Original Studies

Bloodstream Infections in Hospitalized Children

Epidemiology and Antimicrobial Susceptibilities
Beatriz Larru, MD, PhD,* Wu Gong, MS, MPH,† Neika Vendetti, MPH,* Kaede V. Sullivan, MD,‡
Russell Localio, PhD,§¶ Theoklis E. Zaoutis, MD, MSCE,*¶ and Jeffrey S. Gerber, MD, PhD*¶

Background: Bloodstream infection is a major cause of morbidity and

mortality. Much of our understanding of the epidemiology and resistance
B loodstream infection is one of the most common serious bacte-
rial infections in hospitalized children and a major cause of
morbidity and mortality.1,2 Understanding the epidemiology and the
patterns of bloodstream infections comes from studies of hospitalized
adults.
drug resistance patterns of bloodstream infections in hospitalized
Methods: We evaluated the epidemiology and antimicrobial resistance of
children is critical for appropriately directing empiric antimicro-
bloodstream infections occurring during an 11-year period in a large, ter-
bial prescribing and to guide antimicrobial-resistance containment
tiary care children’s hospital in the US. All positive blood cultures were
strategies.3–5 Such data are scarce, however, and the majority of
identified retrospectively from clinical microbiology laboratory records. We
recent pediatric studies have focused on specific subpopulations or
excluded repeat positive cultures with the same organism from the same
specific microorganisms, limiting their generalizability. Therefore,
patient within 30 days and polymicrobial infections.
we sought to determine the epidemiology and antimicrobial resist-
Results: We identified 8196 unique episodes of monomicrobial bacte-
ance patterns of bloodstream infections occurring over a decade in
remia in 5508 patients. Overall, 46% were community onset, 72% were
a large, tertiary care children’s hospital in the US.
Gram-positive bacteria, 22% Gram-negative bacteria and 5% Candida
spp. Coagulase negative Staphylococcus was the most common isolated METHODS
organism. ESKAPE pathogens (Enterococcus faecium, Staphylococcus
aureus, ­Klebsiella ­pneumoniae, Acinetobacter baumannii, Pseudomonas Study Population
aeruginosa and Enterobacter spp.) accounted for 20% of episodes. No This retrospective study was conducted at a 520-bed aca-
S. aureus isolate was resistant to vancomycin or linezolid, and no increase demic tertiary care children’s hospital with more than 26,000
in vancomycin minimum inhibitory concentration among methicillin- admissions per year. We included all positive blood cultures
resistant S. aureus was observed during the study period. Clinically signifi- obtained from children ≤19 years old between 1 January 2002 and
cant increases in vancomycin-resistant E ­ nterococcus, ceftazidime-resist- 31 December 2012, identified using clinical microbiology labora-
ant P. aeruginosa or carbapenem-resistant Enterobacteriaceae were not tory records. We excluded repeat positive cultures with the same
observed during the study period; however, rates of methicillin-resistant organism from the same patient within 30 days of the index culture
S. aureus increased over time (P < 0.01). and polymicrobial blood cultures.
Conclusions: Gram-positive and ESKAPE organisms are leading
causes of bacteremia in hospitalized children. Although antimicrobial Microbiology
resistance patterns were favorable compared with prior reports of hos- Venous blood was inoculated into BacT/ALERT Pediatric
pitalized adults, multicenter studies with continuous surveillance are FAN blood culture bottles (bioMérieux, Durham, NC). Bottles were
needed to identify trends in the emergence of antimicrobial resistance incubated in a BacT/ALERT 3D continuous monitoring blood culture
in this setting. instrument for a maximum of 5 days. Positive blood culture bottles
were subcultured to routine media, and a Gram stain was performed.
Key Words: antimicrobial, resistance, bacteremia Colonial morphology, routine biochemical tests and/or the VITEK 2
(Pediatr Infect Dis J 2016;35:507–510) instrument (bioMérieux) were used for organism identification. Anti-
biotic susceptibility testing was performed by VITEK 2 for Staphy-
lococcus spp., Enterococcus spp., riaceae and most nonglucose
fermenting organisms (eg, Pseudomonas aeruginosa); by E-Test
Accepted for publication November 18, 2015. (bioMérieux) for Streptococcus spp.; and by the YeastOne susceptibil-
From the *Division of Infectious Diseases, Center of Pediatric Clinical Effective- ity microtiter system (TREK Diagnostic Systems, Cleveland, OH) for
ness, †Health Care Analytics Unit, Center of Pediatric Clinical Effectiveness, Candida spp. Minimum inhibitory concentrations were interpreted
‡Pathology & Laboratory Medicine, The Children’s Hospital of Philadel- using Clinical and Laboratory Standards Institute (CLSI) criteria.
phia, Pennsylvania; §Department of Biostatistics and Epidemiology, and
¶Center for Clinical and Epidemiology and Biostatistics, Perelman School of
Medicine, University of Pennsylvania, Philadelphia, Pennsylvania. Definitions
This study was accepted as an abstract for the ID week Conference: Advanc- An episode of monomicrobial bacteremia was defined as iso-
ing Science, Improving Care (October 17–22, San Diego) and received The lation of a single bacterial and/or fungal pathogen in a blood culture
Pediatric Fellow Poster Award by The Pediatric Infectious Diseases Society.
B.L. was supported by The Thrasher Research Foundation Early Career Award bottle. Polymicrobial bacteremia was defined as isolation of ≥2 bac-
Program grant and The Center for Pediatric Clinical Effectiveness Pilot teria and mixed bloodstream infection when a blood culture yielded
Grant Award. All other authors have no conflicts of interest to disclose. both bacteremia and candidemia. Bloodstream infections were con-
Address for correspondence: Beatriz Larru, MD, PhD, Division of Infectious Dis- sidered hospital onset (HO) when the initial positive blood culture
eases, The Children’s Hospital of Philadelphia, 3615 Civic Center Blvd, ARC
Suite 1202, Philadelphia, PA 19104-4318. E-mail: LarruB@chop.email.edu. was drawn more than 3 days after hospital admission and community
Supplemental digital content is available for this article. Direct URL citations onset (CO) when drawn within 2 days of hospital admission.6 Car-
appear in the printed text and are provided in the HTML and PDF versions of bapenem-resistant Enterobacteriaceae (CRE) was defined according
this article on the journal’s website (www.pidj.com). to Centers for Disease Control and Prevention criteria.7 Bacterial
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
ISSN: 0891-3668/16/3505-0507 pathogens were defined as resistant/intermediate/susceptible to anti-
DOI: 10.1097/INF.0000000000001057 microbial therapy based on CLSI interpretative breakpoints.

