Epidemiology of Shiga toxin-producing E. coli infections, focusing on North America

1
Epidemiology of Shiga toxin-producing E. coli
infections, focusing on North America

• Patricia M. Griffin, MD
• Enteric Diseases Epidemiology Branch
• Centers for Disease Control and Prevention
• FDA, April 2007

2
(No Transcript)
3
Topics

• Clinical illness
• Surveillance
• HUS
• Outbreaks
• Transmission
• Impact of surveillance and outbreak investigations

4

• Sequence of events in
• E. coli O157H7 infection

non-bloody diarrhea, abdominal cramps
1-2 days
bloody diarrhea
8
92
6 days
resolution
HUS
5
Surveillance for E. coli O157 in FoodNet

• Count
• every E. coli O157 isolated by clinical labs in
  the catchment area
• Calculate
• annual population-based rates
• trends over time

6
FoodNet Catchment, 200544 million persons15 of
U.S. population
7
Percent of clinical labs screening all stools
for E. coli O157, 1985-2003
of labs
FoodNet surveys
National sample
69
Western states outbreak
Boyce, J Clin Micro 1995, and FoodNet,
unpublished preliminary data
8
Relative rate (compared with 1996-1998 baseline)
of E. coli O157 infections, FoodNet, thru 2005
Relative rate
Decreased 29 since 96-98
9
Relative rate (compared with 1996-1998 baseline)
of E. coli O157 infections, FoodNet, thru 2005
Relative rate
Crude incidence 1.06 illnesses per 100,000
persons
10
Relative rate (compared with 1996-1998 baseline)
of E. coli O157 infections, FoodNet, thru 2005
Relative rate
Crude incidence 1.06 illnesses per 100,000
persons
11
Incidence of E. coli O157 infections, by state,
1999-2002
Isolates /100,000 pop/ year
3.0 6.2 1.7 2.9 0.9 1.6 0.2 0.8
12
Rate of E. coli O157 infectionsby setting, US
Cases/ 100,00 persons/ year
Mead, unpublished data from 1993-1996
13
Estimates of annual frequency of E. coli O157
infections, US

• FoodNet sites (2005 data)
• 1.06 culture-confirmed infections / 100,000
  persons
• But, many ill persons dont have stool cultured
  and many labs do not routinely test for E. coli
  O157
• Multipliers derived from surveys help us estimate
  the true number of infections
• Estimates for United States (Mead et al. EID
  1999)
• 73,000 infections
• 2,000 hospitalized
• 60 deaths

14
Surveillance for non-O157 STEC the Shiga toxin
enzyme immunoassay (EIA) has been a blessing

• Improves detection of non-O157 STEC illnesses and
  outbreaks
• Clinical labs can submit Shiga toxin-positive
  broths to State Health labs so STEC are isolated
• CDC serotypes STEC for State Health labs

15
Human isolates of non-O157 STEC serotyped by CDC,
1983-2002
of isolates
N 940 isolates
55 O groups, each lt1
Brooks, JID 20051921422
16
Human isolates of non-O157 STEC serotyped by CDC,
1983-2002
of isolates
N 940 isolates
55 O groups, each lt1
70
Brooks, JID 20051921422
17
Surveillance for non-O157 STEC the Shiga toxin
EIA has been a curse (as well as a blessing)

• In adopting the EIA, some labs have abandoned use
  of selective media to isolate E. coli O157
• Some clinical labs discard Shiga toxin-positive
  specimens without obtaining an isolate, so
• serotype not determined
• E. coli O157 strains not sub-typed for
  surveillance and outbreak detection
• Clinical labs do not use the EIA routinely
• in 2003, only 3 of clinical labs in FoodNet had
  ever used it
• however, use is increasing

18
(No Transcript)
19

• Clinical laboratories should strongly consider
  including STEC O157 in their routine bacterial
  enteric panel
• The best way to identify all STEC infections is
  to screen all stool samples..for Shiga toxins
• Laboratories that use a Shiga toxin EIA.should
  culture all positive broths.
• When a Stx-positive broth does not yield STEC
  O157, the broth...should be quickly forwarded to
  the statelaboratory for identification of
  non-O157 STEC.
• All non-O157 STECshould be sentto CDC.

