National survey for Salmonella in pigs, cattle and sheep at slaughter in Great Britain (1999–2000)


R.H. Davies, Veterinary Laboratories Agency (VLA) – Weybridge, New Haw, Addlestone, Surrey, KT15 3NB, UK (e-mail:


Aims:  The objective of these surveys was to estimate the prevalence of faecal carriage of Salmonella in healthy pigs, cattle and sheep at slaughter, and of pig carcase contamination with Salmonella. These data can be used as a baseline against which future change in Salmonella prevalence in these species at slaughter can be monitored.

Methods and Results:  In this first randomized National Survey for faecal carriage of Salmonella in slaughter pigs, cattle and sheep in Great Britain, 2509 pigs, 891 cattle and 973 sheep were sampled in 34 pig abattoirs and 117 red meat abattoirs in England, Scotland and Wales. Carriage of Salmonella in 25 g caecal contents was identified in 578 (23·0% pigs) but in only 134 (5·3%) of carcase swabs. The predominant Salmonella serovars found in both types of sample were S. Typhimurium (11·1% caeca, 2·1% carcases) and S. Derby (6·3% caeca, 1·6% carcases). The main definitive phage types (DT) of S. Typhimurium found were DT104 (21·9% of caecal S. Typhimurium isolates), DT193 (18·7%), untypable strains (17·6%), DT208 (13·3%) and U302 (13·3%). Three isolates of S. Enteritidis (PTs 13A and 4) and one enrofloxacin-resistant S. Choleraesuis were also isolated. A positive ‘meat-juice ELISA‘ was obtained from 15·2% of pigs at 40% optical density (O.D.) cut-off level and 35·7% at 10% cut-off. There was poor correlation between positive ELISA results or carcase contamination and the caecal carriage of Salmonella. The ratio of carcase contamination to caecal carriage rates was highest in abattoirs from the midland region of England and in smaller abattoirs. In cattle and sheep 1 g samples of rectal faeces were tested. Two isolates (i.e. 0·2%) were recovered from cattle, one each of S. Typhimurium, DT193 and DT12. One sheep sample (0·1%) contained a Salmonella, S. Typhimurium DT41. In a small subsidiary validation exercise using 25 g of rectal faeces from 174 cattle samples, three (1·7%) isolates of Salmonella (S. Typhimurium DT104, S. Agama, S. Derby) were found.

Conclusions:  The carriage rate of Salmonella in prime slaughter cattle and sheep in Great Britain was very low compared with pigs. This suggests that future control measures should be focused on reduction of Salmonella infection on pig farms and minimizing contamination of carcases at slaughter.

Significance and Impact of the Study:  This work has set baseline figures for Salmonella carriage in these species slaughtered for human consumption in Great Britain. These figures were collected in a representative way, which enables them to be used for monitoring trends and setting control targets.


In many countries, especially those where relatively large volumes of raw smoked or lightly cooked pig meat products are consumed, pork is thought to be a source of a significant proportion of human Salmonella infections (Berends et al. 1998; Hald and Wegener 1999; Anon. 2002). In the UK it is not easy to determine the role of contaminated pork in human salmonellosis but some outbreaks, including a small number of deaths, have been associated with contaminated pig meat (Maguire et al. 1993; Smerdon et al. 2001). In recent years emerging definitive phage types of S. Typhimurium, such as U310, have been specifically linked with pig meat products (Ward and Threlfall 2001) but overall there has been a 69% fall in outbreaks linked with the consumption of pig meat between 1992 and 1999 (Smerdon et al. 2001). As most Salmonella in pig herds is subclinical it is not easy to assess trends in infection by means of reports of clinical investigation, which is much more relevant to cattle where clinical disease is more common (Wray and Davies 2000).

There is little evidence linking human salmonellosis with contaminated beef or lamb, and when this does occur there is often evidence of cross-contamination and poor food hygiene practices which make it difficult to accurately identify the original source (Smerdon et al. 2001). For example, S. Enteritidis which has primarily been associated with poultry meat and eggs, and not with pigs, has been commonly found in pork-linked outbreaks.

For progress to be made in Salmonella control it is necessary to establish the current status in a statistically sound and unbiased way, and to monitor progress against targets (Swanenburg et al. 2001). The national abattoir surveys described in this paper set the foundations for this process.

