A decision tree, with a one-year time horizon, and a Markov model, with a 60-year time horizon, were developed to assess the costs and outcomes for each strategy for a hypothetical cohort of 100,000 individuals. The disease states included chronic hepatitis, cirrhosis, hepatocellular carcinoma, and death. The authors reported that a government perspective was adopted.
The effectiveness data were from various sources. A screening database of the research foundation that initiated the intervention provided the data for the screening rates and positive test rates with the intervention. A review of medical records and a questionnaire survey of one regional hospital in central Taiwan, selected from the 15 hospitals that participated in the intervention, were used for the screening rates in usual practice. Published studies were used for the annual incidence of hepatocellular carcinoma and its associated mortality. The main outcome measures were the number of hepatocellular carcinoma cases detected and the number of hepatocellular carcinoma deaths averted.
Monetary benefit and utility valuations:
The utility weights were from a published study of the burden of disease and injury in Australia. This study used Dutch weights to measure the severity of a wide range of health conditions.
Measure of benefit:
The measures of benefit were life-years and quality-adjusted life-years (QALYs) gained. These were discounted at an annual rate of 5%.
The analysis included the costs of screening, diagnosis, treatment, and follow-up of hepatocellular carcinoma. The unit cost estimates were from a number of national databases including the National Health Insurance Benefit Schedule and the screening programme's administrative data. The price year was 1997, and all costs were reported in Taiwan dollars (TWD). The costs were discounted at an annual rate of 5%.
Analysis of uncertainty:
One-way sensitivity analysis was performed and the results were presented in a tornado diagram. Multivariate sensitivity analysis was performed, using disease progression assumptions from previous studies. Probabilistic sensitivity analyses, with 2,000 iterations, were performed. A scenario was analysed to estimate the impact of expanding the programme to a national level.