20
Surveillance for STEC, Minnesota
Diarrhea Diarrhea Diarrhea DHUS
Site Minneapolis/St Paul HMO Rural hospital Total Entire state
Year 1996-02 2000-02 1997-05
No. 114 77 191 89
O157 44 65 52 91
Non-O157 56 35 48 9
Number of patients with STEC serogroup
determined Includes 3 siblings in an outbreak
Minnesota Dept of Health, unpublished data
21
Topics

• Clinical illness
• Surveillance
• HUS
• Outbreaks
• Transmission
• Impact of surveillance and outbreak investigations

22
National Prospective DHUS Study, 1987-1991

• Convenience sample of nephrologists and
  hematologists
• Inclusion criteria
• Hct lt30
• microangiopathic changes
• increased creatinine
• diarrhea in 21d before diagnosis
• Requested stool, acute and convalescent sera,
  questionnaire

Banatvala, JID 2001
23
DHUS Patients (N83)
U.S. National HUS Study, 1987-1991

• 73 children, 10 adults
• 55 lt5yrs old (range, 3 mos 64 years)
• From 16 states, all 4 census regions

Banatvala, JID 2001
24
Clinical features of 83 DHUS patients
U.S. National HUS Study, 1987-1991
Diarrhea 100 (by definition)
Bloody stools 73
Dialysis 55
Red cell transfusion 37
Seizure 22
Death 6 children, 20 adults
Banatvala, JID 2001
25
Stool culture results (n70 with stool culture)
U.S. National HUS Study, 1987-1991

• 43 of patients had stool cultures that yielded
  STEC
• Low isolation rate likely partly due to freezing,
  shipping, culture delays
• Serotypes
• 86 were E. coli O157H7
• 14 were non-O157 STEC

Banatvala, JID 2001
26
Serology (n66 with serology)
U.S. National HUS Study, 1987-1991

• 81 had IgG or IgM antibodies to O157 LPS
• Titer remained elevated at 2 months

Banatvala, JID 2001
27
Overall (N83 DHUS patients)
U.S. National HUS Study, 1987-1991

• 73 children 73 had STEC infection by culture or
  serology
• 10 adults 70 had STEC infection by culture or
  serology
• All 3 post-partum women had E. coli O157H7
  infection

Banatvala, JID 2001
28
Patients with both serology and stool culture
(n55)
U.S. National HUS Study, 1987-1991

• 18 had no evidence of STEC infection. Compared
  to those with STEC infection, those with no
  evidence of STEC infection
• less likely had bloody diarrhea
• less likely had onset in warm months (48 vs 75)
• 1 had pancreatic cancer
• 82 had evidence of STEC infection
• 98 of these had evidence of E. coli O157
  infection
• 3 of 4 with non-O157 STEC isolated from stool had
  antibodies to O157 LPS

Banatvala, JID 2001
29
Other studies -- stool cultures

• Among HUS cases tested within 6 days of onset of
  diarrhea, proportion with E. coli O157H7
  isolated
• United States (25 cases) 96
  (Tarr, J Infect Dis 1990)
• Canada (30 cases) 87
  (Rowe, Epidemiol Infect 1993)

30
Other studies -- serology

• Proportion of DHUS cases with O157 LPS
  antibodies
• England 73 (Chart, Lancet 1991)
• Central Europe 73 (Bitzan, Epidemiol Infect
  1993)
• France 67 (Decludt, Epidemiol Infect 2000)

31
Summary STEC serotypes in diarrhea vs HUS

• Diarrhea
• E. coli O157 and non-O157 STEC are isolated with
  similar frequency
• HUS
• E. coli O157 causes gt90 of STEC-associated HUS

32
Major predictive factors for STEC-associated HUS

• Host factors
• Extremes of age (lt5 or gt65 years)
• Bloody diarrhea
• Vomiting (early in illness)
• White blood cell count gt13,000 (early in illness)
• Microbial factors
• E. coli O157 strain produces only Shiga toxin 2
  (not both Shiga toxin 1 and 2, not only Shiga
  toxin 1)
• Treatment factors
• Treatment of diarrhea with antimotility agent
• Treatment of diarrhea with antibiotic

33
Cohort study of E. coli O157H7 infection--
factors related to the risk of hemolytic uremic
syndrome

• Conducted in 10 FoodNet Sites
• Evaluate risk of HUS by
• microbiologic characteristics of the infecting
  strain
• patient clinical and laboratory features
• antimicrobial therapy
• other treatment
• Began January 2006
• all patients with E. coli O157 infection
• medical chart review
• interview

