine, semiquantitative, and blood cultures. Arch. Intern. Med. 143: CLIN. MICROBIOL. REV Paya, C. V., L. Guerra, H. M. Marsh, M. B. Farnell, J. A. Washington II, and R. L. Thompson Limited usefulness of quantitative culture of blood drawn through the device for diagnosis of intravascular-device-related bacteremia. J. Clin. Microbiol. 27: Pazin, G. J., K. L. Peterson, F. W. Griff, J. A. Shaver, and M. Ho Determination of site of infection in endocarditis. Ann. Intern. Med. 82: Randriambololona, R., and A. Dodin Etude de l'evolution de la bacteriemie dans onze cas de salmonellose traites. Ann. Inst. Pasteur (Paris) 99: Raucher, H. S., A. C. Hyatt, A. Barzilai, M. B. Harris, M. A. Weiner, N. S. Le Leiko, and D. S. Hodes Quantitative blood cultures in the evaluation of septicemia in children with Broviac catheters. J. Pediatr. 104: Reyes, M. P., W. A. Palutke, R. F. Wylin, and A. M. Lerner Pseudomonas endocarditis in the Detroit Medical Center, Medicine (Baltimore) 52: Ruderman, J. W., M. A. Morgan, and A. H. Klein Quantitative blood cultures in the diagnosis of sepsis in infants with umbilical and Broviac catheters. J. Pediatr. 112: Ryan, J. A., Jr., R. M. Abel, N. M. Abbott, C. C. Hopkins, T. M. Hesney, R. Colley, K. Phillips, and J. E. Fischer Catheter complications in total parenteral nutrition: a prospective study of 200 consecutive patients. N. Engl. J. Med. 290: Salazar-Mallen, M., E. L. Hube, and M. Brenes Comparative study of blood cultures made from artery, vein, and bone marrow in patients with subacute bacterial endocarditis. Am. Heart J. 33: Santosham, M., and E. R. Moxon Detection and quantitation of bacteremia in childhood. J. Pediatr. 91: Shenep, J. L., P. M. Flynn, F. F. Barrett, G. L. Stidham, and D. F. Westenkirchner Serial quantitation of endotoxemia and bacteremia during therapy for gram-negative bacterial sepsis. J. Infect. Dis. 157: Sitges-Serra, A., J. Linares, and J. Garau Catheter sepsis: the clue is the hub. Surgery 97: Snydman, D. R., S. A. Murray, S. J. Kornfeld, J. A. Majka, and C. A. Ellis Total parenteral nutrition-related infections: prospective epidemiologic study using semiquantitative methods. Am. J. Med. 73: Stanaszek, P. M Rapid bacteremia diagnosis using field monitor membrane filtration. Am. J. Med. Technol. 37: Stockman, L., G. D. Roberts, and D. M. 1lstrup Effect of storage of the DuPont lysis-centrifugation system on recovery of bacteria and fungi in a prospective clinical trial. J. Clin. Microbiol. 19: Strand, C. L., and J. A. Shulman Bloodstream infections. Laboratory, detection and clinical considerations, p American Society of Clinical Pathologists, Chicago. 82. Sullivan, N. M., V. L. Sutter, W. T. Carter, H. R. Attebery, and S. M. Finegold Bacteremia after genitourinary tract manipulation: bacteriological aspects and evaluation of various blood culture systems. Apple. Microbiol. 23: Sullivan, N. M., V. L. Sutter, and S. M. Finegold Practical aerobic membrane filtration blood culture technique: development of procedure. J. Clin. Microbiol. 1: Sullivan, N. M., V. L. Sutter, and S. M. Finegold Practical aerobic membrane filtration blood culture technique: clinical blood culture trial. J. Clin. Microbiol. 1: Sullivan, T. D., and L. J. LaScolea, Jr Neisseria meningitidis bacteremia in children: quantitation of bacteremia and spontaneous clinical recovery without antibiotic therapy. Pediatrics 80: Sullivan, T. D., L. J. LaScolea, Jr., and E. Neter Relationship between the magnitude of bacteremia in children and the clinical disease. Pediatrics 69: Thomson, R. B., Jr., S. J. Vanzo, N. K. Henry, K. L. Guenther, and J. A. Washington II Contamination of cultures processed with the Isolator lysis-centrifugation blood culture tube. J. Clin. Microbiol. 19:97-99.

11 VOL. 3, Tidwell, W., and L. L. Gee Use of membrane filters in blood cultures. Proc. Soc. Exp. Biol. Med. 88: Tilton, R. C The laboratory approach to the detection of bacteremia. Annu. Rev. Microbiol. 36: Warren, M., and W. W. Herrick Analysis of one hundred and thirty-four cases of bacteremia. Am. J. Med. Sci. 151: Washington, J. A., II Blood culture principles and techniques. Mayo Cln. Proc. 50: Washington, J. A., II Blood cultures: principles and new approaches. J. Med. Technol. 1: Washington, J. A., II Conventional approaches to blood culture, p In J. A. Washington II (ed.), The detection of bacteremia. CRC Press, West Palm Beach, Fla. 94. Washington, J. A., II, and D. M. Ilstrup Blood cultures: issues and controversies. Rev. Infect. Dis. 8: Weil, M. H., and W. W. Spink The shock syndrome associated with bacteremia due to gram-negative bacilli. Arch. Intern. Med. 101: Weinstein, M. P., L. B. Relier, J. R. Murphy, and K. A. Lichtenstein The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults. I. Laboratory and epidemiologic observations. Rev. Infect. Dis. 5: Weiss, H., and R. Ottenberg Relation between bacteria and temperature in subacute bacterial endocarditis. J. Infect. Dis. 50: Welch, D. F., R. K. Scribner, and D. Hensel Evaluation of a lysis direct plating method for pediatric blood cultures. J. Clin. Microbiol. 21: Werner, A. S., C. G. Cobbs, D. Kaye, and E. W. Hook QUANTITATIVE ASPECTS OF SEPTICEMIA 279 Studies on the bacteremia of bacterial endocarditis. J. Am. Med. Assoc. 202: Wheat, L. J., and M. Bartlett Histoplasma capsulatum fungemia documented using the DuPont Isolator system. Diagn. Microbiol. Infect. Dis. 2: Whimbey, E., T. E. Kiehn, P. Brannon, D. Benezra, and D. Armstrong Clinical significance of colony counts in immunocompromised patients with Staphylococcus aureus bacteremia. J. Infect. Dis. 155: Whimbey, E., B. Wong, T. E. Kiehn, and D. Armstrong Clinical correlations of serial quantitative blood cultures determined by lysis-centrifugation in patients with persistent septicemia. J. Clin. Microbiol. 19: Wing, W. J., C. W. Norden, R. K. Shadduck, and A. Winkelstein Use of quantitative bacteriologic techniques to diagnose catheter-related sepsis. Arch. Intern. Med. 139: Winn, W. R., M. L. White, W. T. Carter, A. B. Miller, and S. M. Finegold Rapid diagnosis of bacteremia with quantitative differential membrane filtration culture. J. Am. Med. Assoc. 197: Wong, B., F. F. Edwards, T. E. Kiehn, E. Whimbey, H. Donnelly, E. M. Bernard, J. W. M. Gold, and D. Armstrong Continuous high Mycobacterium avium intracellulare bacteremia in patients with the acquired immune deficiency syndrome. Am. J. Med. 78: Zierdt, C. H., D. L. Peterson, J. C. Swan, and J. D. MacLowry Lysis-filtration blood culture versus conventional blood culture in a bacteremia rabbit model. J. Clin. Microbiol. 15: Downloaded from on March 11, 2019 by guest