Anatomy of an Outbreak

Kirk Smith, DVM, MS, PhD

Supervisor, Foodborne, Vectorborne, and Zoonotic Diseases Unit

Acute Disease Investigation and Control Section

Minnesota Department of Health

kirk.smith@state.mn.usOffice phone: 651-201-5240

Some Recent Notable Multi-state Foodborne Outbreaks of Salmonellosis

2005 2006 2007 2008

Pot pies•401 cases•42 states

PCA peanut butter•691 cases•46 states

Cake Mix• 25 cases• 9 states

Tomatoes•183 cases•21 states

Hot peppers•1,442 cases•44 states

Veggie Booty•70 cases•23 states

Peter Pan peanut butter•714 cases•48 states

Diseases Reportable to the Minnesota Department of Health

Botulism (Clostridium botulinum)

Campylobacteriosis (Campylobacter sp.)*

Cholera (Vibrio cholerae)*

Cryptosporidiosis (Cryptosporidium sp.)

Enteric Escherichia coli infection (E. coli O157:H7 and other pathogenic E. coli from gastrointestinal infections)*

Giardiasis (Giardia lamblia)

Hemolytic uremic syndrome

Listeriosis (Listeria monocytogenes)*

Salmonellosis, including typhoid (Salmonella sp.)*

Shigellosis (Shigella sp.)*

Toxoplasmosis

Yersiniosis (Yersinia sp.)*

FOODBORNE AND WATERBORNE DISEASES

* Submit isolates or clinical materials to the Minnesota Department of Health

Submission of isolate to public

health lab

Report of case to public health

Becomes ill

Confirmation/ serotyping,

PFGE subtyping

Interview

Lab and epi data

combined

Person eats contaminated food

Stool sample positive

Goes to doctor, stool sample collected

2 - 3 days

1 - 7 days (incubation)

2 - 5 days 1 - 5 days

2 - 4 days

Reportable Bacterial Enteric Pathogen Surveillance in Minnesota

• Isolates must be submitted to the Minnesota Department of Health

• Real-time pulsed-field gel electrophoresis (PFGE) subtyping of all isolates

• Routine, real-time interviews of all cases

Molten agarose

Pulsed-Field Gel Electrophoresis (PFGE)

Lysis

Enzyme digestion (XbaI)

Bacteria

Pulse electrophoresis

DNA

18 hours

1.5 hours

1.5 hours

The National Molecular Subtyping Network for Foodborne Disease Surveillance

Area Labs

National

Database

FoodNet Sites

FDA, USDA Lab

PulseNet Laboratory NetworkPulseNet Laboratory Network

Local Databases

PulseNet National

Databases (CDC)

PulseNet National

Databases (CDC)

Participating LabsParticipating Labs PFGE PatternsPFGE Patterns

Bacterial Isolate Flow from Clinical Labs to Public Health Labs

• Completeness and timeliness of isolate submission to public health labs, and timeliness of serotyping/PFGE subtyping at public health labs, determines the sensitivity of outbreak detection

– i.e., need this for optimal detection of outbreaks (local and multistate) caused by Salmonella, E. coli O157:H7

Minnesota Surveillance Philosophy

• Interview all cases, ASAP

• Collect details on specific exposures

– Restaurant, grocery store names

– Brand names

– Open-ended food histories

• Investigation of all PFGE clusters

– Intensity/resource expenditure depends on the exact nature of the cluster

– Follow leads aggressively

Standard Questionnaire for Salmonella, E. coli O157 cases

Minnesota Surveillance Philosophy

• Interview all cases, ASAP

• Collect details on specific exposures

– Restaurant, grocery store names

– Brand names

– Open-ended food histories

• Investigation of all PFGE clusters

– Intensity/resource expenditure depends on the exact nature of the cluster

– Follow leads aggressively

Response for PFGE Clusters

• Minimum: Compare case interviews

• Maximum: Case-control study

• Food Testing: Before, during, or after case control study

• “Informational” product tracing

Minnesota Approach to Investigation of PFGE Clusters:

Dynamic Cluster Investigation Model

Case #1 Case #2 Case #3 Case #4

Dynamic Cluster Investigation - Pot Pies

Consumed Banquet PP

4

10/4afternoon

“trawling” questionaire

1

Initial trawling questionnaire interview date

9/10

2

9/27

3

10/3night

Re-interviewed cases about frozen foods and pot pies

10/4morning

10/4evening

PP

Exposure

added

Team Diarrhea Fall 2007

Epidemiologic Follow-up of Cases

• Determines the likelihood of identifying the source of an outbreak

Epidemiologic Data are Dirty

• Not all exposed people get sick

• Some people get sick without being exposed

• Not all “exposed people” are really exposed

• Not all “unexposed people” are really unexposed

• Not all sick people are really sick

Presentations of Outbreaks due to Commercially Distributed Food Items

• Cases in community, no obvious common exposure

– Retail food (grocery stores)

• Cases occur among patrons of restaurant(s)

