Journal Pre-Proof: Clinical Microbiology and Infection
Journal Pre-Proof: Clinical Microbiology and Infection
Journal Pre-Proof: Clinical Microbiology and Infection
Seven versus 14-days course of antibiotics for the treatment of bloodstream infections
by Enterobacterales. A randomized, controlled trial
PII: S1198-743X(21)00491-2
DOI: https://doi.org/10.1016/j.cmi.2021.09.001
Reference: CMI 2678
Please cite this article as: Molina J, Montero-Mateos E, Praena-Segovia J, León-Jiménez E, Natera
C, López-Cortés LE, Valiente L, Rosso-Fernández CM, Herrero M, Aller-García AI, Cano Á, Gutiérrez-
Gutiérrez B, Márquez-Gómez I, Álvarez-Marín R, Infante C, Roca C, Valiente-Méndez A, Pachón J,
Reguera JM, Corzo-Delgado JE, Torre-Cisneros J, Rodríguez-Baño J, Cisneros JM, on behalf of the
SHORTEN trial team, Seven versus 14-days course of antibiotics for the treatment of bloodstream
infections by Enterobacterales. A randomized, controlled trial, Clinical Microbiology and Infection, https://
doi.org/10.1016/j.cmi.2021.09.001.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
© 2021 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All
rights reserved.
1 Tittle page
3 Title: Seven versus 14-days course of antibiotics for the treatment of bloodstream infections
7 M.D.1,2, Eva León-Jiménez, M.D., Ph.D.4, Clara Natera, M.D., Ph.D.5, Luis E López-Cortés,
8 M.D., Ph.D.2,6, Lucía Valiente, M.D.7, Clara M. Rosso-Fernández, M.D., Ph.D.2,8, Marta
of
Herrero, M.D.1,2, Ana I Aller-García, M.D., Ph.D.4, Ángela Cano, M.D.5, Belén Gutiérrez-
ro
11 Ph.D.1,2, Carmen Infante, M.D., Ph.D.1,2, Cristina Roca, M.D.1,2, Adoración Valiente-Méndez,
12
-p
M.D.2,6, Jerónimo Pachón, M.D., Ph.D.2,9, José María Reguera, M.D., Ph.D.7, Juan Enrique
re
13 Corzo-Delgado, M.D., Ph.D.4, Julián Torre-Cisneros, M.D., Ph.D.5,10, Jesús Rodríguez-Baño,
lP
14 M.D., Ph.D.2,6,9 and José Miguel Cisneros, M.D., Ph.D.1,2,9 * on behalf of the SHORTEN trial
15 team11.
na
16 1
Unit of Infectious Diseases, Microbiology and Preventive Medicine, Virgen del Rocío
ur
18 2
Institute of Biomedicine of Seville (IBiS), Virgen del Rocío and Virgen Macarena University
20 3
Department of Pathology and Institute of Biomedical Research of Salamanca (IBSAL),
22 4
Unit of Infectious Diseases and Microbiology, Virgen de Valme University Hospital, Seville,
23 Spain.
24 5
Maimonides Institute for Research in Biomedicine of Córdoba (IMIBIC). Service of
26 6
Clinical Unit of Infectious Diseases and Microbiology, Virgen Macarena University
30 8
Spanish Clinical Research Network, Virgen del Rocío University Hospital, Seville, Spain.
31 9
Department of Medicine, University of Seville, Seville, Spain.
32 10
Departement of Medical and Surgical Sciences. University of Cordoba (UCO), Córdoba,
33 Spain.
34 11
Members of the SHORTEN trial team who have made substantial contributions are listed in
36
of
37
ro
* Corresponding author:
39
-p
Department of Infectious Diseases, Microbiology and Preventive Medicine, University
re
40 Hospital Virgen del Rocio, Av. Manuel Siurot s/n, 41013, Seville, Spain.
lP
41 Telephone/fax: +34-955-01-21-85
42
na
e-mail: jmcisnerosh@gmail.com
43
ur
45 antimicrobial stewardship.
