Professional Documents
Culture Documents
REVIEW ARTICLE
While guidance on appropriate volume of blood and number for nonneutropenic adult inpatients and developed an algo-
of blood cultures (BCx) to be collected exist [1–3], guidance rithm to promote wise use of BCx.
regarding indications for drawing BCx is limited. In clinical
practice, a large proportion of BCx are ordered to evaluate ei- METHODS
ther new or persistent fever, or leukocytosis [4, 5]; however,
Data Search and Study Selection
several studies have shown a lack of correlation between these
The review was conducted in accordance with PRISMA-ScR
clinical parameters and bacteremia [5–7]. Unnecessary BCx
(Preferred Reporting Items for Systematic Reviews and Meta-
are associated with increases in hospital length of stay, antibi-
Analyses Extension for Scoping Reviews) guidelines [11]. A sys-
otic use, and other laboratory testing [8, 9]. They may also lead
tematic search was constructed on PubMed (Supplementary Table
to anemia, patient discomfort, and adverse events associated
1) for English-language articles published between 1 January 2004
with antibiotics started for contaminants. A survey of medicine
and 1 June 2019. Two reviewers (V. F., S. L. S.) independently
residents and attending physicians reported that major factors
screened titles and abstracts for eligibility and evaluated the full
influencing BCx ordering decisions were lack of clinical guid-
text of studies assessed as relevant or unclear. Disagreements were
ance and the expectation that BCx are standard components of
resolved by consensus with a third reviewer (S. E. C.). Studies were
a fever workup [10].
included if they reported either the yield or the utility of BCx in
To develop guidance, we performed a scoping review of the
patients with fever and/or select infectious syndromes. Exclusion
literature to highlight existing evidence on indications for BCx
criteria were outpatients, children, neutropenic and bone marrow
transplant recipients, sample size < 50 patients, candidemia (as
we wanted to focus on immunocompetent hosts), uncommon
Received 28 October 2019; editorial decision 7 January 2020; accepted 13 January 2020;
infections in developed countries (eg, dengue), tuberculosis, or
published online January 14, 2020. transient bacteremia after a procedure (eg, dental extraction),
Correspondence: V. Fabre, Johns Hopkins University School of Medicine, Department of
as the focus was on common scenarios in the inpatient setting.
Medicine, Division of Infectious Diseases, 600 N Wolfe St, Osler 425, Baltimore, MD 21287
(mfabre1@jhmi.edu). Bibliographic references found in articles reviewed were also
Clinical Infectious Diseases® 2020;71(5):1339–7 examined to identify pertinent primary literature.
© The Author(s) 2020. Published by Oxford University Press for the Infectious Diseases Society
Studies were grouped in 2 categories: initial (BCx obtained for
of America. All rights reserved. For permissions, e-mail: journals.permissions@oup.com.
DOI: 10.1093/cid/ciaa039 the workup of new onset of signs and symptoms) or follow-up
(BCx ordered after a positive BCx to document clearance of of BCx on clinical management and patient outcomes, and the
bacteremia). Among initial BCx, we recorded the reported pro- utility of repeating BCx after initial bacteremia with gram-posi-
portion of positive BCx and impact on clinical management. tive (GP) and gram-negative (GN) organisms. Finally, we incor-
porate this knowledge into an algorithm to promote rationale
Development of a BCx Order Clinical Decision Support Tool use of BCx.
