Wirz et al.

Critical Care (2018) 22:191

https://doi.org/10.1186/s13054-018-2125-7

REVIEW Open Access

Effect of procalcitonin-guided antibiotic

treatment on clinical outcomes in intensive
care unit patients with infection and sepsis
patients: a patient-level meta-analysis of
randomized trials
Yannick Wirz1†, Marc A. Meier1†, Lila Bouadma3, Charles E. Luyt4, Michel Wolff3, Jean Chastre4, Florence Tubach5,
Stefan Schroeder6, Vandack Nobre7, Djillali Annane8, Konrad Reinhart9, Pierre Damas10, Maarten Nijsten11,
Arezoo Shajiei11, Dylan W. deLange12, Rodrigo O. Deliberato13, Carolina F. Oliveira14, Yahya Shehabi15,16,
Jos A. H. van Oers17, Albertus Beishuizen18, Armand R. J. Girbes19, Evelien de Jong19, Beat Mueller1,2
and Philipp Schuetz1,2*

Abstract
Background: The clinical utility of serum procalcitonin levels in guiding antibiotic treatment decisions in patients
with sepsis remains unclear. This patient-level meta-analysis based on 11 randomized trials investigates the impact
of procalcitonin-guided antibiotic therapy on mortality in intensive care unit (ICU) patients with infection, both
overall and stratified according to sepsis definition, severity, and type of infection.
Methods: For this meta-analysis focusing on procalcitonin-guided antibiotic management in critically ill patients with
sepsis of any type, in February 2018 we updated the database of a previous individual patient data meta-analysis which
was limited to patients with respiratory infections only. We used individual patient data from 11 trials that randomly
assigned patients to receive antibiotics based on procalcitonin levels (the “procalcitonin-guided” group) or the current
standard of care (the “controls”). The primary endpoint was mortality within 30 days. Secondary endpoints were duration
of antibiotic treatment and length of stay.
Results: Mortality in the 2252 procalcitonin-guided patients was significantly lower compared with the 2230
control group patients (21.1% vs 23.7%; adjusted odds ratio 0.89, 95% confidence interval (CI) 0.8 to 0.99;
p = 0.03). These effects on mortality persisted in a subgroup of patients meeting the sepsis 3 definition and
based on the severity of sepsis (assessed on the basis of the Sequential Organ Failure Assessment (SOFA)
score, occurrence of septic shock or renal failure, and need for vasopressor or ventilatory support) and on
the type of infection (respiratory, urinary tract, abdominal, skin, or central nervous system), with interaction
for each analysis being > 0.05. Procalcitonin guidance also facilitated earlier discontinuation of antibiotics,
with a reduction in treatment duration (9.3 vs 10.4 days; adjusted coefficient −1.19 days, 95% CI −1.73 to
−0.66; p < 0.001).
(Continued on next page)

* Correspondence: [email protected]
†
Yannick Wirz and Marc A. Meier contributed equally to this work.
1
Medical University Department, Kantonsspital Aarau, Tellstrasse, CH-5001
Aarau, Switzerland
2
Faculty of Medicine, University of Basel, Basel, Switzerland
Full list of author information is available at the end of the article

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Wirz et al. Critical Care (2018) 22:191 Page 2 of 11

(Continued from previous page)

Conclusion: Procalcitonin-guided antibiotic treatment in ICU patients with infection and sepsis patients results in
improved survival and lower antibiotic treatment duration.
Keywords: Sepsis, Procalcitonin, Antibiotic stewardship, Meta-analysis,

Background definitions among trials and to investigate the impact of

Sepsis, defined as life-threatening organ dysfunction PCT guidance in different patient subgroups.
caused by a dysregulated host response to infection, re- To address this significant drawback of earlier
mains a major healthcare problem worldwide and affects meta-analyses, we performed a meta-analysis of individ-
millions of people each year [1, 2]. Early identification ual patient data from 11 randomized-controlled trials
and appropriate initial management including the start (RCTs) to assess the safety of using PCT to guide anti-
of antibiotic treatment and fluid resuscitation improves biotic decisions in ICU patients with infection and dif-
outcomes [2, 3]. In addition, monitoring of patients dur- ferent sepsis severities and with the involvement of
ing treatment both for timely escalation of therapy in different organs.
case of treatment failure and de-escalation in case of a
favorable treatment response has an important impact Methods
on patient recovery. [1, 2] This also includes early Definition of patient population and trial selection
de-escalation or cessation of antibiotic treatment once a For this meta-analysis focusing on PCT-guided antibiotic
patient’s condition has stabilized, with signs indicating management in critically ill patients with sepsis, we
progression towards the resolution of infection. updated the database of a previous individual patient
Given that clinical signs for monitoring patients with data meta-analysis which was limited to patients with re-
sepsis can be ambiguous, the use of additional bio- spiratory infections only [20, 21]. Trial selection and
markers mirroring specific physiopathological pathways data collection were performed following the initial
has been proposed. In this context, serum procalcitonin protocol published in the Cochrane Library [22] and the
(PCT) has emerged as a sensitive biomarker that pro- report was prepared according to PRISMA individual
vides prognostic information in patients with infections, participant data (IPD) guidelines [23, 24].
and thus may improve sepsis management [4, 5]. Mul- For this analysis, we included all patients residing in
tiple studies have demonstrated that PCT levels increase an ICU with any type of systemic infection. We thus ex-
in response to bacterial infection and decrease during cluded patients not treated in the ICU and also one trial
recovery [4, 5]. Reflecting the host response to a bacter- that focused on ventilator-associated pneumonia because
ial infection, PCT thus provides important adjunctive these patients may not have systemic bacterial infections
information in addition to traditional clinical and diag- [25]. Furthermore, pediatric trials and those not using
nostic parameters [6]. PCT to guide initiation and duration of antibiotic
Multiple trials have investigated the benefits of using treatment were excluded from this analysis.
serum PCT levels to guide whether and for how long Given that the definition of sepsis has changed over
antibiotic therapy is used—a process referred to as time [26], and is used differently among researchers and
PCT-guided antibiotic stewardship—in patients with in- clinicians, we decided to focus primarily on trials involv-
fection in the intensive care unit (ICU) [1, 7–17]. While ing patients presenting with infections to the critical
most trials have focused on the effects of PCT guidance care unit. However, we performed a subgroup analysis
on antibiotic usage, a recent large trial from the on patients meeting the sepsis 3 definition [26]. We also
Netherlands reported lower mortality following stratified the analysis based on sepsis severity and type
PCT-guided therapy [10]. However, conclusive evidence of organ involved. Individual patient data were collected
on the safety of this approach across different types of from eligible RCTs which assessed adults meeting these
infection and sepsis severities has been limited due to criteria.
largely insufficient statistical power in most of the indi-
vidual trials. Moreover, meta-analyses focusing on the Trial search and data collection
impact of PCT-guided antibiotic stewardship in sepsis The search strategy for this review was updated in Feb-
patients on clinical outcomes overall and within patient ruary 2018 in collaboration with personnel from the
subgroups based on infection type and severity have Cochrane collaboration and executed in all databases
been inconclusive [18, 19]. A noteworthy limitation of from the date of their inception to February 2018. All
these meta-analyses was that they were based on aggre- references were also screened for eligibility. Databases
gate data, limiting opportunities to harmonize outcome searched included the Cochrane Central Register of
Wirz et al. Critical Care (2018) 22:191 Page 3 of 11

