Antimicrobial consumption on Austrian dairy farms: an observational study of udder disease treatments based on veterinary medication records

Study population

Veterinary treatment data were collected as part of a larger interdisciplinary project entitled “ADDA—Advancement of Dairying in Austria”, which included both academic and industry partners (e.g., cattle breeding associations, milk processors, farmers’ unions). A convenience sample of veterinary practices primarily involved in dairy production medicine was contacted directly by the authors. Seventeen veterinary practices (29 veterinarians) provided AMC data in this study. The number of veterinarians per practice ranged from one to five (mean 1.7, median 1), with 13 practices being run by a single practitioner. Almost half of the practices treated only farm animals (46.7%), with the remainder (53.3%) being mixed practices (i.e., treating both farm and small animals).

Enrolled veterinarians actively recruited interested farmers from their catchment areas (respondent-driven sampling). As such, this was not a randomised sample of veterinarians and dairy farmers throughout Austria, but was a convenience sample within an observational study. The authors did not influence the veterinarians’ choice of farmer. No restrictions were made with respect to farm size, production system, freestalls or tie-stalls, alpine or valley farms, etc. However, all veterinarians and dairy farmers were members of the Austrian Animal Health Service, and the dairy farmers also participated in the national milk-recording scheme (German: Landeskontrollverband, LKV). The study was discussed and approved by the institutional ethics and animal welfare committee of the University of Veterinary Medicine, Vienna (Ref No. ETK-13/11/2015), in accordance with good scientific practice (GSP) guidelines and national legislation. A total of 283 dairy farmers in the federal states of Lower Austria, Upper Austria, Styria, Carinthia and Tirol initially agreed to take part in the study; 253 of these signed an informed consent form and approved the use of data from their farms for research purposes within the context of this study.

Data collection

The data collection period with respect to antimicrobial treatments began on 1st October 2015 and concluded on 30th September 2016. An online interface was created, using the Electronic Herdbook program, to which the respective veterinarians could upload AMC data from their practice management software. To simplify the process for participating veterinarians, electronic veterinary treatment data for all bovine animals (cows, calves, youngstock) were collected. Table 1 illustrates the type of electronic data collected from veterinarians with respect to each treatment. As well as the farm identification number, an animal identification (ear tag) number was entered for each animal, meaning that analysis at both herd and animal level was possible. Production type for the treated animal was also recorded as the small-scale structure of Austrian dairy farms means that some male calves/youngstock are often kept on dairy farms for fattening.

These data were then validated based on animal and farm identification data from the central cattle database (German: Rinderdatenverbund, RDV system), and a comparison of medication licensing number with the database of the Austrian Medicines and Medical Devices Agency. Antimicrobial use data were then extracted from the dataset into a Microsoft (Microsoft Corporation, Redmond, WA, USA) Excel spreadsheet for further analysis. Statistical analysis was carried out using Microsoft Excel and the R software package, Version 3.3.2 (R Core Team, 2016). Where necessary, data were given pseudonyms (according to local statutory requirements) so that neither the veterinarian nor the farm was identifiable.

According to an Austrian law passed in 2002, diagnoses must be documented for any veterinary medication dispensed or administered to livestock (BMGF, 2002). Diagnoses were reported by the herd veterinarian according to the official cattle disease diagnosis code system (as defined by the Austrian Federal Ministry of Health in 2006), which includes more than 60 possible clinical diagnoses (plus “not otherwise specified” for complexes which do not fit into other listed categories) (Egger-Danner et al., 2012). With respect to udder disease, possible diagnoses included “acute mastitis”, “chronic mastitis” or “other udder disorders” (including disorders of the teat or udder skin, udder oedema, teat injuries etc.). (NB. The current analysis did not include prophylactic drying off). Diagnoses were assigned by the herd veterinarian according to his/her clinical assessment of the animal, no additional external criteria were applied.

If electronic data were incomplete or the veterinary practice software proved incompatible with the online interface, data were provided by the participating veterinarians as paper records, which were then entered into a Microsoft Access database by one of the authors. This database was then exported into a Microsoft Excel spreadsheet and combined with the electronic treatment records for further analysis.

