Annex 1: Final report for Defra project OD2006.

Annex01:The detection of antimicrobial-resistant faecal bacteria on 25 pig and poultry-meat farms in the UK.

Eve Pleydell, Lourdes Migura, Simon Barnes, Felicity Clifton-Hadley and Rob Davies

Aims

  • Collation and analysis of antimicrobial drug usage across the 25 farms.
  • Comparisons of the detection of antimicrobial resistant Escherichia coli and Enterococcus faecium across the 25 farms.
  • Statistical modelling of detection of antimicrobial resistant commensal organisms with drug usage and farm management variables.

Materials and Methods

Project foci

Two organisms were investigated in this study, Escherichia coli (E. coli) and Enterococcus faecium (Ent. faecium). These organisms were chosen as representatives of the aerobic and facultatively anaerobic Gram-negative and Gram-positive bacterial gut flora. Each species has been shown to be adept at acquiring resistance to antimicrobials and is also capable of transferring mobile resistances to other bacterial species within the gut. Therefore these organisms may act as markers of resistances present within a particular environment at a given time. In addition E. coli represents the Enterobacteriaceae family which includes the Salmonella species’, thus potential complications arising from the statutory reporting of Salmonella isolations were avoided. Salmonella are also more clonal in nature than E. coli in livestock and the farm environment and therefore Salmonella may not represent the best markers of the full range of resistances present on a farm.

The antimicrobial drug resistances that were studied were chosen to reflect drugs of importance in human or veterinary medicine, or both, and encompassed a variety of resistance mechanisms (see Table 1).

Recruiting farms

Suitable commercial farms were sought by contacting specialist veterinary practices and by contacting researchers that had already established links with farms for other research studies. The final group of 25 willing participants encompassed a variety of farming practices and were located across southern and middle England.

Data collection

Sampling was conducted from December 2001 through to November 2003. Wherever possible, each farm was visited at least once in the first sampling year and once in the second year. Over this two-year period, the research team conducted a total of 57 farm visits and carried-out over 24,000 bacterial isolations.

Upon arrival at the farm the principle investigator and the farm manager completed a questionnaire together. The areas of farm husbandry that were investigated were as follows:

  • Herd/flock size and structure
  • Recent production figures for the unit
  • Proximity of possible sources of resistant bacteria to the farm
  • Cleaning and disinfecting regime
  • Feed and feed additives
  • Antimicrobial use over the 12 months preceding the visit.

Table 1. Details of the antimicrobial drug resistances that were investigated within this project.

Drug

(commercially available) /

Relevance

/ Basis of Resistance Mechanisms
Escherichia coli / Ampicillin / Used frequently in human and veterinary medicine. / Beta-lactamase genes: chromosomal and/or on mobile genetic elements(6).
Gentamicin / Used more frequently in human medicine, cross-resistance seen with the veterinary therapeutic drug, apramycin. / Genes encoding for drug inactivation enzymes, present on plasmids/transposons(7).
Ciprofloxacin / Used in human medicine, cross-resistance seen with veterinary therapeutic fluoroquinolones. / Chromosomal mutation/s; possible role of efflux pumps; and reports of plasmid-borne resistance in Enterobacteriaceae from China, USA and UK((4;5;9;10).
Enterococcus faecium / Erythromycin / Used more frequently in human medicine, cross-resistance seen with veterinary drugs such as tylosin. / Modification of drug-target site encoded by plasmid- and transposon-borne genes(7).
Vancomycin / Used in human medicine, cross-resistance seen with the growth promoting agent avoparcin. / The vanA operon of genes prevents disruption of cell wall synthesis by vancomycin; carried on transposons and plasmids(2).

An average of 60 pooled-faecal-samples were collected on each visit. Each pooled-sample was comprised of a generous pinch of faeces from each of eight fresh droppings that were located within a specified area. The aim on each visit was to collect samples from all the age groups or production sectors present on the farm on that day. The number of samples collected from each area of the farm was chosen to reflect the proportion of the total stock that was present in that area. Where appropriate, environmental samples such as dust, surfaces of feed equipment, cracks in the floor or walls were also collected using either gauze swabs which were then placed into jars of buffered peptone water or wand swabs in charcoal transport media. All samples were held at 4ºC during transportation and prior to processing in the laboratory.

Table 2. Details of the farms that participated in this study.

