Statistics can be used to monitor critical variable in the running of a procedure and in the performance of the instrumentation that performs it. By applying Nelson’s rules this data can be evaluated as performing as expected or performing in ways that are statistically not compatible with acceptable performance. This can often be done while the overall procedure is still within compliance, so the procedure can still be in use as intended.
Most laboratories monitor the results of a procedure by tracking the results of running standards and using a control chart. This, however, is a very simplistic and limited way of assessing performance. When the procedure fails, the laboratory is not in compliance. This is a very dire situation.
Many of the activities in the procedure follow Gaussian statistics. This means that choosing a metric for key activities or parts of the procedure can show the causes of the overall problem. The use of the 3 sigma occurrence as the only evaluation is also a poor approach. Statistically, there are many other tests for good Gaussian behavior. These are given by the set of Nelson’s rules. This allows monitoring symptoms and catching problems as they develop.