Using native human sera to assess the suitability of external quality assessment materials based on testosterone,...

The commonly processed nature of the control materials used in the external quality assessment (EQA) of medical laboratories requires that their suitability is evaluated against native patient samples to ensure valid evaluation of EQA results across several measurement procedures (MPs). The aim of this study was to examine the extent to which EQA analyses of a small number of native human samples in a regular EQA survey can help to assess the suitability of routinely distributed processed EQA materials (EQAMs), demonstrated by testosterone, progesterone, and 17β-estradiol measurements in human sera.

Laboratory results for the native and processed EQAMs were comparatively analyzed using a mixed-effects model. This estimated the sample type-specific mean natural logarithm (ln) bias to the reference measurement value and mean differences in ln bias between the two sample types, each with a 95%-confidence interval. Applying quite strict commutability criteria defined in consideration of clinical guidelines enabled the identification of analyte/reagent manufacturer pairings exhibiting only minor differences in bias between sample types. For pairings showing differences in bias between the two sample types that exceeded the commutability criterion, efforts were made to differentiate between material- and MP-related contributions to the model outcomes - referred to as non-commutability and trueness bias, respectively.

Strongly scattered bias of the native EQAMs for some MPs revealed issues with measurement performances, such as inappropriate calibration or specificity. Consistent bias but distinct differences in bias between sample types indicated issues with commutability. The two-stage assessment revealed that the processed EQAMs were suitable for EQA purposes in most pairings. In two exceptions (testosterone/DG and progesterone/BE), the processed materials showed insufficient commutability. The limited number of EQAMs may result in inconclusive commutability assessments for individual analyte/MP pairings, four in this study, but allows for targeted follow-up with additional native EQAMs or comprehensive commutability studies.

Compared to conventional commutability studies, these proposed EQA comparability studies that use comparative native EQAMs offer EQA providers a more efficient and resource-conserving approach to assess the suitability and commutability of EQAMs by leveraging existing infrastructures. Nevertheless, even this more pragmatic approach remains resource-intensive and is inherently restricted to specific questions and selected analytes.