DIAGN MICROBIOL INFECT DIS 225 1992;15:225-231

Pediatric Blood Culture Evaluation of the BACTEC PEDS Plus and the Du Pont Isolator 1.5 Systems

Kathleen Eisenach, John Dyke, Martha Boehme, Barbara Johnson, and Mary Beth Cook

A new nonradiometric BACTEC medium has been developed for culturing pediatric blood samples. The BACTEC PEDS Plus me- dium (BACTEC PED) consists of 20 ml of an enriched broth with resins. In contrast to the other aerobic BACTEC media, there is a lower concentration of sodium polyanetholesulfonate in the medium and more C02 in the headspace of the vial. This study was conducted in two different pediatric settings to compare the performance of the BACTEC PED medium with the Du Pont Isolator 1.5 system (1so 1.5) in terms of overall or- ganism recovery and time to detection. Equal volumes of up to 1.5 ml were tested in both systems. A total of 4063 culture sets were analyzed, yielding 301 (7.4%) clinically significant isolates. Of these, 86 (29%) were recovered only from the

BACTEC PED and 12 (4%) only from the ISO 1.5 (p (0.001). BACTEC PED recovered significantly more staphylococci and Enterobacteriaceae than Iso 1.5 (p < 0.001 and p < 0.005, respectively). Detection times of isolates recovered in both systems were comparable. For those patients on antibiotic therapy at the time of culture, 21 (37%) were positive only in the BACTEC PED, whereas two (4%) were positive only in the ISO 1.5. The nontherapy group had 61 (27%) organisms that were detected in BACTEC PED only and 9 (4%) in Iso 1.5 only. These results indicate that BACTEC PED is a significant advance in blood culture systems that will provide a sensitive method for detecting pediatric bacteremias.

I N T R O D U C T I O N

The laboratory approach to the recognit ion of bac- teremia in children poses unique problems that are not encoun te red in adults. Large volumes of blood cannot be drawn from children, and often only one culture is obtained. Originally the concern of low- volume draws in y o u n g children was offset by the observation that bacteremia in neonates tends to be of greater microbial magni tude than in adults. However , several studies indicate that the concen- tration of organisms in neonata l sepsis (Dietzman

From the Department of Pathology (K.E., B.J., M.B.C.), Ar- kansas Children's Hospital, Little Rock, Arkansas; and Sparrow Hospital (J.D., M.B.), Lansing, Michigan, USA.

Address reprint requests to Dr. K.D. Eisenach, Arkansas Chil- dren's Hospital, 800 Marshall Street, Little Rock, AR 72202, USA.

Received 3 October 1990; revised and accepted 8 May 1991. © 1992 Elsevier Science Publishing Co., Inc. 655 Avenue of the Americas, New York, NY 10010 0732-8893/92/$5.00

et al., 1974; Neal et al., 1986) and chi ldhood bac- teremia (Durbin et al., 1978; Sullivan et al., 1982; Szymczak et al., 1979) is often low; therefore, cul- turing small vo lumes of b lood may be inadequate. In contrast to adults, some of the organisms most commonly recovered in pediatric pat ients are fas- tidious and require special nutri t ional considera- tions. Also, pediatric pat ients are f requent ly re- ceiving antibiotics for localized infections, such as otitis media, which may influence the recovery of the organism.

The Du Pont Isolator 1.5 sys tem (ISO 1.5) (E.I. du Pont de Nemours and Company , Wilmington, DE) was one of the first commercial blood culture sys- tems designed specifically for culturing pediatric blood specimens. The ISO 1.5 has been compared with other blood culture systems for its efficiency in detect ing sepsis in children (Campos and Spainhour , 1985a and b; Carey, 1984; Welch et al., 1985). These studies did not demons t ra te any statistical difference in the overall organism recovery rates or in the recovery

226 K. Eisenach et al.

of specific organisms or organism groups. However, the time to detection with ISO 1.5 was enhanced by multiple observations of the inoculated plates within the first 48 hr.

