UBC Faculty Research and Publications

7 versus 14 days of antibiotic treatment for critically ill patients with bloodstream infection: a pilot… Daneman, Nick; Rishu, Asgar H; Pinto, Ruxandra; Aslanian, Pierre; Bagshaw, Sean M; Carignan, Alex; Charbonney, Emmanuel; Coburn, Bryan; Cook, Deborah J; Detsky, Michael E; Dodek, Peter; Hall, Richard; Kumar, Anand; Lamontagne, Francois; Lauzier, Francois; Marshall, John C; Martin, Claudio M; McIntyre, Lauralyn; Muscedere, John; Reynolds, Steven; Sligl, Wendy; Stelfox, Henry T; Wilcox, M. E; Fowler, Robert A Feb 17, 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13063_2018_Article_2474.pdf [ 731.14kB ]
JSON: 52383-1.0363938.json
JSON-LD: 52383-1.0363938-ld.json
RDF/XML (Pretty): 52383-1.0363938-rdf.xml
RDF/JSON: 52383-1.0363938-rdf.json
Turtle: 52383-1.0363938-turtle.txt
N-Triples: 52383-1.0363938-rdf-ntriples.txt
Original Record: 52383-1.0363938-source.json
Full Text

Full Text

RESEARCH Open Access7 versus 14 days of antibiotictreatment for critically ill patients withbloodstream infection: a pilot randomizedclinical trialNick Daneman1*, Asgar H. Rishu2, Ruxandra Pinto2, Pierre Aslanian3, Sean M. Bagshaw4, Alex Carignan5,Emmanuel Charbonney6, Bryan Coburn7, Deborah J. Cook8, Michael E. Detsky9, Peter Dodek10, Richard Hall11,Anand Kumar12, Francois Lamontagne13, Francois Lauzier14, John C. Marshall15, Claudio M. Martin16,Lauralyn McIntyre17, John Muscedere18, Steven Reynolds19, Wendy Sligl4, Henry T. Stelfox20, M. Elizabeth Wilcox21,Robert A. Fowler22 and on behalf of the Canadian Critical Care Trials GroupAbstractBackground: Shorter-duration antibiotic treatment is sufficient for a range of bacterial infections, but has not beenadequately studied for bloodstream infections. Our systematic review, survey, and observational study indicatedequipoise for a trial of 7 versus 14 days of antibiotic treatment for bloodstream infections; a pilot randomizedclinical trial (RCT) was a necessary next step to assess feasibility of a larger trial.Methods: We conducted an open, pilot RCT of antibiotic treatment duration among critically ill patients with bloodstreaminfection across 11 intensive care units (ICUs). Antibiotic selection, dosing and route were at the discretion of the treatingteam; patients were randomized 1:1 to intervention arms consisting of two fixed durations of treatment – 7 versus 14 days.We recruited adults with a positive blood culture yielding pathogenic bacteria identified while in ICU. We excluded patientswith severe immunosuppression, foci of infection with an established requirement for prolonged treatment, single cultureswith potential contaminants, or cultures yielding Staphylococcus aureus or fungi. The primary feasibility outcomes wererecruitment rate and adherence to treatment duration protocol. Secondary outcomes included 90-day, ICU and hospitalmortality, relapse of bacteremia, lengths of stay, mechanical ventilation and vasopressor duration, antibiotic-free days,Clostridium difficile, antibiotic adverse events, and secondary infection with antimicrobial-resistant organisms.Results:We successfully achieved our target sample size (n= 115) and average recruitment rate of 1 (interquartile range (IQR)0.3–1.5) patient/ICU/month. Adherence to treatment duration was achieved in 89/115 (77%) patients. Adherence differed byunderlying source of infection: 26/31 (84%) lung; 18/29 (62%) intra-abdominal; 20/26 (77%) urinary tract; 8/9 (89%) vascular-catheter; 4/4 (100%) skin/soft tissue; 2/4 (50%) other; and 11/12 (92%) unknown sources. Patients experienced a median (IQR)14 (8–17) antibiotic-free days (of the 28 days after blood culture collection). Antimicrobial-related adverse events includedhepatitis in 1 (1%) patient, Clostridium difficile infection in 4 (4%), and secondary infection with highly resistant microorganismsin 10 (9%). Ascertainment was complete for all study outcomes in ICU, in hospital and at 90 days.(Continued on next page)* Correspondence: nick.daneman@sunnybrook.ca1Division of Infectious Diseases and Clinical Epidemiology, SunnybrookHealth Sciences Centre, University of Toronto and Adjunct Scientist, Institutefor Clinical Evaluative Sciences, Sunnybrook Health Sciences Centre, 2075Bayview Ave, Toronto, ON M4N 3M5, CanadaFull 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 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe 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.Daneman et al. Trials  (2018) 19:111 DOI 10.1186/s13063-018-2474-1(Continued from previous page)Conclusion: It is feasible to conduct a RCT to determine whether 7 versus 14 days of antibiotic treatment is associated withcomparable 90-day survival.Trial registration: ClinicalTrials.gov, identifier: NCT02261506. Registered on 26 September 2014.Keywords: Bacteremia, Bloodstream infection, Critical care, Intensive care, Duration of treatmentBackgroundBoth antibiotic use and the acquisition of antibiotic-resistant organisms are high in intensive care units (ICUs),where critically ill patients are vulnerable to bacterial infec-tions and antibiotic complications. Shortened treatmentdurations offer an appealing opportunity to maximize thebenefits while minimizing the harms and reducing the costsof antibiotic therapy [1]. Randomized clinical trials (RCTs)have established that shorter-duration treatments are suffi-cient for a wide range of bacterial infections [1–5], includ-ing some infections among critically ill patients [6], but theoptimal treatment duration for bloodstream infections re-mains understudied.Our systematic review of the academic literaturerevealed no trials of shorter- versus longer-duration treat-ment among adult patients with bloodstream infection,but did uncover 24 studies (of 7595 patients) of shorterversus longer antibiotic treatment for infections com-monly complicated by bacteremia. Cure rates