- Email: [email protected]
Effect of number of blood cultures and volume of blood on detection of bacteremia in children
Effect of number of blood cultures and volume of blood on detection of bacteremia in children Daniel J. Isaacman, MD, Raymond B. Karasic, MD, Ellen A. Reynolds, RN, MS, and Susanne I. Kost, MDa From the Department of Pediatrics, University of PNsburgh School of Medicine; University of Pittsburgh School of Nursing; and Children's Hospital of PiHsburgh, PNsburgh, Pennsylvania
Objective: To determine whether bacteremia can be detected more rapidly and completely by (I) obtaining two blood cultures instead of one and/or (2) collecting a larger volume of blood. Study design: Prospective comparison of different strategies in 300 patients undergoing blood culture for suspected bacteremia. Each patient had two samples of blood, A (2 ml) and B (9.5 ml), obtained sequentially from separate sites. The B sample was divided into three aliquots: BI (2 ml), B2 (6 ml), and ISO (I .5 ml, quantitative culture). Results:A pathogen was isolated from one or more blood cultures in 30 patients (10% of cases). When measured at 24 hours, the pathogen recovery rate for the B2 sample (72%) was higher than that for the individual small-volume samples (A = 37%, BI = 33%; p <0.01 for each comparison) and for the combination of the two small-volume samples (A + BI = 47%; p = 0.04). At final (7-day) reading the pathogen recovery rate for the B2 sample (83%) was higher than that for BI (60%; p = 0.02) and similar to the recovery rate observed with the combination of the two small-volume cultures (A + BI = 73%; p = 0.55). Conclusions: Increasing the volume of blood inoculated into blood culture bottles improves the timely detection of bacteremia in pediatric patients and spares the patients the cost and pain of an additional venipuncture. (J PEDIATR1996; 128:190-5) The optimal number of blood cultures and volume of blood needed to detect bacteremia in children are not known. Nevertheless, clinicians often rely on a single culture containing Supported in part by a grant from the General Clinical Research Center of Children' s Hospital of Pittsburgh (NIH 5M01 RR00084) and by a grant from the Research Advisory Committee of Children' s Hospital of Pittsburgh. Presented in part at the Annual Meeting of the Ambulatory Pediattic Association, May 1994, in Seattle, Wash., and at the Annual Meeting of the Society for Academic Emergency Medicine, May 1994, in Washington, D.C. Submitted for publication June 6, 1995; accepted Sept. 8, 1995. Reprint requests: Daniel J. Isaacman, MD, Chief, Division of Pediatric Emergency Medicine, Children's Hospital of the King's Daughters, 601 Children's Lane, Norfolk, VA 23507. aNow at the A. I. duPont Institute, Wilmington, Del. Copyright © 1996 by Mosby-Year Book, Inc. 0022-3476/96/$5.00 + 0 9•20•69218
190
as little as 1 ml of blood per vial to test for bacteremia in children. 1-5 This "minimalist" approach may have evolved because of the inherent difficulty in obtaining blood from children. To propose that a single low-volume blood culture is sufficiently sensitive, one must make two assumptions reCFU ED
Colony-forming units Emergency department
garding bacteremia in children that may not be true. First, the collection of a single blood culture presupposes that bacteremia is continuous; on the contrary, in most patients bacteremia is an intermittent event. 6 Second, to justify obtaining a small volume of blood for culture, one must assume that the bacteremia is of sufficient density to be detected. Although bacterial meningitis and some other focal infections are associated with high-density bacteremia (100
The Journal of Pediatrics Volume 128, Number 2
organisms or more per milliliter), most cases of bacteremia in pediatric outpatients are of low density (-<10 to 15 organisms per milliliter).7' 8 This study was designed to determine whether obtaining two blood cultures instead of one improves the detection of bacteremia in children and to examine the effect of the volume of blood collected on the recovery of pathogens from blood culture. METHODS Children 18 years of age or less seen at the Children's Hospital of Pittsburgh Emergency Department between February 1991 and February 1994 were eligible for study if, in the judgment of the examining physician, they required a blood culture as part of their evaluation. Because the investigation focused on patients with bacteremia, an effort was made to select children with clinical characteristics associated with a high likelihood of having bacteremia: age between 3 and 36 months; fever, with a temperature higher than 39.5 ° C; and/or clinical toxicity. Investigators were available to enroll patients during daylight hours (8 AM until 6 r'M) and selected evening and weekend hours. Patients were excluded if they had received systemic antibiotics in the 48 hours preceding study entry, had a history of severe skin disease or hypersensitivity to iodophor, or required immediate intravenous access (before notification of the investigator). Patients were also excluded if venipuncture was unsuccessful (defined as either no blood culture specimen obtained or only a single small-volume specimen obtained). For practical reasons, febrile infants less than 2 months of age were not enrolled. Informed consent was obtained from each patient's parent or guardian. Each enrollee then had two blood samples obtained sequentially, in random order, from separate sites. Sample A, containing 2 ml of blood, was inoculated in the standard manner into aerobic and anaerobic culture bottles (1 ml per vial). Sample B, containing 9.5 ml of blood, was divided into three aliquots: B1 (2 ml), B2 (6 ml), and ISO (1.5 ml). The three aliquots were instilled in random order as follows: B 1 was inoculated into one set of culture bottles (1 ml per vial); B2 was inoculated into one set of culture bottles (3 ml per vial); and ISO was injected into an Isolator tube (Wampole Laboratories, Cranbury, N.J.) and then quantitatively plated onto semisolid media. All blood cultures were processed by the microbiology laboratory of Children's Hospital of Pittsburgh by means of the Bactec NR-660 automated system (Becton Dickinson Diagnostic Instrument Systems, Towson, Md.). The Bactec 6A (aerobic) and 7A (anaerobic) vials used with this system contain 30 ml of media. Bacterial growth was monitored twice daily for 7 days. When bacterial growth was detected, isolates were subcultured onto appropriate culture media.
Isaacman et al.
