Historical trends reveal a pathway for predictive analysis

5 March 2014

A new model supports public health experts to carry out surveillance of antimalarial use before drug resistance emerges or spreads to Africa

Currently over half of the world’s population is at risk of malaria infection, whilst more than 600,000 people die from malaria each year, most of whom are children living in Africa. Once resistance to certain antimalarial drugs emerges, newer drugs are often sought as a replacement and the likelihood of resistance emerges. This occurred in the 1950s with chloroquine (CQ) and more recently in Southeast Asia as highly effective artemisinin combination therapies, or ACTs, have been challenged in areas along the Thai, Myanmese and Cambodian borders. Given this emerging resistance, researchers are calling for urgent attention to the threat of ACT resistance spreading or emerging across Africa and compromising the global progress made toward eliminating malaria.

In response to these concerns, the WorldWide Antimalarial Resistance Network (WWARN) is developing a series of mathematical modelling techniques that analyse how government policy changes relate to trends in malaria drug usage in forty African countries. If artemisinin resistance does spread to Africa, these new approaches will help to model actual and estimated drug treatment efficacy.

What is the current situation?

The extensive and persistent use of drugs is a known driver of drug resistance. Over the last decade, resistance to first line antimalarials such as sulphadoxine-pyrimethamine (SP) has driven many African health ministries to change their treatment policies to prioritise ACTs. Although increased access to ACTs is confirmed in most countries, use of non-recommended antimalarials like CQ and SP has continued for many years after these policy changes were confirmed.

Why is this important?

The time interval between the emergence of antimalarial resistance and the implementation of policy changes greatly affects malaria disease and mortality rates. A paper recently published by the American Journal of Tropical Medicine and Hygiene has provided a detailed analysis of historical trends that suggest policy change sometimes occurs many years after the emergence of resistance.

A WWARN statistical group that led this study has provided a detailed analysis of historical trends that suggest policy change sometimes occurs many years after the emergence of resistance. As data on ACT treatment in Africa become available, it will be possible to use these models to estimate actual ACT drug use and analyse how this contributes to the spread of drug resistance. By developing these approaches now, we could monitor key changes quickly and more effectively in the future.

How is it possible to model trends of drug usage and policy in Africa

In collaboration with ACTwatch[1], the WWARN team has quantified the trends in drug use of CQ and SP both from 1999-2011, and also after the change in policy in each country.  In addition, the methodology correlates measures of drug use and drug resistance.

To measure drug use, the team used data on antimalarial treatments taken by children under 5 years to treat fever [2] . CQ treatment levels were then modelled in relation to the national policy changes on a country-by-country basis.

Estimated SP Drug use in AfricaEstimated SP Drug use in Africa

What did the models confirm?

Overall, the analysis shows the routine usage of CQ and SP in children under 5 years old across Africa [3]

The results demonstrate that no obvious spatial pattern exists in the rate of reduction of CQ use after it was no longer the first-line drug in a given country. This suggests that drug use depends less on official country policy and more on factors that vary from country to country, such as prescription habits, costs of treatments and the integrity of the supply chain.


Continuous monitoring will be essential to guide malaria control and resistance containment policies until eradication is achieved. Support for initiatives such as the Demographic Health Surveys (DHS), Multiple Indicator Cluster Surveys (MICS) and ACTwatch is critical to ensure that appropriate data are available to analyse and use in order to strengthen our collective understanding of the ongoing drivers of resistance.

Download the paper: Flegg J, Metcalf CJE, Gharbi M et al. Trends in antimalarial drug use in Africa. American Journal of Tropical Medicine and Hygiene 2013; doi: 10.4269/ajtmh.13-0129

See two short videos demonstrating the spatial and temporal changes in SP resistance, as measured by the dhps540E marker prevalence and a second visualises changing treatment policy in Africa 1999-2011. 

This article was written by Dr Jennifer Flegg, outgoing Senior Postdoctoral Mathematical Modeller at WWARN. Eric Grist PhD, previously at the Menzies School of Health Research, Charles Darwin University in Australia, will be joining WWARN as a Senior Modeller this month.

[1] ACTwatch is a research project of Population Services International in collaboration with the London School of Hygiene & Tropical Medicine.

[2]  Household surveys used for this study were conducted between 1999-2011 by the Demographic Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) in Africa.

[3]  Graph legend (Figure 7): CQ (red), SP (blue) and ACT (green) usage data over the time period 1999-2011 and logistic regression model fits to the data. The upper subplot shows the number of countries per two-year interval that changed national antimalarial policy away from CQ (red) and SP (blue).