were similarin patients receiving shorter (3–7 days) versus longer (7–21 days) treatment (risk ratio 1.00, 95% confidence inter-val (CI) 0.98–1.01) [7]. Although, subgroup outcome datawere only available for small numbers of bacteremicpatients across these studies (n = 115), shorter-durationtreatment was associated with similar clinical cure (45/52versus 47/49, risk ratio 0.88, 95% CI 0.77–1.01) andsurvival rates (15/17 versus 26/29, risk ratio 0.97, 95% CI0.76–1.23) as longer-duration treatments. Given the ab-sence of evidence to guide treatment durations for thesebacteremic patients, we conducted a national practice sur-vey of infectious disease physicians and critical care physi-cians [8]. Among these clinicians, 14 days was the mostcommonly reported treatment duration, but shorter treat-ments (usually 7 or 10 days) were recommended by nearlyhalf of practitioners [8]. In single [9] and multicenter [10]observational studies of actual practice, we confirmed thatcritically ill patients with bacteremia receive a median of14 days of treatment, but with wide variability (interquar-tile range (IQR) 9–17.5 days), further justifying the needfor a trial comparing 7 versus 14 days of treatment. If sucha trial could establish that 7 days is non-inferior to 14 daysof treatment, this could lead to massive reductions in anti-microbial use, resistance and complications, and an esti-mated annual cost savings of $678 to 798 million in NorthAmerica and $1.4 to 1.6 billion across Europe [11].Prior to embarking on a large multicentre trial, weconducted a pilot RCT to test the feasibility of this trialdesign. We hypothesized that the co-primary feasibilityoutcomes (recruitment rate and adherence to treatmentduration protocol) would confirm that a definitive trialcould be conducted to compare 7 versus 14 days of anti-biotic treatment for critically ill patients with blood-stream infections.MethodsGeneral study designWe conducted an open, pilot RCT of 7 versus 14 days ofantibiotic treatment, for critically ill patients with blood-stream infection, to test the feasibility and inform the designof an international, non-inferiority RCT. The trial protocolof the Bacteremia Antibiotic Length Actually Needed forClinical Effectiveness (BALANCE) pilot RCT has been pre-viously published [12]; however, the key design elements aresummarized below. We have adhered to the ConsolidatedStandards of Reporting Trials (CONSORT) guideline forpilot trials [13].Study settingThe study was conducted through the Canadian CriticalCare Trials Group (CCCTG) at 11 participating ICUs,across 10 Canadian cities in five provinces. The ICUsstarted in staggered fashion (Sunnybrook Health SciencesCentre, Toronto, ON (October 2014); Kingston GeneralHospital, Kingston, ON (January 2015); Queen Elizabeth II,Halifax, NS (March 2015); The Ottawa Hospital, Ottawa,ON (June 2015); Université de Sherbrooke, Sherbrooke,Québec, QC (July 2015); St. Boniface Hospital, Winnipeg,MB (January 2016); L’Hôpital de l’Enfant-Jesus, Laval,Québec, QC (January 2016); CHUM, Montréal, Québec,QC (April 2016); Mount Sinai Hospital, Toronto, ON(April 2016); London Health Sciences Centre, London, ON(May 2016); and University of Alberta Hospital, Edmonton,AB (June 2016)). Patient enrollment was completed in July2016, and thus individual ICUs participated for durationsranging from 1 to 22 months. Institutional Research EthicsBoard approval was obtained at these participating sites.Inclusion criteriaThe inclusion criteria were intended to be broad, so as tobe generalizable to most critically ill patients withDaneman et al. Trials  (2018) 19:111 Page 2 of 9bloodstream infection. All adult patients (aged over18 years) with any positive blood culture yielding a patho-genic bacterium while in the ICU were included. Theblood culture need not have been collected in the ICU,but at the time of detection of the positive result thepatient needed to be under ICU care.Exclusion criteriaWe excluded patients who were severely immunocom-promised (only neutropenia or bone marrow, solid organor stem cell transplantation); had prosthetic valves or syn-thetic endovascular grafts; had a suspected or documentedsyndrome with an established requirement for prolongedtreatment (e.g., infective endocarditis, osteomyelitis/septicarthritis, undrainable/undrained abscess or unremovable/unremoved prosthetic infection); had a single positiveblood culture with coagulase-negative staphylococci, Ba-cillus spp., Corynebacterium spp., etc. (because these mayhave represented contamination rather than infection);had positive blood cultures with Staphylococcus aureus orfungal organisms (because observational data suggest thatthese warrant more prolonged treatment); or patientspreviously enrolled in this trial.Intervention: 7 versus 14 days of adequate antimicrobialtreatmentEligible consenting patients were randomized 1:1 toreceive a shorter duration of adequate antimicrobial treat-ment (7 days) versus a longer duration (14 days). Ad-equate antimicrobial treatment was defined as a regimenwith in vitro activity against the organism(s) responsiblefor the bloodstream infection. The duration of adequatetreatment was determined as the cumulative number ofcalendar days for which adequate treatment was deliveredincluding and beyond the date of collection of the indexblood culture specimen [12]. The selection of antimicro-bial agents, doses, intervals and routes of delivery were leftto the discretion of the clinical team; only the durationwas determined by randomization.Randomization, blinding, and allocation concealmentmethodsRandomization was accomplished via a central, web-basedsystem (http://www.randomize.net) using variable blocksizes, stratified by ICU site; patients were allocated 1:1 into7- versus 14-day treatment arms. Although placebo con-trols have been used in some RCTs of antibiotic