Causal-Discovery Signals Between Climatic and Socioeconomic Candidate Correlates and Antimicrobial Resistance in Escherichia coli

Özlem Baydaroğlu 

Antimicrobial resistance (AMR) is a rapidly escalating global health crisis threatening human, animal, and environmental well-being. While its spread is often linked to antibiotic use, its environmental and socioeconomic drivers remain poorly understood. This study investigates observational causal-discovery signals between AMR and climatic and socioeconomic drivers across 49 U.S. states, using annual state-level data from 2011 to 2024. Resistance data for four Escherichia coli phenotypes (carbapenem-, multidrug-, cephalosporin-, and fluoroquinolone-resistant) have been pooled as independent annual realizations, increasing the effective sample size given the limited (14-year) depth of state-level surveillance. Three complementary methods (PCMCI, CCM, and TE) have been applied in parallel, with false-discovery-rate correction and cross-method consensus used to assess robustness. Results reveal a heterogeneous landscape of directional associations, with maximum and minimum temperature, precipitation, and per capita income associated with AMR in a subset of states. After Benjamini–Hochberg correction, no state–driver association has remained significant in more than one method (FDR p<0.05: PCMCI, 7 pairs; CCM, 1; TE, 0). The single association identified by all three methods at the nominal level (maximum temperature in West Virginia) therefore represents a candidate signal for follow-up rather than a confirmed finding, reflecting the limited statistical power of short annual series. These findings nonetheless suggest that climatic and socioeconomic conditions merit consideration alongside antimicrobial-use factors, supporting phenotype-specific, regionally informed AMR surveillance within a One Health framework.