Analytical approach:
This economic evaluation was based on the modified version of a published Markov model which depicted the natural course of depression and the impact of the treatment strategies. The time horizon of the analysis was two years. The authors stated that the perspectives of the health-care payer and society were adopted.
Effectiveness data:
The clinical data were mainly derived from a double-blind, multi-centre, randomised controlled trial (RCT) assessing the efficacy and tolerability of venlafaxine, namely the Prevention of Recurrent Episodes of Depression with venlafaxine XR for Two Years study. This study was used to estimate the probabilities of changing health states (risk of a depressive episode and probability of remission). A meta-analysis was used in the sensitivity analysis to estimate risk of relapse and recurrence. The death rates were taken from Swedish life tables.
Monetary benefit and utility valuations:
The utility data were derived from a Swedish naturalistic observational study, namely the Health Economic Aspects of Depression In Sweden (HEADIS) study using the EuroQol-5D health status questionnaire.
Measure of benefit:
Quality-adjusted life-years (QALYs) were used as the summary benefit measure and the annual discount rate was 3%.
Cost data:
The economic analysis included the costs of drugs, hospitalisations, outpatient visits, and productivity losses. The data on costs and quantities of resources were derived from the HEADIS study. The cost of venlafaxine was based on the drug price listed in the National Pharmaceutical Drug Price list. The costs were in US dollars ($) and the price year was 2005. All costs were discounted at an annual rate of 3%.
Analysis of uncertainty:
A probabilistic sensitivity analysis was undertaken in order to address the issue of uncertainty in the model inputs. This was conducted using a bootstrapping technique that generated confidence intervals around mean cost and benefits and allowed the construction of an acceptability curve. A deterministic one-way sensitivity analysis investigated the robustness of the model outcomes to variations in key model inputs. The costs of adverse events, which were not statistically different between groups, were included in the sensitivity analysis. Similarly, the sensitivity analysis also considered a scenario in which patients could recover from a relapse in the maintenance treatment phase. Assumptions on re-relapses and recovery were made using data from a meta-analysis of RCTs. Finally, a long-term analysis was also carried out based on an extended model.