1. 1
EVALUATING THE DIRECT AND INDIRECT PROTECTIVE EFFECTS OF MALARIA
INTERVENTIONS
By
Autumn B. Bridges
An Honors thesis Submitted to the Department of Biological Sciences
in partial fulfillment of the requirements for the
degree of Bachelor of Arts
Meredith College
Raleigh, North Carolina
April 2016
Honors Student______________________________ Date ____________________
Thesis Director______________________________ Date ____________________
Honors Director______________________________ Date ____________________
2. 2
Publication Agreement
I hereby grant to Meredith College the non-exclusive right to reproduce, and/or distribute this
work in whole or in part worldwide, in any format or medium for non-commercial, academic
purposes only.
Readers of this work have the right to use it for non-commercial, academic purposes as
defined by the "fair use" doctrine of U.S. copyright law, so long as all attributions and
copyright statements are retained.
Meredith College may keep more than one copy of this submission for purposes of security,
backup and preservation.
Autumn B. Bridges
April 27, 2016
Copyright 2016 by Autumn B. Bridges
3. 3
Acknowledgements
I would like to acknowledge and thank my research mentor Dr. Carolina Perez-Heydrich for
her ongoing guidance and assistance throughout this project. This would not have been
possible without her patience and enthusiasm. I would also like to thank my undergraduate
faculty advisor, Dr. Cynthia Edwards, for her guidance and support the last four years, and
Dr. Brent Pitts for his support through the Meredith College Honors Program. Lastly, I would
like to thank the Meredith College Department of Undergraduate Research for allowing me
the opportunity to participate in this research.
4. 4
Abstract
The purpose of this study is to determine if two popular antimalarial interventions,
insecticide-treated bed net (ITN) usage and indoor residual spraying (IRS), confer herd
protection; that is, do these interventions, when used at the community level, provide
protection to those not using them? Malaria is a major problem in sub-Saharan Africa, killing
0.6-1.2 million people each year. Those living in low income countries are most affected by
the disease, and children are especially at risk of infection. The data used in the study was
provided by the Demographic and Health Survey (DHS). A total of seven countries were
examined: Angola, Burundi, Liberia, Madagascar, Malawi, Nigeria, and Tanzania. A crude
analysis of the data was conducted to estimate intervention-specific malaria prevalence,
calculate odds ratios, and identify potential confounding variables for a final regression
model. Maternal education, household wealth index, and urbanity were all associated with
both use of ITN or IRS and malaria parasitemia. A mixed-effects logit model was then fit to
evaluate the effects of ITN use, IRS use, ITN coverage, and IRS coverage on malaria status,
and to calculate adjusted overall odds ratios. The results of the crude analysis showed that in
Burundi (OR: 0.51, 95% CI: 0.43, 0.61) and Malawi (OR: 0.72, 95% CI: 0.59, 0.87) the use
of an ITN had a significant protective effect, while in Tanzania (OR: 1.20, 95% CI: 1.02,
1.41) and Madagascar (OR: 1.26, 95% CI: 1.02, 1.57) it was a significant risk factor. The use
of IRS in Madagascar (OR: 0.36, 95% CI: 0.29, 0.44) significantly decreased the odds of an
individual contracting malaria, while in Tanzania (OR: 1.23, 95% CI: 1.05, 1.43) it
significantly increased the odds. The results of the adjusted analysis showed that there is no
indirect protection offered for non-users of ITNs (OR: 2.19, 95% CI: 1.38, 3.47). As ITN
5. 5
coverage increased, malaria prevalence among non-users also increased. There was,
however, indirect protection offered for users of ITNs (OR: 0.62, 95% CI: 0.41, 0.93); that is,
as ITN coverage within a neighborhood increased, the odds of an ITN user within that
neighborhood contracting malaria decreased. The contradictory nature of these findings
highlight a need to further evaluate mechanisms that may be driving these observed patterns.
6. 6
Introduction
Ecology of Malaria
Mosquitoes, arthropods in the Culicidae family, are responsible for the spread of
numerous infectious diseases, including Dengue fever, West Nile Virus, and yellow fever. A
major disease spread by mosquitoes is malaria, which sickens approximately 200 million and
kills 0.6-1.2 million people per year [1]. Those living in low income countries are most
affected by the disease, and children are especially at risk of infection [1].