Controlled Trials (CENTRAL; January 2017, issue 1), Patients and endpoints
MEDLINE Ovid (1966 to February 2017), and Embase.- All patients with a suspected or proven infection as the
com (1980 to February 2017). We applied no language main diagnosis treated in the ICU who were included in
or publication restrictions. a previous trial and randomized either to PCT-guided
Two authors (YW and MAM) independently assessed care or a control group were included in the overall ana-
trial eligibility based on titles, abstracts, and full-text re- lysis. The primary safety endpoint was all-cause mortal-
ports, with further information being obtained directly ity within 30 days of randomization. For trials with a
from investigators as needed. Study protocols, case re- shorter follow-up period, the available information was
port forms, and unedited databases containing individual used (e.g., mortality at the time of hospital discharge).
patient data were requested from investigators of all eli- The main efficacy endpoint was antibiotic treatment
gible trials. Data from each trial were first checked duration. Other clinical endpoints included length of
against reported results and queries were resolved with hospital stay and length of ICU stay.
the principal investigator, trial data manager, or statisti-
cian. Data were assessed in a consistent manner across Statistical analysis
all trials, with standard definitions and parameters, and The analysis follows a study protocol similar to a
thus mortality rates differed slightly from previous re- previous protocol published in the Cochrane Library
ports. In accordance with the Cochrane methodology, [22] but we specified the main analysis and subgroup
we used the Grading of Recommendations, Assessment, analysis for the sepsis patient population. For the pri-
Development, and Evaluation (GRADE) approach to as- mary endpoint (mortality), odds ratios (ORs) and 95%
sess risk of selection bias, performance bias, detection confidence intervals (CIs) were calculated using multi-
bias, attrition bias, reporting bias, and other types of bias variable hierarchical logistic regression [28, 29]. Vari-
[27]. The grading was performed by two authors (YW ables in the multivariate analysis included treatment
and MAM) and, in case of conflicting results, grading arm, age, gender, and type of infection. To control for
was discussed with another author (PS) and within the within- and between-trial variability, a “trial” variable
meta-analysis group. was added to the model as a random effect.

Fig. 1 Study flow diagram. ICU intensive care unit, PCT procalcitonin, RCT randomized controlled trial
Wirz et al. Critical Care (2018) 22:191 Page 4 of 11

Corresponding linear and logistic regression models were infected. We tested for subgroup effects by adding inter-
fitted for continuous and binary secondary endpoints, re- action terms to the model. All statistical analyses were
spectively. Analyses followed the intention-to-treat performed using Stata version 9.2 (College Station,
principle by analyzing patients in groups to which they Texas, USA) and Review Manager version 5.3.
were randomized. Censoring was used for patients with a
follow-up < 30 days for the time-to-event analyses. Results
We performed a prespecified subgroup analysis on pa- Results of the search and characteristics of included trials
tients meeting the sepsis 3 definition, i.e., defining sepsis Data from 4482 individual patients enrolled in 11 of 32
as life-threatening organ dysfunction due to a dysregu- potentially eligible trials were obtained and included in
lated host response to infection, and, as such, requires at the meta-analysis (Fig. 1). A total of 21 trials whose pa-
least one organ dysfunction (i.e., at least one organ with tients did not have sepsis treated in the ICU and which
a Sequential Organ Failure Assessment (SOFA) score thus did not meet our inclusion criteria were excluded.
above or equal to 2 [26]). Additional prespecified sub- Included trials were conducted in seven countries in-
group analyses were performed for sepsis severity (septic cluding Switzerland, Germany, France, the Netherlands,
shock), treatment modality in the ICU, and type of organ Brazil, Belgium, and Australia (Table 1). There were two
Table 1 Characteristics of included trials
First Country Setting, Patients Follow-up Clinical diagnosis Type of procalcitonin Compliance
author type of included algorithm; procalcitonin with the
(year) trial cutoffs used (μg/L)a PCT
protocol
Annane France ICU, 62 Hospital stay Severe sepsis without overt Initiation and duration; R 63%
(2013) [7] multicenter source of infection and negative against AB: < 0.5 (< 0.25); R adherence
blood culture for AB: > 0.5 (> 5.0)
Bloos Germany ICU, 1089 3 months Severe sepsis or septic shock Discontinuation at days 4, 7, 49.6%
(2016) [8] multicenter (SIRS and documented infection and 10; R against AB: < 1.0 adherence
+ criteria for severe sepsis/septic or > 50% drop over previous
shock) value
Bouadma France ICU, 621 2 months Critically ill patients with Initiation and duration; R 47%
(2010) [9] multicenter assumed/proven bacterial against AB: < 0.5 (< 0.25); R adherence
infection for AB: > 0.5 (> 1.0)
De Jong The ICU, 1546 1 year Critically ill patients with Duration; R against AB: < 0.5 44%
(2016) [10] Netherlands multicenter assumed infection or > 80% drop over peak adherence
value
Deliberato Brazil ICU, single 81 ICU discharge or Sepsis patients with Duration; R against AB: < 0.5 47.6%
(2013) [11] center 14 days microbiologically confirmed or > 90% drop over peak adherence
postrandomization bacterial infection value
Hochreiter Germany Surgical 110 Hospital stay Sepsis (SIRS and documented Duration; R against AB: < 1 not
(2009) [14] ICU, single infection) or > 65% drop over 3 days reported
center
Layios Belgium ICU, single 379 1 month Critically ill patients with Initiation; R against AB: < 0.5 46.3%
(2012) [15] center assumed infection (< 0.25); R for AB: > 0.5 (> 1.0) adherence
Nobre Switzerland ICU, single 79 1 month Severe sepsis or septic shock Duration; R against AB: < 0.5 81%
(2008) [17] center (< 0.25) or > 80% drop over adherence
peak value; R for AB: > 0.5
(> 1.0)
Oliveira Brazil ICU, 94 28 days or Severe sepsis or septic shock Discontinuation; Initial < 1.0: 87.8%
(2013) [16] multicenter hospital discharge (SOFA score > 10 and/or R against AB: 0.1 at day 4; adherence
bacteremia) Initial > 1.0: R against: > 90%
drop over peak value
Schroeder Germany Surgical 27 Hospital stay Severe sepsis following Duration; R against AB: < 1 not
(2009) [13] ICU, single abdominal surgery (SIRS and or > 65% drop over 3 days reported
center documented infection + criteria
for severe sepsis/septic shock)
Shehabi Australia ICU, 394 3 months Sepsis (SIRS and documented Duration; R against AB: < 0.25 97%
(2014) [1] multicenter infection) (< 0.1) or > 90% drop over adherence
peak value
AB antibiotic, ICU intensive care unit, PCT procalcitonin, R recommendation, SIRS systemic inflammation response system, SOFA Sequential Organ
Failure Assessment
a
Cutoffs are listed as recommendation (strong recommendation)
Wirz et al. Critical Care (2018) 22:191 Page 5 of 11

