SEARCH

SEARCH BY CITATION

Keywords:

  • Salmonella;
  • surveillance;
  • pigs;
  • simulation;
  • meat-juice;
  • carcass swab;
  • pork

Summary

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix

Salmonella in pork can be combated during pre- or post-harvest. For large slaughterhouses, post-harvest measures like decontamination might be cost-effective while this is less likely with small-to-medium sized slaughterhouses. In this study, pre-harvest measures might be more relevant. We describe an extended surveillance-and-control programme for Salmonella in finisher pigs, which, to establish equivalence to the Swedish control programme, is intended for implementation on the Danish island, Bornholm. The effect of the programme on food safety was estimated by analysing Salmonella data from pig carcasses originating from herds that would have qualified for the programme during 2006–2008. Food safety was interpreted as prevalence of Salmonella on carcasses as well as the estimated number of human cases of salmonellosis related to pork produced within the programme. Data from the Danish Salmonella programme were obtained from Bornholm. We used a simulation model developed to estimate the number of human cases based on the prevalence of Salmonella on carcass swabs. Herds are only accepted in the programme if they have one or less seropositive sample within the previous 6 months. In this way, the Salmonella load is kept to a minimum. The programme is not yet in operation and pigs that qualify for the programme are currently mixed at slaughter with those that do not qualify. Therefore, we had to assess the impact on the carcass prevalence indirectly. The prevalence of Salmonella in carcass swabs among qualifying herds was 0.46% for the 3 years as a whole, with 2006 as the year with highest prevalence. According to the simulation the expected number of human cases relating to pork produced within the programme was below 10. When the programme is in operation, an extra effect of separating pigs within the programme from those outside is expected to lower the prevalence of Salmonella even further.


Impacts

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix
  • • 
    For large slaughterhouses, post-harvest interventions might be cost-effective to achieve a low Salmonella prevalence in pork. However, for small-to-medium sized slaughterhouses, pre-harvest measures might be more feasible.
  • • 
    We describe an example of a Salmonella surveillance-and-control programme using data from the Danish island of Bornholm.
  • • 
    Our analysis shows that it is possible to obtain a low prevalence of Salmonella in pork produced within the programme.

Introduction

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix

Denmark exports around 85% of its pork production, which is why it is relevant to focus on the requirements listed by the importing countries, where Finland and Sweden play a role. Both countries have a recognized unique position with regard to Salmonella within the European Union (EU). This has been approved by the EU and enables the two countries to require additional guarantees when importing fresh meat to ensure a low risk of Salmonella in imported fresh meat. The official control requires that a large number of samples are taken from every batch of meat prior to export, and positive batches are rejected. For example, of a batch of 200 units, 50 must be sampled (Commission Regulation 1688/2005; Anonymous, 2005).

In accordance with Article 8 of Regulation 853/2004, fresh meat of bovine and porcine animals intended for Finland and Sweden must be tested for Salmonella (Anonymous, 2004). However, according to sub-paragraph (d) of Article 8, these microbiological tests need not be carried out for fresh meat from an establishment subject to a control programme recognized by the EU as equivalent to that approved for Finland and Sweden (Anonymous, 2005).

In Denmark, a serological Salmonella surveillance programme for finisher pigs has been in place since 1995. Accordingly, monthly meat-juice samples are taken in connection with slaughter. The samples are examined for Salmonella antibodies by use of the Danish Mix-ELISA (Nielsen et al., 1998). The cut-off for individual meat-juice samples is OD% 20 (Alban et al., 2002). The herds are assigned to one of three herd-levels based on the proportion of Salmonella-positive samples found in the previous 3 months (Alban et al., 2002). Herds allocated to level 2 or 3 are subjected to a penalty fee, which motivates farmers to seek advice from veterinarians to reduce the within-herd prevalence of Salmonella.

However, within the last couple of years, it has proved difficult to lower the prevalence of Salmonella in Danish market-age pigs further. The recently conducted EU baseline study of Salmonella in finisher pigs showed that – measured by use of lymph nodes – Danish finisher pigs had a prevalence of Salmonella of 7.7% which was a little lower than the EU average (10.3%). When measured as carcass prevalence, Danish pork had a prevalence of 3.3% which is lower than the EU average of 8.3%. In comparison, Sweden and Finland showed prevalences of 1.3% and 0.0%. This indicate that the Danish control programme is not able to ensure a Salmonella prevalence in pigs – and thus a food safety level – which is equivalent to that in Sweden or Finland (http://www.efsa.europa.eu/EFSA/efsa_locale-1178620753812_1178713190037.htm).