19 1
For each positive blood culture result an assessment was made, with the use of predetermined criteria, as to whether the isolate represented a tree pathogen or a contaminant: 9 Bacteremia was defined as growth of a pathogen in any of the culture bottles obtained. Data analysis. To determine whether the collection of two blood cultures instead of one increased the detection of bacteremia, we compared the pathogen recovery rate of a single small-volume culture (A or B 1) with that of the combination of the two small-volume cultures; the combination (A + B 1) was considered to show positive results if either sample A or sample B 1 showed positive results. To ascertain whether increasing the volume of the inocuhim from 1 to 3 ml enhanced the detection of bacteremia, we compared pathogen recovery between samples B1 and B2, two specimens that were drawn simultaneously, in which volume was the only variable. Because the study design included the use of the Isolator system for the performance of quantitative cultures, we were also able to compare the detection of bacteremia using this system with that using the Bactec system. All comparisons of the recovery rates of pathogens were done with the Exact McNemar Test. In each case, significance was considered at p less than 0.05. RESULTS Three hundred forty-two eligible patients were approached for enrollment during the study period. Thirty-five patients declined to participate, and seven were excluded because venipuncture was unsuccessful (either no blood culture specimen was obtained or only a single small-volume specimen was obtained); thus our study population consisted of 300 patients. Fifty-seven percent of patients were male. Sixty-two percent of enrollees were black, 35% were white, and 3% were of other races. The mean age of the study patients was 20 months (median, 15 months), with a range of 2 to 209 months. In 177 children the A sample was collected first, and in 123 children the B sample was collected first. Complete sets of culture specimens (A, B1, B2, and ISO) were not obtained from all patients because of failed venipuncture (sample A not obtained) or collection of an inadequate volume of blood (B 1, B2, or ISO specimens not obtained); the numbers of specimens collected for study were 292 sample A, 288 sample B1, 270 Sample B2, and 240 sample ISO specimens. Thirty patients, or 10% of the study population, had bacteremia (Table). In 13 children (43%) the pathogen was isolated with all of the culture methods tested (concordant cases); in 17 patients (57%), one or more of the cultures yielded no growth (discordant cases). The predominant etiologic agent was Streptococcuspneumoniae (25 cases); in addition, there was one case each of bacteremia caused by Salmonella species, group A [3-hemolytic streptococcus, viridans strep-
192
Isaacman et al.
The Journal of Pediatrics February 1996
100 o~
24._b
29
80
:::i 21"
29
('-
::::
'ili!
~
=.,-,° 60
"
/~
I1)
X~
iiii
le° t5
i2
-
'.........
13 :i:!:! 25 !:i:i:
14
>, 40 -'" .
20-
0
.
.
17
::!:i 2,: ~'~-'~ '*::~:..
~x,,N ~g
"~
•
22
?:ii: N
0
if) t"
~-
:i:::l
///
. e 23
.
--
.
~
,-
2s .
_5_
~
~
24 hrs Final All p a t h o g e n s
Jt
///
g
24 hrs
Final
S. pneumoniae
Culture type
[--IA F-/-AB1~B2 ~A+B1 F=tlSO Figure. Sensitivity of detecting bacteremia for each culture strategy. The percentage of children with bacteremia who were identified by each culture type is represented on the vertical axis. Results are shown for all pathogens and for S. pneumoniae alone at 24 hours and at final (7-day) reading. The comparisons were performed with the exact McNemar test: aB2 versus A, p = 0.007; B2 versus B1, p = 0.0005; B2 versus A + BI, p = 0.04; and B2 versus ISO, p less than 0.001. bB2 versus B1, p = 0.02; and B2 versus ISO, p = 0.02. CB2 versus A, p = 0.04; B2 versus B 1, p = 0.002; and B2 versus ISO, p = 0.002. aB2 versus B1, p = 0.02; and B2 versus ISO, p = 0.04. tococci, Haemophilus influenzae type b, and Escherichia coli. Of the 30 patients with bacteremia, one patient did not have a B2 (large-volume) specimen collected and two patients did not have an ISO specimen obtained. Thirteen children with bacteremia had an identified focus of infection, and 17 had no focus of infection. Because blood cultures are most helpful to clinicians when the results return in a timely fashion, we compared the blood culture results 24 hours after collection, using each strategy (Figure). The sensitivity for detecting all pathogens at 24 hours was highest for the B2 sample (21/29, or 72%) and was significantly greater than that of the A sample (11/30, or 3 7 % ; p = 0.007), the B1 sample (10/30, or 3 3 % ; p = 0.0005),
the ISO sample (6/28, or 21%; p<0.001), and the combination of the two small-volume samples (A + B1) (14/30, or 47%; p = 0.04). No differences were detected between a single small-volume culture (A or B 1) and the combination of both small-volume cultures (A + B1). Because S. pneumoniae was the principal pathogen identified, we analyzed the pathogen recovery data separately for this organism. We found that the recovery rate for S. pneumoniae closely paralleled that observed for all pathogens, although the recovery rates were slightly lower for S. pneumoniae. The pathogen recovery rates at the final (7-day) reading for each of the blood culture types are also presented in the Figure. The sensitivity of detection of bacteremia remained
The Journal
of Pediatrics
128, N u m b e r
Volume
Isaacman
et al.
1 9 3
2
Table. Culture results for the 30 patients with bacteremia Culture results Patient No.
Age (mo)
A Organism
B1
At 24 hr
Final
At 24 hr
B2
ISO
Final
At 24 hr
Final
At 24 hr
Final
+
Concordant cases 3
5
4 16 7 24 10 17 12 15 14 7 15 14 17 12 18 7 21 128 24 24 27 9 29 15 Discordant cases 1 30 2 43
S. p n e u m o n i a e
+
+
+
+
+
+
+
S. p n e u m o n i a e
+
+
-
+
+
+
+
+
S. p n e u m o n i a e
+
+
+
+
+
+
ND
S. p n e u m o n i a e
+ +
+* +
+
+t +
+ ND
+* ND
+
ND + +
S. p n e u m o n i a e
-
+t
+
+
+
+
+
S. p n e u m o n i a e
+
+
+
+
+
-
+
S. p n e u m o n i a e
-
+
+
+
+
-
+
GABHS
-
+
+
+
+
-
+
S. v i r i d a n s $
-
+
+
+
+
+
+
S. p n e u m o n i a e
-
+
-
+
-
+
ND
ND
S. p n e u m o n i a e
+
+
+
+
+
+
+
+
S. p n e u m o n i a e
-
+t
-
+t
+
+
-
+
S. p n e u m o n i a e
-
+?
