Limits...
Mapping and Modelling the Geographical Distribution and Environmental Limits of Podoconiosis in Ethiopia.

Deribe K, Cano J, Newport MJ, Golding N, Pullan RL, Sime H, Gebretsadik A, Assefa A, Kebede A, Hailu A, Rebollo MP, Shafi O, Bockarie MJ, Aseffa A, Hay SI, Reithinger R, Enquselassie F, Davey G, Brooker SJ - PLoS Negl Trop Dis (2015)

Bottom Line: BRT modelling indicated that the probability of podoconiosis occurrence increased with increasing altitude, precipitation and silt fraction of soil and decreased with population density and clay content.The resultant maps can be used to guide programme planning and implementation and estimate disease burden in Ethiopia.This work provides a framework with which the geographical limits of podoconiosis could be delineated at a continental scale.

View Article: PubMed Central - PubMed

Affiliation: Brighton and Sussex Medical School, Falmer, Brighton, United Kingdom; School of Public Health, Addis Ababa University, Addis Ababa, Ethiopia.

ABSTRACT

Background: Ethiopia is assumed to have the highest burden of podoconiosis globally, but the geographical distribution and environmental limits and correlates are yet to be fully investigated. In this paper we use data from a nationwide survey to address these issues.

Methodology: Our analyses are based on data arising from the integrated mapping of podoconiosis and lymphatic filariasis (LF) conducted in 2013, supplemented by data from an earlier mapping of LF in western Ethiopia in 2008-2010. The integrated mapping used woreda (district) health offices' reports of podoconiosis and LF to guide selection of survey sites. A suite of environmental and climatic data and boosted regression tree (BRT) modelling was used to investigate environmental limits and predict the probability of podoconiosis occurrence.

Principal findings: Data were available for 141,238 individuals from 1,442 communities in 775 districts from all nine regional states and two city administrations of Ethiopia. In 41.9% of surveyed districts no cases of podoconiosis were identified, with all districts in Affar, Dire Dawa, Somali and Gambella regional states lacking the disease. The disease was most common, with lymphoedema positivity rate exceeding 5%, in the central highlands of Ethiopia, in Amhara, Oromia and Southern Nations, Nationalities and Peoples regional states. BRT modelling indicated that the probability of podoconiosis occurrence increased with increasing altitude, precipitation and silt fraction of soil and decreased with population density and clay content. Based on the BRT model, we estimate that in 2010, 34.9 (95% confidence interval [CI]: 20.2-51.7) million people (i.e. 43.8%; 95% CI: 25.3-64.8% of Ethiopia's national population) lived in areas environmentally suitable for the occurrence of podoconiosis.

Conclusions: Podoconiosis is more widespread in Ethiopia than previously estimated, but occurs in distinct geographical regions that are tied to identifiable environmental factors. The resultant maps can be used to guide programme planning and implementation and estimate disease burden in Ethiopia. This work provides a framework with which the geographical limits of podoconiosis could be delineated at a continental scale.

No MeSH data available.


Related in: MedlinePlus

Marginal effect curves for covariates included in 120 ensembles of boosted regression tree (BRT) models.The grey envelopes are the 95% bootstrap confidence intervals and the black line indicates the mean marginal effect. The figure in the parentheses indicates the relative contribution of each variable, which adds up to 100. The y-axis is the untransformed logit response and x-axis is the full range of covariate values.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4519246&req=5

pntd.0003946.g002: Marginal effect curves for covariates included in 120 ensembles of boosted regression tree (BRT) models.The grey envelopes are the 95% bootstrap confidence intervals and the black line indicates the mean marginal effect. The figure in the parentheses indicates the relative contribution of each variable, which adds up to 100. The y-axis is the untransformed logit response and x-axis is the full range of covariate values.

Mentions: Fig 2 shows the marginal effect of each covariate on the predicted suitability of occurrence for podoconiosis, averaging across the effects of all other variables, and its relative contribution to the final BRT model. Major predictors of the occurrence of podoconiosis were annual precipitation (accounting for 30.7% of the variation explained by the model), elevation (22.6%), EVI (15.4%) and population density (12.7%). Slope only contributed 8.2% to the predicted occurrence. Annual precipitation causes an increase in probability of occurrence starting from precipitation values of around 1,000 millimeters (mm) per year. High suitability for podoconiosis is also positively associated with elevation, increasing between 1,000–2,000 m asl and then sharply declining after 2,000 m asl. EVI is linearly correlated to the risk of podoconiosis occurrence up to 0.5 and declines sharply thereafter. Population density is negatively correlated with the probability of podoconiosis occurrence, with population density greater than 10,000 population/ km2 causing no effect on the probability of occurrence of podoconiosis. Although silt fraction and clay fraction contributed little to the final BRT model, the occurrence of podoconiosis was found to be associated with decreasing clay fraction and increasing silt fraction.


