A Markov cost-utility analysis of escitalopram and duloxetine for the treatment of major depressive disorder

ABSTRACT

Objective: To estimate the costs and quality-adjusted life weeks of duloxetine and escitalopram.

Research design: A probabilistic Markov cost-utility analysis with a time horizon of 1 year using data from placebo controlled randomized clinical trials for both products.

Methods: Efficacy was defined as remission of depressive symptoms and converted to utilities. Side effects were incorporated using rates from clinical trials and converted to utilities to define treatment effective­ness. The effectiveness outcome was quality adjusted life weeks (QALWs). Estimates of effectiveness (efficacy and side effects) used beta distributions and costs used gamma distributions. Using a managed care perspective, medication costs and physician office visits were included in the model, along with costs associated with treatment failure. Antidepressant costs were obtained using average wholesale price minus 20%. Physician visit costs were obtained from the 2006 US Medicare fee schedule for physician services. A Monte Carlo simulation was conducted using 1000 trials with both first- and second-order sampling.

Results: Over 1 year, the estimated mean quality-adjusted life weeks was 41.0 (95% confidence interval [CI]: 40.7–41.3) for escitalopram and 38.2 (95% CI: 37.9–38.4) for duloxetine. The mean annual total medical cost for escitalopram was $907 (95% CI: $894–$919) and $1633 (95% CI: $1614–$1654) for duloxetine. Limitations to this analysis include using separate studies examining the efficacy and adverse events of either escitalopram or duloxetine, assuming the switch, augmentation, and titration rates for duloxetine to be similar to escitalopram, and using utility estimates from published literature for the antidepressant adverse events.

Conclusion: This analysis suggests that escitalopram was more effective in terms of QALWs and less costly than duloxetine for treatment of depression.