Characterizing the spatial and temporal variation of malaria incidence in Bangladesh, 2007.
ABSTRACT Malaria remains a significant health problem in Bangladesh affecting 13 of 64 districts. The risk of malaria is variable across the endemic areas and throughout the year. A better understanding of the spatial and temporal patterns in malaria risk and the determinants driving the variation are crucial for the appropriate targeting of interventions under the National Malaria Control and Prevention Programme.
Numbers of Plasmodium falciparum and Plasmodium vivax malaria cases reported by month in 2007, across the 70 endemic thanas (sub-districts) in Bangladesh, were assembled from health centre surveillance reports. Bayesian Poisson regression models of incidence were constructed, with fixed effects for monthly rainfall, maximum temperature and elevation, and random effects for thanas, with a conditional autoregressive prior spatial structure.
The annual incidence of reported cases was 34.0 and 9.6 cases/10,000 population for P. falciparum and P. vivax respectively and the population of the 70 malaria-endemic thanas was approximately 13.5 million in 2007. Incidence of reported cases for both types of malaria was highest in the mountainous south-east of the country (the Chittagong Hill Tracts). Models revealed statistically significant positive associations between the incidence of reported P. vivax and P. falciparum cases and rainfall and maximum temperature.
The risk of P. falciparum and P. vivax was spatially variable across the endemic thanas of Bangladesh and also highly seasonal, suggesting that interventions should be targeted and timed according to the risk profile of the endemic areas. Rainfall, temperature and elevation are major factors driving the spatiotemporal patterns of malaria in Bangladesh.
- SourceAvailable from: Waziul Haque[Show abstract] [Hide abstract]
ABSTRACT: Background Malaria is endemic in 13 of 64 districts in Bangladesh. About 14 million people are at risk. Some evidence suggests that the prevalence of malaria in Bangladesh has decreased since the the Global Fund to Fight AIDS, Tuberculosis and Malaria started to support the National Malaria Control Program (NMCP) in 2007. We did an epidemiological and economic assessment of malaria control in Bangladesh. Methods We obtained annually reported, district-level aggregated malaria case data and information about disbursed funds from the NMCP. We used a Poisson regression model to examine the associations between total malaria, severe malaria, malaria-attributable mortality, and insecticide-treated net coverage. We identifi ed and mapped malaria hotspots using the Getis-Ord Gi* statistic. We estimated the cost-eff ectiveness of the NMCP by estimating the cost per confi rmed case, cost per treated case, and cost per person of insecticide-treated net coverage. Findings During the study period (from Jan 1, 2008, to Dec 31, 2012) there were 285 731 confi rmed malaria cases. Malaria decreased from 6·2 cases per 1000 population in 2008, to 2·1 cases per 1000 population in 2012. Prevalence of all malaria decreased by 65% (95% CI 65–66), severe malaria decreased by 79% (78–80), and malaria-associated mortality decreased by 91% (83–95). By 2012, there was one insecticide-treated net for every 2·6 individuals (SD 0·20). Districts with more than 0·5 insecticide-treated nets per person had a decrease in prevalence of 21% (95% CI 19–23) for all malaria, 25% (17–32) for severe malaria, and 76% (35–91) for malaria-associated mortality among all age groups. Malaria hotspots remained in the highly endemic districts in the Chittagong Hill Tracts. The cost per diagnosed case was US$0·39 (SD 0·02) and per treated case was $0·51 (0·27); $0·05 (0·04) was invested per person per year for health education and $0·68 (0·30) was spent per person per year for insecticide-treated net coverage. Interpretation Malaria elimination is an achievable prospect in Bangladesh and failure to push for elimination nearly ensures a resurgence of disease. Consistent fi nancing is needed to avoid resurgence and maintain elimination goals. Funding None.The Lancet Global Health. 02/2014; 2(2).
- [Show abstract] [Hide abstract]
ABSTRACT: The System for Early-warning based on Emergency Data (SEED) is a pilot project to evaluate the use of emergency call data with the main complaint acute undifferentiated fever (AUF) for syndromic surveillance in India. While spatio-temporal methods provide signals to detect potential disease outbreaks, additional information about socio-ecological exposure factors and the main population at risk is necessary for evidence-based public health interventions and future preparedness strategies. The goal of this study is to investigate whether a spatial epidemiological analysis at the ecological level provides information on urban-rural inequalities, socio-ecological exposure factors and the main population at risk for AUF. Our results displayed higher risks in rural areas with strong local variation. Household industries and proximity to forests were the main socio-ecological exposure factors and scheduled tribes were the main population at risk for AUF. These results provide additional information for syndromic surveillance and could be used for evidence-based public health interventions and future preparedness strategies. Copyright © 2014. Published by Elsevier Ltd.Health & Place 12/2014; 31C:111-119. · 2.44 Impact Factor
- The Lancet Global Health. 02/2014;
Characterizing the spatial and temporal variation
of malaria incidence in Bangladesh, 2007
Heidi L Reid1ˆ, Ubydul Haque2, Shyamal Roy3, Nazrul Islam3and Archie CA Clements1*
Background: Malaria remains a significant health problem in Bangladesh affecting 13 of 64 districts. The risk of
malaria is variable across the endemic areas and throughout the year. A better understanding of the spatial and
temporal patterns in malaria risk and the determinants driving the variation are crucial for the appropriate targeting
of interventions under the National Malaria Control and Prevention Programme.
