To identify patients, Wyoming health officials contacted area physicians and health officials in neighboring states. Interviews identified ill persons in Alpine, as well as ill persons from outside Wyoming who had attended a large family reunion in Alpine June 26-28. Patients were routinely interviewed about a wide variety of potential exposures, and stool samples were collected for laboratory confirmation. Environmental health inspectors in Wyoming conducted an assessment of retail food distribution in Alpine. This information was used to develop questionnaires for the subsequent cohort studies.

On July 11 and 12, we conducted a cohort study of persons from out-of-state families who had attended the family reunion in Alpine from June 26 to 29. Using a questionnaire administered by telephone, we asked reunion attendees about illness and exposure to certain foods and municipal water during their stay in Alpine. All family members were questioned individually except children too young to be interviewed; parents responded for these children. A case was defined as diarrhea (>3 loose stools in a 24-hour period) with onset after June 26 in a person attending the family reunion.

To confirm the results of the smaller family reunion cohort and to estimate the attack rate in the community, we conducted a cohort study of all residents of Alpine during July 13-16 . Using a questionnaire administered by telephone, we asked residents about recent gastrointestinal symptoms and exposure to various meats and Alpine municipal water during the 7-day period June 25 to July 1. We collected information on consumption of tap water or beverages made with tap water, the average number of glasses drunk in the week, and water filtration practices. Telephone numbers were called at least twice (once during the day and once at night) before being deemed a “no answer.” A case was defined as a) a stool culture yielding E. coli O157:H7, or b) diarrhea (>3 stools in a 24-hour period) with onset after June 25 in an Alpine resident who was in town between June 25 and July 1.

All stool samples were sent on ice to the Wyoming state laboratory, where they were plated on sorbitol-MacConkey agar (5). Samples producing sorbitol-negative colonies were selected and tested for O157 antigen, and positive samples were then assayed for H7 antigen by latex agglutination (6). Specimens identified as E. coli O157:H7 were sent to the Utah Department of Health State Laboratory for subtyping by pulsed-field gel electrophoresis (PFGE) (7). Additional stool samples were collected serially from day-care attendees, food handlers, and health-care workers whose first stool culture yielded E. coli O157:H7; subsequent stool samples were tested for Shiga toxin by using only the Premier EHEC assay (Meridian Diagnostics Inc., Cincinnati, OH) (8). Serum was collected from town residents to test for immunoglobulin (Ig) G and IgM antibodies to O157 lipopolysaccharide by enzyme-linked immunosorbent assay (ELISA) at CDC (9). An antibody >1:320 for IgM or >1:160 for IgG was considered positive.

Representatives from the Wyoming State Department of Agriculture and the U. S. Environmental Protection Agency (EPA) inspected the Alpine municipal water system and reviewed testing and safety measures. On July 14, 2001, water samples were collected from the outlet of the storage tank and several points in the distribution line. Deer and elk fecal samples were collected in the area above the spring collection system. Water samples were sent overnight to the Cincinnati EPA laboratory and analyzed for bacterial contamination within 24 hours of collection. Water samples were tested for total and fecal coliforms and E. coli by using m-ENDO Medium (10). Animal fecal samples were tested for E. coli O157:H7 at the Wyoming state laboratory as described above for human samples.

In the univariate analysis, relative risks (RR) and 95% confidence intervals (CI) were computed by using Epi Info (version 6.04, CDC, Stone Mountain, GA). Variables found to be significantly associated with disease at the univariate level were examined in a logistic regression model (LogXact version 2, Cytel Software Corporation, Cambridge, MA). The minimum duration of fecal shedding was estimated as the time between the onset of symptoms to the last positive Shiga toxin result before two negative results. Data for children and adults were examined by using Kaplan-Meier survival analysis; curves were compared by using the log-rank test (SAS version 6.12, SAS Institute Inc., Cary, NY).

We identified 157 ill persons from 15 states. The dates of illness onset ranged from June 25 to July 27 for patients with culture-confirmed infection and June 22 to July 30 for patients with culture-negative stools (Figure 1). Hemolytic uremic syndrome developed in four persons: a 15-month-old girl, a 17-month-old boy, a 4-year-old boy, and a 36-year-old woman. All have since recovered. Fifty-seven isolates were subtyped by PFGE; the patterns were indistinguishable.

Review of initial interviews showed that ill persons reported only infrequent exposure to some food and drink items known to be associated with E. coli O157:H7 transmission, including raw milk and apple cider. Wyoming state environmental health inspectors determined that no recently recalled food products, including ground beef, had been shipped to stores in Wyoming. This information was verified by U.S. Food and Drug Administration and U.S. Department of Agriculture officials.

Twelve families consisting of 44 persons attended the family reunion in Alpine. Of the 41 (93%) persons interviewed, 12 met the case definition; one patient's stool sample yielded E. coli O157:H7. Initial interviews showed that all food and drink items, except water, were purchased out of state.

