Article Text

Download PDFPDF

Investigation of the presence of ESBL-producing Escherichia coli in the North Wales and West Midlands areas of the UK in 2007 to 2008 using scanning surveillance
  1. L. C. Snow, BSC, MSc, PhD1,
  2. H. Wearing, BSc, MSc2,
  3. B. Stephenson, BSc, MPhil1,
  4. C. J. Teale, MRCVS, MSc3 and
  5. N. G. Coldham, BSc, PhD2
  1. Animal Health Veterinary Laboratories Agency (AHVLA) – Centre for Epidemiology and Risk Analysis, New Haw, Addlestone, Surrey, KT15 3NB, UK
  2. AHVLA – Department of Bacteriology, Addlestone, Surrey, KT15 3NB, UK
  3. AHVLA – Shrewsbury, Kendal Road, Harlescott, Shrewsbury, Shropshire, SY1 4HD, UK
  1. Correspondence to Coldham, e-mail: nick.coldham{at}

Between November 5, 2007 and November 4, 2008, faecal samples from cattle and sheep submitted for diagnostic purposes to the Aberystwyth and Shrewsbury Veterinary Laboratories Agency (VLA) (now AHVLA) regional laboratories (covering North Wales and the West Midlands) were screened for the presence of Escherichia coli that produces CTX-M extended-spectrum β-lactamase (ESBL) using the selective medium CHROMagar CTX. Samples from 113 farms were tested and eight ESBL-positive farms identified. Of these, six farms were identified via submissions of cattle faeces and two from sheep. Gene sequencing revealed both group 1 and group 9 CTX-M enzymes corresponding to CTX-M-14, CTX-M-14B (group 9) and CTX-M-15/28 (group 1). Analysis of these isolates by nanoarray revealed that some were carrying a range of virulence genes including ireA, iroN and prfB, which have been associated with extraintestinal pathogenic E coli, and were multidrug resistant. Geographical analysis with choropleth maps suggested that these CTX-M genes are relatively widespread in the North Wales and West Midlands study area. This work was carried out concurrently with the running of a VLA ESBL surveillance system, which has subsequently identified many more CTX-M positive farms in the UK.

View Full Text

Statistics from

EXTENDED-spectrum β-lactamases (ESBLs) are enzymes produced by bacteria that confer the ability to resist a wide range of therapeutic β-lactam antibiotics upon their host organism. A number of different families of ESBLs exist and they are classified according to the differences in their genetic structure; in human beings, the first to be identified were the TEM- and SHV-derived enzymes followed more recently by the CTX-M family. This latter family has now emerged as the dominant ESBL globally with over 113 different CTX-M sequence types recognised (Lahey clinic 2011). In England and Wales, over the last few years, cases of infection with ESBL-producing bacteria in human beings have risen posing a serious threat to the use of third-generation cephalosporins for the treatment of severe infections in human beings (Livermore and others 2007).

In 2004, a CTX-M-14 ESBL-producing Escherichia coli was isolated from calves on a dairy farm in Wales (Teale and others 2005). This was the first report of an ESBL in food-producing animals in Great Britain, and at the time, the likely prevalence of ESBL-producing bacteria in livestock farms in the UK was unknown. The initial detection of the CTX-M gene was an incidental finding and there were uncertainties regarding whether this was a new occurrence or emerging condition with limited geographical spread. The aim of the study was to determine whether there were other ESBL-positive farms in the area around the index farm, which was identified as being at high risk. Therefore, a passive surveillance scheme was introduced to screen for the presence of CTX-M and/or other ESBLs on farms in the north Wales and adjoining west Midlands regions. If appropriate, this information could then be used to limit further dissemination from the index farm by the application of control measures, such as cattle movement restrictions.