trials from surgical ICUs with postoperative sepsis pa- Table 2 Baseline characteristics of included patients
tients, while the remaining nine trials focused on med- Parameter Control group PCT group
ical or mixed ICU patients with sepsis. All trials focused (n = 2230) (n = 2252)
on sepsis patients but, as demonstrated in Table 1, defin- Demographics
ition of patients varied somewhat. The largest trials were Age (years) 64.1 ± 15.0 63.5 ± 15.2
those by De Jong (n = 1546) [10], Bloos (n = 1089) [8], Male gender 1281 (57.5%) 1273 (56.5%)
and Bouadma (n = 621) [9]. The PCT algorithms used in
Primary focus of infection
the different trials were similar and mainly focused on
Respiratory 1101 (49.4%) 1102 (48.9%)
early discontinuation of antibiotics if PCT dropped
below 0.5 μg/L or by 80% from the peak level. Urinary 129 (5.8%) 118 (5.2%)
The quality of trials according to GRADE criteria was Abdominal 417 (18.7%) 391 (17.4%)
moderate to high (see Additional file 1). There was a Skin/soft tissue 41 (1.8%) 32 (1.4%)
low risk of selection bias, attrition bias, and reporting CNS 35 (1.6%) 38 (1.7%)
bias. Trials did not blind caregivers and patients with re-
Other/unknown 440 (19.7%) 519 (23.0%)
gard to the intervention and only about one-third of the
Genital/gynecologic 8 (0.4%) 3 (0.1%)
trials featured blinded outcome assessment. There was
no evidence of publication bias based on inspection of Catheter-related 14 (0.6%) 16 (0.7%)
the funnel plot (see Additional file 1). We also found Bloodstream 36 (1.6%) 25 (1.1%)
variable adherence to the PCT protocols ranging from Upper respiratory 9 (0.4%) 8 (0.4%)
44% to 97% (Table 1). Vital signs
Baseline characteristics of individual patients were
Temperature (°C) 37.7 ± 1.2 37.8 ± 1.1
similar between the PCT and control groups. About 50%
Sepsis score
of patients had sepsis due to infection of the lung,
followed by abdominal infection (18%) and urinary tract Meeting sepsis 3 definition 1630 (73.1%) 1605 (71.3%)
infection (5%). The mean SOFA score was 7.4 points SOFA score (points) 7.4 ± 4.0 7.3 ± 4.1
and more than two-thirds of patients were on Additional sepsis support
vasopressors and/or ventilation support. Table 2 lists Vasopressor use 1593 (76.3%) 1606 (76.7%)
additional baseline characteristics stratified according to
Ventilation support 1434 (68.1%) 1478 (69.4%)
randomization. There were no statistical differences
Renal replacement 767 (34.4%) 757 (33.6%)
between the two groups.
Values are presented as mean ± standard deviation or n (%) as appropriate
CNS central nervous system, PCT procalcitonin, SD standard deviation, SOFA
Primary safety endpoint: mortality Sequential Organ Failure Assessment
There were 529 deaths among 2230 control group pa-
tients (23.7%) compared with 475 deaths among 2252 As an additional sensitivity analysis, we restricted the
PCT-guided patients (21.1%), resulting in significantly analysis to trials reporting ≥ 28 days mortality, thereby
lower mortality in the PCT group (adjusted OR 0.89, excluding three trials where this information was not
95% CI 0.80 to 0.99; p = 0.03) (Fig. 2 and Table 3). This assessed (Table 1). Again, results were similar to the
effect was consistent in patients meeting the sepsis 3 main results.
definition, and across different severity groups based on
SOFA score, presence or absence of septic shock, Primary efficacy endpoint: antibiotic use
ventilatory failure, and renal support (p > 0.05 for each A moderate reduction in total antibiotic treatment
interaction, indicating no significant difference due to duration resulted from PCT guidance (mean 10.4 ±
subgroup effect). The effects on mortality were also con- 9.7 vs 9.3 ± 9.2 days; adjusted regression coefficient
sistent among different types of infections including re- −1.19 days, 95% CI −1.73 to −0.66; p < 0.001)
spiratory tract, urinary, abdominal, skin/soft tissue, and (Table 3). Effects were similar in patients meeting
central nervous system (CNS) infections (p > 0.05 for the sepsis 3 definition. There was some evidence of
each interaction) (Table 4). effect modification with stronger effects in patients
As an additional sensitivity analysis, a meta-analysis of with lower sepsis severity based on SOFA score
the aggregate results of all trials was conducted (see (−2.62 days in patients with SOFA score of 0–6
Additional file 1). The point estimate for mortality in points vs 0.01 days in patients with SOFA score of
this analysis was similar to that seen in the individual 7–9 and −0.63 days in patients with SOFA scores of
patient data analysis but did not reach statistical signifi- 10–24 (p for the interaction < 0.05). Similarly, in pa-
cance (OR 0.88, 95% CI 0.76 to 1.01). Heterogeneity was tients with abdominal infection, PCT-guided therapy
low, suggesting robust effects (I2 = 0%). did not reduce duration of treatment compared with
Wirz et al. Critical Care (2018) 22:191 Page 6 of 11

Fig. 2 Forest plot showing 30-day mortality. Association of procalcitonin (PCT)-guided antibiotic stewardship and mortality in predefined
subgroups. CI confidence interval, CNS central nervous system, SOFA Sequential Organ Failure Assessment

