Examining Predictive Malaria Dynamics Factors at the Local Level

Measures to control malaria have created significant progress in addressing disease spread across many countries in the past few decades. Between 2001 and 2013, there has been a 47% decrease in malaria mortality rates globally, but despite this improvement, countries with limited healthcare resources and access often struggle with malaria control and surveillance. A majority of the burden from malaria is in rural areas of low-income countries, where healthcare access is most limited, and climate change and population growth are expected to increase the number of people at risk for malaria in these areas. A main goal of the World Health Organization’s strategic plan to address malaria is to improve global access to malaria surveillance testing and treatment. Surveillance measures are vital to reach this goal, but limited resources make these difficult to implement in low-income countries. Most malaria infections occur in these areas with little access to healthcare, and a majority of surveillance measures only capture cases reported at health facilities, which usually are only located in large cities. 

CEID’s Michelle Evans and a research team investigated how malaria risk varies over space and time, and whether there are predictable factors (socio-economic, climatic, and environmental) that may influence this variability. With the majority of research predicting malaria occurs on a global or regional scale, this study hopes to instead provide a framework to inform malaria control at the local level, where control measures are typically implemented.

This research took place in Ifanadiana, a rural district in southeastern Madagascar. Ifanadiana has a population of about 200,000 people, with 195 administrative village units. Data from January 2014 to December 2017 was utilized, and a total of 326,334 cases of malaria were estimated to occur during this time period, with significant variation in cases across space and time. Measures were adjusted to account for missing data from each administrative village unit. There were an average of 42 cases per 1,000 people per month, but Malaria incidence was seasonal, with higher incidence occurring from November to May (36-92% per month on average) and lower incidence from June to October (7-22% per month on average). The months of November to May were also associated with high temperatures and high levels of precipitation, and there were no significant differences in malaria incidence associated with using bed nets or forest loss. Certain land-use variables were also found to be an important predictor of malaria transmission in Infanadiana. The proportion of rice fields and irrigated agriculture in an area are also associated with higher malaria incidence in this model with a lower malaria incidence associated with a larger proportion of residential areas. 

Though the data was adjusted to account for biases in access to care due to financial status or geographic location, an association was still found with average household wealth and distance from primary health centers. There were lower incidence rates in village units farther from a primary health center and higher incidence rates in wealthier village units, but this is likely reflective of disparities in access to care and therefore, a malaria diagnosis, rather than an actual difference in incidence. With many communities being located between two and six hours away from the nearest primary health center, lack of access to care can prevent diagnosis.

This study provides a framework to predict malaria over small spatial scales. Future research can build on this information to create data driven malaria forecasting models in areas with high rates of transmission. 

Click here to learn more. 

By: Brenna Daly