*Article* **Antimicrobial Resistance, Serologic and Molecular Characterization of** *E. coli* **Isolated from Calves with Severe or Fatal Enteritis in Bavaria, Germany**

**Andrea Feuerstein <sup>1</sup> , Nelly Scuda <sup>1</sup> , Corinna Klose <sup>1</sup> , Angelika Hoffmann <sup>1</sup> , Alexander Melchner <sup>2</sup> , Kerstin Boll <sup>1</sup> , Anna Rettinger <sup>1</sup> , Shari Fell <sup>3</sup> , Reinhard K. Straubinger <sup>4</sup> and Julia M. Riehm 2,\***


**Abstract:** Worldwide, enterotoxigenic *Escherichia coli* (ETEC) cause neonatal diarrhea and high mortality rates in newborn calves, leading to great economic losses. In Bavaria, Germany, no recent facts are available regarding the prevalence of virulence factors or antimicrobial resistance of ETEC in calves. Antimicrobial susceptibility of 8713 *E. coli* isolates obtained from 7358 samples of diseased or deceased diarrheic calves were investigated between 2015 to 2019. Considerably high rates of 84.2% multidrug-resistant and 15.8% extensively drug-resistant isolates were detected. The resistance situation of the first, second and third line antimicrobials for the treatment, here amoxicillinclavulanate, enrofloxacin and trimethoprim-sulfamethoxazole, is currently acceptable with mean non-susceptibility rates of 28.1%, 37.9% and 50.0% over the investigated 5-year period. Furthermore, the ETEC serotypes O101:K28, O9:K35, O101:K30, O101:K32, O78:K80, O139:K82, O8:K87, O141:K85 and O147:K89, as well as the virulence factors F17, F41, F5, ST-I and stx1 were identified in a subset of samples collected in 2019 and 2020. The substantially high rates of multi- and extensively drugresistant isolates underline the necessity of continuous monitoring regarding antimicrobial resistance to provide reliable prognoses and adjust recommendations for the treatment of bacterial infections in animals.

**Keywords:** *E. coli*; calves; enteritis; antimicrobial resistance; serotypes; virulence; multidrug-resistant; extensively drug-resistant

### **1. Introduction**

*Escherichia coli* account to the major enteric and systemic pathogens of the Gramnegative rods within the family Enterobacteriaceae. Most of the *E. coli* colonizing the intestinal tract of animals and humans are commensal, but facultative pathogenic strains may cause intestinal disorder or even severe and life-threatening extraintestinal disease [1,2]. In calves, enterotoxigenic *E. coli* (ETEC) pose a leading cause of intestinal disease, especially within the first four days of life [3–5]. ETEC encode lipopolysaccharide structures (LPS) that may act as endotoxins, fimbrial adhesins and finally enterotoxins. The endotoxins within the blood stream cause fever, damage of endothelial cells and disseminated intravascular coagulation (DIC), that leads to acute shock and sudden death [1]. The serological LPS characterization in calves comprise the *E. coli* serogroups O8, O9 and O101, and respective serotypes O9:K35 and O101:K30, as these are known for endotoxin effect [6]. Further, the

**Citation:** Feuerstein, A.; Scuda, N.; Klose, C.; Hoffmann, A.; Melchner, A.; Boll, K.; Rettinger, A.; Fell, S.; Straubinger, R.K.; Riehm, J.M. Antimicrobial Resistance, Serologic and Molecular Characterization of *E. coli* Isolated from Calves with Severe or Fatal Enteritis in Bavaria, Germany. *Antibiotics* **2022**, *11*, 23. https://doi.org/10.3390/ antibiotics11010023

Academic Editor: Clair L. Firth

Received: 23 November 2021 Accepted: 24 December 2021 Published: 27 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

serotype O78:K80 plays a major role in systemic disease, septicemia and endotoxic shock of newborn calves [1,6,7]. In piglets, the serotype O141:K85 in combination with F4 fimbria is specific for the postweaning diarrhea syndrome [6]. As well, three further serotypes O139:K82, O8:K87 and O147:K89 play an important role as pathogens for swine [6,8]. Proteinaceous fimbrial adhesins precipitate the bacterial attachment to the enteric mucosa that avert the mechanical shedding of virulent strains from the gut by peristalsis [1,4,9]. Former studies showed that the fimbrial adhesins F5, F17 and F41 are associated with calf diarrhea [4]. For ETEC, two different types of enterotoxins contribute to diarrhea in calves, the heat-stable toxin (ST) and heat-labile toxin (LT), respectively [1,10,11]. On a molecular level, the toxins increase the second messengers cyclic adenosine/ guanosine monophosphate (cAMP/cGMP), that effect an active secretion of fluid and electrolytes in the small intestine leading to extreme loss of fluid within the organism [11,12]. Further, ruminants are known to be a major reservoir of human pathogenic Shiga toxin-producing *E. coli* (STEC) [13–16]. Shiga toxins (stx1, stx2) may lead to enterocyte damage, subsequent bloody diarrhea and endothelial damage leading to internal hemorrhages and septicemia in susceptible neonatal calves [1,17,18]. Enterohemorrhagic *E. coli* (EHEC), a subset of STEC, further include intimin, an adhesin coded from the enterocyte effacement pathogenicity island (eaeA) [19,20] and enterohemolysin, a toxin encoded by the ehxA gene [21]. As published in several case reports, a majority of human EHEC disease outbreaks are caused by the serotype O157:H7 originating from contaminated ground beef [13,22,23]. This serotype is responsible for the hemorrhagic colitis and the life-threatening hemolytic uremic syndrome with the occurrence of thrombocytopenia, hemolytic anemia and thrombotic microangiopathy that may lead to acute renal failure and death [23–26].

Worldwide, neonatal diarrhea is still a major economic problem on cattle farms and the therapy with antimicrobials is crucial in routine practice [27]. However, the medication with bactericide antibiotics is solely, but highly indicated exclusively in the case of lifethreatening sepsis [28,29]. The Swiss antibiotic therapy guidelines for veterinarians recommend amoxicillin-clavulanate as a first line, sulfonamide-trimethoprim as a second line and fluoroquinolones as a third line choice, here enrofloxacin [29]. A study from 2014 revealed that veterinarians in Europe mainly used polymyxins (44%), (fluoro)quinolones (18%), penicillins (13%), aminoglycosides (9%) and third and fourth generation cephalosporins (8%) in calves with diarrhea emphasizing the problem of an inappropriate use of antibiotics [30]. This contributes to a higher level of antimicrobial resistant bacteria in young animals compared to adults [31–33]. In addition, the emergence of multidrug- and pandrug-resistant *E. coli* in fecal samples of diarrheic calves has been recently and repeatedly reported [33,34]. According to the expert proposal for standard definitions for acquired resistance from the European Centre for Disease Prevention and Control (ECDC), strains are classified as "multidrug-resistant" if these are non-susceptible (resistant or intermediate) to at least one antimicrobial agent in more than three categories. Isolates meet the definition "extensively drug-resistant" if these are non-susceptible in all agents but two or fewer categories. Finally, isolates non-susceptible to all agents in all antimicrobial categories are ranked as "pandrug-resistant" [35].

Previous data show that the prevalence of extended-spectrum β-lactamase (ESBL) producing *E. coli* in calves increased from 7% to 29% between 2006 and 2013 in Germany [27]. ESBL-producing strains do encode for numerous resistance genes and may transduce these to other, even commensal, bacteria [36]. Animals hosting these *E. coli* bacteria constitute a resistance gene reservoir that may affect the health of man and animals [36,37].

Only few data are available on the identification of ETEC from calves in Bavaria. However, the discrimination between the physiological intestinal flora and pathogenic *E. coli* is crucial [1,6,38]. The aim of the present study was to provide recent information about the most prevalent pathotypes of *E. coli*. These include the investigation of the current virulence factors, serotypes and trends in antimicrobial resistance [9,39–42].

