REVIEW
published: 01 November 2016
doi: 10.3389/fmicb.2016.01728
Edited by:
Jose L. Martinez,
Spanish National Research Council,
Spain
Reviewed by:
Jose Luis Balcazar,
Catalan Institute for Water Research,
Spain
Celia Manaia,
Catholic University of Portugal,
Portugal
*Correspondence:
Andrew C. Singer
acsi@ceh.ac.uk
Specialty section:
This article was submitted to
Antimicrobials, Resistance
and Chemotherapy,
a section of the journal
Frontiers in Microbiology
Received: 12 July 2016
Accepted: 17 October 2016
Published: 01 November 2016
Citation:
Singer AC, Shaw H, Rhodes V and
Hart A (2016) Review of Antimicrobial
Resistance in the Environment and Its
Relevance to Environmental
Regulators. Front. Microbiol. 7:1728.
doi: 10.3389/fmicb.2016.01728
Review of Antimicrobial Resistance
in the Environment and Its Relevance
to Environmental Regulators
Andrew C. Singer
1
*
, Helen Shaw
2
, Vicki Rhodes
3
and Alwyn Hart
3
1
NERC Centre for Ecology & Hydrology, Wallingford, UK,
2
Department for Environment, Food and Rural Affairs, London,
UK,
3
Environment Agency, Bristol, UK
The environment is increasingly being recognized for the role it might play in the global
spread of clinically relevant antibiotic resistance. Environmental regulators monitor and
control many of the pathways responsible for the release of resistance-driving chemicals
into the environment (e.g., antimicrobials, metals, and biocides). Hence, environmental
regulators should be contributing significantly to the development of global and national
antimicrobial resistance (AMR) action plans. It is argued that the lack of environment-
facing mitigation actions included in existing AMR action plans is likely a function of our
poor fundamental understanding of many of the key issues. Here, we aim to present the
problem with AMR in the environment through the lens of an environmental regulator,
using the Environment Agency (England’s regulator) as an example from which parallels
can be drawn globally. The issues that are pertinent to environmental regulators are
drawn out to answer: What are the drivers and pathways of AMR? How do these
relate to the normal work, powers and duties of environmental regulators? What are the
knowledge gaps that hinder the delivery of environmental protection from AMR? We offer
several thought experiments for how different mitigation strategies might proceed. We
conclude that: (1) AMR Action Plans do not tackle all the potentially relevant pathways
and drivers of AMR in the environment; and (2) AMR Action Plans are deficient partly
because the science to inform policy is lacking and this needs to be addressed.
Keywords: antimicrobial resistance, AMR, antibiotics, metals, biocide, plasmid, genes
INTRODUCTION
Many of the hurdles to combating antibiotic-resistant infections in the clinic are well understood
and, as such, have been used to inform existing antimicrobial resistance (AMR) Action Plans
(
European Commission, 2011; Department of Health/Defra, 2013; World Health Organisation,
2015). It is argued, that our inability to satisfactorily answer fundamental questions about AMR in
the environment is responsible for the lack of any signific ant environmental focus in existing AMR
Action Plans and the O’Neill AMR Reviews (O’Neill and The Review on Antimicrobial Resistance,
2016). It is further argued that without inclusion or consideration of all the drivers and pathways
of AMR into the environment, AMR Action Plans are incomplete and at risk of not achieving the
desired goals of ensuring and improving the efficacy of existing and future antibiotics.
Frontiers in Microbiology | www.frontiersin.org 1 November 2016 | Volume 7 | Article 1728
Singer et al. Relevance of AMR to Regulators
In this review, we aim to present the AMR challenge through
the lens of an environmental regulator, using the Environment
Agency (England’s regulator) as an example from which parallels
can be drawn globally. We argue that there is an evidence
gap that hinders the ability of policymakers and environmental
regulators from delivering environmental prote ction from AMR.
The following five questions exemplify the approach taken by
an environmental regulator when tackling the AMR challenge.
The inability to answer these questions, in this point in time,
highlights some of the evidence gaps that need to be prioritized
to facilit ate a holistic, evidence-based AMR Action Plan:
(1) What are the benefits of controlling AMR in the
environment over and above mitigating the potential
transmission of resistance to humans?
