Methods

In April 1999 the California State Microbial Diseases Laboratory (MDL) noted an increase in Salmonella Thompson cases in Southern California. At the same time, Los Angeles County Department of Health Services identified an outbreak of Salmonella Thompson among patrons of one restaurant. Although a comparison study of the Los Angeles cluster could not be completed because of the high attack rate, all ill patrons had eaten a condiment mix made of onions, tomatoes, cilantro, and cabbage

Serotyping of all Salmonella isolates is mandatory in California, and all records of serotypes are maintained at the MDL. We reviewed MDL records to identify cases. A confirmed case was defined as Salmonella Thompson infection identified in California in March 1999. Local health departments collected demographic information on case patients, using a standard form

To determine possible risk factors for the outbreak, a matched case-control study was done. A case was defined as Salmonella Thompson infection in a resident of Southern California, with onset in March 1999. Case patients were excluded from the study if they were identified after the study ended, were <18 years old, were non-English speaking, could not be reached by telephone, or refused to participate. One case patient from the restaurant-associated outbreak was included in the study

We conducted exploratory interviews with 5 case patients, using a standard questionnaire that listed >75 food items. The final case-control questionnaire included 20 food items mentioned by at least 3 of these 5 case patients; the 5 case patients with exploratory interviews were included in the study. Control subjects were age group– and telephone exchange–matched to case patients. The age groups were 18–29, 30–49, and 50–69 years

A study investigator initially contacted all case patients to confirm their telephone numbers and to answer any questions about the outbreak. The case patients and control subjects were then interviewed about food exposures by a unit that specializes in telephone interviews. Case patients were asked about foods eaten during the week before onset of illness, and control subjects were asked about foods eaten during the week before the corresponding case patient’s date of onset of illness. To obtain control subjects, telephone numbers were randomly generated, on the basis of each case patient’s area code and prefix. Respondents were queried until an appropriate age match was identified. Potential control subjects who had diarrhea during March were excluded. We obtained at least 2 control subjects per case patient

Epi Info 6.04 was used for the analysis [5]. In a matched-set analysis, we calculated Mantel-Haenszel matched odds ratios (MORs) and 95% confidence intervals (CIs) for the MORs. In an age-group–stratified analysis, we generated adjusted odds ratios (ORs) and 95% CIs for the ORs

We did a traceback investigation of the source of cilantro served at restaurants with clusters of case patients and at restaurants where case patients ate cilantro but not salsa. We reviewed restaurant and distributor invoices to trace cilantro to specific farms. Local health departments surveyed restaurants where case patients ate fresh salsa or cilantro about the type of service provided (table vs. counter), the cutting and storage of cilantro, and the storage of freshly made salsa

We tested 14 outbreak-associated and 14 background specimens (temporally and geographically distinct from outbreak-associated strains) by pulsed-field gel electrophoresis (PFGE) with XbaI. We also tested 7 outbreak-associated specimens with SpeI and AvrII

To investigate the growth of Salmonella Thompson on cilantro, we immersed 4 bundles of cilantro leaves, which were purchased from a grocery store, in a suspension (2×10⁵ cells/mL) of an outbreak strain of Salmonella Thompson. Immediately after inoculation, we chopped half of the cilantro leaves with sterile scissors and placed them in glass petri dishes. Half of the chopped cilantro was incubated at 2°C. The other half of the chopped cilantro and the unchopped cilantro were each incubated at 26°C

To estimate the population of Salmonella Thompson present on the cilantro, 7 leaves (for the unchopped cilantro studies) or ∼100 mg of cilantro (for the chopped cilantro studies) were placed in 10 mL of 10 mM potassium phosphate buffer (pH 7), sonicated in a Fisher Ultrasonic bath for 1 min, and mixed in a vortex for 30 s. A small volume of an undiluted and a 10-fold diluted suspension were plated onto Salmonella and Shigella agar to distinguish salmonellae from the natural bacterial flora of cilantro. The plates were incubated overnight at 37°C, and colonies typical of Salmonella species were enumerated. Salmonella populations were estimated immediately after inoculation and 1, 2, and 3 days after inoculation. We also tested Salmonella growth in freshly made salsa (consisting of [wt/wt] 1 cilantro, 4.5 onions, 21 tomatoes, and 14.5 lime juice [pH 3.4]) stored at 4°C and 24°C