• Cases clustered in institution(s)

• Any combination of above three

Presentations of Outbreaks due to Commercially Distributed Food Items

• Cases in community, no obvious common exposure

– Retail food (grocery stores)

• Cases occur among patrons of restaurant(s)

• Cases clustered in institution(s)

• Any combination of above three

Dole Prepackaged Salad O157 Outbreak

September 27, 2005

• Three O157 isolates with indistinguishable PFGE patterns identified by Minnesota Public Health Laboratory

• PFGE pattern new in Minnesota, rare in United States

– 0.35% of patterns in National Database

Outbreak Investigation - Methods

September 28–29, 2005

• Additional O157 isolates received and subtyped by PFGE

– 7 isolates demonstrated outbreak PFGE subtype

• Supplemental interview form created

• Case-control study initiated

– Age-matched community controls recruited through sequential digit dialing anchored on case’s telephone number

Case-Control Study Results

Exposure Cases Controls p-valueMatched OR* 95% CI†

Any lettuce 9/10 17/26 3.5 0.5–25.0

* OR = odds ratio† CI = confidence interval

9/10Prepackaged lettuce salad 10/26 8.4 1.2–59.6

Brand A prepackaged lettuce salad 9/10 5/23

0.17

0.01

0.00210.1 1.5–67.3

E. coli O157:H7 Cases Associated with Brand A Prepackaged Lettuce by Date of

lllness Onset

Date of Onset2005

15 16 17 18 19 20 21 22 23 24 25

Nu

mb

er o

f C

ases

26 27 28 29 30 1 2 3 414

1

2

3

4

5

6

7

September October

Initial cluster of 3 isolates among MN residents identified.

Case-control study initiated.

Case-control study implicated Brand A salad.

E. coli O157:H7 Cases Associated with Brand A Prepackaged Lettuce by Date of

lllness Onset

Date of Onset2005

15 16 17 18 19 20 21 22 23 24 25

Nu

mb

er o

f C

ases

26 27 28 29 30 1 2 3 414

1

2

3

4

5

6

7

September October

Initial cluster of 3 isolates among MN residents identified.

Case-control study initiated.

Case-control study implicated Brand A salad.

E. coli O157:H7 Cases Associated with Brand A Prepackaged Lettuce

(n=26)

Date of Onset2005

15 16 17 18 19 20 21 22 23 24 25

Nu

mb

er o

f C

ases

26 27 28 29 30 1 2 3 414

1

2

3

4

5

6

7

September October

WI

WI

Minnesota

Additional states

OR

Presentations of Outbreaks due to Commercially Distributed Food Items

• Cases in community, no obvious common exposure

– Retail food (grocery stores)

• Cases occur among patrons of restaurant(s)

• Cases clustered in institution(s)

• Any combination of above three

Salmonella Saintpaul Patron Cases Associated with Restaurant A by Date of Isolate Receipt in

MDH Laboratory, June 2008

JuneDate of Isolate Receipt

Nu

mb

er o

f C

ases

23 24 25 26 27 28 29 30 1 2 3 422

2

3

4

5

6

1

7 Two cases name Restaurant A

Restaurant A Outbreak

June 30, 2008

• MDH and Ramsey County staff visited restaurant

– Interviewed management and employees

– Collected invoices for ingredients used in dishes consumed by cases

– Requested credit card receipts from same time period

– Obtained copies of menu

Salmonella Saintpaul Patron Cases Associated with Restaurant A by Date of Isolate Receipt in

MDH Laboratory, June 2008

JuneDate of Isolate Receipt

Nu

mb

er o

f C

ases

23 24 25 26 27 28 29 30 1 2 3 422

2

3

4

5

6

1

7 Second case names Restaurant A

Visit restaurant

Initial case-control study/traceback results to CDC

Univariate and Multivariate Results of Minnesota Case-Control study

Ingredient

No. cases exposed/total OR p aOR p

Red salsa 13/18 14.7 <0.001 Not significant

Avocado salsa 14/19 7.5 <0.001 Not significant

Mexican garnish

17/19 69 <0.001 Not significant

Red peppers 17/19 43 <0.001 Not significant

Cilantro 18/19 21.4 <0.001 Not significant

Fresh tomatoes

6/19 0.5 0.2 Not significant

Jalapenos 17/19 69 <0.001 62 <0.001

Presentations of Outbreaks due to Commercially Distributed Food Items

• Cases in community, no obvious common exposure

– Retail food (grocery stores)

• Cases occur among patrons of restaurant(s)

• Cases clustered in institution(s)

• Any combination of above three

December 3, 2008

1st 11 cases in MN

Institutional link,Implication of PB

S. Typhimurium Investigation, 2008-2009

November 17-24, 2008

– MDH received 3 outbreak isolates

Early December

– Leading hypothesis in national investigation was chicken

• Restaurant-associated outbreak in another state with three PFGE patterns

Ultimately shown to be a “red herring”