.
S. p n e u m o n i a e
+
+
+
Salmonella
sp
+
+
.
.
.
.
.
NA
-
5
11
S. p n e u m o n i a e
.
.
.
.
+
+
-
-
6
79
S. p n e u m o n i a e
.
.
.
.
+
+*
-
-
8
5
14. i n f l u e n z a e
+
+
+
+
+
-
-
9
32
S. p n e u m o n i a e
+
+
.
11
3
S. p n e u m o n i a e
.
.
.
.
+
+t
-
-
13 16 19 20 22 23 25 26 28 30
14 19
S. p n e u m o n i a e
.
.
.
.
+
+
-
-
S. p n e u m o n i a e
.
.
.
.
+
+
-
+
11
S. p n e u m o n i a e
.
.
.
.
+t
-
-
14
S. p n e u m o n i a e
+
+*
.
.
19
S. p n e u m o n i a e
-
+
.
.
8
S. p n e u m o n i a e
-
-
+
+
+
+
-
+
E. coli
-
-
+
+
+
+
-
+
26 16
S. p n e u m o n i a e
.
S. p n e u m o n i a e
-
+
+?
-
17
S. p n e u m o n i a e
.
+
-
5
.
.
+ .
.
+
+
.
.
.
.
.
.
.
+t
.
.
.
.
+
.
.
.
.
+
.
-
+, Positive result; -, negative result; N D , not done; N A , data not available; G A B H S , group A [3-hemolyticstreptococci. *Culture positive in aerobic vial only. ?Culture positive in anaerobic vial only. :~Organismisolatedfrom a 10-year-oldgirl with Down syndromeand ventricularseptaldefect.An echocardiogramrevealeda vegetationon the mitral valve leaflets.
greatest in the B2 sample (24/29, or 83%) and was higher
alone, and in four sets the pathogen was isolated from the
than that observed in the B1 sample (18/30, or 60%;
aerobic vial alone (Table). In four patients (Nos. 1, 11, 19,
p - - 0 . 0 2 ) or in the ISO sample (15/28, or 54%; p = 0.02)
and 28) the pathogen was recovered exclusively from the
collected at the same time. The recovery rate of the B2 cul-
anaerobic culture, and in two patients (Nos. 6 and 20) the
ture was not significantly different from that of the A culture
pathogen was recovered only from the aerobic culture.
(19/30, or 63%; p = 0.18) or the combination of the two
O f the 1090 cultures obtained from the study patients, 19
small-volume cultures (A + B1) (22/30, or 73%; p = 0.55).
(1.7%) yielded contaminants. In no instance was a contam-
No significant differences were found in pathogen recovery
inant isolated from more than one culture in any patient. Or-
when either of the small-volume samples alone (A or B1)
ganisms detected included
was compared with the combination o f the two small-
cultures); one isolate each of L e u c o n o s t o c species, viridans streptococci, C l o s t r i d i u m perfringens, and Bacillus species;
volume samples (A + B1). There were 13 culture sets (from nine patients) in which
Staphylococcus
epidermidis
(15
aerobic and anaerobic results were discordant; in nine cul-
and a combination of diphtheroids and viridans streptococci. Contaminants were found in 5 of 292 A cultures (1.7%), 2
ture sets the pathogen was recovered from the anaerobic vial
o f 288 B1 cultures (0.7%), 2 o f 270 B2 cultures (0.7%), and
194
Isaacman et al.
10 of 240 ISO cultures (4.2%). The contamination rate observed by means of the Isolator system was significantly higher than the rates observed with the B1 and B2 Bactec bottles (p -- 0.02 for each comparison). None of the 13 discordant cases yielded quantitative cultures with growth of more than 10 CFU/ml, whereas 5 of 15 concordant cases had quantitative cultures with growth of more than 10 CFU/ml (p = 0.03, Fisher Exact Test). DISCUSSION
The standard practice for diagnosing bacteremia in adults is to obtain two or more blood culture specimens containing a minimum of 10 ml of blood per bottle. 1-3' 10 This strategy is supported by studies demonstrating enhanced detection of bacteremia with increasing numbers of blood cultures. 1, to Despite these findings, the practice of obtaining multiple culture specimens or a large volume of blood is seldom extended to children. Although a single small-volume blood culture sample may be easier to obtain than multiple samples or even a single large-volume sample, the findings of this study and those of a previous investigation 11 indicate that collecting a single small-volume sample for culture will miss a significant proportion of children with bacteremia. In our study the collection of a relatively large sample of blood (3 ml/bottle) improved the detection of bacteremia, with a marked improvement in sensitivity at 24 hours. Although we were able to collect a large-volume sample successfully from 87.9% of children, the experience of the investigators in drawing blood from young children may have contributed to the high success rate. In addition, because our selection process omitted febrile infants less than 2 months of age, we cannot extrapolate the success rate observed to very young infants, in whom we might expect blood collection to be more problematic. Although collecting the additional volume of blood may be difficult, we believe that the improved sensitivity for detecting bacteremia, particularly at 24 hours, justifies the effort. Some investigators have questioned the routine inclusion of an anaerobic bottle during the collection of blood for culture, because of an apparent decline in the incidence of anaerobic bacteremia, t2-15 Nevertheless, we included anaerobic culture bottles because of the concern that aerobic cultures alone might fail to detect certain facultative organisms, such as S. pneumoniae. 16 This decision appears to be justified; we identified four patients in whom S. pneumoniae was recovered from the anaerobic vial only (vs two patients in whom S. pneumoniae was recovered from the aerobic bottle only). The rationale for including a quantitative blood culture in our design was to determine whether discordant cultures were related to low-density bacteremia. None of the discordant cultures yielded growth of more than 10 CFU/ml. Our
The Journal of Pediatrics February 1996
