Calibrator uncertainty (i.e., the uncertainty of the value assigned to the calibrator) and uncertainty of calibration procedure should not be mixed up

To the Editor,

In reading the letter by Åsberg and Skadberg [1], we noted that the Authors seem to mix up the measurement uncertainty (MU) of the values assigned to the end-user calibrator (u_cal) with the source of variation related to the calibration procedure of the in vitro diagnostic measurement procedure (IVD-MP) when the calibrator lot is changed. u_cal belongs to the MU of the selected calibration hierarchy and related higher-order references together with the MU associated to the manufacturer’s protocol for assignment of the unbiased values to the IVD calibrator [2], 3], while, according to the ISO 20914:2019 Technical Specification, the latter (i.e., the experimental variability that is introduced during each calibration procedure of the IVD-MP in the daily practice) is accounted for in the estimate of ‘long-term imprecision’ (u_Rw – uncertainty component under conditions of within-laboratory precision, obtained under defined conditions in same laboratory for a period sufficient to include all routine changes to measuring conditions, e.g., different lots of reagents, calibrators, and instrument maintenance; clause 3.40), derived from the internal quality control (IQC) data [3], [4], [5], [6], [7]. These are of course not the same source of MU and the sentence by the Authors “If the effect of u_cal is already accounted for in u_Rw, it should not be added again … as the square root of the sum of the relative variances” is mistaken.

What is demonstrated by the simulation provided by the Authors is not surprising, with the u_Rw increasing when more calibrator lots are used, confirming the experimental background of the ISO 20914 clause 3.40 asking for the monitoring of the measurement variability for an extended time period, which should include a suitable number of calibrator lots and calibration procedures, simulating what is currently happening in daily laboratory practice [8], 9]. If u_Rw can be derived from IQC (and the presented simulation shows the influence of one of the most important sources of variability, i.e., the change of calibrator lots), u_cal must include all uncertainties introduced by the selected calibration hierarchy for the measurand, beginning with the highest available reference down to the assigned value of the calibrator for the commercial IVD-MP, including the uncertainty of bias correction (u_bias), if a not negligible bias has been detected and corrected by the manufacturer when implementing metrological traceability (Figure 1).

Figure 1:

Sources of measurement uncertainty (u) across the entire metrological traceability chain contributing to the estimate of u of clinical sample [u(y)]. u_ref, uncertainty of selected higher-order reference; u_bias, uncertainty of the bias correction; u_{cal value assign}, uncertainty associated to the manufacturer’s protocol for value assignment to the end-user calibrator; u_cal, uncertainty of end-user calibrator; u_Rw, assay precision under intermediate reproducibility conditions obtained by an individual laboratory using the in vitro diagnostic measurement procedure.

In conclusion, the authors’ message as presented appears to convey a common misconception about the use of the reproducibility of a measurement result (even when important sources of measurement variability such as the change in calibrator lots are considered) as expression of the overall MU of a clinical sample measurement [10]. The underrated point is that this MU must always include (“combine”) the uncertainty contributions from each higher calibration step in the employed hierarchy. Particularly, the MU at the level of clinical sample results must be the combination of all MU contributions accumulated across the entire traceability chain (i.e., √[u_cal² + u_Rw²], where u_cal=√[u_ref² + u_{cal value assign}² + u_bias², if any]). Only the square root of the sum of the squares of all sources of MU gives the accurate evaluation of MU of clinical sample results. This is the main message of the ISO 20914:2019 guideline [3]. We understand that the information about u_cal from IVD manufacturers is sometimes difficult to obtain by end-user laboratories and not rarely incomplete, because, in practice, few manufacturers provide u_cal estimated as combined with u_ref of the selected higher-order reference, as indicated above. What IVD manufacturers are usually providing is just MU of value assignment of commercial calibrators (u_{cal value assign}), so that the medical laboratory should independently retrieve the corresponding u_ref and combine it to the former to obtain the correct u_cal estimate to be finally combined with u_Rw in the evaluation of MU of laboratory results [11].