Geomorphology Laboratory, Department of Geography, University of Hawaii at Manoa, 2424 Maile Way, Honolulu, HI 96822, USA.
Anal Chim Acta. 2010 Nov 8;680(1-2):10-20. doi: 10.1016/j.aca.2010.09.016. Epub 2010 Sep 17.
A detailed quantitative analysis was performed on data presented in the literature that focused on the sequential extraction of cadmium (Cd), chromium (Cr), copper (Cu), nickel (Ni), lead (Pb) and zinc (Zn) from the certified reference material BCR-701 (lake sediment) using the three-step harmonized BCR(®) procedure. The accuracy of data reported in the literature, including precision and different measures of trueness, was assessed relative to the certified values for BCR-701. Forty data sets were accepted following extreme outlier removal, and statistically summarized with measures of central tendency, dispersion, and distribution form. In general, literature data were similar in their measurement precision to the expert laboratories used to certify the trace element contents in BCR-701. The overall median precision for literature reported data was 10% (range 6-19%), compared to certifying laboratories of 9% (range 4-33%). One measure of literature data trueness was assessed via a confirmatory approach using a robust bootstrap method. Only 22% of the comparisons indicated significantly different (all were lower) concentrations reported in the literature compared to certified values. The question of whether the differences are practically significant for environmental studies is raised. Bias was computed as a measure of trueness, and literature data were more frequently negatively biased, indicating lower concentrations reported in the literature for the six trace elements for the three-step sequential procedure compared to the certified values. However, 95% confidence intervals about the average bias for the 18 comparisons indicated only four instances when a mean bias of 0 (i.e., measured=certified) was not incorporated-suggesting statistical difference. Finally, Z-scores incorporating a Horwitz-type function were used to assess the general trueness of laboratory data. Of the 468 laboratory Z-score values computed, 92% were considered to be satisfactory, 5% were questionable, and 3% were unsatisfactory. A detailed examination of the methodology sections of the various studies showed that despite claiming adherence to the harmonized BCR sequential extraction protocol, significant deviations were commonly observed; particularly in moisture correction, sample mass, centrifugation specifics, shaking specifics, and incorporation of filtration. It is likely that failure to strictly adhere to the protocol adversely impacted accuracy, by increasing the degree of imprecision and resulting in more discrepant trueness values.
对文献中重点介绍的从认证参考物质 BCR-701（湖泊沉积物）中采用三步协调 BCR(®)程序连续提取镉（Cd）、铬（Cr）、铜（Cu）、镍（Ni）、铅（Pb）和锌（Zn）的详细定量分析进行了详细的定量分析。相对于 BCR-701 的认证值，评估了文献中报告数据的准确性，包括精密度和不同的准确度衡量标准。经过极端异常值剔除后，接受了 40 个数据集，并使用集中趋势、分散度和分布形式的度量标准进行了统计总结。总体而言，文献数据在测量精度方面与用于认证 BCR-701 微量元素含量的专家实验室相似。文献报告数据的整体中位数精度为 10%（范围为 6-19%），而认证实验室为 9%（范围为 4-33%）。通过稳健的自举法进行确认性评估，对文献数据的一个准确度衡量标准进行了评估。只有 22%的比较表明文献中报告的浓度明显（全部更低）与认证值不同。提出了对于环境研究而言，这些差异是否具有实际意义的问题。偏差作为准确度的衡量标准进行计算，并且文献数据更频繁地呈负偏差，表明与认证值相比，三步连续程序中六种痕量元素的文献报告浓度更低。然而，关于 18 次比较的平均偏差的 95%置信区间仅在四个实例中未包含平均偏差为 0（即，测量值=认证值）的情况-表明存在统计差异。最后，使用包含 Horwitz 型函数的 Z 分数评估实验室数据的总体准确度。在所计算的 468 个实验室 Z 分数值中，92%被认为是令人满意的，5%是可疑的，3%是不满意的。对各个研究的方法部分进行了详细检查，结果表明，尽管声称遵守协调的 BCR 连续提取方案，但通常观察到明显的偏差；特别是在水分校正、样品质量、离心细节、摇动细节和过滤的结合方面。未能严格遵守协议很可能会因增加不精确性并导致更多不一致的准确度值而对准确性产生不利影响。
