https://orcid.org/0000-0003-1333-5603
Abstract
To characterize and compare resistance trends in clinical Escherichia coli isolates from humans, food-producing animals (poultry, cattle and swine) and pets (dogs and cats).
Antibiogram results collected between January 2014 and December 2017 by MedQual [the French surveillance network for antimicrobial resistance (AMR) in bacteria isolated from the community] and RESAPATH (the French surveillance network for AMR in bacteria from diseased animals) were analysed, focusing on resistance to antibiotics of common interest to human and veterinary medicine. Resistance dynamics were investigated using generalized additive models.
In total, 743 637 antibiograms from humans, 48 170 from food-producing animals and 7750 from pets were analysed. For each antibiotic investigated, the resistance proportions of isolates collected from humans were of the same order of magnitude as those from food-producing animals or pets. However, resistance trends in humans differed from those observed in pets and food-producing animals over the period studied. For example, resistance to third-generation cephalosporins and fluoroquinolones was almost always below 10% for both humans and animals. However, in contrast to the notable decreases in resistance observed in both food-producing animals and pets, resistance in humans decreased only slightly.
Despite several potential biases in the data, the resistance trends remain meaningful. The strength of the parallel is based on similar data collection in humans and animals and on a similar statistical methodology. Resistance dynamics seemed specific to each species, reflecting different antibiotic-use practices. These results advocate applying the efforts already being made to reduce antibiotic use to all sectors and all species, both in human and veterinary medicine.
Introduction
Based on their shared bacterial communities and on the evidence of point-in-time transfer of resistance determinants between bacteria carried by humans and animals, a relationship is assumed to exist between antimicrobial resistance (AMR) in bacteria isolated from humans and animals. However, does the development of AMR in bacteria from humans and animals necessarily go hand in hand?
To date, the dynamics of AMR in humans and animals have very rarely been analysed together or directly compared.1 Given the global challenge posed by increasing resistance, it seems essential to carry out such a joint analysis in order to guide control strategies. This comparison is also part of the integrative approach advocated by the WHO/Food and Agriculture Organization of the United Nations (FAO)/World Organisation for Animal Health (OIE) tripartite alliance to fight AMR in a global and transversal way, in keeping with the ‘One Health’ concept.2 Comparing AMR described by different studies concerning humans and animals remains difficult; differences in the design of isolate collection methods, the bacterial population considered and the analyses carried out can all limit the relevance of comparisons made. Drawing parallels requires, as a minimum, the use of data collected in similar ways, at similar times and analysed using a similar methodology.3
Escherichia coli is the most common bacterial species isolated in the human community and the most frequent bacterium isolated in pathological contexts from both food-producing animals and pets. E. coli is also considered to be an excellent sentinel of AMR for a wide range of species4,5 and is therefore a good candidate for comparing resistance trends between humans and animals. In addition, most antibiotic classes used to treat colibacillosis are shared between veterinary medicine and human medicine in the community, regardless of whether these are first-generation antibiotics or critically important antibiotics (CIAs).6,7
In order to document the assumed relationship between AMR dynamics in human and animal health, this study was designed to characterize and compare the resistance proportions and trends in clinical E. coli isolates from humans, food-producing animals and pets.
Methods
Following the ‘One Health’ concept,2 the first step was to determine which data could be relevant for drawing parallels, then how to conduct the analysis in a way that provided meaningful comparisons. Both MedQual8 (the French surveillance network for AMR in bacteria isolated from the community) and RESAPATH9 (the French surveillance network for AMR in bacteria from diseased animals) collect antibiogram results in a pathological context from their member laboratories. The laboratories participating in MedQual comply with European recommendations regarding antibiotic susceptibility testing [Antibiogram Committee of the French Society of Microbiology (CA-SFM) and EUCAST].10 RESAPATH laboratories also use standard methods, as described in the veterinary section of CA-SFM.11
Antibiogram results involving E. coli isolates collected by the two networks between January 2014 and December 2017 were analysed, focusing on resistance to five antibiotics of interest to both human and veterinary medicine according to their spectrum of activity, their use in the treatment of colibacillosis and their influence on public health (Table S1, available as Supplementary data at JAC Online). For each species, the most frequent pathological context and production type involving E. coli were selected for analyses (Table S2). This split into subpopulations was considered the best choice in order to describe meaningful trends, instead of grouping together E. coli isolates from different production types or pathological contexts with different antibiotic uses and recourse to antimicrobial susceptibility testing.12
From an epidemiological point of view, the event of interest is the non-susceptibility to a particular antibiotic, indicating that the isolate is no longer a WT strain. Therefore, intermediate isolates were grouped together with resistant isolates. At first, proportions of resistance were calculated over the whole 2014–17 period for each species and each antibiotic.13 The second step was to implement generalized additive models (GAMs),14 which had previously been used to describe resistance evolution in isolates from animals.3,15 Including at least 25 strains per time step,13 the proportion of the monthly (bimonthly for isolates from cats) number of resistant strains out of the total number of strains collected monthly (or bimonthly) was modelled using binomial regression. The analysis was carried out as described by Boireau et al.13 The models included non-parametric smooth functions of calendar time designed to control for trend and seasonality. Smoothing parameters were estimated using cross-validation. We considered a P value of ≤0.05 as a statistically significant difference. If trend variations were not significant, the trend was stationary.
