Transcript emchvaccinetalk

Spatial Analysis in Vaccine Trials:
Spatial Effect Modifiers and Herd
Immunity Measurement
Michael Emch
Portland State University
Collaborators
• Dr. John Clemens, Director, International Vaccine Institute (IVI),
South Korea
• Dr. Mohammad Ali, Scientist, IVI
• Dr. Ira Longini, Professor, Biostatistics, Rollins School of Public
Health, Emory University
• Dr. Mohammad Yunus, Senior Scientist & Head, Matlab Health
Research Centre, International Centre for Diarrhoeal Disease
Research, Bangladesh (ICDDR,B)
• Dr. David Sack, Director, ICDDR,B & Professor, Johns Hopkins
School of Public Health
Outline
• Introduction and Objectives
• Phase III Vaccine Trial Background & Traditional Efficacy
Calculations
• Local Efficacy Measurement & Spatial Effect Modifiers
• Herd Immunity Measurement Using Spatial Information
• Study Data and Methods
• Findings and Conclusions
Introduction and Objectives
•Clemens et al. (1996) questioned the utility
of conventional vaccine trial methods.
•Traditional protective efficacy measures don’t
provide information to make decisions about
whether or not to vaccinate diverse
populations.
Clemens, J., Brenner, R., Rao, M., Tafari, N., and Lowe, C. (1996) Evaluating new vaccines
for developing countries: Efficacy of Effectiveness? Journal of the American Medical
Association, 275(5): 390-7.
Justification Example
Ty21a typhoid fever vaccine trials have
produced conflicting results in different
settings because of varying exposure levels
to disease.
Vaccine Trial
Efficacy
Egypt (Wahdan et al., 1982)
96%
Chile (Levine et al., 1990)
77%
Indonesia (Simanjuntak et al., 1991)
53%
Wahdan, M., Sarie, C,. and Derisier, Y. (1982) A controlled field trial of live Salmonella typhii strain Ty21a oral vaccine against typhoid.
Journal of Infectious Diseases, 145: 292-5.
Levine, M., Ferreccio, C., and Black, R. (1990) Large-scale field trial of Ty21a live oral typhoid vaccine in enteric-coated capsule formulation.
Lancet, 336: 891-94.
Simanjuntak, C.H., Paleologo, F.P., and Punjabi, N.H. (1991) Oral immunization against typhoid fever in Indonesia with Ty21a vaccine,
Lancet 338: 1055-9
Introduction and Objectives
This paper describes methods to calculate different
efficacy values for different exposure levels within a
phase III cholera vaccine trial area in rural Bangladesh.
Research Questions
Question 1: How did the effectiveness of a cholera
vaccine vary spatially in the study area?
Question 2: How does protective efficacy vary by
potential effect modifiers (i.e., risk factors for the
disease)?
Question 3: How does cholera incidence in the placebo
group vary in areas with different vaccine coverage
rates; i.e., was herd immunity important, and if so, at
what level of coverage did it become important?
Vaccine Trial Background
The evaluation of new vaccines conventionally
proceeds through several stages of testing:
•Phase I: studies of safety and immunogenicity in lowrisk individuals, usually healthy adults.
•Phase II: studies of safety and immunogenicity in the
population to be targeted for the vaccine in practice.
•Phase III: randomized clinical trials that evaluate the
safety and clinical protection of a vaccine in its target
population.
Phase III Vaccine Trials
•Phase III trials serve as the final “gatekeepers” in the movement of new vaccines
into public health practice.
•The conventional design for phase III trialsthe randomized, double-blind, clinical trialserves as the “gold standard” method for
evaluating vaccines.
Vaccine Efficacy Measurement
Protective efficacy is the proportionate reduction of the
incidence of the target infection by vaccination
Equation 1:
vaccinee  incidence
E  1
x100
nonvaccinee  incidence
Local Efficacy Measurement
•Conventional efficacy calculations are global efficacy
measures (i.e., efficacy of the vaccine for the entire
study area).
•Conventional trials stratify protective efficacy by
individual characteristics (e.g., age and sex).
•If locational data are available a local efficacy
measure can be calculated for different subsets of the
trial area.
Local Efficacy Measurement
11
2
10
00
m
3
et
er
s
5
12
1
1
6 000
10
9
7
8
4
m
et
er
s
Identification
Number
Vaccinee
Population
Placebo
Population
Vaccinee
Cholera
Cases
Placebo
Cholera
Cases
1
12
7
0
1
2
2
6
0
0
3
23
25
0
0
4
24
22
1
2
5
25
32
0
0
6
12
25
1
1
7
25
45
0
0
8
22
23
0
0
9
34
25
0
1
10
25
20
0
0
Total
204
230
2
5
Vaccinee
Incidence
0.0098
Placebo
Incidence
0.022
Efficacy
0.55
Local Efficacy Measurement
The mathematical expression for computing an incidence rate in a raster system is:
Equation 2
n
i 
c
j1
j
 1000
n
p
j1
 kj
j
 kj
where,
j = incidence rate for pixel i
cj = number of cases in pixel j
pj = number of people in pixel j
kj = kernel values of cell j of the filter
n = number of cells in the kernel filter
Incidence can be calculated for both vaccinee and non-vaccinee
groups to calculate efficacy.
