Comparison of univariate and multivariate methods used to determine reference ranges

Kaynar, E.K., Comparison of Univariate and Multivariate Reference Interval Methods, Biostatistics Master's Thesis, Ankara, 2021 The multivariate reference interval method was proposed in the clinical laboratory literature over three decades ago as an alternative method to univariate reference intervals. Early articles on the subject summarized that the multivariate reference interval method was effective in reducing the rate of 'false positive' results found using univariate reference intervals. However, despite the fact that it was recommended for many years, the multivariate reference interval method remains an uncertain method in clinical practice. This study proposes two different approaches to reduce the false positive rate in cases where the physician must decide, due to the uncertainty of the multivariate reference interval method recommended to be used when evaluating more than one interrelated analyte, and the difficulty of implementation. In this study, a simulation study was made in R studio program under different scenarios for the related Ferritin and Transferrin Saturation percentage variables used in the diagnosis of anemia, and the reference interval methods currently used in practice were compared with the recommended reference interval methods. By developing a Shiny application, it is aimed to easily implement the suggested methods by users.