A discrete-event simulation had been developed for donepezil in the UK (Getsios, et al. 2010, see 'Other Publications of Related Interest' below for bibliographic details) and this was adapted for Germany. The time horizon was 10 years. The authors stated that the perspectives of the German statutory health insurance and society were adopted.
The clinical and effectiveness data were from published studies. The disease progression and the treatment effects, including the effects of disease on behaviour and function, were based on data from the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) and seven clinical trials of donepezil for Alzheimer’s Disease. These trials had to be Phase III, had to include a measure of baseline MMSE, and had to include at least one outcome for this model.
Monetary benefit and utility valuations:
The utility for patients and their carers was included. The patients’ utilities were estimated based on a published regression equation, which used data collected, using the European Quality of life (EQ-5D) questionnaire, from 272 patients in Nordic countries. The carers' utilities were from a donepezil trial where the carers' quality of life was assessed, using the Short Form (SF-36) health survey, and transformed to utilities based on a published study.
Measure of benefit:
Quality-adjusted life-years (QALYs) were the measure of benefit. Future benefits were discounted at an annual rate of 3%.
The direct costs were those of memantine, donepezil, out-patient visits, dementia tests, laboratory tests, hospital care, and institutional care. The indirect costs were those of informal care, which was the time spent by unpaid carers, valued using German hourly wage rates. The drug costs were from the German Rote list. The health care costs were from German Tariffs or published studies. The price year was 2008 and all costs were in Euros (EUR). Future costs were discounted at an annual rate of 3%.
Analysis of uncertainty:
One-way sensitivity analysis was performed on the key model parameters. A probabilistic sensitivity analysis was undertaken, by fitting probability distributions to model parameters. The results were presented in tables and scatter plots.