treatmentduration [3, 14–19], we believe that they are not practicalor appropriate for bacteremia treatment in ICU because ofvariable pathogens, sources of bacteremia, mono- andcombination antibiotic treatment regimens, and the poten-tial for critically ill patients to develop secondary nosoco-mial infections requiring ongoing re-assessment oftreatment choices. We elected not to protocolize treatmentchoices given the variety of pathogens and infectioussyndromes causing the bacteremia, so as to maximizegeneralizability of study findings. As a consequence, patientsand clinicians were not blinded to treatment assignment. Tomitigate selection bias, and prevent clinicians from providingdifferential treatments to patients in the two arms, we main-tained allocation concealment until the end of day 7, anapproach used successfully in the PneumoA RCT of shorter-versus longer-duration antibiotic treatment for ventilator-associated pneumonia [6].Primary feasibility outcomesThe co-primary outcomes of this pilot RCT related tofeasibility of the main trial: (1) recruitment rates and (2)adherence to assigned treatment duration protocol. Wetargeted overall recruitment rates of one patient per ICUsite per month; we targeted protocol adherence rates of90% of antibiotic treatment courses within 7 ± 2 days inthe shorter-duration arm, and 14 ± 2 days in the longer-duration arm.Secondary clinical outcomesSecondary outcomes included 90-day mortality, which isthe planned primary outcome of the larger RCT, and thebest measure of net treatment efficacy, as well as othermeasures of net efficacy including ICU and hospital mortal-ity, relapse of bacteremia, ICU and hospital lengths of stay(LOS), mechanical ventilation and vasopressor duration.Other secondary outcomes include antibiotic-free days, aswell as measures of antibiotic-related harms, including ratesof Clostridium difficile infection, antibiotic adverse events(allergy, anaphylaxis, acute kidney injury, hepatitis) andsecondary nosocomial infection with antimicrobial-resistantorganisms. Antibiotic-free days were defined as the numberof calendar days within 28 days after blood culture collec-tion on which the patient did not receive any antibiotictreatments; any patient dying within 28 days of bloodculture collection was assigned zero antibiotic-free days. Acomposite definition of highly resistant microorganismswas modified from the description by de Smet et al. [20] aspreviously described [21].Mechanistic sub-studiesWithin the framework of this study, we also piloted twomechanistic sub-studies with plans for expansion withinthe larger RCT. First, in a procalcitonin (PCT) sub-study,we obtained blood PCT measurements on the day ofrandomization, and days 7, 10, and 14 from the indexblood culture collection. The results were not made avail-able to the treating team because this could have undulyinfluenced protocol adherence; rather, they were batch-tested at the end of the trial using VIDAS® B.R.A.H.M.S.PCT™, BioMérieux (Marcy L’Etoile, France). The outcomeof interest was the proportion of patients for whom PCTDaneman et al. Trials  (2018) 19:111 Page 3 of 9levels exceeded the usual threshold (0.25 IU/mL) forrecommending antibiotic treatment at day 7 and day 14.At a single ICU site, we also collected rectal swabs on theday of enrollment, day 7, day 14, and at either hospitaldischarge or 28 days post enrollment, for analysis of gutmicrobial diversity and bacterial community compositionby 16S rRNA gene sequencing, with the main outcomebeing taxonomic diversity (Shannon Diversity Index) [22].Data collection and follow-up detailsPatients were followed daily during the hospital stay, withentry of baseline characteristics, and outcome informationinto a secure, electronic Case Report Form. The secondaryoutcome (90-day mortality) was collected by follow-uptelephone call at 90 days from the index bacteremia.Statistical analysisThe analytic plan for the pilot RCT was primarily descrip-tive – determining the recruitment rate, overall and byICU site, and calculating the protocol adherence rate over-all, by ICU site, and by source of infection. A pilot RCT isnot adequately powered to identify a clinically importantdifference in safety endpoints or in overall mortalityamong patients receiving shorter- versus longer-durationtreatments – rather this is the intended goal of the BAL-ANCE main RCT. Therefore, as per the approach of mostpilot trials conducted by the CCCTG, we did not un-blindtreatment assignment in the BALANCE pilot RCT [23].Avoiding such underpowered analyses for clinical end-points decreases the risk of over-interpreting such results,which could unduly influence investigators, clinicians,peer-reviewers, and research ethics boards [23].Sample size calculationOur pilot RCT sample size (n = 115) was calculated suchthat we would be able to estimate our protocol adherencerate within a margin of error of ± 5%, with 95% confi-dence, with an adherence rate of 90%; or ± 7%, with 95%confidence, with an anticipated adherence rate of 80%.ResultsScreened, eligible and randomized patientsIn total, 1159 critically ill patients with a positive bloodculture were screened for eligibility, of whom 358 (31%)were eligible for enrollment (Fig. 1 CONSORT flowdiagram; Additional file 1). Among 801 patients who metthe exclusion criteria, the most common reasons for non-eligibility were single positive cultures with a commoncontaminant organism (49%), Staphylococcus aureusbacteremia (19%), suspected or documented syndromewith well-defined requirement for prolonged treatment(16%), severe immuncompromise (9%), prosthetic valve orsynthetic endovascular graft (4%) or fungemia (3%). Of theeligible patients, 115/358 (32%) were randomized into thestudy; the most common reasons for non-randomizationwere lack of consent by the ICU physician (29%) or thepatient or substitute