patients with all other infected organs, with the with PCT-guided therapy, confirming the results of a
strongest effects being seen in patients with skin and large Dutch trial [10]. This effect was consistent in sepsis
soft tissue infections (−4.6 days, 95% CI −10.36 to patients based on the sepsis 3 definition, and across dif-
1.23; p = 0.122). ferent severities and types of sepsis based on whether
PCT-guided treatment was employed, and also across
Length of stay different types of infection. Moreover, PCT guidance was
Length of hospital stay (adjusted regression also associated with a modest reduction in exposure to
coefficient 0.09 days, 95% CI −1.51 to 1.7; p = 0.908) antibiotics through shorter treatment durations and earl-
and ICU stay (adjusted regression coefficient ier discontinuation. However, PCT guidance did not
0.04 days, 95% CI −0.9 to 0.99; p = 0.928) were simi- have an effect on length of ICU or hospital stay.
lar in the PCT and control groups overall and in Early diagnosis combined with initiation of appropriate
most subgroup analyses. Patients in the highest antibiotic treatment remains the cornerstone of sepsis
SOFA categories had longer lengths of ICU and hos- care [30]. Once treatment is initiated, close monitoring
pital stay if PCT guidance had been used (3.21 days, of patients is of the utmost importance to identify pa-
95% CI −0.76 to 7.18, and 2.08 days, 95% CI −0.08 tients with a favorable disease course who are at low risk
to 4.24, respectively). In addition, patients with renal for complications so that one may consider early cessa-
failure (renal support) had longer lengths of hospital tion of antibiotic therapy. Daily assessment of patient
stay while patients with no renal failure had shorter risk using objective prognostic parameters is therefore
stays (2.97 days vs −1.41 days). important. In addition to clinical parameters, blood
markers may aid in patient monitoring [31–34]. While
Discussion serum lactate is a commonly used biomarker that may
This meta-analysis of individual patient data from 11 help guide resuscitation measures [35], PCT has been
randomized trials that included 4482 patients with infec- previously demonstrated to be helpful in assessing the
tion treated in ICUs revealed lower mortality associated response to antimicrobial treatment and in aiding
Wirz et al. Critical Care (2018) 22:191 Page 7 of 11

Table 3 Clinical endpoints overall and stratified by sepsis severity and use of sepsis support
Control group PCT group Adjusted OR or difference p value
(95% CI)a, p value for
2230 2252
interaction
Overall
Antibiotic therapy (days) 10.4 ± 9.7 9.3 ± 9.2 −1.19 (−1.73 to −0.66), p < 0.001
30-day mortality 529 (23.7%) 475 (21.1%) 0.89 (0.8 to 0.99), p = 0.03
Length of hospital stay (days) 28.7 ± 27.9 28.6 ± 27.9 0.09 (−1.51 to 1.7), p = 0.908
Length of ICU stay (days) 14.7 ± 16.3 14.8 ± 16.5 0.04 (−0.9 to 0.99), p = 0.928
Subgroup by sepsis definition
Meeting sepsis 3 definition 1630 1605
Antibiotic therapy (days) 10.5 ± 9.2 9.3 ± 8.9 −1.22 (−1.82 to −0.62), p < 0.001 0.915
30-day mortality 397 (24.4%) 338 (21.1%) 0.86 (0.76 to 0.98), p = 0.022 0.388
Length of hospital stay (days) 29.5 ± 27.9 29.6 ± 29.4 0.07 (−1.89 to 2.03), p = 0.946 0.891
Length of ICU stay (days) 14.1 ± 15.5 14.5 ± 17.0 0.37 (−0.74 to 1.48), p = 0.51 0.246
Not meeting sepsis 3 definition 600 647
Antibiotic therapy (days) 10.2 ± 10.9 9.1 ± 9.9 −1.13 (−2.27 to 0.01), p = 0.052 0.915
30-day mortality 132 (22.0%) 137 (21.2%) 0.96 (0.78 to 1.19), p = 0.717 0.388
Length of hospital stay (days) 26.4 ± 27.9 26.2 ± 23.7 0.16 (−2.53 to 2.84), p = 0.909 0.891
Length of ICU stay (days) 16.5 ± 18.3 15.6 ± 15.2 −0.82 (−2.65 to 1.01), p = 0.378 0.246
Subgroup by organ failure
SOFA 0 to 6 763 776
Antibiotic therapy (days) 10.9 ± 10.3 8.1 ± 8.2 −2.62 (−3.51 to −1.73), p < 0.001 < 0.001
30-day mortality 105 (13.8%) 91 (11.7%) 0.85 (0.66 to 1.1), p = 0.227 0.991
Length of hospital stay (days) 28.6 ± 28.6 26.8 ± 25.8 −1.96 (−4.65 to 0.72), p = 0.152 0.138
Length of ICU stay (days) 12.9 ± 16.7 12.1 ± 15.6 −0.92 (−2.52 to 0.69), p = 0.263 0.083
SOFA 7 to 9 474 445
Antibiotic therapy (days) 10.1 ± 8.2 10.2 ± 9.4 0.01 (−1.1 to 1.11), p = 0.988 0.003
30-day mortality 112 (23.6%) 96 (21.6%) 0.92 (0.73 to 1.17), p = 0.512 0.601
Length of hospital stay (days) 30.7 ± 26.4 29.4 ± 25.9 −1.14 (−4.52 to 2.24), p = 0.508 0.543
Length of ICU stay (days) 14.0 ± 14.3 14.1 ± 15.3 0.23 (−1.66 to 2.12), p = 0.813 0.862
SOFA 10 to 24 486 486
Antibiotic therapy (days) 10.7 ± 9.1 10.0 ± 8.8 −0.63 (−1.71 to 0.45), p = 0.256 0.125
30-day mortality 190 (39.1%) 161 (33.1%) 0.86 (0.72 to 1.01), p = 0.065 0.576
Length of hospital stay (days) 29.4 ± 27.5 32.6 ± 35.5 3.21 (−0.76 to 7.18), p = 0.113 0.024
Length of ICU stay (days) 15.6 ± 15.4 17.7 ± 19.2 2.08 (−0.08 to 4.24), p = 0.059 0.036
No septic shock (no vasopressor use) 495 489
Antibiotic therapy (days) 11.2 ± 9.9 9.8 ± 9.2 −1.35 (−2.53 to −0.17), p = 0.025 0.586
30-day mortality 79 (16.0%) 76 (15.5%) 0.97 (0.73 to 1.28), p = 0.823 0.512
Length of hospital stay (days) 24.9 ± 26.3 23.2 ± 20.6 −1.49 (−4.35 to 1.38), p = 0.309 0.258
Length of ICU stay (days) 13.2 ± 15.6 12.7 ± 14.2 −0.43 (−2.25 to 1.39), p = 0.642 0.47
Septic shock (with vasopressor use) 1593 1606
Antibiotic therapy (days) 10.4 ± 9.8 9.3 ± 9.4 −1.03 (−1.68 to −0.39), p = 0.002 0.586
30-day mortality 420 (26.4%) 376 (23.4%) 0.89 (0.79 to 1), p = 0.051 0.512
Length of hospital stay (days) 30.1 ± 27.0 30.6 ± 26.3 0.6 (−1.23 to 2.43), p = 0.52 0.258
Length of ICU stay (days) 15.0 ± 16.1 15.4 ± 17.0 0.34 (−0.8 to 1.49), p = 0.557 0.47
No respiratory failure (invasive ventilation support) 672 651
Wirz et al. Critical Care (2018) 22:191 Page 8 of 11

