ABSTRACT
Unlike most drug products, some drug products must be stored at specific temperatures, such as −20◦C (frozen temperature), 5◦C (refrigerator temperature), and 25◦C (room temperature), to maintain stability until use (Mellon, 1991). Drug products of this kind are usually referred to as frozen products. Unlike the other drug products, a typical shelf-life statement for frozen drug products usually consist of multiple phases with different storage temperatures. For example, a commonly adopted shelflife statement for frozen products could be 24 months at −20◦C followed by 2 weeks at 5◦C. As a result, the drug shelf-life is determined based on a two-phase stability study. However, no discussion of the statistical methods for estimating two-phase shelf-life is available in either the FDA stability guidelines or the ICH stability guidelines. Mellon (1991) suggested that data obtained from the two-phase stability study be analyzed separately to obtain a combined shelf-life for the frozen products. Mellon’s method does not account for the fact that stability at the second phase may depend on the stability at the first phase. That is, an estimated shelf-life at the second phase following 3 months of the first phase may be longer than that following 6 months of the first phase. To overcome this problem, Shao and Chow (2001a) proposed a method for a two-phase stability study using a two-phase linear regression based on the statistical principle described in both the FDA and ICH stability guidelines.
