A Systematic Review on the Effectiveness of Pre-Harvest Meat Safety Interventions in Pig Herds to Control Salmonella and Other Foodborne Pathogens

This systematic review aimed to assess the effectiveness of pre-harvest interventions to control the main foodborne pathogens in pork in the European Union. A total of 1180 studies were retrieved from PubMed® and Web of Science for 15 pathogens identified as relevant in EFSA’s scientific opinion on the public health hazards related to pork (2011). The study selection focused on controlled studies where a cause–effect could be attributed to the interventions tested, and their effectiveness could be inferred. Altogether, 52 studies published from 1983 to 2020 regarding Campylobacter spp., Clostridium perfringens, Methicillin-resistant Staphylococcus aureus, Mycobacterium avium, and Salmonella spp. were retained and analysed. Research was mostly focused on Salmonella (n = 43 studies). In-feed and/or water treatments, and vaccination were the most tested interventions and were, overall, successful. However, the previously agreed criteria for this systematic review excluded other effective interventions to control Salmonella and other pathogens, like Yersinia enterocolitica, which is one of the most relevant biological hazards in pork. Examples of such successful interventions are the Specific Pathogen Free herd principle, stamping out and repopulating with disease-free animals. Research on other pathogens (i.e., Hepatitis E, Trichinella spiralis and Toxoplasma gondii) was scarce, with publications focusing on epidemiology, risk factors and/or observational studies. Overall, high herd health coupled with good management and biosecurity were effective to control or prevent most foodborne pathogens in pork at the pre-harvest level.


Introduction
In Europe, the current most important foodborne hazards in pork include microbiological agents (for example, Salmonella) [1,2]. The new risk-based meat inspection legislation was laid out in agreement with these hazards and proposes a risk-informed visual-only inspection where the focus is on the prevention and control of meat-borne hazards before slaughter, such as on-farm or at transport [3]. This integration of measures along the food chain requires cooperation between the different stakeholders and has the potential to consistently reduce the risks associated with meat-borne hazards.

Materials and Methods
This systematic review is part of a set of three reviews on the effectiveness of preharvest interventions to control foodborne pathogens in broilers, pigs, and bovine. Such work was framed in the context of the RIBMINS Cost Action (please refer to the Acknowledgements section). Likewise, the methods followed are similar to those described by Pessoa et al. [4] and the work presented here was conducted by the same two review coordinators, and two volunteer researchers. The backbone of the methodology used followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement [5] and EFSA's guidelines for conducting systematic reviews for food and feed safety assessments [6]. PRIMA's checklist for systematic reviews has been completed and is available as Supplementary Material (S1).
All literature searches were conducted on two online databases (PubMed ® and Web of Science) on 8 February 2021. Only peer-reviewed studies written in English and published before 31 December 2020 on the effectiveness of pre-harvest meat safety interventions to control 15 foodborne pathogens (those highlighted by EFSA [1]) in pigs were included. Searches were restricted to title and abstract. Figure 1 shows the composition of the search strings used in PubMed ® and in Web of Science. Keywords and search strings specific to each pathogen are presented in Table 1. The detailed search strings employed in each database are available as Supplementary Material (S2).  [4], which employed the same methods). The search strings used are available as supplementary material. Legend: MRSA-Methicillin-resistant Staphylococcus aureus (MRSA); VTEC-Verocytotoxin-producing Escherichia coli (VTEC). The "*" corresponds to the code/character used for the searches in the online databases. By using a * we indicate that the search motor should retrieve all words that start like the example given, regardeless of how they end.
EndNote was used to import all search results. All duplicates were removed. The set of inclusion and exclusion criteria to filter titles and abstracts is presented in Table 2. One of these criteria was to only select scientific papers with experimental/controlled study designs. This decision was made to highlight the presumptive causal effect of the interventions tested. One co-author screened all 1180 records using these criteria and selected 87 papers for further analysis. After that, the selected papers were retrieved and two co-authors (in parallel and blinded to each other's decisions) read the full texts using the same eligibility criteria ( Table 2). Exclusion of records had to be agreed by both co-authors. Records upon which agreement was not reached were reviewed by a third co-author, producing a final decision. Table 1 shows the list of records included in all stages of the systematic review process. The data within the final 52 records included in this work were extracted onto a database (stored in a Microsoft Office Excel ® spreadsheet). Studies (i.e., any peer-reviewed original research in which the authors collected, analysed, and reported their own data) were documented and classified based on the pre-harvest intervention. Other information (country of study, year it took place, type of experiment, subject type, number of experimental units, sample type, outcome measured, and estimate of effectiveness) was also retrieved. Some studies assessed the efficacy of multiple interventions. For Salmonella studies, the comparison of each treatment (intervention) with the control was recorded as a trial and, if possible, detailed information was collected for each trial. For each Salmonella-related study, the results of the interventions tested were summarised according to whether there was a reduction of Salmonella shedding, reduction of Salmonella counts or improvement of protective immunity. Whenever the outcome of an intervention was measured through several time-points, data collected at the end of the study (i.e., closer to the slaughter date) were preferred.

