Journal of Infection 82 (2021) 125–134

Contents lists available at ScienceDirect

Journal of Infection
journal homepage: www.elsevier.com/locate/jinf

Review

Impact of the timeliness of antibiotic therapy on the outcome of

patients with sepsis and septic shock
Sandra A. Asner a,b, Florian Desgranges b, Irene T. Schrijver b, Thierry Calandra b,∗
a
Pediatric Infectious Diseases and Vaccinology Unit, Department Mother-Woman-Child, Lausanne University Hospital, University of Lausanne, Switzerland
b
Infectious Diseases Service, Department of Medicine, Lausanne University Hospital, University of Lausanne, Rue du Bugnon 46, Lausanne CH-1011,
Switzerland

a r t i c l e i n f o s u m m a r y

Article history: Objectives: To review the impact of the timeliness of antibiotic therapy on the outcome of patients with
Accepted 8 March 2021 sepsis or septic shock.
Available online 17 March 2021
Methods: We searched MEDLINE, EMBASE, the Cochrane Library, Web of Science, Open-SIGLE databases,
Keywords: ClinicalTrials.gov and the metaRegister of Controlled Trials on July 27, 2020 for relevant studies on the
Review timing of antibiotic therapy in adult patients with sepsis or septic shock. The primary outcome measure
Sepsis was all-cause crude or adjusted mortality at reported time points.
Septic shock Results: We included 35 sepsis studies involving 154,330 patients. Nineteen studies (54%) provided in-
Antibiotic formation on the appropriateness of antibiotic therapy in 20,062 patients of whom 16,652 patients (83%)
Timing received appropriate antibiotics. Twenty-four studies (68.6%) reported an association between time-to-
Mortality antibiotics and mortality. Time thresholds associated with patient’s outcome varied considerably between
studies consisting of a wide range of time cutoffs (1 h, 125 min, 3 h or 6 h) in 14 studies, hourly delays
(derived from the analyses of time intervals ranging from to 1 to 24 h) in 8 studies or time-to-antibiotic
in 2 studies. Analyses of subsets of studies that focused on patients with septic shock (11 studies, 12,756
patients) or with sepsis (6 studies, 24,281 patients) yielded similar results.
Conclusions: While two-thirds of sepsis studies reported an association between early administration of
antibiotic therapy and patient outcome, the time-to-antibitiocs metrics varied significantly across studies
and no robust time thresholds emerged.
© 2021 The Authors. Published by Elsevier Ltd on behalf of The British Infection Association.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)

Introduction outcome.18–21 This prompted us to review the literature on the

timeliness of antibiotic therapy on the outcome of patients with
Current global estimates indicate that 49 million cases of sep- sepsis with the aim to identify targets for time-to-antibiotics asso-
sis occur annually worldwide with about 11 million deaths.1 Sepsis ciated with favorable outcomes.
results in mortality rates of about 10% in patients with sepsis and
of more than 40% in patients with septic shock.2 Administration
Materials and methods
of appropriate antimicrobial agents is a cornerstone of sepsis man-
agement guidelines and bundles.3–8 However, there is considerable
Search strategy
controversy on the target for time-to-antibiotic therapy in patients
with sepsis.9–14 Several studies indicated that early administration
The literature search was performed on July 27, 2020 in MED-
of antibiotics reduces sepsis mortality.3–5 , 15–17 Yet, other studies
LINE, EMBASE.com, and the Cochrane Library Wiley for relevant
found no association between early antibiotic therapy and patient’s
studies published any year and in any language. We also searched
proceedings of conference from 2008 to 2020 using the Web of
Science and unpublished studies using Open-SIGLE databases, the
Abbreviations: ICU, Intensive care units; ED, Emergency Department; LOS, Length US National Institutes of Health Ongoing Trials Register Clinical-
of hospital stay; RR, Risk ratio; OR, Odds ratio; CI, Confidence interval; IQR, In-
Trials.gov (https://clinicaltrials.gov) and the metaRegister of Con-
terquartile range; SOFA Score, Sequential Organ Failure Assessment, IRB, Institute
for Research in Biomedicine. trolled Trials (www.controlled-trials.com). We also reviewed the
∗
Corresponding author. reference lists of articles for additional relevant studies. The search
E-mail address: [email protected] (T. Calandra). criteria used the following Medical Subject Headings terms: (sep-

https://doi.org/10.1016/j.jinf.2021.03.003
0163-4453/© 2021 The Authors. Published by Elsevier Ltd on behalf of The British Infection Association. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/)
S.A. Asner, F. Desgranges, I.T. Schrijver et al. Journal of Infection 82 (2021) 125–134