34
Topics

• Clinical illness
• Surveillance
• HUS
• Outbreaks
• Transmission
• Impact of surveillance and outbreak investigations

35
1993 genesis of PulseNet, 1993

• 1993 investigated western states E. coli O157
  outbreak in which gt700 persons ill and 4 children
  died
• Developed pulsed-field gel electrophoresis (PFGE)
  subtyping method

• Followed by increased demand for subtyping
• CDC unable to subtype isolates in a timely manner
  

of outbreaks
Large western states outbreak
36
In outbreak investigations, time is not on your
side

• The faster you track and control the source of an
  outbreak, the more cases you prevent
• In a large country like the United States
• sending isolates to a central lab and subtyping
  them there is time consuming
• determining subtypes in state labs can be fast
• Subtype patterns from many state labs can be
  compared quickly
• by transmitting patterns over the internet

37
PulseNet USA

• A national network of public health and food
  regulatory laboratories dedicated to molecular
  surveillance of foodborne infections
• detects and investigates clusters of isolates
  with same molecular subtype
• coordinated by CDC, which has central database
• current method is PFGE

38
Ground beef outbreak, 2002

• First cases found in Colorado
• PulseNet
• posted outbreak strain pattern
• identified 45 persons with outbreak strain in 11
  states
• Beef identified as cause
• Outbreak strain isolated from ground beef
• beef came from one meat processing plant
• Outbreak stopped after recall of 18.6 million
  pounds of ground beef
• entire meat industry worried about bad publicity

39
1993 western States E. coli O157 Outbreak (before
PulseNet)
outbreak detected 1993 732 ill, 4 deaths
39 d
2002 Colorado E. coli O157 Outbreak (after
PulseNet)
outbreak detected 2002 45 ill, no deaths
18 d
40

• Outbreaks of E. coli O157 Infections by State,
  1998-2005

Includes multi-state outbreaks
41
E. coli O157 outbreaks by year, U.S., 1982 -
2005(N 489 outbreaks)
Epidemiologists stimulated better outbreak
reporting
Lab created PulseNet
E. coli O157 became nationally notifiable
37
Large western states outbreak
Rangel, Emerg Infect Dis, 2005 and preliminary
data, CDC
42
Median number of ill persons per E. coli O157
outbreak, U.S., 1982-2005
Median no. ill persons per outbreak
5 ill persons/outbreak
Rangel, Emerg Infect Dis, 2005
43
Outbreaks of non-O157 STEC infections, U.S.,
1990-2005
N 12 outbreaks
Number of outbreaks
Data from 2004 on is preliminary
44
Serogroups of non-O157 STEC outbreaks, U.S.,
1990-2006(N12)

• O111 (5 outbreaks)
• O121 (3 outbreaks)
• O26, O45, O103, O104 (1 outbreak each)

data from 2004 to 2006 is preliminary
45
Suspect modes of transmission (few confirmed) in
non-O157 STEC outbreaks, U.S., 1990-2006)
(N 12)

• Food milk, salad, lettuce, apple cider, punch (5
  outbreaks)
• Lake recreational water (2 outbreaks)
• Person-to-person in child care center (1)
• Undetermined (4)

data from 2004 to 2006 is preliminary
46
Topics

• Clinical illness
• Surveillance
• HUS
• Outbreaks
• Transmission
• Impact of surveillance and outbreak investigations

47
Study areas, Argentina
South America
Argentina
Mendoza
Buenos Aires
48
Case-control study of risk factors for E. coli
O157 and other STEC infections, Argentina,
2001-2002

• Cases were children (lt16 years old) with
• culture-confirmed STEC diarrhea, or
• diarrhea-associated HUS
• Enrolled 150 cases and 299 well control children

Rivas, Sosa-Estani, Rangel, Caletti, Valles,
Roldan, Balbi, deMollar, Amoedo, Milwebsky,
Chinen, Hoekstra, Mead, Griffin, manuscript in
preparation
49
Major risk factors for STEC infection, Argentina,
2001-2002

• Eating undercooked beef, including
• teething on pieces of tender beef
• drinking jugo de carne
• Residing in or visiting a place with farm animals
• Contact with a child lt5 years old with diarrhea

Rivas, Sosa-Estani, Rangel, Caletti, Valles,
Roldan, Balbi, deMollar, Amoedo, Milwebsky,
Chinen, Hoekstra, Mead, Griffin, manuscript in
preparation
50
Major protective factors, Argentina, 2001-2002

