MS : Received 28 March 2010/Accepted 5 July 2010 ABSTRACT

Transcription

1 1967 Journal of Food Protection, Vol. 73, No. 11, 2010, Pages Analysis of Tomato and Jalapeño and Serrano Pepper Imports into the United States from Mexico before and during a National Outbreakof Salmonella Serotype Saintpaul Infections in 2008 KARL C. KLONTZ, 1 * JOSHUA C. KLONTZ, 2 RAJAL K. MODY, 3 AND ROBERT M. HOEKSTRA 3 1 Office of Food Defense, Communications, and Emergency Response, HFS-005, Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, College Park, Maryland 20740; 2 Harvey Mudd College, 340 East Foothill Boulevard, Claremont, California 91711; and 3 Division of Foodborne, Bacterial, and Mycotic Diseases, Centers for Disease Control and Prevention, 1600 Clifton Road, Atlanta, Georgia 30333, USA MS : Received 28 March 2010/Accepted 5 July 2010 ABSTRACT Case-control studies conducted during a multistate outbreak of Salmonella enterica serotype Saintpaul infections in 2008 revealed associations between illness and the consumption of jalapeño peppers, Serrano peppers, and tomatoes. Traceback investigations of implicated jalapeño and Serrano peppers led to farms in Tamaulipas and Nuevo León, Mexico. We conducted a novel analysis of a U.S. Food and Drug Administration database of tomatoes and jalapeño and Serrano peppers imported from Mexico during the first half of 2008 to describe the temporal and spatial flow of these items into the United States. Shipments of all three produce items followed a south-to-northwest corridor; 87% of peppers and 97% of tomatoes produced in Mexican states located west of the Sierra Madre Occidental were transported to ports in California and Arizona, and 90% of peppers and 100% of tomatoes produced in states east of the Sierra Madre Occidental were transported to ports east of Arizona. We found a significant correlation between state-specific infection rates and quantity of imported Mexican jalapeño and Serrano peppers to U.S. states by the first-level consignee but not for imported Mexican tomatoes. We localized production regions of interest by finding that quantities of both peppers and tomatoes imported from the states of Nuevo León and Tamaulipas were correlated with infection rates. In outbreaks possibly caused by agricultural commodities, analysis of import databases may foster a better understanding of growing seasons, harvest sites, shipment itineraries, and consignee destinations, thereby adding valuable insight into findings derived from epidemiologic studies. The outbreak of Salmonella enterica serotype Saintpaul infections that began in the United States in April 2008 involved more than 1,400 reports of illness from the United States and Canada (1). The majority of affected persons fell ill in May and June, with the highest incidence reported in New Mexico and Texas. An initial multistate case-control study of culture-confirmed cases in May revealed an association between illness and consumption of tomatoes but no other item. Tomato traceback investigations failed to converge on a likely source, and tomato and environmental samples collected from entries at the United States Mexico border and at selected farms in Mexico and Florida did not yield the outbreak strain of Salmonella. By July 2008, case-control studies involving patrons who experienced diarrhea after eating at restaurants revealed an association between infection with Salmonella Saintpaul and consumption of raw jalapeño and Serrano peppers. Traceback investigations were conducted by the U.S. Food and Drug Administration (FDA), the Centers for Disease Control and Prevention, and state and local health * Author for correspondence. Tel: ; Fax: ; karl.klontz@fda.hhs.gov. departments on peppers implicated in restaurant-associated illness. Results converged on produce importers located near the Hidalgo, TX port of entry and led to the recovery of the outbreak strain from a jalapeño pepper collected from a produce market in neighboring McAllen, TX. After learning that the contaminated pepper had been grown in Mexico and shipped by a Mexican firm (firm F), FDA investigators visited the firm in the state of Nuevo León. During the inspection, the investigators learned that firm F grew and obtained jalapeño and Serrano peppers from fields located in both Nuevo León and Tamaulipas. In late July, more than 2 months into the outbreak, investigators recovered the outbreak strain of Salmonella Saintpaul on one of these farms in Tamaulipas from a Serrano pepper that was collected from a previously harvested field and from a sample of water used for drip-tape irrigation and pesticide application in the field. In this article, we describe an analysis of FDA data pertaining to the entry of tomatoes and jalapeño and Serrano peppers into the United States from Mexico from January through mid-july 2008 to assess the use of routinely collected produce import data to inform foodborne outbreak investigations.

