The Use of Decision-Analytic Models in Parkinson's Disease

Article excerpt

Parkinson's disease (PD) is a chronic neurodegenerative disorder affecting approximately 1% of the population aged >60 years worldwide, with a rising prevalence with age.[1] The progressive course of PD is associated with increasing morbidity and mortality. Disease progression in PD is caused by ongoing degeneration of dopaminergic and other neurons, characterized by symptoms of resting tremor, rigidity, bradykinesia (slowness of movement) and postural instability.[2] In the absence of recognized, reliable biological markers, disease states and disease progression in PD have been defined and measured by the presence and/or severity of motor symptoms. The natural history of PD is also characterized by an increasing incidence of non-motor symptoms, including dementia, depression, behavioural problems and sleep disorders, and autonomic dysfunction, such as gastrointestinal and genitourinary problems.[3]

There are currently no interventions that prevent or modify the lifelong progression of PD. Many PD motor symptoms can initially be controlled by dopamine replacement therapy using medications such as levodopa, which is converted into dopamine in the brain. However, long-term use of levodopa is associated with motor fluctuations (fluctuating periods of motor performance where the tremor, rigidity and bradykinesia recur at times such as end-of-dose 'wearing off' or random 'on/off' fluctuations) and dyskinesias (jerky, involuntary movements) that eventually add to the burden of PD. Non-pharmacological approaches are also available to manage the motor and non-motor symptoms of PD. These include surgical stimulation of brain neurons, rehabilitation, physiotherapy, occupational therapy, speech therapy and behavioural and psychological interventions.[4]

The physical disability, morbidity and psychosocial difficulties associated with disease progression in PD negatively impact the health-related quality of life (HR-QOL) of people with PD. As populations age, PD is expected to represent an increasingly significant healthcare burden on individuals and healthcare systems.[5] Against the growing demand for healthcare and support from those with PD, it is inevitable that future health technology developments (as well as existing health technologies) for PD will be subject to scrutiny due to healthcare resource constraints and difficult priority-setting conditions.

Cost-effectiveness analysis is now commonly used to inform decisions about competing treatment options based on a comparison of relative benefits and costs. However, the longer-term empirical data needed to inform healthcare decisions for PD (such as disease progression, morbidity and mortality, disability in activities of daily living, home care requirements, institutionalization, HR-QOL) are rarely provided by clinical trials or observational studies. Indeed, it is unlikely that data from randomized controlled trials, even those with concurrent collection of health economic data, will be sufficient to satisfy the needs of decision makers who require a synthesis of all the evidence relevant to their specific decision-making context.[6] In the absence of empirical data, modelling methods have been used to estimate the future costs and future consequences of alternative treatment strategies. Where data are not available on longer-term outcomes, decision-analytic models can provide a logical mathematical framework to extrapolate intermediate outcomes (i.e. clinical endpoints) from a range of sources, including clinical trials, to final outcomes relevant to decision makers.[7,8]

The literature on decision-analytic modelling in PD is relatively undeveloped. Two recent reviews, one of decision-analytic models in PD[9] and another of the cost effectiveness of pharmacotherapies in early PD,[10] identified only eight published studies. Both reviews found considerable methodological differences between the published models that impeded any comparison of findings. …