The Pediatric Infectious Disease Journal  •  Volume 35, Number 5, May 2016 www.pidj.com | 507

Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Larru et al The Pediatric Infectious Disease Journal  •  Volume 35, Number 5, May 2016

TABLE 1.  Frequency of Top 5 Gram-positive Isolates, Top 5 Gram-negative Isolates and
Candida spp

Overall Community-onset Hospital-onset

No. (%) of No. (%) of No. (%) of

Pathogen Pathogens Rank Pathogens Rank Pathogens Rank

Gram-positive
 Staphylococcus CoNS 3533 (43.1) 1 1390 (36.7) 1 2143 (48.6) 1
 Staphylococcus aureus 721 (8.8) 2 391 (10.3) 2 330 (7.5) 2
 Enterococcus spp. 395 (4.8) 3 87 (2.3) 9 308 (7.0) 3
 Streptococcus viridans 317 (3.9) 6 195 (5.1) 3 122 (2.8) 8
 Streptococcus pneumoniae 209 (2.5) 10 190 (5.0) 4 19 (0.4) 11
Gram-negative
 Klebsiella spp. 363 (4.4) 5 141 (3.7) 6 222 (5.0) 5
 Escherichia coli 290 (3.5) 7 168 (4.4) 5 122 (2.8) 8
 Enterobacter spp. 271 (3.3) 8 87 (2.3) 9 184 (4.2) 6
 Pseudomonas aeruginosa 262 (3.2) 9 95 (2.5) 7 167 (3.8) 7
 Serratia spp. 113 (1.4) 11 15 (0.4) 11 98 (2.2) 10
Fungus
 Candida spp. 374 (4.6) 4 98 (2.6) 8 276 (6.3) 4
Total episodes 8196 3791 (46.2)* 4405 (53.8)

