Tag Archives: international outbreak museum

Venison Jerky

Outbreak: Venison Jerky 
Product: Venison JerkyInvestigation Start Date: 11/20/1995
Location: Benton County, OregonEtiology: E. coli (STEC) O157:H7
Earliest known case onset date: 11/13/1995Latest case onset date: 11/20/1995
Confirmed / Presumptive Case Counts: 6 / 5Positive Samples (Food / Environmental / Water): 2 / 2 / 0

At a time when ground beef was thought to be the source of all foodborne E. coli O157 infections, this was the first documentation of venison-associated E. coli O157:H7 outbreak. Epidemiologists showed that deer can be colonized by E. coli O157:H7 and ultimately be a source of human infections. Game should be handled with the same caution indicated for commercially slaughtered meat, and jerky dehydration does not reliably kill bacteria in meat.


To investigate a 1995 outbreak of Escherichia coli O157:H7 infections in an Oregon community and to assess the safety of meat dehydration methods.

Survey subsequent to routine surveillance report, environmental investigations, and lab experimentation.

Members of an extended household and their social contacts with confirmed or presumptive E. coli O157:H7 infections.

A total of 6 confirmed and 5 presumptive cases were identified. Homemade venison jerky was implicated as the source of transmission. E. coli O157:H7 with the same distinctive, pulsed-field gel electrophoresis pattern seen in the case isolates was recovered from leftover jerky, uncooked meat from the same deer, a saw used to dismember the carcass, and fragments of the deer hide. In a subsequent survey, E. coli O157:H7 was recovered from 3 (9%) of 32 deer fecal pellets collected in nearby forest land. In the laboratory, inoculated venison was dried at several time and temperature combinations, ranging up to 10 hours at 62.8 degrees C. Viable organisms were recovered under all conditions tested.

Deer can be colonized by E. coli O157:H7 and can be a source of human infections. Conditions necessary to ensure the safety of dried meat deserve further review. Game should be handled with the same caution indicated for commercially slaughtered meat.

Foundation Farm Raw Milk

Outbreak: Foundation Farm Raw Milk 
Product: Foundation Farm Raw MilkInvestigation Start Date: 04/10/2012
Location: Clackamas, Multnomah and Washington Counties, OregonEtiology: E. coli (STEC) O157:H7
Earliest known case onset date: 04/01/2012Latest case onset date: 04/14/2012
Confirmed / Presumptive Case Counts: 11 / 5Positive Samples (Food / Environmental): 2 / 13
Hospitalizations: 4Deaths: 0

This outbreak highlighted the persistent hazard of raw milk consumption, already well documented in the medical literature. Four of the 11 cases were hospitalized — some critically ill and all with hemolytic-uremic syndrome (HUS), subsequently shown to have been caused by E. coli O157:H7. Although most sales of raw milk are illegal in Oregon, the farm implicated here is an example of one method by which those determined to get the product can skirt the law: instead of “purchasing milk,” families bought “shares” in Foundation Farm cows, their dividends being the milk produced by them and then consumed without pasteurization.

The cause of the outbreak wasn’t long a mystery: Oregon local public health nurses routinely interview each O157 case and ask about “usual suspect” exposures — one of which is whether the case consumed unpasteurized milk. A single case having a high-risk exposure like this one is sufficient to raise suspicion and to warrant further investigation. As the investigation progressed, it transpired that the families that participated in this “cow-share” arrangement had already been aware of many illnesses and communicating among themselves about them; and the dairy farmer had therefore advised the shareholders not to drink the milk. The farm ceased production, and several parents publicly advised the unwary not to serve unpurified milk to their children.


On April 10, 2012, the Multnomah County Health Department was notified by a clinician about a 1 year-old child with HUS (culture pending; later shown to be O157:H7). The child attended a day-care center in Washington County and had a history of regular raw milk consumption, both of which indicated a need for public health follow-up.

Washington County public health officials investigated the child-care facility and identified no other suspected cases there. Meanwhile, Oregon Health Authority (OHA) officials conferred with those from the Oregon Department of Agriculture (ODA) to discuss action that could be taken regarding the herd-share operation that was reportedly the source of the raw milk. It was determined that the operation (dba Foundation Farm), although unlicensed and unknown to Agriculture officials, had been operating legally, with 4 cows, 3 of which were in production. (Dairies of this type are not required to be licensed or regulated in Oregon.) ODA officials contacted the farm owner to request a list of its customers so that they could be contacted and asked about potential illness.

On the morning of April 11th, the farmer reported to ODA that, in the course of contacting his customers through Facebook, he had learned of several additional illnesses. He also reported that he had voluntarily stopped distribution of milk and had recommended to his customers that they discard any available product. Upon request, he sent a list of the 48 household contacts for his customers, and authorized ODA to visit the premises and to collect specimens from the working part of the farm.
That afternoon, while local and state public health officials commenced case finding among herd-share households, OHA investigators and an ODA official visited the dairy farm. They collected 63 samples including rectal swabs on all 4 dairy cows, surfaces in and around the milking parlor, and fecal pat samples from adjacent pastures. No milk was available at the farm for testing.

Over the ensuing days, OHA investigators attempted to interview all households on the list using a standardized questionnaire regarding history of Foundation Farm milk consumption, several other possible food exposures, and diarrheal illness in household members.


A press release was issued to warn the public about the risk and to stimulate reporting. One of the earliest identified cases was from a household that was not on the customer list: that child had consumed the milk while visiting a family friend.

Leftover raw milk samples were collected for testing on April 12 and 13 from two shareholder households.

