Quality Management

QC Validator is a software program that chooses control rules to meet any quality requirement you specify. Version 2.0, described in this lesson, has automatic QC selection with user-defined selection criteria and logic. If you want to know more about its features and possibilities, read this lesson.

Please note: EZ Rules 3 is now the current version of QC Design software offered by Westgard QC. The features described in this article are included in EZ Rules 3, along with substantial, additional improvements.

QC Validator^® is a computer program that helps you select statistical QC procedures that are appropriate for the performance of the methods and the quality required for tests in your laboratory. The program runs on IBM compatible PCs that operate under Windows® 3.1 or Windows® 95.

How do you decide what control rules and number of control measurements are right for a test in your laboratory?

Remember, a QC procedure is an error detector with an alarm. It functions much like a fire alarm, giving both both true and false alarms. Ideally, you want to tune the detector so it gives mainly true alarms, i.e., it goes off when there are errors that are important to detect, but is not so sensitive that it goes off randomly when nothing is wrong. QC Validator^® provides the information about the expected rate of alarms for most of the common single and multirule procedures of interest and the range of N's that are practical in healthcare laboratories.

The heart of the program is a data-table that shows the probabilities of rejecting analytical runs when different control rules and different N's are used. In terms of probabilities, you would like to reject runs whose performance is unstable and causes medically important errors, which means achieving a probability of error detection (P_ed) of 0.90 or greater. P_ed will get higher as you increase N, narrow the control limits, and add rules in a multirule QC procedure.

You would also like to accept all the good runs whose performance is stable. Unfortunately, some good runs also get rejected, thus you have to be careful to pick control rules and N's to maintain a low probability of false rejections (P_fr), preferably 0.05 or less. P_fr decreases as you decrease N, widen the control limits, and go to fewer rules in a multirule procedure.

Your job in selecting control rules and N's is to balance the error detection and false rejection characteristics, keeping P_ed high and P_fr low. QC Validator^® can help you do this by showing you the P_eds for medically important errors and the P_frs when no errors occur except for the inherent random error of the method (stable imprecision). This information is provided through graphical tools, such as power function graphs, critical-error graphs, and OPSpecs^® charts.

To use the program, you enter parameters describing your method imprecision and inaccuracy, define the analytical or clinical quality required by the test, then QC Validator^® calculates the size of the medically important random and systematic errors. You then use the graphical tools to evaluate QC performance and pick the best rules and N's.

Version 1.1 of the program made these graphical tools available for manual application, i.e., you select the graphs from a menu and look to see what performance is possible. You try different control rules and N's until you find the performance you want.

Version 2.0 provided an automatic process that sorts through the control rules and N's and identifies the best choice. The automatic process is governed by a set of selection criteria and logic that can be edited and changed to meet your preferences for the type of control rules that can be implemented in your laboratory, the range of N that is economical, the level of false rejections that is tolerable, and the level of error detection that is desired.

Both versions documented the choices of control rules and N's with both electronic files and hard copy reports that included the graphs and charts you selected.