Baseline spatial distribution of malaria prior to an elimination programme in Vanuatu.
ABSTRACT The Ministry of Health in the Republic of Vanuatu has implemented a malaria elimination programme in Tafea Province, the most southern and eastern limit of malaria transmission in the South West Pacific. Tafea Province is comprised of five islands with malaria elimination achieved on one of these islands (Aneityum) in 1998. The current study aimed to establish the baseline distribution of malaria on the most malarious of the province's islands, Tanna Island, to guide the implementation of elimination activities.
A parasitological survey was conducted in Tafea Province in 2008. On Tanna Island there were 4,716 participants from 220 villages, georeferenced using a global position system. Spatial autocorrelation in observed prevalence values was assessed using a semivariogram. Backwards stepwise regression analysis was conducted to determine the inclusion of environmental and climatic variables into a prediction model. The Bayesian geostatistical logistic regression model was used to predict malaria risk, and associated uncertainty across the island.
Overall, prevalence on Tanna was 1.0% for Plasmodium falciparum (accounting for 32% of infections) and 2.2% for Plasmodium vivax (accounting for 68% of infections). Regression analysis showed significant association with elevation and distance to coastline for P. vivax and P. falciparum, but no significant association with NDVI or TIR. Colinearity was observed between elevation and distance to coastline with the later variable included in the final Bayesian geostatistical model for P. vivax and the former included in the final model for P. falciparum. Model validation statistics revealed that the final Bayesian geostatistical model had good predictive ability.
Malaria in Tanna Island, Vanuatu, has a focal and predominantly coastal distribution. As Vanuatu refines its elimination strategy, malaria risk maps represent an invaluable resource in the strategic planning of all levels of malaria interventions for the island.

Article: Further shrinking the malaria map: how can geospatial science help to achieve malaria elimination?
[Show abstract] [Hide abstract]
ABSTRACT: Malaria is one of the biggest contributors to deaths caused by infectious disease. More than 30 countries have planned or started programmes to target malaria elimination, often with explicit support from international donors. The spatial distribution of malaria, at all levels of endemicity, is heterogeneous. Moreover, populations living in lowendemic settings where elimination efforts might be targeted are often spatially heterogeneous. Geospatial methods, therefore, can help design, target, monitor, and assess malaria elimination programmes. Rapid advances in technology and analytical methods have allowed the spatial prediction of malaria risk and the development of spatial decision support systems, which can enhance elimination programmes by enabling accurate and timely resource allocation. However, no framework exists for assessment of geospatial instruments. Research is needed to identify measurable indicators of elimination progress and to quantify the effect of geospatial methods in achievement of elimination outcomes.The Lancet Infectious Diseases 08/2013; 13(8):70918. · 19.45 Impact Factor  SourceAvailable from: Thomas A Smith[Show abstract] [Hide abstract]
ABSTRACT: Background: Malaria transmission is measured using entomological inoculation rate (EIR), number of infective mosquito bites/person/unit time. Understanding heterogeneity of malaria transmission has been difficult due to a lack of appropriate data. A comprehensive entomological database compiled by the Malaria Transmission Intensity and Mortality Burden across Africa (MTIMBA) project (20012004) at several sites is the most suitable dataset for studying malaria transmissionmortality relations. The data are sparse and large, with smallscale spatialtemporal variation. Objective: This work demonstrates a rigorous approach for analysing large and highly variable entomological data for the study of malaria transmission heterogeneity, measured by EIR, within the Rufiji Demographic Surveillance System (DSS), MTIMBA project site in Tanzania. Design: Bayesian geostatistical binomial and negative binomial models with zero inflation were fitted for sporozoite rates (SRs) and mosquito density, respectively. The spatial process was approximated from a subset of locations. The models were adjusted for environmental effects, seasonality and temporal correlations and assessed based on their predictive ability. EIR was calculated using modelbased predictions of SR and density. Results: Malaria transmission was mostly influenced by rain and temperature, which significantly reduces the probability of observing zero mosquitoes. High transmission was observed at the onset of heavy rains. Transmission intensity reduced significantly during Year 2 and 3, contrary to the Year 1, pronouncing high seasonality and spatial variability. The southern part of the DSS showed high transmission throughout the years. A spatial shift of transmission intensity was observed where an increase in households with very low transmission intensity and significant reduction of locations with high transmission were observed over time. Over 68 and 85% of the locations selected for validation for SR and density, respectively, were correctly predicted within 95% credible interval indicating good performance of the models. Conclusion: Methodology introduced here has the potential for efficient assessment of the contribution of malaria transmission in mortality and monitoring performance of control and intervention strategies.Spatial and Spatiotemporal Epidemiology 01/2012; submitted:.  SourceAvailable from: PubMed Central[Show abstract] [Hide abstract]
ABSTRACT: The goal of malaria elimination faces numerous challenges. New tools are required to support the scale up of interventions and improve national malaria programme capacity to conduct detailed surveillance. This study investigates the cost factors influencing the development and implementation of a spatial decision support system (SDSS) for malaria elimination in the two elimination provinces of Isabel and Temotu, Solomon Islands.Malaria Journal 08/2014; 13(1):325. · 3.49 Impact Factor
Page 1
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Open Access
RESEARCH
Baseline spatial distribution of malaria prior to an
elimination programme in Vanuatu
© 2010 Reid et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons At
tribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any
medium, provided the original work is properly cited.
