Background and rationale
Depression affects at least 350 million people worldwide [
1] and is a leading cause of non-fatal burden of disease [
2]. It is costly to individuals in terms of relationships and functioning and to society in terms of direct medical costs and costs due to loss of individual productivity [
3]. Despite significant investments in mental health globally, there is no evidence of a reduction in the burden of disease associated with depression [
4]. One of the biggest challenges facing mental healthcare systems is the need to develop efficient methods of allocating clinically effective treatment in a cost-effective way to the people that need them most [
5].
The majority of depression cases are identified, treated, and followed up in primary care [
6]. However, general practitioners (GPs) have been criticized for both under- and over-diagnosing, and treating, depression [
7‐
10]. For example, only 16% of Australians with case level depression or anxiety receive an adequate “dose” of an evidence-based intervention consistent with treatment guidelines [
9]. On the other hand, antidepressant prescriptions far outnumber patients for whom such medication is indicated [
11].
Multi-country studies report that 24–55% of patients in primary care waiting rooms meet screening criteria for being “probably depressed” [
12]. However, among this population of “probably depressed,” a range of illness trajectories exist which contribute to the difficulty experienced by practitioners in making a diagnosis and treatment recommendation [
13‐
18]. It may be that the heterogeneity of clinical presentation which characterizes depression in the primary care setting is leading to the simultaneous problems of both over- and under-diagnosis and treatment.
Currently, there is a mismatch in primary care between patient need and the depression care received, possibly as a result of poor treatment allocation. For example, delivery of intensive interventions to people with minimal or mild symptoms is unnecessarily costly and risks medicalizing normal fluctuations in mood [
19]. Conversely, without a targeted intensive intervention, people likely to experience severe and chronic symptoms are likely to experience significant disability, which could have been avoided [
20,
21].
Stepped care models, in which patients are, in the first instance, provided with the least time- and resource-intensive intervention that will be effective [
22], have been promoted as a potential solution to the problem of poor treatment allocation. Although limited empirical evidence exists as to their effectiveness [
23], these models make intuitive sense and feature in both clinical guidelines and policy directives in Australia [
24] and the UK [
22]. Currently, a key obstacle to the implementation of stepped care models is the lack of effective treatment allocation tools to guide GPs in matching the intensity of treatment to a patient’s needs. In particular, current recommendations focus on matching treatment to the patient’s current symptom severity, rather than patient’s likely course of illness in the future. This is out of step with the management of other conditions (e.g. cardiovascular disease or cancer), which routinely take prognostic factors into account when deciding upon treatment recommendations. Further, it contrasts with calls for research, policy, and practice to make prognosis-based intervention a priority [
25]. To date, there has been no quick and systematic way for GPs to identify depression outcomes that a particular person is likely to experience in the future and recommend treatment accordingly.
One option for systematizing treatment recommendations is to use a clinical prediction tool (CPT). CPTs are based on a prognostic model that uses clinical and non-clinical information to estimate an individual’s risk of a specific outcome [
26]. The prognostic model is applied in clinical practice using the CPT which stratifies patients into different treatments according to their estimated risk [
27]. While CPTs are common in many fields of medicine, they are not readily available for use in mental-healthcare settings. [
28]
To address this gap, we wanted to develop a simple, easy-to-use CPT to assist primary care clinicians to triage patients presenting with depressive symptoms and allocate to appropriate treatment. First, we investigated whether an existing prognostic model for depression could be used to build the CPT. We identified several prognostic models that have been developed to predict current [
29,
30] or future major depression [
31‐
34] or treatment response [
35‐
37]. However, none of these prognostic models were found to be suitable for incorporating into a CPT which could be easily administered in routine care [
38].
Therefore, we developed a novel prognostic model using data from the
diamond cohort study [
39] to predict depressive symptom severity at three months [
38]. It comprises 17 items assessing depressive symptom severity at baseline as measured by the Patient Health Questionnaire-9 (PHQ-9) [
40]: sex; current anxiety; history of depression; presence of chronic illness affecting daily functioning; self-rated health; living alone; and perceived ability to manage on available income. Based on an individual’s score, he or she is stratified into one of three groups based on predicted depressive symptom scores at three months; namely, minimal/mild (those predicted to have a PHQ score of ≤ 10 at three months), moderate (PHQ > 10 and < 13), and severe (≥13). Cutoffs for the three groups were established during the development of the
diamond CPT and are explained in full elsewhere [
38]. In the intervention being tested in the current study, individuals are then:
(1)
Presented with feedback reflecting their responses to the CPT;
(2)
Provided an opportunity to set priorities and reflect on motivation to change; and
(3)
Presented with an evidence-based treatment recommendation matched to group classification.
The presentation of feedback and treatment recommendation was informed by the principles of motivational interviewing [
41] and an iterative development process employing user-centered design principles to ensure the information is presented in a way that is meaningful and engaging for participants [
42].
Objectives
The Target-D randomized controlled trial (RCT) aims to test whether using the
diamond CPT to tailor treatment recommendations to an individual’s predicted depressive symptom severity is a clinically effective and economically efficient way of reducing depressive symptoms, relative to usual care. This paper presents the study protocol for the Target-D RCT, adhering to the SPIRIT guidelines for intervention trial designs ([
43]; see Additional file
2 for SPIRIT checklist).
The primary objective of the Target-D trial is to determine if using the diamond CPT to triage individuals with depressive symptoms into symptom severity-appropriate treatment reduces depressive symptoms at three months compared with usual care.
Secondary objectives are to: (1) test whether individuals in the intervention and comparison arms differ in depressive symptom severity at 12 months, quality of life, anxiety symptoms, self-efficacy, and health service use at three and 12 months; (2) determine whether the outcomes differ between the two study arms within each of the three depressive symptom severity groups; and (3) evaluate the cost-effectiveness of the new model of care compared to usual care.