data agree with those of Eisenrach et al.] 7 who found the Isolator system to be significantly less sensitive than the Bactec system for the detection of bacteremia. The increased rate of contamination observed with the Isolator system was a further disadvantage of this method. We found the contamination rate in the large-volume samples to be similar to that of the small-volume samples and lower than the contamination rate in the quantitative cultures. These rates were comparable to those reported in two previous investigations at our institution, in which phlebotomy was performed by trained emergency department nursing staff,9, 11 and were lower than contamination rates noted by other authors. 2, 18-20We believe that scrupulous skin cleansing is the most important factor in reducing contamination. This investigation was conducted in the ED setting and primarily involved children with high feyer and no localizing signs; thus we cannot generalize these findings to other clinical situations. Another limitation of the study relates to the process of patient selection, which required both the identification of candidates by physicians in the ED and the availability of an investigator; as a result, our sample represented a minority of the patients who had blood culture specimens obtained in the ED during the study period. Nevertheless, we believe that our findings were not substantially influenced by sampling bias, because the criteria used by clinicians for selecting patients for blood culture were not altered by the conditions of the study. We conclude that a single small-volume blood culture fails to identify a significant proportion of children with bacteremia seen in theED setting. Our data indicate that previous studies, which have based the diagnosis of bacteremia on a single small-volume blood culture, have underestimated the prevalence of bacteremia in children. We propose that collecting a single large-volume sample represents the most practical approach to the diagnosis of bacteremia in children, in that it maximizes pathogen recovery while sparing the patient the cost and pain of an additional venipuncture. We thank the pediatric house staff, nursing staff, and microbiology staff of Children's Hospital of Pittsburgh for their assistance with this project; Barbara H. Hanusa, PhD, for her guidance with the statistical analysis; Kenneth D. Rogers, MD, and Ellen R. Wald, MD, for their thoughtful review of the manuscript; and David Kazimer and Diane Weidner for their editorial assistance. REFERENCES
1. Washington JA. Blood cultures: principles and techniques. Mayo Clin Proc 1975;50:91-8. 2. Washington JA. Conventional approaches to blood culture. In: Washington JA, ed. The detection of septicemia. West Palm Beach, Florida: CRC Press, 1978:41-87. 3. Washington JA, Ilstmp D. Blood cultures: issues and controversies. Rev Infect Dis 1986;8:792-801.
The Journal of Pediatrics Volume 128, Number 2
Isaacman et al.
4. Chandrasekar P, Brown WJ. Clinical issues of blood cultures. Arch Intern Med 1994;154:841-9. 5. Durban WA, Szynaczak EG, Goldmann DA. Quantitative blood cultures in childhood bacteremia. J PED~A~ 1978;92:77880. 6. Bennett IL Jr, Beeson PB. Bacteremia: a consideration of some experimental and cfinical aspects. Yale J Biol Med 1954;26:241 62. 7. Bell LM, Alpert G, Campos JM, Plotkin SA. Routine quantitative blood cultures in children with Haemophilus influenzae or Streptococcus pneumoniae bacteremia. Pediatrics 1985;76: 901-4. 8. Sullivan TD, LaScolea LJ, Neter E. Relationship between the magnitude of bacteremia in children and the clinical disease. Pediatrics 1982;62:699-702. 9. Isaacman DJ, Karasic RB. Lack of effect of changing needles on contamination of blood cultures. Pediatr Infect Dis J 1990; 9:274-8. 10. Weinstein MP, Reller LB, Murphy JR, Lichtenstein KA. The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults: laboratory and epidemiologic investigations. Rev Infect Dis 1983;5:35-53. 11. Isaacman DJ, Karasic RB. Utility of collecting blood cultures through newly inserted intravenous catheters. Pediatr Infect Dis J 1990;9:815-8. 12. Dor Sher CW, Rosenblatt JE, Wilson WR, Ilstrup DM. Anaer-
13.
14.
15.
16. 17.
18.
19.
20.
195
obic bacteremia: decreasing rate over a 15-year period. Am J Med 1991;13:633-6. Lombardi DP,. Engleberg NC. Anaerobic bacteremia: incidence, patient characteristics, and clinical significance. Am J Med 1992;92:53-60. Murray PR, Traynor P, Hopson D. Critical assessment of blood culture techniques: analysis of recovery of obligate and facultative aerobic and anaerobic blood culture bottles. J Clin Microbiol 1992;30:1462-8. Zaidi AKM, Knaut AL, Mirett S, Reller LB. Value of routine anaerobic blood cultures for pediatric patients. J PEDIATa 1995; 127:263-8. Martin WJ. Routine anaerobic blood cultures: reasons for continued use. Clin Microbiol Newsletter 1992; 14: i33-4. Eisenrach K, Dyke J, Boehme M, Johnson B, Cook MB. Pediatric blood culture evaluation of the Bactec PEDS Plus and DuPont Isolator 1.5 systems. Diagnostic Microbiol Infect Dis 1992;15:225-31. Beaman KD, Kasten BL, Gavan TL. Rate of detection of bacteremia: retrospective evaluation of 23,393 blood cultures. Cleveland Clin Q 1977;44:129-36. MacGregor RR, Beaty HN. Evaluation of positive blood cultures: guidelines for early differentiation of contaminated from valid positive cultures. Arch Intern Med 1972;130:84-7. Roberts FJ. A review of positive blood cultures: identification and source of microorganisms and patterns of sensitivity to antibiotics. Rev Infect Dis 1980;2:329-39.
B O U N D V O L U M E S AVAILABLE TO SUBSCRIBERS B o u n d volumes of the 1996 issues of THE JOURNAL OF PEDIATRICS are available to subscribers (only) from the Publisher, at a cost of $81.00 for domestic, $103.79 for Canadian, and $97.00 for international subscribers, for Vol. 128 (January-June) and Vol. 129 (July-December), shipping charges included. Each b o u n d volume contains subject and author indexes, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram, with the Journal name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact M o s b y - Y e a r Book, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, M O 63146-3318, USA/800-453-4351, or 314-4534351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular Journal subscription.
