Background: The conventional approaches to analyse scarce paediatric PK data with rich adult data often result in the estimated central tendency of the paediatric PK parameters being over-influenced by the adult data.
Aims: The purpose of this study is to develop a Bayesian modelling approach with balancing informative prior to adequately analyse the limited PK information in the paediatric age groups together with rich data from adults, leading to the development of robust optimal dosing regimen for paediatric population.
Methods: A balancing informative prior selection approach within Bayesian analysis set-up is developed to obtain robust confidence bounds for individual PK parameters across age groups with a special consideration for scarce paediatric data. From a data set of 570 subjects (1-25 years) treated with tobramycin , 8 subsets of 63 adults (18-25 years), 3 children (10-13 years) and 3 infants (1-2 years) with cystic fibrosis (CF) were randomly selected for the model building. The resulting distributions of individual PK parameters and the corresponding optimal doses across age were compared with the results obtained from the analysis of 570 subjects. The analyses were performed using Winbugs 1.4.3 and R 2.15.0.
Results: The estimated 95% posterior prediction intervals for individual PK parameters from all subsets of data covered the individual estimates for the whole data by 97 to 99%. The corresponding predicted minimum optimal dosage (11 mg/kg) based on limited paediatric PK information was same as that was estimated from the complete data of 570 subjects.
Conclusion: This study demonstrates that using a very limited paediatric and rich adult PK information, one can still conservatively predict the individual PK parameters across age and produce a robust optimal dosing recommendation.
1, Hennig S, Standing JF, Staatz CE, Thomson, A. Population pharmacokinetics of tobramycin in patients with and without cystic fibrosis. Clin Pharmacokinetics 2012 (Accepted: October 2012)