A Transitional Probability Model for Parkinson's Disease Motor States With Applications to Missing Data.

Background: Parkinson's disease (PD) is a progressive neurodegenerative disorder with significant disability. Subjects with advanced PD often suffer from motor complications that may interfere significantly with their daily activities. Levodopa (LD) in combination with a dopa decarboxylase inhibitor such as carbidopa (CD) is considered the gold standard in the treatment of PD. However, long-term treatment with LD often leads to the development of motor complications. Motor complications include motor fluctuations and dyskinesia. Motor fluctuations are states where the subject cycles between periods of "on" state where subjects are in improved mobility and "off" state where subjects are in impaired mobility. Dyskinesia are the involuntary and irregular twisting and/or turning movements.

Methods: A Markov transitional probability model is proposed to estimate the likelihood of staying in one state versus transitioning from one state to another.

Results: An application of the model to an example from a clinical trial investigating the effect of an extended-release carbidopa-levodopa (CD-LD) product versus an immediate-release CD-LD product is illustrated.

Conclusion: A Markov transitional probability model can be used to model the likelihood of staying in one state versus transitional from one state to another. The model can also be used as a basis for multiple imputation of missing data.