Mapping and Modelling the Geographical Distribution and Environmental Limits of Podoconiosis in Ethiopia.

Deribe K, Cano J, Newport MJ, Golding N, Pullan RL, Sime H, Gebretsadik A, Assefa A, Kebede A, Hailu A, Rebollo MP, Shafi O, Bockarie MJ, Aseffa A, Hay SI, Reithinger R, Enquselassie F, Davey G, Brooker SJ - PLoS Negl Trop Dis (2015)

Marginal effect curves for covariates included in 120 ensembles of boosted regression tree (BRT) models.The grey envelopes are the 95% bootstrap confidence intervals and the black line indicates the mean marginal effect. The figure in the parentheses indicates the relative contribution of each variable, which adds up to 100. The y-axis is the untransformed logit response and x-axis is the full range of covariate values.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4519246&req=5

pntd.0003946.g002: Marginal effect curves for covariates included in 120 ensembles of boosted regression tree (BRT) models.The grey envelopes are the 95% bootstrap confidence intervals and the black line indicates the mean marginal effect. The figure in the parentheses indicates the relative contribution of each variable, which adds up to 100. The y-axis is the untransformed logit response and x-axis is the full range of covariate values.
Mentions: Fig 2 shows the marginal effect of each covariate on the predicted suitability of occurrence for podoconiosis, averaging across the effects of all other variables, and its relative contribution to the final BRT model. Major predictors of the occurrence of podoconiosis were annual precipitation (accounting for 30.7% of the variation explained by the model), elevation (22.6%), EVI (15.4%) and population density (12.7%). Slope only contributed 8.2% to the predicted occurrence. Annual precipitation causes an increase in probability of occurrence starting from precipitation values of around 1,000 millimeters (mm) per year. High suitability for podoconiosis is also positively associated with elevation, increasing between 1,000–2,000 m asl and then sharply declining after 2,000 m asl. EVI is linearly correlated to the risk of podoconiosis occurrence up to 0.5 and declines sharply thereafter. Population density is negatively correlated with the probability of podoconiosis occurrence, with population density greater than 10,000 population/ km2 causing no effect on the probability of occurrence of podoconiosis. Although silt fraction and clay fraction contributed little to the final BRT model, the occurrence of podoconiosis was found to be associated with decreasing clay fraction and increasing silt fraction.

Bottom Line: BRT modelling indicated that the probability of podoconiosis occurrence increased with increasing altitude, precipitation and silt fraction of soil and decreased with population density and clay content.The resultant maps can be used to guide programme planning and implementation and estimate disease burden in Ethiopia.This work provides a framework with which the geographical limits of podoconiosis could be delineated at a continental scale.

View Article: PubMed Central - PubMed

Affiliation: Brighton and Sussex Medical School, Falmer, Brighton, United Kingdom; School of Public Health, Addis Ababa University, Addis Ababa, Ethiopia.

ABSTRACT

Background: Ethiopia is assumed to have the highest burden of podoconiosis globally, but the geographical distribution and environmental limits and correlates are yet to be fully investigated. In this paper we use data from a nationwide survey to address these issues.

Methodology: Our analyses are based on data arising from the integrated mapping of podoconiosis and lymphatic filariasis (LF) conducted in 2013, supplemented by data from an earlier mapping of LF in western Ethiopia in 2008-2010. The integrated mapping used woreda (district) health offices' reports of podoconiosis and LF to guide selection of survey sites. A suite of environmental and climatic data and boosted regression tree (BRT) modelling was used to investigate environmental limits and predict the probability of podoconiosis occurrence.

Principal findings: Data were available for 141,238 individuals from 1,442 communities in 775 districts from all nine regional states and two city administrations of Ethiopia. In 41.9% of surveyed districts no cases of podoconiosis were identified, with all districts in Affar, Dire Dawa, Somali and Gambella regional states lacking the disease. The disease was most common, with lymphoedema positivity rate exceeding 5%, in the central highlands of Ethiopia, in Amhara, Oromia and Southern Nations, Nationalities and Peoples regional states. BRT modelling indicated that the probability of podoconiosis occurrence increased with increasing altitude, precipitation and silt fraction of soil and decreased with population density and clay content. Based on the BRT model, we estimate that in 2010, 34.9 (95% confidence interval [CI]: 20.2-51.7) million people (i.e. 43.8%; 95% CI: 25.3-64.8% of Ethiopia's national population) lived in areas environmentally suitable for the occurrence of podoconiosis.

Conclusions: Podoconiosis is more widespread in Ethiopia than previously estimated, but occurs in distinct geographical regions that are tied to identifiable environmental factors. The resultant maps can be used to guide programme planning and implementation and estimate disease burden in Ethiopia. This work provides a framework with which the geographical limits of podoconiosis could be delineated at a continental scale.

No MeSH data available.


Related in: MedlinePlus