Sampling Strategies for Process Validation of Low Volume Devices


Hello - I am using acceptance sampling plans (Confidence and Reliability) for the process validation of medical products. I typically use sampling plans with 95% confidence and reliability levels from 90% to 99% based on the risk of the product characteristic. This approach works well for medium to high volume production since the sample sizes, especially for attribute sampling plans can be high. Has anyone in the group used a different statistically based approach to sample sizes for low volume production? Say a company builds 100 pieces a year. It is pretty challenging to run a PQ run requiring 299 samples. Any ideas on how to approach this?


Involved In Discussions
First you could decide to measure 100% of the products. In this case, you do not need to validate the production process -- though you will need to qualify your measurement process.
Second, if your response variable is continuous, you could assume a parametric distribution and base your sample calculation and the evaluation on this assumption. Often a normal distribution is assumed and this method is referred to as the k-factor method. In fact, this is equivalent to calculating a P_{pk} value.
Top Bottom