Aspect of Model	Key Assumptions	Justification
Model states and transitions	Disease progression (eg, Hoehn and Yahr scale stage) and mortality are not affected by “on-demand” treatment	“On-demand” treatment is used when needed as an acute treatment of symptoms and is not expected to influence overall disease progression and mortality
	Probability of death was assumed to be based on age and disease stage at baseline and at the start of each model cycle	Presumably, patients that are older and/or have more severe PD (as per modified Hoehn and Yahr scale) have a higher probability of death
Costs	Costs associated with “OFF” time were independent of comparator arms, except indirectly through the effectiveness of “on-demand” treatment in reducing “OFF” time	It is not expected that costs associated with “OFF” time would be otherwise associated with treatment arms
	The underlying costs of PD were assumed to be the same across comparator arms and were not modeled explicitly	It is not expected that there would be systematic differences in the underlying costs of PD associated with different treatment arms
	The utilization and cost of maintenance (eg, carbidopa/levodopa) and “ON-extender” (ie, adjunctive) treatments (eg, dopamine agonists, COMT inhibitors, MAO-B inhibitors) were assumed to be the same across treatment comparator groups	The utilization of maintenance and “ON-extender” treatments may vary from patient to patient; however, it is not expected that there would be any systematic differences among the cohorts of patients initiating each of the respective treatments
	DACON assumed to be the same for all “on-demand” treatments	“On-demand” treatments are used when needed as an acute treatment of symptoms. Given that treatment costs are a function of how often patients use the treatment, the DACON is assumed equal to allow for more fair comparisons
Utilities	Utilities associated with “OFF” time are independent of comparator arms, except indirectly through the effectiveness of treatments in reducing “OFF” time	It is not expected that utilities associated with “OFF” time would be otherwise associated with treatment arms
	The underlying utility of PD was assumed to be the same across comparator arms and was not modeled explicitly	It is not expected that there would be differences in the underlying utility of PD associated with different treatment arms