The importance of Campylobacter jejuni and Campylobacter coli as foodborne pathogens has steadily increased during the last years. With an incidence of 92 reported cases per 100'000 inhabitants, Campylobacter spp. are the leading cause of bacterial zoonosis in Switzerland 1. Inadequately cooked meat, particularly poultry, unpasteurised milk and contaminated drinking water are the most common sources for epidemic and sporadic foodborne cases 23. Due to the often self-limiting diarrhoea in humans, antibiotic treatment is normally not required. Nevertheless, antibiotic treatment is indicated for severe and prolonged enteritis, septicaemia, and for persons at risk such as very young or immuno-compromised patients. Besides erythromycin, which is the antibiotic of choice, fluoroquinolones and tetracycline are commonly used 4. Particularly fluoroquinolones are first-line drugs for empiric therapy of acute diarrhoea, as they are effective against most major pathogens causing bacterial enteritis 56. Therefore, most cases of campylobacteriosis receiving antibiotic treatment will initially be treated with fluoroquinolones. Worrying is the fact, that in recent years a rapidly increasing proportion of Campylobacter isolates from humans and animals all over the world were found to be resistant to fluoroquinolones 7891011. This coincided with initiation of the use of the fluoroquinolones to food animals in many countries 12131415.

The objectives of this study were i) to determine the prevalence of antibiotic resistant Campylobacter strains in raw poultry meat samples at retail level, and ii) to identify possible risk factors associated.

For sampling, 122 retail stores throughout Switzerland and Liechtenstein were randomly selected. From March until July 2002, food safety inspectors of the cantonal laboratories collected 415 samples of raw poultry meat (whole chickens as well as parts such as cutlet, meat cut into strips, legs, drumsticks, wings, and breast) from selected stores. Additionally, information about housing type of the poultry (as labelled), the country of origin (as labelled), refrigeration (chilled, frozen), and packing of the samples was recorded.

Of each sample, 10 g was inoculated into 100 ml selective enrichment broth (Brucella broth (211088, Becton Dickinson, Sparks, USA) with Campylobacter growth and selective supplement (SR084E, SR069E, Oxoid, Basingstoke, Hampshire, UK)) and incubated at 42°C for 24 h under microaerobic conditions provided by commercial gas packs (CampyGen from Oxoid). After enrichment the samples were streaked on selective agar media (Brucella agar (211086 Becton Dickinson) with 6% horse blood (SR0048C, Oxoid)) as well as Butzler Campylobacter selective supplement (SR085E, Oxoid) and incubated at 42°C for another 24 h under microaerobic conditions. Translucent white, moist and glistening colonies were picked and taken for further identification. Identification of Campylobacter strains was performed using the following standard tests: gram-negative stain, characteristic motion, catalase- and oxidase reactions, aerobic growth. One strain from each Campylobacter-positive sample was selected for susceptibility testing.

The disc diffusion method was performed as recommended by the National Committee for Clinical Laboratory Standards 16. The following antibiotic impregnated discs (bioMérieux SA, France) were used: erythromycin (15 μg), ciprofloxacin (5 μg), tetracycline (30 μg), gentamicin (10 μg), streptomycin (10 μg), ampicillin (10 μg), amoxicillin (25 μg) and sulphonamide (20 μg). Three to five isolated colonies of the same morphological type were selected from the agar plate culture and transferred into trypticase soy broth (211768, Becton Dickinson). After incubation at 42°C for 24 h under microaerobic conditions a sterile cotton swab was dipped into the suspension and streaked on the entire surface of a Mueller-Hinton agar (CM 337, Oxoid) with 5% sheep blood. Four antibiotic discs were placed on each plate and after 48 h of microaerobic incubation at 42°C the diameter of the inhibition zone was measured with calipers. E. coli ATCC 25922 and S. aureus ATCC 25923 were used as reference strains. Zones of growth inhibition were evaluated according to the NCCLS standards.

The E-test (AB Biodisk, Solna, Sweden) was performed for erythromycin, ciprofloxacin and tetracycline on Mueller-Hinton agar supplemented with 5% sheep blood according to the manufacture's instructions. Inocula were prepared by incubating the strains for 24 h at 42°C under microaerobic conditions in trypticase soy broth. After application of the E-test strips, plates were incubated in microaerobic conditions at 42°C for 48 h. The minimal inhibition concentration (MIC) was read directly from the test strip at the point where the elliptical zone of inhibition intersected the MIC scale on the strip. The following NCCLS zone diameter (mm) and MIC breakpoints for resistance were applied: erythromycin ≤ 13 mm and MIC ≥ 8 mg/l, ciprofloxacin ≤ 15 mm and MIC ≥ 4 mg/l, tetracycline ≤ 14 mm and MIC ≥ 16 mg/l, streptomycin ≤ 11 mm, ampicillin ≤ 13 mm, gentamicin ≤ 12 mm, amoxicillin ≤ 13 mm and sulphonamide ≤ 12 mm.

