Salmonella enterica Infections in Spanish Swine Fattening Units

Introduction

Infection with Salmonella enterica is one of the main causes of human gastroenteritis worldwide and over the last decade pork and pork products have been identified as an important source of human outbreaks of salmonellosis (Schwartz, 1999; Fedorka-Cray and Wray, 2000). In 1999 salmonellosis in humans was attributable to pork or pork products in 10–15%, 14–19% and 18–23% of the cases in Denmark, the Netherlands and Germany, respectively (Hald and Wegener, 1999) while in England and Wales, in the period 1992–1999, 32% of meat related food-borne outbreaks were associated with pig meat and 48% of them were caused by Samonella (Smerdon et al., 2001).

Clinical porcine salmonellosis is a major cause of economic loss in pig production and can be separated into two syndromes, one associated with enterocolitis and involving mainly S. Typhimurium and the other usually associated with septicaemia and caused by S. Cholerasuis (Fedorka-Cray and Wray, 2000). However, swine salmonellosis has gained recognition as a public-health problem. Pigs are susceptible to a wide variety of serovars of Salmonella and although infected pigs usually remain as healthy carriers, sub-clinically infected pigs may excrete Salmonella spp. in faeces or keep the bacteria in the digestive tract, the closely associated lymph nodes or the tonsils (Fedorka-Cray and Wray, 2000). Salmonella carrier pigs entering the slaughterhouse have been demonstrated to be the most important sources of carcasses and products contamination (Bahnson et al., 2006).

In response to the ongoing need to protect the health of consumers, the European Union (EU) is now requiring all member states to develop monitoring programmes for Salmonella in swine. Directive 2003/99/EC on the monitoring of zoonosis and zoonotic agents and Regulation (EC) No. 2160/2003 on the control of Salmonella and other zoonotic agents provide detailed regulations for monitoring, and targets for reducing the Salmonella prevalence in pig herds. Several countries with medium or high Salmonella prevalence among swine farms: the UK, Ireland, Germany, Holland and particularly Denmark have implemented Salmonella control programmes. The Danish Salmonella Control Program was the first and started in January 1995 with the main objective of reducing the occurrence of Salmonella spp. in pig herds and pork products (Mousing et al., 1997). The key element of the programme is a classification scheme of the herds based on a serological survey using a mix LPS–ELISA containing the O-antigen factors 1, 4, 5, 6, 7 and 12 (Nielsen et al., 1995), which are in serogroups B and C1, and which were previously determined as being the most prevalent ones among swine farms in Denmark (Baggesen et al., 1996). On the other hand, Salmonella Control Programs from Sweden, Norway and Finland, countries with a low prevalence of infection, are based in the rejection of Salmonella contaminated animals, feed and food products.

The development of pig production in Spain has been spectacular. It has increased by 38% since 1997 to reach 3.3 million Tm in 2006. This is the second highest output in the EU and pig meat exports have doubled in the last 5 years to reach almost 670.000 Tm in 2005 (data from the Ministry of Agriculture and Livestock). In spite of the importance of the Spanish swine industry, there is a lack of information concerning Salmonella status in Spanish swine farms. The aim of the present study was to determine the bacteriological prevalence of sub-clinical Salmonella infections in Spanish swine farms and to determine the most frequently isolated serovars of the bacteria.

Material and Methods

Results

A total of 2320 faecal pen floor specimens and 232 farms producing market pigs and located in 12 different regions of Spain were sampled through the study. The estimated number of slaughtered pigs per year in each farm ranged from 100 to 187 000 with a median of 3500 (mean 7316.6).

Salmonella spp. were recovered from 290 of 2320 faecal floor specimens (12.5%; 95% CI: 11.2–13.8%) and 100 out of 232 farms housing fattening pigs had at least one positive sample (43.1%; 95% CI: 37–49.1%). The number of positive samples per positive farm ranged from 1 to 10 with a median of 2 (mean 2.9).

The apparent herd prevalence in different regions of Spain is shown in Table 1. In particular, Castilla y León, a region located in the north-west part of Spain, showed a lower prevalence of Salmonella-positive herds. The risk of a herd being positive for Salmonella enterica in Castilla y León was four times (OR = 4.04, 95% CI: 1.54–11.22%) lower as compared with the rest of the country (χ² = 9.1; P = 0.002). There was no statistical significant correlation across the different regions between the prevalence of positive herds or positive samples and the total number of fattening units or the number of fatteners (>50 kg/km²) in the area.