2 1968 KLONTZ ET AL. J. Food Prot., Vol. 73, No. 11 MATERIALS AND METHODS The FDA Operational and Administrative System for Import Support (OASIS) is an automated system for processing and making admissibility determinations for shipments of foreignorigin FDA-regulated products seeking to enter domestic commerce. Filers (usually import brokers) input data for all prospective entries of FDA-regulated products, including produce. A filing must include a description of the product with the FDA product code and the business information of the filer, importer of record, and product manufacturer or distributor. Although the FDA does not require importers to declare the value and volume of products submitted for entry into U.S. commerce, this information is provided for the vast majority of entries. We reviewed OASIS data for the period 1 January 2008 through 15 July 2008 pertaining to imports from Mexico of tomatoes and jalapeño and Serrano peppers and used this information to evaluate differences in importation patterns between the period preceding the outbreak and the period from the start through the peak of the outbreak. We defined this second period as the outbreak period (1 April through 15 July 2008). Among peppers, we focused solely on jalapeños and Serranos because epidemiologic findings implicated only these two types as risk factors for infection with the outbreak Salmonella strain. Using filters, we examined the product description variable for peppers to identify various spellings of the words jalapeño and Serrano to combine each into a single product category; we used jal and ser to identify jalapeño and Serrano peppers, respectively. In like fashion, we created subcategories to delineate whether the imported jalapeño and Serrano peppers had arrived canned, frozen, brined, in vinegar, sliced or diced, or pureed. When none of the aforementioned terms were stated, we assumed the peppers were fresh. We determined that 99% of jalapeño and Serrano peppers from Mexico entered the United States fresh; the quantities (by weight) arriving as canned, frozen, brined, or in vinegar were 0.1, 0.7, 0.1, and 0.1%, respectively. Unless specified otherwise, all values reported refer to fresh jalapeños and Serranos. For tomatoes, we restricted our analysis to Roma, red round, and plum varieties, the types for which national alerts were issued during the initial phase of the outbreak (5). We excluded from analysis tomatoes that had one or more of the following descriptors: cherry, grape, tomatillo, beef, saladette, strawberry, splendido, on vine, organic, or bola. Tomatoes were designated as red round when they contained the descriptor round or when their sole descriptor was tomato. Because the descriptor greenhouse serves as an umbrella term for a variety of types of tomatoes grown under cover, for imports labeled with that term we included in our analysis only those shipments that contained a designation plum, Roma, or round in addition to greenhouse. Unless specified otherwise, all values and conclusions refer to these designations. We converted the format of each data file (peppers and tomatoes) from an MS Excel spreadsheet (Microsoft, Redmond, WA) to an Epi Info file (version 3.5.1) (2) to facilitate analysis of the following variables: manufacturer name and city (although manufacturer was the term used in the OASIS data set, we employ the term firm in its place given the agricultural nature of the products at hand); entry port, date, and number; consignee name, city, and state; and product quantity (kilograms). To eliminate potentially spurious values for quantity, we identified the median shipment quantity of jalapeños, Serranos, and tomatoes of each type and then eliminated from analysis shipments whose weights were greater than 10 times the median value. In addition to analyzing the pattern of pepper and tomato shipments for the