Heart & Lung xxx (2016) 1e6

Contents lists available at ScienceDirect

Heart & Lung

journal homepage: www.heartandlung.org

De-escalation of empiric antibiotics in patients with severe

sepsis or septic shock: A meta-analysis
Ying Guo, MD a, Wei Gao, MD a, Hongxia Yang, MD a, Cheng’en Ma, MD a,
Shujian Sui, MD b, *
a
Department of Intensive Care Unit, The Second Hospital of Shandong University, China
b
Department of Cardiovascular Medicine, The Second Hospital of Shandong University, China

a r t i c l e i n f o a b s t r a c t

Article history: Objective: To evaluate the impact of de-escalation therapy on clinical outcomes in patients with severe
Received 31 January 2016 sepsis and/or septic shock.
Received in revised form Methods: We performed a systematic literature search in PubMed, EMBASE, Web of Science, and CEN-
29 May 2016
TRAL on The Cochrane Library. The search terms used were “sepsis,” “septic shock” and “de-escalation.”
Accepted 1 June 2016
Available online xxx
The relative risk (RR) with 95% confidence intervals (CI) was used to evaluate the impact of de-escalation
therapy on clinical outcomes.
Results: Nine individual studies (1873 patients) were included. Mortality trended lower in the de-esca-
Keywords:
Antibiotics
lation group as compared with the continuation of broad-spectrum antibiotics group. However, the
Severe sepsis results were not statistically significant (RR ¼ 0.74, 95% CI 0.54e1.03).
Septic shock Conclusion: Antibiotic de-escalation therapy has no detrimental impact on mortality in patients with
De-escalation severe sepsis and/or septic shock, as compared to the continuation of broad-spectrum antibiotics. Since
Empirical de-escalation affords an opportunity to limit overuse of broad-spectrum antibiotics, it should be
considered as an option in clinical practice.
Ó 2016 Elsevier Inc. All rights reserved.

Introduction Numerous studies have been conducted to evaluate the clinical

Severe sepsis and septic shock are common causes of mortality
in hospital, especially in critically ill patients.1 Early, aggressive,
empiric and broad-spectrum antimicrobial therapy, targeted at the
most likely pathogens before microbiological information is avail-
able, is recommended by the current guidelines.2 Adequate anti-
microbial therapy has been shown to reduce the mortality rates.3
However, overuse of broad-spectrum antibiotics therapy may
lead to increased selection pressure, promoting antimicrobial
resistance. Therefore, antibiotics de-escalation, a strategy consist of
switching from broad-spectrum antimicrobials (initial treatment)
to a narrower spectrum as soon as the microbiological results are
available, has been recommended to minimize the emergence of
drug-resistant organisms, as well as costs, during the treatment of
patients with sepsis.2
impact and safety of de-escalation therapy in patients with severe
sepsis and/or septic shock. However, the current evidence avail-
able is not consistent. Several studies concluded that de-
escalation therapy was a protective factor associated with lower
mortality in patients with severe sepsis,4,5 while other studies
have shown that de-escalation did not effect the mortality,6,7 but
increased the number of antibiotic days and the risk of superin-
fection.6 Overall, there is a lack of adequate and firm evidence on
whether de-escalation of empiric antibiotics therapy is effective
for patients with severe sepsis and/or septic shock, due to the
absence of sufficiently powered randomized controlled trials
(RCTs).
Therefore, we performed a meta-analysis of available compar-
ative studies to evaluate the impact of de-escalation therapy on
clinical outcomes in adult patients with severe sepsis and/or septic
shock.

Conflict of interest: None. Materials and methods

* Corresponding author. Department of cardiovascular Medicine, The Second
Hospital of Shandong University, 247 Beiyuan Road, Jinan 250033, China.
Tel.: þ86 15153169977. The process of our meta-analysis followed a prior established
E-mail address: sj_sui@sina.com (S. Sui). protocol.

