Professional Documents
Culture Documents
t
ip
1
Department of Pediatrics, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA;
cr
2
Department of Emergency Medicine, Johns Hopkins University School of Medicine, Baltimore,
us
Maryland, USA; 3Department of Anesthesia and Critical Care Medicine, Children’s Hospital of
ptamma1@jhmi.edu
© The Author(s) 2021. Published by Oxford University Press for the Infectious Diseases
Society of America. All rights reserved. For permissions, e-mail:
journals.permissions@oup.com.
Summary: In a multicenter, observational cohort of 3,376 patients ≥13 years with bloodstream
infections, the administration of a β-lactam agent prior to vancomycin was protective against 48-
hour and 7-day mortality, using inverse probability of treatment weighting incorporating propensity
t
ip
cr
us
an
M
e d
pt
ce
Ac
Abstract
Objective: Prompt initiation of antibiotic therapy improves the survival of patients with bloodstream
infections (BSI). We sought to determine if the sequence of administration of the first dose of
Methods: We conducted a multicenter, observational study of patients ≥13 years with BSIs to
t
evaluate the association of the sequence of antibiotic administration with 7-day mortality using
ip
inverse probability of treatment weighting (IPTW) incorporating propensity scores. Propensity scores
cr
were generated based on: demographics, Pitt bacteremia score, ICU status, highest lactate, highest
us
WBC count, Charlson Comorbidity index, severe immunocompromise, administration of active
empiric therapy, combination therapy, and time from emergency department arrival to first
an
antibiotic dose.
M
Results: Of 3,376 eligible patients, 2,685 (79.5%) received a β-lactam and 691 (20.5%) received
vancomycin as their initial antibiotic. In the IPTW cohort, exposed and unexposed patients were
d
similar on all baseline variables. Administration of a β-lactam agent prior to vancomycin protected
e
against 7-day mortality (aOR 0.48 (95% CI: 0.33-0.69)]. Similar results were observed when
pt
evaluating 48-hour mortality (aOR 0.45 [95% CI: 0.24-0.83]). Administration of vancomycin prior to a
ce
β-lactam was not associated with improved survival in the subgroup of 524 patients with methicillin-
β-lactam over vancomycin may reduce early mortality, underscoring the significant impact of a
Bloodstream infections (BSI) are associated with significant morbidity and mortality1,2. Prompt
initiation of antibiotic therapy improves patient survival3. While attention has been dedicated to the
the sequence of antibiotic administration (i.e., the antibiotic administered first when the suspicion
t
ip
cr
β-lactam agents such as cefepime, piperacillin-tazobactam, or meropenem are generally
administered to critically ill patients with suspected BSIs. In addition to providing broad Gram-
us
negative activity, these β-lactam agents provide sufficient coverage for a number of clinically
an
relevant Gram-positive organisms such as Streptococcus pyogenes and methicillin-susceptible
even in regions with relatively high rates of MRSA infections, the number of BSIs caused by
d
pathogenic organisms likely to be adequately covered with a β-lactam antibiotic (e.g., cefepime) is
e
Gram-negative BSIs are associated with increased early mortality relative to those caused by Gram-
Ac
positive organisms, likely because of the endotoxin-induced inflammatory response and the
association between certain host conditions such as immunocompromise and increased risk for
minutes or even as a bolus infusion - vancomycin is standardly infused over 60 minutes. The
reaction9 potentially further delaying β-lactam administration by several hours when additional lines
are being preserved for fluids, vasopressors, or other medications, resulting in inadequate coverage
for Gram-negative bacterial pathogens. Our objective was to determine whether administration of a
broad-spectrum β-lactam before the infusion of vancomycin, as the first dose of antibiotic therapy, is
Methods
t
ip
Study Population. We conducted a retrospective observational study of patients 13 years of age and
cr
older with a bacterial BSI between July 2016 and June 2020 and hospitalized at any of five hospitals
in the Johns Hopkins Health System located in the Baltimore-District of Columbia region. Unique
us
patients with a BSI caused by Gram-negative or Gram-positive organisms were included. This study
an
was approved by the Institutional Review Boards of each of the participating institutions, with
The primary exposure was receipt of a β-lactam agent (including expanded-spectrum penicillins,
e
the first dose of antibiotic therapy was administered. Unexposed patients were those who received
ce
vancomycin prior to the β-lactam. Antibiotic administration times were determined according to
timestamps created by nursing barcode medication administration. The primary outcome was
Ac
mortality within 7 days from the time of blood culture collection. The secondary outcome was
mortality within 48 hours from the time of blood culture collection. Day 1 was defined as the day the
bacterial genus and species causing bacteremia and before it is even known if a patient has
known to clinicians when antibiotic administration decisions are made, we included all bacterial BSIs,
including polymicrobial BSIs, regardless of whether they were susceptible to the empiric therapy
t
administered. The lack of or poor activity of the empiric antibiotic regimen (whether due to intrinsic
ip
or acquired resistance) was accounted for in the generation of propensity scores.