decision-maker (SDM) (20%), or thelack of an available SDM (14%) (Fig. 1). There was no lossto follow-up and no missing outcome data.Fig. 1 Consolidated Standards of Reporting Trials (CONSORT) flow diagram describing eligibility screening and randomization assessmentsDaneman et al. Trials  (2018) 19:111 Page 4 of 9Patient, infection, and pathogen characteristicsPatients had a median age of 67 (IQR, 57–78) years andwere severely ill with a median Acute Physiology AndChronic Health Evaluation (APACHE II) score of 22 (IQR,18–26); 63 (55%) were male (Table 1). Most bloodstreaminfections were community-acquired (60%), with the re-mainder acquired on hospital wards (17%) or in the ICU(24%). The most common sources of bacteremia werelung (27%), abdomen (25%), and urinary tract (23%); themost common pathogens were Escherichia coli (22%),Klebsiella spp. (14%), and Enterococcus spp. (13%), butthere was a total of 25 different bacteria isolated amongthe index blood cultures of these patients (Table 1).Treatment characteristicsA total of 31 different antimicrobials were used in treatingthese patients, with the most common being piperacillin-tazobactam (in 73.9% of patients), vancomycin (56.5%),ceftriaxone (50.4%), ciprofloxacin (36.5%), and meropenem(30.4%) (Table 2). Patients received a median of 2 (IQR, 1–2) adequate antimicrobials in the empiric window prior toblood culture finalization, and 3 (IQR 2–3) in the period upto 30 days after blood culture collection. A source controlprocedure was required for 76/115 (66.1%) patients.Recruitment rateThe overall average recruitment rate was one patient persite per month. Across individual sites the median (IQR)recruitment rate was 0.7 (0.3–1.5) patients per month,with a range of 0–2.1 patients per month.Adherence to treatment duration protocolThe overall adherence to assigned treatment durationprotocol was 89/115 (77%). Across individual sites themedian (IQR) adherence to treatment duration protocolwas 71% (50–85%). Adherence to treatment duration variedaccording to underlying source of infection: 26/31 (84%)lung; 18/29 (62%) intra-abdominal; 20/26 (77%) urinarytract; 8/9 (89%) vascular-catheter; 4/4 (100%) skin and softtissue; 2/4 (50%) other; and 11/12 (92%) unknown sourcesof infection. The most common reasons for protocol non-adherence were re-starting of antibiotics for documented/suspected recurrence or persistence of the index infection(6), clinical reasons for extended treatment that were notpresent or unrecognized at time of enrollment (4), a newinfection unrelated to the index infection (4), ward phys-ician not agreeing with ICU orders after transfer (2), recentsurgery (1), and recent prosthetic valve insertion (1). In the14-day treatment arm, the most common reasons forprotocol non-adherence were clinical reasons for extendedtreatment that were not present or unrecognized at time ofenrollment (2), a new infection unrelated to the index infec-tion (2), physician error (1), new gastrointestinal tractperforation (1), life-sustaining treatment withdrawal (1),and recurrence or persistence of index infection (1). Therewas a trend towards higher protocol adherence among thesecond half of the patient group enrolled at each ICU site(82%) as compared to the first half of the enrolled patientgroup at each site (73%) (p = 0.24).Other feasibility outcomesNo patients were withdrawn from the trial, and there wereno losses to follow-up, with complete ascertainment of allstudy outcomes in ICU, in hospital, and at 90 days.Clinical outcomesThe overall mortality rate among randomized patientswas 7% in ICU, 13% in hospital, and 15% at 90 days(Table 3). Relapse of bacteremia with the same pathogenoccurred in 4 (3%) patients. Patients experienced a me-dian of 14 antibiotic-free days, with wide variability (IQR8–17 days), and bimodal distribution (modes at 14 and21 antibiotic-free days). No patients experienced allergy/anaphylaxis or acute kidney injury. There was one epi-sode of acute hepatitis (1%), four patients (3%) withClostridium difficile infection, and 10 (9%) with second-ary infection with a highly resistant microorganism, in-cluding four with methicillin-resistant Staphylococcusaureus, two with vancomycin-resistant Enterococci, twowith extended-spectrum beta-lactamase (ESBL)-produ-cing Enterobacteriaceae, two with multi-drug resistantEnterobacteriaceae and one with multi-drug resistantnon-Enterobacteriaceae.Mechanistic sub-studiesThe majority of patients (101/115, 88%) consented tooptional blood testing for procalcitonin (PCT) measure-ments. The median (IQR) PCT levels were 5.0 (2.0–22.5) IU/mL on the day of randomization, 1.3 (0.3–4.3)on day 7, 0.46 (0.15–1.55) on day 10, and 0.28 (0.10–1.19) on day 14. Only a minority of patients had PCTlevels below the 0.25 threshold at which PCT algorithmswould recommend antibiotic discontinuation, including12/71 (17%) on the day 7 measurement, and 21/47(45%) on the day-14 measurement. Most patients (27/37, 73%) at the central study site consented to rectalswabs for microbiome testing, which confirmed the ad-equacy of this sampling method for bacterial communitytesting, and considerable range in the Shannon DiversityIndex, with a median (range) of 4.1 (0.2–5.5) across allspecimens.DiscussionThe BALANCE pilot RCT has confirmed the feasibility ofconducting a large, multicenter RCT comparing 7 versus14 days of antibiotic treatment for critically ill patientswith bloodstream infection. It was feasible to recruitpatients into this trial, and adherence to treatmentDaneman et al. Trials  (2018) 19:111 Page 5 of 9Table 1 Patient, infection, and pathogen characteristics amongcritically ill patients with bloodstream infectionCharacteristica N (%), Median (IQR)Patient characteristicMale sex 63 (55)Age in years 67 (57–78)APACHE II Score 22 (18–26)Baseline vasopressor use 60 (52)Admission categoryMedical 