0147-9563/$ e see front matter Ó 2016 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.hrtlng.2016.06.001
2 Y. Guo et al. / Heart & Lung xxx (2016) 1e6
Literature search
Our meta-analysis was performed according to the PRISMA
(Preferred Reporting Items for Systematic Review and Meta-
Analyses) statement.8 We performed a systematic literature
search in PubMed, EMBASE, Web of Science, and CENTRAL on The
Cochrane Library without time or language limitation. The search
terms used were “sepsis,” “septic shock” and “de-escalation.” Two
reviewers conducted the search strategy independently. We care-
fully reviewed the titles and abstracts of all returned articles and
selected the ones to be read in detail. After further review, all
proper studies satisfied the inclusion criteria (including the refer-
ences listed in these articles) were prudently picked out. Dis-
agreements were resolved by full discussions between the two
reviewers or with another author. The search strategy was lastly
updated on 07 December 2015.
Inclusion criteria
Studies were included if they fulfilled the following criteria: (1)
Participants: adult patients (age  18 years) with severe sepsis and/
or septic shock demanding an empiric antibiotics therapy. Ac-
cording to the current guidelines,2 severe sepsis is defined as the
presence of infection with systemic manifestations (sepsis) plus
sepsis-induced organ failure or tissue hypoperfusion; septic shock
is defined as “severe sepsis plus hypotension not reversed with
fluid resuscitation.” (2) Interventions and comparisons: continua-
tion of broad-spectrum antibiotics therapy and de-escalation of
empiric antibiotics therapy. De-escalation is defined as elimination
of an antibiotics or narrowing the spectrum of antibiotic therapy.9,10
(3) Outcomes: patient mortality at the end of follow-up. (4) Study:
comparative studies, including observational studies and random-
ized clinical trials.
Exclusion criteria
Studies were excluded if they met the following criteria: (1)
Participants were not restricted to adult patients. (2) De-escalation
of empiric antibiotics therapy was not used. (3) Articles reported
data on patients with sepsis but did not report data specifically on
patients with severe sepsis and/or septic shock. (4) Articles did not
report sufficient data to calculate mortality after broad-spectrum
antibiotics therapy or de-escalation of empiric antibiotics therapy.
(5) Review articles, letters without original data, or abstracts with
data that were later published as full text articles.
Data extraction
Two independent reviewers extracted the following data from
included articles: (1) General characteristics, including the study
design, first author, publishing year, geographical origin, sample
size, median age, gender, patients enrollment duration, etc. (2)
Number of patients who died or were alive at the end of follow-up
for both continuation of broad-spectrum antibiotics group and de-
escalation group. (3) Other relevant data, such as the duration of
ICU stay and the number of patients with superinfections, if avail-
able, in the continuation of broad-spectrum antibiotics group and
the de-escalation group.
Quality assessment
Two reviewers independently assessed the methodological
quality of included studies by using the Newcastle-Ottawa Scale
(NOS), which awards a maximum of 9 points for (1) quality of se-
lection (up to 4 points), (2) comparability (up to 2 points) and (3)
exposure or outcome (up to 3 points) of study participants.11 The
qualities of included studies were defined as poor (score, 0e3), fair
(score, 4e6), or good (score, 7e9).12 Disagreements were resolved
by discussions between the two reviewers or with another author.
Statistical analysis
We planned a primary analysis including all patients by using
the relative risk (RR) with 95% confidence intervals (CI) to evaluate
the impact of de-escalation therapy on clinical outcomes in patients
with severe sepsis or septic shock. Random-effects model (the Der
Simonian and Laird method) was used regardless of the results of
heterogeneity. Heterogeneity was examined by Q-statistic test and
I2 index, and was considered significant if p < 0.10 (Q-statistic) or I2
value was 50% and greater.13 The rates of patients with de-
escalation therapy were also calculated by using random-effects
model. Sensitivity analyses were conducted by using the “meta-
inf” program. Publication bias was assessed if enough studies were
included.14All statistical analyses were performed by using the
statistical software Stata 12.0 (StataCorp LP, College Station, TX,
USA).
Results
The initial literature search returned a total of 285 potentially
relevant articles, of which 257 articles were excluded based on the
inclusion and exclusion criteria. Twenty-eight articles were
browsed in detail. We further excluded 9 articles because of
insufficient data, and 1 article because of duplicates. Another 9
articles, which included patients with sepsis but the data provided
was not sufficient to calculate the mortality of patients with severe
sepsis and/or septic shock, were also excluded. Finally, 9 studies
were included for our meta-analysis,4e7,15e19 overall comprising a
total of 1873 patients (729 patients in de-escalation group, 847
patients in no change group, 272 patients in escalation group, and
25 patients in mixed changes group). Fig. 1 shows the flow diagram
of literature search and studies selection.
Relevant arcles returned by the preliminary
stage of literature search (n=285)
257 arcles excluded
-Duplicates (n=73)
-Not fit for inclusion criteria (n=184)
Arcles browsed in detail (n=28)
19 arcles excluded
-Insufficient data (n=9)
-Including paents with sepsis but no
necessary data about paents with
severe sepsis and/or sepc shock
(n=9)
-Duplicates (n=1)
Studies included for meta-analysis
(n=9)
Fig. 1. Flowchart of literature search and studies selection.
 Y. Guo et al. / Heart & Lung xxx (2016) 1e6 3

