Discussion

This population-and laboratory-based study highlights distinct profiles of CA and HCA ESBL-E UTI using individual-and community-level analyses. CA ESBL-E UTI were more common in younger patients with fewer medical conditions, many of whom lacked traditional clinical risk factors for these infections. Furthermore, CA prevalence was relatively unaffected by changes in social vulnerability. In contrast, HCA prevalence increased in communities with higher levels of household vulnerability defined by factors such as older age, disability, single-parent households, and limited English proficiency. These patterns mirror the demographic and clinical profiles of HCA cases in our data: older adults with more comorbidities who are more likely to engage with healthcare facilities, thus potentially increasing exposure risk.

We observed several demographic differences between CA and HCA ESBL-E UTI cases. Asian individuals were overrepresented among all ESBL-E UTI cases and CA in particular, accounting for over 9% of cases despite representing only 4% of the Monroe County population. Travel may partially explain this finding, as Asian individuals accounted for 25% of CA cases with documented travel; 82% (27/33) of that travel was to Southeast Asia. Travel to regions with high ESBL-E prevalence, such as Asia, South America, and the Middle East, is a known risk factor for ESBL-E acquisition. ^{Reference Tängdén, Cars, Melhus and Löwdin9} Carriage rates can reach 70% in some parts of Asia. ^{Reference Tängdén, Cars, Melhus and Löwdin9–Reference Otter, Natale and Batra11} Cultural practices and food sources may also influence ESBL-E colonization and infection risk in patients without additional healthcare risk factors. Further investigation with larger sample sizes and disaggregated data may better inform this finding. ^{Reference Islam, Khan, Kwon, Jang, Ro and Trinh-Shevrin12}

In our study, 85% of CA cases were identified and managed in the outpatient setting, a group that is less represented in the existing literature. Over two-thirds of cases lacked recognized risk factors for UTIs, and most cases did not have documented recent antibiotic exposure. Although antibiotic exposure is a known risk factor for AMR in general, studies are inconsistent for ESBL-E and are limited to specimens collected at the hospital. Leistner et al. found no association between recent antibiotic exposure and ESBL-E colonization among patients identified on hospital admission in Germany. ^{Reference Leistner, Meyer and Gastmeier13} In contrast, Goyal et al. observed a significant association between antibiotic exposure in the prior 3 months and ESBL-E growth in urine cultures at admission in a single US health system. ^{Reference Goyal, Dean and Neill14} While recent antibiotic exposure was uncommon in our CA cohort, antibiotic use may still influence community ESBL-E epidemiology through selective pressure. Inappropriate prescribing of fluoroquinolones and cephalosporins has been associated with geographic variation in ESBL-producing E. coli. ^{Reference Kubes and Fridkin15} These findings suggest that factors beyond recent individual antibiotic exposure likely contribute to CA ESBL-E UTIs.

At the community level, social vulnerability had a varying impact on CA and HCA ESBL-E UTI prevalence in our study. No association was observed between SVI and CA ESBL-E UTI prevalence, whereas HCA ESBL-E UTI prevalence increased with greater vulnerability. This was largely driven by SVI Theme 2, which comprises household composition and disability. One potential explanation is antibiotic “spillover,” the phenomenon where antibiotic use raises AMR prevalence in connected populations through transmission networks. ^{Reference Olesen, Lipsitch and Grad16} In communities with higher household vulnerability, increased healthcare contact may facilitate transmission networks among high healthcare utilizing populations such as older adults and individuals with disabilities, contributing to higher HCA ESBL-E burden in the population. In contrast, the relatively stable CA ESBL-E UTI prevalence suggests that spillover into the broader community may be relatively limited.