78 (68)Surgical 24 (21)Trauma 6 (5)Neurological/neurosurgical 6 (5)Burns 1 (1)ComorbidityCoronary artery disease 23 (20)Congestive heart failure 16 (14)Arrhythmia 15 (13)Peripheral vascular disease 14 (12)Diabetes mellitus 40 (35)Renal insufficiency 13 (11)Dialysis dependency 4 (4)Chronic obstructive pulmonary disease 16 (14)Liver disease 8 (7)Obesity 16 (14)Solid malignancy 18 (16)Leukemia/lymphoma 1 (1)Steroids/immunosuppression 10 (9)Infection characteristicsAcquisition of bacteremiaCommunity-acquired 69 (60)Hospital-acquired 19 (17)ICU-acquired 27 (24)Source of bacteremiaLung 31 (27)Intra-abdominal/hepato-biliary 29 (25)Urinary tract 26 (23)Vascular-catheter-related 9 (8)Skin and/or soft tissue 4 (3)Other 4 (3)Undefined/unknown 12 (10)Top 10 most commonly isolated pathogens in blood culturesbEscherichia coli 28 (22)Klebsiella spp. 18 (14)Enterococcus spp. 17 (13)Streptococcus pneumoniae 13 (10)Table 1 Patient, infection, and pathogen characteristics amongcritically ill patients with bloodstream infection (Continued)Characteristica N (%), Median (IQR)Coagulase-negative staphylococci 12 (9)Enterobacter spp. 6 (5)Pseudomonas spp. 4 (3)Serratia spp. 4 (3)Citrobacter spp. 3 (2)Streptococcus anginosus group 3 (2)aAll data are presented as medians and interquartile ranges (IQR) unlessotherwise specifiedbA total of 24 different bacterial species were isolated among the index bloodcultures of the 115 patients; the denominator for these percentages is allorganisms isolated from patients’ index blood culturesTable 2 Treatment characteristics among critically ill patientswith bacteremiaTreatment characteristicsa N (%), Median(IQR)Empiric antimicrobial treatmentbNumber of unique empiric antimicrobialsadministered2 (2–3)Number of unique adequated empiricantimicrobials administered2 (1–2)Overall antimicrobial treatmentscNumber of unique antimicrobialsadministered3 (3–5)Number of unique adequated antimicrobialsadministered3 (2–3)Top 10 most commonly administeredantimicrobialsPiperacillin-tazobactam 85 (73.9%)Vancomycin 65 (56.5%)Ceftriaxone 58 (50.4%)Ciprofloxacin 42 (36.5%)Meropenem 35 (30.4%)Metronidazole 22 (19.1%)Cefazolin 20 (17.4%)Ampicillin 19 (16.5%)Tobramycin 14 (12.2%)Amoxicillin-clavulanate 11 (9.6%)Underwent a source control procedure 76 (66.1%)aAll data are presented as medians and interquartile ranges (IQR) unlessotherwise specifiedbEmpiric treatment window defined as period between blood culturecollection and finalizationcOverall treatments received within 30 days after blood culture collectiondAdequate antimicrobials defined by in vitro activity against the bloodculture pathogen(s)Daneman et al. Trials  (2018) 19:111 Page 6 of 9duration protocols was adequate to ensure wide separ-ation in antibiotic treatment between groups. No patientswere withdrawn from the trial, and there were no losses tofollow-up in ascertainment of any of the study outcomesin ICU, in hospital, or at 90 days.Although, there was some heterogeneity in recruitmentrates across sites, we achieved our overall target recruit-ment average of one patient per ICU per month. Theprinciple BALANCE trial will seek to test whether 7 daysof treatment is associated with a non-inferior survival rateas compared to 14 days of treatment, with a non-inferiority margin of 4%; after inflation for two interimanalyses and potential losses to follow-up, the sample sizerequirement will be 3598. Therefore, at the rate of recruit-ment in our pilot RCT we will require approximately 60ICUs, recruiting patients over a 4- to 5-year period. Astudy of this magnitude cannot be conducted within asingle country. Through Canadian Institutes of Health Re-search (CIHR) funding, we have initiated enrollment in 20Canadian ICUs (http://balance.ccctg.ca). The BALANCEresearch program has also garnered enthusiastic collabor-ation from critical care trials groups in other countries in-cluding the United States, Australia, New Zealand, SaudiArabia, France, Germany and the United Kingdom, whereenrollment and potential parallel funding applications willhelp to foster accrual of the remaining 40 ICUs.The adherence to treatment duration protocol waslower than our initial target of 90%, but it was similar tothe landmark PneumoA study of 8 versus 15 days oftreatment for ventilator-associated pneumonia, whichdetected 18% protocol non-adherence in the 8-day treat-ment arm [6], and was far lower than the non-adherencerates of as high as 50% seen in some studies ofbiomarker-guided treatment for infections in critically illpatients [24]. Still, non-adherence to treatment durationprotocol, particularly in patients in the 7-day treatmentarm, represents the main threat to the validity of theprinciple BALANCE trial. Non-adherence cannot becompletely eliminated because some ICU patients willdevelop secondary infections or persistent sepsis forwhich antibiotics will be re-initiated or continued, and itis appropriate for those patients to receive treatment, aswould occur outside of a trial. We aim for a pragmaticapproach to intervention duration, and for patient care.However, several lessons learned from this pilot RCTwill enable us to minimize non-adherence in the largerRCT, including: use of delayed randomization whenimaging is pending to rule out abscesses or confirmsuccess of initial source control procedures; daily re-search coordinator contact with the clinical team tomonitor for premature or delayed antibiotic discontinu-ation; and real-time involvement of both infectiousdiseases and ICU site co-investigators in discussing casesof potential non-adherence. The fact that adherencerates were higher in the second versus the first half ofthe enrollments at each site suggests that adherencerates may improve significantly over time as the mainRCT unfolds. Nevertheless, in the principle BALANCERCT it will be important to conduct both a primaryintention-to-treat