Cost-Utility and Cost-effectiveness of MoodSwings 2.0, an Internet-Based Self-management Program for Bipolar Disorder: Economic Evaluation Alongside a Randomized Controlled Trial

Background: Internet-delivered psychosocial interventions can overcome barriers to face-to-face psychosocial care, but limited evidence supports their cost-effectiveness for people with bipolar disorders (BDs). Objective: This study aimed to conduct within-trial cost-effectiveness and cost-utility analyses of an internet-based intervention for people with BD, MoodSwings 2.0, from an Australian health sector perspective. Methods: MoodSwings 2.0 included an economic evaluation alongside an international, parallel, and individually stratified randomized controlled trial comparing an internet-based discussion forum (control; group 1), a discussion forum plus internet-based psychoeducation (group 2), and a discussion forum plus psychoeducation and cognitive behavioral tools (group 3). The trial enrolled adults (aged 21 to 65 years) with a diagnosis of BD assessed by telephone using a structured clinical interview. Health sector costs included intervention delivery and additional health care resources used by participants over the 12-month trial follow-up. Outcomes included depression symptoms measured by the Montgomery-Åsberg Depression Rating Scale (MADRS; the trial primary outcome) and quality-adjusted life years (QALYs) calculated using the short-form 6-dimension instrument derived from the 12-item version of the short-form health survey. Average incremental cost-effectiveness (cost per MADRS score) and cost-utility (cost per QALY) ratios were calculated using estimated mean differences between intervention and control groups from linear mixed effects models in the base case. Results: In total, 304 participants were randomized. Average health sector cost was lowest for group 2 (Aus $9431, SD Aus $8540; Aus $1=US $0.7058) compared with the control group (Aus $15,175, SD Aus $17,206) and group 3 (Aus $15,518, SD Aus $30,523), but none was statistically significantly different. The average QALYs were not significantly different among the groups (group 1: 0.627, SD 0.062; group 2: 0.618, SD 0.094; and group 3: 0.622, SD 0.087). The MADRS scores were previously shown to differ significantly between group 2 and the control group at all follow-up time points (P<.05). Group 2 was dominant (lower costs and greater effects) compared with the control group for average incremental cost per point decrease in MADRS score over 12 months (95% CI dominated to Aus $331). Average cost per point change in MADRS score for group 3 versus the control group was dominant (95% CI dominant to Aus $22,585). Group 2 was dominant (95% CI Aus $43,000 to dominant) over the control group based on lower average health sector cost and average QALY benefit of 0.012 (95% CI –0.009 to 0.033). Group 3, compared with the control group, had an average incremental cost-effectiveness ratio of dominant (95% CI dominated to Aus $19,978). Conclusions: Web-based psychoeducation through MoodSwings 2.0 has the potential to be a cost-effective intervention for people with BD. Additional research is needed to understand the lack of effectiveness for the addition of cognitive behavioral tools with the group 3 intervention.