Research
Heidi Reid*1, Andrew Vallely1, George Taleo2, Andrew J Tatem4, Gerard Kelly1, Ian Riley1, Ivor Harris3, Iata Henri2,
Sam Iamaher2 and Archie CA Clements1,5
Abstract
Background: The Ministry of Health in the Republic of Vanuatu has implemented a malaria elimination programme in
Tafea Province, the most southern and eastern limit of malaria transmission in the South West Pacific. Tafea Province is
comprised of five islands with malaria elimination achieved on one of these islands (Aneityum) in 1998. The current
study aimed to establish the baseline distribution of malaria on the most malarious of the province's islands, Tanna
Island, to guide the implementation of elimination activities.
Methods: A parasitological survey was conducted in Tafea Province in 2008. On Tanna Island there were 4,716
participants from 220 villages, georeferenced using a global position system. Spatial autocorrelation in observed
prevalence values was assessed using a semivariogram. Backwards stepwise regression analysis was conducted to
determine the inclusion of environmental and climatic variables into a prediction model. The Bayesian geostatistical
logistic regression model was used to predict malaria risk, and associated uncertainty across the island.
Results: Overall, prevalence on Tanna was 1.0% for Plasmodium falciparum (accounting for 32% of infections) and 2.2%
for Plasmodium vivax (accounting for 68% of infections). Regression analysis showed significant association with
elevation and distance to coastline for P. vivax and P. falciparum, but no significant association with NDVI or TIR.
Colinearity was observed between elevation and distance to coastline with the later variable included in the final
Bayesian geostatistical model for P. vivax and the former included in the final model for P. falciparum. Model validation
statistics revealed that the final Bayesian geostatistical model had good predictive ability.
Conclusion: Malaria in Tanna Island, Vanuatu, has a focal and predominantly coastal distribution. As Vanuatu refines its
elimination strategy, malaria risk maps represent an invaluable resource in the strategic planning of all levels of malaria
interventions for the island.
Background
In recent years, the momentum behind malaria elimina
tion has gathered speed with thirtynine countries across
the world now making progress toward malaria elimina
tion. One of the key strategies is to shrink the global
malaria map from the endemic margins inward [1]. While
some nations are committed to nationwide elimination,
others are pursuing spatially progressive elimination
within their borders. With support from international
donors, the Ministry of Health in the Republic of Vanuatu
has started to implement a malaria elimination pro
gramme in Tafea Province which is comprised of five
islands, Fatuna, Aneityum, Erromango, Aniwa and Tanna
(Figure 1). Interrupted malaria transmission has already
been achieved on the island of Aneityum through the use
of approaches such as mass drug administration and
insecticide treated bed nets, and with the enthusiastic
support of the local population [2]. Tafea Province repre
sents the most southern and eastern limit of malaria in
the South West Pacific and thus a strategic starting point
for elimination activities.
While the optimal strategy for elimination is being
debated, possible distinctions between control and elimi
nation efforts are the geographical scale and intensity of
key interventions [3]. During the control phase, interven
tions tend to be widely applied to the target areas, often
* Correspondence: heidilouisereid@gmail.com
1 Pacific Malaria Initiative Support Centre (PacMISC), Australian Centre for
International and Tropical Health (ACITH), School of Population Health,
University of Queensland, Queensland, Australia
Full list of author information is available at the end of the article
Page 2
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 2 of 9
with similar strategies between zones of varying ende
micity. As the prevalence declines remaining transmis
sion is increasingly restricted to specific geographical foci
and more precision in the application of interventions is
needed, with more intense targeting of resources to iden
tify and eliminate the last remaining sites of transmission
and/or resistance [4]. An effective elimination campaign
must be capable of identifying these foci. Mapping offers
great potential and the greatest need for malaria maps is
at the periphery of stable, endemic areas where there is
often less empirical information regarding risks and
intensity of infection [5].
The advancement of geographical information systems
(GIS) and spatial statistics has greatly improved our
understanding of malaria dynamics, including its depen
dence on ecological factors [512]. More recently, Bayes
ian geostatistics has been embraced for disease mapping
with the advantage that both environmental covariates
and spatial autocorrelation are able to be estimated
simultaneously and full posterior distributions produced,
which can be used to quantify uncertainties in parame
ters of interest (e.g. predicted prevalence of infec
tion)[13]. Spatial prediction models have been used to
produce malaria risk maps at national [11,1419] sub
continental [2022] and global scales [23,24].