Ultimately, 11 lab-confirmed cases—including 1 Washington State resident who consumed the milk while visiting family in Oregon—and 5 presumptive cases were identified. Four other household members reported very mild concurrent diarrheal symptoms and were not counted as cases. Four cases were hospitalized—all with HUS. Three cases were hospitalized for >25 days. Representatives of 30 (62%) of the 48 herd-share households, comprising 91 individuals, were interviewed. All 16 cases had consumed FF milk on one or (usually) multiple occasions. No other suggestive common exposures were identified.

The farm setup was described by ODA as fairly typical for a family farm. The layout was poorly designed from a food-safety perspective. The premises appeared to be difficult to maintain in a hygienic manner for food production. There were no facilities for rapid chilling of milk. Milk bottles were reportedly stored in an ordinary (non-commercial) refrigerator.

Culture of 13 (21%) of 63 samples collected at Foundation Farm yielded E. coli O157:H7; positive samples included 10 of animal manure, 2 cattle rectal swabs and 1 swab of fencing at the milking station. One of two milk samples collected from customer households was also positive. All but one of the O157 strains isolated from cases, milk, and the farm were indistinguishable by 2-enzyme pulsed-field gel electrophoresis (PFGE); 1 case appeared to have an extra band with XbaI digestion.

This outbreak was caused by consumption of unpasteurized contaminated milk. Consumers should be warned that, without a kill step such as pasteurization, there is no reliable way to render milk free of harmful microbes. In particular, E. coli O157 infection can cause bloody diarrhea, hemolytic uremic syndrome and even death.

Berry Stand Strawberries

Outbreak: Berry Stand Strawberries 
Product: StrawberriesInvestigation Start Date: 08/03/2011
Location: Clackamas, Multnomah, Washington, Clatsop, & Yamhill Counties, OregonEtiology: E. coli (STEC) O157:H7
Earliest known case onset date: 07/01/2011Latest case onset date: 07/29/2011
Confirmed / Presumptive Case Counts: 14 / 1Positive Samples (Food / Environmental): 4 / 100
Hospitalizations: 6Deaths: 2

This investigation implicated a novel vehicle—strawberries—as the cause of an outbreak of E. coli O157 infections. An association with locally produced strawberries quickly became apparent as cases were interviewed; but the fact that almost all had purchased them at roadside stands and farmer’s markets led to concern about potential confounding of some overlooked item sold at similar venues. The concern was laid to rest by visits to the stands, a case-control study, and traceback of the strawberries to a single Oregon farmer. Learning how the strawberries became contaminated was the fruit of field work: epidemiologists investigated the strawberry fields and found them heavily contaminated with deer feces, with a herd of deer seen grazing in the area. Culture of 50 environmental samples that contained visible deer pellets yielded E. coli O157 in 10 (20%), with PFGE patterns matching those of the cases.

This outbreak once again confirmed deer as key reservoirs for the pathogen. Interestingly, an additional PFGE-matched case was identified months later; the case hadn’t eaten the commercial strawberries but lived in the area where they were grown. Culture of a sample from this case household’s vacuum cleaning bag yielded a PFGE-matching strain of E. coli O157:H7, a testimony to the pathogen’s ability to survive for prolonged periods when dried.


The following is the abstract from the published article found here. Editorial commentary can be found here.

An outbreak of Escherichia coli O157:H7 was identified in Oregon through an increase in Shiga toxin-producing E. coli cases with an indistinguishable, novel pulsed-field gel electrophoresis (PFGE) subtyping pattern.

We defined confirmed cases as persons from whom E. coli O157:H7 with the outbreak PFGE pattern was cultured during July–August 2011, and presumptive cases as persons having a household relationship with a case testing positive for E. coli O157:H7 and coincident diarrheal illness. We conducted an investigation that included structured hypothesis-generating interviews, a matched case-control study, and environmental and traceback investigations.

We identified 15 cases. Six cases were hospitalized, including 4 with hemolytic uremic syndrome (HUS). Two cases with HUS died. Illness was significantly associated with strawberry consumption from roadside stands or farmers’ markets (matched odds ratio, 19.6; 95% confidence interval, 2.9–∞). A single farm was identified as the source of contaminated strawberries. Ten of 111 (9%) initial environmental samples from farm A were positive for E. coli O157:H7. All samples testing positive for E. coli O157:H7 contained deer feces, and 5 tested farm fields had ≥ 1 sample positive with the outbreak PFGE pattern.

The investigation identified fresh strawberries as a novel vehicle for E. coli O157:H7 infection, implicated deer feces as the source of contamination, and highlights problems concerning produce contamination by wildlife and regulatory exemptions for locally grown produce. A comprehensive hypothesis-generating questionnaire enabled rapid identification of the implicated product. Good agricultural practices are key barriers to wildlife fecal contamination of produce.

Milk Crates, Cartons, & Jugs

Outbreak: Milk Crates, Cartons, & Jugs 
Product: Milk Crates, Cartons, & JugsInvestigation Start Date: 01/27/2010
Location: Roseburg, OregonEtiology: Salmonella Braenderup
Earliest known case onset date: 10/21/2009Latest case onset date: 10/01/2010
Confirmed / Presumptive Case Counts: 25 / 0Positive Samples (Food / Environmental): 15 / 500
Hospitalizations: 0Deaths: 0

This outbreak illustrates the value of descriptive epidemiology and the virtue of epidemiologist persistence! Over a 12-month period during 2009–2010, a total of 25 infections by Salmonella Braenderup with matching PFGE patterns were reported—no more than 4 in any given month. After 3 scattered cases over 3 months, the investigation began with recognition of a 3-case cluster in a single Oregon county in January 2010. Extensive hypothesis-generating (“shotgun”) interviews were thereafter conducted as each case was recognized, but no obvious source emerged. In desperation, on July 26, 2010, Oregon’s epidemiologists gathered to review the available information. The descriptive epidemiology proved pivotal. Person: although cases ranged in age from 1 to 88 years, there was a predilection for children: median case age was 13 years. Place: cases resided in 9 contiguous Oregon counties—but none from the populous Portland metropolitan area. Time: the prolonged duration of the outbreak ruled out most perishable items. What food or other vehicle could jibe with this information?

