Presentation (CTI)

OBJECTIVES: More and more countries are implementing antimicrobial stewardship programs due to the serious global issue of antimicrobial resistance. In 2013, Taiwan introduced the National Health Insurance (NHI) PharmaCloud System to improve drug safety and reduce duplication. However, only a limited number of local studies have evaluated the effectiveness of this system from a nationwide perspective. Thus, this study aims to analyze trends and impacts on antibiotic prescription rates following the implementation of the NHI PharmaCloud System, particularly in clinics with high antibiotic use.
METHODS: We analyzed retrospective and longitudinal data from the NHI Administration’s Healthcare Quality Information Disclosure Platform on nationwide antibiotic prescription rates from 2006 to 2024. A Difference-in-Differences (DID) approach was used within a linear regression to examine trend effects in our study. Additionally, Generalized Estimating Equations (GEE) were employed to account for repeated measures across clinics and to assess the system’s impact on antibiotic use.
RESULTS: Among 7,364 clinics, the findings show that there was a significant effect of implementing the NHI PharmaCloud System for antibiotic stewardship (β = -0.008, P < 0.001), initially; however, this benefit diminished over time (β = 0.002, P < 0.001), indicating limited long-term effectiveness across all clinics. Nonetheless, antibiotic use rates decreased by 0.001 per year (P < 0.001) as the intervention progressed in public clinics, and the intervention’s impact was more pronounced in clinics with high antibiotic use (β = -0.030, P < 0.001).
CONCLUSIONS: The clinical advantage of antibiotic stewardship using the NHI PharmaCloud System is primarily observed in public clinics and high-utilization settings, whereas its impact in private clinics is minimal. We suggest that health authorities enhance targeted interventions and promotional efforts to encourage the rational use of antibiotics within the private primary care sector.