The aim of this present study was to produce accurate,
validated predicted prevalence maps for P. vivax and P.
falciparum on Tanna Island, Vanuatu. Additionally, it is
envisaged that the maps be used to spatially define an
implementation strategy for malaria elimination in Tanna
Island. The applicability of the methods and the implica
tions of the results are discussed in the context of malaria
elimination strategies, which are beginning to take shape.
Methods
Survey data
Data pertaining specifically to Tanna Island (N = 4763)
were extracted from the results of a parasitological survey
conducted in Tafea Province in 2008 by the National Vec
tor Borne Disease Control Program (VBDCP), Vanuatu,
in collaboration with the Pacific Malaria Initiative Sup
port Centre (PacMISC), a Brisbanebased consortium
consisting of the School of Population Health at the Uni
versity of Queensland; the Australian Army Malaria Insti
tute (AMI); and the Queensland Institute of Medical
Research (QIMR)[25]. Within Tanna, the schoolbased
survey covered all 80 schools on the island. Blood sam
ples were collected from children between two and 12
years of age by finger prick using a lancet and the samples
were examined microscopically for malaria parasites.
Dried blood spot specimens were also collected and
transported to the AMI in Brisbane for analysis by poly
merase chain reaction (PCR), considered the gold stan
dard for malaria diagnosis, the results of which were used
to develop the spatial models presented in this report.
Each child was interviewed to collect information such
as school [N = 80], home village [N = 233] and village in
which they usually sleep ('sleep' village, N = 233). Village
coordinates were not taken at the time of the survey but
school, home and sleep villages were later matched to a
government list provided by the Vanuatu Ministry of
Lands from a 1999 census. To most accurately match
infection to place of transmission, sleep village was used
as the geographical reference because this was considered
a more accurate representation of exposure sites than
home or school village. For sleep villages not able to be
georeferenced, the home and then school village were
used as the geographical reference. If villages were not
included in the census list (4% of children), the local
malaria control officer on Tanna Island was consulted to
determine the closest listed village to the sleep village and
this was chosen as the geographical reference. A total of
4,716 children (99%) could be geolocated in this manner
to a total of 220 'sleep', 'home' or 'school' villages.
Four villages in the Green Hill area in the north of the
island (circled in Figure 2) showed an unexpectedly high
Figure 1 Map of Vanuatu showing the location of Tafea Province
within the country and the location of Vanuatu with respect to
neighbouring countries in the Western Pacific region (inset).
Page 3
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 3 of 9
proportion of malariapositive individuals given their
inland location. As this finding was hypothesized to be an
artefact of the postsurvey georeferencing method, these
four villages were removed during a secondary analysis to
assess their impact on the significance of environmental
variables.
Analysis of spatial structure
Statistical analyses of spatial structure in the prevalence
data were done in R version 2.9.0 (The R foundation for
statistical computing) using the GeoR package. For these
analyses the sample variogram was defined as:
Where p(xi) represents a value of prevalence observed
at location xi and p(xi h) represents a second observation
at a distance h from the first. By discretising lags into a
series of bins of width b, such that each value of h actually
represents a distance interval h ± 1/2b, semivariances
are computed as the mean semivariance amongst the set
of n(h) pairs of observations separated be distances
within that interval [24].
Assembling and testing ecological and climate variables
Images from Landsat Enhanced Thematic Mapper (ETM)
were downloaded from the United States Geological Sur
vey (USGS) Earth Resources Observation and Science
Center [26]. These images were processed to develop
estimates of normalized difference vegetation index
(NDVI) [27] representing the amount of vegetation per
30 metre spatial resolution pixel and Thermal InfraRed
(TIR) as a surrogate for surface temperature at 60 m spa
tial resolution.
Gridded elevation data at 90 m spatial resolution were
obtained from the Shuttle Radar Topography Mission
digital elevation dataset, processed and made available at
the Consultative Group on International Agricultural
Research Consortium for Spatial information [28]. Finally
distance to coast was calculated using the spatial analyst
extension of the GIS software ArcView version 9.3 (ESRI,
Redlands, CA). The same software was used to extract
NDVI, TIR and elevation for the 220 village locations.
