BACKGROUND:

The rapid identification of organisms reported in positive blood cultures via polymerase chain reaction (PCR) can accurately identify a nonpathogenic bacterium and decrease time to definitive identification, as compared with traditional microbiologic methods. How this technology effects clinical and antimicrobial management in children with nonpathogenic bacteria identified in a blood culture without decision support has not been evaluated.

METHODS:

A retrospective study of the management of children with positive blood culture results for nonpathogenic organisms before and after implementation of PCR technology. Each cohort’s antibiotic management, frequency of repeat cultures, and return visits to an emergency department (ED) were compared.

RESULTS:

A total 136 patients during this time (49% [n = 67] pre-PCR and 51% [n = 69] post-PCR) had a blood culture positive for nonpathogenic bacterium. Admitted patients had a second specimen sent for testing on fewer occasions (P = .04); however, total antibiotic exposure did not differ significantly (P = .3) after introduction of PCR technology. There was no significant difference in length of stay postintervention (P = .12). Patients discharged directly from the ED had fewer return visits (P = .02) and received fewer repeat blood cultures (P = .04), and antibiotics were administered on fewer occasions after return (P = .04) postintroduction of PCR technology.

CONCLUSIONS:

With the addition of PCR technology, patients with blood cultures positive for nonpathogenic bacteria received less antibiotics, fewer repeat blood cultures, and fewer repeat ED evaluations.

A blood culture is frequently obtained in febrile children to identify those children with bacteremia and guide definitive antimicrobial treatment when a bacterial pathogen is identified. Most providers would consider a blood culture to have a negative result when there is no growth after 48 hours.1  A dilemma may arise, in regard to management of a patient, when a positive blood culture result is reported before final identification. Bacteria colonizing the skin may inadvertently be introduced to the blood culture bottle during collection and lead to contamination.2  Whether an identified organism is deemed a true pathogen or a contaminant is usually adjudicated by a clinician on the basis of the organism, clinical and laboratory data, and presence or absence of an indwelling catheter. The Centers for Disease Control and Prevention and the National Healthcare Safety Network have recently published guidelines to help the clinician classify bacteria identified in a blood culture as a recognized pathogen or a contaminant.3  For example, coagulase-negative staphylococci would generally be considered a nonpathogen in most immunocompetent pediatric patients without an indwelling catheter. In reports, researchers suggest coagulase-negative staphylococci identified in positive blood culture results may eventually represent nonpathogenic bacteria in up to 90% of cases.4 

With improvements in polymerase chain reaction (PCR) technology, rapid blood culture identification (BCID) has been introduced to the microbiology laboratory’s armamentarium to more expeditiously identify organisms found in a positive blood culture result. Conventional methods for identification of bacteria from a positive blood culture result can take ∼24 to 48 hours. The time to definitive bacterial identification can be reduced to ∼1 to 3 hours after the recognition of a positive blood culture result with BCID technology. This has led to improved antibiotic stewardship in adult patients.57  For example, without rapid BCID, a positive blood culture result, which is eventually determined to be a contaminant, may have led a clinician in the interim to order a repeat blood culture and an additional 24 hours of antibiotics until the availability of a final identification.

BCID via PCR can also accurately identify a subset of pathogens in the blood and decrease time to identification, as compared with traditional blood culture microbiologic techniques, in the pediatric population.8  Although studies in adult patients have revealed improved antibiotic stewardship, the clinical effects of BCID in the management in pediatric patients is limited, specifically in regard to how the decreased time to contaminant identification that is afforded with BCID technology would affect a clinician’s decision-making process. Messacar et al9  reported an overall decreased median time to optimal antimicrobial therapy with the use of BCID technology when it was coupled with real-time antimicrobial decision support. They also specifically reported a decrease in unnecessary antibiotic initiation for children when contaminants were identified in their blood cultures. However this study’s design did not allow the differentiation of the clinical impact of the rapid BCID panel alone versus its use when bundled with real-time antimicrobial stewardship decision support. The objective of this retrospective study was to (1) assess the effect of the implementation of rapid BCID on antimicrobial decision-making by hospitalists who cared for patients on an inpatient unit when a contaminant was identified in a blood culture and to (2) assess the effect of the implementation of BCID technology on the clinical decision-making and antimicrobial use of emergency medicine physicians when children treated in the emergency department (ED) are identified as having a blood culture contaminant.

