Category: Basic QC Concepts | Estimated reading time: 8 minutes
Why Manufacturer Ranges Are Not Enough
A common conclusion reached once a laboratory begins taking QC seriously: manufacturer kit insert ranges cannot be used exactly as they are. Every laboratory has a unique combination of instruments, reagents, operator techniques, and environmental conditions — and that means we need to establish our own laboratory Target Value (TV) and SD.
The question is: how?
The procedure is called Preliminary Testing, and it is actually simpler than it sounds — four steps, sequential, each with a solid scientific rationale behind it.
Step 1: Measure Control Material 30× Under Normal Conditions
The number 30 is not arbitrary. In statistics, 30 measurements is the generally accepted minimum threshold to generate mean and SD estimates that are sufficiently stable. Below 30, SD estimates can be highly unstable — values from 10–15 measurements can differ significantly from the values that would be obtained if measurements continued to 30 or more.
CLSI EP5-A3 (2014) — the international standard document for the evaluation of laboratory method precision — recommends a minimum of 20 measurements for basic precision estimation, with 30 or more being a more robust figure. Many high-standard laboratories use 30 as a standard because it provides more reliable estimates.
Special note for hematology: hematology control materials have different stability characteristics than clinical chemistry. Some parameters — especially those based on live cells — have a shorter stability period, so 30 measurements in the same period might not be feasible. Refer to the ICSH (International Council for Standardization in Haematology) guidelines that are more appropriate for this context.
"Normal conditions" are just as important as the number 30 itself. What is meant by normal conditions:
- The instrument is not in a post-service state — post-service performance is not necessarily representative of daily conditions.
- The instrument has not just been calibrated — new calibration creates conditions "fresher" than the actual daily state.
- Operated by operators who usually run the instrument — not a supervisor or senior MT (Medical Technologist) who is not a routine operator, as their technique may differ from daily conditions.
Ignoring these normal condition requirements will result in a TV and SD that do not represent the actual daily performance of the instrument — and the QC system built upon it will not be responsive to real deviations.
Step 2: Calculate the Mean (Target Value) and SD
From the 30 measurements, the two most important figures are calculated.
Mean (Target Value / TV) — the mean of these 30 measurements becomes our laboratory's Target Value: the central value reflecting our laboratory's specific measurement conditions, rather than the values from the manufacturer's kit insert. This TV becomes the center line on the Levey-Jennings chart — all daily QC evaluations are based on the distance of values from this TV.
SD (Standard Deviation) — reflects the normal variation of the instrument under standard operational conditions; a measure of how much fluctuation we can expect from day to day under non-problematic conditions. Westgard (2016) emphasizes that the SD used in a QC system must be realistic and representative — not too small (because it was measured under "perfect" conditions), and not too large (because there are outliers that haven't been excluded).
Two things need to be checked before the SD is accepted as the final value:
- Check for outliers. If there are measurements with values more than 3SD from the mean, check if there were any abnormal conditions when that measurement was performed (newly opened control material, technical interference, etc.). If there is a clear reason, that value can be excluded and replaced with a new measurement. If there is no reason, that value may reflect the natural variation of the instrument and should likely be kept.
- Check the CV. The CV from the preliminary test should be compared with Biological Variation (BV) or the allowable CV based on TEa. If the CV already exceeds the allowable limit from the start — even before the QC system is running — there is a precision issue that needs to be resolved before proceeding to the next step.
Step 3: Calculate CV, UCL, and LCL
From the established TV and SD, three derived values are calculated:
CV (Coefficient of Variation) — a relative expression of SD, stating how much variation exists as a percentage of the mean. This makes the CV easier to compare across parameters and instruments because it does not depend on the unit of measurement. The CV here is our laboratory's operational CV — which will be used in the calculation of TE (TE = |d%| + 2CV) and Sigma metric.
UCL (Upper Control Limit) and LCL (Lower Control Limit) — the ±3SD control limits that become the hard rejection boundaries in the Westgard system (1₃s rule). A QC value falling outside the UCL or LCL is an immediate signal for rejection. Statistically, if the data follows a normal distribution, only 0.27% of measurements are expected to fall outside ±3SD under stable conditions — so if a QC value is outside the UCL/LCL, it is highly likely that something has actually changed in the process.
Step 4: Compare TV with Manufacturer Range
This is a verification step that is often skipped — even though it is very important.
After the laboratory TV is established, compare it with the range listed in the manufacturer's kit insert:
- If the TV is within the manufacturer's range → proceed. Our TV is consistent with the characteristics of the control material stated by the manufacturer, and the QC system can be built on this TV.
- If the TV is outside the manufacturer's range → investigate first. There is an inconsistency between our laboratory conditions and the conditions expected by the manufacturer — it doesn't mean either party is wrong, but there is a difference that needs to be understood before the QC system is implemented. Investigation includes: verifying instrument calibration, verifying control material storage conditions, verifying reagent lots, and if necessary, contacting the manufacturer for clarification.
One important thing: even if the TV is outside the manufacturer's range, it does not mean our TV is wrong. In some cases, our laboratory's specific conditions consistently produce values slightly outside the manufacturer's range — and as long as those conditions are stable and reproducible, the QC system built on our TV remains internally valid.
When Must Preliminary Testing Be Repeated?
There are three situations that require preliminary testing to be repeated:
Change of reagent lot — a new lot, even from the same brand and catalog, can have slightly different characteristics. These differences are small but consistent, and can shift the TV significantly. Without re-verification, this shift will not be detected until it has been running for quite some time.
After major maintenance — major repairs to an instrument change its technical conditions. A new calibration after maintenance does not necessarily return performance exactly to how it was before. Repeating the preliminary test ensures that the TV and SD being used reflect the instrument's post-maintenance condition.
CV suddenly spikes — a spike in CV that cannot be explained by current conditions is a signal of a fundamental change in instrument performance. Continuing with old control limits in this condition can provide a false sense of security — QC looks "in" because the limits are too wide, even though the instrument performance has actually declined.
Conclusion
Preliminary Testing is not a complicated procedure. But it is fundamental — because everything built upon it, from Levey-Jennings charts to Sigma metric evaluation, is only as strong as the foundation provided by the TV and SD established here.
Laboratories that skip this step — or perform it incorrectly — are building a QC system on an unstable foundation. It may look like it is running just fine. But when a real problem occurs, that system may not be sensitive enough to detect it — or it may be too sensitive, triggering false alerts too often.
Four steps. Thirty measurements. One solid foundation.