Colinearity between each pair of environmental vari
ables (NDVI, TIR, elevation and distance to coastline)
was assessed in Stata/SE Version 10 (Stata Corporation,
College Station, TX, USA) statistical software package
with 0.9 defined as the cutoff. If colinearity was
observed separate models incorporating the different
variables would be constructed. Backwards stepwise
regression analysis was conducted on the remaining vari
ables to determine their inclusion into the final spatial
g( )
( )
n h
2
( ( )
p x
())^
h
( )
n h
=∑
hp x
ii
i
=−+
1
2
1
Figure 2 Geographic distribution of P. vivax (top) and P. falcipar
um (bottom) prevalence on Tanna Island, Vanuatu based on PCR
results from the 2008 parasitological survey and their associated
variograms displaying spatial autocorrelation over a similar
range of 0.05 decimal degrees (corresponding to approximately
5 km).
distance
semivariance
5
10
15
20
25
0.05 0.10
distance
semivariance
1
2
3
4
0.050.10
Page 4
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 4 of 9
prediction model. Those variables with a pvalue < 0.1
were retained.
Bayesian geostatistical model
A spatial prediction model (Figure 3) was constructed
based on the principle of modelbased geostatistics [29]
in the Bayesian statistical software WinBUGS version
14.1 (MRC Biostatistics Unit, Cambridge, UK). The
model comprises two components: a deterministic com
ponent consisting of villagelevel fixed effects; and a sto
chastic component consisting of an isotropic, stationary
autocorrelation function describing villagelevel spatial
variation (i.e. a geostatistical random effect). To predict
the prevalence at unsampled locations, a grid was gener
ated of 1958 prediction locations with a spacing of 0.01
decimal degrees (approximately 1 km), covering the
island. Using the inbuilt spatial.unipred function in
WinBUGS, the geostatistical random effect was interpo
lated to all prediction locations and predicted prevalence
was calculated by adding the random effect to the sum of
the products of the coefficients for the covariates and the
values of the covariates at each prediction location.
Model validation
Validation of predicted prevalence was undertaken by
partitioning the data into four random subsets, running
the model using three of the four subsets and validating
the model with the remaining subset. Four separate mod
els were run, each with a different subset excluded for val
idation. The accuracy of the prediction was determined
in terms of area under curve (AUC) of the receiveroper
ating characteristic (ROC), with observed prevalence,
dichotomised at 0 and ≥0, taken as the comparator. This
gave an indication of the ability of the model to discrimi
nate between areas where transmission did and did not
occur. As a general rule, an AUC between 0.5 and 0.7
indicates a poor discriminative capacity; 0.70.9 indicate
a reasonable capacity; and >0.9 indicate a very good
capacity.
Results
The overall prevalence of P. falciparum and P. vivax infec
tion was 1.0% (95% CI: 0.791.21%) and 2.2% (95% CI:
1.063.34%) respectively [25] with the spatial distribution
broadly similar between the species (Figure 2).
Semivariograms (Figure 2) revealed spatial autocorrela
tion was a feature of raw prevalence of P. vivax and P. fal
ciparum. Colinearity of 0.91 (measured using Pearson's
correlation coefficient) was observed between elevation
and proximity to coastline. NDVI and TIR were not sta
tistically significant predictors and thus were not
included in the final models. Exclusion of the four afore
mentioned villages did not affect the significance of envi
ronmental variables and thus a separate model was not
deemed necessary.
Tables 1 and 2 present the results of the Bayesian geo
statistical models for P. vivax and P. falciparum. For P.
vivax, the model with distance to coastline as the fixed
effect gave the best fit with better predictive ability and
lower mean error and mean square errors values than the
model with elevation as the fixed effect. For P. falci
parum, the model with elevation as fixed effect gave the
best fit. On the basis of the models and validation it was
possible to predict the distributions of P. vivax and P. fal
ciparum risk across the island (Figure 4 and Figure 5
respectively). The strong relationship between proximity
to coastline and P. vivax gave a smooth risk map with
endemicity classes following the contour of the coastline.
The risk maps for P. vivax and P. falciparum include the
upper and lower Bayesian credible limits and show the
extent of uncertainty in predicted malaria risk in 2008.
Discussion
The Bayesian prediction models clearly show that trans
mission is not homogeneous, with malaria risk displaying
a predominately coastal distribution, concentrated within
welldelimited foci or 'hotspots'. While the presence of
malaria foci on the island is supported by anecdotal evi
dence the presence of an inland 'hotspot' in the Green
Hill area is contentious. An entomological survey carried
out on the island concurrently with the parasitological
survey did not reveal any vector breeding sites in the
Green Hill area and concluded that transmission is
Figure 3 Spatial prediction model based on the principles of
modelbased geostatistics.
Assume that the number found positive for malaria parasitaemia at location i is Yiout of Ni
examined then Yiis a binomial random variable:
??
iii
pN BinY
,
?
The bivariate ordinary logistic regression model is given by:
i
i
i
env
1
?
p
p
?
0
1
log
? ??
??
?
?
??
?
?
Where ?o is the intercept, envi is the environmental covariate, ?1 is the corresponding
regression parameter. The spatial correlation is modelled by inclusion of a random effect Si:
ii
i
i
S env
1
?
p
p
?
???
??
?
?
??
?
?
0
1
log
?