Salmonella in pork can be combated during pre- or post-harvest. Recently, it has been shown that at least for large slaughterhouses, post-harvest measures like hot-water decontamination might be cost-effective and will most likely have a significant impact on food safety as described by Alban and Sørensen (2009) and Barfod et al. (2009). Even though equivalence to Sweden and Finland in terms of food safety most likely can be achieved by this intervention, the focus in the currently running negotiations is equivalence of control programmes at the pre-harvest level and their ability to produce a low prevalent output from the slaughterhouses.

The concept of regionality opens for establishing equivalence for well-defined regions within countries. On the island of Bornholm, the prevalence of Salmonella in finishers and in fresh pig meat has consistently been lower than in the rest of Denmark for several years. This conclusion is based on the routine surveillance data from serological analysis of meat-juice and bacteriological examination of carcass swabs. Thus, pig production on Bornholm is already in a situation where equivalence to Sweden and Finland is close to be achieved at the pre-harvest level. The question is, therefore, whether it is possible to define robust surveillance and control programme covering fresh pig meat from the slaughterhouse on Bornholm, which can be accepted by EU as equivalent and will enable the slaughter plant on Bornholm to be exempt from carrying out microbiological testing of pork meat prior to export to Sweden and Finland.

Therefore, for Bornholm a surveillance-and-control programme, known as SALCONMEAT, has been designed on top of the existing Danish programme. Only herds that fulfil the requirements are included in the programme. The programme covers feed, housing, replacement of stock as well as requirements for sow herds, finisher herds, and the slaughterhouse on Bornholm.

The purpose of the present work was to describe the programme and to assess its effect on food safety. By analysing historical data from Bornholm from the period 2006 to 2008, it has been possible to identify the herds that theoretically fulfilled the requirements of the programme as well as the number of herds that did not meet the criteria. First, we unravelled the proportion of Salmonella-positive carcasses among pigs originating from herds that qualified for the SALCONMEAT programme. Hereby, we were able to estimate the effect of the SALCONMEAT programme in terms of the output of Salmonella positive carcasses from the slaughterhouse. By applying a new risk model developed by Barfod et al. (2009), we were also able to study the food safety implications of the SALCONMEAT programme by simulating the number of human cases related to the pork produced within the programme. We make our evaluation based on data from the current pre-harvest surveillance (meat-juice results used to identify positive herds) and slaughterhouse monitoring (carcass swab sampling). Hence, we are interested in seeing how low a prevalence of Salmonella you can reach by use of a system like the Danish where you only allow one seropositive reactor per herd.

Materials and Methods

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix

Description of the production premises on Bornholm and the SALCONMEAT Salmonella control programme

Bornholm is an island located in the Baltic Sea, east of the rest of Denmark and south of Sweden. A total of 45 000 inhabitants live on Bornholm, which is connected to Denmark, Sweden and Poland by four ferry services. There is currently only one feed company located on Bornholm. In 2008, there were 203 pig herds producing approximately 408 000 finishers annually. The most common size is a production of 1000–2000 finishers per herd per year.

The current Danish Salmonella surveillance-and-control programme applies to the slaughterhouse on Bornholm as it does to finisher herds and abattoirs in the rest of the country. This includes requirements for the feed factory, serological monitoring of nucleus and multiplier herds as well as finisher herds (Nielsen et al., 2001). In the Danish Salmonella programme, a large number of pen-faecal samples in herds in level 2 and level 3 are collected and tested microbiologically (Alban et al., 2002). The distribution of Salmonella serotypes in the pig production is monitored this way and it is documented that more than 99% of the isolates carry an antigenic structure which is covered by the Danish Mix ELISA used in the Danish serological surveillance system (Baptista et al., 2009).

To qualify for the SALCONMEAT programme, extra requirements must be met by producers, including the feed suppliers. Serological monitoring of the finisher herds is conducted using a sample size large enough to detect a prevalence of at least 5% (95% confidence level). Samples are dispatched daily and results are available 2 days later and before the next delivery of finishers from the herd. Only herds which have a maximum of one seropositive sample within the previous 6 months – after the last delivery – are included in the programme. These herds are allowed to deliver animals in the following week. Herds with one seropositive sample within the previous 6 months are carefully observed week by week, because they have a higher probability of having an additional seropositive sample which will result in the herd having to leave the programme. Herds with two or more seropositive samples are excluded from the programme and are informed about this as soon as the test-results are received at the slaughterhouse. A sow herd and a finisher herd are most often contractually connected.

