Prevalence and risk factors for bacterial food-borne zoonotic hazards in slaughter pigs: a review.

The Hygiene Package and Regulation EC-2160/2003 require information flow from farm to slaughterhouse to enhance European consumers protection in a 'farm to fork' approach. This obligation especially concerns food-borne zoonotic hazards transmitted to humans through pork consumption, such as thermophilic Campylobacter spp., Listeria monocytogenes, Salmonella enterica and Yersinia enterocolitica. Prevalence estimates of these four hazards are affected by the sampling strategy and diagnostic procedure. Individual prevalence estimates for pig carriage (from digestive contents or lymph nodes collected at slaughterhouse) were higher than individual prevalence estimates for pig shedding (from faeces). Among risk factors described in the literature, poor pen cleaning and disinfection after pig departure to slaughterhouse and poor bio-security measures are of major significance. Moreover, whereas wet feed increases the risk of pig infection by L. monocytogenes, dry feed is a risk factor for Salm. enterica. Mixing batches of pigs, notably in fattening herds, represents a risk for the transmission of Salm. enterica and Y. enterocolitica. Whereas small herds are more infected by thermophilic campylobacters and Y. enterocolitica, higher prevalence of Salmonella is observed in large herds due to a more frequent mixing of batches. Antibiotic treatment during the finishing period increases the risk of transmission of Salm. enterica. The forenamed elements should be taken into account to characterize farms in a risk assessment approach and to improve zoonotic hazard management in the pork food chain.

Impacts

Food‐borne zoonoses are infectious diseases of major health and economic significance in developed countries. In order to protect consumers’ health and to enhance the management of food‐borne zoonotic agents from primary production to consumption, new regulations have been issued in the European Union. These Regulations notably require information concerning each step from farm to slaughterhouse. Today, pork is the most frequently consumed meat in Europe.

In this context, the purpose of this review was to collect information on risk factors on pig farms regarding the prevalence of four bacterial hazards responsible for frequent and/or serious pork‐borne diseases: Campylobacter spp., Listeria monocytogenes, Salmonella enterica and Yersinia enterocolitica. Among the risk factors described in the literature, feed, herd management and biosecurity measures have been shown to greatly impact the prevalence of these hazards. These risk factors may be used as information on the primary production of the pork food chain transmitted from farm to slaughterhouse.

The application of good hygiene practices in herds is paramount to reduce the risk of presence of food‐borne pathogens. As a priority in biosecurity measures, limiting the mixing of pig batches is needed. These measures must be implemented to reduce the presence of pathogens in the first step in the pork food chain.

Introduction

Pork is the most frequently consumed meat in the European Union (Devine, 2003) and the European pig herd is the second largest in the world after the Chinese herd (ITP, 2003). The management of hazards transmitted to humans by the consumption of pork is therefore of major health and economic significance. The European Commission issued in 2002 the General Food Law , a regulation whose main objective is to apply risk analysis to food safety legislation, from primary production to consumption, and in 2003 Regulation EC‐2160/2003 , whose purpose is the control of food‐borne pathogens in Europe. In this context, the management of hazards in the food chain necessitates the interpretation of scientific data on the characteristics and prevalence of hazards defined as ‘biological, chemical or physical agents in, or condition of, food or feed with the potential to cause an adverse health effect’ (Regulation EC‐178/2002, article 3–14). Biological hazards responsible for food‐borne zoonoses are of particular concern since their management on farms is possible notably by reducing their digestive carriage – i.e. the presence of hazards in digestive contents – and their faecal shedding (Blaha, 1999; Collins and Wall, 2004; Maunsell and Bolton, 2004; Humphrey and Jørgensen, 2006; Nørrung and Buncic, 2007).

Moreover, the Hygiene Package, which was issued in 2004 and inspired by the requirements of the General Food Law, defines the obligations of food business operators. This package, notably Regulation EC‐854/2004 , requires Member States to develop new meat inspection methods in order to improve consumer health protection. This could be based on the use of food chain information. Indeed, risk‐based meat inspection schemes, including information flow from farm to slaughterhouse, seem to be possible in an integrated system (Blaha, 1999; Schruff and Blaha, 2004). Since pig rearing and fattening techniques are standardized, particularly with homogenous market weight ranges and herd breeding methods, new ante mortem indicators of meat inspection could be developed from the integration of data collected on farms, as on‐farm prevalence of hazards or risk factors for hazards infection in slaughter pigs. This information may be used to identify high‐risk batches for which a thorough macroscopical examination of carcasses (notably to detect faecal contamination) or bacteriological analyses may be carried out. In a risk assessment approach, it is important to collect information on risk factors on farms for hazards which have the greatest impact on public health.

In Europe, twenty‐seven biological hazards may be transmitted from pork to consumers (Fosse et al., 2008a). Recent studies on the quantification of the informative value of meat inspection to detect biological hazards transmitted to humans by pork consumption have shown that high risk hazards can not be detected by a macroscopic examination of carcasses (Hamilton et al., 2002; Jelsma et al., 2006; Fosse et al., 2007). Among these hazards, Yersinia enterocolitica, Salmonella enterica, thermophilic Campylobacter spp. and Listeria monocytogenes are characterized by the highest scores of risk for pork consumers (Fosse et al., 2008a). Thermophilic Campylobacter are gram‐negative bacteria growing only within the narrow temperature range from 30 and 47°C (Stanier et al., 1986; Doyle, 1990). They are widely carried in poultry, pig and cattle digestive tracts, without clinical signs (Ono et al., 1995; Weijtens et al., 1999; Magras et al., 2005). Listeria monocytogenes is a psychrotroph gram‐positive bacterium with an optimal growth temperature of 30–37°C (Stanier et al., 1986; Bahk and Marth, 1990). Numerous studies have shown the asymptomatic carriage of L. monocytogenes by pigs (Skovgaard, 1990; Buncic, 1991; Iida et al., 1998). Salmonella enterica is a gram‐negative mesophilic bacterium with a 37°C optimal growth temperature (Stanier et al., 1986; Doyle and Cliver, 1990a). Salmonella infection is mainly subclinical in pigs, with rare septicaemia or enterocolitis reported (Barker and Van Dreumel, 1985). Yersinia enterocolitica is a psychrotrophic gram‐negative bacteria which can grow at temperatures as low as 0°C and as high as 44°C (Stanier et al., 1986; Doyle and Cliver, 1990b). Pathogenic strains of Y. enterocolitica belong to biotypes 1B, 2, 3, 4 and mainly serotypes O:3, O:5, O:8, O:9. They are asymptomatically carried in pigs in the digestive tract and in tonsils (Tauxe et al., 1987; Simonet and Catteau, 2005). Pork consumption was shown to be the main cause of human yersiniosis (Tauxe et al., 1987; De Boer and Nouws, 1991; Fredriksson‐Ahomaa et al., 2006).

These four bacterial hazards are carried by pigs without clinical signs, i.e. pig infection is characterized by non‐apparent digestive carriage – defined here as the presence of hazards in digestive contents other than rectal (gastric, caecal, ileal, colonic) and/or digestive tissues (digestive lymph nodes, tonsils, digestive epithelia) – and faecal shedding defined by the presence of hazards in faeces. Hazards detection by an ante mortem examination is thus not possible (Fosse et al., 2008a). Consequently, the identification of risk factors for food‐borne pathogens on a pig farm may be used to improve their control in the pork food chain and to complete hazard management related to meat inspection at slaughterhouses.

The purpose of this review was to sum up the information on prevalence and herd factors statistically linked with prevalence of pigs infected with four bacterial hazards for consumers in Europe: thermophilic Campylobacter spp., L. monocytogenes, Salm. enterica and Y. enterocolitica. Prevalences were summarized as: (i) prevalence linked with shedding (prevalence obtained from faeces and/or rectal contents collected on a farm or at a slaughterhouse, respectively); (ii) prevalence linked with digestive carriage (prevalence obtained from digestive contents other than rectal or digestive tissues collected at a slaughterhouse); (iii) serological prevalence (prevalence obtained from antibody detection in blood samples or meat juice). For serological prevalence, only data with the same threshold of detection were included. A summary of risk factors for on farm infection with the hazards was proposed in order to identify characteristics of pig herds which may be taken into account as food chain information from farm to slaughterhouse.

Material and Methods

Literature search methods

A review of the literature was carried out to collect: (i) data on prevalence of pig infections by Campylobacter spp., L. monocytogenes, Salm. enterica and Y. enterocolitica on farms, at the end of the fattening period, or upon entering a slaughterhouse; (ii) data on herd prevalence; (iii) information on sources of pig contamination; (iv) risk factors for presence and/or for higher prevalence of hazards on farms and pathogen transmission among pigs.

A literature search was conducted using the Commonwealth Abstract Bulletin (CAB) database and Medline for papers indexed since 1990 and the database of the four French National Veterinary Schools libraries for congress proceedings. Searches were performed in April, 2007, November, 2007 and March, 2008. The keyword combination used in the search was: campylobacter or listeria or salmonella or yersinia; and swine or pig; and herd or farm; and risk; or prevalence; in title and/or in abstracts. Searches were restricted to publications from January, 1990 onwards and languages were restricted to English and French. The papers taken into account had: (i) to be original articles; (ii) to report individual or herd prevalence of the food‐borne pathogens in indoor‐reared fattening pigs, at herd level or upon entering a slaughterhouse, or to define herd risk factors associated with hazards; (iii) to be carried out on samples pointing to faecal shedding, digestive carriage or serological prevalence. Prevalence data on piglets, sows, outdoor‐reared pigs or pork carcasses or retail pork were excluded. This search was systematically completed by looking in the reference lists of relevant papers.