cr
us
Patients meeting the following criteria were excluded: (1) those that did not complete
an
administration of at least one dose of both a β-lactam and vancomycin within the first 6 hours after
the first positive blood culture was collected; (2) those who received a Gram-negative agent and
M
vancomycin with overlapping infusion times; (3) those already receiving antibiotic therapy at the
time blood cultures were obtained; (4) those with organisms likely to be contaminants (i.e.,
d
bacteria are unlikely to contribute to early death; and (5) those with non-bacterial organisms
pt
Data Collection. All five hospitals use Epic electronic health record software. Data were collected by
electronic data extraction. Verification of the accuracy of all electronically extracted data elements
was conducted on a subset of 150 patient charts. For variables that could not be obtained
electronically or for variables where there were discrepancies between electronic extraction and
chart review, manual chart review was performed. The following data were collected on all patients:
(1) demographic data; (2) pre-existing medical conditions (Charlson Comorbidity Index10, severe
immunocompromise); (3) severity of illness at the time of blood culture collection (Pitt bacteremia
cardiac arrest, and mental status], ICU admission on day 1 of blood culture collection, highest lactate
and species, antibiotic susceptibility data); (5) detailed antibiotic data (antibiotic infusion times,
administration of active empiric antibiotic therapy within the first 24 hours of therapy (based on
adjudication by two physicians [JA & PDT]), use of combination Gram-negative therapy (i.e.,
t
ip
administration of two antibiotics such as a β-lactam and aminoglycoside within the first 24 hours of
cr
therapy, regardless of whether one or both agents were active against the pathogen); and (6) vital
status up to day 7. Current Clinical and Laboratory Standards Institute criteria were used to define
Severe immunocompromise was defined by the presence of at least one of the following criteria: (1)
hematopoietic stem cell transplantation within the previous 12 months or active treatment for graft-
d
versus-host disease, (2) active chemotherapy within the prior 6 months, (3) previous solid organ
e
transplantation, (4) human immunodeficiency virus infection with a CD4 count under 200 cells/mm 3,
pt
(5) absolute neutrophil count under 500 cells/mm3 at the time of or within 7 days after blood culture
ce
collection, or (6) receipt of corticosteroids at a dose equivalent to 10 mg daily of prednisone for ≥14
for patients suspected of having BSIs occurs at random or is a deliberate decision. Regardless, to
patients who received vancomycin first) were as similar as possible at the time of antibiotic
t
was used to account for selection bias regarding the prescribed treatment sequence.
ip
cr
The following covariates were included in generating the propensity score based on information
us
from day 1: age, gender, Charlson Comorbidity Index, severe immunocompromise, highest Pitt
an
bacteremia score, highest lactate, highest WBC, administration of in vitro active empiric antibiotic
therapy, use of empiric combination therapy, and time from emergency department arrival to the
M
Patients in the exposed group were weighted by the inverse of the propensity score and patients in
pt
the unexposed group were weighted by the inverse of 1 minus the propensity score (i.e., each
ce
person was weighted by the inverse probability of receiving the treatment he/she actually received).