Nucleus and multiplier herds are monitored with an increased intensity by use of both serology and bacteriology. A maximum of one positive serological reactor is allowed while all bacteriological tests must be negative. The sow herds are monitored indirectly through finishing herds that they supply with stock.

If Salmonella is found in a sow pen-faecal sample, both the sow herd and the finisher herd are excluded from the programme. If a positive sample is found in a finisher herd, the sow herd supplying piglets is inspected through trace-back and trace-forward (the other finisher herds that have received pigs from the sow herd).

Herds that have one seropositive sample are listed monthly and observed carefully week by week. It should be noted that since January 2009, a total of three herds did not fulfil the criterion, and six herds were under observation.

The programme is not yet in operation because of the extra costs of running it. However, the existing serological data allows us to investigate how many of the slaughter pig herds on Bornholm that would have been excluded from the program in January 2009 if SALCONMEAT had been in operation (Appendix 1).

Data sources

By use of data from the Danish serological surveillance programme covering a 3-year period from January 2006 to December 2008, we were able to divide the herds into those fulfilling the requirements of the SALCONMEAT programme and those that did not. This assignment of herds was dynamic because the Salmonella status of the herds changed from month to month.

In Danish pig slaughterhouses, a Salmonella monitoring programme is in place and includes daily swabbing of 300 cm2 of five carcasses. Three areas of 100 cm2 are swabbed by using the same gauze tampon. The method is recognized by the US Food Safety and Inspection Service (FSIS, 1996). The only difference between the method recommended by FSIS and the Danish method is that five carcasses are sampled in the Danish programme instead of one required by FSIS. The swab samples from five carcasses are analysed as one pooled sample. The single sample prevalence was calculated from the pooled sample prevalence, by using a conversion factor of three (Sørensen et al., 2007) instead of five, hereby taking loss of sensitivity related to pooling into account. The herds of origin of the five carcasses represented in the pooled swab samples are registered.

The data were divided into three groups: according to the herd origin of each of the five samples in the pooled sample.

  • 1
     All five samples originated from herds that all qualified for the SALCONMEAT programme.
  • 2
     Lack of serological samples from at least one herd, hence its status with regards to the SALCONMEAT programme was unknown.
  • 3
     At least one of five samples originated from a herd that did not qualify for the SALCONMEAT programme.

Data analysis

For calculating the chi-square value, the VassarStats web site for statistical computation was accessed and here Fisher Exact Probability Test for 2 × 2 contingency tables was used (please see http://faculty.vassar.edu/lowry/VassarStats.html).

A simulation model developed by Barfod et al. (2008) was used to estimate the number of human cases of salmonellosis to be expected from pork (fresh pig meat) produced within the SALCONMEAT programme. The model provides an opportunity to estimate the number of human cases to be expected from a pig population with certain Salmonella carcass prevalence at slaughter. Thus, the model was used to estimate the number of human cases to be expected with/and without SALCONMEAT in place.

The model is based, among others, on the parameters included in Table 1. In a year, the Danish population consisting of around five million inhabitants consumes a little more than two million carcasses. A serving size of 200 g was used. As pork is not eaten raw, the model holds the assumption that the number of human cases is proportionate to the number of bacteria present in the raw meat. To adjust for this assumption, the model contains a reduction factor that deals among others with the effect of preparing food including heat-treatment. This factor is analogous to the so-called a-factor in the Danish Salmonella-attribution model (Hald et al., 2004) and anchors the model output to the observed number of human cases in 2006 in Denmark. The reporting rate was set to around 5%. That implies that it is assumed that only about 5% of the true Salmonella cases are reported; presumably the mildest cases are not.

Table 1.   Parameters used in simulation modela assessing the number of human cases related to Salmonella in pork, Denmark, 2008
ParameterValueOrigin of information
  1. aSimulation model developed by Barfod et al. (2008).