The spatial component, Si, is defined by the isotropic, exponentially decaying correlation
function:
?? )
ij
.(exp);(
ij
ddf
????
???????????????????????????????????????????????????????????????????????di,j= xixjmeasures the
Euclidian distance between locations xi and xj.
Page 5
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 5 of 9
Table 1: Results of Bayesian geostatistical models to predict prevalence of P. vivax and P. falciparum for Tanna Island, 2008.
Coefficient, posterior meant
Odds ratio, posterior mean (95%
Bayes credible intervals#)
DIC
Model of P. vivax with distance to coastline fixed effect
α (intercept)
4.680 (5.317 4.137)
Distance from coastline (OR per 1 km)
0.690 (1.151 0.242)0.730 (0.5910.895)
Φ (rate of decay of spatial correlation)*251.5(51.26569.4)
σ2 (variance of geostatistical random effect)** 0.214 (0.056  0.624)
DIC
306.9
Model of P. vivax with elevation fixed effect
α (intercept)
4.611 (5.327  3.931)
Elevation (OR per 100 m)
0.547 (0.992 0.115) 0.654 (0.468, 0.917)
Φ (rate of decay of spatial correlation)*167.8 (33.71, 461.2)
σ2 (variance of geostatistical random effect)**2.471 (1.271, 4.304)
DIC
308.9
Model of P. falciparum with distance to coastline fixed effect
α
5.238 (6.027  4.625)
Distance from coastline (OR per 1 km)
0.101 (0.534  0.334)0.955 (0.783, 1.165)
Φ
289.5 (51.61  575.4)
σ2
0.584 (0.057  4.713)
DIC
219.5
Model of P. falciparum with elevation fixed effect
α
5.129 (5.976  4.416)
Elevation (OR per 100 m)
0.207 (0.673  0.146)0.864 (0.603, 1.149)
Φ
238.2 (53.28  478.9)
σ2
1.753 (0.4521  4.079)
DIC
* The unit is change in spatial autocorrelation per decimal degree. A lower Φ indicates that spatial correlation occurs over longer distances (i.e.
spatial clusters are larger).
** A higher variance indicates a greater tendency toward spatial clustering.
# Bayes credible intervals can be interpreted as having a similar meaning to confidence intervals used in frequentist statistics.
218.9
Page 6
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 6 of 9
restricted to within 2 km of the coast. It appears that the
Green Hill area, which is approximately 5 km from the
coast, does not support the local vector and that malaria
infections were contracted whilst inhabitants visited
coastal areas. The Green Hill area focus requires further
investigation, particularly with respect to patterns of
mobility for work and recreation. In addition, a better
understanding of the transmission patterns could be
obtained from surveying Anopheles farauti breeding sites
on the island during the dry season.
The risk maps have been used to define zones within
which interventions are scaled and planned according to
malaria transmission intensity. This will involve indoor
residual spraying (IRS) on the coastal fringe but not the
hinterland/middle bush area. Additionally the maps pro
vide the base for the design of the surveillance strategy
and will be fully implemented by targeted surveys which
will in turn inform modifications in local implementation
policies such as inclusion of a particular inland hotspot
for IRS.
Whilst the maps of the upper and lower Bayesian credi
ble intervals are important for scientific honesty (i.e.
highlighting areas where the predictions are imprecise)
they are also a useful tool for priority setting and thus aid
ing in the sound and rational deployment of interven
tions. The maps showing the lower Bayesian credible
intervals are particularly useful in highlighting those
locations in which there is high certainty of aboveaver
age malaria prevalence. Active surveillance within high
prevalence foci is a cornerstone for success in interrupt
ing malaria transmission [30]. The upper Bayesian credi
ble intervals are useful in providing an indication of the
maximum extent of malaria transmission or risk.
At the community level the risk maps have been highly
successful in planning meetings. The importance of
grassrootslevel involvement and local ownership of the
elimination goal is well recognized [2]. The island has had
a long history of malaria control and such maps are an
effective way for local citizens to visualize and under
stand the strategy for the elimination programme.
Initial statistics analysis of environmental covariates
revealed no association between malaria risk and NDVI
or TIR. At the small geographical scale of our study (550
km2), NDVI and TIR are relatively homogeneous and
thus not indicative of the presence or absence of malaria.
NDVI has been positively associated with malaria preva
lence throughout Africa [8,18,22,31,32] and the Middle
East [14], but these studies covered wide geographical
areas exhibiting substantial variations in climate. The rel
atively small size of the study area meant that the com
monly used source of vegetation cover, Globcover [33],
was not available at a small enough resolution to usefully
define vector habitats of An. farauti.
The present study found P. vivax risk to be associated
with proximity to the coastline, which has been identified
by others as a desirable An. farauti habitat [34,35]. It is
believed the relatively small number of P. falciparum
cases detected on the island were not sufficient to yield
conclusive results concerning associations with environ
mental variables. Thus despite both P. vivax and P. falci
parum being supported by the same vector species
slightly different relationships with environmental vari
ables were observed.
