Article Text

Download PDFPDF

Paper
Development and evaluation of a system to assess antimicrobial drug use in farm animals: results of an Austrian study
  1. C. Ferner, Mag. med. vet.1,
  2. W. Obritzhauser, DVM3,
  3. K. Fuchs, Univ.-Doz. DI Dr.4 and
  4. I. Schmerold, DVM, Univ.-Prof (emeritus)2
  1. 1Postplatz 7, Tamsweg 5580, Austria
  2. 2Department for Biomedical Sciences, University of Veterinary Medicine Vienna, Institute of Pharmacology and Toxicology, Veterinärplatz 1, Vienna 1210, Austria
  3. 3Department for Farm Animals and Veterinary Public Health, University of Veterinary Medicine Vienna, Institute of Veterinary Public Health, Veterinärplatz 1, Vienna 1210, Austria
  4. 4Department for Data, Statistics and Risk Assessment, Austrian Agency for Health and Food Safety, Beethovenstrasse 8, Vienna 1210, Austria
  1. E-mail for correspondence: ivo.schmerold{at}vetmeduni.ac.at

Abstract

The objective of this study was to develop and evaluate a feasible system for the collection of antimicrobial consumption data in farm animals in Austria. An electronic registry of all antibacterial pharmaceuticals approved in Austria for use in farm animals was created, listing product name, marketing authorisation number, active ingredient, package unit, strength, target species (cattle, swine, poultry), route of administration and indication, and allocating the corresponding code of the World Health Organization (WHO) Anatomical Therapeutic Chemical classification system for veterinary medicines to each substance (ATCvet-code). Different units (absolute quantities, animal daily dose, assumed daily product dose) enabled computation of the amounts of antimicrobials as pure substance, the constituents of a veterinary medicinal product, or the number of administrations. Two data collection systems were evaluated: (1) data transfer from the management software of veterinary practices or the Austrian Poultry Health Service; and (2) on-site data collection by manual data input from prescription records into an electronic registry. A total of 14,267 data sets provided by 18 practices were documented during the period January 2008 to March 2010. The total weight of active substances reported amounted to more than 5.4 tonnes for all species studied. The systems proved suitable for routine data acquisition and were considered in a recent national regulation on the surveillance of sale and consumption of veterinary antimicrobial substances.

View Full Text

Statistics from Altmetric.com

Introduction

Each instance of the medical use of antimicrobial substances may lead to the selection of resistant pathogens (Wilke 2000). Therefore, the prudent use of antimicrobials in both human and veterinary medi­cine is of the greatest importance. The treatment of animals with antimicrobials may be a cause of antimicrobial resistance in human pathogens (van den Bogaard and others 2000). One essential condition for the rational use of antimicrobials is an understanding of bacterial resistance. For analysis of its emergence, data on the consumption of antibiotics are essential (Chauvin and others 2002b). Therefore, the World Health Organization (WHO), in cooperation with the Food and Agriculture Organization (FAO) and the World Organization for Animal Health (OIE), has repeatedly recommended that each country should implement a national monitoring programme to assess the amount of antimicrobials administered to food-producing animals (WHO 2000, 2001, 2003a).

In the European Union (EU), Directive No. 2001/82/EC, as amended, provides the legal basis for the collection of sales data of antimicrobial substances from the pharmaceutical industry and wholesalers by competent national authorities. A method, however, to assess the consumption of antibiotics for specific target species, indications, or application routes based on prescriptions and treatment protocols, which have to be recorded by veterinary practitioners, would allow for a much more detailed analysis. Nonetheless, the routine collection of such data by the authorities requires additional legislation (EMA 2010a).

This report describes a practicable system for the acquisition of antimicrobial consumption data in livestock in Austria. It was initiated and supported by the Ministry of Health for future use by the competent authorities.