Most of the participating farms were multi-site operations (41.9%) while the rest were individual finishing farms (33.3%) or part of farrow-to-finish operations (24.8%). The apparent herd prevalence of salmonellosis was 42%, 42.8% and 36.5%, respectively, among multi-site, finishing and farrow to finish farms. These differences did not reach statistical significance (χ² = 4.8; P = 0.088).

The most frequently isolated serogroups were B and C1, detected in 62% and 31% of the positive samples respectively. Other serogroups identified were E1 (3.4%), G (1.7%), C2 (1.4%), D1 (0.3%) and K (0.3%). The predominant serovar was S. Typhimurium, comprising 91 of the isolates (31.4%), followed by 69 S. Rissen (23.8%) and 47 S. Derby (16.2%). Altogether, a total of 24 different serovars were detected (Table 2).

At herd level, 90% of the positive herds were infected by Salmonella serogroups B or C1. S. Typhimurium was found in 38 farms (38%). Other frequent serovars were S. Rissen, S. Derby, S. 4,12:i:-, S. 4,5,12:i:- and S. Bredeney with 25%, 14%, 8%, 7% and 7% of the positive herds respectively. In 66% of the farms only one Salmonella serovar was recovered while in 27% and 7% of the farms two and three different serovars respectively were detected.

Results of phage typing were available for isolates of S. Typhimurium, including Typhimurium var. Copenhagen (Table 3), the O:5-negative variant of Salmonella serovar Typhimurium, and also for the 22 isolates of S. 4,5,12:i:- and S. 4,12:i:-. A total number of nine different phage types of S. Typhimurium were identified with DT 193 being the most frequently identified (31.9%), followed by DT 104b (9.9%). Twenty-two isolates (24.2%) were non-typable and 6 (6.6%) were atypical or reaction does not conform which indicated that the tested bacterial strain reacted with some of the typing phages but did not conform to a standard phage type. Two phage types occurred in seven farms (15.2% of the S. Typhimurium infected farms) while three phage types occurred in one herd (2.4%). Phage typing of the 12 S. 4,5,12:i:- isolates classified seven of them as DT U302 (58.3%) while among the 9 S. 4,12:i:- isolates, phage type DT 193 was the most common (33.3%).

Discussion

Serological detection using mix LPS–ELISAs, which combine different O-antigens, have been proposed as the best option to establish the Salmonella prevalence in pig herds (Nielsen et al., 1995; Van der Heijden et al., 1998; Van der wolf et al., 1999). However, bacteriological surveys are necessary to establish the prevalence of different Salmonella serogroups, determined by O-antigens, in each particular area and consequently to determine the sensitivity of these serological tests for all the Salmonella infections occurring in the population (EFSA, 2006). On the other hand, estimates of bacteriological prevalence of Salmonella infections in pigs can be established at either the farm or the slaughterhouse with very different results. Williams and Newell (1967) have already demonstrated that different results were obtained at the farm and the slaughterhouse with samples collected from the same pigs. More recently, Fedorka-Cray et al. (1995) demonstrated that S. Typhimurium can be isolated from the caecal content within 4–6 h after oral exposure of pigs to the bacterium and Hurd et al. (2002) showed a 7-fold-higher S. enterica isolation rate from pigs necropsied at the slaughterhouse as compared with those slaughtered on the farm.

The present study is the first large-scale screening for Salmonella spp. conducted in Spain. Apparent bacteriological prevalence of Salmonella infection was estimated using faecal samples collected on swine farms housing fattening pigs distributed through the whole country and was found to be moderately high, both at the sample (12.5%) and the herd level (43.1%). These data are higher than the 6% sample and 38% farm prevalence reported by Fedorka-Cray et al. (1996) on swine farms in the USA, the 23.7% farm prevalence in the Netherlands described by Van der wolf et al. (1999) or the 11.4% farm prevalence described by Christensen et al. (2002) 4 years after the implementation of the Salmonella Control Program in Denmark. In contrast, the results of this study are similar or lower to the 14.3% and 66.7% reported by Rajic et al. (2005) on 90 Alberta finishing farms at the sample and farm level respectively, 51% of positive farms found in Ireland (Rowe et al., 2003) or 57.3% of positive swine farms detected across five states of the USA (Rodriguez et al., 2006).

Comparisons among different studies should be considered very carefully as prevalence estimates are affected by the sampling strategy (sample size and type of sampling) and isolation procedure (Davies et al., 2000; Funk et al., 2000; Rajic et al., 2005). Funk et al. (2000) demonstrated an increase in sensitivity associated with increased faecal sample size whereas Davies et al. (2000) reported differences in the sensitivity of bacteriological culture depending on testing protocols. So, besides other factors, the use of 10 pen floor samples per farm with a minimum of 25 g each, which were processed individually, and the microbiological procedure which included pre-enrichment, enrichment and selective steps may account for the higher prevalence found in this study in comparison with previous ones. Moreover, each faecal sample was a pooled sample as it was collected from five different locations within each pen. According to Arnold et al. (2005), the sensitivity of microbiological methods increases with the number of samples in the pool.