A new weighted pseudopopulation was created in which individuals in the exposed and unexposed
Ac
groups were upweighted or downweighted to ensure that at baseline, both groups were as similar as
possible on all variables, except the sequence of antibiotic administration. Standardized mean
differences were calculated in the new pseudopopulation to ensure similarity between exposed and
Wilcoxon Rank Sum test for continuous variables. In the IPTW cohort, odds ratios and 95%
confidence intervals for 7-day mortality was estimated using weighted regression, adjusting for
between exposed and unexposed groups for that specific variable. A doubly robust estimation was
used to increase the precision of the odds ratio (OR) estimates by additional adjustment for any
potential confounders14. An identical analysis was conducted on the IPTW cohort evaluating the
t
ip
odds of mortality within 48 hours. Statistical analysis was completed using STATA version 17.0
cr
(StataCorp, College Station, United States) and R version 3.6.1 (R Foundation for Statistical
us
an
M
Results
d
Overall Cohort. A total of 5,514 unique patients 13 years of age or older with BSI between July 2016
e
to June 2020 were screened for inclusion and 3,376 patients met eligibility criteria. Common reasons
pt
for exclusion were overlapping administration times of both a β-lactam and vancomycin (1,282,
59.9%) and administration of one agent but not the other (412, 19.3%). Of the 3,376 eligible
ce
patients, 2,685 (79.5%) received a β-lactam agent and 691 (20.5%) received vancomycin as their first
Ac
dose of antibiotic therapy. Initial blood cultures were collected in the emergency department for
2,709 (80.2%), in hospital wards for 421 (12.5%), and in ICUs for 246 (7.3%) patients. The most
the three most common bacteria being Staphylococcus aureus (1,239, 22.5%), Escherichia coli
(1,148, 20.8%), and Klebsiella pneumoniae (764, 13.9%). Overall, 606 (11%) of patients experienced
days 1 through day 7. Table 1 displays 48-hour, 7-day, and 30-day mortality by infecting bacterial
species. The bacterial species associated with the highest mortality within the first 7 days after blood
culture collection were A. baumannii (16.3%), P. aeruginosa (12.1%), and Streptococcus pyogenes
t
ip
(12.1%).
cr
us
Baseline Characteristics. Baseline characteristics of patients with a BSI in the exposed and
unexposed groups in the cohort of 3,376 eligible patients are displayed in Table 2. Some notable
an
baseline differences existed in the full cohort. Exposed patients had lower Pitt bacteremia scores (2
M
[IQR 1-3] vs. 2 [IQR: 1-4]); higher lactate levels (2.5 mmol/L [IQR: 1.6-4.3] vs. 2.4 mmol/L [IQR: 1.5-
4.5] – although of questionable clinically significance; and lower Charlson Comorbidity indices (6
d
[IQR 3-10] vs. 7 [3-10]), compared with unexposed patients. In the new cohort derived using IPTW
e
propensity score analysis, none of these differences persisted and the two groups were similar on all
pt
co-variates with standardized mean differences <0.10 (Figure 1). Approximately 41% of patients in
ce
both study arms in the IPTW cohort were admitted to the ICU at the time of or within 24 hours of
died within 7 days including 186 (6.9%) in the β-lactam group and 74 (10.7%) in the vancomycin
side of conservative estimates, a doubly robust analysis accounting for all variables originally
included in the generation of propensity scores to derive an adjusted OR for 7-day mortality was
t
used.
ip
cr
Administration of a β-lactam agent prior to administration of vancomycin was protective against 7-
us
day mortality with an OR of 0.68 (95% CI: 0.50-0.92); Table 3. The benefit of initial β-lactam
an
administration persisted in the adjusted model (aOR 0.48 [95% CI: 0.22-0.69]. This analysis was
repeated using an identical approach to determine the odds of mortality within 48 hours of blood
M
culture collection and similar results were obtained (aOR 0.45 [95% CI 0.24-0.83]). Not surprisingly,
other variables independently associated with an increased odds of 7-day mortality included (1) each
d
additional unit of the Pitt bacteremia score (aOR 1.26 [95% CI 1.17-1.37]) and (2) each unit increase
e
There were 524 patients with MRSA BSI, of which 380 (72.5%) received a β-lactam first and 144
Ac
(27.5%) received vancomycin first (p=0.12). In the subgroup of patients with MRSA BSI, the aOR of
death within 7-days for patients who received a β-lactam prior to vancomycin was 0.93 (95% CI 0.33-
2.63). Although administering a β-lactam agent first is not protective against mortality for patients
with MRSA BSI, neither is administering vancomycin first – supporting the overall study findings of
prioritizing β-lactam therapy when there is a heightened suspicion for a BSI and the infecting
pathogen is unknown.