This was a single-center, retrospective study that was undertaken at a freestanding children’s hospital. All study participants were children between the ages of 2 months and 18 years. Each of the participants had a blood culture that was positive for a nonpathogenic bacteria and were treated at the children’s hospital in 1 of 2 defined periods of time. The time periods included a 1-year period before implementation of the BCID PCR technology (April 2016 to March 2017) and a 1-year period post-PCR implementation (April 2017 to May 2018). The microbiology laboratory added BCID technology to its blood culture identification process in April 2017. At the time of its introduction, there was no formal educational program for providers regarding the technology. Clinicians received notification of a positive blood culture result through direct communication from the microbiology laboratory and a simultaneous report in the electronic medical record. BCID results were communicated only through the electronic medical record.

The investigation included both patients treated as inpatients and those discharged from the ED. Patients were excluded from the analysis if they had indwelling catheters, had congenital heart disease, or were immunocompromised. The organism identified in a positive culture result was considered to be a contaminant if it was a typical commensal organism not associated with the clinical or laboratory features of clinically significant bacteremia.3,10 

The FilmArray BCID panel (bioMerieux, Marcy-l’Étoile, France), an automated PCR technology to identify nucleic acid sequences of 24 bacteria, was used to analyze all positive blood culture results post implementation.11 

Clinical, laboratory, and sociodemographic data were abstracted from the electronic health record and managed with the use of a Research Electronic Data Capture electronic database. Collected demographic data included age and sex. Clinical data collected included admitting diagnosis, surgical history, and past medical history. Relevant laboratory data included blood culture results and/or BCID results and time to a reported positive result. Time to BCID result was not captured in the electronic medical record but, per hospital protocol, is 1 to 3 hours after the recognition of a positive culture result.

For those admitted to the pediatric inpatient unit, data collected included use and type of antibiotic, whether antibiotics continued to be administered after report of a positive blood culture result, duration of antibiotic treatment, frequency of repeat blood cultures, and length of stay (measured from time of admission to time of discharge). Duration of antibiotic use was calculated by adding the time from the administration of the first dose to the time of administration of the last dose of each antibiotic. This would allow comparison of antimicrobial regimes with different dosing intervals.9  For example, if a child received 24 hours of ceftriaxone and 16 hours of vancomycin, the total duration of antibiotic hours would be considered 40 hours. For those who were discharged directly from the ED, data collected included frequency of return visits, number of repeat blood cultures, and additional antibiotic administration. The primary outcome measure for inpatients was duration of antibiotics. Secondary outcomes included the frequency of repeat blood cultures, persistence of antibiotics after the initial positive result, and length of stay. The primary outcome measure for patients initially discharged from the ED was frequency of return visits. Secondary outcomes included the number of repeat blood cultures and rate of additional antibiotics. The study was approved by our medical center’s institutional review board.

Patients who were admitted to the hospital and those who were discharged from the ED were evaluated separately. Descriptive statistics were reported for each explanatory variable by pre-BCID and post-BCID. Continuous variables were reported as medians and interquartile ranges (IQRs) (25th to 75th percentile). Categorical variables are reported as frequencies and percentages. The χ2 test was used to test for association between pre-BCID and post-BCID and categorical variables. The Wilcoxon rank test was used to test for association between pre-BCID and post-BCID and continuous variables. All tests were considered significant if P < .05. Because this study was exploratory in nature, no adjustments for multiple testing were made. All analyses were conducted by using SAS version 9.4 (SAS Institute, Inc, Cary, NC).

A total of 136 patients met criteria for selection for analysis. During the defined period, 49% (n = 67) were from the pre-BCID implementation and 51% (n = 69) were from the post-BCID implementation. All these patients had positive blood culture results for ≥1 bacteria that were considered nonpathogenic. The final identification of organisms via traditional methods and via BCID were concordant in all instances. Of the 136 patients, 61% (83 of 136) were admitted to the hospital, and 38% (53 of 136) were discharged from the ED. A similar proportion of patients was admitted to hospital during each time frame (49% pre-BCID and 51% post-BCID).

