A pilot randomized controlled trial of 7 versus 14 days of antibiotic treatment for bloodstream infection on non-intensive care versus intensive care wards

Abstract

Background: The optimal treatment duration for patients with bloodstream infection is understudied. The Bacteremia Antibiotic Length Actually Needed for Clinical Effectiveness (BALANCE) pilot randomized clinical trial (RCT) determined that it was feasible to enroll and randomize intensive care unit (ICU) patients with bloodstream infection to 7 versus 14 days of treatment, and served as the vanguard for the ongoing BALANCE main RCT. We performed this BALANCE-Ward pilot RCT to examine the feasibility and impact of potentially extending the BALANCE main RCT to include patients hospitalized on non-ICU wards.

Methods: We conducted an open pilot RCT among a subset of six sites participating in the ongoing BALANCE RCT, randomizing patients with positive non-Staphylococcus aureus blood cultures on non-ICU wards to 7 versus 14 days of antibiotic treatment. The co-primary feasibility outcomes were recruitment rate and adherence to treatment duration protocol. We compared feasibility outcomes, patient/pathogen characteristics, and overall outcomes among those enrolled in this BALANCE-Ward and prior BALANCE-ICU pilot RCTs. We estimated the sample size and non-inferiority margin impacts of expanding the BALANCE main RCT to include non-ICU patients.

Results: A total of 134 patients were recruited over 47 site-months (mean 2.9 patients/site-month, median 1.0, range 0.1-4.4 patients/site-month). The overall recruitment rate exceeded the BALANCE-ICU pilot RCT (mean 1.10 patients/site-month, p < 0.0001). Overall protocol adherence also exceeded the adherence in the BALANCE-ICU pilot RCT (125/134, 93% vs 89/115, 77%, p = 0.0003). BALANCE-Ward patients were older, with lower Sequential Organ Failure Assessment scores, and higher proportions of infections caused by Escherichia coli and genito-urinary sources of bloodstream infection. The BALANCE-Ward pilot RCT patients had an overall 90-day mortality rate of 17/133 (12.8%), which was comparable to the 90-day mortality rate in the ICU pilot RCT (17/115, 14.8%) (p = 0.65). Simulation models indicated there would be minimal sample size and non-inferiority margin implications of expanding enrolment to increasing proportions of non-ICU versus ICU patients.

Conclusion: It is feasible to enroll non-ICU patients in a trial of 7 versus 14 days of antibiotics for bloodstream infection, and expanding the BALANCE RCT hospital-wide has the potential to improve the timeliness and generalizability of trial results.

Trial registration: Clinicaltrials.gov, NCT02917551. Registered on September 28, 2016.

Keywords: Bacteremia; Bloodstream infection; Critical care; Duration of treatment; Intensive care.

Fig. 1

CONSORT flow diagram describing eligibility screening and randomization assignments

Fig. 2

Estimating the final proportion of patients enrolled on non-ICU wards, as a function of the percentage of participating sites which expand to include non-ICU ward enrolments. This analysis assumes average rates of enrolment in ICU based on current BALANCE trial data and in non-ICU wards based on the BALANCE-Ward pilot. The colored lines depict projections accounting for current number of registered sites (red), as well as under assumptions of adding additional sites (five per year, blue; ten per year, black) over the duration of the trial. The projected number of months remaining until trial completion are listed above each data point

Fig. 3

Sample size implications of expanding the BALANCE main RCT to include non-ICU ward patients, as a function of the final percentage of patients enrolled from non-ICU wards and fixing the non-inferiority margin at 4%. The point estimates (solid black line) assume a mortality rate of 17.3% among ICU patients and 12.8% in non-ICU ward patients, with 95% certainty estimates around those estimates (gray shaded area)

Fig. 4

Non-inferiority margin implications of expanding the BALANCE main RCT to include non-ICU ward patients, as a function of the final percentage of patients enrolled from non-ICU wards and fixing the sample size at 3626. The point estimates (solid black line) assume a mortality rate of 17.3% among ICU patients and 12.8% in non-ICU ward patients, with 95% certainty estimates around those estimates (gray shaded area)