The characteristics of study samples and design likely to influence the external validity of the results (sample size, method used for detection – with or without enrichment – individual or pooled analysis, consistency between material and methods and results obtained) were systematically checked and recorded twice by two abstractors, one epidemiologist and one bacteriologist. Data were coded by one abstractor and randomly assessed by a second one. Publications were systematically excluded when prevalence reported was also published by the same authors in other articles (publications or conference proceedings) in order not to repeat data and thus artificially add weight to some values. Besides, when inconsistencies were observed between prevalence or samples size mentionned in abstract, material and method, results or tables, we decided to take into account only data from tables. Cohort studies were also excluded when prevalence in finishing pigs could not be calculated. Therefore, from 236 papers quoted in CAB and Medline databases and analysed, only 106 papers were taken into account to estimate prevalence and summarize risk factors.

Prevalence estimates

For each study, individual or pool and herd apparent prevalence (p) were calculated using the number of positive units and the total sampled units reported in each study. A 95% confidence interval was calculated for each P‐value using the formula (Bouyer, 2000):

Reported apparent prevalence equal to 1 were systematically replaced by a 0.999 value to calculate 95% CI. For each hazard and for each type of material collected (faeces, digestive contents or lymph nodes, blood), median individual and herd prevalence were calculated. For Y. enterocolitica, prevalence was calculated only for pathogenic strains.

To calculate individual and herd prevalence summary estimates (p s), each apparent prevalence was logit‐transformed and the standard errors (σ) for logit prevalence were calculated as follows: and with p: apparent prevalence; n: sample size.

Reported apparent prevalence equal to 1 were systematically replaced by a 0.999 value. Then a summary estimate of prevalence (p s) was determined with its 95% confidence interval (95% CI) for each hazard and each type of material collected using the general variance‐based method described by Petitti (1994). The formulae for meta‐analysis were the following:

where p is the apparent prevalence of the i th study and w is the weight assigned to the i th study calculated from the standard error (σ ). Serological prevalence estimates were calculated from data obtained with the same optical density cut‐off. Individual prevalence estimates were calculated only from individual analyses. Herd prevalence estimates were calculated from individual and pooled analyses. In order to assess the relevance of the summary estimates and the heterogeneity of the samples studied, the Q parameter was calculated and compared with a χ2 distribution with a number of degrees of freedom equal to the number of studies minus 1 (Petitti, 1994). Q was calculated as:

Risk factor calculations and summarization

Our purpose was to identify risk factors significantly higher than one. Published risk factors with their 90% confidence intervals were thus collected, notably adjusted odds ratio (OR) when they were available, and when data were lacking, univariate ORs were calculated with their 90% IC as (Bouyer et al., 1995):

with (a) the number of exposed positive units; (b) the number of non exposed positive units; (c) the number of exposed negative units; (d) the number of non‐exposed negative units. Protective factors (OR <1) identified in papers where transformed into risk factors by calculating their reverse value. When 95% IC OR = [α; β] where mentionned in papers, 90% IC = [α′; β′] were assessed using the relation:

Risk factors were classified into four categories: biosecurity measures, feed and watering, herd management and health management. For each hazard and each risk factor, ORs were reported and/or calculated in order to identify relevant factors. According to those factors, characteristics of farms associated with were summarized in order to try to distinguish among herds which ones may be considered as high‐risk herds.

Thermophilic Campylobacter spp.

Prevalence on farms or at slaughterhouse

Campylobacter coli is the main species identified in pigs compared with C. jejuni or C. lari (Table 1). Prevalence of thermophilic Campylobacter spp. in finishing pig ranges according to the nature of the samples studied. As examples, the results of 17 studies in 11 European and North American countries are listed in Table 2. For Campylobacter shedding, an individual prevalence estimate of 0.655 was calculated versus 0.710 for digestive carriage (Table 3). Two studies carried out on stomach carriage showed an individual prevalence estimate of 0.515 (Nesbakken et al., 2003; Payot et al., 2004) whereas a prevalence estimate obtained from intestinal contents was higher (0.766). A prevalence estimate obtained from lymph nodes was very low (0.247). One study showed a serological prevalence value of 0.812 (Altrock et al., 2007). Herd prevalence of Campylobacter was 1.000 (16 values) except in one study carried out on gastric mucosa with a herd prevalence of 0.830 (Payot et al., 2004).

Table 1

 Distribution of Campylobacter species in finishing pigs reported in eight studies

Type of material sampled	References	Pigs tested	Herds tested	Type of analysis	Distribution of Campylobacter species (in %) among Campylobacter positive samples
					C. coli	C. jejuni	C. lari
Faeces collected on farm	Schuppers et al., 2005	1,280	64	Pool, Cul	96.3	1.2	–
	Varela et al., 2007a, 2007b	800	80	Ind, Cul	99.2	0.2	0.6
	Fosse et al., 2008b	215	6	Pool, Nb	95.3	4.7	0
Rectal contents collected at slaughterhouse	Minvielle et al., 2007	250	50	Ind, Nb	100	0	0
	Fosse et al., 2008b	430	6	Pool, Nb	100	0	0
Caecal contents collected at slaughterhouse	Steinhauserova et al., 2005	595	–	Ind, Cul	92.8	7.2	0
	Boes et al., 2005	1,244	247	Ind, Cul	97.5	2.5	0
	Harvey et al., 1999	595	4	Ind, Nb	65.7	33.9	0.4

Ind, individual analysis; Pool, pooled samples analysed; Cul, culture; Nb, culture with numbering; –, lack of data.

Table 2

 Prevalence of Campylobacter spp. in finishing pigs reported in 17 studies in 11 European and North‐American countries

Shedding or carriage	Country	References	Individual or pool prevalence			Pigs tested	Herd prevalence			Herds tested	Type of material sampled	Type of analysis
			p	p 95% CI			p	p 95% CI
				Lower limit	Upper limit			Lower limit	Upper limit
Shedding (prevalence in faeces or rectal contents)	Belgium	Botteldoorn et al., 2001	0.340	0.264	0.416	150	1.000	0.968	1.000	4	Rectal swabs collected on farm	Ind, Cul
	Canada	Varela et al., 2007a, 2007b	0.999	0.997	1.000	800	1.000	0.992	1.000	80	Fresh faeces (10 g) collected on farm	Ind, Cul
	France	Minvielle et al., 2007	1.000	0.995	1.000	250	1.000	0.990	1.000	50	Rectal contents (5 g) collected at slaughterhouse	Ind, Nb
		Fosse et al., 2008b	1.000	0.995	1.000	215	1.000	0.974	1.000	6	Faeces (25 g) collected on farm	Pool, Nb
			1.000	0.996	1.000	430	1.000	0.974	1.000	6	Rectal contents (25 g) collected at slaughterhouse	Pool, Nb
	Germany	Alter et al., 2005	0.791	0.747	0.835	330	1.000	0.977	1.000	8	Faeces (5 g) collected on farm	Ind, Cul
		Altrock et al., 2007	0.697	0.666	0.726	900	1.000	0.988	1.000	30	Faeces collected on farm	Ind, Cul
		Wehebrink et al., 2007	0.647	0.617	0.675	1,040	–	–	–	16	Faeces or rectal swabs collected on farm	Ind, Cul
	Italy	Cereser et al., 2007	0.114	0.040	0.188	70	1.000	–	–	1	Rectal (25 g) contents collected at slaughterhouse	Ind, Cul
	Norway	Nesbakken et al., 2003	1.000	0986	1.000	24	1.000	0.963	1.000	3	Rectal contents (10 g) collected at slaughterhouse	Ind, Cul
	Switzerland	Schuppers et al., 2005	0.957	0.944	0.967	1,280	1.000	0.991	1.000	64	Pooled faeces (5 × 5 g) collected on farm	Pool, Cul
	United States	Gebreyes et al., 2005	0.558	0.502	0.614	300	1.000	0.979	1.000	10	Fresh faeces (10 g) collected on farm	Ind, Cul
Digestive carriage (prevalence in digestive contents or tonsils and lymph nodes collected at slaughterhouse)	Czech Republic	Steinhauserova et al., 2005	0.491	0.451	0.531	595	–	–	–	–	Caecal contents	Ind, Cul
	Denmark	Boes et al., 2005	0.920	0.904	0.934	1,244	1.000	0.969	1.000	247	Caecal contents (100 ml)	Ind, Cul
	France	Payot et al., 2004	0.504	0.441	0.567	240	0.833	0.684	0.982	24	Gastric mucosa samples (9 cm2)	Ind, Cul
	Italy	Cereser et al., 2007	0.073	0.012	0.134	70	1.000	–	–	1	Mesenteric lymph nodes (25 g)	Ind, Cul
	Norway	Nesbakken et al., 2003	0.458	0.259	0.657						Tonsils (10 g)	Ind, Cul
			0.292	0.110	0.474						Mesenteric lymph nodes (5 g)
			0.625	0.431	0.819	24	–	–	–	–	Stomach contents (10 g)
			0.958	0.878	1.000						Ileal contents (10 g)
			0.958	0.878	1.000						Caecal contents (10 g)
			0.958	0.878	1.000						Colon contents (10 g)
	the Netherlands	Weijtens et al., 1993	0.850	0.772	0.928	80	1.000	0.977	1.000	8	Ileal contents (20 g)	Ind, Nb
	United States	Harvey et al., 1999	0.916	0.915	0.917	595	1.000	0.326	0.994	4	Caecal contents	Ind, Nb
Serological prevalence	Germany	Altrock et al., 2007	0.812	0.785	0.836	900	1.000	0.988	1.000	30	Blood samples	Ind, AD

P, apparent prevalence; Ind, individual analysis; Pool, pooled samples analysed; Cul, culture; AD, antibody detection; Nb, culture with numbering; –, lack of data.

