Upload
clara-merritt
View
215
Download
0
Tags:
Embed Size (px)
Citation preview
Community Validation of
Influenza-like Illness as a Predictor of
InfluenzaJonathan L. Temte, MD/PhD & Alexis Eastman, MS-2
University of Wisconsin School of Medicine and Public Health
Peter A. Shult, PhD, Carol J. Kirk & Mary Wedig
Wisconsin State Laboratory of Hygiene
Madison, Wisconsin
Influenza-like Illness
Definition Fever of 100oF (37.8oC) or higher Cough and/or Sore Throat Not due to any other illness
Utility Simple and elegant Clinically relevant Easily ascertained
ILI uses
Clinical identification of influenza infection High PPV from research protocols
Adults Children
Age 65+
Age 25-64
Age 5-24
Age 0-4
ILI in WisconsinOct. 2007 thtough Sept. 2008
Community surveillance of influenza
Factors Affecting Symptoms
AgeAge
ImmuneImmuneStatusStatus
UnderlyingUnderlyingDiseaseDisease
Viral StrainViral Strain
Viral SubtypeViral Subtype
Host Factors Viral Factors
Purpose of Study
Review the contents of a large database Surveillance data emerging from a
partnership between a public health laboratory and primary care clinicians
Symptoms and virus identification
Validate ILI for influenza infection Community—not research—perspective
The Surveillance Database
Partnership of WSLH and UW-DFM since 1994 Major modification of symptom check off in 1997
Opportunistic sampling with “fee-exempt” virus culture physicians obtain specimens, record demographic and symptom
data, sample is transported to WSLH by courier. Standard culture methods with isolation rate = 45% Limited, de-identified data used
1997-2007 IRB approved
3,796 episodes of acute respiratory illness care available
Preferential Collection from
Children and Young Adults
0
50
100
150
200
250
300
350
0 5 10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Age of Patient
No
. o
f S
pecim
en
s Range: 0 – 103 years
55.6% female
PredictorsWorking definition of ILI
F+CorST F = Fever on symptom checklist
No requirement for level or documentation
CorST = Cough and/or Sore Throat
sF+CorST (includes seasonality) December through March Period with > 90% of influenza cases
Outcomesinfluenza isolation
Paradigm 1: “clinical primary care” Influenza (+) vs. all other specimens
Influenza = 1230 Non-influenza + no virus isolated = 2566
Paradigm 2: “ideal virus capture” Influenza (+) vs. non-influenza virus (+)
Influenza = 1230 Non-influenza = 523
Distribution of outcomes
Reference
population
Season
included
Criteria
used
Influenza (+)
Influenza (-)
All ARI specimens
Yes
sF+CorST (+)
1020 1034
sF+CorST (-)
210 1532
No
sF+CorST (+)
1082 1529
sF+CorST (-)
148 1037
Reference
population
Season
included
Criteria
used
Influenza (+)
Influenza (-)
Virus (+) specimens
Yes
sF+CorST (+)
1020 188
sF+CorST (-)
210 335
No
sF+CorST (+)
1082 302
sF+CorST (-)
148 221
Distribution of outcomes
Performance Characteristics
Criteria Referencepopulation
OR
flu
Sens Spec PPV NPV
F+CorST All ARI 4.96 0.88 0.40 0.41 0.88
F+CorST Virus (+) 5.25 0.88 0.42 0.78 0.60
sF+CorST All ARI 7.20 0.83 0.60 0.50 0.88
sF+CorST Virus (+) 8.66 0.83 0.64 0.84 0.62
Additional Fiddlingassessing effects of age
Concentrate on seasonal data Clinician informed by surveillance
Concentrate on virus (+) specimens Symptomatic patient Early in illness Collection technique good
Concentrate on age categories 0-4 5-24 25-64 65+
Effects of age(reference age = 25-64 years)Binary logistic regression via Minitab –
Release 13.1Factor Odds Ratio 95% CI
sF+CorST 7.55* 5.81 – 9.80
0-4 years 0.10* 0.07 – 0.14
5-24 years 1.21 0.90 – 1.65
25-64 years reference
65+ years 1.67 0.86 – 3.25
* P<0.001
What about little kids?the percent of ILI cases due to:
Virus 0-4 years 5+ years
Influenza 34.8 84.7
Adenovirus 6.6 3.2
Parainfluenza 14.4 3.3
Rhinovirus 1.7 3.7
RSV 37.0 1.4
Herpes simplex 1.1 2.0
Enterovirus 2.9 0.8
Conclusions ILI (F+CorST) performs well
Public health tool for surveillance Early detection of influenza High sensitivity ( 0.88) Limited by low specificity (0.40)
but fined tuned by virological methods
ILI (sF+CorST) performs well Clinician tool for diagnosis of influenza Informed by public health surveillance High PPV (0.84); moderate NPV Excluding young children raises PPV to 0.90
Conclusions
Influenza is the primary cause of ILI in patients age 5+ years
Many viruses can cause ILI in children 0-4 years of age. ILI should not be used for diagnosis alone in this group.
ILI for predicting influenza infection has been validated in a primary care, community-based population
Final WordsIf influenza is in the community and
your patient is over 4 years oldIs it influenza?
F+CorST
“Of Course”
Acknowledgements
Wisconsin Primary Care Clinicians UW-DFM residency clinics Numerous private physicians
UW-DFM Summer Student Research and Clinical Assistantship Program Ms. Alexis Eastman
Wisconsin State Laboratory of Hygiene
Additional Material
Basic Characteristics of Surveillance System
Mean age of patient = 26.6 years Range [ 0 to 103 years]
Sex Female = 55.6% Male = 44.4%
Time between illness onset and collection Mean = 3.86 days Median = 2 days
Rate of virus isolation = 44.6%
Specimens Collected during
“Respiratory Virus” Season
0
50
100
150
200
250
300
350
0 4 8 12 16 20 24 28 32 36 40 44 48 52
Weeks after July 1st
Nu
mb
er
of
Sp
ecim
en
s
Collection Day(Monday through Thursday
Preferred)
0
5
10
15
20
25
SUN MON TUE WED THU FRI SAT
% o
f S
pec
imen
s
Most SpecimensCollected at Optimal Time
0102030405060708090
100
0 3 6 9 12 15 18 21 24 27 30
Days after Onset of Illness
Cu
mu
lati
ve P
erce
nta
ge
Percent of Specimens with Positive Virus
Isolation
0
0.1
0.2
0.3
0.4
0.5
0.6
0 1 2 3 4 5 6 7
Days after Onset of Illness
% p
osit
ive