The study population in the present study consists of children under age 17 in northern Taiwan with loose, watery or bloody diarrhea and divides into 5 age-groups as follows: (1) < 4 years old (infant stage), (2) 4–7 (kindergarten stage), (3) 7–10 (elementary school stage 1), (4) 10–13 (elementary school stage 2) and (5) > 13 (junior high school stage and above). The fecal specimens were collected from more than 30 clinics and hospitals in northern Taiwan, including Taipei, Taoyuan, Hsin-Chu, Yilan, and Keelung. All the samples which were transported in Cary-Blair transport medium (BD Diagnostic Systems, Sparks, MD) or buffered glycerol saline (Creative Microbiologicals, Ltd., Taipei, Taiwan) were previously tested the presence of other enteric pathogens such as Shigella spp., Salmonella spp. and Vibrio spp. Specimens free of the above pathogens were chosen in our study. Therefore, mixed infection with multiple pathogens in this study has been excluded. A total of 894 clinical specimens from different patients gathered between December 2003 and February 2005 were chosen.

After arrival at our laboratory, the specimens were directly inoculated onto two prepared selective media (usually delay 1–7 days between stool collection and plating): modified cefoperazone charcoal deoxycholate agar (mCCDA, Creative Microbiologicals, Ltd., Taipei, Taiwan) and charcoal-containing selective medium (CSM, BD Diagnostic Systems, Sparks, MD), followed by incubation at 42°C for 48 hours in a microaerobic atmosphere containing 5% O₂, 10% CO₂, and 85% N₂ generated by CampyPak (BD Diagnostic Systems, Sparks, MD). If enrichment was needed to recover low numbers of organisms due to delayed transport to the laboratory, swabs or stools were first inoculated into brain heart infusion broth (BHI, Creative Microbiologicals, Ltd., Taipei, Taiwan) and Campylobacter enrichment broth, each of which contained cefoperazone (8 mg/L), amphotericin B (5 mg/L), teicoplanin, and 5% lysed horse blood, followed by incubation at 37°C for 48 hours. Finally, the broth suspensions were inoculated onto mCCDA or CSM agar for another 48 hours.

For further confirmation of Campylobacter, suspected colonies, which were morphologically smooth, gray, flat, and moist, were sub-cultured from each plate onto fresh blood agar plates. They were checked by Gram staining and several biochemical reactions including oxidase production, catalase production, and sodium hippurate hydrolysis. Oxidase- and catalase-positive colonies exhibiting a characteristic Gram stain appearance (e.g., Gram-negative, curved rods) could be reported as Campylobacter spp. Hydrolysis of sodium hippurate is the major identification test used for C. jejuni. Campylobacter species giving a positive hippurate hydrolysis result could be reported as C. jejuni, with no other additional tests required. Isolates giving a negative result for hippurate hydrolysis were further identified by PCR in order to distinguish some isolates of C. jejuni that express the hippuricase gene (hipO), but are negative for the hippurate hydrolysis phenotype from C. coli, another important species of Campylobacter that lacks hippuricase activity, as described below.

To confirm isolates that were negative for hippurate hydrolysis activity as C. jejuni or C. coli, conventional PCR was conducted. DNA was extracted from bacterial suspensions by boiling at 100°C for 15 min. Two primer pairs [17,18] CJF (5'-ACTTCTTTATTGCTTGCTGC-3'), CJR (5'-GCCACAACAAGTAAAGAAGC-3') and COL1 (5'-ATGAAAAAATATTTAGTTTTTGCA-3'), COL2 (5'-ATTTTATTATTTGTAGCAGCG-3') each were used to amplify the hipO gene from C. jejuni and the ceuE gene from C. coli, respectively. PCR was performed using conditions described, as previously reported [17,18].

To determine the susceptibility of Campylobacter isolates to antimicrobial agents, they were suspended in Mueller-Hinton broth (Creative Microbiologicals, Ltd., Taipei, Taiwan) to the concentration of McFarland No. 0.5 and bacterial cells were inoculated on Mueller-Hinton agar plates (Creative Microbiologicals, Ltd., Taipei, Taiwan) containing 5% sheep blood after incubation for 48 hours in a microaerobic atmosphere. Discs containing ampicillin (10 μg), chloramphenicol (30 μg), nalidixic acid (30 μg), ciprofloxacin (5 μg), cephalothin (30 mg), clindamycin (2 μg), erythromycin (15 μg), gentamicin (10 μg), streptomycin (10 μg), and tetracycline (30 μg) (Oxoid, Hampshire, UK) were placed on the inoculated plates. The plates were then incubated at 42°C for 48 hours in a microaerobic atmosphere. Interpretation of the susceptibility test, including susceptible (S), intermediate (I) and resistant (R), was carried out (Available from ) [19,20]. Staphylococcus aureus ATCC 29213, Escherichia coli ATCC 25922, and Pseudomonas aeruginosa ATCC 27853 were used as positive controls.