Results and discussion
Based on graphical analyses, no seasonal cycle was observed in resistance evolution. Even if the resistance levels in E. coli isolates from humans were of the same order of magnitude as the resistance proportions in animals (Table S2), the resistance trends evolved in a way that was specific to each species (Figures 1 and 2).
Trends of resistance to first-generation antibiotics in E. coli isolates from humans, dogs, cats, piglets, broilers, hens, turkeys and calves between 2014 and 2017, based on MedQual and RESAPATH data. Dates are given as year-month-day. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Trends of resistance to CIAs in E. coli isolates from humans, dogs, cats, piglets, broilers, hens, turkeys and calves between 2014 and 2017, based on MedQual and RESAPATH data. Dates are given as year-month-day. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
For most of the species considered, the trends in resistance to first-generation antibiotics (here amoxicillin, trimethoprim/sulfamethoxazole and gentamicin) remained stationary or decreased (Figure 1). Resistance to CIAs during the 2014–17 period was almost always below 10% for both humans and animals (Figure 2). For third-generation cephalosporins (3GCs), the resistance trend in dogs remained stationary at 6.4% for the whole period, while most resistance trends in animals were either stationary below 3% or decreased sharply. In parallel, the proportions of resistance among isolates from humans decreased slightly from 4.2% (95% CI=3.8%–4.7%) in January 2014 to 3.1% (2.9%–3.4%) in December 2017. For fluoroquinolones, most resistance trends in animals decreased sharply over the period, except for broilers, whereas the resistance proportions in humans oscillated, but slightly decreased overall from 10.5% (10.0%–11.1%) in January 2014 to 8.9% (8.6%–9.2%) in December 2017.
CIAs were subject to several communication campaigns, guidelines and measures between 2014 and 2017 in France, both in human and veterinary medicine. In humans, for example, the fosfomycin/trometamol combination has been recommended since 2014 as the first-line treatment for simple cystitis and pivmecillinam as a second-line therapy.16 Fluoroquinolones can be used for cases of complicated cystitis, but only as a third-line treatment or based on antibiogram results.16 The use of fluoroquinolones in the community decreased between 2014 and 2016 from 1.74 to 1.51 daily doses/1000 inhabitants/day,17 but this variation did not noticeably influence the resistance trend in humans (Figure 2b). The stable use of 3GCs in the community over the 2014–16 period17 may partially explain the slight decrease in 3GC resistance proportions (Figure 2a). The proportion of ESBL-producing E. coli in human community urinary tract infections was 0.3% in 1999, but had increased to 3.3% by 2013.18,19 However, in our study, the proportions of 3GC-resistant E. coli, which can be considered as an overestimation of the proportion of ESBL-producing E. coli, remained below 5% throughout the whole period. It can therefore be hypothesized that ESBL-producing E. coli are still limited in the community and that the trend seen prior to 2013, whereby these E. coli proportions were increasing, had by the time of our study petered out.
At the same time, in veterinary medicine various ministerial orders and decrees were published to restrict the use of CIAs as a first-line treatment.3 According to the annual sales of antibiotics in veterinary medicine, all production types and ages considered, exposure to last-generation cephalosporins decreased by 94.9% for cattle, 93.7% for swine and 65.5% for pets, while exposure to fluoroquinolones dropped by 93.0% for cattle, 93.9% for swine and 69.3% for pets between 2013 and 2017.20 With a few exceptions, these substantial decreases match our findings.