Objective: To determine whether bacteremia can be detected more rapidly and completely by (I) obtaining two blood cultures instead of one and/or (2) collecting a larger volume of blood. Study design: Prospective comparison of different strategies in 300 patients undergoing blood culture for suspected bacteremia. Each patient had two samples of blood, A (2 ml) and B (9.5 ml), obtained sequentially from separate sites. The B sample was divided into three aliquots: BI (2 ml), B2 (6 ml), and ISO (I .5 ml, quantitative culture). Results:A pathogen was isolated from one or more blood cultures in 30 patients (10% of cases). When measured at 24 hours, the pathogen recovery rate for the B2 sample (72%) was higher than that for the individual small-volume samples (A = 37%, BI = 33%; p <0.01 for each comparison) and for the combination of the two small-volume samples (A + BI = 47%; p = 0.04). At final (7-day) reading the pathogen recovery rate for the B2 sample (83%) was higher than that for BI (60%; p = 0.02) and similar to the recovery rate observed with the combination of the two small-volume cultures (A + BI = 73%; p = 0.55). Conclusions: Increasing the volume of blood inoculated into blood culture bottles improves the timely detection of bacteremia in pediatric patients and spares the patients the cost and pain of an additional venipuncture. (J PEDIATR1996; 128:190-5) The optimal number of blood cultures and volume of blood needed to detect bacteremia in children are not known. Nevertheless, clinicians often rely on a single culture containing Supported in part by a grant from the General Clinical Research Center of Children' s Hospital of Pittsburgh (NIH 5M01 RR00084) and by a grant from the Research Advisory Committee of Children' s Hospital of Pittsburgh. Presented in part at the Annual Meeting of the Ambulatory Pediattic Association, May 1994, in Seattle, Wash., and at the Annual Meeting of the Society for Academic Emergency Medicine, May 1994, in Washington, D.C. Submitted for publication June 6, 1995; accepted Sept. 8, 1995. Reprint requests: Daniel J. Isaacman, MD, Chief, Division of Pediatric Emergency Medicine, Children's Hospital of the King's Daughters, 601 Children's Lane, Norfolk, VA 23507. aNow at the A. I. duPont Institute, Wilmington, Del. Copyright © 1996 by Mosby-Year Book, Inc. 0022-3476/96/$5.00 + 0 9•20•69218
190
as little as 1 ml of blood per vial to test for bacteremia in children. 1-5 This "minimalist" approach may have evolved because of the inherent difficulty in obtaining blood from children. To propose that a single low-volume blood culture is sufficiently sensitive, one must make two assumptions reCFU ED
Colony-forming units Emergency department
garding bacteremia in children that may not be true. First, the collection of a single blood culture presupposes that bacteremia is continuous; on the contrary, in most patients bacteremia is an intermittent event. 6 Second, to justify obtaining a small volume of blood for culture, one must assume that the bacteremia is of sufficient density to be detected. Although bacterial meningitis and some other focal infections are associated with high-density bacteremia (100
The Journal of Pediatrics Volume 128, Number 2
organisms or more per milliliter), most cases of bacteremia in pediatric outpatients are of low density (-<10 to 15 organisms per milliliter).7' 8 This study was designed to determine whether obtaining two blood cultures instead of one improves the detection of bacteremia in children and to examine the effect of the volume of blood collected on the recovery of pathogens from blood culture. METHODS Children 18 years of age or less seen at the Children's Hospital of Pittsburgh Emergency Department between February 1991 and February 1994 were eligible for study if, in the judgment of the examining physician, they required a blood culture as part of their evaluation. Because the investigation focused on patients with bacteremia, an effort was made to select children with clinical characteristics associated with a high likelihood of having bacteremia: age between 3 and 36 months; fever, with a temperature higher than 39.5 ° C; and/or clinical toxicity. Investigators were available to enroll patients during daylight hours (8 AM until 6 r'M) and selected evening and weekend hours. Patients were excluded if they had received systemic antibiotics in the 48 hours preceding study entry, had a history of severe skin disease or hypersensitivity to iodophor, or required immediate intravenous access (before notification of the investigator). Patients were also excluded if venipuncture was unsuccessful (defined as either no blood culture specimen obtained or only a single small-volume specimen obtained). For practical reasons, febrile infants less than 2 months of age were not enrolled. Informed consent was obtained from each patient's parent or guardian. Each enrollee then had two blood samples obtained sequentially, in random order, from separate sites. Sample A, containing 2 ml of blood, was inoculated in the standard manner into aerobic and anaerobic culture bottles (1 ml per vial). Sample B, containing 9.5 ml of blood, was divided into three aliquots: B1 (2 ml), B2 (6 ml), and ISO (1.5 ml). The three aliquots were instilled in random order as follows: B 1 was inoculated into one set of culture bottles (1 ml per vial); B2 was inoculated into one set of culture bottles (3 ml per vial); and ISO was injected into an Isolator tube (Wampole Laboratories, Cranbury, N.J.) and then quantitatively plated onto semisolid media. All blood cultures were processed by the microbiology laboratory of Children's Hospital of Pittsburgh by means of the Bactec NR-660 automated system (Becton Dickinson Diagnostic Instrument Systems, Towson, Md.). The Bactec 6A (aerobic) and 7A (anaerobic) vials used with this system contain 30 ml of media. Bacterial growth was monitored twice daily for 7 days. When bacterial growth was detected, isolates were subcultured onto appropriate culture media.
Isaacman et al.