Materials and methods

Study population

Faecal samples from cattle and sheep submitted for diagnostic purposes to the Aberystwyth and Shrewsbury Veterinary Laboratories Agency (VLA) regional laboratories covering the North Wales and West Midlands area (this includes in North Wales the counties of Gwynedd, Anglesey, Conway, Denbighshire, Wrexham and Flintshire) were anonymised and screened for ESBL-producing E coli. Only the first four digits of the postcode were retained for location information along with the standard information relating to the livestock contained on the VLA submission form. Only one sample per farm was selected at random from all samples submitted from that farm in any one submission. There was no filtering or preselection, and all farms making submissions on separate occasions were tested. The laboratory screening for ESBL was in batches and was distributed throughout the year. The study area was selected because a cattle farm positive for ESBL E coli had previously been identified within it and was considered to be at highest risk for CTX-M ESBL E coli. Initially, the authors aimed to screen 300 faecal samples that would have been sufficient to provide 95 per cent confidence of detecting at least one positive sample if CTX-M or other ESBL genes were present in 1 per cent of submissions. Sampling throughout the year was undertaken to eliminate seasonal bias. Although other cattle farms positive for ESBL E coli had also been identified in other parts of Wales and in England, resources for a national investigation across the UK were not sufficient.

Laboratory methods

For the isolation of presumptive ESBL E coli from the samples, 1 g of faeces was placed in 9 ml of buffered peptone water (BPW). Ten microlitres of the faecal dilution was plated onto the ESBL E coli selective medium CHROMagar CTX (CHROMagar) and incubated with the BPW cultures overnight at 37°C. Semi-quantitative counts were then performed on the agar plates and the colour of the colonies was noted. Where no growth was observed, 10 µl of the enriched cultures (BPW) were streaked onto CHROMagar CTX and incubated overnight at 37°C (Randall and others 2009).

When growth was observed, presumptive E coli, a single blue and/or white colony from the end of the streak, were replicated onto fresh media. From this subculture, a DNA extraction was performed and tested for the presence of the CTX-M gene with LightCycler PCR. The CTX-M gene was sequence typed as described elsewhere (Randall and others 2010). All CTX-M positive isolates were identified as E coli using the API 20E kit from bioMérieux. All isolates were also stored on Dorset egg slopes.

The virulence and antibiotic resistance gene complement of the 10 CTX-M isolates was determined using the Clondiag strip array (version 2). Bacterial cells were grown aerobically overnight in LB broth at 37°C and genomic DNA was extracted by direct digestion with a lysis buffer containing proteinase K. Genomic DNA was used as a template in a multiplex linear amplification and labelling reaction with primer pairs for 122 virulence and 155 antibiotic resistance genes. IconoClust software was used to calculate quantitative staining values and normalise to appropriate control probes. A signal intensity of greater than 0.4 was considered positive (Anjum and others 2007).

Data analysis

Geographical analysis mapped the location of the farms in ArcGis (ArcGIS version 9.2 (ESRI)) to observe the distribution of sampled farms and pattern of positive farms. To maintain confidentiality, only the first part of the postcode was used. StatXact8 was used for the descriptive statistical analysis, for the Fisher's exact test and the calculation of odds ratios.


Between November 5, 2007 and November 4, 2008, 128 submissions of faecal samples (101 cattle, 27 sheep) were obtained and screened for the CTX-M ESBL E coli, corresponding to 113 farms (92 cattle, 24 sheep, three both). While most farms only submitted samples on one occasion, one farm submitted samples from cattle or sheep on five different occasions over the year. A farm was considered positive if at least one submission received over the year was positive for CTX-M E coli. This identified eight farms (7.1 per cent, 95 per cent CI 3.1 to 13.5 per cent) that were positive for ESBL-producing E coli; in all cases, an ESBL belonging to the CTX-M group was found. Of these eight farms, six were identified via positive faecal samples from cattle and the other two from positive samples from sheep. Sequence typing of all the CTX-M isolates revealed a mixture of group 9 (type 14 and 14B) and group 1 (type 15/28) (Table 1). For cattle, the ESBL-positive submissions (n=6) were from dairy (three positive out of a total of 38 submissions tested), rearing (1/1), and suckler (2/50) herds of animals (Table 1). The Fisher's exact test was used to determine the association between herds, which indicated that there was a significant difference (p=0.048) by type, but conclusions are limited by there being only a single observation for rearing herds. The odds ratios were for dairy versus rest, p=1, OR 1.4, and suckler versus rest, p=0.40, OR 0.37.