sis OR severe sepsis OR septic shock) AND (antimicrobial agents OR (IQR) were reported for continuous outcomes. Given that we did
antibacterial agents OR antibiotics) AND (Time-factor OR time-to- not perform a systematic review with meta-analysis, we do not
treatment OR time-to-antibiotic OR timing). provide information on cumulative OR, RR or on the heterogene-
ity of studies (I2 ). Statistical analyses and figure design were per-
Data extraction and eligibility criteria formed using IBM SPSS Statistics version 26.0 (IBM Corp., New
York, USA) and R statistical software 3.6.0 (R Foundation for Sta-
Two authors (SAA and TC) independently reviewed the articles tistical Computing, Vienna, Austria).
selected for full-text screening of eligibility criteria. We included
studies of adult patients (≥ 18 years) with sepsis, severe sep- Results
sis or septic shock as defined in the 1991 ACCP/SCCM Consensus
Conference definitions (retrospectively labelled Sepsis-1), the 2001 Selection of studies
SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Confer-
ence (retrospectively labelled Sepsis-2), The Third International Fig. 1 shows the flow diagram for the selection of studies. We
Consensus Definitions for Sepsis and Septic Shock (Sepsis-3), codes screened the abstracts of 4409 records and selected 94 articles for
(995.92 and 785.52) of the International Classification of Diseases, full-text screening. The Cohen’s kappa coefficient for agreement
9th Edition and the predisposition, infection, response and organ between the two reviewers was 0.78.23 After full-text screening,
failure (PIRO) score.2 , 22–24 For the analyses of the impact of the we included 35 studies, of which 34 were observational cohorts
timing of antibiotic therapy in subgroups of sepsis patients, we (retrospective: 20 studies, prospective: 14 studies) and one was an
assumed that patients with “severe sepsis” (according to Sepsis-1 analysis of patients enrolled in a randomized controlled trial.
and Sepsis-2) and the patients with “sepsis” (according to Sepsis-3)
had similar grades of sepsis severity. Eligible studies included ran-
domized controlled trials, observational cohort studies and anal- Study characteristics
yses of prospectively collected datasets providing information on
the timing of antibiotic therapy. We excluded animal studies, case Table 1 shows a summary of the 35 studies that included
reports, case studies, meta-analyses, editorials and letters to the 154,330 patients (median: 1058 patients per study, range: 117 to
editor. 49,331). The enrolment period ranged from 1989 to 2020 (Fig. 2).
The median duration of studies was 2.1 years (range: 0.4 to 15.5).
Patient characteristics The criteria used for the selection of patients varied between stud-
ies (Supplementary Table 1). The set point or “time zero”, the time
We recorded information on the selection of patients, inclu- cutoffs (1, 3, 6, 12, 24 or 48 h) and the time intervals (hourly in-
sion criteria, the duration and time period of the study, the setting crements up to 24 h) used for assessing the timing of antibiotic
(emergency department versus intensive care units versus wards), therapy varied considerably between studies (Table 1). The set-
the study design and the total number of patients in each sepsis tings were emergency departments in 21 studies, ICUs in 8, hospi-
severity stratum. We also extracted data on other key study char- tal wards in 1 and a combination of emergency departments, ICUs
acteristics such as the set point (i.e. time zero), the time cutoffs and wards in 5 (Table 1). Septic shock and sepsis with or with-
or intervals used for assessing the timing and impact of antibi- out organ dysfunction occurred in 51,094 patients (33.1%), 61,094
otic therapy, the assessment of the appropriateness of antibiotic patients (39.6%) and 13,444 patients (8.7%), respectively. In nine
therapy and the study endpoints. Criteria used for the analysis of studies, 28,698 patients (18.6%) with sepsis, severe sepsis and sep-
antibiotic appropriateness were based on in vitro susceptibility of tic shock were grouped together. All but one study used mortal-
causative pathogens in case of microbiologically-documented infec- ity endpoints, which was adjusted mortality in 29 studies (83%),
tions or on antibiotic therapy management guidelines in case of all-cause crude mortality in five (14.3%) and a combination of ad-
clinically-documented infections.25 justed and unadjusted mortality in one. The time points for assess-
ing mortality were the end of the ICU or the hospital stay in 24
Subgroup analyses studies (68.6%), day 28 or day 30 in 7 (20.0%), various time points
in three (8.6%) and one year in one (2.8%).
We performed subgroup analyses to investigate the effects of
sepsis severity strata (sepsis versus septic shock) on study end- Impact of the timeliness of antibiotic therapy on mortality
points. We also conducted post-hoc stratified analyses of studies
that addressed the appropriateness of antibiotic administration. In 24 studies (68.6%) time-to-antibiotics was associated with in-
hospital mortality (18 studies),3–5 , 15–17 , 27 , 28 , 31–40 mortality at other
Outcome measures time points (5 studies)38 , 41–44 and the ICU or hospital length of
stay (one study)30 (Table 1). The time thresholds associated with
The primary outcome was all-cause mortality at the time points patient’s outcome consisted of various time cutoffs (1 h, 125 min,
reported in the study. All but five studies26–30 adjusted mortality 3 h or 6 h) in 14 studies, hourly delays (based on the analyses of
using multivariable analyses which included covariates considered time intervals ranging from to 1 to 24 h) in 8 studies and time-
to be potential confounders such as age, gender, Acute Physiol- to-antibiotic in two studies. Fig. 3 shows the odds ratio plots for
ogy and Chronic Health Evaluation II (APACHE II), Charlson index mortality according to three time-to-antibiotics parameters (hourly
measures, site of infection, hypotension, SOFA score, lactate clear- delays, 1 h and 3 h time cutoffs). Arrival or registration in the
ance, fluid resuscitation and the use of vasopressors. Secondary emergency department (ED) was identified as time zero in 10 stud-
outcomes included admission to the ICU, length of ICU or of hospi- ies,16 , 20 , 27 , 29 , 34 , 36 , 38 , 45–47 of which 4 reported an association be-
tal stay (LOS), progression from sepsis/severe sepsis to septic shock tween time cutoffs (hourly delays, 1 h, 3 h) and patients out-
and mortality at fixed time-points (one, three or twelve months) come in a multivariable analysis.16 , 34 , 36 , 38 Recognition of symp-
after hospital discharge. The principal summary measures for di- toms or signs of sepsis was considered as time zero in 9 stud-
chotomous outcomes were odds ratio (OR) with 95% confidence in- ies,17 , 26 , 31 , 33 , 40 , 41 , 44 , 48 , 49 of which 5 reported an association be-
terval (CI) (as provided by the authors or calculated based on avail- tween time cutoffs (1 h, 125 min, 3 h) and mortality in multivari-
able data) and risk ratios (RR). Medians with interquartile range able analyses.17 , 31 , 33 , 40 , 44

126
S.A. Asner, F. Desgranges, I.T. Schrijver et al. Journal of Infection 82 (2021) 125–134

Fig. 1. Flowchart of study selection

Fifty-nine studies were excluded after full text screening. Thirty-two studies did not evaluate the timing of antimicrobial therapy, 18 reported on conditions other than sepsis
(bacteremia in 10, febrile neutropenia in 5 and pneumonia in 3). Nine studies were excluded as they focused on predictors for delayed antibiotic administration (n = 5) or
enrolled pediatric patients (n = 4).