Looking at the broader picture, this study suggests that practices and antibiotic use in a given species only influence resistance in that species and that transmission events are not frequent enough to have a major influence on intra-sector trends. In recent years, regulatory measures in veterinary medicine have greatly impacted antibiotic use in France, thereby contributing to the decreases seen in resistance.3 By contrast, measures in human health, which were non-regulatory, have not had such an impact on antibiotic use and therefore on the resistance trends. In addition, the decrease in the veterinary sector does not seem to affect resistance trends in human medicine.
Two standards were used to analyse data, as they were adapted to origins of samples (human versus animals). However, as the methodology differed (different inoculum and different incubation temperature), the antibiotic tested could also differ (Table S1) and, as different pathological contexts were considered according to the species, direct comparisons between levels of resistance remain limited in our study. Two ongoing surveillance systems for AMR made this study possible. However, it is worth noting that the use of surveillance data could provide an imperfect picture of the overall situation concerning resistance epidemiology in clinical E. coli isolates in France (due to selection bias), since laboratories voluntarily join surveillance networks. Furthermore, though more than 790 000 antibiograms were analysed, most came from humans (93.0%). The large discrepancies in numbers of samples collected by species impact the precision of the resistance levels. The difference between the number of data collected by the two networks can be explained by the lower cost of analysis to the patient in humans, the existence of other alternatives to treatment in animal health (culling for instance) and because veterinary medicine is mainly a population medicine in food-producing animals: one antibiogram is performed to treat all the animals in an infected batch.12 The sampled animal is considered representative of the sick batch. Even if we assume that the vast majority of samples were performed on untreated individuals,12 the proportions of samples from previously untreated compared with treated individuals were unknown. These elements can lead to misestimation of resistance levels. Besides, these findings might have overestimated resistance levels by grouping intermediate and resistant isolates. However, considering these biases to be constant over the period, the resistance trends remain meaningful. Furthermore, the strength of the parallel performed in this study was based on similar data collection in humans and animals and on a strictly similar methodology. Hence, in comparison with the ECDC/European Food Safety Authority (EFSA)/EMA report providing an ecological analysis of E. coli resistance levels in Europe,21 our study allowed an original additional parallel between resistance trends of bacteria isolated in the clinical context from humans, food-producing animals and pets, thanks to the methodology used.
Even if close and direct contact between humans and pets suggests potential similarities in their resistance dynamics, human trends did not seem more similar to dog and cat resistance trends than to resistance trends in food-producing animals. These findings underscore the importance of monitoring AMR in both pets and food-producing animals. They also advocate applying the efforts already being made to reduce antibiotic use to all sectors and all species, both in human and veterinary medicine.
Acknowledgements
We would like to thank all the MedQual and RESAPATH laboratories that have been collecting and transmitting antibiogram results for several years. We are particularly grateful to Christelle Philippon (RESAPATH secretary) for her meticulous follow-up of laboratories and careful data collection and Jean-Luc Vinard (RESAPATH data architect) for his careful management of the database. Finally, we are grateful to Jérôme Robert and Yves Péan from the French national observatory for epidemiology of bacterial resistance to antibiotics (ONERBA) for fostering intellectual exchange between the different AMR surveillance networks in France.
Funding
This study was supported by internal funding. No external funding was received. The funding bodies had no role in the study’s design, data collection and analysis, decision to publish or preparation of the manuscript.
Transparency declarations
None to declare.
The corresponding author affirms that this manuscript is an honest, accurate and transparent account of the study being reported, that no important aspects of the study have been omitted and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.
Supplementary data
Tables S1 and S2 are available as Supplementary data at JAC Online.
References
The AMR One Health Surveillance Committee. Nippon AMR One Health Report (NAOR) 2017.
WHO. Critically Important Antimicrobials for Human Medicine, 5th Revision.
ANSES. RESAPATH French Surveillance Network for Antimicrobial Resistance in Pathogenic Bacteria of Animal Origin.
CA-SFM, EUCAST. Comité de l’Antibiogramme de la Société Française de Microbiologie Recommandations 2018.
CA-SFM. Antibiogramme Vétérinaire du Comité de l’Antibiogramme de la Société Française de Microbiologie.
SPILF. Diagnostic et antibiothérapie des infections urinaires bactériennes communautaires de l’adulte.
ANSM. La consommation d’antibiotiques en France en 2016.
ANSES-ANMV. Suivi des ventes de médicaments vétérinaires contenant des antibiotiques en France en 2017.
ECDC, EFSA, EMA. ECDC/EFSA/EMA Second Joint Report on the Integrated Analysis of the Consumption of Antimicrobial Agents and Occurrence of Antimicrobial Resistance in Bacteria From Humans and Food-Producing Animals.