(2) What is t he relative contribution from the release of
antibiotics, metals, biocides, and antibiotic resistance
genes (ARGs) to the emergence, maintenance and spread
of AMR in the environment?
(3) How well do current technologies and approaches limit
AMR in the environment?
(4) If reducing AMR needs to be reflected within our
regulatory framework, how could this best be tackled?
(5) Is there evidence to suggest that expansion of existing
regulation/control me asures for pollutants to AMR will
translate into a decline in AMR in the environment?
Here, we review the pertinent issues that lie at the root of
the aforementioned questions, such as: What are the drivers and
pathways of AMR? And, How do these relate to the normal work,
powers and duties of environmental regulators? The discussion
is then turned to: What are the knowledge gaps that hinder the
delivery of environmental protection from AMR? Finally, we
offer several thought experiments for how different mitigation
strategies might proceed in the light of a holistic understanding
of AMR drivers and pathways. It is the intention that this review
would be a catalyst for future discussions among scientists,
policymakers, clinicians, veterinarians, and regulators. It is also
our expe ctation that this review will stimulate a more systematic
review and meta-analysis of the literature to further examine
and critique the evidence base and the changing state of our
knowledge gaps.
BACKGROUND OF GLOBAL, REGIONAL
AND NATIONAL AMR ACTION PLANS
The World Health Organisation (WHO) and its Global
Action Plan broadly outlines five strategic objectives to
tackle AMR: (1) to improve awareness and understanding
of AMR; (2) to strengthen knowledge through surveillance
and research; (3) to reduce the incidence of infection; (4)
to optimize the use of antimicrobial agents; and (5) to
ensure sustainable investment in countering AMR (
World
Health Organisation, 2015). The WHO highlights the role
of the environment in Objective 4 of the Action Plan:
“[To] Develop standards and guidance...for the presence of
antimicrobial agents and their residues in the environment,
especially in water, wastewater and food (including aquatic
and terrestrial animal feed).” A similar action plan was
published by the European Commission (2011). However, the
only mention of the role of the environment was in ‘Action
Number 8,’ which details the need to “Initiate cooperation on
reduction of the environmental pollution
by antimicrobial
medicines particularly from production f acilities.” The
highlighting of pharmaceutical manufacturing is echoed in
the O’Neill AMR Review, discussed below. Although rivers that
receive effluent from drug manufacturers have been shown to be
a very important local issue with global implications (Larsson,
2014), it will be shown in subsequent sections to be only one
pathway among several (see Figure 1). Moreover, it will be shown
that antibiotics are only one of the many drivers of AMR in the
environment.
The final Review on Antimicrobial Resistance entitled
“Tackling drug-resistant infections globally: Final report and
recommendations” (
O’Neill and The Review on Antimicrobial
Resistance, 2016) hig hlighted the need to reduce environmental
pollution of antibiotics in much the same way as have done
the WHO and EC AMR Action Plans. In brief, it highlig hts
three pathways: (1) animal waste, (2) human waste, and (3)
manufacturing waste. In every case the focus was on antibiotics,
without any mention of other resistance-driving chemicals. It also
prioritizes two pathways: hospital effluent and pharmaceutical
manufacturing plants. It is difficult to argue against improving
these two pathways, as they are both highly relevant at a
spatially resolved scale of analysis, however, these inputs are
relatively small in number as compared to the overwhelming
input of all resistance-driving chemicals from all the sources
(see Figure 1). Moreover, it is unproven that WWTPs receiving
effluent from hospitals discharge significantly more resistance-
driving chemicals into the environment, with me asurably greater
impact, than other similarly sized WWTPs. Although hospitals
are an easy t arget and in many ways a tractable target, it does
not necessarily make them the highest priority for reducing
the prevalence of AMR in the environment. It is argued that
the vision for these action plans is not sufficiently holistic
to safeguard our natural environment. To appreciate these
critiques, one must first understand what drives AMR in the
environment.
FUNDAMENTAL QUESTIONS OF AMR IN
THE ENVIRONMENT: WHAT ARE THE
DRIVERS AND PATHWAYS OF AMR IN
THE ENVIRONMENT?
