Hospital water sources have long been associated with outbreaks due to various pathogens (1
). However, in recent years, hospital wastewater in particular has gained recognition as a reservoir and source for nosocomial infections, including multidrug-resistant Gram-negative bacteria (2
). Many studies have explored risk factors for patient acquisition of carbapenemase-producing Enterobacterales
), but risk factors for CPE establishment in hospital wastewater plumbing are less well defined. Designs that promote or disturb drain biofilm, misuse of sinks, and placement of patient care materials adjacent to sinks have all been associated with sink-related infections (6
). Factors that facilitate biofilm formation, such as nutrient exposure (7
), also plausibly increase risk of CPE establishment and persistence in the wastewater environment. Low frequency of water use and longer columns of stagnant water have also been associated with higher bacterial CFU counts in tap water (8
Exposure to colonized patients could be another important factor in CPE establishment in hospital wastewater plumbing, but current evidence supporting this is largely anecdotal. Use of sinks to dispose of patient secretions has been associated with sink colonization (9
), and environmental surface contamination from CPE-colonized patients appears to be frequent, particularly among “super spreaders” (10
). Selective pressure from antibiotic excretion in the urine and feces has been proposed as a potential contributor to the success of multidrug-resistant organisms in hospital plumbing. While studies have demonstrated higher levels of antibiotic residues and relative abundance of antimicrobial resistance genes in hospital wastewater (11–13
), studies investigating associations between antibiotic concentrations and specific resistance phenotypes have produced mixed results (14–17
Persistent low-level transmission of Klebsiella pneumoniae
carbapenemase-producing organisms (KPCOs) occurred in our institution for several years and was ultimately linked to wastewater reservoirs (18
). Detection of a wastewater source was achieved through a robust perirectal KPCO patient screening program and early adoption of the Centers for Disease Control and Prevention’s (CDC) toolkit to prevent transmission (19
), as well as the establishment of environmental sampling protocols and a database to track results. We used these resources together with clinical and patient movement data to investigate the effects of KPCO-positive patients and other clinical factors on KPCO positivity in the wastewater environment. In particular, we used whole-genome sequencing (WGS) to estimate the frequency with which KPCO-positive patients seeded the wastewater environment, and we investigated the impact of exposure to KPCO-positive patients, factors that increase KPCO shedding (e.g., antimicrobial exposure), and patient and staff behaviors that influence interactions with the plumbing on environmental KPCO positivity.
In this study, we found that exposure to KPCO-positive patients was associated with environmental KPCO positivity for a patient room overall, but only because of an effect on toilet/hopper positivity, with no evidence of effect on drain or P-trap positivity (Fig. 3A
). This is plausible, as toilets (and hoppers, which are toilet-like waste disposal units) are the elements most frequently exposed to patient fecal matter, where KPCO patient carriage is most prevalent. However, we did find examples of seeding of all tested wastewater sites from KPCO-positive patients based on genomic data. We additionally found that KPCO-positive patients seeded at least one element of the wastewater environment in at least 6% of opportunities. This is likely an underestimate of the frequency of KPCO-positive patient isolates becoming established in the wastewater environment, as we restricted our definition of seeding events to clonal identity between patient and environmental isolates. This will miss transmission due to horizontal gene transfer via plasmids and mobilization of blaKPC
between plasmids via transposition and homologous recombination, which contribute to interspecies and intergenus dissemination of blaKPC
as previously demonstrated (20
). Additionally, within phenotypically identical but genotypically mixed populations, colony picks for WGS may have missed environmental isolates that were genetically linked to patient isolates, which could limit confirmation of seeding events. We also excluded a long-standing S. marcescens
clone with frequent carriage in both patients and the environment so as to not overestimate contributions from a previously established environmental clone.
Furthermore, we found that previous positivity of a site was consistently and strongly predictive of KPCO positivity upon subsequent sampling (Fig. 3B
). This suggests that once a wastewater site is “seeded” with KPCOs, the organisms often thrive and persist. The difficulty many institutions have experienced in clearing the wastewater environment of resistant organisms supports this observation (6
). Room type, with non-ICU rooms being much less likely to be positive, was also consistently associated with environmental KPCOs, being most strongly predictive for the room, drain, and toilet/hopper. We have noted this tendency throughout our experience with environmental sampling at our institution: potential explanations for increased KPCO positivity in ICU rooms include decreased patient mobility (resulting in increased direct nursing care and contact with patient bodily fluids), higher severity of illness contributing to microbiome disruption, and higher intestinal load of resistant organisms, such as KPCOs. Of note, increased exposure to KPCO-positive patients and increased exposure to antibiotics (creating more selective pressure) do not seem to explain this room type effect, since these were both included in the multivariate models.