46 Abstract
47
48 Objective: To prove that 7-day courses of antibiotics for bloodstream infections caused by
51 Methods: A randomized trial was performed. Adult patients developing eBSI with appropriate
52 source control were assigned to 7 or 14 days of treatment, and followed 28 days after treatment
53 cessation; treatments could be resumed whenever necessary. Primary end-point was days of
54
of
treatment at the end of follow-up. Clinical outcomes included clinical cure, relapse of eBSI and
55
ro
relapse of fever. A superiority margin of 3-days was set for the primary end-point, and a non-
56 inferiority margin of 10% for clinical outcomes. Efficacy and safety were assessed together
57
-p
with a Desirability of Outcome Ranking and Response Adjusted for Duration of Antibiotic Risk
re
58 (DOOR/RADAR) analysis.
lP
59 Results: 248 patients were assigned to 7 (n=119) or 14 days (n=129) of treatment. In the
60
na
intention-to-treat analysis, median days of treatment at the end of follow up was 7 and 14 days
61 (difference 7 [95% CI 7-7]). The non-inferiority margin was also met for clinical outcomes,
ur
62 except for relapse of fever (-0.2% [CI95% -10.4 to 10.1]). The DOOR/RADAR showed that 7-
Jo
63 days schemes had a 77.7% probability of achieving better results compared to 14-days
64 treatments.
65 Conclusions: 7-day schemes allowed reducing antibiotic exposure of patients with eBSI
66 achieving similar outcomes. The possibility of relapsing fever in a limited number of patients -
67 without relevance on final outcomes- may not be excluded, but was overcome by the benefits
68 of shortening treatments.
69
70 Introduction
71
72 The duration of antimicrobial treatment for bloodstream infections caused by Enterobacterales
73 (eBSI) has traditionally been supported by experts’ opinions. Different scientific societies
74 propose to treat catheter-related eBSI for a variable duration of 7 to 14 days [1,2], and no
75 recommendations exist for other sources. The current scenario of rapidly spreading bacterial
76 resistance at a global level mandates initiatives able to stop this threat [3], and shortening the
77 duration of antibiotic treatments is probably one of the most effective measures to avoid the
of
79 shorter courses of antibiotics should be priority, especially for common clinical situations.
ro
80 Hitherto, two non-inferiority trials have been published showing similar outcomes in patients
81 -p
receiving 7 versus 14 days of antibiotic treatment for Gram-negative bacilli bacteremia [5,6].
re
82 However, the question remains open for specific subgroups of patients, for whom short courses
83
lP
may not be equally effective, like immunocompromised patients or male urinary tract infections
84 [7,8]. Thus, it would be still useful additional evidence, which allows balancing more accurately
na
85 the theoretical benefits of shortening antibiotic treatments (i.e. reduced risk of adverse reactions
86
ur
87 The aim of the present trial was to prove that a 7-day course of antibiotics will allow to reduce
88 antibiotic exposure of patients, while achieving similar clinical outcomes than traditional 14-
90
93 Participants. Adults over 18 years with a diagnosis of eBSI were recruited. Hospitalized and
94 outpatients were eligible. Exclusion criteria were: a) pregnancy; b) eBSI with a non-controlled
95 source and no expectation of being controlled in the following 24 hours; c) patients undergoing
96 chemotherapy with neutropenia < 500 cells/mm3 expected for more than 7 days; d) eBSI
97 secondary to infections requiring prolonged antibiotic treatment (e.g. osteomyelitis, meningitis,
98 prostatitis, etc.); e) concomitant infection requiring antibiotic treatment at the time of the
100 polymicrobial bacteremia; and h) expectation of survival lower than 48 hours. In the initial
101 study protocol, patients diagnosed after randomization of a previously unnoticed exclusion
102 criteria were excluded from the analysis; to avoid potential biases, these patients were included
104 Setting. Five Spanish hospitals participated in the trial between September 2014 and September
of
105 2016.
ro
106 Intervention and follow-up. Patients were randomized to receive either 7 days or 14 days of
107 -p
any fully-active antibiotic treatment -oral or parenteral- against the microorganism isolated, and
re
108 according to local guidelines. Follow-up blood cultures were obtained after 48 to 72 hours of
109
lP
treatment. To stop the antibiotic treatment, patients were required have a negative follow-up
110 blood culture, and to have remained apyretic and without symptoms of infection for at least 72
na
111 hours. If these requirements were not met on the day of the scheduled stop, the treatment was
112
ur
continued and the patient re-evaluated each 48 to 72 hours until all criteria were fulfilled.
113 Patients were followed until 28 days after stopping the antibiotic treatment: for all patients, in-
Jo
114 person visits were performed on days +7 and +14 after the initiation of the treatment, and
115 additional telephonic interviews were performed on day +14 and +28 after the end of antibiotic
116 treatment (figure 1). Treatment could be resumed or prolonged whenever considered necessary
117 by the physician in charge. Decisions on the antibiotic agent, oral step-down, hospital
118 discharge, and the management of eventual complications were also decided by the physician
120 Microbiological assays were performed following the usual routine of Microbiology
121 laboratories from the respective participating centers (supplementary file 1).
