DIAGN MICROBIOLINFECTDIS 1991;14:111-118
111
Multicenter Comparison of the High Volume (10 ml) NR BACTECPLUS System and the Standard (5 ml) NR BACTEC System Franklin P. Koontz, Kristine K. Flint, Janet K. Reynolds, and Stephen D. Allen
This multicenter study was designed to compare the new BACTE¢ PLUS system (nonradiometric), which utilizes an 8- to 10ml blood inoculum in a resin-containing medium, to the standard BACTEC (nonradiometric) without resins and 5-ml blood inoculum. There were 12,341 compliant sets studied, yielding 1331 positives, with 1099 sets deemed clinically significant. Overall the BACTECPLUS showed an enhanced recovery of 33% (p < 0.001) over its standard counterpart, with significant yield increased in the staphylococci (p < 0.001), streptococci (p < 0.002), pseudomonads (p < 0.002), Enterobacteri-
aceae (p < 0.001), and other aerobic Gram negatives (p < 0.02). The enhanced performance increased to 53% if the patient was receiving any antibiotics at the time the blood was cultured. In patients known to be free of antibiotics at the time of blood draw, there was still an increased yield of 18%. The new system detected positivity at least one reading sooner than twice as often as the converse, and confirmed septic episodes significantly more often (21% overall) (41% on antibiotics) (15% no antibiotics). The BACTECPLUS has distinct advantages over its low blood volume, nonresin counterpart.
The blood culture literature has indicated that the prime variables in the successful detection of bacteremia and fungemia are the volume of blood inoculum per culture set (Tenney et al., 1982; Ilstrup and Washington, 1983; Plorde et al., 1985; Washington and Ilstrup, 1986) and the number of blood culture sets submitted per septic episode (Weinstein et al., 1983; Washington and Ilstrup, 1986). Although the BACTEC systems (radiometric and nonradiometric) have been shown to be valuable tools in the detection of septicemia (Plorde et al., 1985; Jungkind et al., 1986; Jungkind et al., 1989), the major controversies, if any, have been the low volume of blood cultured per bottle (5 ml), and the value of incorporating resins into the blood culture system
(Reynolds et al., 1984; Washington and Ilstrup, 1986; Allen et al., 1989). In the study reported herein, the standard nonradiometric BACrEC system, without added resins, was compared to a new nonradiometric BACTECsystem utilizing a 10-ml blood inoculum into a medium containing resins. The objectives of the 12-month study were to determine if the new system would (a) increase the yield of clinically significant isolates, 0v) diagnose septic episodes more rapidly, and (c) clarify the role of the absorbent resins in conjunction with an increased blood inoculum. For this study the non-resin, low-volume BACTECsystem was considered the standard.
From the Departments of Pathology,Universityof Iowa Hospital (F.P.K., K.K.F),Iowa City, Iowa; and Indiana University Hospitals 0.K.R., S.D.A.), Indianapolis, Indiana, USA. Address reprint requests to Dr. F.P. Koontz, Department of Pathology, Universityof Iowa Hospital, Iowa City, IA 52242, USA. Received 30 May 1990; revised and accepted 26 July 1990. © 1991 Elsevier Science PublishingCo., Inc. 655 Avenue of the Americas, New York, NY 10010 0732-8893/91/$3.50
MATERIALS A N D M E T H O D S During the period of the study, house officers, nurses, and phlebotomists were instructed to collect 24-30 ml of blood from a single draw. Aliquots were injected at bedside, randomly, into the four bottle set as follows: 10 ml into each of the BACTECPLUS (BP) pair, 26A (aerobic) and 27A (anaerobic); and 5 ml into each of the BACTEC STANDARD (BS) pair, 6A
112
(aerobic) and 7A (anaerobic). If < 30 ml was obtained from the single blood draw, the random distribution was adjusted to 9:9:4:4 or 8:8:4:4. Submitters were instructed to adjust the BP pair downward, but never to put < 4 ml into each of the BS pair. Specimens with less than the 8:8:4:4 ratio as measured visually in the laboratory were processed, but excluded from the study, as were the incomplete sets (less than the four designated bottles). All bottles were tested on the BACTEC NR660 nonradiometric analyzer. Aerobic bottles (6A and 26A) were shaken during day-1 incubation and read twice daily for days 1 and 2, and daily thereafter for the 7-day testing cycle. Anaerobic bottles (7A and 27A) were not shaken and were read once daily for the 7-day testing cycle. Cultures from suspected fungemias, mycobacteremias, or bacteremias due to fastidious organisms, for example, Brucella spp., were incubated for 3 weeks, but were only included in the study for the first 7 days. All negative bottles were visibly inspected before being discarded. The initial thresholds for the BP aerobic and anaerobic bottles were at 40 and 30 growth value (GV) units (U), respectively. The BS pair GV thresholds were 30 U (6A aerobic) and 25 U (7A anaerobic). Thresholds were adjusted as the study progressed. The date, time, and GV of the positive bottles were recorded, as well as the final identification of the isolate. Patient demographics were automatically recorded via laboratory computer. Pathology residents (Iowa) or laboratory personnel (Indiana) determined the clinical significance of the isolates utilizing a combination of the methods of Tenney et al. (1982) and Weinstein et al. (1983). Nonsignificance was indicated utilizing the approach of Jungkind et al. (1986). Paired comparisons of the BS and the BP were done only on the complete four-bottle sets adequately filled with 4-5 ml and 8-10 ml of blood respectively, and yielding clinically significant isolates in any or all of the four-bottle set. Significance testing was accomplished by the chi-square method of McNemar (1962).