Materials and methods

All abattoirs slaughtering pigs, cattle and sheep in Great Britain in 1998 were contacted and asked to volunteer to participate in a survey of slaughter animals for foodborne zoonotic organisms and commensal organisms used as indicators of antimicrobial resistance. A detailed description for the cattle and sheep abattoir enrolment and sampling protocol is provided by Paiba et al. (2002). The pig abattoirs were recruited by the Meat and Livestock Commission (MLC) who contacted the abattoir owners directly to request their participation. Sampling frames were set up such that the identity of the abattoirs and origin of slaughter animals was anonymous with respect to the results of testing. Sampling schedules were constructed according to the abattoir's throughput for 1998. For pigs a maximum of five carcases per abattoir were sampled on a different day each week throughout the year in a randomized way according to the abattoir throughput. For each species sampling was conducted over a 12-month period, which began in February 1999 for pigs and in January 2000 for cattle and sheep. Sampling in all abattoirs was begun in the same week. Samples were taken by Food Standards Agency Meat Hygiene Service staff according to a standardized protocol. Samples were removed and dispatched on ice to the testing laboratories on the day of collection.

Three samples (caecum, carcase swab and neck muscle) were collected from each pig. Caeca were tied off and sectioned from the remaining guts; at the laboratory 10 g contents were aseptically removed for testing. Carcases were swabbed over an area of 0·1 m2 on the skin of the chest and abdomen using a large gauze swab. Portions of muscle, to obtain tissue fluid for testing by ELISA, were removed from the major neck muscle and frozen on arrival at the laboratory.

In cattle and sheep rectal faeces were milked into a section of the terminal rectum which was tied off and sectioned from the colon (Paiba et al. 2002). At the laboratory 1 g rectal faeces was removed for culture. All samples were transported to the laboratory on ice by overnight carrier and culture begun on the day after collection.

Pig caecal contents and carcase swabs were added to 100 ml buffered peptone water (BPW; 1 g to 10 ml for cattle and sheep faeces) (Oxoid CM509) and incubated at 37°C for 18 h. Two hundred microlitres of broth was then inoculated into 20 ml DIASSALM (Lab537; LabM, Bury, UK) in a Petri dish; this was incubated at 41·5°C for 48 h. After 24 and 48 h 0·1 μl turbid culture from the edge of the growth zone in DIASSALM was inoculated onto Rambach Agar (Merck 7500, Poole, UK) which was incubated at 37°C for 18–24 h. Suspect Salmonella colonies were confirmed serologically and biochemically after which they were serotyped and phage typed. A diffusion technique using Sensitest agar (Oxoid) and antimicrobial-containing discs (Oxoid) was used (Wray et al. 1991). The discs contained the following antimicrobial agents: amikacin (30 μg), amoxycillin/clavulanic acid (30 μg). ampicillin (10 μg), apramycin (15 μg), cefoperazone (30 μg), cefuroxime (30 μg), chloramphenicol (10 μg), chlortetracycline (10 μg), colistin (25 μg), furazolidone (15 μg), gentamicin (10 μg), nalidixic acid (30 μg), neomycin (10 μg), streptomycin (25 μg), sulphamethoxazole/trimethoprim (25 μg) and sulphonamide compounds (300 μg). A growth inhibition zone with a diameter of <13 mm was recorded as resistant (Sojka et al. 1972), and the cultures were classified as either resistant or sensitive. Minimum inhibitory concentrations (MICs) to enrofloxacin were determined for nalidixic acid-resistant strains by a plate dilution method (based on National Committee for Clinical Laboratory Standards 1997). Isolates with MICs of 4 mg l−1 or more were considered to be fully resistant. Tissue fluid extracted from the neck muscle by freezing and thawing was tested using an in-house indirect group B/C1 lipopoly saccharide antigen (LPS) ELISA (Clouting and Davies 2001).

Data was double entered into a Microsoft Access database and a 10% visual check was performed between data entry clerks. Initial analysis was conducted using queries in Microsoft Access. Proportions, 95% binominal confidence intervals (CI95%), (χ2) and χ2 tests for trend were calculated using Epi-Info 6·04 (CDC, Atlanta, GA, USA).