Table 1 describes the baseline characteristics of the included 9

NOS score

studies. Six studies were performed in Western pop-

ulations,4,6,7,15,16,19 while the other 3 studies in Asian populations.
Patients enrollment periods ranged from December 200415 to May
5

5
9

4
5

7
2013.17 Three studies included subpopulation patients with severe

randomized trial
sepsis and/or septic shock (including patients with septic shock
Retrospective

Retrospective
Retrospective

Retrospective
Study design

Non-blinded
related to community-onset monomicrobial Enterobacteriaceae
Prospective

Prospective
Prospective

Prospective
bacteremia,18 cancer patients with severe sepsis7 and neutropenic
cancer patients with severe sepsis19).There was only 1 randomized
trial in the included 9 studies.6
We used NOS to assess the methodological quality of included
(male %)
Gender

studies. Only 4 studies4,6,16,19 were rated as good quality(7e9

60.5d
66.7b
62.7a

67.3
59.8

64b
64a
56c
NR

points). The primary reason for low methodological quality was the
56

50

52
imbalances between the continuation of broad-spectrum antibi-
Patient’s age

66.8  14.9b

61.7  12.8b
66 (53e75)d
57.9  17.0a

62.5  13.2a
60 (45e71)c

otics group and the de-escalation of empiric antibiotics group in

57  21.5

64.2  16.3

58 (48e67)
61  14

baseline characteristics and severity of illness.

(years)

Our meta-analysis showed that the mortality trended lower in

67.8
NR

the de-escalation group, as compared with the continuation of

broad-spectrum antibiotics group. However, there was no signifi-
Hospital-acquired severe sepsis

enterobacteriaceae bacteremia
Severe sepsis and septic shock
(with or without septic shock)

cant statistical difference (RR ¼ 0.74, 95% CI 0.54e1.03, p ¼ 0.005,

Severe sepsis or septic shock

Neutropenic cancer patients

Cancer patients with severe

I2 ¼ 63.8%. Fig. 2). We performed several subgroup analyses

Septic shock related to

because of the considerable heterogeneity. When we analyzed the 4

studies with high methodological quality (NOS score S 7
with severe sepsis
community-onset

points),4,6,16,19 we found no significant statistical difference in

monomicrobial
Severe sepsis

Severe sepsis

Severe sepsis

mortality between the de-escalation group and the continuation

group, which was consistent with the result of total population.
Object

sepsis

And the test of heterogeneity for this subgroup yielded a p value of

0.373, conventionally interpreted as being non-significant
(RR ¼ 0.85, 95% CI 0.67e1.08, p ¼ 0.373, I2 ¼ 3.9%). We also did
2004.12e2008.12

2012.02e2013.04

2013.01e2013.05
2008.01e2012.05

2010.01e2012.12
2005.01e2012.12

2008.01e2013.03

2008.01e2010.05

subgroup analysis on the 5 prospective studies4,6,15,17,19 and got

Included period

similar results (RR ¼ 0.91, 95% CI 0.75e1.12, p ¼ 0.651, I2 ¼ 0.0%).

