Non-clinical pharmaceutical statistics

Non-clinical statistics is the application of statistics in pharmaceutical R&D and manufacturing and that is not directly related to planning and analyzing clinical trials.

There are numerous statistical issues to deal with in these areas, and often regulatory expectations require the use of statistical methods.

In this episode we discussed example areas for Non-Clinical Statistics in the Manufacturing and Quality area (also known as Chemistry, Manufacturing and Controls (CM&C))

Analytical Methods for assessing quality
- Understanding the science and technology is important (understanding how the data is actually generated or "how data happens").
- Determining appropriate specifications for a drug product and ensuring that the natural variation in the measurements will not lead to failing specifications.
- Qualifying and determining the defined content of chemical reference standards (using appropriate experimental designs)

Ongoing Manufacturing Processes
- Using control charts for on-going monitoring of manufacturing processes
- Understanding sources of variability that impact the output of the manufacturing process
- When something goes wrong: for example there is a bad batch that fails to meet its quality specification, and when this happen, an investigation has to be conducted, and statistical thinking, data gathering and combining, data visualization, data mining and experimental design can be used as part of the investigation
- Building tools to automatically build data sets from manufacturing data and visualize that data
Designing new manufacturing processes
- Applying the principles of good process design.
- Highly recommend book for any statistician: Juran on Quality by Design: The New Steps for Planning Quality into Goods and Services
- More on this will be covered in a later episode.