Table 3

 Summary of prevalence of the four high‐risk food‐borne zoonotic hazards in finishing pigs reported in 80 studies

	Nature of sample tested or type of shedding	Individual prevalence									Herd prevalence
		Median	Min	Max	ps	ps 95% CI		Q	Pv	n	Median	Min	Max	ps	ps 95% CI		Q	Pv	n
						Lower limit	Upper limit								Lower limit	Upper limit
Campylobacter spp.	Shedding (faeces or rectal contents)	0.697	0.114	0.999	0.655	0.646	0.664	8.222	0.412	9	1.000	1.000	1.000	0.999	0.996	1.000	0.000	1.000	11
	Digestive carriage	0.738	0.073	0.958	0.710	0.699	0.720	19.222	0.057	12	0.999	0.830	1.000	0.877	0.809	0.923	0.011	1.000	5
	Gastric contents	0.565	0.504	0.625	0.515	0.484	0.545	0.075	0.784	2	0.830	–	–	0.830	0.739	0.794	–	–	1
	Intestinal contents	0.920	0.549	0.958	0.766	0.755	0.776	14.618	0.023	7	1.000	1.000	1.000	0.999	0.993	1.000	0.000	1.000	3
	Tonsils and digestive lymph nodes	0.292	0.073	0.458	0.247	0.203	0.297	0.418	0.811	3	1.000	–	–	0.999	–	–	–	–	1
	Serological prevalence	0.812	–	–	0.812	0.799	0.825	–	–	1	1.000	–	–	0.999	0.994	1.000	–	–	1
Listeria monocytogenes	Shedding (faeces or rectal contents)	0.009	0	0.017	0.017	0.009	0.030	0.000	1.000	2	0.140	–	–	0.140	0.108	0.180	–	–	1
	Digestive carriage:	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
	Gastric contents	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
	Intestinal contents	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
	Tonsils and digestive lymph nodes	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
	Serological prevalence	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
Salmonella enterica	Shedding (faeces or rectal contents)	0.039	0.007	0.340	0.062	0.060	0.064	8.300	0.600	11	0.416	0.101	0.999	0.218	0.211	0.225	14.214	0.942	25
	Digestive carriage:	0.145	0.018	0.547	0.155	0.153	0.157	84.818	0.000	29	0.554	0.028	0.999	0.234	0.227	0. 240	23.436	0.219	20
	Gastric contents	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
	Intestinal contents	0.202	0.018	0.547	0.246	0.242	0.251	35.028	0.001	14	0.500	0.028	0.999	0.190	0.183	0.197	8.304	0.599	11
	Tonsils and digestive lymph nodes	0.108	0.033	0.332	0.109	0.107	0.111	31.115	0.005	15	0.640	0.191	0.919	0.372	0.355	0.389	10.334	0.242	9
	Serological prevalence:	0.142	0.016	0.470	0.081	0.081	0.081	281.337	<0.001	12	0.773	0.047	0.951	0.124	0.120	0.128	62.064Δ	‐	15
	Blood samples (10% OD cut‐off)	0.313	0.156	0.470	0.367	0.361	0.373	28.376	<0.001	2	0.920	0.890	0.950	0.898	0.882	0.912	0.012	0.913	2
	Blood samples (25% OD cut‐off)	0.253	–	–	0.253	0.233	0.275	–	–	1	0.812	0.773	0.850	0.781	0.747	0.812	0.014	0.906	2
	Blood samples (30% OD cut‐off)	0.099	–	–	0.099	0.096	0.102	–	–	1	0.367	–	–	0.367	0.349	0.386	–	–	1
	Blood samples (40% OD cut‐off)	–	–	–	–	–	–	–	–	–	0.230	–	–	0.230	0.190	0.276	–	–	1
	Meat juice samples (20%   OD cut‐off)	0.082	–	–	0.082	0.082	0.082	–	–	1	0.833	–	–	0833	0.625	0.937	–	–	1
	Meat juice samples (25%   OD cut‐off)	0.274	0.115	0.432	0.296	0.279	0.313	3.622	0.057	2	0.900	–	–	0.900	0.810	0.950	–	–	1
	Meat juice samples (30%   OD cut‐off)	0.283	–	–	0.283	0.278	0.288	–	–	1	0.450	–	–	0.450	0.400	0.501	–	–	1
	Meat juice samples (40%   OD cut‐off)	0.062	0.016	0.152	0.055	0.054	0.055	3.159	0.368	4	0.210	0.047	0.667	0.049	0.047	0.051	0.911	0.823	4
Pathogenic strains of Yersinia enterocolitica	Shedding (faeces or rectal contents)	0.125	0.005	0.196	0.122	0.118	0.125	0.672	0.999	8	0.666	0.357	0.999	0.599	0.557	0.640	0.684	0.998	8
	Digestive carriage:	0.147	0.038	0.625	0.194	0.189	0.199	11.156	0.800	17	0.584	0.122	0.999	0.269	0.245	0.295	2.510	0.999	14
	Gastric contents	0.042	–	–	0.042	0.016	0.108	–	–	1	0.333	–	–	0.333	0.128	0.630	–	–	1
	Intestinal contents	0.125	0.040	0.173	0.165	0.156	0.175	0.119	0.998	5	0.500	0.122	0.667	0.141	0.119	0.167	0.402	0.940	4
	Tonsils and digestive lymph nodes	0.219	0.038	0.625	0.203	0.197	0.209	10.542	0.482	12	0.585	0.250	0.999	0.396	0.357	0.436	1.139	0.999	10
	Serological prevalence	0.668	0.541	0.875	0.699	0.325	0.918	0.007	0.997	3	0.883	0.640	0.999	0.653	0.627	0.678	0.148	0.929	3

Min, minimal value; Max, maximal value; ps, prevalence estimate; n, number of values per category; ‐, lack of data; Pv, P‐value for the calculated Q parameter in comparison with a χ2 distribution; OD, optical density.

Sources of infection

A Dutch study showed that parturition enhances faecal shedding of campylobacters in sows with an early contamination of piglets, during the first week of life (Weijtens et al., 1997). Restriction Fragment Length Polymorphism (RFLP)‐typing of the isolates collected from sows and piglets showed strong genomic homology between isolates and suggested transmission from sows to offspring (Magras et al., 2004; Laroche et al., 2007). Further studies have shown a higher diversity of strains isolated from finishing pigs than from piglets, suggesting infection of pigs during the fattening period by strains present in pens or intercontamination (Weijtens et al., 1999; Laroche et al., 2007). Thus, piglets are infected at a young age and spread after weaning. The implication of environmental contamination was shown by Weijtens et al. (2000) with the repopulation of an infected farm after an ‘empty and clean’ period with specific pathogen free pigs, i.e. pigs which have not been contaminated by sows (Weijtens et al., 2000). In this herd, the individual prevalence of faecal shedding was 0.22 versus 0.98 in the control herd, and this prevalence was observed during the whole 20‐month long study. Introduction of infection to pigs through vehicular and vectors such as boots and clothes or by contact to rodents and birds was suggested. However, whereas numerous studies have shown the substantial risk of campylobacters infection by contaminated water and feed in poultry (Pearson et al., 1993; Byrd et al., 2001), this relationship has not been shown for pigs (Magras et al., 2004; Leblanc‐Maridor et al., 2008).

Risk factors for shedding and transmission

Few publications have identified risk factors for infection with thermophilic campylobacters in pigs. One recent study carried out by Wehebrink et al. (2007) on 12 fattening farms showed that the prevalence of Campylobacter spp. was significantly lower on farms with over 1,000 fattening facilities than on farms with under 1,000 pig fattening facilities (prevalence of 0.743 and 0.800, respectively). Thus, management factors correlated with herd size may have an influence on the occurrence of campylobacters, with higher prevalences in smaller herds. Moreover, in the same study, the following risk factors were reported: housing pigs in separate stalls had a preventive effect, similar to antibiotic treatment at the beginning of the fattening period, whereas anthelmintic treatment seemed to increase the risk of detection of campylobacters (Wehebrink et al., 2007).