entire country of Mexico, we conducted a subassessment of shipments from Tamaulipas and Nuevo León because these states had been identified as the source of three samples that yielded the outbreak Salmonella strain: a jalapeño pepper obtained from a market in McAllen, TX, that had been distributed by firm F (headquartered in Nuevo León), a Serrano pepper collected off the ground postharvest from a field in Tamaulipas, and a sample of water used for drip-tape irrigation and pesticide application in the field where the contaminated Serrano pepper had been collected. We used two approaches to calculate the combined quantity of jalapeño and Serrano peppers and the combined quantity of plum, Roma, and red round tomatoes that entered each state during the outbreak period based on the location of first-line consignees. First, imports originating from all Mexican states were combined. Second, import volumes from individual Mexican states were evaluated independently. We divided state import quantities and reported Salmonella Saintpaul infection case counts by state census populations to arrive at quantity of imported peppers and tomatoes per person per state and infection rates per state (1, 4). Using JMP (version 8.0; SAS, Cary, NC), we created scatter plots of statespecific infection rate (cases per million people) by kilograms of imported produce type per 100,000 people for every U.S. state. States with no first-line consignees were included and treated as having 0 kg of produce per person. Spearman correlation coefficients and P values were calculated for each of these comparisons (3). RESULTS Peppers. From 1 January through 15 July 2008, the United States received 11,331 shipments of jalapeño peppers (83.2 million kg) from 436 Mexican firms and 5,308 shipments of Serrano peppers (11.7 million kg) from 307 firms (Table 1). The median shipment weight was 4,590 kg for jalapeños and 1,666 kg for Serranos. Exclusion of shipments with weights 10 times greater than the median value led to the elimination of 0.3% of jalapeño shipments and 6.2% of Serrano shipments. Just over two-thirds of jalapeño and Serrano peppers originated from five Mexican states (Sinaloa, Sonora, Tamaulipas, Nuevo León, and Chihuahua). The quantity of peppers that entered the United States by month from Mexico and from Tamaulipas and Nuevo León alone is summarized in Figure 1. Ninety-two percent of jalapeño and Serrano peppers entered through the ports of Hidalgo, TX, Nogales, AZ, or Otay Mesa, CA (Table 2). Pepper shipments followed a south-to-northwest corridor; 87% of peppers produced in Mexican states located west of the Sierra Madre Occidental were transported to the ports of Otay Mesa, CA, Calexico-Cargo, CA, and Nogales, AZ, and 90% of peppers produced in states east of the Sierra Madre Occidental were shipped to ports east of Arizona. Mexican jalapeño and Serrano peppers were shipped to 289 first-line consignees in 20 U.S. states (Table 3); 93% of peppers went to first-line consignees in Arizona, Texas, and California, and 1.2% went to first-line consignees in New Mexico. Sixty firms in Nuevo León and Tamaulipas shipped a total of 13.9 million kg of jalapeño and Serrano peppers to the United States during the study period (1,247 total shipments). No jalapeño or Serrano peppers produced in

