Shorter Versus Longer Courses of Antibiotics for Infection in Hospitalized Patients: A Systematic Review and Meta-Analysis
BACKGROUND: Infection is a leading cause of hospitalization with high morbidity and mortality, but there are limited data to guide the duration of antibiotic therapy.
PURPOSE: Systematic review to compare outcomes of shorter versus longer antibiotic courses among hospitalized adults and adolescents.
DATA SOURCES: MEDLINE and Embase databases, 1990-2017.
STUDY SELECTION: Inclusion criteria were human randomized controlled trials (RCTs) in English comparing a prespecified short course of antibiotics to a longer course for treatment of infection in hospitalized adults and adolescents aged 12 years and older.
DATA EXTRACTION: Two authors independently extracted study characteristics, methods of statistical analysis, outcomes, and risk of bias.
DATA SYNTHESIS: Of 5187 unique citations identified, 19 RCTs comprising 2867 patients met our inclusion criteria, including the following: 9 noninferiority trials, 1 superiority design trial, and 9 pilot studies. Across 13 studies evaluating 1727 patients, no significant difference in clinical efficacy was observed (d = 1.6% [95% confidence interval (CI), −1.0%-4.2%]). No significant difference was detected in microbiologic cure (8 studies, d = 1.2% [95% CI, −4.1%-6.4%]), short-term mortality (8 studies, d = 0.3% [95% CI, −1.2%-1.8%]), longer-term mortality (3 studies, d = −0.4% [95% CI, −6.3%-5.5%]), or recurrence (10 studies, d = 2.1% [95% CI, −1.2%-5.3%]). Heterogeneity across studies was not significant for any of the primary outcomes.
CONCLUSIONS: Based on the available literature, shorter courses of antibiotics can be safely utilized in hospitalized patients with common infections, including pneumonia, urinary tract infection, and intra-abdominal infection, to achieve clinical and microbiologic resolution without adverse effects on mortality or recurrence.
© 2018 Society of Hospital Medicine
Acute infections are a leading cause of hospitalization and are associated with high cost, morbidity, and mortality.1 There is a growing body of literature to support shorter antibiotic courses to treat several different infection types.2-6 This is because longer treatment courses promote the emergence of multidrug resistant (MDR) organisms,7-9 microbiome perturbation,10 and Clostridium difficile infection (CDI).11 They are also associated with more drug side effects, longer hospitalizations, and increased costs.
Despite increasing support for shorter treatment courses, inpatient prescribing practice varies widely, and redundant antibiotic therapy is common.12-14 Furthermore, aside from ventilator-associated pneumonia (VAP),15,16 prior systematic reviews of antibiotic duration have typically included outpatient and pediatric patients,3-6,17-19 for whom the risk of treatment failure may be lower.
Given the potential for harm with inappropriate antibiotic treatment duration and the variation in current clinical practice, we sought to systematically review clinical trials comparing shorter versus longer antibiotic courses in adolescents and adults hospitalized for acute infection. We focused on common sites of infection in hospitalized patients, including pulmonary, bloodstream, soft tissue, intra-abdominal, and urinary.20,21 We hypothesized that shorter courses would be sufficient to cure infection and associated with lower costs and fewer complications. Because we hypothesized that shorter durations would be sufficient regardless of clinical course, we focused on studies in which the short course of antibiotics was specified at study onset, not determined by clinical improvement or biomarkers. We analyzed all infection types together because current sepsis treatment guidelines place little emphasis on infection site.22 In contrast to prior reviews, we focused exclusively on adult and adolescent inpatients because the risks of a too-short treatment duration may be lower in pediatric and outpatient populations.
METHODS
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses.23 The review was registered on the Prospero database.24
Information Sources and Search Strategy
We performed serial literature searches for articles in English comparing shorter versus longer antibiotics courses in hospitalized patients. We searched MEDLINE via PubMed and Embase (January 1, 1990, to July 1, 2017). We used Boolean operators, Boolean logic, and controlled vocabulary (eg, Medical Subject Heading [MeSH] terms) for each key word. We identified published randomized controlled trials (RCTs) of conditions of interest (MeSH terms: “bacteremia,” “sepsis,” “pneumonia,” “pyelonephritis,” “intra-abdominal infection,” “cellulitis,” “soft tissue infection”) that compared differing lengths of antibiotic treatment (keywords: “time factors,” “duration,” “long course,” “short course”) and evaluated outcomes (key words: “mortality,” “recurrence,” “secondary infections”). We hand searched references of included citations. The full search strategy is presented in supplementary Appendix 1.
Study Eligibility and Selection Criteria
To meet criteria for inclusion, a study had to (1) be an RCT; (2) involve an adult or adolescent population age ≥12 years (or report outcomes separately for such patients); (3) involve an inpatient population (or report outcomes separately for inpatients); (4) stipulate a short course of antibiotics per protocol prior to randomization and not determined by clinical response, change in biomarkers, or physician discretion; (5) compare the short course to a longer course of antibiotics, which could be determined either per protocol or by some other measure; and (6) involve antibiotics given to treat infection, not as prophylaxis.
Two authors (SR and HCP) independently reviewed the title and/or abstracts of all articles identified by the search strategy. We calculated interrater agreement with a kappa coefficient. Both authors (SR and HCP) independently reviewed the full text of each article selected for possible inclusion by either author. Disagreement regarding eligibility was adjudicated by discussion.
Data Abstraction
Two authors (SR and HCP) independently abstracted study methodology, definitions, and outcomes for each study using a standardized abstraction tool (see supplementary Appendix 2).
Study Quality
We assessed article quality using the Cochrane Collaboration’s tool,25 which evaluates 6 domains of possible bias, including sequence generation, concealment, blinding, and incomplete or selective outcome reporting. The tool is a 6-point scale, with 6 being the best score. It is recommended for assessing bias because it evaluates randomization and allocation concealment, which are not included in other tools.26 We did not exclude studies based on quality but considered studies with scores of 5-6 to have a low overall risk of bias.