Two authors (V. F., S. E. C.) developed an algorithm with indi-
cations for initial and follow-up blood cultures (FUBCx) based Incidence of Bacteremia in Select Clinical Scenarios
on the available evidence. Subsequently, an internal multidis- Yield of Blood Cultures in Selected Infectious Syndromes
ciplinary group at the Johns Hopkins Hospital, with partici- Based on the reported incidence of bacteremia, studies were
pation from antimicrobial stewardship, clinical microbiology, categorized into 5 groups according to the pretest probability
infectious diseases, hospital epidemiology, critical care, surgery of bacteremia (Table 1). Very low (< 5%) probability includes
and hospital medicine, reviewed and modified the algorithm, general medicine patients with isolated fever [5, 6] and fever
resulting in the flowchart presented in Figure 1. within the first 48 hours after a surgery [12–14]. Low (< 10%)
probability included cellulitis [15–18], cystitis/prostatitis [19,
RESULTS
20], nonsevere community-acquired pneumonia (CAP), and
Of 2893 studies screened, 50 met inclusion criteria (Figure 2). healthcare-associated pneumonia (HCAP) [21–24]. Low to
Herein, we describe the incidence of bacteremia in different moderate (10%–20%) includes ventilator-associated pneumonia
clinical scenarios, the additive value of BCx to yield from cul- (VAP) [25, 26] and cellulitis in patients with severe comorbidities
tures obtained from the primary source of infection, the impact [18, 27, 28]. Moderate (20% to < 50%) probability includes
Table 1. Pretest Probability of Bacteremia in Common Clinical Scenarios (Percentages as Reported in the Studies)
Between 10% and < 20% Between 20% and < 50%
< 5% (Very Low) < 10% (Low) (Low-moderate) (Moderate) ≥ 50% (High)
Fever within first Uncomplicated cellulitis Cellulitis in patients with severe Severer sepsis Discitis and VO [39, 40, 47]
48 h of surgery [6, 15–17, 43, 44], including comorbidities [18, 27, 28] Epidural abscesses [40, 41]
[12–14, 42, 55] periorbital cellulitis [45, 46] Acute nontraumatic native septic
joints [48]
Isolated fever [5, 6] Lower urinary tract infection … Acute pyelonephritis Meningitis [6]
[19, 20] [29, 30, 49, 50]
… … … Cholangitis [32, 33] …
Pyogenic liver abscess [34]
… CAP [6, 22, 23, 51–53] VAP [25, 26] Severe CAP [31] …
HCAP [21, 22, 52, 56]
… … … Nonvascular shunt Ventriculoatrial shunt infections
infections [35] [35]
… … … Severe sepsis [54, 57] Septic shock [6]
Shaking chills in febrile pa- Catheter-related bloodstream
tient [6] infections
Abbreviations: CAP, community-acquired pneumonia; HCAP, healthcare-associated pneumonia; VAP, ventilator-associated pneumonia; VO, vertebral osteomyelitis.
Major Limi-
Study Details Inclusion Criteria Exclusion Criteria No. Analyzed Results Conclusions tations
Shi et al [67], 2019, Age ≥ 18 y admitted with UTI Patients without 306 patients • Clinical characteristics: 72% women, median age 70 y, 33% malig- Routine FUBCx in UTI are not needed, Retrospec-
single hospital, and had > 1 BCx drawn bacteremia or nancy, 30% complicated UTI, 18% admission to ICU consider in those without clinical tive
Korea other source of • Positive FUBCx: 18% (55/306) response
bacteremia • Factors associated with positive FUBCx:
Staphylococcus aureus, malignancy, ICU admission, time to
defervescence > 48 h
Canzoneri et al [65], Age ≥ 18 y; repeat BCx ≥ 24 h Fungemi, false- 383 patients • Clinical characteristics: 43% in ICU; 43% had a CVC; 24% on he- To obtain 1 positive FUBCx in GN Low inci-
2017, retrospec- after an initial true-positive positive BCx modialysis; 9% neutropenic bacteremia, 17 FUBCx were needed dence of
tive observa- BCx between January and • Positive FUBCx: overall 14% (55/383); GN bac-
tional, tertiary December 2015 78% (43/55) GP; 15% (8/53) GN teremia;
care hospital, • Factors associated with positive FUBCx with GP: diabetes mellitus; source