Discussion
Using a contemporary cohort of 3,376 patients with BSIs across five hospitals, we found that when
vancomycin, the odds of 7-day mortality may be reduced by 52%. Data from the SENTRY
Antimicrobial Surveillance Program indicate that there are approximately 13,245 BSI per year6.
t
Extrapolating our study findings, prioritizing initial β-lactam administration has the potential to save
ip
737 lives per year, underscoring the significant impact of a relatively simple practice change.
cr
us
Before undertaking this work, we hypothesized that prioritizing β-lactam administration would
an
improve survival in patients with BSIs for three major reasons. First, the prevalence of bacterial
pathogens causing BSIs susceptible to broad-spectrum β-lactams is higher than the prevalence of
M
pathogens not susceptible to broad-spectrum β-lactams6. Second, significant endotoxin release with
d
the first dose of antibiotic therapy is a known phenomenon that may occur during the treatment of
e
Gram-negative pathogens15-21. The likelihood of a severe cytokine storm from endotoxin release is
pt
likely to be more dramatic and contribute to poor outcomes as the bacterial burden increases, as
theoretically occurs in the absence of prompt antibiotic therapy15-21. Finally, patients with severe
ce
immunocompromise, particularly those with profound neutropenia, are at higher risk of mortality
Ac
when infected with bacterial pathogens compared to patients with competent immune systems and
the former are more likely to be infected with Gram-negative bacteremia rather than MRSA
For patients presenting with septicemia, the likelihood that a broad-spectrum β-lactam will provide
adequate coverage for the causative pathogen is higher than for vancomycin. Evaluating 20 years of
data from the SENTRY Antimicrobial Surveillance Program, broad-spectrum β-lactam coverage is
anticipated to be effective for over 90% of all pathogens recovered in blood cultures. As an example,
cefepime covers a range of Gram-negative bacteria but also has activity against Streptococcal
organisms, and this includes organisms like Enterococcus faecalis and Staphylococcus epidermidis
which are unlikely to be independently associated with mortality, and less so early mortality35,36.
Data indicate that since 2005, there has been a decline in the rates of MRSA BSI in the United States,
t
ip
likely due to successful efforts in reducing device-associated infections and interruptions in
cr
comprised 44% of bacterial BSIs, less than 10% of all BSIs were caused by MRSA.
us
an
Second, there is the potential for significant endotoxin release associated with antibiotic-induced
coli sepsis rat model demonstrated a 20-fold increase in plasma endotoxin concentrations after the
d
initiation of antibiotic therapy compared with no antibiotic therapy31. Bacterial endotoxin release
e
can result in a pro-inflammatory cascade, in turn resulting in metabolic acidosis, acute hypotension,
pt
and severe clinical disease15,16. Theoretically, the impact of these downstream effects likely increases
ce
as delays in antibiotic therapy result in increases in the bacterial burden, highlighting the importance
Finally, some of the early mortality observed with Gram-negative BSIs are likely attributable to the
antibiotic therapy, such as those with hematologic malignancies or bone marrow transplant
recipients, can rapidly progress leading to life-threatening complications, such that prophylactic
fluoroquinolone use is often employed to mitigate this risk32,33. Over the past two decades there has
been a clear shift to a significantly higher portion of Gram-negative organisms infecting the
t
ip
cr
There are several limitations to our study. First, we selected 7-day mortality as the primary outcome
us
since this endpoint is more likely to be attributable to the initial choice of antibiotic therapy, in
contrast to a 30-day or 90-day endpoint where the first dose of antibiotics is of less significance.