No. pork carcasses for national consumption2 170 800Danmarks Statistik/Danish Agricultural & Food Council
Carcass weight75 kgDMA
Size of serving0.2 kgAuthors opinion
Servings per carcass375Calculated
No. servings814 050 000Calculated
Population5 427 000Danmarks Statistik
Yearly consumption of pork30 kg/personEurostat (corrected)
Reduction factor (RF)0.0142Estimated RF = Observed_incidence/incidence_predicted_without_factor
Reporting rate0.0539 RiskNormal(0.05; 0.005)Chalker and Blaser, 1988

A stochastic simulation model was constructed in @Risk (Palisade Corporation, Middlesex, UK) using Monte Carlo simulation (Latin hypercube sampling). The model simulates whether a serving is contaminated with Salmonella or not, based on quantitative input data from 2005 to 2006 from five Danish abattoirs. In short, carcass swab samples of 2800 cm2 and faecal samples were obtained from around 2000 pig carcasses and tested quantitively for Escherichia coli and Salmonella (unpublished data, S. Aabo). These data provided the basis for modelling the quantitative number of Salmonella bacteria on the carcass by simulation based on a combination of a faecal carcass contamination model (Barfod et al., 2008;.) and the concentration of Salmonella bacteria, in the faeces (Salmonella/g faeces). In short, the faecal contamination was modelled by sampling from the lognormal distribution of E. coli in swabs and in faeces and the number of Salmonella/carcass was achieved by sampling from the binomial distribution for Salmonella prevalence followed by using the RiskDiscrete function on faecal Salmonella concentration data. From the simulated distribution describing the number of Salmonella per carcass, a distribution of the number of Salmonella bacteria for Salmonella-positive serving was modelled. Next, for the positive servings, it was subsequently assessed whether a given quantitative number of Salmonella bacteria in a serving would result in disease. This was performed by use of a β-poisson dose–response model, as recommended by FAO/WHO (2002):

  • image( (Equation 1))

It should be noted from Fig. 1 that with a dose of Salmonella of approximately 104, around 50% of exposures would lead to sickness. It should also be noted that for very low doses of Salmonella (<10) the probability of becoming ill becomes almost insignificant.

image

Figure 1.  Graphical description of dose–response model for Salmonella recommended by FAO/WHO (2002) and used in a modelling of the number of human cases related to pork produced according to the SALCONMEAT programme, 2009, Denmark.

Download figure to PowerPoint

To evaluate the effect of the SALCONMEAT programme, the national surveillance prevalence of 1.0%Salmonella positive carcasses was used as reference point in the risk model. By entering the Bornholmean surveillance prevalences of 0.5% and 0.2%, the number of cases, which could be expected if Bornholm delivered all pork meat to the Danish marked in 2006 could be modelled.

Results

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix

The number of herds that would have qualified for the SALCONMEAT programme varied over the previous 3 years (Fig. 2). Initially, approximately 170 herds qualified. The number of qualified herds declined to 120 during the first half of 2007 and increased during the second half of 2007 to 140. By the end of 2008, approximately 130 herds had qualified. The latter decline is probably a result of herds stopping production. The number of herds outside the programme has varied between 10 and 50. Figure 2 shows the monthly number of herds inside and outside the programme, despite that, as described in Materials and Methods, a weekly status is assigned to each herd.

image

Figure 2.  The number of finisher herds and the total number of meat-juice samples positive for antibodies against Salmonella divided according to whether the herd would have qualified for the SALCONMEAT programme for Salmonella during 2006–2008, Bornholm.

Download figure to PowerPoint

The fluctuations in the number of herds qualifying for the programme is also reflected in the total number of positive meat-juice samples found in herds that qualified/did not qualify for the programme (Fig. 2). A maximum was observed in January 2007 when 33 positive samples were found among the herds outside the programme and 15 samples among the herds inside the programme. Since then, the total number of positive meat-juice samples has declined substantially. In all 3 years, the total number of positive meat-juice samples in the herds within the programme is much lower than the number of positive samples in the herds outside the programme – despite the fact that the number of herds within the programme is much higher than the number of those outside the programme.

For Salmonella, there is an association between herd serology and bacteriology measured on, for example, the carcass; however, there is also a variation between the slaughterhouses in their ability to deal with Salmonella (Sørensen et al., 2004). It is, therefore, important to look at Salmonella data from the slaughterhouse monitoring programme, specifically from Bornholm (Table 2). On the basis of the herd origin of the carcass swab samples, the data were divided according to whether the herd would have been within or outside the programme at the time the samples were taken. Only pools where all five carcass samples came from herds that qualified for the programme were used for the estimation of the effect of the programme.

Table 2.   Prevalence of Salmonella in 3550 carcass swab samples divided according to whether the finisher pig herd qualified for the SALCONMEAT programme or not, by year, 2006–2008, Bornholm
 Prevalence of Salmonella in given year
200620072008
Pools of 5CarcassaPools of 5CarcassaPools of 5Carcassa
Pos/total(%)Pos/total(%)Pos/total(%)
  1. aA conversion factor of 3 was used to convert the prevalence of Salmonella measured in a pool of samples to the prevalence on the individual carcass sample (Sørensen et al., 2007).