Minnesota S. Typhimurium Investigation

December 10-19, 2008

• MDH received 8 additional outbreak isolates

• All chicken for first 4 cases traced back - source did not converge with other state’s investigation or with each other

• First 8 interviewed cases reported eating peanut butter

– Suspicious, but not enough evidence to implicate one product, or even peanut butter overall, as the vehicle

Minnesota S. Typhimurium Investigation

December 22, 2008

• Medical director of LTCF (LTCF A) in northern MN reports confirmed Salmonella infections in 3 residents

• Specimens from 2 other residents pending

– All five cases confirmed with outbreak strain of S. Typhimurium

• Outbreak cases identified in other institutions

Minnesota S. Typhimurium Investigation

• LTCF A, LTCF B, elementary school all purchased food from a common distributor in Fargo, North Dakota

• Only food common to the 3 institutions was King Nut Creamy Peanut Butter

• Open tub of King Nut peanut butter collected from LTCF A by Minnesota Department of Agriculture on January 5

Week

June July Aug Sept

Cas

es

0

2

4

6

8

10

12

14

16

18

20

22

Cases of Salmonella Typhimurium, by Week of Specimen Collection, Minnesota, June -September, 1995

June July Aug Sept

Cas

es

0

2

4

6

8

10

12

14

16

18

20

22

PFGE Subtype Restaurant A

PFGE Subtype Restaurant B

PFGE Subtype Restaurant C

Other PFGE Subtypes

PFGE Subtype Patterns of Salmonella Typhimurium, by Week of Specimen Collection, Minnesota, June -September, 1995

Week

Selected Enteric Pathogens Reported to MDH, 1996-2008

0

200

400

600

800

1000

1200

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Year

Nu

mb

er o

f C

ases

Campylobacter

Salmonella

E. coli O157:H7

Cryptosporidium

Shigella

1 confirmed case of salmonellosis = 38 actual cases

Interview Options

• In person, telephone, mail, e-mail, web-based

• Rate-limiting step in most outbreak investigations is number of interviewers available to conduct interviews

– staff epidemiologist or sanitarian

– public health nurses and other professionals

– Team Diarrhea

Examine Descriptive Epidemiology

• Age, gender distribution of cases can give clues to vehicle

– e.g., predominately female, median age in 30’s suggests a produce item like lettuce, tomatoes, or sprouts

– e.g., predominately school-aged children, young adults often associated with vehicles like ice cream, microwaveable chicken products, etc.

Generate and Test Hypotheses

• Analytic study designed to test hypothesis

– study design based on study questions, resources, and target population

– regardless of form of study, intent is to determine whether given exposure led to the occurrence of the disease.

Comparison Groups

• Gathering/event (e.g., wedding reception)– non-ill attendees

• Restaurant – well-meal companions, credit card names

• Cluster of bacterial cases identified through routine surveillance– age-matched controls obtained through

sequential digit dialing using case’s telephone prefix

– Friends, neighboring households

– Population surveys

– Cases with similar (but not exact) illness

Analytical Study – Need More than Just a Statistical Association

• A true exposure should account for high proportion of cases (although in some outbreaks there can be multiple vehicles)

• Biologic plausibility

– right incubation, plausible vehicle, etc.

• Distribution of cases vs. distribution of food

• Converging tracebacks

• Explanation of outliers

• Watch out for co-linearity

Analytic Study Problems

•Background rate of consumption is high

– chicken, eggs

•Risk is diffuse – product used in dozens of products

•Food is “cryptic” or eaten as an ingredient

– e.g., spices

•Not enough interviews completed

•Interviewer variability

•Confounding

Sub-Cluster Analyses

• In large outbreaks, there may be sub-clusters of epi-linked cases

– e.g., multiple cases at a restaurant, nursing homes, or school

•Look for common suppliers and food items

•Specific analytical studies in these settings

Why Epidemiologic Links May Not be Identified for Cases in a Cluster

• Cases have imperfect recall

• Cases may not know they were exposed

Secondary

transmission

Why Epidemiologic Links May Not be Identified for Cases in a Cluster

Why Epidemiologic Links May Not be Identified for Cases in a Cluster

• Case did not consume product but may have handled it or was exposed through cross-contamination

Why Epidemiologic Links May Not be Identified for Cases in a Cluster

• Common exposures could be difficult to link

– Lack of specific brand information from patient on a common commodity (e.g., eggs, chicken)

– Traceback inadequacies

• Establishment record-keeping

• Resources available

• There isn’t a common source for all of the cases (or any of them)

– Stable, endemic strains of various bacteria are present in the animal population

• e.g., E. coli O157 PFGE subtype associated with Jack-in-the-Box outbreak is still identified in Minnesota each year

Why Epidemiologic Links May Not be Identified for Cases in a Cluster

Daily Report from MDH Lab to Epi

Cultures Confirmed Yesterday

Daily Report from MDH Lab to Epi

Cultures Confirmed or Subtyped in the Past 30 Days