Patient Age Into the Model

age is identified as important when the age range in the database is large. Data pooled from Phase 1 studies, where the typical age is from 18 years to the mid-40s, has insufficient signal to identify age as a covariate since the range is so narrow. But, when data from Phase 1 is pooled with data from elderly subjects or pediatric patients, then the signal is usually large enough to detect an age effect, if it is important. 

Age can be treated as either a continuous variable in the model, wherein it is typically centered around the mean or median, or it can be treated as a categorical variable. When age is treated as a continuous variable in adults, its value is typically an integer. So if someone was 

32 years and 11 months old, they would be treated as simply 32 years old. This truncation is not problematic for adults. But truncation error becomes significant as patients get younger and younger so that age is usually not truncated with pediatric patients. Indeed, for the very young, a few months may have a significant impact on the pharmacokinetics of a drug. 

Categorization of age can be done using regulatory documents as guidelines. The (1994) ICH Guidance on studies in pediatric patients (Ell) breaks down pediatric patients into four groups  

An important concept in the absorption modeling of any extravascular administered drug is that the absorption process is independent of distribution and elimination. Hence, in accounting for these different formulations all that needs to be accounted for is the absorption process because distribution and elimination is usually the same regardless of how the drug is absorbed. Hence, different formulations are usually accorded their own absorption model through a series of IF-THEN statements with one formulation being the reference formulation. Thus, a solution and capsule may be modeled as 

---