Discussion

The prevalence of C. upsaliensis reported in this study (38%) is in the middle of the range reported for similar dog populations (17%, to 59%) (Rossi et al., 2008; Sandberg

et al., 2002). Different prevalence rates identified in studies may be due to differences in the underlying populations, or in the methods used, which have only recently been optimised forC. upsaliensis as well as C. jejuni detection (Byrneet al., 2001; Guest et al., 2007; Kulkarni et al., 2002; Labarca et al., 2002; Lastovica and Le Roux, 2003). Despite the dog population in this study being a vet-visiting one; the majority of these dogs were healthy (no diarrhoea). These results suggest that dogs may be an important reservoir forC. upsaliensis.

Previous studies have reported considerably higher prevalences of C. jejuni (3 to 40%) in dogs (Hald and Madsen, 1997; Koene et al., 2004; Lopez et al., 2002; Tsai et al., 2007; Workman et al., 2005) than this study (1.2%), this may be due to a number of factors including; the source of the dogs, age and detection methods used. Some studies have found an association with C. jejuniand disease in dogs (Burnens et al., 1992; Fox

et al., 1983; Nair et al., 1985), which might imply that C. jejuni infection in dogs is sporadic, potentially the result of eating contaminated food, as with humans. However other studies have found no association between C. jejuni infection and diarrhoea in dogs (Damborg et al., 2004; Hald et al., 2004; Lopez et al., 2002). Most studies have found younger dogs are more likely to carry Campylobacter spp. than older dogs, particularly whenC. upsaliensisis the most common species isolated (Haldet al., 2004; Sandberget al., 2002; Wielandet al., 2005). This association with younger dogs has not been reported where C. jejuni is taken into account, or when the C. upsaliensis

prevalence is low (Tsaiet al., 2007; Wieland et al., 2005). If indeedC. jejuniis sporadic in dogs, then this disassociation with age might be expected. The low prevalence in this present study suggests that this vet visiting population of dogs are unlikely to be an important source ofC. jejuniinfection for humans.

3.5.1 Comparison of Detection Methods

Of the three culture methods, plating after enrichment detected fewer Campylobacter

spp. than the other two culture methods. This has also been noted by another study (Westgarth et al., 2009). One explanation for this might be that the enrichment stage allows for contaminating bacteria to increase and out compete Campylobacter spp. (Abulreesh et al., 2005; Korhonen and Martikainen, 1990). There appeared to be no significant difference in the numbers of Campylobacter spp. detected by direct plating onto mCCDA, and CAT media with prior filtration. Other studies have found similar

findings for the detection ofCampylobacter spp. (Bourkeet al., 1998),C. jejuniandC. coli(Engberget al., 2000), andC. upsaliensis(Hald and Madsen, 1997).

3.5.2 Culture Versus Direct PCR

In this study, direct PCR was found to be more sensitive than culture for detecting C. upsaliensis. Transportation time appears to have a significant effect on culturable

Campylobacter spp., and this has been observed in another study (Koeneet al., 2004). An additional factor may be the existence of viable, but non-culturable forms of C. upsaliensis(Murphyet al., 2006; Persson and Olsen, 2005), which are thought to occur more frequently when bacteria are exposed to adverse conditions (transportation). Alternatively low level shedding or a past infection may be detectable by direct PCR but not by culture. In the instance of dog 06012, which was cultured as C. jejuni, but was found to have C. upsaliensis by direct DNA extract, the C. jejuni may have out competed C. upsaliensis in culture due to its faster growing time (Labarca et al., 2002; Lastovica and Le Roux, 2003).

Persson and Olsen, (2005) found that when isolating C. coli and C. jejuni, direct PCR was inferior compared to culture, particularly with fresh samples. In the current study, the three C. jejuni isolates were all detected by culture, whereas only one of these isolates was detected by direct PCR. Samples obtained with minimal transportation time between collection and processing, may yield a different outcome to the one observed in this study. A disadvantage to direct PCR is that specific species primers need to be used instead of degenerate primers, which in this study limited the detection to C. jejuni and

C. upsaliensis. Despite this, it is possible to expand this method with the use of additional primers, multiplex PCR (Grove-Whiteet al., 2009) or real-time PCR (Chaban

However, eight faecal samples that yielded C. upsaliensis isolates in culture, did not yield an amplification product when tested by direct PCR. This observation has been made previously for C. jejuni (Lawson et al., 1999) and has been attributed to the degradation of DNA and/or the presence of inhibitory substances present in the faeces that may reduce the sensitivities of the PCR assay. Currently, no ‘gold standard’ exists for the detection of Campylobacterspp., and therefore direct PCR and culture methods should both be used to maximise recovery.

3.5.3 Dog Age, Clinical Signs, and Campylobacterspp.Status

Similar to other studies (Acke et al., 2006; Engvall et al., 2003; Guest et al., 2007; Sandberget al., 2002; Wielandet al., 2005), younger dogs were found to have a greater risk forC. upsaliensiscarriage than older dogs. Haldet al, (2004) found that the carriage rate of Campylobacter spp. in pet dogs in Denmark peaked at 13-15 months of age, especially for C. upsaliensis, which is similar to the findings of this present study. Similarly, Guest et al, (2007) also found that dogs negative for Campylobacter spp. were older (with an average age of 42.5 months) than the positive dogs who had an average age of 13.5 months. The most likely explanation for this effect is that older dogs have probably been exposed toCampylobacterspp. previously, and therefore developed a certain level of immunity to the bacterium. Immunity toCampylobacter spp. has been observed in Macaca nemestrina monkeys based on increasing immunoglobulin titres, which suggested that after an initial Campylobacter spp. infection, the host retained some immunity if exposed to the bacteria again (Russellet al., 1989).