RESULTS
Of the 12,341 compliant sets processed, 1331 (11%) sets were positive, of which 1099 (9%) were clinically significant, 85 (0.7%) were definitively nonsignificant, and 147 (1.2%) where significance status could not be determined by the methods utilized (Weinstein et al., 1983; Tenney et al., 1982). The total positive culture sets yielded 1470 isolates, with 1212 (82%) significant, 99 (7%) definitively nonsignificant, and 159 (11%) isolates of undetermined significance, These isolates, from 725 patients, represent 875 episodes including 642 clinically significant, 145 undeterminable significance, and 88 definitive nonsignificance
F.P. Koontz et al.
(contaminant, transitory bacteremia, and so on). Admittedly, the undeterminable group consists of some isolates that were probably clinically significant, others that were transitory bacteremias and the remainder that were contaminants. The reason for the haziness of the significance data is simply that the major blood culture isolate in nosocomial sepsis, and the number one contaminant (nonsignificant) in blood cultures are the same organism, Staphylococcus epidermidis. Only the confirmed clinically significant organisms will be discussed in detail in this report. There were 1212 bacterial and fungal isolates associated with septic episodes, 1097 (91%) detected by the BP system, and 831 (68%) by the BS. These data represent a 33% increased yield by the BP over the BS. Both systems detected 717 significant isolates with 530 (74%) of these detected at the same time. Overall, the BP had 141 (20%) detections at least one reading (6--10 hr) sooner, whereas the BS pair was faster in 77 (11%) instances. Table 1 presents the list of isolates by the comparative systems, BS versus BP, detailed by specie or group. Overall, the isolate yield from all clinically significant cultures statistically favored the new BP system (p < 0.001), although some groups of organisms did equally well in both systems, for example, viridans streptococci, Gram-positive rods, and the less common coagulase-negative staphylococci (CNS). There were also instances of increased yield that were not sufficient to be significant, as seen with the Klebsiella and the yeasts. The most significant increase in detection was observed with the staphylococci (p < 0.001) and was almost equally divided between the two major species when isolates were identified to species level utilizing the Vitek Automicrobic System (Iowa). The other center involved in this study (Indiana) did not identify the CNS beyond that enzyme test level. With the streptococci (p 0.002), the significant difference was due totally to better yield of the enterococci. The viridans streptococci were recovered equally well by both systems. It should be noted that although there were 80 isolations of the viridans group, they did not reflect a high incidence of endocarditis but rather multiple culture sets positive per patient. The protocol of three sets of blood cultures per day for 3 days in cases of suspect bacterial endocarditis could lead to six or more isolations per patient. The significantly increased detection of the Enterobacteriaceae (p < 0.001) was due to several species, and Klebsiella fell just one or two isolations short of significance. The pseudomonad significance was due solely to the increased isolation of P. aeruginosa (p < 0.002) in the BP system. The potpourri of organisms in the "other Gram-negative" category was also significant as a group (p K 0.02), however, no one genus or specie was isolated in sufficient numbers to determine individual p values.