Of the 67 patients who presented to the ED during the pre-BCID implementation, 41 were admitted to the pediatric inpatient unit, and of the 69 who presented to the ED during the post-BCID implementation, 42 were admitted to the pediatric inpatient unit. The median time to reach positive blood culture result was 19 hours (IQR 20–26) for the pre-BCID group and 22 hours (IQR 18–25) for the post-PCR group (P = not significant). The most common nonpathogenic bacteria identified was Staphylococcus epidermidis (67%, n = 56). The final identification of other organisms that were identified as nonpathogens are listed in Table 1. Two patients in the post- BCID period had blood cultures positive for both S epidermidis and Streptococcus viridans (Table 1).

TABLE 1

Patient Characteristics and Clinical Variables for Patients Pre-BCID and Post-BCID Panel PCR Technology

VariablesBlood Culture Era, n = 67BCID PCR Era, n = 69
Age, median (IQR), y 2 (0.5–7) 1 (0.17–6) 
Girls, n (%) 19 (46) 16 (38) 
Time to positivity, median (IQR), h 22 (20–26) 22 (18–25) 
Discharge diagnosis, n (%)   
 Upper respiratory infection 11 (27) 9 (21) 
 Fever 3 (7) 3 (7) 
 UTI 6 (12) 2 (5) 
 Pneumonia 5 (8) 5 (12) 
 Dehydration 2 (3) 9 (21) 
 Other 14 (34) 14 (33) 
Pathogen, n (%)   
S epidermidis 29 (71) 26 (62) 
Streptococcus mitis 2 (5) 1 (2) 
S viridans 2 (5) 8 (19) 
 Micrococcus luteus 2 (5) 2 (5) 
Staphylococcus hominis 3 (7) 0 (0) 
 Other 3 (7) 5 (12) 
VariablesBlood Culture Era, n = 67BCID PCR Era, n = 69
Age, median (IQR), y 2 (0.5–7) 1 (0.17–6) 
Girls, n (%) 19 (46) 16 (38) 
Time to positivity, median (IQR), h 22 (20–26) 22 (18–25) 
Discharge diagnosis, n (%)   
 Upper respiratory infection 11 (27) 9 (21) 
 Fever 3 (7) 3 (7) 
 UTI 6 (12) 2 (5) 
 Pneumonia 5 (8) 5 (12) 
 Dehydration 2 (3) 9 (21) 
 Other 14 (34) 14 (33) 
Pathogen, n (%)   
S epidermidis 29 (71) 26 (62) 
Streptococcus mitis 2 (5) 1 (2) 
S viridans 2 (5) 8 (19) 
 Micrococcus luteus 2 (5) 2 (5) 
Staphylococcus hominis 3 (7) 0 (0) 
 Other 3 (7) 5 (12) 

UTI, urinary tract infection.

Before the introduction of the BCID, 56% of the admitted patients (23 of 41) received antibiotics, whereas post-BCID 31% of patients (13 of 42) received antibiotics (P = .021). Pre-BCID, 23 patients received antibiotics at some point during their stay, although 52% (12 of 23) were treated after the clinician received a positive Gram-stain result. Post-BCID, 13 patients received antibiotics during their stay, but only 15% of the positive Gram-stain results continued to receive treatment (P = .03). The median duration of antibiotic exposure by the pre-BCID group was 48 hours (IQR 24–64) compared to the post-BCID group (32 hours, IQR 24–48; P = .3). The proportion of patients who had a second blood culture specimen sent was significantly lower in the post-BCID group compared to that of the pre-PCID group (60% vs 80%, respectively; P = .04). All repeat blood culture specimens had negative culture results. There was not a significant difference in length of stay after BCID implementation (P = .12) (Table 2).