analysis and a secondary per-protocolanalysis; the strength of the inference of non-inferioritywill depend on whether the results of the two analyticapproaches are in harmony.We have not measured treatment group-separatedclinical and safety outcomes: the pilot RCT is inadequatelypowered to assess these outcomes. Spurious differencesbetween groups could lead to unnecessary concerns aboutembarking on the main RCT, while similar outcomesbetween groups could lead to premature acceptance ofnon-inferiority and the adoption of shorter-duration treat-ment [23]. Instead, this is an internal pilot and the patientsenrolled to date will be included and analyzed within theprinciple RCT. Nevertheless, from the aggregate clinicaloutcomes in the cohort of 115 randomized patients wehave still been able to confirm low rates of antibiotic-related adverse events, and a wide bi-modal distribution ofantibiotic-free days. Importantly, the overall rates of Clos-tridium difficile and secondary infections with antibiotic-resistant organisms are high enough that we may be able todetect potentially important benefits of shorter treatmentTable 3 Clinical outcomes among critically ill patients withbacteremiaOutcomea N (%), Median (IQR)Mortalityin ICU 8 (7)in hospital 15 (13)at 90 days 17 (15)Length of stay (in days)in ICU 8 (4–20)in hospital 20 (12–42)Duration of life support (in days)mechanical ventilation 8 (3–21)Relapse of bacteremia 4 (4)Antibiotic-free days (by day 30) 14 (8–17)Antimicrobial-related adverse outcomesAllergy 0 (0)Anaphylaxis 0 (0)Acute kidney injury 0 (0)Acute hepatitis 1 (1)Clostridium difficile infection 4 (4)Secondary infection with highly resistantmicroorganisms10 (9)aAll data are presented as medians and interquartile ranges (IQR) unlessotherwise specified. ICU intensive care unitDaneman et al. Trials  (2018) 19:111 Page 7 of 9in the principle RCT. For example, the main trial will have85% power to detect a reduction in Clostridium difficile to3% from 5% with 7 versus 14 days of antibiotics.One important limitation of the BALANCE pilot andmain RCT is that we will only be able to compare out-comes among patients with two fixed durations of treat-ment, 7 versus 14 days, rather than giving individualpatients the minimum duration necessary for curing theirparticular infection. However, we believe that there arecurrently no adequate clinical indicators of cure in critic-ally ill patients with bloodstream infection since host signsof sepsis can persist beyond cure of the triggering infec-tion. Although biomarkers, such as PCT, offer a promisingmeans to individualize and reduce antibiotic treatmentdurations [25, 26], there has been poor adherence to PCTguidance in key trials [24], and this technology has not yetbeen widely adopted. The results of the BALANCE RCTcould be potentially synergistic with prior and subsequentbiomarker-focused trials of treatment duration; if 7 days isnon-inferior to 14 days of treatment this may lead togreater physician adherence to biomarker-guided stoppingrules in the future. Moreover, the blinded PCT resultsfrom our study were above the 0.25-IU/mL treatmentthreshold in the majority of patients at both day 7 and atday 14, suggesting that PCT-guided treatment has thepotential to lead to treatment prolongation, rather thanreduction, in this clinical context. Finally, as in other stud-ies that compare two arbitrary and fixed interventionthresholds (such as the TRICC trial of transfusion thresh-olds) [27], an adequately powered trial will provide abenchmark duration of therapy around which to makeindividualized treatment decisions.ConclusionsIn the BALANCE pilot RCT, we have documented thatrandomization of patients to 7 versus 14 days of treatmentis acceptable and feasible. Based on these findings, CIHRhas funded the BALANCE main RCT to be conducted inboth Canadian and international ICUs, with parallel fund-ing applications in progress or under review in collaborat-ing critical care trials groups in other countries. TheBALANCE main RCT will provide foundational evidenceregarding treatment duration for our sickest and most vul-nerable patients. If 7 days is non-inferior to 14 days of treat-ment, the result could be a paradigm shift in antibiotictreatment durations for these patients, with immediate sub-stantial healthcare savings, and downstream reductions inClostridium difficile and antimicrobial resistant pathogens.Additional fileAdditional file 1: CONSORT 2010 Checklist of information to includewhen reporting a pilot or feasibility trial*. (DOC 226 kb)AbbreviationsAPACHE: Acute Physiology And Chronic Health Evaluation; CCCTG: CanadianCritical Care Trials Group; ICU: Intensive care unit; IQR: Interquartile range;LOS: Lengths of stay; PCT: Procalcitonin; RCT: Randomized clinical trialAcknowledgementsWe would like to acknowledge the extraordinary efforts of the researchcoordinators at each participating site in screening and enrolling eligiblepatients, including: Miranda Hunt, Ilinca Georgescu, Eileen Campbell, LisaJulien, Rebecca Porteous, Irene Watpool, Brigette Gomes, Marie-HeleneMasse, Hélène Fournier, Marie-Claude Tremblay, Danny Barriault/Gabrielle,Orla Smith, Gyan Sandhu, Stacy Ruddell, Joshua Booth, Nadia Baig, NicoleMarten, Suzette Willems, Victoria Alcuaz, Martine Lebrasseur, Sumesh Shah,Lia Stenyk, Barbara Kosky. We also like to acknowledge the crucial contribu-tions of Lisa Buckingham and Nicole Zytaruk at the CLARITY methods centerin helping to develop the electronic Case Report Form and database.FundingSupported by grants from Ontario Ministry of Health and Long-Term Care AcademicHealth Sciences Centre Alternative Funding Plan (AFP) Innovation Fund Awards andCanadian Institute of Health Research (CIHR). Dr. Fowler’s work was supported by apersonnel award from the Heart and Stroke Foundation, Ontario Provincial Office.Availability