At least 94% of cases affirmed consumption of milk—though pasteurized and of different labels. But Umpqua brand milk was known to be distributed predominantly south of the Portland area, and no other plausible hypothesis survived the analysis of the assembled epidemiologists.

Experts in the Oregon Department of Agriculture were able to confirm that the alternative milk brands reported by cases or served in the case children’s schools were, in fact, produced by Umpqua. With the epidemiologists, they visited the production facility and confirmed that the pasteurization process was in order: records indicated that the milk was heated, stored, and packaged appropriately, with no apparent opportunity for post-pasteurization contamination. But culture of 132 environmental specimens obtained at the facility yielded Salmonella Braenderup in five: one from a milk crate, three from sites on the crate’s conveyor system, and one from a floor drain. The crate washer was situated on the roof of the facility, open to birds and perhaps other wildlife.

The investigation tendered a surprising conclusion: the milk itself was never contaminated. Rather, the bacteria were on the milk crates, occasionally finding their way onto individual cartons, and, from perhaps one in 10,000 cartons, into someone’s mouth. This conclusion that the milk crates and cartons acted as fomites (fō’ mĭ tēz)—objects that can transfer infectious microorganisms to humans—explained the prolonged duration of the outbreak, the low attack rate, the case demographics, and the cases’ exposure histories. Finding the cause allowed for remediation and resumption of business for Umpqua: after closing for a week to sanitize and to rebuild its crate system, the plant resumed production, and the outbreak ended.


25 cases dribbled out from Oct 2009 to Oct 2010; eventually linked to product from Umpqua Dairy in Roseburg. Crate washer was contaminating milk crates and (in turn) the sealed containers (not the milk per se).

Umpqua milk products produced at the Roseburg plant have been confirmed as this source of the 25-case, year-long outbreak.

In a surprising twist, however, it appears that the milk itself may NEVER have been contaminated. Rather, the weight of evidence suggested that the problem stemmed from external contamination of the milk cartons and jugs from an outdoor crate washer.

In other words, the containers were fomites, meaning that by most definitions this was not a “foodborne” outbreak. Milk crates used to move product around the plant and to distribute product to retailers around the state come back to the Roseburg plant, where they recycled up a conveyor line through an outdoor crate washer and thence were carried inside the plant where they were loaded with cartons of milk.

As the crates went through the washer, they were in effect being sprayed down with a solution of Salmonella Braenderup. Matching bacteria were recovered from the washer and multiple points downstream, including crates and external surfaces of sealed containers.

The plant was shut for 5 days while the crate washer was disassembled, cleaned, and rebuilt, and other parts of the plant were cleaned and sanitized. Crates were boiled. New sanitizer stations were added to reduce or eliminate the risk of external contamination of the crates when they are filled with product.

Lian How White Pepper

Outbreak: Lian How White Pepper 
Product: White PepperInvestigation Start Date: 03/02/2009
Location: Multi-state outbreak–OR, CA, NV, WAEtiology: Salmonella Rissen
Earliest known case onset date: 11/01/2008Latest case onset date: 05/27/2009
Confirmed / Presumptive Case Counts: 87 / 0Positive Samples (Food): 33
Hospitalizations: 8Deaths: 0

Salmonella Rissen is a very rare serotype in the US. When Oregon, California, and Nevada all reported it within days of each other, it began a coordinated investigation between those states (later to be joined by Washington and Idaho). This outbreak focused FDA attention on the role spices may play in salmonellosis outbreaks. Additionally, a large proportion of cases either had urine isolates or no gastrointestinal symptoms, making interviewing about food histories much more complicated.


Salmonella Rissen is a very rare serotype in the U.S. with approximately 6 cases/year; in 2006 there was a sizable jump to 18 cases. Rissen is a fairly common serotype in Thailand, however; it is commonly associated with pork and chicken products.

On March 2, 2009, the Nevada State Public Health Laboratory posted a notification of a local cluster of Salmonella Rissen; CA and OR state health departments responded the same day to report similar local clusters—definitely above the expected annual baseline. In fact, this was the first reported outbreak of Salmonella Rissen in the United States. Washington and Idaho would later join the investigation with locally reported cases. CA and NV cases were clusters in the San Francisco Bay and Carson City/Reno areas, respectively.

Methods and Results:
By consensus, cases with an identifiable onset date were interviewed with shotgun questionnaires. CDC reported that the only PFGE match in the PulseNet database was an FDA report of an isolate from Vietnamese black peppercorns in 2006. Although the significance of this finding was uncertain, we were aware of it and added pepper-related questions in all interviews.

Oregon noted that they had one case who was Vietnamese and another who frequently ate at Vietnamese restaurants—fairly tenuous but about all we had to start with. An absence of travel histories implied a commercial product distributed primarily or exclusively in the West.