  2. bA specific number of meat-juice samples are required from each herd before it can be allocated to the programme.

Within programme3/1100.910/10802/1480.45%
Outside programme or lackb of samples taken2/1320.516/1241.61%2/880.76%

Table 2 shows that the prevalence of Salmonella-positive carcasses varied slightly during 2006–2008. Among the herds that would have qualified for the programme, a carcass prevalence of 0.91%, 0% and 0.45% was obtained for 3 years, respectively. When the data for 3 years are combined, a carcass prevalence of 0.46% was obtained for the herds within the programme and 0.97% for the herds outside the programme (P = 0.19). If only data from 2007 and 2008 are used, then a prevalence of 0.26% was obtained within the programme compared with 1.26% for the herds outside the programme (P = 0.049).

According to the simulation, a total of 107 human cases (90% CI 0–366) were attributable to pork in Denmark in 2006 (Table 3). If we assume that all pork consumed in Denmark originate from SALCONMEAT alone, we set the individual carcass Salmonella prevalence to 0.5 or 0.2%, which resembled the findings of 2006–2008 or 2007–2008, respectively. For these scenarios, the model predicted 52 and 21 human cases, respectively.

Table 3.   Simulateda number of human cases of Salmonella in Denmark ascribed to pork assuming three different prevalence levels of Salmonella and origin of production, 2008
ScenarioOrigin of porkPrevalenceb in pork based on carcass swabs (%)Average no. human cases of Salmonella (90% CI)
Assuming all pork in Denmark is of specific originAdjusting for production capacity in area of origin
  1. aBased on simulation model developed by Barfod et al. (2008).

  2. bPrevalence calculated as the individual carcass prevalence.

1Entire Denmark1.0107 (0–366)107 (0–366)
2aOnly   
 SALCONMEAT0.552 (0–298)  9
2bOnly   
 SALCONMEAT0.221 (0–92)  4

The initial results presented in scenario 2a and 2b in Table 3 assume that all pork consumed in Denmark originates from SALCONMEAT. However, to estimate the number of human cases caused specifically by pork from SALCONMEAT, an adjustment to the actual production of pork within the programme was needed. On Bornholm, 406 000 finishers are slaughtered annually, and of these, 393 000 carcasses are expected to be produced within the SALCONMEAT programme. In Denmark, around 2.2 million carcasses are consumed annually. Hence, production within the SALCONMEAT programme is 393 000/2 170 800 = 18% of the national production.

Table 3 presents the estimated number of human cases of salmonellosis related to the production of pork within the SALCONMEAT programme. It is observed that the number is nine or four – depending upon whether data from 2006to 2008 or 2007 to 2008 were used.

As model control, we used the national carcass prevalence of Salmonella in 2006 which was 1.0%. In that year, 101 human cases were ascribed to Danish pork according to a mathematical model that compared the number of human cases caused by different Salmonella sero- and phage types with the distribution of the Salmonella types isolated from the various animal-food types (Anonymous, 2006). Consequently, if a prevalence of 0.5% is assumed for pork from SALCONMEAT and, as a maximum, production constitutes 18% of what is consumed in Denmark, then around 101 × (0.005/0.01) × 0.18 = 10 human cases can be expected. Similarly, when a carcass prevalence of 0.2% is assumed, the number of human cases is expected to be around four.

Discussion

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix

The programme may commence once it has been approved by the EU as equivalent to the programme in Sweden and Finland. However, the effect of the programme can be estimated by examining the data from the herds which would have qualified for the programme.

For a given programme to be successful, it needs to be able to correctly allocate herds into those which qualify and those which do not qualify for the programme. And this not only applies at the beginning of the programme. Herds that at some stage no longer qualify for the programme should be removed from the programme immediately. Hereafter, the herds can receive advice on risk-mitigating action with respect to Salmonella. Once the action taken leads to observable results that qualify the herds, then they can re-enter the programme.

In the SALCONMEAT programme, the criterion for entering and leaving the programme is clearly stated. It is noted that the number of herds within the programme has varied between 2006 and 2008. In the first half of 2007, however, a larger number of herds left the programme because of the detection of Salmonella-positive meat-juice samples. Half a year later, the herds re-entered the programme (Fig. 2).