For the 6 studies enrolling general patients (patients with severe
sepsis or septic shock who met inclusion criteria but were not
2007

limited to several subpopulation patients7,18,19 as we mentioned

above) with severe sepsis and/or septic shock,4e6,15e17 the analysis
changes

result was consistent with the result of all 9 studies with moderate
Mixed

heterogeneity (RR ¼ 0.82, 95% CI 0.59e1.14, p ¼ 0.066, I2 ¼ 51.6%.

0

17

8
0

0
0

Fig. 3).
Escalation

We failed to compare the duration of ICU stay and the number of

NOS score, Quality assessment score using the Newcastle-Ottawa Scale. NR, Not Reported.

patients with superinfections because few studies focused on these

64

15

0
163

22

topics.
Because of the randomized nature of Leone et al,6 we excluded
change

this study in order to accurately report rates of de-escalation

140

58

57

106
246

58
81

44

57
No

therapy witnessed exclusively in the observational studies. The

rate of patients in which de-escalation was accomplished was 39.5%
No. of patients

De-escalation

(95% CI 0.293e0.497, p ¼ 0.000, I2 ¼ 95.2%). Our subgroup analyses

assessed the influence of heterogeneity on de-escalation rates: (1)
For the 6 studies including general severe sepsis and/or septic
116

79

59

13
219

52
86

61

44

shock patients,4e6,15e17 the pooled de-escalation rate was 34.0%

(95% CI 0.214e0.465, p ¼ 0.000, I2 ¼ 96.0%). (2) For the 5 studies
Study type

performed in Western populations,4,7,15,16,19 the pooled de-

Abstract

Abstract
Full text

Full text

Full text

Full text
Full text

Full text

escalation rate was 42.3% (95% CI 0.353e0.493, p ¼ 0.000,

Letter

I2 ¼ 82.9%). (3) For the 3 studies enrolling general severe sepsis and/
Baseline characteristics of included studies.

or septic shock patients from Western populations,4,15,16 the pooled

Geographical

Saudi Arabia

de-escalation rate was 37.1% with no significant heterogeneity (95%

CI 0.330e0.413, p ¼ 0.156, I2 ¼ 46.2%).
Belgium

Taiwan,
France

France
origin

China
Spain

Spain

Japan

USA

De-escalation group.

Sensitivity analyses
No change group.
Garnacho-Montero
Escoresca-Ortega

Salahuddin 2013

Paskovaty 2015

Death group.
Alive group.

We performed sensitivity analyses for the comparison of mor-

Heenen 2012

Oshima 2015

Mokart 2014
Leone 2014

tality in the de-escalation group and the continuation of broad-

Lee 2015
2009

2014

spectrum antibiotics group, and our results remained stable after

Study
Table 1

omission of any individual study, indicating the credibility of the

b
c
d
a

outcomes in our meta-analysis (see Supplement Fig. 1s).

4 Y. Guo et al. / Heart & Lung xxx (2016) 1e6

Fig. 2. Forest plots showing the results of meta-analyses comparing mortality in the de-escalation group and the continuation group (total populations).