This study found that the apparent prevalence of Salmonella infected herds was lower in Castilla y León, one of the largest regions in the EU, which represented more than 10% of the Spanish fattener and nearly 20% of the Spanish breeding sow census (data from the Ministry of Agriculture and Livestock http://www.mapa.es). No correlation could be demonstrated between the prevalence of Salmonella-positive herds or samples and the total number of fattening units or the density of fatteners in each of the sampled areas. Nevertheless, it is important to emphasize that Aragón, Catalonia and Murcia regions, where more than 50% of the Spanish fattening units are concentrated, and are the most highly concentrated areas of pig production in Spain, reached very high values of Salmonella-positive herds and samples.

A total of 24 different serovars were identified, indicating a great diversity among swine Salmonella isolates, probably as a result of multiple infective sources (Baggesen et al., 1996). A variety of Salmonella serovars in swine farm samples, higher than those of beef, dairy or poultry farms, has been reported by Rodriguez et al. (2006). As previously described in swine farms of other countries (Baggesen et al., 1996; Van der wolf et al., 1999; Rowe et al., 2003; Rajic et al., 2005), S. Typhimurium was the predominant serovar at the herd (40%) and individual level (31.4%). Among them, its O:5-negative variant, designated variant Copenhagen, which was primarily reported to be found in pigeons but it is now also isolated from cattle, swine and other animals (Frech et al., 2003), was the most frequent at the herd (16.2%) and individual level (21%). Although the variant Copenhagen has rarely been related with diseases in humans, during 2003 and 2004 it was identified in the NRLSS in 8.2% and 9.7% of the Spanish human isolates respectively (data do not shown).

According to the results reported by Mejía et al. (2003) on swine farms in Catalonia and Astorga et al. (2007) on Andalusian swine production units, S. Rissen, detected in 25% of the farms, was the second serovar isolated from Spanish swine fattening units. This serovar has also emerged since the year 2000 among Spanish human isolates of Salmonella and ranked 6th in importance in 2002–2003 (Echeita et al., 2005), being mostly associated with pork products (De Frutos et al., 2005). Conversely, S. Derby, a serovar which it has been suggested is able to establish a persistent infection in pig herds (Baggesen et al., 1996), was isolated in 16% of the samples and 14% of the positive farms. Regarding this, a clone of S. Derby in swine samples, pork-derived products and subsequently in humans has recently been described in Spain (Valdezate et al., 2005).

The S. 4,5,12:i:-, a serovar that emerged and spread among human isolates of Salmonella in 1997 and has been associated with a swine reservoir (Echeita et al., 1999), ranks 4th in our study. It has been suggested that this serovar originates from S. Typhimurium DT U302 strains (Echeita et al., 1999, 2001; De la Torre et al., 2003), and is frequently associated with the multi-resistance profile R-ACSSuT (ampicillin, chloramphenicol, streptomycin, sulfonamide and tetracycline) as well as resistance to gentamicin and trimethoprim–sulfamethoxazole (De la Torre et al., 2003).

Knowledge of prevailing serogroups of Salmonella is important because it is an important factor influencing the sensitivity of mix LPS–ELISAs used for serological detection of Salmonella infected farms. Most of these techniques are designed to detect antibodies only against serogroups B, C1 and D1 (Nielsen et al., 1995; Van der Heijden et al., 1998) and therefore 91% of the infected farms in this study should be identified using the serological approach (Table 2). A similar result has been described for Denmark (Baggesen et al., 1996), the Netherlands (Van der wolf et al., 1999) or Canada (Letellier et al., 1999; Rajic et al., 2005).

The present investigation has also demonstrated a high diversity of phage types among S. Typhimurium isolates with nine different types regardless of the high number of non-typable strains. Phage type DT 193 was the most frequently identified, followed by DT 104, DT 104b and DT U302. Moreover, these phage types were also the most frequent among human isolates of S. Typhimurium in Spain (Echeita et al., 2005). It is very important to emphasize that DT 104 that is now acknowledged as an internationally distributed zoonotic pathogen (Helms et al., 2005), DT U302, closely related to DT 104 and DT 104b phage types have been associated with several multi-resistant patterns and account for 23% and 29% of the Typhimurium total isolates or Typhimurium infected farms respectively.