122 Outcomes. The primary end-point was the total number of days of antibiotic treatment
123 prescribed to the patient for any reason, from the day of the first positive blood samples
124 collection until the end of the follow-up. Clinical outcome was assessed through relapse of the
125 eBSI, relapse of fever and clinical cure (defined as resolution of all signs and symptoms of
126 infection) at the end of follow-up. Additional secondary end-points included crude mortality,
127 superinfections (defined as infections different to the initial episode occurring during the
128 follow-up), and adverse events at the end of the follow-up. A superiority design was defined
129 for the primary end-point, and a non-inferiority design for clinical outcomes.
130 Survival was recorded for all randomized patients due to safety reasons, including those lost to
131 follow-up; in these cases, this information was obtained from regional healthcare system
of
132 databases.
ro
133 A full list of definitions for the main clinical variables can be found in supplementary file 1.
134
-p
Sample size. Sample size was calculated for the primary end-point and for clinical outcome
re
135 end-points, according to the only meta-analysis available at the moment of designing the trial
lP
136 including bacteraemic infections from different sources (pneumonia, pyelonephritis, peritonitis,
137 etc.) [9]. This study showed a rate of clinical failure of 13.5% for short treatments vs. 4.1% for
na
138 prolonged treatments, without significant heterogeneity among all syndromes analysed. The
ur
139 definition of clinical failure varied among studies in the meta-analysis and included survival,
140
Jo
relapses and resolution of symptoms. Assuming a mean of 14 days (SD 7.53) of antibiotic use
141 for the prolonged-treatment arm and 10% of patients lost to follow-up, 40 patients would
142 provide a 80% power at two-sided α=0.05, to detect a difference of at least 3 days of treatment
143 between both arms. To test the non-inferiority for variables of clinical outcome, 119 patients in
144 each group would be necessary with a 10% non-inferiority margin, 1-sided α=0.025, and the
145 same power and losses, increasing target sample size to 238 patients. A pre-scheduled interim
146 analysis was set when half the sample was recruited.
147 Randomization. Simple randomization was performed in a 1:1 ratio, stratified by sites through
148 a predesigned randomization list. Randomization was performed up to 72 hours after the
149 identification of Enterobacterales in the blood samples (typically, up to 3-4 days after blood
150 cultures were taken). The process was centralized in the coordinating center, and performed
151 online through an automatic system integrated in the electronic case report form (eCRF). The
152 randomization list was computer-generated (Epidat 4.0 software). Only after the eCRF was
153 fulfilled with inclusion and exclusion criteria, the system provided the group allocation. The
154 Information Technology department responsible for the eCRF and the clinical trials unit were
157 Statistical methods. We tested the superiority of the short regimen by calculating the difference
of
158 between group medians (95% CI) using the Hodges-Lehmann estimator. Regarding non-
ro
159 inferiority endpoints, one-sided 97.5% CI for the difference between treatments in the
160
-p
proportion of patients were computed with the Newcombe-Wilson Score method. Outcomes
re
161 were compared between groups with χ2 test, Fisher exact test or Mann-Whitney U as
lP
162 appropriate.
163
na
164 Duration of Antibiotic Risk (DOOR/RADAR) analysis was performed post hoc, an innovative
ur
165 methodology published after the design of this trial [10]. In the case of our trial, this analysis
Jo
166 seemed useful to balance the expected benefits of shortening antibiotic treatments (i.e. reducing
167 adverse effects due to prolonged antibiotic exposure) with the potential risk of impaired
168 effectiveness. To do so, we designed an ordinal scale with five outcome categories: (1) cure
169 without incidences, (2) cure with relapsing fever, (3) cure with a severe adverse event, (4) not
170 cured, (5) death. The comparison between arms is established in terms of the probability of
171 having a better DOOR score for the experimental group compared with the controls, so that if
172 the short-treatment strategy was better than the prolonged treatments, this probability would be
174 Analysis of missing data was performed through multiple imputations (five imputed datasets)
177 Research and the Spanish Agency of Medicines and Medical Devices (EudraCT: 2013-002148-
178 95.), and was conducted following the principles of the Helsinki Declaration and national
179 regulations (RD 223/2004). All patients signed an informed consent previously to their
180 recruitment. The trial methodology was registered before its initiation in clinicaltrials.gov
181 (NCT02400268). All items from the WHO Trial Registration Data Set are included in the
182 registry.
