- Email: [email protected]
A trend analysis of antimicrobial resistance in commensal Escherichia coli from several livestock species in Belgium (2011–2014)
Accepted Manuscript Title: A trend analysis of antimicrobial resistance in commensal Escherichia coli from several livestock species in Belgium (2011-2014) Author: Jean-Baptiste Hanon Stijn Jaspers Patrick Butaye Pierre Wattiau Estelle M´eroc Marc Aerts Hein Imberechts Katie Vermeersch Yves Van der Stede PII: DOI: Reference:
S0167-5877(15)30006-4 http://dx.doi.org/doi:10.1016/j.prevetmed.2015.09.001 PREVET 3867
To appear in:
PREVET
Received date: Revised date: Accepted date:
2-2-2015 30-8-2015 1-9-2015
Please cite this article as: Hanon, Jean-Baptiste, Jaspers, Stijn, Butaye, Patrick, Wattiau, Pierre, M´eroc, Estelle, Aerts, Marc, Imberechts, Hein, Vermeersch, Katie, Stede, Yves Van der, A trend analysis of antimicrobial resistance in commensal Escherichia coli from several livestock species in Belgium (2011-2014).Preventive Veterinary Medicine http://dx.doi.org/10.1016/j.prevetmed.2015.09.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A trend analysis of antimicrobial resistance in commensal Escherichia coli from several livestock species in Belgium (2011-2014) Jean-Baptiste Hanona, Stijn Jaspersb, Patrick Butayea,d,f, Pierre Wattiaua, Estelle Méroca, Marc Aertsb, Hein Imberechtsa, Katie Vermeerschc, Yves Van der Stedea,e* [email protected] a
Veterinary and Agrochemical Research Centre (CODA-CERVA), Groeselenberg 99, 1180
Brussels, Belgium b
Interuniversity Institute for Biostatistics and statistical Bioinformatics (I-BioStat), Hasselt
University, Agoralaan building D, 3590 Diepenbeek, Belgium c
Federal Agency for the Safety of the Food Chain (FASFC), Food Safety Centre, Boulevard
du Jardin Botanique 55, 1000 Brussels, Belgium d
Ghent University, Department of Pathology, Bacteriology and Poultry Diseases, Faculty of
Veterinary Medicine, Salisburylaan 133, 9820 Merelbeke, Belgium e
Ghent University, Department of Virology, Parasitology and Immunology, Faculty of
Veterinary Medicine, Salisburylaan 133, 9820 Merelbeke, Belgium f
Ross University, Department of Biosciences, Basseterre, St Kitts and Nevis, West Indies
*
Corresponding author at: Veterinary and Agrochemical Research Centre (CODA-CERVA),
Groeselenberg 99, 1180 Brussels, Belgium. Tel.: + 32 379 0 626.
1
ABSTRACT A temporal trend analysis was performed on antimicrobial resistance data collected over 4 consecutive years (2011-2014) in the official Belgian antimicrobial resistance monitoring programme. Commensal E. coli strains were isolated from faecal samples of four livestock categories (veal calves, young beef cattle, broiler chickens and slaughter pigs) and the trends of resistance profiles were analysed. The resistance prevalence remained high (> 50%) during the study period for ampicillin in veal calves and chickens, for ciprofloxacin and nalidixic acid in chickens, for sulfamethoxazole in veal calves, chickens and pigs and for tetracycline in veal calves. Using logistic regression and Generalized Estimating Equation and after p value adjustment for multiple testing (Linear step-up method), statistically significant decreasing temporal trends were observed for several of the 11 tested antimicrobials in several livestock categories: in veal calves (10/11), in chickens (6/11) and in pigs (5/11). A significant increasing trend was observed for the prevalence of resistance to ciprofloxacin in chickens. Multi-resistance, considered as the resistance to at least three antimicrobials of different antibiotic classes, was observed in the four livestock categories but was significantly decreasing in veal calves, chickens and pigs. Overall, the prevalence of resistance and of multi-resistance was lowest in the beef cattle livestock category and highest in broiler chickens. These decreasing temporal trends of antimicrobial resistance might be due to a decrease of the total antimicrobial consumption for veterinary use in Belgium which was reported for the period between 2010 and 2013. The methodology and statistical tools developed in this study provide outputs which can detect shifts in resistance levels or resistance trends associated with particular antimicrobial classes and livestock categories. Such outputs can be used as objective evidence to evaluate the possible efficacy of measures taken by animal health authorities and stakeholders in the livestock sector to limit antimicrobial resistance occurrence. 2
Keywords: Antimicrobial resistance; Commensal bacteria; Escherichia coli; Trend analysis; Surveillance; Monitoring.
3
INTRODUCTION Antimicrobial resistance in humans and animals has become a major public health concern leading to the set-up of national and supra-national surveillance programmes to monitor its evolution (AFSSA, 2009, CVI, 2014, DANMAP, 2014, EFSA/ECDC, 2015, WHO, 2014). Antimicrobial resistance in bacteria from animals has not only implications on animal therapy: both resistant bacteria and resistance genes can spread from animals to humans, through direct contact, by contaminating the environment or through the food chain (Geenen et al., 2010). Intestinal commensal bacteria present in animals and humans are considered as good indicators to monitor the general level of antimicrobial resistance as they are subjected to a selection pressure driven by the antimicrobials to which their hosts are exposed (Szmolka and Nagy, 2013). Moreover, they are abundant in intestinal flora and relatively easy to isolate. This study was based on data obtained from a national annual monitoring programme for antimicrobial resistance in commensal bacteria from livestock (Butaye, 2011, 2012, 2013, unpublished results), which started in Belgium in 2011 under the supervision of the Federal Agency for the Safety of the Food Chain (FASFC). The aim was to determine the effect of measures taken to reduce antimicrobial resistance. The European Union (EU) made it mandatory for its Member States to monitor and report antimicrobial resistance in zoonotic Salmonella spp. and Campylobacter spp. isolates (EU Directive 2003/99/EC). Guidelines were provided by the EU and the European Food Safety Authority (EFSA) in 2008 for the harmonized monitoring and reporting of resistance in indicator E. coli and Enterococcus spp. (EFSA, 2008). Furthermore, since 2014, the monitoring and reporting of the resistance in commensal E. coli has become mandatory in the EU (EU Decision 2013/652/EU) (ECDC/EFSA/EMA, 2015).
4
The objective of this study was to perform a trend analysis of the prevalence of antimicrobial resistance in commensal Escherichia coli, which can be considered as representative of Gram negative bacterial species of the commensal flora from livestock. The bacterial isolates were obtained from faecal samples taken from animals of several livestock species living in Belgian farms: veal calves, beef cattle, broiler chickens and slaughter pigs. A secondary objective of the study was to evaluate the level of multi-resistance (resistance against at least three antibiotics of different antibiotic classes in a single strain) and its trend over the same period.
MATERIAL AND METHODS Study period This study was based on annual data, collected from 2011 to 2014, in the context of the antimicrobial resistance Belgian monitoring programme. Sampling Faecal samples were collected each year by official veterinarians of the Federal Agency for Safety of the Food Chain (FASFC) according to standardised technical sampling instructions as part of the national monitoring programme. Faecal samples were taken from the following categories of livestock species: - Veal calves: young cattle kept in specialised units for fattening. In 2011 faecal samples were taken at the farm level (one sample consisted of a pool of faeces collected from different spots on the floor and representing 10 animals < 7 months old) but from 2012 the samples were taken directly from the rectum of the animals at the slaughterhouse (one animal sampled/farm; average age at slaughter 7-8 months).
5
- Beef cattle (meat production): young animals (< 7 months old) from farms raising beef cattle for meat production. Faecal samples were taken from the floor at the farm (one sample consisted of a pool of faeces collected from different spots on the floor and representing 10 animals). - Broiler chickens: samples were taken at the slaughter house (pools of pairs of caeca from 10 chickens/batch). - Slaughter pigs: faecal samples of fattening pigs (1 animal sampled/farm) were taken from the rectum at the slaughter house (average age at slaughter 5-6 months). The sample size was designed in order to reach a target of 170 isolates per year for each livestock category and each bacterial species, following EFSA recommendations. Such a sample size allows the detection of significant changes of prevalence of resistance (in function of the observed prevalence) and with an acceptable precision (± 8% in the worst case scenario). It also allows the detection of annual trends such as a 5% decrease per year if starting from an initial 50% prevalence of resistance or a 2% decrease per year if starting from an initial 0.1% prevalence (EFSA, 2008). To improve representativeness, the sampling was stratified according to the different Belgian provinces. The number of samples taken in each province was proportional to the number of registered cattle herds for beef cattle and to the number of slaughtered animals for other livestock categories. Laboratory testing Isolates of E. coli (n=3198) were obtained from the faecal samples (one isolate/faecal sample) at the laboratories of the two Belgian regional animal health association, ARSIA (Association Régionale de Santé et d’Identification Animales) and DGZ-Vlaanderen 6
(Dierengezondheidszorg), following their specific standard operating procedures (SOP) as described by Butaye in the annual reports of the monitoring programme (unpublished results). The isolated strains were sent to the National Reference Laboratory (CODACERVA) for species confirmation and susceptibility testing using a micro-dilution technique (Sensititer®). A panel of 14 antimicrobials as recommended by EFSA (EFSA, 2008, EFSA, 2014) was used from 2011 to 2013. However, in 2014 the panel changed, following specifications laid down in decision 2013/652/EU when harmonised monitoring at EU level became mandatory for E. coli. In this latter panel, new antimicrobials were added (azithromycin, meropenem, tigecycline) while others were removed (florfenicol, kanamycin, streptomycin), in order to better evaluate the resistance to antimicrobial substances used in human medicine. Therefore, this trend analysis was restricted to the 11 antimicrobials tested during all 4 years of the study period (Table 1). For each isolated strain and each antimicrobial substance tested, the Minimal Inhibitory Concentration (MIC) was read through a semi-automated procedure and stored in a database. MIC is defined as the lowest concentration of antimicrobial that inhibits the visible growth of bacteria (Andrews, 2001). To differentiate between susceptible and resistant strains, quantitative MIC values were converted into binary qualitative values (Resistant/Susceptible) based on the susceptibility breakpoints or Epidemiological Cut-offs (ECOFF) defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (EUCAST, 2014). A bacterial strain is considered resistant to a specific antimicrobial if the MIC value is greater than the corresponding cut-off value. Statistical analysis Descriptive statistics
7
For each livestock category and each specific antimicrobial, the percentage (with 95% confidence intervals) of resistant isolates observed annually was calculated and plotted on graphs (data not shown). The proportion (with 95% confidence interval) of multi-resistant strains observed annually was also calculated for each animal category and plotted on graphs. In addition, diversity indices (entropy, weighted entropy) were calculated using the R software (http://cran.rproject.org/) to describe the degree of diversity of multi-resistance: diversity indices are quantitative measures that reflect how many different types of resistance are present in a dataset, and simultaneously take into account how evenly the observed resistance types in question are distributed. These indices take a value between 0 and 1 (0 if no diversity = all isolates resistant to the same number of antimicrobials; 1 if maximum diversity). We used the weighted entropy index (Guiasu, 1971) which is a diversity index that takes a higher value (closer to 1) if the isolates are more distributed to the right end of the scale, i.e. resistant to a higher number of antimicrobials. The weighted entropy was calculated per year for each animal category. Trend analysis Using SAS 9.2 software (SAS Institute Inc., Cary, NC, USA), several statistical methods were used to model the observed trends. Univariate models, based upon categorical data (logistic regression, generalised logit models) or upon continuous data (models for intervalcensored data, mixture models) were used to analyse the time trend of resistance for each antimicrobial agent separately. Multivariate models (Generalised Estimating Equation models (GEE)) were used to take into account a possible correlation between antimicrobial agents in the time trend of resistance. After evaluating these various models for their capacity to accurately reflect the data and after comparing the results and their possible interpretation, it 8
was decided to restrict the trend analysis to the logistic regression (univariate analysis) and to the GEE (multivariate analysis). These two approaches offered the best convergence and gave outputs which were convenient to be interpreted and compared and which could be easily plotted on graphs for presentation to authorities and decision makers. If, in the logistic regression model, i represents the probability for an isolate to be resistant to a certain antimicrobial in year of reporting ti and f ti represents a function of time, then we can consider a linear time trend and use the logit link function, which is the logarithm of the odds of the probability as expressed in Eq.(1).
log i 0 1ti 1 i
(1)
where 0 is an intercept .The results were described in the form of Odds Ratio (OR) as in the logistic regression β1 = ln OR . In this model, an OR > 1 means that the probability to be resistant increases with time and an OR < 1 means it decreases with time. GEE was used to perform multivariate analysis and take into account possible correlations between antimicrobials through the specification of one of a variety of possible working correlation matrix structures. An unstructured working correlation matrix was used, as correlations between any two responses are unknown and need to be estimated. For multi-resistance a logistic model was used to check whether there was a significant trend (increase or decrease) over the years regarding the prevalence of multi-resistant strains, for each animal category. In this model an OR >1 means that the probability for a strain to be multi-resistant increases with time.
9
In order to take into account the multiple comparisons/testing performed in this study, two correction methods were used to adjust the 5% significance level: the Bonferroni correction and the linear step-up method of Benjamini and Hochberg (1995) based on the control of the false discovery rate. The original p values refer to the effects of each different antimicrobial assessed separately. The adjusted p values provide a family wise significance level in order to make a statement on the entire pool of antimicrobials jointly. In the results of this study, adjusted p values obtained with both correction methods were compared to the original uncorrected p values.