Results

We identified 35 “sporadic” cases of Salmonella Thompson that were independent of the Los Angeles restaurant-associated outbreak. Onset of illness was between 6 March and 31 March 1999. In contrast, fewer than 9 cases of Salmonella Thompson occur monthly in California. Nineteen (54%) of the 35 case patients were women. The case patients were 1–66 years of age (median, 29 years). All but 2 case patients resided in Southern California, including Los Angeles. Three case patients were hospitalized; there were no deaths

We enrolled 19 case patients and 43 matched control subjects in the case-control study. Ten case patients (53%) and 10 control subjects (24%) reported eating salsa at a restaurant (MOR, 4.8; 95% CI, 1.1–19.9). On the basis of interviews with case patients and restaurant managers, we determined that restaurant-made salsa usually contained tomatoes, onions, and cilantro. We then examined exposure to each of these ingredients eaten individually and in fresh salsa. In the matched-set analysis, there was a trend toward an association with eating cilantro (63% of case patients vs. 34% of control subjects; MOR, 4.5; 95% CI, 0.9–22.1) but not tomatoes or onions in a restaurant (table 1). In the age-group–stratified analysis, illness was significantly associated both with eating restaurant-made salsa (OR, 3.5; 95% CI, 1.1–11.0) and with eating any cilantro in a restaurant (OR, 3.5; 95% CI, 1.1–11.4). There was no association between illness and eating onions or tomatoes in a restaurant (table 1). Two patients ate cilantro but not salsa in a restaurant

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Mean growth of Salmonella serotype Thompson on whole (dashed line) and chopped (solid line) cilantro leaves. The whole cilantro leaves were incubated at 26°C (solid diamonds) and the chopped leaves were incubated at both 26°C (solid squares) and 2°C (open circles) The concentration of Salmonella Thompson on soft rotten, chopped cilantro leaves 2 days after inoculation and incubation at 26°C is designated by the solid triangle. Error bars indicate SE

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Salmonella serotype Thompson outbreak, California, 1999: selected results comparing case patients and community control subjects

In addition to the restaurant associated with the large outbreak in Los Angeles (restaurant A), we identified 2 other restaurants (restaurants B and C) with clusters of cases. Two case patients had eaten salsa at restaurant B. Four case patients had eaten at restaurant C during the week before onset of illness but did not report eating salsa or cilantro at this restaurant; however, restaurant C does make salsa with fresh cilantro

We attempted to trace the cilantro served by restaurants A, B, and C and by 2 additional restaurants where case patients had eaten cilantro but not salsa (restaurants D and E). A common grower of cilantro (grower M) was found for restaurants A and B. Restaurants A and E had 3 suppliers in common, and 1 of these suppliers bought cilantro grown in the same region as that bought from grower M. Restaurants C and D shared a common supplier; however, the growers could not be ascertained, because of the lack of records. Determining if there was a grower in common to all 5 restaurants was impossible because of poor record keeping; thus, no farm investigations were done

Six of the 10 restaurants surveyed served patrons at a table rather than a counter. All restaurants washed cilantro in water and chopped it with a knife. Seven restaurants used the chopped cilantro within 2 days; 1 reported use within a maximum of 2–3 days, 1 reported use after “at most 4 days,” and 1 said that cilantro “can last up to a week.” All restaurants reported storing whole-leaf cilantro, chopped cilantro, and freshly prepared salsa under refrigerated conditions