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
0
0
0
0
0
0
0
3
0
0
0
0
3
1
5
4
0
2
1
1
0
0
3
2
0
0
1
1
1
0
0
0
0
2
3
0 10 0
5
0
0
0
6
2
0
4
Attributes of case events for the pixels
(white cells are outside study area)
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
0
0
0
0
0
0
0
10
1
0
5
0
31
1
31
12
0
Unitary kernel filter
1
1
1
1
Unitary kernel filter
Local Efficacy Measurement
in a raster GIS
8
6
6
5
1
6
13
0
0
5
8
3
1
0
0
0 10 13 1 51 0 10
0
2
0 15 9
1 16
Attributes of population for the pixels
(white cells are outside study area)
Event and Population Image Used to Calculate Incidence
Spatial Effect Modifiers
•Effect modifiers are variables that modify the effect of
a vaccine.
•There may be differences in the protective efficacy of
vaccines in populations with one characteristic versus
another.
•Some effect modifiers are spatial in nature.
•For instance, protective efficacy of the cholera vaccine
might be worse near rivers because there are more
cholera bacteria in rivers and the vaccine might not
work as well when exposure to the bacteria is greater.
Differential Exposure and Spatial
Effect Modifiers
•Efficacy might differ in different parts of a trial area
since exposure to the disease varies within the study
area.
•We can calculate relationships between socioenvironmental risk factors and efficacy by
neighborhood.
•Example: local sanitation environment is a risk factor
for cholera and may modify the effect of the vaccine.
Differential Exposure (Force of Infection)
and Spatial Effect Modifiers
Efficacy stratified by:
1.
total risk (force of infection)
2.
specific risk factors
Spatial Effect
Modifiers
Are Risk
Factors that
Vary in Space
Satellite remote
sensing allows us
to model environmental
variables
GIS allows us to model
neighborhood-level
variables and
integrate
environmental, health &
demographic
data
John Jensen, 2004
High Incidence Areas
of O139 and El Tor
HIA=High Incidence Area
LIA=Low Incidence Area
Risk Factors
Sig. b
Dependent Variable:
Incidence Rate of O139
(constant)
.000
Population Density
.000
Distance to Nearest
Waterbody
.015
Flood control
.006
Dependent Variable:
Incidence Rate of El Tor
(constant)
.000
Educational status
.000
Sanitation
.001
Distance to Nearest
Waterbody
.021
Flood control
.000
Anthropogenic and Environmental Risk Factors of Cholera
Cholera transmission
Multiple households use latrines
Environmental variable
Flood-controlled area
Variable
P value
Number of open latrines
0.00**
Number of non-septic latrines
0.27
Number of ring septic latrines
0.35
Number of concrete septic latrines
0.65
Number of other households using latrines
0.11
Latrines per person (excluding open)
0.26
Number of tube wells in bari
0.15
Number of households sharing a common tube
well in bari
0.00**
Tube wells per person
0.68
Household area (sq. ft.)
0.02*
Bari population
0.01**
Population density around baris
0.00**
Total household assets
0.35
Annual income
0.27
Mid-arm circumference (children under 5 years
old)
0.10
Distance from main river
0.30
Flood control
0.00**
Environmental variable
Large number of households use tubewells
Cultural/ behavioral and
environmental variable
Household area small
Socioeconomic and
environmental variable
Bari population is large
Cultural/ behavioral, environmental,
and socioeconomic variable
Population density is high within
a half kilometer radius of bari
Cultural/ behavioral, environmental,
and socioeconomic variable
Ecological Spatial Variables Calculated Using a GIS
Spatial Patterns
of Sanitation
Status
Cloud Penetrating Radar Satellite Image:
Can easily differentiate between flooded/non-flooded areas
Herd Immunity
•
Herd immunity is protection of an individual from a
disease because others are immune to the disease.
•
When enough people are vaccinated, the
probability that the disease agent will come into
contact with an individual who is not vaccinated is
lower.
•
This is called herd immunity because nonimmunized people in the population are protected
since most people in the population, i.e., the herd,
are protected.
Herd Immunity Measurement
We measured whether herd immunity is important for
the cholera vaccine by using spatial information.
This was done in three steps:
1. calculated incidence in the placebo group by
neighborhood
2. calculated vaccine coverage rate by neighborhood
3. measured relationship between incidence in the
placebo group and vaccine coverage rate.
a
ghn
Me
'] Matlab
er
Riv
Matlab
Research
Area
Study Area
Study Data
•In 1985, a community based
individually randomized oral
cholera vaccine trial was
conducted in Matlab, Bangladesh.