Appropriateness of antibiotic therapy or with a start of antibiotics more than 3 h after triage in
the emergency department.50 No association between early an-
Nineteen studies (54%) provided information on the appropri- tibiotic therapy and outcome was noted in the other six stud-
ateness of antibiotic therapy in 20,062 patients (13.0% of the total ies.18 , 26 , 41 , 46–48
patient population) of whom 16,652 patients (83%) received ap-
propriate antibiotics. Among these 19 studies, 11 based the assess- Sepsis
ment on in vitro susceptibility criteria and 8 on combined clinical Twenty studies (57.1%) enrolled patients with sepsis and or-
and microbiological criteria. Ten of these 19 studies reported an gan dysfunction but without shock, of which six analysed the re-
association between time-to-antibiotics and mortality with hourly lationship between time-to-antibiotic and mortality in 24,281 pa-
delays or using various timecutoffs ranging from 1 h, 3 h or 6 h tients.16 , 21 , 26 , 32 , 37 , 47 In three studies that included 92.0% of the pa-
delays and onwards. Three studies reported on associations be- tients, mortality increased with an hourly delay in antibiotic ther-
tween one-hour delay or hourly delays and mortality among pa- apy16 , a longer time-to-antibiotic32 or a delay of more than 6 h in
tients with septic shock.3 , 30 , 50 All but one study30 conducted mul- the administration of empirical antibiotic therapy.37 Of note, one of
tivariable analyses. these studies reported an 8.0% risk of progression to septic shock
when antibiotic therapy was delayed.32 There was no association
Septic shock and sepsis studies between early antibiotic therapy and mortality in the other three
studies.
We then examined the effects of the timing of antibiotic ther-
apy in studies that included patients with septic shock or with sep- Discussion
sis.
Overall, two-thirds of the studies included in this review re-
Septic shock ported an association between time-to-antibiotics and mortality
Twenty-one studies (60%) enrolled patients with septic shock, (Table 1). Yet, the time metrics for antibiotic delivery associated
of which 11 specifically analysed the impact of the time-to- with patient’s outcome varied considerably among studies and the
antibiotic on mortality in 12,756 patients.3 , 16 , 18 , 26 , 30 , 40 , 41 , 46–48 , 50 time thresholds were wide ranging from one to six hours. The
In five studies that included 79.6% of the patients, the in-hospital studies that identified an increase in mortality with each hour de-
mortality or the length of stay in the ICU increased significantly lay in the start of antibiotics used risk-adjusted linear models over
with each hour delay in the administration of antibiotics3 , 16 , 30 , 40 time intervals of 6 to 12 h, which are likely to be influenced by the

127
 Table 1
Study characteristics.

S.A. Asner, F. Desgranges, I.T. Schrijver et al.

Assessment of
Study design, Number of Classification by Set point for antibiotic therapy
Author, year, number of centers patients sepsis severity assessing timing of Time cutoffs and (percent with
reference Setting or units included (number of patients) antibiotic therapy intervals Primary endpoint appropriate therapy) Main findings

Abe, 201948 ICU Prospective 1124 Severe Sepsis (421) Recognition of sepsis 0–60, 61–120, In-hospital mortality No No association between time
observational, 59 Septic shock (703) at the ED, ward or 121–180,181–240, to antibiotic administration
centers ICU 241–360, (1 h or 3 h cutoffs or time as
361–1440 min and continuous variable) and
continuous variable in-hospital mortality
Alam, 201826 ED Randomized 2631∗ Sepsis (1003) Severe Recognition of sepsis Hourly up to 4 h and All-cause crude No No impact of prehospital
controlled sepsis (1525) Septic or shock by more than 4 h mortality at day 28 antibiotic administration or of
open-label trial, 34 shock (103) paramedics time to antibiotics
centers administration on day 28
mortality
Ascuntar, ED and Prospective 2454 Sepsis, septic shock Admission to the ED 1 h, 3 h In-hospital mortality Yes (76.5%) No association between
202046 ICU observational (869) antibiotic administration
cohort study, 3 within 1 h or 3 h and
centers in-hospital mortality
Ballester, ED Retrospective 153 Sepsis, severe sepsis ED arrival Median In-hospital crude No Door-to-antibiotic time were
201827 observational and septic shock door-to-antibiotic mortality associated with mortality.
study, single center (153) time
Bloos-1, ICU Prospective 1011 Severe sepsis (379) Documentation of 0 to 1 h, 1 to 3 h, 3 All-cause crude Yes (58%) No linear association between
201419 observational Septic shock (632) first infection-related to 6 h, more than 6 h mortality at day 28 time to antibiotic and day 28
multicenter cohort organ dysfunction mortality
study, 44 centers
Bloos-2, ICU Prospective 4183 Severe sepsis (1001); Documentation of Hourly (time range All-cause day 28 Yes 2% increase in 28-day
201742 multicenter cohort septic shock (3182) first infection-related not specified) mortality mortality for each 1 h delay in
study 44 centers organ dysfunction initiation of antimicrobial
therapy
128

