Yersinia enterocolitica O:3 Infections in Infants and Children, Associated with the Household Preparation of Chitterlings

Between November 1 and December 13, 1988, eight cases of Y. enterocolitica gastroenteritis in black infants and children were reported to the Division of Public Health, Georgia Department of Human Resources, by a large county hospital and a university pediatric hospital in Atlanta. The county hospital had previously cultured stool specimens for Y. enterocolitica only at the request of a physician. After isolating Y. enterocolitica from one blood culture and two stool specimens from three children with diarrhea in November 1988, the county hospital began to culture all stool specimens for Y. enterocolitica on November 28. The university hospital had been routinely culturing stool specimens for yersinia since 1980. On December 5, 1988, the Georgia Department of Human Resources notified all hospitals in metropolitan Atlanta that several cases of yersiniosis had been detected in black infants and children, and advised pediatricians who saw patients with symptoms and signs compatible with yersiniosis to submit a stool sample for culture. Hospital laboratories were asked to report any case of culture-confirmed Y. enterocolitica infection.

To identify risk factors associated with the illness, we conducted a case–control study. A case was defined as a culture-confirmed Y. enterocolitica infection reported to the Georgia Department of Human Resources by an Atlanta hospital between November 1, 1988, and January 20, 1989. For each case patient, we selected two controls matched according to race, age within two weeks, and enrollment in the Women, Infants, and Children (WIC) Program, which provides food coupons to low-income mothers with infants or young children. We selected controls for the first cluster of eight cases from records of infants enrolled at the two county WIC offices serving the Atlanta area. We later learned that one WIC office only kept records of infants considered to be at higher risk for poor health. We selected controls for the subsequent cluster of cases from birth records at the case patient's hospital of birth to avoid this potential source of bias. We used a standardized questionnaire to collect information on exposure to potential risk factors in the case patients' households and matched control households in the 2 weeks before the patient became ill, and on gastrointestinal illnesses in household members in the 2 weeks before and 1 1/2 weeks after the case patient became ill. Stool specimens from members of the patients' households were not collected for culture. Interviews were conducted in person with the mother or guardian of each patient and control, as well as with other persons involved in child care or preparation of meals in the home. We obtained clinical information through interviews and by reviewing medical records. To detect associations between illness and categorical risk factors, we used an exact test described by Thomas and Gart.²² Conditional logistic regression was used to detect associations between illness and continuous risk factors. All analyses were performed in a matched fashion with a two-sided P value considered to indicate statistical significance at a level of 0.05 or less.

Y. enterocolitica was isolated from stool specimens by clinical laboratories on MacConkey or cefsulodin-Irgasan-novobiocin (CIN) agar, a selective differential medium.²³ Y. enterocolitica was isolated from chitterlings at the Enteric Diseases Branch of the CDC on CIN agar after a modified alkali treatment with potassium hydroxide^{24 , 25} and enrichment in phosphate-buffered saline or modified Rappaport broth.²⁶ Isolates were serotyped by the Yersinia Reference Laboratory of the CDC. An agar disk-diffusion test was used to determine resistance to chloramphenicol, cephalothin, tetracycline, trimethroprim—sulfamethoxazole, sulfisoxazole, nalidixic acid, ampicillin, carbenicillin, kanamycin, streptomycin, gentamicin, and trimethoprim.²⁷ Plasmid profiles were determined by a modified Birnboim procedure.²⁸ Phage types were determined by the International Yersinia Reference Center (Pasteur Institute, Paris). Each isolate was tested for the following in vitro markers of virulence: pyrazinamidase activity,²⁹ ability to bind Congo red dye and exhibit calcium-dependent growth at 37°C on Congo red—magnesium oxalate agar,³⁰ and ability to hydrolyze esculin and acidify salicin.³¹ To assess the national distribution of Y. enterocolitica serotypes, we compared information on clinical isolates of Y. enterocolitica submitted to the Yersinia Reference Laboratory of the CDC for identification and serotyping by state health department laboratories between 1986 and 1988 with previously reported data.²¹

Figure 1. Figure 1. Date of Onset of Illness in Patients with Culture-Confirmed Y. enterocolitica Infections in Atlanta from November 1, 1988, to January 10, 1989.

From November 1988 to January 1989, two clusters of Y. enterocolitica infections were detected in Atlanta: one around Thanksgiving and one around the Christmas and New Year's holidays (Fig. 1). Y. enterocolitica O:3 was isolated from 14 patients, and Y. enterocolitica O:1,2,3 from one patient. All the patients were black, and all but one were infants; the median age was 3 months (range, 17 days to 10 years). None of the patients were related, and they lived in various areas of the city. Eleven (73 percent) of the patients received federal assistance through the WIC Program.

