Automated Data Collection on Antimicrobial Use in Dogs and Cats in a Tertiary Hospital and Private Practices

Judicious antimicrobial use in veterinary medicine is important because improper
antimicrobial use can contribute to the evolution of antimicrobial resistance in bacterial
pathogens, which makes subsequent use of these drugs less effective in both human
and veterinary medicine. There is very little on-the-ground information about veterinary
clinicians’ antimicrobial use (AMU) practices in companion animal practice in the US.
veterinary medicine. To improve our understanding of antimicrobial use in dogs and
cats, we propose to create a nationwide digital surveillance system to collect critical
AMU data using existing electronic practice information management systems (PIMS) in
collaboration with veterinary industry partners. The system will automatically harvest
AMU and patient data from digital PIMS. The proposed system will harvest data
collected in routine veterinary examinations from existing PIMS systems and therefore
will not require any additional effort from practitioners to participate in the program.
Natural language processing, a machine learning method used to classify unstructured
text, will be used to review electronic medical records to determine patients’ diagnosis.
We aim to prototype the system in our native digital PIMS at North Carolina State
University’s College of Veterinary Medicine Teaching hospital. We will then enroll
additional private veterinary practices, including general practice, specialty hospitals,
and emergency clinics, as sentinels and collect the same detailed PIMS data from a
more representative set of clinics. Working closely with the sentinel clinics will provide a
deep understanding of how our system operates in private clinics, and in the final stage
we aim to expand the fully automated system to PIMS nationwide. The combination of
sentinel clinics with the nationwide survey of clinics will create a powerful broad and
deep surveillance system for antimicrobial use in veterinary clinics. A broad suite of
AMU parameters will be estimated from this data, and the results reported to the FDA in
an annual report. Additionally, we will share the data with other researchers through an
web-based portal and GitHub repositories. This system will provide the critical data and
analysis to understand veterinary AMU in the US.