COS 132-9 - Malaria and socioeconomic conditions: A time series model to address double causality

Friday, August 12, 2011: 10:50 AM
10B, Austin Convention Center
Andres Baeza, Ecology & Evolutionary Biology, University of Michigan, Annemarie ter Venn, Royal Dutch Tropical Institute, Menno J. Bouma, Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, University of London, London, United Kingdom, Aaron A. King, Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI and Mercedes Pascual, Ecology and Evolutionary Biology, University of Michigan,Howard Hughes Medical Institute, Santa Fe Institute, Ann Arbor, MI

In agricultural communities, malaria and socioeconomic conditions are intimately connected in such a way that causality operates in both directions. On one hand, malaria incidence can affect income level by decreasing labor productivity. On the other hand, low income level can affect prevention and control, resulting in the exacerbation of the disease burden, and in so doing reinforcing the negative feedback loop between poverty and disease. Theoretical studies for infectious diseases in general have proposed a ‘poverty trap’ resulting from this double causality or feedback loop. Although multiple empirical studies have addressed these linkages for malaria in isolation, for one direction of influence or the other, these have not considered the explicit dynamics of the system, especially when both effects are acting in concert. An understanding of the relative importance of these two effects from a dynamic perspective is relevant to malaria control in developing countries.

In this work we investigate the dynamic feedback between malaria and socioeconomic conditions by relying on historical data for the eradication of malaria in Mississippi between 1914 and 1927. We focus on Mississippi because this state was highly malarious and harbored the largest number of tenant farmers, most of whom lived in poverty. Moreover, the main source of income in Mississippi was cotton production. To statistically test the feedback hypothesis, we constructed a series of nested time series models and parameterized these based on maximum likelihood for point estimates and trajectory matching, with yearly data for malaria incidence, cotton productivity and income level. The full model was compared via likelihood ratio test to sub-models that specifically disconnect part of the loop.


Results show that the models that include the explicit feedback loop between malaria, cotton productivity, and income do significantly better than those that do not. However, we also found that malaria had only a weak effect on cotton productivity. Furthermore, results suggest that the eradication of the disease was not a consequence of the feedback loop itself, but was instead mostly driven by macro-level determinants related to the overall improvement of socioeconomic conditions. We conclude that at the scale of the state, the poverty trap was too weak to account for the dramatic decrease in cases. We discuss the implications of our findings for malaria today in parts of the world with similar socioeconomic and ecological conditions.

Copyright © . All rights reserved.
Banner photo by Flickr user greg westfall.