RANGE VERIFICATION

[gilmanch]

Hello everyone,

 

I have a question for what other places do to fullfill the CAP Hem 25870 stating: If commercially assayed controls are used for CBC instruments, control values correspond to the methodology and target values (mean and QC ranges) are verified or established by the laboratory. Note: most commercial controls have expected recovery ranges for each parameter, provided by the manufacturer. The mean of such ranges may not be the exact target value in a given laboratory. Each laboratory must assign its own initial target value, based on initial analysis of the material; this target value should fall within the recovery range supplied by the manufacturer, but need not exactly match the package insert mean. The lab must establish specific recovery ranges that accommodate known changes in product attributes, assuming the calibration has not changes.

 

Do you guys tighten the ranges? Make your SD smaller so the range is smaller? The package insert range is usually quite wide and if it is tighter problems from trends and shifts can be seen sooner.

 

I have tightned ours, but it seems it maybe too much?

 

Thanks for the help!

[nziegler]

This is the calculation Beckman gives users to determine their "Lab Limits" (aka: SD)

taking 10 months of data (6 would be fine), average your %CV for each measured parameter

avg %CV x package insert mean (usually roughly the same lot to lot) / 100 * 3 = 2 SD limit

 

I have no idea how they came up with that calculation. The only parameter the calculation won't work for is MCV because of the upward trend as the control ages.  The ranges will be tight, but like you said, you WILL catch shifts/trends/poor reproducibility easier.  I've caught diluent problems based on high MPV %CV and random low hgb results that wouldn't have been seen with a wider range.

 

For my mean lot to lot, I run about 10 runs of the new lot while finishing up the old lot. After eliminating the outliers (usually due to an empty vial) I set my target as my observed mean, and tighten the ranges to my "lab limits".

 

Hope that helps!

[SMILLER]

Simular to the above.  We also have Coutlers.  Once we reset the means, we use an SD based on the analyzer's expected performance.

 

Scott