The BACTEC blood culture system (Becton Dick- inson Diagnostic Instrument Systems, Sparks, MD) is a well-established semiautomated method for blood culture. Institutions using the BACTEC non- radiometric system generally use only the aerobic 6A medium when culturing small blood volumes. Recently the sodium polyanetholesulfonate (SPS) concentration of the 6A medium was increased to 0.035%. In anticipation of the concern that the in- creased SPS concentration may be inhibitory for Neisseria meningitidis and with the recent develop- ment of the high-blood-volume media, a new BAC- TEC medium was formulated. The BACTEC PEDS Plus medium (BACTEC PED) was developed primarily for pediatric patients and differs from the 6A medium by having a smaller volume of broth (20 ml), a de- creased concentration of SPS (0.025%), and the ad- dition of resins.

In this study, the BACTEC PED medium was eval- uated at two different pediatric sites. The purpose was to compare the performance of BACTEC PED and the Iso 1.5 systems in terms of overall organism re- covery and time to detection. The effect of antimi- crobial therapy on these parameters was also inves- tigated.

MATERIALS AND METHODS

Specimen Collection

A total of 4063 blood samples were collected from outpatients and inpatients at Arkansas Children's Hospital (ACH) and the Pediatric Unit of Sparrow Hospital (SH). Blood specimens were drawn from a single venipuncture following appropriate skin dis- infection. Equivalent blood volumes were inoculated into a BACTEC PED vial and an ISO 1.5 tube at bedside in random order, with the maximum volume of 1.5 ml in each. Both samples were then inverted several times to ensure adequate mixing. Only those paired specimens containing equal volumes of blood were included in the study.

BACTEC PED Medium

The BACTEC PED vial contains 20 ml of soybean-cas- ein digest broth with yeast extract, hemin, vitamin K, 0.025% SPS, nonionic adsorbing resins, and cat- ionic exchange resins. The CO2 content in the head-

space of the vial is higher than that in the other aerobic BACTEC media.

Processing of BACTEC PED

The BACTEC PED vials were placed on an orbital shaker for incubation and agitation for the first 48 hr. Vials were read on either a BACTEC NR730 or BACTEC NR660 instrument. Arkansas Children's Hospital read the vials twice on days I and 2 and once daily thereafter for 7 days. Sparrow Hospital read the vials once on day 1, twice on days 2 and 3, and once daily there- after for 7 days. BACTEC PED cultures were consid- ered positive when the growth value (GV) was ~20, the GV (8 GV) increased by ~5 between consecutive readings, or there was visual evidence of growth. Gram-stained smears were examined and subcul- tures performed from the positive vials.

Processing of Iso 1.5

Iso 1.5 tubes were processed within 4 hr of collec- tion. They were vortexed several times, and -0.35 ml of the lysate was inoculated onto each plate. Plat- ing media consisted of 1 CDC (Centers for Disease Control) anaerobic blood agar plate and 1-3 choco- late agar plates, depending upon the volume of lys- ate. Chocolate agar plates were incubated for at least 4 days in 5%-10% CO2, and the CDC blood agar plate was incubated anaerobically for 6 days at 35 °- 37°C. Chocolate plates were examined for growth at approximately the same time as the reading of the companion BACTEC PED vial, for as long as the plates were held. The CDC plates were examined once a day on days 3-6. Iso 1.5 cultures were considered positive when ~1 colonies appeared on the inocu- lure and/or streak area. Colonies outside of this area were considered to be laboratory contaminants and were excluded from the study.

Identification of Isolates and Clinical Assessment

Isolates from the positive cultures were identified using standard methods and were classified as sig- nificant on the basis of a review of the patient's med- ical record and/or a consult with the patient's phy- sician. Criteria used to designate an isolate as significant included fever, leukocytosis, positive cul- ture with the same organism from another body site, and response to specific antimicrobial therapy. When a clinically significant organism was isolated, an in- quiry was made to determine whether the patient was receiving antibiotics at the time of blood collec- tion. A contaminant or clinically insignificant isolate was defined as an organism that was not considered to be responsible for the illness of the patient.

