Development and validation of an automated antimicrobial surveillance system based on indications for antimicrobial administration.

Abstract

Introduction: We developed an antimicrobial and patient background surveillance system (APBSS), an automated surveillance system that can calculate surveillance data such as antimicrobial use and detection of antimicrobial resistance for each indication of antimicrobial administration. We evaluated the validity of the APBSS data.

Methods: Eligible patients were hospitalized at the Toyota Kosei Hospital on July 7, 2022. Evaluated surveillance data included antimicrobial administration, indications for antimicrobial administration, and diagnosis. In the APBSS, surveillance data were calculated using the Diagnosis Procedure Combination data and Japan Nosocomial Infections Surveillance laboratory data. Using surveillance data collected by the Point Prevalence Survey (PPS) as a reference standard, the agreement between the results calculated based on the APBSS was evaluated using Cohen's kappa coefficient. Indications for antimicrobial administration and diagnosis were analyzed in patients identified for antimicrobial administration in PPS or APBSS.

Results: A total of 582 patients were included in this study, 223 of whom were evaluated for indications for antimicrobial administration and diagnosis. For the indications of antimicrobial administration, the Cohen's kappa coefficient was almost perfect (0.81-1.00) for all items. Cohen's kappa coefficient for the diagnosis of healthcare-associated infections was low. However, in major diseases (pneumonia and intra-abdominal, and symptomatic upper urinary tract infections) among community-acquired infections (CAIs) diagnosis, Cohen's kappa coefficient was substantial (0.61-0.80).

Conclusions: The APBSS can identify indications for antimicrobial administration and major CAIs with high accuracy. Therefore, the APBSS can calculate surveillance data, such as antimicrobial use and detection of antimicrobial resistance, for each of these items.