RESULTS Resistance to specific antimicrobials Descriptive statistics A total of 3198 isolates analysed for antimicrobial susceptibility were included in the data set. The distribution of the number of isolates per year and per livestock category with the corresponding prevalence of isolates resistant to at least one antimicrobial are presented in Table 2. Detailed data and descriptive statistics for the years 2011-2013 are reported in the annual reports published by CODA-CERVA (Butaye, 2011, 2012, 2013). High prevalence values (> 50%) of resistant E. coli strains were observed during the study period for several antimicrobials in all animal categories except in beef cattle: for ampicillin in veal calves and chickens, for ciprofloxacin and nalidixic acid in chickens, for sulfamethoxazole in veal calves, chickens and pigs and for tetracycline in veal calves. In broiler chickens, a resistance prevalence > 40% was observed each year for more than half of
10
the antimicrobials tested (6/11) (Table 3). The resistance prevalence to ciprofloxacin was > 40% in veal calves in 2011 and 2012 and remained >20% in 2013 and 2014 while in beef cattle and veal calves it remained < 20% during the whole study period. The resistance prevalence to cephalosporins (cefotaxime and ceftazidime) was < 5% during the whole study period for veal calves, beef cattle and pigs except in 2012 for veal calves and beef cattle. In broiler chickens it was much higher than in other species, especially in 2011/2012 with prevalence values between 15 and 30%, which decreased later in 2013/2014 to below 10%. The prevalence of resistance to chloramphenicol remained above 20% during the whole study period in all livestock categories except in beef cattle for which it was close to 15%. Trend analysis All trends observed for the four livestock categories are summarized in Table 3. A mention is made when the resistance prevalence remained high during the whole study period. A significant decreasing resistance trend was observed for several antimicrobials in all four livestock categories: in veal calves for all tested antimicrobials except for gentamicin and tetracyclin (9/11), in beef cattle for ampicillin and nalidixic acid (2/11), in chickens for ampicillin, cefotaxime, sulfamethoxazole, ceftazidime, tetracycline and trimethoprim (6/11) and in pigs for ciprofloxacin, cefotaxime, nalidixic acid, ceftazidime and tetracycline (5/11). However, a significant increasing resistance trend was observed for ciprofloxacin in chickens. When looking at the annual resistance prevalence values for each antimicrobial in the different livestock species, it was noticed that in several cases, after an increase in 2012 compared to 2011, the prevalence decreased in 2013 and 2014 (data not shown), resulting in an overall significant decreasing trend for the study period. The same pattern was observed for ciprofloxacin in chickens but the decrease of resistance prevalence in 2013 (74.8%; 95% C.I. [68.8-80.0]) and 2014 (69.6%; 95% C.I. [61.9-76.3]) was not sufficient to compensate
11
for the previous increase of 2012 (79.6%; 95% C.I. [74.2-83.2]) compared to 2011 (62.9%; 95% C.I. [58.1-67.4]), resulting in an overall increasing trend in the study period. The trends observed with the univariate and multivariate analysis are presented, in figure 1 and 2 respectively (uncorrected p value) for each animal category and for each antimicrobial. When adjusting the p values due to multiple testing, using the linear step-up method, all trends remained significant except in beef cattle; however, when adjusting the p values using the Bonferroni method, several trends became insignificant (Table 3). Multi-resistance Multi-resistance was defined as resistance by an isolate to at least three antimicrobials of the panel, belonging to different antimicrobial classes : ampicillin (penicillins class), cefotaxime and/or ceftazidime (cephalosporins class), chloramphenicol (phenicols class), ciprofloxacin and/or acid nalidixic (quinolones class), colistine (polymyxins class), gentamicin (aminoglycosides class), sulfamethoxazole (sulphonamides class), tetracycline (tetracyclines class) and trimethoprim (trimethoprim class). The prevalence of multi-resistance remained above 40% during the whole study period for all livestock categories except for beef cattle; the highest levels of multi-resistance were found in broiler chickens (> 60%) (Fig 3a). Significant decreasing trends in the prevalence of multi-resistant E. coli strains were observed in veal calves and in broiler chickens in the study period (Fig. 3b). The comparison of diversity indices (weighted entropy) shows that multi-resistance against a large number of antibiotics was most frequent in E. coli from veal calves and broiler chickens and least frequent in isolates from fattening pigs; weighted entropy index decreased (i.e. multi-resistance shifted to a lower number of antimicrobials) in 2013 and 2014 in veal calves, broiler chickens and pigs (Fig. 4).
12
DISCUSSION Persisting high levels of resistance to some specific antimicrobials were observed continuously during this 4-year study period for several livestock categories. This was the case for ampicillin, ciprofloxacin, nalidixic acid, sulfamethoxazole and tetracycline, for which the prevalence of resistant E. coli strains remained continuously high (> 50%) in veal calves, chickens and pigs. Such high levels of resistance have also been observed previously, both in pathogenic and non-pathogenic strains of E. coli isolated from livestock, in other studies and other monitoring programmes implemented in several European countries (AFSSA, 2009, EFSA/ECDC, 2015, CVI, 2014, Garcia-Migura, 2014, Hendriksen et al, 2008a, Hendriksen et al, 2008b, Persoons et al., 2010). Therefore, the use of such antimicrobials should be carefully monitored, especially in livestock species with more intensive treatment practices and more frequent antimicrobial use (veal calves, broiler chickens, slaughter pigs), for which the highest resistance prevalence values were observed in this study. Indeed several studies have shown that higher exposure of livestock to antimicrobials is associated with higher resistance prevalence in commensal E. coli (Berge et al., 2005, Bosman et al., 2014, Burow et al., 2014, Li et al., 2014, Simoneit et al., 2014). The highest levels of antimicrobial resistance revealed by this study were found in E. coli strains isolated from broiler chickens: in this animal species, a high prevalence (>50%) of resistance to more than half of the tested antimicrobials was observed each year during the study period. In addition it was shown that the level of multi-resistance in E. coli isolates was the highest in broiler chickens compared to other livestock categories. The explanation for the difference in terms of prevalence of antimicrobial resistance between livestock categories is possibly due to different antimicrobial treatment regimens (group vs individual and oral (feed) vs parenteral treatment) following various management and animal husbandry
13
practices. Moreover, in poultry farming, broiler chickens are slaughtered very young, at the age of 6 weeks, when they harbour more resistant bacteria than older animals (Butaye et al., 1999). Much higher levels of resistance to cephalosporins were observed in chickens compared to other livestock categories, especially in 2011/2012. Such an observation has already been reported in other studies (Persoons et al., 2010; Persoons et al., 2011a). A possible explanation is the off label use of celtiofur sprayed on 1-day old chicks, a practice in use in some hatcheries in Belgium until 2010, when it came to an end after enforced veterinary inspection of hatcheries was carried out by FASFC (FASFC, 2011, p.252). Another hypothesis for such high levels of resistance to cephalosporins observed in chickens is the exposure of 1-day old chicks to antimicrobials before they are imported into Belgium. Indeed, 15 million 1-day chicks were imported in 2014 from 7 different EU countries, of which 80% came from the Netherlands and France (FASFC, personal communication). However the use of cephalosporins in poultry has been banned in the EU since 2010 (Council regulation EC 470/2009 and Commission regulation EC 37/2010), therefore this exposure is now unlikely to happen. The interpretation of the resistance prevalence must be specific to each antimicrobial and take into consideration not only the risk of therapeutic failure in veterinary medicine but also the possible impact on public health in case resistant strains encountered in animals spread to humans. Antimicrobials belonging to the (fluoro)quinolones and cephalosporins of 3rd and 4th generation classes (represented in this trend analysis respectively by ciprofloxacin on one hand and by cefotaxime and ceftazidime on the other hand) are listed among substances considered as critically important drugs to human medicine by WHO and critically important drugs to animal heath by OIE. These lists are based on criteria such as the importance of the drug for therapeutic use in serious infections in human and in veterinary medicine or as the absence of alternative treatment (OIE, 2015, WHO, 2012, WHO, 2014). This categorization 14
of critically important drugs has been subsequently recognised by other international institutions involved in public and animal health (ECDC/EFSA/EMA, 2015, EFSA/ECDC 2015, EMA, 2014, FAO, 2008). Therefore, a moderate (10 – 20%) prevalence of resistance to these critically important antimicrobials can be considered as more worrisome than the high prevalence (>50%) observed for other antimicrobials such as tetracycline which are less important in human therapy and for which alternative antimicrobials are available. Both statistical models (logistic regression and GEE) used in this study show significant decreasing trends of the resistance prevalence in all livestock categories for several antimicrobials. The comparison of annual data shows that the decrease in most cases started in 2013 and continued in 2014 after a rise in 2012 compared to 2011. This pattern of an increasing trend followed by a decreasing one was confirmed and found statistically significant when using non-linear models for resistance prevalence to chloramphenicol, ciprofloxacin, colistin and nalidixic acid in chickens (data not shown). The only increasing trend detected in this study was for resistance to ciprofloxacin in chickens but as mentioned in the results this was due to an important increase in prevalence in 2012 (+16,2%) followed by a moderate decrease in 2013 and 2014 (-9.4%). Data collected in the coming years will provide a better estimation of the actual evolution. The detection of significant decreasing antimicrobial resistance and decreasing multi-resistance in poultry and in veal calves, for which the prevalence of resistance was highest and therefore easiest to detect, may indicate that there is indeed an actual decrease in the overall resistance prevalence amongst commensal E. coli in Belgian livestock. There are several possible explanations for an increase or a decrease of antimicrobial resistance in a given population. The most obvious is the selection pressure on bacteria when exposed to antimicrobials. Indeed, significant correlations or positive associations between 15
veterinary antimicrobial use and antimicrobial resistance in food-producing animals have already been demonstrated at the national and European level (Chantziaras et al., 2014, ECDC/EFSA/EMA, 2015, Garcia-Migura, 2014) as well as at farm level (Bosman et al., 2014). However other explanations have also been put forward to explain short-term variations in resistance: e.g. the spread of a more vital clone with Salmonella serotypes (Bertrand et al., 2006, Butaye et al., 2006, Imberechts et al., 2000), the horizontal transfer of extended-spectrum beta-lactamases (ESBL) resistance genes without selective pressure in animal strains of E. coli (Smet et al.2011) or the rapid spread of methicillin resistant Staphylococcus aureus (MRSA) in pigs without any antimicrobial selective pressure (Crombé et al., 2011). Resistance may also be co-selected by the use of totally unrelated antibiotics, giving a discrepancy between the presence of resistance and the actual usage of antimicrobials. This phenomenon was observed with the persistence of vancomycin resistance in pigs after its ban as a growth promoter, due to a physical link to a macrolide resistance gene (Bager et al., 1999). It was also seen in the persistence of chloramphenicol resistance without resistance to florfenicol due to resistance genes found in multi-resistance plasmids and in integrons, harbouring also other resistance genes (Schwarz, 2004). In our study we observed indeed that resistance to chloramphenicol is still present, although this antimicrobial has been prohibited for veterinary use in Europe since January 1997 (Council Regulation (EEC) 2377/90 and amendments 1570/98 and 508/1999). A significant decreasing trend for the resistance prevalence was observed in veal calves while no trend was observed in other livestock categories. Nevertheless, the prevalence of resistance to chloramphenicol remained above 20% of E. coli isolates for the whole study period in all livestock categories except in beef cattle for which it was close to 15%. Suspected illegal use of this drug by livestock keepers can be ruled out as an explanation of the persistence of such resistance levels. Indeed there is 16
