A reproducible One Health evidence database of antimicrobial resistance in Escherichia coli: 2,671 PubMed records graded for transmission evidence

Published: 8 October 2026| Version 1 | DOI: 10.17632/f8gfnhxjwn.1
Contributor:

Description

This dataset is a curated, reproducible evidence base on antimicrobial resistance (AMR) in Escherichia coli across the One Health continuum, assembled from five PubMed search strategies covering One Health AMR, resistance genes and mobile genetic elements, human–animal–environment transmission, environmental selection, and genomic transmission attribution. The five searches returned 3,525 records. After removing 854 cross-dataset duplicates on PMID, 2,671 unique records remained: 2,262 primary studies, 176 reviews held in a secondary layer, and 233 records flagged at screening. Every primary study was coded across 191 fields covering geography (121 countries), One Health compartment, sample and phenotype data, resistance genes (4,421 observations), plasmids and mobile genetic elements (1,790), E. coli lineages (1,454), transmission evidence (1,195) and selective pressures (906). The dataset's distinguishing feature is a seven-level transmission evidence hierarchy applied to every study, separating resistance detection, gene detection, gene mobility, plasmid association, clonal relatedness, cross-compartment dissemination and demonstrated transmission. The basis for each grade is recorded so that assignments can be audited and re-graded. A derived 27-sheet analysis workbook contains country, compartment, gene, plasmid, lineage, temporal, gap and statistical analyses, with a codebook and a data-quality-control sheet. All coding derives from PubMed titles, abstracts and metadata; full texts were not retrieved, and unreported fields are coded NR. Counts describe published studies, not AMR prevalence. No study met all seven criteria required for a comparable prevalence estimate, so the dataset supports no pooled prevalence, country comparison or prevalence map.

Files

Steps to reproduce

Five PubMed queries were run and exported in abstract format (query strings are in the Codebook sheet). Records were parsed and deduplicated on PMID; full metadata was re-fetched from PubMed by PMID via NCBI E-utilities rather than re-keyed from the exports. Each record was coded across 191 fields using documented rule-based extraction: country from stated sampling location only (never author affiliation); resistance phenotypes only where susceptibility testing was reported; resistance genes only where named; ARG–element linkage only where both appeared in one sentence. Transmission evidence was graded 0–6, with the basis recorded per study. A shared gene never raised a study above Level 2; a shared sequence type alone was not accepted as genetic relatedness. Analyses used study-level incidence to avoid pseudoreplication. Statistics: Cochran–Armitage trend tests, chi-square and Fisher exact tests with Holm adjustment, Wilson confidence intervals, Shannon and Chao2 diversity, Jaccard similarity, and bipartite network centrality. The scripts/ folder reproduces the chain from PubMed records to the database, workbook and figures.

Categories

Medical Microbiology, Bioinformatics, Antimicrobial Resistance, Escherichia coli

Licence