Spatial Analysis of Malaria on The Geo-Additive Bayesian Model

Dawit Getnet Ayele1*, Temesgen Zewotir1, Henry Mwambi1

1School of Mathematics, Statistics and Computer Science, University of KwaZulu-Natal, Pietermaritzburg, Private Bag X01, Scottsville 3209, South Africa

Copyright: © 2016 Dawit Getnet Ayele, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.


One of the major public health issues in Ethiopia is malaria. From the total population about 4-5 million Ethiopians affected each year. The objective of this study is to identify the dependence of malaria situation on spatial factors and socio-economic, demographic and geographic variables. The investigation in this study uses the household cluster malaria survey which was conducted from December 2006 to January 2007. Geo-additive Bayesian model using Kebele as the geographic unit of the study was used. From the investigation, it can be seen that households in the SNNP region were found to be at more risk than Amhara and Oromiya regions. Households with better facilities including bed nets have less chance to be infected by malaria. The study also suggested that including spatial variability is essential to understand and plan the most suitable policies to decrease the threat of malaria. Semi-parametric models were used to modeling the effects of socioeconomic, demographic and geographic covariates and spatial effects on malaria distribution in Ethiopia. The results recommend the strong positive relations between malaria rapid diagnosis test and socio-economic, demographic and geographic factors. The spatial variability showed important spatial patterns of malaria.


Malaria, Rapid Diagnostic Test, Spatial Statistics, Bayesian Model, MCMC, Geo-additive


The relationship between malaria and socio-economic statusin Ethiopia dictated the use of a spatial model to identify the risks. About 4-5 million people are affected by malaria because most (75%) of Ethiopia is malarious during the rainy season. Currently, strong associations between malaria and climate; and demographic, geographic and socio-economic factors have been found. A further significantly positively correlated relationship between the number of malaria cases, temperature and rainfall was documented by Pemola and Jauhari in 2006 [1-3]. 

A number of researchers examining the same topic indicated that factors other than climate may explain the distribution of malaria [4-6]. For instance, Ayele, Zewotir and Mwambi (2012, 2013, 2014. 2015)
noted high rates of malaria morbidity could result from poor access to socio-economic services. Consequently, the problems are associated with key socio-economic, demographic and geographic factors, and in particular, with poverty levels of households [7-11]. In addition to this, environmental factors, population growth, limited access to healthcare systems, and lack of unsuccessful malaria control measurescontribute to malaria transmission [12].

In previous studies malaria risk factors were examined using spatial statistics analysis and semiparametric methods separately [13,14]. But, the factors affecting malaria RDT result might have both spatial variability and nonlinear relationships with malaria RDT result. These effects were not done previously. Therefore, in this study, a geo-additive model is suggested to identify the risk factors of malaria on spatial effects and socio-economic, demographic and geographic factors in three regions of Ethiopia. The method incorporates both the spatial variability and the nonlinear relationships between covariates and response variables.


Ethiopia’s land size is estimated to be about 1.1 million square kilometers. The country is the Federal Democratic Republic divided into nine national regional states. These are: Tigray, Afar, Amhara, Oromia, Somali, Benishangul-Gumuz, Southern Nations Nationalities and People Region (SNNPR), Gambella and Harari and two administrative regions (Addis Ababa City administration and Dire Dawa City Council). From these eleven regions, Amhara,
Oromiya and SNNP regions constitute more than 58% of the total size of Ethiopia. The Amhara region is found in the north western and north central part of Ethiopia. This region has ten administrative zones, one special zone, 105 woredas, and 78 urban centres. The region is divided into the highlands and lowlands. The highlands (northern and eastern parts of the region) are 1500 meters above sea level and are characterized by chains of mountains and plateaus. The region of Oromiya covers the biggest portion of the country. This region has of twelve administrative zones and 180 woredas. The landscape of the region includes tall and rocky mountains. The region is located 500 meters above sea level to high ranges that culminate into more than 4000 meters. But the altitude of over 1500 meters is dominant in the region. The region of Southern Nations, Nationalities and Peoples’ comprises 10% of the total area of the country. The region is divided in to nine zones, 72 woredas and five special woredas. The region lies in the southern part of the country and has an elevation range from 376 to 4, 207 meter above sea level. About 56 % of the total area is found below 1,500m. The remaining 44% is temperate in climate. These three regions were selected for this study.


Data description

The aim of the study was to identify the problem of malaria on aspects, such as socio-economic, demographic and geographic variables and spatial correlated and uncorrelated spatial effects in three regions of Ethiopia. Baseline household survey was conducted from December 2006 to January 2007 by The Carter Center (TCC). In the survey, Kebele (smallest administrative unit) was considered as the sampling frame in each of the rural populations of Amhara, Oromiya and SNNP regions. From the three regions, 5,708 households located in 224 clusters were selected, i.e, 4,101 (71.85%) for Amhara, 809 (14.17%) for Oromiya and 798 (13.98%) for SNNP. From each Kebele, twelve even numbered households were selected for malaria tests. Socioeconomic, demographic and geographic factors of interest Outcome variable: For this study, the malaria rapid diagnosis test (RDT) result (binary) was considered as an outcome variable. RDT is a method which helps to the diagnosis of malaria. RDT are used instead of microscopy if there are no good quality services. Predictor variables: The predictor variables or covariates were the baseline socioeconomic status, demographic and geographic variables. These variables are described in the following table (Table 1).

Model construction

In previous studies, assuming that socioeconomic, demographic and geographic variables were assumed to have a nonlinear effect on malaria rapid diagnosis test [15-17]. Because age, household size, number of rooms per person, number of nets per person, altitude and number of months the room sprayed are continuous variables, the relationship with malaria rapid diagnosis test might be nonlinear [14]. In addition to nonlinear effects, there was spatial variability was found in the previous study [13]. Therefore, using the algorithm described in [18], Generalized Additive Mixed Model (GAMM) with spatial covariance structure [19] was suggested to investigate the effect of malaria rapid diagnosis test on socioeconomic, demographic and geographic variables.