an ongoing official monitoring programme in Belgium for the detection of chloramphenicol residues (Royal decree 2013/18154) at the slaughter house (injection sites) and at farm level (urine and faecal samples) (FASFC, 2014). The detection of such residues are extremely rare: between 2011 and 2014, 1626 samples were taken at farm-level or at the slaughter house from the various livestock categories (cattle, n= 198; veal calves, n= 89; pigs, n=463; broiler chickens, n= 876) and only one sample was detected with very low levels of chloramphenicol residues, at a concentration much below the authorized EU reference limit (FASFC, personal communication, 2015). Cross-resistance with other phenicol antimicrobials (e.g. florfenicol), which are authorized for veterinary use, is not an explanation either since most of the strains are only resistant to chloramphenicol indicating that the mechanism involved relies on the persisting circulation of ‘old’ resistance genes (Kim and Aoki, 1996, Schwartz et al., 2004). The most probable explanation for the persistence of chloramphenicol resistance is the mechanism of co-selection which selects bacterial strains resistant to several antimicrobials due to the clustering of several resistance genes at the same genetic locus in the genome (plasmids, transposons or integrons). This phenomenon was reported in Salmonella spp. and in vitro for E. coli (Barraud and Ploy, 2015, Hall, 2010, Montero 2013). A decrease in consumption of antimicrobial pharmaceuticals for veterinary use has been reported in Belgium between 2010 and 2013 (AMCRA, 2014, EMA, 2014). While the total biomass of livestock has not changed much during this period in Belgium, the total sales in tons based on the figures recorded by the Belgian Federal Agency for Medicines and Health Products (FAMHP) and reported by the Belgian Center of Expertise on Antimicrobial Consumption and Resistance in Animals (AMCRA) have decreased by 13.3 % in 2013 as compared to 2010. The decrease was observed for all antimicrobial classes (-6.8% to -53.5 % according to the antimicrobial classes), except for cephalosporins (+12.5%) and for phenicols (+9.4%); a decrease was also observed for the penicillin class between 2013 and 2011 (-8%). 17
However, these figures represent the overall antimicrobial consumption for veterinary use but the consumption stratified by animal species is currently unavailable. Hence it is impossible to correlate directly the observed decreasing trends of antimicrobial resistance in the different livestock categories to the decreasing consumption of antimicrobials for veterinary use. This shows the necessity in the future that animal health authorities collect data about antimicrobial consumption at the national level by animal species and not only bulk figures. Efforts are currently underway in Belgium in the pig and poultry sector to better record antimicrobial use at the farm level and a national data collection system is planned for 2016. Since 2012, recommendations have been provided by AMCRA to farmers and veterinarians, to promote the responsible use of antimicrobials with the objective to reach a substantial reduction by 2020 (-50% compared to 2011). Although the actual impact of this intervention has not been formally assessed, one can expect that it has contributed to a decrease of antimicrobial consumption and indirectly to a decrease of antimicrobial resistance in the commensal flora due to a lower exposure of livestock to antimicrobials. Similar observations of decreasing prevalence of antimicrobial resistance in livestock following an effective decreasing consumption of antimicrobial veterinary drugs has already been reported in other European countries such as the Netherlands and Denmark (CVI, 2014; DANMAP, 2014). This trend analysis of antimicrobial resistance in commensal E. coli, isolated from different livestock species, was based upon data collected over a 4-year period of national monitoring in Belgium (2011-2014). Though EFSA performs trend analysis on data provided by EU Member States after at least 5 years of monitoring (EFSA/ECDC, 2015), it was decided to undertake this analysis after 4 years as it appeared to be sufficient to detect some significant trends. These results will be confirmed or adjusted when more data are available in the coming years. Besides the time period, the sample size is also an important parameter that determines the power and robustness of this trend analysis. In most of the cases a sufficient 18
number of E. coli strains were isolated as proposed by EFSA (> 170 isolates/year/animal category) in order to detect annual trends (EFSA, 2008). However, in 2011 and 2014 the number of bacterial isolates per year was lower for some of the animal categories (Table 1), leading to larger uncertainty in the estimation of the prevalence of resistant strains and therefore decreasing the power to detect some trends. For this reason, it is recommended that animal health authorities shall implement and control a harmonised sampling protocol (sample size, sampling method and sample matrix) during further monitoring of E. coli in the different livestock species, following EU specifications. In this study, several dependent or independent statistical tests were performed simultaneously on a single data set. Therefore, p values were adjusted using correction methods for multiple testing/comparison, such as the Bonferroni method and the linear stepup method of Benjamini and Hochberg. The use of the correction methods for the p values is conservative as it decreases the risk of detecting a trend when in reality there is no trend (false positive results: type I error). However, such correction methods might also decrease the power of the statistical tests and prevent the detection of truly existing trends (false negative results: type II error). The Bonferroni correction is a well-known but very conservative method by which the significance level (α) is divided by the number of comparisons being made (Bland and Altman, 1995). It is usually not recommended when a large number (>5) of hypotheses are being tested. The Benjamini and Hochberg linear stepup method is less conservative and more appropriate to the current analysis and data set. The interpretation of the adjusted p values depends on the attitude of decision-makers. If the aim is to identify possible minor increasing trends of antimicrobial resistance because these are considered as a source of concern for public health (e.g. for critically important antimicrobials), one would try to minimize the false negatives and just consider the original p values. On the contrary, if false negatives can be accepted and the interest is in looking at 19
“strong” trends, then the linear step-up correction method should be used in order to avoid excessive sensitivity. Some biases may have affected the results of this study. For veal calves, a possible bias is a change in the sampling method in 2012 using faecal samples taken at the slaughter house from one single animal /farm (average age at slaughter 7-8 months) instead of pooling faecal samples taken at the farm from 10 animals younger than 7 months, as was done in 2011. The use of pooled faecal samples is known to be an appropriate sampling strategy to estimate the prevalence of antimicrobial resistance compared to individual samples (Bosman et al., 2012). However, several studies showed that optimal precision in the estimation of the resistance prevalence in a study population can be reached by testing one isolate obtained from one animal of each herd/flock sampled (Butaye et al., 1999, Persoons et al., 2011b, Yamamoto et al., 2014). In 2011, pooled samples were used but only one E. coli isolate was taken from each pool therefore the change to individual samples in 2012 probably did not affect the detection of resistant strains compared to the previous year. On the other hand, the age of sampled animals might also have affected the resistance prevalence: in 2011 samples were taken from calves slightly younger than in the following years and it is well known that the prevalence of resistance is generally higher among isolates of younger animals, whatever the species (Berge et al., 2010, Butaye et al., 1999, Langlois, 1988). Therefore the change in the sampling method should have resulted in a reduced prevalence of the resistance in 2012 for veal calves, as samples were taken from older animals. Yet, the opposite was observed, possibly due to a lack of precision of the estimated prevalence in 2011 because of an insufficient number of samples during that specific year (31 isolates). Finally, it should be mentioned that the observation of high levels of antimicrobial resistance or of increasing/decreasing trends does not necessarily reflect the efficacy of antimicrobial
20
treatment at field level. No information was available about an increase or decrease in treatment failures observed by farmers or veterinarians to support the observations made in our study. This aspect should be investigated in future studies to correlate antimicrobial resistance observed in vitro with field observations. In conclusion, after 4 years of monitoring of antimicrobial resistance in commensal E. coli isolates from different livestock species, descriptive statistics showed persisting high levels of resistance (>50%) for several antimicrobials in all livestock categories except in beef cattle. Particularly relevant was the high level of resistance to ciprofloxacin observed in E. coli strains isolated in broiler chickens. Moreover, the methodology and statistical models applied in this study allowed the detection of significant trends, mostly decreasing, in the prevalence of antimicrobial resistance. These results however need to be confirmed and maybe adjusted by data covering a longer period. The latest figures released by AMCRA indicate a slight increase (+ 1.1%) in the overall consumption/ kg of animal biomass of antimicrobials for veterinary use in Belgium for the year 2014 compared to 2013 (AMCRA, 2014). Therefore one can expect a possible increase in the prevalence of antimicrobial resistance in 2015 in Belgian livestock and a change in the trends observed previously. It is therefore recommended that the official monitoring programme should continue in order to provide sound data, relevant results and scientific information about the level of antimicrobial resistance to the public health and animal health authorities as well as to the stakeholders of the livestock sector. Such information can support decision-makers to provide guidelines and take appropriate control measures aimed at reducing the use of antimicrobials in livestock. However such measures should also take into consideration resistance mechanisms and the risk for animal and human health associated with specific antimicrobial classes. The interpretation of the observed trends should not be restricted to statistical
21
analysis; it should also consider biological aspects and be discussed and validated by experts on the topic. Another possible outcome of the continuous monitoring of antimicrobial resistance is the evaluation of the efficacy and of the impact of actions taken to decrease the prevalence of antimicrobial resistance. Given the close link observed and demonstrated elsewhere between antimicrobial consumption and antimicrobial resistance it is advised to integrate both monitoring programmes in the future in order to correlate those data.
ACKNOWLEDGEMENTS This study was commissioned by the Belgian Federal Agency for the Safety of the Food chain (FASFC) and carried out by the epidemiology unit and the bacteriology department of CODA-CERVA (Veterinary and Agrochemical Research Centre, Brussels) in close collaboration with the Center for Statistics (CenStat, Hasselt University, Belgium) which developed the statistical models adapted to the available data and the expected outputs. Our thanks also go to the staff of the veterinary laboratories of the two Belgian regional animal health association (ARSIA and DGZ) for their work done to isolate the bacterial strains.
22
REFERENCES AFSSA, 2009. Farm 2005-2006: French antimicrobial resistance monitoring program in bacteria of animal origin. Available at http://www.anses.fr/Documents/SANT-RaFARM2006.pdf AMCRA, 2014. BelVet-Sac, Belgian Veterinary Surveillance of Antibacterial Consumption, National consumption report 2013. Available at http://www.amcra.be/en/rapportenpublicaties-en-wetgeving/national-reports AMCRA, 2015. BelVet-Sac, Belgian Veterinary Surveillance of Antibacterial Consumption, National consumption report 2014 Andrews, JM., 2001. Determination of minimum inhibitory concentrations. J Antimicrob Chemother. 2001 Jul; 48 Suppl 1:5-16 Bager, F., Aarestrup, FM., Madsen, M, Wegener, HC., 1999. Glycopeptide resistance in Enterococcus faecium from broilers and pigs following discontinued use of avoparcin. Microb Drug Resist. 1999 Spring; 5(1):53-6. Barraud, O., Ploy, MC., 2015. Diversity of Class 1 Integron Gene Cassette Rearrangements Selected Under Antibiotic Pressure. J Bacteriol. 2015 Apr 20. pii: JB.02455-14. [Epub ahead of print] Benjamini, Y. and Hochberg, Y., 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Statist Soc Ser B (Methodological). 57, 289-300.
23
Berge, A.C., Atwill, E.R., Sischo, W.M., 2005. Animal and farm influences on the dynamics of antibiotic resistance in faecal Escherichia coli in young dairy calves. Prev Vet Med. 2005 Jun 10;69(1-2):25-38. Epub 2005 Feb 26. Berge, A.C., Hancock, D.D., Sischo, W.M., Besser, T.E.. 2010. Geographic, farm, and animal factors associated with multiple antimicrobial resistance in fecal Escherichia coli isolates from cattle in the western United States. J Am Vet Med Assoc. 2010 Jun 15;236(12):1338-44. doi: 10.2460/javma.236.12.1338. Bertrand, S., Weill, FX, Cloeckaert, A., Vrints, M., Mairiaux, E., Praud, K., Dierick, K., Wildemauv,e C., Godard, C., Butaye, P., Imberecht,s H., Grimont, PA., Collard JM., 2006. Clonal emergence of extended-spectrum beta-lactamase (CTX-M-2)-producing Salmonella enterica serovar Virchow isolates with reduced susceptibilities to ciprofloxacin among poultry and humans in Belgium and France (2000 to 2003). J Clin Microbiol. 2006 Aug; 44(8):2897-903. Bland, J.M., Altman, D.G., 1995. Multiple significance tests: The Bonferroni method. BMJ. 310 (6973):170. Bosman, AB., Wagenaar, JA., Stegeman, A., Vernooij, H., Mevius, D., 2012. Quantifying antimicrobial resistance at veal calf farms. PLoS One. 2012; 7(9):e44831. doi: 10.1371/journal.pone.0044831. Bosman, AB., Wagenaa,r JA., Stegeman, JA., Vernooij, JC., Mevius, DJ., 2014. Antimicrobial resistance in commensal Escherichia coli in veal calves is associated with antimicrobial drug use. Epidemiol Infect. 2014 Sep; 142(9):1893-904. doi:10.1017/S0950268813002665.