Table 1

Submission details from CTX-M positive faecal samples

Unfortunately, although requested, the postcodes were not recorded on the VLA submission form for all farms by the submitting private veterinary surgeon; of the 23 sheep farms from which samples were received, 20 farms had postcodes available, for the other three, these data were missing including one of the CTX-M positive farms. This farm was located somewhere in Shropshire but further details are not available. Thus, Fig 1 shows only one of the two farms with CTX-M positive samples from sheep. Nine of the 89 cattle farms did not have a postcode.

Fig 1

Map of the North Wales and West Midlands regions showing postcode regions from which faecal samples from sheep were received and tested for ESBL E coli

Figs 1 and 2 show the distribution by postcode region of sampled and positive sheep and cattle farms. Visual examination reveals little in the way of clustering with ESBL-positive farms identified in Staffordshire, Gwynedd and Clwyd but further work using a geographically representative sampling method would be required to confirm this. It is likely that ESBL E coli are widespread in the North Wales and West Midlands area in livestock and not confined to the locality of a previously identified positive farm. Furthermore, these results show that ESBLs are also present in sheep in the region. No positive farms submitted both cattle and sheep samples so it is unknown whether both species are likely to be infected when present on the same farm. Work currently being carried out may clarify this.

Fig 2

Map of the North Wales and West Midlands regions showing postcode regions from which faecal samples from cattle were received and tested for ESBL E coli

Seven of the 10 isolates contained at least one of the virulence genes which are commonly found in extraintestinal pathogenic E coli (ExPECs). These included iroN and ireA, mchF, mcmA and perfB, which are classical ExPEC virulence genes (Table 2), but can also be found in commensal strains of E coli (Anjum and others 2007, Wragg and others 2009). In addition to the appropriate CTX-M group genes, the isolates also contained those encoding resistance to tetracyclines (tetA (n=1 isolate) and tetB (n=7), sulphonamides (sul1 (n=5) and sul2 (n=9) and streptomycin (strB (n=9)). Similar profiles of antibiotic resistance and virulence genes were found for isolates from the same but not different farms.

Table 2

Virulence genes found in CTX-M E coli isolates


This work aimed to determine whether the presence of ESBL E coli in sheep and cattle in an area including North Wales was a new and emerging phenomenon or whether they were present but underidentified via current surveillance. UK has experienced a sudden increase in ESBL rates in human beings, mainly due to the spread of CTX-M-producing E coli (Livermore and Hawkey 2005). A similar trend has not been observed in cattle (Hunter and others 2010). In a study carried out in 2008/2009, 37.5 per cent of randomly selected dairy herds in the North-West of England were found to have CTX-M ESBLs present (Snow, L. C., Warner, R. G., Cheney, T., Wearing, H., Stokes, M., Harris, K., Teale, C. J., Coldham, N. G., unpublished observations). The work presented here suggests that in 2007 and 2008, ESBL E coli were already geographically widespread in North Wales and more than one strain was present. This conclusion is supported by the nanoarray data which showed that the isolates were carrying different virulence determinants and antibiotic resistance genes. Due to the small number of positive farms identified, there is insufficient evidence from these data to report any evidence of clustering of CTX-M. The passive scanning surveillance of farms in the study area from clinical submissions was not geographically representative and therefore, estimates of disease levels or prevalence cannot be determined. Previously, the authors have reported (Teale and others 2005, Watson and others 2011) that the prevalence of CTX-M-14 ESBL E coli was much higher in calves compared with adults but the low number of positive samples in this study prevents any age-related conclusions.