19 1
For each positive blood culture result an assessment was made, with the use of predetermined criteria, as to whether the isolate represented a tree pathogen or a contaminant: 9 Bacteremia was defined as growth of a pathogen in any of the culture bottles obtained. Data analysis. To determine whether the collection of two blood cultures instead of one increased the detection of bacteremia, we compared the pathogen recovery rate of a single small-volume culture (A or B 1) with that of the combination of the two small-volume cultures; the combination (A + B 1) was considered to show positive results if either sample A or sample B 1 showed positive results. To ascertain whether increasing the volume of the inocuhim from 1 to 3 ml enhanced the detection of bacteremia, we compared pathogen recovery between samples B1 and B2, two specimens that were drawn simultaneously, in which volume was the only variable. Because the study design included the use of the Isolator system for the performance of quantitative cultures, we were also able to compare the detection of bacteremia using this system with that using the Bactec system. All comparisons of the recovery rates of pathogens were done with the Exact McNemar Test. In each case, significance was considered at p less than 0.05. RESULTS Three hundred forty-two eligible patients were approached for enrollment during the study period. Thirty-five patients declined to participate, and seven were excluded because venipuncture was unsuccessful (either no blood culture specimen was obtained or only a single small-volume specimen was obtained); thus our study population consisted of 300 patients. Fifty-seven percent of patients were male. Sixty-two percent of enrollees were black, 35% were white, and 3% were of other races. The mean age of the study patients was 20 months (median, 15 months), with a range of 2 to 209 months. In 177 children the A sample was collected first, and in 123 children the B sample was collected first. Complete sets of culture specimens (A, B1, B2, and ISO) were not obtained from all patients because of failed venipuncture (sample A not obtained) or collection of an inadequate volume of blood (B 1, B2, or ISO specimens not obtained); the numbers of specimens collected for study were 292 sample A, 288 sample B1, 270 Sample B2, and 240 sample ISO specimens. Thirty patients, or 10% of the study population, had bacteremia (Table). In 13 children (43%) the pathogen was isolated with all of the culture methods tested (concordant cases); in 17 patients (57%), one or more of the cultures yielded no growth (discordant cases). The predominant etiologic agent was Streptococcuspneumoniae (25 cases); in addition, there was one case each of bacteremia caused by Salmonella species, group A [3-hemolytic streptococcus, viridans strep-
192
Isaacman et al.
The Journal of Pediatrics February 1996
100 o~
24._b
29
80
:::i 21"
29
('-
::::
'ili!
~
=.,-,° 60
"
/~
I1)
X~
iiii
le° t5
i2
-
'.........
13 :i:!:! 25 !:i:i:
14
>, 40 -'" .
20-
0
.
.
17
::!:i 2,: ~'~-'~ '*::~:..
~x,,N ~g
"~
•
22
?:ii: N
0
if) t"
~-
:i:::l
///
. e 23
.
--
.
~
,-
2s .
_5_
~
~
24 hrs Final All p a t h o g e n s
Jt
///
g
24 hrs
Final
S. pneumoniae
Culture type
[--IA F-/-AB1~B2 ~A+B1 F=tlSO Figure. Sensitivity of detecting bacteremia for each culture strategy. The percentage of children with bacteremia who were identified by each culture type is represented on the vertical axis. Results are shown for all pathogens and for S. pneumoniae alone at 24 hours and at final (7-day) reading. The comparisons were performed with the exact McNemar test: aB2 versus A, p = 0.007; B2 versus B1, p = 0.0005; B2 versus A + BI, p = 0.04; and B2 versus ISO, p less than 0.001. bB2 versus B1, p = 0.02; and B2 versus ISO, p = 0.02. CB2 versus A, p = 0.04; B2 versus B 1, p = 0.002; and B2 versus ISO, p = 0.002. aB2 versus B1, p = 0.02; and B2 versus ISO, p = 0.04. tococci, Haemophilus influenzae type b, and Escherichia coli. Of the 30 patients with bacteremia, one patient did not have a B2 (large-volume) specimen collected and two patients did not have an ISO specimen obtained. Thirteen children with bacteremia had an identified focus of infection, and 17 had no focus of infection. Because blood cultures are most helpful to clinicians when the results return in a timely fashion, we compared the blood culture results 24 hours after collection, using each strategy (Figure). The sensitivity for detecting all pathogens at 24 hours was highest for the B2 sample (21/29, or 72%) and was significantly greater than that of the A sample (11/30, or 3 7 % ; p = 0.007), the B1 sample (10/30, or 3 3 % ; p = 0.0005),
the ISO sample (6/28, or 21%; p<0.001), and the combination of the two small-volume samples (A + B1) (14/30, or 47%; p = 0.04). No differences were detected between a single small-volume culture (A or B 1) and the combination of both small-volume cultures (A + B1). Because S. pneumoniae was the principal pathogen identified, we analyzed the pathogen recovery data separately for this organism. We found that the recovery rate for S. pneumoniae closely paralleled that observed for all pathogens, although the recovery rates were slightly lower for S. pneumoniae. The pathogen recovery rates at the final (7-day) reading for each of the blood culture types are also presented in the Figure. The sensitivity of detection of bacteremia remained
The Journal
of Pediatrics
128, N u m b e r
Volume
Isaacman
et al.
1 9 3
2
Table. Culture results for the 30 patients with bacteremia Culture results Patient No.
Age (mo)
A Organism
B1
At 24 hr
Final
At 24 hr
B2
ISO
Final
At 24 hr
Final
At 24 hr
Final
+
Concordant cases 3
5
4 16 7 24 10 17 12 15 14 7 15 14 17 12 18 7 21 128 24 24 27 9 29 15 Discordant cases 1 30 2 43
S. p n e u m o n i a e
+
+
+
+
+
+
+
S. p n e u m o n i a e
+
+
-
+
+
+
+
+
S. p n e u m o n i a e
+
+
+
+
+
+
ND
S. p n e u m o n i a e
+ +
+* +
+
+t +
+ ND
+* ND
+
ND + +
S. p n e u m o n i a e
-
+t
+
+
+
+
+
S. p n e u m o n i a e
+
+
+
+
+
-
+
S. p n e u m o n i a e
-
+
+
+
+
-
+
GABHS
-
+
+
+
+
-
+
S. v i r i d a n s $
-
+
+
+
+
+
+
S. p n e u m o n i a e
-
+
-
+
-
+
ND
ND
S. p n e u m o n i a e
+
+
+
+
+
+
+
+
S. p n e u m o n i a e
-
+t
-
+t
+
+
-
+
S. p n e u m o n i a e
-
+?