Table 1 shows the overall results obtained. From 2509 pigs tested, 578 [23·0% (CI95% 21·4–24·7)] of caecal contents contained Salmonella, just under half of which (48·1%), 278 [11·1% (CI95% 9·8–12·3)], were S. Typhimurium. The figures for carcase swabs were 134 [5·3% (CI95% 4·5–6·2)] with 52 (2·1%) being S. Typhimurium. Only 2403 neck muscle ‘meat juice’ samples were tested as some samples yielded insufficient volume of tissue fluid. At the 40% O.D. cut-off level (equivalent to that used in the Danish Salmonella Control Programme at the time that this survey was carried out) 365 (15·2%) of pigs were classified as positive whereas with the 10% O.D. cut-off point, which is the standard cut-off used for research purposes (Alban et al. 2002), 857 (35·7%) of samples were classified as positive.

Table 1.  Prevalences and 95% confidence intervals for Salmonella in caecal contents, carcase swabs and tissue fluid (ELISA)
SpeciesSampleOrganismNo. positiveNo. testedPrevalence (%)95% CI
  1. MJE, meat juice ELISA.

CaecumSalmonella Typhimurium278250911·19·8–12·3
Carcase swabSalmonella13425095·34·5–6·2
Carcase swabSalmonella Typhimurium5225092·11·5–2·6
Caecum or carcase swabSalmonella666250926·524·8–28·3
Caecum or carcase swabSalmonella Typhimurium322250912·811·5–14·1
Neck muscle (MJE > 40% O.D.)Salmonella365240315·213·7–16·6
Neck muscle (MJE > 10% O.D.)Salmonella857240335·733·7–37·6
CattleRectal faecesSalmonella28910·20·0–0·5
SheepRectal faecesSalmonella19730·10·1–0·3

In cattle two of 891 samples [0·2% (CI95% 0·0–0·5)] and in sheep one of 973 [0·1% (CI95% 0·1–0·3)] contained Salmonella. In a subsidiary study using the same samples, 174 samples of cattle rectal faeces were tested in parallel using 1 and 25 g samples, three Salmonella isolates [1·7% (CI95% 0·0–3·7)] were found in the 25-g samples only. In cattle one isolate each of S. Typhimurium DT193 and 12 were found in 1 g samples and S. Typhimurium DT104, S. Agama and S. Derby in 25 g samples. Salmonella Typhimurium DT41 was found in sheep. The S. Typhimurium DT12 isolate from cattle showed the typical DT104 pentaresistance profile of resistance to ampicillin, chloramphenicol, streptomycin, sulphonamide and tetracycline, and the DT193 isolate was resistant to furazolidone only. The DT41 isolate from sheep was fully sensitive to the 16 antibiotics used in the test panel.

A breakdown of the ranking of predominant Salmonella serovars found in pigs is shown in Table 2. Salmonella Typhimurium was the predominant serovar found in caeca (48·1%) and carcases (38·8%). Salmonella Mbandaka and S. Infantis were amongst the top six serotypes found on carcases but these were uncommon in caecal contents.

Table 2. Salmonella serotype distribution in pigs
SerotypeNo. positive (%)
 Typhimurium278 (11·1)
 Derby157 (6·3)
 Kedougou 23 (0·9)
 Goldcoast 23 (0·9)
 Panama 15 (0·6)
 Brandenberg 9 (0·4)
 Others 73 (2·9)
 All serotypes578/2509 (23·0)
Carcase swab
 Typhimurium 52 (2·1)
 Derby 40 (1·6)
 Goldcoast 8 (0·3)
 Panama 4 (0·2)
 Mbandaka 4 (0·2)
 Infantis 4 (0·2)
 Others 22 (0·8)
 All serotypes134/2509 (5·3)

The predominant definitive phage type of S. Typhimurium in pigs (Table 3) was DT104 [61/278 (21·9%) caecal isolates, 2·8% of pigs] but DT193 was more common in carcase swabs. A large proportion (18·2% of isolates) could not be phage typed with the standard panel of phages. Six isolates of DT135 which had only rarely been found in porcine samples previously, were made from caecal contents. Similarly U188 had not been recognized in Great Britain previously in any livestock species. On later checking of the untypable strains with extra typing phages which were not available at the time of the survey several isolates of U310, a newly recognized phage type were also found. Three isolates of S. Enteritidis (PTs 13a and 4) were found in caecal contents and one isolate of S. Choleraesuis var kunzendorf, which was resistant to enrofloxacin, was found in a carcase swab.

