Risk-Based versus Statistical Sampling
Defined, sampling is the process of taking a small but representative portion of a much larger stream, where the sample collected accurately represents the content of the larger stream. Effective sampling is a very important part of process validation in pharmaceutical manufacturing. Sampling is a mandated element of FDA Good Manufacturing Practices, but the FDA does not legislate a specific approach to establishing a sampling plan. Whatever approach is selected must be clearly and easily justified. We use two main sampling techniques in the industry today: a risk-based method and a statistical approach. What’s the difference between the two? In risk-based sampling, the design of the sampling plan is based upon sound principles and the experience of the Subject Matter Experts. Facilities using this method will have a baseline number for sample size based upon risk and performance, and that number can change based on prior inspection results – it may be reduced due to good results, or tightened due to poor results. Implementing a risk-based method can help companies spend less time and money, but it often makes sampling systems difficult to track. We also have the option of using statistically-based sampling to validate our processes. Rather than basing samples on risks and performance, this method gathers data from each individual lot from a probability standpoint to ensure that the sample is unbiased representation of the entire batch. This method may provide a more concrete representation, however, using this approach can prove difficult for companies that manufacture a variety of products. Risk-based and statistical sampling both have their benefits and drawbacks. So when should one method be used over another? Find out in just a few weeks at the ISPE/PQRI Process Validation Conference, where Global Lead Statistician James Crichton will speak on the most appropriate and effective uses for each method. Join us from October 7 – 8 in Silver Spring, Maryland. Register Now!
