Evaluation of a Bayesian Belief Network tool for translating science and informing policy: the case of antibiotic misuse in Tanzania
Introduction: Developing appropriate antimicrobial stewardship policies is complicated by the many contextual factors that influence antimicrobial misuse. Bayesian Belief Networks (BBNs) can model and visualise disparate data types, making them an engaging and accessible means of translating interdi...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | International Journal of Infectious Diseases |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S120197122400657X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Introduction: Developing appropriate antimicrobial stewardship policies is complicated by the many contextual factors that influence antimicrobial misuse. Bayesian Belief Networks (BBNs) can model and visualise disparate data types, making them an engaging and accessible means of translating interdisciplinary complexity to non-experts. BBNs are considered statistically intuitive, are data-derivable and can be used to perform scenario-analyses to support evidence-based decision-making. To assess BNNs’ capacity to live up to this promise, we make a previously derived BBN for antibiotic misuse into a web-based interactive tool and evaluated it with non-specialists. Methods: A BBN of determinants of antibiotic misuse in Tanzanian urinary tract infection patients was incorporated into a web-based application, which included an interactive tool plus other relevant information, such as educational materials. Within the tool-specific pages users could interact with the BBN network and assess probability outcomes of different scenarios. Feedback from eight subject experts and stakeholders was gathered between January to May 2023. Content-analysis of written and oral feedback was used to assess the tool's translational value and relevance within policy decision-making processes. Results: The interactive visualisations promoted playful engagement with research and its implications. The tool also efficiently conveyed system complexities to non-experts which helped to discourage disciplinary siloing. However, users found it challenging to independently use and understand the tool due to technicalities and unfamiliarity with BBNs. Demonstrations and assistance mitigated these issues. Discussion: The time to acquaint oneself with the tool and BBNs in general was considered too high for adoption directly by policymakers. Instead, use of the tool was recommended for other areas including co-designing interventions and developing advocacy campaigns. Thus, the supportive presence of researchers during engagement proved an essential component speeding users’ use of the tool and establishing the tools additional value as a visual aid in co-productive discussions. Conclusions: The study shows the value of BBN visuals for translating system-complexities to non-expert audiences can be further enhanced by interactive engagement (afforded by a web-based application) and the support of researchers during model exploration. Furthermore, simultaneous multi-user and multi-disciplinary engagement with these interactive BBNs can provide a useful ‘discursive space’ for clarifying model understandings and the co-production of research. |
|---|---|
| ISSN: | 1201-9712 |