A decision tree was developed to determine the positive and negative predictive values of disease. Markov models were then used to determine the costs and utilities of the natural history of dementia with different causes. All models were probabilistic. The causes were Alzheimer's disease, vascular dementia, normal-pressure hydrocephalus, subdural haematoma, and brain tumour. The time horizon was lifetime. The authors stated that the perspective was that of the Ontario Ministry of Health and Long-Term Care.
The clinical data included disease prevalence, the sensitivity and specificity of the tests, disease progression probabilities, and the absolute and treatment relative risks of outcomes, such as stroke or death. The transition probabilities were from a published model that used data from the Consortium to Establish a Registry for Alzheimer's Disease (CERAD; see Other Publications of Related Interest). The diagnostic accuracy of the CCC rules came from a primary diagnostic study, which had CT as the gold standard. Prevalence was estimated based on a published meta-analysis and a Canadian cohort study. Diagnostic utility was estimated based on a clinical evidence-based analysis, conducted by Health Quality Ontario.
Monetary benefit and utility valuations:
Disease progression and the treatment effect were modelled in terms of their impact on cognitive function and rates of admission to an institution. The utilities for the six health states, representing severity of illness (mild, moderate or severe) and institutional status (community or nursing home) were modified from the CERAD publication to create three states for severity of illness. These scores were weighted by the proportion of patients in an institution, from another publication. The utilities were derived using the Health Utilities Index 2. The disutility for a stroke was applied in the vascular and mixed dementia models.
Measure of benefit:
The summary measure of benefit was quality-adjusted life-years (QALYs), which were discounted at a rate of 5% annually.
The costs included the drugs, imaging, procedures and care for dementia of various causes. The drug costs were from Ontario Drug Benefit data, and included donepezil and aspirin. The imaging costs were from the Ontario Case Costing Initiative. The costs of procedures and care for dementia of each cause were from a variety of published sources, including the Ontario Case Costing Initiative. All costs were reported in 2012 Canadian dollars (CAD), inflated where necessary. They were discounted at a rate of 5% per year.
Analysis of uncertainty:
Univariate, multivariate, and probabilistic sensitivity analyses were undertaken to assess the effects of uncertainty on the results. The probabilistic sensitivity analyses varied all model parameters simultaneously.