Report of the Norwegian EQA Validator Workshop
Dr. Heidi Steensland, head of the the Norwegian EQA, presents us with the results of a QC Validator workshop. Over 40 QC Validator users from more than 25 laboratories in Norway gathered to discuss the use and application of the QC Validator program. They shared their results with us, and now we share them with you.
Norwegian Clinical Chemistry EQA Program
Department of Clinical Chemistry
Ullevaal University Hospital
Kirkeveien Oslo 0407
Note: QC Validator has now been replaced by EZ Rules 3. All functions described here are also available with EZ Rules 3.
As the two elements of quality, EQA and IQC, are closely related, the EQA organisations often also engage in IQC. The Norwegian EQA organisation (NKK) has frequently touched upon IQC issues during its annual meetings. Thus the 1997 meeting spent one day presenting the basis for Westgard's Validator program as well as his support via Internet, - and as a follow up NKK organised a common purchase for 25 labs who were interested. So, prior to NKK's 1998 meeting, before Christmas, 42 participants were gathered in a Validator workshop to share experiences and views on how to use the Validator to establish/remodel the IQC program. So here is how the meeting went:
Pål Rustad is NKK's statistician and works in a large outpatient lab in Oslo (Fürst Medical Laboratory). His approach to IQC is somewhat different from those of us who run hospital labs with immediate release of results all the time. Pål starts out with the analytical quality that his various analysis systems actually have. In his lab the analysis work is mainly done batchwise so he much prefers mean and range rule. The probability for false rejection and number of control measurements are chosen which determines the control limits. By choosing probability of detecting the critical systematic error, the critical systematic error is calculated together with TEa. This in turn could be compared to different proposed limits for TEa.
The other approach is first to define the total allowable error and go from there, and Pål outlined these two approaches. In practice we do a little bit of both, playing back and forth, to find out what quality we can actually achieve, or in other words what error we will have to live with (the "minimal achievable error"). This is very nicely visualised by using the Validator.
Kristine Solem reported from a working group at the University Hospital in Trondheim; she decribed the process of getting started with a real time process control program, and how to involve and teach the personnel in a big lab. She also told how they had manipulated the Hitachi IQC input parameters to show the Levy Jennings plots according to their (simple) control rules.
Tove Aas (Vestfold Central Hospital) has two Vitros instruments, and she presented her approach to IQC on dry chemistry systems, emphasising the various factors that had to be considered (Calibration stability, change of electrolyte solution, slide box turn over, etc). Like Kristine, she had also defined the limits in the Vitros IQC program (F2 and F3) to give alarm according to her rules.
Lars Eikvar at Ullevaal University Hospital went through practical factors/limitations that influence the planning of the IQC program for the Roche Integra instrument, - the use of same set of control rules for all tests, occasional bias due to lot variations of calibrators, problems with on board evaporation from control samples, etc.
Both Kristine, Tove and Lars discussed these practical limitations very nicely, together with the fact that the labs want to release the results to the ward as soon as they are available from the instrument. But they all pointed out that the most difficult part of the planning process was the establishment of TEa. What quality is "good enough" to satisfy the medical needs and at the same time take into account the routines and the technical limitations in the laboratory? And, - important when we consider the recommendations for TEa in the literature is that we generally claim that we have traceable calibration, and therefore the target value of the control material is set by the lab prior to its use, which means that the bias is set to zero. So, the discussion of TEa was the recurrent question of the day!
Sveinung Rørstad (Sandefjord klinisk kjemiske laboratorium) told us how he had used the Validator to establish an IQC program for the outpatient lab where he works, - as the lab was applying for accreditation (which they actually got!). Sveinung raised one question about the Validator, where only one CV (i.e. total over a month) is registered. The Validator program does not ask for within and between series CV in the application of mean and range rules. So Sveinung also used graphs from Tove Dreyer and Per Hyltoft's text book, Kvalitet og Kontroll i Klinisk Kemi, Nordkem 1990 (in Scandinavian) in addition to the Validator.
Also a short presentation of the QC Selection Grid was included, as the Grid may still serve a useful check that you are on the right spot.
It was indeed nice to notice the enthusiasm around the IQC planning process, - the Validator being the useful guide to the understanding on how the statistics actually work. IQC has always been a mysterious, impenetrable problem nagging us with an uneasy feeling of something undone. This was a good push....we hope that the process will go on!