Since April 2010, the European Medicines Agency (EMA) has been leading the European Surveillance of Veterinary Antimicrobial Consumption (ESVAC) project (EMA 2011a). The main objective of this project is to develop a harmonised approach across the EU for collecting and reporting data on the use of antimicrobial agents in animals (EMA 2010b). Such data are intended to facilitate the interpretation of patterns and trends regarding antimicrobial resistance, the setting of risk management priorities, the evaluation of the effectiveness of control measures, and the comparison of usage of veterinary antimicrobial agents between human and veterinary medicine as well as between time periods and between countries (EMA 2012a).

By 2011, the ESVAC project had made possible the publication of sales data for antimicrobial substances for nine countries covering the period of 2005–2009 (Czech Republic, Denmark, Finland, France, the Netherlands, Norway, Sweden, Switzerland, UK) (EMA 2011b). In 2012, the EMA published a second ESVAC report with sales figures for 2010 of veterinary antimicrobial agents from 19 EU/European Economic Area (EEA) countries and in 2013 a third report with sales data for 2011 from 25 EU/EEA countries (EMA 2012b, 2013a). In most cases, sales data, provided by pharmaceutical wholesalers, marketing authorisation holders (MAHs), feed mills and pharmacies, allow for analysis of trends on the overall sales of veterinary antimicrobials but offer no accurate details on the amounts of antimicrobials used in different species and weight groups, on the number of dosages, or of treatments (Finres-vet 2007, EMA 2013b). A few European countries (Belgium, Denmark, Finland, France, Germany, Italy, the Netherlands, Norway, Spain, Sweden, Switzerland, UK), therefore, started to collect data on the animal or herd level (EMA 2013b).

In France, MAHs provide information for each antimicrobial product by target species. This information, however, contains estimated measurements and in some cases dosages actually applied in the field and durations are not indicated (ANSES 2013). Antimicrobial consumption data on the basis of veterinarian prescriptions already exist (Chauvin and others 2002a), but implementation of further studies at the animal level was recommended (ANSES 2013).

Switzerland implemented a retrospective study using computerised records of prescriptions and treatments of veterinary practices to obtain information about antibiotic use in different species. Prescription data showed good correlations with sales data of substances used, but only 45 per cent of the prescriptions corresponded to the recommendations given by manufacturers (Regula and others 2009).

Denmark's monitoring programme for veterinary drug use (VetStat) establishes it as a leader in the monitoring of antibiotic consumption in livestock (WHO 2001). The data are collected from different sources, such as pharmacies, feed mills and veterinarians, and are validated using wholesalers' statistics. Detailed data comprising farm identity, animal species, age group, disease, medicinal product, amount and date of purchase, and the prescribing veterinarian are obtained for all antimicrobial agents used in production animals (Hammerum and others 2007). By using the WHO Anatomical Therapeutic Chemical classification system for veterinary medicines (ATCvet) and the animal daily dose (ADD) as a consumption unit, the Danish monitoring approach allows for the calculation of drug use in different species and production classes, as well as in diseases, and permits the modelling of associations between usage and antimicrobial resistance (Jensen and others 2004, WHO 2010).

In Austria, a national antimicrobial resistance monitoring programme was started in 2004. The official AURES report, published yearly, provides information about antimicrobial resistance in the human, veterinary, and phytosanitary sector. In regard to veterinary statistics, data on selected indicator and/or zoonotic bacteria from livestock are used (AURES 2012). Nevertheless, in the past, AURES suffered from a lack of quantitative antibiotic consumption data in livestock, as no such data were available in Austria.

Recently, data concerning the consumption of antimicrobials intended for use in cattle, pigs, poultry, sheep and goats were published for Austria (Fuchs 2012). However, these data merely reflect sales data reported from the pharmaceutical industry and not data detailing the species treated, the number of treatments, or other drug-treatment related aspects.

To complete these essential data, the Austrian Ministry of Health assigned the Austrian Agency for Health and Food Safety (AGES) to develop methods for the monitoring of the consumption of antibiotics in the Austrian cattle, pig and poultry populations in accordance with the Austrian Law of Zoonosis (Zoonosengesetz 2005). The aim of this study, as a part of this programme, was to develop a system to collect accurate data on the use of antibiotics in livestock in Austria, focusing on evaluation of the appropriateness of data acquisition systems.

