Lessons Learned from the National Syndromic Surveillance Conference

1
Lessons Learned from the National Syndromic
Surveillance Conference

• Sponsored by the
• Centers for Disease Control and Prevention
• NYC Department of Health and Mental Hygiene
• New York Academy of Medicine
• September 23-24, 2002
• New York City

2
What is Syndromic Surveillance?

• Passive Systems
• Minimal burden
• Designed to detect and monitor large usual/mild
  illnesses
• Active Systems-
• Educational Outreach Tool
• Designed to detect and report small
  unusual/severe syndromes

3
Legal MandateWho Should be Doing This?

• Public Health Practice
• Local health officers shall exercise due
  diligence in ascertaining the existence of
  outbreaks of illness or the unusual prevalence of
  diseases, and shall immediately investigate the
  causes of same
• New York State Sanitary Code, 10 NYCRR Chapter 1,
  Section 2.16(a)
• Research Development
• Non-traditional data sources
• Academia (training) contractors
• Authorized agents of public health departments

4
Privacy and Confidentiality

• Health departments have strong tradition of
  maintaining security of confidentiality
  information
• Public health provisions in HIPAA
• Data collected under auspices of bioterrorism
  surveillance de-linked from any identifiers for
  non-BT surveillance

5
Goals

• Early detection of large outbreaks
• Characterization of size, spread, and tempo of
  outbreaks once detected
• Monitoring of disease trends

6
Potential Syndromic Surveillance Data Sources

• Day 1- feels fine
• Day 2- headaches, fever- buys Tylenol
• Day 3- develops cough- calls nurse hotline
• Day 4- Sees private doctor flu
• Day 5- Worsens- calls ambulance
• seen in ED
• Day 6- Admitted- pneumonia
• Day 7- Critically ill- ICU
• Day 8- Expires- respiratory failure

7
Potential Syndromic Surveillance Data Sources

• Day 1- feels fine
• Day 2- headaches, fever- buys Tylenol
• Day 3- develops cough- calls nurse hotline
• Day 4- Sees private doctor flu
• Day 5- Worsens- calls ambulance
• seen in ED
• Day 6- Admitted- pneumonia
• Day 7- Critically ill- ICU
• Day 8- Expires- respiratory failure

Pharmaceutical Sales
Nurses Hotline
Managed Care Org
Absenteeism
Ambulance Dispatch (EMS)
ED Logs
Traditional Surveillance
8
Data Transfer
EMS
9
Data requirements

• Core variables
• Hospital name
• Date of visit
• Time of visit
• Age
• Sex
• Chief complaint (free text)
• Home zip code
• /- Unique identifier
• Discharge diagnosis not generally available in
  timely manner
• Need to consider response protocols patient
  identification, logistics

10
Electronic coding of chief complaints into
clinical syndromes

• Performed in SAS
• Text-string recognition
• Mutually exclusive vs. overlapping
• Hierarchy of coding
• Iterative refinement of syndrome definition
• Entire dataset can be recoded easily allows for
  changes in definition and addition of new
  syndromes

11
Electronic ED logs
AGE SEX TIME CHIEF COMPLAINT
ZIP 15 M 0104 ASSAULTED YESTERDAY,
RT EYE REDDENED.11691 1 M 0117 FEVER 104 AS
PER MOTHER. 11455 42 F 0320
11220 4 F
0145 FEVER, COUGH, LABORED BREATHING.
11507 62 F 2251 ASTHMA ATTACK.
10013 48 M 1304 SOB AT HOME.
10027 26 M 0602 C/O
DIFFICULTY BREATHING. 66 M
1701 PT. MOTTLED AND CYANOTIC.
10031 Text Recognition with SAS
IF index(cc,"FEV")gt0 or index(cc,"HIGH
TEMP")gt0 or index(cc,"NIGHT SWEAT")gt0
or (index(cc,"CHILL")gt0 and index(cc,"ACHILLES"
)0) or index(cc,"780.6") etc. then
FEVER1
12
Data Summary

Syndromic Grouping Call-Type Chief Complaint Drug Class
Geographic Grouping Pickup Zip Home Zip Hospital Store Zip
Other Information Age Gender
Follow-up Possible Yes