Listeria monocytogenes

Prevalence on farms or at slaughterhouses

In Denmark, Skovgaard and Nørrung (1989) showed L. monocytogenes faecal shedding prevalence of 0.017 from 172 pigs collected at slaughterhouses. A French study carried out by Beloeil et al. (2003a) showed the presence of L. monocytogenes in 14% of pig batches (n = 93) with a detection of bacteria on pooled perianal swabs after the enrichment step. Cereser et al. (2007) did not show the presence of L. monocytogenes in the rectal content of 70 pigs collected at slaughterhouses. Nevertheless, the method used was the numbering ISO 11290‐1 method, without a preliminary enrichment step and thus a lower threshold of detection. In a Japanese study carried out on caecal contents of 250 pigs from the same herd collected at a slaughterhouse, digestive carriage prevalence of 0.3 was observed (Yokoyama et al., 2005).

The main source of pig contamination by L. monocytogenes described is feed. Indeed, a few studies have shown that wet feed is associated with a higher prevalence of shedding than dry feed (Skovgaard and Nørrung, 1989; Beloeil et al., 2003a). It may be explained by the more frequent presence of L. monocytogenes in wet feed than in dry feed (Beloeil et al., 2003b) due to the contamination of pipelines used for the distribution of wet feed. The modification of pig digestive bacterial flora due to wet feed with development of L. monocytogenes may be another explanation. Environmental contamination may also be possible, because of the telluric origin of L. monocytogenes (Beloeil et al., 2003a).

Wet feed and inefficient biosecurity measures are described as risk factors regarding the presence of L. monocytogenes (Table 4). Pipeline cleaning and disinfection is notably associated with a higher prevalence of Listeria shedding (Beloeil et al., 2003a). Indeed, disinfection may destroy bacterial pipeline biofilm which may inhibit the development of L. monocytogenes (Royer et al., 2004).

Table 4

 Summary of risk factors for Yersinia enterocolitica shedding on pig farms reported in two studies.

	Risk factor of presence of Yersinia enterocolitica		Reference	OR [90%CI]
Biosecurity measures	Ventilation	Lack of under‐pressure ventilation	Skjerve et al., 1998	3.0* [1.5–6.2]
	Presence of domestic animals	Daily observation of cats with kittens	Skjerve et al., 1998	2.4 [1.3–4.6]
Feed	Water and feed	Detection of Yersinia enterocolitica in water or feed	Pilon et al., 2000	NE
	Distribution of feed	Manual distribution of feed correlated to a small herd size (<1000 pigs)	Skjerve et al., 1998	2.3* [1.3–3.9]
Herd management	Litter	Straw on floor	Skjerve et al., 1998	2.3 [1.2–4.3]
	Herd type	Mix of pigs from different origins in fattening herds versus farrow‐to‐finish herds	Skjerve et al., 1998	6.7* [3.5–12.7]
	Transport	Carriage of pigs to slaughterhouse in personnal vehicles (correlated to a small herd size)	Skjerve et al., 1998	12.9 [2.2–74.2]

NE: not possible to estimate.

*Calculated OR from published data.

Salmonella enterica

Prevalence on farm or at slaughterhouse

A meta‐analysis carried out from 98 references showed the influence of sampling design and a diagnostic test used on prevalences published (Sanchez et al., 2007). Using pooled faecal samples for bacteriological detection of Salmonella proved to be more sensitive than individual detection (Arnold et al., 2006). Thus, individual prevalence in Europe and North America showed great differences depending on the nature and the volume of samples analysed and the type of analysis. As examples, the results of 46 studies in 15 European and North American countries are listed in Table 5. An individual prevalence estimate of 0.062 was calculated for Salmonella shedding with a herd prevalence estimate of 0.218 (Table 3). Salmonella enterica prevalence in digestive lymph nodes was estimated at 0.109 versus 0.242 in digestive contents collected at slaughterhouses. An individual serological prevalence estimate of 0.081 was obtained. Individual serological prevalence estimates ranged from 0.099 to 0.367 for blood samples according to the optical density cut‐off. Such a variation is observed for meat juice (from 0.055 to 0.296). A serological herd prevalence estimate of 0.124 was calculated (from 0.773 to 0.950 for blood samples and from 0.047 to 0. 667 for meat juice, according to optical density cut‐off). Inconsistently, farm bacteriological data showed a mean herd prevalence of 0.209.

Table 5

 Summary of risk factors for Listeria monocytogenes shedding on pig farms reported in two studies

	Risk factor of presence of Listeria monocytogenes		OR [90%CI]	Reference
Biosecurity	Change room	Lack of change room at the entrance of facilities	7.7 [1.2–49.7]	Beloeil et al., 2003a
	Boot disinfection	Frequency of boot disinfection inferior to once a week	4.7 [1.4–15.6]	Beloeil et al., 2003a
	‘Empty and clean’ period	An ‘empty and clean’ period of one day or less between two fattening batches	3.5 [1.4–11.8]	Beloeil et al., 2003a
Feed	Wet feed	L. monocytogenes was isolated from 19.3% of the 57 wet feed batches studied versus 5.9% of the 36 dry feed batches	4.4 [1.1–17.2]	Beloeil et al., 2003a
		Listeria spp. was isolated from 93% of the 27 wet fed pens studied versus 50% of 20 dry fed pens	12.5* [7.0–22.4]	Beloeil et al., 2003b
	Cleaning of pipelines used for wet feed	Pipeline cleaning and disinfection	8.4* [4.7–15.1]	Beloeil et al., 2003a

*Calculated OR from published data.

Numerous studies have shown an increase in prevalence from farm to slaughterhouse (Craven and Hurst, 1982; Warris, 1992; Mulder, 1995; Hurd et al., 2004; Fosse et al., 2008b) notably explained by the impact of transport and the lairage period. Stress during transport may enhance shedding of Salmonella by non‐apparent carriers and then the infection of trucks or interinfection of pigs during lairage (Fravalo et al., 1999).

An increase in Salmonella shedding at weaning was observed in sows (Nollet et al., 2005) and in piglets, notably due to feed transition and a decrease in sow colostral antibodies (Kranker et al., 2003). A progressive increase in Salmonella shedding was also suggested in one American study involving a cohort of finishing pigs (Davies et al., 1999). Transmission of Salmonella to pigs through contaminated feed or environment was described (Hurd et al., 2001; Langvad et al., 2006). A study carried out by Fablet et al. (2003b) showed a close link between residual contamination of fattening rooms after a ‘clean and empty’ period and the level of infection of fattening pigs before slaughtering. The rapid infection of pigs from 2 to 3 h after contact with Salmonella was also described (Hurd et al., 2001). Contact with persons, contaminated slurry or sharing contaminated equipment were also proven to be risk factors for the transmission of Salmonella between pigs herds and from cattle to pigs herds (Langvad et al., 2006). Salmonella enterica was also isolated from rodents in pig herds (Le Moine et al., 1987).

Moreover, Fablet et al. (2003b) showed that dry feed enhanced the risk of Salmonella shedding compared to wet feed. The acidification of intestinal content inhibiting the development of Salmonella due to wet feed is an explanation (Fablet et al., 2003b; Royer et al., 2004). The direct infection of pigs through contaminated feed may be another explanation. A Canadian study showed the presence of Salmonella in 25 of the 420 (0.059) dry feed samples versus 3 of the 400 (0.008) wet ones (Friendship et al., 2006). This study also showed the presence of Salmonella in 38% of the 21 herds using dry feed versus 15% of the 20 herds using wet feed. Nevertheless, an American study carried out by Funk et al. (2001a) showed that only 2 out of 800 feed samples tested were contaminated by Salmonella (0.003).

Among all factors explaining Salmonella status in pig herds, Fablet et al. (2003a) showed in a study carried out in 105 herds that hygiene measures in farrowing rooms had an impact on Salmonella occurrence. Indeed, not emptying pits below floors of farrowing rooms after the removal of previous batches of sows and the removal of manure less than twice a day were associated with higher Salmonella shedding at the end of the fattening period. Studies highlighted the implication of infection status of sows on Salmonella infection in fattening pigs (Kranker et al., 2001; Lurette et al., 2007). Besides, a serological study carried out by Merialdi et al. (2007) showed a high level of seropositivity in sows (from 0.938 to 1) and a progressive increase of seropositivity in pigs during farrowing, post‐weaning and fattening periods. The infection of pigs at the beginning of the fattening period was suggested by this study. Moreover, a variation in individual susceptibility was suggested to explain differences in levels of infection between herds (Kranker et al., 2003).

Among all the risk factors of Salm. enterica shedding on farms described in the literature, dry feed and lack of biosecurity measures are mainly reported (Table 6). A Danish study showed that Salm. enterica may increase the risk of diarrhoea in pigs (Møller et al., 1998). Thus, digestive clinical signs may be considered risk factors for the presence of Salmonella in pigs. Preventive antibiotic treatment during the fattening period is also described to enhance the risk of Salmonella shedding (Rossel et al., 2006). Nevertheless, on the contrary, an American study showed a higher prevalence of Salmonella in antimicrobial‐free production systems than in conventional systems (Gebreyes et al., 2006). Variations of antibiotic doses used for therapy or as probiotics may explain such differences.

