ORIGINAL RESEARCH

An Empiric Antibiotic Protocol Using Risk Stratification Improves

Antibiotic Selection and Timing in Critically Ill Children
Todd J. Karsies1, Cheryl L. Sargel2, David J. Marquardt1, Nadeem Khan1, and Mark W. Hall1
1
Department of Pediatrics, Nationwide Children’s Hospital/The Ohio State University College of Medicine, Columbus, Ohio;
and 2Department of Pharmacy, Nationwide Children’s Hospital, Columbus, Ohio

Abstract was assessed, as was time from initial culture to appropriate

Rationale: Timely and appropriate empiric antibiotics can
improve outcomes in critically ill patients with infection. Evidence
and guidelines to guide empiric antibiotic decisions are lacking for
critically ill children.
Objectives: To evaluate the impact of an empiric antibiotic
protocol on appropriateness of initial antibiotics and time to
appropriate antibiotics in critically ill children with suspected
infection.
Methods: A computer order entry–based, pediatric intensive
care unit–specific, empiric antibiotic protocol including risk
stratification for healthcare-associated infections was implemented
in a tertiary pediatric intensive care unit. Antibiotic and culture
data were evaluated for a total of 556 infectious episodes in
491 patients from 2004 (preprotocol, n = 252) and 2007
(protocol, n = 304) with suspected infection. Antibiotics
appropriateness based on risk factors and culture results
antibiotics.
Measurements and Main Results: Patients treated using
the protocols were more likely to receive appropriate empiric
antibiotics based on risk factors (76 vs. 15%; P , 0.0001) and
culture results (89 vs. 64%; P , 0.0001). Patients treated after
protocol implementation had a shorter time to appropriate antibiotics
(median, 5.9 vs. 9.6 h; P , 0.0001), particularly in those who grew
healthcare-associated pathogens (5.8 vs. 24 h; P = 0.0001). No
significant baseline characteristic differences were seen.
Conclusions: An empiric antibiotic protocol in the pediatric
intensive care unit incorporating risk stratification for healthcare-
associated infections resulted in increased appropriateness of
empiric antibiotics and in decreased time to appropriate antibiotics
in critically ill children with infection.
Keywords: pediatric intensive care unit; treatment protocol;
antibiotic resistance

(Received in original form August 21, 2014; accepted in final form October 7, 2014 )
This work was supported by an internal grant from the Center for Clinical and Translational Research at Nationwide Children’s Hospital Research Institute
(D.J.M.).
Author Contributions: Conception and design: C.L.S., D.J.M., N.K., and M.W.H. Analysis and interpretation: T.J.K., C.L.S., and M.W.H. Drafting the
manuscript for important intellectual content: T.J.K., C.L.S., D.J.M., N.K., and M.W.H.
Correspondence and requests for reprints should be addressed to Todd J. Karsies, M.D., Department of Pediatrics, The Ohio State University College of
Medicine, 700 Children’s Drive, Columbus, OH 43215. E-mail: todd.karsies@nationwidechildrens.org
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org
Ann Am Thorac Soc Vol 11, No 10, pp 1569–1575, Dec 2014
Copyright © 2014 by the American Thoracic Society
DOI: 10.1513/AnnalsATS.201408-389OC
Internet address: www.atsjournals.org

Multiple studies in adults propose
a critical role of early, appropriate empiric
antibiotics in sepsis and septic shock, with
mortality and morbidity benefits (1–8).
Data also suggest a mortality benefit with
early and appropriate antibiotics in adult
healthcare-associated pneumonia and
ventilator-acquired pneumonia and
a morbidity benefit in children with
severe community-acquired pneumonia
(9–12). Pediatric data outside the
intensive care unit (ICU) also suggest that
inappropriate empiric antibiotics are
independently associated with mortality in
bacteremia (13). However, there are very
limited data examining inappropriate
empiric antibiotic rates, time to appropriate
antibiotics, or the clinical impact of
inappropriate empiric antibiotics in the
pediatric ICU (PICU).
The rise of antibiotic-resistant
bacteria has made antibiotic decisions
more challenging. Particularly challenging
are organisms such as methicillin-
resistant Staphylococcus aureus
(MRSA), Pseudomonas aeruginosa,
and other so-called ESKAPE organisms