Malaria is a vector-borne disease, transmitted by the Anopheles mosquito [2]. There
are several species of Plasmodium, the genus of single-celled eukaryotes that malaria
parasites belong to, Plasmodium falciparum and Plasmodium vivax [1]. While both species
commonly infect humans, P. falciparum is responsible for the majority of human deaths [1].
The malaria pathogen enters humans through the bite of an infected mosquito and
immediately colonizes a liver cell; it then divides many times, eventually rupturing the cell
and producing a new form of the pathogen that infects red blood cells. Once in the red blood
cell, the Plasmodium continues to multiply and produce pathogen forms that infect red blood
cells [1]. During the Plasmodium life cycle, when cells containing male and female gametes
are ingested by Anopheles mosquitoes, mating occurs to form a mobile zygote [1]. The
zygote attaches to the gut wall and penetrates the salivary gland, thereby infecting more
humans or animals at the time of a blood meal [1].
Individual Level Interventions
7. 7
There are numerous factors that have an effect on the risk of individuals of becoming
infected with malaria, including household wealth index, education, age, and number of
household members [3]. Insecticide-treated bed nets (ITNs) and indoor residual spraying
(IRS) are two vector control methods currently used to prevent malaria transmission [4].
ITNs are cost-effective antimalarial interventions that offer a mix of personal protection: they
provide a physical barrier between humans and the malaria vector while diverting mosquitoes
to non-human hosts [5]. There have been several studies that have researched the impact that
ITNs have had on reducing the mortality rate of malaria at the individual level. One study
found that, following an ITN distribution effort in Uganda, the percentage of households
owning at least one ITN increased from 22 percent to 69 percent and the ITN ownership of
eligible households was 84 percent, close to the distribution goal of 90 percent [6].
Indoor residual spraying is the application of long-lasting insecticides on the walls
and roofs of households within a community [7]. IRS kills mosquitos in the malaria vector
that land on sprayed surfaces; it also reduces the life span of vector mosquitos so they can no
longer transmit malaria parasites from person to person [7].
Community Level Interventions
In addition to documenting individual-level benefits of ITNs and IRS, studies have
also addressed the indirect protection conferred by these interventions within a community.
Bayoh et al measured the changes in relative abundance of two primary malaria vectors in
two adjacent communities in Kenya with a large disparity in ITN ownership [8]. This study
found that the density of Anopheles mosquitos, and proportion of habitats within 200m of all
8. 8
housing compounds containing Anopheles larvae, were lower in the community with nearly
100% ITN ownership relative to the community with < 5% ITN coverage [8].
A review of the effects of ITN use and IRS on the mortality of malaria found that in
areas with a high coverage of ITNs, the number of lives saved per ITN was high:
approximately 5.5 lives saved per 1000 children protected per year [9]. The results presented
in this review were from highly controlled conditions, leading to high coverage and use rates
of both ITN and IRS [9].
A systematic review of numerous studies, including randomized controlled trials,
quasi-experimental trials, and evaluation studies, suggested that the delivery of community-
based interventions increased ITN usage and ownership [10]. The review proposes that the
community distribution of ITNs and IRS effectively reduced malaria prevalence, parasitemia,
anemia prevalence, and all-cause mortality [10].
Children are especially at risk of contracting malaria [1]; therefore educating parents
about how to prevent and manage the disease is crucial in reducing child mortality [11].
Research conducted in Ghana examined the community effects of health education through
various platforms, including social media, formal education, and mass media determined that
participants with higher education and literacy achievement were more likely to have higher
levels of health education [11]. Additionally, literacy at the community level impacts
individuals’ health education knowledge: living in a community with high levels of literacy
can positively impact a person’s health education knowledge, even if they themselves are
illiterate [11].
An additional consideration is the use of both ITNs and IRS within the same
household; while many households use both interventions to further decrease malaria
9. 9
transmission, there is limited evidence to suggest that co-application provides greater
protection than either intervention alone [12]. A review of possible outcomes of co-
application found that there is no clear added advantage to using both interventions together
[12]. Co-application is suggested for hastening vector control in high transmission areas
where ITN or IRS use alone would not be effective [12]. The use of both interventions can
also be effective in households when one intervention is weakened; for example, using ITNs
when previously applied IRS has begun to decay [12].