BACTEC PEDS Plus vs Du Pont Isolator 1.5 227

Statistical Analysis

Data was analyzed with the SPSS/PC+ software (version 3.0; SPSS, Chicago, IL). Tests to determine statistical significance were performed using the McNemar modification of the ×2 test (McNemar, 1962).

RESULTS

During the 11-month evaluation, 4063 blood culture sets were compared. There were 628 positive cul- tures yielding 675 isolates and, of these, 301 (45%) were clinically significant. The mean volume of blood cultured in the positive sets was 1.0 ml per system at Arkansas Children's Hospital and 0.8 ml at Spar- row Hospital.

Table I lists the clinically significant organisms by species or group and compares the rate of recovery with both systems. Of the organisms, 203 (67%) were isolated from both systems, 86 (29%) in the BACTEC PED only, and 12 (4%) in the ISO 1.5 only. Thus, the BACTEC PED system detects significantly more blood culture isolates (p K 0.001). Of the isolates, 51% were staphylococci, and significantly more staphylococci were isolated in the BACTEC PED then in the Iso 1.5 (p < 0.001). Ten of the 47 coagulase-positive staph- ylococci were not recovered from the Iso 1.5, whereas all of them were recovered in the BACTEC PED. Al- though the recovery of individual members of the Enterobacteriaceae was low in both systems, as a group the isolation rate in BACTEC PED was signifi- cantly higher than in ISO 1.5 (p < 0.005). Two My- cobacterium chelonae were isolated, both of which grew only in BACTEC PED.

The effect of antibiotic therapy on the rate of re- covery is shown in Table 2. Of the 57 isolates re- covered from patients receiving antibiotics, 34 (60%) were from both systems. Of the remaining 23 iso- lates, 21 (37%) were recovered only in BACTEC PED compared with two (4%) in ISO 1.5 only (p < 0.001). Of the 227 specimens collected from patients who were not on antibiotics, 157 (69%) isolates were re- covered from both systems, 61 (27%) from BACTEC

PED only, and nine (4%) from ISO 1.5 only (p < 0.001). Thus, the increased recovery rate in the BACTEC PED was no different in either the antimicrobial therapy group or the nontherapy group.

The average time interval between inoculation of media and detection of clinically significant organ- isms was 1.9 days for both systems and was the same for both study sites. Among the 203 isolates re- covered from both systems, 146 (72%) were detected at the same time, 28 (14%) earlier in BACTEC PED, and 29 (14%) earlier in ISO 1.5 (Table 3). Thus, the time to detection was comparable for both systems. The effect of antimicrobial therapy on detection times is also shown in Table 3. Of the 34 isolates from pa-

tients receiving antibiotics, 22 (65%) isolates were detected at the same time, seven (20%) earlier in BACTEC PED, and five (15%) earlier in ISO 1.5. The detection times for both systems were essentially the same regardless of the presence or absence of anti- microbial agents at the time of culture.

Colony counts from ~so 1.5 specimens yielding clinically significant isolates revealed that 36% con- tained ~ 10 colony-forming units per milliliter (CFU/ ml), 21% contained 11-100 CFU/ml, 12% contained 101-500 CFU/ml, 30% contained 501-1000 CFU/ml, and 1% contained > 1000 CFU/ml (median, 50 CFU/ ml) (Table 4). The staphylococci and Enterobacteri- aceae were present in all quantitation groups, whereas nine of 15 Streptococcus pneumoniae, five of 11 Hae- mophilus influenzae, five of eight yeast, and the one N. meningitidis grew in the ~10 CFU/ml range. It is noteworthy that 12 of 215 isolates grew in 1so 1.5 only and, of these, four S. pneumoniae, two Pseudo- monas aeruginosa, one Serratia marcescens, and one coagulase-negative staphylococcus grew in the 0-10 CFU/ml range.