Antimicrobial consumption data

AMC data were analysed by diagnosis code (including all codes relating to udder disease, but excluding dry cow therapy), mode of application (e.g., intramammary instillation, injection) and whether the substances used were classed as “highest priority critically important antimicrobials” (HPCIAs) (World Health Organization, 2017). Further analyses were done by antimicrobial class, herd and per veterinary practice.

Amounts of antimicrobial active substance were calculated by multiplying the volume of veterinary medicinal product administered to the animal (usually given in millilitres [ml]) by the concentration of the antimicrobial active substance (e.g., mg/ml) to give the total mass of antimicrobial active substance in milligrams (mg). The number of DDD_vet administered were calculated using the DDD_vet values assigned to the individual antimicrobial substances and animal species by the European Medicines Agency (EMA, 2016b) as follows:

${\displaystyle {\mathrm{nDDD}}_{\mathrm{vet}}=\frac{\text{amount active substance (mg)}}{{\mathrm{DDD}}_{\mathrm{vet}}\text{ for that antimicrobial active substance}}}$

To calculate the nDDD_vet/cow/year, the following formula was used:

${\displaystyle {\mathrm{nDDD}}_{\mathrm{vet}}∕\text{cow/year}=\frac{{\mathrm{nDDD}}_{\mathrm{vet}}}{\text{Production days}\times \text{Standardised liveweight (500 kg)}}\times 365}$

Only cows were included in the analysis of antimicrobial use for udder disease treatments. The standardised liveweight of 500 kg for a dairy cow was taken from European Medicines Agency guidelines (EMA, 2013). Production days were defined as the actual number of days an animal was kept in the herd during the observational period (according to livestock movement data from the central cattle database). The number of cow-years per herd was calculated using the total number of cow production days per herd divided by 365.

Intramammary treatment injectors were classed as “one treatment per tube”. In accordance with the EMA definitions for DDD_vet for intramammary treatments, where the defined daily dose is given as one treatment per teat (EMA, 2016b), no calculations were necessary with respect to the amount of active substance. The DDD_vet for intramammary treatments as defined by the EMA is based on daily dose per cow, rather than per kilogram liveweight. The calculation of the DDD_vet/cow/year for intramammary treatments did not, therefore, include the standardised liveweight of the cows treated. The number of production days per farm, based on animal movement data, was, however, always included in the number of daily doses per cow and year.

The number of defined daily doses per cow and year for intramammary therapies was calculated as follows:

${\displaystyle {\mathrm{nDDD}}_{\mathrm{vet}}∕\text{cow/year(intramam.)}=\frac{{\mathrm{nDDD}}_{\mathrm{vet}}}{\text{Production days}}\times 365}$

Oxytetracycline sprays for the local treatment of udder disorders are not included in this analysis as no DDD_vet value has been assigned by the EMA to antimicrobial sprays. This decision led to the exclusion of six spray cans of oxytetracycline dispensed over the one-year period. Similarly, dry cow treatments (DCT) are not included here, as no DDD_vet values have been allocated to these products by the European authorities (EMA, 2015; EMA, 2016b). Calculation of antimicrobial treatment with dry cow injectors is instead standardised by the EMA using the Defined Course Dose (DCD_vet), which is classified as four treatments per udder. To avoid confusion with DDD_vet values, it was decided not to include DCT in the current analysis, which concentrated on the frequency of antimicrobial treatment of mastitis and other udder diseases, rather than the primarily prophylactic use of antimicrobials at drying off.

Statistical analysis

The total number of DDD_vet/cow/year was calculated for each herd and the mean, standard error of the mean (SEM) and median of the herd values were then determined for each udder disease diagnosis code as assigned by the treating veterinarian. The analysis was based on the treated population of dairy cows, however, descriptive statistics were calculated only for those herds where animals were treated for a particular diagnosis. The number of treatments calculated on the basis of defined daily doses are presented as boxplots, illustrating the minimum, maximum, mean, median, 25th and 75th percentiles and outliers of DDD_vet/cow/year values.

The median DDD_vet/cow/year values for acute and chronic mastitis treatments were compared using a non-parametric statistical test, the Wilcoxon rank sum test with continuity correction for unpaired samples. The diagnosis group “other udder disorders” was not included in the statistical analysis as the number of cases reported was too small to allow for a meaningful analysis.

