Factor assays: The percentage deviation is not used because the spread of results is too large e.g. a factor VIII assay could generate results ranging from 150 IU/dL. For these reasons an overall consensus median is determined as a central reference point and individual results are ranked into 5 unequal quantiles above or below the median. This is summarised in the illustration below:
A/a: The nearest 25% above [A] or below [a] the median value [i.e. 50% of the results]
B/b: The next 10% above or below the median [i.e. 20% of the results]
C/c: The next 5% above or below the median [10% of results]
D/d: The next 5% above or below the median[10% of results]
E/e: The furthest 5% of results above or below the median [10% of results]
'A-E' refers to values above the median and 'a-e' to values below the median.
This approach identifies centres furthest from the median although there are some inherent problems with this methodology:
- For imprecise assays there will be a wide ‘acceptable’ range e.g. in an exercise involving measurement of VWF:RCo an 'A' grade was awarded to any lab reporting a result between 63-96 U/dl.
- For more precise assays with a tighter range of results the acceptable range will be smaller e.g. in n exercise involving measurement of Factor X, an 'A' grade was awarded to any lab reporting a result between 46-52 U/dL.
- Finally because of this statistical approach 10% of labs in any test will get an 'E/e' grade.
The following example illustrates the results obtained for a FVIII assay and the distribution of grades:
Performance 'Outwith Consensus' and 'Persistently Outwith Consensus'
The terms 'outwith consensus' and 'persistently outwith consensus' refer to the grades and cumulative grades [in a number of exercises] that a lab receives for a specific test