Clinically insignificant isolates (contaminants) were recovered from 9.2% (374 of 4063) of all blood culture sets and from 55% (374 of 628) of the positive cul- tures. The majority of these isolates were skin bac- teria. With pediatric blood cultures, an exceptionally high rate of contamination with skin flora is a com- mon occurrence (Campos, 1989). The contamination rate was 4.8% for the BACTEC PED and 6.7% for the Iso 1.5. The ISO 1.5 contamination rate was the same at both test sites (6.6%, ACH; 6.7%, SH). Only the organisms recovered from the Iso 1.5 that were con- sidered potentially significant were included in this analysis. A higher number of contaminants, 179 (5.8%) of 3099, were isolated at Arkansas Children's Hos- pital with the BACTEC PED than at Sparrow Hospital, 17 (1.8%) of 964. This difference is unexplainable; at ACH, however, the BACTEC PED contamination rate is comparable to the average rate obtained with the BACTEC 6A medium.

DISCUSSION

Blood cultures play central and critical roles in the diagnosis of systemic infections in pediatric patients. Although there are no standards for the number of blood cultures to obtain and the volume of blood to culture, in practice, most pediatric patients have only one blood sample drawn with a volume ranging from 0.5 to 1.5 ml. Therefore, it is important to use a blood culture system that provides a sensitive method for detecting pediatric bacteremias.

Washington and Ilstrup (1986) and Kiehn (1988) have suggested that combinations of blood culture systems are necessary to optimize the recovery of

228 K. E i senach et al.

TABLE 1 C o m p a r i s o n of Cl in ica l ly S igni f ican t Isola tes f rom BACTEC PED a n d I so 1.5 Blood C u l t u r e Sys t ems

BACTEC Total No. PED BACTEC PED ISO 1.5

Microorganism Isolates + Iso 1.5 Only Only

Staphylococci 155 107 43 a 5 Coagulase positive 47 37 10 0 Coagulase negative 108 70 33 5

Streptococci 52 36 12 4 S. pneumoniae 17 11 2 4 Group A 3 3 0 0 Group B 3 3 0 0 Group D 9 6 3 0 Viridans group 16 9 7 0 Leuconostoc 4 4 0 0

Haemophilus influenzae 14 11 3 0

Neisseria meningitidis 1 1 0 0

Branhamella catarrhalis 1 0 1 0

Enterobacteriaceae 49 32 16 b 1 Klebsiella 17 13 4 0 Enterobacter 12 9 3 0 Escherichia 9 4 5 0 Serratia 7 3 3 1 Salmonella 3 2 1 0 Morganella 1 1 0 0

Pseudomonas spp. 9 3 4 2

Acinetobacter spp. 7 5 2 0

Candida spp. 10 8 2 0

Mycobacterium chelonae 2 0 2 0

Lactobacillus spp. 1 0 1 0

Total 301 203 86 c 12

" = p<0.001. b = p<0.005. c = p<0.001.

TABLE 2 Effect of An t ib io t i c T h e r a p y o n Recovery of Cl inical ly S igni f icant O r g a n i s m s f rom BACTEC PED a n d I so 1.5 Blood C u l t u r e Sys t ems

Total No. BACTEC PED BACTEC PED Therapy Isolates + ISO 1.5 Only

Iso 1.5

Only

ap < 0.001.

Yes 57 34 21 a 2 No 227 157 61 a 9 Unknown 17 12 4 1

Total 301 203 86 ~ 12

BACTEC PEDS Plus vs Du Pont Isolator 1.5 229

TABLE 3

Therapy

Effect of Antibiotic Therapy on Time to Detection of Clinically Significant Organisms from BACTEC PED and Iso 1.5 Blood Culture Systems

BACTEC Iso PED + BACTEC PED BACTEC PED 1,5 Iso 1.5 + iso 1.5 Earlier Earlier

Yes 34 22 7 5 No 157 114 21 22 Unknown 12 10 0 2

Total 203 146 28 29

bacteria and fungi from various patient populations. Specifically they have suggested that broth culture systems be supplementedwith a lysis-concentration system, such as Isolator. In many settings this ap- proach is not feasible due to the limited blood vol- ume, not practical from the standpoint of necessi- tating several different specimen processing procedures, and not economical.