NR BACTEC5 ml vs 10 ml
TABLE 1
113
Comparison of Clinically Significant Isolates from the BACTECSTANDARDand BACTECPLUS Blood Culture Systems Total No. Isolates
Microorganisms
BS
Pair
BP
Pair
p
203 277
111
Multicenter Comparison of the High Volume (10 ml) NR BACTECPLUS System and the Standard (5 ml) NR BACTEC System Franklin P. Koontz, Kristine K. Flint, Janet K. Reynolds, and Stephen D. Allen
This multicenter study was designed to compare the new BACTE¢ PLUS system (nonradiometric), which utilizes an 8- to 10ml blood inoculum in a resin-containing medium, to the standard BACTEC (nonradiometric) without resins and 5-ml blood inoculum. There were 12,341 compliant sets studied, yielding 1331 positives, with 1099 sets deemed clinically significant. Overall the BACTECPLUS showed an enhanced recovery of 33% (p < 0.001) over its standard counterpart, with significant yield increased in the staphylococci (p < 0.001), streptococci (p < 0.002), pseudomonads (p < 0.002), Enterobacteri-
aceae (p < 0.001), and other aerobic Gram negatives (p < 0.02). The enhanced performance increased to 53% if the patient was receiving any antibiotics at the time the blood was cultured. In patients known to be free of antibiotics at the time of blood draw, there was still an increased yield of 18%. The new system detected positivity at least one reading sooner than twice as often as the converse, and confirmed septic episodes significantly more often (21% overall) (41% on antibiotics) (15% no antibiotics). The BACTECPLUS has distinct advantages over its low blood volume, nonresin counterpart.
The blood culture literature has indicated that the prime variables in the successful detection of bacteremia and fungemia are the volume of blood inoculum per culture set (Tenney et al., 1982; Ilstrup and Washington, 1983; Plorde et al., 1985; Washington and Ilstrup, 1986) and the number of blood culture sets submitted per septic episode (Weinstein et al., 1983; Washington and Ilstrup, 1986). Although the BACTEC systems (radiometric and nonradiometric) have been shown to be valuable tools in the detection of septicemia (Plorde et al., 1985; Jungkind et al., 1986; Jungkind et al., 1989), the major controversies, if any, have been the low volume of blood cultured per bottle (5 ml), and the value of incorporating resins into the blood culture system
(Reynolds et al., 1984; Washington and Ilstrup, 1986; Allen et al., 1989). In the study reported herein, the standard nonradiometric BACrEC system, without added resins, was compared to a new nonradiometric BACTECsystem utilizing a 10-ml blood inoculum into a medium containing resins. The objectives of the 12-month study were to determine if the new system would (a) increase the yield of clinically significant isolates, 0v) diagnose septic episodes more rapidly, and (c) clarify the role of the absorbent resins in conjunction with an increased blood inoculum. For this study the non-resin, low-volume BACTECsystem was considered the standard.
From the Departments of Pathology,Universityof Iowa Hospital (F.P.K., K.K.F),Iowa City, Iowa; and Indiana University Hospitals 0.K.R., S.D.A.), Indianapolis, Indiana, USA. Address reprint requests to Dr. F.P. Koontz, Department of Pathology, Universityof Iowa Hospital, Iowa City, IA 52242, USA. Received 30 May 1990; revised and accepted 26 July 1990. © 1991 Elsevier Science PublishingCo., Inc. 655 Avenue of the Americas, New York, NY 10010 0732-8893/91/$3.50
MATERIALS A N D M E T H O D S During the period of the study, house officers, nurses, and phlebotomists were instructed to collect 24-30 ml of blood from a single draw. Aliquots were injected at bedside, randomly, into the four bottle set as follows: 10 ml into each of the BACTECPLUS (BP) pair, 26A (aerobic) and 27A (anaerobic); and 5 ml into each of the BACTEC STANDARD (BS) pair, 6A
112
(aerobic) and 7A (anaerobic). If < 30 ml was obtained from the single blood draw, the random distribution was adjusted to 9:9:4:4 or 8:8:4:4. Submitters were instructed to adjust the BP pair downward, but never to put < 4 ml into each of the BS pair. Specimens with less than the 8:8:4:4 ratio as measured visually in the laboratory were processed, but excluded from the study, as were the incomplete sets (less than the four designated bottles). All bottles were tested on the BACTEC NR660 nonradiometric analyzer. Aerobic bottles (6A and 26A) were shaken during day-1 incubation and read twice daily for days 1 and 2, and daily thereafter for the 7-day testing cycle. Anaerobic bottles (7A and 27A) were not shaken and were read once daily for the 7-day testing cycle. Cultures from suspected fungemias, mycobacteremias, or bacteremias due to fastidious organisms, for example, Brucella spp., were incubated for 3 weeks, but were only included in the study for the first 7 days. All negative bottles were visibly inspected before being discarded. The initial thresholds for the BP aerobic and anaerobic bottles were at 40 and 30 growth value (GV) units (U), respectively. The BS pair GV thresholds were 30 U (6A aerobic) and 25 U (7A anaerobic). Thresholds were adjusted as the study progressed. The date, time, and GV of the positive bottles were recorded, as well as the final identification of the isolate. Patient demographics were automatically recorded via laboratory computer. Pathology residents (Iowa) or laboratory personnel (Indiana) determined the clinical significance of the isolates utilizing a combination of the methods of Tenney et al. (1982) and Weinstein et al. (1983). Nonsignificance was indicated utilizing the approach of Jungkind et al. (1986). Paired comparisons of the BS and the BP were done only on the complete four-bottle sets adequately filled with 4-5 ml and 8-10 ml of blood respectively, and yielding clinically significant isolates in any or all of the four-bottle set. Significance testing was accomplished by the chi-square method of McNemar (1962).
