Use of Syndromic Surveillance in the investigation of Salmonella Wandsworth Outbreak

1
Use of Syndromic Surveillance in the
investigation ofSalmonella Wandsworth Outbreak

• Erin L. Murray, MSPH
• New York City Department of Health and Mental
  Hygiene

2
Overview

• Syndromic surveillance in NYC
• Multi-state S. Wandsworth outbreak
• Diarrhea activity in NYC, late June 2007
• Epidemiologic investigation

3
NYC Syndromic Surveillance Systems

• Emergency Department (ED)
• 46/61 EDs
• 90 of ED visits
• 8500 visits/day
• Analyzed and monitored 7 days/week
• Over-the-counter (OTC) sales
• 27,500 units sold/day
• Analyzed 7 days/week

4
NYC Experience Investigating Syndromic
Surveillance Signals

• Infrequently indicate an outbreak
• Only 2 investigated spatial signal linked to real
  outbreak
• Investigations are resource intensive
• Rarely conduct large scale investigations

5
Multi-state S. Wandsworth Outbreak

• May 16 CDC announced
• 23 states involved
• 70 cases
• Onsets February 26 July 4, 2007
• 77 reported bloody diarrhea
• 92 3 years of age
• Linked to Veggie Booty snack food
• FDA recall on June 28

6
Diarrhea Activity
Notification of multi-state S. Wandsworth outbreak
CDC
ED
Results assessed
OTC
21 T
25 M
26 T
20 W
27 W
28 T
23 S
22 F
24 S
June 2007
7
Diarrhea Trends June 20-26

• 1 area with higher proportion of ED visits for
  diarrhea than others
• 3 of 6 hospitals had larger increases than other
  hospitals

8
Diarrhea Activity
FDA
Veggie Booty recall
Health Alert on S. Wandsworth
outbreak
NYCDOHMH
Notification of multi-state S. Wandsworth outbreak
CDC
ED
OTC
21 T
25 M
26 T
20 W
27 W
28 T
23 S
22 F
24 S
June 2007
9
Objective

• Determine if increases in diarrheal illness
  detected through syndromic surveillance in June
  2007 were associated with a multi-state outbreak
  of S. Wandsworth

10
Investigation Methods

• Case-control study initiated July 3 among lt5 year
  olds
• Cases
• Children who visited ED for diarrhea at 3
  hospitals with noted increases in diarrheal
  illness from June 20-26
• Controls
• Children who visited ED for Other conditions at
  same hospitals from June 20-26
• Other Chief complaints that did not mention
  respiratory, GI, asthma, or severe illness
  symptoms

11
Investigation Methods

• Patient contact information obtained from
  hospitals
• Questionnaire administered to parent or guardian
  of cases and controls
• History of condition resulting in ED visit
• Included diarrhea and stool culture questions
• Food exposures
• Veggie Booty
• Other exposures
• Ill family members

12
Investigation Analysis

• Two case/control designations
• ED chief complaint of diarrhea
• Interview report of diarrhea
• Odds ratios calculated
• 0.5 correction in every cell if zero-cell present
• 95 confidence intervals calculated

13
Results

• 51 of 39 cases in diarrhea syndrome completed
  interviews (n20)
• 43 of 89 controls completed interviews (n38)
• Overall response 45

14
Case/Control Designations
Interview
Chief Complaint
1 childs diarrhea symptom information was
missing on interview
15
Results by Case/Control Status

• Chief complaint (diarrhea syndrome)
• OR 1.7
• 95 CI (0.1, 29.4)

4 controls had missing Veggie Booty consumption
information
16
Results by Case/Control Status

• Interview (reported diarrhea)
• OR 8.3
• 95 CI (0.4, 182.7)

4 controls had missing Veggie Booty consumption
information 1 childs diarrhea symptom
information was missing on interview
17
Laboratory Findings

• 7 of 20 cases reported providing stool cultures
  in ED
• None were children who ate Veggie Booty
• Only confirmed 2 had cultures taken
• 1 negative
• 1 positive for Campylobacter

18
Limitations

• Delay from signals to interviews - Recall bias
• 11 days from OTC signal day to request for
  contact information from EDs
• 6 additional days to obtain all patient contact
  information
• Small sample size - Limited power
• Could not differentiate excess diarrhea from
  baseline diarrhea - Bias towards null

19
Retrospective Assessment

• Could an integration of other data sources have
  prompted an earlier detection and response to
  this outbreak?
• Why didnt we detect an association between
  Veggie Booty and diarrhea ED visits?

