OBJECTIVES: To compare the real-world cost-effectiveness of two different depression treatment programs in daily practice.

METHODS: A naturalistic study was used to compare the cost and outcomes in terms of OQ45 improvement scores of two depression treatment programs. One uses a flexible approach with fixed treatment duration and is shared-decision based; the other is algorithm-guided treatment. Data were analyzed from two Dutch mental healthcare providers providing specialized mental healthcare for patients with moderate to severe depression; the dataset contained data from 2010 to 2017 (n=2,710). Propensity score matching was conducted to correct for selection bias and confounding by indication. Matching was done in a 3:1 ratio based on gender, age and baseline-symptom score. Costs were calculated from the healthcare perspective, while effects were calculated as the differences of two consecutive routine outcome monitoring scores. Incremental cost-effectiveness ratios were calculated and the uncertainty of the results was assessed using bootstrapping and univariate and sensitivity analysis.

RESULTS: The flexible program incurred lower costs while generating inferior outcomes than the algorithm-guided treatment program. The ICER was €343.73 per unit on the outcome score, indicating that for each drop-in OQ45 improvement accepted, a cost saving of approximately €340 was realized. Uncertainty analyses confirmed the robustness of these findings.

CONCLUSIONS: The results showed no clear superiority in terms of the cost-effectiveness of either program. To further assess this, the willingness to pay for one unit of OQ45 improvement would be needed. The costs of the two programs were significantly different. This disparity might be due to the pre-defined treatment duration program indeed showing a shorter average treatment length and, therefore, lower costs. This came ,however, at a loss in effectiveness after correcting for selection by indication with the help of propensity score matching.