an
While other factors in the first seven days could also have impacted mortality (e.g., poor fluid
M
management, lack of source control, failure to tailor the antibiotic regimen based on antibiotic
susceptibility data), we found similar results when evaluating 48-hour mortality, bolstering the
d
argument that prioritizing β-lactam therapy is protective against early mortality. Second, due to the
e
retrospective nature of our study, we were unable to account for all variables that impact the
pt
We attempted to mitigate some of the confounding by indication with the development and
incorporation of propensity scores. The use of IPTW incorporating propensity scores led to the
Ac
exclusion of some patients in the overall cohort; however, it ensured the two comparator groups
were very similar with regards to demographic data, pre-existing medical conditions, severity of
illness at the onset of bacteremia, and the administration of effective antibiotics. We believe the
cohort analyzed in this study is generalizable to the population at risk for BSI writ large. However, we
cannot exclude the possibility of additional, unmeasured residual factors that could impact the
association between the first administered antibiotic and mortality. Finally, antibiotics administered
at other facilities prior to the first day of blood culture collection at the five participating hospitals
were not captured. This could potentially have resulted in misclassification of exposure surrounding
t
Our study suggests that when presented with a critically ill patient who may be experiencing a BSI, it
ip
is prudent to administer the β-lactam agent first, if both agents are clinically indicated and cannot be
cr
administered simultaneously. Future studies are needed to investigate operational factors
us
associated with antibiotic sequence, if findings similar to ours are identified in a larger,
JA, EYK, KC, and PDT report no disclosures. SEC receives consulting fees from Novartis, Theravance
and Basilea, outside the submitted work. None of the other authors report any conflicts of interest.
This work was funded by the CDC’s Prevention Epicenter Program under grants U01CK000554-02-02
t
ip
CK20-004 and U54CK000617-01-00. The content is solely the responsibility of the authors and does
cr
not necessarily represent the official view of the funding agency.
us
an
M
e d
pt
ce
Ac
References
1. Nielsen SL. The incidence and prognosis of patients with bacteremia. Dan Med J.
2015;62(7).
2. Massart N, Wattecamps G, Moriconi M, Fillatre P. Attributable mortality of ICU
t
sepsis: does sequence impact outcome? Crit Care Med. 2015;43(3):e102.
ip
5. Strich JR, Heil EL, Masur H. Considerations for Empiric Antimicrobial Therapy in
Sepsis and Septic Shock in an Era of Antimicrobial Resistance. J Infect Dis.
cr
2020;222(Suppl 2):S119-S131.
6. Diekema DJ, Hsueh PR, Mendes RE, et al. The Microbiology of Bloodstream Infection:
20-Year Trends from the SENTRY Antimicrobial Surveillance Program. Antimicrob
7.
Agents Chemother. 2019;63(7).
us
Thaden JT, Park LP, Maskarinec SA, Ruffin F, Fowler VG, Jr., van Duin D. Results from
a 13-Year Prospective Cohort Study Show Increased Mortality Associated with
an
Bloodstream Infections Caused by Pseudomonas aeruginosa Compared to Other
Bacteria. Antimicrob Agents Chemother. 2017;61(6).
8. Eng RH, Smith SM, Fan-Havard P, Ogbara T. Effect of antibiotics on endotoxin release
M
10. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying
prognostic comorbidity in longitudinal studies: development and validation. J Chronic
e
Dis. 1987;40(5):373-383.
pt
11. Rhee JY, Kwon KT, Ki HK, et al. Scoring systems for prediction of mortality in patients
with intensive care unit-acquired sepsis: a comparison of the Pitt bacteremia score
and the Acute Physiology and Chronic Health Evaluation II scoring systems. Shock.
ce
2009;31(2):146-150.
12. CLSI. performance standards for antimicrobial susceptibility testing. CLSI M100.
clinical and laboratory standards institute. 31st ed., 2021.
Ac
13. Austin PC, Stuart EA. Moving towards best practice when using inverse probability of
treatment weighting (IPTW) using the propensity score to estimate causal treatment
effects in observational studies. Stat Med. 2015;34(28):3661-3679.
14. Amoah J, Stuart EA, Cosgrove SE, et al. Comparing Propensity Score Methods Versus
Traditional Regression Analysis for the Evaluation of Observational Data: A Case
Study Evaluating the Treatment of Gram-Negative Bloodstream Infections. Clin Infect
Dis. 2020;71(9):e497-e505.
15. Narita K, Koide N, Morikawa A, et al. Differential release of endotoxin from
Pseudomonas aeruginosa treated with beta-lactam antibiotics and its effect on the
lethal activity. Jpn J Med Sci Biol. 1997;50(6):233-239.