183
of
184 Results
ro
185 Baseline characteristics. Among 248 randomized patients, 231 (93.1%) were assessed for the
186
-p
primary and secondary outcomes, with 17 patients lost to follow-up (9 in short-treatment arm
re
187 and 8 in prolonged-treatment arm) (figure 2, supplementary files 3-4). Baseline characteristics
lP
188 were in general well balanced between groups, except for respiratory source and chronic kidney
189 disease, which were more frequent in the control group (table 1).
na
190 Outcomes. The median length of antibiotic treatment at the end of the follow-up was 7 (7-14)
ur
191 in the experimental group and 14 (14-16) in control arm (difference 7 [95% CI 7-7]) in the ITT
Jo
192 population. No significant differences were observed regarding the other end-points at the end
193 of follow-up, including mortality, relapse of eBSI, relapse of fever, superinfections, or drug-
194 related adverse events. The non-inferiority margin was met for all clinical outcomes, except for
195 relapse of fever, which was more frequent in the experimental group (difference in absolute
196 risk -0.2% (CI97.5% -∞ to 10.1) (table 2, figure 3). No significant differences were observed
197 between groups for the causes of relapsing fever (supplementary file 6). The multiple
198 imputation analysis of missing data produced similar results (table 2).
199 The DOOR/RADAR analysis showed that patients receiving 7-day courses had 77.7% more
200 probabilities to achieve better results compared to those receiving 14 days, considering
201 altogether clinical cure, adverse events, mortality and antibiotic exposure (table 3,
203 Safety. No statistically significant differences were detected in different safety variables,
204 including severe adverse events, or drug-related reactions (table 2, supplementary file 7).
205
206 Discussion
207 The results of this trial suggest that 7-day courses of antibiotics may be the preferential strategy
208
of
for treating bacteraemic infections produced by Enterobacteriaceae, whenever an adequate
ro
210 -p
In order to ensure the safety of the intervention, a number of secondary end-points were settled.
211 Compared to 14-day treatments, non-inferiority was showed for clinical cure and relapse of
re
212 eBSI. The predefined non-inferiority margin was barely unmet for the relapse of fever. It should
lP
213 be noted that not proving non-inferiority is not the same as proving inferiority. However, real-
na
214 life practice tells us that there may be individual patients for whom 7 days of treatment might
215 be insufficient, whilst no differences in final outcomes were proved in our trial. The need to
ur
216 retreat a limited number of cases should be balanced with the effects of systematically give
Jo
217 prolonged antibiotic treatments to all patients, since the ecological costs of doing so may be
218 unaffordable in the current era of antibiotic crisis [12,13]. The risk of superinfections or other
219 drug-related adverse events should also be balanced when choosing the duration of treatments.
220 Although the trial was not designed to assess this aspect, a trend towards an increased risk of
221 treatment-related adverse events was observed among patients with 14-days treatments, as
223 The aforementioned reasons justified the addition of the DOOR/RADAR analysis. This novel
224 analysis is helpful for randomized trials to define the optimal therapeutic strategy, since
225 considering exclusively the primary end-point may not allow researchers to accurately balance
226 a proven benefit (i.e. reducing treatment duration) with other potential harms (i.e. impaired
227 effectiveness or side effects) [10,16]. In this case, this analysis pointed towards the 7-day
228 treatment as the strategy of choice, showing that patients receiving short treatments had 77.7%
229 probabilities of achieving better results, considering altogether clinical cure, adverse effects and
231 Previous to ours, there are two randomised trials addressing the optimal duration of the
232 treatment of eBSI [5,6]. Consistently with our results, the former non-inferiority trials did not
233 find differences in outcomes of patients treated during 7 days compared to those receiving 14
234 days of treatment. Our trial adds an insight into the magnitude of the beneficial effect of
of
235 shortening antibiotic treatment in this scenario through the DOOR/RADAR analysis, which
ro
236 could encourage the adoption of this strategy in routine clinical practice.