#### **2. Results**

#### *2.1. Antimicrobial Susceptibility*

Within the study period 8713 *E. coli* were isolated from 7358 diarrheic calves at the federal state veterinary laboratory in Bavaria, Germany (Table S1). This number matches an average count of 1740 isolates per year that is in accordance with previous years (data not shown). The results on antimicrobial susceptibility testing revealed mean non-susceptibility values of 28.1% for amoxicillin-clavulanate, 37.9% for enrofloxacin and 50% for trimethoprim-sulfamethoxazole (Figures 1 and 2, Table S1). The highest nonsusceptibility value of a substance within each antimicrobial class revealed 11.9% for tulathromycin (macrolides), 18.3% for colistin (polymixins), 61.9% for tetracycline (tetracyclines), 62.2% for spectinomycin (aminoglycosides), 69.7% for ampicillin (penicillins), 80.5% for cephalothin (cephalosporins) and 96.8 % for florfenicol (phenicols) (Figure 1). A 5-year tendency from 2015 to 2019, evaluated for amoxicillin-clavulanate, enrofloxacin and trimethoprim-sulfamethoxazole, revealed a statistically significant decrease of the nonsusceptibility rates for amoxicillin-clavulanate and enrofloxacin (*p* < 0.05) (Figure 2, Table 1). Regarding trimethoprim-sulfamethoxazole a significant decrease was assessed from 51.9% to 47.8% between 2015 and 2017 regarding the non-susceptible *E. coli* isolates (*p* < 0.05). A subsequent increase was further revealed from 47.8% to 52.5% in the years 2017 to 2019 (*p* < 0.05) (Figure 2, Table 1). Categorizing the 8713 isolates according to the ECDC expert proposal, 84.2% of the isolates (7336/8713) were multidrug-resistant, 15.7% (1368/8713) were extensively drug-resistant, eight isolates (0.1%) were pandrug-resistant and one isolate was susceptible to all antimicrobials tested. As we only tested antimicrobials licensed for the veterinary use, and none of the latest antimicrobials available on the market, we rededicated the eight presumably pandrug-resistant as extensively drug-resistant summing up to 1376 isolates in this specification (Figure 3).

**Table 1.** Statistic parameters regarding the increase or decrease of resistance values within the five-year period for the three clinically relevant antimicrobials (Figure 2).


OR: odds ratio, CI: confidence interval, <sup>1</sup> *p*-value (Wald test) < 0.05.

*Antibiotics* **2022**, *11*, x FOR PEER REVIEW 4 of 17


**Figure 1.** Minimum inhibitory concentration (MIC) distribution of 8713 *E. coli* isolates on 12 antimicrobial agents from 11 antimicrobial classes. The three first lines represent the clinically relevant substances, first to third treatment choices in buiatrics. The red line demarcates the breakpoint to‐ wards resistance, the green line a breakpoint towards intermediate. Regarding the two combination compounds, only the concentration of the former substance is presented; the ratio of amoxicillin:clavulanic acid is 2:1 (1), concentration ratio of trimethoprim:sulfamethoxazole is 1:19 (2). Tulathro‐ mycin has not been tested in the first quarter of 2015 (3). The summation of intermediate and resistant isolates was named non‐susceptible (4). Some **Figure 1.** Minimum inhibitory concentration (MIC) distribution of 8713 *E. coli* isolates on 12 antimicrobial agents from 11 antimicrobial classes. The three first lines represent the clinically relevant substances, first to third treatment choices in buiatrics. The red line demarcates the breakpoint towards resistance, the green line a breakpoint towards intermediate. Regarding the two combination compounds, only the concentration of the former substance is presented; the ratio of amoxicillin:clavulanic acid is 2:1 (1), concentration ratio of trimethoprim:sulfamethoxazole is 1:19 (2). Tulathromycin has not been tested in the first quarter of 2015 (3). The summation of intermediate and resistant isolates was named non-susceptible (4). Some results were not evaluable (5).

results were not evaluable (5).

**Figure 2.** The mean value (bold) and the five‐year trend on non‐susceptible *E. coli* isolated from calves revealed the highest proportion of isolates against trimethoprim‐sulfamethoxazole, followed by enrofloxacin and amoxicillin‐clavulanate. The trends regarding enrofloxacin and amoxicillin‐ clavulanate remain at a stable level and rather tend towards a decrease regarding the number of non‐susceptible isolates. The graph of non‐susceptible isolates regarding trimethoprim‐sulfameth‐ oxazole reveals a decrease, 2016–2017, followed by a steep increase of non‐susceptible isolates in 2019. The corresponding statistic parameters are presented in Table 1. **Figure 2.** The mean value (bold) and the five-year trend on non-susceptible *E. coli* isolated from calves revealed the highest proportion of isolates against trimethoprim-sulfamethoxazole, followed by enrofloxacin and amoxicillin-clavulanate. The trends regarding enrofloxacin and amoxicillinclavulanate remain at a stable level and rather tend towards a decrease regarding the number of nonsusceptible isolates. The graph of non-susceptible isolates regarding trimethoprim-sulfamethoxazole reveals a decrease, 2016–2017, followed by a steep increase of non-susceptible isolates in 2019. The corresponding statistic parameters are presented in Table 1. *Antibiotics* **2022**, *11*, x FOR PEER REVIEW 6 of 17

**Figure 3.** The classification of 8713 *E. coli* into extensively drug‐resistant and multi drug‐resistant isolates was carried out according to the expert proposal for standard definitions for acquired re‐ sistance. We categorized eight potential pandrug‐resistant isolates in the category extensively drug resistant, as we only tested antimicrobials licensed for the veterinary use and did not include the latest antimicrobials available on the market. **Figure 3.** The classification of 8713 *E. coli* into extensively drug-resistant and multi drug-resistant isolates was carried out according to the expert proposal for standard definitions for acquired resistance. We categorized eight potential pandrug-resistant isolates in the category extensively drug resistant, as we only tested antimicrobials licensed for the veterinary use and did not include the latest antimicrobials available on the market.

#### *2.2. Serologic Characterization*

were not detected at all.

2, Table S2).

**Additionally Known for Pathogenicity in**

box).

**Serotype**

*2.2. Serologic Characterization* Serotyping of a randomly chosen subset of 108 *E. coli* isolated in 2019 and 2020 re‐ vealed 38 unequivocally typeable (35.2%), 29 untypeable (26.8%) and 41 seronegative Serotyping of a randomly chosen subset of 108 *E. coli* isolated in 2019 and 2020 revealed 38 unequivocally typeable (35.2%), 29 untypeable (26.8%) and 41 seronegative (38%) strains

(38%) strains (Table 2, Table S2). The most frequently detected serotypes were O101:K28

once each (Table 2, Table S2). Finally, the serotypes O138:K81, O149:K91 and O157:H7

**Table 2.** The serologic and molecular characterization revealed 13 different serotypes known to be pathogenic for cattle and other species. Furthermore, four different genotypes were detected with five different coding sequences for fimbria and/or toxins in one or more isolates. Some of the isolates were untypeable/ seronegative and did not reveal any of the investigated virulence factors (green

**Molecular results**

**F5 F41 ST‐I**

**stx1**

**F17 F5**

**ST‐I**

The fimbrial antigen F5 agglutinated in 6.5% of the isolates (*n* = 7) in combination with the serotypes O101:K30, O101:K28 and O9:K35. The fimbrial antigen F4 agglutinated in 4.6% of the isolates (*n* = 5), and exclusively combined with the serotype O139:K82 (Table

**Number of** 

**Isolates Non‐Virulent**

O9:K35 6 5 1 O9:K35/F5 1 1 O101:K28 6 6 O101:K28/F5 3 3 O101:K30 1 1 O101:K30/F5 3 3 (Tables 2 and S2). The most frequently detected serotypes were O101:K28 (8.3%; *n* = 9), O9:K35 and O139:K82 (6.5%; *n* = 7), O101:K30 (3.7%; *n* = 4), O101:K32, O78:K80 and O8:K87 (2.8%; *n* = 3). The serotypes O141:K85 and O147:K89 were detected once each (Tables 2 and S2). Finally, the serotypes O138:K81, O149:K91 and O157:H7 were not detected at all.

The fimbrial antigen F5 agglutinated in 6.5% of the isolates (*n* = 7) in combination with the serotypes O101:K30, O101:K28 and O9:K35. The fimbrial antigen F4 agglutinated in 4.6% of the isolates (*n* = 5), and exclusively combined with the serotype O139:K82 (Tables 2 and S2).

**Table 2.** The serologic and molecular characterization revealed 13 different serotypes known to be pathogenic for cattle and other species. Furthermore, four different genotypes were detected with five different coding sequences for fimbria and/or toxins in one or more isolates. Some of the isolates were untypeable/ seronegative and did not reveal any of the investigated virulence factors (green box).