Results
A total of 1180 unique studies published between 1968 and 2020 were retrieved through the search strings run on PubMed ® and Web of Science for the 15 pathogens in-cluded in this systematic review. After the review process described, a final list of 52 studies published between 1983 and 2020 were retained. This list is available as Supplementary Material (S3). In the full text analysis and selection, the authors had an agreement rate of 95.4% (83/87). Four studies were reviewed by a third author to decide upon its selection. Due to the decision of the third reviewer, one study out of the four was retained in the systematic review. All authors were blinded to each other's final decisions. The list of studies excluded during the full-text evaluation (n = 35) is available as Supplementary Material (S4). Only five of the pathogens listed had studies meeting the defined criteria (Table 1).

Campylobacter
The two studies retained for Campylobacter spp. tested the efficacy of probiotics to reduce the colonisation of this pathogen as competitive exclusion and consequently reduce the risk of carcass contamination during slaughter. Bratz et al. [7] tested the inhibitory activity of the strain E. faecium NCIMB 10,415 against C. coli in vivo. This probiotic was administered as a diet supplementation in sows (three weeks before parturition) and to their progeny from 12 days of age until the end of the trial. Sows and piglets from the control group were not fed any supplements. The authors reported that all piglets were already naturally colonised with C. coli before the challenge trial, which was a unique dosage of 7 × 10 7 cfu strain C. coli 5981 via an intragastric application. The excretion load of C. coli was monitored for 28 days and the results indicate that the tested probiotic did not significantly affect C. coli excretion levels in pigs. In the other Campylobacter spp.-related study, Hasan et al. [8] tested the effects of diet supplementation of resin acid-enriched composition (RAC) in the last week of gestation on colostrum yield, composition and gut microbiota. Three trials in three different commercial herds were performed. Apart from the colostrum yield and composition improvements, the diet supplementation with RAC seemed to shift the relative abundance of opportunistic and pathogenic agents, such as Campylobacter, potentially reducing the risk of piglet infection.

Clostridium Perfringens
Five studies assessing the efficacy of vaccinations (n = 4) and probiotics (n = 1) were retained. Of the vaccination studies, two of them tested sow and gilt vaccination strategies to control necrotizing enteritis (C. perfringens type C; [9]) and C. perfringens type A-associated diarrhoea in piglets [10]. Two other studies assessed the efficacy of piglet vaccination to control neonatal diarrhoea caused by Clostridioides difficile [11] and necrotizing enteritis (C. perfringens type C; [12]).
One study assessed the efficacy of competitive exclusion by administering a probiotic to control diarrhoea in piglets [13]. The cocktail tested contained living strains of attenuated C. perfringens type A and non-pathogenic Escherichia coli and it was administered per os to newborn piglets in a commercial farm with a history of neonatal diarrhoea caused by C. perfringens type A.
All studies reported positive outcomes for the interventions tested. Two studies reported a reduction in piglet mortality, which corresponded to a numerical but not statistical reduction in the study by Kelneric et al. [9] and to a statistically significant mortality rate reduction in the study by Unterweger et al. [13]. Hammer et al. [10] documented an increase of neutralizing antitoxins against C. perfringens type A in piglets born from vaccinated dams compared to those born of dams not vaccinated, and Richard et al. [12] reported higher titres against C. perfringens type C in vaccinated piglets when compared to those not vaccinated. Finally, in the study by Oliveira et al. [11], the authors documented a reduction of the isolation of C. perfringens in diarrhoea samples after administering a non-toxigenic strain of C. difficile to one-day-old piglets on a commercial pig farm.

