Relative importance of climatic, geographic and socio-economic determinants of malaria in Malawi

Abstract

Background: Malaria transmission is influenced by variations in meteorological conditions, which impact the biology of the parasite and its vector, but also socio-economic conditions, such as levels of urbanization, poverty and education, which impact human vulnerability and vector habitat. The many potential drivers of malaria, both extrinsic, such as climate, and intrinsic, such as population immunity are often difficult to disentangle. This presents a challenge for the modelling of malaria risk in space and time.

Methods: A statistical mixed model framework is proposed to model malaria risk at the district level in Malawi, using an age-stratified spatio-temporal dataset of malaria cases from July 2004 to June 2011. Several climatic, geographic and socio-economic factors thought to influence malaria incidence were tested in an exploratory model. In order to account for the unobserved confounding factors that influence malaria, which are not accounted for using measured covariates, a generalized linear mixed model was adopted, which included structured and unstructured spatial and temporal random effects. A hierarchical Bayesian framework using Markov chain Monte Carlo simulation was used for model fitting and prediction.

Results: Using a stepwise model selection procedure, several explanatory variables were identified to have significant associations with malaria including climatic, cartographic and socio-economic data. Once intervention variations, unobserved confounding factors and spatial correlation were considered in a Bayesian framework, a final model emerged with statistically significant predictor variables limited to average precipitation (quadratic relation) and average temperature during the three months previous to the month of interest.

Conclusions: When modelling malaria risk in Malawi it is important to account for spatial and temporal heterogeneity and correlation between districts. Once observed and unobserved confounding factors are allowed for, precipitation and temperature in the months prior to the malaria season of interest are found to significantly determine spatial and temporal variations of malaria incidence. Climate information was found to improve the estimation of malaria relative risk in 41% of the districts in Malawi, particularly at higher altitudes where transmission is irregular. This highlights the potential value of climate-driven seasonal malaria forecasts.

Figure 1

Malaria SMR and average climate in Malawi for the period July 2004 - June 2011. (a) Malaria standardised morbidity ratios (SMR) for the under five (dashed curve) and five years and over (solid curve) age categories and (b) average precipitation (solid bars) and average temperature (dashed curve) in Malawi for the period July 2004 - June 2011.

Figure 2

Spatial distribution of malaria SMR, geographic and socio-economic indicators across Malawi for the period July 2004 - June 2011. Map of (a) malaria SMR for under fives, (b) malaria SMR for five years and over, (c) ecological zones, (d) mean altitude, (e) population density, (f) proportion of households with only one room for sleeping, (g) mean ITN distribution rate and (h) the number of health facilities per 1000 inhabitants in each district over the period July 2004 - June 2011.

Figure 3

Kernel density estimates for significant explanatory variables. Kernel density estimates for the marginal posterior distributions for the parameters associated with (a) average precipitation, (b) precipitation squared, (c) average temperature (d) temperature squared, (e) health facilities per inhabitant and (f) ITN distribution rate.

Figure 4

Multiplicative contribution of spatially unstructured and structured random effects to malaria relative risk. Spatial distribution of multiplicative contribution of posterior mean spatially (a) unstructured ${\widehat{\varphi }}_s$ and (b) structured ${\widehat{\upsilon }}_s$ random effects.

Figure 5

Multiplicative contribution of temporally unstructured and structured random effects to malaria relative risk. Temporal distribution of auto-correlated random month effects ${\widehat{\omega }}_{t(t)}$ (accounting for annual cycle) and random year effects τ_t(t) (to account for unexplained trend) for under five and five years and over age categories.

Figure 6

Multiplicative contribution of climate variables to malaria relative risk. Surface of malaria relative risk given varying average precipitation and temperature values. Note that the maximum relative risk is found at the maximum temperature of 28°C and a precipitation threshold of 6.24 mm day^-1.

Figure 7

Observed versus predicted malaria SMR in space and time. Scatter plot (a, e) and time series (b, f) of space aggregated observed versus posterior predictive mean malaria SMR for the 84 month time period. Root mean squared error (RMSE) of observed and posterior mean malaria SMR for the 27 districts of Malawi for the period July 2004 - June 2011 (c, g) for the five years and over (upper panel) and under five years (lower panel) age groups. The lower the RMSE, the better the model fit. Difference between RMSE for the model including climate information and RMSE for a model fit without climate information (d, h). Districts with negative values of RMSE_clim - RMSE_noclim (white) suggest that climate information improves the model in these areas. Districts with positive values of RMSE_clim - RMSE_noclim (grey) suggest that climate information does not improve the model.