Questions

This question and answer covers the older version of our software, QC Validator. The concepts still apply with the use of our current version, EZ Rules 3.

This month's question comes from Gisborne Hospital, New Zealand:

In selecting a QC procedure for a calcium method which has a CV of 1.41%, according to QC Validator® 2.0, the rule of choice is a mean and range procedure (x_0.01/R_0.01) with N=6. How can I set up mean and range charts which are simple enough to allow bench workers to understand and use them?

The easiest way to apply mean and range rules when using 2 or 3 different control materials would be to first calculate a z-value for each control measurement (i.e., the number of standard deviations of each measurement from the mean for that material), then average the z-values for the group of measurements to obtain a mean z-value and take the difference between the highest and lowest z-values to obtain the range. The average z-value would be plotted on a mean chart and the high-low difference on a range chart. Control limits would be set using the factors found on page 140 of the QC Validator® program manual, e.g., plus and minus 1.05 for the mean z-value and an upper limit of 4.76 for the range.

In many laboratories, these calculations and charts may be too complicated to be practical in daily work, therefore the use of a multirule procedure such as 1_3s/2of3_2s/R_4s/3_1s/6_x may be more useful and will provide similar power, but at a somewhat higher false rejection rate (approximately 0.07 or 7% compared to 1-2% for the mean/range procedure).

The difficulty with calcium is often the tight requirement for quality (approximately 5% in many proficiency testing surveys outside the USA, compared to 10% by the USA CLIA PT criterion), which means a method with a CV near 1.0% is often needed. One way to achieve this analytical imprecision might be to perform calcium measurements in duplicate. The effect of duplicate measurements can be studied by entering a value of 2 for the "number of replicate samples analyzed". With duplicate measurements, a 1.41% CV, and a 5% quality requirement, a 1_3s rule with N=3 would be sufficient. This N=3 means that each of your 3 control materials would be analyzed in duplicate, just like all your patient specimens are being analyzed in duplicate.

If these alternatives are too costly, then consider whether the method is sufficiently stable that 50% error detection is satisfactory. By setting the frequency of errors to <2%, modifying the QC selection criteria to delete "90% preferred" (but keeping "50% OK for f<2%") and limiting N to 3, the program will look for rules for N=3 that provide only 50% error detection. Possible selections may be a 1_2.5s rule with N=3 or the 1_3s/2of3_2s/R_4s/3_1s multirule with N=3. When only 50% error detection is achieved, remember that your Total QC strategy should include a strong emphasis on preventive maintenance, instrument function checks, etc., to maintain a low frequency of problems.