Action Tendencies and Prognosis in Major Depressive Disorder

Overview

Predicting the prognosis and treatment responses in individuals with major depressive disorder (MDD) is currently based on trial and error, because some treatments work for some individuals, but not others. Novel predictors of prognosis and treatment response in MDD can add value to the development of targeted treatments and the stratified approaches to improve long-term outcomes of individuals with MDD. This study uses a novel virtual-reality-based measure of blame-related action tendencies and combines this with established predictors of treatment response and prognosis in individuals with MDD.

Developing clinical and psychological markers which characterise the large subgroup of patients with major depression who do not benefit from serotonergic antidepressants offered as first line treatment is of utmost clinical importance but has so far not been achieved. Clinical and psychological indicators would be ideal in predicting non-response to antidepressant medications, given their wide availability. So far, however, these measures have failed to provide accurate predictions at the individual level. Novel predictors of prognosis and treatment response in major depressive disorder (MDD) can add value to the development of targeted treatments and stratified approaches to improve long-term outcomes of individuals with MDD. The psychological underpinnings of patients' response to treatment have been an important direction of research. As proposed by the revised learned helplessness model, one central cognitive vulnerability to MDD is the tendency to excessively blame oneself for negative events occurring in one's life. Consistent with the theory, previous studies have demonstrated the importance of self-blaming bias both as a vulnerability factor and as a symptom of depression. For example, it has been shown that individuals with remitted MDD had increased self-contempt biases compared to healthy control participants in a recent study. As MDD is a life-long diagnosis, understanding the differences between remitted MDD and healthy controls could help to identify vulnerability traits associated with MDD. Thus, the finding of self-blaming biases in remitted MDD demonstrate the potential role of self-blame as a vulnerability trait and a novel cognitive marker for MDD that remain present during remission and possibly constitute vulnerability for recurrence. However, one limitation in previous studies is that people might have experienced difficulties distinguishing their moral emotions such as shame and guilt under certain circumstances. Previous measures assessing these emotions largely depended on participants' subjective rating and are thus problematic in terms of differentiating the moral emotions. In addition, these measures fail to address the adaptive or maladaptive nature of moral emotions. As proposed by Tangney, moral emotions can be either adaptive and maladaptive, and this difference is possibly determined by an individual's different action tendencies associated with their moral emotions. Action tendencies describe an implicit cognitive and motivational state before an action is taken. It was suggested that adaptive action tendencies, such as feeling like apologizing, were associated with self-blaming emotions such as guilt, and maladaptive action tendencies such as feeling like hiding and creating a distance from oneself were associated with shame. However, an empirical investigation of the associations between action tendencies and self-blaming emotions is lacking. Further investigations of this topic are important for understanding the potential role of action tendencies as a novel measure of self-blame and its association to the vulnerability to MDD. It is important to develop measures of action tendencies with a high ecological validity. In previous studies, our research group has developed a computerised task that measures action tendencies and used it to predict prognosis in MDD. It was found that this task can predict recurrence risk in people with MDD, showing a large effect size (Cohen's d=.96). However, there were two major limitations. First, the task used in a verbal format and included abstract descriptions of scenarios (e.g. "You act stingily towards your friend"), which makes the task dependent on how well participants can imagine the scenarios. Second, the lack of immersiveness of the task made it difficult to engage, which may limit the task's ecological validity. Virtual reality (VR)-based assessment is a new paradigm for cognitive evaluation compared to the traditional paper-and-pencil or computerized assessment. VR scenarios were suggested to be promising tools for cognitive assessments and have been demonstrated as safe for the assessment of anxiety disorders and depression. Importantly, the interactive and immersive nature of virtual reality makes it possible to develop a cognitive task that is engaging and has a higher ecologically validity, which would be ideal for identifying novel cognitive markers of MDD outcomes. Thus, this study will aim to employ a virtual reality task to measure blame-related action tendencies. There are three major research questions of this study 1. Is MDD associated with a higher proneness towards maladaptive action tendencies, such as self-distancing and hiding, compared with a non-MDD control group? 2. Are maladaptive self-blame-related action tendencies associated with a poor prognosis for current major depressive disorder when treated as usual in primary care? 3. Can maladaptive self-blame-related action tendencies be used to predict prognosis in MDD at the individual level when combined with other predictors using a nested elastic-net regularised doubly-cross-validated regression model? (https://github.com/AndrewLawrence/dCVnet). This will use both primary and secondary predictors in the same model. Our proposed primary predictors of prognosis for major depressive disorder are the following (these will be used in a non-regularised multiple regression model): 1. Percentage of trials during which hiding was chosen as measured by the VR-task 2. Percentage of trials during which self-distancing was chosen as measured by the VR-task 3. Autonomy total score as measured by the Personal Style Inventory 4. Sociotropy total score as measured by the Personal Style Inventory 5. Maudsley Staging Model total score 6. Compliance with treatment as measured on an ordinal scale (how regularly have you taken your antidepressants over the last month at the prescribed dose? 0=Never, 1=Some of the time, 2=More than half the time, 3=Most of the time, 4=Almost every day, 5=Every day) 7. Social support received as measured by the Social Support Scale 8. Baseline depression score as measured by the Self-rated Quick Inventory of Depressive Symptomatology (QIDS-SR-16) and the Maudsley-Modified Patient Health Questionnaire -9 (MM-PHQ-9, two separate models will be run for using either QIDS-SR-16 or the MM-PHQ-9 as the outcome variable). 9. Baseline anxiety symptoms as measured by the Generalised Anxiety Disorder assessment 10. Optimisation of antidepressant medication during the follow-up period on an ordinal scale (0=no new antidepressant/stopping current antidepressant or lowering its dose, 1=increase from effective dose to a higher dose, 2=increase from ineffective to effective dose /or change to another antidepressant at effective dose) Other potential predictors in secondary analyses for a non-regularised regression model: 1. Coping mechanism as measured by the Brief COPE 2. Type of treatment obtained during the four months (e.g. SSRI or Non-SSRI) 3. Affective Lability 4. Early life trauma 5. Physical co-morbidity 6. Age 7. Gender 8. Education 9. Age of onset 10. Number of previous episodes

Conditions

Interventions

Eligibility

Locations (1)

Outcomes