237 -p
Finally, our sample included a considerable rate of immunosuppressed patients (over 10%),
re
238 cephalosporin-resistant eBSI (over 15%), or infections with a severe clinical presentation (over
239
lP
13%), reinforcing the reproducibility of its results in real-life condition. The new data provided
240 by this trial added to those previously published [5,6] may enable proper meta-analyses which
na
241 could confirm their hypothesis also for these subsets of patients.
ur
242 A number of limitations should be addressed. The intervention of the trial was closely related
243
Jo
to the primary endpoint, and was determinant to explain the differences in antibiotic exposure
244 between groups. Setting this endpoint responded to the aim of the trial -to reduce unnecessary
245 antibiotic duration-, and was consistent with the methodology of the few preceding trials with
246 similar purposes [12]. It must be noticed that the end-point included any antibiotic treatment
247 received from randomization and until the end of the follow-up, and thus it depended on the
248 clinical course of the infection. The need of frequent retreatments would have attenuated the
249 differences between the trial arms if the 7-day strategy would have been ineffective. On the
250 contrary, a median reduction of treatments duration of 50% was achieved in the experimental
252 The power of the sample is limited to prove the non-inferiority for less-frequent clinical
253 outcomes, like mortality, which was low, probably due to the syndrome tackled by the trial.
254 The prevalence of events assessing recurrent illness (relapses, relapsing fever, or absence of
255 clinical cure) were comparable to the clinical failure rate reported in the meta-analysis of Havey
256 et al. [9], and thus, we believe our sample is properly empowered to evaluate this key point
258 Our study included several differences in the follow-up compared to previous trials. First,
259 randomization was carried out early after the diagnosis of the eBSI, in order to avoid a potential
260 risk of bias through the selection of patients with the best responses to treatment. This may have
261 enabled the recruitment of patients with initially unnoticed uncontrolled sources; considering
262
of
this, we believe that the favorable outcomes achieved even in the ITT analyses, provides
263
ro
robustness to the conclusions. Second, follow-up was set 28 days starting from treatment
264 cessation. Since relapses are unlikely to occur during antibiotic treatment, follow-up was set
265
-p
equally for both groups after its discontinuation. To avoid any interference because of the
re
266 differences in the follow-up, adverse effects were assessed in absolute terms but also pondered
lP
268
na
In conclusion, this trial points to 7-day course of antibiotics as the preferential treatment for
269 eBSI, as long as the source is properly controlled. The potential impact of implementing this
ur
270 recommendation into clinical practice would be significant in the fight against bacterial
Jo
271 resistance. A possible need of retreating a limited number of patients after short courses without
272 clinical impact on the final outcomes cannot be discarded by this trial, but seemed to be
274
275 Acknowledgements.
276 The SHORTEN trial was a non-commercial, investigator-driven clinical trial supported by a
277 grant from the Regional Ministry of Health of Andalusia (PI-0161/2013) and by Plan Nacional
278 de I+D+i and Instituto de Salud Carlos III, Subdirección General de Redes y Centros de
282 Europe" ERDF and by Plan Nacional de I+D+i and Instituto de Salud Carlos III, Subdirección
284 Competitividad, Spanish Clinical Research Network (SCReN, PT13/0002/0010). The REIPI
285 also collaborated in the provision of human resources required for the trial tasks, and also
286 provided the platform for the electronic case report form. In particular, we acknowledge the
287 endless availability of Alejandro González-Herrero, responsible for the technical support of the
of
289 Collaborators.
ro
290 Also participiated in the SHORTEN trial team: Blanca Solano, Verónica González-Galán;
291
-p
Esteban Hinojosa; Francisco López-Bernal; Marta Suñer; José Ángel Noval; Álvaro Giráldez;
re
292 Antonio Navarro; María Jesús Rodríguez-Hernández; Yolanda Borrego; Paloma Gil; José
lP
293 Antonio Lepe; Isabel Morales; Pilar Retamar, Marina de Cueto, Juan José Castón and Elisa
294 Vidal.
na
296 J.M.C. and J.M. conceived and designed the study. C.R.F and B.S. supervised and coordinated
Jo
297 the accomplishment of legal procedures required for the trial as well as its monitoring, J.M.,
298 J.P.S., M.H., E.L.J, C.N., E.L, R.A.M, L, A.I.A.G, A.C., B.G.G, J.E.C., I.M.G. and A.V.M.
299 recruited and performed the clinical follow-up of patients. C.R. and C.I. recorded data in the
300 trial database. E.M. performed the statistical analysis. J.M. J.M.C., J.P., J.M.R., J.E.C., J. T.
301 and J.R.B. collaborated in the achievement of the public funds and provided the team with the
302 human resources required for the development of the study. All authors contributed in the
303 discussion of the results and approved the final version of this manuscript.