24
Burow, E., Simoneit, C., Tenhagen, BA., Käsbohrer, A., 2014. Oral antimicrobials increase antimicrobial resistance in porcine E. coli--a systematic review. Prev Vet Med. 2014 Mar 1; 113(4):364-75. doi: 10.1016/j.prevetmed.2013.12.007. Epub 2013 Dec 27. Review Butaye, P., Devriese, L.A., Goossens, H., Ieven, M., Haesebrouck, F., 1999. Enterococci with acquired vancomycin resistance in pigs and chickens of different age groups. Antimicrob Agents Chemother. 1999 Feb;43(2):365-6 Butaye, P., Michael, GB., Schwarz, S., Barrett, TJ., Brisabois, A., White, DG., 2006. The clonal spread of multidrug-resistant non-typhi Salmonella serotypes. Microbes Infect. 2006 Jun;8(7):1891-7. Epub 2006 Mar 22. Review. Butaye, P., unpublished results. Annual Reports on antimicrobial resistance in commensal E. coli from poultry, pigs, cows and veal calves. (2011, 2012, 2013). Available at http://www.coda-cerva.be Chantziaras, I., Boyen, F., Callens, B., Dewulf, J., 2014. Correlation between veterinary antimicrobial use and antimicrobial resistance in food-producing animals: a report on seven countries. J Antimicrob Chemother. 2014 Mar; 69(3):827-34. doi: 10.1093/jac/dkt443. Crombé, F., Willems, G., Dispas, M., Hallin, M., Denis, O., Suetens, C., Gordts, B., Struelens, M., Butaye, P., 2012. Prevalence and antimicrobial susceptibility of methicillin-resistant Staphylococcus aureus among pigs in Belgium. Microb Drug Resist. 2012 Apr; 18(2):125-31. doi: 10.1089/mdr.2011.0138. Epub 2011 Nov 16 CVI (Central Veterinary Institute of Wageningen), 2014. MARAN 2014: Monitoring of antimicrobial resistance and antibiotic usage in animals in the Netherlands in 2013. Central Veterinary Institute, Wageningen. Available at www.cvi.wur.nl 25
DANMAP, 2014. DANMAP 2013 (Danish Integrated Antimicrobial Resistance Monitoring and Research Programme): Use of antimicrobial agents and occurrence of antimicrobial resistance in bacteria from food animals, food and humans in Denmark. ISSN 1600-2032. Available at http://www.danmap.org ECDC/EFSA/EMA, 2015. ECDC (European Centre for Disease Prevention and Control), EFSA (European Food Safety Authority) and EMA (European Medicines Agency) first 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. EFSA Journal 2015; 13(1):4006, 114 pp. doi:10.2903/j.efsa.2015.4006 EFSA (European Food Safety Authority), 2008. Report from the Task Force on Zoonoses Data Collection including guidance for harmonized monitoring and reporting of antimicrobial resistance in commensal Escherichia coli and Enterococcus spp. from food animals. EFSA Journal (2008) 141, 1-44 EFSA (European Food Safety Authority), 2014. Technical specifications on randomised sampling for harmonised monitoring of antimicrobial resistance in zoonotic and commensal bacteria. EFSA Journal 2014; 12(5):3686, 33 pp. doi:10.2903/j.efsa.2014.3686 EFSA/ECDC, 2015. The European Union summary report on antimicrobial resistance in zoonotic and indicator bacteria for humans, animals and food in the EU in 2013. EFSA Journal 2015; 13(2):4036, 178 pp., doi:10.2903/j.efsa.2015.4036 EMA, 2014. European Medicines Agency. Sales of veterinary antimicrobial agents in 26 EU/EEA countries in 2012, fourth European Surveillance of Veterinary Antimicrobial Consumption (ESVAC) report. (EMA/333921/2014). 26
EUCAST (European Committee on Antimicrobial Susceptibility Testing), 2014. Data from the EUCAST MIC distribution website. Available at http://www.eucast.org/mic_distributions/ FAO, 2008. Joint FAO/WHO/OIE Expert Meeting on Critically Important Antimicrobials. Report of the FAO/WHO/OIE Expert Meeting FAO Headquarters, Rome 26-30 November 2007. Available at http://www.fao.org/publications/card/en/c/3c566f76-caa85125-bb06-19f08dc27550/ FASFC, 2011. Federal Agency for the safety of the Food Chain, Rapport d’activités 2010. Available at http://www.afsca.be/rapportsannuels/ FASFC, 2014. Federal Agency for the safety of the Food Chain, Rapport d’activités 2013. Available at http://www.afsca.be/rapportsannuels/ Garcia-Migura, L., Hendriksen, R.S., Fraile, L., Aarestrup, F.M., 2014. Antimicrobial resistance of zoonotic and commensal bacteria in Europe: the missing link between consumption and resistance in veterinary medicine. Vet. Microbiol. 2014 May 14; 170 (1-2):1-9. doi: 10.1016/j.vetmic.2014.01.013. Epub 2014 Feb 3. Review. Geenen, P.L., Koene, M.G.J., Blaak, H., Havelaar, A.H. and van de Giessen, A.W., 2010. Risk profile on antimicrobial resistance transmissible from food animals to humans. RIVM-rapport 330334001, 2010. Available at http://www.rivm.nl Guiasu, S. (1971) Weighted Entropy. Reports on Mathematical Physics, 2, 165-179. Hall, RM., 2010. Salmonella genomic islands and antibiotic resistance in Salmonella enterica. Future Microbiol. 2010 Oct; 5(10):1525-38. doi: 10.2217/fmb.10.122. Review.
27
Hendriksen, R.S., Mevius, D.J., Schroeter, A., Teale, C., Jouy, E., Butaye, P., Franco, A., Utinane, A., Amado, A., Moreno, M., Greko, C., Stark, K.D., Berghold, C., Myllyniemi, A.L., Hoszowski, A., Sunde, M. and Aarestrup, F.M., 2008a. Occurrence of antimicrobial resistance among bacterial pathogens and indicator bacteria in pigs in different European countries from year 2002 - 2004; the ARBAO-II study. Acta Vet. Scand. 50, 19 Hendriksen, R.S., Mevius, D.J., Schroeter, A., Teale, C., Jouy, E., Butaye, P., Franco, A., Utinane, A., Amado, A., Moreno, M., Greko, C., Stark, K.D., Berghold, C., Myllyniemi, A.L., Wasyl, D., Sunde, M. and Aarestrup, F.M., 2008b. Prevalence of antimicrobial resistance among bacterial pathogens isolated from cattle in different European countries: 2002-2004. Acta Vet. Scand. 50, 28 Imberechts, H., D'hooghe, I., Bouchet, H., Godard, C., Pohl, P., 2000. Apparent loss of enrofloxacin resistance in bovine Salmonella typhimurium strains isolated in Belgium, 1991 to 1998. Vet Rec. 2000 Jul 15; 147(3):76-7. Kim, E., Aoki, T., 1996. Sequence analysis of the florfenicol resistance gene encoded in the transferable R-plasmid of a fish pathogen, Pasteurella piscicida. Microbiol Immunol. 1996; 9: 665–669 Langlois, BE., Dawson, KA., Leak, I., Aaron, DK., 1988. Effect of age and housing location on antibiotic resistance of fecal coliforms from pigs in a non-antibiotic-exposed herd. Appl Environ Microbiol. 1988 Jun; 54(6):1341-4. Li, P., Wu, D., Liu, K., Suolang, S., He, T., Liu, X., Wu, C., Wang, Y., Lin, D., 2014. Investigation of antimicrobial resistance in Escherichia coli and enterococci isolated
28
from Tibetan pigs. PLoS One. 2014 Apr 18; 9(4):e95623. doi: 10.1371/journal.pone.0095623. eCollection 2014 Montero, I., Herrero-Fresno, A., Rodicio, R., Rodicio, MR., 2013. Efficient mobilization of a resistance derivative of pSLT, the virulence plasmid specific of Salmonella enterica serovar Typhimurium, by an IncI1 plasmid. Plasmid. 2013 Jul; 70(1):104-9. doi:10.1016/j.plasmid.2013.03.002. Epub 2013 Mar 27. Persoons, D., Dewulf, J., Smet, A., Herman, L., Heyndrickx, M., Martel, A., Catry, B., Butaye, P., Haesebrouck, F., 2010. Prevalence and persistence of antimicrobial resistance in broiler indicator bacteria. Microb Drug Resist. 2010 Mar; 16(1):67-74. doi:10.1089/mdr.2009.0062 Persoons, D., Haesebrouck, F., Smet, A., Herman, L., Heyndrickx, M., Martel, A., Catry, B., Berge, A.C., Butaye, P., Dewulf, J., 2011. Risk factors for ceftiofur resistance in Escherichia coli from Belgian broilers. Epidemiol Infect. 2011 May;139(5):765-71. doi:10.1017/S0950268810001524. Epub 2010 Jun 29. Persoons, D., Bollaerts, K., Smet, A., Herman, L., Heyndrickx, M., Martel, A., Butaye, P., Catry, B., Haesebrouck, F., Dewulf, J., 2011. The importance of sample size in the determination of a flock-level antimicrobial resistance profile for Escherichia coli in broilers. Microb Drug Resist. 2011 Dec;17(4):513-9. doi: 10.1089/mdr.2011.0048 Schwarz, S., Kehrenberg, C., Doublet, B., Cloeckaert, A., 2004. Molecular basis of bacterial resistance to chloramphenicol and florfenicol. FEMS Microbiol Rev. 2004 Nov; 28(5):519-42. Review. Simoneit, C., Burow, E., Tenhagen, BA., Käsbohrer, A., 2014. Oral administration of antimicrobials increase antimicrobial resistance in E. coli from chicken--a systematic 29
review. Prev Vet Med. 2015 Jan 1; 118(1):1-7. doi: 10.1016/j.prevetmed.2014.11.010. Epub 2014 Nov 20 Smet, A., Rasschaert, G., Martel, A., Persoons, D., Dewulf, J., Butaye, P., Catry, B., Haesebrouck, F., Herman, L., Heyndrickx, M., 2010. In situ ESBL conjugation from avian to human Escherichia coli during cefotaxime administration. J Appl Microbiol. 2011 Feb; 110(2):541-9. doi: 10.1111/j.1365-2672.2010.04907.x. Epub 2010 Dec 10. Szmolka, A., Nagy, B., 2013. Multidrug resistant commensal Escherichia coli in animals and its impact for public health. Front Microbiol. 2013 Sep 3; 4: 258. doi: 10.3389/fmicb.2013.00258. OIE, 2015. World Organisation for Animal Health: OIE list of antimicrobials of veterinary importance, adopted in November 2007, latest update May 2015. Available at http://www.oie.int/en/our-scientific-expertise/veterinary-products/antimicrobials/ WHO, 2012. World Health Organisation: Critically important antimicrobials for human medicine - 3rd revision 2011. Available at http://apps.who.int/iris/bitstream/10665/77376/1/9789241504485_eng.pdf. 31. WHO, 2014. World Health Organisation: Antimicrobial resistance: global report on surveillance. Available at http://apps.who.int/iris/bitstream/10665/112642/1/9789241564748_eng.pdf?ua=1. Yamamoto, T., Hayama, Y., Hidano, A., Kobayashi, S., Muroga, N., Ishikawa, K., Ogura, A., Tsutsui, T., 2014. Sampling strategies in antimicrobial resistance monitoring: evaluating how precision and sensitivity vary with the number of animals sampled per farm. PLoS One. 2014 Jan 23;9(1):e87147. doi: 10.1371/journal.pone.0087147. eCollection 2014
30
Figure Captions
Veal Calves
Beef cattle
Chickens
Pigs
Fig. 1. Odds Ratios and their 95 % CI (univariate analysis: Logistic regression; uncorrected p value = 0.05) of the temporal trends to be resistant to a specific antimicrobial for E. coli isolates, per livestock category, between 2011 and 2014. OR > 1 indicates an increasing trend, OR < 1 indicates a decreasing trend. See legend for antimicrobials symbols in Table 1.
31
Fig. 2. Graphical representation (multivariate analysis: Generalised Estimating Equation model – GEE; uncorrected p value = 0.05) of the temporal trends of the probability to be resistant to a specific antimicrobial for E.coli isolates, per livestock category, between 2011 and 2014. Significant trends are in continuous lines. See legend for antimicrobials symbols in Table 1
32
100%
veal calves beef cattle chickens pigs
90% 80%
Prevalence
70% 60% 50% 40% 30% 20% 10% 0%
Fig. 3a. Prevalence (with 95% C.I.) of multi-resistant E. coli isolates, per animal category and per year. Multi-resistance was considered as resistance by an isolate to at least 3 antimicrobials belonging to different antimicrobial classes.
Fig.3b. Odds Ratios and their 95 % CI (univariate analysis: Logistic regression) of the temporal trends to be multi-resistant for E. coli isolates, per livestock category, between 2011 and 2014. OR > 1 indicates an increasing trend, OR < 1 indicates a decreasing trend.
33
weighted entropy
1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00
veal calves
2011
beef cattle
2012
broiler chickens slaughter pigs
2013
2014
Fig. 4. Weighted entropy indices per animal category and per year for E. coli isolates, describing the diversity of multi-resistance. A high weighted entropy index (closer to 1) means that overall, isolates of the corresponding year and animal category are generally resistant to a high number of antimicrobials.
34
Tables Table 1. Panel of 11 antimicrobials used for susceptibility testing of E. coli isolates during the study period (2011-2014). Antimicrobial Ampicillin Chloramphenicol Ciprofloxacin Colistin Cefotaxime Gentamicin Nalidixic acid Sulfamethoxazole Ceftazidime Tetracycline Trimethoprim
Symbol AMP CHL CIP COL FOT GEN NAL SMX TAZ TET TMP
35
Table 2. Number (N) of E. coli isolates tested for antimicrobial resistance for each year and each livestock category and corresponding percentage of isolates resistant to at least one antimicrobial (% Res)
2011 2012 2013 2014 Total
Veal calves N % Res 34 85.3 181 87.8 202 82.2 188 73.9 605 81.5
Beef cattle N % Res 154 38.3 175 52.0 204 37.3 164 28.7 697 39.2
Chickens N % Res 420 93.3 320 94.4 234 95.3 158 88.6 1132 93.4
Pigs N % Res 157 75.2 217 71.9 206 69.9 184 71.2 764 71.9
Total N % Res 765 78.2 893 79.3 846 72.0 694 65.9 3198 74.2
36
Table 3. Summary of the observed trends and of the persisting high prevalences of antimicrobial resistance for E. coli isolates, per livestock category from 2011 till 2014. All indicated trends (↑, ↓) were statistically significant (p = 0.05) both in univariate (logistic regression) and multivariate (GEE) analysis, even after using correction methods for multiple testing (Bonferroni and Linear step-up method), unless otherwise mentioned in the table 1, 2, 3. Antimicrobial Ampicillin Chloramphenicol Ciprofloxacin Colistin Cefotaxime Gentamicin Nalidixic acid Sufamethoxazole Ceftazidime Tetracycline Trimethoprim
veal calves
beef cattle 1, 2
↓ ++ ↓ ↓ ↓2 ↓
↓
↓ ↓ ++ ↓ ↓2, 3 ++ ↓
↓1, 2
chickens
pigs
2
↓ ++ ↑2 ++
↓
↓
↓2
++ ↓ ++ ↓ ↓+ ↓2 +
↓ ++ 2
↓ ↓2 + +
++ : Prevalence of antimicrobial resistance > 50% during 4 consecutive years + : Prevalence of antimicrobial resistance > 40% during 4 consecutive years
↑ : significant increasing trend of antimicrobial resistance prevalence ↓ : significant decreasing trend of antimicrobial resistance prevalence 1
: trend not significant after p value adjustment with linear step-up method : trend not significant after p value adjustment with Bonferroni method 3 : trend not significant in multivariate analysis (GEE) If no symbol is indicated for a corresponding antimicrobial and livestock category: no significant trend were detected and no persisting high prevalence of antimicrobial resistance during the 4 years period were observed 2
37
S0167-5877(15)30006-4 http://dx.doi.org/doi:10.1016/j.prevetmed.2015.09.001 PREVET 3867
To appear in:
PREVET
Received date: Revised date: Accepted date:
2-2-2015 30-8-2015 1-9-2015
Please cite this article as: Hanon, Jean-Baptiste, Jaspers, Stijn, Butaye, Patrick, Wattiau, Pierre, M´eroc, Estelle, Aerts, Marc, Imberechts, Hein, Vermeersch, Katie, Stede, Yves Van der, A trend analysis of antimicrobial resistance in commensal Escherichia coli from several livestock species in Belgium (2011-2014).Preventive Veterinary Medicine http://dx.doi.org/10.1016/j.prevetmed.2015.09.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A trend analysis of antimicrobial resistance in commensal Escherichia coli from several livestock species in Belgium (2011-2014) Jean-Baptiste Hanona, Stijn Jaspersb, Patrick Butayea,d,f, Pierre Wattiaua, Estelle Méroca, Marc Aertsb, Hein Imberechtsa, Katie Vermeerschc, Yves Van der Stedea,e* [email protected] a
Veterinary and Agrochemical Research Centre (CODA-CERVA), Groeselenberg 99, 1180
Brussels, Belgium b
Interuniversity Institute for Biostatistics and statistical Bioinformatics (I-BioStat), Hasselt
University, Agoralaan building D, 3590 Diepenbeek, Belgium c
Federal Agency for the Safety of the Food Chain (FASFC), Food Safety Centre, Boulevard
du Jardin Botanique 55, 1000 Brussels, Belgium d
Ghent University, Department of Pathology, Bacteriology and Poultry Diseases, Faculty of
Veterinary Medicine, Salisburylaan 133, 9820 Merelbeke, Belgium e
Ghent University, Department of Virology, Parasitology and Immunology, Faculty of
Veterinary Medicine, Salisburylaan 133, 9820 Merelbeke, Belgium f
Ross University, Department of Biosciences, Basseterre, St Kitts and Nevis, West Indies
*
Corresponding author at: Veterinary and Agrochemical Research Centre (CODA-CERVA),
Groeselenberg 99, 1180 Brussels, Belgium. Tel.: + 32 379 0 626.
