Aaron KitePowell, MS - PowerPoint PPT Presentation

Title: Aaron KitePowell, MS


1
Syndromic Surveillance of ED Chief Complaints to
Enhance Situational Awareness during Wildfires,
Florida, 2008

• Aaron Kite-Powell, MS
• Lauren Ball, DO, MPH
• Richard Hopkins, MD, MSPH
• Florida Department of Health
• Bureau of Epidemiology
• International Society for Disease Surveillance
• Raleigh, NC
• December 2008

2
Objectives

• Describe the free text chief complaint query
  builder mechanism in ESSENCE and demonstrate
  increased correlation with environmental
  variables relative to predefined syndromes
• Explore relationship of ESSENCE signals with
  particulate matter 2.5 (µg/m3) , NOAA Satellite
  Fire Detection Maps, and 1-hour surface smoke
  images in an area affected by wildfires in south
  Florida.

3
BackgroundWildfires in Florida, May 2008

• Florida
• Jan 1, 2008 to May 31, 2008
• 1,419 wildfires
• 82,090 acres burned
• May 2008
• 440 wildfires
• 62,553 acres burned
• 76 of acres burned occurred in May
• (FL Division of Forestry)
• Federal lands
• Mustang Corners Fire Everglades
• 39,465 acres burned
• (National Park Service)

4
Background

• Governor declares State of Emergency and
  Emergency Operations Center (EOC) was activated.
• Health Medical Emergency Support Function
  (ESF-8) requests information on hospital census
  and bed availability data.
• Where available, enhanced surveillance was
  conducted using ESSENCE to enhance situational
  awareness of ED visits in impacted areas.

5
Methods

• Three data sources were used
• Electronic Surveillance System for the Early
  Notification of Community-based Epidemics
  (ESSENCE)
• ED data from 14 Broward County hospitals (chief
  complaint-based analysis)
• Air quality data (FL Dept. of Environmental
  Protection)
• Particulate matter (2.5 µg/m3)
• Satellite Fire Detections (NOAA Satellite and
  Information Service)
• Smoke Plume Image Maps
• Time period under study
• 5/1/2008 5/31/2008
• Control period
• 5/1/2007 5/31/2007

6
Methods

• Syndrome and subsyndrome categories
• Respiratory illness, bronchitis, asthma, cough.
• Free text query
• Boolean operators AND / OR / ANDNOT
• Keyword search excluded terms suggestive of
  infection
• Example
• cough,andnot,lung,andnot,migraine,andnot,sp
  utum,andnot,rash,andnot,sorethroat,andnot,co
  ngested,andnot,run nose,andnot,fever,andnot,
  febrile,andnot,flu,andnot,cold,andnot,chills
  ,or,trouble breathing,or,asthma..

7
Methods

• Statistical analysis
• ESSENCE-based time series
• Regression/EWMA switch algorithm
• Yellow alert p-value between 0.01 to 0.05
• Red alert p-value lt 0.01
• Pearsons correlation coefficients (r2)
• Compared syndromic queries with PM2.5 µg/m3 data
• Non-parametric Mann-Whitney test
• Compared observed/expected ratios during control
  and event periods.

8
Results Comparison between ER Visits and PM 2.5

• Bronchitis subsyndrome category
• Time-series analysis No statistical signals

r2 -0.02 p-value gt0.05
9
Results Comparison between ER Visits and PM 2.5

• Respiratory illness syndrome category
• Time-series analysis No statistical signals

r2 0.31 p-value gt0.05
10
Results Comparison between ER Visits and PM 2.5

• Cough subsyndrome category
• Time-series analysis No statistical signals

r2 0.43 p-value lt0.05
11
Results Comparison between ER Visits and PM 2.5

• Asthma subsyndrome category
• Time-series analysis Alerts on 5 days

r2 0.53 p-value lt0.05
12
Results Comparison between ER Visits and PM 2.5

• Free text query
• Time-series analysis Alerts on 6 days

r2 0.62 p-value lt0.05
13
Results Pre-Fire and Fire Period ER Visit
Comparison
14
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 11-12
15
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 12-13
16
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 13-14
17
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 14-15
18
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 15-16
19
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 16-17
20
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 17-18
21
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 18-19
22
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 19-20
23
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 20-21
24
Results ER data, PM 2.5 µg/m3, and Smoke Plume
Maps - May 21-22
25
Limitations

• Analysis based on ER chief complaints
• Diverse chief complaint recording methods across
  hospitals.
• Specificity?
• Smoke mapping methodology
• Smoke in the atmosphere vs surface smoke
  measurements.

26
Conclusions

• ESSENCE detected increases in respiratory-related
  chief complaints when PM2.5 µg/m3 was increased
  and smoke plumes showed smoke in the study area.
• Free text querying gave us the ability to refine
  our analysis.

27
Recommendations

• Continue to refine the use of ESSENCE to monitor
  ED usage for wildfire event-related chief
  complaints to enhance situational awareness by
  providing situation status reports to FL ESF-8
  Planning Section.
• Explore impact of smoke, air quality, and
  emergency health services in greater detail
• surface smoke forecasting generated with data
  from satellite imagery and air quality
  measurements
• effects of biomass and particulate matter size on
  health
• disposition of patients from ER
• effects of sheltering in place