An extraordinary proportion of cases (63%) lacked what we could call a “clean” onset date—they either had vague symptoms going back for an indefinite period or had only urine isolates, sometimes with no referable GI symptoms.

Cases broke out into three distinct populations:

•Approximately half of the cases either had exposure to Asian restaurants or foods or had Asian surnames who cooked traditionally Asian foods at home (Generally younger and healthy)

•Cases exposed to same large casino buffet in Nevada (exposures occurred between 12/08-2/09)

•35% of cases were hospitalized prior to or during time of specimen collection (generally older with underlying health conditions); they were at different facilities all under the same HMO organization.

At least 18 hypothesis-generating interviews were conducted using a hybrid of the Oregon shotgun and other more open-ended questions to explore longer time periods and more vague food exposures. Instead of using the standard time period of 7 days to go back and assess foods consumed prior to illness, we expanded the question over weeks and sometimes months of time; so many cases had little or no GI symptoms we wanted to cast a wide net when asking about exposures.

Nothing strongly suggestive emerged from those analyses other than the obvious clustering of NV and CA cases who had visited and eaten at Casino A in Reno. Efforts to identify common foods there foundered on the combination of poor recall and all-you-can-eat buffet overload. Some product testing was attempted, but initial results were all negative, including black pepper. NV initially suspected cilantro to be the culprit and were waiting to get confirmation that a buffet reservation had contained it at an ingredient; the decision was made not to test the black and white ground pepper that had been collected from the buffet in the first round of testing. All buffet ingredients that were initially tested came back negative.


• Bean sprouts: 33% cases vs. 12% FoodNet;
    – OR 3.8, exact 95% CI 0.8-14.5, p=0.04
    – Cumulative (binomial) probability = 0.0154
• Cilantro: 55% cases vs. 35% FoodNet;
    – OR 2.3, exact 95% CI 0.6-9.5, p=0.20
    – Cumulative (binomial) probability = .0597
• Black pepper: 67% cases, no FoodNet data available

Oregon decided to pursue brute force testing at two different restaurants frequented by one case (i.e., not really “implicated” in any way). This case had a urine isolate of Salmonella Rissen that was tested as follow up to a urinary tract infection and only reported one instance of loose stools (i.e., diarrhea); the case ate the named restaurant in the 3 days before illness. Food and spice samples were collected and split between IEH Laboratories in Seattle and the Oregon State Public Health Laboratory. Both labs isolated Salmonella Rissen from an open 5# container of Lian How brand white pepper collected at a Chinese/Vietnamese restaurant where the case added white pepper from a shaker into Pho (Vietnamese noodle soup). The Oregon PHL also recovered Rissen from a similar and adjacent container of black pepper. California Food and Drug Branch and the FDA worked with Union International Food Co. (Union City CA), which quickly led to a “voluntary” recall of Lian How and Uncle Chen brand pepper products on March 30, 2009—and soon expanded include everything that they packaged or prepared because of gross environmental contamination with Salmonella.

We worked with local and ODA environmental health staff to trace product distribution in Oregon and to reach out to restaurant operators who may have received the product. Language barriers and other considerations made us chary of relying solely on the media or the efforts of product distributors.

The final official tally was 87 cases in OR, WA, ID, NV, and CA; eight of the cases were Oregon residents. Clean onset dates occurred between 12/09/2008 and 04/29/2009 and could only be establish for 32 (37%) of cases. Rissen was isolated from stool specimen in 54% of the cases and from urine specimen in 39% of the cases. Ages ranged from 5 months to 94 years old with a median of 52 years old. Females accounted for 64% of the cases. Eight people were hospitalized due to GI illness (in all states).

Quantitative testing of three white pepper samples (recovered from 3 different Oregon restaurants) was performed by IEH Laboratories; results were 14 MPN/g, 2,400 MPN/g, and 14,900 MPN/g.

Contaminated white pepper was the cause of this outbreak. The positive black pepper sample was never replicated at any other lab and may have resulted from cross-contamination (most likely at the restaurant).

This outbreak (and the later Daniele salami outbreak) has focused FDA attention on the role spices may play in salmonellosis outbreaks; there are ongoing efforts in this arena.

Bagged Spinach

Outbreak: Bagged Spinach 
Product: SpinachInvestigation Start Date: 09/08/2006
Location: Multi-StateEtiology: E. coli (STEC) O157:H7
Earliest known case onset date: 08/25/2006Latest case onset date: 09/13/2006
Confirmed / Presumptive Case Counts: 5 / 0Positive Samples (Food / Environmental / Water): 0 / 0 / 0
Hospitalizations: 98Deaths: 3

This outbreak highlighted Inter-state health department cooperation. Oregon epidemiologists identified that 4 out of 5 cases in Oregon consumed bagged spinach using their homegrown “Shotgun” hypothesis-generating questionnaire. Contact with WI and NM showed cases already underway with spinach showing signs of being the culprit. Contact with Utah allowed them to quickly re-interview on spinach. The rest played out nationally following a conference call with NM, UT, OR, CDC, and FDA. FDA went public later that day.


A multi-state outbreak linked to bagged spinach, Dole and others from Natural Selections, Calif.

On Friday, 8 Sept 2006, ~1800, Janie called to report 3 that we had a match of 3 O157s, including 1 from Cowlitz Co. We had 1 form in hand already (GW), so we contacted Debby Uri at home to get the story on her case and attempted to reach Cowlitz Co on-call people. The Linn Co person had been in Idaho the whole exposure period, and presumably was exposed there. GW worked at a hospital in Oregon, but had no contact with any known O157 cases. The Washington guy had been in Oregon only briefly during during the 10 d before onset. His father visiting from Florida had also been ill but no MD. We agreed to pursue jointly on Monday.