Methicillin-Resistant Staphylococcus Aureus (MRSA)
Only one study on MRSA met the inclusion criteria. This study reported the results of a randomised control trial to test the efficacy of a thorough cleaning disinfection protocol for sows and the environment (farrowing house and nursery unit) to reduce the prevalence of livestock-associated MRSA in sows and their progeny [14]. Two farrow-to-finish commercial farms with a 3-week batch system were enrolled in the study, and, in each farm, six sow batches were tested (three batches tested and three batches as control, all with approximately 20 sows). Results showed that the tested disinfection protocol reduced temporarily the sow and piglet MRSA status, but it did not equate to a final reduction in MRSA at weaning or in the nursery unit.

Mycobacterium Avium Complex
Hines et al. [15] tested the efficacy of vaccination for Mycobacterium avium with two different vaccines in preventing infection and disease in experimentally challenged pigs. The study tested a killed "whole cell" M. avium serovar 2 as a vaccine, and a conjugated MIF-A3 subunit vaccine. The results showed that the killed vaccine did not prevent infection but attenuated its severity with regard to gross and macroscopic lesions, when compared to the pigs vaccinated with the subunit vaccine. The latter did not prevent infection and the lesions observed were very similar to pigs vaccinated with a sham vaccine (saline solution).

Salmonella
In total, 43 studies testing different pre-harvest interventions for the control of Salmonella infections in pigs were found . Table 3 compiles a description of the studies retained, with a summary of the trials reported in them and their results. Forty-one studies were designed to investigate on-farm interventions, and four studies tested transport interventions [19,20,29,48].  1 Two studies tested both on-farm and transport interventions. 2 One study had two trials, one performed in a commercial farm setting, and another performed under controlled research laboratory conditions. 3 The comparison of each treatment (intervention) with the control was recorded as a trial. Some trials consisted of a combination of approaches (i.e., acids in-feed and in water simultaneously). The trials are repeated across different categories if they fit in more than one type of intervention. 4 Examples of other interventions tested are off-site early weaning, washing and disinfecting of lorries, split marketing approaches, and different space allowances. * Trials which reported at least one positive result (i.e., reduction of Salmonella shedding, increase of protective immunity).
In total, 86 trials were identified among the 43 Salmonella studies selected. The most tested type of intervention was in-feed and/or water treatments. Out of the 32 trials that tested different acids in-feed or water (i.e., sorbic acid, sodium butyrate, or blends of citric acid, formic acid and essential oils) and other feed-related interventions like fermentation or herbal extracts, including prebiotics, 23 (72%), reported positive results.
Most vaccination trials (88%, 21/24) reported positive results. The most common challenges reported were the lack of cross-protection of some vaccines against other serotypes and the potential interference of vaccination-induced antibodies in the meat juice sampling for Salmonella control purposes at slaughter. The three trials reporting vaccination interventions without positive results were: (1) a trial where an oral vaccine administered to piglets at 3 weeks of age lowered transmission (numerically) but failed to reduce excretion of Salmonella Typhimurium [36]; (2) a trial where the vaccination of sows with a commercial vaccine to control of Salmonella Typhimurium infections failed to decrease the prevalence of Salmonella Typhimurium field strain positive lymph nodes at slaughter in finisher pigs born to those sows [58]; and (3) a trial where a commercial oral vaccine based on Salmonella serovar Choleraesuis variety Kurzendorf was administered between 24 and 72 h after birth and was not effective in reducing the within-herd spread of Salmonella during the finishing phase or the frequency of carcass contamination at slaughter, with Salmonella Typhimurium being isolated from lymph nodes of vaccinated pigs [57].