Table 3. Salmonella Typhimurium (STM) definitive types isolated from caeca and carcase swabs in pigs
Phage typeCaecumCarcase swab
No. positive% of total STMNo. positive% of total STM

An apparent seasonal variation in Salmonella in pigs was noted (Fig. 1), with peaks in isolation from caeca in March, August and December but with peaks of meat juice ELISA results in June and October.

Figure 1.

Percentage of Salmonella tests from caeca and swabs positive by month of sampling. (◆)All Salmonella, (•) caccum, ( inline image) MJE pos 10% O.D., (×) MJE pos 40% O.D., (○) carcase swab

The correlation between the isolation results for Salmonella and the meat juice ELISA results is summarized in Table 4(i and ii). When compared with the results of caecal culture, the sensitivity of the MJE was 0·289 (CI95% 0·252–0·329) and the specificity was 0·890 (CI95% 0·874–0·903) at the 40% O.D. cut-off value. The positive and negative predictive values were 0·441 and 0·806, respectively. At the 10% cut-off level the sensitivity increased to 0·515 (CI95% 0·473–0·558) and the specificity reduced to 0·691 (CI95% 0·670–0·712). There was a statistically significant difference between culture and ELISA results in pigs infected with any Salmonella or with S. Typhimurium specifically.

Table 4.  Sensitivity and specificity of the MJE
  1. MJE, meat juice ELISA.

  2. *Sensitivity = 0·289 (0·252–0·329); specificity = 0·890 (0·874–0·903); positive predictive value = 0·441; negative predictive value = 0·806.

  3. †Sensitivity = 0·515 (0·473–0·558); specificity = 0·691 (0·670–0712); positive predictive value = 0·335; negative predictive value = 0·825.

(i) MJE (40% O.D. cut-off)*
(ii) MJE (10% O.D. cut-off)†

The difference in results for pigs between regions of slaughter is shown in Table 5. Salmonella isolation from caeca was highest (29·2%) from abattoirs in the North and Scotland but carcase contamination was lower than the other two regions, as was the proportion of pigs positive by the meat juice ELISA test. Salmonella from carcases (8·7%) and the carcase swab to caecal contents ratios was significantly (P < 0·05) higher in samples from abattoirs in the midlands than for the other regions. The percentage of Salmonella-positive caecal samples was lowest in the South (19·1%) but the ELISA results were higher than in the other regions at 17·6% at the 40% O.D. cut-off or 39·5% at the 10% cut-off.

Table 5.  Percentage of tests positive for each organism/sample – comparison of abattoirs by regions
 Percent positiveMean
MidlandsNorth and ScotlandSouth
  1. *P < 0·05.

Salmonella from caeca21·329·219·123·0
Salmonella from carcase8·73·54·45·3
MJE > 40% O.D.14·013·417·615·2
MJE > 10% O.D.34·332·439·535·7
Swab/caecum ratio0·4*0·10·20·2

Table 6 shows a comparison of the results from pig abattoirs by annual throughput. The highest level of Salmonella in caecal contents (27·9%) was found in the high throughput abattoirs (>600 000 pigs) but the level of carcase contamination (4·3%) and the carcase/caecal positive ratio (0·2) was lowest for this category. In contrast, despite a lower prevalence of positive caeca (19·4%), smaller abattoirs with a throughput of <15 000 pigs per year had a higher level of carcase contamination (16·7%) and a higher carcase/caecal positive ratio (0·9). This difference was not statistically significant because of the small number of samples (72) taken from the smaller abattoirs compared with the largest throughput category (1070) and the high level of variability between individual abattoirs. However, a trend of reducing carcase contamination with increasing throughput was apparent across all the throughput categories.

Table 6.  Percentage of tests positive for each organism/sample – comparison of abattoirs by annual throughput
Percentage positiveThroughput: pigs per yearMean
>15 000>100 000>150 000>600 000
  1. MJE, meat juice ELISA.

