Mapping malaria transmission in West and Central Africa
Mapping malaria transmission in West and Central Africa
复制标题
DOI：
10.1111/j.1365-3156.2006.01640.x
复制
发表时间：
2006-07-01
影响因子：
3.3
通讯作者：
Vounatsou, Penelope
中科院分区：
文献类型：
--
作者：
Gemperli, Armin;Sogoba, Nafomon;Vounatsou, Penelope
我们利用非洲疟疾风险绘图数据库制作了西非和中非恶性疟原虫疟疾传播地图，该数据库包括这些地区所有可进行地理定位的疟疾流行率调查。分析的1846次疟疾调查是在不同季节进行的，并使用不同年龄组的人口进行报告。为了进行比较，我们使用了Garki疟疾传播模型，将976个采样点的疟疾流行率数据转换为传播强度E的单一估计值，并利用基于归一化差异植被指数（NDVI）、温度和降雨量数据的季节性模型。我们使用进一步的环境协变量将贝叶斯地统计模型拟合到E，并应用贝叶斯克里金法获得E的平滑地图，从而获得特定年龄的患病率。该产品是包括中部非洲在内的第一份详细的疟疾传播强度变化经验地图。它已得到专家意见的证实，并在总体上证实了已知的疟疾传播模式，提供了一个基线，可据以评价驱虫蚊帐方案等干预措施和抗药性趋势。模型估计的精确度在地理上有很大差异，在西非的一些地区，预测结果与其他风险图的预测结果相差很大。由此产生的不确定性表明，最迫切需要进一步调查数据的地区。根据不同调查数据汇编的疟疾风险图对分析方法高度敏感。
We have produced maps of Plasmodium falciparum malaria transmission in West and Central Africa using the Mapping Malaria Risk in Africa (MARA) database comprising all malaria prevalence surveys in these regions that could be geolocated. The 1846 malaria surveys analysed were carried out during different seasons, and were reported using different age groupings of the human population. To allow comparison between these, we used the Garki malaria transmission model to convert the malaria prevalence data at each of the 976 locations sampled to a single estimate of transmission intensity E, making use of a seasonality model based on Normalized Difference Vegetation Index (NDVI), temperature and rainfall data. We fitted a Bayesian geostatistical model to E using further environmental covariates and applied Bayesian kriging to obtain smooth maps of E and hence of age-specific prevalence. The product is the first detailed empirical map of variations in malaria transmission intensity that includes Central Africa. It has been validated by expert opinion and in general confirms known patterns of malaria transmission, providing a baseline against which interventions such as insecticide-treated nets programmes and trends in drug resistance can be evaluated. There is considerable geographical variation in the precision of the model estimates and, in some parts of West Africa, the predictions differ substantially from those of other risk maps. The consequent uncertainties indicate zones where further survey data are needed most urgently. Malaria risk maps based on compilations of heterogeneous survey data are highly sensitive to the analytical methodology.