A comparison of linear and exponential models for drug expiry estimation

While a large portion of pharmaceutical stability data is known to follow an exponential model decay, linear modeling of this data for expiry estimation is the norm. Expiry predictions based on linear and exponential fits to stability data were made to estimate the bias due to the linear fitting. It was found that within the usual expiry limits on drug potency, the difference between the model fits is relatively trivial. In cases of loss of potency greater than 15%, small assay variability, or great disparity between the length of the study and the time of expiry prediction, however, there is a nontrivial difference in the predictions and the exponential model is preferable.