304 Funding. The trial obtained competitive public funds from the Regional Ministry of Health of
305 Andalusia (PI-0161/2013), and received technical support from the Spanish Network of
306 Research in Infectious Diseases (REIPI) and the Spanish Clinical Research Network (SCReN)
307 of the Instituto de Salud Carlos III, Ministerio de Economía y Competitividad, Spanish
308 Government (PT13/0002/0010). These organizations did not participate in the analysis of the
309 study.
310 Conflict of interests. LELC declared consulting fees from Novartis and MSD, honoraria for
311 lectures from Correvio, ViiV and Gilead, and participation on advisory board for Angelini. AC
312 declared honoraria for lectures from Pfizer and Shionogu. The rest of authors declared no
of
ro
-p
re
lP
na
ur
Jo
314 Bibliography.
315 [1] Capdevila JA, Guembe M, Barberán J, de Alarcón A, Bouza E, Fariñas MC, et al.
316 2016 Expert consensus document on prevention, diagnosis and treatment of short-term
317 peripheral venous catheter-related infections in adults. Cir Cardiovasc 2016;23:192–8.
318 https://doi.org/10.1016/j.circv.2016.06.001.
319 [2] Mermel LA, Allon M, Bouza E, Craven DE, Flynn P, O’Grady NP, et al. Clinical
320 practice guidelines for the diagnosis and management of intravascular catheter-related
321 infection: 2009 Update by the Infectious Diseases Society of America. Clin Infect Dis
322 2009;49:1–45. https://doi.org/10.1086/599376.
323 [3] European Centre for Disease Prevention and Control (ECDC). Antimicrobial
324 resistance strategies and action plans 2016. http://ecdc.europa.eu/en/healthtopics/Healthcare-
325 associated_infections/guidance-infection-prevention-control/Pages/antimicrobial-resistance-
326 strategies-action-plans.aspx.
327 [4] Drusano GL, Louie A, MacGowan A, Hope W. Suppression of Emergence of
328 Resistance in Pathogenic Bacteria: Keeping Our Powder Dry, Part 1. Antimicrob Agents
329
of
Chemother 2015;60:1183–93. https://doi.org/10.1128/AAC.02177-15.
330 [5] Yahav D, Franceschini E, Koppel F, Turjeman A, Babich T, Bitterman R, et al. Seven
331
ro
versus fourteen Days of Antibiotic Therapy for uncomplicated Gram-negative Bacteremia: a
332 Non-inferiority Randomized Controlled Trial. Clin Infect Dis Off Publ Infect Dis Soc Am
333 2018. https://doi.org/10.1093/cid/ciy1054.
334
335
[6] -p
von Dach E, Albrich WC, Brunel A-S, Prendki V, Cuvelier C, Flury D, et al. Effect
of C-Reactive Protein-Guided Antibiotic Treatment Duration, 7-Day Treatment, or 14-Day
re
336 Treatment on 30-Day Clinical Failure Rate in Patients With Uncomplicated Gram-Negative
337 Bacteremia: A Randomized Clinical Trial. JAMA 2020;323:2160–9.
338
lP
https://doi.org/10.1001/jama.2020.6348.
339 [7] Yahav D, Mussini C, Leibovici L, Paul M. Reply to De Greef et al. Clin Infect Dis
340 2020;70:351–3. https://doi.org/10.1093/cid/ciz393.
341
na
https://doi.org/10.1016/j.cmi.2018.01.013.
345 [9] Havey TC, Fowler RA, Daneman N. Duration of antibiotic therapy for bacteremia: a
346 systematic review and meta-analysis. Crit Care 2011;15:R267.
Jo
347 https://doi.org/10.1186/cc10545.
348 [10] Evans SR, Rubin D, Follmann D, Pennello G, Huskins WC, Powers JH, et al.
349 Desirability of Outcome Ranking (DOOR) and Response Adjusted for Duration of Antibiotic
350 Risk (RADAR). Clin Infect Dis Off Publ Infect Dis Soc Am 2015;61:800–6.
351 https://doi.org/10.1093/cid/civ495.
352 [11] Anonymous. Missing data in confirmatory clinical trials. Eur Med Agency 2018.
353 https://www.ema.europa.eu/en/missing-data-confirmatory-clinical-trials (accessed May 28,
354 2019).