Karsies, Sargel, Marquardt, et al.: Empiric Antibiotic Protocol in the PICU 1569

(i.e., Enterococcus faecium, S. aureus,
Klebsiella pneumoniae, Acinetobacter
baumannii, P. aeruginosa, and Enterobacter
species), as highlighted in a recent
Infectious Diseases Society of America
review (14). Adult data indicate that
infections with these organisms are
associated with poor outcomes, especially
if initial antibiotics are incorrect (9, 15).
These organisms are a growing problem
in children, although this problem is less
well described. During a recent year in
our PICU, nearly 60% of patients with
positive cultures grew resistant or
healthcare-associated organisms, most
commonly MRSA or Pseudomonas
(unpublished data), and a recent study
suggested worse outcomes in bacteremic
children infected with resistant
Pseudomonas (16).
Faced with these dual challenges,
one potential approach would be to use
extremely broad-spectrum antibiotics
empirically in all critically ill children
with suspected infection. Although this
would increase the likelihood of covering
healthcare-associated organisms such as
Pseudomonas, many patients would receive
unnecessarily broad antibiotic coverage.
There is concern that excessive broad-
spectrum antibiotic use could lead to
increased antibiotic resistance and
increased toxicity (17, 18). This has led
to a commonly held consensus opinion,
offered by most major adult and pediatric
infectious disease societies, that more
narrow-spectrum antibiotics should be
used when it is safe to do so (19–21).
Although there are published
adult guidelines outlining risk factors for
resistant organisms to aid antibiotic decision
making (19, 20, 22), no similar pediatric
guidelines exist. Studies that have examined
risk factors for healthcare-associated
infections in critically ill children primarily
focus on hospital-acquired infection
(23–25), although we know that many
children present from the community
infected with ESKAPE-type organisms.
We are unaware of any prior studies
using patient-specific risk factors to
drive empiric antibiotic selection for all
critically ill children in a PICU.
We therefore designed and
implemented a prospective intervention
involving the use of a computer-based,
PICU-specific, empiric antibiotic protocol
that incorporated risk factors for
healthcare-associated infections into
1570
the prescription process. The objective of
this analysis was to define the baseline
rate of empiric antibiotic appropriateness
and time from culture to appropriate
antibiotics and to evaluate the impact of
our protocol on antibiotic appropriateness
and timing in critically ill children with
suspected infection. Our hypothesis was
that this protocol would improve initial
antibiotic appropriateness and decrease
time from culture to appropriate antibiotics
in these patients.
Methods
Study Population and Data Collection
This is a retrospective cohort study
evaluating all patients in a single, tertiary
care PICU with an infection-related
diagnosis for one calendar year before
(2004) and after (2007) implementation
of a PICU empiric antibiotic protocol
(implemented in the 2005–2006 academic
year). Approval by the institutional review
board of Nationwide Children’s Hospital
with a waiver of consent was obtained.
We included all PICU patients during
2004 and 2007 with a primary or secondary
ICD-9 diagnosis at discharge of
pneumonia, sepsis, septic shock, urinary
tract infection, or urosepsis confirmed by
chart review. Data were collected from
chart review and included demographic,
diagnostic, microbiologic, medication
utilization, resource utilization, and
outcomes data.
Empiric Antibiotic Protocol
We developed a computerized physician
order entry–based empiric antibiotic
protocol for patients in our PICU suspected
to have infection (Figure 1). The protocol
involved risk stratification to place
patients at high or low risk for healthcare-
associated bacteria, particularly gram-
negative organisms, using risk factors
based on those described in adults for
hospital-acquired or healthcare-associated
pneumonia due to multi–drug-resistant
pathogens (19). The risk factors were listed
in the PICU antibiotic order set to allow
for appropriate antibiotic selection, and
specific age- and risk-based options were
available as drop-down menus. Additional
protocol details can be found in the
online supplement. Clinicians were
encouraged but not required to use the
protocol.
AnnalsATS Volume 11 Number 10 | December 2014
ORIGINAL RESEARCH
Due to a high community MRSA
rate, all patients treated with empiric
antibiotics were to be prescribed
vancomycin or other MRSA coverage.
Patients with at least one high-risk criterion
were also prescribed an antipseudomonal
b-lactam and an aminoglycoside (Figure 1).
Protocol-targeted gram-negative organisms
included (but were not limited to)
Pseudomonas species, Acinetobacter species,
Enterobacter species, Klebsiella oxytoca,
and extended-spectrum b-lactamase–
producing species. Empiric antibiotics
had automatic 72-hour stop times, with
definitive treatment left to the attending
physician based on culture and antibiotic
susceptibility results and clinical response.
Before protocol implementation, an
escalation strategy for antibiotic use was
typical in our unit (further details are
provided in the online supplement).
Study Definitions
Initial empiric antibiotics providing
adequate coverage based on patient risk
factors (e.g., a patient with any high-risk
criterion should get high-risk antibiotics)
were considered risk appropriate. For
patients with positive cultures, antibiotics
were considered culture appropriate if
the cultured organism was susceptible
to at least one prescribed antibiotic. In
polymicrobial infections, each organism
had to be susceptible to at least one
prescribed antibiotic for coverage to be
considered appropriate. If correct antibiotics
were not prescribed until after a positive
culture result, empiric antibiotics were
considered inappropriate. The time to
correct antibiotics was defined from the
time of the initial culture (obtained either
locally or at a referring facility) to the time
the first dose of an appropriate parenteral
antibiotic was administered. A positive
culture was defined as any growth of
a pathogenic organism from a sterile site
or from a lower respiratory specimen.
Although some patients did grow fungi,
these were neither targeted by the antibiotic
protocol nor included in analysis of
antibiotic appropriateness.
Statistical Analysis
Primary outcomes were protocol adherence
(risk-based antibiotic appropriateness) and
culture-based antibiotic appropriateness
for culture-positive patients. Secondary
outcomes included time to appropriate
antibiotics, antibiotic appropriateness at
 ORIGINAL RESEARCH