As the need to define the limits of malaria distribution
and to predict the distribution within these limits
increases, the utility of ecological and climatic variables
in aiding this endeavour have been questioned. While an
issue that cannot be resolved within this brief discussion
suffice to say that as vectors' biological niches continue to
be determined by ecological and climatic factors (along
with human related socioeconomic factors) understand
ing and identifying these variables will continue to be
Table 2: Summary of validation statistics for the geostatistical models described in Table 1.
Model AUC
Mean Error# (% prevalence)
Mean Absolute Error* (% prevalence)
Model of P. vivax with distance to coastline fixed effect
0.8675.07 1.30
Model of P. vivax with elevation fixed effect
0.8575.461.33
Model of P. falciparum with distance to coastline fixed
effect
0.8210.390.55
Model of P. falciparum with elevation fixed effect
0.8560.340.50
AUC between 0.5 and 0.7 indicates a poor discriminative capacity; 0.70.9 indicate a reasonable capacity; and >0.9 indicate a very good
capacity.
# Mean error is a measure of the bias of predictions (the overall tendency to over or under predict).
* Mean absolute error is a measure of overall precision (the average magnitude of error in individual predictions).
Page 7
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 7 of 9
important. However, at differing spatial scales and for dif
ferent vector species the significance of ecological and
climatic variables will vary. For example, increases in alti
tude are commonly associated with cooler temperatures
and thus a less suitable environment for malaria trans
mission for much of Africa [3638], while in the forested
hills of SouthEast Asia increased altitude has been found
to be associated with increased forest cover and thus a
more suitable transmission environment for the vector
species in this ecological niche [39,40]. Looking at smaller
spatial scales, variables such distance to local water bod
ies and a households' location with respect to the village
periphery become important determinants of malaria
risk [10,12].
Accurate data on distribution of population across the
island was not available at the time of this study but has
been identified as a priority for the malaria elimination
programme. However, the population is described to be is
spread over the entire island with an inland plateau
region more densely populated [25]. Once more accurate
population data is available a better understanding of dis
ease burden is possible.
While these maps will serve as baseline maps for the
elimination programme such intensive surveying (i.e. 76%
of all children aged 212 years [25]) will not be necessary
to update the maps. As surveillance operations become
more streamlined the routine data collected from periph
eral aid posts can be used. Additionally, the data collected
throughout the year can be adjusted for seasonality [24]
to provide more accurate estimates of average annual
prevalence.
The current study does come with some limitations,
particularly the georeferencing method, which was
applied retrospectively using independently sourced
coordinates (rather than during the survey using a global
positioning system). Additionally, the crosssectional sur
Figure 4 a) Median predicted spatial distribution of P. vivax prevalence on Tanna Island, Vanuatu in 2008 at approximately 1 km2 resolu
tion, b) lower 25% predicted prevalence, and c) upper 75% predicted prevalence.
Page 8
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 8 of 9
vey design meant that prevalence data were obtained at a
single time point and thus the data only represent a snap
shot of malaria risk, which is known to have important
temporal dynamics. Despite these limitations, as data are
updated with the input from surveillance operations, the
spatial predictions presented here represent the baseline
of a dynamic, and ideally shrinking, malaria risk map for
Tanna Island.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PacMISC, AMI and the VBDCP conceived the study and participated in its
design and data collection. AMI assembled the bulk of the incidence data. AC
and HR devised and implemented the analytical methods. AT provided pro
cessed satellite images for analysis. HR wrote the first draft of the manuscript.
All authors participated in the interpretation of results and in the writing and
editing of the manuscript.
Acknowledgements
Our special thanks go to the people of Tafea Province who participated in the
study and without whom this research would not have been possible. We
thank our colleagues at the National Vector Borne Disease Control Program,
Vanuatu, (particularly Jenifer Iavro, Malao John Kalomuana and Ken Mera) and
at the Australian Army Malaria Institute in Brisbane (especially Christine Atkins,
Alison Auliff, Lisa Bain, Qin Cheng, Robert Cooper, Andrew Ebringer, Mike
Edstein, Nathan Elmes, KarenAnn Gray, John Humphries, MarieLouise John
son, Darren Krause, Ken Lilley, Brady McPherson, Rob Perrin, Wesley Sharrock,
John Staley and Norman Waters), for conducting the field survey in Tafea and
for providing the malaria point prevalence data that was used to develop the
risk maps presented in this paper. We would additionally like to thank Marcel
Tanner and Dennis Shanks from the AusAID Malaria Reference Group for their
guidance and mentorship. AJT is supported by a grant from the Bill and
Melinda Gates Foundation (#49446). This work forms part of the output of the
Pacific Malaria Initiative principally funded by AusAID.