The symptoms and signs reported for the 15 patients included diarrhea (100 percent), fever (100 percent), irritability (66 percent), small streaks of blood in the stools (60 percent), and vomiting (27 percent). In three patients, fever preceded the onset of diarrhea by a day. The illness lasted for a median of 11 days (range, 3 to 28 days), and seven patients were hospitalized. Y. enterocolitica was isolated from stool specimens from 14 patients, and from blood in one patient, a 17-day-old infant. Blood cultures performed on two other case infants were negative for Y. enterocolitica. No deaths occurred. In six infants under three months of age who were at two different hospitals and for whom hematologic information was available, the ratio of immature to total neutrophils was between 0.84 and 0.95 during the first 4 days of the illness in five infants and, in one infant, 11 days after the illness began.

Thirteen of the 15 patients were entered into the case–control study. Exposure data for two patients were incomplete because we were unable to contact the principal care provider. All 12 of the case patients who were infants and 23 of the 24 controls who were infants were fed commercial formula. The methods of formula preparation in the case and control households were similar. No brand or source of formula common to the case patients was identified, and no association was found between illness and the use of a specific WIC clinic or grocery store. In 12 of the 13 case households (92 percent), however, chitterlings had been prepared in the two weeks before the onset of the patient's illness, as compared with 5 of the 26 control households (19 percent; 95 percent confidence interval of the odds ratio, 4.96 to infinity; P = 0.00009). Illness and the preparation of chitterlings were still significantly associated when each of the two patient clusters was analyzed separately (first cluster: 95 percent confidence interval, 2.64 to infinity; P = 0.005; second cluster: 95 percent confidence interval, 1.59 to infinity; P = 0.033). No association was found between illness and the frequency with which other pork products were prepared in the household.

Chitterlings were prepared by boiling the large intestines of pigs after removing fat and particles of fecal matter. Case households prepared between 4.5 and 32 kg (median, 14 kg) of raw chitterlings over a period of one to four days. It took a median of five hours to clean the raw chitterlings, and in 11 (92 percent) of the 12 exposed housholds, this was done by a woman who was simultaneously caring for the case patient. Although none of the 12 case infants had direct contact with the raw chitterlings, five of the chitterling preparers in case households recalled specific instances of touching the infant while cleaning the chitterlings. None of the chitterling preparers in the five exposed control households recalled touching the infant. The mother of the only patient who was not an infant, a 10-year-old boy, reported that the boy had touched the raw chitterlings while she was cleaning them and had not washed his hands afterward. The median incubation period in cases with a single exposure to the preparation of chitterlings was 4.5 days (range, 1 to 14). In one household, the twin sister of the case infant had diarrhea three days after her twin became ill; otherwise, no other family members reported having a gastrointestinal illness in the 2 weeks before and the 1 1/2 weeks after the case patient became ill.

In 10 of the 12 exposed case households, the chitterlings were prepared for a Thanksgiving, Christmas, or New Year's meal. A large grocery-store chain in Atlanta reported that sales of chitterlings are largely restricted to the period from October through January, and peak in November. This grocery chain sold an estimated 110,000 kg of chitterlings in the Atlanta area during November 1988; the heaviest sales occurred in the three days before Thanksgiving. On the basis of these findings, an educational plan was developed describing the risk of transmitting yersiniosis to infants from raw chitterlings and the importance of good hygiene during the preparation of chitterlings. This information was incorporated into an existing nutritional-counseling program for mothers enrolled in the WIC Program. The months of October through January were targeted for more frequent use of a supplemental lesson plan.

Table 1. Table 1. Resistance to Antimicrobial Agents and Phage Types of Y. enterocolitica Strains Isolated from Patients and Chitterling Samples.

Y. enterocolitica O:3 was isolated from 3 of 4 unopened containers of chitterlings from one case household and from 5 of 11 containers purchased from Atlanta stores that had been the source of chitterlings for some case households. The eight containers harboring Y. enterocolitica O:3 had come from pigs slaughtered in the Midwest, South, and Northeast; three of the eight containers also yielded Y. enterocolitica O:1,2,3 or Y. enterocolitica O:5,27. All isolates expressed in vitro markers of virulence. Plasmid profiles of seven Y. enterocolitica O:3 isolates, four from patients and three from samples of chitterlings, were indistinguishable, with a single 44-megadalton plasmid. Isolates were generally resistant to ampicillin, carbenicillin, and cephalothin and sensitive to all other antimicrobial agents. Two phage types, 8 and 9b, were seen among the isolates of serotype O:3 (Table 1).