For serotyping of C. jejuni, a commercial antisera kit (Denka Seiken, Tokyo, Japan) designed to detect the heat-stable antigen was utilized by the passive hemagglutination method (PHA). For preparation of the sensitized bacterial antigen solution, the bacteria and 0.25 ml of each extraction solution provided in the kit were mixed. To prepare sensitized red blood cells for use in the PHA test, 0.5 ml of each bacterial antigen was incubated with fixed chick red blood cells at 37°C for 30 min. The PHA test was conducted by placing one drop of each antiserum and 25 μl of sensitized red blood cells into microplate wells and incubating for 30 min. Examination for agglutination in each well was utilized to determine the serotypes of the C. jejuni isolates.

The sentinel surveillance for diarrhea syndrome is one of the national disease surveillance systems established by Taiwan CDC. Sentinel physicians in individual hospitals, clinics and medical centers reported outpatients suspected of catching diarrhea and sent specimens for enteric bacterial pathogens identification. The present study was carried out through sentinel surveillance and reviewed by Taiwan CDC, which does not require oversight by an ethics committee.

To study the prevalence rate of Campylobacter in children in northern Taiwan, a total of 894 fecal specimens from different diarrheal patients were tested for Campylobacter. Sixty-one isolates (61/894, 6.8%) including 58 C. jejuni (58/61, 95.1%) and 3 C. coli (3/61, 4.9%) isolates were identified by direct plating, followed by biochemical reactions and PCR. Two selective culture media (mCCDA and CSM) were used and the recovery rates from both were the same (data not shown). During the 15 months of the study, the highest and lowest positive rates were observed in February 2005 (4/22, 18.2%) and August 2004 (1/82, 1.2%), respectively. Most isolates were recovered in winter (December to February, 25/225, 11.1%), followed by spring (March to May, 15/214, 7.0%), autumn (September to November, 11/225, 4.9%), and summer (June to August, 10/230, 4.3%). The positive rate of Campylobacter and its relationship with the average temperature during each month are shown in Table 1 and Figure 1, respectively.

Of the 61 Campylobacter isolates, 29 (47.5%) and 32 (52.5%) each were gathered from male and female children, respectively (Figure 2). The age of these 61 patients ranged from 10 months to 17 years old (Figure 2). We further divided them into 5 age-groups according to the school age (see Methods). Children in the < 4 group had the highest positive rate (28/61, 45.9%), followed by the 4–7 group (16/61, 26.2%), group 7–10 (9/61, 14.8%), group > 13 (7/61, 11.5%) and group 10–13 (1/61, 1.6%). Bloody diarrhea is a classical symptom in children with campylobacteriosis [21]. Of the infected children, 25 out of 61 (41%) had bloody diarrhea (Table 2). Most of these 25 children belonged to the < 4 age group (14/25, 56%), followed by group 4–7 (4/25, 16%), group 7–10 (3/25, 12%), group > 13 (3/25, 12%) and group 10–13 (1/25, 4%). Thirty-nine out of 61 (39/61, 63.9%) presented with fever, and most of these 39 children belonged to the < 4 age group (14/39, 35.9%), followed by group 4–7 (9/39, 23.1%), group 7–10 (9/39, 23.1%), group > 13 (6/39, 15.4%), and group 10–13 (1/39, 2.5%).

The Clinical and Laboratory Standards Institute (CLSI) has approved the agar dilution method as a standard susceptibility testing method for Campylobater. Due to the time-consuming procedures, it is rarely carried out in routine diagnostic laboratories. The easier and less expensive disc diffusion method represents a good correlation while comparing with agar-based dilution method and provides an alternative choice for antibiotic susceptibility test of Campylobacter [20,22]. Since the disc diffusion method was found to be a reliable and easy tool for monitoring the prevalence of Campylobacter strains, we applied this method for antimicrobial susceptibility test in this study.

The results of the susceptibility testing for 61 Campylobacter isolates are shown in Table 3. All C. jejuni and C. coli isolates in this study were resistant to cephalothin (100%), and most showed resistance to the four other drugs tested, including tetracycline (93.4%), nalidixic acid (91.8%), ciprofloxacin (90.2%), and ampicillin (85.5%). Five out of the ten antimicrobial agents tested, including streptomycin, chloramphenicol, gentamicin, clindamycin, and erythromycin, tended to result in a lower resistance rate as listed in Table 3. Cross-resistance between nalidixic acid and ciprofloxacin was found in all quinolone-resistant isolates.

For serotyping, 24 out of the 58 C. jejuni isolates (24/58, 41.4%) were typable and belonged to 9 different serotypes according to Penner's scheme (Table 4). Five of the above 9 serotypes contained only one isolate and the other four types contained two to eight isolates. The most prevalent serotypes among these isolates were Penner's type 2 (group B), followed by type 6,7 (group F) and type 11 (group J). Thirty-four (58.6%, 34/58) strains were defined as "untypable", including 14 (41.2%, 14/34) strains that failed to agglutinate in the presence of any of the antisera used in the present study and 20 that reacted with two or more types of antisera. Serotyping using Penner's scheme was sufficient for identification of about half of the human clinical strains.