From Data to Signals to Screenshots: Recent Developments in NYCDOHMH ED Syndromic Surveillance.

1
From Data to Signals to Screenshots Recent
Developments in NYCDOHMHED Syndromic
Surveillance.

• Marc Paladini
• New York City Department of Health and Mental
  Hygiene

2
Questions

• What are we looking for?
• How are we looking for it?
• What do we need to find it?
• What do we do after we find it?
• How does this help us do our job more effectively?

3
Outline

• Introduction to NYC emergency department
  syndromic surveillance.
• Analysis
• Signal Investigation
• Data Visualization
• Future Directions

4
Analysis
5
Hospital map
6
ED Syndromic Surveillance in NYC

• Chief complaint hierarchical syndrome coding
• Syndromes
• 1. Sepsis
• 2. Respiratory
• 3. Rash
• 4. Fever
• 5. Cold
• 6. Diarrhea
• 7. Asthma
• 8. Vomit
• Other

7
Hospital Statistics - 2005

• 3,136,772 total visits. (7000-10,000 / day)

8
Data Analysis

• Ratio of syndrome/other
• Citywide and spatial (zip code, hospital)
• Age 13 respiratory and fever
• All ages diarrhea and vomit
• Age group
• 0-4
• 5-17
• 18-59
• 60
• 7 days/week

9
SIGNALS
10
Syndromic Analyst

• Run surveillance usually _at_ 2 hours
• Review output for signals
• Review signal details
• Linelist
• Baseline vs. signal frequency tables
• Consult with DOH Physician on call (Cluster Doc)

11
Cluster Doc

• Review data with analyst
• Ask for further details
• Decide on follow up
• guidelines/protocol
• fingerprint of signal

12
Signals 2005
13
Investigation of Signals

• Review line list
• Check complimentary systems
• Acquire interim data (12 hour log)
• Call to EDs
• Chart reviews
• Patient follow up (phone calls)
• Augment lab testing/collect specimens
• On site epi teams
• Special studies (case-control)

14
Routine Steps

• First day (Resp/Fev) vs. 2nd day (Vom/Dia)
• Perform descriptive statistics, midday log
• Examine CUSUM, other systems
• Call hospitals with CUSUM alarms
• What did they see yesterday?
• What are the seeing today?
• Clinical clusters, unexpected severe illness?
• Augment lab testing
• Alert ED staff

15
Concerning Features of a Syndromic Signal

• Sustained
• Multiple hospitals
• Large number of excess cases
• Uniformity of chief complaints
• Young adults or age/sex clustering
• Overlapping syndrome signals
• Coincident clinician call
• Coincident with high profile public event
• Concordance with other surveillance systems

16
Results of Investigations

• Some clear seasonal patterns evident
• Sharp spikes associated with known events
• Rich ecological associations
• Difficult to investigate
• Used to reinforce public health messages
  (influenza, viral GI, heat wave, blackout, asthma)

17
Proposed Prospective Investigation Protocol

• All significant resp and fever signals
• Chart reviews
• Patient interviews
• Classify if cases in signal are related
• by lab diagnosis, i.e. strep pharyngitis
• by clinical diagnosis, i.e. pharyngitis
• by risk factor, i.e. subway travel

18
DATA VISUALIZATION
19
Intranet
20
Respiratory intranet
21
Signal details 1
22
Signal details 2
23
Signal details 3
24
Signal details 4
25
FUTURE DIRECTIONS
26
ILI age 1
27
ILI age 2 - serfling
28
ILI age 3 - TERS
29
ILI age 4 - daily
30
Resp CC
31
Fev/resp/resp non-feb
32
Diar/vom age 0-4
33
Fev resp/fev cold
34
Questions and Projects

• What are we looking for?
• Are more fields better?
• Discharge diagnosis / ICD-9 code
• Disposition
• Recorded temperature
• Quantify effect of school closings
• Day of Week
• SaTScan mapping
• Age as space