Analytical approach:
A published Markov model, with a lifetime horizon, was updated to simulate the clinical and economic impact of the three strategies. The authors stated that the perspective of the third-party payer was adopted.
Effectiveness data:
The relevant sources of evidence were selected to include clinical trials, meta-analyses, and observational studies. The efficacy of each preventive treatment, which was the reduction in CHD events for aspirin and the reduction in upper gastrointestinal bleeding for the PPI, was the key input. The relative reduction in CHD with aspirin was from a meta-analysis of clinical trials, while the reduction in gastrointestinal bleeding with the PPI was from one small randomised controlled trial. Framingham risk equations were used to model the natural history of CHD without aspirin. The authors selected the most appropriate estimates from the available evidence. The mortality data were from US life tables.
Monetary benefit and utility valuations:
The utility values were mainly from the literature and most of the studies used the time trade-off technique. Authors’ opinions were used where there was no published evidence available.
Measure of benefit:
Quality-adjusted life-years (QALYs) were the summary benefit measure and they were discounted at an annual rate of 3%.
Cost data:
The economic analysis included the medical costs of aspirin, omeprazole, and out-patient visits, and the annual costs for gastrointestinal bleeding, angina, stroke, and myocardial infarction. The calculation of the annual costs was based on data from the Healthcare Cost and Utilization Project's Nationwide Inpatient Sample, which contained national representative hospitalisation data. The other costs were from published sources and average wholesale prices. They were in US dollars ($), for the price year 2009, and a 3% annual discount rate was applied.
Analysis of uncertainty:
One- and two-way sensitivity analyses were carried out on selected parameters, using published or assumed ranges of values. A probabilistic sensitivity analysis was carried out, using second-order Monte Carlo simulation, with predetermined probability distributions for the model inputs. Cost-effectiveness acceptability curves were generated.