Farm demographics

The number of herds enrolled (and whose herd veterinarians provided AMC data) in the study is described in Fig. 1. Based on the 253 herds where farmers provided informed consent, herd size ranged from nine to 240 head of cattle (mean 55.6; median 45), with numbers of dairy cows ranging from five to 98 (mean 26.2; median 21). The vast majority (73.8%) of cows were of the Simmental (Austrian Fleckvieh) breed, with the remainder made up of Brown Swiss (Braunvieh, 10.9%) and Holstein-Friesians (12.7%). Approximately two-thirds (65.5%) of farms kept their cows in freestalls, primarily cubicles or straw-yards, whereas 34.5% used a tie-stall system. As a consequence of these housing systems, just over half (52.6%) of the farms had milking parlours, with 32.9% milking via a vacuum-line in the barn. Twelve farms (4.8%) used automated milking systems and a small number (4%) still milked directly into a bucket.

Veterinary medication records

Medication data on all administered and dispensed antimicrobial products were collected electronically for 248 enrolled dairy farms from the practice management software systems of 17 veterinary practices (Fig. 1). Where electronic AMC data were incomplete, paper records were included in the dataset. This was necessary for a total of 48 (19.4%) farms and made up 10.6% of total treatment data.

Overall, 6,334 single antimicrobial treatments of dairy cows were collated during the one-year observational study period.

Reported diagnoses for udder disease

Once calves and youngstock were excluded, 232 herds were determined to have treated dairy cows with antimicrobial agents at least once over the one-year study period (for details see Fig. 1). The most commonly reported initial diagnosis treated with antimicrobial substances within the group of “udder diseases” over the one-year period was acute mastitis (Table 2). Diagnoses are reported as designated by the prescribing veterinarian.

Antimicrobial consumption

A total of 2,304 antimicrobial treatments were designated as being for udder diseases, this group made up 36.4% (2,304/6,334) of all antimicrobial treatments in this study population. Overall, 92% (214/232) of all herds where antimicrobial substances were administered to lactating cows treated these animals for “udder disease”, with 79% of all farms using HPCIA for these treatments (Table 3). Concentrating the analysis solely on these 214 herds where cows were treated for udder disease, the mean DDD_vet/cow/year was determined to be 1.33 (±0.09 SEM) for all udder disease treatments, with a median of 0.86, ranging from 0.04 to 9.99 DDD_vet/cow/year (Table 3 and Fig. 2). Cows suffering from acute mastitis were reported to have been treated on 205 farms, with a mean DDD_vet/cow/year of 1.21 (±0.09) for all antimicrobial consumption, and a mean DDD_vet/cow/year of 0.71 (±0.08) for HPCIAs (Fig. 3A and Table 3). Chronic mastitis cases were treated much less frequently, on only 72 farms, with a mean DDD_vet/cow/year of 0.48 (±0.05) for all antimicrobial consumption and a mean DDD_vet/cow/year of 0.25 (±0.04) for HPCIAs (Fig. 3B and Table 3).

The highest level of use by antimicrobial class was determined to be penicillins at a mean of 0.16 (±0.02) DDD_vet/cow/year administered systemically and 0.63 (±0.08) DDD_vet/cow/year for intramammary use (Table 4). With respect to the maximum level of use on individual farms, however, the fourth generation cephalosporin, cefquinome, was reported to be administered at the highest level for both intramammary and systemic use (ranging from 0.04 to 6.65 DDD_vet/cow/year and 0.01 to 1.11 DDD_vet/cow/year, respectively) (Table 4).

The Wilcoxon rank sum test determined that the difference between the medians of the herd level DDD_vet/cow/year with respect to acute mastitis and chronic mastitis treatments was significant at the 5% level, (acute mastitis n = 205: median 0.71, 95% CI [0.61–0.89] and chronic mastitis n = 72: median 0.34, 95% CI [0.23–0.52]), with a test statistic W = 10,734, p-value < 0.001.

Treatments by veterinary practice

Antimicrobial treatments for udder disease by veterinary practice are shown in Fig. 4. To ensure the anonymity of practices participating in this study, the number of veterinarians per practice is not shown. The proportion of defined daily doses by cow and year to treat udder disease with non-critical antimicrobial substances varied considerably between veterinary practices, with the use of HPCIAs extending from a minimum of 8% of all defined daily doses for antimicrobial substances administered and/or dispensed on the study farms by practice #502 to >80% HPCIAs by practice #845.