#### *2.3. Molecular Characterization*

Within the molecular characterization, 14 PCR assays targeted genes for the expression of fimbria, adhesin, hemolysin and toxins. A positive result was obtained for 24 isolates and 35 single assays, respectively (Tables 2 and S2). The most frequently detected genes coded for the fimbria F17 (13.9%; 15/108), F41 (3.7%; 4/108) and F5. The latter was always detected in combination with the toxin gene coding for ST-I (6.5%; 7/108). Finally, the gene coding for stx1 was detected in two of 108 isolates (1.9%). Seven of 108 isolates (6.5%) carried more than one type of virulence-associated genes (Tables 2 and S2). The fimbrial antigens F4, F6, F18, O157, adhesin eaeA, hemolysin ehxA and the toxins LT, ST-II and stx2 were not detected in any isolate. The occurrence of F4 fimbria in the serotyping assays could not be confirmed in the PCR investigation (Tables 2 and S2). In all, 84 of 108 isolates were negative in all PCR assays (Tables 2 and S2).

#### **3. Discussion**

Antibiotic treatment is the fundamental therapy regarding serious or life-threatening bacterial infections in man and animals [28,29]. Records regarding antimicrobial susceptibility on single substances are collected in many countries all over the world [43]. Worldwide this is a critical topic in line with the One Health issue [44]. Monitoring on the application and more important efficacy of antimicrobials regarding bacterial infections of farm animals is possible on principle in industrial countries. However, it is costly and difficult to standardize [36]. Published data from Canada in 2018 revealed a 51.6% susceptibility rate of 489 *E. coli* against trimethoprim-sulfamethoxazole, which is in consensus with our

data (50%) (Figures 1 and 2) [45]. Tetracycline was accounted to be effective in 36.8% and resembles our findings at 38.1% (Figure 1) [45]. Further, authors from the United States and Germany determined similar high resistance rates for tetracycline, with 71.1% and 70.9%. These data rather resemble the rate of 61.4% revealed in the present study (Figure 1) [46,47].

The antimicrobial class of fluoroquinolones includes enrofloxacin which is one of the substances of choice for the treatment of diarrhea in young cattle [29,48]. In Germany, the usage of fluoroquinolones has risen from 2011 to 2013 in human and veterinary medicine. This trend needs close monitoring to preserve the efficacy of the agent [27]. Fluoroquinolones are assessed as highest priority clinically important antimicrobials and as one of the few options for the treatment of serious *Salmonella* and *E. coli* infections in children recommended by the World Health Organization (WHO) [49]. The legislation reacted and passed a law in 2017 including obligatory antimicrobial susceptibility testing in case of the application of fluoroquinolones or third or fourth generation cephalosporines in Germany [50]. In the present study, the investigated *E. coli* isolated revealed a resistance rate of 34.1% regarding enrofloxacin (Figure 1). This finding correlates with published results from South America in 2017, with 36.4% [51].

Antimicrobial substances or closely related compounds may likewise be licensed for the use in man and animals. The application in an organism does trigger the development of antimicrobial resistance in present bacteria [49]. Legal restrictions regarding the use of cephalosporines, especially from the third and fourth generation, aim at a high prioritization of critically important antimicrobials in human medicine [49]. This is again in accordance with the terms of One Health [27,44]. The use of cephalosporines for the therapy of *E. coli* diarrhea in calves is a malpractice, as the effective therapeutic concentration is not reached within the gut [29]. Nonetheless, cephalosporin is the fifth-most commonly prescribed antimicrobial in the case of diarrhea with 8% according to a recent survey in Europe [30]. Regarding the third generation cephalosporine ceftiofur, a susceptibility rate of 86.4% could be determined in a study from Canada between 1994 and 2013 [45]. Significantly, our findings revealed 76.8% (Figure 1). Compared to data from the USA collected within the years 1960 until 2002 and in 2007, the resistance rate was at 7.4% and 11%, whereas in the present study the resistance rate of ceftiofur revealed 20.4% (Figure 1) [46,52]. This result is concerning, and the use of ceftiofur must be scrutinized critically, if not avoided completely. The resistance rates of the first generation cephalosporine, cephalothin, were lower in a comparable study regarding data within the period of 1960 to 2002, with 20.1%, in contrast to our results with an average rate at 46.1% from 2015 to 2019 (Figure 1) [46]. Currently, the standard antimicrobial therapy of mastitis in cows includes penicillins as well as first and second generation cephalosporines in the EU. Traces of antibiotics may reach the calves through the feeding of antibiotic contaminated waste milk [36]. To predict a reliable trend regarding the prevalence of ESBL-producing *E. coli*, PCR and sequencing methods should be applied to investigate the existence of ESBL- encoding genes as these are probably more accurate than the phenotypic characterization [53]. A study from 2013 revealed high rates (32.8%, 196 of 598 samples) of ESBL-encoding *E. coli* on dairy and beef cattle farms in Bavaria [54].

Completely inconsistent data are publicly available regarding the resistant rates for *E. coli* isolates and the substance florfenicol within the phenicol group. A 78% share of resistant isolates was determined in a study from the USA in 2006, only a 28% share from Canada in 2018, and a share of 35% from Bavaria, Germany, in 2002 [45,52,55]. In the present study, a rather higher resistance rate of 60.6% was determined for florfenicol (Figure 1). There was no information about ages of animals within the American and Canadian studies [45,52]. Since lower resistance rates were previously published in older animals for the substances ampicillin, tetracycline, streptomycin, sulfamethoxazole and chloramphenicol, this might accordingly apply for florfenicol [32]. This argument, however, still does not explain the diverse results of the Bavarian study from 2002 and the present study (Figure 1) [55].

With a 9% share of the most frequently listed antimicrobials, aminoglycosides remain at the fourth top position for the treatment of diarrhea in calves [30]. As these are almost solely used in the therapy of enterococcal endocarditis and multidrug-resistant tuberculosis in humans, they account to the high priority, clinically important antimicrobials in human medicine [49]. An application in veterinary medicine should therefore be prudent and well considered. Gentamicin belongs to the aminoglycoside antimicrobial class and has a withdrawal time for meat of more than 200 days in Germany for cattle and the indication of gastrointestinal disease. As this is economically hardly acceptable, the application of gentamicin is quite limited [48]. However, resistance to gentamicin among *E. coli* isolated from animals has been increasing from 0% to 40% between 1970 and 2002 within the United States [46]. Another long-term investigation from Germany revealed a further decrease of resistance rates including data from 2010 until 2013, and 2016 until 2017, respectively [47]. In the present study, the resistance rate of *E. coli* against Gentamicin was at 14.1% (Figure 1). Likewise, spectinomycin is an aminoglycoside antibiotic as well, and frequently used in combination with lincomycin for oral application in the treatment of simultaneous infection of the respiratory and the gastrointestinal tract in calves. The meat withdrawal time of 21 days is acceptable for farmers and practitioners and may be an explanation for the frequent prescription [48]. Within the present study and correspondingly a resistance rate of 48.9% was revealed in calves (Figure 1).

As stated by the WHO, the antimicrobial class of polymyxins accounts for the highest priority in critically important antimicrobials regarding the treatment of serious infections with Enterobacteriaceae and *Pseudomonas aeruginosa* in human medicine [49]. Despite rather frequent prescription of polymyxins in the treatment of diarrhea in animals, investigated *E. coli* isolates are still highly susceptible [30]. In the present study, the resistance rate against colistin revealed to be only 1.8% (Figure 1). Corresponding to this suggestion, another study revealed that only 3.8% of the isolates were resistant to colistin [47].

The aminopenicillin family, as well as the preparation amoxicillin-clavulanate, belong to the high priority critically important antimicrobials for the therapy of *Listeria* and *Enterococcus* spp. infections in humans according to the WHO [49]. For the aminopenicillin, ampicillin, an alarming resistance rate of 76.3% was determined in *E. coli* published in a most recent study from Germany [47]. Regrettably, a rate of 69.5% was determined in the present work as a similar result (Figure 1). Consequently, the recommendation on the usage of ampicillin for the treatment of calf diarrhea cannot further be continued. The amoxicillin-clavulanate susceptibility rate averaged at 57% in Germany in 2013 [27]. In the present study, the average susceptibility rate was 71.9%, and the resistance rate was 8.6% (Figure 1). Accordingly, a recently published study reported 7% of resistant *E. coli* isolates in Germany in 2018 [34]. Analogical to the report on the resistance monitoring study 2018 of the Federal Office of Consumer Protection and Food Safety, Germany, we determined decreasing non-susceptibility rates regarding the clinically important antimicrobial amoxicillin-clavulanate [34]. In conclusion, the resistance rates of *E. coli* against amoxicillin-clavulanate have decreased since 2013 and remained on a constant level within the years 2015 and 2019. This is a positive trend is beneficial for the One Health point of view [27].