3 J. Food Prot., Vol. 73, No. 11 PEPPER AND TOMATO IMPORTS DURING AN OUTBREAK OF SALMONELLA INFECTION 1969 TABLE 1. Origin of jalapeño and Serrano peppers and Roma and red round tomatoes shipped from Mexican states to U.S. ports of entry, 1 January through 15 July 2008 Peppers (kg 1,000) Tomatoes (kg 1,000) Mexican state of origin Jalapeño Serrano Total (%) Roma Red round Total (%) Aguascalientes (0.9) (,0.1) Baja California 2, ,443 (3.6) (,0.1) Baja California Sur 1, ,628 (1.7) (0.2) Chiapas 1, ,060 (1.1) Chihuahua 5, ,821 (6.1) (0.2) Coahuila 1, ,738 (1.8) 1, ,480 (0.6) Colima 1,632 1,142 2,774 (2.9) (0.3) Distrito Federal (,0.1) Durango (0.2) (,0.1) Guanajuato (0.9) (,0.1) Guerrero (,0.1) Hidalgo (0.2) Jalisco 4, ,781 (5.0) 6, ,361 (2.7) México (0.4) (,0.1) Michoacán (0.2) (,0.1) Morelos (0.1) Nayarit 3, ,260 (4.5) Nuevo León 5, ,652 (7.0) 2, ,237 (0.9) Oaxaca (,0.1) Puebla (0.8) (0.2) Querétaro (0.4) (0.2) San Luis Potosí (0.8) 151 5,111 5,262 (2.2) Sinaloa 21,662 2,363 24,025 (25.3) 111,331 96, ,126 (87.2) Sonora 20, ,567 (21.7) 3,117 4,097 7,214 (3.0) Tabasco (,0.1) Tamaulipas 6,290 1,003 7,293 (7.7) 3, ,269 (1.4) Veracruz 1, ,512 (2.6) 1, ,093 (0.5) Zacatecas (0.9) (0.2) Not stated 1, ,779 (2.9) (,0.1) Total 83,236 11,663 94,896 (100.0) 130, , ,552 (100.0) either state entered ports of entry in California, Arizona, or New Mexico; peppers from Nuevo León and Tamaulipas entered the United States via Hidalgo, TX (90%), Rio Grande City, TX (9%), and Progresso, TX (1%). Table 4 summarizes the location of first-line consignees of jalapeño and Serrano peppers produced in Nuevo León and Tamaulipas. Ninety-six percent of first-line consignees were located in Texas and Illinois, and none were located in New Mexico. Of the 11.8 million kg of peppers shipped to firstline consignees in Texas, 80% went to consignees located along the Mexican border (e.g., McAllen, Edinburg, Hidalgo, and Harlingen) and 19% went to consignees in or near Houston. Forty-two firms in Nuevo León and Tamaulipas shipped peppers to the United States at least some time during the outbreak period, 1 April though 15 July (Fig. 2). Firm F, the firm linked to the samples that yielded the outbreak strain (i.e., jalapeño pepper from a produce market in McAllen, TX, Serrano pepper from a field in Tamaulipas, and water sample obtained from a canal beside the pepper field in Tamaulipas), was the sixth-largest pepper firm in Nuevo León or Tamaulipas to ship peppers to the United States during the outbreak period. All of firm F peppers were shipped through the port of Hidalgo, TX, to a single first-line consignee in that city. In addition to shipping peppers to the United States during the outbreak period (Fig. 2), firm F and the majority of the other 41 firms also shipped peppers to the United States from January through March Figure 3 depicts the locations (headquarters addresses) of the leading 10 pepper firms in Tamaulipas and Nuevo León that shipped peppers to the United States during the outbreak period. In some instances, growing fields were distant from the headquarters location; e.g., the Serrano pepper and irrigation water sample that yielded the outbreak strain were collected from a growing field that firm F owned in Tamaulipas, approximately 160 km south of the firm s headquarters in Allende, Nuevo León. Tomatoes. From 1 January through 15 July 2008, the United States received 9,724 shipments of Roma tomatoes and 19,994 shipments of red round tomatoes from 148 firms in 20 Mexican states; the total weight of tomato shipments was million kg. The median weight was 16,330 kg for Roma tomato shipments and 3,548 kg for red round tomatoes (the median weight for shipments described solely as tomatoes was 3,661 kg, and no shipments were described as plum ). Exclusion of shipments with weights 10 times greater than the median value led to the elimination

4 1970 KLONTZ ET AL. J. Food Prot., Vol. 73, No. 11 FIGURE 1. Entry of peppers and tomatoes into the United States from Mexico, by month, January through June (A) Jalapeño peppers (filled bars) and Serrano peppers (hatched bars) from Mexico. (B) Jalapeño peppers (filled bars) and Serrano peppers (hatched bars) from Tamaulipas and Nuevo León. (C) Roma and red round tomatoes from Mexico. (D) Roma and red tomatoes from Tamaulipas and Nuevo León. of 0.4% of Roma tomato shipments and 2.5% of red round tomato shipments. Sinaloa state alone accounted for 87% of tomatoes shipped to the United