TABLE 2

Admitted (Inpatient) Patients Pre-BCID and Post-BCID Panel PCR Technology

VariablesBlood Culture Era, n = 41BCID PCR Era, n = 42P
Received antibiotics, n (%) 23 (56) 13 (31) .02 
Antibiotics postresult, n (%) 12 (52) 2 (15) .03 
Cumulative h of antibiotics, median (IQR) 48 (24–64) 32 (24–48) .3 
Repeat blood culture, n (%) 33 (80) 25 (60) .04 
Length of stay, median (IQR), h 78 (42–142) 65.5 (44–96) .12 
VariablesBlood Culture Era, n = 41BCID PCR Era, n = 42P
Received antibiotics, n (%) 23 (56) 13 (31) .02 
Antibiotics postresult, n (%) 12 (52) 2 (15) .03 
Cumulative h of antibiotics, median (IQR) 48 (24–64) 32 (24–48) .3 
Repeat blood culture, n (%) 33 (80) 25 (60) .04 
Length of stay, median (IQR), h 78 (42–142) 65.5 (44–96) .12 

Of the 67 patients who presented to the ED during the pre-BCID implementation, 26 were discharged from the ED, and of the 69 who presented to the ED during the post-BCID implementation, 27 were discharged from the ED. Time to a positive blood culture results, discharge diagnosis, and types of identified organisms were similar to the inpatient population (Table 3). Both pre-BCID and post-BCID groups received antibiotics at their initial visit to the ED at a similar frequency (P = .09). All patients pre-BCID and 96% of patients post-BCID received a phone call from a midlevel provider in regard to the positive result. The proportion of patients who were asked to return to the ED was significantly lower in the post-BCID group compared to the pre-BCID group (59% vs 88%, respectively; P = .02). The proportion of patients who had a second blood culture specimen sent after the initial positive culture result was significantly lower in the post-BCID group compared to the pre-BCID group (56% vs 85%, respectively; P = .02). All repeat specimens had negative culture results. For those patients who did return to the ED, the post-BCID group was less likely to receive antibiotics when compared to the pre-BCID group on their return visit (31% vs 61%, respectively; P = .04) (Table 4).

TABLE 3

Patient Characteristics and Clinical Variables for Patients Pre-BCID and Post-BCID Panel PCR Technology for Discharged Patients

VariablesBlood Culture Era, n = 67BCID PCR Era, n = 42
Age, median (IQR), y 0.9 (0.25–4) 1 (0.25–5) 
Girls, n (%) 11 (42) 12 (44) 
Time to positivity, median (IQR), h 22 (18–23) 24 (19–41) 
Discharge diagnosis, n (%)   
 Upper respiratory infection 6 (23) 8 (30) 
 Fever 12 (46) 9 (33) 
 UTI 2 (8) 3 (11) 
 Pneumonia 0 (0) 1 (4) 
 Dehydration 0 (0) 0 (0) 
 Other 6 (23) 6 (22) 
Pathogen, n (%)   
S epidermidis 20 (77) 21 (78) 
S mitis 2 (8) 0 (0) 
S viridans 1 (4) 1 (4) 
Micrococcus luteus 1 (4) 1 (4) 
S hominis 0 (0) 0 (0) 
 Other 2 (8) 4 (15) 
VariablesBlood Culture Era, n = 67BCID PCR Era, n = 42
Age, median (IQR), y 0.9 (0.25–4) 1 (0.25–5) 
Girls, n (%) 11 (42) 12 (44) 
Time to positivity, median (IQR), h 22 (18–23) 24 (19–41) 
Discharge diagnosis, n (%)   
 Upper respiratory infection 6 (23) 8 (30) 
 Fever 12 (46) 9 (33) 
 UTI 2 (8) 3 (11) 
 Pneumonia 0 (0) 1 (4) 
 Dehydration 0 (0) 0 (0) 
 Other 6 (23) 6 (22) 
Pathogen, n (%)   
S epidermidis 20 (77) 21 (78) 
S mitis 2 (8) 0 (0) 
S viridans 1 (4) 1 (4) 
Micrococcus luteus 1 (4) 1 (4) 
S hominis 0 (0) 0 (0) 
 Other 2 (8) 4 (15) 

UTI, urinary tract infection.

