Sampling Techniques for Data Audits

The data audit is a critical element in any quality assurance program and an area in which considerably more methods development is needed. To assist the QAU in organizing the requirements of the data audit for bioassay data and in developing a more rational and flexible sampling strategy, a decision tree has been designed. Surprisingly few QA auditors are aware of the many options available in performing a data audit. The general rule of thumb is the 10% audit, but other sampling plans may best represent the potential error rate of the whole data set. Presented here is a step-by-step approach to assist the quality assurance unit (QAU) in organizing the requirements of the data audit for bioassay data and in choosing the best available option. Steps in the development of a sampling plan include (1) defining, identifying, and characterizing data, (2) determining what constitutes an error and how errors are categorized, and, if possible, (3) specifying acceptable error rates.