Table 3 Clinical endpoints overall and stratified by sepsis severity and use of sepsis support (Continued)
Control group PCT group Adjusted OR or difference p value
(95% CI)a, p value for
2230 2252
interaction
Antibiotic therapy (days) 11.0 ± 11.4 9.5 ± 9.6 −1.43 (−2.49 to − 0.37), p = 0.008 0.467
30-day mortality 104 (15.5%) 83 (12.7%) 0.83 (0.64 to 1.08), p = 0.158 0.633
Length of hospital stay (days) 25.4 ± 21.7 25.3 ± 23.3 −0.01 (−2.33 to 2.31), p = 0.993 0.788
Length of ICU stay (days) 11.7 ± 13.5 12.4 ± 14.2 0.53 (− 0.86 to 1.92), p = 0.456 0.433
Respiratory failure (invasive ventilation support) 1434 1478
Antibiotic therapy (days) 10.2 ± 9.0 9.2 ± 9.2 −1.09 (−1.73 to −0.45), p = 0.001 0.467
30-day mortality 401 (28.0%) 373 (25.2%) 0.89 (0.79 to 1), p = 0.058 0.633
Length of hospital stay (days) 30.9 ± 30.9 30.3 ± 26.0 −0.39 (−2.42 to 1.65), p = 0.708 0.788
Length of ICU stay (days) 16.2 ± 17.5 16.0 ± 17.5 −0.3 (−1.56 to 0.95), p = 0.637 0.433
No renal replacement therapy 1463 1495
Antibiotic therapy (days) 9.4 ± 9.2 8.4 ± 8.6 −1.02 (−1.64 to −0.39), p = 0.001 0.475
30-day mortality 301 (20.6%) 265 (17.7%) 0.86 (0.74 to 1), p = 0.046 0.553
Length of hospital stay (days) 29.8 ± 31.1 28.4 ± 25.5 −1.41 (−3.4 to 0.59), p = 0.168 0.026
Length of ICU stay (days) 14.6 ± 17.4 14.1 ± 15.9 −0.64 (−1.82 to 0.53), p = 0.284 0.077
Renal replacement therapy 767 757
Antibiotic therapy (days) 12.3 ± 10.2 10.9 ± 10.0 −1.45 (−2.44 to −0.46), p = 0.004 0.475
30-day mortality 228 (29.7%) 210 (27.7%) 0.96 (0.83 to 1.11), p = 0.584 0.553
Length of hospital stay (days) 26.6 ± 20.5 29.1 ± 32.2 2.97 (0.31 to 5.63), p = 0.028 0.026
Length of ICU stay (days) 14.9 ± 14.0 16.2 ± 17.5 1.43 (−0.16 to 3.02), p = 0.079 0.077
Values are presented as mean ± standard deviation or n (%) as appropriate
CI confidence interval, ICU intensive care unit, OR odds ratio, PCT procalcitonin, SOFA Sequential Organ Failure Assessment
a
Multivariable hierarchical regression with outcome of interest as dependent variable and trial as a random effect

antibiotic stewardship decisions [36–38]. Based on mul- previous meta-analyses based on aggregate data.
tiple randomized trials integrated in an aggregate patient While the effects seem similar between different sub-
data meta-analysis, the US Food and Drug Administra- groups, we did not find PCT to be associated with
tion (FDA) has recently approved the PCT assay for the reduced antibiotic use in patients with abdominal in-
purpose of guiding antibiotic therapy in the context of fections and those with renal impairment. It is pos-
sepsis [18, 39]. However, since sepsis is not a sible that, in these two patient groups, the PCT
well-defined disease but a heterogenous syndrome aris- kinetics are impaired by bacterial translocation due to
ing from different possible organs being infected and mucositis even if the initial infection has resolved,
with different clinical presentations based on severity, and by the slower decline of PCT in patients with
the question of safety and efficacy of PCT guidance in renal impairment. Additional investigations will be
patients with sepsis overall, and in subgroups based on needed to understand whether PCT algorithms re-
severity and organs involved, is relevant. In this context, quire further adaptation in these two patient groups.
our finding that lower mortality and shorter antibiotic Although we have no definitive explanations for the
courses are associated with PCT-guided care is positive effects of PCT-guided antibiotic treatment on
reassuring. mortality, several mechanisms seem plausible. First,
We have recently published a similar individual patient PCT provides additional prognostic information in
data meta-analysis looking at patients with different patients with sepsis and may influence site-of-care de-
types and severities of respiratory infections [20, 40]. cisions and timing of discharge [17]. The MOSES
Similar to what was seen in the current analysis, pa- study investigating PCT kinetics over 72 h in several
tients with respiratory infections also benefited from US emergency departments validated the prognostic
PCT-guided therapy, experiencing lower mortality and potential of PCT independent of other prognostic
needing significantly reduced antibiotic exposure. The scores [4]. Hence, early identification of treatment
current analysis further expands these findings, also nonresponders has the potential to prevent adverse
showing similar effects in subgroups of other types of events, although this was not true in the PASS trial, a
infections—an analysis that was not possible in all study investigating whether escalation of diagnostic
Wirz et al. Critical Care (2018) 22:191 Page 9 of 11

Table 4 Clinical endpoints stratified by type of infection

Control group PCT group Adjusted OR or difference (95% CI)a, p value p value for interaction
Subgroup by type of infection (suspected infection site)
Respiratory 1101 1102
Antibiotic therapy (days) 9.9 ± 7.8 8.5 ± 7.8 −1.36 (−1.99 to −0.73), p < 0.001 0.582
30-day mortality 262 (23.8%) 243 (22.1%) 0.92 (0.79 to 1.07), p = 0.299 0.466
Length of hospital stay (days) 28.2 ± 27.7 27.7 ± 24.7 −0.21 (−2.36 to 1.94), p = 0.849 0.668
Length of ICU stay (days) 15.1 ± 16.6 15.3 ± 17.5 0.19 (−1.24 to 1.61), p = 0.798 0.858
Urinary 129 118
Antibiotic therapy (days) 12.5 ± 12.4 11.0 ± 12.2 −1.62 (−4.6 to 1.36), p = 0.286 0.786
30-day mortality 21 (16.3%) 11 (9.3%) 0.59 (0.3 to 1.16), p = 0.128 0.215
Length of hospital stay (days) 29.5 ± 25.4 25.1 ± 21.7 −4.08 (−9.7 to 1.54), p = 0.154 0.209
Length of ICU stay (days) 14.3 ± 20.5 11.2 ± 13.6 −2.49 (−6.68 to 1.7), p = 0.244 0.123
Abdominal 417 391
Antibiotic therapy (days) 10.5 ± 10.6 11.0 ± 11.9 0.55 (−0.96 to 2.06), p = 0.477 0.005
30-day mortality 109 (26.1%) 89 (22.8%) 0.87 (0.68 to 1.11), p = 0.266 0.821
Length of hospital stay (days) 30.5 ± 27.7 32.1 ± 27.8 1.62 (−2.18 to 5.41), p = 0.404 0.361
Length of ICU stay (days) 15.1 ± 15.3 15.7 ± 16.4 0.42 (−1.76 to 2.6), p = 0.704 0.634
Skin/soft tissue 41 32
Antibiotic therapy (days) 12.9 ± 15.8 8.6 ± 8.5 −4.57 (−10.36 to 1.23), p = 0.122 0.159
30-day mortality 11 (27%) 8 (25%) 0.94 (0.43 to 2.06), p = 0.874 0.918
Length of hospital stay (days) 26.0 ± 25.8 20.9 ± 23.4 −4.17 (−15.4 to 7.06), p = 0.467 0.442
Length of ICU stay (days) 10.4 ± 9.4 10.9 ± 12.7 0.83 (−4.18 to 5.84), p = 0.747 0.916
Central nervous system 35 38
Antibiotic therapy (days) 11.7 ± 7.7 10.4 ± 7.7 −1.7 (−5.04 to 1.63), p = 0.317 0.958
30-day mortality 3 (9%) 2 (5%) 0.61 (0.11 to 3.46), p = 0.58 0.692
Length of hospital stay (days) 31.3 ± 22.8 30.8 ± 25.7 −0.44 (−11.48 to 10.6), p = 0.938 0.954
Length of ICU stay (days) 10.8 ± 14.0 13.7 ± 13.1 2.91 (−3.22 to 9.04), p = 0.352 0.457
Values are presented as mean ± standard deviation or n (%) as appropriate
CI confidence interval, ICU intensive care unit, OR odds ratio, PCT procalcitonin
a
Multivariable hierarchical regression with outcome of interest as dependent variable and trial as a random effect