Farm ID / Species / System / Type and Age of Stock on Farm[a]
1 / Chicken / Organic / Parent Broiler Breeding Flock: 1-434 days
2 / Chicken / Organic / Broilers: 1-70 days
3 / Chicken / Organic / Broilers: 21-80 days
4 / Chicken / Organic / Broilers: 1-72 days
5 / Chicken/Turkey / Organic / Broilers: 1-80 days / Turkeys: 1-180 days
6 / Chicken / Organic / Broilers: 1-84 days
7 / Chicken / Organic / Broilers: 26-73 days
8 / Chicken / Conventional / Housed Broilers: 1-60 days
9 / Turkey / Conventional / Housed Turkeys: 1-148 days
10 / Chicken / Conventional / Housed Broilers: 1-42 days
11 / Chicken / Conventional / Housed Broilers: 1-42 days
12 / Chicken / Conventional / Housed Broilers: 1-56 days
13 / Chicken / Conventional / Free-Range Broilers: 1-63 days
14 / Pig / Organic / Breeding/Rearing/Finishing Unit: 0-168 days
15 / Pig / Organic / Breeding/Rearing/Finishing Unit: 0-200 days
16 / Pig / Organic / Breeding/Rearing/Finishing Unit: 0-140 days
17 / Pig / Organic / Breeding/Rearing/Finishing Unit: 0-175 days
18 / Pig / Organic / Breeding/Rearing/Finishing Unit: 0-200 days
19 / Pig / Conventional / Finishing Unit: 84-161 days
20 / Pig / Conventional / Breeding/Rearing Unit: 0-91 days
21 / Pig / Conventional / Breeding/Rearing/Finishing Unit: 0-140 days
22 / Pig / Conventional / Breeding/Rearing/Finishing Unit: 0-147 days
23 / Pig / Conventional / Finishing Unit: 30-100 days
24 / Pig / Conventional / Finishing Unit: 91-154 days
25 / Pig / Conventional / Rearing Unit: 28-84 days

Laboratory methods

Detection of resistant bacteria within the samples was carried out by diluting the pooled-faecal samples using an equal weight to volume of buffered peptone water and then homogenising the resultant mixture using a vortex mixer. The homogenised faeces were then plated onto selective agar media that incorporated antimicrobials at specified breakpoint concentrations. The environmental samples were incubated in buffered peptone water overnight at 37°C prior to plating.

In an attempt to harmonise the laboratory procedures used in this project with those used by other research groups, the breakpoints[b] utilised for the E. coli were chosen to harmonise with the Public Health Laboratory Service (PHLS, UK) 2001 breakpoints and those for the Ent. faecium were chosen to harmonise with the Danish Veterinary Laboratories (DVL) 2001 breakpoints.

Table 3. A comparison of the breakpoints chosen for OD2006 with those utilised by four different organisations in the year 2001.

Breakpoints (mg/L)

DVL[c] / PHLS[d] / BSAC[e] / NCCLS[f] /

OD

2006

E. coli

/

Ampicillin

/ 16
(32) / 8 / 16 / 32 / 8
Gentamicin
/ 8 / 4 / 2
(4) / 16 / 4
Ciprofloxacin / 2
(4) / 1 / 2
(1) / 4 / 1
Ent. faecium
/ Erythromycin / 4
(8) / - / - / 8 / 4
Vancomycin / 16
(32) / 4 / 8 / 32 / 16

The E. coli were grown upon Chromagar ECC agar and incubated at 37ºC for 12-18 hours. The enterococci were grown upon Slanetz and Bartley agar for 48 hours at a primary incubation temperature of 42ºC.

After incubation a plate was scored as positive if there were more than three colonies of expected morphology growing outside the initial area of inoculation. One to three colonies were then sub-cultured onto the same media incorporating the appropriate antimicrobial. To allow for further characterisation of the bacteria grown, a selection of isolates from each visit were catalogued and stored on to Dorset egg slopes or frozen with glycerol at -80ºC.

Calculation of quantities of antimicrobials used upon each farm

Data on drug use was collected in a variety of ways on each farm:

  • Direct conversation with the farm manager
  • Veterinary invoices
  • Farm medication records
  • Feed-bag labels

Based upon these figures estimates were made for each farm of the amount of active therapeutic agent administered over the 12-months preceding the first visit to that farm for each of the two sampling years of the project.

The quantities of growth promoters administered were calculated for those farms using these drugs, by estimating the total amount of food consumed over a 12-month period by the sector of stock consuming growth promoters. These figures were calculated using the farm production figures and supplementing this information with published production figures where necessary.(1) The quantity of growth promoting agents used was then calculated using the feed-incorporation concentrations detailed on the feed labels from that farm.