Table 6

 Prevalence of Salmonella enterica in finishing pigs reported in 46 studies in 15 European and North American countries

Shedding or carriage	Country	Reference	Individual or pool prevalence			Pigs tested	Herd prevalence			Herds tested	Type of material sampled	Type of analysis
			p	p 95% CI			p	p 95% CI
				Lower limit	Upper limit			Lower limit	Upper limit
Shedding (prevalence in faeces or rectal contents)	Belgium	Botteldoorn et al., 2001	0.070	0.036	0.104	215	0.250	0.000	0.674	4	Rectal swabs collected on farm	Ind, Cul
	Canada	Rajic et al., 2005	0.143	0.124	0.162	1,344	0.511	0.408	0.614	90	Faeces collected (5 × 5 g) on farm on floors	Pool, Cul
		Farzan et al., 2006	0.034	0.022	0.046	820	0.136	0.061	0.211	81	Faeces (25 g) collected on farm	Ind, Cul
	Denmark	Stege et al., 2000	–	–	–	–	0.101	0.035	0.167	79	Faeces (5 × 5 g) collected on farm	Pool, Cul
	France	Fablet et al., 2003a	–	–	–	–	0.362	0.270	0.454	105	Faeces collected on fattening rooms floors by swabbing (10 m2)	Pool, Cul
		Fablet et al., 2007	–	–	–	–	0.189	0.093	0.285	64
		Fosse et al., 2008b	0.023	0.003	0.043	215	0.167	0.000	0.465	6	Faeces (25 g) collected on farm	Pool, Cul
			0.314	0.270	0.358	430	0.833	0.535	1.000	6	Rectal contents (25 g) collected at slaughterhouse	Pool, Cul
	Germany	Steinbach et al., 2002	0.037	0.034	0.040	11,960	0.200	0.170	0.230	679	Rectal contents collected at slaughterhouse	Ind, Cul
	Hungary	Biksi et al., 2007	0.215	0.181	0.249	558	0.548	0.373	0.723	31	Faeces collected on farm	Pool, Cul
	Ireland	Rowe et al., 2003	–	–	–	–	0.508	0.380	0.636	59	Faeces (25 g) collected on farm	Pool, Cul
		Duggan et al., 2007	0.340	0.263	0.417	147	0.889	0.684	1.000	9	Rectal contents (25 g) collected at slaughterhouse.	Ind, Cul
	Italy	Cereser et al., 2007	0.028	0.000	0.067	70	1.000	–	–	1	Rectal contents (25 g) collected at slaughterhouse	Ind, Cul
		Cibin et al., 2007	0.276	0.231	0.321	381	–	–	–	–	Faeces (5 g) collected at slaughterhouse	Ind, Cul
	Spain	Mejía et al., 2006	0.021	0.015	0.027	2,187	0.204	0.130	0.278	113	Pooled faeces (25 g) collected on floor on farm	Pool, Cul
		Garcia‐Feliz et al., 2007	0.125	0.112	0.138	2,320	0.431	0.367	0.495	232	Faeces (pools of 5 × 5 g) collected on floor on farm	Pool, Cul
	Denmark, Germany, Greece, The Netherlands	Lo Fo Wong et al., 2003	0.093	0.078	0.108	1,455	0.416	0.306	0.526	77	Pen faecal samples (5 × 5 g) collected on farm	Pool, Cul
	United‐Kingdom	Davies et al., 2003	0.328	0.280	0.376	369	0.900	0.769	1.000	20	Faeces (30 g) collected on floor	Pool, Cul
	United‐States	Barber et al., 2002	–	–	–	720	0.750	0.505	0.995	12	Faeces (1 g) collected on farm	Ind, Cul
		Hurd et al., 2004	0.011	0.000	0.023	280	0.667	0.290	1.000	6	Faeces (1 g) collected on farm	Ind, Cul
			0.007	0.000	0.017	281	0.167	0.000	0.465	6	Rectal contents (1 g) collected on farm	Ind, Cul
			0.252	0.202	0.302	286	1.000	0.974	1.000	6	Faeces (1 g) collected at slaughterhouse	Ind, Cul
		Bahnson et al., 2006b	0.049	0.035	0.063	934	0.159	0.069	0.249	63	Faeces collected on farm	Ind, Cul
			0.039	0.027	0.051	937	0.222	0.119	0.325	63	Rectal contents (20 g) collected at slaughterhouse	Ind, Cul
Digestive carriage (prevalence in digestive contents or tonsils and lymph nodes collected at slaughterhouse)	Belgium	Korsak et al., 2003	0.473	0.401	0.545	186	–	–	–	13	Colon contents (25 g)	Pool, Cul
		Nollet et al., 2004	0.332	0.311	0.353	1,860	0.919	0.851	0.987	62	Gut lymph nodes (10 g)	Ind, Cul
	Canada	Letellier et al., 1999a	0.052	0.040	0.064	1,420	0.028	0.006	0.050	223	Caecal contents (1 g)	Ind, Cul
	Denmark	Baggesen et al.,1996	0.062	0.058	0.066	13,468	0.222	0.200	0.244	1,363	Caecal contents (5 g)	Pool, Cul
		Christensen et al., 2002	–	–	–	–	0.114	0.100	0.128	1,962	Caecal contents (25 g)	Ind, Cul
		Sørensen et al., 2004	0.130	0.114	0.146	1,658	–	–	–	–	Caecal contents (25 g)	Ind, Cul
			0.108	0.093	0.123	1,666	–	–	–	–	Caecal lymph nodes (25 g)
	France	Beloeil et al., 2004b	0.248	0.222	0.274	1,030	0.703	0.614	0.792	101	Caecal contents (25 g)	Ind, Cul
	Germany	Käsbohrer et al., 2000; Steinbach et al., 2002	0.033	0.030	0.036	11,960	0.191	0.161	0.221	679	Gut lymph nodes	Ind, Cul
		Nowak et al., 2007	0.165	0.128	0.202	393	0.344	0.179	0.509	32	Jejunum lymph nodes (10 g)	Ind, PCR
			0.155	0.093	0.217	129	–	–	–	–	Tonsils	Ind, Cul
	Ireland	Quirke et al., 2001	0.146	0.112	0.180	419	0.500	0.359	0.641	48	Caecal contents (1 g)	Ind, Cul
		Duggan et al., 2007	0.469	0.388	0.550	147	0.889	0.684	1.094	9	Caecal contents (25 g)	Ind, Cul
	Italy	Bonardi et al., 2003	0.367	0.290	0.444	150	–	–	–	–	Caecal contents (10 g)	Ind, Cul
		Cibin et al., 2007	–	–	–	1,557	0.607	0.514	0.700	107*	Mesenteric lymph nodes (5 g)	Ind, Cul
	Portugal	Vieira‐Pinto et al., 2005	0.139	0.072	0.206	101	–	–	–	–	Ileum contents (25 g)	Ind, Cul
			0.188	0.112	0.264	101	–	–	–	–	Ileocolic lymph nodes (25 g)	Ind, Cul
			0.129	0.064	0.194	101	–	–	–	–	Mandibular lymph nodes (10 g)	Ind, Cul
			0.099	0.041	0.157	101	–	–	–	–	Tonsils (10 g)	Ind, Cul
	United‐Kingdom	Davies et al., 2003	0.243	0.202	0.284	420	0.800	0.625	0.975	20	Caecal contents (25 g)	Ind, Cul
		Davies et al., 2004	0.230	0.214	0.246	2,509	–	–	–	–	Caecal contents (25 g)	Ind, Cul
		Mac Dowell et al., 2007	0.314	0.274	0.354	513	–	–	–	–	Caecal contents (50 g)	Ind, Cul
Digestive carriage (prevalence in digestive contents or tonsils and lymph nodes)	United‐States	Carlson and Blaha, 2001	0.037	0.031	0.043	3,442	0.640	0.452	0.828	25	Ileocaecal lymph nodes	Ind, Cul
		Hurd et al., 2004	0.018	0.002	0.034	281	0.333	0.000	0.710	6	Caecal contents (30 ml) collected on farm	Ind, Cul
			0.136	0.096	0.176	286	1.000	0.974	1.000	6	Caecal contents (30 ml) collected at slaughterhouse	Ind, Cul
			0.036	0.014	0.058	281	0.833	0.535	1.000	6	Ileocaecal (5 g) and inguinal (10 g) lymph nodes collected on farm	Ind, Cul
			0.091	0.058	0.124	286	0.833	0.535	1.000	6	Ileocaecal (5 g) and inguinal (10 g) lymph nodes collected at slaughterhouse	Ind, Cul
		Bahnson et al., 2006a	–	–	–	300	0.500	0.190	0.810	10	Ileocaecal lymph nodes (5 × 0.5 g)	Pool, Cul
		Bahnson et al., 2006b	0.174	0.150	0.198	942	0.635	0.516	0.754	63	Caecal contents (20 g)	Ind, Cul
			0.145	0.123	0.167	941	0.492	0.369	0.615	63	Ileocolic lymph nodes (10 g)	Ind, Cul
		Bahnson et al., 2006c	0.070**	0.062	0.078	4,380	0.685	0.610	0.760	146	Ileocolic lymph nodes (5 × 4 g)	Pool, Cul
		Bahnson et al., 2007	0.084**	0.077	0.091	6,330	–	–	–	211	Ileo‐colic lymph nodes (10 g)	Pool, Cul
		Rostagno et al., 2007	0.279	0.253	0.305	1,110	–	–	–	33	Mesenteric lymph nodes	Ind, Cul
			0.547	0.518	0.576	1,110	–	–	–	33	Caecal contents	Ind, Cul
Serological prevalence	Canada	Rajić et al., 2007	0.132	0.119	0.145	2,700	0.833	0.756	0.910	90	Blood collected at slaughterhouse	Ind, AD
	Denmark	Mousing et al., 1997	0.054	0.053	0.055	604,006	0.047	0.043	0.051	13,036	Meat juice (40% OD cut‐off)	Ind, AD
		Christensen et al., 1999	0.283	0.274	0.292	9,654		–	–	1,248	Meat juice (30% OD cut‐off)	Ind, AD
		Stege et al., 2000	–	–	–	4,800	0.230	0.146	0.314	96	Blood collected at slaughterhouse (40% OD cut‐off)	Pool, AD
			–	–	–	–	0.450	0.350	0.550	96	Meat juice (30% OD cut‐off)	Pool, AD
		Benschop et al., 2008	0.082	0.082	0.082	6,768,845	–	–	–	22,344	Meat juice (20% OD cut‐off)	Ind, AD
	Germany	Czerny et al., 2001	0.016	0.012	0.020	3,048	0.231	0.116	0.346	52	Meat juice (40% OD cut‐off)	Ind, AD
		Steinbach et al., 2002	0.099	0.094	0.104	11,896	0.367	0.331	0.403	679	Blood collected at slaughterhouse (30% OD cut‐off)	Ind, AD
		Nowak et al., 2007	0.070	0.044	0.096	383	0.188	0.053	0.323	32	Meat juice (40% OD cut‐off)	Ind, AD
	Greece	Leontides et al., 2003	0.156	0.143	0.169	2,950	0.880	0.797	0.963	59	Blood collected at slaughterhouse and on farm (10% OD cut‐off)	Ind, AD
	Spain	Mejía et al., 2006	–	–	–	–	0.773	0.704	0.842	141	Blood `collected at slaughterhouse (25% OD cut‐off)	Pool, AD
	The Netherlands	Van der Wolf et al., 2001	–	–	–	–	0.890	0.857	0.923	355	Blood (at least 40 samples per herd) collected at slaughterhouse (10% OD cut‐off)	Pool, AD
	Denmark, Germany, Greece, The Netherlands	Lo Fo Wong et al., 2003	0.470	0.455	0.485	4,194	–	–	–	77	Blood (10% OD cut‐off)	Ind, AD
	United‐Kingdom	Davies et al., 2003	0.253	0.212	0.294	430	0.850	0.694	1.000	20	Blood (25% OD cut‐off)	Ind, AD
			0.432	0.385	0.479	421	0.900	0.769	1.000	20	Meat juice (25% OD cut‐off)	Ind, AD
		Davies et al., 2004	0.152	0.138	0.166	2,509	–	–	–	–	Meat juice (40% OD cut‐off)	Ind, AD
		Mac Dowell et al., 2007	0.115	0.087	0.143	513	–	–	–	–	Meat juice (25% OD cut‐off)	Ind, AD
	United‐States	Hurd et al., 2004	–	–	–	557	0.833	0.535	1.000	6	Meat juice (20% OD cut‐off)	Pool, AD
			–	–	–	557	0.667	0.290	1.000	6	Meat juice (40% OD cut‐off)	Pool, AD
		Farzan et al., 2006	–	–	–	–	0.950	0.902	0.998	80	Blood (10% OD cut‐off)	Pool, AD

