Specimens such as stool, rectal swab, and nose swab were collected by local health authorities from patients and food handlers involved in the food-borne disease outbreaks and sent to the laboratories of the Center for Disease Control, where bacterial pathogen examinations were conducted. Briefly, the specimens were streaked in selective differential media described for Vibrio cholerae and V. parahaemolyticus (7), Salmonella spp. (16), Staphylococcus aureus (5), Bacillus cereus (13), Shigella spp. (14), Aeromonas spp. (15), and Escherichia coli O157:H7 (6). Suspected colonies were initially identified by biochemical tests with triple sugar iron agar (Eiken Chemical Co., Tokyo, Japan), SIM medium (Eiken), and lysine iron agar (Difco Laboratories, Detroit, Mich.). For screening E. coli O157:H7, suspected E. coli colonies were tested with Clig agar slants (Kyokuto, Tokyo, Japan) instead of triple sugar iron agar. After the screening tests, suspected isolates were further identified with API biochemical test kits (bioMérieux, l'Etoile, France), i.e., API 20E for V. cholerae, V. parahaemolyticus, Salmonella spp., Shigella spp., and Aeromonas spp., API ID32 Staph for S. aureus, and API 50CH for B. cereus. Serologic tests for identification and serotyping were performed routinely with the antisera of O antigens and H antigens for E. coli O157:H7 (Difco) and Salmonella spp. (Denka, Seiken Corp., Tokyo, Japan), the antisera of O1 and O139 antigens for V. cholerae (Denka), the antisera of O antigens for Shigella spp. (Denka), the antisera of K antigens for V. parahaemolyticus (Denka), and the antisera of enterotoxins (SEA to SEE) for S. aureus (Denka). Some isolates of serovars K6, K8, K10, K12, K63, and K68 of V. parahaemolyticus were also serotyped with antisera of O antigens (Denka). Laboratory procedures of the Centers for Disease Control and Prevention, Atlanta, Ga., were followed for examination of Clostridium botulinum (2).

Epidemiological data relating to the food-borne disease outbreaks investigated were obtained from standardized case report forms filled in by the county public health authorities. These reports included the dates, times, and locations of suspected food ingestion, the suspected foods, the persons who consumed the food and were poisoned, and basic information about each patient's sex, age, date and time of onset, residency, clinical symptoms, and medical treatment.

The etiology of a food-borne disease outbreak was confirmed based on comparisons of the clinical symptoms, disease incubation period, implicated foods, and laboratory findings from the specimens from the implicated persons.

Table ​Table11 presents the etiologies of food-borne disease outbreaks occurring in Taiwan during the period from 1995 to 1999. A total of 850 outbreaks were reported, and specimens from these outbreaks were examined by the Center for Disease Control. Bacterial agents were responsible for 610 outbreaks (71.8%), with the remainder (28.2%) due to chemical or unknown agents. Among the bacterial agents, V. parahaemolyticus, Salmonella spp., and S. aureus accounted for 63.8, 5.2, and 2.5% of the total outbreaks, respectively. B. cereus, V. cholerae (non-O1, non-O139 strains), S. sonnei, and Aeromonas sp. combined caused only 1.5% (13) of all bacterial outbreaks. The outbreak caused by S. sonnei was imported. No E. coli O157:H7 and C. botulinum outbreaks were identified during this period.

Compared to 1995, outbreaks increased abruptly from 121 to 177 cases in 1996 and have remained high since this time. The highest incidence of disease outbreaks was in 1997, with a total of 242 outbreaks occurring. The number of outbreaks due to chemical and unknown agents continued to remain steady, ranging from 43 to 57, but outbreaks caused by bacterial agents fluctuated from 72 to 185. This fluctuation in bacterial outbreaks is related to the number of V. parahaemolyticus outbreaks. Outbreaks caused by this agent range from 54 (44.6%) to 171 (70.7%) over the period from 1995 to 1999. In contrast, six other bacterial agents, including Salmonella spp. and S. aureus, caused 13 to 20 outbreaks. These data indicate that the high incidence of food-borne disease outbreaks during 1996 through 1999 can be attributed to V. parahaemolyticus.

To investigate whether a particular V. parahaemolyticus serovar was involved in the sudden increase in food-borne disease outbreaks, the K serotypes of the recovered V. parahaemolyticus isolates were analyzed. Furthermore, some isolates of several major K serovars were also serotyped with antisera of O antigens. Table ​Table22 presents the prevalence of V. parahaemolyticus K serovars in food-borne disease outbreaks in Taiwan from 1995 to 1999 (inclusive). A total of 3,743 V. parahaemolyticus isolates were recovered and serotyped. Of these, 3,729 isolates were grouped into 40 K serovars and 14 isolates were nontypeable. K6 (O3:K6), K12 (O4:K12), K8 (O4:K8), K10 (O4:K10), and K63 (O3:K63) were ranked as the leading serovars, accounting for 82.6% of the total V. parahaemolyticus isolates. The O3:K6 serovar was the most prevalent over this time period. There were 2,234 isolates of O3:K6 identified in 351 of the 542 V. parahaemolyticus outbreaks over the period investigated. In 1995, O3:K6 was identified in a single outbreak with three isolates recovered, accounting for 0.6% of the V. parahaemolyticus identified in that year. However, in 1996, 372 O3:K6 isolates were identified in 66 outbreaks, accounting for 50.1% of the V. parahaemolyticus. The highest incidence of O3:K6 occurred in 1997 and accounted for 83.8% of the total V. parahaemlyticus isolates detected. The incidence of O3:K6 infections decreased in 1998, but this serovar still accounted for 71.5% of the V. parahaemolyticus and 61.3% in 1999. These data indicate that the prevalence of O3:K6 is correlated with the increase in total V. parahaemolyticus outbreaks in Tawian.