Broadly speaking, antibiotic use in humans and animals
carries an inherent risk of selecting for ARGs. These ARGs
are often, if not always, found in bacteria with other genes
promoting resistance to other potentially harmful chemicals
(
Alekshun and Levy, 1999; Levy, 2002; Enault et al., 2016;
Subirats et al., 2016). It is most helpful to see these genes
as ‘resistance genes’ (not ARGs), because antibiotics are
not the only chemicals t hat select for resistance genes.
Frontiers in Microbiology | www.frontiersin.org 2 November 2016 | Volume 7 | Article 1728
Singer et al. Relevance of AMR to Regulators
FIGURE 1 | Schematic of the hot-spots and drivers of antimicrobial resistance (AMR). The environmental compartments that are currently monitored or
regulated by the Environment Agency (EA; England) are denoted by an asterisk in red. WFD, Water Framework Directive.
It is an unfortunate historical convention to label all these
genes antimicrobial resistance genes (ARGs), when in
fact they can potentially confer resistance to many more
chemicals.
A review of the literature highlights three well-characterized
classes of resistance-driving chemicals: (1) antimicrobials,
of which there are four subclasses, antibiotics, antifungals,
antivirals, and antiparasitics; (2) heavy metals; (3) biocides (i.e.,
disinfectants and surfactants). However, many other chemicals,
natural [e.g., plant-derived (
Friedman, 2015)], and xenobiotic
[e.g., solvents such as octanol, hexane and toluene (Ramos et al.,
2002; Fernandes, 2003)], are also known to select for resistance
genes. The prevalence of resistance genes in the environment
are the result of a complex combination of factors, that reflect a
dynamic b alance of fitness costs and benefits: costs of carrying
the ARG in the context of the host genome and environment
(
Maher et al., 2012; Roux et al., 2015); relative to the severity
and frequency of hazard (Gullberg et al., 2011, 2014); relative
to some physical environmental factors, such as temperature
(Gifford et al., 2016) and microbial ecology (Amini et al., 2011),
among others.
Co-selection of genes that confer resistance to chemical
hazards [solvents (
Fernandes, 2003), biocides (Pal et al., 2015;
Wales and Davies, 2015; Wesgate et al., 2016), antibiotics (Levy,
2002; Blanco et al., 2016), and metals (Percival et al., 2005;
Pal et al., 2015)] is a potentially ecologically and clinically
important phenomenon. Co-selection is achieved in two ways:
(1) co-resistance, whereby selection for one gene fosters the
maintenance of another resistance gene, one that does not
necessarily offer a sele ctive advantage to the chemical in question
(
Johnson et al., 2016); and (2) cross-resistance, whereby one
resistance gene can offer protection from multiple toxic chemicals
(
Curiao et al., 2016). Co-resistance is analogous to bringing a
toolbox to a worksite; one might only need one or two tools from
the toolbox at any one time, but there are many tools ‘fortuitously’
available for use should the need arise. The genomic architecture
(e.g., ARGs found within: plasmids, transposons, and integrons)
is the toolbox and the resistance genes that are co-located in the
genome are the tools. Cross-resistance is analogous to having a
tool t hat is capable of multiple functions, such as a claw hammer,
which can hammer in as well as extract a nail, thereby offering
multiple functions from the same tool. Efflux pumps c an often
provide cross-resistance to multiple chemicals.
The literature describes three major pathways for resistance-
driving chemicals into the environment (see Figure 1):
(1) Municipal and industrial wastewater;
(2) Land spreading of animal manure and sewage sludge;
and
(3) Aquaculture.
Additional pathways will be discussed (e.g., aerosols and
mining), howe ver, they were not included in Figure 1 as it was
felt that the evidence base for their importance needs further
strengthening.
DRIVERS OF RESISTANCE: ANTIBIOTICS
Scale of Human Antibiotic Use
In 2013, the total measured consumption of antibiotics in
England was 27.4 DDD per 1000 inhabitants per day [general
practice 79%, hospital 15% and other community consumption
Frontiers in Microbiology | www.frontiersin.org 3 November 2016 | Volume 7 | Article 1728
Singer et al. Relevance of AMR to Regulators
(predominantly dentists) 6%], in line with the median across
Europe in 2011 of 21.3 median DDD per 1000 inhabitants per day
(Public Health England, 2014), where the DDD is the assumed
average maintenance dose per day for a drug used for its main
indication in adults. A total of 66 different antibiotics were
prescribed, with the top 15 antibiotics in general practice and
hospitals accounting for 98 and 88% of consumption, respectively
(
Public He alth England, 2014).
