Large-scale drivers of malaria and priority areas for prevention and control in the Brazilian Amazon region using a novel multi-pathogen geospatial model

Abstract

Background: Most of the malaria burden in the Americas is concentrated in the Brazilian Amazon but a detailed spatial characterization of malaria risk has yet to be undertaken.

Methods: Utilizing 2004-2008 malaria incidence data collected from six Brazilian Amazon states, large-scale spatial patterns of malaria risk were characterized with a novel Bayesian multi-pathogen geospatial model. Data included 2.4 million malaria cases spread across 3.6 million sq km. Remotely sensed variables (deforestation rate, forest cover, rainfall, dry season length, and proximity to large water bodies), socio-economic variables (rural population size, income, and literacy rate, mortality rate for children age under five, and migration patterns), and GIS variables (proximity to roads, hydro-electric dams and gold mining operations) were incorporated as covariates.

Results: Borrowing information across pathogens allowed for better spatial predictions of malaria caused by Plasmodium falciparum, as evidenced by a ten-fold cross-validation. Malaria incidence for both Plasmodium vivax and P. falciparum tended to be higher in areas with greater forest cover. Proximity to gold mining operations was another important risk factor, corroborated by a positive association between migration rates and malaria incidence. Finally, areas with a longer dry season and areas with higher average rural income tended to have higher malaria risk. Risk maps reveal striking spatial heterogeneity in malaria risk across the region, yet these mean disease risk surface maps can be misleading if uncertainty is ignored. By combining mean spatial predictions with their associated uncertainty, several sites were consistently classified as hotspots, suggesting their importance as priority areas for malaria prevention and control.

Conclusion: This article provides several contributions. From a methodological perspective, the benefits of jointly modelling multiple pathogens for spatial predictions were illustrated. In addition, maps of mean disease risk were contrasted with that of statistically significant disease clusters, highlighting the critical importance of uncertainty in determining disease hotspots. From an epidemiological perspective, forest cover and proximity to gold mining operations were important large-scale drivers of disease risk in the region. Finally, the hotspot in Western Acre was identified as the area that should receive highest priority from the Brazilian national malaria prevention and control programme.

Figure 1

Spatial distribution of the locations that originated the malaria data used in this article.

Figure 2

The ‘Pathogen + Spatial’ model had the best out-of-sample predictive skill. Left panel depicts 95% CI coverage, defined as the proportion of out-of-sample observations that fell within the estimated predictive 95% CI. Dotted horizontal line depicts the target CI coverage. Right panel depicts the Mean Squared Error (MSE) for each model. A description of each model is provided in Table 1.

Figure 3

Malaria risk factors for Plasmodium vivax (black) and Plasmodium falciparum (red). Each panel depicts the estimated relationship between each covariate and log-malaria risk α _iyj. Significant relationships are depicted with lines, which represent the median of the posterior distribution. Polygons are 95% CIs. Boxplots show the distribution of the corresponding covariate.

Figure 4

Smooth surfaces of predicted malaria risk in the Brazilian Amazon. Upper and lower panels show results for P. vivax and P. falciparum, respectively, while results for 2004 to 2008 are shown from left to right. Letters in the upper left panel refer to state names (AC = Acre, AM = Amazonas, RR = Roraima, AP = Amapa, PA = Para, and RO = Rondonia). The same colour scheme was used for all panels, where warmer colours indicate higher malaria risk. Numbers in colour key are predicted malaria incidence per capita per month.

Figure 5

Coldspots (CS) and hotspots (HS) reveal substantially different spatial patterns than Figure 4 . Upper and lower panels show results for P. vivax and P. falciparum, respectively. Persistent hotspots (red) are areas that were classified as hotspots throughout the five-year study period (2004-2008). Intermittent hotspots (light red) are areas that were classified at least once as a hotspot but not in all five years. Analogous definitions were used for persistent (blue) and intermittent (light blue) coldspots. Finally, potential hotspots are shown in grey. Letters indicate state names (AC = Acre, AM = Amazonas, RR = Roraima, AP = Amapa, PA = Para, and RO = Rondonia).

Figure 6

Malaria risk is declining throughout most of the Brazilian Amazon region except for Amazonas state. Areas where the estimated malaria risk significantly increased or declined from 2004 to 2008 are shown with red and blue triangles, respectively. Areas with no significant change (‘stable’) are shown with empty squares. Intermittent and persistent hotspots from Figure 5 are shown in grey in the background. Letters indicate state names (AC = Acre, AM = Amazonas, RR = Roraima, AP = Amapa, PA = Para, and RO = Rondonia).