Materials and methods

In Austria, veterinary medicinal products (VMP) containing antimicrobial substances are prescription only medicines. Only veterinarians running a veterinary dispensary are authorised to dispense, prescribe and sell veterinary drugs to farmers (Tierärztegesetz 1975). Also public pharmacies are allowed to sell veterinary antimicrobials to farmers but only on prescription from a veterinarian. Veterinary practitioners have to keep records of animal treatments and veterinary drugs dispensed to farmers. The type and quantity of drugs used and the diagnosis made have to be recorded. Making use of these obligatory records, two different systems to collect data on antimicrobial consumption were tested and evaluated: (1) data transfer either from the management software of veterinary practices or from the database of the Austrian Poultry Health Service (PHS); and (2) on-site data collection at veterinary practices by manual data input from prescription records (handwritten or printouts of non-transferable electronic files) into an electronic registry.

In this study, a data set is defined as one prescription of a veterinarian representing a single treatment of individual animals, a group treatment (eg, pigs) and/or dispensed drugs. In cases of combination therapies with more than one antimicrobial product administered, each one was recorded as separate data set. Combination products containing more than one antimicrobial substance were entered as a single data set, but the amount of each active ingredient was taken into account.

Recruitment of participants

Only veterinarians personally known to one of the authors were asked to participate in study (convenience sample). All practices specialising in poultry were selected by the management of the PHS. In total, veterinarians from 18 practices voluntarily participated in the study; all were members of the Austrian animal health service (AHS).

Privacy policy

To ensure confidentiality, a privacy policy was posted to all veterinary practitioners participating in the study. To maintain anonymity, a ‘vet-code’ was assigned to each veterinarian.

Drug registry

The processing of collected data was based on a registry of all antibacterial pharmaceuticals approved in Austria for use in farm animals, listing product name, marketing authorisation number, active ingredient, package unit (millilitre, gram, number of packages, bags, injectors), strength (mg/unit), target species (cattle, swine, chicken, turkey), and form of administration (per oral, parenteral, intramammary, intrauterine, topical/external).

The antimicrobial substances were classified according to the WHO ATCvet system. For every VMP, minimum and maximum dosages and treatment periods as given in the Summary of Product Characteristics (SPC) and minimum and maximum weights for each species were defined to allow for the computation of plausibility limits and doses prescribed (Table 1).

TABLE 1:

Excerpt of the drug registry used for the collection of prescription data for the target species cattle

The variables listed in the drug registry are similar to those of the ESVAC template (EMA 2013a); the drug registry of this study also included fields for the calculation of dosages, of plausibility limits and of ADDs and assumed daily product doses (DPDs) (see below).

Data acquisition systems

Electronic data recording

Ten veterinary practices used an electronic practice management system suitable for electronic data transfer. Six of them specialised in cattle and/or swine and four specialised in poultry only. The prescription data, recorded during different periods of time between January 2008 and March 2010, comprised the species treated, animal identification number, number of treated animals, date of treatment, diagnosis, farm identification number, vet-code, marketing authorisation number of the medicinal product, and amount of drug prescribed (eg, millilitre, gram, original package, sachets, piece).

In Austria, most poultry farms are members of the PHS. This organisation runs a databank in which prescription data must be recorded by the contracted veterinary practitioners. The prescription data of four practices were made available by the PHS for the period between July 2008 and June 2009.

On-site data collection

In eight veterinary practices, data on prescriptions or treatments of cattle and swine were solely recorded on-site. Records consisting of handwritten prescriptions or printouts of non-transferable electronic files from June and October 2008 and from February and May 2009 were collected. An electronic data entry page with the following input fields was generated: date of data acquisition, vet-code, date of treatment, farm and animal identification number, species, number of treated animals, diagnosis, name of medicinal product, amount of drug prescribed using the above mentioned units, and treatment period. Wherever possible, drop-down fields facilitated data input. Also acquisition times for the input of data sets were recorded (Table 2).