1
ABSTRACT A temporal trend analysis was performed on antimicrobial resistance data collected over 4 consecutive years (2011-2014) in the official Belgian antimicrobial resistance monitoring programme. Commensal E. coli strains were isolated from faecal samples of four livestock categories (veal calves, young beef cattle, broiler chickens and slaughter pigs) and the trends of resistance profiles were analysed. The resistance prevalence remained high (> 50%) during the study period for ampicillin in veal calves and chickens, for ciprofloxacin and nalidixic acid in chickens, for sulfamethoxazole in veal calves, chickens and pigs and for tetracycline in veal calves. Using logistic regression and Generalized Estimating Equation and after p value adjustment for multiple testing (Linear step-up method), statistically significant decreasing temporal trends were observed for several of the 11 tested antimicrobials in several livestock categories: in veal calves (10/11), in chickens (6/11) and in pigs (5/11). A significant increasing trend was observed for the prevalence of resistance to ciprofloxacin in chickens. Multi-resistance, considered as the resistance to at least three antimicrobials of different antibiotic classes, was observed in the four livestock categories but was significantly decreasing in veal calves, chickens and pigs. Overall, the prevalence of resistance and of multi-resistance was lowest in the beef cattle livestock category and highest in broiler chickens. These decreasing temporal trends of antimicrobial resistance might be due to a decrease of the total antimicrobial consumption for veterinary use in Belgium which was reported for the period between 2010 and 2013. The methodology and statistical tools developed in this study provide outputs which can detect shifts in resistance levels or resistance trends associated with particular antimicrobial classes and livestock categories. Such outputs can be used as objective evidence to evaluate the possible efficacy of measures taken by animal health authorities and stakeholders in the livestock sector to limit antimicrobial resistance occurrence. 2
Keywords: Antimicrobial resistance; Commensal bacteria; Escherichia coli; Trend analysis; Surveillance; Monitoring.
3
INTRODUCTION Antimicrobial resistance in humans and animals has become a major public health concern leading to the set-up of national and supra-national surveillance programmes to monitor its evolution (AFSSA, 2009, CVI, 2014, DANMAP, 2014, EFSA/ECDC, 2015, WHO, 2014). Antimicrobial resistance in bacteria from animals has not only implications on animal therapy: both resistant bacteria and resistance genes can spread from animals to humans, through direct contact, by contaminating the environment or through the food chain (Geenen et al., 2010). Intestinal commensal bacteria present in animals and humans are considered as good indicators to monitor the general level of antimicrobial resistance as they are subjected to a selection pressure driven by the antimicrobials to which their hosts are exposed (Szmolka and Nagy, 2013). Moreover, they are abundant in intestinal flora and relatively easy to isolate. This study was based on data obtained from a national annual monitoring programme for antimicrobial resistance in commensal bacteria from livestock (Butaye, 2011, 2012, 2013, unpublished results), which started in Belgium in 2011 under the supervision of the Federal Agency for the Safety of the Food Chain (FASFC). The aim was to determine the effect of measures taken to reduce antimicrobial resistance. The European Union (EU) made it mandatory for its Member States to monitor and report antimicrobial resistance in zoonotic Salmonella spp. and Campylobacter spp. isolates (EU Directive 2003/99/EC). Guidelines were provided by the EU and the European Food Safety Authority (EFSA) in 2008 for the harmonized monitoring and reporting of resistance in indicator E. coli and Enterococcus spp. (EFSA, 2008). Furthermore, since 2014, the monitoring and reporting of the resistance in commensal E. coli has become mandatory in the EU (EU Decision 2013/652/EU) (ECDC/EFSA/EMA, 2015).
4
The objective of this study was to perform a trend analysis of the prevalence of antimicrobial resistance in commensal Escherichia coli, which can be considered as representative of Gram negative bacterial species of the commensal flora from livestock. The bacterial isolates were obtained from faecal samples taken from animals of several livestock species living in Belgian farms: veal calves, beef cattle, broiler chickens and slaughter pigs. A secondary objective of the study was to evaluate the level of multi-resistance (resistance against at least three antibiotics of different antibiotic classes in a single strain) and its trend over the same period.
MATERIAL AND METHODS Study period This study was based on annual data, collected from 2011 to 2014, in the context of the antimicrobial resistance Belgian monitoring programme. Sampling Faecal samples were collected each year by official veterinarians of the Federal Agency for Safety of the Food Chain (FASFC) according to standardised technical sampling instructions as part of the national monitoring programme. Faecal samples were taken from the following categories of livestock species: - Veal calves: young cattle kept in specialised units for fattening. In 2011 faecal samples were taken at the farm level (one sample consisted of a pool of faeces collected from different spots on the floor and representing 10 animals < 7 months old) but from 2012 the samples were taken directly from the rectum of the animals at the slaughterhouse (one animal sampled/farm; average age at slaughter 7-8 months).
5
- Beef cattle (meat production): young animals (< 7 months old) from farms raising beef cattle for meat production. Faecal samples were taken from the floor at the farm (one sample consisted of a pool of faeces collected from different spots on the floor and representing 10 animals). - Broiler chickens: samples were taken at the slaughter house (pools of pairs of caeca from 10 chickens/batch). - Slaughter pigs: faecal samples of fattening pigs (1 animal sampled/farm) were taken from the rectum at the slaughter house (average age at slaughter 5-6 months). The sample size was designed in order to reach a target of 170 isolates per year for each livestock category and each bacterial species, following EFSA recommendations. Such a sample size allows the detection of significant changes of prevalence of resistance (in function of the observed prevalence) and with an acceptable precision (± 8% in the worst case scenario). It also allows the detection of annual trends such as a 5% decrease per year if starting from an initial 50% prevalence of resistance or a 2% decrease per year if starting from an initial 0.1% prevalence (EFSA, 2008). To improve representativeness, the sampling was stratified according to the different Belgian provinces. The number of samples taken in each province was proportional to the number of registered cattle herds for beef cattle and to the number of slaughtered animals for other livestock categories. Laboratory testing Isolates of E. coli (n=3198) were obtained from the faecal samples (one isolate/faecal sample) at the laboratories of the two Belgian regional animal health association, ARSIA (Association Régionale de Santé et d’Identification Animales) and DGZ-Vlaanderen 6
(Dierengezondheidszorg), following their specific standard operating procedures (SOP) as described by Butaye in the annual reports of the monitoring programme (unpublished results). The isolated strains were sent to the National Reference Laboratory (CODACERVA) for species confirmation and susceptibility testing using a micro-dilution technique (Sensititer®). A panel of 14 antimicrobials as recommended by EFSA (EFSA, 2008, EFSA, 2014) was used from 2011 to 2013. However, in 2014 the panel changed, following specifications laid down in decision 2013/652/EU when harmonised monitoring at EU level became mandatory for E. coli. In this latter panel, new antimicrobials were added (azithromycin, meropenem, tigecycline) while others were removed (florfenicol, kanamycin, streptomycin), in order to better evaluate the resistance to antimicrobial substances used in human medicine. Therefore, this trend analysis was restricted to the 11 antimicrobials tested during all 4 years of the study period (Table 1). For each isolated strain and each antimicrobial substance tested, the Minimal Inhibitory Concentration (MIC) was read through a semi-automated procedure and stored in a database. MIC is defined as the lowest concentration of antimicrobial that inhibits the visible growth of bacteria (Andrews, 2001). To differentiate between susceptible and resistant strains, quantitative MIC values were converted into binary qualitative values (Resistant/Susceptible) based on the susceptibility breakpoints or Epidemiological Cut-offs (ECOFF) defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (EUCAST, 2014). A bacterial strain is considered resistant to a specific antimicrobial if the MIC value is greater than the corresponding cut-off value. Statistical analysis Descriptive statistics
7
For each livestock category and each specific antimicrobial, the percentage (with 95% confidence intervals) of resistant isolates observed annually was calculated and plotted on graphs (data not shown). The proportion (with 95% confidence interval) of multi-resistant strains observed annually was also calculated for each animal category and plotted on graphs. In addition, diversity indices (entropy, weighted entropy) were calculated using the R software (http://cran.rproject.org/) to describe the degree of diversity of multi-resistance: diversity indices are quantitative measures that reflect how many different types of resistance are present in a dataset, and simultaneously take into account how evenly the observed resistance types in question are distributed. These indices take a value between 0 and 1 (0 if no diversity = all isolates resistant to the same number of antimicrobials; 1 if maximum diversity). We used the weighted entropy index (Guiasu, 1971) which is a diversity index that takes a higher value (closer to 1) if the isolates are more distributed to the right end of the scale, i.e. resistant to a higher number of antimicrobials. The weighted entropy was calculated per year for each animal category. Trend analysis Using SAS 9.2 software (SAS Institute Inc., Cary, NC, USA), several statistical methods were used to model the observed trends. Univariate models, based upon categorical data (logistic regression, generalised logit models) or upon continuous data (models for intervalcensored data, mixture models) were used to analyse the time trend of resistance for each antimicrobial agent separately. Multivariate models (Generalised Estimating Equation models (GEE)) were used to take into account a possible correlation between antimicrobial agents in the time trend of resistance. After evaluating these various models for their capacity to accurately reflect the data and after comparing the results and their possible interpretation, it 8
was decided to restrict the trend analysis to the logistic regression (univariate analysis) and to the GEE (multivariate analysis). These two approaches offered the best convergence and gave outputs which were convenient to be interpreted and compared and which could be easily plotted on graphs for presentation to authorities and decision makers. If, in the logistic regression model, i represents the probability for an isolate to be resistant to a certain antimicrobial in year of reporting ti and f ti represents a function of time, then we can consider a linear time trend and use the logit link function, which is the logarithm of the odds of the probability as expressed in Eq.(1).
log i 0 1ti 1 i
(1)
where 0 is an intercept .The results were described in the form of Odds Ratio (OR) as in the logistic regression β1 = ln OR . In this model, an OR > 1 means that the probability to be resistant increases with time and an OR < 1 means it decreases with time. GEE was used to perform multivariate analysis and take into account possible correlations between antimicrobials through the specification of one of a variety of possible working correlation matrix structures. An unstructured working correlation matrix was used, as correlations between any two responses are unknown and need to be estimated. For multi-resistance a logistic model was used to check whether there was a significant trend (increase or decrease) over the years regarding the prevalence of multi-resistant strains, for each animal category. In this model an OR >1 means that the probability for a strain to be multi-resistant increases with time.
9
In order to take into account the multiple comparisons/testing performed in this study, two correction methods were used to adjust the 5% significance level: the Bonferroni correction and the linear step-up method of Benjamini and Hochberg (1995) based on the control of the false discovery rate. The original p values refer to the effects of each different antimicrobial assessed separately. The adjusted p values provide a family wise significance level in order to make a statement on the entire pool of antimicrobials jointly. In the results of this study, adjusted p values obtained with both correction methods were compared to the original uncorrected p values.