Methods: Numbers of Plasmodium falciparum and Plasmodium vivax malaria cases reported by month in 2007,
across the 70 endemic thanas (sub-districts) in Bangladesh, were assembled from health centre surveillance reports.
Bayesian Poisson regression models of incidence were constructed, with fixed effects for monthly rainfall, maximum
temperature and elevation, and random effects for thanas, with a conditional autoregressive prior spatial structure.
Results: The annual incidence of reported cases was 34.0 and 9.6 cases/10,000 population for P. falciparum and
P. vivax respectively and the population of the 70 malaria-endemic thanas was approximately 13.5 million in 2007.
Incidence of reported cases for both types of malaria was highest in the mountainous south-east of the country
(the Chittagong Hill Tracts). Models revealed statistically significant positive associations between the incidence of
reported P. vivax and P. falciparum cases and rainfall and maximum temperature.
Conclusions: The risk of P. falciparum and P. vivax was spatially variable across the endemic thanas of Bangladesh
and also highly seasonal, suggesting that interventions should be targeted and timed according to the risk profile
of the endemic areas. Rainfall, temperature and elevation are major factors driving the spatiotemporal patterns of
malaria in Bangladesh.
Keywords: Malaria, Bangladesh, Spatial, Temporal, Seasonality, Bayesian, Plasmodium falciparum, Plasmodium vivax
Spatial and temporal variability in malaria transmission are
known to be driven partly by ecological factors that affect
the survival and population size of anopheline vectors.
Temperature, humidity and water available for breeding
habitats have been shown to be important primary eco-
logical factors . Fluctuation in malaria risk has significant
public health implications for ensuring adequate provision
of anti-malarials, the delivery of intermittent preventative
therapy (IPT) and optimizing the timing and frequency of
indoor residual spraying with insecticides and/or distribu-
tion of long-lasting insecticidal nets (LLIN), and thus war-
At a macro-scale, attempts have been made to under-
stand the mechanisms driving malaria risk. The Mapping
Malaria Risk in Africa (MARA) collaboration developed
maps that describe the expected duration and timing of
the transmission season by country based on climate suit-
ability models [2,3]. Insight into the underlying ecological
mechanisms both inhibiting and propelling malaria trans-
mission has led to research into malaria early warning sys-
tems (MEWS) to be able to predict disease patterns based
on known relationships between the disease and eco-
logical variables [4-11].
A host of issues, however, make characterizing the nat-
ural phenomena underlying spatial and temporal patterns
in malaria risk difficult. Differences in ecological require-
ments between malaria vectors mean that particular envir-
onmental events, such as the rainy season, can lead to an
increase in vector capacity for most vectors but an initial
* Correspondence: email@example.com
1Infectious Disease Epidemiology Unit, Level 4 Public Health Building, School
of Population Health, University of Queensland, Herston, QLD 4006, Australia
Full list of author information is available at the end of the article
© 2012 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.
Reid et al. Malaria Journal 2012, 11:170
decrease for others [12-14]. Variability due to ecological
drivers can be further complicated by patterns in host im-
munity, which is a possible explanation for intra-annual
and inter-annual variation [15-17]. In addition, the longi-
tudinal datasets required for analysis of both spatial and
temporal patterns are commonly obtained from national
health information system systems, from which the accur-
acy of estimates is highly variable .
The advancement of geographical information systems
(GIS) and spatial statistics has greatly improved the under-
standing of malaria dynamics, including its dependence
on ecological factors. Bayesian geostatistical models are
now commonly employed to generate malaria risk maps
providing a valuable evidence base for programmatic
decision-making [19-26]. For malaria data aggregated by
administrative area, as is often the case with health infor-
mation system data, Bayesian conditional autoregressive
(CAR) models have been used to model spatial and tem-
poral patterns [17,27-29]. This approach has the advan-
tage of accounting for spatial correlation in the data and
smoothing out variability associated with small popula-
tions in some areas.
