Katholi Charles R, Unnasch Thomas R
Department of Biostatistics, University of Alabama at Birmingham, Birmingham, Alabama 35294-0022, USA.
Am J Trop Med Hyg. 2006 May;74(5):779-85.
Measuring transmission of a vector-borne infection is essential to understanding infection dynamics. When infection prevalence in the vector population is low, transmission is often measured by pool screening (also referred to as group testing). Several investigators have developed statistical methods to recover infection prevalence estimates from pool screen data. These are based on models that contain certain assumptions, and a pool screening approach must be designed to take these into account if accurate estimates of infection prevalence are to be obtained. Here we describe these assumptions and discuss appropriate sampling protocols. The sources of error inherent in pool screening are described, and we show that, under most conditions in which one would want to use group testing, most of the error results from sampling and not the pooling process. Issues involved in developing a sampling protocol, including the total number of insects to be screened and optimal pool size, are explored. The meaning of confidence intervals associated with prevalence estimates and the appropriate interpretation of these intervals are discussed.
测量媒介传播感染的传播情况对于理解感染动态至关重要。当媒介种群中的感染流行率较低时，传播情况通常通过混合筛查（也称为分组检测）来测量。几位研究人员已经开发出统计方法，以便从混合筛查数据中恢复感染流行率估计值。这些方法基于包含某些假设的模型，并且如果要获得准确的感染流行率估计值，必须设计一种混合筛查方法来考虑这些假设。在这里，我们描述这些假设并讨论适当的抽样方案。描述了混合筛查中固有的误差来源，并且我们表明，在大多数想要使用分组检测的情况下，大部分误差来自抽样而非混合过程。探讨了制定抽样方案时涉及的问题，包括要筛查的昆虫总数和最佳混合样本大小。讨论了与流行率估计相关的置信区间的含义以及对这些区间的适当解释。
