Objective: The overall objective of the present study is to examine the cost-effectiveness of 1) unguided Internet-based treatment for depression, 2) guided Internet-based treatment for depression, and 3) blended treatment for depression compared to TAU for depression over a 5-year time-horizon from a societal perspective using a Markov model.
Hypothesis: Internet-based treatments will be at least as effective as TAU in improving depressive symptoms and Quality-of-Life, and may result in decreased costs at the long-term due to the decreased time therapists spend with patients.
Study design: A Markov-model representing the course of depression over a 5-year time horizon will be developed. Societal costs and effects (e.g. QALYs) are assigned to each health state in the model.
Study population/Datasets: Adults with elevated depressive symptoms will be simulated in the model. Existing individual-participant data (IPD) will be used to populate the model. The model parameters after one year will be identified through a systematic literature search in electronic databases.
Interventions: We will include three different forms of Internet-based treatments in the model that are distinguished based on the different levels of contact with a therapist offered to patients. In the unguided Internet-based treatment, patients progress independently through the online treatment modules without support from a therapist. In the guided Internet-based treatment, patients receive feedback from a therapist in the form of e-mail after completing each online module. Finally, the blended treatment combines guided Internet-based treatment and face-to-face psychological treatment.
We will combine existing data from different Internet-based psychotherapies (e.g. cognitive behavioral therapy (CBT) and problem solving treatment) in the main analysis, and we will conduct sensitivity analyses stratified for type of therapy. Previous research examined the patients’ perspective of Internet-based treatments and they positively evaluated their experience (4, 5). Finally, a recent study including various stakeholders in Europe (such as government bodies, care providers, service-users, funding/insurance bodies, technical developers and researchers) indicated that stakeholders demonstrated great acceptability towards digital treatments and identified cost-effectiveness as the primary incentive for implementation (6).
Usual care/Comparison: TAU for depression generally follows existing national guidelines. In most studies, there is unrestricted access to treatment from mental and non-mental healthcare providers. In the model, we will stratify between patients receiving treatment in primary and in specialized mental healthcare. As a first step, only data from TAU offered in the Netherlands will be included in the model. As a sensitivity analysis, pooled TAU data from different countries will be used to investigate the robustness of the results.
Outcome measures: Clinical and cost outcomes will be included in the model. Clinical outcomes regard remission from depression and Quality-Adjusted Life-Years (QALYs). Costs will be measured from the societal perspective (healthcare costs and productivity losses) and the healthcare provider perspective (only healthcare costs).
Sample size calculation/Data analysis: Patients simulated in the model will move from one health state to another based on transition probabilities, which will be different for each of the three forms of Internet-based treatment and TAU. Probabilistic and deterministic sensitivity analysis will be performed to account for the uncertainty around the model parameters.
Cost-effectiveness analysis/Budget impact analysis: Cost-effectiveness analyses will be performed from both a societal and healthcare perspective. We will calculate incremental cost-effectiveness ratios (ICERs) by dividing the difference in total costs between the intervention and TAU by the difference in clinical effects after five years. Probabilistic sensitivity analyses will be performed to estimate uncertainty surrounding the outcomes. Cost-effectiveness planes will illustrate this uncertainty. We will estimate cost-effectiveness acceptability curves, which demonstrate the probability that the interventions are cost-effective in comparison with TAU for a range of different values of willingness to pay (7). The Markov model will also be used to estimate the budget impact in the long term from a government perspective.
Time schedule: 1-5 months: Building the structure of the model; 6-12 months: Data preparation, model development and validation; 13-22 months: Data analyses, sensitivity analyses; 23-28 months: Results presentation, publications.
Implementation: An implementation plan will be developed in order to maximize potential wider roll-out and sustainable delivery of the most cost-effective internet-based depression treatment in the future.