%Salmonella from caeca19·420·818·727·923·0
%Salmonella from carcase16·76·25·24·35·3
% MJE > 40% O.D.10·016·515·514·715·2
% MJE > 10% O.D.30·036·138·833·435·7
Swab/caecum ratio0·90·30·30·20·2
% Positive of all tests (MJE > 40% O.D.)21·620·118·621·720·3
% Positive of all tests (MJE > 10% O.D.)25·223·622·725·124·0

Table 7 shows a breakdown of the antimicrobial resistances in the various Salmonella serotypes isolated from pigs. Overall, 75·6% were resistant to tetracycline (T), 44·8% to sulphonamide (SU), 28·1% to ampicillin (AM), 27·1% to streptomycin (S), 24·3% to trimethoprim/sulphonamide (TM) 17·9% to chloramphenicol (C), 7·4% to cefoperazone (CF) and 4·1% to nalidixic acid (NA). Resistance to other antimicrobials in the panel was uncommon and no resistance to cefuroxime (CX), amikacin (AK), amoxycillin/clavulanic acid (AMC) or colistin (CT) was identified.

Table 7.  Percentage of Salmonella serotypes from pigs resistant to individual antimicrobials percentage resistant
  1. No isolates were resistant to cefuroxime (3 mcg), amikacin (30 mcg), amoxycillin/clavulanic acid (30 mcg) or colistin (25 mcg).

  2. NA, nalidixic acid (30 mcg); T, tetracycline (10 mcg); N, neomycin (10 mcg); AM, ampicillin (10 mcg); FR, furazolidone (15 mcg); TM, sulphamethoxazole/trimethoprim (25 mcg); C, chloramphenicol (10 mcg); CN, gentamicin (10 mcg); S, streptomycin (25 mcg); SU, sulphonamide (300 mcg); CF, cefoperazone (30 mcg); APR, apramycin (15 mcg).

All serotypes4·175·63·128·11·024·317·91·127·144·87·41·1716
 Gold coast3·219·40·012·90·06·50·00·06·56·50·00·031
Total rough strains50·083·30·066·70·033·350·00·066·766·716·70·06
Total incomplete antigenic structures0·054·50·00·04·518·20·00·09·127·30·00·022

The most resistant phage types of S. Typhimurium were DT104, 104B and U302, most of which showed the typical pentaresistant (ampicillin, chloramphenicol, streptomycin, sulphonamide, tetracycline) phenotypes. Half of S. Typhimurium strains were also resistant to cefoperazone. Resistance to nalidixic acid was found in 7·5% of DT104 and 17·7% of DT193 (Table 8). Nalidixic acid resistance was found in 30 isolates, more commonly S. Typhimurium DT193 (12 isolates).

Table 8.  Percentage of Salmonella Typhimurium definitive types from pigs resistant to each antimicrobial
  1. DT, definitive type; Stm DT, Salmonella Typhimurium DT.

  2. NA, nalidixic acid; T, tetracycline; N, neomycin; AM, ampicillin; FR, furazolidone; CX, cefuroxime; TM, trimethoprim/sulphonamide; C, chloramphenicol; AK, amikacin; AMC, amoxycillin/clavulanic acid; CN, gentamicin; S, streptomycin; SU, sulphonamide; CF, cefoperazone; APR, apramycin; CT, colistin.

All DTs7·493·62·855·50·60·031·036·20·00·01·549·168·116·01·50·0326

When tested for the MIC of enrofloxacin only one isolate of S. Choleraesuis var Kunzendorf showed full resistance at 4 mg l−1 and one isolate of DT103 had an MIC of 1 mg l−1, but most strains had MICs of 0·5 mg l−1.



This study was the first structured national survey of Salmonella in pigs, cattle and sheep in Great Britain. The caecal Salmonella carriage rate in pigs was high at 23·0%, although carcases were only moderately contaminated (5·3%). The meat juice ELISA results also indicated significant exposure to infection with 15·2% of tissue fluid samples being positive at the 40% cut-off level and 35·7% at the 10%‘experimental’ cut-off level. Danish testing and interpretation criteria have changed since this survey such that a new cut-off point of 20% O.D. is now used (Alban et al. 2002), but unfortunately it was not possible to reinterpret data from the current study to relate to this because of a database failure.

It is difficult to compare these results with findings in other countries as there have been few truly representative national surveys and there are considerable variations in sampling frames and methods as well as in Salmonella culture techniques.

In Denmark, in 1993–94, before the start of the Danish National Salmonella Control Policy (Mousing et al. 1997) Salmonella was found in 6·2% of 5 g samples of 13 468 caecal contents taken from all herds producing over 2600 slaughter pigs per year plus a sample of other herds producing between 500 and 550 pigs. This obviously differed from the UK survey where abattoirs were sampled randomly according to throughput and although pig herds were sampled anonymously this may have led to a greater representation of pigs from larger herds than in the UK survey. In addition the UK survey used a larger volume of caecal material (10 g) and a culture method, which has been shown to be more sensitive than standard culture methods (Davies and Wray 1997; Raes and Voogt 2000). The rate of Salmonella detection is also substantially raised by increasing the amount of caecal or faecal material examined (Funk et al. 2000).