Most evaluations of the Isolator have been with the large blood volume (7.5 ml and 10 ml) systems. In general, the Isolator has recovered more staph- ylococci, yeast, and fungi, and the broth systems have recovered more streptococci, especially S. pneu- moniae, and anaerobes. The ISO 1.5 differs from the Iso 7.5 and ~so 10 in that the lysate is plated directly on the media instead of concentrating the lysed cells and removing the supernatant before plating.

Studies with ISO 1.5 have demonstrated some- what different and variable trends. Welch et al. (1985) reported more staphylococci and yeast in ISO 1.5 than in a conventional broth culture system. In con- trast, Campos and Spainhour (1985a) reported a de- creased recovery of staphylococci and an increased recovery of streptococci with 1so 1.5 when compared with a two-bottle broth culture system. In a subse- quent study, they described a modified ISO 1.5 pro- cedure in which a portion of the lysate was inocu- lated into 10 ml of eugonic broth (Campos and Spainhour, 1985b). About 15% of the isolates were recovered in broth only. They advocate the inclusion of a broth culture to eliminate the time and expense of anaerobic agar media and to aid the interpretation of laboratory contaminants. When compared with the radiometric BACTEC 6B and 7C, an increase in S. pneumoniae and N. meningitidis and a decrease in H. influenzae were observed with ISO 1.5 (Carey, 1984). The distribution of blood among the systems eval- uated varied.

In this study we evaluated a new nonradiometric medium, BACTEC PEDS Plus, formulated for pediatric and other low volume blood specimens. The BACTEC

PED was compared with the Iso 1.5, with the blood

volume being equally distributed between the sys- tems. To optimize the recovery of organisms, es- pecially S. pneumoniae, from Iso 1.5, tubes were pro- cessed within 4 hr of collection. The BACTEC PED was strikingly more efficient than the ISO 1.5 in the over- all recovery of organisms (p < 0.001).

More than one-half of the clinically significant iso- lates were staphylococci, with the coagulase-nega- tive staphylococci being the most frequently isolated organism. Statistically, more staphylococci were re- covered from the BACTEC PED than ISO 1.5 (p < 0.001). Coagulase-negative staphylococci, particularly Staphylococcus epidermidis, are increasingly important causes of nosocomial infection. Of particular concern is the rising incidence of bacteremia due to coagu- lase-negative staphylococci in association with the use of in-dwelling central venous catheters. Most of the isolates in this study were recovered from high- risk patients in the neonatal and pediatric intensive care units and were associated with intravenous cen- tral line bacteremia.

The Enterobacteriaceae was the other group of organisms in which there was a higher recovery rate in the BACTEC PED (p < 0.005). The "classic" pedia- tric pathogens, S. agalactiae, S. pneumoniae, H. in- fluenzae, and N. meningitidis, were not recovered in sufficient numbers, individually or as a group, to determine significant differences between the two systems or to postulate isolation trends. Only a few Candida species were isolated in this study and no fungi, such as Malassezia furfur, which is of current interest in pediatrics, were recovered. More cul- tures need to be evaluated to determine how the two systems compare for the isolation rates of these organisms.

Anaerobic bacteremia is relatively rare in pedia- tric patients (Campos, 1989; Noel et al., 1988; Win- chester et al., 1977), and no anaerobes were re- covered from the CDC anaerobic blood agar plates in our study. The inclusion of an anaerobic system in pediatric settings on a routine basis is not war- ranted.

230 K. Eisenach et al.

TABLE 4 Correlation of Clinically Significant Isolates and Colony Counts Obtained in the Iso 1.5 Blood Culture System

No. CFU/ml

Microorganism 0-10 11-100 1 0 1 - 5 0 0 501-1000 > 1000

Staphylococci 37 29 11 33 Streptococci 18 5 4 13

S. pneumoniae 9 2 1 3 Haemophilus spp. 5 2 0 4 Neisseria meningitidis 1 0 0 0 Enterobacteriaceae 7 9 8 9 Other Gram 5 0 0 5

negatives Yeasts 5 0 2 1

Total 78 45 25 65

2 0

0 0 0 0 0

The presence of antimicrobials did not affect the overall recovery rate of organisms in either system. The BACTEC PED was superior to ISO 1.5 (p < 0.001) in the recovery of organisms regardless of whether or not the patients were receiving antibiotics. The use of BACTEC resin-containing media has been shown to increase the recovery of organisms from all pa- tients, not just those receiving antibiotics (Allen et al., 1989; Koontz et al., 1991; Morello et al., 1991).

