Our evaluation of 287 sink-traps across 29 UK hospitals reveals that important pathogenic species and ARGs are near-ubiquitous in UK hospital sink-traps, with substantial heterogeneity in quantitative burden. Some sink-omes consist almost entirely of major clinical pathogens, and notably, high pathogen burdens correlate with high ARG burdens. 40% of sink-traps harboured either CTX-M type ESBLs or major carbapenemases. The variability in pathogen and ARG sink-trap burden was driven to some degree by hospital, potentially related to differences in patient populations and hospital-level factors. However, most of the variability was driven by specific sink/sink-trap-related factors, similar to findings in a previous study considering isolate-level genomic variation in Enterobacterales in patients and hospital environmental reservoirs¹⁴. Exploring this further, we identified significant associations between sink-trap pathogen/ARG burden and sink-level features such as sink design, location, sink-trap chemistry and cleaning protocols, which likely influence biofilm formation, and pathogen and ARG selection and persistence. A key question is whether these factors could represent targets for interventions to mitigate the pathogen/ARG burden observed in sink-traps and thereby the risk to patients, although importantly our results reflect associations that are not necessarily causal.

The detection of at least five major clinical pathogens in sampled sink-traps strongly supports the known role of the hospital wastewater environment as a pathogen reservoir contributing to patient colonisation and infection⁵. Notably, Pseudomonas aeruginosa, Stenotrophomonas spp. and clinically-relevant Enterobacterales were found in almost all (98-100%) sinks and are common healthcare infection-associated pathogens. Pseudomonas spp. and Stenotrophomonas were frequently the most abundant species in sink-omes, in keeping with at least one other smaller study¹⁵, likely driven by the ability of these bacteria to form resilient biofilms in moist environments¹⁶. Studies focused on sequencing P. aeruginosa and drug-resistant Enterobacterales isolates to understand transmission between patients and sinks have estimated sink-to-patient acquisition accounts for 7-40% of transmissions^14,17,18 (although these were mostly asymptomatic gastrointestinal colonisation), with patient-to-sink dissemination accounting for 45% of cases¹⁴. The latter finding would be consistent with Crassvirales (a human faeces-specific marker) detection in ∼50% of sinks in this study.

Like the distribution of major clinical pathogens, nearly all sinks harboured ARGs encoding resistance to the 12 antibiotic classes considered, with strong positive associations in cross-class ARG abundance, albeit with highly variable total ARG abundance observed sink-to-sink. Interestingly, ARGs co-occurred irrespective of the dominance of major clinical pathogens in some sink-omes, suggesting they are maintained together regardless of the species-level competition shaping these communities, and consistent with the fact that many may be horizontally mobilised on mobile genetic elements. Our finding of significant associations between reduced microbiome diversity and increased ARG and major clinical pathogen burdens suggests that sink-ome dysbiosis may be both a consequence and driver of colonisation by ARG-associated clinical pathogens. These dynamics closely resemble the disruption of colonisation resistance in human gut microbiomes which can promote infections such as C. difficile and colonisation with drug-resistant Enterobacterales^19,20, and where recurrence has been effectively treated with attempts to restore a “healthy microbiome” through biotherapeutic products and microbiota transplantations²¹. Approaches aimed at positively modulating the sink-trap microbiome could serve as effective interventions for both preventing and mitigating sink-trap colonisation with antimicrobial-resistant and/or healthcare associated pathogens.

Based on our observations, sink-trap chemistry likely plays a major role in shaping pathogen and ARG burdens in sink-trap microbiomes and could represent an opportunity for intervention. Sinks positive for beta-lactam residues exhibited increased major clinical pathogen, total ARG burdens and high-risk clinical ARG burden, and likely represent a major selection pressure for drug-resistant pathogens in these reservoirs. In this study, 95/287 (33%) sink trap aspirates had at least one antibiotic class detected, similar to another study²², and 87/287 (30%) of sink-traps contained detectable beta-lactams²³. Strategies to ensure that medicines (including antibiotics) and patient waste containing antimicrobial metabolites are not disposed of via sinks should be prioritised for implementation.

Iron and alkalinity were also associated with sink-trap microbial community structure. Sink-traps with iron measurements at 5mg/L (versus 0) had higher major clinical pathogen, total ARG, and clinically-high-risk ARG burdens. Elevated iron levels can facilitate bacterial growth and biofilm formation²⁴. Iron/steel pipes in premise plumbing can leach iron to water in a range of 0.2–18 mg/L²⁵; given the long timeframe over which NHS infrastructures have been developed it is plausible that some pipework in UK hospitals contains iron. This could also represent a consideration for any future infrastructure projects. We also observed a concentration-dependent effect for alkalinity whereby high alkalinity was independently associated with higher burdens of clinically-high-risk ARGs. High alkalinity may reduce the effectiveness of cleaning approaches which are often pH-dependent such as chlorine-releasing agents (used in 75% of sinks in this study)²⁶.