States during this period (Table 1). The quantities of tomatoes that entered the United States by month from all Mexican states and from Tamaulipas and Nuevo León alone are summarized in Figure 1. Eighty-seven percent of tomatoes entered the United States through the port of Nogales, AZ (Table 2). As with peppers, tomato shipments followed a south-tonorthwest corridor; 97% of tomatoes produced in Mexican states located west of the Sierra Madre Occidental were transported to ports in California or Arizona, and 100% of tomatoes produced in Mexican states east of the Sierra Madre Occidental were shipped to ports in Texas. Overall, nearly 73% of tomatoes went to first-line consignees in Arizona, just under 10% were shipped to consignees in Texas, and less than 0.1% were shipped to consignees in New Mexico (Table 3). Twenty-five firms in Nuevo León and Tamaulipas shipped a total of 5.5 million kg of Roma and red round tomatoes to the United States during the study period (552 total shipments). Virtually all tomatoes (99.5%) from these two states entered the United States through the port of Hidalgo, TX (the remainder entered via Brownsville and Laredo, TX). Table 4 summarizes the U.S. location of firstline consignees for tomatoes produced in Nuevo León and Tamaulipas; 88% went to consignees in Texas, but none went to New Mexico. Nine firms in Nuevo León and Tamaulipas shipped a total of 675,000 kg of tomatoes to the United States during the first 3 weeks of April; six of these firms also shipped jalapeño and/or Serrano peppers during the same period (two of these firms are depicted in Fig. 2 as firms A and B; firm A shipped 26,400 kg of tomatoes and firm B shipped 96,800 kg, all going to first-line consignees in Texas). Correlations. We evaluated associations between state-specific Salmonella infection rates and overall quantity of Mexican peppers and tomatoes imported to first-line consignees in individual U.S. states during the outbreak period. The combined quantity of Mexican-grown jalapeño and Serrano peppers imported into a U.S. state was significantly associated with a state s infection rate (Spearman r, ; P ~ 0.011). The combined quantity of Mexican-grown Roma and red round tomatoes imported into a U.S. state was not associated with a state s infection rate (Spearman r, ; P ~ 0.28). To determine whether the infection rate was associated with any particular region(s) of Mexico where the imported peppers and/or tomatoes were harvested, we calculated a series of similar correlations specific to produce harvested in individual Mexican states. The correlation between statespecific infection rates and the quantity of peppers from 2 of 25 Mexican states that exported jalapeño and Serrano peppers to first-level consignees in U.S. states approached

5 J. Food Prot., Vol. 73, No. 11 PEPPER AND TOMATO IMPORTS DURING AN OUTBREAK OF SALMONELLA INFECTION 1971 TABLE 2. Jalapeño and Serrano peppers and Roma and red round tomatoes arriving at U.S. ports of entry from Mexico, 1 January through 15 July 2008 Peppers (kg 1,000) Tomatoes (kg 1,000) U.S. port of entry Jalapeño Serrano Total (%) Roma Red round Total (%) Nogales, AZ 39,951 1,171 41,122 (43) 106, , ,259 (87.3) Hidalgo, TX 22,291 3,490 25,781 (27) 23, ,307 (10.1) Otay Mesa, CA 13,900 6,399 20,299 (21) El Paso, TX 2, ,692 (3) (,0.1) Rio Grande City, TX 1, ,088 (2) Columbus, NM 1, ,228 (1) Santa Teresa, NM 1, ,099 (1) Calexico-Cargo, CA (,1) (0.1) Progresso, TX (,1) Laredo, TX (,1) 45 5,631 5,676 (2.4) Newark, NJ (,1) Detroit, MI (,1) Brownsville, TX 4 4 (,0.1) Douglas, AZ (0.1) Eagle Pass, TX (,0.1) Total 83,233 11,663 94,896 (100.0) 130, , ,552 (100.0) significance. The two Mexican states involved were Nuevo León (Spearman r, ; P ~ 0.054) and Tamaulipas (Spearman r, ; P ~ 0.052). The correlation between state-specific infection rates and the quantity of tomatoes from 3 of 16 Mexican