Eight trials tested the efficacy of administering antimicrobials to control Salmonella infections [16,24,27,39,55]. The only trial with a positive result was reported in a study combining intramuscular administration of ceftiofur with off-site early weaning at 10-15 days of age, where Nietfeld et al. [18] concluded that this intervention prevented Salmonella spp. infection in grow-finish pigs.
Of the seven trials reporting results on the efficacy of cleaning and disinfection interventions, two exclusively tested it on transport [19,29], and only one other trial reported the effect of cleaning and disinfection on-farm alone [50]. Rajkowski et al. [19] tested the effect of washing and sanitizing lorries after each load and it significantly reduced levels of Salmonella detected on lorries. Similarly, Mannion et al. [29], who tested bacterial loads and isolates on lorries carrying pigs from high-and low-risk farms during and after transport, and also after washing, commented on the need for better cleaning of lorries after each transport (or load), especially when transporting pigs from "high-risk herds of Salmonella". The authors found isolates identical to those on farm on lorries after washing. Finally, Martelli et al. [50] tested the application of a rigorous disinfection protocol of finisher facilities on-farm with the objective of eliminating Salmonella, comparing it to the normal procedures followed by farmers. The authors found that this protocol significantly reduced the prevalence of Salmonella in pigs prior to slaughter. Other studies reporting trials testing cleaning and disinfection have been described before or did not obtain positive results [17,24].
Several trials investigated the efficacy of a combination of different interventions. These included a combination of cleaning and disinfection with off-site early weaning [17]; feed withdrawal and duration of transport [20]; a combination of chlorate treatment and topical disinfection administered to piglets together with early weaning [28]; and a combination of different particle size and acids in feed [32]. All of the cited studies reported positive results in one or more of the interventions tested. Four trials (within three studies) testing combined interventions did not report positive effects [24,27,32].
Nineteen trials tested other types of interventions, either alone or in combination with the intervention types described above. Examples of other interventions tested are off-site early weaning [17,18], washing and disinfecting of lorries [19,29], split marketing [31], different space allowances [27,40], and feed withdrawal and transport times [20,48].

Discussion
Over the years, several studies have been published on pre-harvest interventions to control foodborne zoonoses in pork. In this systematic review, we aimed to identify controlled studies that could provide a certain level of confidence regarding the effectiveness of the interventions tested, rather than identifying risk factors for the control of infections by the biological hazards listed. The papers selected for full text analysis were published over an extended timeframe (1983 to 2020). Across these years, Salmonella was one of the biggest concerns, with related publications representing 49% (785/1606) of the initial search returns. This is a direct consequence of its high relevance as a biological hazard of concern in pork meat (EFSA, 2011) and of the vast research undertaken to address this issue. Indeed, Salmonella is currently the second most reported foodborne pathogen in the EU, having been the most reported pathogen for many years, and is commonly associated with the consumption of pork [59]. The fact that few studies were retrieved for other pathogens within the criteria defined highlights the need for further research on the effectiveness of pre-harvest interventions to control these hazards. Another possible explanation is that some of these pathogens may be more cost-efficiently controlled by post-harvest interventions.