Statistical Analysis Hispanic. Overall, 46% of bloodstream infections were CO, 72%
Descriptive analyses used medians for continuous variables were Gram-positive bacteria, 22% Gram-negative bacteria and 5%
and percentages for categorical data. Difference among proportions Candida spp. (Table 1). The distribution of pathogens is illustrated
over time was tested using a χ2 test for trend. A P value of <0.05 by year and by age group in Figure (Supplemental Digital Content
was considered statistically significant. All data were analyzed 1, http://links.lww.com/INF/C383). Coagulase-negative Staphylo-
using STATA software (Stata Inc, College Station, TX). coccus (CoNS) was the most commonly isolated organism. Chil-
dren younger than 3 months had the highest percentage of all iso-
lated CoNS (31%, P < 0.01). ESKAPE pathogens (Enterococcus
RESULTS faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acineto-
Over the 11-year observation period, 8196 episodes of bacter baumannii, P. aeruginosa and Enterobacter spp.) accounted
monomicrobial bacteremia occurred in 5508 patients. Median for 20% of all episodes and 40% of hospital-acquired bacteremia
patient age was 2 years (interquartile range: 0–9); 57% were male, when excluding CoNS. The median time from hospital admission
50% were Caucasian, 35% were African-American and 10% were to onset of bacteremia (Fig. 1) ranged from 14 (for Streptococcus
viridans) to 33 days (for Enterococcus spp.)
Antibiotic susceptibility results were available for 7100
episodes. Antimicrobial resistance patterns for the Gram-negative
organisms most frequently isolated during the study period are
shown in Table 2. All S. aureus isolates were susceptible to van-
comycin or linezolid. We found a significant increase in the inci-
dence of methicillin-resistant S. aureus (MRSA) during the study
period (P < 0.01). During the first 4-year interval MRSA episodes
were equally distributed between HO and CO infections, whereas
over the last 7 years, a greater proportion of CO infections were
observed. There was no increase in vancomycin minimum inhibi-
tory concentration for MRSA over time. Ampicillin-resistant E. coli
was common throughout the study period. There was no significant
increase in the incidence of vancomycin-resistant Enterococcus or
ceftazidime-resistant P. aureginosa (Fig. 2). Only 7 bloodstream
infections with CRE were identified: 4 with K. pneumoniae, 2 with
Enterobacter cloacae and 1 with Enterobacter aerogenes. Four of
the 7 children with CRE bacteremia died (mean length from culture
to death = 22.5 days; standard deviation: 34.7).

DISCUSSION
Analyzing more than 8000 unique episodes of monomi-
crobial bacteremia in a decade at a large, tertiary care children’s
hospital, we defined the epidemiology and resistance patterns of
FIGURE 1.  Time interval between hospital admission and bloodstream infections among these hospitalized children. Overall,
onset of blood stream infections for HAI (only includes Gram-positive organisms predominated. After CoNS, S. aureus was
mono-microbial episodes with onset >72 hours from the most commonly isolated Gram-positive pathogen. ESKAPE
date of admission). Time interval (days): median ±95% organisms accounted for almost half of all hospital-acquired bac-
confidence interval. teremia. Ampicillin-resistant E. coli was common throughout the

508  |  www.pidj.com © 2016 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
The Pediatric Infectious Disease Journal  •  Volume 35, Number 5, May 2016 Bloodstream Infections

study period, MRSA incidence increased with time, and CRE

% Nonsusceptible
was rare.
Our data add to the limited understanding of the epidemiol-
ogy of pediatric bacteremia, which has evolved because of changes

8.0
8.9
1.8
18.7
12.5
8.0
15.2

0
47
Serratia spp.

—
in prenatal screening and infant immunizations.8 In decades past,
Streptococcus pneumoniae, Haemophilus influenzae type b and
Neisseria meningitidis were considered the major causes of bac-
teremia in children.9 More recently, changes in the epidemiology

20 (17.9)
61 (54.5)
61 (54.5)
83 (74.1)
52 (46.4)

112 (100)
112 (100)
112 (100)

of bacteremia have been reported. Consistent with this, our study

No (%) of
Isolates

93 (83)
Tested

has identified S. aureus and Enterobacteriaceas as more common

—
TABLE 2.  Rates of Antimicrobial Resistance Among Gram-negative Organisms Most Frequently Isolated During the Study Period

pathogens. Continuous surveillance is required to inform empiric

therapy for presumed bacteremia as isolated pathogens also vary
between geographic locations and clinical settings. For instance,
% Nonsusceptible