20
S. Wandsworth in NYC

• 8 NYC residents culture confirmed
• Onset dates March 4 May 19, 2007
• All lt5 years of age
• Residents of 3 NYC counties

21
Bronx
Hospital3/4/2007
Queens
Manhattan
Zip Code 3/14/2007
Brooklyn
22
Zip Code 4/15/2007
Bronx
Queens
Manhattan
Brooklyn
23
Retrospective Assessment

• Could an integration of other data sources have
  prompted an earlier detection and response to
  this outbreak?
• Why didnt we detect an association between
  Veggie Booty and diarrhea ED visits?

24
Estimated Total Number of Cases in NYC

• FoodNet (CDC)
• 39 cases of Salmonella for each culture confirmed
  case
• Expected cases in NYC ?
• 8 x 39 312
• 4 per day (March 4 May 19)

25
Expected Number of ED Visits

• FoodNet
• 12 of people with acute gastrointestinal illness
  seek care (all ages)
• Assumptions
• May be higher among young children
• Higher among those with bloody diarrhea
• Most would seek care from primary care provider
• 25 of NYC cases visited ED
• Of ED visits, some would not be identified by
  diarrhea syndrome

26
Expected Number of ED Visits

• S. Wandsworth outbreak ? very few (1/day)
  additional ED diarrhea visits
• Limited analysis 7 day period (6/20-26)
• Limited analysis to 3 (of 61) EDs

27
Conclusions

• Outbreak unlikely to be captured by syndromic
  surveillance
• Small size
• Misalignment of syndromic signals and cases
• Spatially
• Temporally
• Delay in laboratory reporting

28
Recommendations

• Create better methods for visualizing syndromic
  surveillance data on daily basis
• Integrate data from multiple systems
• Improve geographic visualization
• Incorporate laboratory data
• Increase the number of laboratories reporting
  electronically

29
Recommendations

• Develop rapid methods for obtaining patient
  information
• Identify appropriate contact person for data
  extractions during an emergency
• Identify methods on both sides that will ease
  burden of data extraction
• Create secure transfer methods for patient records

30
Acknowledgements

• Richard Heffernan
• Lara Kidoguchi
• Bernadette Fascina
• Don Olson
• Debjani Das
• Andrea Fischer
• Giselle Merizalde
• Glenette Houston
• Staff from hospitals involved in investigation

• Alice Yeung
• Don Weiss
• Alejandro Cajigal
• Martha Felder
• Timothy Clark
• Kristin Wall
• Chris Goranson
• Trang Nguyen
• University of Pittsburgh (RODS)

31
(No Transcript)
32
(No Transcript)
33
(No Transcript)
34
Expected Number of ED Visits

• S. Wandsworth outbreak ? very few (1/day)
  additional ED diarrhea visits
• Limited analysis 7 day period (6/20-26)
• Expected 7 cases in EDs citywide
• Limited analysis to 3 (of 61) EDs
• Expect .1 cases/hospital ? .3 cases total
• Observed 2 possible cases

35
Limitations

• S. wandsworth outbreak began in late-February
• Compared proportion of diarrhea visits in late
  June to those in early/mid-June
• Comparison baseline contained Salmonella outbreak

36
Findings

• 8-fold increase in odds of eating Veggie Booty
  among children with diarrhea
• Not significant very large confidence interval
• No laboratory confirmation of Salmonella
• Chief complaint doesnt capture all symptoms

37
Findings

• Exposed difficulties in obtaining high volume of
  patient contact information from hospitals
• Identifying correct person to handle request
• Ability to create electronic database of specific
  records
• Security of transferring files to NYCDOHMH

38
NYC Syndromic Surveillance Systems

• Emergency department visits
• Pharmacy Sales
• RODS
• Over-the-counter (OTC)
• NYC specific system
• OTC
• Prescription
• EMS calls
• Hospital outpatient visits
• School nurse clinic visits

39
Investigation Analysis

• Two case/control designations
• ED chief complaint of diarrhea
• Interview report of diarrhea

• Odds ratios calculated
• 0.5 correction in every cell if zero-cell present
• 95 confidence intervals calculated

40
Case/Control Designations
41
NYC Experience Investigating Syndromic
Surveillance Signals

• 375 statistically significant spatial signals
  from 2002-2005
• 75 investigated
• 13 (of 117) for diarrhea syndrome
• 1 diarrhea signal determined to be true outbreak
• Signal investigations are resource intensive and
  infrequently indicate an outbreak
• Rarely conduct large scale investigation