Predicting shelf-lives of pharmaceutical products

The computer simulation package SIMAN^® is used to carry out Monte Carlo simulations with a view to estimating and predicting the shelf-lives of pharmaceutical products. The input data take account of assay variance and low extents of decomposition typical of data sets submitted by pharmaceutical companies in support of their product licence applications. It is shown that Monte Carlo methods provide estimates of the predicted shelf-life with a narrower and more symmetrical distribution than obtainable with integral methods. The results indicate that the median may be a more reliable estimate of shelf-life than the mean particularly if assay variance is high. Despite its usefulness SIMAN is a difficult package to use and is not generally recommended.