P, apparent prevalence; Ind, individual analysis; Pool, pooled samples analysed; Cul, culture; Nb, culture with numbering; AD, antibody detection; OD, optical density; –, lack of data.

*Number of batches.

**Individual estimated prevalence from regression.

Yersinia enterocolitica

Prevalence on farms and at slaughterhouses

Prevalence figures in pigs greatly depend on the type of material sampled and the diagnostic test applied (Table 7). A herd prevalence estimate for Y. enterocolitica shedding was 0.599 whereas an individual prevalence estimate was 0.194 (Table 3). On intestinal contents samples, individual and herd prevalence estimates were 0.165 and 0.141, respectively. Tonsil samples showed prevalence ranging from 0.147 to 0.625 with a prevalence estimate of 0.324 whereas prevalence reported from other lymph nodes were lower (0.038–0.052; three values) which shows the extent of tonsil contamination. Serological herd prevalence ranged from 0.640 to 1.000 and serological individual prevalence range from 0.541 to 0.875. Differences between serological and faecal bacteriological prevalence suggest an intermittent faecal shedding in pigs (Altrock et al., 2007).

Table 7

 Summary of risk factors for Salmonella enterica shedding on pig farms reported in 23 studies

	Risk factor of presence of Salmonella enterica		OR [90%CI]	Reference
Biosecurity measures and equipment	Cleaning measures and ‘empty and clean’ period Hygiene and clothes	Frequency of sow manure removal in farrowing rooms during the lacation period lower than once a day	2.9 [1.2–6.7]	Beloeil et al., 2004a
		Lack of emptying pits below slatted floors after removal of previous sows batches	2.6 [1.1–6.4]	Fablet et al., 2003a; Beloeil et al., 2004a
		Disinfection of rooms (without preliminary cleaning) is associated with a higher serological prevalence	1.4* [1.1–1.9]	Van der Wolf et al., 2001
		Washing room without disinfection (risk factor concerning all enteric pathogens in swine)	3.3* [1.1–9.7]	Pearce, 1999
		Residual Salmonella contamination of the room before loading of the batch followed	3.1 [1.4–7.1]	Beloeil et al., 2004a
		Duration of ‘empty and clean’ period lower than 6 days in farrowing room	3.1* [1.7–5.5]	Beloeil et al., 2004a
		Duration of ‘empty and clean’ period lower than 7 days in post‐weaning	3.2 [1.3–8.2]	Fablet et al., 2003a
		Duration of ‘empty and clean’ period lower than 3 days in fattening room	2.0* [1.1–3.5]	Beloeil et al., 2004a
		Detection of Salmonella on boots or environmental samples and/or lack of boot‐dip at the entrance of the facilities	NE	Letellier et al., 1999b;Rajic et al., 2005
			2.1* [1.2–3.8]	Beloeil et al., 2004a
		Lack of hand hygiene before tending to pigs; lack of toilet	11.1* [1.8–70.2]	Funk et al., 2001b
			1.5* [1.1–2.1]	Lo Fo Wong et al., 2004
	Infection through people or equipment	More than two humans at finisher daily	4.8* [1.4–17.1]	Funk et al., 2001b
		Allowing visitors with same day contact to other herds	2.2 [1.3–3.9]	Bahnson et al., 2007
		Sharing equipment	NE†	Langvad et al., 2006
	Floors	Solid floors or straw on floors versus slatted floor	1.5 [1.4–1.6]	Rossel et al., 2006
			NE‡	Hill et al., 2007
		Partially slatted floor versus fully slatted	8.9* [5.0–15.9]	Nollet et al., 2004
	Pen separation	Pierced pen separation between batches or possibilities of snout contacts	1.7 [1.1–2.6]	Lo Fo Wong et al., 2004
Feed and watering	Acidification or fermented liquid feed	No acidification of feed or water	1.4* [1.3–1.6]	Rossel et al., 2006
		Use of dry or liquid feed in comparison with fermented liquid feed	5.0* [2.8–8.9]	Van der Wolf et al., 2001
	Dry feed	Numerous studies have showed higher bacteriological or serological prevalences in pigs herds using dry feed versus herds using wet feed	1.1* [0.2–6.8]	Kranker et al., 2001
			3.2 [1.4–7.1]	Beloeil et al., 2004a
			4.9 [1.9–12.7]	Bahnson et al., 2006b
			4.1* [1.4–12.2]	Farzan et al., 2006
	Pelleted feed	Serological prevalence is higher in pigs fed with pelleted ration versus wet or dry non‐pelleted ration	4.2* [2.0–8.7]	Kranker et al., 2001
			12.5* [2.2–71.6]	Leontides et al., 2003
			2.4* [1.5–3.7]	Lo Fo Wong et al., 2004
			2.1 [1.2–3.8]	Bahnson et al., 2007
	No. feeds	Distribution of more than two kinds of feeds between post‐weaning and fattening period	NE§	Fablet et al., 2003a
		Change in the feed diet during the follow‐up	3.4 * [1.9–6.1]	Beloeil et al., 2004a
	Drinker design	Bowl drinkers are associated with higher prevalence than nipple drinkers	8.0 [3.4–19.0]	Bahnson et al., 2006b
Herd management	Infection of sows	Infection of sow herds is associated with a higher serological prevalence in finishers	3.2* [1.6–6.5]	Kranker et al., 2001
	Herd size	Risk of salmonella shedding seems to be associated with a higher herd size	NE	Baggesen et al., 1996; Carstensen and Christensen, 1998
			2.0 [1.3–3.0]	Bahnson et al., 2007
			1.3* [1.0–1.7]	Kranker et al., 2001
	Herd type	Post‐weaning to fattening herds and fattening herds were more contaminated than farrow‐to‐finish herds	1.6 [1.4–1.8]	Rossel et al., 2006
			NE‡	Clough et al., 2007
	Stocking density	Space allowance inferior to 0.75m2 per pig	4.5* [1.3–15.7]	Funk et al., 2001b
	Gain of weight during fattening	Link between 10 kg gain of weight and Salmonella shedding	1.2 [1.0–1.5]	Bahnson et al., 2007
	Mixing batches	Mixing batches during the fattening period	1.5 [1.4–1.6]	Rossel et al., 2006
		Continuous production of pigs compared to all‐in/all‐out	3.7* [2.1–6.4]	Lo Fo Wong et al., 2004
			3.9* [1.4–10.5]	Farzan et al., 2006
	Origin of pigs	Recruitment of pigs from more than 3 different supplier herds	3.3 [1.8–6.0]	Lo Fo Wong et al., 2004
	Other breedings and contacts with domestic species or wild animals	Poultry breeding on the farm	1.2 [1.1–1.3]	Rossel et al., 2006
		Other domestic species at the site or indirect contacts with other herds	4.7* [1.2–18.0]	Funk et al., 2001b
			NE	Langvad et al., 2006
		Contact with rodents	NE	Le Moine et al., 1987; Letellier et al., 1999b
Health management	Antibiotic	Curative antibiotic treatment at the end of the fattening period	NE§	Fablet et al., 2003b
		Preventive antibiotic treatment during fattening enhances serological prevalence	1.5 [1.4–1.7]	Rossel et al., 2006
		Using tylosine as growth promoter at the end of the fattening period	1.6 [1.2–2.2]	Van der Wolf et al., 2001
		Using chlortetracycline as growth promotor during the fattening period	6.9 [2.8–17.1]	Funk et al., 2007
		Serological prevalence is higher in pigs fed with a chlortetracycline, procaine penicillin and sulphamethazine supplemented ration versus approved growth promotor or probiotic	4.1† [2.1–8.1]	Leontides et al., 2003
	Intercurrent diseases	Diarrhoea in growing pigs	NE	Møller et al., 1998
		Infections by PRRSV (Porcine Reproductive and Respiratory Syndrome Virus)	3.0 [1.3–6.7]	Beloeil et al., 2004a
		Infections by PRCV (Porcine Respiratory Coronavirus)	6.9 [2.2–21.6]	Fablet et al., 2003a; b
		Infections by Lawsonia intracellularis	2.1 [1.2–3.7]	Beloeil et al., 2004a
		Liver infestations by Ascaris suum with high level (>16%) of liver condemnation at slaughterhouse during meat inspection	2.1 [1.2–3.8]	Van der Wolf et al., 2001

