1. Combining Nordtest method and bootstrap resampling for measurement uncertainty estimation of hematology analytes in a medical laboratory
- Author
-
Shaoqing Ju, Ming Cui, Huimin Wang, Rongrong Jing, Lili Xu, and Shuizhu Xu
- Subjects
0301 basic medicine ,Quality Control ,Analyte ,Clinical Biochemistry ,Medical laboratory ,03 medical and health sciences ,0302 clinical medicine ,External quality assessment ,Statistics ,Humans ,Computer Simulation ,Mathematics ,Alternative methods ,business.industry ,Mathematical statistics ,Uncertainty ,Small sample ,General Medicine ,Hematology ,Internal quality ,030104 developmental biology ,Dimensional Measurement Accuracy ,030220 oncology & carcinogenesis ,Measurement uncertainty ,business ,Software - Abstract
Background Measurement uncertainty (MU) is a metrological concept, which can be used for objectively estimating the quality of test results in medical laboratories. The Nordtest guide recommends an approach that uses both internal quality control (IQC) and external quality assessment (EQA) data to evaluate the MU. Bootstrap resampling is employed to simulate the unknown distribution based on the mathematical statistics method using an existing small sample of data, where the aim is to transform the small sample into a large sample. However, there have been no reports of the utilization of this method in medical laboratories. Thus, this study applied the Nordtest guide approach based on bootstrap resampling for estimating the MU. Methods We estimated the MU for the white blood cell (WBC) count, red blood cell (RBC) count, hemoglobin (Hb), and platelets (Plt). First, we used 6 months of IQC data and 12 months of EQA data to calculate the MU according to the Nordtest method. Second, we combined the Nordtest method and bootstrap resampling with the quality control data and calculated the MU using MATLAB software. We then compared the MU results obtained using the two approaches. Results The expanded uncertainty results determined for WBC, RBC, Hb, and Plt using the bootstrap resampling method were 4.39%, 2.43%, 3.04%, and 5.92%, respectively, and 4.38%, 2.42%, 3.02%, and 6.00% with the existing quality control data ( U [ k = 2]). For WBC, RBC, Hb, and Plt, the differences between the results obtained using the two methods were lower than 1.33%. The expanded uncertainty values were all less than the target uncertainties. Conclusions The bootstrap resampling method allows the statistical analysis of the MU. Combining the Nordtest method and bootstrap resampling is considered a suitable alternative method for estimating the MU.
- Published
- 2017