Malaria remains a significant health problem in Bangla-
desh affecting 13 of 64 districts. In 2006, the Global Fund
to Fight AIDS, Tuberculosis, and Malaria awarded Bangla-
desh $39.6 million US dollars to support the national
malaria-control programme . Three key objectives of
the programme are to provide effective diagnosis and
treatment to 80% of estimated malaria cases by 2012, to
promote the use of LLINs in 80% of households in these
districts, and to use selective indoor residual spraying for
containment of outbreaks .
While spatial patterns of Plasmodium falciparum preva-
lence have been previously described in Bangladesh , spa-
tiotemporal patterns of P. falciparum and Plasmodium vivax
incidence have not. The aims of this study were to describe
the spatiotemporal patterns of reported malaria cases and
identify environmental drivers of the spatiotemporal patterns
of malaria risk in Bangladesh. This information may affect
the timing and geographical targeting of interventions such
as indoor residual spraying and distribution of LLINs.
Through the more focussed implementation of interventions,
the findings presented here have the potential to enhance the
effectiveness of the national malaria control programme and
provide valuable baseline epidemiological information upon
which to chart the progress of the programme.
Malaria surveillance data and descriptive methods
The Bangladesh Ministry of Health lists malaria as a noti-
fiable disease and maintains routine surveillance of clinical
malaria nationwide. Cases of malaria diagnosed at health
facilities are reported to the thana health complex, and
collated up through the district and division administrative
levels. From 2004 to 2009, cases in Bangladesh were con-
firmed as uncomplicated malaria presumptive (UMP), un-
complicated malaria confirmed (UMC), as well as severe
malaria (SM) and vivax malaria (VM). Neither microscopy
nor rapid diagnosis tests (RDT) were performed for UMP,
but for the others, either microscopy or a RDT was used
for diagnosis. According to official national figures, 98% of
confirmed cases were diagnosed using microscopy in 2007
and 2% using a RDT, with approximately 76% of cases
being UMP . Since 2007, the situation has changed,
with 63% of cases diagnosed using a RDTand only 10% of
cases being UMP in 2010 .
For the purposes of the present study, reported numbers
of P. vivax and P. falciparum malaria cases across all age
groups were assembled by month for each of the 70 tha-
nas in the 13 endemic districts between January and De-
cember 2007. Population data from the national census
conducted in 2001 were adjusted according to national
growth figures , assuming an even population growth
across all thanas. The population was then used as a de-
nominator to calculate malaria incidence in each thana. In
the 13 malaria-endemic districts, the incidence of reported
P. falciparum malaria cases for the whole of 2007 was
compared to prevalence determined by population-based
surveys using RDTs conducted in the same year . Pair-
wise, Pearson cross correlations between incidence of P.
falciparum and P. vivax for each month (January–Decem-
ber) were calculated to determine whether the distribution
of the two species was related.
Climate and elevation data
A number of ecological and climatic factors affect both the
extrinsic life cycle of the malaria parasite and that of the
anopheline vectors . Based on published literature, vari-
ables pertinent to malaria transmission in Bangladesh were
selected for statistical analysis. High-resolution (1 sq km)
raster maps of interpolated long-term average monthly
rainfall, and minimum and maximum temperature, were
obtained from the WORLDCLIM project . Digital ele-
vation data (in meters above sea level for cells of a 1 sq km
grid), were obtained from the same source.
Data concerning vegetation cover were obtained from the
GlobCover Land Cover product, which is derived from sat-
ellite imagery from 2005 to 2006 . Vegetation was la-
belled according to the United Nations (UN) Land Cover
Classification System  at a resolution of 300 m. For this
analysis, vegetation cover was dichotomized into forested,
which included multiple specific forest categories, and not
forested, which included all other vegetation categories.
Rainfall, temperature, elevation and vegetation maps
were imported into the geographical information system
(GIS) ArcView version 9 (ESRI, Redlands, CA, USA) and
linked spatially to a digitized boundary map of the 70
malaria-endemic thanas. For each thana, the mean values
Reid et al. Malaria Journal 2012, 11:170
Page 2 of 8
of the raster cells contained in the thana for rainfall,
temperature and elevation, and percentage forest cover,
were computed in the GIS to define covariates in subse-
quent models. Environmental variables were standardized
by centreing the covariates on their mean and dividing by
the standard deviation to assist mixing and convergence of
the spatiotemporal models.
Bayesian spatiotemporal Poisson regression models were
constructed using the WinBUGS software, version 1.4.3
(MRC Biostatistics Unit, Cambridge, UK). The outcome
was the monthly number of reported cases of P. vivax or
P. falciparum malaria (January to December 2007) in
each thana and the offset was the expected number of
cases based on the population of each thana.