In our study the advantage of BACTEC PED was greater recovery of both staphylococci and members of the Enterobacteriaceae. Similar results were ob- served by Morello et al. (1991) in their comparison of BACTEC PED and BACTEC 6A. Although it is unclear why these differences in systems were observed, the enhanced recovery in BACTEC PED was most likely due to the high dilution of blood (average blood- broth ratio was 1:20) and the presence of resins.

In general, resins have increased detection of staphylococci and have yielded highly variable re- covery rates with other organisms. The major action of resins has been the removal of antibiotics from blood specimens; however, recent studies indicate that resins have other effects. It is postulated that shaking the resins during the first 48 hr promotes mechanical lysis of the white blood cells, thereby releasing intracellular organisms and breaking up the clumps of organisms. Recently Jungkind et al. (1989) demonstrated an increase in white blood cell lysis with shaking the BACTEC NR16A resin medium. The extent of lysis was determined by measuring the amount of indium-I l l released into the medium from indium-Ill-labeled white cells. Koontz et al. (1991) noted that Gram staining of the BACTEC Plus resin-containing medium at 4--6 hr demonstrated to- tal lysis of both red and white blood cells. The lysing and dispersing action of resins is a plausible expla-

nation for the increased recovery of staphylococci and those Gram-negative rods that tend to conglu- tinate. Other possible mechanisms of action of resins have yet to be elucidated.

A major advantage of the Iso systems is early detection of bacteremia, providing that multiple ob- servations of the plated media are made during the first 24-48 hr. In other studies with ISO 1.5, the de- tection of organisms has been a few hours faster when compared to broth culture; however, the over- all time difference has not been significant. Thus it was not too surprising that BACTEC PED was as rapid as ISO 1.5 in detecting bacteremia.

Santosham and Moxon (1977) have reported a wide variation in colony counts in blood of bacteremic children. Others have indicated that the concentra- tion of organisms is occasionally very low, not only in pediatric bacteremias but also in neonatal sepsis (Campos et al., 1985a; Dietzman et al., 1974; Szym- czak et al., 1979). In our study, 36% of the ISO 1.5 cultures yielded clinically significant organisms with colony counts of ~<10 CFU/ml, which is slightly higher than the 27% reported by Campos. Szymczak et al. (1979) advocate using a conventional two-bottle sys- tem with 1 ml per bottle and state that pediatric bacteremias will be missed if <1 ml of blood is in- oculated. They reported the concentration of organ- isms in blood was as low as 0.5 CFU/ml with a mean of 51 CFU/ml.

These observations reemphasize the necessity of using a sensitive system that can detect organisms in small blood volumes. It seems prudent if ~3 ml of blood is obtained, the entire sample should be inoculated into the BACTEC PED vial. It is important to encourage the physicians to collect as much blood (5-10 ml) as possible from children 2 years and older. If >3 ml of blood is drawn, the volume could be

BACTEC PEDS Plus vs Du Pont Isolator 1.5 231

divided be tween two vials, and the choice of the second m e d i u m wou ld be d e p e n d e n t on the clinical setting. Ano the r al ternat ive wou ld be to inoculate the entire sample into the aerobic, h igh -vo lume me- d ium wi th resins (BACTEC PLUS 26A).

In conclusion, w h e n cul tur ing pediatr ic bactere- mias, the BACTEC PEDS Plus is the m e d i u m of choice for those us ing the BACTEC nonrad iomet r ic system. It provides a rapid and sensit ive m e t h o d for detect- ing bacteremia in small vo l um es of blood.