states that exported Roma or red TABLE 3. Shipments of jalapeño and Serrano peppers and Roma and red round tomatoes from Mexico to first-line consignees in the United States, 1 January through 15 July 2008 Consignee state No. first-line consignees Peppers Quantity (kg 1,000) No. first-line consignees Tomatoes Quantity (kg 1,000) AZ 37 32, ,205 CA , ,566 FL ,808 GA IA 1 15 IL 8 2, IN 1 37 MI 1,1 3 6,405 MO 1 14 NC ,565 NJ 4 1, NM 4 1, NY OH ,179 OK 1 18 OR 1 74 PA 1 1 TX 84 30, ,883 UT 1,1 VA 1 69 WA 1 33 WI 1 13 Not stated 15 Total , ,552 round tomatoes to first-level consignees in U.S. states was significant or close to significant. The three Mexican states involved were Nuevo León (Spearman r, ; P ~ 0.005), Tamaulipas (Spearman r, ; P ~ 0.053), and Coahuila (Spearman r, ; P ~ 0.028), located immediately northwest of and adjacent to Tamaulipas (Table 5). DISCUSSION Our analysis of OASIS entry data for tomatoes and peppers illustrates the complexity of international trade in agricultural commodities. During the first half of 2008, more than 16,000 shipments of jalapeño and Serrano peppers weighing roughly 94 million kg and approximately TABLE 4. Shipments of jalapeño and Serrano peppers and Roma and red round tomatoes from Nuevo León and Tamaulipas to firstline consignees in the United States, 1 January through 15 July 2008 Consignee state No. first-line consignees Peppers Quantity (kg 1,000) No. first-line consignees Tomatoes Quantity (kg 1,000) AZ CA GA IA 1 15 IL 6 1, NC NJ NY OK 1 14 TX 54 11, ,822 Not stated 2 15 Total 82 13, ,503

6 1972 KLONTZ ET AL. J. Food Prot., Vol. 73, No. 11 FIGURE 2. Firms in Tamaulipas and Nuevo León, Mexico, shipping the highest quantities of jalapeño and/or Serrano peppers to the U.S. market from 1 April though 15 July Firms are designated A through T. Horizontal bars indicate weeks when firm shipped jalapeño and/or Serrano peppers to the United States; quantities for each firm varied over time. * Firm associated with peppers and water sample that yielded outbreak strain of Salmonella Saintpaul. 35,000 shipments of Roma and red round tomatoes totaling 266 million kg arrived in the United States from Mexico. The tomatoes and peppers were grown by hundreds of firms throughout Mexico. Adding to the complexity was the fact that Mexican tomatoes and peppers were shipped to hundreds of first-line consignees in 20 U.S. states, from where the produce items moved deeper into domestic U.S. commerce. During the outbreak of Salmonella Saintpaul infections, investigators noted a relative dearth of human illness in California, a state with roughly 10% of the nation s population and a large market for hot peppers. By the end of August 2008, only 16 cases of Salmonella Saintpaul infection had been reported in California (0.4 case per million people) compared with 550 cases in Texas (23 cases per million) and 115 cases in New Mexico (58 cases per million) (1). These differences may be explained in part by the finding that produce harvested in different parts of Mexico traveled along distinct geographic lanes of commerce to the United States. The vast majority of peppers produced in western Mexico, a region where investigators did not recover the outbreak strain, were shipped to California and Arizona ports, whereas peppers produced in eastern Mexico, including those grown in Nuevo León and Tamaulipas, went to ports in Texas and points east. Our analysis revealed that the production and flow of jalapeño and Serrano peppers differed substantially from that of tomatoes. Two Mexican states, Sinaloa and Sonora, FIGURE 3. Location of 10 firms in Tamaulipas and Nuevo León, Mexico, shipping the highest quantities of jalapeño and/ or Serrano peppers to the U.S. market from 1 April though 15 July Firms are designated A through J. Black lines depict location of firm headquarters which, for some firms, are in same city.