With regard to sow herds, the requirement for re-entering is that no positive pen-faecal samples are found within the previous 6 months. For the finisher herds, a maximum of one seropositive meat-juice sample within the previous 6 months is required. This implies that herds that do not qualify for the programme often remain excluded for 3–6 months before the requirement is met.

As soon as more than one seropositive sample is found, the herd can no longer participate in the programme. In this study, we have presented the data as if the assignment of herds took place on a monthly basis because the data are stored as monthly records. A cut-off of OD% 20 is used in the Danish programme, despite that the correlation between serology and bacteriology is slightly higher at cut-off OD% 11 (Alban et al., 2002). The reason is that a level 0 is used in the Danish programme for herds with continuous negative samples. It has been considered that would be detrimental for level 0 if cut-off of OD% 11 was chosen, because this would increase the proportion of false-positive results (Alban et al., 2002). However, it would be of interest to study what the effect of lowering the cut-off in the SALCONMEAT programme would be on the ability to produce pork with a low prevalence of Salmonella.

To promote reassignment of herds outside the programme, the herd owners receive advice from the responsible slaughterhouse consultant in relation to (i) reducing the risk of introducing Salmonella into the herd and (ii) reducing the spread of Salmonella within the herd if introduced. The advice on Salmonella introduction deals with rules that any person, including visitors, must abide by when entering the herd as well as rules for bringing animals and feed into the herd. The advice regarding mitigating internal spread concerns segregating the herd into smaller units. The aim was to ensure that each unit consists of a more or less uniform group of pigs (based on weight of pigs or function of sows) with a specified hygiene level. The unit with weaned piglets should have the highest hygiene level. Any passage from 1 unit to the next should be carried out most carefully. Furthermore, care should be taken to ensure that the other slaughter pig units (young and older) maintain their hygiene level. Information about cleaning procedures is crucial for the maintenance of a high level of hygiene. And it is recommended that washing, disinfection and liming are carried out.

Usually, herd owners are motivated to implement the advice given. To take into account the variation in how pig production is conducted, the advice given is based on risk factors present in the individual herd. Some known risk factors play a minor role for the herds in the SALCONMEAT programme, for example, feed and trade, both because of local tradition and because requirements are already a part of the programme. Experience shows that protection against infections like Salmonella is not always taken sufficiently seriously in all herds at all times. However, in most cases, it is possible to obtain visible results after an advisory visit to the herd, resulting in a risk-mitigating effect on Salmonella.

Based on data from the slaughterhouse monitoring, the SALCONMEAT programme seem to have an effect because the prevalence of Salmonella-positive carcasses is lower among the herds that would have qualified for the programme (0.46%) than among the remaining herds (0.97%). However, this is not statistical significant when evaluating the 3 years (P = 0.19), but significant when evaluating the latest 2 years (P = 0.049) (Table 2). Because of export of pork to the US, the Danish slaughterhouse monitoring is made in accordance with the US authorities, FSIS. If the area to swab was increased from the current 300 cm2 to, for example, 1400 cm2, then the sensitivity would increase.

One of the main points of difference between the SALCONMEAT programme and the Danish Surveillance-and-Control programme for Salmonella in finisher herds is that Salmonella-positive herds are removed from the SALCONMEAT programme and transported and slaughtered on separate days. The aim of this work was to see how well such a system would perform. The strategy in the programme is somewhat equivalent to the Swedish sanctions against Salmonella-positive herds.

The programme is not currently in operation because of the high costs involved. Consequently, the evaluation of the expected effect in respect of Salmonella was based on historic data. Once the programme is in operation, pigs from herds outside the programme will be slaughtered separately from pigs from herds within the programme. This reduces cross-contamination between carcasses harbouring Salmonella and carcasses not harbouring Salmonella. In general, logistic slaughter has a limited effect among herds with lower levels of Salmonella (Alban et al., 2007). However, recent results show that if the number of seropositive pigs slaughtered in a day can be kept below around 50, then roughly a halving of the prevalence can be expected (Dahl, 2009). The study showed that for slaughterhouses where more than 50 seropositive pigs per day were introduced, approximately half the positive carcasses could be attributed to cross-contamination. The other half were positive because of Salmonella from the sampled pig itself. Reducing the number of seropositive pigs below the threshold of 50 seropositive pigs per day is feasible in the SALCONMEAT programme because:

  • 1
     the slaughterhouse only slaughters between 1700 and 2100 pigs a day;
  • 2
     the herds in the programme are only allowed to have a maximum of one positive sample within the previous 6 months.