We found an intriguing protective effect of C. difficile
patient-days on KPCO positivity at the level of the sink drain and, to a lesser degree, the P-trap. As there is significant overlap between risk factors for C. difficile
and CPE, it seems unlikely that this is due directly to the presence of C. difficile
. We hypothesize that it is due to differences in the way sinks are used in the rooms of patients known to be positive for C. difficile
. While handwashing made up only 4% of activities in a previous observational study of behaviors around ICU sinks, it was anecdotally noted that use of the sink for hand hygiene increased markedly when a C. difficile
patient was admitted to the room (21
). Current hospital policy considers alcohol gel to be acceptable for hand hygiene for most patients, but for C. difficile
-positive patients, soap-and-water hand hygiene is required. The frequent flushing of the pipes with fresh municipal water during hand hygiene may protect against biofilm formation, which was previously demonstrated to be the route for drain colonization following P-trap colonization (7
). Tube feed days were also associated with increased KPCO positivity of the drain and, to a slightly lesser extent, the P-trap, which may be due to increased nutrient availability to support biofilm growth when nutrient-rich substances are disposed of down the sink (7
). Urinary catheter days were associated with increased KPCO positivity of the P-trap and, to a slightly lesser extent, the sink drain, and we hypothesize that this may reflect decreased patient mobility (and thus less sink usage).
While selective pressure due to antibiotics is frequently mentioned as a factor contributing to the presence of multidrug-resistant organisms in the hospital wastewater, antibiotic days were not an independent predictor in the multivariate models at any level in our study. We focused on systemic antimicrobials, many of which are excreted relatively intact in urine and thus into the wastewater. However, we may have had limited resolution, since we considered only total days of exposure to antibiotics, and different antibiotics may have various influences on KPCO survival in the environment. Additionally, a previous study demonstrated that antibiotics may accumulate in biofilm and be released over time after flushing of a wastewater siphon (13
); thus, our 7-day look-back period may not be optimal for examining the relationship between antibiotic use and resistant organisms in the environment.
The sink trap heater-vibration unit, which has been previously described (6
), was associated with increased risk for toilet/hopper positivity and decreased risk for sink drain positivity; notably, no P-traps with heaters harbored KPCOs, meaning that this factor could not be included in P-trap models. This likely reflects the nature of the device, which targeted elimination of KPCOs from the P-trap and hence could plausibly affect the associated sink drain but would not be expected to directly affect the rate of toilet/hopper positivity. Of note, the positive association between heater presence and toilet/hopper positivity likely reflects the high background positivity in the unit in which the devices were deployed; we were not able to adjust for this further since heaters were only deployed in this unit.
Our study has several limitations. Some rooms underwent repeated sampling, which we attempted to address with an analysis using a mixed-effects model with room number as a random effect; however, the large number of rooms (123) and high proportion with one or few sampling events led to issues with convergence. Thus, we used a multivariate model with previous positivity and room type, two characteristics most likely to contribute to similarity between samplings of the same room, as covariates. As noted above, 7 days may not be the optimal time frame for assessing the influence of the factors. Finally, our definition of a KPCO-positive patient (any patient with any history of a KPCO-positive culture) may have led to underestimation of the impact of KPCO-positive patient exposure, as several of the KPCO-positive patients had a remote history of KPCOs. However, this is consistent with the definition used at our institution for infection control purposes.
In conclusion, the factors that affect KPCO positivity in the hospital wastewater environment are complex and vary between specific wastewater sites; this is important for those involved in outbreak investigations to consider. KPCO-positive patients seed the wastewater environment at least 6 to 8% of the time, and sites that become positive for KPCOs are likely to be positive thereafter. Therefore, interventions that interrupt transmission to patients or are able to prevent seeding and establishment in wastewater sites may be more successful. Additionally, use of sinks for hand hygiene may be protective, whereas disposal of nutrient-rich substances down sinks may be detrimental. This work provides the basis for several potential infection control and behavioral interventions which could be deployed to reduce the risk of having detectable KPCOs in wastewater reservoirs.
This study was funded by the Centers for Disease Control and Prevention (grant BAA 200-2017-96194). S.C.P. was supported by a National Institutes of Health (NIH) Infectious Diseases Training Grant (no. 5T32AI07046-42). D.W.C., T.E.A.P., and A.S.W. are supported by the National Institutes of Health Research (NIHR) Health Protection Research Unit in Healthcare Associated Infection and Antimicrobial Resistance at the University of Oxford in partnership with Public Health England (PHE) (HPRU-2012-10041) and the NIHR Oxford Biomedical Research Centre. T.E.A.P. and A.S.W. are NIHR Senior Investigators. N.S. is funded by a University of Oxford/Public Health England Clinical Lectureship.
The views expressed in this publication are those of the authors and not necessarily those of the National Health Service, the NIHR, the Department of Health, or PHE.
The funders of the study (Centers for Disease Control and Prevention) had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.
S.M.K., K.E.B., S.D., and A.J.M. collected the data. D.W.C., K.E.B., and N.S. were involved with the sequencing of isolates. S.C.P., A.S.W., N.S., D.W.C., K.V., H.P., and A.J.M. analyzed and interpreted the data. A.J.M., A.S.W., K.E.B., and S.M.K. made substantial contributions to the conception and design of the study. S.C.P., A.J.M., A.S.W., N.S., and H.P. wrote the manuscript. All authors contributed to the revision of the manuscript. The corresponding author had full access to all data and had final responsibility for the decision to submit for publication.
A.J.M. participated in the Tango II trial with meropenem-vaborbactam and was a former consultant to the Medisans Company.