Salmonella
S. Typhimurium is the most common Salmonella in pig herds in most European countries, and this agent is known to be introduced into the herds by healthy carriers among the breeding animals and also by contaminated feed [60]. However, there is an extensive list of additional risk factors connected to biosecurity that should be tackled at the herd level, such as birds, rodents, insects, water, manure, humans entering the piggery and environment, etc. [61]. Unsurprisingly, several types of interventions to control Salmonella were found in the literature. In line with the risk factors identified in the literature for Salmonella infections, the most common pre-harvest interventions identified were in-feed and/or water treatments as well as vaccination. Among the most effective interventions, cleaning and disinfection and vaccination appeared to have high success rates. Nevertheless, across all trials, the results for Salmonella are very encouraging, with 76% (65/86) of the trials assessed reporting positive results. Although there is no scope in this paper to debate the reasons for intervention failure, including vaccination failure, reported in the studies evaluated, other studies have systematically assessed the effect of vaccination as a control strategy against Salmonella infection in pigs [62], the efficacy and quality of evidence for five on-farm interventions for Salmonella reduction in grow-finish swine [63], and the evidence for effectiveness of primary production interventions to control Salmonella in pork [64]. In spite of these positive results and vast literature published, the endemic Salmonella spp. infections in pig herds across the world reflect how challenging it is to control this pathogen.
At national level, Finland, Norway and Sweden have documented that the successful control of Salmonella in cattle, pigs and poultry through pre-harvest interventions is possible. Heat-treatment of feed, and starting with breeding animals free from Salmonella at the top of the breeding pyramid have probably been the most important measures [61]. The food safety authorities have an important role following up positive herds to prevent transmission to other herds, humans and food, by prohibiting the purchase and transportation of animals and foods from infected farms. This highlights that prevention rather than control is a feasible pre-harvest intervention when targeting this hazard in pork.

Campylobacter
Multiple studies have shown that pigs are an important reservoir of C. coli and that it is difficult to control this species at the herd level [65][66][67]. It seems more cost-efficient to control this agent post-harvest. Given the sensitivity of Campylobacter to both freezing and drying, blast chilling has proved to significantly reduce this agent on carcasses' surface [68]. Even after traditional slow chilling there is a significant decline of this agent [69]. Accordingly, pig carcasses and pork are not regarded as an important source of Campylobacter in a public health context as confirmed by most epidemiological studies [70,71]. According to Roux et al. [72], "The aetiology of human C. coli infections is similar in a number of respects to C. jejuni but there are important differences. There is an increased risk of C. coli infection in the older people, in people who live in rural areas and during the summer months. Public health together with national and international food safety agencies should take these differences into account when considering interventions to reduce the incidence of this gastrointestinal pathogen".

C. perfringens
All of the C. perfringens-related studies reported outcomes referring to the control of disease in piglets and none reported or discussed the possible effects of the tested intervention to control shedding of this pathogen in the faeces. Thus, in spite of the apparent efficacy of the pre-harvest interventions tested, such as vaccination and competitive exclusion, these were not meant to control the risk of foodborne infections by C. perfringens acquired by pork consumption. This is likely to be related to the low risk this pathogen represents since the "risk of disease seems not to be correlated with occurrence in raw meat but rather to improper hygiene and storage" [1], meaning that this pathogen is mostly controlled by post-harvest interventions.

MRSA
The tested disinfection protocol in one study temporarily reduced the sow and piglet MRSA status, but it did not equate to a final reduction in MRSA at weaning or in the nursery unit.
Other similar trials testing thorough cleaning and disinfection of the facilities or sow washing and disinfection were captured in this review, but the absence of control groups dictated their exclusion.
However, more comprehensive measures have been successful. Norway has established a unique control strategy for MRSA in their pig population, which includes population-wide annual surveillance, in addition to contact tracing upon detection of MRSA in pig farms and farm workers. Restrictions prohibit trade of live pigs carrying MRSA, other than directly to slaughter. Following depopulation, the farm owner is responsible for thorough washing and disinfection of farm premises. After a mandatory down-time, the farm is repopulated with pigs from MRSA-negative herds [73]. The surveillance programme in 2019 detected only one pig herd with MRSA. In total, 722 herds were included in the survey [74].

Mycobacterium avium
The authors of the study [15] reported that it was not possible to determine if the vaccine tested had significantly reduced the bacterial load of the animals challenged, since low numbers of organisms were cultured. More importantly, the authors also note that the vaccines were not effective in controlling the foodborne zoonotic potential of M. avium given that the elimination of the organism was not achieved.