TABLE 2:

Electronic data entry page for on-site collection of prescription data (excerpt)

Each data set collected on-site (as well as electronic data recording) was checked for plausibility to evaluate the correctness of data acquisition. Non-plausible records were marked for subsequent review.

Consumption units

Active ingredient

Amounts of active ingredients were recorded in milligrams of pure substance. Dosages expressed as international units were converted into mg of active agent using internationally accepted conversion factors (EMA 2011b).

Animal daily dose (ADD, ADDkg, ADDLU)

The unit ADD is derived in this study from the unit defined daily dose (DDD) used in human medicine (WHO 2009). While DDD designates the average maintenance dose per day for its main indication in adults, the ADD analogously refers to the calculated (see below) average maintenance dose per day for a drug used for its main indication per animal of a given species.

ADDkg: In this project, ADDkg was defined in accordance with Jensen and others (2004) as the average maintenance dose of an active ingredient for its main indication and was calculated as mg per kg body mass (bm) per day (mg/kg bm/day). In a first step, the arithmetic mean of the recommended minimum and maximum dose per kg bm per day was calculated for each medicinal product. From these arithmetic means, the median was set as ADDkg. The ADDkg for an antimicrobial substance was set specifically for each target species, route of application, and main indication. This approach yielded different ADDskg for any one given active ingredient in different product groups (oral formulation, premixes, injectables, etc) depending on the parameters mentioned.

ADDLU: The livestock unit (LU) serves as a conversion key in order to compare consumption data between different species based on the body mass. The criteria for the calculation of LUs were set by the Austrian Agency for Marketing of Agricultural Products (AMA) and obtained from this source. One LU was considered to correspond to 500 kg bm for all species (one adult cow ∼1 LU, one fattening pig ∼0.15 LU, one layer ∼0.004 LU) (AMA/OPÜL 2007). For the calculation of an ADDLU, the ADDkg was multiplied by 500 (mg/500 kg bm/day).

Number of animal daily doses per livestock unit (nADDsLU)

Dividing the number of ADDsLU by the number of (treated) LUs yields the number of ADDsLU standardised per LU (nADDsLU).

Calculation of the number of treated LUs: The treated population is defined as the total number of LUs produced in one year by all farms, in which at least one treatment was recorded. The number of LUs produced per farm can be calculated using census data in cattle and pig farms and production data from the poultry farms.

Assumed daily product dose (DPD, DPDkg, DPDLU)

The unit assumed DPD was considered as a second unit for daily doses to illustrate the prescribing patterns of veterinarians. The DPD was calculated for each VMP by adjusting the recommended maximum daily dose by an assumed factor of 0.8, correcting for the fact that the maximum doses are not used in every treatment. This calculation resulted in species and product specific DPDs per animal (DPD), per kg bm (DPDkg) or per LU (DPDLU). For intramammary preparations, including products for dry-off, no arithmetical adjustments were made and the recommended dose was taken as DPD.

Results

Data acquisition

Of the 18 veterinary practices participating in this project, eight specialised in cattle, four each focused on swine and poultry, and two practices provided data from swine as well as from cattle.

In total, data from 2078 animal stocks including all animal species considered for this study were collected, leading to 14,267 data sets acquired during the investigation period from January 2008 through March 2010 (Table 3). The total weight of active substances reported amounted to more than 5.4 tonnes for all species studied, corresponding to a total of 576,242 ADDsLU and 631,939 DPDsLU.

TABLE 3:

Source and number of data acquired between January 2008 and March 2010

Electronic prescription data

The 10 practices operating an electronic practice management system provided a total of 9141 data sets (Table 3).

On-site data collection at surgeries

A total of 5126 data sets from eight practices were recorded on-site (Table 3). The acquisition times from 2087 data sets were automatically determined and amounted to an average of 40 seconds per data set.

Plausibility

As a result of this study, 2707 out of 5126 prescriptions acquired by the on-site data collection failed the plausibility check and were marked permanently. Data records were considered non-plausible when plausibility limits were violated, due to non-conforming prescription recording. In most cases, plausibility limits were exceeded during data input and were mainly restricted to practices specialising in swine. ‘Under-dosing’ was very rare and remained negligible.