of data and materialsThe datasets used and/or analyzed during the current study are availablefrom the corresponding author on reasonable request.Authors’ contributionsND and RAF conceived and designed the study, obtained funding,developed the statistical analysis plan, database development, drafted themanuscript, and are responsible for overall management and supervision.AHR participated in the study design, helped develop the paper andelectronic CRF, and web randomization and drafting of the manuscript. RPcontributed to the study design, provided statistical and methodologicalexpertise, and helped draft the manuscript. AP, SMB, AC, EC, BC, DC, MD, PD,RH, AK, FL, FL, JM, CMM, LM, JM, SR, WS, HTS, and MEW participated in thestudy design, contributed to writing grant applications and helped inreviewing and revising the manuscript for intellectual content. All theauthors have reviewed and approved the manuscript for publication.Ethics approval and consent to participateInformed consent as approved by the following Research Ethics Boards was obtainedfrom all the participants or substitute decision-maker (SDM) as appropriate.Research Ethics Committee approval was obtained at the followingparticipating sites:Sunnybrook Health Sciences Center Research Ethics Board, Toronto, ON.Research Ethics Board, Queen’s university, Kingston, ON.Capital Health Research Ethics Board, Halifax, NS.Ottawa Hospital Research Ethics Boards, Ottawa, ON.Comité d’éthique de la recherché du CIUSSS de l’Estrie – CHUS, Sherbrooke, QC.University of Manitoba Health Research Ethics Board, Winnipeg, MB.Comité d’évaluation scientifique, Centre de Recherche CHUS, Québec, QC.Research Ethics Board, Mount Sinai Hospital, ON.Research Ethics Board, University of Western Ontario, London, ON.Research Ethics Office, University of Alberta, Edmonton, ABConsent for publicationNot applicableCompeting interestsDr. Nick Daneman is supported by a Clinician Scientist salary award from theCanadian Institutes of Health Research (CIHR). Dr. Rob Fowler is supported bya personnel award from the Heart and Stroke Foundation, Ontario ProvincialOffice. Dr. Deborah Cook holds a Canada Research Chair of Research Transferin Intensive Care. Drs. Lauzier and Lamontagne hold a career award from theFonds de Recherche du Québec-Santé. Dr. John Marshall is an associateeditor for Critical Care. Dr. Sean Bagshaw holds a Canada Research Chair inCritical Care Nephrology. Dr. HT Stelfox is supported by a Population HealthInvestigator Award from Alberta Innovates and an Embedded Clinician ResearcherAward from CIHR. This pilot randomized controlled trial was supported by operatingfunds from the Ontario Ministry of Health and Long-Term Care Academic HealthDaneman et al. Trials  (2018) 19:111 Page 8 of 9Sciences Alternate Funding Plan Innovation Fund Award, and bridge funding fromthe Canadian Institutes of Health Research. The BALANCE main trial is supported bya Project Grant from the Canadian Institutes of Health Research.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1Division of Infectious Diseases and Clinical Epidemiology, SunnybrookHealth Sciences Centre, University of Toronto and Adjunct Scientist, Institutefor Clinical Evaluative Sciences, Sunnybrook Health Sciences Centre, 2075Bayview Ave, Toronto, ON M4N 3M5, Canada. 2Department of Critical CareMedicine, Sunnybrook Health Sciences Center, Toronto, ON, Canada. 3Servicede Soins Intensifs et Centre de Recherche, Centre Hospitalier de l’Universitéde Montréal, Montréal, QC, Canada. 4Department of Critical Care Medicine,Faculty of Medicine and Dentistry, University of Alberta, Edmonton, AB,Canada. 5Department of Microbiology and Infectious Diseases, Université deSherbrooke, Sherbrooke, QC, Canada. 6Department of Critical Care Medicine,Hôpital du Sacré-Coeur de Montreal and Hôpital de Trois-Rivières, Universityof Montreal, Montreal, QC, Canada. 7Division of Infectious Diseases, Universityof Toronto, Toronto, ON, Canada. 8Division of Critical Care Medicine,Department of Medicine, McMaster University, Hamilton, ON, Canada.9Division of Critical Care, Department of Medicine, Sinai Health System,Toronto, ON, Canada. 10Division of Critical Care Medicine and Center forHealth Evaluation and Outcome Sciences, St. Paul’s Hospital and University ofBC, Vancouver, BC, Canada. 11Departments of Critical Care Medicine andAnesthesiology, Pain Management and Perioperative Medicine, DalhousieUniversity, Halifax, NS, Canada. 12Section of Critical Care Medicine, Universityof Manitoba, Winnipeg, MB, Canada. 13Centre de Recherche du CHU deSherbrooke and Department of Medicine, Université de Sherbrooke,Sherbrooke, QC, Canada. 14Centre de Recherche du CHU deQuébec-Université Laval, Axe Santé des Populations et Pratiques Optimalesen Santé, Division de Soins Intensifs, Québec, QC, Canada. 15Departments ofSurgery and Critical Care Medicine, St. Michael’s Hospital, University ofToronto, Toronto, ON, Canada. 16Department of Medicine, University ofWestern Ontario, London, ON, Canada. 17Division of Critical Care, Departmentof Medicine, The Ottawa Hospital, Ottawa, ON, Canada. 18Department ofCritical Care Medicine, Queen’s University, Kingston, ON, Canada.19Department of Biophysiology and Kinesiology, Simon Fraser University,Burnaby, BC, Canada. 20Department of Critical Care Medicine, Institute ofPublic Health, University of Calgary, Calgary, AB, Canada. 21Division of CriticalCare, Department of Medicine, Toronto Western Hospital, Toronto, ON,Canada. 