We did not observe a statistically significant association between Campylobacter spp. carriage and clinical presentation/history, as has been reported in other studies (Acke et al., 2006; Engvall et al., 2003; Sandberget al., 2002; Workmanet al., 2005). However,

some studies have foundCampylobacter spp. associated with clinical signs (Ackeet al., 2009; Guestet al., 2007), particularly in younger dogs (Burnenset al., 1992; Fox et al., 1983; Nair et al., 1985), and often for C. jejuni (Fox et al., 1983; Nair et al., 1985). Furthermore the way in which dogs were sampled by the practitioners may have lead to bias, i.e. samples were taken based on the practitioners decision, not by random selection. A case control study would be a more appropriate method to explore this variable.

3.5.4 Dogs Living With Other Dogs and Cats

Dogs that lived with other dogs (not necessarily carrying Campylobacterspp.) tended to be more likely to carry C. upsaliensis in multivariable analysis, and this has also been found in other work (Westgarth et al., 2009). There was a significant association between a dog carrying C. upsaliensis and living with another positive dog in univariable analysis, although numbers for this group were small. Acke et al, (2006) suggested that dogs who live in groups, such as kennels have a higher prevalence of

Campylobacterspp. carriage, possibly due to cross-infection, and Damborget al, (2008) found indistinguishable amplified fragment length polymorphism (AFLP) patterns in strains that were isolated from dogs living in the same house or kennel, suggesting transmission. Previous studies have found no association between a dog’s

Campylobacterspp. status, and whether or not they lived with any other animals (Hald

et al., 2004; Lopezet al., 2002); these findings are supported by the current work as we did not find any association between canine C. upsaliensiscarriage and cohabiting with a cat, possibly because cats predominantly carryC. helveticus rather thanC.upsaliensis

3.5.5 Salmonella

In vet visiting and household dogs, aSalmonellaspp. prevalence of 1-2% has previously been found (Bagcigil et al., 2007; Hald et al., 2004; Tsai et al., 2007), which is supported by the findings in this study where the prevalence ofSalmonella spp. in dogs was very low (0.8%, 95% CI -0.3, 1.9). Other studies have found higher prevalences of

Salmonella spp. in dogs from various populations, ranging from 1-69%, although the majority of studies find a prevalence of less than 10%. (Bagcigil et al., 2007; Cantor et al., 1997; Hackett and Lappin, 2003; Hald et al., 2004; Schotte et al., 2007; Tsaiet al., 2007).

3.5.5.1 Salmonella Serovar Newport

Salmonella Newport was the only serovar found in this study and was found in two dogs. Although there appears to be no one serovar dominant in dogs, Newport has previously been reported as either the most common, or second most common serovar (Hald et al., 2004; Oloya et al., 2007; Seepersadsingh et al., 2004). The most likely source of infection for this serovar is thought to originate from cattle, but it has also been isolated from horses, reptiles, and seafood (CDC, 2008a; Gaertner et al., 2008; Karonet al., 2007; Khanet al., 2009; Oloyaet al., 2007; Talbotet al., 2006). There was evidence of a shared source of infection, or possible transmission between a calf and a dog during an outbreak of S. Newport on a farm (Daly and Neiger, 2008). Of the two dogs carrying S. Newport in this current study, one experienced diarrhoea in the past week prior to sampling, but the other had no recent history of diarrhoea. Both dogs in this study had received recent antibiotic treatment, which has been associated with increasedSalmonellaspp. isolation (Warnicket al., 2003).

3.5.5.2 Salmonella spp.Infections in Humans

The most commonly identified serotypes found in humans appear to be S. Enteritidis and S. Typhimurium in the UK and some states of America (CDC, 2008c; DEFRA, 2007). In some states of America,S. Newport accounted for 10% of salmonellosis cases, which was second only to the servovars previously mentioned (CDC, 2008c; Jones et al., 2008), and in some situations S. Newport appears to be the second most commonly isolated serovar after Typhimurium (Oloyaet al., 2007; Oloyaet al., 2009) (discussed in Chapter 5). Although S. Newport is less invasive and results in fewer deaths compared toS. Typhimurium, salmonellosis caused byS.Newport can still result in hospitalisation (Jones et al., 2008), and outbreaks of this serovar in humans have occurred (CDC, 2008a; Greene et al., 2008; Irvine et al., 2009). The sources of these outbreaks varies, but exposure to cattle, farms, unpasteurised milk, Mexican-style cheese, ham, mung bean sprouts, tomatoes, and lettuce (presumably contaminated with animal faeces whilst growing) have been identified as possible sources of infection (CDC, 2008a; Greene et al., 2008; Irvine et al., 2009; Karon et al., 2007; Lyytikainen et al., 2000; Mohle- Boetaniet al., 2009).

3.5.5.3 Salmonella Zoonoses

Associations between reptiles, pet rodents and salmonellosis in humans have been documented, but little is known about the relationship between dogs and humans, regarding Salmonella spp. transmission (CDC, 2003; Friedman et al., 1998). Interestingly, dog food/treats have been implicated in human cases of salmonellosis, some of which involved S. Newport (CDC, 2008b; Pitout et al., 2003). The results of this study suggest that this population of dogs is not a significant source of Salmonella