Cameron A R, Baldock F C
Lao-Australian Animal Health Project, Vientiane, Laos.
Prev Vet Med. 1998 Feb 6;34(1):1-17. doi: 10.1016/s0167-5877(97)00081-0.
Surveys to substantiate freedom from disease are becoming increasingly important. This is due to the changes in rules governing international trade in animals and animal products, and to an increase in disease eradication and herd-level accreditation schemes. To provide the necessary assurances, these surveys must have a sound theoretical basis. Until now, most surveys have been based on the assumption that the screening test used was perfect (sensitivity and specificity both equal to one), and/or that the study population was infinite. Clearly, these assumptions are virtually always invalid. This paper presents a new formula that calculates the exact probability of detecting diseased animals, and considers both imperfect tests and finite population size. This formula is computationally inconvenient, and an approximation that is simpler to calculate is also presented. The use of these formulae for sample-size calculation and analysis of survey results is discussed. A computer program, 'FreeCalc', implementing the formulae is presented along with examples of sample size calculation for two different scenarios. These formulae and computer program enable the accurate calculation of survey sample-size requirements, and the precise analysis of survey results. As a result, survey costs can be minimised, and survey results will reliably provide the required level of proof.
用以证实无疾病状态的调查正变得越来越重要。这是由于动物及动物产品国际贸易规则的变化，以及疾病根除和畜群水平认证计划的增加。为提供必要的保证，这些调查必须有坚实的理论基础。到目前为止，大多数调查基于这样的假设：所使用的筛查测试是完美的（敏感性和特异性均等于1），和/或研究总体是无限的。显然，这些假设几乎总是无效的。本文提出了一个新公式，用于计算检测患病动物的确切概率，同时考虑了不完美测试和有限总体规模。这个公式计算起来不方便，因此还给出了一个更易于计算的近似公式。讨论了如何使用这些公式进行样本量计算和调查结果分析。给出了一个实现这些公式的计算机程序“FreeCalc”，以及两种不同场景下样本量计算的示例。这些公式和计算机程序能够准确计算调查样本量要求，并精确分析调查结果。因此，可以将调查成本降至最低，且调查结果将可靠地提供所需的证明水平。
