Considerations when determining a routine sample size for a retest procedure

During an evaluation of a pharmaceutical drug substance or drug product, a result may be obtained that is unusual when compared with the expected result. This kind of result is referred to as a suspect result and is defined as a result that does not meet a prespecified criterion. When a suspect result is observed, a thorough analytical and manufacturing investigation is conducted to assess whether the result is due to an assignable cause. In some cases, an assignable cause is found, and the appropriate actions are performed. In other cases, the investigations cannot find any assignable cause, and additional testing is required. Recently, regulatory agencies have focused intensely on how investigations of suspect data are performed and documented. This article provides a rationale to assist in determining the number of nonsuspect retest results that must be generated to overcome an initial suspect result. Scientists should choose the number of retests that will allow them to adequately discriminate between material of acceptable and unacceptable quality.