Compromise in uncertainty estimation by modelling and validation approaches for an HPLC-UV method for measurement of biochemical indicators of vitamins A and E

It is common practice nowadays to associate the measurement uncertainty to the measurand, in order to judge the quality of a result related to the measurement process. However, the improvement of this parameter as well as the adaptation of its estimation modes always remain an analytical challenge, especially in chemical testing. In this paper, we outline a measurement uncertainty estimation mode based on the one-point linear calibration equation to fully establish a "bottom-up" approach for estimating the measurement uncertainty of a multi-point calibration-based HPLC-UV quantitative method. To demonstrate this estimation mode, we have followed as an example of interest the influences resulting from the simultaneous determination of two biochemical indicators, namely the human plasma vitamers retinol and α-tocopherol. Results from this estimation showed consistency when compared to those obtained from the validation-based alternative method, where the relative expanded uncertainties were found, at a 95% confidence level, to be less than 15% for the low concentration ranges of the two molecules. However, the modelling approach shows all the benefits of its use to identify and quantify all the uncertainty contributions arising from the different steps of the analytical process and seems to be quite achievable for comparative HPLC methods.