RESULTS Resistance to specific antimicrobials Descriptive statistics A total of 3198 isolates analysed for antimicrobial susceptibility were included in the data set. The distribution of the number of isolates per year and per livestock category with the corresponding prevalence of isolates resistant to at least one antimicrobial are presented in Table 2. Detailed data and descriptive statistics for the years 2011-2013 are reported in the annual reports published by CODA-CERVA (Butaye, 2011, 2012, 2013). High prevalence values (> 50%) of resistant E. coli strains were observed during the study period for several antimicrobials in all animal categories except in beef cattle: for ampicillin in veal calves and chickens, for ciprofloxacin and nalidixic acid in chickens, for sulfamethoxazole in veal calves, chickens and pigs and for tetracycline in veal calves. In broiler chickens, a resistance prevalence > 40% was observed each year for more than half of
10
the antimicrobials tested (6/11) (Table 3). The resistance prevalence to ciprofloxacin was > 40% in veal calves in 2011 and 2012 and remained >20% in 2013 and 2014 while in beef cattle and veal calves it remained < 20% during the whole study period. The resistance prevalence to cephalosporins (cefotaxime and ceftazidime) was < 5% during the whole study period for veal calves, beef cattle and pigs except in 2012 for veal calves and beef cattle. In broiler chickens it was much higher than in other species, especially in 2011/2012 with prevalence values between 15 and 30%, which decreased later in 2013/2014 to below 10%. The prevalence of resistance to chloramphenicol remained above 20% during the whole study period in all livestock categories except in beef cattle for which it was close to 15%. Trend analysis All trends observed for the four livestock categories are summarized in Table 3. A mention is made when the resistance prevalence remained high during the whole study period. A significant decreasing resistance trend was observed for several antimicrobials in all four livestock categories: in veal calves for all tested antimicrobials except for gentamicin and tetracyclin (9/11), in beef cattle for ampicillin and nalidixic acid (2/11), in chickens for ampicillin, cefotaxime, sulfamethoxazole, ceftazidime, tetracycline and trimethoprim (6/11) and in pigs for ciprofloxacin, cefotaxime, nalidixic acid, ceftazidime and tetracycline (5/11). However, a significant increasing resistance trend was observed for ciprofloxacin in chickens. When looking at the annual resistance prevalence values for each antimicrobial in the different livestock species, it was noticed that in several cases, after an increase in 2012 compared to 2011, the prevalence decreased in 2013 and 2014 (data not shown), resulting in an overall significant decreasing trend for the study period. The same pattern was observed for ciprofloxacin in chickens but the decrease of resistance prevalence in 2013 (74.8%; 95% C.I. [68.8-80.0]) and 2014 (69.6%; 95% C.I. [61.9-76.3]) was not sufficient to compensate
11
for the previous increase of 2012 (79.6%; 95% C.I. [74.2-83.2]) compared to 2011 (62.9%; 95% C.I. [58.1-67.4]), resulting in an overall increasing trend in the study period. The trends observed with the univariate and multivariate analysis are presented, in figure 1 and 2 respectively (uncorrected p value) for each animal category and for each antimicrobial. When adjusting the p values due to multiple testing, using the linear step-up method, all trends remained significant except in beef cattle; however, when adjusting the p values using the Bonferroni method, several trends became insignificant (Table 3). Multi-resistance Multi-resistance was defined as resistance by an isolate to at least three antimicrobials of the panel, belonging to different antimicrobial classes : ampicillin (penicillins class), cefotaxime and/or ceftazidime (cephalosporins class), chloramphenicol (phenicols class), ciprofloxacin and/or acid nalidixic (quinolones class), colistine (polymyxins class), gentamicin (aminoglycosides class), sulfamethoxazole (sulphonamides class), tetracycline (tetracyclines class) and trimethoprim (trimethoprim class). The prevalence of multi-resistance remained above 40% during the whole study period for all livestock categories except for beef cattle; the highest levels of multi-resistance were found in broiler chickens (> 60%) (Fig 3a). Significant decreasing trends in the prevalence of multi-resistant E. coli strains were observed in veal calves and in broiler chickens in the study period (Fig. 3b). The comparison of diversity indices (weighted entropy) shows that multi-resistance against a large number of antibiotics was most frequent in E. coli from veal calves and broiler chickens and least frequent in isolates from fattening pigs; weighted entropy index decreased (i.e. multi-resistance shifted to a lower number of antimicrobials) in 2013 and 2014 in veal calves, broiler chickens and pigs (Fig. 4).
12
DISCUSSION Persisting high levels of resistance to some specific antimicrobials were observed continuously during this 4-year study period for several livestock categories. This was the case for ampicillin, ciprofloxacin, nalidixic acid, sulfamethoxazole and tetracycline, for which the prevalence of resistant E. coli strains remained continuously high (> 50%) in veal calves, chickens and pigs. Such high levels of resistance have also been observed previously, both in pathogenic and non-pathogenic strains of E. coli isolated from livestock, in other studies and other monitoring programmes implemented in several European countries (AFSSA, 2009, EFSA/ECDC, 2015, CVI, 2014, Garcia-Migura, 2014, Hendriksen et al, 2008a, Hendriksen et al, 2008b, Persoons et al., 2010). Therefore, the use of such antimicrobials should be carefully monitored, especially in livestock species with more intensive treatment practices and more frequent antimicrobial use (veal calves, broiler chickens, slaughter pigs), for which the highest resistance prevalence values were observed in this study. Indeed several studies have shown that higher exposure of livestock to antimicrobials is associated with higher resistance prevalence in commensal E. coli (Berge et al., 2005, Bosman et al., 2014, Burow et al., 2014, Li et al., 2014, Simoneit et al., 2014). The highest levels of antimicrobial resistance revealed by this study were found in E. coli strains isolated from broiler chickens: in this animal species, a high prevalence (>50%) of resistance to more than half of the tested antimicrobials was observed each year during the study period. In addition it was shown that the level of multi-resistance in E. coli isolates was the highest in broiler chickens compared to other livestock categories. The explanation for the difference in terms of prevalence of antimicrobial resistance between livestock categories is possibly due to different antimicrobial treatment regimens (group vs individual and oral (feed) vs parenteral treatment) following various management and animal husbandry
13
practices. Moreover, in poultry farming, broiler chickens are slaughtered very young, at the age of 6 weeks, when they harbour more resistant bacteria than older animals (Butaye et al., 1999). Much higher levels of resistance to cephalosporins were observed in chickens compared to other livestock categories, especially in 2011/2012. Such an observation has already been reported in other studies (Persoons et al., 2010; Persoons et al., 2011a). A possible explanation is the off label use of celtiofur sprayed on 1-day old chicks, a practice in use in some hatcheries in Belgium until 2010, when it came to an end after enforced veterinary inspection of hatcheries was carried out by FASFC (FASFC, 2011, p.252). Another hypothesis for such high levels of resistance to cephalosporins observed in chickens is the exposure of 1-day old chicks to antimicrobials before they are imported into Belgium. Indeed, 15 million 1-day chicks were imported in 2014 from 7 different EU countries, of which 80% came from the Netherlands and France (FASFC, personal communication). However the use of cephalosporins in poultry has been banned in the EU since 2010 (Council regulation EC 470/2009 and Commission regulation EC 37/2010), therefore this exposure is now unlikely to happen. The interpretation of the resistance prevalence must be specific to each antimicrobial and take into consideration not only the risk of therapeutic failure in veterinary medicine but also the possible impact on public health in case resistant strains encountered in animals spread to humans. Antimicrobials belonging to the (fluoro)quinolones and cephalosporins of 3rd and 4th generation classes (represented in this trend analysis respectively by ciprofloxacin on one hand and by cefotaxime and ceftazidime on the other hand) are listed among substances considered as critically important drugs to human medicine by WHO and critically important drugs to animal heath by OIE. These lists are based on criteria such as the importance of the drug for therapeutic use in serious infections in human and in veterinary medicine or as the absence of alternative treatment (OIE, 2015, WHO, 2012, WHO, 2014). This categorization 14
of critically important drugs has been subsequently recognised by other international institutions involved in public and animal health (ECDC/EFSA/EMA, 2015, EFSA/ECDC 2015, EMA, 2014, FAO, 2008). Therefore, a moderate (10 – 20%) prevalence of resistance to these critically important antimicrobials can be considered as more worrisome than the high prevalence (>50%) observed for other antimicrobials such as tetracycline which are less important in human therapy and for which alternative antimicrobials are available. Both statistical models (logistic regression and GEE) used in this study show significant decreasing trends of the resistance prevalence in all livestock categories for several antimicrobials. The comparison of annual data shows that the decrease in most cases started in 2013 and continued in 2014 after a rise in 2012 compared to 2011. This pattern of an increasing trend followed by a decreasing one was confirmed and found statistically significant when using non-linear models for resistance prevalence to chloramphenicol, ciprofloxacin, colistin and nalidixic acid in chickens (data not shown). The only increasing trend detected in this study was for resistance to ciprofloxacin in chickens but as mentioned in the results this was due to an important increase in prevalence in 2012 (+16,2%) followed by a moderate decrease in 2013 and 2014 (-9.4%). Data collected in the coming years will provide a better estimation of the actual evolution. The detection of significant decreasing antimicrobial resistance and decreasing multi-resistance in poultry and in veal calves, for which the prevalence of resistance was highest and therefore easiest to detect, may indicate that there is indeed an actual decrease in the overall resistance prevalence amongst commensal E. coli in Belgian livestock. There are several possible explanations for an increase or a decrease of antimicrobial resistance in a given population. The most obvious is the selection pressure on bacteria when exposed to antimicrobials. Indeed, significant correlations or positive associations between 15
veterinary antimicrobial use and antimicrobial resistance in food-producing animals have already been demonstrated at the national and European level (Chantziaras et al., 2014, ECDC/EFSA/EMA, 2015, Garcia-Migura, 2014) as well as at farm level (Bosman et al., 2014). However other explanations have also been put forward to explain short-term variations in resistance: e.g. the spread of a more vital clone with Salmonella serotypes (Bertrand et al., 2006, Butaye et al., 2006, Imberechts et al., 2000), the horizontal transfer of extended-spectrum beta-lactamases (ESBL) resistance genes without selective pressure in animal strains of E. coli (Smet et al.2011) or the rapid spread of methicillin resistant Staphylococcus aureus (MRSA) in pigs without any antimicrobial selective pressure (Crombé et al., 2011). Resistance may also be co-selected by the use of totally unrelated antibiotics, giving a discrepancy between the presence of resistance and the actual usage of antimicrobials. This phenomenon was observed with the persistence of vancomycin resistance in pigs after its ban as a growth promoter, due to a physical link to a macrolide resistance gene (Bager et al., 1999). It was also seen in the persistence of chloramphenicol resistance without resistance to florfenicol due to resistance genes found in multi-resistance plasmids and in integrons, harbouring also other resistance genes (Schwarz, 2004). In our study we observed indeed that resistance to chloramphenicol is still present, although this antimicrobial has been prohibited for veterinary use in Europe since January 1997 (Council Regulation (EEC) 2377/90 and amendments 1570/98 and 508/1999). A significant decreasing trend for the resistance prevalence was observed in veal calves while no trend was observed in other livestock categories. Nevertheless, the prevalence of resistance to chloramphenicol remained above 20% of E. coli isolates for the whole study period in all livestock categories except in beef cattle for which it was close to 15%. Suspected illegal use of this drug by livestock keepers can be ruled out as an explanation of the persistence of such resistance levels. Indeed there is 16