Late Tuesday afternoon (again around 1800) Janie reported 2 additional matches, so Melissa took over the task of shotgunning these people. She called until late Tuesday night, and finished Wed morning. The only surprising finding was 4/5 reporting eating bagged spinach, vs. an estimated background for ANY spinach of 17% (P = 0.0036). [Testing months later showed the non-spinach eater to have a different MLVA type.] She had a mix of brands being reported, although 3/5 said they shopped at Winco. Around noon she left to have lunch with a friend and was going to swing by Winco to get sample bags.

Soon after she left I decided to notify CDC that we had a probable hit on this cluster, and to find out if there were any other clusters around the US. Our lab had been unable to upload the patterns because their CDC “fob” had expired, and we were temporarily cut off from the PulseNet data exchange. I sent Chris Braden the picture of the PFGE, and he quickly responded that it appeared by eye to match cases in WI, NM, and UT that were under investigation. He said that WI suspected some kind of fresh produce—possibly based on their demographic profile alone, at which point I emailed that we suspected bagged spinach specifically. (In the earlier message I had said we had a tentative product, but didn’t say what it was. All these exchanges happened within 15-20 minutes). I followed up the 2nd email with a phone call, and he connected us to WI (Jeff Davis and John Archer at least), who had ~17 cases already and were underway with a c-c study.

At that point they had 8 case interviews only, but all 8 had eaten spinach, which for me certainly iced the cake. After that call ended I contacted Utah and NM (not vv as stated in the MMWR). Utah did not have spinach on the Q they were using, but agreed to quickly re-interview on spinach. NM was already interested in spinach from their interviews, and was already trying to collect leftover spinach from 1 hh. During that phone call I got a voice mail from Marilee Poulson that their cases were all reporting bagged spinach consumption.

That is about the end of it. The rest played out nationally following a conference call with NM, UT, OR, CDC, and FDA Thursday morning (14 Sept). FDA went public later that day.

Paramount & Raw Almonds

Outbreak: Paramount & Raw Almonds 
Product: Paramount & Raw AlmondsInvestigation Start Date: 02/23/2004
Location: Multi-state outbreakEtiology: Salmonella Enteritidis
Earliest known case onset date: 02/01/2004Latest case onset date: 04/17/04
Confirmed / Presumptive Case Counts: 7 / 0Positive Samples (Food / Environmental / Water): 1 / Lots / 0
Hospitalizations: 7Deaths: 0

This investigation was rapidly solved by the use of Oregon’s “Shotgun” hypothesis-generating questionnaire, a detailed food history interview tool that asks about over 400 different foods and places to eat. A single interviewer at the Oregon state health department used this standardized questionnaire to interview all 5 of the first cases detected as part of the same outbreak. Though the questionnaire only asked about “almonds in the shell,” when the first two cases recalled eating raw almonds the interviewer took an iterative approach and modified the questions she asked the next 3 cases to include raw almonds.


Outbreak of Salmonella Serotype Enteritidis Infections Associated with Raw Almonds — United States and Canada, 2003–2004

On June 4, this report was posted as an MMWR Dispatch on the MMWR website: (http://www.cdc.gov/mmwr).

On May 12, 2004, the Oregon State Public Health Laboratory identified a cluster of five patients infected with Salmonella enterica serotype Enteritidis (SE) isolates that were matched by using two-enzyme pulsed-field gel electrophoresis (PFGE). The five patients were from four Oregon counties; their onsets of illness occurred during February–April 2004. A subsequent investigation, still ongoing, has identified a total of 29 patients in 12 states and Canada with matching SE isolates, since at least September 2003. Seven patients have been hospitalized; no one has died. Raw almonds distributed throughout the United States and internationally have been implicated as the source of the SE infections. As of May 21, approximately 13 million pounds of raw almonds had been recalled by the producer.

Routine interviews of the initial five patients with salmonellosis had not indicated a common exposure. However, prompted by the May 12 laboratory data, the patients were reinterviewed by using a standard hypothesis–generating questionnaire that included questions about consumption of approximately 400 specific food items and their shopping and eating venues during the 5 days before illness onset. Using binomial distribution, consumption rates for selected foods were compared with background rates estimated from a 2002– 2003 population-based survey of residents of Oregon (1). The initial five patients from Oregon all reported consuming Kirkland Signature brand raw almonds, purchased at Costco warehouse stores. Survey data (1) indicated that an estimated 9% of Oregon residents (86 of 921 surveyed) consumed raw almonds from any source in the preceding week. Even assuming that 20% of all Oregon residents ate Kirkland Signature brand raw almonds each week, the binomial probability of finding five of five sporadic cases with that history is <0.001. No other foods or food sources were associated with illness.
After determining that the raw almonds were distributed widely, U.S. and Canadian epidemiologists and state and federal regulatory agencies were notified on May 13 via electronic information networks. Through PulseNet, the national molecular subtyping network (2), laboratories were queried for reports of isolates matching the outbreak PFGE patterns (XbaI: JEGX01.0049; BlnI: JEGA26.0008 or JEGA26.0009, reflecting minor variation later observed with the second enzyme).   (CONTINUED BELOW ↵)

Laboratories that did not routinely screen SE isolates by using PFGE were encouraged to do so for isolates collected since February 1, 2004. Phage typing was performed by standard methods. As additional PFGE-matching isolates were identified, a brief, customized questionnaire was used in interviews with persons about their nut consumption.