The final regression model was:
where Oijis the observed number of cases in thana i,
month j and
where Eiis the expected monthly number of cases in
thana i, which did not vary by month because the popu-
lation was considered static. The mean log relative risk
was modelled as:
?¼ log Ei
ð Þ þ θij;
θij¼ α þ βñeleviþ δñprecijþ φñtempijþ si
where α is the intercept, β, δ and ∅ are the coefficients for
mean elevation (elev), precipitation (prec), and maximum
temperature (temp), respectively, and siis a spatial county-
level random effect. Note, minimum temperature and per-
cent forest cover were excluded because of collinearity
(Pearson’s correlation> |0.7|) with maximum temperature
and elevation, respectively. Spatial structuring in si was
modelled using a conditional autoregressive prior structure,
where spatial relationships between counties were modelled
using a simple adjacency weights matrix . If two coun-
ties were adjacent, the weight=1 and if they were not adja-
cent, the weight=0. Non-informative priors were used for
the intercept (an unbounded uniform, or flat, prior), coeffi-
cients (Gaussian priors with mean=0 and variance=
10,000) and the variance of si(a gamma prior with shape
and scale parameters=0.01). Models were created with,
and without, the environmental covariates to determine
their ability to capture spatial and temporal effects in the
data. The residuals of the models were examined to deter-
mine whether any temporal patterns, including seasonality
and temporal autocorrelation, remained after adjusting for
the effects of rainfall and temperature. Additionally, models
were compared using the deviance information criterion
(DIC)to determinewhether incorporationofthe
environmental covariates improved the fit of the models
(a lower DIC indicates a better-fitting model).
The population of the 70 malaria-endemic thanas was ap-
proximately 13.5 million in 2007. In these thanas, there
was a total of 45,761 reported cases of P. falciparum mal-
aria in 2007, representing a crude incidence of 34.0 cases
per 10,000 people; and 12,968 reported cases of P. vivax
malaria, representing a crude incidence of 9.6 cases per
10,000 people. Maps revealed high reported incidence of
both species in the south-east of the country (i.e. the Chit-
tagong Hill Tracts) (Figure 1).
The monthly incidence of reported P. vivax and P. fal-
ciparum cases is shown by thana in Figure 2. It is evident
that in some thanas there was a strong seasonal pattern
but that incidence was highly over-dispersed. Seasonal
comparisons between the two types of malaria are clearer
on a plot of the overall incidence for the 13 endemic dis-
tricts (Figure 3). P. falciparum showed a peak in incidence
in July while P. vivax peaked in June. There was a moder-
ately high correlation (range 0.30–0.62) between incidence
of reported P. falciparum and P. vivax in any given month
(Table 1). At the district level, there was also a strong posi-
tive association between the incidence of reported cases of
P. falciparum malaria and the prevalence of P. falciparum
infection determined by the population-based surveys
conducted in the same year (Figure 4).
The P. falciparum and P. vivax models revealed simi-
lar positive associations with maximum temperature and
monthly precipitation (Tables 2 and 3). The impact of
elevation on the risk of P. falciparum and P. vivax dif-
fered, with a higher risk of P. falciparum and a lower
risk of P. vivax at increasing elevation.
Residual analysis indicated that incorporation of environ-
mental variables was effective in removing seasonality from
the data, with only a small amount of unstructured tem-
poral variation and no temporal autocorrelation evident in
the residuals (indicating that it was not necessary to incorp-
orate a temporal autoregressive term in the model). The en-
vironmental covariates were also able to explain some of
the spatial variability in malaria counts, measured through
the variance of the spatial random effect (si). Inclusion of
environmental variables into the models reduced the vari-
ance of the spatial random effect by 32.4% for P. falciparum
and 6.7% for P. vivax compared to models without environ-
mental covariates (Table 2). Incorporation of the environ-
mental models improved the fit of the models for both
types of malaria, as indicated by a lower DIC.
Malaria in Bangladesh exhibits distinct spatial and temporal
patterns, with varying levels of endemicity and significant,
biologically tenable relationships to environmental variables.
Reid et al. Malaria Journal 2012, 11:170
Page 3 of 8
Figure 1 Incidence of reported cases of Plasmodium falciparum and Plasmodium vivax per 1,000 people across the endemic thanas in
Figure 2 Thana-level monthly incidence (grey lines) and overall reported incidence across endemic thanas (black lines) of reported
cases of A. Plasmodium falciparum and B. Plasmodium vivax malaria in Bangladesh, 2007.
Reid et al. Malaria Journal 2012, 11:170
Page 4 of 8
Importantly, the analysis provides insight into the relatively
unexplored epidemiology of P. vivax within the country.