Hepatitis E Virus
One unexpected result was the absence of Hepatitis E virus-related papers retained for analysis, even after a relatively high number of papers were detected in the initial search (n = 77). This pathogen has been earning attention in the last few years. However, none of these papers fulfilled the criteria for inclusion in this systematic review. According to Meester et al. [75], pigs are the main reservoir of the HEV (genotypes 3 and 4) worldwide, and humans can become infected by consumption of pork or contact with pigs. As HEV is persistently present on most pig farms, current risk mitigation strategies should focus on lowering transmission within farms, especially between farm compartments. Vaccination of pigs may aid HEV control in the future [76].

Y. enterocolitica
Due to the exclusion criteria, studies on Y. enterocolitica were not retained. However, the risk assessment by EFSA [1] identified Y. enterocolitica as one of the most relevant biological hazards in the context of meat inspection of swine. Accordingly, this agent should be covered by preventive measures in the meat chain. At the farm level, some risk factors have been identified as contributors for seropositive herds, namely:
Use of non-municipal water sources and having a continuous production (instead of applying an all-in/all-out strategy) [79].
One study indicated that clusters (health and breeding pyramids) of pig herds free from animal diseases (Specific Pathogen Free (SPF) herds) also seem to be free from Y. enterocolitica [81]. Some of these SPF herds were even free from Campylobacter spp. [82].
However, there are several control options at the slaughterhouse [83]. However, after slaughter, control measures seem ineffective, since Y. enterocolitica can survive and grow during cold storage and under modified atmospheres [83,84].

T. gondii
The risk assessment by EFSA [1] also identified T. gondii as one of the most relevant biological hazards in the context of meat inspection of swine, but no studies of T. gondii were retained. Former studies show that the prevalence of T. gondii in pigs has decreased considerably in areas with intensive farm management [85,86]. However, pork originating from outdoor pig husbandry systems including those that are more welfare friendly such as free roaming, poses a higher risk compared to the indoor system [87], and this was not the focus of this systematic review. Other interesting approaches to interrupt the zoonotic circle of T. gondii are the vaccination of cats [88] or the control of the cat population in endemic regions [89]. So far, no commercial vaccine for cats is available.

Limitations of This Review
The results of this review and the implications inferred from them are valid within the context of the inclusion and exclusion criteria as defined a priori. This means that papers which did not contain a control group and from which a causal effect of the intervention tested could not be inferred were rejected. This decision was made to minimize bias and to eliminate confounding factors. However, observational studies are prevalent in the literature and the quality of the evidence provided by some of these studies should be graded up, provided that their results are robust. For example, identifying a strong correlation between high biosecurity and cleaning standards and low Salmonella shedding across multiple farms is a strong indication that such interventions are likely to be effective under the various scenarios of each farm. Conversely, randomized control trials reporting positive effects (P < 0.05) rarely declare the magnitude of this effect (i.e., adjusted Rsquared with the proportion of the variability explained by the factor tested in the outcome variable). Mapping and summarizing the risk factors for each foodborne pathogen and the pre-harvest interventions proposed to tackle them is a task yet to be undertaken.

Conclusions
Some foodborne pathogens appear to be best controlled at a post-harvest level. However, overall, high herd health status coupled with good management and biosecurity were effective to control or prevent most foodborne pathogens in pork at the pre-harvest level. In spite of not having been included in the review, the SPF herd principle, stamping out and repopulating with disease-free animals, has been reported as a feasible and effective intervention to control foodborne pathogens like Salmonella, Y. enterocolitica and MRSA.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/microorganisms9091825/s1, S1: PRISMA checklist for systematic reviews; S2: Detailed search strings used in each database; S3: List of studies included in the systematic review (n = 52); S4: List of studies excluded after full text evaluation (n = 35). Funding: This research received no external funding. This publication is based upon work from COST Ac-tion 18105 RIBMINS (www.ribmins.com) supported by COST (European Cooperation in Science and Technology; www.cost.eu), funded by the Horizon 2020 Framework Programme of the Eu-ropean Union. The publication of this article was funded by RIBMINS COST Action (CA18105).

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data generated and analysed during this study are included in this published article (and its Supplementary Information Files).

Conflicts of Interest:
The authors declare no conflict of interest.