• Caregiver always washing hands after handling raw
  beef
• Child eating more than the average variety of
  fruits and vegetables

Rivas, Sosa-Estani, Rangel, Caletti, Valles,
Roldan, Balbi, deMollar, Amoedo, Milwebsky,
Chinen, Hoekstra, Mead, Griffin, manuscript in
preparation
51
Swaziland outbreak
52
Massive outbreak of E. coli O157 infections,
Swaziland, 1992

• 62 of Swaziland is cattle pasture
• Early 1990s
• Years of drought
• Cattle aggregated close to water and vegetation

53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
Swaziland, 1992 (continued)

• September 1992
• Inadequate water and pasture for cattle
• 7x increase in cattle deaths
• Cattle defecated and died in stream beds
• October 1992 heavy rains
• November 1992
• marked increase in human bloody diarrhea
• E. coli O157 isolated from human stools

57
Epidemiologic findings

• Risk factors for human illness
• Consumption of beef
• Drinking untreated water
• Estimated thousands of persons ill

Effler, Isaacson, Arntzen, Heenan, Canter,
Barrett, Lee, Mambo, Levine, Zaidi, Griffin, EID
2001
58
Specimens that yielded E. coli O157 outbreak
strain, Swaziland and South Africa

• Human feces
• Cattle feces
• Water
• surface
• river
• borehole
• water stored in home
• Cooked maize

Sites of positive cultures spanned several
hundred kilometers
59
(No Transcript)
60
E. coli O157H7 August 1998
61
Proportion of illnesses due to each mode of
transmission in 350 E. coli O157 outbreaks,
U.S.,1982-2002
Mode Illnesses in outbreaks (N8,598 illnesses)

Foodborne 61
Drinking water 15
Unknown 9
Person-to-person 8
Animal contact 4
Recreational water 3
Lab acquired lt1
Rangel, EID-April 2005
62
Proportion of illnesses due to each mode of
transmission in 350 E. coli O157 outbreaks,
U.S.,1982-2002
Mode Illnesses in outbreaks (N8,598 illnesses)

Foodborne 61
Drinking water 15
Unknown 9
Person-to-person 8
Animal contact 4
Recreational water 3
Lab acquired lt1
Rangel, EID 2005
63
Agricultural fairs

• Traditionally held in autumn in rural areas
• Bring farm animals, food, and people into close
  proximity
• Outbreaks at 10 agricultural fairs, 1999-2005
• 8 states
• gt1400 persons ill
• many children with HUS

64
(No Transcript)
65
North Carolina county fairgrounds
66
Fall 2004 108 ill, mostly children 14
developed HUS
67
Petting zoo implicated

• Illness associated with visiting petting zoo,
  touching animal environment, touching animals
• Cultures of patients, animals, and environment
  yielded outbreak strain

Well
HUS
68
Proportion of illnesses to each mode of
transmission in 350 E. coli O157 outbreaks due,
U.S.,1982-2002
Mode Illnesses in outbreaks (N8,598 illnesses)

Foodborne 61
Drinking water 15
Unknown 9
Person-to-person 8
Animal contact 4
Recreational water 3
Lab acquired lt1
Rangel, EID 2005
69
Prime suspects
70
Proportion of illnesses in foodborne E. coli
O157 outbreaks due to various foods, 1982-2002
Vehicle illnesses in foodborne outbreaks (N5,269 illnesses)

Ground beef 33
Other beef 11
Produce 34
Other 4
Dairy products 6
Unknown 12

1 bovine products, 50

Rangel, EID 2005
71
Percent of foodborne E. coli O157 outbreaks due
to beef, 1982 - 2005
of outbreaks
35
Rangel EID 2005, and CDC unpublished data
72
Proportion of illnesses in foodborne E. coli
O157 outbreaks due to various foods, 1982-2002
Vehicle illnesses in foodborne outbreaks (N5,269 illnesses)

Ground beef 33
Other beef 11
Produce 34
Other 4
Dairy product 6
Unknown 12
2 produce, 34
Rangel, EID 2005.
73
Cattle feedlot
74
Leafy green vegetables implicated in E. coli O157
outbreaks, U.S., 1973 2005

• None implicated until 1995
• 26 outbreaks between 1995 and 2005
• Lettuce and lettuce salads 21 outbreaks
• Cabbage 3 outbreaks
• Parsley 2 outbreaks
• Spinach 1 outbreak