355 [12] Chastre J, Wolff M, Fagon J-Y, Chevret S, Thomas F, Wermert D, et al. Comparison
356 of 8 vs 15 days of antibiotic therapy for ventilator-associated pneumonia in adults: a
357 randomized trial. JAMA 2003;290:2588–98. https://doi.org/10.1001/jama.290.19.2588.
358 [13] Chotiprasitsakul D, Han JH, Cosgrove SE, Harris AD, Lautenbach E, Conley AT, et
359 al. Comparing the Outcomes of Adults With Enterobacteriaceae Bacteremia Receiving Short-
360 Course Versus Prolonged-Course Antibiotic Therapy in a Multicenter, Propensity Score-
361 Matched Cohort. Clin Infect Dis Off Publ Infect Dis Soc Am 2018;66:172–7.
362 https://doi.org/10.1093/cid/cix767.
363 [14] Branch-Elliman W, O’Brien W, Strymish J, Itani K, Wyatt C, Gupta K. Association
364 of Duration and Type of Surgical Prophylaxis With Antimicrobial-Associated Adverse
365 Events. JAMA Surg 2019. https://doi.org/10.1001/jamasurg.2019.0569.
366 [15] Aguilar-Guisado M, Espigado I, Martín-Peña A, Gudiol C, Royo-Cebrecos C,
367 Falantes J, et al. Optimisation of empirical antimicrobial therapy in patients with
368 haematological malignancies and febrile neutropenia (How Long study): an open-label,
369 randomised, controlled phase 4 trial. Lancet Haematol 2017;4:e573–83.
370 https://doi.org/10.1016/S2352-3026(17)30211-9.
371 [16] Molina J, Cisneros JM. A Chance to Change the Paradigm of Outcome Assessment of
372 Antimicrobial Stewardship Programs. Clin Infect Dis Off Publ Infect Dis Soc Am 2015.
373 https://doi.org/10.1093/cid/civ496.
374
of
ro
-p
re
lP
na
ur
Jo
375
Experimental Control
(n=119) (n=129)
of
HUVM 13/119 (10.9%) 12/129 (9.3%)
- HRM 2/119 (1.7%) 7/129 (5.4%)
ro
Patient care
- Outpatient 25/116 (21.6%) 36/125 (28.8%)
- Inpatient
-p
91/116 (78.4%) 90/125 (71.2%)
Charlson index ≥ 3
re
54/119 (45.4%) 56/129 (43.4%)
Comorbidities
lP
of
Other relevant risk factors
ro
- Immunosuppressant drugs 17/118 (14.4%) 14/129 (10.9%)
- Previous ICU stay (30 days) 13/118 (11.0%) 8/129 (6.2%)
-
-
Previous surgery (30 days)
Permanent indwelling urinary
-p
13/118 (11.0%)
15/118 (12.7%)
10/129 (7.8%)
15/129 (11.6%)
re
catheter
- Previous urinary obstruction 7/118 (5.9%) 14/129 (10.9%)
lP
Q1-Q3: Quartile 1 – Quartile 3; BSI: Bloodstream infection; ESBL: Extended spectrum betalactamase; ICU:
379 intensive care unit; HUVR: Virgen del Rocío University Hospital; HUVV: Virgen del Valme University
380 Hospital; HRS: Reina Sofía University Hospital; HUVM; Virgen Macarena University Hospital; HRM:
381 Regional Hospital of Malaga.
ur
382 a The stratification of the Charlson index was set post hoc to identify the standardized definition of patients with
383 high or very high comorbidiy (Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying
384 prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40(5):373-83).
Jo
385 Table 2. Primary and secondary end-points, measured at the end of the follow-up (28 days
386 after antibiotic treatment interruption).