The overall attack rate was 29%: 33% among males, 24% among females, and 50% among children <5 years old. The only exposure significantly associated with illness was drinking Alpine municipal water. Persons who drank Alpine municipal water were nine times more likely to become ill than were persons who did not drink the water (95% CI=1.3-63.3). The attack rate among reunion members who drank Alpine municipal water during the weekend was 50%. Attack rates for persons who drank Alpine municipal water did not vary by age (14) (Table 1)

The 1997-98 Alpine telephone directory listed 490 phone numbers with the Alpine prefix; of the 287 household numbers, we reached 146 (51%). Twelve of the households reached were excluded because the family had been out of town during the exposure period, June 25 to July 1. A total of 319 persons were living in the remaining 134 households during the exposure period; 284 (89%) were available for interview. Of the 284 respondents, 139 (49%) were male, and the median age was 39 years (range <1 to 92). For 192 (68%) interviewees, Alpine municipal water was their source of tap water at home.

Fifty-four (19%) persons met the case definition; 18 (33%) had stool samples yielding E. coli O157:H7. The attack rate was almost two times higher in women (25%) than in men (13%); however, in the week before illness, women were no more likely than men to drink any or more water.

In univariate analysis, 5 of 16 exposure variables were significantly associated with illness (except when noted, all exposures refer to the period June 25 through July 1): playing in a sprinkler, playing with water guns, being serviced by municipal water at home, drinking municipal water, and drinking municipal water during June 26 to 28 (Table 2). Since drinking Alpine municipal water was a broader exposure than, and not as strong a risk factor for illness as, drinking municipal water during June 26 to 28, it was excluded in multivariate analysis. In the multivariate model, the only variable that remained significantly associated with illness was drinking municipal water on the weekend of June 26 to 28 (Table 3; OR 10.1; 95% CI=1.8-56.4).

Overall, the attack rates among residents who drank Alpine municipal water were 23% during June 25 to July 1 and 27% during the 3-day period of June 26-28. Attack rates among females or males for persons who drank Alpine municipal water did not differ significantly. However, although not significant, the risk for illness decreased with increasing age (Table 1). Among persons who drank municipal water, the average number of glasses of water and the risk for illness were not statistically associated (Kruskal-Wallis test, p=0.7).

Twenty persons with culture-confirmed infection were followed with serial stool cultures to examine the duration of fecal shedding; 14 were children <10 years old in day care and 6 were adults aged 16 to 60 years (including 4 adult food handlers and 1 health-care worker). The median length of fecal shedding was 9 days among children (range 1 to 35) and 7.5 days among adults (range 2 to 36), a difference that was not statistically significant (p=0.63). Among children <5 years old, the median length of fecal shedding was 9 days (range 1 to 35).

Serum samples were collected from 129 town residents 16 to 27 days after the most likely time of exposure, July 27. Of the 129 persons, 57 (44%) were male, and the mean age was 42 years (range 3 to 78 years). Twenty-six (20%) persons had illness that met the Alpine cohort case definition (all drank Alpine municipal water), and nine of these were either culture positive or had bloody diarrhea. Of the other 103 (80%) well persons, 95 drank municipal water and 8 did not. IgM and IgG geometric mean titers were highest in persons with culture- positive E. coli O157:H7 or bloody diarrhea, followed by persons with diarrhea, and then well persons (Figure 2). The geometric mean IgM titer was <1:320 and the IgG titer >1:160 for ill persons. Three of eight well persons who did not drink Alpine municipal water during the outbreak had IgG titers >1:160.

The Alpine municipal water system was supplied by an underground spring. A series of small boxes connected by perforated pipes 7 to 10 feet below ground level collected water from an unconfined aquifer and routed it to an underground concrete storage tank. Pipes feeding off this tank then delivered unchlorinated water to the town. Sanitary surveys conducted in 1992 and 1997 indicated a potential risk of contamination from wildlife as well as surface water. Although Alpine was in compliance with the Total Coliform Rule, which requires one safe total coliform result each month, there were several positive readings in April 1998 (1/5 positive), May 1998 (4/7 positive), and June 1998 (2/10) just before the outbreak. On inspection after the outbreak, the spring was found to be under the influence of surface water; a large pool of water was found in the area over the water collection pipes, probably the result of a late snow melt combined with heavy rains and ground water outfalls. Numerous deer and elk feces were present in the area; the animals came to the pool to drink. E. coli O157:H7 was not isolated from any of the five deer or elk fecal samples taken on July 14. Water taken from the storage tank on July 14 yielded a total coliform count of 108 CFU/100 mL. Enterococcus faecium was isolated from the same water sample, indicating fecal contamination. E. coli O157:H7 was not isolated.