Uncertainty in measurement for 43 biochemistry, immunoassay, and hemostasis routine analytes evaluated by a method using only external quality assessment data

• Published in: Clinical chemistry and laboratory medicine Vol. 53; no. 11; pp. 1725 - 1736
• Main Authors: Matar, Gladys, Poggi, Bernard, Meley, Roland, Bon, Chantal, Chardon, Laurence, Chikh, Karim, Renard, Anne-Claude, Sotta, Catherine, Eynard, Jean-Christophe, Cartier, Regine, Cohen, Richard
• Format: Journal Article
• Language: English
• Published: Germany De Gruyter 01.10.2015
• Subjects: external quality assessment; Hemostasis; Humans; Immunoassay; internal quality control; Linear Models; long-term analytical coefficient of variation; proficiency testing; Quality Assurance, Health Care - methods; Quality Control; Uncertainty; uncertainty in measurement
• ISSN: 1434-6621; 1437-4331
• DOI: 10.1515/cclm-2014-0942

International organizations require from medical laboratories a quantitative statement of the uncertainty in measurement (UM) to help interpret patient results. The French accreditation body (COFRAC) recommends an approach (SH GTA 14 IQC/EQA method) using both internal quality control (IQC) and external quality assessment (EQA) data. The aim of this work was to validate an alternative way to quantify UM using only EQA results without any need for IQC data. This simple and practical method, which has already been described as the long-term evaluation of the UM (LTUM), is based on linear regression between data obtained by participants in EQA schemes and target values. We used it for 43 routine analytes covering biochemistry, immunoassay, and hemostasis fields. Data from 50 laboratories participating in ProBioQual (PBQ) EQA schemes over 25 months were used to obtain estimates of the median and 90th percentile LTUM and to compare them to the usual analytical goals. Then, the two UM estimation methods were compared using data from 20 laboratories participating in both IQC and EQA schemes. Median LTUMs ranged from 2.9% (sodium) to 16.3% (bicarbonates) for biochemistry analytes, from 12.6% (prothrombin time) to 18.4% (factor V) for hemostasis analytes when using the mean of all participants, and were around 10% for immunoassays when using the peer-group mean. Median LTUMs were, in most cases, slightly lower than those obtained with the SH GTA 14 method, whatever the concentration level. LTUM is a simple and convenient method that gives UM estimates that are reliable and comparable to those of recommended methods. Therefore, proficiency testing (PT) organizers are allowed to provide participants with an additional UM estimate using only EQA data and which could be updated at the end of each survey.