Suspected Bacterial Infection

post-protocol (257/304) (P , 0.0001).
In culture-positive patients, only 64% of
pre-protocol patients received appropriate
Culture Suspected Source(s)
empiric antibiotics based on the cultured
organism compared with almost 90% of
protocol patients (Table 2). Post-protocol,
there was an increase in the number of
Any Risk Factor for Resistant Organisms?
patients who received their first antibiotic,
-Hospitalization > 48 hours -Chronic Lung Disease first risk-appropriate antibiotic, and first
-Current/Recent antibioticsa -Congenital Heart Disease culture-appropriate antibiotic before
-Immunosuppressionb -Resident of chronic care facility ICU admission, but this still represented
-Hospitalization within past month -Indwelling central venous line
a minority of patients overall (Table 3).
-History of antibiotic-resistant infection -History of prematurity (if pt < 2y)
Post-protocol, only 40% of patients
received risk-appropriate antibiotics
No Yes pre-ICU, and 36% of culture-positive
patients received culture-appropriate
Low Risk Antibiotics High Risk Antibiotics antibiotics before PICU admission.
1. Vancomycin 1. Vancomycin Among all culture-positive patients,
2. Cefotaxime 2. Cycled anti-Pseudomonal β-lactamc nearly 50% grew gram-positive organisms
3. Cycled aminoglycosided
3. Other agents as clinically indicated (161/324), whereas 66% grew gram-negative
4. Other agents as clinically indicated
organisms (215/324). Nearly 12% of
culture-positive patients grew MRSA,
Figure 1. Pediatric intensive care unit empiric antibiotic pathway including risk factors for infection
with no difference in frequency seen
due to healthcare-associated bacteria. aMinimum 7 days in previous 6 weeks. bMalignancy,
chemotherapy, chronic steroid/immunosuppressants, organ transplant, immunodeficiency, or between high- and low-risk patients. The
acute steroids .5 days in the past month. cPiperacillin-tazobactam, cefepime, and meropenem. majority of positive cultures for MRSA
d
Gentamicin, tobramycin, and amikacin. were lower respiratory cultures, including
50% of low-risk patients with MRSA and
select time points in culture-positive (304 episodes). No differences were seen in 88% of high-risk patients with MRSA.
patients, and risk-based and culture-based demographics or baseline characteristics, No differences were seen in rates of
appropriateness for subgroups. Data were and no difference was seen in mortality gram-positive organisms, gram-negative
analyzed using SAS 9.2 (SAS Institute Inc., (Table 1). organisms, or MRSA between pre- and post-
Cary, NC). Further details can be found protocol patients. Empiric antibiotics were
in the online supplement. Antibiotic Appropriateness appropriate for 79% of patients growing
Protocol patients were more likely to gram-positive organisms (127/161),
receive risk-appropriate empiric antibiotics including 59% pre-protocol and 93%
Results than pre-protocol patients regardless of post-protocol (P , 0.0001). Empiric
risk category (Table 2). This includes antibiotics were appropriate for 78% of
A total of 213 preprotocol patients a significant increase in the number of patients growing gram-negative organisms,
representing 252 infectious episodes were patients receiving empiric MRSA coverage with 67% appropriate pre-protocol and
compared with 278 protocol patients from 29% pre-protocol (72/252) to 85% 88% post-protocol (P , 0.0001). Empiric
antibiotics were appropriate for 69%
of patients growing MRSA, including
Table 1. Demographics and baseline data
36% pre-protocol and 88% post-protocol
(P = 0.0013). No difference was seen in the
Characteristics Pre-Protocol* Post-Protocol P Value frequency of patients growing high-risk
protocol-targeted organisms between pre-
Age, yr, mean (SD) 7.5 (8.7) 7.7 (8.9) 0.77 protocol and protocol patients (34 vs. 28%;
Male sex, n (%) 103 (48) 155 (56) 0.1
Admit location 0.1 P = 0.14).
ED, n (%) 113 (53) 140 (50) 0.1
Floor, n (%) 63 (30) 104 (37) 0.1 Time to Antibiotics
Other, n (%) 37 (17) 34 (13) 0.1 No difference was seen in time to first
High-risk patients, n (%) 171/252 (68) 206/304 (68) 0.98 antibiotic between pre-protocol and
PRISM score, mean (SD) 7.4 (7.3) 6.6 (6.9) 0.19
PELOD score, mean (SD) 10.8 (10) 11 (9.5) 0.87 protocol patients (median time after
Mechanical ventilation, n (%) 140 (66) 162 (58) 0.14 culture, 1.55 h for both groups; P = 0.99).
Mortality, n (%) 25 (11.7) 22 (7.9) 0.17 However, protocol patients had a
significantly shorter time from culture
Definition of abbreviations: ED = emergency department; PELOD = Pediatric Logistic Organ
Dysfunction; PRISM = Pediatric Risk of Mortality.
to first risk-appropriate antibiotic than
*Pre-protocol n = 213 and post-protocol n = 278 except where noted due to analysis per episode pre-protocol patients (median time, 4 vs.
rather than per patient. 5.9 h) (Figure 2A). Among patients with