Publication bias strategy currently used in the management of sepsis, especially in

We did not assess the publication bias because the number of
studies included in our meta-analysis was not sufficient to ensure
high statistical power.14
Discussion
The spread of bacterial resistance and the high costs of antimi-
crobial therapy call for a decrease of antibiotic exposure and the
rational use of antibiotics. De-escalation of empiric antibiotics, a
patients with severe sepsis, is one possible way to achieve these
objectives. The lack of adequate and firm evidence requires further
research on this topic, especially high-quality RCTs. However,
considering the diverse conditions of critically ill patients with
severe sepsis and/or septic shock, such RCTs need to be well-
designed. Meta-analysis of non-randomized controlled studies
plays a more and more important role in evidence supports.20
Given the lack of well-designed RTCs in this specific area, meta-
analysis provides a feasible way to assess the impact of de-
escalation.

Fig. 3. Forest plots showing the results of meta-analyses comparing mortality in the de-escalation group and the continuation group (subgroup analysis including 6 studies
enrolling general patients with severe sepsis and/or septic shock).
 Y. Guo et al. / Heart & Lung xxx (2016) 1e6
Nine studies were included in our meta-analysis, overall
comprising of 1873 patients. Although the mortality trended lower
in the de-escalation group as compared to the continuation group,
there was no significant statistical difference. Our study demon-
strates that de-escalation therapy do not negatively effect the
mortality as compared to the continuation of broad-spectrum an-
tibiotics. In view of the high costs and potential side effects of
broad-spectrum antibiotics therapy, this is an encouraging
conclusion.
As part of a global management of antibiotics therapy in patients
with severe sepsis, the strategy termed “de-escalation therapy” was
realized in approximately 35e45% of cases.4,16,21 In our meta-
analysis, the pooled de-escalation rate was 39.5%. This relatively
low rate could be explained by several plausible reasons: reluctance
to change a proven effective antimicrobial regimen, inaccessible
microbiological information, perplexities of the effectiveness and
methods of de-escalate therapy.10 The non-inferiority of de-
escalation therapy as demonstrated in our meta-analysis may
give clinicians a stronger reason to consider this strategy for pa-
tients with severe sepsis and/or septic shock whenever clinically
appropriate. Furthermore, a better marker to guide antibacterial
therapy might help physicians in deciding the duration of antibiotic
therapy. De Jong et al22 performed a randomized, controlled, open-
label trial to access the efficacy and safety of procalcitonin in
guiding antibiotic duration in critically ill patients and arrived a
positive conclusion. This effort may be helpful to the realization of
de-escalation strategy.
In addition to severe sepsis and/or septic shock, de-escalation
therapy has also been assessed in other patient populations. For
example, Shime et al23 concluded that in immunocompetent pa-
tients presenting with bacteremia caused by antibiotic-sensitive
pathogens, de-escalation was safe and associated with a trend
to increase the chance of survival given that the initial empirical
therapy was early and adequate. Another study by the same
research group24 also supported the use of a de-escalation policy
for bacteremia due to difficult-to-treat Gram-negative bacilli. In a
recent publication, Koupetori et al25 reported that in patients with
sepsis due to bloodstream infections, de-escalation therapy did
not negatively affect the final outcome and might even lead to
survival benefit in the second study period. Regarding ventilator-
associated pneumonia, in which the potential risk of multi-drug-
resistant microorganisms is relatively high, numerous studies
have shown that de-escalation therapy was rational and safe in
these patients and was even associated with a trend towards
lower mortality.26,27
The diverse conditions of critically ill patients with severe sepsis