The purpose of the present study is to determine if community-wide use of ITNs and
IRS provide indirect protection with respect to malaria interventions. Indirect protection is
defined as the difference in disease rates among individuals from neighborhoods that vary
with respect to community-wide use of interventions. If ITN and IRS use confer indirect
protection via a neighborhood-wide reduction in vector density, then malaria rates are
expected to decline with increasing coverage of these interventions across neighborhoods. A
cross-sectional study was conducted using population-representative data from seven
different countries to evaluate the association between community use of interventions and
occurrence of malaria among children under five. Statistical methodology and results are
described in subsequent sections.
10. 10
Methods
Data Sources
Analyses included datasets provided by the Standard Demographic and Health Survey
(DHS). For consistency, only countries with GPS data and complete phase DHS-VI data
were included. A total of seven countries were included: Angola, Burundi, Liberia,
Madagascar, Malawi, Nigeria, and Tanzania. The analysis was restricted to children under
five years of age, and a total of 33,845 observations were used in the final analysis.
The primary outcome of interest was malaria infection status, as determined by rapid
diagnostic and blood smear tests administered by the DHS Program. A rapid diagnostic test
detects parasite antigens [13]. The test requires only a drop of blood from a finger or heel
prick, and results are available within fifteen minutes [13]. Blood smear tests are a form of
microscopy tests, and there are two types of smears that can be taken: thick smears and thin
smears [13]. Thick smears are used to identify the presence or absence of a malaria parasite,
as well as the type of Plasmodium species; thin smears are used to quantify the malaria
parasites [13]. Rapid diagnostic tests typically report a higher prevalence of malaria than
blood smear tests; as such, they tend to be more reliable [13]. A previous study evaluating the
diagnostic performances of rapid and blood smear tests found that rapid tests had a sensitivity
of 97%, compared to 85% for blood smear tests [14]. The outcome variable of malaria for
this study was defined using both measures of infection status: A person was classified as
having malaria if they tested positive for either the blood smear or rapid test.
The primary predictor variables for this study relate to the two primary antimalarial
interventions: individual use of insecticide-treated bed nets (ITNs) and household indoor
11. 11
residual spraying (IRS). When constructing the ITN variable, there were several missing
responses. It was determined that these responses matched to participants who did not sleep
under an ITN. These responses were changed to “no”. Participants who answered “don’t
know” to whether they slept under an ITN or used IRS were also changed to “no”. A
preliminary analysis was conducted to determine if there were confounding variables that
needed to be accounted for in the final regression model. These potential confounders were
maternal education, wealth index, and urbanity.
Because of the clustered nature of the data, a mixed-effects logit model was fit to the
data to evaluate the effects of predictors on the outcome variable individual malaria status.
The predictor variables included in the model were ITN use, IRS use, ITN coverage, and IRS
coverage. ITN coverage for an individual j was defined as the percentage of people within
individual j’s household cluster who used an ITN, excluding individual j. Mathematically,
this is represented as∑ 𝐼(𝑥𝑖 = 1)/(𝑛 − 1)
𝑛−1
𝑖=1,𝑖≠𝑗
, where 𝐼(𝑥𝑖 = 1) indicates whether
individual i used intervention x, and n represents the number of individuals in a cluster. IRS
coverage for a household was defined analogously. The confounders that were controlled for
included urbanity, wealth index, maternal education, annual malaria prevalence, and country.
Annual malaria prevalence for each cluster was obtained from estimates provided by the
Malaria Atlas Project [15]. Adjusted overall odds ratios were calculated to quantify the effect
of ITN use, ITN coverage, IRS use, and IRS coverage on individual malaria status.
12. 12
Results
Crude Analysis
For the preliminary analysis, numerous χ² tests were conducted to identify
confounding variables for the final regression model. Ultimately, maternal education,
urbanity, wealth index, and annual malaria prevalence for each country were all controlled
for in the regression model. The results of the chi-square tests are shown in Table 1.
Table 2 shows the crude odds ratios and 95% confidence intervals for the association
between malaria status, based on the combined results of blood smear and rapid tests, and use
of ITNs and IRS. Across all countries, with the exception of Burundi, Liberia, and Malawi,
ITN use was associated with increased odds of contracting malaria. The use of ITNs offered
a significant protective effect in Burundi (OR: 0.51, 95% CI: 0.43, 0.61) and Malawi (OR:
0.72, 95% CI: 0.59, 0.87), whereas it was a significant risk factor in Tanzania (OR: 1.20,
95% CI: 1.02, 1.41) and Madagascar (OR: 1.26, 95% CI: 1.02, 1.57). Similar results were
found for the use of IRS: Burundi, Madagascar, and Malawi were the only countries where
IRS had a protective effect. In Madagascar the use of IRS significantly reduced the odds of
an individual contracting malaria (OR: 0.36, 95% CI: 0.29, 0.44), and in Tanzania the use
significantly increased the odds (OR: 1.23, 95% CI: 1.05, 1.43).