Karsies, Sargel, Marquardt, et al.: Empiric Antibiotic Protocol in the PICU 1571

Table 2. Patients with appropriate initial empiric antibiotics stratified by risk factors or
culture results
Patient Groups Pre-Protocol,
n/ Total (%)
Risk-appropriate
All patients 38/252 (15)
High-risk patients 28/171 (16)
Low-risk patients 10/81 (12)
Culture-appropriate
All patients 95/148 (64)
High-risk patients 72/116 (62)
Low-risk patients 23/32 (72)
positive cultures, protocol patients had
a shorter time from culture to appropriate
antibiotics compared with pre-protocol
patients (Figure 2B) (median time to
appropriate antibiotics, 5.9 vs. 9.6 h). For
patients growing high-risk protocol-targeted
organisms, the median time to appropriate
antibiotics was 5.8 hours for protocol
patients, compared with 24 hours for
pre-protocol patients (P = 0.0001).
Risk Factor Performance and
Protocol Compliance
When considering all patients, the selected
risk factors had a sensitivity of 99% for
the presence of any risk factor in predicting
resistant gram-negative organisms, with
only two patients without any risk factor
growing one of these organisms (both grew
K. oxytoca). The risk-factor specificity
was 43%, with a positive predictive value
of 38% and a negative predictive value of
99%. Risk factor performance was similar
when only culture-positive patients were
considered (Table 4 and 5). Of the 142
high-risk patients with cultures growing
high-risk gram-negative rods, 69 (49%)
presented from the community without
recent hospital exposure.
Table 3. Patients receiving first antibiotic, first risk-appropriate antibiotic, and first culture-appropriate antibiotic (culture-positive
patients) before ICU admission for all patients for whom antibiotic administration location was available, stratified by risk category
Patient Groups
First antibiotic
Low-risk patients
High-risk patients
First risk-appropriate antibiotic
Low-risk patients
High-risk patients
First culture-appropriate antibiotic
Low-risk patients
High-risk patients
1572
ORIGINAL RESEARCH
the largest study to evaluate rates of empiric
antibiotic appropriateness and timing of
appropriate empiric antibiotics in the
Post-Protocol, P Value PICU. The decreased time to appropriate
n/ Total (%) antibiotics occurred despite the fact that,
for many patients, initial cultures were
obtained outside of the ICU and the
231/304 (76) ,0.0001 protocol was often not applied until several
158/206 (77) ,0.0001 hours after initial presentation.
73/98 (74) ,0.0001 Growing evidence suggests that
157/176 (89) ,0.0001 inappropriate or delayed initial empiric
120/136 (88) ,0.0001 antibiotics can have significant
37/40 (93) 0.0268 consequences in critically ill patients,
including increased morbidity and mortality
in a variety of conditions (3, 6–9, 11, 26).
The empiric antibiotic protocol was Additional studies, mostly in adults,
correctly followed in 76% of cases (231/304). have demonstrated that delay of empiric
Of the 73 protocol patients who did not antibiotics in critically ill patients with
receive appropriate risk-based empiric infection is associated with adverse
therapy, the most common reasons included outcomes, even if the antibiotics prescribed
classification in incorrect risk category are appropriate (1, 2, 11, 12). However,
(23%), low clinical suspicion of bacterial other than a single retrospective analysis of
infection (21%), use of definitive therapy children with severe community-acquired
for known infection (17%), and initial pneumonia, minimal research or guidelines
antibiotic selection by non-PICU service have been published on the subject of
(17%). Of the 53 culture-positive, pre- timely and appropriate empiric antibiotic
protocol patients with inappropriate therapy for critically ill children (12).
antibiotics, 85% would have had appropriate Our approach was consistent with
coverage using the antibiotic protocol. This current consensus adult guidelines for
included 18 patients growing Pseudomonas, treatment of community-acquired and
10 patients with MRSA, and 11 patients healthcare-associated pneumonia (19,
with other high-risk gram-negative rods. 20, 22). It included many antibiotic
stewardship strategies, including initial
combination therapy for patients at risk
Discussion for resistant gram-negatives, tailoring
antibiotic selection to local antibiograms,
This is the first study to our knowledge to and use of antibiotic order sets (27–29).
evaluate a PICU empiric antibiotic protocol This protocol improved initial antibiotic
applied to all patients incorporating risk appropriateness not just for those at
stratification for resistant gram-negative risk for resistant organisms but also for
infections, and it demonstrated that such those suspected of having community-
a protocol can improve initial antibiotic acquired infections, likely due to
appropriateness and time to appropriate increased coverage of community-
antibiotics in critically ill children. It is also acquired MRSA and community-onset
Pre-Protocol, n/ Total (%) Post-Protocol, n/ Total (%) P Value
118/252 (47) 176/304 (58) 0.01
52/81 (64) 71/98 (72) 0.26
66/171 (39) 105/206 (51) 0.02
80/252 (32) 123/304 (40) 0.03
49/81 (60) 67/98 (68) 0.28
31/171 (18) 56/206 (27) 0.049
46/139 (33) 59/166 (36) 0.72
12/29 (41) 20/37 (54) 0.33
34/110 (31) 39/129 (30) 1.0
AnnalsATS Volume 11 Number 10 | December 2014
 ORIGINAL RESEARCH