Castaño, ED Prospective cohort 705 Severe sepsis (632) ED arrival Hourly (time range Inpatient mortality Yes (75% - 85.6% No association between hourly
201945 study, three Septic shock (73) not specified) and length of dependent upon delay in antibiotic
centers hospital stay criteria used) administration and in-patient
mortality
De Groot, ED Prospective 1168 Sepsis (1168) PIRO Registration in the 0 to 1 h, 1 to 3 h, Number of surviving Yes (76%) No association between time
201520 observational scores (1–7: 413; ED more than 3 h days outside the to antibiotics and surviving
cohort study, three 8–14: 532; >14: hospital at day 28 days outside hospital or
centers 223) mortality.
Ferrer-1, ICU Prospective 2796 Severe sepsis and Symptom 0-1 h, 1–3 h, 3–6 h, Propensity-adjusted No Early antibiotic therapy
200933 observational septic shock (2796) recognition no antibiotic in the hospital mortality (within 1 h vs. no treatment
study, 77 ICUs first 6 h within 6 h) associated with
lower hospital mortality
Ferrer-2, ED or Retrospective 17,990 Severe sepsis (6432) Triage (ED) or One-hour time Adjusted hospital No More than 2-h delay in
201415 ICU analysis of a and septic shock symptoms period up to 6 h mortality antibiotic administration
prospectively (11,558) recognition (ICU) associated with increase
collected dataset hospital mortality
Gaieski, ED Retrospective 261 Severe sepsis (126) ED triage One-hour time point Adjusted in-hospital Yes (85%) Association between time (one
20104 cohort study, Septic shock (135) up to 5 h mortality hour) from triage to

Journal of Infection 82 (2021) 125–134

single center appropriate antibiotic
administration and in-hospital
mortality.
Husabo, ED Retrospective 1559 Sepsis (1559) ED triage 0-1 h, 1-2 h, 2-3 h, 30-day all-cause No Lower mortality rates when
202043 cohort study, 24 3-4 h and > 4 h; mortality antibiotics administred
centers time to first between 2 and 3 h after ED
antibiotic dose admission compared to 2 h or
3h
Jalili, 201329 ED Prospective cohort 145 Sepsis (145) ED arrival 0 to 1 h, 1 to 2 h, Crude sepsis-related Yes Association between
study, one single more than 2 h death or discharge door-to-antibiotic time and
center; mortality, but only in patients
with APACHE II scores of 21 or
higher
(continued on next page)
 S.A. Asner, F. Desgranges, I.T. Schrijver et al.
Table 1 (continued)

Author, year, Setting Study design, Number of Classification by Set point for Time cutoffs and Primary endpoint Assessment of Main findings
reference number of centers patients sepsis severity assessing timing of intervals antibiotic therapy
or units included (number of patients) antibiotic therapy (percent with
appropriate therapy)

Joo, 201434 ED Retrospective 591 Severe sepsis and ED arrival 0–3 h versus greater Adjusted in-hospital No Association between timely
cohort study septic shock (591) than 3 h mortality (within 3 h) antibiotic
(sepsis registry), administration and improved
single center outcome (survival, reversal of
organ failure and shorter
length of stay)
Kim, 201835 ED Retrospective, 117 Sepsis and septic Time from triage to One-hour intervals Unadjusted and No Association between time
observational shock (117) antibiotic(s) up to 3 h and >3 h adjusted in-hospital from triage to antibiotic
cohort study, mortality administration and in-hospital
single center mortality
Ko, 202050 ED Prospective 2229 Septic shock (2229) ED triage 0 - 1 h, 1-2 h, 2-3 h, In-hospital mortality Yes Early antibiotic (<1 h)
observational > 3 h and 0–3 h associated with lower
cohort study, 10 in-hospital mortality with
centers propensity score analysis. No
linear association between
hourly delays and in-hospital
mortality.
Kumar, 20063 ICU Retrospective 2154 Septic shock (2154) Initial onset of One-hour time point Adjusted survival to Yes (100%) Association between hourly
analysis of three recurrent or up to 6 h, plus 4 hospital discharge delay in appropriate antibiotic
cohorts, 10 centers persistent additional time administration and mortality
(14 ICUs) hypotension cutoffs up to more
than 36 h
129

Liu, 201716 ED Retrospective 35,000 Sepsis (12,122) ED registration 30-min increment Risk-adjusted No For each sepsis severity
study, 21 hospitals Severe sepsis from 0 to 6 h hospital mortality stratum, hourly delays in
(18,210) Septic shock antibiotic administration (up
(4668) to 6 h) associated with
increased adjusted odds ratios
for hospital mortality
Londoño, ED Prospective cohort 884 Severe sepsis and ED arrival Less than 1 h or 3 h, Adjusted in-hospital Yes (83%) Administration of antibiotic
201836 study, three septic shock (884) hourly up to 24 h mortality within 1 or 3 h associated
hospitals with reduced mortality
Lueangarum Medical Retrospective 229 Sepsis (31) Severe Sepsis recognition 0 to 1 h, 1 to 6 h, Crude 28-day and Yes (70%) More than 3-h delay in
201244 wards cohort study, sepsis (58) Septic more than 3 or 6 h overall mortality antibiotic administration
single center shock (140) associated with higher overall
mortality
Nygard, Infectious Prospective 220 Severe sepsis (220) Hospital admission < 6 h, ≥ 6h In-hospital mortality Yes (76−82%) Delay in administration of
201437 diseases observational antibiotics of 6 h or more
or cardi- study, single center associated with increased
ology mortality
ward and

Journal of Infection 82 (2021) 125–134

ICUs
Peltan, ED Retrospective 10,811 Sepsis and septic ED arrival Door-to-antibiotic Adjusted one-year No Door-to-antibiotic time (3-h
201938 cohort study, fours shock (10,811) time, ≤ 1 h vs. > 1 h, mortality. plus cutoff) associated with
centers ≤ 1 h vs. each hour in-hospital, 30-day, increased adjusted odds of
beyond first hour up 90-day and 1-year one-year mortality
to 6 h, ≤ 3 h vs. > 3 mortality
h
Peng, 201831 ICU Retrospective 541 Sepsis (382) Septic Diagnosis of sepsis ≤ 1 h, 1–24 h, ICU and hospital Yes Delayed appropriate antibiotic
study, single center shock (159) 24–48 h, > 48 h mortality therapy associated with higher
ICU and in-hospital mortality
(continued on next page)
 S.A. Asner, F. Desgranges, I.T. Schrijver et al.
Table 1 (continued)

