A state-transition Markov model, with a one-year cycle and a lifetime horizon, was developed to capture the ongoing fracture risk over time. The authors reported that the perspective of the German statutory health insurer was adopted.
: A literature search for meta-analyses and randomised controlled trials (RCTs), published between 2005 and 2007, was conducted in the PubMed database. The efficacy results from RCTs were pooled using the Mantel-Haenszel method. Other model parameters, such as the incidence of fractures, mortality, and sensitivity and specificity of diagnostic tests, were from published studies and official reports. The main measure of effectiveness was the incidence of a fracture.
Monetary benefit and utility valuations:
The utility values were from published studies. For the health state of no fracture, the utility values were derived, using a time trade-off questionnaire, from a sample of the general population. For the health states of fractures, the utility values were derived, using the European Quality of Life (EQ-5D) questionnaire and expert opinion.
Measure of benefit:
The primary measure of benefit was the quality-adjusted life-year (QALY) and these were discounted at an annual rate of 3%.
The analysis included the cost of screening and diagnosis, medication, treatment of fractures, and general health care during additional years of life. The costs of screening, diagnosis, and medication were from estimates in German guidelines, price lists for out-patient treatment, and a public database of medication costs. The medication resource use was adjusted according to an adherence rate derived from German prescription data. In-patient costs were a weighted average of diagnosis-related group data. Out-patient treatment costs were from a large random-sample survey. The price year was 2006 and all costs were reported in Euros (EUR). They were adjusted for inflation using the German consumer price index and they were discounted at an annual rate of 3%.
Analysis of uncertainty:
A one-way sensitivity analysis of all the model parameters was carried out. A probabilistic sensitivity analysis was also performed, by assigning probabilistic distributions to the model parameters. A Monte Carlo simulation was completed and the results were presented on cost-effectiveness acceptability curves.