Background

Antimicrobial resistance (AMR) arises when microorganisms adapt to survive antimicrobial drugs through spontaneous genetic mutations, ^{Reference Friedman, Temkin and Carmeli1,Reference Jindal, Pandya and Khan2} a process exacerbated by human activities such as the misuse of antimicrobials and poor infection control practices. ^{Reference Knight, Costelloe and Deeny3,Reference Klein, Milkowska-Shibata and Tseng4} In 2019, approximately 4.95 million deaths occurred globally in direct association with AMR. ^{Reference Laxminarayan5} In Australia, AMR-associated deaths have risen to over 1,000 annually, and the 5 most common AMR pathogens were attributed to the loss of 27,705 quality-adjusted life years in 2020. ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6} The most prominent of these were methicillin-resistant Staphylococcus aureus (MRSA) infections in the respiratory tract, bloodstream, and skin, accounting for $3.5 billion of Australian healthcare expenditure. ^{Reference Cameron, Paterson and Britton7} Globally, AMR threatens the efficacy of life-saving antibiotics, fueling community-wide colonization with resistant pathogens that were once confined to hospital settings. ^{Reference DeLeo, Otto, Kreiswirth and Chambers8}

At this time, there is sparse literature that evaluates low socioeconomic status (SES) as a key driver of AMR, limiting its recognition as a potential target for policy and intervention. Numerous primary studies have examined the prevalence of resistant strains of specific pathogens in clinical settings, although current literature lacks consensus on SES as an independent contributor to AMR in high-income countries. SES is a composite of contributing determinants, and the term SES represents the factors collectively (household crowding, education level, etc) unless otherwise specified.

At the community level, low-SES neighborhoods may be disproportionately vulnerable to AMR due to reduced gut microbiome diversity linked to food insecurity. ^{Reference Eggers, Kates, Safdar, Suen and Malecki9} Determinants of SES such as household crowding (per Canadian National Occupancy Standard) increase exposure to microorganisms and facilitate resistance development. Education level directly correlates with health literacy, influencing antimicrobial use and treatment adherence. ^{Reference Coughlin, Vernon, Hatzigeorgiou and George10} Furthermore, lower SES communities have a greater reliance on Medicare and its international counterparts, which stresses stewardship and delays treatment in the community. ¹¹

In rural and remote areas, lack of stringent antimicrobial stewardship (AMS) contributes significantly to inappropriate prescription and antibiotic use, ^{Reference Yau, Thor, Tsai, Speare and Rissel12} directly increasing AMR. Additionally, crowded housing, inadequate hygiene practices, limited healthcare access, and literacy add to increased AMR in rural communities. ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6} Less studied, however, are populations of low SES in urban and suburban communities with similar AMR prevalence and related disease burdens. Local studies show direct correlations between low SES and the prevalence of specific strains of AMR pathogens, suggesting that similar social determinants drive these disparities in both rural and urban areas. ^{Reference Yau, Thor, Tsai, Speare and Rissel12,Reference Prestinaci, Pezzotti and Pantosti13}

Examining SES in specifically high-income countries is a crucial distinction when identifying global patterns in AMR. Low- and middle-income countries (LMICs) have severe limitations to antimicrobial supply and stewardship and therefore have significant challenges gathering truly representative data on resistance prevalence. ^{Reference Seale, Hutchison, Fernandes, Stoesser, Kelly and Lowe14} These nations also exhibit economic conditions not comparable to SES distributions elsewhere. Because much of their population earns less than $1 USD/day, ^{Reference Fleece, Nshama, Walongo, Kimathi, Gratz and McQuade15} socioeconomic measures from LMICs would have limited applicability to those of high-income countries.

Research question

This review investigates the association between SES and AMR rates across high-income countries based on the World Bank classification, testing the hypothesis that SES is negatively correlated with AMR prevalence. Addressing these disparities will inform policies to mitigate AMR and protect global health.