RESULTS
Of the 12,341 compliant sets processed, 1331 (11%) sets were positive, of which 1099 (9%) were clinically significant, 85 (0.7%) were definitively nonsignificant, and 147 (1.2%) where significance status could not be determined by the methods utilized (Weinstein et al., 1983; Tenney et al., 1982). The total positive culture sets yielded 1470 isolates, with 1212 (82%) significant, 99 (7%) definitively nonsignificant, and 159 (11%) isolates of undetermined significance, These isolates, from 725 patients, represent 875 episodes including 642 clinically significant, 145 undeterminable significance, and 88 definitive nonsignificance
F.P. Koontz et al.
(contaminant, transitory bacteremia, and so on). Admittedly, the undeterminable group consists of some isolates that were probably clinically significant, others that were transitory bacteremias and the remainder that were contaminants. The reason for the haziness of the significance data is simply that the major blood culture isolate in nosocomial sepsis, and the number one contaminant (nonsignificant) in blood cultures are the same organism, Staphylococcus epidermidis. Only the confirmed clinically significant organisms will be discussed in detail in this report. There were 1212 bacterial and fungal isolates associated with septic episodes, 1097 (91%) detected by the BP system, and 831 (68%) by the BS. These data represent a 33% increased yield by the BP over the BS. Both systems detected 717 significant isolates with 530 (74%) of these detected at the same time. Overall, the BP had 141 (20%) detections at least one reading (6--10 hr) sooner, whereas the BS pair was faster in 77 (11%) instances. Table 1 presents the list of isolates by the comparative systems, BS versus BP, detailed by specie or group. Overall, the isolate yield from all clinically significant cultures statistically favored the new BP system (p < 0.001), although some groups of organisms did equally well in both systems, for example, viridans streptococci, Gram-positive rods, and the less common coagulase-negative staphylococci (CNS). There were also instances of increased yield that were not sufficient to be significant, as seen with the Klebsiella and the yeasts. The most significant increase in detection was observed with the staphylococci (p < 0.001) and was almost equally divided between the two major species when isolates were identified to species level utilizing the Vitek Automicrobic System (Iowa). The other center involved in this study (Indiana) did not identify the CNS beyond that enzyme test level. With the streptococci (p 0.002), the significant difference was due totally to better yield of the enterococci. The viridans streptococci were recovered equally well by both systems. It should be noted that although there were 80 isolations of the viridans group, they did not reflect a high incidence of endocarditis but rather multiple culture sets positive per patient. The protocol of three sets of blood cultures per day for 3 days in cases of suspect bacterial endocarditis could lead to six or more isolations per patient. The significantly increased detection of the Enterobacteriaceae (p < 0.001) was due to several species, and Klebsiella fell just one or two isolations short of significance. The pseudomonad significance was due solely to the increased isolation of P. aeruginosa (p < 0.002) in the BP system. The potpourri of organisms in the "other Gram-negative" category was also significant as a group (p K 0.02), however, no one genus or specie was isolated in sufficient numbers to determine individual p values.
NR BACTEC5 ml vs 10 ml
TABLE 1
113
Comparison of Clinically Significant Isolates from the BACTECSTANDARDand BACTECPLUS Blood Culture Systems Total No. Isolates
Microorganisms
BS
Pair
BP
Pair
p
203 277
