The Analytical Column

Most working labs run a check sample day after day and plot its results, or should: a quality-control (QC) sample of known value, its results on a chart with a centre line and a pair of limits. While the points stay inside the limits, the method is treated as in control and the samples are released. A point beyond the action limit, or a pattern that breaks one of the chart's rules, means you investigate before treating the run as normal; your quality system says whether work pauses while you do.

That is how the chart is usually read, and it leaves most of its information on the table. In my experience, overlaying the lab's temperature on a QC chart often shows the results moving with it: the chart meant to show whether the method is fit for use is partly tracking the weather. This piece is about reading the chart for what it is telling you.

How the chart is drawn

First, what you are looking at. Each point is one result for the QC sample, in time order. The centre line is the average of those results over a period when nothing unusual was happening. The limits come from how much the results normally move from one run to the next in that same period; a common way, for one result at a time, is to use the average difference between consecutive results (the moving range) to estimate the standard deviation, and to set the action limits at three of those standard deviations either side of the centre line. Your method or practice gives the factor that turns one into the other. Limits drawn from a period that included problems come out too wide and hide the next problem.

A chart like this answers one question: is the method behaving the way it did when the limits were set? It does not tell you whether the method is right. A method can sit calmly inside its limits and still be biased; that is the job of a certified reference material, drawn on the chart as a separate reference line (the piece on whether the calibration is still holding covers this).