75
(No Transcript)
76
Patient and strain characteristics

• Patients (N 206)
• 70 female
• 105 (51) hospitalized
• 31 (15) HUS
• 3 (1.5) died (toddler and 2 elderly women)
• Strain produced only Shiga toxin 2

77
Number of patients with illness and HUS, by age
group
Number of patients
8 HUS
24 HUS
13 HUS
32 HUS
Age Group
78
Summary of spinach outbreak

• Large international outbreak with high rate of
  HUS
• Rapid action by health authorities
• too late to have much impact
• Prevention measures needed to improve safety of
  leafy green vegetables consumed raw

79
Topics

• Clinical illness
• Surveillance
• HUS
• Outbreaks
• Transmission
• Impact of surveillance and outbreak investigations

80
Reminder -- decreased incidence of E. coli O157
infections in 2003
Year
81
Impact of E. coli O157 surveillance and outbreak
investigations on beef safety

• 1993 fast-food hamburger outbreak caused gt700
  illnesses and 4 children died
• 1993 Model Food Code for restaurants
  incorporated temperature guidelines for cooking
  ground beef
• 1994 E. coli O157 made an adulterant in ground
  beef
• mandatory recalls

82
Impact of E. coli O157 surveillance and outbreak
investigations on beef safety

• 1996 New meat inspection system created based on
  Hazard Analysis and Critical Control Points
  (HACCP)
• focused on cleaner carcasses
• included microbiologic testing for Salmonella

• 1996 FoodNet created to track incidence of E.
  coli O157 infections
• issues annual report card

83
Impact of E. coli O157 surveillance and outbreak
investigations on beef safety (continued)

• 2001 much bad publicity due to beef recalls
• beef industry decided to collaborate, not
  compete, on food safety
• 2002 ground beef from one plant caused
  multi-state outbreak, with gt18 million pounds of
  beef recalled
• Industry began test and hold
• test all lots of beef trim for E. coli O157
• positive lots not distributed

84
Possible reason for the decrease in incidence of
E. coli O157 infections -- ground beef safer

• Fast-food hamburger are safer
• Fast-food chains are
• requiring suppliers to provide cleaner beef
• cooking hamburgers better
• Eating from a fast-food restaurant was risk
  factor in a 1990 study, not in a 1996 study
• Consumers cooking ground beef better
• Ground beef is less contaminated

85
Percent of people who recently consumed pink
ground beef, 1996-2002
Percent
15
12
10
6
5
0
1996-97
1998-99
2000-01
2002-03
Year
FoodNet population survey, weighted estimates
86
Percent of ground beef samples that yielded E.
coli O157H7, 2000-2005
positive
Outbreak with recall of 18 million ground beef
www.fsis.usda.gov
87
Other impacts of E. coli O157 surveillance and
outbreak investigations

• 1996 unpasteurized apple juice caused HUS, 1
  death.
• 2001 apple juice shipped interstate must either
  be treated to kill pathogens or have a warning
  label
• Since then, marked decrease in juice outbreaks

88
Other impacts of E. coli O157 surveillance and
outbreak investigations

• 1999-2004 contact with animals and their
  environment at county fairs caused HUS. As a
  result..

89
(No Transcript)
90
(No Transcript)
91
(No Transcript)
92
Summary of major messages

• Surveillance
• Decline in incidence of E. coli O157 infections
  by 2003, but recent increase
• HUS
• E. coli O157 is major cause
• Pathogen and host risk factors still being
  determined
• Outbreaks
• Median size decreased to 5 ill persons

93
Summary of major messages (continued)

• Transmission
• Petting zoos can be hazardous
• Beef is still an important source
• Produce consumed raw is a major source
• Surveillance and outbreak investigations prompt
  industry changes that decrease illnesses
• Ground beef is safer
• Apple cider and juice is safer
• Petting zoos now have guidelines
• Produce needs work

94
This talk was made possible by the efforts of
people in many groups, including

• PulseNet
• OutbreakNet
• Enteric Diseases Laboratory Branches
• State Public Health Departments

95
and current and former members of the Enteric
Diseases Epidemiology Branch
96
Thank you
Produce the next frontier
Return to Main Menu.
The conclusions and opinions expressed herein
are those of the presenter and do not
necessarily represent the views and policies of
CDC and DHHS.