A run of control results, before anyone reaches for a formula

A Level 2 sodium control runs five times in one morning on the same indirect ion-selective electrode analyzer, same reagent lot, same operator, no specimen involved: 139.8, 140.1, 140.0, 139.9, and 140.2 mmol/L, assigned target 140.0 mmol/L. None of the five results carries a flag. Looking at the numbers alone, the run looks fine, but 'looks fine' is not a number a laboratory can act on or record.

Common descriptive statistics answer what control performance actually looks like: where is the center of this run, how spread out are the results, and how far does any one result sit from the rest. Those three questions have names: mean, standard deviation (SD), and a standardized distance. A fourth question, how spread out is a result relative to its own size, has a name too: coefficient of variation (CV). A fifth question, how far the run's center sits from the assigned target, is bias.

None of these numbers alone say the run is acceptable. Acceptability depends on a laboratory's own acceptance limits and QC rule set, which are a local policy decision layered on top of the statistics, not produced by the statistics. Working through the numbers first gives that decision something real to stand on.

Before a number gets a name, plot the run and ask what you are looking at; a mean and an SD describe a specific, defined dataset, not a universal fact about the analyzer.