16. Morrison DC, Bucklin SE. Evidence for antibiotic-mediated endotoxin release as a
contributing factor to lethality in experimental gram-negative sepsis. Scand J Infect
Dis Suppl. 1996;101:3-8.
17. Walterspiel JN, Kaplan SL, Mason EO, Jr. Protective effect of subinhibitory polymyxin
B alone and in combination with ampicillin for overwhelming Haemophilus
influenzae type B infection in the infant rat: evidence for in vivo and in vitro release
t
endotoxin by bacteria retained on intravenous inline filters. Am J Hosp Pharm.
ip
1975;32(4):378-380.
21. Strutz F, Heller G, Krasemann K, Krone B, Muller GA. Relationship of antibodies to
cr
endotoxin core to mortality in medical patients with sepsis syndrome. Intensive Care
Med. 1999;25(5):435-444.
22. Perez F, Adachi J, Bonomo RA. Antibiotic-resistant gram-negative bacterial infections
23.
us
in patients with cancer. Clin Infect Dis. 2014;59 Suppl 5:S335-339.
Trecarichi EM, Tumbarello M. Antimicrobial-resistant Gram-negative bacteria in
febrile neutropenic patients with cancer: current epidemiology and clinical impact.
an
Curr Opin Infect Dis. 2014;27(2):200-210.
24. Montassier E, Batard E, Gastinne T, Potel G, de La Cochetiere MF. Recent changes in
bacteremia in patients with cancer: a systematic review of epidemiology and
M
29. Klein EY, Jiang W, Mojica N, et al. National Costs Associated With Methicillin-
Susceptible and Methicillin-Resistant Staphylococcus aureus Hospitalizations in the
United States, 2010-2014. Clin Infect Dis. 2019;68(1):22-28.
30. Perlin JB, Hickok JD, Septimus EJ, Moody JA, Englebright JD, Bracken RM. A bundled
approach to reduce methicillin-resistant Staphylococcus aureus infections in a
system of community hospitals. J Healthc Qual. 2013;35(3):57-68; quiz 68-59.
31. Horii T, Kimura T, Nadai M, Kobayashi M. Lincomycin-induced endotoxin release in
Escherichia coli sepsis: evidence for release in vitro and in vivo. Int J Infect Dis.
2000;4(3):118-122.
32. Gafter-Gvili A, Fraser A, Paul M, et al. Antibiotic prophylaxis for bacterial infections in
afebrile neutropenic patients following chemotherapy. Cochrane Database Syst Rev.
2012;1:CD004386.
33. Leibovici L, Paul M, Cullen M, et al. Antibiotic prophylaxis in neutropenic patients:
new evidence, practical decisions. Cancer. 2006;107(8):1743-1751.
34. Mikulska M, Viscoli C, Orasch C, et al. Aetiology and resistance in bacteraemias
t
staphylococcal bacteremia in febrile neutropenia: A cohort study. Can J Infect Dis
ip
Med Microbiol. 2014;25(1):e14-17.
