Endocrine Abstracts

JOINT1226

Introduction: Reference intervals (RIs) for androstenedione should be specific to the population and immunoassay method. Generally, laboratories are responsible for verifying RIs established by an external source (e.g. a manufacturer) or determining their own. Initially, the laboratory adopted a manufacturer-defined RI for reproductive-aged women calculated by the direct method. As we noted many results above the upper reference limit, we needed to implement more suitable RIs for our population. We opted for an indirect approach using routine laboratory data stored in the laboratory information system. This study aims to compare (i) manufacturer-defined and calculated indirect RI and (ii) calculated indirect RIs from one and five years of data.

Methods: Manufacturer-defined 90%RI for reproductive-aged women was 1.71-4.58 nmol/l measured by Roche electrochemiluminescent immunoassay. All unique visits from women aged 20 to 45, with androstenedione, testosterone, sex hormone binding globulin (SHBG), and follicle-stimulating hormone (FSH) results, were extracted from the database. All patients with testosterone, SHBG and FSH outside the RI (0.29-1,67 nmol/l, 32-128 nmol/l, < 25.8 IU/l, respectively) were excluded. After the data-clearing step: (i) the initial data set included 520 subjects from one year of data, which is considered a small sample size for an indirect approach, and (ii) the subsequent data set included 2664 subjects from 5 years of data. Outliers were tested for each group separately and subsequently eliminated. RIs were calculated for the total age group (20 – 45 years) and subgroups (20-30 and 30-45 years) with the parametric method after Box-Cox transformation.

Results: For the first data set, the 95%RI in the total age group (20-45 years) was 2.48-8.89 nmol/l with a median of 5.13 nmol/l. The age-stratified 95%RIs were 3.02-9.43 nmol/l (20-30 years) and 2.23-7.75 nmol/l (30-45 years). The number of subjects was 518, 267, and 250, respectively. For the second data set, the 95%RI for the total age group (20-45 years) was 2,13-8,74 nmol/l with a median of 4.73 nmol/l. The age-stratified 95%RIs were 2.81-9.25 nmol/l (20-30 years) and 1.88-7.96 nmol/l (30-45 years). The number of subjects was 2640, 1524, and 1391, respectively.

Conclusion: A larger dataset drawn from a broader data range showed RIs comparable to those calculated initially from a smaller sample size. This study confirms significant differences between calculated and manufacturer-reported RIs. Both calculated medians exceeded the manufacturer-declared upper reference limit, indicating that more than half of our subjects would have shown elevated androstenedione concentrations had the manufacturer-defined RI been applied.