These results, in conjunction with the molecular findings, also suggest that the current cases have arisen from a number of different sources. Both group 1 and group 9 CTX-M enzymes have been previously identified in livestock in the UK (Teale and others 2005, Defra 2007) and CTX-M-14 is now considered the second most common CTX-M-type in human beings in the UK after CTX-M-15 (Tarrant and others 2007). The group 1 isolates were presumed to be CTX-M-15 rather than 28 as the latter sequence type is a rare variant in human beings and has not been identified in UK livestock to date. Further typing would need to be carried out to verify this. The two types, CTX-M-28 and CTX-M-15, differ from each other in only two single nucleotide substitutions. The presence of certain virulence genes has been used to define the pathotype of E coli (Wu and others 2010). The ireA and iroN (iron utilisation), prfB (fimbriae) and f17-G (adhesion) which has previously (Anjum and others 2007) been associated with extraintestinal pathogen E coli (ExPEC). The iron and mchF genes have been associated with non-haemorrhagic diarrhoea in calves (Wu and others 2010). The antimicrobial resistance genes detected in the CTX-M isolates are those commonly found in E coli (Batchelor and others 2008).

Scanning surveillance cannot provide estimates of prevalence due to the intrinsic biases in the system. However, it is clear that CTX-M is prevalent in cattle and sheep in this area; in the selected sample population, approximately 7.1 per cent of the farms included in the study were positive. It is important to note that this sample population is biased as it only reflects clinically diseased animals that are receiving veterinary care and where the consulting veterinary surgeon considers that a laboratory diagnosis is necessary. Large areas were not sampled, which limits any conclusions to those regions that were mainly in the north-west of Wales. However, the data are of value to assess the change in the study area over time or to determine the effect of control measures. A systematic survey of holdings may reveal other patterns and would enable a true estimate of prevalence. The scanning surveillance system has other limitations that have been made clear in this work, namely the incompleteness of some sample records. The geographical analysis is reliant on the private veterinary surgeon (PVS) supplying the correct information on the submission form and as has been highlighted with the absence of postcode information for some samples; this can limit the usefulness of the data for epidemiological analyses. The use of antimicrobials on farms, including those containing third- or fourth-generation cephalosporins, was not recorded but they would be expected to exert a pressure favouring selection of CTX-M ESBL E coli (Cavaco and others 2008). The prevalence of CTX-M-1 ESBL E coli was reported to be lower on an organic dairy farm which had no use of third- and fourth-generation cephalosporins compared with a conventional farm where they were in routine use (Dolejska and others 2011).

This study examined one randomly selected faecal sample from each diagnostic submission for ESBL E coli. The samples were examined by a selective method, which would be expected to detect ESBL E coli present as a minor component of the total faecal flora. The detection of CTX-M E coli is dependent on the sensitivity of Chromagar CTX, which is approximately 10 cfu/g faeces (Randall and others 2009). This differs from the methods by which diagnostic samples are routinely examined, where they are cultured on non-selective media and a randomly selected E coli is subjected to susceptibility testing. In the latter case, ESBL E coli present as minor components of the total bacterial faecal flora might not be detected and the sensitivity is lower compared with the former method. In this study, more ESBL E coli positive farms could probably have been detected had multiple faecal isolates from different animals from the same submission been examined, but that would skew the study, because the total number of individual submissions from a farm was variable and ESBL E coli status might then also be related to the number of samples examined from a farm. The on-farm prevalence of ESBL E coli is variable and a statistically valid sample size would be required from each farm to be confident that all negative samples from this study were representative of the farm status. This study suggests that the sensitivity of scanning surveillance to detect ESBLs may be low.

A surveillance system for ESBLs in E coli and other bacteria in diagnostic samples from livestock in England and Wales has been implemented since 2006 and has identified other ESBL E coli positive farms. This surveillance screens all E coli isolates, which are subjected to susceptibility testing for diagnostic purposes for susceptibility to certain indicator cephalosporins (cefotaxime and ceftazidime or cefpodoxime), and follow-up tests to detect ESBL producers are performed when resistance is detected. Preliminary results have been published (Hunter and others 2010). This surveillance relies on randomly selected E coli isolates from primary culture plates, whereas the current study incorporated selective media to screen the samples; great differences in sensitivity are to be expected between these methods.

The role of livestock in the dissemination of the ESBL genes, which are generally located on plasmids, and their potential public health importance is not known and further epidemiological and molecular studies are required to fully understand the epidemiology of these genes.


The authors thank all staff in the VLA regional laboratories who were involved in this study (project reference OD2023) and Defra for funding the work.


View Abstract


  • 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.