The malaria prevalence of each country is shown in Figure 1. Figure 2 shows the
prevalence of malaria stratified according to ITN use. Figure 3 shows the prevalence of
malaria stratified according to IRS use.
Adjusted Analysis
13. 13
Table 2 shows the overall adjusted odds ratios for IRS use, IRS coverage, ITN use,
and ITN coverage, adjusting for maternal education, wealth index, urbanity, and annual
malaria prevalence. As cluster-level ITN coverage increased, the prevalence of malaria
decreased; non-users of ITNs always had higher prevalence of malaria than users (Figure 4).
These results do not coincide with what was observed with IRS coverage, where there was
more variability across cluster-level IRS coverage (Figure 5).
The results of the adjusted analysis showed that there is no indirect protection offered
for non-users of ITNs. As ITN coverage increased, malaria prevalence among non-users also
increased (OR: 2.19, 95% CI: 1.38, 3.47). There was, however, indirect protection offered for
users of ITNs; that is, as ITN coverage within a neighborhood increased, the odds of an ITN
user within that neighborhood contracting malaria decreased (OR: 0.62, 95% CI: 0.41, 0.93).
IRS use did not provide protection at either the individual or community-level against
malaria (Table 3).
14. 14
Discussion
The findings from this study provide important insight on how community level
interventions, ITN and IRS use, affect the individual. The initial crude analysis showed that,
unexpectedly, sleeping under an ITN increased the odds of contracting malaria in Angola,
Madagascar, Nigeria, and Tanzania; however, after adjusting for confounders, ITNs had an
indirect protective effect for ITN users in a high coverage area. In Angola, Liberia, Nigeria,
and Tanzania, IRS was also a risk factor for malaria. Surprisingly, in Angola, Nigeria, and
Tanzania the use of both interventions increased the odds of an individual contracting
malaria, although not significantly. This could be due in part to Tanzania possessing the
majority of Lake Tanganyika, the second largest freshwater lake in the world. Anopheles
mosquitos lay their eggs directly on stagnant water, and the presence of the lake increases the
amount of breeding grounds. This may lead to an increase of mosquito reproduction rates. In
theory, adjusting for baseline malaria prevalence should have controlled for this potential
confounding factor.
ITN use, IRS use, and IRS coverage all offered protective effects, although odds
ratios associated with these measures were not statistically significant. ITN coverage,
however, was a significant risk factor. This implies that, on average and with all confounders
held constant, as ITN coverage increases the individual odds of contracting malaria also
increase. It is important to note the significant interaction between ITN use and ITN
coverage. This suggests that if an individual is using an ITN and the overall ITN coverage
increases, the individual’s odds of contracting malaria decreases.
15. 15
It was surprising to find that cluster-level ITN coverage was associated with increased
odds of malaria among ITN non-users. It is possible that if an unprotected household was in a
neighborhood with high intervention coverage the individuals in the unprotected household
would be more likely to contract malaria due to a shift in the location of the mosquito
population. Previous studies evaluating the behaviors of the Anopheles population found that
mosquitos engage in contact avoidance and exit behaviors that decrease their exposure to
insecticides, including their contact with ITNs and IRS [16]. Simulation analyses have also
suggested that these behaviors allow mosquitos to search until they find unprotected hosts to
feed from, or enter houses protected by ITNs and IRS but safely leave [17].
Across all country-specific regression models, ITN and IRS coverage had significant
regression coefficients. It is important to note, however, that only Madagascar and Angola
had negative regression coefficients, for IRS coverage in both countries. This means that in
every country except for Madagascar and Angola, there was a significant increase in malaria
prevalence for changes in ITN and IRS coverage. These results conflict with the findings of
Lengeler [9] which suggested a decrease in child mortality as more people are protected by
ITNs; however, other studies have also found similar non-protective effects of ITN and IRS
use and attributed this to factors associated with differences in mosquito biting behaviors. A
previous study examining the entomological factors of ITN effectiveness in Western
Myanmar found that the malaria vectors engaged in more early evening biting during the dry
cool season, and biting between 5 and 6 PM was considerably higher than other times [18].
These findings suggest that the effectiveness of ITNs may be decreased if the malaria vectors
bite when people are not in or near bed [18].