Discussion

For a correct calculation of the consumption of antimicrobials in connection with the analysis of resistance data, the application of appropriate units for drug amounts and dose units is crucial. According to FAO/OIE/WHO recommendations, measurements of antimicrobial drug consumption in husbandry should include (WHO 2003a):

  • at a minimum, data concerning the overall use of each antimicrobial agent in kilograms of active agent;

  • stratified data on usage per animal species;

  • data using a ‘defined daily dose’ (DDD)-equivalent concept.

In this study, the amount of antimicrobials is expressed as milligrams of pure active ingredient, which is in line with previous publications on consumption statistics (Stege and others 2003, FINRES-Vet 2007, Moulin and others 2008). The assignment of the ATCvet-code to active agents facilitates grouping, the evaluation of usage patterns of consumed antimicrobials, and a direct comparison with human consumption data on antimicrobial substances, to which consonant ATC-codes are allocated (Grave and others 2006). For combination preparations, the ATCvet-code for the combination was used and for the calculation of the overall amounts consumed all active ingredients were considered.

Grave and others (1999), as well as Arnold and others (2004), have emphasised that data only presenting the quantity of active ingredient might give a false impression of the actual treatment frequency. In human medicine statistics, the DDD is controlled by the WHO Collaborating Centre for Drug Statistics Methodology and is an internationally accepted measure of drug usage (Jensen and others 2004). In veterinary medicine, an internationally accepted equivalent to the DDD has not yet been established (Moulin and others 2008).

Taking into consideration the specific provisions of use for veterinary drugs, the ADD of a VMP was separately defined in this study, also assuming the average maintenance dose per day for its main indication (mg substance/kg bm/day). However, if a drug was licensed for different species, variable indications, or routes of administration including those routes exclusively used in veterinary medicine, (eg, intramammary or intrauterine use), such particulars were distinctively considered in our drug registry.

The conversion of the quantity of active components in ADD allows the frequency of administration of various differently dosed substances to be compared. ADD should be determined on the basis of national recommendations as well as on information given in the pharmaceutical and academic literature (Jensen and others 2004). As ADDs are specifically defined in this study for a given active substance, any varying dosage recommendations for different medicinal products containing the same substance are without influence on the calculation of the nADDs. So far, however, the unit ADD has not been officially established.

Analogously, division of the total weight of a given active substance consumed within a certain time period by the corresponding ADDLU (resp. DPDLU) allows the calculation of the nADDsLU (resp. nDPDsLU) administered for this period. In combination with detailed data sets collected in this study, this strategy permits an estimation of the average frequency of administration for a specific diagnosis, route of administration, and target species (even when split into production type, if appropriate) of a particular antimicrobial substance (Fuchs and others 2012).

In Austria, the legal basis for the dosing of veterinary drugs is the approved SPC. Doses may differ between different VMPs and from the ADDkg as calculated in this study. For comparative analyses of the number of administrations of a particular product, Muller and others (2006) suggested use of the number of prescribed daily doses (nPDDs). In human medicine, the PDD was defined by the WHO as the average daily dose prescribed according to a representative sample of prescriptions (WHO 2003b).

Prescriptions often lacked the exact number of animals as well as the exact weight of the animals being treated. The actually prescribed dose per body mass could, therefore, often not be identified. For this reason, the DPD was considered as a second unit for a daily dose.

As the DPD was determined using the recommendations of the manufacturer of a specific veterinary pharmaceutical product, it most likely represents a more realistic dosage unit and may help to clarify true prescribing patterns.

The 80 per cent limit specified for the definition of this unit (80 per cent of the highest dose stated by the SPC of the medicinal product) was taken because it mirrors with high probability the actual average dose applied by practitioners; this dose must realistically be assumed to be higher than the ADD and somewhat lower than the maximum dose.

