Premium
Randomized dose ranging trials; assessment of statistical power for various design and analysis strategies
Author(s) -
Grahnen A.,
Karlsson K.,
Karlsson M. O.,
Jonsson N. E.
Publication year - 2004
Publication title -
clinical pharmacology and therapeutics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.941
H-Index - 188
eISSN - 1532-6535
pISSN - 0009-9236
DOI - 10.1016/j.clpt.2003.11.334
Subject(s) - nonmem , statistics , pharmacodynamics , randomized controlled trial , statistical power , randomization , pharmacokinetics , mathematics , sample size determination , medicine , econometrics , pharmacology
Purpose To investigate the hypothesis that the most beneficial trial design and analysis strategy is to randomize as far as possible from, and use an independent variable as close as possible to the clinical endpoint in the mechanistic pathway of drug response. Methods The hypothesis is tested using simulations where the basic setup consists of a one compartment pharmacokinetic model at steady state conditions and a sigmoidal Emax pharmacodynamic model. The studies are simulated with randomization schemes on dose and concentration, and the analysis of the data is performed with dose and concentration as independent variables, respectively. The significance criterion used for the statistical power of the studies is the log‐likelihood ratio test. All the simulations and the analyses were made in the NONMEM software. Results. The randomized dose‐controlled trial with concentration as the independent variable in the analysis yield the greatest power followed by the randomized concentration‐controlled trial with concentration as the independent variable in the analysis. Conclusions. Our hypothesis seems to hold under the simulation conditions used. Real‐life factors that may influence this result are: correlation between pharmacokinetic and pharmacodynamic parameters, patient drop‐outs and model misspecification/complexity. From a safety point of view it may be necessary to have close control of the drug concentration levels that patients are exposed to and this may necessitate a randomized concentration‐controlled trial design despite its somewhat expected lower statistical power. Clinical Pharmacology & Therapeutics (2004) 75 , P87–P87; doi: 10.1016/j.clpt.2003.11.334