Between-group absolute
7 days 14 days
risk difference
(n=119) (n=129)
(1-sided CI 97.5%)
Days of treatment (median, Q1-Q3)
ITT populationa 7 (7-14) 14 (14-16) 7 (7-7)
(n=110) (n=124)
PP populationb 7 (7-10.5) 14 (14-15) 7 (7-7)
(n=93) (n=108)
MI analysis 8 (7-16.4) 14 (14-17) 7 (6-7)
a
Death
ITT population 3/119 (2.5%) 9/129 (7.0%) -4.5% (-∞ to 1.2)
PP population 1/93 (1.1%) 6/108 (5.6%) -4.5% (-∞ to 1.1)
of
Relapse of the BSI
ro
ITT population 7/108 (6.5%) 6/121 (5.0%) 1.5% (-∞ to 8.4)
PP population 5/93 (5.4%) 5/107 (4.7%) 0.7% (-∞to 7.8)
MI analysis 9/119 (7.6%)
-p
7/129 (5.4%) 2.1% (-∞ to 8.9)
re
Relapse of feverb
ITT population 21/110 (19.1%) 23/119 (19.3%) -0.2% (-∞ to 10.1)
lP
Superinfections
ITT population 16/110 (14.5%) 23/121 (19.0%) -4.5% (-∞ to 5.4)
PP population 11/93 (11.8%) 20/107 (18.7%) -6.9% (-∞ to 3.4)
MI analysis 19/119 (16.0%) 26/129 (20.2%) -4.2% (-∞ to 5.5)
Safety
Adverse eventsc 51/119 (42.9%) 53/129 (41.1%) 1.8% (-∞ to 13.9)
Severe adverse 15/119 (12.6%) 27/129 (20.9%) -8.3% (-∞ to 1.1)
events
Readmissions or 15/119 (12.6%) 27/129 (20.9%) -8.3% (-∞ to 1.1)
prolongation of
hospitalisation
Drug-related adverse 7/119 (5.9%) 12/129 (9.3%) -3.4% (-∞ to 3.5)
reactiond
Acute kidney injury 3/119 (2.5%) 1/129 (0.8%) 1.7% (-∞ to 6.4)
Diarrhea 2/119 (1.7%) 3/129 (2.3%) -0.6% (-∞ to 3.9)
Rash 1/119 (0.8%) 4/129 (3.1%) -2.3% (-∞ to 2.0)
387
388
389
390 Q1-Q3: Interquartile range; PP: per protocol; ITT: intention to treat; MI: Multiple imputation; BSI:
391 Bloodstream infection.
392
393 aSurvival was recorded for all randomized patients. In the case of patients lost for follow-up, this data was
394 obtained by access to healthcare databases.
395
396 b Causes for relapsing fever are detailed in supplementary file 6.
397
398 cAdverse events were defined as any adverse health incidence in a patient or subject of a clinical trial treated
399 with a drug, even if it does not necessarily have a causal relationship with such treatment.
400
401 d Adverse reactions with a definite, probable, or possible relationship with the study drug were considered for
402 the analysis. All safety analyses were performed in the intention to treat cohort.
of
ro
-p
re
lP
na
ur
Jo
403 Table 3. Distribution of patients per Desirability of Outcome Ranking (DOOR) in the per
404 protocol cohort.
405
7 days 14 days
(N=93) (N=108)
n (%) n (%)
of
Probability of a better
DOOR/RADAR score in the 77.7% (CI95% 76.8 to 78.5)
ro
experimental arm*
406
407 -p
*Detailed score calculations are provided in supplementary file 8.
re
lP
na
ur
Jo
Figure 1. Randomization and follow-up process.
28 days End of
Experimental Antibiotic follow-
group treatment stops* up
f
oo
pr
Day +0 Day +1 Randomization
e-
Pr
Blood cultures Enterobacterales
al
obtained identified in blood 28 days End of
rn
Empirical cultures Control Antibiotic follow-
u
antibiotic group treatment stops* up
Jo
treatment starts
Day +0 Day +1 and Day +7 Day +14 Day +35 Day +42
up to day +4
Assessment of total number of days of antibiotic treatment until de end of the follow-up
* Antibiotic treatment was stopped at this point if negativization of blood cultures was confirmed and the patient had remained apyretic and without symptoms of infection for
at least 72 hours. Antibiotic treatment could be resumed after this point whenever considered necessary by the clinician in charge of the patient if an unfavourable course
was observed.
1,496 patients were screened for the study
Criteria of exclusion:
327 patients presented an
uncontrolled source.
302 failed to be randomised within
72h from the diagnosis, or
impossibility for the follow-up.
173 presented BSI due to, or
associated with infections requiring
prolonged antibiotic use.
111 had a expected survival inferior
of
to 48h.
ro
89 patients had polymicrobial BSI
248 patients were randomized
76 patients presented prolonged
-p
neutropenia.
75 patients refused to participate.
of
Relapse of BSI
ro
ITT 1.5% (-4.8 to 8.4)
-p
PP 0.7% (-5.9 to 7.8)
re
Relapse of fever
lP
ITT -0.2% (-10.4 to 10.1)
na
PP 0.4% (-10.3 to 11.3)
Absence of clinical cure
ur
ITT -2.6% (-10.0 to 5.1)
Jo
PP -5.4% (-11.8 to 0.4)