Syndromic Surveillance: Is it Worth the Effort Chance, Feb' 2004

1
Syndromic Surveillance Is it Worth the
Effort?Chance, Feb. 2004

• Michael A. Stoto, Mattias Schonlau,
• and Louis T. Mariano
• For presentation at NACCHO Meeting
• St. Paul, MN, July 16, 2004

2
Everyone Wants a Bioterrorism "Early Warning
System"

• The sooner you know about a terrorist attack, the
  more effective your response can be
• Smallpox example
• Isolate and quarantine to prevent spread
• Vaccinate unexposed
• Prepare hospitals for cases
• Help identify perpetrators

3
Syndromic Surveillance Offers the Possibility of
Early Detection

• Suggested by Cryptosporidium outbreak in
  Milwaukee
• Focus on symptoms rather than confirmed diagnoses
• Especially flu-like symptoms typical of initial
  states of many bioterrorist agents (anthrax,
  smallpox, etc.)
• Builds on existing data systems
• Health care, medication sales, absenteeism, etc.
• Usually computerized, often massive
• Statistical analyses used to detect sudden changes

4
Why is This Important?

• Ability to detect bioterrorist events earlier
  than otherwise presumably can enable a timely and
  effective public health response
• Major investments of federal funds provided to
  state and local public health departments
• Vendors with systems to sell
• Surveillance for non-BT health outcomes
• Including flu
• General availability of health data

5
How Do We Evaluate Whether Syndromic Surveillance
Works?

• All alarm systems face tradeoffs
• Sensitivityability to detect attacks when they
  occur
• False positive ratesounding an alarm when there
  is no attack
• Timeliness
• Any of the three can be improved, but at the
  expense of the other two

6
Research Questions

• What kinds of attacks are reasonably detectable?
• Can more sophisticated statistical methods do
  better than simple methods? Why?
• How should syndromic surveillance be integrated
  into public health practice?
• How can syndromic surveillance help with other
  health goals?

7
We Used Flu Symptom Data from a Typical Urban
Hospital
20
15
Numberof cases
10
5
0
01jan1998
01jan1999
01jan2000
01jan2001

Day
8
We Simulated an Attack
15
10
Number of cases
5
0
0
10
20
30
40
Day
9
We Tested Four Statistical Detection Methods

• One days data
• Integrated methods average over several recent
  days
• Exponentially Weighted Moving Average
• Cumulative Summation (CUSUM)
• Mean-adjusted CUSUM

10
Fast Attacks Can be Detected by Day 2
11
Slow Attack Cant be Detected Until Day 9
12
Timely Detection Is Difficult

• How long does it take to detect a medium-sized
  attack?
• 50 chance of detection for ILI at
• Day 2 of 3 for fast attack
• Day 9 of 9 for slow attack
• Not at all during the flu season
• Better for Viral Not otherwise specified
• Day 1 or 4
• But only works if Viral NOS is coded
• Little difference between detection methods
• Integrated method necessary for slow attacks

13
Can This Performance Be Improved?

• Choose a syndrome that is less common
• Pool data over multiple hospitals
• Analyze more indicators or hospitals
• Improve baseline model to reduce noise
• Look for geographic or other patterns

14
Infectious vs. non-infectious agent

• 90 people exposed to a non-infectious agent
• 24 people exposed to an infectious agent, with
  second wave

15
Integrating Syndromic and Public Health
Surveillance

• Syndromic surveillance cannot be expected to
  detect attacks with few cases (e.g. anthrax in
  2001)
• Syndromic surveillance intended to alert public
  health officials to possible bioevent
• Must be followed with
• Epidemiologic investigation
• Policy decisions regarding intervention
• Syndromic surveillance must be linked to other
  surveillance systems in advance

16
Appropriate Physician Involvement Is Essential

• Syndromic surveillance efforts often minimize
  involvement of physicians
• May have consequences after an alert when
  physician communication and cooperation is needed
  for
• Active surveillance
• Epidemiologic investigation
• Mass prophylaxis
• Consider active syndromic surveillance by phone
  or computer

17
Conclusions

• Much impressive work has been done
• Information technology Real-time integration of
  many disparate data streams
• Analysis Development of background models,
  detection algorithms, visualization
• Can be very costly
• Cost of gathering data can be reduced through
  improved IT
• Major cost is for investigation and response

18
Conclusions

• Benefits of syndromic surveillance are not yet
  established
• Intrinsic limits to statistical detection ability
• Many current systems are not integrated with
  public health and health care

19
Recommendations

• Go slow until systems can be improved and
  evaluated
• Clarify the tradeoffs (sensitivity, specificity,
  timeliness) for different systems and attack
  scenarios (agent, extent, intensity, time)
• Integrate syndromic surveillance with existing
  surveillance and data systems
• Evaluate active syndromic surveillance
• Identify advantages of better data systems for
  public health and health care purposes