Determinants of MDA impact and designing MDAs towards malaria elimination

  1. Bo Gao
  2. Sompob Saralamba
  3. Yoel Lubell
  4. Lisa J White
  5. Arjen M Dondorp
  6. Ricardo Aguas  Is a corresponding author
  1. Centre for Tropical Medicine and Global Health, Nuffield Department of Medicine, University of Oxford, United Kingdom
  2. Mahidol-Oxford Tropical Medicine Research Unit, Faculty of Tropical Medicine, Mahidol University, Thailand
6 figures, 2 tables and 1 additional file

Figures

Figure 1 with 6 supplements
Model structure illustration.

(A) Flow diagram representing the natural history of Plasmodium falciparum infections in human populations. Uninfected individuals (S) can be infected at rate λ, with the probability of developing …

Figure 1—figure supplement 1
Synthetic population demographics.

Depicts the age distribution of the simulated individuals.

Figure 1—figure supplement 2
Model fit to immunological data.

Displays the simultaneous fit to 4 measured relationships between age and clinical immunity. The black dots refer to the data points offering a measurement for clinical immunity in each of these …

Figure 1—figure supplement 3
Calibrated relationship between EIR and Data.

The black dots depict the data points in the Guerra et al. (2007) database. The colored dots show 100 thousand individual model predictions for the observable prevalence in a village (aggregated …

Figure 1—figure supplement 4
Calibrated relationship between prevalence and clinical malaria incidence.

Compares the predicted clinical malaria case incidence for varying levels of prevalence (red), with the data (black) published in Patil et al. (2009). Much like in the previous figure, 100 thousand …

Figure 1—figure supplement 5
Clinical age profiles for different endemic levels.

Shows how the distribution of malaria clinical cases across age bins changes with increasing all-age true prevalence.

Figure 1—figure supplement 6
MDA effect size for different values of starting prevalence.

The black triangles and dots refer to the prevalence measured at baseline and at month 12 post MDA in the MDA trials performed in the Thai-Myanmar border (Landier et al., 2018). We refrained from …

Figure 2 with 5 supplements
Multivariate sensitivity analysis of the predicted intervention impact on malaria prevalence and artemisinin resistance.

The box plots show the median and interquartile ranges of the proportional reduction in malaria prevalence for all simulated parameter sets. Each parameter set consists of a combination of initial …

Figure 2—figure supplement 1
Multivariate model sensitivity analysis independent of transmission heterogeneity.

Model outcome (measured as the proportional decrease in parasite prevalence) sensitivity to logistical (Teams – speed of coverage, and number of MDA campaigns), epidemiological (initial mean …

Figure 2—figure supplement 2
Interplay between population mobility and transmission heterogeneity.

Intervention outcome sensitivity to spatial transmission heterogeneity in settings with different population mobility patterns. The vectorial capacity distributions associated with these simulations …

Figure 2—figure supplement 3
Interplay between logistics and human population topologies.

Model predicted interaction of logistical MDA considerations (speed of coverage – teams) with different population mobility levels (given in black background over each panel) in determining …

Figure 2—figure supplement 4
Factors influencing the spread of artemisinin resistance.

In all similar to Figure 2, it shows how the different model parameters modulate the predicted proportional decrease in artemisinin resistance after MDA. Negative numbers then reflect a rise in …

Figure 2—figure supplement 5
Population movement and mosquito distributions determine artemisinin resistance spread.

Illustrates how human population mobility (y-axis) interacts with how transmission risk is distributed over space (colored groups) to modulate the extent to which artemisinin resistance is able to …

Figure 3 with 3 supplements
Intervention implementation speed in different epidemiological contexts.

Demonstrates under what conditions using 400 implementation teams is preferable over 15 teams when deploying an MDA campaign. We investigate different epidemiological contexts, characterized by …

Figure 3—figure supplement 1
Elimination likelihood over time in different settings.

Illustrates the interaction between human population mobility, seasonality patterns, and MDA operational parameters (number of teams and campaign start time). Colors indicate the proportion of …

Figure 3—figure supplement 2
Relationship between elimination probability and MDA campaign village sequencing.