TABLE 4

Discharged (ED) Patients Pre-BCID and Post-BCID Panel PCR Technology

VariablesBlood Culture Era (n = 26), n (%)BCID PCR Era (n = 27), n (%)P
Received call 26 (100) 26 (96) .98 
Returned visit 23 (88) 16 (59) .02 
Repeat blood culture 22 (85) 15 (56) .02 
Received antibiotics (initial visit) 14 (54) 8 (30) .09 
Received antibiotics (return visit) 14 (61) 5 (31) .04 
VariablesBlood Culture Era (n = 26), n (%)BCID PCR Era (n = 27), n (%)P
Received call 26 (100) 26 (96) .98 
Returned visit 23 (88) 16 (59) .02 
Repeat blood culture 22 (85) 15 (56) .02 
Received antibiotics (initial visit) 14 (54) 8 (30) .09 
Received antibiotics (return visit) 14 (61) 5 (31) .04 

Rapid PCR identification of typically nonpathogenic bacteria in pediatric patients with positive blood culture results led to less health care usage. For those children who were admitted to the hospital, antibiotics were initiated on fewer occasions, and fewer repeat blood cultures were performed. Although the duration of total antibiotic exposure between the 2 groups did not reach statistical significance, there was trend toward a shorter duration from a median of 48 to 32 hours. This lack of significance could be secondary to the study’s modest sample size. Alternatively, the lack of a robust decision support service, as described by Messacar et al,9  may have led hospitalists to delay their decision-making for a couple of hours while they sort further direction. A formal educational program focused on the utility of this new technology in the management of antibiotics should be considered as part of a pediatric hospitalist’s faculty development.12 

Length of stay was not significantly decreased for patients in the BCID PCR era, as compared with the earlier group. We hypothesize that although the length of stay for some patients may be affected by blood culture results, it is more likely that factors not related to bacteremia concerns may have affected length of stay in this patient subset. This is consistent with previous studies in both adult and pediatric populations in which researchers have reported mixed results in regard to the impact of BCID PCR length of stay.13,14  It is also possible, as mentioned above, this may be related to the individual hospitalists’ understanding of BCID.

For patients who were discharged from the ED, the overall rate of initial antibiotic administration was similar between the 2 groups. This was expected because the clinical practice of prescribing empirical antibiotics pending the results of a blood culture had not changed over this time period. However, patients were less likely to be asked to return to the hospital for reevaluation after the report of a positive result post BCID PCR implementation. Presumably, this was because the clinician was simultaneously aware that the organism identified was not typically a pathogen. Furthermore, even for those patients who were asked to return to the ED for reevaluation, they were less likely to receive antibiotics on their return visit. Again, presumably, this was because at the time of the reevaluation, the organism was identified by the clinician as nonpathogenic. Before the introduction of BCID technology, an ED physician might have felt compelled to repeat a blood culture and treat those patients with an initial positive blood culture result while they awaited definitive identification.

Nevertheless, although significantly fewer patients (88% vs 59%) were recalled to the ED after the introduction of BCID PCR technology, more than half were still asked to return for evaluation. This most likely was due to clinician variable practice patterns in the management of initial positive blood culture results in the ED. The precision of this novel technology was likely not fully understood by all ED staff.

The finding of 100% concordance with the BCID technology in our laboratory with traditional methods should provide reassurance to the clinician. This confirms previously published data regarding the accuracy of the test.5  Further education of ED staff will be necessary to modify the traditional reflexive process of an automatic revisit for a positive blood culture result when identification of the organism is now rapidly available.

There have been several studies in adult patients in which researchers measured the clinical impact of BCID. They have reported results similar to what was found in this investigation. Specifically, BCID has been established to be as accurate as traditional blood culture analysis in both pediatric and adult patients5  and has also been demonstrated to have improved clinical management of adult patients with blood cultures positive for pathogenic organisms.6,7  Furthermore, rapid identification of blood pathogens in adults has been shown to decrease health care costs15  and, in some studies, has revealed improved antibiotic stewardship.14,15 

There are several limitations to our study. As a single-centered study, these data may not be applicable to other centers. In addition, although not investigated, there were probably differences in how individual clinicians interpreted the BCID results. Lastly, the precise time the hospitalist was notified of the BCID results could not be captured but inferred from hospital protocols.

The use of a rapid BCID PCR identification test for pediatric patients with positive blood culture results allows for the identification of nonpathogenic organisms in a timelier manner. This led to a decrease in health care use and antibiotic use. Although speculative, this decrease would probably be further reduced by additional education of clinicians as to the utility of BCID.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. We thank Jane Cerise for her assistance with the statistical analysis.

Dr Hughes conceptualized and designed the study, drafted the initial manuscript, and conducted the analyses; Dr Barone conceptualized the study and reviewed and revised the manuscript; and both authors approved the final manuscript as submitted.

