Basic QC Statistics

A total QC system must control both trueness and precision. The goal is for a test to have both low bias and low imprecision. Problems with imprecision are less likely to cause analytical failure in the modern laboratory due to computerization and mechanization of the analytical process. Therefore, bias is brought to the forefront.

The following illustration represents low bias and low imprecision as a target, where the center of the target represents the target value.

The illustrations below show two other possibilities.

• The illustration on the left represents a situation where the standard deviation (SD) for a test is small (good precision) but is shifted away from the target value (high bias).
• The illustration on the right represents poor precision but surprisingly, low bias because the average of the results is close to the target. Of course, individually none of the points are close, and their individual z-scores would reflect this.

Note: Bias and imprecision are most important at the clinical decision levels. For example, ß-hCG clinical decision levels are at low concentrations (corresponding to early pregnancy in the female and early testicular cancer in the male) or at moderate concentrations (to diagnose the progression of pregnancy).