MD&DI February 1997 Feature Article
FDA Makes Quality the Rule (continued)
12. STATISTICAL TECHNIQUES
The requirements of the section on statistical techniques (subpart 0) have not changed from the 1995 working draft. Sampling plans must be based on statistical rationales, and statistical techniques must be used when appropriate for controlling processes and product characteristics. The section on sampling plans has been amended to add the phrase "to ensure that when changes occur the sampling plans are reviewed" in order to make it clear when sampling plans should be reviewed (820.250(b)). However, sampling plans should also be reviewed for change in other situations, such as when they appear to be ineffective.
Since the phrase "where appropriate" is used, there is an opportunity for an investigator to second-guess the manufacturer concerning the use of statistical methods. For example, if a manufacturer is not using statistical process control (SPC) for a continuous milling process, the investigator may question why not. To avoid such situations, manufacturers should review all processes and determine which should be subject to SPC. Where such methods are not used, but could apply, a written rationale should be provided for not applying statistical controls. This should be done as part of the facility's process validation strategy and plan documentation.