United States ESRD on hemodialysis; fever when cultures drawn, presence of a control
CVC was not
• Factors associated with positive FUBCx with GN: fever when cul- reported
tures drawn
• Mortality and need for ICU care were similar among 3 groups: bac-
teremia cleared, persistent BCx with either GP or with GN
Miyamoto et al [63], Adult patients with ≥ 2 BCx None 64 patients • Clinical characteristics: 43% diabetes, 26% kidney disease Lack of documentation of MRSA Small
2017, retrospec- positive for MRSA within • Probability of survival at 14 d of initial positive BCx: 100% for those bacteremia clearance is associ- sample
tive, academic 2–4 d of initial BCx col- who had a negative FUBCx; 57% for those who had only positive ated with a 3-fold increased risk of size
hospital, United lection between 2011 and FUBCx; 41% for those who had no FUBCx mortality
States 2016
Wiggers et al [66], Age ≥ 17 y with a FUBCx Subsequent 701 patients • Clinical characteristics: 18% admitted to hematology/oncology FUBCx useful in patients with S. au-
2016, retrospec- between 2010 and 2014. bacteremias for • Positive FUBCx: overall 16.8% (118/701); 76.3% (90/118) GP; reus bacteremia, endovascular or
tive and a nested Case: positive FUBCx with a single patient 22.9% (27/118) GN; 1.7% (2/118) anaerobes; 7.6% (9/118) spinal infections, lack of source
case-control anal- the same bacteria 2–7 d polymicrobial control
ysis, academic after initial positive BCx; • Factors associated with positive FUBCx: S. aureus, endovascular
hospital, Canada control: negative FUBCx sources of infection (catheter-associated/device/graft infection);
epidural abscess/discitis; lack of source control within 48 h of initial
positive BCx
• Factors associated with negative FUBCx: UTI source; Escherichia
coli, Streptococcus pneumoniae, Viridans group streptococci,
β-hemolytic streptococci
• Clinical variables at time of FUBCx collection with no impact on
FUBCx results: SIRS; leukocytosis; fever; physician concern for
instability
• Positive FUBCx increased risk of 7- and 30-d mortality
Kang et al [68], Age ≥ 18 y with a first Recurrent 862 patients • Positive FUBCx: 7.2% (62/862) Routine FUBCx are not recom-
2013, 1:3 case- episode of Klebsiella Klebsiella or (186 con- • Factors associated with positive FUBCx: IAI; high CCI; SOT; unfa- mended for most K. pneumoniae
control study pneumoniae polymicrobial trols, 62 vorable treatment response on day 2 following initial BCx (fever, bacteremias; consider in those with
according to age between 2007 and 2011 infections positive leukocytosis and lack of CRP decrease constituted “unfavorable high CCI score, SOT, and/or unfavor-
and sex, 2 tertiary FUBCx treatment response”) able treatment response on day 2
care hospitals, cases) • A scoring system was created (IAI: 1 point; nosocomial after initial BCx
Korea K. pneumoniae: 2 points; fever: 3 points; lack of CRP decrease: 2
points): positive FUBCx: > 50% when score > 5 and < 5% when
score was 0–1
Abbreviations: BCx, blood culture; CCI, Charlson Comorbidity Index; CRP, C-reactive protein; CVC, central venous catheter; ESRD, end-stage renal disease; FUBCx, follow-up blood culture; GN, gram-negative; GP, gram-positive; IAI, intra-abdominal infection;
were not
occurred
Major Limi-
reported
or whether
control
source
status
by im-
mune
either positive or negative FUBCx (likely due to small sample
Results
tations
size). Persistent bacteremia was observed in 7% of patients
with Klebsiella pneumoniae bacteremia and was associated with
ICU, intensive care unit; MRSA, methicillin-resistant Staphylococcus aureus; SIRS, systemic inflammatory response syndrome; SOT, solid organ transplant; UTI, urinary tract infection.
teremia; transplant recipient; neutropenia; nosocomial origin
DISCUSSION
cases (2 BCx)
(199 BCx)
Not reported
single hospital,
López Dupla et al