cr
us
an
M
e d
pt
ce
Ac
Figure Legend
Figure 1: Standardized mean biases comparing the full eligible cohort and the inverse probability of
treatment weighted propensity score incorporated cohort in a study of patients 13 years of age and
t
ip
cr
us
an
M
e d
pt
ce
Ac
Table 1: Mortality associated with the most frequently recovered bacterial organisms causing bloodstream
1,2,3
infections in a cohort of 5,514 patients
t
Klebsiella pneumoniae 764 (13.9) 23 (3.0) 65 (8.5) 153 (20.0)
ip
Proteus mirabilis 128 (2.3) 2 (1.6) 11 (8.6) 18 (14.1)
Pseudomonas aeruginosa 380 (6.9) 16 (4.2) 46 (12.1) 73 (19.2)
cr
Serratia marcescens 168 (3.0) 5 (3.0) 16 (9.5) 25 (14.9)
Stenotrophomonas maltophilia 136 (2.5) 2 (1.5) 5 (3.7) 20 (14.7)
Gram-positive organisms 2,476 (44.9) 84 (3.4) 162 (6.5) 295 (11.9)
Enterococcus faecalis
Enterococcus faecium
Methicillin-susceptible Staphylococcus
424 (7.7)
261 (4.7)
715 (13.0) us 7 (1.7)
6 (2.3)
35 (4.9)
14 (3.3)
13 (5.0)
69 (9.7)
23 (5.4)
28 (10.7)
121 (16.9)
an
aureus
Methicillin-resistant Staphylococcus 524 (9.5) 14 (2.7) 29 (5.5) 60 (11.5)
aureus
M
1
Excludes organisms highly likely to be contaminants (i.e., coagulase-negative Staphylococcus, Diphtheroids,
e
at the time of initial blood culture collection or within the subsequent 7 days; numbers in table are greater
than 5,514
3
Organisms identified in <1% of blood cultures not included
ce
Ac
Table 2: Characteristics of 3,376 patients 13 years of age and older with bloodstream infections
t
(IQR) years (51-73) 5 years (54-76) 0
ip
(52-76) (53-
76)
Female, n (%) 1,194 299 0.57 -0.061 1,159 296 0.87 0.007
cr
(44.5) (43.3) 2 (55.3) (54.9) 4
Pitt bacteremia 2 2 0.00 -0.170 2 2 0.62 0.023
us
score, median (1-3) (1-4) 0 (1-3) (1-3) 3
(IQR)
Highest lactate, 2.5 2.4 0.04 -0.072 2.6 2.3 0.84 0.010
median (IQR) mmol/ mmol/L 5 mmol mmol/L 3
an
L (1.5-4.5) /L (1.4-4.1)
(1.6- (1.6-
4.3) 4.6)
M
26,300) (8,80
e
0-
26,90
pt
0)
Severe 317 81 0.95 -0.002 257 67 0.93 -0.006
immunocompro (11.8) (11.7) 1 (12.3) (12.4) 5
ce
mise, n (%)
Intensive care 956 286 0.00 -0.143 875 223 0.84 0.010
unit, n (%) (35.6) (41.4) 5 (41.7) (41.2) 2
Ac
Time in hours 1.6 2.7 <0.0 -0.145 2.0 2.1 0.88 -0.008
from emergency (1.0- (1.4-3.5) 01 (1.1- (1.3-2.9) 7
department 2.6) 2.7)
arrival to first
dose of
antibiotic
administration2,
median (IQR)
Charlson 6 7 0.01 -0.079 7 7 0.99 -0.000
Comorbidity (3-10) (3-10) 8 (3-10) (3-10) 6
Index, median
(IQR)
Active antibiotic 1,971 525 0.17 -0.043 1535 393 0.88 0.007
therapy, n (%) (73.4) (76.0) 0 (73.2) (72.9) 5
t
ip
cr
us
an
M
e d
pt
ce
Ac
Table 3: Univariable and multivariable analysis of mortality at 7 days for 3,376 patients 13 years of
age and older with bloodstream infections, using an inverse probability of treatment weighted
cohort based on propensity scores
t
blood cell count
ip
Severe 0.83 (0.53-1.30) 0.412 1.18 (0.69-2.03) 0.544
immunocompromise
cr
Intensive care unit 3.38 (2.53-4.52) <0.001 1.30 (0.88-1.92) 0.188
admission
Charlson Comorbidity 0.99 (0.95-1.00) 0.138 0.96 (0.92-1.00) 0.051
us
Index
Severe 0.83 (0.53-1.30) 0.412 1.18 (0.69-2.03) 0.544
immunocompromise
an
Active antibiotic therapy 1.47 (1.05-2.05) 0.025 1.25 (0.84-1.86) 0.275
Combination Gram- 1.48 (0.72-3.04) 0.283 0.94 (0.44-1.99) 0.865
negative therapy
1
Doubly robust analysis using the inverse probability of treatment weighted cohort based on
M
propensity scores, with additional adjustment for all co-variates included in the table.
e d
pt
ce
Ac
Downloaded from https://academic.oup.com/cid/advance-article/doi/10.1093/cid/ciab865/6381106 by Universidad de Antioquia user on 20 October 2021
t
ip
cr
us
an
Figure 1
M
d e
pt
ce
Ac