.
S. p n e u m o n i a e
+
+
+
Salmonella
sp
+
+
.
.
.
.
.
NA
-
5
11
S. p n e u m o n i a e
.
.
.
.
+
+
-
-
6
79
S. p n e u m o n i a e
.
.
.
.
+
+*
-
-
8
5
14. i n f l u e n z a e
+
+
+
+
+
-
-
9
32
S. p n e u m o n i a e
+
+
.
11
3
S. p n e u m o n i a e
.
.
.
.
+
+t
-
-
13 16 19 20 22 23 25 26 28 30
14 19
S. p n e u m o n i a e
.
.
.
.
+
+
-
-
S. p n e u m o n i a e
.
.
.
.
+
+
-
+
11
S. p n e u m o n i a e
.
.
.
.
+t
-
-
14
S. p n e u m o n i a e
+
+*
.
.
19
S. p n e u m o n i a e
-
+
.
.
8
S. p n e u m o n i a e
-
-
+
+
+
+
-
+
E. coli
-
-
+
+
+
+
-
+
26 16
S. p n e u m o n i a e
.
S. p n e u m o n i a e
-
+
+?
-
17
S. p n e u m o n i a e
.
+
-
5
.
.
+ .
.
+
+
.
.
.
.
.
.
.
+t
.
.
.
.
+
.
.
.
.
+
.
-
+, Positive result; -, negative result; N D , not done; N A , data not available; G A B H S , group A [3-hemolyticstreptococci. *Culture positive in aerobic vial only. ?Culture positive in anaerobic vial only. :~Organismisolatedfrom a 10-year-oldgirl with Down syndromeand ventricularseptaldefect.An echocardiogramrevealeda vegetationon the mitral valve leaflets.
greatest in the B2 sample (24/29, or 83%) and was higher
alone, and in four sets the pathogen was isolated from the
than that observed in the B1 sample (18/30, or 60%;
aerobic vial alone (Table). In four patients (Nos. 1, 11, 19,
p - - 0 . 0 2 ) or in the ISO sample (15/28, or 54%; p = 0.02)
and 28) the pathogen was recovered exclusively from the
collected at the same time. The recovery rate of the B2 cul-
anaerobic culture, and in two patients (Nos. 6 and 20) the
ture was not significantly different from that of the A culture
pathogen was recovered only from the aerobic culture.
(19/30, or 63%; p = 0.18) or the combination of the two
O f the 1090 cultures obtained from the study patients, 19
small-volume cultures (A + B1) (22/30, or 73%; p = 0.55).
(1.7%) yielded contaminants. In no instance was a contam-
No significant differences were found in pathogen recovery
inant isolated from more than one culture in any patient. Or-
when either of the small-volume samples alone (A or B1)
ganisms detected included
was compared with the combination o f the two small-
cultures); one isolate each of L e u c o n o s t o c species, viridans streptococci, C l o s t r i d i u m perfringens, and Bacillus species;
volume samples (A + B1). There were 13 culture sets (from nine patients) in which
Staphylococcus
epidermidis
(15
aerobic and anaerobic results were discordant; in nine cul-
and a combination of diphtheroids and viridans streptococci. Contaminants were found in 5 of 292 A cultures (1.7%), 2
ture sets the pathogen was recovered from the anaerobic vial
o f 288 B1 cultures (0.7%), 2 o f 270 B2 cultures (0.7%), and
194
Isaacman et al.
10 of 240 ISO cultures (4.2%). The contamination rate observed by means of the Isolator system was significantly higher than the rates observed with the B1 and B2 Bactec bottles (p -- 0.02 for each comparison). None of the 13 discordant cases yielded quantitative cultures with growth of more than 10 CFU/ml, whereas 5 of 15 concordant cases had quantitative cultures with growth of more than 10 CFU/ml (p = 0.03, Fisher Exact Test). DISCUSSION
The standard practice for diagnosing bacteremia in adults is to obtain two or more blood culture specimens containing a minimum of 10 ml of blood per bottle. 1-3' 10 This strategy is supported by studies demonstrating enhanced detection of bacteremia with increasing numbers of blood cultures. 1, to Despite these findings, the practice of obtaining multiple culture specimens or a large volume of blood is seldom extended to children. Although a single small-volume blood culture sample may be easier to obtain than multiple samples or even a single large-volume sample, the findings of this study and those of a previous investigation 11 indicate that collecting a single small-volume sample for culture will miss a significant proportion of children with bacteremia. In our study the collection of a relatively large sample of blood (3 ml/bottle) improved the detection of bacteremia, with a marked improvement in sensitivity at 24 hours. Although we were able to collect a large-volume sample successfully from 87.9% of children, the experience of the investigators in drawing blood from young children may have contributed to the high success rate. In addition, because our selection process omitted febrile infants less than 2 months of age, we cannot extrapolate the success rate observed to very young infants, in whom we might expect blood collection to be more problematic. Although collecting the additional volume of blood may be difficult, we believe that the improved sensitivity for detecting bacteremia, particularly at 24 hours, justifies the effort. Some investigators have questioned the routine inclusion of an anaerobic bottle during the collection of blood for culture, because of an apparent decline in the incidence of anaerobic bacteremia, t2-15 Nevertheless, we included anaerobic culture bottles because of the concern that aerobic cultures alone might fail to detect certain facultative organisms, such as S. pneumoniae. 16 This decision appears to be justified; we identified four patients in whom S. pneumoniae was recovered from the anaerobic vial only (vs two patients in whom S. pneumoniae was recovered from the aerobic bottle only). The rationale for including a quantitative blood culture in our design was to determine whether discordant cultures were related to low-density bacteremia. None of the discordant cultures yielded growth of more than 10 CFU/ml. Our
The Journal of Pediatrics February 1996
