Antibacterial resistance (ABR) is a global health crisis, particularly threatening in low-resource settings like East Africa. Current interventions primarily focus on antibiotic use, neglecting broader factors influencing vulnerability and behavior. Multi-drug resistant (MDR) pathogens, prevalent in low- and middle-income countries (LMICs), pose a significant clinical and public health challenge. This study adopts a One Health framework, acknowledging the interconnectedness of human, animal, and environmental factors in driving ABR. It aims to investigate how diverse factors jointly contribute to the risk of MDR UTIs in East Africa, examining the interplay between environmental, social, economic, behavioral, and demographic variables. The research questions are: (1) What factors are jointly associated with a higher risk of MDR UTI? and (2) How do environmental, social, economic, behavioral, and demographic factors relate to the burden of MDR UTI in East Africa?

Existing research often highlights antibiotic use in humans and animals as a core driver of ABR and MDR. Individual antibiotic use correlates with resistant infections, and easy access to antibiotics in low-resource settings contributes to overuse and misuse. However, ABR drivers extend beyond antibiotic use to include socioeconomic status, poor sanitation, close proximity to waste, and animal contact. While some studies have explored individual risk factors like age, gender, and health status, few have holistically investigated the joint contribution of these diverse factors. Ecological studies have identified macro-level drivers such as poor WASH, socioeconomic inequality, and weak regulatory standards, but often lack the detailed individual-level data needed to understand complex interrelationships. Previous household-based studies in Tanzania and Malawi have highlighted cultural-ecological factors and advanced age as significant drivers, but relied on methods that don't capture the synergistic effects of multiple risk factors. This study aims to address these gaps by employing a study design and statistical approach that explicitly accounts for these interrelationships.

This cross-sectional study used data collected by the HATUA (Holistic Approach To Unravelling ABR) Consortium across nine sites in Kenya, Tanzania, and Uganda. The study linked environmental, social, economic, behavioral, and microbiological data at the individual level. A total of 6827 adult outpatients with UTI symptoms were initially recruited. After exclusions (detailed in Figure 5), the final analysis sample comprised 1610 individuals with confirmed UTI and complete data. Variables included contextual factors (location, healthcare system, etc.), household-level factors (WASH, livestock practices, socioeconomic status), and individual-level factors (sociodemographics, health status, antibiotic use, treatment-seeking behavior). Bivariate analysis (chi-square tests with FDR control) was used to assess individual associations between variables and MDR UTI. Bayesian profile regression was then applied to identify clusters of individuals with similar risk profiles and to assess how these profiles relate to MDR UTI risk. The primary outcome was MDR UTI, defined as resistance to at least one agent in three or more antimicrobial categories (modified to include nitrofurantoin and trimethoprim). A secondary outcome, EA ABR, was also defined based on resistance to ABs recommended in national treatment guidelines. Sensitivity analyses were conducted using subsamples of women and Gram-negative infections, and by reclassifying intermediate AST results as susceptible. The analysis was performed using R and the PReMiuM package.

The overall proportion of MDR UTI in the sample was 48%, varying across sites (Figure 1). Bivariate analysis revealed significant associations between MDR UTI and 23 of 67 variables (Supplementary Data 2 and Figure 2). Bayesian profile regression identified 24 clusters, ranging in size and MDR UTI risk (Figure S1). Figure 3 summarizes how factors jointly predict membership of the 10 high-risk clusters. High-risk clusters were more likely to include patients from Tanzania or Uganda; from households with animals, antibiotic use in livestock, and exposure to animal manure and waste; with lower education levels, poorer health, lower asset ownership, unprotected water sources, and less handwashing with soap; and with older patients, those with less education, treatment failures, and delayed treatment seeking. Individuals with all low-risk characteristics had a predicted median MDR UTI rate of 31%, versus 64% for those with all high-risk characteristics (Figure 4, Figure S3). Sensitivity analyses largely confirmed these findings. The study found that factors like recent antibiotic use were not strongly associated with higher MDR UTI risk, indicating the importance of broader social and environmental contexts.

This study demonstrates the complex interplay of over 40 environmental, social, and economic factors influencing MDR UTI risk in East Africa. The findings challenge behaviorist approaches that focus solely on antibiotic use and misuse. The study shows that antibiotic use is only one part of a larger system of vulnerabilities. Socioeconomic deprivation, coupled with environmental and behavioral factors, creates intersectional inequalities in ABR risk. Older age, lower education, poorer health, and poor WASH conditions increase both exposure to and vulnerability to MDR infections. The findings highlight that interventions to address ABR need to consider the broader social and environmental context, addressing multidimensional poverty and improving access to healthcare, sanitation, and hygiene.

This study, using linked One Health data and Bayesian profile regression, reveals the significant intersectional factors contributing to MDR UTI in East Africa. It emphasizes the need to move beyond behaviorist approaches that focus solely on antibiotic use and misuse and highlights the importance of addressing the broader social and environmental determinants of ABR. Future research should utilize longitudinal data to better understand the causal relationships between these factors and ABR and to identify effective interventions.

The study's cross-sectional design limits causal inferences. Self-reported data may introduce bias. Site-level variations, potentially reflecting unmeasured structural factors, influenced MDR rates. The COVID-19 pandemic affected recruitment, particularly in Kenya. Finally, the sample was predominantly from public outpatient settings and may not be fully representative of the general population or those with other infections.