Both units, ADD and DPD, are calculated units and enable estimation of the frequency of administration (on the basis of total consumption data), even if information on the exact number of treated animals and/or their body mass is missing in the prescription recordings, with nADDs referring to an active substance and nDPDs to a specific product. In cases of the ADD being significantly higher than the DPD the dosage information as given in the SPC might be too low, leading to an underdosing and vice versa.

A large number of prescriptions acquired from the electronic and on-site data collection failed the plausibility check. Since prescriptions for pigs often imply group treatments, from such records the number of treated animals could only be roughly estimated, leading to an ‘overdosing of antimicrobials’ in the majority of cases during the evaluation process.

In cattle, predominantly drug dispensing for follow-up treatment of chronic mastitis and group treatments of respiratory diseases resulted in the plausibility limits being exceeded. Again, information on the number of animals treated was missing and/or the treatment duration was not specified. For the computation of plausibility limits the entry of the number of treated animals and the duration of treatment are mandatory.

In this regard, veterinarians have to pay particular attention to their prescription recordings. In this study, the plausibility check was primarily a safety measure for correct data input during the on-site data collection.

It became evident that for the collection of antimicrobial consumption data in animal husbandry, transfer of electronic data ­provided by veterinary practices was most appropriate. Nevertheless, on-site data acquisition was also a feasible method to collect prescription data. With an average time needed to record a data set (representing one prescription) of 40 seconds, and allowing for an average recording of 90 data sets per hour, practices not equipped with an electronic management system can be included in monitoring programmes. Nonetheless, the time needed for compiling data sets on-site decreased greatly with the quality of the prescription records. Poor handwriting and incomplete records resulted in a prolonged acquisition time. Usually, printed versions of prescription data were more accurately arranged, considerably more often complete, and easier to manage than hand written prescriptions. Both acquisition systems, however, are substantially dependent on complete data records.

Only prescription records from veterinary surgeons, but not drug or treatment related farm accounts, were surveyed in this study. Discrepancies may exist between prescription records of veterinarians and entries in the management diary of farmers or their actual drug administration to animals. The frequency and significance of such divergences could not be considered within the scope of this study.

Participation of veterinary practitioners in this study was voluntary. The number of cooperating practitioners as well as the farms attended by these practitioners were not selected randomly but were based on the design of this study, aiming to ascertain effective surveillance techniques for antimicrobial drug consumption on a national scale rather than to obtain statistically sound results. The data collected do not represent a random sample and do not meet the requirements for a probability sampling and are, therefore, not shown in detail. A reliable assessment of the amount of used antibiotics can only be done with data from official consumption surveillance programmes. Austrian antimicrobial overall sales data for 2011 amount to 53.4 tonnes (EMA 2013a), that is, approximately 10 times the amount recorded in this study. At present, solid extrapolation data on overall consumption in a wider Austrian farm animal population are not available.

Conclusion

The following conclusions can be drawn:

Collection of electronic data in combination with on-site collection emerged as a practicable system to monitor the consumption of antimicrobials in livestock in Austria.

The structure of the drug registry designed for this study proved to be operable in respect to (1) practicability, (2) the assortment of units for antimicrobial consumption and daily dosages, and (3) applicability as a data source for further analyses using a statistical database for data evaluation.

Primary compatibility tests showed that prescription data obtained from veterinary surgeries and electronic data provided by the PHS are qualitatively and quantitatively transferable in statistical data banks and, thus, can be matched with presently less detailed pharmaceutical sales data provided to the competent authorities (Obritzhauser and others 2014, personal communication).

The dose units and the methods of acquiring antimicrobial consumption data elaborated in this study were considered in a national regulation on the surveillance system on sale and consumption of veterinary antimicrobial substances which came into force very recently (Veterinär-Antibiotika-MengenströmeVO 2014). The first veterinary reports for official statistical data evaluation will be due March 31, 2016.

Acknowledgments

This study was made possible by the financial support of the Federal Ministry of Health. We thank the Austrian Poultry Health Service and the Austrian Animal Health Service for collaboration. We sincerely thank the participating veterinarians for providing their treatment records.

References

View Abstract

Footnotes

  • Provenance: not commissioned; externally peer reviewed

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.