Statistical analysis consisted of initial bivariate screening for possible risk factors associated with resistance to the tested antibiotics. Variables screened for were:

• production system: free range production or production with bedding and winter garden, labelled as 'animal friendly' vs. conventional production

• foreign production vs. Swiss production

• whole chicken vs. parts of chicken

• products sold pre-packed vs. products sold in open displays

• chilled products vs. frozen products

A χ² test was applied to determine significance in bivariate screening. Factors significantly associated with antibiotic resistance (p < 0.05) were included into a multiple logistic regression model. Variable selection was performed by stepwise backward selection. Only significant variables (p < 0.05) were retained in the model. For each tested antibiotic, a separate regression analysis was performed. In addition, risk factors for multiple resistance, defined as resistance to at least two of the tested antibiotics, were determined in a separate regression model. All statistical analyses were performed with the statistical software package NCSS 2000 (Number Cruncher Statistical Systems, Kaysville, Utah, USA).

The high prevalence of Campylobacter spp. in raw poultry meat samples found in this study agrees with data from other studies 171819. Compared to the situation in other countries, the data of this study show a favourable resistance situation for Campylobacter strains isolated from raw poultry meat produced in Switzerland. In the USA, 90% of Campylobacter strains isolated from poultry meat had resistance to at least one and 45% to at least two antibiotics 20.

In our study, 28.7% of the isolated strains were resistant to ciprofloxacin. This prevalence is comparable to the results reported in the cited American study 20. A much higher prevalence of ciprofloxacin resistant strains was found in Austria (47%) 21, while a lower prevalence was reported in Denmark (6%) 22. However, it has to be considered, that in our study the resistance to ciprofloxacin was significantly higher in Campylobacter spp. isolated from foreign products than from Swiss products. Nevertheless, the prevalence of ciprofloxacin resistance in Campylobacter strains from Swiss poultry meat products was 19% and thus higher than for any other antibiotic tested. Data on resistance patterns of Campylobacter isolated from humans in Switzerland only have limited validity (small number of non-representative samples). Nevertheless, the highest rates for resistance, in the last four years, were observed for nalidixid acid (over 80% of isolates) 23. As there is strong cross resistance to fluoroquinolones which are most frequently used for empiric antimicrobial treatment of acute diarrhoea, this fact gives cause for concern. In Switzerland Baytril^®(enrofloxacin) is licensed for therapeutic use in poultry and therefore, ongoing monitoring is required. Twelve point six percent of the isolated strains in our study were resistant against tetracycline. A much higher prevalence of tetracycline resistant strains was found in USA (66%), and in Austria (47%) 2021. Only 7% of the strains in the Danish study were tetracycline resistant 22.

Comparable to our data, a low resistance prevalence against erythromycin was also reported in Austria and Denmark 2122, whereas in the USA 20% erythromycin resistant strains were found 20.

In a recent Swiss study resistance patterns of Campylobacter strains isolated from finishing pigs were investigated 24. Whereas prevalence of resistance against ciprofloxacin, tetracycline, and gentamicin in these strains was similar to the results of this study, prevalence of resistance against erythromycin and streptomycin was 19.8% and 79.1%, respectively, and therefore much higher than in strains isolated from poultry meat. This finding is consistent with Danish data 25 finding a higher prevalence of resistance to streptomycin in Campylobacter coli, predominantly isolated in pigs, than in Campylobacter jejuni, the predominantly strain in poultry meat. Generally, the comparison of different studies turned out to be difficult because of different sample collection and varieties in culture methods and susceptibility test methods.

A noteworthy finding was the apparent clustering of retail shops with at least one sample with Campylobacter resistant to multiple antibiotics in the south of Switzerland (figure 1). This could be explained by the high proportion of imported poultry meat products in this region. However, due to sample size limitations, analysis by country of origin did not produce meaningful results. Moreover, the results of this study indicate an association between Campylobacter spp. being resistant to amoxicillin and poultry products sold in open displays. One could speculate that this astonishing result may be based on a different origin of the products. Samples from products sold in open displays were predominantly from small local butcheries. Pre-packed products, on the other hand, were primarily sampled in supermarkets. However, no data are available on the patterns of antibiotics use in poultry farms delivering to small butcheries or to supermarkets.

It is known that poultry from free range housing systems have a higher risk of being infected with Campylobacter spp. than poultry from conventional housing systems 26. However, there was no significant association between these systems and the occurrence of antibiotic resistance in the isolated strains. While in bivariate analysis, both "conventional production vs. 'animal-friendly' production" and "foreign production vs. Swiss production" were significant risk factors for multiple resistance, only "foreign production vs. Swiss production" remained significant in the logistic regression model. This means that the univariate results were caused by confounding as the majority of 'animal-friendly' samples was produced in Switzerland.

In this study, the overall prevalence of ciprofloxacin resistance in Campylobacter strains from raw poultry meat (29%) was higher than the prevalence of ciprofloxacin resistance in Campylobacter strains from poultry meat produced in Switzerland (19%) and foreign versus Swiss production was a significant risk factor for multiple resistant strains in the logistic regression model. Therefore, a national monitoring program for antibiotic resistance needs to include both domestically produced and imported meat.

None declared.

UL performed the statistical analysis and drafted the manuscript. GR participated in the study design, coordinated the collaboration with the cantonal laboratories and supervised data analysis. RS was responsible for isolation and susceptibility testing of Campylobacter species. JD and BB participated in the design of the study. JD initiated the collaboration with the cantonal laboratories, and consulted on all practical aspects of sample collection. KS conceived of the project, and provided valuable scientific input during the design, implementation and data analysis phase. All authors read, commented on and approved of the final manuscript.