Raw almonds from an opened package recovered from one patient’s household were tested for Salmonella by enzyme immunoassay. Unopened packages of nuts from the supplier’s warehouse and environmental samples collected at the almond processor and at huller-shellers supplying the processor were tested for Salmonella by using standard microbiologic methods.

As of June 2, a total of 29 patients with SE infections matching both XbaI and BlnI PFGE patterns had been identified in 12 states and one Canadian province. Symptom onsets ranged from September 2003 to April 2004 (Figure). Patients ranged in age from 11 months to 91 years (median: 40 years); 17 (59%) were female. Seven patients were hospitalized; no one died. Multiple other cases with matching PFGE patterns and onsets earlier in 2003 remain under review. To date, nine isolates from the current outbreak have been phage typed; all are type 9c, which is uncommon. Among 26 patients interviewed, 24 recalled eating raw almonds during the week before illness onset; 20 patients identified brands packaged or supplied by Paramount Farms (Lost Hills, California). One infant patient was presumed secondarily infected. Through retailer computer records linked to membership cards or customer receipts, dates and places of almond purchase were verified for 10 households of patients. The dates of verified almond purchases ranged from November 3, 2003, to January 28, 2004.

Efforts to identify specific production lots associated with illness, based on almond purchase dates and locations and store inventory data, are ongoing. On May 18, Paramount announced a nationwide recall* of all raw almonds sold under the Kirkland Signature, Trader Joe’s, and Sunkist labels. Costco mailed 1,107,552 letters to members known to have purchased the recalled product in the United States. The recall was expanded subsequently to include nuts sold in bulk to approximately 50 other commercial customers, some of whom repackaged almonds for sale under other brand names. In addition to sales in the United States, almonds were exported to France, Italy, Japan, Korea, Malaysia, Mexico, Taiwan, the United Kingdom. The majority of the recalled almonds likely were consumed months ago; however, raw almonds have a shelf life of >1 year, and consumers might still have the implicated products.

Tests of raw almonds recovered from a patient’s household and samples collected at Paramount were negative for Salmonella; however, Salmonella was isolated from one environmental sample collected at Paramount and from three samples from two huller-shellers that supplied Paramount during the period of interest. Serotype and PFGE analyses of these isolates have not been completed, and additional sampling continues.

Hydro-Harvest alfalfa sprouts

Outbreak: Hydro-Harvest alfalfa sprouts 
Product: Alfalfa sproutsInvestigation Start Date: 02/01/1999
Location: Multi-State: OR, WA, ID.Etiology: Salmonella Mbandaka
Earliest known case onset date: 01/09/1999Latest case onset date: 06/12/1999
Confirmed / Presumptive Case Counts: 43 / 0Positive Samples (Food / Environmental / Water) 12 / 0 / 0

The increasing popularity of sprouts as health foods was sharply followed by the epidemiologic discovery that sprouts are prone to transmitting foodborne pathogens like Salmonella. This was the largest “sproutbreak” of foodborne salmonellosis in Oregon since the multistate and multinational outbreak of Salmonella Newport infection in 1996, in which alfalfa sprouts were also implicated as the vehicle of transmission.


Since 1995, sprout products (alfalfa, clover, radish, mung bean) have become recognized as common vehicles for enteric pathogens including Salmonella, Escherichia coli O157, and Listeria. Outbreaks have been recognized repeatedly in Europe (notably Scandinavia), North America, and Japan. Most of these outbreaks have been traced to the use of contaminated seed, in which pathogens proliferate during germination. Seed lots are often quite large (≥18,000 kg) and may be distributed to numerous growers in widely dispersed locations. Outbreaks often manifest as cases scattered across several states or countries, and can persist for months, reflecting the stability of the pathogen on dried seeds.

The only completely reliable way to prevent sprout-associated illness is abstinence. In the United States, several state health departments, the FDA, and other agencies have issued statements discouraging raw sprout consumption. Because of the increased risk for severe infection with at least some sprout-associated pathogens (notably Salmonella and Listeria), these warnings are often targeted at immunocompromised individuals. Most persons who become ill in outbreaks are immunocompetent, however.

Many people enjoy eating sprouts, and many people make a living growing, processing, and distributing these products. In consequence, food scientists and members of the sprout industry have investigated procedures to improve the safety of sprouts. These efforts have concentrated on chemical disinfection of seeds immediately prior to germination.

Laboratory studies suggest that treatments with sodium hypochlorite or calcium hypochlorite can reduce Salmonella loads on artificially inoculated seeds to undetectable levels. Calcium hypochlorite (at 20,000 ppm) has been strongly recommended, because higher disinfectant levels are achievable without unacceptable reductions in germination rates.

Some outbreaks provide natural experiments that shed light on the effectiveness of seed disinfection. Oregon Public Health Division epidemiologists investigated one such outbreak in early 1999.

By law, all Salmonella isolates identified by Oregon laboratories are forwarded to the Public Health Laboratory for serotyping. Similar law or custom obtain in Washington, California, and Idaho. Salmonella Mbandaka is uncommon in Oregon (1988–98 mean, 1.5 cases/year). When 3 Mbandaka isolates were identified within a week in January 1999, it was not difficult to surmise that a common-source outbreak was occurring. Surrounding states and CDC were notified immediately. Initially, reported cases were confined to Oregon.    (CONTINUED BELOW ↵)

Preliminary interviews were conducted with cases or household informants by local health department nurses using standardized case investigation forms. Sprouts—a perennial suspect in geographically dispersed outbreaks of salmonellosis—quickly became a focus of the investigation.