Statistical models

Due to recent developments in mixed-effect statistical modelling it is now possible for multivariable models to be fitted to clustered data. In this study pooled-faecal samples were clustered within visits, and visits were clustered within farms. As the response variable was binary (detection or non-detection of the specified resistant bacteria within a sample) generalised linear mixed-effects models were fitted using the glmmPQL function with a binomial family in R 2.1.1. Initially a series of single fixed-effect models was run for each animal-bacterium-drug combinations that were investigated. The fixed-effects fitted were as follows:

1.For Poultry and Pigs

  • Farming method:
  • Conventional.
  • Organic.
  • Husbandry system:
  • Livestock are permanently housed.
  • Livestock have access to pasture or range.
  • The number of animals finished per year.
  • Use of therapeutic antimicrobials:
  • Rare or never.
  • Used within the 12-months preceding the visit.
  • Use of antimicrobial growth promoters:
  • Not used on farm.
  • Administered to some/all livestock.
  • Total quantity of antimicrobial drugs administered on farm over 12-months.
  • Quantity of drugs used in 12-months per kilogram liveweight finished product.
  • Total quantity of therapeutic antimicrobials administered over 12-months.
  • Use of beta-lactam antimicrobials within 12-months of visit.
  • Use of enrofloxacin within 12-months of visit.
  • The number of different antimicrobial drugs used by the farm.
  1. For Poultry
  • Livestock species:
  • Chicken.
  • Turkey.
  • Stock rotation:
  • Single-age, all-in-all-out system.
  • Multi-age continuous replacement system.
  • The age of the birds sampled at the time of the visit.
  • The number of birds per group/house.
  • The type of house:
  • Conventional house.
  • Mobile barn.
  • Old farm-shed.
  • Use of lincomycin-spectinomycin within 12-months of visit.
  • Disinfecting regime:
  • Rarely or never used.
  • Houses disinfected between batches of birds.
  • Regime for disinfecting waterlines:
  • Rarely or never undertaken.
  • Waterlines flushed and disinfected between batches of birds.
  1. For Pigs
  • Type of unit:
  • Rearers or Finishers only.
  • Breeding stock present.
  • The status of the pigs sampled at the time of the visit:
  • Growing / Finishing.
  • Farrowing Sow / Sucking Piglets / Weaned Piglets.
  • Dry sows and replacement gilts.
  • Service area.
  • Presentation of food:
  • Pellets.
  • Other (meal or liquid).
  • Use of aminoglycosides in 12-months preceding visit.
  • Use of tetracyclines as above.
  • Use of macrolide or lincosamide drugs as above.
  • Frequency of disinfection:
  • Never.
  • On average once every 6-months.
  • Annually or less often.
  • Depopulation practised as a management tool.
  • Sampling method:
  • Eight pinches of faeces pooled in a single sample pot.
  • Wand swab inserted into eight faecal masses and transported in charcoal media.

Using the results of the single fixed-effect models, multivariable models were then fitted using forward stepwise and backward elimination strategies to determine the models that fitted the data most accurately. Unlike linear mixed-effect models fitted using maximum likelihood, one cannot use Aikaike’s Information Criteria or Log Likelihood to compare glmmPQL models.Therefore for this work the models of best fit were judged to be those models whereby statistically significant associations (at the 0.05% level) were found between the outcome variable and each of the fixed-effect variables that were incorporated into the model.

Results

Use of antimicrobial drugs

Tables 4 and 5 detail the variety of drugs used across the 25 farms and provide estimates of the total amount of each drug that was administered within a 12-month period.

Table 4. The antimicrobial agents administered across the 13 poultry farms.

Antimicrobial Agent / Number of Farms Using Drug / Amount Administered over 12-months (kilograms) / Percentage of total poultry drug use
Avilamycin: growth promoter / 4 / 192.5 / 61%
Amoxicillin / 6 / 86.5 / 27%
Tylosin / 1 / 23.5 / 7%
Lincomycin-Spectinomycin / 3 / 8.2 / 3%
Potentiated Sulphonamides / 1 / 3.6 / 1%
Tilmicosin / 1 / 1.9 / 1%
Enrofloxacin / 1 / 0.4 / <1%
Total / 317kg

For both pigs and poultry, the most heavily used drug (50-61% of the total drug use) was the growth promoting agent avilamycin[g]. Furthermore on 3/12 pig units tetracycline and tylosin were routinely incorporated into the feed for a given age of growing pig for 10-35 days depending on the farm. For both drugs this routine incorporation into the feed accounted for 96% of the total use of these two drugs, and all growing pigs on these units would have been exposed to these drugs. Most of the remaining porcine drugs are being used at much lower levels and are variously added to the feed or water of specific groups of ill animals or administered by injection on an individual animal basis as necessary. On the organic pig farms all antimicrobial use was via injectable antimicrobials to individual animals.