and therapeutic measures based on high PCT levels severities and types of infection. Limitations of this
would decrease mortality [41]. Second, prolonged analysis include incomplete adherence to the PCT al-
antibiotic exposure has a toxic effect and thus may gorithm among the included trials, exclusion of im-
increase risk of treatment failure in control patients munocompromised patients in most trials, and
[42, 43]. Unexpected PCT results may also prompt heterogeneity among trials with regard to patient pop-
physicians to further examine patients and look for ulations and follow-up of patients. In addition,
additional explanations and illnesses. Finally, several cost-effectiveness analyses need to be undertaken to
observational studies have reported lower mortality determine whether using PCT-guided care is a
and risk of treatment failure associated with early cost-effective intervention [45].
antibiotic de-escalation in patients with sepsis, al-
though pathophysiological mechanisms are incom- Conclusion
pletely understood [37, 44]. In conclusion, PCT-guided antibiotic treatment in ICU
This analysis is based on a comprehensive literature patients with infection results in improved survival and
search and retrieval of all relevant trials and a net- shorter antibiotic treatment duration. Effects were simi-
work that permitted inclusion of individual patient lar in sepsis patients and among subgroups based on
data from most eligible trials. We also standardized sepsis severity, sepsis treatment modalities, and type of
definitions across trials and were able to perform ap- infection. Whether the reduction in antibiotic exposure
propriate subgroup analyses looking at different sepsis fully explains the mortality effects seen in our data needs
Wirz et al. Critical Care (2018) 22:191 Page 10 of 11

to be investigated in future trials. These findings have Germany. 7Department of Intensive Care, Hospital das Clinicas da
substantial clinical and public health implications. Universidade Federal de Minas Gerais, Belo Horizonte, Brazil. 8Critical Care
Department, Hôpital Raymond Poincaré, Assistance Publique - Hôpitaux de
Paris, Garches, France. 9Department of Anesthesiology and Intensive Care
Additional file Medicine, Jena University Hospital, Jena, Germany. 10Department of General
Intensive Care, University Hospital of Liege, Domaine universitaire de Liège,
Liege, Belgium. 11University Medical Centre, University of Groningen,
Additional file 1: Figure S1. Assessment of risk of bias in included
Groningen, The Netherlands. 12University Medical Center Utrecht, Utrecht,
trials. Figure S2. Forrest plot based on aggregate data. Figure S3.
The Netherlands. 13Laboratory for Critical Care Research, Critical Care Unit,
Funnel plots regarding possible publication bias. (PDF 57 kb)
Hospital Israelita Albert Einstein, São Paulo, Brazil. 14Department of Internal
Medicine, School of Medicine, Universidade Federal de Minas Gerais, Belo
Acknowledgements Horizonte, Brazil. 15Critical Care and Peri-operative Medicine, Monash Health,
We thank all patients involved in the individual trials and the investigators Melbourne, Australia. 16Faculty of Medicine Nursing and Health Sciences,
for sharing their data. Assistance with the preparation of the manuscript was School of Clinical Sciences, Monash University, Melbourne, Australia.
17
provided by Prasad Kulkarni, PhD, CMPP of Asclepius Medical Elisabeth Tweesteden Hospital, Tilburg, The Netherlands. 18Medisch
Communications LLC, Ridgewood, NJ, USA and was funded by BioMérieux Spectrum Twente, Enschede, The Netherlands. 19VU University Medical
Inc. Centre, Amsterdam, The Netherlands.

Funding Received: 2 June 2018 Accepted: 10 July 2018

This work was made possible by an unrestricted research grant from
BioMérieux Inc.