This implies that there are very few Salmonella-positive pigs slaughtered on the days when the programme is in action. Only recently infected herds that have not yet been identified as infected might be slaughtered together with the non-infected herds. The impact will be that the prevalence of Salmonella will be lower than seen today; and most likely less than 0.2% measured as the individual carcass prevalence.

The simulation model applied is both flexible and robust and is designed to contrast the effect of different levels of slaughter hygiene and concentration of enteric pathogens. According to the simulation, a total of 107 human cases could be ascribed to Danish pork in 2006. This is very close to the 101 cases reported as the official number ascribed to pork in that year (Anonymous, 2006). Although the majority of simulations give estimates of human illness close to the observed annual incidences, occasional sampling of extreme values from the input distributions can cause prediction of a high number of human cases, corresponding to outbreaks naturally occurring. Consequently, the output distributions are not normally distributed but skewed towards the right (observable from the 90% CIs in Table 3).

The prevalence of Salmonella in pork produced within the SALCONMEAT programme is low. It might have been higher if the area swabbed was increased to 1400 cm2. However, the simulation model that was used is based on the current monitoring where 300 cm2 is swabbed. The estimated number of human cases relating to the pork produced within the SALCONMEAT programme is very low; <10. This is a result of (i) a low prevalence of Salmonella on the carcasses and (ii) a limited amount of pork produced within the programme. It is questionable whether such a small number will show up in the national statistics of the importing country. It is up to a risk manager to decide whether the limited number of expected human cases relating to the pork produced under the SALCONMEAT programme is considered a low or a negligible risk. It might be speculated that if the exposure of a population to Salmonella is very low, then the immunity is low as well. However, the difference in salmonellosis notification rate (human cases per 100 000 inhabitants) between the two countries is not that large: 45.6 for Sweden and 67.0 for Denmark (please see http://www.efsa.europa.eu/en/scdocs/doc/1496.pdf).

EU countries with very low prevalence of Salmonella in pork (Sweden and Finland – and with them also Norway which is not a member of EU) are currently allowed to require testing of meat prior to export (Anonymous, 2005). However, an outbreak of Salmonella Typhimurium U288 occurred in Norway and Sweden during November and December 2008, which was caused by Danish pork that had tested negative in export tests (Bruun et al., 2009). This emphasizes that food safety is not an outcome of testing but an outcome of management. Testing will only verify the quality of a management programme with a certain probability.

In a few of years, EU will set targets for the accepted level of Salmonella in the pork production. It is unknown at present whether the targets will be set at pre- or post-harvest level – or both. For small-to-medium sized slaughterhouses, post-harvest measures like hot-water decontamination might not be cost-effective and pre-harvest measures, therefore, might be the solution. The principles behind a surveillance-and-control programme like SALCONMEAT might, thus, be a feasible option if the targets are set at a low prevalence. Our results show that for a medium-sized slaughterhouse, it is possible to keep the prevalence of Salmonella in pork very low by setting strict pre-harvest requirements based among others on a serological testing programme.

Acknowledgements

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix

Because the SALCONMEAT project has been developed over a number of years, a large number of people have contributed to the process of designing the programme. They include Bent Nielsen (Bayer) and Claes Enøe (National Veterinary Institute). Lene Lund Sørensen and Vibeke Møgelmose (Danish Meat Association), Flemming Bager (Danish Veterinary Institute), Thorkild Laursen and Anders Jensen (Danish Crown), Finn Poulsen (DLG), Therese Brøndsted (Danish Plan Directorate), Stig Andersen and Kirsten Vogt Kyndesen (pig Advisory Services, Bornholm), as well as Henrik Hjorth Jensen (BAF).

