Seid Ahmed, Gadisa Endalamaw, Tsegaw Teshome, Abera Adugna, Teshome Aklilu, Mulugeta Abate, Herrero Merce, Argaw Daniel, Jorge Alvar, Kebede Asnakew, Aseffa Abraham
Geospat Health. 2014 May;8(2):377-87. doi: 10.4081/gh.2014.27.
Cutaneous leishmaniasis (CL) is a neglected tropical disease strongly associated with poverty. Treatment is problematic and no vaccine is available. Ethiopia has seen new outbreaks in areas previously not known to be endemic, often with co-infection by the human immunodeficiency virus (HIV) with rates reaching 5.6% of the cases. The present study concerns the development of a risk model based on environmental factors using geographical information systems (GIS), statistical analysis and modelling. Odds ratio (OR) of bivariate and multivariate logistic regression was used to evaluate the relative importance of environmental factors, accepting P ≤ 0.056 as the inclusion level for the model's environmental variables. When estimating risk from the viewpoint of geographical surface, slope, elevation and annual rainfall were found to be good predictors of CL presence based on both probabilistic and weighted overlay approaches. However, when considering Ethiopia as whole, a minor difference was observed between the two methods with the probabilistic technique giving a 22.5% estimate, while that of weighted overlay approach was 19.5%. Calculating the population according to the land surface estimated by the latter method, the total Ethiopian population at risk for CL was estimated at 28,955,035, mainly including people in the highlands of the regional states of Amhara, Oromia, Tigray and the Southern Nations, Nationalities and Peoples' Region, one of the nine ethnic divisions in Ethiopia. Our environmental risk model provided an overall prediction accuracy of 90.4%. The approach proposed here can be replicated for other diseases to facilitate implementation of evidence-based, integrated disease control activities.
皮肤利什曼病（CL）是一种与贫困密切相关的被忽视的热带病。治疗存在问题且尚无可用疫苗。埃塞俄比亚在以前未知的地方出现了新的疫情，通常还伴有人类免疫缺陷病毒（HIV）合并感染，感染率达病例的5.6%。本研究涉及利用地理信息系统（GIS）、统计分析和建模开发基于环境因素的风险模型。二元和多元逻辑回归的优势比（OR）用于评估环境因素的相对重要性，接受P≤0.056作为模型环境变量的纳入水平。从地理表面的角度估计风险时，基于概率和加权叠加方法，坡度、海拔和年降雨量被发现是CL存在的良好预测指标。然而，将埃塞俄比亚作为一个整体考虑时，两种方法之间观察到细微差异，概率技术的估计值为22.5%，而加权叠加方法为19.5%。根据后一种方法估计的陆地面积计算人口，埃塞俄比亚CL风险总人口估计为28955035人，主要包括阿姆哈拉、奥罗米亚、提格雷等州高地以及埃塞俄比亚九个民族分区之一的南方民族、民族和人民地区的人口。我们的环境风险模型提供了90.4%的总体预测准确率。这里提出的方法可用于其他疾病，以促进循证综合疾病控制活动的实施。