The inclusion of ecological covariates in the model pro-
vided some explanation as to the temporal and spatial vari-
ation in malaria risk. Temperature and rainfall showed an
expected positive relationship with increased malaria risk,
which is consistent with similar analyses of incidence data
from Ethiopia, Peru and China [17,40,41]. These two vari-
ables were effective in explaining the seasonal patterns in
the data. The environmental covariates could explain a siz-
able percentage of spatial variability observed in distribution
of P. falciparum but only a relatively modest percentage
seen with P. vivax. Other unmeasured factors including
local malaria control activities , socio economic indica-
tors  and population movement  are likely to con-
tribute to variability in malaria risk, to differing degrees for
the two species. The positive relationship between P. falcip-
arum and elevation (after adjusting for the effects of
temperature) could be explained by higher elevations pro-
viding denser forest cover and a more favourable environ-
ment for forest breeding vectors known to sustain
transmission in these areas . However, given that the
vectors for the two species overlap in Bangladesh, the op-
posite finding for P. vivax was unexpected. With little
known about the epidemiology of P. vivax in the country,
this finding requires further investigation. However, of note,
the odds ratio for elevation in the P. vivax model was of a
small magnitude and was only marginally significant.
Plasmodium falciparum and P. vivax showed broadly
similar spatial distributions which is consistent with more
recent entomological information that suggests the same
vectors are able to transmit both species. A study con-
ducted in three ecologically distinct areas of the Chitta-
gong Hill Tracts identified eight Anopheles species to be
positive for circumsporozoite protein with a number of
Figure 3 Monthly incidence of reported cases of Plasmodium falciparum and Plasmodium vivax across endemic thanas in Bangladesh,
Table 1 Pearson cross correlations comparing the thana-
level incidence of Plasmodium vivax with Plasmodium
falciparum for each month
MonthCorrelation between the incidence of P. falciparum and
P. vivax in each month
P. falciparum/P. vivax
Reid et al. Malaria Journal 2012, 11:170
Page 5 of 8
these vectors infected by both P. falciparum and P. vivax
. Prior to this study no data were available document-
ing vectors of P. vivax in Bangladesh. The moderately high
correlation between the incidence of P. falciparum and P.
vivax for each of the months supports the entomological
evidence, although this could be affected by factors such
as geographical differences in the quality of laboratory
practices that are used to differentiate malaria due to the
different parasites. The correlation between P. falciparum
and P. vivax, and the effect estimates in the P. vivax mod-
els, could also be sensitive to relapsing of vivax malaria.
Unfortunately the surveillance system was not able to dif-
ferentiate relapsing infections from primary P. vivax infec-
tions and it was not possible to test the potential impact
of relapsing infections. Inter-species interactions, includ-
ing cross-protective immunity, might influence observed
heterogeneity of malaria infections in areas co-endemic
for P. falciparum and P. vivax . Additionally, back-
ground or ongoing malaria control could influence the
ability to accurately quantify correlations between P. fal-
ciparum and P. vivax malaria and effect estimates (such as
between elevation and malaria incidence) in the analysis.
The results presented here suggest that P. vivax is more
widespread than found in cross-sectional surveys con-
ducted in the same year . Depending on the accuracy
of diagnosis of P. vivax malaria in the surveillance system,
this has significant implications for treatment of P. vivax,
with national treatment guidelines recommending the use
of chloroquine and primaquine . Primaquine has been
associated with acute haemolytic anaemia in patients with
G6PD deficiency and, therefore, knowledge of the preva-
lence and geographical distribution of this inherited blood
condition in P. vivax endemic areas is important . The
only cross-sectional data on G6PD deficiency in Bangla-
desh found a prevalence of 10.7% .
An important consideration for this analysis is that the
data were obtained through the national passive surveil-
lance system, which is known to grossly under-report mal-
aria infections . One contributing factor is that many
infections are asymptomatic. Asymptomatic infections have
been shown to be significant in Bangladesh, with two to 40
asymptomatic cases per 1,000 population reported in the
Chittagong region . Additionally, cases who seek treat-
ment through private health services are not registered by
Figure 4 A comparison of incidence of reported cases of Plasmodium falciparum malaria by district and the prevalence of P. falciparum
infection from population-based surveys using rapid diagnostic tests in the 13 malaria-endemic districts of Bangladesh, 2007.
Table 2 Bayesian Poisson regression models of Plasmodium falciparum, Bangladesh, 2007
−1.23 (−1.28 - -1.19)
−1.02 (−1.06 - -0.98) Intercept
Elevation (10 m increase)1.45 (1.39–1.49)
Maximum temperature (1°C increase)1.012 (1.011–1.013)
Precipitation (100 mm increase)1.15 (1.14–1.15)
Variance CAR random effect (si) 0.10 (0.07–0.14) 0.15 (0.10 – 0.21)
Deviance Information Criterion 11774.6 23356.4
*Estimates for the covariates are relative risks; results show mean and 95% credible interval (CrI).