Shows how the relationship between human population mobility (rows), seasonality patterns (blocks of columns), MDA operational parameters – number of teams (block of rows), campaigns and campaign …

Figure 3—figure supplement 3
Relationship between elimination probability and the relative infectivity of asymptomatic infections.

Illustrates how a lower infectiousness assumption for asymptomatic malaria (red line) drives the likelihood of malaria elimination up for independently of the remaining model parameters.

Figure 4 with 3 supplements
Elimination probability surfaces (Normal distribution of transmission risk over space).

These surface plots show the proportion of simulations (out of 100) in which elimination was achieved within a 5 year time horizon. Hypothetical interventions that decrease the vectorial capacity by …

Figure 4—figure supplement 1
Integrated control elimination surfaces for the Log-normal distribution of transmission risk over space.
Figure 4—figure supplement 2
Integrated control elimination surfaces for the Pareto distribution of transmission risk over space.
Figure 4—figure supplement 3
Mean reduction in prevalence in intervention strategies containing either MDA + VC or VC alone.

These surface plots show a drastic improvement in transmission reduction impact when comparing implementations of MDA in setting with varying vector control (VC), against VC only intervention …

Figure 5 with 1 supplement
Elimination probability with a targeted approach.

Illustrates the likelihood of elimination within 5 years of an elimination strategy consisting of 3 MDA rounds and a vector control strategy sustained over 1 year. We compare elimination prospects …

Figure 5—figure supplement 1
Elimination probability with a targeted approach.

The same as in Figure 5, except that vector control is extended to last 5 years.