•This double-blind trial measured
the efficacy of two vaccines, the B
subunit-killed whole cell (BS-WC)
and the killed whole cell only (WC)
vaccine.
•The control agent was
Escherichia coli K12 strain.
Data
•Women over 15 and children aged 2 to 15 were the
target group in the trial.
•Three vaccine doses were given to 62,285 people in
the target group in six-week intervals.
•The vaccine trial used a passive surveillance system
to identify cholera cases from the study area.
•The surveillance took place at one hospital and two
community-based treatment centers.
Data
•During 5 years of follow-up, the protective efficacy for
the BS-WC group was 49% (P < 0.001) and 47% for
the WC group (P < 0.001).
•Protection was lower in children who were vaccinated
at 2 to 5 years than in older persons.
•For children in this age group, protection waned after
4 to 6 months and was not evident during the third
year.
•Persons vaccinated who were older than 5 years were
protected even in the third year of follow-up.
Data
What we do not know from this study:
1)
Whether efficacy varies within the study area.
2)
Whether the spatial variance is related to different
socio-environmental characteristics that are
responsible for spatially heterogeneous exposure.
3)
If there was herd immunity in areas with high
vaccine coverage rates.
Question 1: Findings
Spatial Pattern of Vaccine Efficacy
3 year cumulative
Question 2: Findings
•As local-level population density increases the vaccine is not as effective
•The vaccine is more effective in neighborhoods with higher socio-economic status.
•In neighborhoods with higher vaccine coverage efficacy is higher.
•Vaccine efficacy is lower farther away from passive surveillance facilities.
Regression Model: Dependent Variable Efficacy by 1000 meter neighborhoods
Variable
Prob.
Value
Direction
Population Density by 100 meter neighborhood
(cholera risk factor)
0.0000
Negative
Education: Socio-economic Status in Neighborhood- 250
meters (cholera risk factor)
0.0003
Positive
Vaccine coverage in Neighborhood- 250 meters
0.0419
Positive
Distance to Passive Surveillance Facility (Possible confounder)
0.0195
Negative
Question 3: Findings
Spatial distribution of oral
cholera vaccine coverage
based on 2 doses
Question 3: Findings
•As vaccine coverage rates got larger the cholera incidence rate went down.
•This pattern is true for both the placebo and vaccine group.
Incidence rate and protective efficacy by vaccine coverage (based on two doses)
Vaccine
coverage
(%)
Total
baris
%
baris
Incidence rate/1000
Vaccine group
Rate
% changed
--
Protective
Efficacy
(PE)
Pvalue
--
-1.50
0.6696
Placebo group
Rate
% changed
≤10
320
4.98
9.15
11-20
768
11.96
10.16
1.00*
17.10
1.00*
0.40
0.0804
21-30
1446
22.51
9.70
-4.53
18.63
+8.90
0.48
0.0002
31-40
2412
37.55
5.54
-45.47
12.29
-28.13
0.54
0.0000
41-50
630
9.81
3.80
-62.60
10.74
-37.19
0.65
0.0001
51+
847
13.19
4.98
-50.98
10.75
-37.13
0.54
0.0001
* this is used as the reference group
3.66
Question 3: Findings
•In the original trial, the cholera vaccines were less effective in the
third year (Clemens et al., 1990)
• Using spatial information we found that the vaccines remained
effective (PE60%) in the third year in areas where the coverage is
better (40% and over).
•Non-vaccinees in high coverage areas benefited from reduced
probability of infection.
Conclusions
•Vaccine efficacy varies in space no matter how it is
measured at multiple scales
•Vaccine efficacy varies by different socioenvironmental circumstances (i.e., these variables are
spatial effect modifiers)
•Herd immunity is present is some areas
•Incorporating a spatial component into vaccine
efficacy measurement provides specific information
that is necessary to determine whether or not to
vaccinate different populations.
Conclusions
•All vaccine researchers can use these methods in
future phase III trials.
•Using spatial information policy makers can make
better decisions about the effectiveness of a vaccine in
different circumstances.
•For example, if poor sanitation modifies the effect of
the cholera vaccine then public health officials can
stratify efficacy measures for different sanitation
circumstances.
•They can therefore know how well the vaccine will
perform in another area with a similar sanitation
environment.
Next
•Explore more effect modifiers and total risk (force of
infection). We are just getting started on the spatial
effect modification question.
•Explore herd immunity further
•Create spatially autoregressive models to control for
spatially autocorrelated variables.
•Investigate local relationships by mapping
geographically weighted regression (GWR) results.
Acknowledgements
This project is being funded by:
•National Institutes of Health (NIH): National Institute of
Allergies and Infectious Disease.
•National Science Foundation (NSF): Geography and
Regional Science
Questions?