In July 2008, four teaching hospitals in northern Taiwan consecutively reported to the Department of Health of Taipei City Government (DHTCG) and the Centers for Disease Control (CDC) 8 patients in 2 groups in whom fever, myalgias, and eosinophilia of unknown cause developed after they shared a common food source in May 2008. Group A, comprising 20 Taiwanese, participated in a festive meal in Taipei City at a Japanese food restaurant, and were served raw meat, blood, liver, and eggs of 3 of the 5 soft-shelled turtles provided by the host, a supplier of soft-shelled turtles. The other 2 soft-shelled turtles were refrigerated at 4°C and served at the same restaurant to a group of 3 Japanese customers (group B) 6 days later.

DHTCG and Taiwan CDC jointly investigated this outbreak. The restaurant had never previously served raw or undercooked soft-shelled turtles. Restaurant patrons other than those in groups A and B did not eat raw or undercooked soft-shelled turtles. Five of the 20 Taiwanese in group A and the 3 Japanese in group B exhibited signs and symptoms 1–3 weeks after eating at the restaurant and were defined as case-patients (Table 1). The 15 asymptomatic persons were defined as controls.

The most common symptoms were myalgia (88%), fever (88%), malaise (63%), and periorbital swelling (38%). Seven case-patients whose blood was analyzed all had eosinophilia and increased serum creatine phosphokinase and alanine aminotransferase levels. Five case-patients were hospitalized. Two patients underwent extensive serologic testing for helminths by ELISA, including tests for Dirofilaria immitis, Toxocara canis, Ascaris lumbricoides, Anisakis spp., Gnathostoma spinigerum, Strongyloides stercoralis, Paragonimus westermanii, Paragonimus miyazakii, Fasciola hepatica, Clonorchis sinensis, Spirometra erinacei, Taenia solium, and Trichinella spp. Both patients had weakly positive results for A. lumbricoides, G. spinigerum, and S. stercoralis and strongly reactive results for Trichinella spp.

Serum samples from 5 patients during the acute phase (3–5 weeks postexposure) and from all 8 patients during the convalescent phase (7–9 weeks postexposure) were sent to the Department of Parasitology, Gifu University, Gifu, Japan, for Trichinella serologic diagnosis, with ELISA and immunohistochemical staining. Of the 15 controls, none consented to give a blood sample. Briefly, the ELISA microtiter plates were sensitized with excretory-secretory (ES) antigen from T. spiralis or T. pseudospiralis, probed with a diluted human serum sample (1:200–1:6,400), and incubated with 100 μL of 1:10,000-diluted goat antihuman immunoglobulin G (Fab specific) peroxidase-conjugate (Sigma Chemical Co., St. Louis, MO, USA). Absorbance at 414 nm was monitored with a plate reader. All samples were analyzed in duplicate. The cutoff point was 3× the mean values of the A₄₁₄ for the negative controls. Immunohistochemical staining was performed by incubating skeletal muscle tissues from T. spiralis–infected mice with the serum specimens (1:200 dilution) for 1 h at 37°C and processing the sections with the HistoStain SP kit (Zymed Laboratories Inc., San Francisco, CA, USA).

In both the acute and convalescent phases, all serum samples reacted to T. spiralis and T. pseudospiralis ES antigen and were positive in immunohistochemical staining (Table 1). The diagnosis of trichinosis was confirmed. Mebendazole or albendazole was prescribed for all patients, and their symptoms gradually resolved.

We conducted semistructured interviews with the 8 case-patients and 15 controls in both groups to learn which food items they had eaten at the restaurant. None had eaten raw or undercooked soft-shelled turtles before this outbreak (Table 2). In univariate analysis, consumption of raw soft-shelled turtle meat was strongly associated with infection (p = 0.003). Trichinosis developed in 8 (62%) of the 13 persons who ate raw soft-shelled turtle meat.

We performed an environmental study of the restaurant and the soft-shelled turtle farm. No leftover food was available from the restaurant for analysis. The soft-shelled turtles were bred artificially and hatched on a farm in Taiwan. They were fed only indigenous fish and shellfish. The farm used neither imported feed nor feed containing any mammals or reptiles. Microscopic inspection, with a meat-digesting method, of the soft-shelled turtles obtained from the farm 2 months after the outbreak did not show Trichinella spp. After the investigation, Taiwan CDC issued a press release to describe the outbreak and alert the public of the risk for trichinosis from eating raw or undercooked soft-shelled turtles.

The incubation period, clinical features, and laboratory findings in this outbreak are similar to those of other reported trichinosis outbreaks associated with eating mammals (11,12). T. papuae and T. zimbabwensis are the most likely parasites causing this outbreak because of their ability to infect mammals and reptiles (13). The ELISA method has limited specificity because of cross-reactions with non-Trichinella helminths (14). Moreover, because of similar antigen patterns among all Trichinella spp., the antigens prepared with 1 species can be used to detect specific antibodies in patients infected with any species (1). Therefore, although we detected strongly reactive antibodies to T. spiralis and T. pseudospiralis, we could not determine the etiologic Trichinella sp. in this outbreak without parasitic diagnosis.

A recent study demonstrated that the 53-kDa recombinant proteins in larval ES products could provide species-specific antibody responses in Trichinella-infected mice (15). We assessed the absorbance at 414 nm with a 1:200-diluted serum sample in our patients by using the 53-kDa recombinant proteins expressed from 5 Trichinella species (T. spiralis, T. britovi, T. nativa, T. pseudospiralis, and T. papuae). Our preliminary results showed that convalescent-phase serum specimens from 6 of the 8 case-patients reacted most strongly to the 53-kDa recombinant protein of T. papuae. Although application of this method in species-specific human diagnosis requires further studies, the data suggest our patients were likely to be infected with T. papuae. Because we have not yet determined how soft-shelled turtles were infected by T. papuae in this outbreak, further investigation of the potential infectious source is warranted.

Persons in many parts of the world typically consume raw or uncooked reptile meat. Further investigations are urgently needed to assess the epidemiology of reptile trichinosis and the human risk for trichinosis from reptiles.