Reid et al. Malaria Journal 2012, 11:170
Page 6 of 8
the national surveillance system. Misdiagnosis and under-
reporting of malaria infections might affect the estimate of
mean incidence, the proportion of cases that are P. vivax
and the observed spatial patterns in the data, if there was
systematic spatial variation in the quality of the surveillance
system, as well as the effect estimates of the environmental
covariates. Without being able to measure the quality of the
surveillance system, it is not possible to speculate as to what
the magnitude of these effects might be. However, the com-
parison of incidence reported through the national passive
population-based survey methods, showed strong congru-
ence at the district level (Figure 4). This is consistent with
recent research that indicates incidence (calculated accord-
ing to active case detection data) increases smoothly and
then flattens out as prevalence increases , increasing
confidence in the observed spatial patterns in the incidence
data reported here. Seasonal migration of people, particu-
larly workers, might also affect observed spatial patterns.
Another limitation is the short time series (12 months) used
in this analysis. A longer time series is needed to establish
longer-term temporal variability of malaria, including im-
portant inter-annual patterns, seasonal patterns, and the
predictive value of climatic risk factors.
Health information system data have been used to es-
tablish spatial variation of disease risk for P. falciparum
and P. vivax in Bangladesh. Given the observed spatial
and temporal variation, targeting of interventions has
the potential to enhance the effectiveness of the national
malaria-control programme in Bangladesh.
HLR, UH and ACAC conceived the study. HLR conducted the analysis and
drafted the manuscript. UH compiled the data, assisted with the analysis and
interpretation of the results, and contributed to the final draft. SR and NI
facilitated access to the data and provided contextual information necessary
to interpret the findings. ACAC provided guidance on the analysis and
drafting of the manuscript. All authors read and approved the final
ACAC is supported by a National Health and Medical Research Council
Career Development Award. UH is supported by an A. Ralph and Sylvia E.
Barr Fellowship from the Johns Hopkins Bloomberg School of Public Health
W. Harry Feinstone Department of Molecular Microbiology and Immunology.
This paper is dedicated to Heidi Reid in recognition of her brief but glittering
contribution to malaria epidemiology and for the lasting impression she
made on her colleagues and dear friends.
1Infectious Disease Epidemiology Unit, Level 4 Public Health Building, School
of Population Health, University of Queensland, Herston, QLD 4006, Australia.
2Department of Molecular Microbiology and Immunology, Johns Hopkins
Bloomberg School of Public Health, Baltimore, MD 21205, USA.3Ministry of
Health and Family Welfare, Malaria and Parasitic Disease Control, Director
General of Health Services, Mohakhali, Dhaka, Bangladesh.
Received: 5 April 2012 Accepted: 10 May 2012
Published: 21 May 2012
1.Stresman GH: Beyond temperature and precipitation: ecological risk
factors that modify malaria transmission. Acta Trop 2010, 116:167–172.
2. Craig MH, Snow RW, le Sueur D: A climate-based distribution model
of malaria transmission in sub-Saharan Africa. Parasitol Today 1999,
3.Tanser FC, Sharp B, le Sueur D: Potential effect of climate change on
malaria transmission in Africa. Lancet 2003, 362:1792–1798.
4. Hay SI, Were EC, Renshaw M, Noor AM, Ochola SA, Olusanmi I, Alipui N,
Snow RW: Forecasting, warning, and detection of malaria epidemics: a
case study. Lancet 2003, 361:1705–1706.
5.Bouma MJ, van der Kaay HJ: The El Nino Southern Oscillation and the
historic malaria epidemics on the Indian subcontinent and Sri Lanka: an
early warning system for future epidemics? Trop Med Int Health 1996,
6. Thomson MC, Connor SJ: The development of malaria early warning
systems for Africa. Trends Parasitol 2001, 17:438–445.
7.Abeku TA, de Vlas SJ, Borsboom G, Teklehaimanot A, Kebede A, Olana D,
van Oortmarssen GJ, Habbema JD: Forecasting malaria incidence from
historical morbidity patterns in epidemic-prone areas of Ethiopia: a
simple seasonal adjustment method performs best. Trop Med Int Health
8.Rogers DJ, Randolph SE, Snow RW, Hay SI: Satellite imagery in the study
and forecast of malaria. Nature 2002, 415:710–715.
9.Abeku TA, Hay SI, Ochola S, Langi P, Beard B, de Vlas SJ, Cox J: Malaria
epidemic early warning and detection in African highlands. Trends
Parasitol 2004, 20:400–405.
10.Ceccato P, Connor SJ, Jeanne I, Thomson MC: Application of geographical
information systems and remote sensing technologies for assessing and
monitoring malaria risk. Parassitologia 2005, 47:81–96.
11.Cox J, Abeku TA: Early warning systems for malaria in Africa: from
blueprint to practice. Trends Parasitol 2007, 23:243–246.