an ongoing official monitoring programme in Belgium for the detection of chloramphenicol residues (Royal decree 2013/18154) at the slaughter house (injection sites) and at farm level (urine and faecal samples) (FASFC, 2014). The detection of such residues are extremely rare: between 2011 and 2014, 1626 samples were taken at farm-level or at the slaughter house from the various livestock categories (cattle, n= 198; veal calves, n= 89; pigs, n=463; broiler chickens, n= 876) and only one sample was detected with very low levels of chloramphenicol residues, at a concentration much below the authorized EU reference limit (FASFC, personal communication, 2015). Cross-resistance with other phenicol antimicrobials (e.g. florfenicol), which are authorized for veterinary use, is not an explanation either since most of the strains are only resistant to chloramphenicol indicating that the mechanism involved relies on the persisting circulation of ‘old’ resistance genes (Kim and Aoki, 1996, Schwartz et al., 2004). The most probable explanation for the persistence of chloramphenicol resistance is the mechanism of co-selection which selects bacterial strains resistant to several antimicrobials due to the clustering of several resistance genes at the same genetic locus in the genome (plasmids, transposons or integrons). This phenomenon was reported in Salmonella spp. and in vitro for E. coli (Barraud and Ploy, 2015, Hall, 2010, Montero 2013). A decrease in consumption of antimicrobial pharmaceuticals for veterinary use has been reported in Belgium between 2010 and 2013 (AMCRA, 2014, EMA, 2014). While the total biomass of livestock has not changed much during this period in Belgium, the total sales in tons based on the figures recorded by the Belgian Federal Agency for Medicines and Health Products (FAMHP) and reported by the Belgian Center of Expertise on Antimicrobial Consumption and Resistance in Animals (AMCRA) have decreased by 13.3 % in 2013 as compared to 2010. The decrease was observed for all antimicrobial classes (-6.8% to -53.5 % according to the antimicrobial classes), except for cephalosporins (+12.5%) and for phenicols (+9.4%); a decrease was also observed for the penicillin class between 2013 and 2011 (-8%). 17
However, these figures represent the overall antimicrobial consumption for veterinary use but the consumption stratified by animal species is currently unavailable. Hence it is impossible to correlate directly the observed decreasing trends of antimicrobial resistance in the different livestock categories to the decreasing consumption of antimicrobials for veterinary use. This shows the necessity in the future that animal health authorities collect data about antimicrobial consumption at the national level by animal species and not only bulk figures. Efforts are currently underway in Belgium in the pig and poultry sector to better record antimicrobial use at the farm level and a national data collection system is planned for 2016. Since 2012, recommendations have been provided by AMCRA to farmers and veterinarians, to promote the responsible use of antimicrobials with the objective to reach a substantial reduction by 2020 (-50% compared to 2011). Although the actual impact of this intervention has not been formally assessed, one can expect that it has contributed to a decrease of antimicrobial consumption and indirectly to a decrease of antimicrobial resistance in the commensal flora due to a lower exposure of livestock to antimicrobials. Similar observations of decreasing prevalence of antimicrobial resistance in livestock following an effective decreasing consumption of antimicrobial veterinary drugs has already been reported in other European countries such as the Netherlands and Denmark (CVI, 2014; DANMAP, 2014). This trend analysis of antimicrobial resistance in commensal E. coli, isolated from different livestock species, was based upon data collected over a 4-year period of national monitoring in Belgium (2011-2014). Though EFSA performs trend analysis on data provided by EU Member States after at least 5 years of monitoring (EFSA/ECDC, 2015), it was decided to undertake this analysis after 4 years as it appeared to be sufficient to detect some significant trends. These results will be confirmed or adjusted when more data are available in the coming years. Besides the time period, the sample size is also an important parameter that determines the power and robustness of this trend analysis. In most of the cases a sufficient 18
number of E. coli strains were isolated as proposed by EFSA (> 170 isolates/year/animal category) in order to detect annual trends (EFSA, 2008). However, in 2011 and 2014 the number of bacterial isolates per year was lower for some of the animal categories (Table 1), leading to larger uncertainty in the estimation of the prevalence of resistant strains and therefore decreasing the power to detect some trends. For this reason, it is recommended that animal health authorities shall implement and control a harmonised sampling protocol (sample size, sampling method and sample matrix) during further monitoring of E. coli in the different livestock species, following EU specifications. In this study, several dependent or independent statistical tests were performed simultaneously on a single data set. Therefore, p values were adjusted using correction methods for multiple testing/comparison, such as the Bonferroni method and the linear stepup method of Benjamini and Hochberg. The use of the correction methods for the p values is conservative as it decreases the risk of detecting a trend when in reality there is no trend (false positive results: type I error). However, such correction methods might also decrease the power of the statistical tests and prevent the detection of truly existing trends (false negative results: type II error). The Bonferroni correction is a well-known but very conservative method by which the significance level (α) is divided by the number of comparisons being made (Bland and Altman, 1995). It is usually not recommended when a large number (>5) of hypotheses are being tested. The Benjamini and Hochberg linear stepup method is less conservative and more appropriate to the current analysis and data set. The interpretation of the adjusted p values depends on the attitude of decision-makers. If the aim is to identify possible minor increasing trends of antimicrobial resistance because these are considered as a source of concern for public health (e.g. for critically important antimicrobials), one would try to minimize the false negatives and just consider the original p values. On the contrary, if false negatives can be accepted and the interest is in looking at 19
“strong” trends, then the linear step-up correction method should be used in order to avoid excessive sensitivity. Some biases may have affected the results of this study. For veal calves, a possible bias is a change in the sampling method in 2012 using faecal samples taken at the slaughter house from one single animal /farm (average age at slaughter 7-8 months) instead of pooling faecal samples taken at the farm from 10 animals younger than 7 months, as was done in 2011. The use of pooled faecal samples is known to be an appropriate sampling strategy to estimate the prevalence of antimicrobial resistance compared to individual samples (Bosman et al., 2012). However, several studies showed that optimal precision in the estimation of the resistance prevalence in a study population can be reached by testing one isolate obtained from one animal of each herd/flock sampled (Butaye et al., 1999, Persoons et al., 2011b, Yamamoto et al., 2014). In 2011, pooled samples were used but only one E. coli isolate was taken from each pool therefore the change to individual samples in 2012 probably did not affect the detection of resistant strains compared to the previous year. On the other hand, the age of sampled animals might also have affected the resistance prevalence: in 2011 samples were taken from calves slightly younger than in the following years and it is well known that the prevalence of resistance is generally higher among isolates of younger animals, whatever the species (Berge et al., 2010, Butaye et al., 1999, Langlois, 1988). Therefore the change in the sampling method should have resulted in a reduced prevalence of the resistance in 2012 for veal calves, as samples were taken from older animals. Yet, the opposite was observed, possibly due to a lack of precision of the estimated prevalence in 2011 because of an insufficient number of samples during that specific year (31 isolates). Finally, it should be mentioned that the observation of high levels of antimicrobial resistance or of increasing/decreasing trends does not necessarily reflect the efficacy of antimicrobial
20
treatment at field level. No information was available about an increase or decrease in treatment failures observed by farmers or veterinarians to support the observations made in our study. This aspect should be investigated in future studies to correlate antimicrobial resistance observed in vitro with field observations. In conclusion, after 4 years of monitoring of antimicrobial resistance in commensal E. coli isolates from different livestock species, descriptive statistics showed persisting high levels of resistance (>50%) for several antimicrobials in all livestock categories except in beef cattle. Particularly relevant was the high level of resistance to ciprofloxacin observed in E. coli strains isolated in broiler chickens. Moreover, the methodology and statistical models applied in this study allowed the detection of significant trends, mostly decreasing, in the prevalence of antimicrobial resistance. These results however need to be confirmed and maybe adjusted by data covering a longer period. The latest figures released by AMCRA indicate a slight increase (+ 1.1%) in the overall consumption/ kg of animal biomass of antimicrobials for veterinary use in Belgium for the year 2014 compared to 2013 (AMCRA, 2014). Therefore one can expect a possible increase in the prevalence of antimicrobial resistance in 2015 in Belgian livestock and a change in the trends observed previously. It is therefore recommended that the official monitoring programme should continue in order to provide sound data, relevant results and scientific information about the level of antimicrobial resistance to the public health and animal health authorities as well as to the stakeholders of the livestock sector. Such information can support decision-makers to provide guidelines and take appropriate control measures aimed at reducing the use of antimicrobials in livestock. However such measures should also take into consideration resistance mechanisms and the risk for animal and human health associated with specific antimicrobial classes. The interpretation of the observed trends should not be restricted to statistical
21
analysis; it should also consider biological aspects and be discussed and validated by experts on the topic. Another possible outcome of the continuous monitoring of antimicrobial resistance is the evaluation of the efficacy and of the impact of actions taken to decrease the prevalence of antimicrobial resistance. Given the close link observed and demonstrated elsewhere between antimicrobial consumption and antimicrobial resistance it is advised to integrate both monitoring programmes in the future in order to correlate those data.
ACKNOWLEDGEMENTS This study was commissioned by the Belgian Federal Agency for the Safety of the Food chain (FASFC) and carried out by the epidemiology unit and the bacteriology department of CODA-CERVA (Veterinary and Agrochemical Research Centre, Brussels) in close collaboration with the Center for Statistics (CenStat, Hasselt University, Belgium) which developed the statistical models adapted to the available data and the expected outputs. Our thanks also go to the staff of the veterinary laboratories of the two Belgian regional animal health association (ARSIA and DGZ) for their work done to isolate the bacterial strains.
22
REFERENCES AFSSA, 2009. Farm 2005-2006: French antimicrobial resistance monitoring program in bacteria of animal origin. Available at http://www.anses.fr/Documents/SANT-RaFARM2006.pdf AMCRA, 2014. BelVet-Sac, Belgian Veterinary Surveillance of Antibacterial Consumption, National consumption report 2013. Available at http://www.amcra.be/en/rapportenpublicaties-en-wetgeving/national-reports AMCRA, 2015. BelVet-Sac, Belgian Veterinary Surveillance of Antibacterial Consumption, National consumption report 2014 Andrews, JM., 2001. Determination of minimum inhibitory concentrations. J Antimicrob Chemother. 2001 Jul; 48 Suppl 1:5-16 Bager, F., Aarestrup, FM., Madsen, M, Wegener, HC., 1999. Glycopeptide resistance in Enterococcus faecium from broilers and pigs following discontinued use of avoparcin. Microb Drug Resist. 1999 Spring; 5(1):53-6. Barraud, O., Ploy, MC., 2015. Diversity of Class 1 Integron Gene Cassette Rearrangements Selected Under Antibiotic Pressure. J Bacteriol. 2015 Apr 20. pii: JB.02455-14. [Epub ahead of print] Benjamini, Y. and Hochberg, Y., 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Statist Soc Ser B (Methodological). 57, 289-300.
23
Berge, A.C., Atwill, E.R., Sischo, W.M., 2005. Animal and farm influences on the dynamics of antibiotic resistance in faecal Escherichia coli in young dairy calves. Prev Vet Med. 2005 Jun 10;69(1-2):25-38. Epub 2005 Feb 26. Berge, A.C., Hancock, D.D., Sischo, W.M., Besser, T.E.. 2010. Geographic, farm, and animal factors associated with multiple antimicrobial resistance in fecal Escherichia coli isolates from cattle in the western United States. J Am Vet Med Assoc. 2010 Jun 15;236(12):1338-44. doi: 10.2460/javma.236.12.1338. Bertrand, S., Weill, FX, Cloeckaert, A., Vrints, M., Mairiaux, E., Praud, K., Dierick, K., Wildemauv,e C., Godard, C., Butaye, P., Imberecht,s H., Grimont, PA., Collard JM., 2006. Clonal emergence of extended-spectrum beta-lactamase (CTX-M-2)-producing Salmonella enterica serovar Virchow isolates with reduced susceptibilities to ciprofloxacin among poultry and humans in Belgium and France (2000 to 2003). J Clin Microbiol. 2006 Aug; 44(8):2897-903. Bland, J.M., Altman, D.G., 1995. Multiple significance tests: The Bonferroni method. BMJ. 310 (6973):170. Bosman, AB., Wagenaar, JA., Stegeman, A., Vernooij, H., Mevius, D., 2012. Quantifying antimicrobial resistance at veal calf farms. PLoS One. 2012; 7(9):e44831. doi: 10.1371/journal.pone.0044831. Bosman, AB., Wagenaa,r JA., Stegeman, JA., Vernooij, JC., Mevius, DJ., 2014. Antimicrobial resistance in commensal Escherichia coli in veal calves is associated with antimicrobial drug use. Epidemiol Infect. 2014 Sep; 142(9):1893-904. doi:10.1017/S0950268813002665.