A healthcare-associated infections. This

100% was accomplished without using broad,
antipseudomonal drugs for all patients
but rather by limiting these agents for those
p=0.0138 (log-rank) Protocol patients felt to be at highest risk for resistant
Pre-Protocol gram-negative bacilli, consistent with
% w/o appropriate antibiotics

75%
current guidelines that recommend tailoring
antibiotic therapy based on risk for
particular organisms and using the most
50% narrow-spectrum agents as possible (19–21).
Our protocol was designed to limit
misclassification of patients as low risk
who ultimately grow resistant gram-
negative organisms. The criteria were
25%
similar to the adult guidelines on which they
were based (e.g., structural lung disease,
recent hospitalization or antibiotics, and
immunosuppression) and were highly
0% sensitive (19, 20, 22). This sensitivity did
0 12 24 36 48 come at the expense of lower specificity,
Time (hours) but 36% of all high-risk patients did grow
“high-risk” organisms. Research is ongoing
B to refine our risk factors and to improve
100%
our predictive capability, which is
important in improving antibiotic
decisions, particularly in patients at risk
p<0.0001 (log-rank) Protocol
Pre-Protocol for resistant organisms but without
% w/o appropriate antibiotics

75% a culture from the site of suspected

infection. Many of our low-risk patients
likely have very low likelihood of MRSA
infections, so we are also developing criteria
50% to limit vancomycin use for those patients
unlikely to culture MRSA to limit overuse
of this antibiotic. Some elements of our
antibiotic management protocol designed
25% to minimize the development of resistance
(e.g., antibiotic rotation, use of unit-specific
antibiograms, and combination therapy)
are unproven in critically ill children, but
we suspect strategies successful in critically
0%
ill adults are likely to have application in
0 12 24 36 48
the PICU.
Time (hours) We acknowledge that other units
Figure 2. Kaplan-Meier curves of time to risk-appropriate empiric antibiotics (A) and culture-
may not approach the problem of antibiotic
appropriate antibiotics (for culture-positive patients) (B). appropriateness from the same starting
place. There is a lack of published PICU
data on empiric antibiotic appropriateness.
Although our pre-protocol appropriateness
rate was very low (in our view), it is
Table 4. Contingency table showing growth of high-risk gram-negative rods and risk similar to a study of bacteremic children
factor performance for all culture-positive patients stratified by risk status (total n = 324) (not ICU specific) in which the rate of
appropriate empiric antibiotics was found to
1 High-Risk GNRs 2 High-Risk GNRs be only 62% (50% in children 1–5 yr of age)
and 55% in the PICU (30). This suggests
High-risk patients 142 110 Sensitivity, 99% that our preprotocol practice may not
Low-risk patients 2 70 Specificity, 39% have been unique to our unit. We also
Predictive values PPV = 56% NPV = 97%
designed our protocol based on our
Definition of abbreviations: GNR = gram-negative rod; NPV = negative predictive value; PPV = positive unit-specific antibiogram. This led us to
predictive value. include the use of empiric vancomycin for

