Statistical analysis for long-term stability studies with multiple storage conditions

To determine the shelf life of new medicinal products, long-term stability studies are conducted of at least three registration batches at multiple storage conditions. It is common practice to perform a statistical analysis on the resulting data separately for each storage condition. Although this is a straightforward approach to evaluating the data at each condition, the opportunity exists to combine all of the storage conditions into one single combined analysis that yields benefits without compromise to the estimation of the effect of each storage condition separately. Present external guidance provides little or no direction on treating multiple storage conditions in one combined statistical analysis. Because it has been underexposed in literature, this article presents a combined analysis, and discusses in detail the use of initial results, the imposed restrictions on batch intercepts, and poolability of batches across storage conditions. A case study is presented to demonstrate the differences with an analysis per storage condition and to show the advantages with respect to shelf life estimation.