Availability of data and materials References

Data of the individual trials are available upon reasonable request based on 1. Shehabi Y, Sterba M, Garrett PM, Rachakonda KS, Stephens D, Harrigan P,
the authorization of principal investigators of the trials. Walker A, Bailey MJ, Johnson B, Millis D, et al. Procalcitonin algorithm in
critically ill adults with undifferentiated infection or suspected sepsis. A
Authors’ contributions randomized controlled trial. Am J Respir Crit Care Med. 2014;190(10):
PS, YW, and MAM performed the literature search and all analyses included 1102–10.
in this report and wrote the first draft. All authors shared trial data, gave 2. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, Kumar A,
critical feedback on the manuscript, and approved the final version for Sevransky JE, Sprung CL, Nunnally ME, et al. Surviving Sepsis Campaign:
submission. PS had full access to all of the data in the study and takes international guidelines for management of sepsis and septic shock: 2016.
responsibility for the integrity of the data and the accuracy of the data Intensive Care Med. 2017;43(3):304–77.
analysis. PS and BM oversaw the study and act as guarantors. 3. Kumar A, Ellis P, Arabi Y, Roberts D, Light B, Parrillo JE, Dodek P, Wood G,
Kumar A, Simon D, et al. Initiation of inappropriate antimicrobial therapy
Ethics approval and consent to participate results in a fivefold reduction of survival in human septic shock. Chest. 2009;
All trial included in the meta-analysis enrolled patients providing informed 136(5):1237–48.
consent and had approval from their institutional review boards. 4. Schuetz P, Birkhahn R, Sherwin R, Jones AE, Singer A, Kline JA, Runyon MS,
Self WH, Courtney DM, Nowak RM, et al. Serial procalcitonin predicts
mortality in severe sepsis patients: results from the multicenter procalcitonin
Consent for publication
MOnitoring SEpsis (MOSES) study. Crit Care Med. 2017;45(5):781–9.
Not applicable.
5. Wacker C, Prkno A, Brunkhorst FM, Schlattmann P. Procalcitonin as a
diagnostic marker for sepsis: a systematic review and meta-analysis. Lancet
Competing interests
Infect Dis. 2013;13(5):426–35.
BioMérieux Inc. provided an unrestricted research grant to conduct this
6. Mitsuma SF, Mansour MK, Dekker JP, Kim J, Rahman MZ, Tweed-Kent A,
analysis. However, no personnel from the company were involved in the
Schuetz P. Promising new assays and technologies for the diagnosis and
conduct of the analysis or in the preparation of the manuscript. PS and BM
management of infectious diseases. Clin Infect Dis. 2013;56(7):996–1002.
have received support from Thermo-Fisher and BioMérieux to attend meet-
7. Annane D, Maxime V, Faller JP, Mezher C, Clec'h C, Martel P, Gonzales H,
ings and fulfill speaking engagements. BM has also served as a consultant to
Feissel M, Cohen Y, Capellier G, et al. Procalcitonin levels to guide antibiotic
BioMérieux and Thermo-Fisher. SS has received lecture fees and research
therapy in adults with non-microbiologically proven apparent severe sepsis:
support from Thermo-Fisher. CEL has received lecture fees from Brahms and
a randomised controlled trial. BMJ Open. 2013;3(2):112-15.
Merck Sharp & Dohme-Chibret. JC has received consulting and lecture fees
from Pfizer, Brahms, Wyeth, Johnson & Johnson, Nektar-Bayer, and Arpida. 8. Bloos F, Trips E, Nierhaus A, Briegel J, Heyland DK, Jaschinski U, Moerer O,
MW has received consulting and lectures fees from Merck Sharp & Dohme- Weyland A, Marx G, Grundling M, et al. Effect of sodium selenite
Chibret, Janssen-Cilag, Gilead, Astellas, Sanofi, and Thermo-Fisher. YS has administration and procalcitonin-guided therapy on mortality in patients
received unrestricted research grants from Thermo-fisher, BioMérieux, Orion with severe sepsis or septic shock: a randomized clinical trial. JAMA Intern
Pharma, and Pfizer. The remaining authors declare that they have no com- Med. 2016;176(9):1266–76.
peting interests. 9. Bouadma L, Luyt CE, Tubach F, Cracco C, Alvarez A, Schwebel C, Schortgen
F, Lasocki S, Veber B, Dehoux M, et al. Use of procalcitonin to reduce
patients’ exposure to antibiotics in intensive care units (PRORATA trial): a
Publisher’s Note multicentre randomised controlled trial. Lancet. 2010;375(9713):463–74.
Springer Nature remains neutral with regard to jurisdictional claims in 10. de Jong E, van Oers JA, Beishuizen A, Vos P, Vermeijden WJ, Haas LE, Loef
published maps and institutional affiliations. BG, Dormans T, van Melsen GC, Kluiters YC, et al. Efficacy and safety of
procalcitonin guidance in reducing the duration of antibiotic treatment in
Author details critically ill patients: a randomised, controlled, open-label trial. Lancet Infect
1 Dis. 2016;16(7):819–27.
Medical University Department, Kantonsspital Aarau, Tellstrasse, CH-5001
Aarau, Switzerland. 2Faculty of Medicine, University of Basel, Basel, 11. Deliberato RO, Marra AR, Sanches PR, Martino MD, Ferreira CE, Pasternak J,
Switzerland. 3Service de Réanimation Médicale, Université Paris Paes AT, Pinto LM, dos Santos OF, Edmond MB. Clinical and economic
7-Denis-Diderot, AP-HP, Paris, France. 4Service de Réanimation Médicale, impact of procalcitonin to shorten antimicrobial therapy in septic patients
Université Paris 6-Pierre-et-Marie-Curie, Paris, France. 5Département with proven bacterial infection in an intensive care setting. Diagn Microbiol
d’Epidémiologie Biostatistique et Recherche Clinique, AP-HP, Hôpitaux Infect Dis. 2013;76(3):266–71.
Universitaires Paris Nord Val de Seine, Paris, France. 6Department of 12. Hochreiter M, Schroeder S. Procalcitonin-based algorithm. Management of
Anesthesiology and Intensive Care Medicine, Krankenhaus Dueren, Dueren, antibiotic therapy in critically ill patients. Anaesthesist. 2011;60(7):661–73.
Wirz et al. Critical Care (2018) 22:191 Page 11 of 11