A second survey carried out in Denmark between June 1998 and February 1999, after the control policy had been operating for 4 years, identified Salmonella in 3·4% of 17 987 25 g caecal contents samples from slaughter pigs (Christensen et al. 1999). The national Salmonella sero-prevalence of slaughter pigs, as measured in ongoing control programme samples was also reduced from 5·4% in 1994–95 (Christensen and Rudemo 1998) to 2–3% in 1998 (Christensen et al. 1999).

The only other comparable survey was carried out in Germany between February and June 1996 (Kaesbohrer et al. 2000). Only seven abattoirs participated but 11 942 pigs from 752 batches, mostly from separate finishing farms were examined. Rectal faecal swabs rather than caecal samples were used and the isolation rate was 3·7%. Mesenteric lymph nodes were also examined and 3·3% of these were positive, as were 4·7% of carcase swabs. The meat juice ELISA was positive in 7·7% of pigs. (Protz et al. 1997). In another German study the serological prevalence was 7·3 and 1% of carcase swabs were positive (Altrock et al. 2000). In Canada 5·2% (CI95%4·0–6·4) of caecal contents of 1420 pigs from six abattoirs in 1996 contained Salmonella (Letellier et al. 1999) but Salmonella was found in 80·2% of 200 pig faecal samples at slaughter in Quebec (Mafu et al. 1989). The relatively high rate of Salmonella found in slaughter pigs in the UK does not appear to be so common in other countries. For example, no Salmonella was found in 3388 samples from five pig abattoirs in Sweden (Thorberg and Engvall 2001) and very low infection rates have also been found in surveys in Styria and Austria (Schotzlinger 1994; Pless et al. 2000). Quoted prevalences of infection or contamination of pigs at slaughter usually refer to results of culture of relatively small volume samples of a single sample type. If multiple samples, eg. caecal, colonic, rectal contents, mesenteric lymph nodes, are included then the measured infection rate doubles (Davies et al. 1999a; Swanenburg et al. 2001). There is also evidence that cull sows may also carry high levels of Salmonella and as these may be skinned rather than scalded and singeing may represent a greater public health threat than prime pork (McKean et al. 2001). The apparent seasonal peaks in the prevalence of Salmonella are interesting and, although repeat tests during further years would be needed to confirm this, similar findings have been recorded in Denmark (Hald and Andersen 2001).

High rates of carcase contamination have been found in studies in various countries: Greece 28% (Epling et al. 1993), Canada 17·5% (Lammerding et al. 1988), Netherlands 21% (Oosterom et al. 1985), Belgium 27% (Korsak et al. 1998). These limited studies may be influenced by variations which may occur in the degree of carcase contamination from day to day (Davies et al. 1999a) and result from sampling carried out before introduction of more modern slaughter lines.

The Salmonella isolation results from caeca in the UK slaughter survey are substantially higher than levels found in Denmark and Germany. This may have been partially influenced by efficient sampling and culture methods but is likely to reflect a genuinely high National prevalence of infection.

The average prevalence of Salmonella on carcases was relatively low. However, there was a wide range found between abattoirs and sampling occasions. This may reflect a variation in the prevalence of Salmonella carriage in pigs slaughtered on any one day or the application of measures designed to reduce carcase cross-contamination. It is likely that targeted sampling of a larger surface area including the lowest parts of the carcase and initiation of culture on the same day as collection (Davies et al. 1999b) may be more sensitive. However, the absence of detectable Salmonella on 0·1 m2 carcase is most likely to reflect the risk of contamination associated with prime pork cuts, especially as a reduction in carcase contamination occurs during chilling (Davies et al. 1999a).