Karsies, Sargel, Marquardt, et al.: Empiric Antibiotic Protocol in the PICU 1573
 ORIGINAL RESEARCH

Table 5. Contingency table showing growth of high-risk gram-negative rods and risk
factor performance for all patients (total n = 556)
1 High-risk GNRs*
High-risk patients 142
Low-risk patients 2 177
Predictive values PPV = 38%
Definition of abbreviations: GNR = gram-negative rod; NPV = negative predictive value; PPV = positive
predictive value.
*The “2 high-risk GNRs” group includes patients growing organisms other than high-risk GNRs plus
culture-negative patients.
all critically ill children with suspected
infection (high local MRSA rate), to cover
for P. aeruginosa in high-risk patients,
and to exclude empiric coverage of
Stenotrophomonas maltophila and
vancomycin-resistant Enterococcus (except
in patients with history of these bacteria).
Modification of our approach based
on individual unit-specific antibiograms
rather than copying our particular
antibiotic choices would be appropriate in
units with differing prevalences of these
pathogens and different antibiotic
resistance patterns.
There are some additional limitations
to our study. First, although the antibiotic
protocol was prospectively applied as
part of a quality improvement project,
much of the clinical data were collected
retrospectively from the medical record.
The inclusion criteria were broad and
objective, however, limiting the selection
bias that can be seen in retrospective studies.
Similarly, our outcome measures used
objective microbiologic and pharmacy data.
Second, this represents a single-center study.
However, we analyzed nearly 600 infectious
episodes and were able to demonstrate
strong statistical significance. The risk
factors we used have not been previously
studied or validated in children. Despite
2 High-risk GNRs
235 Sensitivity, 99%
Specificity, 43%
NPV = 99%
this, these risk factors performed quite well,
and we are currently validating them in
a separate population.
Our study was primarily designed
to evaluate the impact of the protocol on
antibiotic appropriateness. Accordingly,
we are not able to comment on the effect
of the protocol in children whose cultures
were negative other than to evaluate risk
appropriateness. Similarly, the study was
not powered to evaluate outcomes such
as mortality. Based on our mortality rates,
a study to assess the impact on mortality
would need to have nearly 1,300 patients
in each arm due to the low overall mortality
rate seen in critically ill children. We are
currently evaluating the relationships
between a protocolized approach to
antibiotic use and morbidity in our critically
ill children. This report does not address
unit-wide patterns of antibiotic resistance
over time. Research is ongoing to investigate
the impact of our protocol on local
antibiotic resistance patterns, especially
in Pseudomonas species.
Lastly, the time from culture to
appropriate empiric antibiotics remained
prohibitively long for many children, even
post-protocol. This is likely due, in part,
to the fact that much of that time was
spent before ICU admission. Although
more postprotocol patients did receive
risk- and culture-appropriate antibiotics
pre-ICU, this number was still very low,
does not explain the overall increase in
antibiotic appropriateness, and represents
an area for further improvement. We are
therefore evaluating earlier intervention
starting in the emergency department
to further improve the impact of our
protocol in children at risk for resistant
organisms.
Conclusions
A computerized physician order entry–
based antibiotic protocol incorporating risk
assessment for resistant organisms can
improve antibiotic decisions and timing in
critically ill children while limiting broad
antipseudomonal antibiotics to those at
greatest risk for resistant bacteria. This
protocol represents a first step in striking
a balance between the need for rapid,
correct empiric antibiotic therapy in
critically ill children with suspected
infection and the growing demand for
antibiotic stewardship. Although the
specific antibiotics included should be
tailored to local antibiograms, pediatric risk
stratification guidelines similar to those used
in adults do appear to improve antibiotic
selection and timeliness in critically ill
children at risk for antibiotic-resistant
bacteria and should likely be considered
in PICUs treating these patients. Further
research is needed to evaluate the impact
of this protocol on patient outcomes and
its role in the broader arena of antibiotic
prescribing practices in the PICU. n
Author disclosures are available with the text
of this article at www.atsjournals.org.
Acknowledgment: The authors thank Wei
Wang from the Biostatistics Core at the
Nationwide Children’s Hospital Research Institute
for valuable assistance with data analysis.

