RESUMEN
BACKGROUND: Biological variation (BV) data may be used to develop analytical performance specifications (APS), reference change values (RCV), and support the applicability of population reference intervals. This study estimates within-subject BV (CVI) for several endocrine biomarkers using 3 different methodological approaches. METHODS: For the direct method, 30 healthy volunteers were sampled weekly for 10 consecutive weeks. Samples were analyzed in duplicate for 17-hydroxyprogesterone (17-OHP), androstenedione, cortisol, cortisone, estradiol, follicle-stimulating hormone (FSH), luteinizing hormone (LH), sex hormone-binding globulin (SHBG), and testosterone. A CV-ANOVA with outlier removal and a Bayesian model were applied to derive the CVI. For estradiol, FSH and LH, only the male subgroup was included. In the indirect method, using the same analytes and groups, pairs of sequential results were extracted from the laboratory information system. The total result variation for individual pairs was determined by identifying a central gaussian distribution in the ratios of the result pairs. The CVI was then estimated by removing the effect of analytical variation. RESULTS: The estimated CVI from the Bayesian model (µCVP(i)) in the total cohort was: 17-OHP, 23%; androstenedione, 20%; cortisol, 18%; cortisone, 11%; SHBG, 7.4%; testosterone, 16%; and for the sex hormones in men: estradiol, 14%; FSH, 8%; and LH, 26%. CVI-heterogeneity was present for most endocrine markers. Similar CVI data were estimated using the CV-ANOVA and the indirect method. CONCLUSIONS: Similar CVI data were obtained using 2 different direct and one indirect method. The indirect approach is a low-cost alternative ensuring implementation of CVI data applicable for local conditions.