Author response image 1

Tables

Table 1
Factors explored by the model and their respective sets of values.
FactorMeaningValues
TeamsNumber of teams performing prevalence surveys and distributing ACTs simultaneously. Translated into coverage speed in square brackets, that is, the number of days it takes to perform one MDA round in a region of 1000 villages.15 [267]
25 [160]
50 [80]
100 [40]
200 [20]
400 [10]
PrevalenceMean initial malaria u-PCR prevalence across all villages.0.5, 0.1, 0.15, 0.2, 0.25, 0.3
ResistanceMean initial proportion of parasites which are artemisinin resistant.0.1, 0.2, 0.3
DistributionTransmission heterogeneity distribution, underlying spatial heterogeneity of malaria transmission, manifested by differential mosquito density distributions.Gaussian, Log-normal, Pareto
MobilityDescribes the intensity of population movement in the general population. Indicates the per person average daily probability of moving to places other than their home village for short-term visits (see Mobility section in the Simulation Protocol). that is a population of individuals whose short-term movement occurs on average 25 times per year (25/365) has a relatively static population with low mobility.5/365
25/365
125/365
250/365
CampaignsThe number of MDA Campaigns1, 2
PeaksThe number of annual seasonal peaks1, 2
Table 2
List of parameters used in the model.
 #NameDescriptionValueReference
 1VSet of villages|V| = 1000
 2NSet of individuals across all villages|N| = 314,795-
 3mlProportion of males in the population0.48(National Institute of Statistics, 2008)
 4maxageMaximum age80 years-
 5betaBiting ratevariable*-
 6previInitial proportion of infectious peoplevariable*-
 7resitInitial proportion of artemisinin resistant infectionsvariable*-
 8netuseProportion of individuals that own an insecticide treated bed net0.5-
 9itneffectproportional decrease of individual susceptibility/infectiousness related to ITN usage0.2-
 10ovstayMean number of nights spent somewhere when undertaking short-term population movement3-
 11critCritical distance below which overnight stays somewhere other than your home are made very unlikely4 km-
 12timecompMean time to complete ACT routine treatment4 daysBest guess
 13fullcourseProportion that receives treatment full course0.8(Yeung et al., 2008)
 14covabProportion of symptomatic cases that receive antimalarials0.6(Yeung et al., 2008)
 15nompRelative probability of receiving treatment in a non-malaria post village0.1Best guess
 16asymtreatRelative probability of receiving treatment without clinical symptom10−4-
 17tauabDaily probability of receiving ACT in a village under MDA1/1.5-
 18gammaMean liver stage duration5 days(Collins and Jeffery, 1999; Eyles and Young, 1951)
 19sigmaMean time to infectiousness after liver emergence15 days(Jeffery and Eyles, 1955)
 20mellowMean duration of symptoms3 days(Church et al., 1997)
 21xa0Daily probability of going below the minimum effective artemisinin concentration1/7(Karbwang et al., 1998)
 22xaiDaily probability losing the DHA effect as part of ACT1/3(Rijken et al., 2011Tarning et al., 2008)
 23xabDaily probability of going below the minimum effective piperaquine concentration1/30(Rijken et al., 2011Tarning et al., 2008)
 24xprDaily probability of going below the minimum effective primaquine concentration1/2(Burgess and Bray, 1961)
 25deltaMean duration of a malaria untreated infection160 days(Eyles and Young, 1951; Babiker et al., 1998; Franks et al., 2001)
 26imm_minMinimum clinical immunity period40 daysBest guess
 27alphaAverage permanence in each immunity level60 days-
 28phicRelative infectiousness of symptomatic infections compared to sub-patent ones1-
 29mdiMosquito daily probability of dying while infectious1/7(Dawes et al., 2009)
 30mdnMosquito daily probability of dying while infected but not yet infectious1/20(Dawes et al., 2009)
 31mgammaMean extrinsic incubation period14 days(Smith et al., 2014)
 32ampAmplitude of mosquito density seasonal variation0.6Best guess
 33processDays needed to administer a full ACT course in one village4 daysOptimistic guess
 34roundsNumber of drug rounds in an MDA campaign3Standard practice
 35btroundsNumber of days between drug rounds in an MDA campaign32Standard practice
 36vcefficacyVector control efficacyvariable*-
 37cbr0aDaily probability of clearing blood stage drug sensitive parasites with circulating dha1/5(Adjuik et al., 2004Pukrittayakamee et al., 2004)
 38cbraaDaily probability of clearing blood stage artemisinin resistant parasites with dha0.27*cbr0a
(0.05)
(Dondorp et al., 2009)
 39cir0aDaily probability of clearing infectious stage drug sensitive parasites with circulating dha1/3(Adjuik et al., 2004Pukrittayakamee et al., 2004)
 40ciraaDaily probability of clearing infectious stage artemisinin resistant parasites with dha0.27*cir0a
(0.09)
(Dondorp et al., 2009)
 41cbroabDaily probability of clearing blood stage drug sensitive parasites with circulating dha- piperaquine1/3(Adjuik et al., 2004; Pukrittayakamee et al., 2004)
 42cbraabDaily probability of clearing blood stage artemisinin resistant parasites with dha- piperaquine0.27*cbroab + (1.0–0.27)*cbr0b
(0.33)
(Dondorp et al., 2009)
 43cir0abDaily probability of clearing infectious stage
drug sensitive parasites with circulating dha- piperaquine
1/3(Bustos et al., 2013)
 44ciraabDaily probability of clearing infectious stage artemisinin resistant parasites with dha- piperaquine0.27*cir0ab + (1.0–0.27)*cir0b
(0.126)
-
 45cbr0bDaily probability of clearing blood stage drug sensitive parasites with circulating piperaquine1/3(Chen et al., 1982)
 46cbrabDaily probability of clearing blood stage artemisinin resistant parasites with piperaquine1/3(Chen et al., 1982)
 47cir0bDaily probability of clearing infectious stage drug sensitive parasites with circulating piperaquine1/20(Myint et al., 2007)
 48cirabDaily probability of clearing infectious stage artemisinin resistant parasites with piperaquine1/20(Myint et al., 2007)
 49cir0pDaily probability of clearing infectious stage drug sensitive parasites with primaquine1/1.5(Burgess and Bray, 1961; Smithuis et al., 2010)
 50cirapDaily probability of clearing infectious stage
artemisinin resistant parasites with primaquine
1/1.5-
 51kSteepness of susceptibility increase with age0.14(Aguas et al., 2008)
 52rAmplitude of susceptibility increase with age0.99(Aguas et al., 2008)
  1. *the values are varied in different simulation settings. Their values are given in the description of each set of experiments and the set of possible values is given in Table 1.

Additional files

Download links