Figure 5 a) Median predicted spatial distribution of P. falciparum prevalence on Tanna Island, Vanuatu in 2008 at approximately 1 km2 res
olution, b) lower 25% predicted prevalence, and c) upper 75% predicted prevalence.
Page 9
Reid et al. Malaria Journal 2010, 9:150
http://www.malariajournal.com/content/9/1/150
Page 9 of 9
Author Details
1Pacific Malaria Initiative Support Centre (PacMISC), Australian Centre for
International and Tropical Health (ACITH), School of Population Health,
University of Queensland, Queensland, Australia, 2National Vector Borne
Disease Control Program (VBDCP), Ministry of Health, Port Vila, Vanuatu,
3Australian Army Malaria Research Institute, Department of Defence,
Government of Australia, Queensland, Australia, 4Emerging Pathogens
Institute and Department of Geography, University of Florida, Gainesville, USA
and 5Australian Centre for Tropical and International Health, Queensland
Institute of Medical Research, Brisbane, Queensland, Australia
References
1.Feachem RA, Phillips A, Targett G: Shrinking the malaria map: A prospectus
on malaria elimination San Francisco: The Malaria Elimination Group; 2009.
2.Kaneko G, Taleo M, Kalkoa S, Yamar T, Kobayakawa A, Björkman A: Malaria
eradication on islands. Lancet 2000, 356:15601564.
3.Sabot O, Tulloch J, Basu S, Dyckman W, Moonasar D, Moonenf B: Getting
to Zero. In Shrinking the malaria map: A prospectus on elimination Edited
by: Feachem RGA, Phillips AA, Targett GA. San Francisco: The Global
Health Group; 2009.
4.Greenwood BM: Control to elimination: implications for malaria
research. Trends Parasitol 2008, 24:449454.
5. Carter R, Mendis KN, Roberts D: Spatial targeting of interventions
against malaria. Bull World Health Organ 2000, 78:14011411.
6.Ernst K, Adoka S, Kowuor D, Wilson M, John C: Malaria hotspot areas in a
highland Kenya site are consistent in epidemic and nonepidemic
years and are associated with ecological factors. Malar J 2006, 5:78.
7.Brooker S, Clements A, Hotez P, Hay S, Tatem A, Bundy D, Snow R: The co
distribution of hookworm and Plasmodium falciparum among African
schoolchildren. Malar J 2006, 5:99.
8.Omumbo JA, Hay SI, Snow RW, Tatem AJ, Rogers DJ: Modelling malaria
risk in East Africa at highspatial resolution. Trop Med Int Health 2005,
10:557566.
9.Kreuels B, Kobbe R, Adjei S, Kreuzberg C, von Reden C, Bater K, Klug S,
Busch W, Adjei O, May J: Spatial variation of malaria incidence in young
children from a geographically homogeneous area with high
endemicity. J Infect Dis 2008, 197:8593.
10. Myers W, Myers A, CoxSingh J, Lau H, Mokuai B, Malley R: Micro
geographic risk factors for malarial infection. Malar J 2009, 8:27.
11. Kazembe L, Kleinschmidt I, Holtz T, Sharp B: Spatial analysis and
mapping of malaria risk in Malawi using pointreferenced prevalence
of infection data. Int J Health Geogr 2006, 5:41.
12. Cohen JM, Ernst K, Lindblade KA, Vulule JM, Chandy CJ, Wilson ML:
Topographyderived wetness indices are associated with household
level malaria risk in two communities in the western Kenyan
highlands. Malar J 2008, 7:.
13. Basáñez M, Marshall C, Carabin H, Gyorkos T, Joseph L: Bayesian statistics
for parasitologists. Trends Parasitol 2004, 20:8591.
14. Brooker S, Leslie T, Kolaczinski K, Mohsen E, Mehboob N, Saleheen S,
Khudonazarov J, Freeman T, Clements A, Rowland M, Kolaczinski J: Spatial
epidemiology of Plasmodium vivax, Afghanistan. Emerg Infect Dis 2006,
12:16001602.
15. Kleinschmidt I, Sharp BL, Clarke GPY, Curtis B, Fraser C: Use of generalized
linear mixed models in the spatial analysis of smallarea malaria
incidence rates in KwaZulu Natal, South Africa. Am J Epidemiol 2001,
153:12131221.
16. Gemperli A, Vounatsou P, Sogoba N, Smith T: Malaria mapping using
transmission models: application to survey data from Mali. Am J
Epidemiol 2006, 163:289297.
17. Noor AM, Clements A, Gething PW, Moloney G, Borle M, Shewchuk T, Hay
SI, Snow RW: Spatial predition of Plasmodium falciparum prevalence in
Somalia. Malar J 2008, 7:.
18. Kleinschmidt I, Bagayoko M, Clarke GP, Craig M, le Sueur D: A spatial
statistical approach to malaria mapping. Int J Epidemiol 2000,
29:355361.
