Exploring the cost-effectiveness of a Dutch collaborative stepped care intervention for the treatment of depression and/or anxiety when adapted to the Australian context: a model-based cost-utility analysis

Aims Depression and anxiety are among the most common mental health conditions treated in primary care. They frequently co-occur and involve recommended treatments that overlap. Evidence from randomised controlled trials (RCTs) shows specific stepped care interventions to be cost-effective in improving symptom remission. However, most RCTs have focused on either depression or anxiety, which limits their generalisability to routine primary care settings. This study aimed to evaluate the cost-effectiveness of a collaborative stepped care (CSC) intervention to treat depression and/or anxiety among adults in Australian primary care settings. Method A quasi-decision tree model was developed to evaluate the cost-effectiveness of a CSC intervention relative to care-as-usual (CAU). The model adapted a CSC intervention described in a previous Dutch RCT to the Australian context. This 8-month, cluster RCT recruited patients with depression and/or anxiety (n = 158) from 30 primary care clinics in the Netherlands. The CSC intervention involved two steps: (1) guided self-help with a nurse at a primary care clinic; and (2) referral to specialised mental healthcare. The cost-effectiveness model adopted a health sector perspective and synthesised data from two main sources: RCT data on intervention pathways, remission probabilities and healthcare service utilisation; and Australia-specific data on demography, epidemiology and unit costs from external sources. Incremental costs and incremental health outcomes were estimated across a 1-year time horizon. Health outcomes were measured as disability-adjusted life years (DALYs) due to remitted cases of depression and/or anxiety. Incremental cost-effectiveness ratios (ICERs) were measured in 2019 Australian dollars (A$) per DALY averted. Uncertainty and sensitivity analyses were performed to test the robustness of cost-effectiveness findings. Result The CSC intervention had a high probability (99.6%) of being cost-effective relative to CAU. The resulting ICER (A$5207/DALY; 95% uncertainty interval: dominant to 25 345) fell below the willingness-to-pay threshold of A$50 000/DALY. ICERs were robust to changes in model parameters and assumptions. Conclusions This study found that a Dutch CSC intervention, with nurse-delivered guided self-help treatment as a first step, could potentially be cost-effective in treating depression and/or anxiety if transferred to the Australian primary care context. However, adaptations may be required to ensure feasibility and acceptability in the Australian healthcare context. In addition, further evidence is needed to verify the real-world cost-effectiveness of the CSC intervention when implemented in routine practice and to evaluate its effectiveness/cost-effectiveness when compared to other viable stepped care interventions for the treatment of depression and/or anxiety.


Introduction
Background and objectives 3 Provide an explicit statement of the broader context for the study.

See Introduction
Present the study question and its relevance for health policy or practice decisions.

Methods
Target population and subgroups 4 Describe characteristics of the base case population and subgroups analysed, including why they were chosen. See Table 1 and  Supplementary Tables S4  and S7   Incremental costs and  outcomes   19 For each intervention, report mean values for the main categories of estimated costs and outcomes of interest, as well as mean differences between the comparator groups. If applicable, report incremental cost-effectiveness ratios.
See Table 1 in the main  manuscript and  Supplementary Table S7 Characterising uncertainty 20a Single study-based economic evaluation: Describe the effects of sampling uncertainty for the estimated incremental cost and incremental effectiveness parameters, together with the impact of methodological assumptions (such as discount rate, study perspective). A total of 12 RCTs examined stepped care models involving a progressive increase in treatment intensity. Of these, we excluded: four involving prevention [3][4][5][6]; one targeting elderly outpatients with diabetes [7]; one not generalizable to the Australian setting [8]; one involving treatment of anxiety only [9]; one involving treatment of depression only and with insufficient data to model changes in resource use [10]; one due to confounding of the treatment effect -i.e., the first step of the stepped care model (watchful waiting) was administered to both intervention and control arms [11]; one involving an active treatment in the control arm [12]; and one with insufficient data to inform cost-effectiveness modelling [13]. First, GBD 2019 data on the prevalence of major depressive disorder, dysthymia and anxiety disorders were adjusted to only include individuals who were eligible to receive the stepped care intervention. This included those who: consulted with a GP for mental health problem in the past 12 months; did not have 12-month psychotic symptoms; did not have a 12-month bipolar disorder; and did not attempt suicide in the past 12 months. Second, the prevalence of depression was calculated by combining data on the prevalence of major depressive disorder and dysthymia, after accounting for comorbidity between the two disorders. Third, the prevalence of anxiety was adjusted to exclude the prevalence of obsessive-compulsive disorder and posttraumatic stress disorder, which were both included in the GBD 2019 prevalence estimates. Fourth, the GBD 2019 prevalence estimates did not account for comorbidity between depression and anxiety. A correction was made by calculating the overlap between the prevalence of depression and anxiety; and assigned all co-morbid cases to a single disorder based on the proportion who reported depression or anxiety as the primary disorder.