7 J. Food Prot., Vol. 73, No. 11 PEPPER AND TOMATO IMPORTS DURING AN OUTBREAK OF SALMONELLA INFECTION 1973 TABLE 5. Correlations between state-specific infection rates and per capita quantity of produce imported into a U.S. state from Mexico Location of export firm in Mexico Jalapeño and Serrano peppers Spearman Spearman r P. r a r Roma and red round tomatoes P. r a All Mexico * Aguascalientes Baja California Baja California Sur Chiapas Chihuahua Coahuila * Colima Guanajuato Guerrero Hidalgo Jalisco México Michoacán Morelos Nayarit Nuevo León * * Oaxaca Puebla Querétaro San Luis Potosí Sinaloa Sonora Tabasco Tamaulipas * * Veracruz Zacatecas a Asterisks indicates correlations at or near significance (as defined by P, 0.05). produced 90% of tomatoes shipped to the United States during the study period, whereas the production of peppers was more decentralized. The routes of entry into the United States also differed for the two categories of produce. Whereas 87% of tomatoes entered through Nogales, AZ, only 43% of peppers did so. Although 82% of tomatoes were shipped to first-line consignees located in Arizona and Texas, only 67% of peppers went to first-line consignees in these two states. Although the analysis of the OASIS database did not allow us to offer definitive insight into the relative roles of tomatoes versus peppers as a cause of illness in the outbreak of Salmonella Saintpaul infections in 2008, several lines of evidence suggest that peppers played a predominant role. Although a significant direct correlation was found between the incidence of infection and the quantity of peppers produced in all of Mexico and received by states with firstline consignees during the outbreak period, the correlation for tomatoes was not significant. The import database indicated that firm F, a firm that shipped from fields where the outbreak strain was identified, shipped peppers but not tomatoes to the United States. However, these observations do not negate a possible role for tomatoes as a cause of illness, as the initial case-control study suggested. Mexican tomatoes destined for the United States may have become cross-contaminated by peppers en route, or tomatoes in the United States may have become cross-contaminated by peppers imported from Mexico. Our study indicates how OASIS data can be used to narrow down the search for growing regions most likely involved in produce-associated foodborne outbreaks. By evaluating import quantities from specific Mexican states, we found that imports of both peppers and tomatoes from two Mexican states were significantly or near significantly correlated with infection rates. These two states, Nuevo León and Tamaulipas, were the states identified through routine traceback investigations. Although OASIS led us to the correct growing region, colinearity in the distribution patterns of various produce types grown in a region make it difficult to determine which of the specific produce items from that region are truly related to the outbreak. However, when conducted early in the process of outbreak investigations, analyses of the spatial and temporal flow of imported agricultural products may identify foods worthy of careful scrutiny. Our study indicated that firm F, the Nuevo León based firm linked to peppers and a water sample that yielded the outbreak Salmonella strain, shipped peppers to the United States before and throughout the outbreak period. Among firms located in Nuevo León and Tamaulipas, firm F was the sixth-largest provider of jalapeño and Serrano peppers to the United States, shipping over 0.5 million kg of peppers to the United States from 1 April through 15 July All firm F peppers were shipped through the port of Hidalgo, TX, to a single first-line consignee in that city. From there, contaminated peppers probably were distributed in a fashion that accounted for illnesses throughout the nation. Although other pepper firms in Nuevo León and Tamaulipas shipped product to the United States before and during the outbreak period and some shipped more peppers than did firm F, traceback investigations identified only firm F as the source of contaminated product. This finding suggests that contamination with the outbreak strain of Salmonella Saintpaul might have been relatively localized in the Tamaulipas field and possibly also in the Nuevo León facility. This study had several limitations. Because we assessed aggregate import data from Mexico and population-level incidence rates of infection in the United States, conclusions drawn from the analysis are subject to ecological fallacy. For example, one or more factors not apparent at the country or state level (e.g., shipment of a single contaminated product lot from a site outside Tamaulipas or Nuevo León) could have accounted for the observations derived from epidemiologic studies, although the recovery of the outbreak strain from peppers and a water sample linked to Nuevo León and Tamaulipas argues against this possibility. Several limitations stem from misclassification of data, which is inherent in the OASIS database. First, the sole use of generic descriptors such as greenhouse introduces the possibility for product misclassification. Second, as the