12.Ratanatham S, Upatham ES, Prasittisuk C, Rojanasunan W, Theerasilp N,
Tremongkol A, Viyanant V: Bionomics of Anopheles minimus and its role in
malaria transmission in Thailand. Southeast Asian J Trop Med Public Health
13.Rosenberg R, Andre RG, Somchit L: Highly efficient dry season
transmission of malaria in Thailand. Trans R Soc Trop Med Hyg 1990,
14.Somboon P, Aramrattana A, Lines J, Webber R: Entomological and
epidemiological investigations of malaria transmission in relation to
population movements in forest areas of north-west Thailand. Southeast
Asian J Trop Med Public Health 1998, 29:3–9.
Table 3 Bayesian Poisson regression models of Plasmodium vivax, Bangladesh, 2007
Variable With environmental
−1.02 (−1.14 - -0.92)
−0.93 (−1.05 - -0.84)Intercept
Elevation (10 m increase)0.98 (0.96 – 1.00)
Maximum temperature (1°C increase)1.010 (1.009 – 1.011)
Precipitation (100 mm increase)1.09 (1.09 – 1.10)
Variance CAR random effect (si)0.11 (0.07–0.15)0.12 (0.08 – 0.17)
Deviance Information Criterion8950.110735.8
Reid et al. Malaria Journal 2012, 11:170
Page 7 of 8
15.Hay SI, Myers MF, Burke DS, Vaughn DW, Endy T, Ananda N, Shanks GD,
Snow RW, Rogers DJ: Etiology of interepidemic periods of mosquito-
borne disease. Proc Natl Acad Sci USA 2000, 97:9335–9339.
Hay SI, Rogers DJ, Shanks GD, Myers MF, Snow RW: Malaria early warning
in Kenya. Trends Parasitol 2001, 17:95–99.
Clements AC, Barnett AG, Cheng ZW, Snow RW, Zhou HN: Space-time
variation of malaria incidence in Yunnan province. China Malar J 2009,
Guerra CA, Gikandi PW, Tatem AJ, Noor AM, Smith DL, Hay SI, Snow RW:
The limits and intensity of Plasmodium falciparum transmission:
implications for malaria control and elimination worldwide. PLoS Med
Kleinschmidt I, Omumbo J, Briet O, van de Giesen N, Sogoba N, Mensah NK,
Windmeijer P, Moussa M, Teuscher T: An empirical malaria distribution
map for West Africa. Trop Med Int Health 2001, 6:779–786.
Gemperli A, Vounatsou P, Sogoba N, Smith T: Malaria mapping using
transmission models: application to survey data from Mali. Am J
Epidemiol 2006, 163:289–297.
Kazembe LN, Kleinschmidt I, Holtz TH, Sharp BL: Spatial analysis and
mapping of malaria risk in Malawi using point-referenced prevalence of
infection data. Int J Health Geogr 2006, 5:41.
Noor AM, Clements AC, Gething PW, Moloney G, Borle M, Shewchuk T, Hay
SI, Snow RW: Spatial prediction of Plasmodium falciparum prevalence in
Somalia. Malar J 2008, 7:159.
Noor AM, Gething PW, Alegana VA, Patil AP, Hay SI, Muchiri E, Juma E,
Snow RW: The risks of malaria infection in Kenya in 2009. BMC Infect Dis
Gosoniu L, Vounatsou P, Sogoba N, Maire N, Smith T: Mapping malaria risk
in West Africa using a Bayesian non-parametric non-stationary model.
Comput Stat Data Anal 2009, 53:3358–3371.
Reid H, Vallely A, Taleo G, Tatem AJ, Kelly G, Riley I, Harris I, Henri I, Iamaher
S, Clements AC: Baseline spatial distribution of malaria prior to an
elimination programme in Vanuatu. Malar J 2010, 9:150.
Reid H, Haque U, Clements AC, Tatem AJ, Vallely A, Ahmed SM, Islam A,
Haque R: Mapping malaria risk in Bangladesh using Bayesian
geostatistical models. Am J Trop Med Hyg 2010, 83:861–867.
Kleinschmidt I, Sharp B, Mueller I, Vounatsou P: Rise in malaria incidence
rates in South Africa: a small-area spatial analysis of variation in time
trends. Am J Epidemiol 2002, 155:257–264.
Nobre AA, Schmidt AM, Lopes HF: Spatio-temporal models for mapping
the incidence of malaria in Pará. Environmetrics 2005, 16:291–304.
Mabaso ML, Vounatsou P, Midzi S, Da Silva J, Smith T: Spatio-temporal
analysis of the role of climate in inter-annual variation of malaria
incidence in Zimbabwe. Int J Health Geogr 2006, 5:20.
Anon: Malaria situation in Bangladesh 2008. Malaria Situation in SEAR
Countries: World Health Organization; 2009.
Anon: Monthly malaria diseases specific report. In. Dhaka. Bangladesh:
Directorate General of Health Services; 2007.
