Quality Control

On the Stats tab, in the Assay Quality Control group: Click Quality Control to monitor the results of repeated assays of standard samples. Shewhart and CUSUM control charts can be generated to help verify whether or not a process is operating within predefined limits.

 

Primarily, these functions are provided so that you can monitor an assay laboratory’s process, but may also be adopted for any process where there is a measurable variable which can be assumed to have a normal distribution when in a state of statistical control.

Normally, standard samples are submitted to a laboratory together with a batch of samples. The assay results for the standard samples are plotted on the control chart over time. If the assay results of the standard samples exceed defined limits, this will be apparent on the control chart and you can take corrective action.

Variation in quality or assays of standard samples can be attributed to two causes:

• Random variation: over which little control can be exercised other than by changing the processing method.
• Assignable causes: over which some control can be exercised. These include the quality of process materials, operator skills, etc.

If repeated assays of a standard sample are such that they come from a single normal distribution, having a mean within a specified range, the whole process can be said to be "in control". If variation from one or more assignable causes is present, the process is said to be "out of control".

Data can be displayed on Shewhart Mean, Shewhart Range, or CUSUM charts.

A Shewhart Mean chart displays the sample group mean, that is, the mean of the current assays will be plotted on the Y axis of the chart. It measures how closely the mean of the individual group corresponds to the expected mean (accuracy). The precision of this method relies on the assays in the group mean having a small range. This chart will quickly show large deviations from the mean since a sample group that has gone out of control can be identified without any reference to previous samples.

On a Shewhart Range chart, the group range is plotted on the Y axis. This is a measure of the precision or the spread of values in a group. The values of the group could have a small range (high precision) but the mean of the group can still differ markedly from the expected mean (accuracy).

On a Cumulative Sum (CUSUM) chart, changes in the mean are detected by keeping a cumulative total of deviations from a reference value. The CUSUM chart quickly shows small deviations from the reference. However it usually takes two or three observations to see a change on a CUSUM chart. CUSUM charts are good for detecting gradual changes in a process over time.

Whether you choose one of the Shewart charts or a CUSUM chart, will depend on the degree of variation you anticipate, the urgency with which you need to identify a loss of process control and the type of process you are monitoring. Both types of Shewart chart should be used in order to understand the precision and the accuracy of the sample group.

For a detailed description of quality control concepts, refer to:

• Wetherill G. Barrie (1982) Sampling Inspection and Quality Control, 2nd Edition, London, Science Paperbacks, Chapman and Hall. Numerous other publications are available on this subject.

The Process

The following process describes the steps you would take when monitoring the assay process for a laboratory.

1. Create a file with a structure that will suit your sample quality monitoring plan.
2. Submit your standard samples along with a batch of samples to the laboratory.
3. When they are returned from the laboratory, enter the assays in your file for each of the standard samples.
4. Decide which of the control charts you want to display. Enter parameters for that chart, then run the program to display it. You may want to run both the Shewart and CUSUM charts.
5. Evaluate the assay laboratory's process based the evidence.