Author, year, Setting Study design, Number of Classification by Set point for Time cutoffs and Primary endpoint Assessment of Main findings
reference number of centers patients sepsis severity assessing timing of intervals antibiotic therapy
or units included (number of patients) antibiotic therapy (percent with
appropriate therapy)

Pruinelli, All hospi- Retrospective 5072 Severe sepsis and Identification of 0 to 6 h Adjusted in-hospital No Association between antibiotic
201817 talized cohort study, six septic shock (5072) sepsis (ICD-9 codes) mortality delivery delays above 125 min
patients hospitals and mortality
Puskarich, ED Pre-planned 291 Septic shock (291) Triage in the ED and Hourly up to 6 h Adjusted in-hospital Yes (91%) No association between time
201118 analysis of RCT, 3 shock recognition mortality to antibiotics (up to 6 h) and
centers in-hospital mortality
Ryoo, 201541 ED Retrospective 426 Septic shock (426) Shock recognition One-hour increment Adjusted 28-day Yes (98%) No association between hourly
cohort study, up to 5 h mortality delay in antibiotic
single center administration and mortality
Seok, 202047 ED Prospective cohort 482 High grade ED arrival 0-1 h, 0–3 h, hourly Day 7, day 14 and Yes (77.8%) No association between time
study, single center infection/sepsis day 28 mortality to antibiotics and outcomes in
(279), septic shock overall and subgroup analyses
(203) including (patients with septic
shock or with appropriate
antibiotics)
Seymour-1, ED Retrospective 49,331 Severe sepsis First medical contact Hourly up to 12 h Risk-adjusted No Longer time to the
20175 cohort study, 149 (26,995) Septic shock (prehospital) and in in-hospital mortality administration of antibiotics
hospitals (22,336) ED associated with higher
risk-adjusted in-hospital
mortality
Seymour-2, ED Retrospective 2683 Sepsis and septic First medical contact 0 to 6 h, > 6 to 12 h, Risk-adjusted No ED antibiotic delay associated
201739 cohort study, nine shock (2683) > 12 h in-hospital mortality with in-hospital mortality
130

hospitals (21 EMS)

Suberviola ICU Prospective 342 Septic shock (342) Documentation of 0 to 1 h, 1 to 6 h, > 6 ICU and in-hospital Yes (88%) Association between antibiotic
Canas, observational septic shock h mortality treatment delay and increased
201540 cohort study mortality
Tan, 201928 ED Retrospective chart 261 Sepsis (41), severe Sepsis recognition ≤ 3 h, > 3 h In-hospital mortality No Lower in-hospital mortality in
review, single sepsis (76) and patients who received
center septic shock (144) antibiotics within 3 h
Whiles, ED Retrospective 3929 Severe sepsis (3929) ED triage or ED One-hour time point Adjusted in-hospital No Longer time to initial
201732 cohort study arrival time up to 5.99 h, 6 to mortality. antimicrobial administration
8.99 h, 9 to 11.99 h, Progression from associated with mortality and
12 to 17.99 and 18 severe sepsis to progression to septic shock
to 24 h septic shock
Wisdom, ED Retrospective 220 Sepsis (102) Severe ED triage by nurse ≤1 h, 1 to 3 h, 3 to Adjusted in-hospital No No association between time
201521 review of patients sepsis (118) 6 h, > 6 h mortality from triage to administration
presenting to ED of antibiotic and mortality in
with sepsis, single the entire cohort; trend in
center patients with severe sepsis
who received antibiotics after

Journal of Infection 82 (2021) 125–134

6h
Yokota, ICU Retrospective 1279 Severe sepsis (403) Severe sepsis 0-1 h Adjusted in-hospital Yes (74%) No association between
201449 cohort study, Septic shock (876) recognition mortality antibiotic administration and
single center reduced mortality
Zhang, All hospi- Retrospective 1058 Severe sepsis (311) Documentation of 0-1 h, hourly up to ICU and hospital Yes (70%) Association between time to
201530 talized cohort study, Septic shock (747) in-vitro susceptibility 24h length of stay appropriate antibiotic therapy
patients single center to antibiotics (1-h increments) and ICU or
hospital LOS
Abbreviations: APACHE II: Acute Physiology, Age, Chronic Health Evaluation II. ED: emergency department. EMS: emergency medical services. H: hour(s). ICU: intensive care unit. LOS: length of hospital stay. Min: minutes.
PIRO: predisposition, infection, response, and organ dysfunction. RCT: randomized controlled trial.
∗
after exclusion of 41 patients with other diagnoses.
S.A. Asner, F. Desgranges, I.T. Schrijver et al. Journal of Infection 82 (2021) 125–134

Fig. 2. Study enrolment periods

The graph displays the patient’s enrolment period of the 34 studies that provided this information. The black or gray filling color symbolizes studies that found (black) or
did not find (gray) an association between time-to-antibiotics and patient’s outcome.

increased odds of mortality associated with long delays in the ini- in the odds of mortality among patients in whom antibiotics were
tiation of antibiotic therapy (Fig. 3).3 , 5 , 16 , 38 , 42 Of note, the largest administred within one hour.53 Yet, their findings were largely in-
studies did not find associations between early antibiotic deliv- fluenced by one study that reported a 7.5% linear increase in the
ery and patient’s outcome.19 , 20 , 45–48 Therefore, the available data risk of mortality after adjusting for numerous covariables includ-
do not allow making recommendation on the timing of antibiotic ing geographic locations.15 Other limitations are the inclusion of
dispensation with a great level of precision in patients with sep- studies that did not provide information on the type of antibiotics
sis.6 , 8–11 administered and tying together patients who received antibiotics
With a 3-fold larger number of included studies, this re- over long time intervals (more than one hour to six hours) after
view supports the results of two systematic reviews and meta- the arrival in the emergency department.
analyses.51 , 52 In a review of 11 studies, Sterling et al. reported The need for stratifying recommendations for time-to-antibiotic
no survival benefit with dispensation of antibiotics within one according to the severity of sepsis is well recognized.10 , 11 Given the
hour of detection of severe sepsis or septic shock or within three extremely high mortality rates of septic shock, antibiotics should
hours of triage in the emergency department.51 Likewise, a meta- be administered immediately in patients with septic shock. Sepsis
regression analysis of 13 studies concluded that there was no dif- is a continuum with no clear-cut zone of rarity,54 hence the diffi-
ference in mortality between patients receiving antibiotics within culty of providing treatment with robust predictive validity across
one or three hours after the onset of sepsis.52 In contrast, in a a broad range of disease probability. Notwithstanding that delays
meta-analysis of 10 studies Johnston et al. found a 33% reduction were also associated with increased mortality among patients with