In 2010, India was the largest consumer of antibiotics when
assessing total tonnage, however, t heir per capita usage (7.5 units
per capita), was low by comparison to Australia and New Zealand
which recorded among the highest usage rates of 87 and 70 units
per capita, respectively (
Van Boeckel et al., 2014). An antibiotic
unit was defined in this study as the number of doses (i.e.,
pill, capsule, or ampoule) sold (Van Boeckel et al., 2014). China
was the second largest consumer of antibiotics, globally when
assessing total tonnage, but similar to India, recorded relatively
low antibiotic usage per capita (7.5 units).
Heterogeneity in antibiotic use is reproduced at seemingly
every geographical scale. For example, nested within the
larger global differences are differences between countries
within smaller regions, e.g., Europe, where nearly 300% more
antibiotics were used per capita in Turkey than Armenia
(
Versporten et al., 2014). Within Europe, the highest defined
daily dose (DDD) of antibiotics (outpatient) per capita is
France (32 DDD) and Greece (30 DDD), with the UK (15
DDD) exceeding that of the lowest use European country,
The Netherlands, by 33% (
Goossens et al., 2005). Differences
between countries is often attributed to the ease at which
one can self-medicate, with eastern and southern Europe an
countries (e.g., Bulgaria, Cyprus, Greece, Lithuania, Romania,
and Spain), demonstrating greater access to antibiotics without
a prescription, thereby increasing use and misuse (
Safrany and
Monnet, 2012). In general, per capita antibiotic use across
Europe is lowest in northern regions, moderate in eastern
regions and highest in southern regions (
Goossens et al.,
2005).
In addition, to there being large differences between countries
within regions, there are significant differences in antibiotic
use within countries. The highest combined general practice
and hospital antibiotic consumption was in Merseyside, with
similar levels reported in Southern Europe with 30.4 DDD
per 1,000 inhabitants per day, over 30% higher than Thames
Valley with the lowest consumption, (22.8 DDD per 1,000
inhabitants per day) (
Public Health England, 2014). Durham,
an area in the northeast of England reported 40% more
antibiotic prescriptions in general practice than in London
(southeast England) (
Public Health England, 2014). The
ESPAUR Report 2014 notes that areas within the UK that
report high antibiotic prescription rates also reported higher
antibiotic resistance rates; a finding the authors acknowledged
necessitates further examination. It remains unexplored as to
whether these regional variations in antibiotic prescriptions
translate into differences in risk to antibiotic resistance
selection and maintenance in the environment. The evidence
of heterogeneous antibiotic use across large and small scales
likely has multiple causes and as such, AMR Action Plans
need to be sufficiently holistic to address the diversity of
aetiologies.
Scale of Animal Antibiotic Use
According to The State of the World’s Antibiotics 2015, two-
thirds of all of the antibiotics produced globally each year
(65,000 tons of 100,000 tons) are used in animal husbandry
(
Gelband et al., 2015). The variability between countries in
veterinary antimicrobial use in food-producing animals just
within the hig h income countries can be significant. Sales of
veterinary antimicrobial agents among the 26 EU countries in
2013 expressed as mg of antibiotic per population correction
unit (PCU) places Norway as having the lowest mg/PCU at:
3.7 mg/PCU, with the UK at 62.1 mg/PCU. PCU is a standard
unit of measure that takes into account t he number of animals
in a country and their average weight at the point t hey are
most likely to be treated, providing an estimate of total kg
of food producing animal in a country. At the upper end of
veterinary antimicrobial use in the EU, was Italy (301.6 mg/PCU),
Spain (317.1 mg/PCU), and Cyprus (425.8 mg/PCU) (
European
Medicines Agency, 2015). The majority of these sales are for
tetracycline and penicillin antibiotic classes, composing between
6–56 and 11–61% of the total antibiotics sold in each country
for food-producing animals, respectively (European Medicines
Agency, 2015). In combination with sulfonamides, these three
classes of antibiotic account for 71% of the total sales in
these 26 European Union (EU)/European Economic Area (EEA)
countries in 2013.