Samples obtained from “dry” sink traps and without a visible pellet post-centrifugation were associated with lower pathogen, total ARG and clinically-high-risk ARG burdens; thus keeping sink-traps dry through improved drainage²⁷ or heating may represent useful interventions, consistent with studies showing reductions in Enterobacterales and P. aeruginosa colonisation following heat-based sink-trap interventions^28–30.

Cleaning audits conducted more frequently than monthly were associated with lower clinically-high-risk ARG burdens suggesting that structured cleaning regimes including frequent audits may be beneficial for mitigating sink-trap colonisation in addition to reducing surface contamination. However, we also found some evidence (p<0.01) that intensively cleaned (>1xday) sink-traps had an increased major clinical pathogen compared with those cleaned daily (Supplementary material), suggesting that excessive cleaning may paradoxically contribute to pathogen and AMR selection; this has also been shown in recent experimental studies^31–34. Capturing more detailed and informative data on specific cleaning regimes and products used was not possible in this study, but is likely to be relevant.

Several sink design features were associated with differential impacts on pathogen and ARG burden. Although these are unlikely to be directly influencing microbial communities, they may be reflecting unmeasured drivers, such as water flow (faucet shape) and temperature (separated hot/cold taps), or other unmeasured aspects of usage. For instance, tap streams directly entering drains may result in aeration of sink-trap wastewater or biofilm disruption contributing to higher major clinical pathogen burdens. Interdisciplinary innovation with engineering teams and physicists would be helpful to optimise infrastructure design to minimise biofilm formation, pathogen overgrowth and dissemination risks from sinks^35,36.

Sink-traps in staff-only medicine/drug preparation rooms exhibited high levels of clinically-high-risk ARGs. This potentially reflects medicine disposal into these sinks³⁷ selecting for antibiotic-resistant organisms¹¹, and consistent with the observation that medicine/drug preparation room sink-traps were most likely to be positive for antibiotics²³. Although patients do not have direct contact with sinks in sluice or medicine/drug preparation rooms, the potential for sink-trap biofilms to spread throughout wards via interconnected plumbing³⁶ and the risk of sink splatter contaminating staff or equipment³⁸ suggests that these sinks may still pose a transmission risk to patients. Regional differences in sink-trap total ARG burden and/or clinically high-risk ARG burden were apparent, with sink-traps in several regions in England, and sink-traps in Wales and Northern Ireland, exhibiting significantly higher ARG burdens compared to those in Scotland. For the six affected regions in England (the North-West, East of England, Yorkshire and the Humber, West Midlands, South-East, and London regions), these were the same regions with the highest AMR rates observed during national bloodstream infection surveillance in 2023-2024². Although most research on sink-trap environmental reservoirs has occurred in critical care settings and for antibiotic-resistant bacteria, the burden and contribution to transmission may be greater in non-critical care wards and relevant for susceptible healthcare-associated pathogens.

Our study has several limitations. It was undertaken during a single time window, with each sink-trap evaluated only at a single timepoint. Other studies have demonstrated that microbial composition can vary markedly over time, and that sinks within wards have more similar taxonomic profiles than those between wards suggesting distinct ward-based wastewater ecologies⁶. Our breadth of sampling across multiple hospitals UK-wide was a trade-off for investigating ward-level sink diversity in more limited detail. Sink-trap pathogen and ARG epidemiology may be different in other settings, as these are influenced by sink-trap specific factors. There are many different approaches to taxonomic and ARG profiling in metagenomes and it remains unclear which is the most accurate approach; however, importantly we applied a consistent approach within-study enabling relevant comparison. We were unable to evaluate the relationship between pathogen/ARG abundance in sink-trap microbiomes and patient colonisation burden or clinical infection-associated data in populations co-located with the sampled sinks, and the threshold at which sink-drain colonisation becomes a risk to patients remains unexplored - this is important future work. Similarly, we were unable to assess whether ward-level wastewater could be used to monitor pathogen/AMR burden in ward inpatients.

Overall, however, our findings have significant implications for optimising the design, cleaning, usage and monitoring of hospital sinks to ensure that selection pressures for pathogens and ARGs can be avoided, minimising the risk to patients. Our findings are also relevant to sinks in community settings where pathogen/ARG burdens may be even higher³⁹, potentially contributing to community-acquired, hospital-onset infections. Our findings of sink-trap microbiome dysbiosis and associations with pathogen/ARG abundance warrant future research into what sink-ome features contribute to colonisation resistance against drug-resistant pathogens and how this might inform interventions. In the absence of removing sinks from healthcare completely, interventional studies to characterise the best strategies to manage them and minimise any negative impact on patients are urgently needed.