References adequate empirical antibiotic therapy on the outcome of patients

1 Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, Suppes
R, Feinstein D, Zanotti S, Taiberg L, et al. Duration of hypotension
before initiation of effective antimicrobial therapy is the critical
determinant of survival in human septic shock. Crit Care Med
2006;34:1589–1596.
2 Gaieski DF, Mikkelsen ME, Band RA, Pines JM, Massone R, Furia FF,
Shofer FS, Goyal M. Impact of time to antibiotics on survival in
patients with severe sepsis or septic shock in whom early goal-
directed therapy was initiated in the emergency department. Crit
Care Med 2010;38:1045–1053.
3 Garnacho-Montero J, Garcia-Garmendia JL, Barrero-Almodovar A,
Jimenez-Jimenez FJ, Perez-Paredes C, Ortiz-Leyba C. Impact of
admitted to the intensive care unit with sepsis. Crit Care Med
2003;31:2742–2751.
4 Garnacho-Montero J, Aldabo-Pallas T, Garnacho-Montero C, Cayuela
A, Jiménez R, Barroso S, Ortiz-Leyba C. Timing of adequate
antibiotic therapy is a greater determinant of outcome than are
TNF and IL-10 polymorphisms in patients with sepsis. Crit Care
2006;10:R111.
5 Paul M, Shani V, Muchtar E, Kariv G, Robenshtok E, Leibovici L.
Systematic review and meta-analysis of the efficacy of appropriate
empiric antibiotic therapy for sepsis. Antimicrob Agents Chemother
2010;54:4851–4863.
6 Kang CI, Kim SH, Park WB, Lee KD, Kim HB, Kim EC, Oh MD,
Choe KW. Bloodstream infections caused by antibiotic-resistant

