A presentation by Kate Zinszer (Université de Montréal) and Emmanuel Bonnet (Institut de recherche pour le développement). Global Health Workshop: Methods For Implementation Science in Global Health. http://www.equitesante.org/implementation-science-methods-in-global-health/

1. Spatial heterogeneity &
intervention effects
METHODS FOR IMPLEMENTATION SCIENCE
IN GLOBAL HEALTH
April 20th 2017
McGill University
Workshop co-organised by REALISME Chair and McGill of Global Health Programs
Emmanuel Bonnet & Kate Zinszer

2. Context
Study 1
• In West Africa, it is common practice to pay for care at point of
health service
• Capacity to pay major determinant of access to care
• User fee exemptions are advocated for indigent populations
• Indigent selection process is a major challenge to ensure universal
access to healthcare
• Community selection process of village selection committee (VSC) and
community management committee (COGES)

3. Objective
• To determine if a community-based process of selecting
indigent populations in the rural district of Ouargaye, Burkina
Faso, resulted in selection bias
• Compare the VSC (n=259) and COGES (n=26) selection of indigents

4. Methods
• Study population: 50% stratified random sample
from VSC list
• 1,230 indigents selected from VSC (570 retained by COGES)
• Spatial analysis of the remoteness of indigents
• GPS for each individual
• Distances and Kernel density analysis

5. Results
Original Article16
Figure 4. Number of indigents in the two groups based on their distance from the health and soc
centres (CSPSs).
COGES: community management committee; VSC: village selection committee.
Table 1. Distance (in metres) between indigents’ place
of residence and the nearest health and social
promotion centre (CSPS)
VSC
N=660
COGES
N=570
Mean 5,444 4,198
Median 5,260 4,062
COGES: community management committee; VSC:
village selection committee.
VSCs and the COGESs. In Figure 5, the differences
in colour express the comparison of densities
between the VSC and COGES indigents. When the
densities are bluer, the number of indigents selected
by COGESs is higher than the number selected by
seen that some polygons contain
colour (polygons T, L, P, etc. for the
R.,etc.fortheVSCs).Thereistherefor
in the selection of indigents acc
catchment areas of the CSPSs.
The ellipse characterizes the gene
of the indigents. It can be seen in F
two greatest concentrations are in
and the southeast. This orientation
the elongated form, seems to corresp
road network. This confirms that t
of accessibility to the road network
to a CSPS was a major determinant
of indigents for all of the communi
more systematically for the COGES
Discussion
Figure 4. Number of indigents in the two groups based on their distance from the health and social promotion
centres (CSPSs).
COGES: community management committee; VSC: village selection committee.
Table 1. Distance (in metres) between indigents’ place
of residence and the nearest health and social
promotion centre (CSPS)
VSC
N=660
COGES
N=570
Mean 5,444 4,198
Median 5,260 4,062
COGES: community management committee; VSC:
village selection committee.
VSCs and the COGESs. In Figure 5, the differences
in colour express the comparison of densities
between the VSC and COGES indigents. When the
densities are bluer, the number of indigents selected
by COGESs is higher than the number selected by
VSCs. The colour red indicates the inverse, with the
number of COGES indigents selected being lower.
The Thiessen polygons represent the CSPSs’
theoretical catchment areas. They highlight the fact
seen that some polygons contain nearly all one
colour (polygons T, L, P, etc. for the COGESs and U,
R.,etc.fortheVSCs).Thereisthereforeaspecialization
in the selection of indigents according to the
catchment areas of the CSPSs.
The ellipse characterizes the general distribution
of the indigents. It can be seen in Figure 5 that the
two greatest concentrations are in the northwest
and the southeast. This orientation, together with
the elongated form, seems to correspond to the main
road network. This confirms that the combination
of accessibility to the road network and proximity
to a CSPS was a major determinant in the selection
of indigents for all of the communities but applied
more systematically for the COGESs.
Discussion
Because this analysis is based on a sample of
indigents, the results should be interpreted with
caution. Further studies, taking into account other

6. Lessons learned for
implementation science
in global health
Selection bias within community-
based process for indigents
identification
• VSC more effective than COGES
Strengths
• First study of community-based selection
process for indigents
• Comprehensive sampling frame
Limitations
• Further understanding of COGES selection
process and influences

7. Context
Study 2
• Zika virus introduced into the Americas via Easter Island in 2014,
following an outbreak in French Polynesia
• Local transmission has been confirmed in 84 countries and
territories
• Despite rapid spread across the Americas, little is known about the
pattern of spread

8. Objective
• To estimate the velocity (direction and speed) of Zika virus
disease spread in Brazil
• Describe the pattern of Zika introduction in Brazil

9. Methods
• Study population: first lab-confirmed case in each
municipality
• Surface trend analysis of first cases
• Interpolated a continuous estimate of disease spread in magnitude
and direction
• Time of dispersal was calculated from the start of the epidemic

10. Results
Average speed was 42.1 km/day
and trend of spread from the
northeast to south and west of
Brazil

11. Lessons learned for
implementation science
in global health
Large heterogeneity in speed and
directions of diffusion across
municipalities
Strengths
• Provides an estimate of speed and
direction of Zika introduction
• Identifies critical locations and corridors for
containment efforts, and pace of
movement
• Targeted interventions, public health
messaging, and travel advisories
Limitations
• Missed earlier undiagnosed or
misdiagnosed Zika cases

12. To go further
• Ridde V, Sombie I. Street-level workers’ criteria for identifying indigents to be exempted
from user fees in Burkina Faso. Trop Med Int Health. 2012; DOI:10.1111/j.1365–
3156.2012.02991.x.4.
• Ridde V, Yaogo M, Kafando Y, et al. Targeting the worst-off for free health care: a
process evaluation in Burkina Faso. Eval Program Plann. 2011; 34: 333–342.8.
• Zinszer K, Morrison K, Anema A, Majumder MS, Brownstein JS. The velocity of Ebola
spread in parts of west Africa. Lancet Infect Dis. 2015;15:1005–7.
• Pioz M, Guis H, Calavas D, Durand B, Abrial D, Ducrot C. Estimating front-wave velocity
ofinfectious diseases: a simple, efficient method applied to bluetongue. Vet Res.
2011;42:60.http://dx.doi.org/10.1186/1297-9716-42-60

13. Let’s discuss together!
Meet the experts
Thank you for your attention