Comparing data originating from other continents and collected over the last 60 years clearly reveals an increase of resistance regarding *E. coli* in nine out of the 12 tested drugs, namely gentamicin, cephalothin, ceftiofur, enrofloxacin, trimethoprim-sulfamethoxazole, ampicillin, amoxicillin-clavulanate, florfenicol and tetracycline [27,34,45–47,51,52,55]. Out of the 12 tested drugs in the present study, eight substances are similarly suitable for the treatment of human patients, namely gentamicin, spectinomycin, cephalothin, ampicillin, tetracycline, amoxicillin-clavulanate, colistin and trimethoprim-sulfamethoxazole (Figure 1) [49]. The application of these in veterinary medicine should be prudent due to the One Health aspect.

In a published study from Canada in 2018, 48.7% of multidrug-resistant *E. coli* were isolated from ruminants [45]. Within another study from the USA covering the years 1950 until 2002, a significantly increasing trend in resistance was observed for ampicillin, sulfonamide and tetracycline antibiotics regarding more than 1700 *E. coli* isolates. Two of these strains were identified as pandrug-resistant and originated from cattle in 2001 [46]. Further, multidrug resistance in *E. coli* increased from 7.2% to 63% between 1950 and 2002. Finally, 59.1% of the strains recovered form cattle were classified as multidrug resistant in the USA [46]. In the present study, we detected an even higher rate of 84.2% regarding multidrug resistance, 15.7% extensively drug-resistance and 0.1% pandrugresistance (Figure 3). Furthermore, there were no exclusively susceptible isolates found amongst 108 isolates recovered in 2019 and 2020 from diarrheic calves in Bavaria (Table S2). Comparably high levels of antimicrobial resistance were published regarding the countries Brazil and Uruguay. Calves aged up to 60 days revealed a multidrug-resistance rate in *E. coli* at 78.7%, and at 61.6%, respectively [51]. As published, these bacteria occurred frequently in herds with high levels of diarrhea symptoms and subsequent antimicrobial therapy, as equally described in the present study [31].

Besides antimicrobial resistance, the determination of virulence regarding infectious agents is crucial in diagnostics. The discrimination from commensal *E. coli* was determined investigating virulence factors and evaluating the pathogenicity of isolates. As published, the *E. coli* serotypes O139:K82, O8:K87 and O147:K89 are pathogenic in swine [6]. However, in the present study, a fair amount of such isolates, six out of 108, were isolated from cattle, respectively (Tables 2 and S2). In laboratory diagnostics, implication of these serotypes should therefore be considered. Three isolates were identified as the serotype O78:K80, which frequently causes septicemia in calves (Table 2) [5,7,56]. However, more than one third, 38%, of the *E. coli* in this study revealed to be entirely seronegative (Tables 3 and S2), as it was as well published previously [57]. Preferably and in accordance with the One Health approach, the screening of *E. coli* isolated from diseased animals should always be of interest to identify zoonotic and human pathogenic serotypes [25]. As a matter of fact, formula associated with severe human syndromes included the serotypes O26, O103, O111, O117, O128, O145 and O146 respectively [13,22,23,58].

**Table 3.** In all, 16 different polyvalent and monovalent (mono) antisera were used for the agglutination and the characterization of *E. coli*. The listed serotypes are known for their pathogenicity in humans and farm animals.


In recent studies, the fimbrial adhesins F17, F41 and F5 were frequently and significantly correlated with diseased calves compared to healthy animals [4,9]. These findings clearly correspond to the results of the present study (Tables 2 and S2). Other selective fimbrial antigens, F4, F6 and F18, occur frequently in isolates from diarrheic piglets [1,10,59]. As to be expected, we did not detect these amongst our strains isolated from calves (Table S2). Even five serologically F4 positive isolates were not confirmed within our molecular investigation (Tables 2 and S2). We assume that none of these isolates carry the specific primer sites, or agglutination was non-specific [9]. However, working at a federal state laboratory, we do research cross species infections especially among farm animals [60]. Furthermore, we consider the One Health approach, here especially the idea from farm to fork, and therefore continuously consider possible correlations between food-borne human pathogens and isolates from farm animals [27,44].

As published, hemolysis in *E. coli* isolates from piglets is a reliable diagnostic marker for virulence and pathogenicity [61–63]. Within the present study, only few (3/108) isolates revealed a hemolytic phenotype that was not even confirmed within the molecular analysis (Table S2). We conclude that hemolysis is not a relevant marker for virulence of *E. coli* isolated from calves in the present study. This statement is in accordance with prior publications [64,65].

Regarding the present study, ST-I was found in similar prevalence at a rate of 6.5% (7/108) compared to published data (Tables 2 and S2) [4,66]. The enterotoxins LT and ST-II were not detected in the present study (Table S2) and this again resembles data of relevant previous studies [4,56]. Concluding published data, ETEC isolated from calves only produced ST-I, whereas ETEC isolated from pigs may encode varying combinations of the enterotoxins LT, ST-I and ST-II [11,67]. In the present study, the detection rate of stx1 was very low and stx2 as well as intimin were not detected at all among the diarrheic calves' isolates (Tables 2 and S2). This finding matches the results of previously published data to a high degree [9,51,68]. Obviously, the detection rate of Shiga toxins rose with the number of colonies isolated from each clinical sample, suggesting the selection of up to 35 colonies [69,70]. In the present investigation however, only up to three colonies were analyzed per clinical sample (Table S2). Other published results suggested a positive correlation between animal age and the amount of Shiga toxin, supporting our findings including animals of young age [69–71]. Targeted infection studies with STEC led to severe disease and bloody diarrhea in neonatal calves, but more recent studies disproved this observation revealing a still controversial discussion [4,72–74].

#### *Limits of the Study*

The antimicrobial susceptibility testing was carried out with a standard panel of antibiotics currently used in veterinary diagnostics in Germany. The results are therefore limited to substances only partially prescribed in human diagnostics and sometimes even in veterinary medicine regarding other countries of the world.

A thorough molecular investigation of single isolates is fairly time consuming and costly compared to the benefit that might be drawn from the results. In routine diagnostics, the molecular methods therefore can hardly be kept up.

#### **4. Materials and Methods**

#### *4.1. Study Design and Bacterial Isolates*

At the Bavarian Health and Food Safety Authority in Germany 7358 fecal samples of diseased or deceased calves with enteritis younger than six weeks of age were analyzed and included in the present study. Samples were collected between January 2015 and December 2019. Clinical symptoms ranged from low general condition, diarrhea, fever, sepsis and sudden death, respectively. A total of 8713 *E. coli* strains were isolated and confirmed through positive fluorescence on ECD agar (Merck Millipore, Burlington, MA, USA) and a positive Kovacs-Indole reaction (Merck Millipore, Burlington, MA, USA). All isolates were subject to antimicrobial resistance testing, further analysis and cryopreservation at the internal vaccine laboratory.

#### *4.2. Antimicrobial Susceptibility Testing*

Antimicrobial susceptibility testing was carried out according to the protocols published in CLSI VET01, 5th edition (Clinical and Laboratory Standards Institute, Wayne, PA, USA) [41]. Breakpoints were adopted from CLSI Vet01S, 5th edition, and national break-

points for farm animals [41,42,75]. We used the microbroth dilution method on the following twelve different antimicrobial agents (antimicrobial class): Amoxicillin-clavulanic acid (betalactam combination agent), enrofloxacin (fluoroquinolone), Trimethoprimsulfamethoxazole (folate pathway inhibitor), gentamicin and spectinomycin (aminoglycosides), cephalothin (cephalosporin I and II), ceftiofur (cephalosporin III and IV), ampicillin (penicillin), florfenicol (phenicol), colistin (polymyxin), tetracycline (tetracycline) and tulathromycin (macrolide). A commercially available set was used according to the manufacturer's instructions (Micronaut-S, Grosstiere 4, Merlin, Bruker, Bornheim, Germany). The minimum inhibitory concentration (MIC) of each isolate and antibiotic substance was metered using a photometric plate reader system (Micronaut Scan and MCN6 software, Merlin/ Sifin, Bruker, Bornheim, Germany). Subsequently, the MIC value was reconciled with supplemented CLSI breakpoints, to categorize the respective *E. coli* isolate into "susceptible", "intermediate" and "non-susceptible" for each antimicrobial substance tested [41,42,75,76]. *E. coli* ATCC 25922 was used as quality control strain [41].