Investigators compared exposure histories of the first 10 Oregon cases reported with those of age- and phone prefix-matched controls. We traced sprouts consumed by cases to their origins.

Information about later cases was collected by public health agencies in Washington, Idaho, California, and Oregon. Environmental investigations at produce retailers and wholesalers and with sprout growers and seed distributors were conducted by the Oregon and Washington Departments of Agriculture, the California Department of Health and Human Services, and the FDA. Sprouts and seed samples were collected and cultured for Salmonella.

Initial suspicions that commercially distributed alfalfa sprouts were the source (based on demographics, case distribution, and preliminary food histories) were corroborated by case-control study results. Nine of the first 10 cases recalled alfalfa sprout consumption vs. 0/20 controls (p=0.002). Among cases who had consumed sprouts, 8/9 reported definite (N=5) or possible (N=3) consumption of Hydro-Harvest sprouts (see Product Traceback figure). No other plausible source was identified.

On February 12, press releases were issued announcing a voluntary recall of all Hydro-Harvest sprouts. S. Mbandaka was later cultured from alfalfa sprouts and ungerminated seed collected at the Hydro-Harvest facility. Outbreak-associated cases were eventually identified in four states (see Epidemic Curve): Oregon (N= 40), California (N= 21), Washington (N= 19), and Idaho (N= 5).

A common outbreak pattern was identified by molecular typing (PFGE and micro-restriction fingerprinting). This pattern differed from those of “sporadic” isolates obtained before the outbreak, which were heterogeneous.

The implicated seed came from an 18,000-kg lot milled from alfalfa grown in Southern California. By the embargo date, seed from that lot had been distributed to 4 growers in California, 1 in Florida and Hydro-Harvest in Washington (see Seed Distribution figure); one grower had not started to use it. Cases were eventually linked to only 2 of the 5 growers that had sold sprouts from this lot, however: Hydro-Harvest (OR, WA, ID cases) and brand Y in San Diego (CA). Although documentation was incomplete, the 3 sprouters    (CONTINUED BELOW ↵)

that were not linked to any cases (and who used 41% of the seed) allegedly disinfected with 20,000 ppm calcium hypochlorite or 500 ppm sodium hypochlorite.

This outbreak provided an imperfect natural experiment to assess the efficacy of alfalfa seed disinfection. The observation that—cases were attributable only to the growers who did not disinfect seed–is consistent with, though not proof of, the hypothesis that seed disinfection can reduce the risk of salmonellosis for sprout consumers.

Seed disinfection procedures could not be verified at any site. No sites, including those who allegedly used the FDA-recommended 20,000 ppm calcium hypochlorite soak, maintained any sort of production log or similar, contemporaneous documentation of disinfection practices. The germination procedures, undertaken by Hydro-Harvest and grower Y, could not be independently verified.

Whether contamination within seed lot 8119 was uniform could not be assessed, but, S. Mbandaka was recovered from almost every sample tested (both from ready-to-eat sprouts and from seed collected Hydro-Harvest, grower Y, and the originating warehouse).

The number of reported cases reflects to some extent publicity about the outbreaks, which differed from state to state. Most importantly, data from other outbreak investigations indicate that even the preferred 20,000 ppm calcium hypochlorite disinfection can be inadequate under at least some circumstances. Seed disinfection may be better than nothing, but it is unclear how much.

Sprout growers should routinely document seed disinfection methods, and outbreak investigators should carefully review and report these practices. As these data accumulate, we will get a better sense of how effective seed disinfection is.

Sprouts continue to pose a risk for all consumers. Despite episodic publicity, many consumers remain unaware of this risk. Public health advisors should consider ways to increase awareness of this hazard, particularly among high-risk individuals.

Cal-Farms Parsley

Outbreak: Cal-Farms Parsley 
Product: ParsleyInvestigation Start Date: 10/20/2005
Location: Deschutes County, OREtiology: E. coli (STEC) O157:H7
Earliest known case onset date: 10/08/2005Latest case onset date: 10/27/2005
Confirmed / Presumptive Case Counts: 3 / 61Positive Samples (Environmental / Water) 10 / 2

This 2005 McGrath’s Fish House outbreak of E. coli O157:H7 infections was traced back to parsley from Cal-Farms. The journey to finding the source was interesting: 206 people interviewed, 64 ill people, extremely muddy field conditions, cattle near the flooded parsley fields, and even evidence of heavy deer presence.


On October 20, 2005 we were notified of two E. coli O157:H7 cases from two adjacent counties, Deschutes and Jefferson. Routine follow-up interviews revealed that both ate at a popular restaurant in Bend on the same day, October 12, 2005. Seeing this as a possible common-source exposure, the Deschutes County Health Department in Bend attempted active case finding by obtaining the names of restaurant patrons who used credit cards on October 12th. An initial quick survey identified several other diners who reported gastrointestinal symptoms after their meal. A formal case-control study was launched using an expanded list of credit card names from October 12-15, and a questionnaire based on restaurant menu as well as characterization of illness, if any.

Names from the credit card list were used to look up telephone numbers on the Internet. Deschutes County staff and MPH students from the Acute and Communicable Disease Program (ACDP) of the Oregon Department of Health Services interviewed people. Several people reported gastrointestinal illness after eating at the restaurant. On October 23, 2005, the Deschutes County Health Department decided to issue a press release. Medical alerts were sent to the local physicians and emergency rooms. The press release stimulated additional reports of illness. E. coli O157:H7 isolates were subtyped by pulse-field gel electrophoresis (PFGE) People who reported GI illness were asked to submit stool specimen for confirmation. Some gave specimens to the local health department, while others gave specimens to their doctors.