24
Burow, E., Simoneit, C., Tenhagen, BA., Käsbohrer, A., 2014. Oral antimicrobials increase antimicrobial resistance in porcine E. coli--a systematic review. Prev Vet Med. 2014 Mar 1; 113(4):364-75. doi: 10.1016/j.prevetmed.2013.12.007. Epub 2013 Dec 27. Review Butaye, P., Devriese, L.A., Goossens, H., Ieven, M., Haesebrouck, F., 1999. Enterococci with acquired vancomycin resistance in pigs and chickens of different age groups. Antimicrob Agents Chemother. 1999 Feb;43(2):365-6 Butaye, P., Michael, GB., Schwarz, S., Barrett, TJ., Brisabois, A., White, DG., 2006. The clonal spread of multidrug-resistant non-typhi Salmonella serotypes. Microbes Infect. 2006 Jun;8(7):1891-7. Epub 2006 Mar 22. Review. Butaye, P., unpublished results. Annual Reports on antimicrobial resistance in commensal E. coli from poultry, pigs, cows and veal calves. (2011, 2012, 2013). Available at http://www.coda-cerva.be Chantziaras, I., Boyen, F., Callens, B., Dewulf, J., 2014. Correlation between veterinary antimicrobial use and antimicrobial resistance in food-producing animals: a report on seven countries. J Antimicrob Chemother. 2014 Mar; 69(3):827-34. doi: 10.1093/jac/dkt443. Crombé, F., Willems, G., Dispas, M., Hallin, M., Denis, O., Suetens, C., Gordts, B., Struelens, M., Butaye, P., 2012. Prevalence and antimicrobial susceptibility of methicillin-resistant Staphylococcus aureus among pigs in Belgium. Microb Drug Resist. 2012 Apr; 18(2):125-31. doi: 10.1089/mdr.2011.0138. Epub 2011 Nov 16 CVI (Central Veterinary Institute of Wageningen), 2014. MARAN 2014: Monitoring of antimicrobial resistance and antibiotic usage in animals in the Netherlands in 2013. Central Veterinary Institute, Wageningen. Available at www.cvi.wur.nl 25
DANMAP, 2014. DANMAP 2013 (Danish Integrated Antimicrobial Resistance Monitoring and Research Programme): Use of antimicrobial agents and occurrence of antimicrobial resistance in bacteria from food animals, food and humans in Denmark. ISSN 1600-2032. Available at http://www.danmap.org ECDC/EFSA/EMA, 2015. ECDC (European Centre for Disease Prevention and Control), EFSA (European Food Safety Authority) and EMA (European Medicines Agency) first 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. EFSA Journal 2015; 13(1):4006, 114 pp. doi:10.2903/j.efsa.2015.4006 EFSA (European Food Safety Authority), 2008. Report from the Task Force on Zoonoses Data Collection including guidance for harmonized monitoring and reporting of antimicrobial resistance in commensal Escherichia coli and Enterococcus spp. from food animals. EFSA Journal (2008) 141, 1-44 EFSA (European Food Safety Authority), 2014. Technical specifications on randomised sampling for harmonised monitoring of antimicrobial resistance in zoonotic and commensal bacteria. EFSA Journal 2014; 12(5):3686, 33 pp. doi:10.2903/j.efsa.2014.3686 EFSA/ECDC, 2015. The European Union summary report on antimicrobial resistance in zoonotic and indicator bacteria for humans, animals and food in the EU in 2013. EFSA Journal 2015; 13(2):4036, 178 pp., doi:10.2903/j.efsa.2015.4036 EMA, 2014. European Medicines Agency. Sales of veterinary antimicrobial agents in 26 EU/EEA countries in 2012, fourth European Surveillance of Veterinary Antimicrobial Consumption (ESVAC) report. (EMA/333921/2014). 26
EUCAST (European Committee on Antimicrobial Susceptibility Testing), 2014. Data from the EUCAST MIC distribution website. Available at http://www.eucast.org/mic_distributions/ FAO, 2008. Joint FAO/WHO/OIE Expert Meeting on Critically Important Antimicrobials. Report of the FAO/WHO/OIE Expert Meeting FAO Headquarters, Rome 26-30 November 2007. Available at http://www.fao.org/publications/card/en/c/3c566f76-caa85125-bb06-19f08dc27550/ FASFC, 2011. Federal Agency for the safety of the Food Chain, Rapport d’activités 2010. Available at http://www.afsca.be/rapportsannuels/ FASFC, 2014. Federal Agency for the safety of the Food Chain, Rapport d’activités 2013. Available at http://www.afsca.be/rapportsannuels/ Garcia-Migura, L., Hendriksen, R.S., Fraile, L., Aarestrup, F.M., 2014. Antimicrobial resistance of zoonotic and commensal bacteria in Europe: the missing link between consumption and resistance in veterinary medicine. Vet. Microbiol. 2014 May 14; 170 (1-2):1-9. doi: 10.1016/j.vetmic.2014.01.013. Epub 2014 Feb 3. Review. Geenen, P.L., Koene, M.G.J., Blaak, H., Havelaar, A.H. and van de Giessen, A.W., 2010. Risk profile on antimicrobial resistance transmissible from food animals to humans. RIVM-rapport 330334001, 2010. Available at http://www.rivm.nl Guiasu, S. (1971) Weighted Entropy. Reports on Mathematical Physics, 2, 165-179. Hall, RM., 2010. Salmonella genomic islands and antibiotic resistance in Salmonella enterica. Future Microbiol. 2010 Oct; 5(10):1525-38. doi: 10.2217/fmb.10.122. Review.
27
Hendriksen, R.S., Mevius, D.J., Schroeter, A., Teale, C., Jouy, E., Butaye, P., Franco, A., Utinane, A., Amado, A., Moreno, M., Greko, C., Stark, K.D., Berghold, C., Myllyniemi, A.L., Hoszowski, A., Sunde, M. and Aarestrup, F.M., 2008a. Occurrence of antimicrobial resistance among bacterial pathogens and indicator bacteria in pigs in different European countries from year 2002 - 2004; the ARBAO-II study. Acta Vet. Scand. 50, 19 Hendriksen, R.S., Mevius, D.J., Schroeter, A., Teale, C., Jouy, E., Butaye, P., Franco, A., Utinane, A., Amado, A., Moreno, M., Greko, C., Stark, K.D., Berghold, C., Myllyniemi, A.L., Wasyl, D., Sunde, M. and Aarestrup, F.M., 2008b. Prevalence of antimicrobial resistance among bacterial pathogens isolated from cattle in different European countries: 2002-2004. Acta Vet. Scand. 50, 28 Imberechts, H., D'hooghe, I., Bouchet, H., Godard, C., Pohl, P., 2000. Apparent loss of enrofloxacin resistance in bovine Salmonella typhimurium strains isolated in Belgium, 1991 to 1998. Vet Rec. 2000 Jul 15; 147(3):76-7. Kim, E., Aoki, T., 1996. Sequence analysis of the florfenicol resistance gene encoded in the transferable R-plasmid of a fish pathogen, Pasteurella piscicida. Microbiol Immunol. 1996; 9: 665–669 Langlois, BE., Dawson, KA., Leak, I., Aaron, DK., 1988. Effect of age and housing location on antibiotic resistance of fecal coliforms from pigs in a non-antibiotic-exposed herd. Appl Environ Microbiol. 1988 Jun; 54(6):1341-4. Li, P., Wu, D., Liu, K., Suolang, S., He, T., Liu, X., Wu, C., Wang, Y., Lin, D., 2014. Investigation of antimicrobial resistance in Escherichia coli and enterococci isolated
28
from Tibetan pigs. PLoS One. 2014 Apr 18; 9(4):e95623. doi: 10.1371/journal.pone.0095623. eCollection 2014 Montero, I., Herrero-Fresno, A., Rodicio, R., Rodicio, MR., 2013. Efficient mobilization of a resistance derivative of pSLT, the virulence plasmid specific of Salmonella enterica serovar Typhimurium, by an IncI1 plasmid. Plasmid. 2013 Jul; 70(1):104-9. doi:10.1016/j.plasmid.2013.03.002. Epub 2013 Mar 27. Persoons, D., Dewulf, J., Smet, A., Herman, L., Heyndrickx, M., Martel, A., Catry, B., Butaye, P., Haesebrouck, F., 2010. Prevalence and persistence of antimicrobial resistance in broiler indicator bacteria. Microb Drug Resist. 2010 Mar; 16(1):67-74. doi:10.1089/mdr.2009.0062 Persoons, D., Haesebrouck, F., Smet, A., Herman, L., Heyndrickx, M., Martel, A., Catry, B., Berge, A.C., Butaye, P., Dewulf, J., 2011. Risk factors for ceftiofur resistance in Escherichia coli from Belgian broilers. Epidemiol Infect. 2011 May;139(5):765-71. doi:10.1017/S0950268810001524. Epub 2010 Jun 29. Persoons, D., Bollaerts, K., Smet, A., Herman, L., Heyndrickx, M., Martel, A., Butaye, P., Catry, B., Haesebrouck, F., Dewulf, J., 2011. The importance of sample size in the determination of a flock-level antimicrobial resistance profile for Escherichia coli in broilers. Microb Drug Resist. 2011 Dec;17(4):513-9. doi: 10.1089/mdr.2011.0048 Schwarz, S., Kehrenberg, C., Doublet, B., Cloeckaert, A., 2004. Molecular basis of bacterial resistance to chloramphenicol and florfenicol. FEMS Microbiol Rev. 2004 Nov; 28(5):519-42. Review. Simoneit, C., Burow, E., Tenhagen, BA., Käsbohrer, A., 2014. Oral administration of antimicrobials increase antimicrobial resistance in E. coli from chicken--a systematic 29
review. Prev Vet Med. 2015 Jan 1; 118(1):1-7. doi: 10.1016/j.prevetmed.2014.11.010. Epub 2014 Nov 20 Smet, A., Rasschaert, G., Martel, A., Persoons, D., Dewulf, J., Butaye, P., Catry, B., Haesebrouck, F., Herman, L., Heyndrickx, M., 2010. In situ ESBL conjugation from avian to human Escherichia coli during cefotaxime administration. J Appl Microbiol. 2011 Feb; 110(2):541-9. doi: 10.1111/j.1365-2672.2010.04907.x. Epub 2010 Dec 10. Szmolka, A., Nagy, B., 2013. Multidrug resistant commensal Escherichia coli in animals and its impact for public health. Front Microbiol. 2013 Sep 3; 4: 258. doi: 10.3389/fmicb.2013.00258. OIE, 2015. World Organisation for Animal Health: OIE list of antimicrobials of veterinary importance, adopted in November 2007, latest update May 2015. Available at http://www.oie.int/en/our-scientific-expertise/veterinary-products/antimicrobials/ WHO, 2012. World Health Organisation: Critically important antimicrobials for human medicine - 3rd revision 2011. Available at http://apps.who.int/iris/bitstream/10665/77376/1/9789241504485_eng.pdf. 31. WHO, 2014. World Health Organisation: Antimicrobial resistance: global report on surveillance. Available at http://apps.who.int/iris/bitstream/10665/112642/1/9789241564748_eng.pdf?ua=1. Yamamoto, T., Hayama, Y., Hidano, A., Kobayashi, S., Muroga, N., Ishikawa, K., Ogura, A., Tsutsui, T., 2014. Sampling strategies in antimicrobial resistance monitoring: evaluating how precision and sensitivity vary with the number of animals sampled per farm. PLoS One. 2014 Jan 23;9(1):e87147. doi: 10.1371/journal.pone.0087147. eCollection 2014
30
Figure Captions
Veal Calves
Beef cattle
Chickens
Pigs
Fig. 1. Odds Ratios and their 95 % CI (univariate analysis: Logistic regression; uncorrected p value = 0.05) of the temporal trends to be resistant to a specific antimicrobial for E. coli isolates, per livestock category, between 2011 and 2014. OR > 1 indicates an increasing trend, OR < 1 indicates a decreasing trend. See legend for antimicrobials symbols in Table 1.
31
Fig. 2. Graphical representation (multivariate analysis: Generalised Estimating Equation model – GEE; uncorrected p value = 0.05) of the temporal trends of the probability to be resistant to a specific antimicrobial for E.coli isolates, per livestock category, between 2011 and 2014. Significant trends are in continuous lines. See legend for antimicrobials symbols in Table 1
32
100%
veal calves beef cattle chickens pigs
90% 80%
Prevalence
70% 60% 50% 40% 30% 20% 10% 0%
Fig. 3a. Prevalence (with 95% C.I.) of multi-resistant E. coli isolates, per animal category and per year. Multi-resistance was considered as resistance by an isolate to at least 3 antimicrobials belonging to different antimicrobial classes.
Fig.3b. Odds Ratios and their 95 % CI (univariate analysis: Logistic regression) of the temporal trends to be multi-resistant for E. coli isolates, per livestock category, between 2011 and 2014. OR > 1 indicates an increasing trend, OR < 1 indicates a decreasing trend.
33
weighted entropy
1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00
veal calves
2011
beef cattle
2012
broiler chickens slaughter pigs
2013
2014
Fig. 4. Weighted entropy indices per animal category and per year for E. coli isolates, describing the diversity of multi-resistance. A high weighted entropy index (closer to 1) means that overall, isolates of the corresponding year and animal category are generally resistant to a high number of antimicrobials.
34
Tables Table 1. Panel of 11 antimicrobials used for susceptibility testing of E. coli isolates during the study period (2011-2014). Antimicrobial Ampicillin Chloramphenicol Ciprofloxacin Colistin Cefotaxime Gentamicin Nalidixic acid Sulfamethoxazole Ceftazidime Tetracycline Trimethoprim
Symbol AMP CHL CIP COL FOT GEN NAL SMX TAZ TET TMP
35
Table 2. Number (N) of E. coli isolates tested for antimicrobial resistance for each year and each livestock category and corresponding percentage of isolates resistant to at least one antimicrobial (% Res)
2011 2012 2013 2014 Total
Veal calves N % Res 34 85.3 181 87.8 202 82.2 188 73.9 605 81.5
Beef cattle N % Res 154 38.3 175 52.0 204 37.3 164 28.7 697 39.2
Chickens N % Res 420 93.3 320 94.4 234 95.3 158 88.6 1132 93.4
Pigs N % Res 157 75.2 217 71.9 206 69.9 184 71.2 764 71.9
Total N % Res 765 78.2 893 79.3 846 72.0 694 65.9 3198 74.2
36
Table 3. Summary of the observed trends and of the persisting high prevalences of antimicrobial resistance for E. coli isolates, per livestock category from 2011 till 2014. All indicated trends (↑, ↓) were statistically significant (p = 0.05) both in univariate (logistic regression) and multivariate (GEE) analysis, even after using correction methods for multiple testing (Bonferroni and Linear step-up method), unless otherwise mentioned in the table 1, 2, 3. Antimicrobial Ampicillin Chloramphenicol Ciprofloxacin Colistin Cefotaxime Gentamicin Nalidixic acid Sufamethoxazole Ceftazidime Tetracycline Trimethoprim
veal calves
beef cattle 1, 2
↓ ++ ↓ ↓ ↓2 ↓
↓
↓ ↓ ++ ↓ ↓2, 3 ++ ↓
↓1, 2
chickens
pigs
2
↓ ++ ↑2 ++
↓
↓
↓2
++ ↓ ++ ↓ ↓+ ↓2 +
↓ ++ 2
↓ ↓2 + +
++ : Prevalence of antimicrobial resistance > 50% during 4 consecutive years + : Prevalence of antimicrobial resistance > 40% during 4 consecutive years
↑ : significant increasing trend of antimicrobial resistance prevalence ↓ : significant decreasing trend of antimicrobial resistance prevalence 1
: trend not significant after p value adjustment with linear step-up method : trend not significant after p value adjustment with Bonferroni method 3 : trend not significant in multivariate analysis (GEE) If no symbol is indicated for a corresponding antimicrobial and livestock category: no significant trend were detected and no persisting high prevalence of antimicrobial resistance during the 4 years period were observed 2
37