#### *4.3. Phenotypic Analysis and Serotyping*

We deeper investigated a subset of 108 *E. coli* isolated in 2019 and 2020 originating from 66 diarrheic calves. The isolates were subcultured on Gassner agar (Oxoid Deutschland GmbH, Wesel, Germany) to differentiate specific colony morphology. The expression of potential virulent F5 fimbria was investigated by subculturing the isolates on pH 7.5 stabile, "minimum of casein" (Minca) agar (Sifin Diagnostics GmbH, Berlin, Germany) as previously published [76]. Finally, potential hemolytic properties of isolates were interpreted as described with subcultures on Columbia Sheep Blood Agar (Sifin Diagnostics GmbH, Berlin, Germany) [77]. Growth incubation was carried out for 18 to 24 h at 37 ◦C at all times. Serotyping for specific O-antigens was carried out using two polyvalent and 14 monovalent agglutination sera in a hierarchical approach according to the manufacturer's instructions (Sifin Diagnostics GmbH Berlin, Germany) (Table 3). If an isolate showed a positive agglutination reaction with a polyvalent serum, but none with any correspondent monovalent or several reactions with various correspondent monovalent sera, it was categorized as untypeable. If an isolate showed no positive agglutination with any serum, it was categorized as seronegative.

#### *4.4. Molecular Investigation*

The molecular characterization of the *E. coli* isolates in the present study aimed at surface antigens, toxins and virulence factors. In all, 14 different target genes were of interest. Amongst were seven fimbrial genes F4, F5, F6, F17, F18, F41 and the outer membrane protein O157:H-. Further, two virulence genes were included, here adhesin intimin (eaeA), and enterohemolysin (ehxA). Finally, PCR targets coding for five toxins were screened, including heat-labile toxin (LT), heat-stabile toxin I (ST-I) and II (ST-II), Shiga toxin 1 (stx1) and stx2 (Table 4). Primer sequences were adopted from published protocols [9,39,40]. All 14 qPCR assays were performed applying a singleplex high resolution melting method, using AccuMelt HRM SuperMix (Quantabio, Beverly MA, USA) in 20 µL volumes according to the manufacturer's instructions. DNA was extracted after thermolysis. The primers were added in a concentration of 0.2 µM each, and 3 µL of template DNA was used. Polymerase chain reaction assays were conducted on a Stratagene MX3000P device (Agilent Technologies, Waldbronn, Germany). The cycling protocol comprised an initial single denaturation step for 10 min at 95 ◦C, followed by 40 cycles of annealing and polymerization for 30 s at 60 ◦C and 10 s at 95 ◦C. After completing amplification, the melting curve analysis was performed. Specific melting temperatures were determined for each molecular target and all tested isolates. Reference strains were used as positive controls and kindly provided from Prof. R. Bauerfeind (Justus-Liebig-Universität, Gießen, Germany), and purchased from the German Collection of Microorganisms and Cell Cultures GmbH (DSMZ, Braunschweig, Germany) (Table 4).

*Antibiotics* **2022**, *11*, 23


#### *4.5. Statistical Analysis*

All statistical analyses were performed using the free software R Studio version 1.2.5033 (RStudio, Inc., Boston, MA, USA). Resistance trends of three clinically relevant antimicrobials amoxicillin-clavulanate, enrofloxacin and trimethoprim-sulfamethoxazole were evaluated by calculating a logistic regression model. The respective year was set as a continuous variable. The resulting odds ratio (OR) > 1 indicated an increased resistance trend, whereat an OR < 1 indicated a decreased antimicrobial resistance. The Wald test was used to determine the statistical significance of the year-antimicrobial trend. A value of *p* < 0.05 was considered significant (Table 1).

#### **5. Conclusions**

We conclude that an extensive monitoring, characterization and the analysis of antimicrobial resistance regarding enteritis causing *E. coli* is crucial to determine the currently raging serotypes, virulent genotypes and most important, the resistance situation. It is then possible to calculate reliable tendencies and prognoses from data collected over long terms in routine diagnostics. This is an important premise for objective and professional treatment recommendations regarding humans and animals within the scope of One Health. A further goal should be a slowdown of the increasing antimicrobial resistance situation that constitutes a global public health threat.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics11010023/s1, Table S1: data set for all 8713 isolates from 2015–2019. Table S2: data set for a subset of 108 isolates in 2019–2020.

**Author Contributions:** Conceptualization, A.F., R.K.S. and J.M.R., methodology, A.F., A.H., K.B. and J.M.R.; software, A.F., A.M. and K.B.; validation, A.F., N.S., C.K., A.H., S.F. and J.M.R.; formal analysis, A.F., A.H., A.R., S.F., R.K.S. and J.M.R.; investigation, A.F., A.M., K.B., A.R. and S.F.; resources, C.K., K.B., A.R., S.F. and J.M.R.; data curation, A.F. and J.M.R.; writing—original draft preparation, A.F., N.S., A.M. and J.M.R.; writing—review and editing, N.S., C.K., K.B., A.R. and R.K.S.; visualization, A.F. and J.M.R.; supervision, R.K.S. and J.M.R.; project administration, J.M.R.; funding acquisition, K.B. and J.M.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Acknowledgments:** The authors are grateful to the veterinary bacteriology staff members for technical support. Furthermore, the authors are grateful to the staff of the StabLab, Ludwig-Maximilians-University Munich, for a basic teaching class in statistical analyses.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Clair L. Firth <sup>1</sup> , Reinhard Fuchs <sup>2</sup> and Klemens Fuchs 2,\***


**Abstract:** Antimicrobial use in livestock production systems is increasingly scrutinised by consumers, stakeholders, and the veterinary profession. In Austria, veterinarians dispensing antimicrobials for use in food-producing animals have been required to report these drugs since 2015. Here, we describe the national monitoring systems and the results obtained for Austrian pig production over a six-year period. Antimicrobial dispensing is described using the mass-based metric, milligrams per population correction unit (mg/PCU) and the dose-based metric, Defined Daily Dose (DDDvet) per year and divided into the European Medicines Agency's prudent use categories. Pig production was divided into breeding units, fattening farms, farrow-to-finish farms, and piglet-rearing systems. Over all six years and all pig production systems, the mean amount of antimicrobials dispensed was 71.6 mg/PCU or 2.2 DDDvet per year. Piglet-rearing systems were found to have the highest levels of antimicrobial dispensing in DDDvet, as well as the largest proportion of Category B antimicrobials, including polymyxins. Although progress has been made in promoting a more prudent use of antimicrobials in veterinary medicine in Austria, further steps need to be taken to proactively improve animal health and prevent disease to reduce the need for antimicrobials, particularly those critically important for human medicine, in the future.

**Keywords:** antimicrobial use; pigs; veterinary; monitoring

#### **1. Introduction**

Globally, antimicrobial use in agriculture, particularly in food-producing animals, is increasingly seen critically by consumers [1]. Although the use of antimicrobials as growth promoters has been banned in the European Union (EU) since 2006, these medications are often still used for disease prophylaxis, and reductions in antimicrobial use (AMU) are both possible and necessary in order to ensure their continued effectiveness against bacterial infections. From 2022, the new Veterinary Medicinal Products Regulation (2019/6) in the EU will legislate new restrictions on AMU in veterinary medicine and requires all member states to monitor and record veterinary AMU in their countries, initially in food-producing animals, but eventually (from 2029) in pets as well [2].

The excessive use of antimicrobials in pig production initially came under criticism in Denmark in the 1990s, and the country was among the first to successively ban a variety of antimicrobials as growth promoters from 1995 onwards [1,3]. Denmark also led the way in benchmarking pig producers and introducing penalty schemes, such as the yellow card for excessive antimicrobial use in 2010 [3]. A number of other European countries, such as the Netherlands, also began to document their veterinary antimicrobial use, and the first EU report of veterinary antimicrobial sales (the ESVAC report) was published by the European Medicines Agency in 2011, using sales data from nine countries [4].