FUNDING: No external funding.

1
Pardo
J
,
Klinker
KP
,
Borgert
SJ
,
Trikha
G
,
Rand
KH
,
Ramphal
R
.
Time to positivity of blood cultures supports antibiotic de-escalation at 48 hours
.
Ann Pharmacother
.
2014
;
48
(
1
):
33
40
2
Hall
KK
,
Lyman
JA
.
Updated review of blood culture contamination
.
Clin Microbiol Rev
.
2006
;
19
(
4
):
788
802
3
Centers for Disease Control and Prevention; National Health Safety Network
.
Definitions for specific types of infections. 2020. Available at: https://www.cdc.gov/nhsn/pdfs/pscmanual/17pscnosinfdef_current.pdf. Accessed November 30, 2020
4
Becker
K
,
Heilmann
C
,
Peters
G
.
Coagulase-negative staphylococci
.
Clin Microbiol Rev
.
2014
;
27
(
4
):
870
926
5
Banerjee
R
,
Teng
CB
,
Cunningham
SA
, et al
.
Randomized trial of rapid multiplex polymerase chain reaction-based blood culture identification and susceptibility testing
.
Clin Infect Dis
.
2015
;
61
(
7
):
1071
1080
6
Kang
CM
,
Chen
XJ
,
Chih
CC
, et al
.
Rapid identification of bloodstream bacterial and fungal pathogens and their antibiotic resistance determinants from positively flagged blood cultures using the BioFire FilmArray blood culture identification panel
.
J Microbiol Immunol Infect
.
2020
;
53
(
6
):
882
891
7
MacVane
SH
,
Nolte
FS
.
Benefits of adding a rapid PCR-based blood culture identification panel to an established antimicrobial stewardship program
.
J Clin Microbiol
.
2016
;
54
(
10
):
2455
2463
8
Zheng
X
,
Polanco
W
,
Carter
D
,
Shulman
S
.
Rapid identification of pathogens from pediatric blood cultures by use of the FilmArray blood culture identification panel
.
J Clin Microbiol
.
2014
;
52
(
12
):
4368
4371
9
Messacar
K
,
Hurst
AL
,
Child
J
, et al
.
Clinical impact and provider acceptability of real-time antimicrobial stewardship decision support for rapid diagnostics in children with positive blood culture results
.
J Pediatric Infect Dis Soc
.
2016
;
6
(
3
):
267
274
10
Weddle
G
,
Jackson
MA
,
Selvarangan
R
.
Reducing blood culture contamination in a pediatric emergency department
.
Pediatr Emerg Care
.
2011
;
27
(
3
):
179
181
11
Blaschke
AJ
,
Heyrend
C
,
Byington
CL
, et al
.
Rapid identification of pathogens from positive blood cultures by multiplex polymerase chain reaction using the FilmArray system
.
Diagn Microbiol Infect Dis
.
2012
;
74
(
4
):
349
355
12
Donner
LM
,
Campbell
WS
,
Lyden
E
,
Van Schooneveld
TC
.
Assessment of rapid-blood-culture-identification result interpretation and antibiotic prescribing practices
.
J Clin Microbiol
.
2017
;
55
(
5
):
1496
1507
13
Tseng
AS
,
Kasule
SN
,
Rice
F
, et al
.
Is it actionable? An evaluation of the rapid PCR-based blood culture identification panel on the management of gram-positive and gram-negative blood stream infections
.
Open Forum Infect Dis
.
2018
;
5
(
12
):
ofy308
14
Ray
ST
,
Drew
RJ
,
Hardiman
F
,
Pizer
B
,
Riordan
A
.
Rapid identification of microorganisms by FilmArray blood culture identification panel improves clinical management in children
.
Pediatr Infect Dis J
.
2016
;
35
(
5
):
e134
e138
15
Pardo
J
,
Klinker
KP
,
Borgert
SJ
,
Butler
BM
,
Giglio
PG
,
Rand
KH
.
Clinical and economic impact of antimicrobial stewardship interventions with the FilmArray blood culture identification panel
.
Diagn Microbiol Infect Dis
.
2016
;
84
(
2
):
159
164

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.