Transcript Slide 1

NYC Syndromic
Surveillance
IFH HIT Meaningful Use Workshop
10/1/2010
Marlena Plagianos, MS
NYCDOHMH
[email protected]
What is Biosurveillance?
• “Collection and integration of timely healthrelated information for public health action
achieved through the early detection,
characterization, and situation awareness
of exposures and acute human health
events of public health significance.”
Aaron T Fleischauer, PhD; Pamela S Diaz, MD; Daniel M Sosin MD . Biosurveillance:
A Definition, Scope and Description of Current Capability for a National Strategy.
Advances in Disease Surveillance 2008;5:175
2
Traditional Surveillance
• Case definitions
• Historically low
compliance
• Laboratory
confirmation can be
slow
• Still important (e.g.
H1N1 in NYC)
3
Laboratory Confirmation
•Making firm diagnosis
commonly relies on lab
result
•Limited in-house testing
in outpatient setting
(minutes)
•Commercial laboratory
testing takes time (daysweeks)
4
Traditional Reporting is
Labor Intensive
5
Traditional Reporting is
Labor Intensive
6
Syndromic Surveillance
• Pre-diagnostic indicators of disease
• Readiness scenarios: bioterrorism,
pandemics
• Objectives:
– Timely, sensitive, specific surveillance
– Detect outbreak before ‘astute clinician’
• Typical Process
Collect
data
Process &
code data
Establish
baseline
Identify
outbreak
Sound
alarm
7
New and Exciting Data
Types
Data source
Level of
data
Data type
Setting
Care phase
Medication sales
School absences
Nurse hotline call
Aggregate
Aggregate
Individual
Drug category
Frequency
Call type
Pre-clinical
Pre-clinical
Pre-clinical
Pre-diagnostic
Pre-diagnostic
Pre-diagnostic
Chief complaint /
Reason for Visit
Individual
Text, brief
Clinical
Pre-diagnostic
EMS call
Temperature
Radiology Report
Chest X-ray
Diagnosis code
Progress Note
Individual
Individual
Individual
Individual
Individual
Individual
Run type
Vital sign
Text, narrative
CPT code
ICD9 code
Text, narrative
Clinical
Clinical
Clinical
Clinical
Clinical
Clinical
Pre-diagnostic
Pre-diagnostic
Pre-diagnostic
Pre-diagnostic
Diagnostic
Diagnosis
8
EHR Syndromic Surveillance
• The Primary Care Information Project (PCIP) uses
different EHR data sources to conduct & pilot its
syndromic surveillance activities
• Some syndromes tracked using EHR data are:
– Influenza-like Illness (ILI)
– Fever
– Gastrointestinal Illness (GI)
• Case definitions for these syndromes based upon
text in these structured fields:
– Chief Complaint
– Measured Temperature
– Diagnosis (ICD-9 CM Code)
9
System Screenshot
10
Aggregate Level Syndromic
Data
• Only “Count” Data is collected
11
Data processing and syndrome
coding
Respiratory
conditions
Misspelling
Shortness of
breath
Difficulty
breathing
Upper
respiratory
infection
%Macro Resp; *Respiratory;
IF
CC=:'COUGH' OR
CC=:'COUGHING' OR
CC=:'SOB' OR
CC=:'DIFFICULTY BREATHING' OR
CC='BREATHING PROBLEMS' OR
CC=:'SHORTNESS OF BREA' OR
CC=:'DIFF BREA'
CC='URI' OR
THEN RESP=1; ELSE DO;
RESP=
INDEX(CC,"COUG") + INDEX(CC,"COUH") +
INDEX(CC,"S.O.B") + INDEX(CC,"SOB") + INDEX(CC,"S O B") +
INDEX(CC,"S O B") + INDEX(CC,"S.OB");
INDEX(CC,"BREAT") + INDEX(CC,"BEATH") + INDEX(CC,"DIB") +
INDEX(CC,"D I B") + INDEX(CC,"D.I.B") + INDEX(CC,"BRATHING") +
INDEX(CC,"DIFF BR") + INDEX(CC,"DIFF, BR") +
INDEX(CC,"URI ") + INDEX(CC,"URI/") + INDEX(CC,"URI;") +
INDEX(CC,"U R I") + INDEX(CC,"URI,") + INDEX(CC,"U.R.I") +
12
Analysis:
Calculate Baseline
Expected disease level
• Approaches: Moving average, regression, time
series methods.
