Causal Conditions Supporting Antibiotic Stewardship Information Dashboards

BACKGROUND: Antibiotic stewardship is key to minimizing antibiotic resistance. To assist antibiotic stewards in dissecting population-level antibiotic use patterns, our study group developed a dashboard that displays consolidated patterns, supports data exploration, and compares facility-level antibiotic use to others. We report fuzzy set qualitative comparative analyses (QCA) of interviews designed to elicit user experiences to uncover different combinations of causal conditions supporting dashboard use. METHODS: Dashboards were iteratively designed based upon longitudinal feedback from stewards. Views include antibiotic use stratified by diagnoses and duration of therapy. Eight VAMCs, each with 0.5 to 2.0 FTE stewards, used the dashboard. One to 2 stewards from each site were interviewed using a structured script that focused on: 1) structure (i.e., program FTE) and functions of the local stewardship program; 2) critical incident or usage story; and 3) perceived knowledge and efficacy. RESULTS: Qualitative codes were developed from the interviews and were scaled in a fuzzy logic framework (i.e., between 0 and 1) to reflect the degree to which the qualitative theme was present in the stewardship program at participating clinical sites. The scaling was assigned using prior knowledge external to the data. The most parsimonious QCA solution identified just the absence of program structure (program FTE) a sufficient causal configuration to the frequency of dashboard use (coverage = 0.612, consistency = 0.813). Intermediate solutions added stewardship activities, dashboard self-efficacy, and trust in the data (coverage = 0.502, consistency = 0.952) as sufficient conditions. The coverage for both solutions exceeded 0.75, which was the lower bound of acceptability. CONCLUSION: The dashboard may be successfully integrated into institutions based on the complicated interplay between program structure (e.g., # FTE) and dashboard self-efficacy, experience with data-activities, and trust of population data. Incorporating user-centered design of dashboards supports the development of fully functional teams and has the potential for important population health impact. DISCLOSURES: All authors: No reported disclosures.