NE, not possible to estimate.

*Calculated OR from published data.

†Epidemiological link showed by molecular strains typing methods.

‡Risk factor shown by modelisation.

§Results obtained from multiple factors analysis.

Gurtler et al. (2005) showed that no piglets (n = 600) shed Y. enterocolitica in faeces in the period from farrowing to post‐weaning whereas 19.6% of the 491 fattening pigs shed the bacteria at the end of the fattening period. Pilon et al. (2000) showed that Pulsed‐Field Gel Electrophonesis (PFGE) profiles of isolates from pig faecal samples and environmental samples were specific for each of the 16 positive herds studied, with no genomic link between strains isolated from neighbouring farms. Rodents and flies were not found positive in one study (Pilon et al., 2000). Moreover, only 3.4% of the 117 environmental samples (feed and drinking troughs, water taps, and boots) were found positive in the study carried out by Altrock et al. (2007). Only one study reported a statistically significant association between the presence of cats on farms and higher serological prevalence (Skjerve et al., 1998). Herd management seems to have an influence on Y. enterocolitica presence. This point was reported by Skjerve et al. (1998) who showed serological individual and herd prevalences of 0.350 and 0.531, respectively, among 179 farrow‐to‐finish herds, versus serological individual and herd prevalences of 0.660 and 0.860, respectively, among 86 fattening herds. The lack of contact between pigs from different origins in farrow‐to‐finish herds may explain such a result.

Risk factors for infection with Y. enterocolitica included herd type (fattening herds versus farrow‐to‐finish herds) and biosecurity measures (Table 8).

Table 8

 Prevalence of pathogenic Yersinia enterocolitica in finishing pigs reported in 14 studies in eight European and North‐American countries

Shedding or carriage	Country	References	Individual or pool prevalence			Pigs tested	Herd prevalence			Herds tested	Type of material sampled	Type of analysis
			p	p 95% CI			p	p 95% CI
				Lower limit	Upper limit			Lower limit	Upper limit
Shedding (prevalence in faeces or rectal contents)	Canada	Pilon et al., 2000	0.128*	0.107	0.149	1,010	0.800	0.625	0.975	20	Pen swabs	Ind, Cul
	Finland	Asplund et al., 1990	0.177	0.115	0.239	147	0.357	0.106	0.608	14	Faeces collected on farms	Ind, Cul
	Germany	Gürtler et al., 2005	0.196	0.161	0.231	491	0.500	0.010	0.990	4	Faeces (5 g) collected on farm	Ind, Cul
			0.005	0.001	0.012	379	0.500	0.010	0.990	4	Faeces (5 g) collected at slaughterhouse
		Altrock et al., 2007	0.084	0.066	0.102	900	–	–	–	30	Faeces collected at slaughterhouse	Ind, Cul
	Norway	Nesbakken et al., 2003	0.125	0.001	0.257	24	1.000	0.963	1.000	3	Faeces (10 g) collected at slaughterhouse	Ind, Cul
	United States	Bhaduri et al., 2005	0.124	0.112	0.136	2,793	0.532	0.421	0.643	77	Faeces (1 g) collected on farms	Ind, Cul
		Bowman et al., 2007	0.107	0.084	0.130	718	0.875	0.645	1.000	8	Rectal and pharyngal swabs	Ind, Cul
Digestive carriage (prevalence in digestive contents or tonsils and lymph nodes collected at slaughterhouse)	Canada	Letellier et al., 1999a	0.173	0.153	0.193	1,420	0.673	0.611	0.735	223	Caecal contents (1g)	Ind, Cul
	Finland	Asplund et al., 1990	0.364	0.321	0.407	481	0.708	0.579	0.837	48	Tonsils	Ind, Cul
		Fredriksson‐Ahomaa et al., 2000	0.330	0.262	0.398	185	–	–	–	–	Tonsils (10 g)	Ind, Cul
	Germany	Gürtler et al., 2005	0.384	0.335	0.433	372	1.000	0.968	1.000	4	Tonsils	Ind, Cul
			0.038	0.018	0.058	346	0.500	0.010	0.990	4	Ileocaecal lymph nodes	Ind, Cul
		Nowak et al., 2006	0.050	0.020	0.080	210	1.000	0.974	1.000	6	Caecal contents	Ind, Cul
			0.290	0.229	0.351	210	1.000	0.974	1.000	6	Tonsils	Ind, Cul
			0.020	0.001	0.039	210	1.000	0.974	1.000	6	Ileocaecal lymph nodes	Ind, Cul
	Italy	Bonardi et al., 2003	0.040	0.009	0.071	150	–	–	–	–	Caecal contents (10 g)	Ind, Cul
			0.147	0.090	0.204	150	–	–	–	–	Tonsils (1 g)	Ind, Cul
	Latvia	Terentjeva et al., 2007	0.310	0.223	0.397	108	1.000	0.974	1.000	6	Tonsils	Ind, Cul
	Norway	Nesbakken et al., 2003	0.042	0.002	0.082	97	0.333	0.001	0.866	3	Submaxillary lymph nodes (5 g)	Ind, Cul
			0.625	0.431	0.819	24	1.000	0.963	1.000	3	Tonsils (10 g)	Ind, Cul
			0.042	0.001	0.122		0.333	0.001	0.866		Mesenteric nodes (5 g)
			0.042	0.001	0.122		0.333	0.001	0.866		Stomachs (10 g)
			0.167	0.018	0.316		0.333	0.001	0.866		Ileum contents (10 g)
			0.125	0.001	0.257		0.667	0.134	1.000		Caecum contents (10 g)
			0.083	0.001	0.193		0.667	0.134	1.000		Colon contents (10 g)
	United States	Funk et al., 1998	0.132	0.121	0.143	3,375	0.282	0.195	0.369	103	Oro‐pharyngal swabs	Ind, Cul
Serological prevalence	Denmark	Skjerve et al., 1998	0.541	0.517	0.565	1,605	0.638	0.580	0.696	265	Blood samples	Ind, AD
	Germany	Altrock et al., 2007	0.668	0.637	0.698	900	0.833	0.700	0.966	30	Blood samples at slaughterhouse	Ind, AD
	Norway	Nesbakken et al., 2003	0.875	0.743	1.000	24	1.000	0.963	1.000	3	Blood samples collected at slaughterhouse	Ind, AD

P, apparent prevalence; Ind, individual analysis; Pool, pooled samples analysed; Cul, culture; AD, antibody detection; –, lack of data.

*Estimate calculated from the proportion of pathogenic strains among fecal and environmental samples.

Summary of Risk Factors for Bacterial Hazards in Slaughter Pigs

The two main biosecurity measures involved with the presence of bacterial food‐borne pathogens in herds were lack of cleaning after batch removal and the absence of cloth‐boot change rooms at entrances to the facilities (Table 9). Straw on floors was a reported risk factor for the presence of Salm. enterica and Y. enterocolitica. Whereas wet feed increased the risk of infection of pigs with L. monocytogenes, dry feed was a risk factor for Salmonella. Mixing pig batches, notably in fattening herds, increased the transmission of Salmonella and Yersinia. Whereas small herds (<1000 pigs) were more contaminated by thermophilic campylobacters and Y. enterocolitica, large herds were associated with higher prevalence of Salmonella. Antibiotic treatment during the fattening period seems to increase the risk of transmission of Salmonella.