Determinants of socioeconomic status

Included studies investigated a plethora of SES determinants: income, household crowding, education, Medicaid use, insurance, employment, and census-derived deprivation scores (Table 4). Although these determinants are collectively a comprehensive measure of SES, direct comparison between studies is made difficult by their heterogeneity. Overall, 12 of 14 studies reported at least 1 significant association between a determinant of SES deprivation and increased AMR prevalence (Table 3). Of the 2 studiesthat did not find any significant positive correlations, ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6,Reference Schubert, Woodman and Mangoni25} Wozniak et al ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6} found that AMR prevalence increased with remoteness, independent of socioeconomic indices or income. Although remoteness may relate to SES, it was not considered as an independent determinant. Schubert et al ^{Reference Schubert, Woodman and Mangoni25} reported no significant increase in H. pylori positivity or any local coefficients for antibiotic resistance with increasing SEIFA Socioeconomic Index. Across the 14 included studies, 9 unique measures including and related to SES were investigated in 27 iterations (N, Table 4). Overall, 20 instances of the 27 were statistically significant correlations in support of the hypothesis (Table 4). Hence, most studies report positive correlations between SES and AMR in a majority of the various SES determinants, but further analysis is required to ascertain which determinants are the largest contributors and develop understanding of a potential dose-response relationship.

Mechanisms of association between SES determinants and AMR

Household crowding was a key determinant of SES with a demonstrated causal mechanism of low SES and increased AMR prevalence. ^{Reference Andreatos, Shehadeh, Pliakos and Mylonakis17,Reference Paumier, Asquier-Khati and Thibaut20,Reference Casey, Rudolph and Robinson26,Reference Blakiston and Freeman29} Blakiston et al ^{Reference Blakiston and Freeman29} identified increased prevalence of ESBL-E. Coli and MRSA with crowded housing, organisms that are known to readily spread within households. However, Wozniak et al ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6} challenged simplistic crowding metrics of “persons per household,” emphasizing the need to assess housing quality rather than simply the number of occupants. Community antibiotic use also played a role, with Andreatos et al ^{Reference Andreatos, Shehadeh, Pliakos and Mylonakis17} and Blakiston et al ^{Reference Blakiston and Freeman29} reporting higher prescription rates in lower SES areas, driving MRSA and ESBL-E. coli resistance. In contrast, Wozniak et al ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6} rejected antibiotic use as a sole driver, citing increased AMR prevalence despite longstanding stewardship programs in remote Australia.

Higher income appeared protective, with Chen et al ^{Reference Chen, Li, Lin, Huang and Dong18} and Gill et al ^{Reference Gill, Ma and Guo22} showing inverse relationships between affluence and MRSA infections. Similarly, McCowan et al ^{Reference McCowan, Bakhshi and McConnachie28} found more antimicrobial-resistant E. coli in Scotland’s most deprived areas but did not propose specific mechanisms. Henig et al ^{Reference Henig, Weber and Hoshen27} and Wozniak et al, ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6} however, framed AMR disparities as a function of healthcare access rather than income alone. Remoteness was a more complex variable. Univariable analysis for remoteness showed a significant association with AMR for Chua and Stewardson, ^{Reference Chua and Stewardson19} but findings were not significant in multivariate regression. Wozniak et al ^{Reference Wozniak, Cuningham, Ledingham and McCulloch6} argued that remoteness itself directly contributed to AMR spread in resource-limited settings, underscoring the need for context-specific, targeted approaches to AMR mitigation rather than one-size-fits-all interventions.

Overall, these studies suggest that no single mechanism fully explains SES-related AMR disparities. Instead, determinants such as household density, antibiotic exposure, income, and healthcare accessibility interact in complex ways, necessitating tailored interventions that consider both local socioeconomic determinants and healthcare policies.

Strengths and limitations

We observed a heterogeneity of study types and settings among included studies, from hospitalized patients to community screening participants. The included studies were well-conducted, with large sample sizes and broad geographic coverage, enhancing the external validity of the findings (Table 5). However, several limitations are present. No formal publication bias assessment was conducted, and therefore, publication bias remains a potential limitation of this review. One study used an ecological design, which is vulnerable to ecological fallacy. Although this does not invalidate the findings, it constrains their applicability to policies targeting AMR at the individual level.

Table 5. Quality assessment of included studies. The Mixed Methods Appraisal Tool Version 2018 ^{Reference Hong, Pluye and Fàbregues16} was used to evaluate the quality of each study

The wide variation in measurement of variables among studies included is the primary limitation in this review. The definition of AMR as the outcome variable varied considerably, not only in the studied strain but also in the outcome classification. Some studies defined AMR through the presence of a resistant microbial in a community screening test, while other studies were based on positive diagnostic findings of hospitalized cases of UTI or bacteremia. A standardization of SES measurements within a nation, from individual to population levels, would also facilitate research and benchmarking across health boards and between different states. We recognize that this would allow robust meta-analyses, which could not be conducted due to the heterogeneity of SES and AMR measures in the current literature.