The quality of studies evaluating antimicrobial stewardship interventions: a systematic review

Abstract

Background: Antimicrobial stewardship aims to optimize antibiotic use and minimize selection of antimicrobial resistance. The methodological quality of published studies in this field is unknown. Aims: Our objective was to perform a comprehensive systematic review of antimicrobial stewardship research design and identify features which limit validity and translation of research findings into clinical practice. Sources: The following online database was searched: PubMed. Study eligibility criteria: Studies published between January 1950 and January 2017, evaluating any antimicrobial stewardship intervention in the community or hospital setting, without restriction on study design or outcome. Content: We extracted data on pre-specified design quality features and factors that may influence design choices including (1) clinical setting, (2) age group studied, (3) when the study was conducted, (4) geographical region, and (5) financial support received. The initial search yielded 17 382 articles; 1008 were selected for full-text screening, of which 825 were included. Most studies (675/825, 82%) were non-experimental; 104 (15%) used interrupted time series analysis, 41 (6%) used external controls, and 19 (3%) used both. Studies in the community setting fulfilled a median of five out of 10 quality features (IQR 3–7) and 3 (IQR 2–4) in the hospital setting. Community setting studies (25%, 205/825) were significantly more likely to use randomization (OR 5.9; 95% CI 3.8–9.2), external controls (OR 5.6; 95% CI 3.6–8.5), and multiple centres (OR 10.5; 95% CI 7.1–15.7). From all studies, only 48% (398/825) reported clinical and 23% (190/825) reported microbiological outcomes. Quality did not improve over time. Implications: Overall quality of antimicrobial stewardship studies is low and has not improved over time. Most studies do not report clinical and microbiological outcome data. Studies conducted in the community setting were associated with better quality. These limitations should inform the design of future stewardship evaluations so that a robust evidence base can be built to guide clinical practice.