Sadler W A, Smith M H
Department of Nuclear Medicine, Christchurch Hospital, New Zealand.
Clin Chem. 1990 Jul;36(7):1346-50.
Published estimates of the imprecision characteristics of immunoassays are often based on quantities of data that seem to be inadequate. The increasing use of imprecision profiles has not necessarily improved the situation. We describe and illustrate a method of computing 95% confidence intervals for profiles estimated directly from replicated assay results. The data used were chosen to mimic the sort of data that are typically available in practice, either within laboratories or from external quality-assessment programs. Use of confidence intervals is an effective pictorial way of incorporating information about the quantity and the distribution of the data used for estimation. This is in keeping with an important property of imprecision profile plots: potentially complex information is summarized in a readily comprehensible way. A computer program is available for estimating and plotting profiles and their 95% confidence intervals.
已发表的免疫分析不精密度特征估计值往往基于似乎不充分的数据量。不精密度图谱的使用日益增加，但情况未必得到改善。我们描述并举例说明了一种根据重复分析结果直接计算图谱95%置信区间的方法。所使用的数据旨在模拟实验室内部或外部质量评估计划中通常可获得的数据类型。使用置信区间是纳入有关用于估计的数据数量和分布信息的一种有效直观方式。这与不精密度图谱的一个重要特性相符：潜在的复杂信息以易于理解的方式进行了总结。有一个计算机程序可用于估计和绘制图谱及其95%置信区间。
