Analytical approach:
This economic evaluation was based on a single study. The time horizon was 24 weeks. The authors stated that a societal perspective was adopted.
Effectiveness data:
The clinical data on treatment effect were derived from a 24-week, double-blind, multinational (nine European countries), fixed-dose, randomised controlled trial (RCT). The sample included 295 patients: 144 allocated in the escitalopram group and 151 in the duloxetine group. However, only 143 escitalopram and 146 duloxetine patients were included in the intention-to-treat analysis. The length of follow-up was 24 weeks. The key clinical outcome was the change in the Sheenan Disability Scale (SDS) score.
Monetary benefit and utility valuations:
None.
Measure of benefit:
The primary benefit measure was the change in the SDS score. Secondary benefit measures were treatment response, defined as a decrease in Montgomery-Asberg Depression Rating Scale (MADRS) score of 50% or more, and remission, defined as a MADRS score of 12 or less.
Cost data:
The economic analysis included the costs associated with physician visits, visits to other health care professionals, hospitalisations (in different wards), drugs, and sick leave. The unit costs were reported, but quantities of resources were not given. The resource use data were derived from the RCT using the full analysis set of patients who had at least one valid health economic assessment after the baseline measurement (113 escitalopram and 110 duloxetine patients). In particular, sick leave due to depression was assessed alongside the clinical trial using a questionnaire implemented at baseline, and at weeks 4, 12 and 24. The health care costs were derived from official UK sources such as National Reference costs or the Personal Social Services Research Unit. The costs of sick leave were estimated using National Statistics on average wage rates. All costs were in UK pounds sterling (£) and the price year was 2006. Instances of missing data and between-group comparisons were dealt with by means of statistical tests. A multivariate analysis was also carried out in order to investigate the impact of treatment choice on total costs as well as on sick leave.
Analysis of uncertainty:
The issue of uncertainty was addressed in three deterministic sensitivity analyses. In the first, specialist visits were reduced. In the second, the acquisition cost of duloxetine was set to zero (to reflect the most conservative case). In the third, the cost of sick leave was varied from £0 to £100 per day. A bootstrapped approach was used to generate 95% confidence interval (CI)s around the incremental costs and benefits.