Related Experiment Videos
Modelling a spontaneously reported side effect by use of a Markov mixed-effects model
Per-Henrik Zingmark1, Matts Kågedal, Mats O Karlsson
1Department of Clinical Pharmacology, AstraZeneca R&D, Södertälje, Sweden. per-henrik.zingmark@astrazeneca.com
Journal of Pharmacokinetics and Pharmacodynamics
|November 12, 2005
Summary
This study introduces a novel method for analyzing side-effect data, incorporating Markov elements to accurately model changes in severity over time. The approach effectively predicts side-effect progression in clinical trials.
Area of Science:
- Pharmacometrics
- Clinical Pharmacology
- Biostatistics
Background:
- Spontaneously reported side-effect data in clinical trials often exhibit changes in severity.
- Analyzing this data requires methods that account for the temporal dependence of observations.
Purpose of the Study:
- To present a novel analytical method for side-effect data with spontaneously reported changes in severity.
- To evaluate the performance of a mixed-effects model incorporating Markov elements for analyzing such data.
Main Methods:
- A clinical study involving 12 healthy volunteers investigated a CNS-specific side-effect.
- Subjects received individualized drug infusions to achieve target concentrations.
- Side-effect occurrence and severity (0=none, 1=mild, 2=moderate/severe) were self-reported.
Main Results:
- A mixed-effects model with Markov elements was developed to estimate the probability of side-effect severity transitions.
- The Markov model demonstrated superior predictive performance compared to a model without Markov elements, as confirmed by posterior predictive checks.
- Observed transitions included 0->1 (24), 0->2 (11), 1->2 (23), 2->1 (1), 2->0 (32), and 1->0 (2).
Conclusions:
- Incorporating Markov elements into the analysis of categorical side-effect data adequately predicts the observed time course.
- This approach is valuable for analyzing categorical data where inter-observation dependence is a significant factor.