ED
Pharmacy
EMS
13
Data Summary

Daily Volume 3,000 calls 6,500 visits 6,000 Rx 26,000 OTC
Coverage gt95 65-70 30
Prospective Data Collection March 1998 October 2001 August 2002
Analytic Methods Cyclical Regression Scan Statistic CUSUM Scan Statistic In development

ED
Pharmacy
EMS
14
Data Summary

Syndromes ILI Respiratory Febrile GI
Detection Limit (city-wide) 50 calls 100 visits
Detection Limit (localized) 10 calls 10-20 visits

ED
EMS
15
Denominator Surveillance is Less Sensitive than
Syndromic
Total Visits
Fever/Respiratory
GI/ Vomiting
16
(No Transcript)
17
Selected Antibiotic and Antiviral
Prescriptions 1997-2002
18
ED Respiratory Visits, Nov-May
19
EMS calls
Pharmacy Antiviral Rx
Subway worker- flu
20
West Nile Virus ActivityThrough September 2001
21
Tabletop Drills
REDEX (2001) Test of 911-EMS System
SANDBOX (2002) Test of ED System
22
Nov 12 9.17 am Flight AA 587 Crashes in
Rockaways Respiratory Zip Code Signal (7 zips)
27 Observed / 10 Expected plt0.001
Hospital Signal 31 Observed/ 16
Expected plt0.05
23
(No Transcript)
24
Investigation

• Key Questions
• True increase or natural variability?
• Bioterrorism or self-limited illness?
• Available Methods
• Drill down
• Query clinicians/ laboratories
• Chart reviews
• Patient followup
• Increased diagnostic testing

25
Investigation

• Checked same-day logs at 2 hospitals
• Increase not sustained
• Chart review in one hospital (9 cases)
• Smoke Inhalation (1 case)
• Atypical Chest Pain/ Anxious (2 cases)
• Shortness of Breath- Psych (1 case)
• Asthma Exacerbation (3 cases)
• URI/LRI (2 cases)

26
Future Directions

• Research Agenda
• More evaluations- Simulation models and spiked
  validation datasets
• Better cluster detection software
• Signal Integration
• Optimizing response protocols- Inexpensive (and
  accurate) rapid diagnostics
• Emergency Department Surveillance
• Chief Complaint and/or Discharge Diagnosis
• HL7 Standards
• Need standard cc-gtsyndrome coder (SAS)

27
Is It Worth the Effort?

• Costs
• Implementation costs can be modest
• Operational coststime of public health staff,
  investigations
• Benefits
• Possibility of huge benefit if early detection
• Characterization
• Strengthening traditional surveillance
• Dual Use

28
Dual Use

• Opportunity to use new syndromic surveillance
  infrastructure other public health activities as
  well as for bioterror events
• Can enhance all public health efforts
• Sets higher standard for all surveillance (e.g.,
  laboratory)

29
Cipro and Doxycycline Prescriptions
30
Drug Overdose

• Epidemiology of drug overdoses
• Detection of outbreaks

Day of Week
Sat
Fri
Day of Month
31
(No Transcript)
32
Suicidal Ideation/AttemptsNov. 2001 to Sept. 19,
2002
33
Asthma ED Visits and EMS Calls
34
Improvement in Asthma Treatment
35
(No Transcript)
36
So What?

• Strengthened surveillance systems in place
• Potential to better monitor all public health
  situations
• Even if there are no more bioterror attacks,
  preparation can strengthen our public health
  infrastructure and ability to respond
• Syndromic surveillance vs. better surveillance

37
Acknowledgements
NYCDOH Syndromic Surveillance Team
Joel Ackelsberg Sharon Balter Katie
Bornschlegel Bryan Cherry Hyunok Choi Debjani
Das Jessica Hartman Rick Heffernan Adam Karpati
Marci Layton Jennifer Leng Karen Levin Mike
Phillips Sudha Reddy Rich Rosselli Polly
Thomas Don Weiss Field teams MIS staff
38
Spatial ScanStatistic

• Developed by Martin Kulldorff
• Flexible windows in time and space
• Probability through Monte Carlo simulations
• Controls for multiple comparisons
• Modified for infectious disease surveillance