data agree with those of Eisenrach et al.] 7 who found the Isolator system to be significantly less sensitive than the Bactec system for the detection of bacteremia. The increased rate of contamination observed with the Isolator system was a further disadvantage of this method. We found the contamination rate in the large-volume samples to be similar to that of the small-volume samples and lower than the contamination rate in the quantitative cultures. These rates were comparable to those reported in two previous investigations at our institution, in which phlebotomy was performed by trained emergency department nursing staff,9, 11 and were lower than contamination rates noted by other authors. 2, 18-20We believe that scrupulous skin cleansing is the most important factor in reducing contamination. This investigation was conducted in the ED setting and primarily involved children with high feyer and no localizing signs; thus we cannot generalize these findings to other clinical situations. Another limitation of the study relates to the process of patient selection, which required both the identification of candidates by physicians in the ED and the availability of an investigator; as a result, our sample represented a minority of the patients who had blood culture specimens obtained in the ED during the study period. Nevertheless, we believe that our findings were not substantially influenced by sampling bias, because the criteria used by clinicians for selecting patients for blood culture were not altered by the conditions of the study. We conclude that a single small-volume blood culture fails to identify a significant proportion of children with bacteremia seen in theED setting. Our data indicate that previous studies, which have based the diagnosis of bacteremia on a single small-volume blood culture, have underestimated the prevalence of bacteremia in children. We propose that collecting a single large-volume sample represents the most practical approach to the diagnosis of bacteremia in children, in that it maximizes pathogen recovery while sparing the patient the cost and pain of an additional venipuncture. We thank the pediatric house staff, nursing staff, and microbiology staff of Children's Hospital of Pittsburgh for their assistance with this project; Barbara H. Hanusa, PhD, for her guidance with the statistical analysis; Kenneth D. Rogers, MD, and Ellen R. Wald, MD, for their thoughtful review of the manuscript; and David Kazimer and Diane Weidner for their editorial assistance. REFERENCES
1. Washington JA. Blood cultures: principles and techniques. Mayo Clin Proc 1975;50:91-8. 2. Washington JA. Conventional approaches to blood culture. In: Washington JA, ed. The detection of septicemia. West Palm Beach, Florida: CRC Press, 1978:41-87. 3. Washington JA, Ilstmp D. Blood cultures: issues and controversies. Rev Infect Dis 1986;8:792-801.
The Journal of Pediatrics Volume 128, Number 2
Isaacman et al.
4. Chandrasekar P, Brown WJ. Clinical issues of blood cultures. Arch Intern Med 1994;154:841-9. 5. Durban WA, Szynaczak EG, Goldmann DA. Quantitative blood cultures in childhood bacteremia. J PED~A~ 1978;92:77880. 6. Bennett IL Jr, Beeson PB. Bacteremia: a consideration of some experimental and cfinical aspects. Yale J Biol Med 1954;26:241 62. 7. Bell LM, Alpert G, Campos JM, Plotkin SA. Routine quantitative blood cultures in children with Haemophilus influenzae or Streptococcus pneumoniae bacteremia. Pediatrics 1985;76: 901-4. 8. Sullivan TD, LaScolea LJ, Neter E. Relationship between the magnitude of bacteremia in children and the clinical disease. Pediatrics 1982;62:699-702. 9. Isaacman DJ, Karasic RB. Lack of effect of changing needles on contamination of blood cultures. Pediatr Infect Dis J 1990; 9:274-8. 10. Weinstein MP, Reller LB, Murphy JR, Lichtenstein KA. The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults: laboratory and epidemiologic investigations. Rev Infect Dis 1983;5:35-53. 11. Isaacman DJ, Karasic RB. Utility of collecting blood cultures through newly inserted intravenous catheters. Pediatr Infect Dis J 1990;9:815-8. 12. Dor Sher CW, Rosenblatt JE, Wilson WR, Ilstrup DM. Anaer-
13.
14.
15.
16. 17.
18.
19.
20.
195
obic bacteremia: decreasing rate over a 15-year period. Am J Med 1991;13:633-6. Lombardi DP,. Engleberg NC. Anaerobic bacteremia: incidence, patient characteristics, and clinical significance. Am J Med 1992;92:53-60. Murray PR, Traynor P, Hopson D. Critical assessment of blood culture techniques: analysis of recovery of obligate and facultative aerobic and anaerobic blood culture bottles. J Clin Microbiol 1992;30:1462-8. Zaidi AKM, Knaut AL, Mirett S, Reller LB. Value of routine anaerobic blood cultures for pediatric patients. J PEDIATa 1995; 127:263-8. Martin WJ. Routine anaerobic blood cultures: reasons for continued use. Clin Microbiol Newsletter 1992; 14: i33-4. Eisenrach K, Dyke J, Boehme M, Johnson B, Cook MB. Pediatric blood culture evaluation of the Bactec PEDS Plus and DuPont Isolator 1.5 systems. Diagnostic Microbiol Infect Dis 1992;15:225-31. Beaman KD, Kasten BL, Gavan TL. Rate of detection of bacteremia: retrospective evaluation of 23,393 blood cultures. Cleveland Clin Q 1977;44:129-36. MacGregor RR, Beaty HN. Evaluation of positive blood cultures: guidelines for early differentiation of contaminated from valid positive cultures. Arch Intern Med 1972;130:84-7. Roberts FJ. A review of positive blood cultures: identification and source of microorganisms and patterns of sensitivity to antibiotics. Rev Infect Dis 1980;2:329-39.
B O U N D V O L U M E S AVAILABLE TO SUBSCRIBERS B o u n d volumes of the 1996 issues of THE JOURNAL OF PEDIATRICS are available to subscribers (only) from the Publisher, at a cost of $81.00 for domestic, $103.79 for Canadian, and $97.00 for international subscribers, for Vol. 128 (January-June) and Vol. 129 (July-December), shipping charges included. Each b o u n d volume contains subject and author indexes, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram, with the Journal name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact M o s b y - Y e a r Book, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, M O 63146-3318, USA/800-453-4351, or 314-4534351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular Journal subscription.