Determination of sigma score based on biological variation for haemostasis assays: fit-for-purpose for daily practice?

• Published in: Clinical chemistry and laboratory medicine Vol. 57; no. 8; pp. 1235 - 1241
• Main Authors: Hollestelle, Martine J., Ruinemans-Koerts, Janneke, Idema, René N., Meijer, Piet, de Maat, Moniek P.M.
• Format: Journal Article
• Language: English
• Published: Germany De Gruyter 01.08.2019; Walter De Gruyter & Company
• Subjects: Antithrombin; Biological variation; Blood Coagulation Tests; Coefficient of variation; Diagnostic systems; Fibrinogen; haemostasis; Hemostasis; Humans; Laboratory tests; Mathematical analysis; Measurement methods; Medical laboratories; Monitoring instruments; Prothrombin; Quality Control; quality control data; Reagents; sigma score; Thromboplastin; Total Quality Management; Variation; sigma score; biological variation; haemostasis; quality control data
• ISSN: 1434-6621; 1437-4331; 1437-4331
• DOI: 10.1515/cclm-2018-0934

Background Internal quality control (QC) rules for laboratory tests can be derived from analytical performance specifications (APS) using the six-sigma method. We tested the applicability of this paradigm to routine haemostasis measurements. Methods Three laboratories using different instruments and reagents calculated sigma scores for their prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen and antithrombin (AT) measurements. Sigma scores were calculated using biological variation (BV) data from the literature in combination with internal and external QC data. Results Wide ranges in sigma scores for the PT (0.1-6.8), APTT (0.0-4.3), fibrinogen (1.5-8.3) and AT (0.1-2.4) were observed when QC data was combined with the minimum, median and maximum value of BV data, due in particular to a large variation in within-subject and between-subjects coefficients of variation. When the median BV values were applied, most sigma scores were below 3.0, for internal QC data; 75% and for external QC data; 92%. Conclusions Our findings demonstrate that: (1) The sigma scores for common haemostasis parameters are relatively low, and (2) The application of the six-sigma method to BV-derived APS is hampered by the large variation in published BV data. As the six-sigma concept is based on requirements for monitoring, and many haemostasis tests are only designed for diagnostic purposes, a fit-for-purpose APS is needed to achieve clinically relevant quality goals.