MINT-data-v20230208

A public repository for the model simulation data used in MINT, the malaria intervention tool, found here.

This dataset contains 2,540,160 model simulations from a malaria transmission dynamics model varying mosquito bionomics, malaria prevalence, transmission seasonality, history of insecticide-treated nets and indoor residual spraying, and projected vector control coverage. The dataset was created by running the model, varying parameters to represent the diversity of settings in malaria-endemic sub-Saharan Africa, and the projections resulting from these simulations are incorporated into Version 2 of the online malaria decision-making tool MINT.

Full details of this update to MINT can be found in the article "Model projections of the epidemiological benefit of pyrethroid-pyrrole insecticide treated nets against malaria" (Churcher et al. 2024). An abstract is found below.

Background. Insecticide treated nets (ITNs) are the most important malaria prevention in Africa but the rise of pyrethroid resistance in mosquitoes is likely impeding control. The World Health Organisation have recommended a novel pyrethroid-pyrrole ITN following evidence of epidemiological benefit in cluster-randomised control trials (cRCTs) in Tanzania and Benin. It remains unclear how effective these more costly pyrethroid-pyrrole ITNs are compared to other tools and there is a need for guidance on use when budgets are limited. Methods. Entomological data are collated from across Africa to inform a malaria transmission dynamics model which is validated against cRCT results. The full impact of new ITNs is quantified for trial sites and simulations for different settings are included within a user-friendly interface MINT (mint.dide.ic.ac.uk) to allow National Malaria Programmes (NMPs) to explore local impact and how budgets should be allocated across regions to avert most cases. Findings. The model projects that distributing pyrethroid-pyrrole ITNs averted 65% of cases over three-years in Tanzania, and 60% in Benin. It indicates that both cRCTs underestimated the benefit of switching to pyrethroid-pyrrole ITNs by at least 6-12%, as participants stopped using allocated nets. In non-trial settings pyrethroid-pyrrole ITNs are projected to reduce malaria prevalence by 20-65% over three-years and that switching from pyrethroid-only ITNs could avert an additional 10-30% more cases depending on where they were deployed. Interpretation. Pyrethroid-pyrrole ITNs are likely to be the most cost-effective indoor vector control intervention in Africa and NMPs can strategise deployment to maximise impact where budgets are restricted. 

Funding. Funding from the Bill & Melinda Gates Foundation (OPP1200155), Innovative Vector Control Consortium through the New Nets Project funded by Unitaid, and a UKRI Future Leaders Fellowship to ES-S from the Medical Research Council (MR/T041986/1). GSC was supported by the Wellcome Trust (220900/Z/20/Z) and TSC, IS, AH, DPD, JC, PW, GC, ER, RF and ES-S received support from the MRC Centre for Global Infectious Disease Analysis (MR/R015600/1), jointly funded by the UK Medical Research Council (MRC) and the UK Foreign, Commonwealth & Development Office (FCDO), under the MRC/FCDO Concordat agreement, the EDCTP2 programme supported by the European Union, and Community Jameel.

Other data and code that have supported the development of MINT can be found here. See below for a description of the data files, which can be downloaded below.

The .rds files contained here, for use in the R programming language, are structured as follows.

The index.rds file contains 5040 unique combinations of baseline characteristics for a region, used to parameterise and calibrate a region that is seen in MINT, with each one allocated an identifier ("index") from 5041 to 10080. The column headings indicate the baseline malaria prevalence ("currentPrevalence"), mosquito behaviours ("bitingPeople", indicating if the human blood index is high or low, and "bitingIndoors", whether mosquito preference for biting indoors is high or low), malaria seasonality ("seasonalityOfTransmission"), pyrethroid resistance ("levelOfResistance") and baseline vector control coverage ("itnUsage" and "sprayInput").

The "table" and "prevalence" folders contain summary outputs from model simulations, with files within both folders named according to the index identifier from the index.rds file.

Each file in the "table" folder contains a number of model simulations, one per row, all of which use the baseline characteristics associated with that index identifier. What varies between rows is the intervention(s) choice and the coverage of that/these intervention(s). The columns identify the expected initial coverage of the intervention after implementation ("netUse" and "irsUse"), the choice of intervention or interventions ("intervention"), the average model-predicted annual prevalence for each of three years post-intervention ("prevYear1", "prevYear2" and "prevYear3"), the mean number of cases per persons over three years ("meanCases"), the absolute difference in cases over three years for this model simulation compared to a simulation with the same index but no intervention ("casesAverted" [per person] and "casesAvertedPer1000" [per 1,000 persons]) as well as the equivalent relative reduction in cases and prevalence ("reductionInCases" and "reductionInPrevalence"). All outputs also have a higher ("ErrorPlus") and lower ("ErrorMinus") uncertainty level, in separate columns. For this version of MINT (V2), these were calculated by running three simulations for each scenario; the mean of these three results was used for the primary outputs in the columns mentioned above while +/-10% of this value forms the basis of the ErrorPlus and ErrorMinus.

The "prevalence" folder contains the data used to make the prevalence graph in the MINT impact tab. Each file, again named by index identifier, contains monthly simulated prevalence data for each intervention(s) ("intervention" column) and intervention(s) coverage ("netUse" and "irsUse"). Here, "month" runs from 13 months prior to intervention start until 47 months after intervention, and "value" is the model-predicted prevalence for that month.