8 1974 KLONTZ ET AL. J. Food Prot., Vol. 73, No. 11 situation with firm F exemplified, in some instances firm addresses listed in the OASIS database did not correspond precisely with physical locations of growing fields. Although names of firms in the OASIS database are described as manufacturers, it is not clear whether the firms were growers, packers, distributors, brokers, importers of record, and/or warehouses or whether they handled products in any physical fashion. Although firm F was headquartered in Allende, Nuevo León, the Serrano pepper and the water sample that yielded the Salmonella Saintpaul outbreak strain were obtained from a growing field approximately 160 km (100 mi) away in the adjacent state of Tamaulipas. Thus, by using the firm headquarters address as a proxy for harvest site, we may not have always identified the true state where the crop was produced. Third, the use of geographic location of first-line consignees as a proxy for location of consumption is prone to a high degree of misclassification because produce can be distributed widely to second- and third-line consignees. For example, New Mexico, the state with the highest infection rate (approximately twice the rate in Texas, which had the second highest rate) had no firstline consignees that received peppers or tomatoes from firms located in Nuevo León or Tamaulipas (1). Contaminated produce obviously dispersed into New Mexico from first-line consignees located in other states, most likely Texas, which received the majority of the produce grown in Nuevo León and Tamaulipas. Given the lack of data on the dispersal of produce beyond first-line consignees, the correlations between state infection rates and state import volumes are inherently biased. Regardless of these limitations, the results of this study illustrate the utility of elucidating the temporal and spatial pathways by which agricultural products reach the United States from abroad and evaluating correlations between these pathways and outbreak infection rates. The methods we present can be expanded upon to investigate future outbreaks with refinements to account for the distribution of produce beyond first-line consignees. Sensitivity analyses could be performed to arrive at the best approximation of dispersal routes. Collection of baseline data on the dispersal patterns of produce from first-line consignees could help inform this type of modeling. If the temporal dynamics of specific produce items from time of import to time of consumption were better elucidated, refinements could be made to account for lag times from importation to consumption when evaluating correlations between import quantity and infection rates. Ideally, analyses of the sort described here should be conducted during the earliest stages of outbreak investigations, as early as the hypothesis generation period. In summary, we present a novel and robust methodology that uses import data to complement traditional foodborne outbreak investigations. In outbreak settings involving agricultural commodities, an understanding of growing seasons, harvest sites, shipment itineraries, and consignee destinations may add valuable insight into findings derived from epidemiologic studies. ACKNOWLEDGMENT The authors thank Abebe Gedamu for preparing Figure 3. REFERENCES 1. Centers for Disease Control and Prevention Outbreak of Salmonella serotype Saintpaul infections associated with multiple raw produce items United States, Morb. Mortal. Wkly. Rep. 57: Centers for Disease Control and Prevention What is Epi Info? Available at: Accessed 27 March Kendall, M., and J. D. Gibbons Rank correlation methods, 5th ed. Oxford University Press, New York. 4. U.S. Census Bureau Annual estimates of the resident population for the United States, regions, states, and Puerto Rico: April 1, 2000 to July 1, NST-EST Available at: popest/states/nst-ann-est.html. Accessed 16 November U.S. Food and Drug Administration FDA warns consumers nationwide not to eat certain types of raw red tomatoes [FDA news release, 7 June 2008]. Available at: Newsroom/PressAnnouncements/2008/ucm htm. Accessed 22 September 2009.