In 2013, the total antibiotic sales for therapeutic use in
animals, was 420 tons, approximately 44% of the total antibiotic
usage in the UK (
Public Health England, 2015). The number of
livestock in the UK in 2013 was: cattle (9.8 m), pigs (4.9 m), sheep
(32.8 m), chicken (162.6 m) (
Veterinary Medicines Directorate,
2014). The milligrams of active ingredient of critically important
antibiotics (CIA) sold for food producing animals per PCU for
2013 was: cattle (8 mg/pcu), pigs and poultry (172 mg/pcu)
(
Veterinary Medicines Dire ctorate, 2014). Tetracyclines were the
most frequently used antibiotics in animals (43.5%), followed
by penicillins (21.7%), mirroring the use patterns in humans
where it was reversed, penicillins (64%) followed by tetracyclines
(10%) (Public Healt h England, 2015). The estimate that roughly
44% of the antibiotics used in the UK are for veterinary use,
is much lower t han estimates for other countries, such as the
U.S., where >70% of antibiotics are estimated to be used in
livestock (O’Neill and The Review on Antimicrobial Resistance,
2015
).
Antibiotics are important for maintaining animal health and
welfare. Antibiotics are often dispensed to treat or prevent
infections in herds/flocks (
Gelband et al., 2015). Antibiotics
are often added to the animal’s water or food as a pragmatic
solution to the fact that they are reared in groups/flocks making
it difficult to isolate and treat only the infected. In addition, t he
effort to isolate animals could be stressful to the animal and
sometimes dangerous to the veterinarian who administers the
antibiotic.
The use of antibiotics for growth promotion has been
banned in Europe and is in decline in North America. Major
Frontiers in Microbiology | www.frontiersin.org 4 November 2016 | Volume 7 | Article 1728
Singer et al. Relevance of AMR to Regulators
food suppliers such as Tyson (Tyson Fresh Meat Inc., 2014),
McDonalds (McDonalds Corporation, 2015), Chick-fil-A (Chick-
fil-A, Inc., 2014), Subway (Subway Ip, Inc., 2016), and Taco Bell
(Taco Bell Corp, 2016) are slowly phasing in new commitments
to greatly restrict antibiotic use in food production, largely in
poultry.
Relevant Pathways for Antibiotics
Municipal and Industrial Wastewater
A large fraction of the antibiotics consumed by humans are
excreted in the urine and feces in t heir biologically active form
(
Singer et al., 2011; Zhang et al., 2015; Verlicchi and Zambello,
2016). The antibiotic s excreted by humans will enter WWTPs,
with one of three fates: (1) biodegradation (Chen et al., 2015);
(2) absorption to sewage sludge (Li and Zhang, 2010; Ahmed
et al., 2015); or (3) exit in the effluent unchanged (Rivera-Utrilla
et al., 2013; Luo et al., 2014). Biologically active metabolites of
the antibiotic which can be generated in the wastewater and in
the wider environment are not being considered in this review,
though they are potentially ecologically relevant (
Gros et al.,
2013
).
The persistence of an antibiotic in a WWTP is a function of:
(1) influent composition (industry, municipal) (Gardner et al.,
2013; Larsson, 2014); (2) salinity; (3) temperature; (4) nature of
WWTP (e.g., trickling bed filter, activated sludge, and membrane
bioreactor); and (5) hydraulic retention time. A survey of 16
UK WWTPs revealed the presence of erythromycin, ofloxacin,
and oxytetracycline (the only three for which they authors
assayed) in e ach of the WWTPs (Gardner et al., 2013). The
median concentration of erythromycin was 2.0 µg/L ± 0.8
standard deviation [(SD; 42% coefficient of variation (CoV)]. The
median concentration of ofloxacin had was 0.18 µg/L ± 0.33 SD
(178% CoV), while oxytetracycline had a median concentration
of 3.6 ± 2.5 SD (70% CoI) (
Gardner et al., 2013). The CoV
for these antibiotics, ranged from 42 to 178% reinforcing the
fact that sewage influent composition has a high degree of
variability. The affinity of oxtetracycline to bind to sludge led
to relatively high concentrations in the sludge as compared to
other pharmaceuticals in the influent (1.15–43 mg/kg) (
Gardner
et al., 2013). The preferential removal of some antibiotics
[e.g., sulfonamides (
Yang et al., 2012), ciprofloxacin (Polesel
et al., 2015)] into the sludge suggests that the risks from
sludge applic ation to land are likely to be different from
the risks from discharging sewage effluent into rivers. The
authors also reported enhanced removal of pharmaceuticals
where there was iron dosing at the primary treatment stage
of a trickling filter WWTP (Gardner et al., 2013). Specifically,
they reported an average loss of erythromycin, ofloxacin and
oxytetracycline of 20, 74, and 51%, respectively, after iron-dosing
(used to remove phosphorus), as compared to undosed primary
treatment of −11, 19, and −4%, respectively. Concentrations
of oxytetracycline within a river catchment receiving WWTP
effluent varied 14-fold across nine WWTP sampling days, from
9.4 to 137 g/d (normalized to flow) (
Comber et al., 2015),
illustrating the variability in antibiotic consumption and the
extent to which environmental exposures to antibiotics can
fluctuate.
