ISSUES IN ACCREDITATION
scientists another, robust means to assess and monitor the performance of our assays and the results being generated from them. Whether MU will be beneficial in informing clinicians of the clinical utility of results is another question. Analytical uncertainty and variation in reported results may arise from analytical bias, biological variation and MU, so it is clear that although MU is not the sole determinant of changes in results, even in the single patient, it is a factor that must be considered.2–10
Don’t we already do this with IQC and EQA? Routinely, IQC is used to assess, and detect at an early stage, analytical error propagated through patient results and suggests early intervention to suitable well-trained laboratory personnel. Analytical bias is routinely monitored as an ongoing process by internal and external quality control and assessment schemes. Corrective actions to minimise and eradicate analytical bias are required to be in place individually for each assay. Such corrective actions cannot eradicate MU. It contributes to all analytical results. As such, MU cannot be compared directly to an assay’s CV or the standard deviation of its IQC. Neither can bias measured by external quality assessment (EQA) performance describe MU on its own.
The process in general The process by which laboratory assay performance assessment is determined follows the same format irrespective of the assay, and is summarised in Figure 1: • develop a model of the measurement • define the measurand • list sources of uncertainty and determine the source of error
• calculate the imprecision uncertainty • combine all uncertainties • expand the uncertainty (Coverage factor) • determine inter-observer variation if appropriate
Measurement uncertainty is evaluated by two methods using the Guide for Uncertainty of Measurement (GUM) principles. These are termed Type A (the Top-Down approach) and Type B (the Bottom-Up approach).
Top-Down The Top-Down approach allows assessment of the entire system using only a final result produced by that closed system. This is best shown in automated assays that run IQC, and it is considered that all errors are contained within the variability of the results generated by the IQC.
Bottom-Up This is a very comprehensive approach whereby the contribution of each and every source of uncertainty throughout the entire process is identified and quantified. By mathematical combination according to
412 Define measurand Review process
Determine most appropriate method of analysis
Top-Down Collect data Bias present?
Yes
Recalibrate if applicable
No
Calculate imprecision
Calculate bias uncertainty
Combined uncertainty
Expanded/final uncertainty
Fig 1. Diagrammatic representation of the process of assessing uncertainty of measurement. The process involves three phases. Initial: A review of the assay and assignment into whether the model is best described using the Top-Down or Bottom-Up approach. Data Collection: Either analyse all results from the end process, such as IQC (Top-Down), or assign uncertainties to each individual component (Bottom-Up). Calculations: Calculate the imprecision and bias uncertainties and combine to give a combined uncertainty. Finally, multiply by a coverage factor to give and expanded (and final) uncertainty.
error propagation rules, these uncertainties are then combined to give an overall MU for the entire process.
Development of the model of the measurement This is performed entirely on an individual assay basis and is related to the methods/techniques used for the analysis. This includes an analysis of the sources of uncertainty and decision as to the most appropriate method by which analysis is to be undertaken. Having performed this analysis for all assays in haematology, the categories and assays associated with them are shown in Table 1.
Definition of the measurand The determination of the contribution of MU to a result depends on what we are telling people we are measuring. It is not sufficient for this purpose merely to state the name of the assay. A more detailed and informative description of the analyte and the system in which it is being assessed is required. The measurand is defined by: • the analyte (eg factor VIII) • the system examined (eg plasma, whole blood etc)
• definition of quantity (eg coagulant activity, concentration etc).
Additionally, it is preferable that the definition of the measurand includes a traceable SI unit. For many assays, however, this is not the case, such as for factor VIII assays where the unit of measure is iu/mL. Prothrombin time and other assays do, however, have SI-traceable units such as seconds.
Coverage factor This is exclusively quoted as 2, to equate to ±1.96 standard deviations (SD) from the result within which the results is expected to reside with 95% confidence
Review of the uncertainty of measurement Measurement uncertainty may be reviewed routinely every six months. In addition, unscheduled assessment must occur if one of the following changes in the measurement procedure is implemented. A change in: • technology • reagents • QC material • calibrators (and their associated lot numbers used for the measurement).
AUGUST 2016 THE BIOMEDICAL SCIENTIST Bottom-Up
Assign uncertainty for each component
Combine uncertainty by rules of error propagation
Calculations
Data collection
Initial assessment
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60