and/or septic shock make it hard to standardize the algorithms
used for empiric antibiotics therapy, the time when de-escalation
should be performed, and the roles of multi-drug resistant patho-
gens, etc. Furthermore, local guidelines for empiric antibiotics
therapy vary dramatically owing to diverse microbiological envi-
ronment, infection sources and the severity of illness. Considering
the problems stated above, we conducted several subgroup ana-
lyses based on study characteristics to explore sources of the
foreseeable significant heterogeneity between studies in our meta-
analysis. For the comparison of mortality in the de-escalation group
and the continuation group, the heterogeneity was not statistically
significant when we analyzed the 4 studies with high methodo-
logical quality (NOS score S 7 points). The result was consistent
with that of the prospective studies subgroup, revealing that the
heterogeneity could be explained by the methodological quality of
included studies. The I2 value also decreased in the subgroup
analysis for the 6 studies enrolling general severe sepsis and/or
septic shock patients, indicating that the heterogeneity could be
partially explained by the etiology of the included patients.
5
The heterogeneity was significant when we calculated the rates
of de-escalation therapy in patients from observational studies,
revealing the discrepancy between different studies. According to
the results of our subgroup analysis for general Western pop-
ulations with severe sepsis and/or septic shock, the heterogeneity
could be partially explained by the geographical origin and the
etiology of included patients.
Several limitations of our meta-analysis should be considered.
First, most studies included in our meta-analysis were observa-
tional studies. The lack of good matched studies might be a source
of associated bias. The only randomized trial included was a non-
blinded trial conducted by Leone et al,6 yet the imbalance be-
tween the de-escalation group and the continuation group makes it
difficult to draw firm conclusions. Second, we failed to compare
other outcomes except for mortality because of the lack of sufficient
data. Third, we included only 6 studies enrolling general severe
sepsis and/or septic shock patients. The type of patients and
enrollment number is a limitation.
Conclusions
In conclusion, our study demonstrates that antibiotic de-
escalation therapy has no detrimental impact on mortality in pa-
tients with severe sepsis and/or septic shock, as compared to the
continuation of broad-spectrum antibiotics. And considering the
reduced selection pressure and decreased costs associated with
antibiotic de-escalation, it should be considered as an option in
clinical practice. However, research in this regard has not been
accomplished. Well-designed RCTs or better-matched observa-
tional studies focusing on other important clinical outcomes are
still required in the near future.
Acknowledgments
Our manuscript was not supported by any commercial company
or grants.
Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.hrtlng.2016.06.001.
References
1. Osmon S, Warren D, Seiler SM, Shannon W, Fraser VJ, Kollef MH. The influence
of infection on hospital mortality for patients requiring > 48 h of intensive
care. Chest. 2003;124:1021e1029.
2. Dellinger RP, Levy MM, Rhodes A, et al. Surviving sepsis campaign: interna-
tional guidelines for management of severe sepsis and septic shock, 2012.
Intensive Care Med. 2013;39:165e228.
3. Garnacho-Montero J, Ortiz-Leyba C, Herrera-Melero I, et al. Mortality and
morbidity attributable to inadequate empirical antimicrobial therapy in pa-
tients admitted to the ICU with sepsis: a matched cohort study. J Antimicrob
Chemother. 2008;61:436e441.
4. Garnacho-Montero J, Gutierrez-Pizarraya A, Escoresca-Ortega A, et al. De-
escalation of empirical therapy is associated with lower mortality in patients
with severe sepsis and septic shock. Intensive Care Med. 2014;40:32e40.
5. Oshima T, Kodama Y, Takahashi W, et al. Empiric antibiotic therapy for severe