The relationship between social vulnerability and ESBL-E in high-income countries is unclear in the existing literature, with studies showing positive, negative, and no associations. Examples include US-and Australia-based studies that found greater community-level socioeconomic deprivation was associated with higher prevalence of ESBL-E, without epidemiologic classification. ^{Reference Brown, Henderson, Ruegsegger, Moody and van Duin17,Reference Ling, Cadavid-Restrepo, Furuya-Kanamori, Harris and Paterson18} However, another study in Australia found no significant relationship between socioeconomic disadvantage and ceftriaxone-resistant E. coli. ^{Reference Chua and Stewardson19} Interestingly, two studies in France reported an inverse correlation between community-onset ESBL-E UTIs and social deprivation. Lower ESBL-E prevalence was observed in more deprived areas, a pattern different from that commonly described for other AMR organisms. ^{Reference Paumier, Asquier-Khati and Thibaut20,Reference Larramendy, Gaultier, Fournier, Caillon, Moret and Beaudeau21}

The absence of epidemiologic classification and differences in the community health indices used may further obscure associations between social vulnerability and ESBL-E UTI risk. A US study investigating community-level drivers of composite AMR (using several organisms, including ESBL-E) found a higher prevalence in socially vulnerable neighborhoods as identified by the Area Deprivation Index, but not when using the composite SVI. These disparate results suggest that different indicators of community vulnerability may capture distinct drivers of AMR. ^{Reference Cooper, Beauchamp and Ingle22}

The inconsistent or absent association between social vulnerability and CA ESBL-E UTI prevalence reflects the multifactorial nature of ESBL-E acquisition outside the healthcare setting. Several known or suspected drivers, such as international travel and antibiotic consumption, are often more common among individuals with higher socioeconomic status. Additional non-healthcare exposures, including food sources and environmental exposures such as agriculture, livestock, and water sources, are important potential vectors and may vary across sociodemographic and economic contexts. ^{Reference Blaak, de Kruijf, Hamidjaja, van Hoek, de Roda Husman and Schets23–Reference Madec, Haenni, Nordmann and Poirel26} Household transmission from colonized family members or pets has also been described. ^{Reference Damborg, Broens and Chomel27,Reference Ludden, Cormican, Vellinga, Johnson, Austin and Morris28} Consequently, commonly used health indices, such as the SVI, may not fully capture risk, particularly in communities experiencing structural vulnerability in other ways, such as adverse environmental exposures. Future studies should consider incorporating a One Health framework and complementary measures of structural vulnerability that better account for environmental drivers.

In summary, this population-based study shows that over half of ESBL-E urinary isolates were CA, largely affecting younger females with few comorbidities and limited traditional healthcare risk factors. These findings suggest that traditional HCA risk factors alone may not adequately identify patients at risk for CA ESBL-E UTI. Although we did not identify community-level SVI variables associated with increased CA ESBL-E UTI prevalence, clinicians should still consider ESBL-E in outpatients with UTI, particularly those with individual-level risk factors such as recurrent infections or recent international travel to high-prevalence regions. Emerging predictive models that incorporate prior infections, resistance history, and demographic factors may help identify patients at higher risk earlier in their course of care. ^{Reference Shields, Cheng and Kponee-Shovein29}

This study has several limitations. It was conducted in a single county in Western NY, which may limit generalizability to rural areas and other regions outside of NY. Secondly, our surveillance relies on clinical isolates, primarily from symptomatic individuals, and may therefore not capture the full burden of ESBL-E urine colonization in the community. Third, data collection relied on EHR documentation, meaning epidemiologic classification and risk factors may be incompletely captured; notably, individual insurance status was not recorded. Furthermore, prior antibiotics were captured only in the month prior to culture, which may have underestimated this risk. Additionally, although SVI has been increasingly applied in AMR research, it was not originally designed to measure infectious disease outcomes and therefore may not fully capture the community-level factors most relevant to AMR risk.

Despite these limitations, to our knowledge, this is the only US population-based study to rigorously define CA ESBL-E UTI using specimens collected across the spectrum of care to evaluate individual-and community-level drivers of disease. Our findings highlight the distinct epidemiologic characteristics of CA ESBL-E UTIs and the lack of a clear relationship between community-level prevalence and SVI. Taken together, these findings suggest that factors driving CA ESBL-E UTIs differ from those associated with healthcare exposure. Efforts to reduce ESBL-E UTIs should consider these differences and broader community and structural factors beyond those captured by the SVI.