Biondi et al5 reported that E. coli was the most common patho-
Pseudomonas aeruginosa

gen from 4255 blood cultures collected among infants aged 1 week
4.3
3.2
13.4
3.2
7.5
4.3
12.2

—
—
—

to 3 months, whereas in our study, S. aureus was the most com-

monly isolated pathogen in this age group (19.5%) followed by
E. coli (11%). Differences are likely attributable to variations in
study populations (predominance of CO vs. HO infections).
126 (49.6)
129 (50.8)
252 (99.2)
No (%) of

254 (100)
254 (100)
Isolates

Healthcare-acquired blood stream infections (HAI) due to

Tested

127 (50)
127 (50)
—
—
—

Gram-positive organisms have become increasingly frequent since

the 1980s because of the increasing use of central venous catheters
and are now the leading cause of nosocomial bacteremia.10,11 Our
% Nonsusceptible

study corroborates the predominance of Gram-positive pathogens,

even when potential contaminants such as CoNS were excluded.
4.5
0.4
1.1
7.5
24.6
1.8
17.5
27.9
17.9
Enterobacter spp.

The interval between hospital admission and onset of HO-blood

—

stream infection in our study correlates to the data from adult

SCOPE patients and the pediatric nosocomial bloodstream infection
series by Wisplinghoff et al, with E. coli and S. aureus occurring
earliest and Enterococcus spp. latest. Almost all HO bloodstream
40 (14.9%)
267 (99.6%)
146 (54.5%)
227 (84.7%)

268 (100%)
No (%) of

146 (54.5)
191 (71.3)
121 (45.2)
isolates

268 (100)
tested

infections occurred after 2 weeks from hospital admission. Detailed

—

knowledge of the epidemiology of pediatric bacteremia is essential

to develop appropriate recommendations for empiric therapy. The
proportion of bloodstream infections caused by Candida spp. in
% Nonsusceptible

our study, 4.6%, was similar to other nationwide surveillance stud-

ies. We also found a significant percentage of non-albicans species,
3.6
0.6
1.9
24.4
3.3
4.4
5.6
11.4
5.8

10

some of which are resistant to first-generation azoles.11

Klebsiella spp.

Our study did not show a significant increase in the preva-

lence of vancomycin-resistant Enterococcus or ceftazidime-resistant
P. aureginosa during the study period. This is in contrast to National
Healthcare Safety Network estimates, which reported that nearly
66 (18.3)
293 (81.4)
359 (99.7)
354 (98.3)
197 (54.7)
256 (71.1)
160 (44.4)
No (%) of

360 (100)
360 (100)
Isolates
Tested

198 (55)

20% of all pathogens causing HAI had a multiresistant phenotype,

and data published by Edelsberg et al who documented alarming
levels of antimicrobial resistance among Enterococcus (E. faecium
resistant to vancomycin represented 87.1%).12–15 The lower rates of
% Nonsusceptible

antimicrobial resistance among pediatric patients found in our study

is consistent with a recent survey of pediatric antibiograms from
9.7
6.6
49.5
1.7
3.5
7.3
8.3
2.4

0
0

55 US pediatric healthcare institutions.4 In comparison to this com-

posite antibiogram, we observed a significant increase in MRSA
E. coli

episodes during the study period particularly among CO infections.

Our longer study length, more detailed clinical information and
288 (99.7)
67 (23.2)
221 (76.5)
288 (99.7)
286 (98.9)
162 (56.1)
165 (57.1)
217 (75.1)

ability to exclude repeated episodes and polymicrobial infections

No (%) of

289 (100)
Isolates
Tested

122 (42)

might explain these differences as variable compliance with CLSI

recommendations for developing antibiograms has been noted.16
Our study has several limitations. Data were derived from
a single institution, limiting its generalizability to other settings.
Piperacillin/tazobactam
Ampicillin/sulbactam

However, the size and scope of this cohort coupled with our abil-
Antimicrobial Drug

ity to isolate clinical episodes of monomicrobial bacteremia over

a decade makes this a unique contribution. Also, because we did
Ciprofloxacin
Ceftazidime

Meropenem
Ceftriaxone
Cefotaxime

not perform detailed chart review for each episode, we could not
Aztreonam
Imipenem
Cefepime

characterize the clinical severity of each case of bacteremia and

based our definition of contaminants on organism class (eg, CoNS).
Finally, antimicrobial susceptibility was available for only 87% of
episodes.

