Ruokonen A, Leskinen E, Nyberg A, Vihko R
Scand J Clin Lab Invest Suppl. 1984;172:125-34.
Since 1979 the Nordic Clinical Chemistry Project (NORDKEM) has organized four external quality assessment surveys of serum hormone determinations in the five Nordic countries. Charcoal-stripped control sera with weighed additions of the following hormones have been used: thyrotropin, human placental lactogen, prolactin, estriol, estradiol, testosterone, cortisol, triiodothyronine, thyroxine, and, in the fourth survey, progesterone. Human serum was usually used but in the first survey the serum matrix for the protein hormones was calf serum. In three years no systematic improvement in the coefficients of variation of the results has occurred. The change from calf serum to human serum also had no major effects on the results. Percentage coefficients of bias usually varied on both sides of the true value, but at low hormone concentrations the results were almost always positively biased. It is therefore useful to know the true hormone concentrations of the control samples because all-laboratory means are relatively often biased from these. Variation produced by different methods of calculation of the radioimmunoassay results and the influence of outlying standard points on these data processing routines were studied in the third survey. It is possible that sometimes the variation produced by different methods of calculation is one of the main components of the total variation in external quality assessment of RIA. There were no major differences between the performance of computerized or manual calculation methods. The influence of different calibration standards on serum thyroxine determination was studied in the fourth survey. Of the total coefficient of variation (10.8% to 17.0%) the component of variation produced by different thyroxine calibration standards was 0.8% to 4.6%. In conclusion, standardization of the data processing methods and further standardization of assay reagents are urgently needed to give possibilities to identify more subtle differences in the analytical performance of the laboratories. External quality assessment of hormone determinations needs improvement of the quality of reference materials.
自1979年以来,北欧临床化学项目(NORDKEM)已在北欧五国组织了四次血清激素测定的外部质量评估调查。使用了添加了以下激素的活性炭处理对照血清:促甲状腺激素、人胎盘催乳素、催乳素、雌三醇、雌二醇、睾酮、皮质醇、三碘甲状腺原氨酸、甲状腺素,在第四次调查中还包括孕酮。通常使用人血清,但在第一次调查中,蛋白质激素的血清基质是小牛血清。三年来,结果的变异系数没有系统性改善。从小牛血清改为人类血清对结果也没有重大影响。偏差百分比系数通常在真实值的两侧变化,但在低激素浓度下,结果几乎总是正偏差。因此,了解对照样品的真实激素浓度很有用,因为所有实验室的均值相对经常与这些值有偏差。在第三次调查中研究了放射免疫分析结果不同计算方法产生的变异以及异常标准点对这些数据处理程序的影响。有时,不同计算方法产生的变异可能是放射免疫分析外部质量评估总变异的主要组成部分之一。计算机化或手动计算方法的性能之间没有重大差异。在第四次调查中研究了不同校准标准对血清甲状腺素测定的影响。在总变异系数(10.8%至17.0%)中,不同甲状腺素校准标准产生的变异分量为0.8%至4.6%。总之,迫切需要数据处理方法的标准化以及分析试剂的进一步标准化,以便能够识别实验室分析性能中更细微的差异。激素测定的外部质量评估需要提高参考物质的质量。