Serovars O4:K8 and O4:K10 caused 107 (19%) outbreaks over the study period. The number of these outbreaks caused by these two serovars showed no significant change during these 5 years. An O4:K68 serovar appeared for the first time in 1997 and had caused 22 outbreaks by 1999. This serovar is rarely implicated in food poisoning, but it had caused six outbreaks in Japan in 1998 (9). The future trend of V. parahaemolyticus O4:K68 infection must be carefully watched.

To determine the cause of the high incidence of food-borne disease outbreaks, we compared the outbreak profiles of the classified groups. That is, we compared the yearly total outbreaks with the agent responsible for that outbreak. The agents were classified into etiological groups, either chemical and unknown agents, or bacterial agents (V. parahaemolyticus, non-V. parahaemolyticus, O3:K6 , and non-O3:K6). Figure ​Figure11 depicts the profiles of the total outbreaks during the study period and also of the six etiology groups. The outbreak profiles of bacterial agents, V. parahaemolyticus, and O3:K6 coincide with the total outbreaks. In contrast, the outbreak profiles for all other groups showed no correlation with the total outbreak profiles. This profile analysis indicates that the O3:K6 V. parahaemolyticus was the cause of the unusually high incidence of food-borne disease outbreaks over the period from 1996 to 1999.

Figure ​Figure22 depicts the numbers of monthly food-borne disease outbreaks caused by the V. parahaemolyticus O3:K6 serovar, mixed serovars (consisting of the O3:K6 serovar and other non-O3:K6 serovars), and non-O3:K6 serovars and compares these with the average temperature in Taiwan. V. parahaemolyticus outbreaks were high between May and September for each year investigated. However, frequent outbreaks caused by V. parahaemolyticus were also reported in April and October. The occurrence of V. parahaemolyticus food-borne disease outbreaks can be correlated to temperature. Outbreaks tend to be more prevalent in late spring to early autumn and less prevalent in the cooler months of winter. Figure ​Figure22 also shows that the first V. parahaemolyticus O3:K6 serovar was identified in October 1995. The same serovar was identified again in April 1996 and has become more prevalent over time.

This study has investigated the causes behind a high incidence of food-borne disease outbreaks from 1996 to 1999 in Taiwan by analyzing the possible etiologic agents. We present data indicating that a single bacterial agent, V. parahaemolyticus, has caused 61 to 71% of the total food-borne disease outbreaks in Taiwan during a 5-year period. By using the technique of outbreak profile analysis based on etiology groups, we have demonstrated that infections with a single V. parahaemolyticus serovar, O3:K6, were responsible for the unusually high incidence of food-borne disease from 1996 to 1999.

Infections caused by V. parahaemolyticus are usually associated with diverse serovars, but this situation has altered in recent years. Okuda et al. (11) first reported that a new strain of V. parahaemolyticus (O3:K6 serovar) had emerged. This serovar first emerged in February 1996 in Calcutta, India, and was responsible for 50 to 80% of the V. parahaemolyticus infections during the surveillance period of February to August 1996. Soon, it appeared that the emergence of this new serovar was a pandemic. The serovar appeared in several Asian countries, including Taiwan, Bangladesh, Laos, Japan, Korea, and Thailand (8), as well as in North America (3, 4). In Japan, O3:K6 has been identified as the major serovar in V. parahaemolyticus food-borne disease outbreaks in recent years (9). This strain has recently also caused two outbreaks associated with the consumption of raw seafood in the United States (3, 4). In Taiwan, the O3:K6 serovar may have appeared as early as October 1995, and the incidence has increased since April 1996 (Fig. ​(Fig.2).2). Our data show that this serovar has caused 50.1 to 83.8% of the V. parahaemolyticus infections during the 1996-to-1999 period, in contrast to only 0.6% in 1995. In 1999, the O3:K6 serovar still accounted for 61.3% of the V. parahaemolyticus infections in Taiwan, indicating that the number of infections with this serovar is still high. The O3:K6 serovar is currently dominant in the Asian region and has also appeared on the North American continent (3, 4). The origins of this serovar are unknown, and it may have spread to other regions of the world. This suggests that an intensive monitoring program for this organism is needed globally to monitor its spread and help understand its origins.

The genome of numerous O3:K6 isolates has been studied by several molecular techniques, including the arbitrarily primed PCR method (8, 11), ribotyping (1), pulsed-field gel electrophoresis analysis (PFGE; 1, 17), and toxRS sequence analysis (8). These studies conclude that the O3:K6 strain which emerged recently in Asia and the United States actually belongs to a single clone. However, Bag et al. (1) reported that a certain degree of genomic reassortment has occurred among the O3:K6 strains. Molecular typing of O3:K6 isolates recovered in Taiwan and several other Asian countries has recently been conducted by Wong and colleagues (17). From this study, eight different but genetically closely related PFGE pulsotypes were identified. In accordance with other studies (1, 8), Wong et al. demonstrated that the newest O3:K6 strains which caused pandemics in many Asian countries, including Taiwan, originated from a common ancestor with minor genomic changes.