19. Kleinschmidt I, Sharp B, Mueller I, Vounatsou P: Rise in malaria incidence
rates in South Africa: A smallarea spatial analysis of variation in time
trends. Am J Epidemiol 2002, 155:257264.
20. Gemperli A, Sogoba N, Fondjo E, Mabaso M, Bagayoko M, Briet OJT, Liebe
J, Anderegg D, Smith T, Vounatsou P: Mapping malaria transmission in
West and Central Africa. Trop Med Int Health 2006, 11:10321046.
21. Kleinschmidt I, Omumbo J, Briët O, van de Giesen N, Mensah NK,
Windmeijer P, Moussa M, Teuscher T: An empirical malaria distribution
map for West Africa. Trop Med Int Health 2001, 6:779786.
22. Gosoniu L, Vounatsou P, Sogoba N, Maire N, Smith T: Mapping malaria
risk in West Africa using a Bayesian nonparametric nonstationary
model. Comput Stat Data Anal 2009, 53:33583371.
23. Guerra C, Gikandi P, Tatem A, Noor A, Smith D, Hay S, Snow R: The limits
and intensity of Plasmodium falciparum transmission: implications for
malaria control and elimination worldwide. PLoS Med 2008, 5:e38.
24. Hay SI, Guerra CA, Gething PW, Patil AP, Tatem AJ, Noor AM, Kabaria CW,
Manh BH, Elyazar IRF, Brooker S, Smith DL, Moyeed RA, Snow RW: A World
Malaria Map: Plasmodium falciparum endemicity in 2007. PloS Med
2009, 6:e1000048.
25. Melanesian survey team: Malaria on isolated melanesian islands prior to
the initiation of malaria elimination activities. Mal J in press.
26. United States Geological Survey (USGS) Earth Resources Observation
and Science Center [http://glovis.usgs.gov]
27. Tucker C: Red and photographic infrared linear combinations for
monitoring vegetation. Rem Sens Env 1979, 8:127150.
28. Consultative Group on International Agricultural Research Consortium
for Spatial Information [http://srtm.csi.cgiar.org/]
29. Diggle PJ, Tawn JA, Moyeed R: Modelbased geostatistics. J Appl Stat
1998, 47:299350.
30. WHO: Guidelines on the elimination of residual foci of malaria transmission
World Health Organization; 2007.
31. Gosoniu L, Vounatsou P, Sogoba N, Smith T: Bayesian modelling of
geostatistical malaria risk data. Geospat Health 2006, 1:127139.
32. Hay S, Snow R, Rogers D: Predicting malaria seasons in Kenya using
multitemporal meteorological satellite sensor data. Trans Roy Soc Trop
Med Hyg 1998, 92:1220.
33. Global RuralUrban Mapping Project (GRUMP) [Alpha version]
Socioeconomic Data and Applications Center, Columbia University.
34. Belkin JN: The mosquitoes of the South Pacific (Diptera, Culicidae) Berkeley
and Los Angeles: University of California Press; 1962.
35. Sweeney AW, Beebe NW, Cooper RD, Bauer JT, Peterson AT:
Envionmental factors associated with the distribution and range limits
of malaria vector Anopheles farauti in Australia. J Med Entomol 2006,
43:10681075.
36. Drakeley C, Carneiro I, Reyburn H, Malima R, Lusingu J, Cox J, Theander T,
Nkya WMM, Lemnge MM, Riley EM: Altitudedependent and
independent variations in Plasmodium falciparum prevalence in
northeastern Tanzania. J Infect Dis 2005, 191:15891598.
37. Lindsay SW, Martens WJM: Malaria in the African highlands: past,
present and future. Bull World Health Organ 1998, 76:3345.
38. Hay SI, Omumbo J, Craig M, Snow RW: Earth observation, geographic
information systems and Plasmodium falciparum malaria in sub
Saharan Africa. Adv Parasitol 2000, 47:173215.
39. Das NG, Talukdar PK: Epidemiological and entomological aspects of
malaria in forestfringed villages of Sonitpur district, Assam. J Vector
Borne Dis 2004, 41:59.
40. Erhart A, Thang ND, Hung NQ, Toi LV, Hung LX, Tuy TQ, Cong LD,
Speybroeck N, Coosemans M, D'Alessandro U: Forest malaria in Vietnam:
a challenge for control. Am J Trop Med Hyg 2004, 70:110118.
doi: 10.1186/147528759150
Cite this article as: Reid et al., Baseline spatial distribution of malaria prior to
an elimination programme in Vanuatu Malaria Journal 2010, 9:150
Received: 2 December 2009 Accepted: 2 June 2010
Published: 2 June 2010
This article is available from: http://www.malariajournal.com/content/9/1/150© 2010 Reid et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Malaria Journal 2010, 9:150