13. Schroeder S, Hochreiter M, Koehler T, Schweiger AM, Bein B, Keck FS, von 32. Henriquez-Camacho C, Losa J. Biomarkers for sepsis. Biomed Res Int. 2014;
Spiegel T. Procalcitonin (PCT)-guided algorithm reduces length of antibiotic 2014:547818.
treatment in surgical intensive care patients with severe sepsis: results of a 33. Schuetz P, Maurer P, Punjabi V, Desai A, Amin DN, Gluck E. Procalcitonin
prospective randomized study. Langenbecks Arch Surg. 2009;394(2):221–6. decrease over 72 hours in US critical care units predicts fatal outcome in
14. Hochreiter M, Kohler T, Schweiger AM, Keck FS, Bein B, von Spiegel T, sepsis patients. Crit Care. 2013;17(3):R115.
Schroeder S. Procalcitonin to guide duration of antibiotic therapy in 34. Schuetz P, Litke A, Albrich WC, Mueller B. Blood biomarkers for personalized
intensive care patients: a randomized prospective controlled trial. Crit Care. treatment and patient management decisions in community-acquired
2009;13(3):R83. pneumonia. Curr Opin Infect Dis. 2013;26(2):159–67.
15. Layios N, Lambermont B, Canivet JL, Morimont P, Preiser JC, Garweg C, 35. Jones AE, Shapiro NI, Trzeciak S, Arnold RC, Claremont HA, Kline JA,
Ledoux D, Frippiat F, Piret S, Giot JB, et al. Procalcitonin usefulness for the Emergency medicine shock research network I. Lactate clearance vs central
initiation of antibiotic treatment in intensive care unit patients. Crit Care venous oxygen saturation as goals of early sepsis therapy: a randomized
Med. 2012;40(8):2304–9. clinical trial. JAMA. 2010;303(8):739–46.
16. Oliveira CF, Botoni FA, Oliveira CR, Silva CB, Pereira HA, Serufo JC, Nobre V. 36. Schuetz P, Mueller B. Procalcitonin: an effective screening tool and safe
Procalcitonin versus C-reactive protein for guiding antibiotic therapy in therapeutic decision making aid for emergency department patients with
sepsis: a randomized trial. Crit Care Med. 2013;41(10):2336–43. suspected sepsis. Ann Emerg Med. 2015;66(3):318–9.
17. Nobre V, Harbarth S, Graf JD, Rohner P, Pugin J. Use of procalcitonin to 37. Schuetz P, Mueller B. Biomarker-guided de-escalation of empirical therapy is
shorten antibiotic treatment duration in septic patients: a randomized trial. associated with lower risk for adverse outcomes. Intensive Care Med. 2014;
Am J Respir Crit Care Med. 2008;177(5):498–505. 40(1):141.
18. Iankova I, Thompson-Leduc P, Kirson NY, Rice B, Hey J, Krause A, Schonfeld 38. Schuetz P, Briel M, Christ-Crain M, Stolz D, Bouadma L, Wolff M, Luyt CE,
SA, DeBrase CR, Bozzette S, Schuetz P. Efficacy and safety of procalcitonin Chastre J, Tubach F, Kristoffersen KB, et al. Procalcitonin to guide initiation
guidance in patients with suspected or confirmed sepsis: a systematic and duration of antibiotic treatment in acute respiratory infections: an
review and meta-analysis. Crit Care Med. 2018;46(5):691–8. individual patient data meta-analysis. Clin Infect Dis. 2012;55(5):651–62.
19. Andriolo BN, Andriolo RB, Salomao R, Atallah AN. Effectiveness and safety of 39. FDA clears test to help manage antibiotic treatment for lower respiratory
procalcitonin evaluation for reducing mortality in adults with sepsis, severe tract infections and sepsis. https://www.fda.gov/NewsEvents/Newsroom/
sepsis or septic shock. Cochrane Database Syst Rev. 2017;1:CD010959. PressAnnouncements/ucm543160.htm. Accessed 15 Jan 2018.
20. Schuetz P, Wirz Y, Sager R, Christ-Crain M, Stolz D, Tamm M, Bouadma L, 40. Schuetz P, Wirz Y, Mueller B. Reassessment of a meta-analysis of
Luyt CE, Wolff M, Chastre J, et al. Procalcitonin to initiate or discontinue procalcitonin-guided antibiotic therapy for lower respiratory tract
antibiotics in acute respiratory tract infections. Cochrane Database Syst Rev. infections—authors’ reply. Lancet Infect Dis. 2018;18(2):141.
2017;10:CD007498. 41. Jensen JU, Heslet L, Jensen TH, Espersen K, Steffensen P, Tvede M.
21. Schuetz P, Wirz Y, Sager R, Christ-Crain M, Stolz D, Tamm M, Bouadma L, Procalcitonin increase in early identification of critically ill patients at high
Luyt CE, Wolff M, Chastre J, et al. Effect of procalcitonin-guided antibiotic risk of mortality. Crit Care Med. 2006;34(10):2596–602.
treatment on mortality in acute respiratory infections: a patient level meta- 42. Roberts RR, Hota B, Ahmad I, Scott RD 2nd, Foster SD, Abbasi F, Schabowski
analysis. Lancet Infect Dis. 2018;18(1):95–107. S, Kampe LM, Ciavarella GG, Supino M, et al. Hospital and societal costs of
22. Schuetz P, Briel M, Christ-Crain M, Wolbers M, Stolz D, Tamm M, Bucher H, antimicrobial-resistant infections in a Chicago teaching hospital:
Mueller B. Procalcitonin to initiate or withhold antibiotics in acute implications for antibiotic stewardship. Clin Infect Dis. 2009;49(8):1175–84.
respiratory tract infections (protocol). Cochrane Database Syst Rev 2008; 43. Classen DC, Jaser L, Budnitz DS. Adverse drug events among hospitalized
(Issue 4). Art. No.: CD007498. Medicare patients: epidemiology and national estimates from a new
23. Stewart LA, Clarke M, Rovers M, Riley RD, Simmonds M, Stewart G, Tierney approach to surveillance. Jt Comm J Qual Patient Saf. 2010;36(1):12–21.
JF, Group P-ID. Preferred reporting items for systematic review and meta- 44. Garnacho-Montero J, Gutierrez-Pizarraya A, Escoresca-Ortega A, Corcia-
analyses of individual participant data: the PRISMA-IPD statement. JAMA. Palomo Y, Fernandez-Delgado E, Herrera-Melero I, Ortiz-Leyba C, Marquez-
2015;313(16):1657–65. Vacaro JA. De-escalation of empirical therapy is associated with lower
24. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, Clarke mortality in patients with severe sepsis and septic shock. Intensive Care
M, Devereaux PJ, Kleijnen J, Moher D. The PRISMA statement for reporting Med. 2014;40(1):32–40.
systematic reviews and meta-analyses of studies that evaluate health care 45. Muller F, Christ-Crain M, Bregenzer T, Krause M, Zimmerli W, Mueller B,
interventions: explanation and elaboration. J Clin Epidemiol. 2009; Schuetz P, Pro HSG. Procalcitonin levels predict bacteremia in patients with
62(10):e1–34. community-acquired pneumonia: a prospective cohort trial. Chest. 2010;
25. Stolz D, Smyrnios N, Eggimann P, Pargger H, Thakkar N, Siegemund M, 138(1):121–9.
Marsch S, Azzola A, Rakic J, Mueller B, et al. Procalcitonin for reduced
antibiotic exposure in ventilator-associated pneumonia: a randomised study.
Eur Respir J. 2009;34(6):1364–75.
26. Shankar-Hari M, Phillips GS, Levy ML, Seymour CW, Liu VX, Deutschman CS,
Angus DC, Rubenfeld GD, Singer M, Sepsis definitions task F. Developing a
new definition and assessing new clinical criteria for septic shock: for the
third international consensus definitions for sepsis and septic shock (Sepsis-
3). JAMA. 2016;315(8):775–87.
27. Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P,
Schunemann HJ, Group GW. GRADE: an emerging consensus on rating quality
of evidence and strength of recommendations. BMJ. 2008;336(7650):924–6.
28. Thompson SG, Turner RM, Warn DE. Multilevel models for meta-analysis,
and their application to absolute risk differences. Stat Methods Med Res.
2001;10(6):375–92.
29. Turner RM, Omar RZ, Yang M, Goldstein H, Thompson SG. A multilevel
model framework for meta-analysis of clinical trials with binary outcomes.
Stat Med. 2000;19(24):3417–32.
30. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, Sevransky
JE, Sprung CL, Douglas IS, Jaeschke R, et al. Surviving Sepsis Campaign:
international guidelines for management of severe sepsis and septic shock,
2012. Intensive Care Med. 2013;39(2):165–228.
31. Jun KR, Lee JN, Song SA, Oh SH, Lee JY, Shin JH, Kim HR. Serial changes in
serum procalcitonin, interleukin 6, and C-reactive protein levels according to
non-specific surgical stimulation. Clin Chemi Lab Med. 2015;53(4):549–58.