Table 2. Table 2. Distribution of Serotypes of Clinical Y. enterocolitica Isolates Submitted by State Health Departments to the Yersinia Reference Laboratory of the CDC for Identification from 1970 to 1980 and from 1986 to 1988.*

A review of clinical isolates of Y. enterocolitica submitted by state health department laboratories to the Yersinia Reference Laboratory of the CDC from 1970 to 1980 and from 1986 to 1988 showed a shift in the predominant serotype: Y. enterocolitica O:3 replaced Y. enterocolitica O:8 (Table 2). The 12 strains of Y. enterocolitica O:3 referred in the period from 1986 to 1988 had been isolated from patients in Georgia, Virginia, Pennsylvania, Ohio, Wisconsin, Iowa, South Dakota, North Dakota, Colorado, and Utah.

In countries in which Y. enterocolitica O:3 has become an important cause of diarrhea, the organism is commonly isolated from pork products, and pigs appear to be an important reservoir. Although Y. enterocolitica O:3 was rarely isolated in previous surveys of pigs and pork products in the United States,^{32 33 34 35} more than half the containers of chitterlings that we tested harbored Y. enterocolitica O:3. The existence of contaminated containers from several regions of the country suggests that this serotype has recently become widespread in the American swine population. The appearance of an animal reservoir and the occurrence of the outbreak reported here may signify the emergence of Y. enterocolitica O:3 as an important enteric pathogen in the United States.

Although infection was rarely reported in Canada before 1970,³⁶ by 1977 the isolation rate of Y. enterocolitica O:3 from Montreal children with diarrhea approached that of salmonella and was 2.7 times that of shigella.² Similar patterns of increase have been seen in many other industrialized countries in which Y. enterocolitica O:3 has become a major cause of diarrhea. In 1983, a number of sporadic infections with Y. enterocolitica O:3 occurred in New York City.³⁷ The source of these infections is unknown, but since that time, Y. enterocolitica O:3 has replaced Y. enterocolitica O:8 as the predominant clinical serotype in that area.³⁸ Since few reference laboratories serotype Y. enterocolitica, the national distribution of serotypes is unclear, but several pieces of evidence suggest that Y. enterocolitica O:3 has emerged in other parts of the country as well. Recent cases of sepsis due to Y. enterocolitica O:3 have been reported in Mississippi, Missouri, Texas, and Wisconsin after transfusions of blood from donors with asymptomatic bacteremia.³⁹ We found that in the past decade Y. enterocolitica O:3 replaced Y. enterocolitica O:8 as the predominant serotype of isolates submitted to the Yersinia Reference Laboratory of the CDC and that these isolates came from all over the United States.

Ingestion of contaminated foods, contact with sick pets, and transfusion of contaminated blood products have been implicated as possible modes of transmission of Y. enterocolitica.^{11 12 13 14 15 16 , 39} The outbreak described in the present report identifies a fourth mode of transmission that may be of particular importance in infants, who ingest a limited variety of foods. Although other household members may have been infected but remained asymptomatic, infants who had no direct contact with the raw chitterlings were the only persons in exposed households to become ill. These infants were probably exposed to infection by their caretakers, who were cleaning the chitterlings while caring for the infants. Similar observations have been made in outbreaks of salmonellosis due to contaminated turtles and marijuana.^{40 , 41} In these outbreaks, infants with little or no direct contact with the contaminated vehicle were often the only ones to become ill. This suggests that intrahousehold transmission of salmonella and Y. enterocolitica O:3 may share a common mechanism in which indirect exposure to contaminated vehicles has an important role.

Only 16 states require the reporting of yersiniosis, and only 2 states serotype isolates of Y. enterocolitica. Many serotypes of Y. enterocolitica and nontypable strains have not been shown to be human pathogens and are commonly found in the environment.^{35 , 42} Since only some serotypes are known to be pathogenic, the clinical relevance of Y. enterocolitica isolates cannot be interpreted without information on serotypes or markers of virulence. Continued surveillance, investigation of cases, and culturing of stool specimens for Y. enterocolitica by clinical laboratories, coupled with the widespread ability of reference laboratories to serotype clinical isolates, will be critical to an understanding of the changing epidemiology of yersiniosis in this country.