1574 AnnalsATS Volume 11 Number 10 | December 2014

 ORIGINAL RESEARCH
gram-negative bacilli: risk factors for mortality and impact of
inappropriate initial antimicrobial therapy on outcome. Antimicrob
Agents Chemother 2005;49:760–766.
7 Rodrı́guez-Baño J, Millán AB, Domı́nguez MA, Borraz C, González MP,
Almirante B, Cercenado E, Padilla B, Pujol M; GEIH/GEMARA/REIPI.
Impact of inappropriate empirical therapy for sepsis due to health
care-associated methicillin-resistant Staphylococcus aureus. J Infect
2009;58:131–137.
8 Shorr AF, Micek ST, Welch EC, Doherty JA, Reichley RM, Kollef MH.
Inappropriate antibiotic therapy in Gram-negative sepsis increases
hospital length of stay. Crit Care Med 2011;39:46–51.
9 Zilberberg MD, Shorr AF, Micek ST, Mody SH, Kollef MH. Antimicrobial
therapy escalation and hospital mortality among patients with
health-care-associated pneumonia: a single-center experience.
Chest 2008;134:963–968.
10 Micek ST, Reichley RM, Kollef MH. Health care-associated pneumonia
(HCAP): empiric antibiotics targeting methicillin-resistant
Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa
predict optimal outcome. Medicine (Baltimore) 2011;90:390–395.
11 Luna CM, Aruj P, Niederman MS, Garzón J, Violi D, Prignoni A, Rı́os F,
Baquero S, Gando S; Grupo Argentino de Estudio de la Neumonı́a
Asociada al Respirador group. Appropriateness and delay to initiate
therapy in ventilator-associated pneumonia. Eur Respir J 2006;27:
158–164.
12 Muszynski JA, Knatz NL, Sargel CL, Fernandez SA, Marquardt DJ, Hall
MW. Timing of correct parenteral antibiotic initiation and outcomes
from severe bacterial community-acquired pneumonia in children.
Pediatr Infect Dis J 2011;30:295–301.
13 Ashkenazi S, Leibovici L, Samra Z, Konisberger H, Drucker M. Risk
factors for mortality due to bacteremia and fungemia in childhood.
Clin Infect Dis 1992;14:949–951.
14 Boucher HW, Talbot GH, Bradley JS, Edwards JE, Gilbert D, Rice LB,
Scheld M, Spellberg B, Bartlett J. Bad bugs, no drugs: no ESKAPE!
An update from the Infectious Diseases Society of America. Clin
Infect Dis 2009;48:1–12.
15 Parker CM, Kutsogiannis J, Muscedere J, Cook D, Dodek P, Day
AG, Heyland DK; Canadian Critical Care Trials Group. Ventilator-
associated pneumonia caused by multidrug-resistant organisms or
Pseudomonas aeruginosa: prevalence, incidence, risk factors, and
outcomes. J Crit Care 2008;23:18–26.
16 Yang MA, Lee J, Choi EH, Lee HJ. Pseudomonas aeruginosa
bacteremia in children over ten consecutive years: analysis of clinical
characteristics, risk factors of multi-drug resistance and clinical
outcomes. J Korean Med Sci 2011;26:612–618.
17 Tacconelli E, De Angelis G, Cataldo MA, Mantengoli E, Spanu T, Pan A,
Corti G, Radice A, Stolzuoli L, Antinori S, et al. Antibiotic usage and
risk of colonization and infection with antibiotic-resistant bacteria:
a hospital population-based study. Antimicrob Agents Chemother
2009;53:4264–4269.
18 Hensgens MP, Goorhuis A, Dekkers OM, Kuijper EJ. Time interval of
increased risk for Clostridium difficile infection after exposure to
antibiotics. J Antimicrob Chemother 2012;67:742–748.
19 American Thoracic Society; Infectious Diseases Society of America.
Guidelines for the management of adults with hospital-acquired,
ventilator-associated, and healthcare-associated pneumonia.
Am J Respir Crit Care Med 2005;171:388–416.
Karsies, Sargel, Marquardt, et al.: Empiric Antibiotic Protocol in the PICU
20 Mandell LA, Wunderink RG, Anzueto A, Bartlett JG, Campbell GD,
Dean NC, Dowell SF, File TM Jr, Musher DM, Niederman MS, et al.;
Infectious Diseases Society of America; American Thoracic Society.
Infectious Diseases Society of America/American Thoracic Society
consensus guidelines on the management of community-acquired
pneumonia in adults. Clin Infect Dis 2007;44:S27–S72.
21 Bradley JS, Byington CL, Shah SS, Alverson B, Carter ER, Harrison C,
Kaplan SL, Mace SE, McCracken GH Jr, Moore MR, et al.; Pediatric
Infectious Diseases Society and the Infectious Diseases Society of
America. The management of community-acquired pneumonia in
infants and children older than 3 months of age: clinical practice
guidelines by the Pediatric Infectious Diseases Society and the
Infectious Diseases Society of America. Clin Infect Dis 2011;
53:e25–e76.
22 Woodhead M, Blasi F, Ewig S, Garau J, Huchon G, Ieven M, Ortqvist A,
Schaberg T, Torres A, van der Heijden G, et al.; Joint Taskforce
of the European Respiratory Society and European Society for
Clinical Microbiology and Infectious Diseases. Guidelines for the
management of adult lower respiratory tract infections: summary.
Clin Microbiol Infect 2011;17:1–24.
23 Mello MJ, Albuquerque MdeF, Lacerda HR, Souza WV, Correia JB,
Britto MC. Risk factors for healthcare-associated infection in
pediatric intensive care units: a systematic review. Cad Saude
Publica 2009;25:S373–S391. (Rio J).
24 de Mello MJ, de Albuquerque MdeF, Lacerda HR, Barbosa MT, de
Alencar Ximenes RA. Risk factors for healthcare-associated infection
in a pediatric intensive care unit. Pediatr Crit Care Med 2010;11:
246–252.
25 Toltzis P, Hoyen C, Spinner-Block S, Salvator AE, Rice LB. Factors
that predict preexisting colonization with antibiotic-resistant
gram-negative bacilli in patients admitted to a pediatric intensive
care unit. Pediatrics 1999;103:719–723.
26 Hamandi B, Holbrook AM, Humar A, Brunton J, Papadimitropoulos EA,
Wong GG, Thabane L. Delay of adequate empiric antibiotic therapy is
associated with increased mortality among solid-organ transplant
patients. Am J Transplant 2009;9:1657–1665.
27 Dellit TH, Owens RC, McGowan JE Jr, Gerding DN, Weinstein RA,
Burke JP, Huskins WC, Paterson DL, Fishman NO, Carpenter CF,
et al.; Infectious Diseases Society of America; Society for Healthcare
Epidemiology of America. Infectious Diseases Society of America
and the Society for Healthcare Epidemiology of America guidelines
for developing an institutional program to enhance antimicrobial
stewardship. Clin Infect Dis 2007;44:159–177.
28 Kollef MH, Golan Y, Micek ST, Shorr AF, Restrepo MI. Appraising
contemporary strategies to combat multidrug resistant gram-negative
bacterial infections: proceedings and data from the Gram-Negative
Resistance Summit. Clin Infect Dis 2011;53:S33–S55, quiz S56–S58.
29 Shlaes DM, Gerding DN, John JF Jr, Craig WA, Bornstein DL, Duncan
RA, Eckman MR, Farrer WE, Greene WH, Lorian V, et al. Society for
Healthcare Epidemiology of America and Infectious Diseases Society
of America Joint Committee on the Prevention of Antimicrobial
Resistance: guidelines for the prevention of antimicrobial resistance
in hospitals. Clin Infect Dis 1997;25:584–599.
30 Ashkenazi S, Samra Z, Konisberger H, Drucker MM, Leibovici L. Factors
associated with increased risk in inappropriate empiric antibiotic
treatment of childhood bacteraemia. Eur J Pediatr 1996;155:545–550.
1575