Manipulations and NONMEM Data Base Creation

Data Manipulations and NONMEM Data Base Creation 

Because patient height was missing for 28% of the data base, it was decided that only using patients with complete data (complete case analysis) would not be an option as this would decrease the sample size to only 70 patients. Furthermore, the percent missing was rather large and so simple mean imputation did not seem a 

Since some subjects were sampled on different occasions, the occasion on which each subject received their dose was determined so that BSV may be further partitioned into interoccasion variability (IOV). Once the data set was created, the first 19 subjects were removed and saved as a validation data set. This represented 20% of the total data. The remaining subjects were used for model development. Thus, the model development data set consisted of 250 observations from 78 patients, whereas the model validation data set consisted of 72 observations from 19 subjects. 

---