北京市49家三级医院医学实验室酶学检验结果测量不确定度的评定与分析

Evaluation and analysis of measurement uncertainty for enzymology test results from medical laboratories of 49 tertiary hospitals in Beijing

目的用"自上而下(top-down)"方法评定北京市49家三级医院临床实验室肌酸激酶(CK)、乳酸脱氢酶(LDH)、γ-谷氨酰基转移酶(GGT)常规检验结果的测量不确定度,鉴别并分析其主要来源。方法应用"top-down"方法,通过酶学参考测量程序赋值的人血清和实验室室内质控数据评估酶学检验结果的测量不确定度。以常规实验室重复测量赋值血清10次数据评估偏移引入的相对测量不确定度分量[ucrel(bias)];以连续3个月实验室室内质控统计数据评定实验室内复现性引入的相对测量不确定度分量[urel(RW)];再通过ucrel(bias)和urel(RW)计算相对合成标准不确定度(ucrel)和相对扩展不确定度(Urel)。结果 49家实验室测量不确定度的评定结果显示:对于CK、LDH、GGT 3个指标,在合成ucrel(bias)的3个分量中,测量平均值与赋值血清定值的相对偏移量值(brel)是ucrel(bias)的主要来源。3个指标ucrel(bias)引入的不确定度均大于urel(RW)引入的不确定度;ucrel(bias)的离散度亦明显大于urel(RW);并依CK、LDH、GGT次序顺序增大。CK、LDH、GGT指标Urel的中位数和四分位数分别为8.72%(6.08%,12.26%)、9.45%(6.66%,16.68%)、9.90%(6.28%,21.29%);若修正偏移,Urel将会有相当程度的改善,中位数和四分位数分别为4.77%(3.84,7.41%)、5.50%(4.06%,7.35%)、4.68%(3.56%,6.31%)。检测系统分组数据显示:依组成分组,其组间平均brel的差异均无统计学意义(P均>0.05);据校准品分组,LDH、GGT指标组间平均brel的差异有统计学意义(P<0.05)。结论 brel是实验室主要不确定度来源;使用某厂家校准品或不使用校准品的分析系统是产生偏移的主要因素。统一评估测量不确定度是改善酶学检验结果实验室间可比性的方式。

Objective To evaluate the measurement uncertainty of creatine kinase（ CK）,lactate dehydrogenase（ LDH） and γ-glutamyl transferase（ GGT） test results from medical laboratories of 49 tertiary hospitals in Beijing by the top-down approach,and analyze its main source. Methods The measurement uncertainty of enzymology test results was evaluated with the frozen human-pooled serum sample assigned by the enzyme reference measurement procedures and internal quality control data by the top-down approach. The relative standard measurement uncertainty from bias [ucrel（ bias） ] was assessed with the repeated values of the assigned sample for 10 times detected by the routine laboratory. The component induced from internal laboratory measurement reproducibility [urel（ RW） ]was assessed with the internal quality control data of three successive months. Then,the relative combined standard uncertainty（ ucrel） and the relative expanded uncertainty（ Urel） were calculated with ucrel（ bias） and Urel（ RW）. Results The results from 49 laboratories showed that the relative bias（ brel） between the mean value detected by routine laboratories and the assigned value was the main contributor to the ucrel（ bias） for CK,LDH and GGT. The uncertainty introduced by ucrel（ bias） was higher than that from urel（ RW） and the dispersions of ucrel（ bias） were also wider than that of Urel（ RW） for CK,LDH and GGT,and the degree increased at sequence from CK,LDH to GGT. The medians and quartiles of the Urelwere 8. 72%（ 6. 08%,12. 26%）,9. 45%（ 6. 66%,16. 68%） and 9. 90%（ 6. 28%,21. 29%）,respectively,for CK,LDH and GGT. When the bias was corrected,the Urelreduced to a large extent,and the medians and quartiles were 4. 77%（ 3. 84,7. 41%）,5. 50%（ 4. 06%,7. 35%） and 4. 68%（ 3. 56%,6. 31%）,respectively.Further grouping analysis showed that there was no difference in the brel of different groups divided by the analysis system composition（ P〉0. 05）,but there was difference in the brel of different groups divided by the calibrate materials for LDH and GGT（ P〈0. 05）.Conclusion The brelis the main source of measurement uncertainty in laboratories. With or without calibrate materials in the analysis system is a major factor influencing the result. The unified estimation of measurement uncertainty may be a way of improving the comparability of enzymology test results between laboratories.