Required section · Section 5 of 6
Classify four QC data sets
Four glucose Level 2 QC data sets share the same assigned target (100 mg/dL) and the same allowable difference (plus or minus 8 mg/dL at this concentration). Only precision and bias vary between them. Use the mean, SD, %CV, and bias in the table to classify each one and to decide what to do next.
Data set A: mean 100.3 mg/dL, SD 1.2 mg/dL, %CV 1.2%, bias plus 0.3 mg/dL (plus 0.3%). Data set B: mean 100.4 mg/dL, SD 4.6 mg/dL, %CV 4.6%, bias plus 0.4 mg/dL (plus 0.4%). Data set C: mean 109.0 mg/dL, SD 1.3 mg/dL, %CV 1.2%, bias plus 9.0 mg/dL (plus 9.0%). Data set D: mean 109.2 mg/dL, SD 4.8 mg/dL, %CV 4.4%, bias plus 9.2 mg/dL (plus 9.2%).
A tight SD and %CV, on their own, describe precision only. A bias that exceeds the allowable difference, on its own, describes an accuracy problem only. Data sets A and B stay inside the plus or minus 8 mg/dL allowable difference; data sets C and D exceed it. Data sets A and C are tightly clustered (low %CV); data sets B and D are widely scattered (high %CV). Put the two axes together and each data set lands in one quadrant of the target board from the mental-model section.
The next investigation depends on which axis is the problem. A precise but biased pattern (data set C) points toward a systematic cause: check calibration, compare against a reference or comparative method, and check for a reagent lot or calibrator change. An imprecise but accurate pattern (data set B) points toward a random cause: check specimen handling, mixing, pipetting, and run repeat testing before touching calibration. A pattern that is both biased and imprecise (data set D) may need both investigations, and the order (fix the shift first, then re-assess spread) matters because a systematic offset can widen apparent scatter if it is drifting during the run.
Classify precision and bias separately before choosing an investigation, because a calibration check will not fix random scatter and a repeat run will not fix a systematic offset.
| Data set | Mean (mg/dL) | SD (mg/dL) | %CV | Bias (mg/dL) | %Bias |
|---|---|---|---|---|---|
| A | 100.3 | 1.2 | 1.2% | +0.3 | +0.3% |
| B | 100.4 | 4.6 | 4.6% | +0.4 | +0.4% |
| C | 109.0 | 1.3 | 1.2% | +9.0 | +9.0% |
| D | 109.2 | 4.8 | 4.4% | +9.2 | +9.2% |
Ordering exercise
A QC result flags outside its limit. Put the investigation steps in the order that avoids wasted work and avoids an unsupported conclusion.
1. Verify calibration or run a comparison check
Only after a genuine, reproducible offset is confirmed does a calibration or comparative-method check make sense; running it first wastes time if the cause was clerical or random.
2. Compare the pattern against QC history
Plot the new result against recent QC data to see whether this looks like ordinary scatter, an abrupt shift, or a gradual drift.
3. Rerun the same control material
A single repeat distinguishes a one-time miss from a reproducible pattern.
4. Check for a clerical or handling error
Confirm the correct control level and lot were used, the dilution and expiration are correct, and the result was transcribed correctly before assuming an analytic problem.
Knowledge checks
Reading and checks are open. Sign in only to save.
Knowledge check 1
Knowledge check 2
Section status
Finish this section
Reading and checks are open. Sign in only to save.
The module finishes after every required section is marked done and every check in those sections is correct.