131
S.A. Asner, F. Desgranges, I.T. Schrijver et al. Journal of Infection 82 (2021) 125–134

Fig. 3. Timeliness of antibiotic therapy and mortality of sepsis. Odds ratio plots for mortality with 95% confidence intervals according to the time-to-antibiotics reported as
hourly delays (panel a), 1 h (panel b) or 3 h (panel c) time cutoffs. Data were available from 28 studies.

sepsis, no clear time threshold emerged from available data for this While prompt antibiotic therapy was recommended for patients
subgroup. Conceivably, time thresholds for the administration of with septic shock and bacterial meningitis, the authors did not
antibiotics in patients with suspicion of sepsis could be tailored find evidence of worse outcome when initiation of therapy was
to the likelihood of infection. This approach is supported by a re- delayed (e.g. by 4 to 8 h) in patients with less severe infections.
cent review by Naucler et al. on the outcome of patients with A perilous tradeoff of delayed therapy is the risk of progression
bacterial infections of different sources and degrees of severity.55 from sepsis to septic shock. This was addressed in only one of the

132
S.A. Asner, F. Desgranges, I.T. Schrijver et al. Journal of Infection 82 (2021) 125–134

studies reviewed here and was found to be 8%.32 Assuming a 10% References
risk of death in sepsis and 40% in septic shock, an 8% risk of tran-
sition from sepsis to septic shock would result in an increase of 1. Rudd KE, Johnson SC, Agesa KM, Shackelford KA, Tsoi D, Kievlan DR,
et al. Global, regional, and national sepsis incidence and mortality,
mortality from 10 to 12.4%. Finding an equipoise between poten- 1990-2017: analysis for the Global Burden Of Disease Study. Lancet
tial morbidity criteria benefit and an increased risk of mortality is 2020;395(10219):200–11.
a delicate balancing act. Immunocompetent, non-neutropenic pa- 2. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M,
et al. The third international consensus definitions for sepsis and septic shock
tients with suspected sepsis without shock and a low probability (sepsis-3). JAMA 2016;315(8):801–10.
of infection could be a suitable patient population to investigate 3. Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, et al. Du-
whether watchful clinical observation and prompt initiation of an- ration of hypotension before initiation of effective antimicrobial therapy is
the critical determinant of survival in human septic shock. Crit Care Med
tibiotic upon documentation of infection is feasible and safe. 2006;34(6):1589–96.
Strengths of the current review are a large sample size, a rig- 4. Gaieski DF, Mikkelsen ME, Band RA, Pines JM, Massone R, Furia FF, et al. Im-
orous assessment of eligibility increasing the reliability of the re- pact of time to antibiotics on survival in patients with severe sepsis or septic
shock in whom early goal-directed therapy was initiated in the emergency de-
sults, subgroup analyses according to sepsis severity defined a pri- partment. Crit Care Med 2010;38(4):1045–53.
ori and post hoc analyses of studies addressing the appropriateness 5. Seymour CW, Gesten F, Prescott HC, Friedrich ME, Iwashyna TJ, Phillips GS,
of antibiotic therapy. Limitations predominantly relate to the risk et al. Time to treatment and mortality during mandated emergency care for
sepsis. N Engl J Med 2017;376(23):2235–44.
of bias in the included studies inherent to their observational de-
6. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, et al. Sur-
sign. A high degree of heterogeneity regarding the inclusion of pa- viving sepsis campaign: international guidelines for management of sepsis and
tients with various duration of sepsis and degrees of sepsis sever- septic shock: 2016. Intensive Care Med 2017;43(3):304–77.
ity, the use of differents definitions of time zero and the lack of 7. Howell MD, Davis AM. Management of sepsis and septic shock. JAMA
2017;317(8):847–8.
information on time-to-adequate antibiotic therapy may also im- 8. Levy MM, Evans LE, Rhodes A. The Surviving sepsis campaign bundle: 2018 up-
pact on the robustness of the findings. Few studies provided in- date. Crit Care Med 2018;46(6):997–10 0 0.
formation on the number of patients in each sepsis severity stra- 9. Singer M. Antibiotics for sepsis: does each hour really count, or is it incestuous
amplification? Am J Respir Crit Care Med 2017;196(7):800–2.
tum, considerably limiting the additional value of cumulative odd 10. Force IST. Infectious diseases society of America (IDSA) POSITION STATEMENT:
ratios or relative risks. Of note, no study provided information on why IDSA did not endorse the surviving sepsis campaign guidelines. Clin Infect
source control and therapeutic drug monitoring. Future studies on Dis 2018;66(10):1631–5.
11. Klompas M, Calandra T, Singer M. Antibiotics for sepsis-finding the equilibrium.