The Austrian health authorities began contributing data on veterinary antimicrobial sales from pharmaceutical companies/wholesale pharmacies to the European Union's

**Citation:** Firth, C.L.; Fuchs, R.; Fuchs, K. National Monitoring of Veterinary-Dispensed Antimicrobials for Use on Pig Farms in Austria: 2015–2020. *Antibiotics* **2022**, *11*, 216. https://doi.org/10.3390/ antibiotics11020216

Academic Editor: Carlos M. Franco

Received: 15 January 2022 Accepted: 5 February 2022 Published: 8 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

annual ESVAC report in 2010. To date, reported sales of veterinary antimicrobial drugs in Austria for food-producing animals have ranged from a maximum of 63 mg/population correction unit (PCU) in 2010 to a minimum of 42.6 mg/PCU in 2019 [5,6]. Since 2015, it has been required by local law for all veterinarians in Austria who dispense antimicrobials for use in food-producing animals to annually report the amounts dispensed to the relevant authorities [7]. In addition, antimicrobials are only available from veterinarians, and, since 2005, injectable (as well as intramammary and intrauterine) antimicrobials have been further restricted and can only be dispensed to farmers who are members of the Austrian Animal Health Service (*Tiergesundheitsdienst*, TGD) and have completed training courses in the use and administration of veterinary medications [8,9]. Antimicrobials administered directly by the veterinarians themselves do not currently have to be reported, although their use is documented in both veterinary practice and on-farm records [7].

Pig production in Austria is not an extremely large industry, when compared internationally. The average herd size is 133 head (ranging from 15,950 holdings keeping only 1–3 pigs to 12 units with more than 3000 pigs) [10,11]. Based on official data available with respect to the reference day of 1st June each year, the pig population included here ranged from a minimum of 2,773,225 pigs in 2019 to a maximum of 2,845,451 in 2015 (mean number of pigs from 2015–2020: 2,802,433; median: 2,799,632) [10]. Pig producers are primarily located in the federal states of Upper Austria (39.8% of total pig numbers in 2020), Lower Austria (27.0%), and Styria (26.8%) [10].

The most recent national data records in metric tonnes in 2020 reported that 73.4% of all veterinary antimicrobials dispensed in Austria were for use in pigs (ranging from 71.8–76.4% between 2016 to 2020), compared to 19.7% in cattle (beef and dairy) and 6.7% in poultry [12]. However, when comparing these figures to other countries, it is important to note that the Austrian national-monitoring system currently only includes antimicrobials dispensed by veterinarians to farmers and does not include those administered by the veterinarians themselves.

The data presented here represent the results of the national monitoring of veterinary antimicrobials dispensed between 2015 and 2020. To allow comparison with other countries and systems, the data analysis focuses on using international metrics, such as mg/PCU (population corrected unit) and Defined Daily Doses (DDDvet), as published by the European Medicines Agency and recommended by European expert groups [13,14].

#### **2. Results**

#### *2.1. Study Population*

Pig production in Austria is divided into farrow-to-finish farms, fattening farms, breeding units, and piglet-rearing units. Figure 1 shows the proportions of the different pig production systems in the study population over the years included here. The study population (i.e., farms where antimicrobials were dispensed and reported to the authorities by herd veterinarians) covers between 81% (in 2015) and 87% (in 2020) of the total national pig production.

Data are provided in standardised livestock units, as defined by the Austrian Ministry of Agriculture [15]. The vast majority of pigs included in this study population were kept in fattening and farrow-to-finish units (mean: 351,261 and 310,933 LSU; median 348,398 and 315,147 LSU, respectively). An extremely small number of pigs are reared in piglet-rearing systems (mean: 7809 LSU; median 7450 LSU) (Figure 1).

#### *2.2. Overall Antimicrobials Dispensed*

#### 2.2.1. Mass-Based Metrics (mg/PCU)

All veterinarians treating farm animals and dispensing antimicrobials to farmers for use in such animals are required by Austrian law to report their annual dispensed amounts [7]. The data included here are taken from these national records of annual antimicrobial monitoring between 2015 and 2020 [12].

**Figure 1.** Comparative numbers of 1000 livestock units (LSU) in Austrian pig production systems (included in the study population, i.e., farms where antimicrobials were dispensed and reported) between 2015 and 2020.

Antimicrobials dispensed by herd veterinarians for use in pig production between 2015 and 2020 are shown in milligrams per population correction unit (mg/PCU, as defined in the European Medicines Agency's ESVAC report and calculated for the entire national pig herd [6]) in Figure 2. (NB. 1 PCU is approximately equivalent to 1 kg livestock biomass). For all pig production systems overall, the antimicrobial use ranged from a maximum of 79.3 mg/PCU in 2018 to a minimum of 66.5 mg/PCU in 2019.

**Figure 2.** Amount of antimicrobials (mg/PCU) dispensed for use in pigs in Austria between 2015 and 2020.

Table 1 shows the proportions of antimicrobial dispensing by veterinarians for use in the various pig production systems. The vast majority of antimicrobial dispensing in mg/PCU over all six years was for use in farrow-to-finish and fattening farms. It is important to note that the decreasing proportion of pig production units that were "not assignable" to a specific production system has fallen dramatically (from 4.6% to 0.8%) since the monitoring system was first initiated in 2015. This is primarily due to improvements to the electronic-reporting system and the data-plausibility checks now in place.


**Table 1.** Proportion in percent per year of antimicrobials dispensed for use in Austrian pigs for different farm types, based on mg/PCU.

#### 2.2.2. Comparison of Mass-Based and Dose-Based Metrics

A variety of antimicrobial monitoring guidelines and recommendations suggests the use of dose-based metrics, such as the European Medicines Agency's DDDvet, to allow for divergences in dosing to be accounted for within AMU records [14,16]. Recording antimicrobial dispensing in mg/PCU often leads to an overestimation of some antimicrobials and an underestimation of others [16,17].

Figure 3 demonstrates the proportions of the total antimicrobial-dispensing data collected in 2020 when analysed by mg/PCU or DDDvet. The differences between tetracyclines in mg/PCU (59.6% of all antimicrobials dispensed) compared to around 43.8% of all dispensed DDDvet are particularly striking. By contrast, aminoglycosides make up 8.3% of antimicrobials dispensed by DDDvet compared to just 1.9% by mg/PCU, and polymyxins make up a much higher proportion of overall use (9.5%) by DDDvet compared to under 5% as mg/PCU (Figure 3).

**Figure 3.** Comparison of the mean proportions of mass-based and dose-based metrics for antimicrobial classes dispensed for use in pigs in Austria in 2020.

When antimicrobial dispensing is presented by the proportion of DDDvet per year for the entire monitoring period (see Table 2), tetracyclines continue to make up the largest proportion each year (ranging from a maximum of 50.43% in 2018 to a minimum 39.77% in 2019). Extended-spectrum penicillins make up a much lower proportion (between 13.72–15.54%) and remain in second place over the study period, while polymyxins and

macrolides alternate for the third most frequently dispensed antimicrobials. By contrast, when antimicrobial dispensing is presented by proportion of mg/PCU, although tetracyclines continue to make up the vast majority of antimicrobial use (generally > 60%), polymyxins have fallen to fifth place and make up only 2.77% to 4.19% of antimicrobial dispensing (compared to a much higher proportion of between 6.93–9.51% when analysed by DDDvet/year) (Table 3).

**Table 2.** Proportion in percent per year of antimicrobial classes dispensed for use in Austrian pigs, based on the European Medicines Agency's DDDvet.


**Table 3.** Proportion in percent per year of antimicrobial classes dispensed for use in Austrian pigs, based on mg/PCU.


#### *2.3. Antimicrobials of Critical Importance to Human Medicine*

Antimicrobial dispensing presented here is divided into categories as defined by the European Medicines Agency's Antimicrobial Expert Group (AMEG) [18,19]. Category A is not included, as antimicrobials in this category are not licensed for use in veterinary medicine in the EU (although they may be used off-label in nonfood-producing animal species). Categories B and C are critically important for human medicine and should be used restrictively (Category B: 3rd and 4th generation cephalosporins, fluoroquinolones, and polymyxins) or with caution (Category C includes e.g., macrolides, extended-spectrum penicillins, amongst others). Category D antimicrobials should be used prudently and

include tetracyclines, sulfonamide/trimethoprim, beta-lactamase-sensitive penicillins, etc. With the exception of macrolides, Category B ("restrict") antimicrobials are comparable to the WHO's highest-priority, critically important antimicrobials (HPCIA) [18,19]. Further details are provided in the Section 5.

In all production systems, the majority of antimicrobials dispensed were in Category D, with the exception of piglet-rearing units, where a substantial proportion of antimicrobials dispensed were in Category B. For details, see Figure 4 and Sections 2.5 and 2.8 below. Again, the differences in mass-based versus dose-based metrics became apparent and can be seen very clearly when comparing Figure 4 (mg/PCU) with Figure 6d (DDDvet for piglet-rearing systems).