16. 16
While attempts were made to control for as many confounders as possible,
unmeasured confounders may still have been present, due to the observational nature of this
study. Additionally, as with all survey-based studies, it is possible that there was response
bias in the answers that participants provided with regard to intervention use. The cross-
sectional nature of this study also has inherent limitations. For example, ITN use was defined
according to survey responses related to whether or not an individual used a particular bed
net the night before the interview. This may or may not have been representative of the
individual’s habitual bed net use.
This study evaluated the effectiveness of two popular antimalarial interventions, ITNs
and IRS, with aims to determine whether they could provide indirect protection. Based on
this analysis, indirect protection was only detected among individuals that used ITNs.
Specifically, ITN users in communities with high ITN coverage had significantly reduced
odds of malaria occurrence; however ITN non-users in these same communities had
increased odds. The contradictory nature of these findings highlight a need to further evaluate
mechanisms that may be driving these observed patterns.
17. 17
Table 1 Results of chi-square test of association for potential confounding variables.
Country Variable Malaria ITN Use IRS Use
Angola MatEd p < 0.05* p < 0.05* p = 0.0581
Wealth p < 0.05* p < 0.05* p < 0.05*
Urbanity p < 0.05* p < 0.05* p = 0.7227
Burundi MatEd p < 0.05* p < 0.05* p = 0.0048*
Wealth p < 0.05* p < 0.05* p < 0.05*
Urbanity p < 0.05* p < 0.05* p = 0.2017
Liberia MatEd p < 0.05* p = 0.0735 p = 0.2478
Wealth p < 0.05* p = 0.0069* p = 0.0232*
Urbanity p < 0.05* p = 0.6289 p = 0.1831
Madagascar MatEd p < 0.05* p < 0.05* p = 0.0092*
Wealth p < 0.05* p < 0.05* p < 0.05*
Urbanity p < 0.05* p = 0.0478* p < 0.05*
Malawi MatEd p < 0.05* p = 0.0004* p = 0.0965
Wealth p < 0.05* p = 0.0296* p < 0.05*
Urbanity p < 0.05* p = 0.6413 p < 0.05*
Nigeria MatEd p < 0.05* p < 0.05* p = 0.0581
Wealth p < 0.05* p < 0.05* p < 0.05*
Urbanity p < 0.05* p < 0.05* p = 0.7227
Tanzania MatEd p < 0.05* p < 0.05* p < 0.05*
Wealth p < 0.05* p < 0.05* p < 0.05*
Urbanity p < 0.05* p = 0.2856 p = 0.1549
χ² df for tests associated with maternal education, wealth, and urbanity were 3, 4, and 1,
respectively.
18. 18
Table 2 Odds ratios and 95% confidence intervals for each country by intervention type
based on both malaria rapid test and blood smear test results.
Country Intervention OR 95% Confidence Interval
Angola ITN 1.08 (0.96, 1.22)
IRS 1.07 (0.61, 1.91)
Burundi ITN 0.51 (0.43, 0.61)*
IRS 0.93 (0.59, 1.40)
Liberia ITN 0.94 (0.82, 1.09)
IRS 1.14 (0.91, 1.44)
Madagascar ITN 1.26 (1.02, 1.57)*
IRS 0.36 (0.29, 0.44)*
Malawi ITN 0.72 (0.59, 0.87)*
IRS 0.94 (0.66, 1.33)
Nigeria ITN 1.08 (0.96, 1.22)
IRS 1.07 (0.61, 1.91)
Tanzania ITN 1.20 (1.02, 1.41)*
IRS 1.23 (1.05, 1.43)*
19. 19
Table 3 Adjusted odds ratios and 95% confidence intervals for interventions and intervention
coverages, adjusted for maternal education, wealth index, urbanity, and annual malaria
prevalence.
Intervention Adjusted OR 95% Confidence Interval
ITN Use 0.93 (0.78, 1.12)
ITN Coverage 2.19 (1.38, 3.47)*
IRS Use 0.75 (0.48, 1.17)
IRS Coverage 0.76 (0.46, 1.25)
ITN Use x ITN Coverage 0.62 (0.41, 0.93)*
IRS Use x IRS Coverage 1.40 (0.65, 2.98)
21. 21
Figure 2 Use of insecticide-treated bed nets and malaria prevalence by country. Angola (n=5153), Burundi (n=3698), Liberia
(n=3192), Madagascar (n=6838), Malawi (n=2105), Nigeria (n=5153), Tanzania (n=7706).