this topic should provide information on these critical parameters. JAMA 2018;320(14):1433–4.
12. Mi MY, Klompas M, Evans L. Early administration of antibiotics for suspected
sepsis. N Engl J Med 2019;380(6):593–6.
Conclusions 13. Chen AX, Simpson SQ, Pallin DJ. Sepsis guidelines. N Engl J Med
2019;380(14):1369–71.
Two-thirds of studies included in this review reported an asso- 14. Singer M, Inada-Kim M, Shankar-Hari M. Sepsis hysteria: excess hype and un-
realistic expectations. Lancet 2019;394(10208):1513–14.
ciation between early antibiotics and mortality. However, the time 15. Ferrer R, Martin-Loeches I, Phillips G, Osborn TM, Townsend S, Dellinger RP,
metrics linking early allocation of antibiotics with mortality varied et al. Empiric antibiotic treatment reduces mortality in severe sepsis and septic
significantly across studies and no robust time threshold emerged shock from the first hour: results from a guideline-based performance improve-
ment program. Crit Care Med 2014;42(8):1749–55.
from the overall study population or from subsets of studies that 16. Liu VX, Fielding-Singh V, Greene JD, Baker JM, Iwashyna TJ, Bhattacharya J,
included patients with sepsis or septic shock. et al. The timing of early antibiotics and hospital mortality in sepsis. Am J Respir
Crit Care Med 2017;196(7):856–63.
17. Pruinelli L, Westra BL, Yadav P, Hoff A, Steinbach M, Kumar V, et al. Delay
Declarations of Competing Interest within the 3 h surviving sepsis campaign guideline on mortality for patients
with severe sepsis and septic shock. Crit Care Med 2018;46(4):500–5.
18. Puskarich MA, Trzeciak S, Shapiro NI, Arnold RC, Horton JM, Studnek JR,
None. et al. Association between timing of antibiotic administration and mortality
from septic shock in patients treated with a quantitative resuscitation protocol.
Crit Care Med 2011;39(9):2066–71.
Acknowledgments 19. Bloos F, Thomas-Rüddel D, Rüddel H, Engel C, Schwarzkopf D, Marshall JC,
et al. Impact of compliance with infection management guidelines on outcome
We thank Cecile Jaques, Jolanda Elmers and Isabelle de Kaenel in patients with severe sepsis: a prospective observational multi-center study.
Crit Care 2014;18(2):R42 -R.
for their invaluable help in the literature search. 20. de Groot B, Ansems A, Gerling DH, Rijpsma D, van Amstel P, Linzel D, et al. The
association between time to antibiotics and relevant clinical outcomes in emer-
gency department patients with various stages of sepsis: a prospective multi–
Author’s contributions center study. Crit Care 2015;19:194.
21. Wisdom A, Eaton V, Gordon D, Daniel S, Woodman R, Phillips C. INITIAT-E.D.:
Conception and design of the study: SAA and TC. Analyses of impact of timing of INITIation of antibiotic therapy on mortality of patients
presenting to an emergency department with sepsis. Emerg Med Australas
the data: SAA, FD, ITS and TC. Contribution of materials/analysis 2015;27(3):196–201.
tools: SAA, FD, ITS and TC. Writing of the manuscript: SAA, FD, 22. Bone RC, Balk RA, Cerra FB, Dellinger RP, Fein AM, Knaus WA, et al. Definitions
ITS and TC. All authors read and approved the final version of the for sepsis and organ failure and guidelines for the use of innovative therapies
in sepsis. The ACCP/SCCM consensus conference committee. American college of
manuscript.
chest physicians/society of critical care medicine. Chest 1992;101(6):1644–55.
23. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, et al. 2001
SCCM/ESICM/ACCP/ATS/SIS international sepsis definitions conference. Crit Care
Funding statement Med 2003;31(4):1250–6.
24. Howell MD, Talmor D, Schuetz P, Hunziker S, Jones AE, Shapiro NI. Proof of
IS was supported by the European Sepsis Academy Horizon principle: the predisposition, infection, response, organ failure sepsis staging
system. Crit Care Med 2011;39(2):322–7.
2020 Marie Skłodowska-Curie Action: Innovative Training Network
25. Calandra T, Cohen J. International sepsis forum definition of infection in the
(MSCA-ESA-ITN, grant number 676129) and received a scholarship ICUCC. The international sepsis forum consensus conference on definitions of
from the Société Académique Vaudoise (Lausanne, Switzerland). infection in the intensive care unit. Crit Care Med 2005;33(7):1538–48.
26. Alam N, Oskam E, Stassen PM, Exter PV, van de Ven PM, Haak HR, et al. Pre-
hospital antibiotics in the ambulance for sepsis: a multicentre, open label, ran-
Supplementary materials domised trial. Lancet Respir Med 2018;6(1):40–50.
27. Ballester L, Martinez R, Mendez J, Miro G, Solsona M, Palomera E, et al. Differ-
ences in hypotensive vs. non-hypotensive sepsis management in the emergency
Supplementary material associated with this article can be department: door-to-antibiotic time impact on sepsis survival. Med Sci (Basel)
found, in the online version, at doi:10.1016/j.jinf.2021.03.003. 2018;6(4):91. doi:10.3390/medsci6.