22Departments of Medicine and Critical Care Medicine, SunnybrookHealth Sciences Center, Adjunct Scientist, Institute for Clinical EvaluativeSciences, Sunnybrook Health Sciences Centre, Institute of Health Policy,Management and Evaluation, University of Toronto, Toronto, ON, Canada.Received: 3 August 2017 Accepted: 16 January 2018References1. Rice LB. The Maxwell Finland Lecture: for the duration-rational antibioticadministration in an era of antimicrobial resistance and Clostridium difficile.Clin Infect Dis. 2008;46(4):491–6.2. el MR, de Borgie CA, den BP V, et al. Effectiveness of discontinuingantibiotic treatment after three days versus eight days in mild to moderate-severe community acquired pneumonia: randomised, double blind study.BMJ. 2006;332(7554):1355.3. Hepburn MJ, Dooley DP, Skidmore PJ, Ellis MW, Starnes WF, Hasewinkle WC.Comparison of short-course (5 days) and standard (10 days) treatment foruncomplicated cellulitis. Arch Intern Med. 2004;164(15):1669–74.4. Eliakim-Raz N, Yahav D, Paul M, Leibovici L. Duration of antibiotic treatmentfor acute pyelonephritis and septic urinary tract infection—7 days or lessversus longer treatment: systematic review and meta-analysis ofrandomized controlled trials. J Antimicrob Chemother. 2013;68(10):2183-91.5. Rafailidis PI, Pitsounis AI, Falagas ME. Meta-analyses on the optimization ofthe duration of antimicrobial treatment for various infections. Infect Dis ClinN Am. 2009;23(2):269–76. Table.6. Chastre J, Wolff M, Fagon JY, et al. Comparison of 8 vs 15 days of antibiotictherapy for ventilator-associated pneumonia in adults: a randomized trial.JAMA. 2003;290(19):2588–98.7. Havey TC, Fowler RA, Daneman N. Duration of antibiotic therapy forbacteremia: a systematic review and meta-analysis. Crit Care. 2011;15(6):R267.8. Daneman N, Shore K, Pinto R, Fowler R. Antibiotic treatment duration forbloodstream infections in critically ill patients: a National Survey of CanadianInfectious Diseases and Critical Care Specialists. Int J Antimicrob Agents.2011;38:480–5.9. Havey T, Fowler R, Pinto R, Elligsen M, Daneman N. Duration of antibiotictherapy in intensive care unit bloodstream infections. Can J Infect Dis MedMicrobiol. 2013;24(3):129–37.10. Daneman N, Rishu AH, Xiong W, et al. Duration of Antimicrobial Treatment forBacteremia in Canadian Critically Ill Patients. Crit Care Med. 2016;44(2):256–64.11. Daneman N, Rishu A, Xiong W, Palmay L, Fowler RA. Antimicrobial costsavings associated with shorter duration treatment for bloodstreaminfections. JAMMI. 2016;1(2):33-34.12. Daneman N, Rishu AH, Xiong W, et al. Bacteremia Antibiotic Length ActuallyNeeded for Clinical Effectiveness (BALANCE): study protocol for a pilotrandomized controlled trial. Trials. 2015;16(1):173.13. Eldridge SM, Chan CL, Campbell MJ, et al. CONSORT 2010 statement: extensionto randomised pilot and feasibility trials. Pilot Feasibility Stud. 2016;2:64.14. Sandberg T, Skoog G, Hermansson AB, et al. Ciprofloxacin for 7 days versus 14days in women with acute pyelonephritis: a randomised, open-label and double-blind, placebo-controlled, non-inferiority trial. Lancet. 2012;380(9840):484–90.15. Jernelius H, Zbornik J, Bauer CA. One or three weeks’ treatment of acutepyelonephritis? A double-blind comparison, using a fixed combination ofpivampicillin plus pivmecillinam. Acta Med Scand. 1988;223(5):469–77.16. Agarwal G, Awasthi S, Kabra SK, Kaul A, Singhi S, Walter SD. Three dayversus five day treatment with amoxicillin for non-severe pneumonia inyoung children: a multicentre randomised controlled trial. BMJ. 2004;328(7443):791.17. Clinical efficacy of 3 days versus 5 days of oral amoxicillin for treatment ofchildhood pneumonia: a multicentre double-blind trial. Lancet. 2002;360(9336):835-841.18. Dunbar LM, Wunderink RG, Habib MP, et al. High-dose, short-courselevofloxacin for community-acquired pneumonia: a new treatmentparadigm. Clin Infect Dis. 2003;37(6):752–60.19. Tellier G, Niederman MS, Nusrat R, Patel M, Lavin B. Clinical andbacteriological efficacy and safety of 5 and 7 day regimens of telithromycinonce daily compared with a 10 day regimen of clarithromycin twice daily inpatients with mild to moderate community-acquired pneumonia. JAntimicrob Chemother. 2004;54(2):515–23.20. de Smet AM, Kluytmans JA, Blok HE, et al. Selective digestive tractdecontamination and selective oropharyngeal decontamination and antibioticresistance in patients in intensive-care units: an open-label, clustered group-randomised, crossover study. Lancet Infect Dis. 2011;11(5):372–80.21. Savage RD, Fowler RA, Rishu AH, et al. The effect of inadequate initialempiric antimicrobial treatment on mortality in critically ill patients withbloodstream infections: a multi-centre retrospective cohort study. PLoS One.2016;11(5):e0154944.22. Dethlefsen L, Relman DA. Incomplete recovery and individualized responsesof the human distal gut microbiota to repeated antibiotic perturbation.Proc Natl Acad Sci U S A. 2011;108(Suppl 1):4554–61.23. Arnold DM, Burns KE, Adhikari NK, Kho ME, Meade MO, Cook DJ. The designand interpretation of pilot trials in clinical research in critical care. Crit CareMed. 2009;37(1 Suppl):S69–74.24. Bouadma L, Luyt CE, Tubach F, et al. Use of procalcitonin to reducepatients’ exposure to antibiotics in intensive care units (PRORATA trial): amulticentre randomised controlled trial. Lancet. 2010;375(9713):463–74.25. Kopterides P, Siempos II, Tsangaris I, Tsantes A, Armaganidis A.Procalcitonin-guided algorithms of antibiotic therapy in the intensive careunit: a systematic review and meta-analysis of randomized controlled trials.Crit Care Med. 2010;38(11):2229–41.26. Prkno A, Wacker C, Brunkhorst FM, Schlattmann P. Procalcitonin-guidedtherapy in intensive care unit patients with severe sepsis and septicshock—a systematic review and meta-analysis. Crit Care. 2013;17(6):R291.27. Hebert PC, Wells G, Blajchman MA, et al. A multicenter, randomized,controlled clinical trial of transfusion requirements in critical care.Transfusion Requirements in Critical Care Investigators, Canadian CriticalCare Trials Group. N Engl J Med. 1999;340(6):409–17.Daneman et al. Trials  (2018) 19:111 Page 9 of 9


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items