22. 22
Figure 3 Use of indoor residual spraying and malaria prevalence by country. Angola (n=5153), Burundi (n=3698), Liberia (n=3192),
Madagascar (n=6838), Malawi (n=2105), Nigeria (n=5153), Tanzania (n=7706).
23. 23
Figure 4 Malaria prevalence as a function of ITN use and coverage. Malaria prevalence declines
as ITN coverage increases.
24. 24
Figure 5 Malaria prevalence as a function of IRS use and coverage. No protective trends were
observed for IRS coverage.
25. 25
Literature Cited
1. Godfray HCJ: Mosquito ecology and control of malaria. Journal of Animal Ecology
2013, 82:15-25.
2. Vector-borne diseases [http://www.who.int/mediacentre/factsheets/fs387/en/]
3. Messina JP, Taylor SM, Meshnick SR, Linke AM, Tshefu AK, Atua B, Mwandagalirwa
K, Emch M: Population, behavioural and environmental drivers of malaria
prevalence in the Democratic Republic of Congo. Malaria Journal 2011, 10.
4. Fullman N, Burstein R, Lim SS, Medlin C, Gakidou E: Nets, spray or both? The
effectiveness of insecticide-treated nets and indoor residual spraying in reducing
malaria morbidity and child mortality in sub-Saharan Africa. Malaria Journal 2013,
12.
5. Birget PLG, Koella JC: An Epidemiological Model of the Effects of Insecticide-
Treated Bed Nets on Malaria Transmission. PloS one 2015, 10:e0144173-e0144173.
6. Wanzira H, Yeka A, Kigozi R, Rubahika D, Nasr S, Sserwanga A, Kamya M, Filler S,
Dorsey G, Steinhardt L: Long-lasting insecticide-treated bed net ownership and use
among children under five years of age following a targeted distribution in central
Uganda. Malaria Journal 2014, 13.
7. Indoor residual spraying. World Health Organization; 2006.
8. Bayoh MN, Mathias DK, Odiere MR, Mutuku FM, Kamau L, Gimnig JE, Vulule JM,
Hawley WA, Hamel MJ, Walker ED: Anopheles gambiae: historical population
decline associated with regional distribution of insecticide-treated bed nets in
western Nyanza Province, Kenya. Malaria Journal 2010, 9.
9. Lengeler C: Insecticide-treated bed nets and curtains for preventing malaria.
Cochrane Database of Systematic Reviews 2004.
10. Salam R, Das J, Lassi Z, Bhutta Z: Impact of community-based interventions for the
prevention and control of malaria on intervention coverage and health outcomes for
the prevention and control of malaria. Infectious Diseases of Poverty 2014, 3.
11. Andrzejewski CS, Reed HE, White MJ: Does where you live influence what you
know? Community effects on health knowledge in Ghana. Health & Place 2009,
15:228-238.
12. Okumu FO, Moore SJ: Combining indoor residual spraying and insecticide-treated
nets for malaria control in Africa: a review of possible outcomes and an outline of
suggestions for the future. Malaria Journal 2011, 10.
13. Florey L: Measures of Malaria Parasitemia Prevalence in National Surveys:
Agreement between Rapid Diagnostic Testing and Microscopy. vol. 43. Rockville,
Maryland, USA: DHS Analytical Studies; 2014.
14. Stauffer WM, Cartwright CP, Olson DA, Juni BA, Taylor CM, Bowers SH, Hanson KL,
Rosenblatt JE, Boulware DR: Diagnostic Performance of Rapid Diagnostic Tests
versus Blood Smears for Malaria in US Clinical Practice. Clinical Infectious Diseases
2009, 49:908-913.
15. Malaria Atlas Project [http://www.map.ox.ac.uk/map/]
16. Killeen GF: Characterizing, controlling and eliminating residual malaria
transmission. Malaria Journal 2014, 13.
26. 26
17. Killeen GF, Chitnis N: Potential causes and consequences of behavioural resilience
and resistance in malaria vector populations: a mathematical modelling analysis.
Malaria Journal 2014, 13.
18. Smithuis FM, Kyaw MK, Phe UO, van der Broek I, Katterman N, Rogers C, Almeida P,
Kager PA, Stepniewska K, Lubell Y, et al: Entomological determinants of insecticide-
treated bed net effectiveness in Western Myanmar. Malaria Journal 2013, 12.