Data Analysis & Performance Assessment

The purpose of External Quality Assessment is to compare performance of individual laboratories or test centres with their peers. EQA is a retrospective form of quality control - participants do not know the target range or expected results for their test, unlike internal quality control.

Target values

Performance for the FVIIIFIX treatment monitoring programme is based on the assay result (IU/dL) which is used to determine the z score, calculated as follows.  The central reference point is taken as the peer group median.  The robust mean and SD are calculated by statistical exclusion of outlying results. The z-score is then calculated as   

[participant result – mean] / SD

A z-score between -2 and +2 is deemed ‘within consensus’.  A z score >+/-2 is considered outwith consensus (warning); a z-score of >+/-3 is considered outwith consensus (action required).   

Performance is determined by comparison of individual laboratory results with the target value for each test. UK NEQAS BC uses median values determined from participants' results to determine the consensus or "target" values against which individual laboratory performance can be assessed. Use of the median avoids the effect of outlying results and the need to remove any outlying data from the analysis.

All participants not using the approved test methods/reagents or are found to be in a group of <10 users will have a comment on their report to highlight that it is not approved/ <10 users. These centres results are not included in the overall statistical median results calculation. Regardless of which reagent/assay kit method has been used, result performance is assessed against the z-score. All centres’ results will be included in the peer group statistical analysis and provided with a peer group assessed median, CV% and range for their reagent versus assay choice. 

We do advise participants who are using non-approved methods to compare their results with the median value of the FVIII product from the approved method/reagent overall median analysis.

Terminology

Results are assessed using statistical measures comparing individual results with peer group or overall consensus values. Therefore, a result falling outside the statistically defined terms is not necessarily a "poor" or "unsatisfactory" result. We deliberately employ the term "outwith consensus" to indicate a result that falls outside the statistically determined limits we have applied.