PFGE results for the 14 outbreak specimens and the 14 background strains produced the same 3 profiles, which differed by only 1 or 2 bands. Testing with SpeI and AvrII did not increase diversity; thus, we could not identify a specific outbreak strain

In the inoculation studies, Salmonella Thompson grew on the whole (unchopped) cilantro leaves held at room temperature, from a mean population of 4.13 log₁₀ cfu per gram of leaf to 5.18 log₁₀ cfu per gram of leaf, within 2 days. In contrast, Salmonella Thompson on chopped leaves incubated at 26°C reached a mean population level of 7.10 log₁₀ cfu per gram of leaf within only 1 day (figure 1). Because the chopped cilantro maintained at 26°C showed signs of spoilage 1 day after inoculation, the healthy leaf tissue was separated from the rotten tissue, and, 2 days after inoculation, Salmonella populations were estimated for both the rotten and healthy leaves. The spoiled chopped leaves had a mean Salmonella Thompson population of 8.45 log₁₀ cfu per gram of leaf, whereas there was minimal further increase on the healthy leaves (figure 1). The chopped cilantro incubated at 2°C showed no meaningful change in Salmonella Thompson concentration during the 3 days of observation. Salmonella Thompson concentrations increased 300-fold in freshly made salsa within 1 day of storage at 24°C, but we detected no growth when it was stored at 4°C

Discussion

We found a strong association between infection with Salmonella Thompson and eating fresh restaurant-made salsa. Freshly made salsa contains cilantro, tomatoes, and onions, but we found an association only with the cilantro. Cilantro was also an ingredient in the condiment mix consumed by all case patients in the restaurant-associated outbreak. The limited duration of this outbreak is typical of outbreaks due to fresh produce

The inoculation studies indicate that refrigeration can limit growth of Salmonella species on cilantro for at least 3 days and in freshly made salsa for at least 1 day. Despite a low pH, salsa will support Salmonella growth when stored at room temperature. The increased growth of Salmonella Thompson on chopped cilantro (compared with whole cilantro) stored at room temperature most likely results from the release of nutrients from the broken plant tissue. This high growth rate is similar to that of Salmonella Thompson in glucose-containing minimal medium (data not shown). Similarly, Shigella sonnei grew more rapidly on chopped parsley than on whole-leaf parsley [6]

The higher growth rate of Salmonella Thompson on spoiled chopped leaves may be due to increased nutrients leaking out of the diseased plant tissue. Similarly, a recent survey found that Salmonella contamination on produce with soft rot was almost twice as high as that on healthy produce [7]

Our traceback investigation was hampered by poor record keeping. However, 3 of the 5 restaurants received cilantro from growers in the same region of Mexico. The other 2 restaurants shared a common distributor. Unfortunately, inadequate record keeping precluded a complete trace, and no environmental investigations of the farms were done. Although cilantro has been rarely associated with enteric illness, cilantro from Mexico was recently associated with an outbreak of shigellosis in California [8]

To prevent outbreaks from raw produce, growers, packers, and shippers should adopt good agricultural practices, as detailed in industry guidelines provided by the US Food and Drug Administration, US Department of Agriculture, and Centers for Disease Control and Prevention [9]. On the farm, efforts should be made to prevent contamination of water and produce by humans and animals. At the packing shed, potable water should be used in washing produce. Chlorinated wash water could also be used to decrease contamination [10]. At all levels, workers should maintain good hygiene. Diseased and decaying produce should be removed during harvesting, processing, packing, and distribution and at retail sale. To identify specific production practices that could contribute to contamination of produce, traceback investigations are essential. Accurate records of the distribution of raw produce should be in place at all times. Irradiation of produce should be considered as a preventive measure [11–13]

The public and the food service industry should be instructed to serve cilantro promptly after chopping it, to refrigerate chopped cilantro and freshly made salsa, and to avoid serving and eating damaged cilantro. The public, especially populations at risk for complications from enteric bacteria, should be informed about the potential risk of salmonellosis from eating raw cilantro