• Length of baseline: Years, months, days
Adjustments
• Long: Seasonal, secular, environmental (e.g. heat,
pollen)
• Short: Day of week, weekend/weekday, holidays,
reporting failures
13
Analysis:
Test Observed vs. Expected
Significance tests
Predetermined number of
standard deviations
Crossing statistical
thresholds  Signal
14
Analysis:
Test Observed vs. Expected
15
Electronic Health Record
Syndromic Surveillance
During 2009 Pandemic
H1N1 in NYC
Friday
18
Saturday
19
Sunday
20
Monday
21
Tuesday
22
Wednesday
23
Thursday
24
Friday
25
Saturday
26
Sunday
27
Monday – Memorial Day
28
Tuesday
29
H1N1 in New York City:
Where did patients seek
treatment?
Emergency Departments
or
Primary Care Clinics?
Objective
• To determine whether the timing of the
increase in patient visits was different at
emergency departments from primary care
clinics during the spring 2009 H1N1
influenza outbreak across the 5 boroughs
of NYC
31
Study Sites
l 58 Primary Care
Providers (PCP):
– 9 Institute for Family
Health (IFH) clinics
– 49 practices enrolled
in the NYCDOHMH
PCIP (30 visits/day)
v 50 Emergency
Departments
– 247 visits/day
32
Methods
• Influenza-like Illness (ILI) as a broad
estimate of H1N1
PCP
ED
• Fever + respiratory related reason
for visit or diagnosis
• Chief complaint of fever + a sore
throat or cough, or a chief complaint
of flu
33
Methods
Compared number of days to a significant
increase at EDs to PCP clinics using a logrank test
• City-wide
• By borough to see if there was a geographic difference
Two Waves:
• 4/24-5/8
• 5/14-6/4
34
ED, IFH and PCIP ILI Visits
35
ED, IFH and PCIP ILI Visits
ED, IFH and PCIP ILI Visits
Facilities with a significant
increase in ILI volume
38
Results, April 24-May 8
Median Days to Facilities with Increase in
Increase in ILI
ILI
ED
1-sided
log rank
Borough
All
4
PCP
ED
PCP
p-value
12
43/50 (86%) 36/58 (62%) <0.0001
Bronx
5
12
8/9 (88%)
Brooklyn
3
14
12/15 (80%) 6/9 (67%)
Manhattan
4
13
13/15 (87%) 11/19 (58%) 0.008
Queens
3
7
8/8 (100%)
6/7 (86%)
Staten Island
14
10
2/3 (67%)
3/3 (100%) 0.902
10/17 (59%) 0.045
0.025
0.007
39
Results, May 14-June 4
Median Days to Facilities with Increase in
Increase in ILI
ILI
ED
PCP
1-sided
log rank
Borough
All
4
8
ED
PCP
p-value
47/50 (94%) 50/58 (86%) <0.0001
Bronx
1
6
9/9 (100%)
Brooklyn
4
12
13/15 (87%) 7/9 (78%)
Manhattan
4
7
14/15 (93%) 17/19 (89%) 0.016
Queens
4
8
8/8 (100%)
5/7 ( 71%)
Staten Island
5
8
3/3 (100%)
3/3 (100%) 0.012
16/17 (82%) 0.004
0.039
0.091
40
Findings
• Emergency Departments experienced an
increase in patients with ILI before Primary
Care Providers
• PCPs were vastly under-utilized during the
outbreak
• NYCDOHMH changed messaging to
encourage visiting PCPs instead of EDs
for mild illness
41
Future of Syndromic
Surveillance
Meaningful Use
• Capability to submit syndromic data to health
departments
Regional Health Information
Organizations (RHIOS), Hubs
Data Validation and Quality Assurance
42
Online Resources
CDC Flu Surveillance
http://www.cdc.gov/flu/weekly/fluactivity.htm
Distribute
http://www.isdsdistribute.org/
43