133
S.A. Asner, F. Desgranges, I.T. Schrijver et al. Journal of Infection 82 (2021) 125–134

28. Tan IRP, Mendoza MT. Early antibiotic therapy (EAT) decreases in-hospital mor- 42. Bloos F, Rüddel H, Thomas-Rüddel D, Schwarzkopf D, Pausch C, Harbarth S,
tality of patients with sepsis at the emergency department. Phillippine J Intern et al. Effect of a multifaceted educational intervention for anti-infectious
Med 2019;57(2):87–92. measures on sepsis mortality: a cluster randomized trial. Intensiv Care Med
29. Jalili M, Barzegari H, Pourtabatabaei N, Honarmand AR, Boreiri M, Mehrvarz A, 2017;43(11):1602–12.
et al. Effect of door-to-antibiotic time on mortality of patients with sep- 43. Husabo G, Nilsen RM, Flaatten H, Solligard E, Frich JC, Bondevik GT, et al. Early
sis in emergency department: a prospective cohort study. Acta Med Iran diagnosis of sepsis in emergency departments, time to treatment, and associa-
2013;51(7):454–60. tion with mortality: an observational study. PLOS ONE 2020;15(1):0227652.
30. Zhang D, Micek ST, Kollef MH. Time to appropriate antibiotic therapy is an in- 44. Lueangarun S, Leelarasamee A. Impact of inappropriate empiric antimicrobial
dependent determinant of postinfection ICU and hospital lengths of stay in pa- therapy on mortality of septic patients with bacteremia: a retrospective study.
tients with sepsis. Crit Care Med 2015;43(10):2133–40. Interdiscip Perspect Infect Dis 2012;2012:765205.
31. Peng MY, Wu YK, Yang MC, Lan CC, Yan HC, Shui HA, et al. Predictors of time to 45. Castano P, Plaza M, Molina F, Hincapie C, Maya W, Catano J, et al. Antimicrobial
appropriate antibiotics for septic patients in intensive care units. Int J Clin Exp agent prescription: a prospective cohort study in patients with sepsis and septic
Med 2018;11(8):8556–65. shock. Trop Med Int Health 2019;24(2):175–84.
32. Whiles BB, Deis AS, Simpson SQ. Increased Time to initial antimicrobial admin- 46. Ascuntar J, Mendoza D, Jaimes F. Antimicrobials administration time in patients
istration is associated with progression to septic shock in severe sepsis patients. with suspected sepsis: is faster better? An analysis by propensity score. J Intens
Crit Care Med 2017;45(4):623–9. Care 2020;8:28.
33. Ferrer R, Artigas A, Suarez D, Palencia E, Levy MM, Arenzana A, et al. Effective- 47. Seok H, Song J, Jeon JH, Choi HK, Choi WS, Moon S, et al. Timing of antibiotics
ness of treatments for severe sepsis: a prospective, multicenter, observational in septic patients: a prospective cohort study. Clin Microbiol Infect 2020.
study. Am J Respir Crit Care Med 2009;180(9):861–6. 48. Abe T, Kushimoto S, Tokuda Y, Phillips GS, Rhodes A, Sugiyama T, et al. Im-
34. Joo YM, Chae MK, Hwang SY, Jin SC, Lee TR, Cha WC, et al. Impact of timely plementation of earlier antibiotic administration in patients with severe sepsis
antibiotic administration on outcomes in patients with severe sepsis and septic and septic shock in Japan: a descriptive analysis of a prospective observational
shock in the emergency department. Clin Exp Emerg Med 2014;1(1):35–40. study. Crit Care 2019;23(1):360.
35. Kim RY, Ng AM, Persaud AK, Furmanek SP, Kothari YN, Price JD, et al. Antibiotic 49. Yokota PK, Marra AR, Martino MD, Victor ES, Durao MS, Edmond MB, et al. Im-
timing and outcomes in sepsis. Am J Med Sci 2018;355(6):524–9. pact of appropriate antimicrobial therapy for patients with severe sepsis and
36. Londono J, Nino C, Archila A, Valencia M, Cardenas D, Perdomo M, et al. An- septic shock–a quality improvement study. PLOS ONE 2014;9(11):104475.
tibiotics has more impact on mortality than other early goal-directed therapy 50. Ko BS, Choi SH, Kang GH, Shin TG, Kim K, Jo YH, et al. Time to antibiotics and
components in patients with sepsis: an instrumental variable analysis. J Crit the outcome of patients with septic shock: a propensity score analysis. Am J
Care 2018;48:191–7. Med 2020;133(4):485–91 e4.
37. Nygard ST, Langeland N, Flaatten HK, Fanebust R, Haugen O, Aetiology SS. An- 51. Sterling SA, Miller WR, Pryor J, Puskarich MA, Jones AE. The impact of timing of
timicrobial therapy and outcome of patients with community acquired severe antibiotics on outcomes in severe sepsis and septic shock: a systematic review
sepsis: a prospective study in a Norwegian university hospital. BMC Infect Dis and meta-analysis. Crit Care Med 2015;43(9):1907–15.
2014;14:121. 52. Rothrock SG, Cassidy DD, Barneck M, Schinkel M, Guetschow B, Myburgh C,
38. Peltan ID, Brown SM, Bledsoe JR, Sorensen J, Samore MH, Allen TL, et al. Outcome of immediate versus early antibiotics in severe sepsis
et al. ED door-to-antibiotic time and long-term mortality in sepsis. Chest and septic shock: a systematic review and meta-analysis. Ann Emerg Med
2019;155(5):938–46. 2020;76(4):427–41.
39. Seymour CW, Kahn JM, Martin-Gill C, Callaway CW, Yealy DM, Scales D, 53. Johnston ANB, Park J, Doi SA, Sharman V, Clark J, Robinson J, et al. Effect of
et al. Delays from first medical contact to antibiotic administration for sepsis. immediate administration of antibiotics in patients with sepsis in tertiary care:
Crit Care Med 2017;45(5):759–65. a systematic review and meta-analysis. Clin Ther 2017;39(1):190–202 e6.
40. Suberviola Canas B, Jauregui R, Ballesteros MA, Leizaola O, Gonzalez-Castro A, 54. Angus DC, Seymour CW, Coopersmith CM, Deutschman CS, Klompas M,
Castellanos-Ortega A. Effects of antibiotic administration delay and inadequacy Levy MM, et al. A framework for the development and interpretation of dif-
upon the survival of septic shock patients. Med Intensiv 2015;39(8):459–66. ferent sepsis definitions and clinical criteria. Crit Care Med 2016;44(3):113–21.
41. Ryoo SM, Kim WY, Sohn CH, Seo DW, Koh JW, Oh BJ, et al. Prognostic value of 55. Naucler P, Huttner A, van Werkhoven CH, Singer M, Tattevin P, Einav S,
timing of antibiotic administration in patients with septic shock treated with et al. Impact of time to antibiotic therapy on clinical outcome in patients with
early quantitative resuscitation. Am J Med Sci 2015;349(4):328–33. bacterial infections in the emergency department: implications for antimicro-
bial stewardship. Clin Microbiol Infect 2021;27(2):175–81.

134