The conventional poultry units used a narrower range of different drugs than the pig units and, other than avilamycin, no drugs were routinely incorporated into the feed for extended periods of time. The section of farms surveyed here relied heavily upon amoxicillin as a first-line therapeutic agent commonly using a course of 3-days and administering the compound via the drinking water. It is also worth noting that two of the three farms using the combination compound lincomycin-spectinomycin were routinely dosing all chicks for the first-three days after their arrival onto the farm as a preventative measure against yolk-sac infection and necrotic enteritis. Therefore, whilst the total amount of lincomycin-spectinomycin used appears to be relatively low, all the birds on these farms were being exposed to the drug. Such prophylactic use of lincomycin-spectinomycin is now very common in the UK broiler industry.

Figure 1 displays the breakdown of the estimates for the total quantity of drugs, in kilograms, used on each farm over a 12-month period. It is apparent that there was a wide discrepancy in drug administration practices across the 25 farms. Figure 2 incorporates an indication of the size of the farm and the species being reared into the calculation of annual drug use. Using the production figures provided by the farm managers, the total number of kilograms of live-weight finished bird or pig over a 12-month period was calculated for each farm and micrograms of active drug administered per kilogram of live-weight product was calculated. This markedly alters the shape of the chart for the poultry farms and shows that the farms using very high levels of antimicrobial (over 40kg) are the farms producing large numbers of finished birds. The shape of the pig chart has also altered to display a wide range of intensities of drug use across the conventional units from 5 to 122µg per kg finished live-weight. The finishing unit applying ~122µg/kg live-weight pig was administering ~9g of active antimicrobial drug per finished pig, it is unknown if the pigs on this farm had received further antimicrobial treatment before arriving on the unit.

Table 5. The antimicrobial agents administered across the 12 pig units.

Antimicrobial Agent / Number of Farms Using the Drug / Amount Administered over 12-months (kilograms ) / Percentage of total porcine drug use
Avilamycin: growth promoter / 5 / 109.2 / 50%
Tetracyclines / 5 / 42.3 / 19%
Tylosin / 3 / 34.1 / 16%
Salinomycin: growth promoter / 1 / 6.1 / 3%
Penicillin / 6 / 5.9 / 3%
Amoxicillin / 4 / 4.6 / 2%
Potentiated Sulphonamides / 5 / 4.6 / 2%
Lincomycin / 6 / 4.6 / 2%
Penicillin-Streptomycin / 7 / 4.5 / 2%
Enrofloxacin / 4 / 0.3 / <1%
Apramycin / 3 / 0.3 / <1%
Tiamulin / 4 / 0.2 / <1%
Lincomycin-Spectinomycin / 1 / 0.2 / <1%
Danofloxacin / 1 / 0.1 / <1%
Ceftiofur / 3 / 0.04 / <1%
Florfenicol / 1 / 0.03 / <1%
Amoxicillin-Clavulanate / 1 / 0.02 / <1%
Streptomycin / 1 / 0.001 / <1%
Total / 217kg

The use of therapeutics on the organic pig farms was very low. 2/5 farms had not used antimicrobials for over 12-months prior to our first visit nor did they use them throughout the period of the study. The remaining three farms used a low-level (0.02 – 0.36kg/12-months) of injectable therapeutics for individual animals where necessary. On the organic poultry farms, only one of the seven units had used therapeutic antimicrobials within the 2-year period over which the calculations were based. This had been in 2001, 9-months prior to the first visit made to this farm, and had been necessitated when two batches of chicks became seriously ill shortly after arrival onto the farm. The attending veterinary surgeon diagnosed yolk-sac infection and the birds were treated with amoxicillin and enrofloxacin. As a point of note, this was the only occasion upon which any of the 13 poultry farms (organic and conventional) reported using fluoroquinolones during this study, or in the 12-months preceding it.

During the course of this work one of the conventional poultry farms tried withdrawing the growth promoter avilamycin from the feed on two occasions. In both cases after a few crop cycles vague symptoms of wet litter, erratic food and drink intakes, listlessness and uneven growth rates manifested. After extensive diagnostic procedures this clinical syndrome was diagnosed as dysbacteriosis, and initially the birds were responsive to treatment with amoxicillin. However on both occasions the number and length of amoxicillin treatments necessary to control the syndrome increased over successive crop cycles until there was less of a discernable clinical response to the drug.

Figure 1. Estimates[h] of kilograms of active antimicrobial compounds that were administered to the stock over a 12-month period for each of the study farms.