Table 9

 Summary of risk factors of food‐borne zoonotic hazards prevalence in pig herds

	Characteristic of pig herd	No. studies showing the risk factor of increase in presence of foodborne pathogens
		Campylobacter spp.	Listeria monocytogenes	Salmonella enterica	Yersinia enterocolitica
Biosecurity measures	Lack of cleaning after batches removal and/or default of duration of the ‘empty and clean’ period	1	1	5	–
	Lack of change rooms at the entrance to the facilities and/or lack of hand or clothes hygiene	–	1	5	–
	Contact with visitors or contaminated equipment	–	–	2	–
	Contact with domestic or wild animals (bird and rodents)	–	–	4	1
	Nature of floors (straw or partially slatted floor versus fully slatted floor)	–	–	4	1
Feed and watering	Wet feed	–	1	–	–
	Dry feed	–	–	4	–
	Pelleted feed	–	–	3	–
	High (basic) pH feed or water	–	–	2	–
	Contaminated feed or water	–	–	–	1
	Change of diet during the following	–	–	1	–
	Cleaning of pipelines used for wet feed and water	–	1	–	–
	Bowl drinkers	–	–	1	–
Herd management	Infection of sows	–	–	2	–
	High density of animals (<0.75 m2 per pig)	–	–	1	–
	Small herd size (<1000 pigs)	1	–	–	1
	Huge herd size (> 1000 pigs)	–	–	3	–
	Fattening herds (versus farrow‐to‐finish herds)	–	–	2	1
	Presence of other species breedings	–	–	3	1
	Batches mixing or contact between pens (continuous production or snout contacts)	1	–	3	1
Health management	Preventive or curative antibiotic treatment during the fattening period	–	–	5	–
	Preventive or curative anthelmintic treatment during the fattening period	1	–	–	–
	Intercurrent diseases (diarrhoea, Porcine Reproductive and Respiratory Syndrome Virus, Porcine Respiratory Coronavirus, Lawsonia intarcallularis, Ascaris suum)	–	–	7	–

–, lack of data (i.e. risk factor for which no study was published).

These risk factors may be used for a risk‐profiling approach of farms. For instance, herds with a lack of cleaning after batches removal could be considered as high risk herds for Campylobacter, L. monocytogenes and Salm. enterica, in opposition to herds for which cleaning after batches removal is systematically carried out. Besides, wet fed pigs herds may be considered as high‐risk herds for L. monocytogenes, in opposition to dry fed pigs herds. Nevertheless, further studies are needed to complete and to explain some of these factors and to improve their use in a pre‐harvest hazard control approach.

Discussion and Conclusion

Meat contamination by food‐borne pathogens mainly occur: (i) on farms, with primary contamination of muscles and tissues, (ii) during slaughtering from digestive contents and/or digestive tract tissues of pigs themselves when the reservoir of hazards is digestive (Fosse and Magras, 2004). To strengthen working relationships between farm and slaughterhouse in a risk assessment approach to control food‐borne zoonoses, new European Union regulations notably require ‘food chain information’. The goal of this effective risk approach is to provide information that is meaningful, relevant and targeted to four high‐risk pork‐borne pathogens (EFSA‐ECDC, 2007; Fosse et al., 2008a) faecal shedding and/or digestive carriage of finishing pigs, which are the primary products of this pork food chain. The primary objective of this review was to better map the knowledge base of prevalence and risk factor data of four high‐risk food‐borne pathogens in finishing pigs. Prevalence estimates are affected by the sampling strategy and diagnostic procedure (Davies et al., 2003; Sanchez et al., 2007). So to describe the between‐study variation in pathogen prevalence and to establish prevalence distributions which could be used in risk assessment, a quantitative meta‐analysis has been carried out.

A systematic literature search was conducted. Among 256 articles that met the inclusion criteria on CAB and Medline databases, 86 articles contained original data suitable for a quantitative meta‐analysis performed on prevalence at the herd and finishing pig levels. Since serological herd‐prevalence is partially linked with digestive carriage herd‐prevalence (Christensen et al., 1999; Davies et al., 2003), these data were also included. Few data were available on L. monocytogenes in pigs (three studies). For thermophilic Campylobacter spp., Salm. enterica and Y. enterocolitica, 17, 46 and 14 papers reporting apparent prevalence and 1, 23 and 2 papers reporting risk factors were used, respectively. To our knowledge, this is the first time that such a review has been conducted. Even if such work may never be considered exhaustive, the authors felt that all the relevant literature was identified. Criteria used to exclude publications (notably language) could have partially biased results, even if scientific publications quoted on databases were mainly written in English. The assessment of quality of included studies was carried out by two abstractors who compared their results and had the same conclusions. Nevertheless, it could also be responsible for bias. Our results could thus only be considered as primary estimates of prevalence.

The second objective of this review was to identify significant risk factors for increase in food‐borne zoonotic hazards prevalence in finishing pigs herds which may be used as food chain information in a farm to fork risk assessment approach. However, very few data were published for risk factors on Campylobacter spp. (Wehebrink et al., 2007), L. monocytogenes (Beloeil et al., 2003a; b) and Y. enterocolitica (Skjerve et al., 1998; Pilon et al., 2000) whereas risk factors for Salmonella have been studied in more detail (23 papers). Consequently, such a summarization may be considered a first qualitative approach to risk factor critical review. Nevertheless, its use in managing food‐borne hazards in the pork food chain in a farm to fork approach may already be taken into account. Moreover, each hazard should be considered separately with a reasoned discussion of risk factors linked with its bacteriological characteristics (growth conditions, survival in a given environment, and ability for biofilm formation). However, when considering global risk factors (Table 9), the two main categories of risk factors are: (i) biosecurity measures with a lack or absence of external and internal measures preventing hazard transmission such as contact with other animals (Skjerve et al., 1998; Letellier et al., 1999b; Langvad et al., 2006) or contaminated feed and watering (Hurd et al., 2001; Beloeil et al., 2003b; Bahnson et al., 2006b); (ii) herd management practices with considerable transmission of hazards (mixing of batches, snout contact).

For L. monocytogenes very few data of prevalence were reported. Nevertheless, pigs infection seems very low. On the opposite, wide ranges of apparent prevalence were reported for Salm. enterica, Campylobacter spp. and Y. enterocolitica. Sanchez et al. (2007) have shown that the three most important factors influencing the apparent prevalence of Salmonella in pigs were diagnostic procedures, sampling design and countries. This review analysed three categories of prevalence estimates: pig shedding prevalence (bacteriological data from fresh faeces or rectal contents), pig carriage prevalence (from digestive tissues or contents), serological prevalence. Consequently, prevalence estimates are less biased by diagnostic procedures and sampling design and observed variations may be explained by particular sanitary situations. The Q parameter values showed that the heterogeneity of data used was not significant for Campylobacter and Y. enterocolitica. Such results suggested that Campylobacter and Y. enterocolitica pigs shedding and/or carriage could be independent of the country status or that herd management practices are not significantly different in European and North‐American countries for this hazard.

Prevalence estimates highly range according to material sampled and analyses carried out (bacteriological or serological analyses). Serological individual prevalence summaries were systimatically higher than bacteriological individual prevalence summaries except for Salm. enterica. For this hazard, we also showed that serological individual or herd prevalence estimates were highly dependent on the optical density cut‐off. Serological analyses with high cut‐off could conduct to underestimates of Salmonella infection in pigs.

For all hazards, individual prevalence estimates for pig carriage (samples collected at slaughterhouses) were higher than individual prevalence estimates for pig shedding (samples collected on farms or at slaughterhouses). Such results may be explained by two points: (i) sampling at slaughterhouses can be targeted to predilection sites for the presence of bacteria: lymph nodes – especially tonsils – for Y. enterocolitica (Tauxe et al., 1987), intestinal and caecal contents for Salmonella (Hurd et al., 2004; Sørensen et al., 2004; Bahnson et al., 2006b; Rostagno et al., 2007); (ii) the transfer of bacteria from digestive tissues to digestive tracts due to stress during transport or lairage at slaughterhouses and/or infection of pigs from herd to slaughterhouse (Fravalo et al., 1999). Campylobacter pig shedding and carriage prevalence were very high. This showed that this hazard is widespread all over the world. Thus, this pig digestive tract bacterium would be an interesting indicator of faecal contaminations during the slaughtering process (Laroche et al., 2007). Salmonella enterica and Y. enterocolitica are characterized by lower individual and herd prevalence.

To summarize, the application of good hygiene and biosecurity practices in herds, notably with respect to cleaning and disinfection procedures and high hygiene standard of clothes, may reduce the contamination pressure at slaughter. As a priority in biosecurity measures, limiting the mixing of pig batches is needed, as well as less antibiotic treatment. These measures can be taken to reduce the presence of food‐borne hazards in the first step in the pork food chain, i.e. the farm, and thus to better protect consumers. Further studies are needed to characterize pig infection at herd level, notably for L. monocytogenes and to quantify the correlation between infection of pigs on farms and contamination of carcasses at slaughterhouses.