We are grateful to the phlebotomy and laboratory staff at both institutions for their cooperation and extra efforts, thus mak- ing the study possible. This study was supported in part by a grant from Becton Dickinson Diagnostic Instrument Systems, Sparks, MD, and was presented in part at the 29th lntersci- ence Conference on AntimicrobiaI Agents and Chemotherapy, Houston, Texas, 1989. We also thank Ann Graziosi of BDDIS for her many helpful suggestions and expert assistance throughout the course of the study.

REFERENCES

Allen SD, Place DA, Reynolds JK (1989) Assessing the relevance of positive BACTEC resin blood cultures [abstr 862]. In Program and Abstracts of the 29th Interscience Con- ference on Antimicrobial Agents and Chemotherapy. Wash- ington, DC: American Society for Microbiology, p 245.

Campos JM (1989) Detection of bloodstream infections in children. Eur J Clin Microbiol Infect Dis 8:815.

Campos JM, Spainhour JR (1985a) Comparison of the Iso- lator 1.5 microbial tube with a conventional blood cul- ture broth system for detection of bacteremia in chil- dren. Diagn Microbiol Infect Dis 3:167.

Campos JM, Spainhour JR (1985b) Rapid detection of bac- teremia in children with a modified lysis direct plating method. J Clin Microbiol 22:674.

Carey RB (1984) Clinical comparison of the Isolator 1.5 microbial tube and the BACTEC radiometric system for detection of bacteremia in children. J Clin Microbio119:634.

Dietzman DE, Fischer GW, Schoenknecht FD (1974) Neo- natal Escherichia coli septicemia-bacterial counts in blood. J Pediatr 85:128.

Durbin WA, Szymczak EG, Goldman DA (1978) Quanti- tative blood cultures in childhood bacteremia. J Pediatr 92:778.

Jungkind DL, Thakur M, Dyke J (1989) Evidence for a second mechanism of action of resin in BACTEC NR 16A aerobic blood culture medium [abst C-225]. In Ab- stracts of the 89th Annual Meeting of the American Society for Microbiology. Washington, DC: American Society for Microbiology, p 431.

Kiehn TE (ed) (1988) Isolator versus broth for blood cul- tures. In Clinical Microbiology Newsletter. Eds, MJ Fer- raro, PA Granato, JA Morello, and RJ Zabransky. New York: Elsevier, pp 53-54.

Koontz FP, Flint JK, Reynolds JK, Allen SD (1991) Mul- ticenter comparison of the high volume (10 ml) NR BAC- TEC Plus system and the standard (5 ml) NR BACTEC sys- tem. Diagn Microbiol Infect Dis 14:111.

McNemar Q, ed. (1962) Psychological Statistics, 3rd ed. New York: John Wiley and Sons, pp 209-239.

Morello JA, Matushek SM, Dunne WM, Hinds DB (1991) Performance of a BACTEC nonradiometric medium for pediatric blood cultures. J Clin Microbiol 29:359.

Neal PR, Kleiman MB, Reynolds JK, Allen SD, Lemons JA, Yu PA (1986) Volume of blood submitted for culture from neonates. J Clin Microbiol 24:353.

Noel GT, Laufer DA, Edelson PJ (1988) Anaerobic bacte- remia in a neonatal intensive care unit: an eighteen- year experience. Pediatr Infect Dis J 7:858.

Santosham M, Moxon ER (1977) Detection and quantita- tion of bacteremia in childhood. J Pediatr 91:719.

Sullivan TD, LaScolea LJ, Neter E (1982) Relationship be- tween the magnitude of bacteremia in children and the clinical disease. Pediatrics 69:699.

Szymczak E G, Barr JT, Durbin WA, Goldman DA (1979) Evaluation of blood culture procedures in a pediatric hospital. J Clin Microbiol 9:88.

Washington JA, Ilustrup DM (1986) Blood cultures: issues and controversies. Rev Infect Dis 8:792.

Welch DF, Scribner RK, Hensel D (1985) Evaluation of a lysis direct plating method for pediatric blood cultures. J Clin Microbiol 21:955.

Winchester PD, Todd JK, Roe MH (1977) Bacteremia in hospitalized children. Am J Dis Child 131:753.