Results

Active case finding. We identified 16 cases of botulism in the greater Dallas—Fort Worth area; onset of symptoms occurred during the period of 25 August through 1 September 2001 (table 1 and figure 1). Fifteen patients were from the church cohort. An additional case of botulism occurred in a Fort Worth resident (hereafter referred to as the “outlier”) who had cranial neuropathy and a descending flaccid paralysis. He had not attended the church supper but had consumed a commercial chili product that was confirmed to contain botulinum toxin. Nine patients had laboratory-confirmed botulinum intoxication. There were no deaths. Neurologic symptoms included dysphagia (69% of patients), blurred vision (56%), slurred speech (50%), and double vision (38%). Gastrointestinal symptoms were abdominal pain (56% of patients), diarrhea (44%), nausea (31%), and vomiting (31%). Botulinum toxin was detected in the stool specimen of 1 asymptomatic person. Ten patients (63%) were hospitalized, 6 (38%) of whom required mechanical ventilatory support. Nine patients (56%) received bivalent AB botulinum antitoxin. No cases of botulism were detected in other parts of Texas or in neighboring states during August and September 2001.

Of the 15 symptomatic patients, 5 did not seek medical care; they were detected by the investigative team with the use of the outbreak questionnaire and the collection of stool specimens. Two of these 5 were confirmed to have botulinum toxin in stool specimens; one patient had diarrhea and abdominal cramps, and the other had mild diplopia. Two of the 3 patients who were identified using the outbreak questionnaire and who did not have laboratory-confirmed illness had nausea and diarrhea; the third had blurred vision and diarrhea.

Among the 10 patients who sought medical care, 6 had severe neurologic impairment, requiring mechanical ventilatory support. Two of the remaining 4 patients had multiple cranial nerve and gastrointestinal symptoms; the other 2 patients had gastrointestinal symptoms. The diagnosis of botulism was delayed in 4 persons who presented with classic neurologic symptoms. Two children received diagnoses of botulism >24 h after initial presentation to medical care; the early symptoms of one included dysarthria, dysphagia, weakness, and diarrhea, and the early symptoms of the other included blurred vision and diplopia. An adult patient from the church cohort was diagnosed with hypertensive urgency and was hospitalized for 5 days with blurry vision and dysphagia. In addition, the 17-year-old outlier sought treatment at hospital emergency departments 4 times over the course of 3 days with classic, progressive multiple cranial neuropathies and descending paralysis before botulism was diagnosed.

Among church supper attendees, we attempted to quantify the relationship between “dose,” defined as the amount of chili consumed, and the severity of illness, in addition to the association between the timing of chili consumption and the development of botulism. However, we were unable to obtain accurate information on how much chili was eaten from the 6 most severely affected patients, all of whom required mechanical ventilatory support and sedation.

Cohort study of church supper. Forty persons attended the church supper or ate leftovers from the supper. Questionnaires were completed for 38 persons, including 37 who attended the supper and 1 who did not attend but who ate leftovers. Fifteen persons (39%), including the asymptomatic patient, met the case definition for botulism. The supper menu was limited to fried chicken from a restaurant and a chili dish prepared at home from commercial products. Fifteen (50%) of the 30 persons who ate any food served at the supper became cases, compared with none of the 8 persons who ate no food (relative risk [RR], undefined; P = .011, by Fisher's exact test). Fifteen (63%) of 24 persons who ate the chili dish became cases, compared with none of the 14 persons who did not eat chili (RR, undefined; P < .001, by Fisher's exact test). The attack rate among those who ate chicken but not chili was 0% (0 of 6 persons). No significant association was detected between the amount of chili consumed and the severity of illness.

Preparation of implicated food. On 25 August, the Saturday afternoon of the church supper, a church congregant purchased four 5-lb (2.27-kg) plastic tubs of frozen chili, an unknown number each of “brand X” and “brand Y,” from a salvage store in Fort Worth. The congregant reported that she had prepared the chili dish by thawing a single frozen plastic tub of chili, but she could not remember which brand she used. She heated the contents of the plastic tub in a pot and removed the pot from the stove as soon as the chili began to boil. The congregant drove 5 min to the church kitchen, where she placed the pot on a burner set to low heat. Hot dogs and “brand Z” commercial chili from an aluminum can were later added to the pot. After the church supper, the remaining chili dish was taken to the cook's home, where several family members ate it over the subsequent 4 days.

One day before the onset of his symptoms, the 17-year-old outlier ate frozen brand X chili, which he had heated but not boiled. The patient reported to his mother that the chili appeared to be spoiled, with a fungus-like appearance in one section of the container. He proceeded to eat the chili despite the indication of spoilage and his mother's advice against consumption. The brand X chili had been purchased by his mother at the Fort Worth salvage store on 24 August, 1 day before the church congregant purchased brands X and Y chili at the same store.

Laboratory investigation. Botulinum toxin type A was detected by mouse bioassay in stool samples obtained from 9 of 14 symptomatic patients from the church cohort. Botulinum toxin type A was identified in the stool of 1 of 2 asymptomatic church supper attendees who submitted specimens. No toxin was identified in the outlier's stool sample. Toxin type A was also identified in the chili leftovers from the church supper and in an opened container of frozen brand X chili eaten by the outlier. An unopened container of frozen brand X chili obtained from the cook in the church cohort had a pH of 5.47. The concentration of botulinum toxin in the previously unopened brand X chili container was 10,000 mouse lethal doses/g and 160 ng of type A toxin/g of chili. C. botulinum organisms were isolated by culture of the food and were verified as type A toxin—producing organisms using ELISA.

Traceback and investigation of food source. The outlier's mother and the cook in the church cohort purchased brand X chili on Friday, 24 August, and Saturday, 25 August, respectively, at the same Fort Worth salvage store. The salvage store sold foods in conditions often rejected by standard grocery stores. Such products included products in dented cans, repackaged foods, and products that had passed their expiration date. Employees at the Fort Worth salvage store sorted through large volumes of damaged foods and selected items for sale to the public. Informal rules were followed to determine which foods could be sold. We found no evidence that standardized training or use of written manuals had guided the selection process.

During an unannounced outbreak-related inspection on a Saturday afternoon, gross mishandling of foods was observed. On Saturdays, the store routinely offered foods at discounts exceeding those of weekdays. Bulging cans, dented cans, and opened plastic containers of food were on display for sale. Foods that required refrigeration or freezing were displayed in carts at room temperature for customers to browse through and buy. Employees reported that the items displayed at ambient temperatures for >7 hours were routinely placed back in refrigerator storage for sale on later days. Other violations included the use of the warehouse area for retail sales and inadequate labeling of repackaged foods. The salvage store was regulated and had been routinely inspected by the TDH. Because routine inspections before the outbreak were usually conducted during weekdays, the mishandling of food observed during Saturday sales had not been detected previously.

Because there was a lack of invoices for products received at the salvage store, the distributors, the dates of receipt, and the condition on delivery of specific lots of frozen brands X and Y chili could not be determined. Because it is not a canned product, brand X chili did not undergo retort cooking or similar spore-killing processes. Inspections of the chili manufacturing plant did not reveal poor production practices or product mishandling. We did not detect botulinum toxin in 2 containers of brand X chili obtained from a first-line distributor. No other brand X chili containers from the implicated lot were found at other distributors or retailers. Brand X chili was distributed to Louisiana and Oklahoma; no cases of botulism were reported in either state in August and September 2001.