References

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix
  • Alban, L., and L. L. Sørensen, 2009: Prevalence of Salmonella Spp after hot-water decontamination of finishers slaughtered in a Danish abattoir during 2004-2008. 8th International symposium on the Epidemiology and Control of Foodborne pathogens in pork (SafePork), 30 September to 2 October 2009, Québec City, Canada, pp. 405408.
  • Alban, L., H. Stege, and J. Dahl, 2002: The new classification system for slaughter-pig herds in the Danish Salmonella surveillance-and-control program. Prev. Vet. Med. 53, 133146.
  • Alban, L., J. Boes, and L. L. Sørensen, 2007: Begrænset reduktion af Salmonella på slagtekroppe ved logistisk slagtning (limited reduction of Salmonella on carcasses due to logistic slaughter). In Danish. Veterinær Information No. 709, March 30, 2007. Available at: http://www.danishmeat.dk/Veterinaerfagligt/VetInfo/2007/2007/VetInfo_709.aspx.
  • Anonymous, 2004: Regulation (EC) No 853/2004 of the European Parliament and of the Council of 29 April 2004 laying down specific hygiene rules for the hygiene of foodstuffs.
  • Anonymous, 2005: Regulation (EC) No 1688/2005 of 14 October 2005 implementing Regulation (EC) No 853/2004 of the European Parliament and of the Council as regards special guarantees concerning salmonella for consignments to Finland and Sweden of certain meat and eggs.
  • Anonymous, 2006: Annual Report on Zoonoses in Denmark 2006. Technical University of Denmark, 49 pp.
  • Baptista, F. M., L. Alban, A. K. Ersbøll, and L. R. Nielsen, 2009: Factors affecting persistence of high Salmonella serology in Danish pig herds. Prev. Vet. Med. 92, 301308.
  • Barfod, K., S. Aabo, A. H. Andersen, and D. Lo Fo Wong, 2008: Modelling Salmonella Carcass Contamination and Decontamination in Slaughter Pigs. In: Proceedings of Food Micro 2008, 1–4 September, Aberdeen, Scotland, 104 pp. Available at: http://www.foodmicro2008.org/fm08/resources/374/1143/pdf/FM2008_0660.pdf.
  • Bruun, T., G. Sørensen, L. P. Forshell, T. Jensen, K. Nygard, G. Kapperud, B. A. Lindstedt, T. Berglund, A. Wingstrand, R. F. Petersen, C. Müller, C. Kjelsø, S. Ivarsson, M. Hjertqvist, S. Löfdahl, and S. Ethelberg, 2009: An outbreak of Salmonella Typhimurium infections in Denmark, Norway and Sweden, 2008. Eurosurveillance 14, 16.
  • Chalker, R. B., and M. J. Blaser, 1988: A review of human salmonellosis: III. Magnitude of Salmonella infection in the United States. Rev. Infect. Dis. 10, 111124.
  • Dahl, J., 2009: Risk factors for positive carcass swabs in Danish slaughterhouses. 8th International symposium on the Epidemiology and Control of Foodborne pathogens in pork (SafePork), 30 September to 2 October 2009, Québec City, Canada, pp. 2629.
  • FAO/WHO, 2002: Risk assessment of Salmonella in eggs and broiler chicken, ISBN 92 9 156229 3. Available at: http://www.fao.org/DOCREP/005/Y4392E/y4392e00.htm.
  • FSIS, 1996: Pathogen reduction; hazard analysis and critical control point (HACCP) systems; final rule. Food Saf. Inspect. Serv. 61, 3891738928.
  • Nielsen, B., L. Ekeroth, F. Bager, and P. Lind, 1998: Use of muscle juice as a source of antibodies for large scale serological surveillance of Salmonella infection in slaughter pig herds. J. Vet. Diag. 10, 158163.
  • Nielsen, B., L. Alban, H. Stege, L. L. Sørensen, V. Møgelmose, J. Bagger, and D. L. Baggesen, 2001: A new Salmonella surveillance and control programme in Danish pig herds and slaughterhouses. Berl. Münch. Tierärztl. Wschr. 114, 323326.
  • Sørensen, L. L., L. Alban, B. Nielsen, and J. Dahl, 2004: The correlation between Salmonella serology and isolation of Salmonella in Danish pigs at slaughter. Vet. Microbiol. 101, 131141.
  • Sørensen, L. L., H. Wachmann, and L. Alban, 2007: Estimation of Salmonella prevalence on individual-level based upon pooled swab samples from swine carcasses. Vet. Microbiol. 119, 213220.

Appendix

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  10. Appendix

Appendix 1

Graphical presentation of serological surveillance results from finisher herds that were assigned to be outside the SALCONMEAT programme in January 2009 or which were under observation because of the finding of one seropositive sample within the previous 6 months, Bornholm 2009

Herd no.No. seropositive samples found in herd in month in year 2008
JanFebMarAprMayJunJulAugSepOctNovDec
  1. The vertical line is used to indicate that only the most recent 6 months’ results are used in the assignment of a herd in January 2009. Figures in bold show the samples which formed part of the herd assignment in January 2009.

Herd could not participate in the SALCONMEAT programme in January 2009 because of >1 seropositive sample within the previous 6 months
1131 1  111 2
2        1  1
3 1  11  11  
Herd was under observation in January 2009 because of one seropositive sample within the previous 6 months
4           1
5           1
6         1  
712122   1   
8        1   
9 2 1  111