Greywater, Reclaimed and Black Water
The global water footprint in the period 1996–2005 was
9087 Gm
3
year
−1
, of which 15% was categorized as gray water
(
UNESCO-IHE, 2011). Greywater is defined as water originating
from the mains potable water supply that has been used for
bathing or washing dishes or laundering clothes, excluding toilet
water (Finley et al., 2009). Reclaimed water is typically water that
originates from WWTP effluent that has undergone additional
treatment to ensure its safe use in a variety of applications,
including irrigation and toilet flushing (
Kinney et al., 2006).
Blackwater is recycled, treated sewage effluent (Otterpohl et al.,
2004). The use of reclaimed water for irrigation purposes would
be subject to Discharge Consent by the Environment Agency
and would need to comply with British Standards (e.g., BS
8595:2013 Code of practice for the sele ction of water reuse
systems; BS 8525-1:2010 and BS 8525-2:2011 Greywater systems.
Domestic greywater treatment equipment. Requirements and
test methods). While sewage wastewater in the U.S. must meet
treatment guidelines set by individual states before used for
irrigation, regulations do not require wastewaters originating
from animal feeding lots to be treated before land application
(Dodgen and Zheng, 2016).
Although the use of reclaimed water is relatively new in
the UK, it’s a practice that is well-established in hot, dry
climates where water pressures have been more acutely sustained
(
Hamilton et al., 2005). The prospect of less reliable water sources
into the future, as a result of a changing climate, might make this
source of water more important in the UK (Prudhomme et al.,
2012). Reclaimed water is currently used to sprinkler irrigate
crops (e.g., lettuce, carrots, and green beans), golf courses, and
landscapes (Kinney et al., 2006; Calderón-Preciado et al., 2013;
Thanner et al., 2016), and as such is being introduced into soil
habitats that might have pre viously been unexposed to significant
quantities of ARGs or resistance-driving chemicals (
Fahrenfeld
et al., 2013; Han et al., 2016). The amplific ation of antibiotic
resistant bacteria within distribution system for reclaimed water
is poorly understood, but poses a potential risk to humans
and the environment (Fahrenfeld et al., 2013). Both treated
sewage wastewater and animal feeding operations wastewater
can have high levels of dissolved organic matter and nutrients
as well as a highly variable microbial community activity and
composition (Dodgen and Zheng, 2016). The interaction between
the microbial community, the antibiotics and the dissolved
organic matter can greatly impact the fate of the antibiotics in the
soil environment through a combination of biodegradation and
adsorption (Dodgen and Zheng, 2016).
Veterinary and Livestock
As in humans, when animals consume antibiotics as much as
30 to 90% is released into the manure and urine (
Sarmah
et al., 2006
; Berendsen et al., 2015). Animal excreta has been
shown to contaminate the environment with antibiotic resistant
bacteria and antibiotics (Udikovic-Kolic et al., 2014; Wichmann
et al., 2014). This phenomenon was recently demonstrated in a
survey of feces from 20 commercial swine and 20 cattle farms
in The Netherlands. The study reported antibiotics in 55% of
the swine feces from 80% of the swine farms and 75% of the
Frontiers in Microbiology | www.frontiersin.org 5 November 2016 | Volume 7 | Article 1728