#### *2.4. Route of Administration for the Dispensed Antimicrobials*

As would be expected, the vast majority of antimicrobials dispensed in all categories for use in Austrian pig production were for oral administration. Category D antimicrobials for oral use ranged from 53 mg/PCU in 2019 to around 66 mg/PCU in 2018, as shown in Figure 5. By mg/PCU, the most frequently dispensed antimicrobial class for oral use in Category D ("prudent use") were tetracyclines (37.2 mg/PCU (2019)–50.8 mg/PCU (2018)) followed by macrolides (4.0 mg/PCU (2016)–5.2 mg/PCU (2018)) in Category C ("use with caution") and polymyxins (2.1 mg/PCU (2017)–2.9 mg/PCU (2020)) ("restrict use") in Category B; for details see Supplementary Figure S1. Injectable antimicrobials were dispensed at very low levels ranging from 0.4–0.5 mg/PCU in Category B to 2.6–2.9 mg/PCU in Category D (Figure 5). Specifically, the most commonly dispensed injectable antimicrobials were found in the classes of aminoglycosides (0.60 mg/PCU (2016)–0.75 mg/PCU (2015), Category C), beta-lactamase-sensitive pencillins (0.54 mg/PCU (2017)–0.60 mg/PCU (2020), Category D), and macrolides (0.27 mg/PCU (2019)–0.31 mg/PCU (2020), Category C), amongst others (Supplementary Figure S1).

#### *2.5. Antimicrobial Use on Piglet production/Breeding Units*

Breeding (piglet production) units made up approximately 20.5% of pig-producing units in Austria from 2015–2020, on average, ranging from 19.6% to 21.3% of pig production by LSU. Antimicrobial use on breeding pig units is shown in Figure 6a. The mean number of pigs kept on breeding units was 173,251 LSU.

**Figure 5.** Antimicrobials (in mg/PCU) dispensed and divided by route of administration over time. INJ = systemic/injectable administration; ORAL = oral administration.

#### *2.6. Antimicrobial Use on Farrow-To-Finish Farms*

Farrow-to-finish farms made up approximately 36.9% of pig-producing units in Austria from 2015–2020, on average, ranging from 35.9% to 38.1% of pig production by LSU (a mean number of pig equivalent to 310,933 LSU). Antimicrobial use on farrow-to-finish farms is shown in Figure 6b

#### *2.7. Antimicrobial Use on Fattening Farms*

Fattening/finishing farms made up approximately 41.7% of pig-producing units in Austria from 2015–2020, on average, ranging from 40.0% to 43.1% of pig production by LSU. The mean number of pigs kept on Austrian fattening farms was 351,261 LSU. Antimicrobial use on fattening farms is shown in Figure 6c, divided by EMA category. The vast majority of antimicrobials dispensed for use on fattening farms fall into Category D (prudent use).

#### *2.8. Antimicrobial Use on Piglet-Rearing Farms*

Piglet-rearing farms made up only a very small proportion of Austrian pig production, with, on average, approximately 0.9% of pigs produced in Austria from 2015–2020 by LSU (with a mean number of pigs equivalent to 7809 LSU). Only 23.3 of such farms reported antimicrobial use to the authorities, on average, over the six-year period (median 23.5 farms). Antimicrobial use on piglet-rearing units is shown in Figure 6d. It is important to note that the antimicrobial use in DDDvet per year on piglet-rearing farms is substantially higher than in other production systems. Furthermore, antimicrobial use in Category B (antimicrobials which are critical for human medicine and should be avoided) increased in 2020 to the highest level recorded since 2016 (median: 1.44 DDDvet/year in 2016 compared to 1.34 DDDvet/year in 2020).

**Figure 6.** National recording of antimicrobials dispensed for use on a variety of pig production systems ((**a**) breeding units; (**b**) farrow-to-finish farms; (**c**) fattening farms; (**d**) piglet-rearing units) by European Medicines Agency antimicrobial category (B, C, D) and DDDvet/year.

#### **3. Discussion**

The data presented here provide a comprehensive overview of veterinary antimicrobial dispensing for use on Austrian pig farms over a six-year period. Given the mandatory nature of reporting and the fact that data were provided for between 81–87% of national pig production in Austria, these analyses can be considered an accurate representation of antimicrobial dispensing for use in pig production in the country. Nevertheless, it is important to note that antimicrobials administered directly by veterinarians themselves (rather than dispensed to farmers), while no doubt making up a small proportion of antimicrobial use in pig production overall, were not included in this dataset.

The most recent data available on total antimicrobial dispensing for all pig production systems in Austria were calculated to be 68.8 mg/PCU. (NB. 1 PCU is approximately equivalent to 1 kg livestock biomass). These figures are comparable with antimicrobial sales reported in a study of veterinary wholesale data in Switzerland in 2015 (77.4 mg/kg) [20], but are higher than those previously reported for a small convenience sample of 75 pig farms in Austria (mean over four years: 33.9 mg/kg) [21]. By contrast, the Austrian national figures are much lower than those recently reported for 67 Irish pig farms (161.9 mg/PCU) or the UK figures for the national pig herd in 2020 (namely 105 mg/kg) [22,23].

With respect to Defined Daily Doses (DDDvet), the mean value of the six-year median DDDvet per year (2.2 DDDvet/year) reported here and covering all pig production systems is difficult to compare with other dose-based metrics, as calculation methods vary. A recent study in Italy (using national DDD metrics) reported annual median values of between 6.24–7.57 DDDita/100 kg on 36 fattening farms [24], which is substantially higher than the Austrian national mean of the six-year median value of 2.17 DDDvet for fattening farms determined here. Meanwhile, a Swiss study of 227 pig farms reported a mean treatment of 4 DDDvet over a one-year period [25], which is also higher than that reported here in Austria.

When analysing antimicrobial use by substance, the Austrian data show that tetracyclines are dispensed in the greatest volumes by mass. However, it is important to note that mass-based calculations are often skewed with respect to older antimicrobial molecules which have higher dosage requirements in mg/kg than other newer drugs which may be

more potent [14,16,26]. Oxytetracycline, for example, is licensed for use in pigs in Austria at a dosage of 40 mg/kg/d, which leads to a requirement of 2000 mg per day for a 50 kg pig. In contrast, the polymyxin, colistin, licensed at a dosage of 5 mg/kg/d, leads to a requirement of 250 mg per day for the same pig. This means that when comparing these antimicrobial drugs using mass-based metrics, oxytetracycline appears to be used at an eight-fold higher amount than colistin, which skews the overall proportions of antimicrobial classes in mg/kg. These discrepancies can be balanced out by using the defined daily dose (DDDvet), which refers to the daily dose as a whole, regardless of the amount of antimicrobial drug administered in milligrams.

For this reason, a comparison using dose-based metrics is essential [16]. Nevertheless, even when analysed by DDDvet metrics, tetracyclines still made up the majority (>55%) of antimicrobials dispensed for use in pig production in Austria between 2015–2020. Other studies have also reported that tetracyclines and penicillins are the most commonly used antimicrobials in pig production, such as a systematic review of 36 international papers [27] and a survey of 36 finishing pig farms in Italy [24]. In 2016, an Irish study of 67 farms, as well as Danish national reporting data, both demonstrated that tetracyclines were most frequently used [22,28], and similar findings have also been reported more recently from Japan [29]. The vast majority of tetracycline use in all these studies, as well as in the Austrian data presented here, was for oral administration. Whilst we do not have access to diagnoses data in Austria, tetracyclines are known to be commonly used for the treatment of gastrointestinal disorders and respiratory disease in pigs of all ages. Although tetracyclines are categorised by the EMA as the lowest level of caution (Category D, prudent use), some countries, such as Denmark, have seen increasing levels of antimicrobial resistance to this antibiotic and are now taking measures to reduce its routine use in pigs [28,30]. Similar resistance patterns have also been reported in studies in Austria, where tetracycline resistance was reported among 66% of *Streptococcus suis* isolates (increasing up to 88% of *Sc. suis* isolates obtained from joints) and 67.7% of *Escherichia coli* isolates obtained from piglets with diarrhoea [31,32].