The results for the ‘meat-juice’ ELISA support the bacteriological findings in that pigs are exposed to a relatively high level of Salmonella during the weeks prior to slaughter. At the 40% O.D. cut-off level, which was used for commercial herd testing in Denmark at the time of the survey (Nielsen et al. 1995), ca 15% of pigs were sero-positive, which was considerably lower than the proportion of bacteriological positive pigs. When the 10% O.D. cut-off level was used in the current study nearly 36% of pigs showed evidence of exposure to infection. This improved the sensitivity of the test, although still failed to correctly identify almost half of the bacteriological positive animals, and the positive predictive value reduced by a quarter (44–33%) when the cut-off point was reduced from 40 to 10%. In Denmark the 40% cut off level was adopted so that pig herds with a high prevalence of sero-reactors could be identified while minimizing false-positive reactions, which are likely to be higher with lower cut-off levels (Wiuff et al. 1999). The Danish cut-off level has now been changed to 20% O.D. to improve sensitivity as the national infection rate falls (Alban et al. 2002). A disadvantage of the ELISA system is the need to take large numbers of individual samples from the same herd for monitoring whereas faecal or caecal samples can be bulked. It is also not possible to determine which Salmonella serovars have provoked the antibody response. This means that herds with a moderate prevalence of virulent zoonotic serovars, such as S. Typhimurium DT104, are not always identified as problem herds (Møgelmose et al. 1999) whereas herds with a high prevalence of S. Derby, which is rarely involved in human disease, may be categorized as high risk.

Cattle and sheep

A very low faecal carriage of Salmonella was detected in cattle and sheep by testing 1 g of faeces, and even when a 25 g sample size the prevalence was still low. Similar findings were obtained in an abattoir survey carried out in Canada (van Donkersgoed et al. 1999).

In most other countries where such surveys have been conducted, the prevalence of Salmonella carriage was much lower in slaughter cattle and sheep compared with pigs. A low frequency of Salmonella was found on beef carcases in Northern Ireland (Madden et al. 2001) and even when calves were infected with Salmonella during rearing most had shed infection by the time of slaughter (Galland et al. 2000). Beef carcase contamination may be worse in smaller nonline abattoirs (Kuplulu 1999; Tahamtan et al. 2002). In the USA feedlot cattle have been shown to have a prevalence of 3–5%Salmonella at slaughter (Sofos et al. 1999; Beach et al. 2002). In contrast, various other studies have shown a high prevalence in cull cattle, especially after transport. In one such study a pre transport prevalence of 1% increased to 21% at the abattoir (Beach et al. 2002). In another survey 21·3% of 5,087 caecal and colonic contents samples from cull dairy cows in the USA contained Salmonella (Troutt et al. 2001). Thus, although not currently permitted in Great Britain, adult cattle slaughtered for human consumption may present greater public health risks, especially as much of this lower grade meat would be used in ground beef products. The cattle tested in this study were all under 30 months of age, as older cattle are not permitted for human consumption in Great Britain, and so the low prevalence among these slaughter stocks may reflect the low prevalence in this age group.

Like cattle, the prevalence of Salmonella in sheep was very low. Little comparable data is available for sheep from other countries. A survey in the USA showed 1·5% lamb carcases positive for Salmonella (Duffy et al. 2001). In Iceland Salmonella is thought to be rare but healthy sheep carrying Salmonella in their tonsils have caused outbreaks of Salmonella food poisoning in human beings consuming sheep's heads (Hjartardottir et al. 2002). The Salmonella subspecies III S.61;k;1,5,(7) has been found in up to 45% of sheep, mainly adults, in larger, high replacement rate flocks, in some regions of Norway (Alvseike and Skjerve 2001).

The purpose of surveillance is to provide a tool for enhanced control. In pigs it is clear that a whole food chain approach is needed as breeding herds seem to have similar Salmonella problems to commercial herds (Christensen et al. 2002). The relative contribution of improvements on farm, transport, lairage, in slaughter scheduling, logistics and methodology remains to be clarified and are probably variable by region (Boes et al. 2001; Quirke et al. 2001) but statistically based abattoir surveillance provides a means of assessing progress to the point of slaughter. The situation in relation to antimicrobial resistance is less clear. Although the relatively high level of resistance in Salmonella and indicator organisms in pigs in various countries is a matter of concern (Moreno et al. 2000; Berends et al. 2001; Farrington et al. 2001; Oliveira et al. 2003), it is not clear that changes in farm medication can bring about reductions (Edrington et al. 2001).


This study was funded by Defra and the Meat and Livestock Commission. The authors would also like to thank abattoir owners, Meat Hygiene Service staff, Meat and Livestock Commission staff, and Laboratory and Epidemiology staff in Regional and Central Laboratories of the VLA.