Where are the killers? Working with national malaria control programmes for improving malaria mosquitoes’ detection in Africa.
Aims
Using open source data and methods to identify where and when to use mosquito traps to provide accurate distribution and predict the abundance of malaria mosquitoes.
Overview
Malaria is one of the major causes of death in Africa, especially in children under 5 years of age. The World Health Organisation recently highlighted that no significant progress in reducing malaria cases had been made recently, due to the lack of tools for surveillance, collection, and analysis of data. Knowing where mosquito populations are and how they change is essential for malaria control and elimination.
Within the Lancaster Ecology and Epidemiology Group at Lancaster Medical School we are working on developing methods for the optimal identification of the best locations that need to be sampled in order to have accurate descriptions of malaria mosquito distribution, abundance and dynamics (e.g., is their ecology changing? Are they spreading in new environments?).
Dr Luigi Sedda, a spatial epidemiologist that works on mosquito populations dynamics, explains: we were concerned about the lack of representativeness of the spatial distribution and seasonality in the mosquito data coming from Africa. These problems can cause large errors in the analyses of malaria mosquitoes, jeopardising any intervention. To support representative malaria mosquito collections, we have developed an algorithm, OSSA, (Optimal Spatial Sampling Algorithm) in collaboration with local stakeholders and communities in Africa.
Results and Outcomes
Tab Content: For Partners and Engagement
Our approach directly addresses the WHO call to improve surveillance systems for effective intervention. Most of the previous malaria mosquito surveillance programmes were carried out in responsive mode, concentrating the effort in areas identified as having outbreaks of malaria, or opportunistically – locations where it was assumed malaria mosquitoes were present. Previous surveillance methods did not allow description of the distribution of malaria mosquitoes in relation to climate and environment.
We have designed OSSA, an algorithm that harvested open source data from the web (land cover, land temperature, elevation, climate, human and animal population densities etc…) and, by delineating the different epi-ecological zones, returned the optimal number and locations of sites that need to be investigated to describe the distribution, abundance and dynamics of malaria mosquitoes.
OSSA is free since it is based on open source data and statistical tools. This makes the algorithm a valuable option for institutions with limited budgets, as those in low and middle income countries. OSSA adoption in Africa started from pilot regions in five countries (Benin, Côte d'Ivoire, Ghana, Kenya and Tanzania) and now it is used for national malaria mosquito surveillance in Senegal and Mozambique.
The development of OSSA included regular meetings with malaria and health NGOs, national malaria control programmers and malaria epidemiology, entomology and field work scientists, but also with local communities, such as village leads. This allowed tailoring of the structure of the algorithm (e.g. inputs, parameters) and the outputs.
Feedback from field workers and National Malaria Control managers endorsed the easy design of the surveillance plan, the flexibility to use different locations if some of them is deemed not investigable and the rich open data behind it (which facilitates subsequent data analyses as well).
We believe that the success of OSSA so far is due to the involvement of local communities, public health managers and researchers from the incipit of surveillance planning design.
Tab Content: For Academics
Mosquito surveillance is usually carried out with a limited if not without a quantitative assessment of the sampling effort which can results in underpowered or overpowered studies. This affects the thousands of mosquito surveys (not only for malaria) that are carried out each year in tropical countries.
By working with local communities, experts, public health managers and other stakeholders we understood the current challenges faced in malaria control and elimination, and was thanks to their input that a (co-)developed algorithm was produced. Their support was essential for the selection of databases and for the structure and outputs of the algorithm.
There are still many operational questions that national malaria montrol programmes faces every day, such as best routes to reduce travel time, number of repeated measurements in each location and for how long the surveillance is necessary, how often to review the sampling design protocol and what the best goal to name few of them.
Most of these challenges needs to be cost-effective due to the limited resources available and the vast scale of malaria disease.
We, statisticians, are powerless given the amount of complexity in infectious disease systems (COVID19 is an example). Therefore, multi-disciplinary and multi-institutional approaches with participation of local communities in all the stages of a surveillance and control programme development, is essential to defeat this prehistoric parasite that still kills half million people each year.
Back to listing