**Background:** Antimicrobial resistance (AMR) is a top global public health threat, with an estimated 1.27 million deaths directly attributable to antibiotic-resistant bacterial infections in 2019. Extended-spectrum beta-lactamase (ESBL) and AmpC-producing Escherichia coli (ESC-EC) are among the most important pathogen-drug combinations, and their dissemination across human, animal, and environmental compartments exemplifies the One Health nature of AMR. Despite growing recognition of the need for cross-sectoral data sharing, existing databases are fragmented, lack standardization, and often exclude environmental or companion animal sources. The DiSCoVeR (Discovering the sources of Salmonella, Campylobacter, VTEC and antimicrobial resistance) project, funded by the One Health European Joint Programme, aimed to build a prototype international strain-level database for ESC-EC to assess the feasibility and challenges of such an endeavor.
**Methods:** Nineteen partners from 13 European Union countries participated in the DiSCoVeR project. Thirteen partners from eight countries (Czech Republic, Denmark, Germany, Ireland, Netherlands, Poland, Portugal, Spain) contributed ESC-EC isolate data. Partners completed a shared template providing isolate ID, year of isolation, institute and country, country of origin of the sample, source information (source group, subgroup, sample type), whole genome sequence status, antimicrobial susceptibility testing method and minimum inhibitory concentration (MIC) data, and ESBL/AmpC gene presence. Most isolates (n=10,035, 93%) were tested using the harmonized EU antibiotic panel for Salmonella/E. coli (EUVSEC, Sensititre™). Phenotypic classification followed EUCAST guidelines and EFSA recommendations. Genotypic data were obtained via PCR or whole genome sequencing (WGS). Data were collected from national monitoring programs (per Commission Implementing Decision 2013/652/EU) and additional research projects.
**Key Results:** The database gathered 10,763 isolates collected between 2013 and 2020. The distribution by source was heavily skewed: livestock (n=9,444, 88%), human (n=711, 7%), environment (n=470, 4%), wild animals (n=103, 1%), pets (n=26, <1%), fruit/vegetables (n=7, <1%), and zoo animals (n=2, <1%). Germany contributed the most isolates (n=3,955, 37%), followed by the Netherlands (n=2,727, 25%), Spain (n=2,228, 21%), Denmark (n=607, 6%), Ireland (n=539, 5%), Portugal (n=437, 4%), Czech Republic (n=151, 1%), and Poland (n=119, 1%). Of the 9,234 isolates tested by microdilution, 8,952 (96.9%) had MIC data for the second panel allowing ESBL phenotype confirmation. Among these, 7,237 (80.8%) had a confirmed ESBL phenotype, 528 (5.9%) had ESBL+AmpC phenotype, 1,146 (12.8%) had AmpC phenotype, and 41 (0.5%) expressed other phenotypes. Genotypically, 5,416 (50.3%) isolates carried at least one blaCTX-M gene, with blaCTX-M-1 being most common (n=2,667, 49.2%), followed by blaCTX-M-15 (n=1,132, 20.9%) and blaCTX-M-14 (n=375, 6.9%). Among 1,606 blaTEM-carrying isolates, blaTEM-1 was most common (n=1,015, 63.2%), followed by blaTEM-52 (n=544, 33.9%). Among 898 blaSHV-carrying isolates, blaSHV-12 dominated (n=860, 95.7%). Only 866 (8.0%) isolates had information on AmpC genotypes, with blaCMY-2 being the most common (480 of 485). Only 1,348 (13%) isolates were analyzed by WGS. Key limitations identified included: lack of geographical representativeness (only 8 of 27 EU member states), large gaps between sectors (human data from only 4 countries, environmental data from only 2 countries), heterogeneous data reporting (e.g., CTX-M groups vs. specific gene types), duplicate isolates, and 2,797 (26%) isolates lacking any ESBL/AmpC gene information.
**Clinical Implications:** The study demonstrates that while structured routine monitoring programs (e.g., EU legislation since 2014) provide harmonized, comparable data for livestock sources, they are insufficient for a comprehensive One Health approach due to the exclusion of human clinical isolates, environmental samples, companion animals, and wildlife. The severe underrepresentation of human and environmental data limits the ability to perform robust source attribution modeling, which is essential for identifying transmission pathways and designing targeted interventions. The authors note that human data are crucial for source attribution models, yet were only available from four countries, with Germany (n=463) and the Netherlands (n=230) providing the vast majority. The paper emphasizes that international data sharing following FAIR principles (findability, accessibility, interoperability, reusability) is essential, but faces challenges including data protection concerns, reluctance to share due to reputational risk, heterogeneous methodologies, and lack of standardized metadata. The authors call for minimum guidelines for analyzing and reporting ESC-EC, particularly for research projects, to enable integration with routine monitoring data. The increased use of WGS is advocated, but standardization of bioinformatics pipelines is needed. The DiSCoVeR database, despite its limitations, represents a proof-of-concept that cross-sectoral, international ESC-EC data can be compiled, and its development has identified specific challenges that must be addressed to build a truly comprehensive One Health AMR surveillance system capable of informing evidence-based mitigation strategies.