Among piglet-rearing (and, to a much lesser extent, breeding) farms, a large proportion of antimicrobial dispensing was made up of polymyxins. This antimicrobial class contains the drug, colistin, which is commonly used to treat gastrointestinal disorders in young piglets (both pre- and post-weaning age), particularly disease caused by enterotoxigenic *Escherichia coli* (ETEC). While it is important to note that piglet-rearing farms make up only a very small proportion of Austrian pig producers (namely a mean number of pigs equivalent to 7809 LSU and between 0.8–1.5% of total antimicrobials dispensed for use in pigs by mg/PCU), polymyxins still made up a relatively large proportion (up to 9% by DDDvet, the third most frequently dispensed class in 2020) of antimicrobials dispensed in Austria overall. Polymyxins are classified by the European Medicines Agency as Category B antimicrobials, the use of which should be restricted as much as possible. Some countries, such as the UK and Denmark, have recently managed to avoid their use altogether among pig producers [23,28]. Although the most recent European Sales of Veterinary Antimicrobial Agents (ESVAC) report in 2021 stated that polymyxin use had fallen by 77% in 31 European countries since 2011, they are still sold at a higher level (based on mg/PCU metrics) in Germany, Poland, Hungary, Portugal, and Cyprus than in Austria [6]. The Netherlands has also reported a 7.3% increase in the use of colistin in all livestock production in 2020 and, as seen in the Austrian data, the vast majority of this use (91% of pig use) was for weaners [33]. Since plasmid-mediated colistin resistance was first detected in China in 2013, and the subsequent discovery of this resistance gene among pigs and humans throughout the world, recommendations have been made to reduce the use of this antimicrobial in livestock production wherever possible [34–36].

As would be expected, and as reported in many other studies [27,37,38], given the primarily intensive nature of pig production, the vast majority of antimicrobials were dispensed for oral administration. Systemically administered antimicrobials are generally used for the treatment of individual animals rather than entire groups and were dispensed at a

very low level. Category D antimicrobials made up the largest proportion of antimicrobials dispensed for use by injection, namely 2.6 to 2.9 mg/PCU (compared to 53–66 mg/PCU for oral use). While dispensed at a much lower level than Category D antimicrobials for oral administration, Category B antimicrobials (including colistin) were more commonly dispensed for oral rather than systemic treatment, which is particularly concerning as a previous Austrian study of 75 pig farms demonstrated that oral treatments are frequently (in 75% of cases) underdosed and only 8% of cases were correctly dosed [21]. Furthermore, a number of studies have reported that the risk of antimicrobial resistance is substantially higher following oral antimicrobial treatment rather than parenteral administration of such drugs, and the European Medicines Agency also classes oral treatment, particular as a group treatment, to be the least preferable route of antimicrobial administration [19,39].

The data presented here have demonstrated that antimicrobials dispensed for use on pig units with a high number of young piglets make up the highest proportion of Category B antimicrobials, drugs which should be limited to restricted use. Here, it is particularly important for herd veterinarians to work together with pig producers to attempt to prevent disease, such as post-weaning diarrhoea, by improving hygiene and biosecurity, reducing stress, and vaccinating either breeding sows or young piglets whenever possible [40]. Given that colistin is critically important for human health (as the first-line drug for carbapenemase-producing *Enterobacteriaceae* infections), is primarily administered orally to pigs, and colistin-resistant bacteria have been isolated from wastewater from pig slaughterhouses in Germany, the use of this antimicrobial substance is an extremely relevant example of an essential One Health drug affecting human, animal, and environmental health [36,41,42]. For this reason, Austrian pig producers should attempt to learn from pig producers in other countries, where the use of colistin has been considerably reduced or stopped completely.

The implementation of the new EU Regulation 2019/6 will bring a number of changes to the use of veterinary antimicrobials in Austrian and European livestock production as a whole. Prophylactic use of antimicrobials will no longer be permitted, and only the metaphylaxis of a group will be allowed when one or more animal is proven to be infected. It is expected that the restrictions on the use of Category B antimicrobials will be tightened and enforced. For this reason, Austrian pig producers and their herd veterinarians will need to alter their antimicrobial use towards a more prudent use of these essential drugs in the future.

#### **4. Conclusions**

Based on mandatory veterinary reporting, antimicrobial dispensing in the pig sector in Austria has not decreased over the past six-year period. While the vast majority of antimicrobials dispensed are in the EU's least restrictive Category D, an alarming proportion of Category B antimicrobials (primarily polymyxins, namely colistin) are dispensed for use in young piglets. National-benchmarking schemes are already in place for herd veterinarians and are currently being rolled out to individual pig producers. In future, partly due to new EU legislation, changes will need to be made to improve pig health and prudent antimicrobial use in this sector.

#### **5. Materials and Methods**

In Austria, pharmaceutical companies, marketing authorisation holders (distributors), and pharmaceutical wholesalers are required by law to provide the authorities with details of the sales of veterinary drugs containing antimicrobials. Additionally, veterinarians with in-house pharmacies must also report the quantities of antibiotics that are dispensed for use in food-producing animals for each farm and livestock species. The legal basis for the collection of these data is the "Veterinary Antibiotics Volumetric Flows Regulation" (*Veterinär-Antibiotika Mengenströme Verordnung*), which was enacted in 2014 [7].

#### *5.1. Pig Population Data*

The number of animals reared on each farm, as well as animal movement and official veterinary authority data, and numbers of animals slaughtered were available from the official veterinary database, namely the "Veterinary Information System (VIS)".

Each farm was categorised into one farm type using the reported "production system type", and the number of pigs in each category (piglets, fattening pigs, breeding sows/boars) are from the VIS database. The categorisation was taken from official records and can broadly be defined as follows. Breeding units refers to farms where sows (and sometimes boars) are kept to produce piglets for sale (it is not known at the veterinary authority level whether these piglets then go on to fattening farms or piglet rearing units). Fattening farms rear grower/finisher pigs from 20–32 kg liveweight up to slaughter. Pigletrearing units keep piglets from weaning (i.e., the sows are not present on this type of farm) until the beginning of the fattening period (approx. 20–32 kg). As the name would suggest, farrow-to-finish farms rear piglets from birth to slaughter.

#### *5.2. Antimicrobial Use Data*

Veterinarians with in-house practice pharmacies are required to provide the amount of dispensed antimicrobials for each marketing authorisation identification number (i.e., each licensed pharmaceutical product) for each farm and livestock species. This is used to calculate the total metric tonnes dispensed of each antimicrobial active ingredient each year. This metric was then converted into mg/PCU for pigs using the standardised method used by the Austrian authorities for the entire national pig herd and described for national reporting for the European Union's ESVAC report [6]. The standardised weight of a slaughtered pig as part of the PCU calculation is 65 kg; further details on the calculation of the PCU are provided elsewhere [6].

Furthermore, the number of Defined Daily Doses (DDDvet) for each antimicrobial substance, as defined by the European Medicines Agency, for the treatment of pigs was calculated as follows. The total number of milligrams of active ingredient dispensed for each antimicrobial substance was divided by the number of DDDvet for that antimicrobial substance with respect to pigs and the route of administration [13] to obtain the potential total number of Defined Daily Doses (DDDvet) for 1 kg animal biomass. To calculate the number of DDDvet per year, the following formula was used:


Livestock numbers were estimated based on the number of reported animals on the farm combined with animal movement and slaughter data. To ensure uniformity, livestock numbers were converted into the Austrian Ministry of Agriculture's livestock units (LSU), e.g., piglets and weaners (up to 20 kg liveweight) are classified as 0.07 LSU, growers and young boars/sows (up to 50 kg liveweight) as 0.15 LSU, and breeding boars/sows as 0.30 LSU [15].

The data were also divided by route of drug administration, such as systemic or oral application, as well as by production group.

#### *5.3. Classification into Prudent Use Categories*

In addition, data were divided into groups based on the European Medicines Agency's classifications of B (restrict use), C (use with caution), and D (use prudently), as well as according to the World Health Organization category of "highest priority critically important antimicrobials" (HPCIAs) [18,43]. For details, see Table 4. (NB. The EMA classification A (avoid) was not included as it does not list any antimicrobial substances licensed for use in food-producing animals).


**Table 4.** Categorisation of veterinary antimicrobials according to the European Medicines Agency.

Based on the EMA AMEG infographic [18].

#### *5.4. Statistical Analyses*

All statistical analyses were carried out using the statistical programming language R [44]. The data were prepared and plots were created using the tidyverse package [45].

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/antibiotics11020216/s1, Figure S1: Antimicrobial classes (in mg/PCU) dispensed, divided by route of administration over time.

**Author Contributions:** Conceptualization, C.L.F., R.F. and K.F.; methodology, R.F. and K.F.; validation, K.F.; formal analysis, R.F.; investigation, C.L.F., R.F. and K.F.; resources, K.F.; data curation, R.F.; writing—original draft preparation, C.L.F.; writing—review and editing, C.L.F., R.F. and K.F.; visualization, R.F. and C.L.F.; supervision, K.F.; project administration, K.F.; funding acquisition, K.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data included in this analysis were collated as part of a mandatory national monitoring programme and are not freely available in the public domain.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