© 2016 Wolters Kluwer Health, Inc. All rights reserved. www.pidj.com  |  509

Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Larru et al The Pediatric Infectious Disease Journal  •  Volume 35, Number 5, May 2016

FIGURE 2.  Rates of antimicrobial resistance. Vertical axis represents % of nonsusceptible monomicrobial episodes.

REFERENCES 9. Long SS, Pickering LK, Prober CG. Principles and Practice of Pediatric
Infectious Diseases. Philadelphia, PA: Elsevier Health Sciences; 2012.
1. Gray J, Gossain S, Morris K. Three-year survey of bacteremia and fungemia
in a pediatric intensive care unit. Pediatr Infect Dis J. 2001;20:416–421. 10. Wisplinghoff H, Seifert H, Tallent SM, et al. Nosocomial bloodstream infec-
tions in pediatric patients in United States hospitals: epidemiology, clinical
2. Pammi M, Zhong D, Johnson Y, et al. Polymicrobial bloodstream infections
features and susceptibilities. Pediatr Infect Dis J. 2003;22:686–691.
in the neonatal intensive care unit are associated with increased mortality: a
case-control study. BMC Infect Dis. 2014;14:390. 11. Wisplinghoff H, Bischoff T, Tallent SM, et al. Nosocomial bloodstream
3. Folgori L, Livadiotti S, Carletti M, et al. Epidemiology and clinical out- infections in US hospitals: analysis of 24,179 cases from a prospective
comes of multidrug-resistant, gram-negative bloodstream infections in nationwide surveillance study. Clin Infect Dis. 2004;39:309–317.
a European tertiary pediatric hospital during a 12-month period. Pediatr 12. Sievert DM, Ricks P, Edwards JR, et al. Antimicrobial-resistant pathogens
Infect Dis J. 2014;33:929–932. associated with healthcare-associated infections: summary of data reported
4. Tamma PD, Robinson GL, Gerber JS, et al. Pediatric antimicrobial sus- to the National Healthcare Safety Network at the Centers for Disease Control
ceptibility trends across the United States. Infect Control Hosp Epidemiol. and Prevention, 2009–2010. Infect Control Hosp Epidemiol. 2013;34:1–14.
2013;34:1244–1251. 13. Lagace-Wiens PRS, Adam HJ, Low DE, et al. Trends in antibiotic resist-
5. Biondi E, Evans R, Mischler M, et al. Epidemiology of bacteremia in febrile ance over time among pathogens from Canadian hospitals: results of the
infants in the United States. Pediatrics. 2013;132:990–996. CANWARD study 2007–11. J Antimicrob Chemother. 2013;68(suppl
1):i23–i29.
6. Siegman-Igra Y, Fourer B, Orni-Wasserlauf R, et al. Reappraisal of commu-
nity-acquired bacteremia: a proposal of a new classification for the spectrum 14. Karlowsky JA, Adam HJ, Desjardins M, et al. Changes in fluoroquinolone
of acquisition of bacteremia. Clin Infect Dis. 2002;34:1431–1439. resistance over 5 years (CANWARD 2007–11) in bacterial pathogens isolated
in Canadian hospitals. J Antimicrob Chemother. 2013;68(suppl 1):i39–i46.
7. Centers for Disease Control and Prevention (CDC). Guidance for con-
trol of infections with carbapenem-resistant or carbapenemase-producing 15. Edelsberg J, Weycker D, Barron R, et al. Prevalence of antibiotic resistance
Enterobacteriaceae in acute care facilities. MMWR Morb Mortal Wkly Rep. in US hospitals. Diagn Microbiol Infect Dis. 2014;78:255–262.
2009;58:256–260. 16. Lautenbach E, Nachamkin I. Analysis and presentation of cumulative anti-
8. Pérerz Lopéz A, Ladhani SN, Breathnach A, et al. Trends in paediat- microbial susceptibility data (antibiograms): substantial variability across
ric nosocomial bacteraemia in a London tertiary hospital. Acta Paediatr. medical centers in the United States. Infect Control Hosp Epidemiol.
2013;102:1005–1009. 2006;27:409–412.

510  |  www.pidj.com © 2016 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.