Environmental Inspection
The Deschutes County Environmental Health section inspected the restaurant. Employees were asked about recent GI illness. Environmental samples and food samples were collected.

We visited Cal-Farms in Oregon City, reviewed their operations, and obtained parsley sales records for September and October. We were shown 2 fields that were the source of parsley harvested in September and October, and collected ~100 convenience samples of soil and parsley stubble on November 7th from both fields. There were no obvious deficiencies in operating conditions to these inexpert observers. Several potential sources of O157 were noted, including livestock in nearby fields (goats, lamas), river water used for irrigation (in drier months), and a deer presence heavy enough to merit repellent next to parsley fields. Because extremely muddy conditions made it difficult to seal sample bags, these were pooled into 6 composite samples – 3 from each field. These samples were enriched and tested by multiplex PCR and cultured on CT-SMAC for O157 at the Institute for Environmental Health (Lake Forest Park, WA; courtesy Dr. Masour Samadpour).

A total of 206 people were interviewed, including people who called in after the press release was issued. Of 120 completed interviews derived from the credit card names, 33 (27%) reported GI illness, of whom 23 (19%) met a working case definition (three or more loose stools in a 24 hour period, and incubation period > 1 day and < 10 days). In total, we identified 64 people who reported three or more loose stools (Figure 1). The median incubation period was 2.6 days. Forty-three (67%) were females, and the median age was 51 (range 4-87). Meal dates for those who were ill were from October 2 – October 22 (Figure 2).
Symptom profile is shown on Table 1. Twenty-one people reported having bloody diarrhea, and only 7 (33%) of these sought medical care. Of those who sought medical care, 6 (86%) were asked to give stool specimen.

Bloody diarrhea1727%

By our initial univariate analysis, crab and artichoke dip, coconut prawns, and vegetables were all significantly associated with illness, but none of these items accounted for more than 20% of the cases.

As we were considering these results, first-enzyme PFGE results on the initial isolate were posted to PulsNet revealing a match to a recently investigated cluster (2 restaurant groups, and 2 sporadic cases) in Washington. Washington epidemiologists had identified parsley (traced to an Oregon supplier and grower) as the likely vehicle for those cases. Armed with this knowledge, we then considered parsley as a potential vehicle for the McGrath’s outbreak. We determined that parsley from the same supplier (Pacific Coast Fruit) and farm (Cal-Farms, Oregon City) had been the sole source of parsley at McGrath’s from XX through ~ October 10 (last delivery date on October 10th). Parsley was used on a large number of menu items, including several side dishes that unfortunately had been omitted from the questionnaire. We reanalyzed our data, using a composite variable standing in for all the food items served with parsley. Limiting the analysis to meal dates from October 12-15 and adding abdominal cramps to the case definition, we found parsley to be significantly associated with illness (OR=3.32, p-value = 0.02). This variable could explain about 80% of the cases. Other food items that were significant included vegetables (OR=3.18, p-value = 0.02) and red potatoes (OR=4.44, p-value = 0.03), both of which had parsley. Interestingly, we learned that the crab and artichoke dip was always served with parsley-topped bruchetta.

Food samples and environmental samples tested negative at the Oregon State Public Health Laboratory (OSPHL). Water and the filter from the ice machine were tested, and were negative.

OR = 3.3295% CI = 1.11-10.39P-value = 0.028
OR = 3.1895% CI = 1.11-9.18P-value = 0.027

Red PotatoesCaseControlTotal
OR = 4.4495% CI = 0.98-21.10P-value = 0.03

Parsley from Cal-Farms was the apparent cause of the McGrath’s outbreak, by extension, all recent Oregon and Washington cases with matching PFGE. The only food item that explained most of the cases was the parsley. The fact that the three confirmed cases matched the Washington cluster suggest a common source. Seven of the 8 cases in Washington had foods that had parsley, and the parsley came from the same grower. The Washington cases had other things in common, such as romaine lettuce, roma tomatoes, red cabbage, and onions, but these food items came from different distributors.

The parsley farm distribute to various places including grocery stores. Pacific Coast Fruit, the distributor for the restaurant, is only one of a number of their customers. It is not clear why this distributor would get a contaminated batch and why only a few restaurants would get contaminated parsley. This wholesaler got about 28% of the sales from 9/1/05 – 10/28/05.
We would have expected to see a lot more cases since the parsley was distributed to other places too.

During the visit to the farm, a number of possible sources of contamination were noted. There were deer repellents, and deer tracts. Samples of parsley and dirt were tested, but were negative.

Only 3 cases were confirmed out of 64 ill people. Five additional cases were tested, but no E. coli O157 were detected. The specimens were collected 9-15 days after onset of diarrhea. Most adults do not excrete the organism after ~2 weeks, and 3 weeks in kids. Three cases were tested through their private physicians, and
were negative. Interestingly, most cases did not seek medical care, even those with bloody diarrhea. Of those with bloody diarrhea, 33% sought care, and of these, 86% were cultured.

The restaurant voluntarily closed on October 24th and was cleaned thoroughly. All the food workers were tested and were negative. The restaurant was reopened on November 1st. We did not hear of new cases after the restaurant was reopened.

The recommendations were:
1) Throw out all left over foods.
2) Emphasize employee hand washing.
3) Ill employees should not work until symptoms resolve, or if they are culture positive for E. coli O157, they should have 2 negative stools before returning to work.