Text S5. Methods used to estimate disability weights for depression, anxiety and background morbidity
Disability weights for remitted and unremitted cases of depression/anxiety were estimated as follows.
Disability weights were obtained by severity level for depression and anxiety from the Global Burden of Disease Study 2019 (GBD 2019) [14]. Data on the proportion of mild, moderate and severe cases (i.e., the severity distribution) were also sourced from GBD 2019 [14]. Both sets of data are presented in the Australian population (see Table 1 in the manuscript) [16] was used to estimate the weighted-average PYLD rate due to other causes of disease and injury (across all sexes and age groups). This was estimated to be 0.146 (SE: 0.005) and became the disability weight for remitted cases of depression/anxiety.

Overview
Intervention pathways for the CSC intervention and the CAU comparator are presented in Figure 1 and Hourly wage rates for psychiatrists and psychologists who provide training were taken from the ABS Survey of Employee Earnings and Hours [21], plus a 30% loading to incorporate on-costs -i.e., wage loadings to account for administration costs, personal leave, superannuation, etc.

CSC intervention -Guided self-help
Guided self-help for people with a mild-to-moderate depression/anxiety (CSC 1a) involved patients participating in a 4-month self-help course comprising five 45-minute sessions, with guidance from a GP mental health nurse at a primary care clinic or at home. Patients with depression participated in the 'Coping with Depression' course [22], while those with anxiety participated in the 'Stresspac' course [23]. Both courses consisted of workbooks with psychoeducation and CBT exercises. This was applied to the Australian context by assuming that all GP mental health nurses and eligible patients had their own copy of the relevant self-help book costed at the recommended retail price (range of prices obtained via Google search).

CSC intervention -Medication algorithms
In addition to receiving guided self-help, patients with a moderate depression/anxiety (CSC 1b) were offered antidepressant medication according to disorder-specific medication algorithms. All patients with depression/anxiety were initially prescribed a selective serotonin reuptake inhibitor (SSRI). In the case of non-response or low tolerability, patients were switched to: a serotonin-norepinephrine reuptake inhibitor (SNRI), tricyclic antidepressant (TCA) or tetracyclic antidepressant (TeCA) if they had depression; an SNRI or other SSRI if they had generalised anxiety disorder; or another SSRI if they had panic disorder with/or without agoraphobia. In the model, the probability of switching medications following first-line treatment with an SSRI was based on data from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study [24]. The timing of the switch was modelled using a uniform distribution -i.e., assuming the switch had an equal probability of occurring across all time points during the intervention step. The monthly unit price for each drug class was estimated using data from the Pharmaceutical Benefits Scheme (PBS) [25]. to those applied to patients in the first step. Furthermore, patients who only received psychotherapy were assumed to see a clinical psychologist 8 times (range: [6][7][8][9][10]; while patients who received psychotherapy plus antidepressants were assumed to see a psychiatrist 8 times (range: [6][7][8][9][10]. The point estimate for the number of psychologist/psychiatrist visits was based on commonly used thresholds for minimally adequate treatment [27], while the range was based on current Medicare guidelines which subsidise a maximum of 10 psychotherapy sessions each year [28].

CAU comparator
The types of treatments received by patients in the CAU comparator, included: treatment with antidepressants only (CAU 1); referral to specialised mental healthcare (CAU 2); and receiving no treatment (CAU 3). Patients who received antidepressant medication were costed using the same medication algorithms as those outlined in the CSC intervention. Likewise, patients receiving specialised mental healthcare were costed in a similar manner to those in the CSC intervention. Patients who dropped out were included among those who received no treatment. All patients who received no treatment were assumed to incur zero costs, which is a conservative assumption given that additional costs in the CAU comparator will lead to a lower (i.e., more cost-effective) ICER. The duration of each treatment in the CAU comparator was 8 months.

Benzodiazepine use
GPs were free to prescribe benzodiazepines to all patients on top of their existing treatments in the CSC intervention and CAU comparator over the 8-month study period (CSC 9 and CAU 9, respectively). The cost of benzodiazepine use was calculated in a similar manner to the medication algorithms outlined previously, but based on the proportion who reported using benzodiazepines in each group.

Unit cost of health professional visits (CSC intervention and CAU comparator)
All medical consultations with a GP, psychologist or psychiatrist were costed using weighted average of all relevant fees from the Medicare Benefits Schedule (MBS) [28].  Psychologist visits among patients who receive psychotherapy only (0-8 months) No. of psychotherapy sessions: 6 (range: 8 -10) Cost per psychologist session: $106.75

Assumption [28]
Parameter Value and uncertainty range (when applicable) Uncertainty distribution