Anon G: Malaria disease specific report. Dhaka. Bangladesh: National Malaria
Control Programme, Directorate General of Health Services; 2010.
Anon: World Population Prospects: The 2010 Revision, Volume 1:
Comprehensive Tables. New York: United Nations Department of Economic
and Social Affairs Population Division; 2010:2011.
Haque U, Ahmed SM, Hossain S, Huda M, Hossain A, Alam MS, Mondal D,
Khan WA, Khalequzzaman M, Haque R: Malaria prevalence in endemic
districts of Bangladesh. PLoS One 2009, 4:e6737.
Molineaux L: The epidemiology of human malaria as an explanation of
its distribution, including some implications for its control. In Malaria:
Principles and Practice of Malariology. vol. 2nd edition. Edited by Wernsdorfer
WH, McGregor I. London: Churchill Livingstone; 1988:913–998.
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Garvis A: Very high resolution
interpolated climate surfaces for global land areas. Int J Climatol 2005,
Bontemps S, Defourny P, Van Bogaert E: Globcover 2009 product description
and validation report. Louvain: European Space Agency and Université
Catholique de Louvain; 2010.
Di Gregorio A, Jansen LJM: Land Cover Classification System (LCSS). Rome:
Food and Agriculture Organization; 2000.
Besag J, York J, Mollié A: Bayesian image restoration, with two
applications in spatial statistics. Ann Inst Stat Math 1991, 43:1–59.
40.Abeku TA, van Oortmarssen GJ, Borsboom G, de Vlas SJ, Habbema JD:
Spatial and temporal variations of malaria epidemic risk in Ethiopia:
factors involved and implications. Acta Trop 2003, 87:331–340.
Chowell G, Munayco CV, Escalante AA, McKenzie FE: The spatial and
temporal patterns of falciparum and vivax malaria in Peru: 1994–2006.
Malar J 2009, 8:142.
Gething PW, Smith DL, Patil AP, Tatem AJ, Snow RW, Hay SI: Climate
change and the global malaria recession. Nature 2010, 465:342–345.
Bui HM, Clements AC, Nguyen QT, Nguyen MH, Le XH, Hay SI, Tran TH,
Wertheim HF, Snow RW, Horby P: Social and environmental determinants
of malaria in space and time in Viet Nam. Int J Parasitol 2010, 41:109–116.
Childs DZ, Cattadori IM, Suwonkerd W, Prajakwong S, Boots M:
Spatiotemporal patterns of malaria incidence in northern Thailand. Trans
R Soc Trop Med Hyg 2006, 100:623–631.
Rosenberg R: Forest malaria in Bangladesh. III. Breeding habits of
anopheles dirus. Am J Trop Med Hyg 1982, 31:192–201.
Alam MS, Khan MG, Chaudhury N, Deloer S, Nazib F, Bangali AM, Haque R:
Prevalence of anopheline species and their Plasmodium infection status
in epidemic-prone border areas of Bangladesh. Malar J 2010, 9:15.
Haghdoost AA, Alexander N: Systematic review and meta-analysis of the
interaction between Plasmodium falciparum and Plasmodium vivax in
humans. J Vector Borne Dis 2007, 44:33–43.
Anon: Revised malaria treatment regimen. In. Dhaka. Bangladesh:
Directorate General of Health Services Ministry of Health and Family
Nkhoma ET, Poole C, Vannappagari V, Hall SA, Beutler E: The global
prevalence of glucose-6-phosphate dehydrogenase deficiency: a
systematic review and meta-analysis. Blood Cells Mol Dis 2009, 42:267–278.
Papiha SS, Roberts DF, Ali SG, Islam MM: Some hereditary blood factors of
the Bengali Muslim of Bangladesh (red cell enzymes, haemoglobins, and
serum proteins). Humangenetik 1975, 28:285–293.
Khan WA, Sack DA, Ahmed S, Prue CS, Alam MS, Haque R, Khyang J, Ram M,
Akter J, Nyunt MM, Norris D, Glass G, Shields T, Haq MZ, Cravioto A, Sullivan
DJ Jr: Mapping hypoendemic, seasonal malaria in rural Bandarban,
Bangladesh: a prospective surveillance. Malar J 2011, 10:124.
Patil AP, Okiro EA, Gething PW, Guerra CA, Sharma SK, Snow RW, Hay SI:
Defining the relationship between Plasmodium falciparum parasite rate
and clinical disease: statistical models for disease burden estimation.
Malar J 2009, 8:186.
Cite this article as: Reid et al.: Characterizing the spatial and temporal
variation of malaria incidence in Bangladesh, 2007. Malaria Journal 2012
Submit your next manuscript to BioMed Central
and take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
Reid et al. Malaria Journal 2012, 11:170
Page 8 of 8