sepsis and septic shock. Surg Infect (Larchmt); 2015;. http://dx.doi.org/10.1089/
sur.2014.096.
6. Leone M, Bechis C, Baumstarck K, et al. De-escalation versus continuation of
empirical antimicrobial treatment in severe sepsis: a multicenter non-blinded
randomized noninferiority trial. Intensive Care Med. 2014;40:1399e1408.
7. Paskovaty A, Pastores SM, Gedrimaite Z, Kostelecky N, Riedel ER, Seo SK.
Antimicrobial de-escalation in septic cancer patients: is it safe to back down?
Intensive Care Med. 2015;41:2022e2023.
8. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for sys-
tematic reviews and meta-analyses: the PRISMA statements. Ann Intern Med.
2009;151:264e269.
6 Y. Guo et al. / Heart & Lung xxx (2016) 1e6
9. Cotta MO, Roberts JA, Tabah A, Lipman J, Vogelaers D, Blot S. Antimicrobial
stewardship of beta-lactams in intensive care units. Expert Rev Anti Infect Ther.
2014;12:581e595.
10. Garnacho-Montero J, Escoresca-Ortega A, Fernandez-Delgado E. Antibiotic de-
escalation in the ICU: how is it best done? Curr Opin Infect Dis. 2015;28:193e198.
11. Wells G, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale (NOS) for
Assessing the Quality of Nonrandomised Studies in Meta-analyses, http://www.
ohri.ca/programs; 2001.
12. Jeong H, Yim HW, Cho YS, et al. Efficacy and safety of stem cell therapies for
patients with stroke: a systematic review and single arm meta-analysis. Int J
Stem Cells. 2014;7:63e69.
13. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in
meta-analyses. BMJ. 2003;327:557e560.
14. Lau J, Ioannidis J, Terrin N, Schmid C, Olki I. The case of the misleading funnel
plot. BMJ. 2006;333:597e600.
15. Escoresca-Ortega A, Garnacho-Montero J, Herrera-Melero M, Marquez-
Vacaro J, Ortiz-Leyba C. De-escalation therapy strategy in patients admitted to
the ICU with severe sepsis. Intensive Care Med. 2009;35(suppl 1):S241.
16. Heenen S, Jacobs F, Vincent JL. Antibiotic strategies in severe nosocomial
sepsis: why do we not de-escalate more often? Crit Care Med. 2012;40:1404e
1409.
17. Salahuddin N, Amer L, Joseph M, et al. De-escalation of empiric antibiotics in
severe sepsis: a situational analysis. Crit Care Med. 2013;41:A234.
18. Lee CC, Lee NY, Chen PL, et al. Impact of antimicrobial strategies on clinical
outcomes of adults with septic shock and community-onset Enter-
obacteriaceae bacteremia: de-escalation is beneficial. Diagn Microbiol Infect Dis.
2015;82:158e164.
19. Mokart D, Slehofer G, Lambert J, et al. De-escalation of antimicrobial treatment
in neutropenic patients with severe sepsis: results from an observational study.
Intensive Care Med. 2014;40:41e49.
20. Rerriter M, Huband N. Does the non-randomized controlled study have a place
in the systematic review? A pilot study. Crim Behav Ment Health. 2005;15:111e
120.
21. Morel J, Casoetto J, Jospe R, et al. De-escalation as part of a global strategy of
empiric antibiotherapy management. A retrospective study in a medico-
surgical intensive care unit. Crit Care. 2010;14:R225.
22. de Jong E, van Oers JA, Beishuizen A, et al. Efficacy and safety of procalcitonin
guidance in reducing the duration of antibiotic treatment in critically ill pa-
tients: a randomised, controlled, open-label trial. Lancet Infect Dis; 2016; http://
dx.doi.org/10.1016/S1473-3099(16)00053-0.
23. Shime N, Satake S, Fujita N. De-escalation of antimicrobials in the treatment of
bacteraemia due to antibiotic-sensitive pathogens in immunocompetent pa-
tients. Infection. 2011;39:319e325.
24. Shime N, Kosaka T, Fujita N. De-escalation of antimicrobial therapy for bac-
teraemia due to difficult-to-treat Gram-negative bacilli. Infection. 2013;41:
203e210.
25. Koupetori M, Retsas T, Antonakos N, et al. Bloodstream infections and sepsis in
Greece: over-time change of epidemiology and impact of de-escalation on final
outcome. BMC Infect Dis. 2014;14:272.
26. Rello J, Vidaur L, Sandiumenge A, et al. De-escalation therapy in ventilator-
associated pneumonia. Crit Care Med. 2004;32:2183e2190.
27. Giantsou E, Liratzopoulos N, Efraimidou E, et al. De-escalation therapy rates are
significantly higher by bronchoalveolar lavage than by tracheal aspirate.
Intensive Care Med. 2007;33:1533e1540.