|
Abstract:
|
The major obstacles towards making inferences from analysis of effects of drugs using observational data include a bias due to non-random assignment of patients to treatment modalities, which theoretically cannot be eliminated, and uncertain dependency relationships between analyzed variables. At any moment during treatment of chronic conditions, we iassess neither a severity of the disorder, nor an outcome of treatment. Rather it is a momentary functional state of the patient under the influence of therapy ("Treatment-Outcome Complex" - TOC). Analysis of cross-sectional and longitudinal patterns of TOC in the framework of Non-Random, RTC, Random, Titration, Incomplete Titration, and Mixed Models explores additional information to characterize relationships between treatment and outcome. The inferential opportunities, as well as limitations of this approach will be discussed.
|
